file
stringlengths 27
28
| text
stringlengths 19
1.51M
|
|---|---|
PMC009xxxxxx/PMC9099898.txt |
==== Front
Sensors (Basel)
Sensors (Basel)
sensors
Sensors (Basel, Switzerland)
1424-8220
MDPI
10.3390/s22093542
sensors-22-03542
Article
Evaluation of Driver’s Reaction Time Measured in Driving Simulator
https://orcid.org/0000-0001-7574-3384
Čulík Kristián 1*
https://orcid.org/0000-0002-3225-2961
Kalašová Alica 1
https://orcid.org/0000-0003-3240-722X
Štefancová Vladimíra 2
Stopka Ondrej Academic Editor
Jaśkiewicz Marek Academic Editor
Poliak Milos Academic Editor
1 Department of Road and Urban Transport, University of Zilina, Univerzitna 1, 01026 Zilina, Slovakia; [email protected]
2 Department of Railway Transport, University of Zilina, Univerzitna 1, 01026 Zilina, Slovakia; [email protected]
* Correspondence: [email protected]; Tel.: +42-141-513-3507
06 5 2022
5 2022
22 9 354218 3 2022
02 5 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
This article evaluates the driver’s reaction times in a driving simulator environment. The research focused mainly on young drivers under the age of 26, who cause many accidents. Each participating driver provided basic information later used for mathematical-statistical analysis. The main advantage of driving simulators is limitless usage. It is possible to simulate situations that would be unacceptable in real road traffic. Therefore, this study is also able to examine drunk driving. The main goal of the article is to evaluate if gender, practice, or alcohol significantly affected the reaction time of 30 drivers. We also focused on drinking before driving for a smaller number of the drivers; ten of them performed driving under the influence of alcohol. For these mathematical-statistical purposes, we used a one-sample t-test, a paired-samples t-test, an independent-sample t-test, and a correlation analysis together with the assessment of its statistical significance.
driver behavior monitoring
driving simulators
road safety
ground vehicle safety
MŠVVŠ SR—VEGA1/0178/22 This paper was developed with support of a project: MŠVVŠ SR—VEGA No. 1/0178/22 Kalašová, A.: Basic research of the sharing economy as a tool for reducing negative externalities.
==== Body
pmc1. Introduction
In general, more than 90% of traffic accidents are caused by human failure [1,2,3]. A pedestrian, cyclist, or another road user can make a mistake. The driver causes the highest number of accidents. Fatigue and stress also contribute to road accidents [4]. Many studies focus on the behavior of professional drivers and their accidents [5,6]. Alcohol-related road accidents are also serious [7]. They often include young drivers, who are reckless and prone to alcohol consumption. Scientific studies define ‘young drivers’ differently. Most often they include drivers between the ages of 18 and 26. However, the upper limit may vary in some studies. For example, the authors in [8] focused on a group of 18 to 20-year-old drivers, the authors of the Romanian study in [9] focused on 18 to 24-year-old drivers, and the authors of the Belgian study in [10] focused on 17 to 24-year-old drivers. On the contrary, Dénommée et al. [11] focuses on 16 to 24-year-old drivers. Several studies have focused on the combination of young drivers and alcohol. In Greece, for example, the legal age for driving under the influence of alcohol is 18. A study of 241 young Greek drivers (aged between 18 and 24) found that young drivers whose dominant lifestyle was alcohol consumption were at a higher risk of being involved in an accident. In addition to alcohol, lack of driving experience also contributes to higher accident rates for young drivers [12].
Another study [13] has showed that male individuals are more likely to drink than female. Additionally, the authors in [14] have pointed out that after alcohol consumption, young male drivers tend to engage in risky behavior and aggression. Other studies, for example [15], have pointed to the differences in the characteristics of male and female drivers. Male risk perception is lower, which means they are more careless drivers [16,17]. A study [18] examined drivers’ reactions under the influence of alcohol. The main aim of this research was to investigate the effects of alcohol consumption on stopping behavior at stop signs and at an intersection with traffic lights. It was a laboratory experiment that also used a driving simulator. The results showed a significant difference between the mean deceleration values for sober drivers and drivers under the influence.
Alcohol impairs the judgment ability of the driver. It causes delayed responses, such as an increased reaction time when encountering a stimulus on the road [19,20]. A study [21] has also shown interesting results; blood alcohol concentration levels of 0.03%, 0.05%, and 0.08% resulted in 36%, 53%, and 94% incremental changes in the reaction times of drivers encountering a pedestrian crossing.
Many studies have showed that blood alcohol concentration is related to accident risk [22,23]. Young drivers who drive under the influence of alcohol have a higher risk of accident involvement at all blood alcohol concentration levels [24,25]. Many authors have proved that sober drivers represent a lower accident risk [26,27].
It is possible to find many theoretical studies that deal with traffic accidents. They focus on traffic accidents correlated with the driver’s age or alcohol consumption. It is problematic to carry out research in real road traffic in this area. It is possible to use a driving simulator to cover this research gap. Young drivers in the age group of 18 to 26 are particularly problematic; hence, this demographic was chosen.
The research described in this article had several main objectives. The first task was to measure the accurate values of the drivers’ reaction times. The evaluation aimed to consider differences between the reaction times of:drivers that do not expect obstacles on the road and drivers on the second attempt when they know the scenario,
male and female drivers,
sober drivers and drivers under the influence of alcohol.
The secondary objective was to describe and evaluate the best possibilities for enhancing the simulation validity. Simulator improvements can bring better results and eliminate simulator sickness.
2. Materials and Methods
One of the research aims was to measure the reaction times of drivers under the influence of alcohol. This measurement would not be possible in real road conditions. Therefore, a driving simulator was used. As it is a complex device, it is described in detail in the following section.
2.1. Driving Simulators
A driving simulator can be simply defined as a device that simulates the driving of a road vehicle in a virtual environment as realistically as possible [28,29,30].
An advanced driving simulator should reproduce all stimuli that the driver perceives when driving. The simulator software should ensure [31]:mathematical model of vehicle behavior,
virtual reality—image and sound,
scene control/event generator,
platform movement control,
driving record,
tools for evaluating driver’s behavior.
The University Science Park at University of Žilina has a training driving simulator available for research purposes. The essential difference between the research driving simulator and the training driving simulator is in their purposes:
(a) Training driving simulators, usually used in driving schools, are devices used for training new drivers. The term ‘training driving simulator’ can be found in Methodical Instruction no. 22/2005 on technical requirements for training driving simulators [32] from 26 September 2005. This guideline sets out the basic requirements for training driving simulators. There is, for example, a requirement for a projection area of at least 180 × 130 mm2 and other conditions for sound, vehicle dynamics, and virtual environment. Researchers use training driving simulators occasionally, for example, in the study [33] or for evaluation of training effectiveness in driving schools [34].
(b) Research driving simulators have a wide range of use in research institutes, universities, and the automotive industry. Their main advantage is the ability to adapt to the current requirements of the experiment. It means that they must have an open system in which it is possible to change the virtual environment, vehicle, and its properties. They can be used, e.g., for driver fatigue research [35,36], crossing intersections [37], lane change behavior [38,39,40], driver error rate [41], driver glare [42], or human-vehicle interaction research in general [43]. Simulators also have great potential in autonomous vehicle research [44,45,46].
Driving simulators have various advantages and disadvantages depending on the individual versions and construction arrangements. The advantages of driving simulators include, for example [20]:Versatility and new developments at reduced cost. Simulators can be easily and economically configured to research many human factor issues.
Experimental control and measurement. Driving simulators allow researchers to control experimental conditions and measure any parameters. For example, a study [47] measured steering wheel angles while changing lanes when the gap between vehicles in the target lane was constant or decreasing, as well as maneuvering times. These were subsequently projected into a graphic form.
Safety. Driving simulators provide a safe environment for driver research.
Driving simulators also have several disadvantages—weaknesses that every researcher should consider as a limitation of the study:Validity. Simulators cannot duplicate the whole world due to its details and complexity. Therefore, this raises the question of to what extent the research on a simulator is credible. Some authors have described this issue in the article [48], which compares 44 studies. Another comprehensive study is in [20]. The virtual environment can be very different or very similar to real conditions. A study [39] has evaluated the similarity between real driving and driving in a simulator. Interestingly, it showed similar results between simulation and reality (similar measured speeds in turning and connecting lanes).
Costs. Driving simulators have relatively high acquisition costs, but very low operating costs.
Simulator sickness [49]. Usually, driving simulators with a motion system or poor graphic quality cause nausea. These impacts on the human body are so-called Simulator Adaption Syndrome (SAS). The authors of [50,51] have written that the source of SAS was the difference between the performances of the driving simulator and the real vehicle. Many studies, for example [52,53], have compared the negative effects of static and motion simulators. According to them, the most common symptoms are nausea (feeling sick), dizziness, vomiting, eye pain, fatigue, and anxiety. Interestingly, they are less common in dynamic (moving) simulators.
In our research, the SNA–211 REN training driving simulator was used for experimental driving [54,55]. The participants in this study had not yet had experience with this driving simulator. Therefore, they had to be trained before their performance could be measured. Drivers had about 10 min to familiarize themselves with the environment and the simulator controls. Each driver could try starting, braking, turning, and shifting between gears.
2.2. Other Usedequipment
The central element of the research is the driving simulator (Figure 1). It is a replica of a truck cabin equipped with a gear stick with a small button. It can switch gears between the lower row (1st to 4th gears) and the upper row (5th to 8th gears).
There were three people in the laboratory during the individual measurements: the supervisor of the driving simulator, the person responsible for data recording, and the tested driver. The other participants involved in the research task were in another room during the measurements. It was necessary to ensure that they had not seen the obstacles and virtual environment before their ride.
The drivers’ reactions encompassed the moment when an unexpected situation occurred until the moment when they activated the brake pedal. The computer program Corel VideoStudio recorded the environment and braking. It was installed on the driving simulator computer. The recording was also made using an external camera (Figure 2a). An AlcoCheck X400L (Figure 2b) was also used to test alcohol in the drivers’ breath. The obtained values were not used in statistical evaluation.
2.3. Measurement Methodology
In our research, we evaluated the reaction times of 30 drivers. Half of them were male drivers and half of them were female drivers. The average ages of the male and female drivers were 22.27 years (SD = 1.58) and 22.33 years (SD = 1.84), respectively. The average mileages of the male and female drivers were 45,200 km (SD = 34,526) and 16,700 km (SD = 15,780), respectively.
Measurement was divided into five parts, which are described in detail in the following lines.
2.3.1. First Part of Experiment: Unexpected Obstacle
For the first measurement, the drivers were focused, but did not expect an obstacle. The obstacle in the virtual environment was an animal running from behind a tree across the road. The reaction time of the driver was recorded as the time interval between the obstacle animation trigger and the moment of brake activation. We did not evaluate the success of the obstacle stop in this article.
2.3.2. Second Part of Experiment: Expected Obstacle
The second measurement took place at the same time as the first. The scenario for measuring reaction times continued, but the drivers were already expecting another obstacle. Therefore, the reaction time should be even shorter than in the case of the first sudden obstacle. For this measurement, the drivers were more careful. They peripherally checked the edge of the road.
2.3.3. Third Part of Experiment: Impressions from the Simulation
Given the need to increase the validity of the simulation in the future, after completing these two measurements, all test drivers completed a questionnaire. It aimed to record the perceived quality of the simulation. For most drivers, this was the final part of the measurement. A minority also took part in the fourth and fifth parts of the experiment: driving after drinking alcohol.
2.3.4. Fourth Part of Experiment: Drunk Driving 1
For technical reasons, not all drivers performed further tests. Therefore, only 10 drivers were chosen for the last part of the measurement. All drivers involved in the fourth and fifth parts of the experiment agreed to drink alcohol. Without accounting for the differences in weight and other factors, each driver consumed 200 mL of 35% alcohol. Each driver had 10 min for this consumption. After a subsequent 10-min break, his or her ride began in a virtual environment. The drivers’ responses to a sudden obstacle was measured again. The scene was the same as in previous measurements.
2.3.5. Fifth Part of Experiment: Drunk Driving 2
This measurement was taken after a long break from the time of first alcohol consumption (30 min) to increase its influence on the drivers’ behavior and attention. The drivers went through the same scene. Here we expected a more significant deterioration of their reactions and a higher level of alcohol in their breath.
Before the measurement process, it was necessary to ensure:Drivers signed the Informed Consent agreement before the experiment.
Familiarization with the course of research.
Drivers who had to undergo drunk driving had to consume no alcohol before driving, be in approximately the same sleep mode (students from the same study group who get up at the same time). The had to consume the same food (lunch together with the same menu).
The procedure of the measurement itself was as follows: Familiarization of the driver with the driving simulator (10 min).
Start of the scenario for measuring reactions no.1 (15 min).
Measurement of the time interval between the trigger start time and the activation of the brake pedal.
3. Start of the reaction measurement taken during scenario no. 2 (5 min).
Measurement of the time interval between the trigger start time and the activation of the brake pedal.
4. Completion of the simulation validity questionnaire (10 min).
The questionnaire was in paper form.
5. Alcohol consumption.
6. 10 min break.
7. Start of the scenario to measure reactions no. 3 (5 min).
Measurement of the time interval between the trigger start time and the activation of the brake pedal.
8. Break 15 min.
9. Start of the scenario to measure reactions no. 4 (5 min).
Measurement of the time interval between the trigger start time and the activation of the brake pedal.
10. End of measurement.
2.4. Evaluation Methods
Using software, such as SPSS or the Data Analysis add-in in MS Excel, it is possible to evaluate reaction times in a modern and simple way. However, in this article, the authors used the methods described below, calculated in the traditional way, to demonstrate the possibilities of statistical evaluation. The following methods will therefore be used to evaluate reaction times [56]:One-sample t-test. We use one-sample t-test in experimental situations where we know the mean value µ0 of the basic set. We can then consider this as a constant. In this experiment, we verify the hypothesis that the experimental sample comes from a population that has the same mean as this known constant. We test the null hypothesis: H0: µ0 = const. We start the test from the data of the monitored sample, which we assume comes from a population with certain parameters µ and s2 and further from the known mean value of the base set m, which is equal to a certain (known) constant.
Two-sample t-test. This test evaluates experiments where we do not know the mean of the base set and compares only two sets of sample data. These data can be represented by either two measurements performed repeatedly on one group of individuals (paired experiment) or by two independent groups of measurements (non-paired experiment). In the case of a two-sample t-test, we test the null hypothesis: H0: µ1 = µ2. A two-sample t-test can be:
Independent-sample t-test, which compares the data formed by two independent selections, i.e., that they come from two different groups of individuals. Typically, this is a comparison of the values of the experimental group (where the experimental intervention was applied) and the control group (where the experimental intervention was not performed).
Dependent-sample t-test, which compares the data that make up “paired variation series,” i.e., where they come from those subjects that were subjected to two measurements.
Correlation analysis. This simple correlation analysis deals with the evaluation of the dependence of two random variables and emphasizes the intensity of the relationship rather than the examination of variables in a cause-effect relationship (regression) [57].
Correlation coefficient significance test. A common task in mathematical statistics is to find out whether the random variables X and Y are correlated or not. The value of the correlation coefficient depends on the elements in the random selection. If the value of the correlation coefficient is close to zero, we want to verify whether it is only random (caused by random selection) or whether it is really a linear independence. The linear independence test is used for verification. We express the hypothesis H0: ρ = 0 against the alternative hypothesis H1: ρ ≠ 0 to find out whether the random variables X, Y are correlated or not [58].
2.5. Reaction Time Values and Hypothesis
In this article, we use a t-test to test the hypotheses described below. First, we verify that the mean value of the reaction times in the first measurement is equal to 0.8 s, which is located in the middle of the table below, signifying the concentrated drivers who do not expect the stimulus (Table 1). We assume that the mean value is less than the table data, and we verify this hypothesis at the significance level α = 5%.
We further verify the hypothesis that the mean reaction time for male and female drivers does not differ at the 5% significance level. We verify this hypothesis using an independent-sample t-test, performed for the first and second measurements.
Third, we use the Dependent-Sample t-test to verify the research hypothesis that the mean reaction time before alcohol consumption is less than the mean value of the reaction time after alcohol consumption. We verify this hypothesis on 20 values measured during the first and second phases, and 20 values measured under the influence of alcohol in the fourth and fifth phases of the research.
In the last part, we calculate the correlation coefficient between the mileages and the driver’s reaction time. In addition, we verify the statistical significance of the correlation coefficient. We decide whether the detected dependence (regardless of the value of the correlation coefficient) is statistically significant or random. We perform all the above tests at the significance level α = 5%.
3. Results
In total, we were able to measure the reaction times of 30 drivers at two points in time. Drivers also completed a questionnaire on simulation impressions. All data measured during these experiments are presented in the individual tests in the following sections.
3.1. One-Sample t-Test
We have organized the individual tests into different sections, so that, in addition to the results themselves, we can also point out our methods for statistical testing. We performed the one-sample t-test with the data visualized in the Figure 3.
The procedure for testing with the one-sample t-test is as follows:Determination of hypotheses:
H0A : The mean value of the reaction times of concentrated drivers is 0.80 s: µ = 0.80 s.
H1A : The mean value of the reaction times of concentrated drivers is less than 0.80 s: µ < 0.80 s.
Calculation of test criterion (1), in which x¯ is the arithmetic mean of all measured values of reaction times (0.732) and μ0 is chosen as 0.80 s. In the equation, n is the number of all measurements (30) and S is standard deviation (0.166). (1) t=x¯−μ0S·n
After substituting, we find that the test criterion has a value −2.252.
The critical field is presented in the formula (2), where α is the level of significance, in our case 0.05. Subsequently, we looked in the quantile tables of the Student’s distribution for the value t0.9529, which is 1.699. (2) Wα=t≤−t1−αn−1
Subsequently, we can complete the formula as follows (3):(3) Wα=−2.252≤−1.699
From this, we can conclude that the critical field is fulfilled and thus, we reject the original hypothesis H0A and accept the alternative hypothesis H1A.
The answer in this case is: the mean value of the reaction times of the concentrated drivers is less than 0.80 s at a significance level of 5%.
3.2. Independent-Sample t-Test
An independent-sample t-test was the second test that we used. This test for independent selections is a commonly used method to evaluate the difference in the averages of the two groups. The test is used for small samples on the assumption that both groups have a normal distribution and the variances of these groups do not differ significantly. The input data are visualized in Figure 4.
The procedure for testing with the independent-sample t-test is as follows:Determination of hypotheses:
H0B : The mean reaction time of male and female drivers is the same: µM = µW.
H1B : The mean value of the reaction time of male and female drivers is not the same: µM ≠ µW.
Calculation of test criterion (4), in which x1¯ and x2¯ are the arithmetic means of all measured values of reaction times of males and females, respectively. The number of measurements is denoted as n1 and n2. In the case of this test, the measurement values may also be different, as they are not paired. S1 and S1 are the standard deviations, which are S1=0.147 and S2=0.136. (4) u=x1¯−x2¯S12n1+S22n2
After substituting, we find that the test criterion has a value +1.910.
The critical field in this case is given by (5), where α is the level of significance (0.05). Subsequently, we looked in the quantile tables of the normal distribution N (0,1) for the value u0.975, which is 1.960.(5) Wα=u≥u1−α2
Subsequently, we can complete the formula of critical field as follows (6):(6) Wα=1.910<1.960
From this, we can conclude that the critical field is not met and therefore we accept the original hypothesis H0B.
The answer in this case is: the mean reaction time of male and female drivers is the same at a significance level of 5%. However, as it can be seen, the test criterion is very close to the critical range.
3.3. Paired-Samples t-Test
The paired-samples t-test was the third one we used. This test compares the values of a variable for the same respondent in two different experimental conditions. In our case, we use this test to compare the reaction time before and after drinking alcohol. Analyzed reaction times are visualized in Figure 5.
The procedure for testing with the paired-samples t-test is as follows:1 Determination of hypotheses:
H0C : The mean value of the reaction times of the drivers in a sober state and under the influence of alcohol is the same: µS = µD.
H1C : The mean value of the reaction times of drivers in a sober state and under the influence of alcohol is not the same: µS ≠ µD.
2. Calculation of test criterion (7), in which D¯ is the arithmetic mean of all mutual deviations (differences) between two experiments. The number of measurements is denoted as n. It is also necessary to calculate the standard deviation SD from all values of the mentioned differences for the calculation of the test criterion. (7) T=n·D¯SD
After substituting, we find that the test criterion has a value −2.618.
3. In this case, the critical field is given by (8), where α is the level of significance (0.05). Subsequently, we looked in the quantile tables of the Student’s distribution for the value t0.9759, which is 2.262. (8) Wα=t≥t1−α2n−1
Subsequently, we can check (9) the fulfillment or non-fulfillment of the critical field:(9) Wα=2.618>2.262
From (9) we conclude that the critical field is fulfilled and thus we reject the original hypothesis H0C and accept the alternative hypothesis H1C.
4. The answer in this case is: at a significance level of 5%, it was shown that the mean values of the reaction times of drivers in a sober state and under the influence of alcohol are not the same.
3.4. Correlation Coefficient Test
The last test, a Correlation Coefficient Test, assesses the statistical significance of the correlation between two variables. It should be noted that relatively low values of correlation coefficients can be expected in traffic psychological research. According to [38], the interpretation of the correlation coefficient depends on the context. In field of physics, a correlation coefficient of 0.8 is very low; on the contrary, in the social sciences, it is a very high value. In 1988, Cohen [60] established the exact tool for the interpretation of correlation coefficients in psychological research:A correlation in the absolute value below 0.1 is trivial,
A correlation in the range of 0.1 to 0.3 is small,
In the interval of 0.3 to 0.5, the correlation is medium,
At values above 0.5, the correlation is high,
A correlation of 0.7 to 0.9 is very high,
A correlation in the range from 0.9 to 1.0 is almost perfect.
The correlation coefficient (10) measures the two-tailed linear dependence of two variables and takes values from the interval 〈−1;1〉. The following implications apply for the correlation coefficient:
rxy = 0 ⇔ variables X and Y are not linearly dependent,
rxy > 0 ⇔ there is a direct linear relationship between the variables X and Y,
rxy < 0 ⇔ there is an indirect linear relationship between the variables X and Y.
The sign of the correlation coefficient determines the direction of the dependence. The absolute value of the correlation coefficient reveals the strength of the linear association between the two variables. The closer the absolute value is to 1, the stronger the dependence. (10) rxy=n∑xy−∑x∑yn∑x2−∑x2·n∑y2−∑y2
In this case, we will calculate the correlation coefficient between the number of traveled kilometers x. (driving experience) and the reaction time of drivers y on the first attempt. The correlation coefficient can be very easily calculated in MS Excel using the CORREL function. In our study, the correlation coefficient was −0.430, which can be considered a non-proportional medium dependence. The described dependence is visualized in the Figure 6.
The procedure for testing the correlation coefficient is as follows:Determination of hypotheses:
H0D : There is no statistically significant linear relationship between the variables y and x.
H1D : There is a statistically significant linear relationship between the variables y and x.
Calculation of test criterion (11), in which r is the correlation coefficient calculated above and n is the number of all data pairs (30). (11) T=r·n−21−r2
After substituting, we find that the test criterion has a value −2.522.
The Critical Field is given by (12), where α is the level of significance (0.05). Subsequently, we looked in the quantile tables of the Student’s distribution for the value value t0.97528, which is 1.699. (12) Wα=t≥t1−α2n−2
Subsequently, we can add to the formula itself as follows (13):(13) Wα=2.522>2.048
From this, we can conclude that the critical field is fulfilled and thus, we reject the original hypothesis H0D and accept the alternative hypothesis H1D.
The answer is: At a significance level of 5%, it was shown that there is a statistically significant linear relationship between the variables y and x.
We have calculated all these tests in statistical software. The most commonly used program is SPSS, but it is also possible to perform these tests in MS Excel with the T.TEST function. The following figure (Figure 7) shows an example of the Independent Sample t-test setting in MS Excel (Array1—reaction times of male drivers, Array2—reaction times of female drivers, Tails 2—two-tailed test, type 2—the variances do not differ).
4. Discussion
This article describes the measurement of drivers’ reaction times in the driving simulator. Its essential goal was to point out the possibilities of statistical evaluation of the measured values. Due to the equipment available for this research, the driving simulator can be considered a limitation of the study. As can be seen from the graph in the following figure (Figure 8), the drivers also took part in a survey of the perception of virtual reality in the third phase of testing. In the survey, they evaluated what could make the ride more realistic (1 = the least significant factor, 5 = the most important improvement). It is clear from the picture that the graphics and behavior of the vehicle most significantly contribute to the perception of reality. Other related factors are the number of frames per second, the quality of the textures, and the traffic in the virtual environment [61]. Drivers also mentioned vehicle behavior and steering response. Another study [62] deals with the possibilities of eliminating such defects in driving simulators.
Figure 8 shows that drivers perceive the vibrations and movements of the simulator’s cab as most insignificant. Financially, these two requirements would be the most economically demanding. The authors of [63,64,65] write that cabin movements can increase the validity of the simulation. However, at the same time, they can cause negative feelings from driving. In these studies, the authors compared three simulator options: a fixed base platform with poor visibility, a fixed base platform with good visibility, and a motion base platform with good visibility. It is clear from the studies that most health problems, such as nausea, oculomotor, and disorientation, occur when using the motion platform.
In terms of the experiment itself, in this study, we were able to evaluate the reaction times of 30 drivers, which is not a sufficiently representative sample. In Slovakia, there are 244,663 registered drivers in the 17–24 age group. According to a sample size calculator, with a population of 244,663, a confidence level of 95%, and a confidence interval of 5, we would need a sample size of 384 people. However, this was not possible in our case. For this reason, we verified the statistical significance by testing hypotheses.
Another goal of the paper was to provide educational value. This article shows how it is possible to statistically evaluate data. Therefore, students at universities can use the described methods for their theses. For this reason, all four tests are performed not by software (through the so-called p-values), but by traditional calculations.
The research described in this article has several limitations. First, it would be necessary to ensure approximately the same physiological training for drivers, especially those who also took part in driving under the influence of alcohol. The drivers who also took part in the second part of the research were students from one study group, thus ensuring they had a similar duration of regular sleep and a similar diet. However, we could not completely monitor all consumed meals.
From the results and the literature, it is obvious that alcohol has a significant effect on drivers’ reaction times and overall behavior. We did not record detailed physiological data on the individual drivers during our research. A detailed comparison of measured values would be problematic. Characteristics, such as gender, physical condition, and especially the weight of a person, significantly affect the coping of the human body with the same dose of alcohol consumed. Therefore, for further research in this area, we recommend the physiological preparation of drivers (sleep, drug exclusion, etc.) and a thorough investigation and recording of the relevant characteristics of the individual persons involved. Other than age and gender, we did not evaluate any other drivers’ attributes. It is a limitation of the study.
Another limitation of the study is that we used the training driving simulator, which is a replica of the truck Renault Midlum. For measurements, a passenger car would be more appropriate. On the other hand, there was no difference in driving, because all drivers used only four high gears. There was no problem driving with only four gears and an unloaded vehicle.
This research provided several results. In the first test, the tabular value of the reaction time of the concentrated driver, who did not expect a stimulus, was not completely confirmed. The reaction times were shorter, probably due to the lack of distractors during the drive.
Psychological research in this paper did not consider the drivers’ hand preferences. According to recent studies, this area seems to be very interesting, and we can address it in our next scientific endeavor. Due to professional contributions, left-handed people used to be routinely excluded from studies. Hand preference is problematic, but at the same time it is a very useful variable that deserves its place in the deeper examination of human behavior.
Statistical testing also confirmed that the reaction time of male and female drivers is approximately the same. However, in this case, it is possible that similar reaction times occurred due to the presence of similar people in the selected sample. All the young drivers were students or graduates of a technical university. It means that this could have an impact on the results.
With the paired-samples t-test, we tested the hypothesis of prolonging the reaction times of drivers under the influence of alcohol. The reaction times were indeed even shorter in some attempts. However, in general, at a significance level of 0.05, it can be stated that the times are different and, of course, shorter in a sober state.
The last point of the evaluation was the correlation analysis. Calculating the correlation coefficient without assessing its statistical significance can bring misleading results. In our case, the correlation coefficient between driving experience and reaction time was a mean inverse value of −0.430. In the field of physics, this would be a low value. However, in traffic psychology, this represents a medium dependence. The Correlation Coefficient Test proved a statistically significant linear relationship between the above variables at the 5% significance level.
5. Conclusions
This article focused on the drivers’ reaction time measurements while driving in the simulator. For the measurement process, we formulated the following recommendations:Measurement accuracy is a critical factor because the reaction time is a short time interval.
It is necessary to avoid time delays caused by slow response time. These delays arose from hardware and should be avoided.
It is also crucial to ensure that individual respondents do not provide information about the process of the experiment.
Drivers should be in approximately the same psycho-physiological condition.
From the research results, we can formulate the following recommendations:
The consumption of alcohol before driving prolongs reaction time and thus increases the risk of an accident. Therefore, it is necessary to protect young drivers through prevention campaigns.
Drivers with higher mileage have a better reaction time, but only in some cases (correlation coefficient 0.430).
Concentration during driving significantly shortens the reaction time. Therefore, the main recommendation of the study is to maintain attention while driving.
From this article, the danger of drunk driving is evident. However, it is also clear from many other studies [65,66,67,68]. In terms of statistics, we have pointed out that the basic characteristics (median, mode, arithmetic mean, or standard deviation) are insufficient in similar research. It is necessary to assess statistical significance.
Author Contributions
Conceptualization, K.Č. and V.Š.; methodology, K.Č.; software, K.Č.; validation, K.Č., V.Š. and A.K.; formal analysis, K.Č.; data curation, K.Č.; writing—original draft preparation, K.Č. and V.Š.; writing—review and editing, K.Č., V.Š. and A.K.; visualization, K.Č.; supervision, A.K.; project administration, A.K.; funding acquisition, A.K. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Informed consent was obtained from all subjects involved in the study.
Data Availability Statement
All used data is available on request from the author.
Conflicts of Interest
The authors declare no conflict of interest.
Figure 1 Training driving simulator in SNA–211 REN. Source: Processed by authors.
Figure 2 Other equipment: (a) External video camera for recording; (b) AlcoCheck X400L. Source: Processed by authors.
Figure 3 Input data of one-sample t-test. Source: Processed by authors.
Figure 4 Input data of independent-sample t-test. Source: Processed by authors.
Figure 5 Input data of paired-samples t-test. Source: Processed by authors.
Figure 6 Input data of correlation analysis. Source: Processed by authors.
Figure 7 Independent sample t-test with T.TEST function in MS Excel. Source: Microsoft Excel.
Figure 8 Possibilities of improving driving simulator validity. Source: Processed by authors.
sensors-22-03542-t001_Table 1 Table 1 Drivers’ reaction times for different conditions. Source: [59].
Reaction Time [s] Driver
0.6–0.7 driver is attentive, focused, awaiting stimulus and ready to brake
0.7–0.9 driver is attentive, but does not expect a stimulus
1.0–1.2 driver has focused his or her attention on other activities related to driving (driving, preventing, sidewalk observation)
1.4–1.8 driver is inattentive (having fun with the passenger, etc.)
1.6–2.4 driver is indisposed (alcohol, illness, fatigue, etc.)
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. De Felice F. Petrillo A. Methodological approach for performing human reliability and error analysis in railway transportation system Int. J. Eng. Technol. 2011 3 341 353
2. Dhillon B.S. Methods for Performing Safety, Reliability, Human Factors, and Human Error Analysis in Nuclear Power Plants, 1st ed CRC Press Boca Raton, FL, USA 2017 63 88
3. Kahn C.A. Gotschall C.S. The economic and societal impact of motor vehicle crashes, 2010 (revised) Ann. Emerg. Med. 2015 66 194 196 10.1016/j.annemergmed.2015.06.011
4. Useche S. Cendales B. Gómez V. Work stress, fatigue and risk behaviors at the wheel: Data to assess the association between psychosocial work factors and risky driving on bus rapid transit drivers Data Brief 2017 15 335 339 10.1016/j.dib.2017.09.032 29214195
5. Cendales B. Useche S. Gómez V. Psychosocial work factors, blood pressure and psychological strain in male bus operators Ind. Health 2014 52 279 288 10.2486/indhealth.2013-0156 24869893
6. Useche S.A. Cendales B. Alonso F. Montoro L. Pastor J.C. Trait driving anger and driving styles among colombian professional drivers Heliyon 2019 5 e02259 10.1016/j.heliyon.2019.e02259 31440599
7. Mann R.E. Stoduto G. Vingilis E. Asbridge M. Wickens C.M. Ialomiteanu A. Sharpley J. Smart R.G. Alcohol and driving factors in collision risk Accid. Anal. Prev. 2010 42 1538 1544 10.1016/j.aap.2010.03.010 20728600
8. Rolison J.J. Moutari S. Combinations of factors contribute to young driver crashes J. Saf. Res. 2020 73 171 177 10.1016/j.jsr.2020.02.017
9. Šimková I. Konečný V. Liščák Š. Stopka O. Measuring the quality impacts on the performance in transport company Transp. Probl. 2015 10 113 124 10.21307/tp-2015-039
10. Jongen E.M.M. Brijs K. Komlos M. Brijs T. Wets G. Inhibitory control and reward predict risky driving in young novice drivers—A simulator study Procedia—Soc. Behav. Sci. 2011 20 604 612 10.1016/j.sbspro.2011.08.067
11. Dénommée J.A. Foglia V. Roy-Charland A. Turcotte K. Lemieux S. Yantzi N. Cellphone use and young drivers Can. Psychol.—Psychol. Can. 2020 61 22 30 10.1037/cap0000175
12. Møller M. Haustein S. Keep on cruising: Changes in lifestyle and driving style among male drivers between the age of 18 and 23 Transp. Res. Part F Traffic Psychol. Behav. 2013 20 59 69 10.1016/j.trf.2013.05.003
13. Nicolai J. Moshagen M. Demmel R. Patterns of alcohol expectancies and alcohol use across age and gender Drug Alcohol Depend. 2012 126 347 353 10.1016/j.drugalcdep.2012.05.040 22748519
14. Graham K. Bernards S. Knibbe R. Kairouz S. Kuntsche S. Wilsnack S.C. Greenfield T.K. Dietze P. Obot I. Gmel G. Alcohol-related negative consequences among drinkers around the world Addiction 2011 106 1391 1405 10.1111/j.1360-0443.2011.03425.x 21395893
15. Useche S.A. Hezaveh A.M. Llamazares F.J. Cherry C. Not gendered… but different from each other? A structural equation model for explaining risky road behaviors of female and male pedestrians Accid. Anal. Prev. 2021 150 105942 10.1016/j.aap.2020.105942 33338915
16. Oppenheim I. Oron-Gilad T. Parmet Y. Shinar D. Can traffic violations be traced to gender-role, sensation seeking, demographics and driving exposure? Transp. Res. Part F Traffic Psychol. Behav. 2016 43 387 395 10.1016/j.trf.2016.06.027
17. Cordellieri P. Baralla F. Ferlazzo F. Sgalla R. Piccardi L. Giannini A.M. Gender effects in young road users on road safety attitudes, behaviors and risk perception Front. Psychol. 2016 7 1412 10.3389/fpsyg.2016.01412 27729877
18. Wan J. Wu C. Zhang Y. Houston R.J. Chen C.W. Chanawangsa P. Drinking and driving behavior at stop signs and red lights Accid. Anal. Prev. 2017 104 10 17 10.1016/j.aap.2017.04.008 28448790
19. Christoforou Z. Karlaftis M.G. Yannis G. Reaction times of young alcohol-impaired drivers Accid. Anal. Prev. 2013 61 54 62 10.1016/j.aap.2012.12.030 23332180
20. Li Y.C. Sze N.N. Wong S.C. Yan W. Tsui K.L. So F.L. A simulation study of the effects of alcohol on driving performance in a Chinese population Accident. Anal. Prev. 2016 95 334 342 10.1016/j.aap.2016.01.010
21. Yadav A.K. Velaga N.R. Modelling the relationship between different Blood Alcohol Concentrations and reaction time of young and mature drivers Transp. Res. Part F Traffic Psychol. Behav. 2019 64 227 245 10.1016/j.trf.2019.05.011
22. Mayhew D.R. Donelson A.C. Beirness D.J. Simpson H.M. Youth, alcohol and relative risk of crash involvement Accid. Anal. Prev. 1986 18 273 287 10.1016/0001-4575(86)90042-4 3741579
23. Peck R.C. Gebers M.A. Voas R.B. Romano E. The relationship between blood alcohol concentration (BAC), age, and crash risk J. Saf. Res. 2008 39 311 319 10.1016/j.jsr.2008.02.030 18571573
24. Du H. Zhao X. Zhang G. Rong J. Effects of alcohol and fatigue on driving performance in different roadway geometries Transportation Research Record: J. Transp. Res. Board 2016 2584 88 96 10.3141/2584-11
25. Zhang X. Zhao X. Du H. Ma J. Rong J. Effect of different breath alcohol concentrations on driving performance in horizontal curves Accid. Anal. Prev. 2014 72 401 410 10.1016/j.aap.2014.07.032 25146494
26. Jaśkiewicz M. Frej D. Tarnapowicz D. Poliak M. Upper Limb Design of an Anthropometric Crash Test Dummy for Low Impact Rates Polymers 2020 12 2641 10.3390/polym12112641
27. Leung S. Starmer G. Gap acceptance and risk-taking by young and mature drivers, both sober and alcohol-intoxicated, in a simulated driving task Accid. Anal. Prev. 2005 37 1056 1065 10.1016/j.aap.2005.06.004 16036208
28. Konstantopoulos P. Chapman P. Crundall D. Driver’s visual attention as a function of driving experience and visibility. using a driving simulator to explore drivers’ eye movements in day, night and rain driving Accid. Anal. Prev. 2010 42 827 834 10.1016/j.aap.2009.09.022 20380909
29. Blana E. A Survey of Driving Research Simulators around the World 1996 Available online: https://eprints.whiterose.ac.uk/2110/ (accessed on 17 March 2022)
30. Carsten O. Jamson A.H. Driving simulators as research tools in traffic psychology Handbook of Traffic Psychology Elsevier Amsterdam, The Netherlands 2011 87 96 10.1016/B978-0-12-381984-0.10007-4
31. Novotný S. Interaktivní Simulátory Dopravních Prostředků Pro Analýzu Spolehlivosti Interakce Řidiče s Vozidlem. České Vysoké Učení Technické 2014 Available online: https://portal.cvut.cz/wp-content/uploads/2017/04/HP2014-30-Novotny.pdf (accessed on 17 March 2022)
32. Ministry of Transport, Posts and Telecommunications of the Slovak Republic Methodical Instruction no. 22/2005 on Technical Requirements for Simulators Ministry of Transport, Posts and Telecommunications of the Slovak Republic Bratislava, Slovakia 2005
33. Martín-delosReyes L.M. Jiménez-Mejías E. Martínez-Ruiz V. Moreno-Roldán E. Molina-Soberanes D. Lardelli-Claret P. Efficacy of training with driving simulators in improving safety in young novice or learner drivers: A systematic review Transp. Res. Part F Traffic Psychol. Behav. 2019 62 58 65 10.1016/j.trf.2018.12.006
34. Van Leeuwen P.M. Happee R. de Winter J.C.F. Changes of driving performance and gaze behavior of novice drivers during a 30-min simulator-based training Procedia Manuf. 2015 3 3325 3332 10.1016/j.promfg.2015.07.422
35. Rossia R. Gastaldia M. Gecchelea G. Analysis of driver task-related fatigue using driving simulator experiments Procedia-Soc. Behav. Sci. 2011 20 666 675 10.1016/j.sbspro.2011.08.074
36. Meng F. Wong S.C. Yan W. Li Y.C. Yang L. Temporal patterns of driving fatigue and driving performance among male taxi drivers in Hong Kong: A driving simulator approach Accid. Anal. Prev. 2019 125 7 13 10.1016/j.aap.2019.01.020 30690275
37. Matsumoto Y. Peng G. Analysis of driving behavior with information for passing through signalized intersection by driving simulator Transp. Res. Procedia 2015 10 103 112 10.1016/j.trpro.2015.09.060
38. Hess S. Choudhury C.F. Bliemer M.C.J. Hibberd D. Modelling lane changing behaviour in approaches to roadworks: Contrasting and combining driving simulator data with stated choice data Transp. Res. Part C Emerg. Technol. 2020 112 282 294 10.1016/j.trc.2019.12.003
39. Calvi A. D’Amico F. Ferrante C. Bianchini Ciampoli L. A driving simulator validation study for evaluating the driving performance on deceleration and acceleration lanes Adv. Transp. Stud. 2020 50 67 80 10.4399/97888255317325
40. Yuan J. Abdel-Aty M. Cai Q. Lee J. Investigating drivers’ mandatory lane change behavior on the weaving section of freeway with managed lanes: A driving simulator study Transp. Res. Part F Traffic Psychol. Behav. 2019 62 11 32 10.1016/j.trf.2018.12.007
41. Papantoniou P. Yannis G. Christofa E. Which factors lead to driving errors? A structural equation model analysis through a driving simulator experiment IATSS Res. 2019 43 44 50 10.1016/j.iatssr.2018.09.003
42. Haycock B.C. Campos J.L. Koenraad N. Potter M. Advani S.K. Creating headlight glare in a driving simulator Transp. Res. Part F Traffic Psychol. Behav. 2019 61 93 106 10.1016/j.trf.2017.10.006
43. Weir D.H. Application of a driving simulator to the development of in-vehicle human–machine-interfaces IATSS Res. 2010 34 16 21 10.1016/j.iatssr.2010.06.005
44. Jang J. Lee H. Kim J. Carfree: Hassle-free object detection dataset generation using carla autonomous driving simulator Appl. Sci. 2022 12 281 10.3390/app12010281
45. Riegler A. Riener A. Holzmann C. AutoWSD: Virtual reality automated driving simulator for rapid HCI prototyping Proc. Mensch Und Comput. 2019 2019 853 857 10.1145/3340764.3345366
46. Costa V. Rossetti R.J.F. Sousa A. Autonomous driving simulator for educational purposes Proceedings of the 11th Iberian Conference on Information Systems and Technologies (CISTI) Coimbra, Portugal 19–22 June 2019 1 5 10.1109/CISTI.2016.7521461
47. Koppel C. van Doornik J. Petermeijer B. Abbink D. Investigation of the lane change behavior in a driving simulator ATZ Worldwide 2019 121 62 67 10.1007/s38311-019-0156-6
48. Wynne R.A. Beanland V. Salmon P.M. Systematic review of driving simulator validation studies Saf. Sci. 2019 117 138 151 10.1016/j.ssci.2019.04.004
49. Helland A. Lydersen S. Lervåg L. Jenssen G.D. Mørland J. Slørdal L. Driving simulator sickness: Impact on driving performance, influence of blood alcohol concentration, and effect of repeated simulator exposures Accid. Anal. Prev. 2016 94 180 187 10.1016/j.aap.2016.05.008 27322638
50. Gálvez-García G. Albayay J. Rehbein L. Tornay F. Mitigating simulator adaptation syndrome by means of tactile stimulation Appl. Ergon. 2017 58 13 17 10.1016/j.apergo.2016.05.004 27633193
51. Gálvez-García G. A comparison of techniques to mitigate simulator adaptation syndrome Ergonomics 2015 58 1365 1371 10.1080/00140139.2015.1005168 25653095
52. Aykent B. Merienne F. Guillet C. Paillot D. Kemeny A. Motion sickness evaluation and comparison for a static driving simulator and a dynamic driving simulator Proc. Inst. Mech. Eng. Part D J. Automob. Eng. 2014 228 818 829 10.1177/0954407013516101
53. Weech S. Moon J. Troje N.F. Influence of bone-conducted vibration on simulator sickness in virtual reality PLoS ONE 2018 13 e0194137 10.1371/journal.pone.0194137 29590147
54. JKZ Spol. s r.o Driving Simulator SNA–211 REN: Operating Instructions JKZ Spol. s r.o Olomouc, Czech Republic 2016 25
55. JKZ Spol. s r.o Driving Simulator SNA–211 REN: Cab Controls and Indicators JKZ Spol. s r.o Olomouc, Czech Republic 2016 25
56. Cit.vfu.cz: Parametric Tests—Student’s Test Available online: https://cit.vfu.cz/statpotr/POTR/Teorie/Predn3/ttest.htm (accessed on 22 February 2022)
57. Evaluation of the Dependence of Two Random Available online: https://is.muni.cz/do/rect/el/estud/prif/js18/korelacna_analyza/web/pages/02-hodnotenie-zavislosti-dvoch-nahodnych-velicin.html (accessed on 22 February 2022)
58. Kubanova J. Statisticke Metody Pro Ekonomickou a Technickou Praxi STATIS Bratislava, Slovakia 2004 80-85659-37-9
59. Ondruš J. Vrábel J. Kolla E. The influence of the vehicle weight on the selected vehicle braking characteristics Transport Means 2018: Proceedings of the 22nd International Conference—Trakai, Lithuania, 3–5 October 2018 Kaunas University of Technology Kaunas, Lithuania 2018
60. Cohen J. Statistical Power Analysis for the Behavioral Sciences 2nd ed. Routledge New York, NY, USA 1988 250 290
61. Jurecki R.S. Influence of the scenario complexity and the lighting conditions on the driver behaviour in a car-following situation Arch. Automot. Eng. Arch. Motoryz. 2019 83 151 173 10.14669/AM.VOL83.ART11
62. Al Qaisi I. Traechtler A. Human in the loop: Optimal control of driving simulators and new motion quality criterion Proceedings of the 2012 IEEE International Conference on Systems, Man, and Cybernetics (SMC) Seoul, Korea 14–17 October 2012 10.1109/ICSMC.2012.6378073
63. Dziuda Ł. Biernacki M.P. Baran P.M. Truszczyński O.E. The effects of simulated fog and motion on simulator sickness in a driving simulator and the duration of after-effects Appl. Ergon. 2014 45 406 412 10.1016/j.apergo.2013.05.003 23726466
64. Domeyer J.E. Cassavaugh N.D. Backs R.W. The use of adaptation to reduce simulator sickness in driving assessment and research Accid. Anal. Prev. 2013 53 127 132 10.1016/j.aap.2012.12.039 23416680
65. Vrábel J. Šarkan B. Vashisth A. Change of driver’s reaction time depending on the amount of alcohol consumed by the driver —The case study Arch. Automot. Eng. Arch. Motoryz. 2020 87 47 56 10.14669/AM.VOL87.ART4
66. Gnap J. Konečný V. Varjan P. Research on relationship between freight transport performance and GDP in Slovakia and EU countries Naše More 2018 65 32 39 10.17818/NM/2018/1.5
67. Dirnbach I. Kubjatko T. Kolla E. Ondruš J. Šarić Ž. Methodology designed to evaluate accidents at intersection crossings with respect to forensic purposes and transport sustainability Sustainability 2020 12 1972 10.3390/su12051972
68. Jurecki R. Poliak M. Jaśkiewicz M. Young adult drivers: Simulated behaviour in a car-following situation Promet 2017 29 381 390 10.7307/ptt.v29i4.2305
|
PMC009xxxxxx/PMC9099899.txt |
==== Front
Int J Mol Sci
Int J Mol Sci
ijms
International Journal of Molecular Sciences
1422-0067
MDPI
10.3390/ijms23094564
ijms-23-04564
Communication
Streptozotocin-Induced Diabetes Causes Changes in Serotonin-Positive Neurons in the Small Intestine in Pig Model
https://orcid.org/0000-0002-1402-5423
Bulc Michał 1*
https://orcid.org/0000-0002-7593-9552
Palus Katarzyna 1
Całka Jarosław 1
https://orcid.org/0000-0003-0177-836X
Kosacka Joanna 2
Nowicki Marcin 3
Hmadcha Abdelkrim Academic Editor
1 Department of Clinical Physiology, Faculty of Veterinary Medicine, University of Warmia and Mazury in Olsztyn, Oczapowski Str. 13, 10-718 Olsztyn, Poland; [email protected] (K.P.); [email protected] (J.C.)
2 Department of Visceral, Transplant, Thoracic and Vascular Surgery, University of Leipzig Medical Center, Liebigstr. 21, 04103 Leipzig, Germany; [email protected]
3 Institute of Anatomy, University of Leipzig, Liebigstraße 13, 04103 Leipzig, Germany; [email protected]
* Correspondence: [email protected]
20 4 2022
5 2022
23 9 456407 3 2022
19 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Serotonin (5-hydroxytryptamine or 5-HT) is an important neurotransmitter of the central and peripheral nervous systems, predominantly secreted in the gastrointestinal tract, especially in the gut. 5-HT is a crucial enteric signaling molecule and is well known for playing a key role in sensory-motor and secretory functions in the gut. Gastroenteropathy is one of the most clinical problems in diabetic patients with frequent episodes of hyperglycemia. Changes in 5-HT expression may mediate gastrointestinal tract disturbances seen in diabetes, such as nausea and diarrhea. Based on the double immunohistochemical staining, this study determined the variability in the population of 5-HT-positive neurons in the porcine small intestinal enteric neurons in the course of streptozotocin-induced diabetes. The results show changes in the number of 5-HT-positive neurons in the examined intestinal sections. The greatest changes were observed in the jejunum, particularly within the myenteric plexus. In the ileum, both de novo 5-HT synthesis in the inner submucosal plexus neurons and an increase in the number of neurons in the outer submucosal plexus were noted. The changes observed in the duodenum were also increasing in nature. The results of the current study confirm the previous observations concerning the involvement of 5-HT in inflammatory processes, and an increase in the number of 5-HT -positive neurons may also be a result of increased concentration of the 5-HT in the gastrointestinal tract wall and affects the motor and secretory processes, which are particularly intense in the small intestines.
serotonin
enteric nervous system
diabetes
pig
==== Body
pmc1. Introduction
Serotonin (5-hydroxytryptamine, 5-HT) was first isolated from the platelets in 1948 and identified as an agent causing vascular smooth muscle contraction. 5-HT is synthesized in the body by the enzymatic conversion (hydroxylation and decarboxylation) of tryptophan [1,2,3]. The synthesis of 5-HT takes place primarily in the central and peripheral nervous system neurons, chromophilic cells of the gastrointestinal tract, and mast cells [4]. In mammalian bodies, the vast majority of serotonin is found in the gastrointestinal tract, mainly in the intestinal region. Much less serotonin is found in the platelets, and only 1–2% is found in the brain [5,6]. In the gastrointestinal tract, enteroendocrine (EE) cells are scattered throughout the mucosal lining of the gut wall and synthesize and release a great number of different natural biologically active substances. Approximately half of all EE cells are responsible for the production of serotonin and are called enterochromaffin (EC) cells. EC cells are the largest source of 5-HT in the body, producing about 95% of total body 5-HT. Significantly lower amounts of 5-HT are secreted in other peripheral tissues and serotonergic neurons within the enteric neurons [6,7,8]. 5-HT synthesized in the intestinal cells is released into the blood and into the intestinal lumen. This amine exerts its biological effect after binding to the appropriate membrane receptor [9,10]. Five different 5-HT receptor types have been described in the gastrointestinal wall. They are found in the enterocytes, enteric neurons, smooth muscle cells, and immune cells [11]. The physiological role of 5-HT in the gastrointestinal tract involves a number of secretory, motor, and sensory processes. It is involved in fluid and mucin secretion within the intestines and affects the secretion of pancreatic juice [12]. Another important function is the involvement in electrolyte resorption. 5-HT also stimulates intestinal motor activity and is involved in shaping the pattern of intestinal contractile activity [13,14]. The involvement of 5-HT is not limited to the regulation of physiological processes since, due to its effects on the immune cells, it is also involved in inflammatory processes and autoimmune disorders such as inflammatory bowel disease (IBD) and coeliac disease [15].
The multidirectional effect of 5-HT on the gastrointestinal tract is also due to the complex innervation of this tract. Neural regulation of the gastrointestinal tract activity is carried out by exogenous neurons represented by the sympathetic and parasympathetic parts of the nervous system and by neurons that are part of the enteric neurons [16]. The enteric neurons are a highly organized structure formed by plexuses organized into ganglia. Within the intestinal region, a distinction can be drawn between the myenteric plexus and submucosal plexus, which, in large animals, is secondarily divided into the inner and outer submucosal plexus (Figure 1) [17]. The myenteric plexus is clearly visible along the entire length of the digestive tract, from the esophagus to the anal sphincter, and primarily controls the activities of the muscular layer, although some neurons supply the mucosa, blood vessels, and glands of the digestive tract and they also create connections with other types of ENS plexuses. The myenteric plexus shows significant variations depending on the species and part of the gastrointestinal tract [16,17,18]. The submucosal plexus in large mammals is divided into two types of plexuses, which are clearly visible mainly in the small and large intestines. Occasionally, they can be found in the stomach and esophagus. In turn, they are absent in the esophagus and forestomachs of ruminants. Similar to the myenteric plexus, the submucosal plexuses show broad species differences. They are involved in the control of reabsorption and fluids secretion as well as blood flow in the mucosal layer. Moreover, they create connections with other types of ENS plexuses and supply cells to the immune and entero-endocrine systems [17,18].
The enteric neurons are a structure characterized by a high degree of functional autonomy. They control both motor processes and processes related to the secretion and resorption of both gastric contents and the digestive enzymes [18]. A characteristic feature of enteric neurons is their neuronal plasticity, which can be defined as an adaptive variability in the synthesis of biologically active substances and their release into the surrounding environment in response to pathological factors [19].
Hyperglycemia is a condition of elevated blood glucose levels and occurs most often in the course of undiagnosed or poorly treated diabetes. A consequence of prolonged hyperglycemia is damage to many organs and tissues [18]. One of the tissues that are most vulnerable is nerve tissue, and such damage may affect both somatic and autonomic nerves. Dysfunction of the nervous system supplying the gastrointestinal tract is referred to as gastroenteropathy, and it can affect any section of the gastrointestinal system. The consequences of enteric neuronal damage are manifold and are manifested by a wide spectrum of clinical symptoms, which include nausea, the feeling of severe constipation, diarrhea, the abnormal passage of intestinal contents, or frequently occurring pain episodes [20,21,22,23]. The pathomechanism of these disorders is diverse. In addition to well-known mechanisms, including the development of abnormal metabolic pathways (sorbitol, polyol pathway), this may involve a change in the amount of biologically active substances synthesized and released by the neural cells of the enteric neurons [23].
The aim of the current study is to determine how hyperglycemia lasting for six weeks affects the quantitative population of 5-HT-synthesising neurons in the porcine small intestinal system. This also determines whether serotonin is a substance involved in the response of the intramural neurons of the gastrointestinal tract to hyperglycemia. The study was conducted using the pig as a model animal due to the numerous similarities between the porcine and human gastrointestinal tract anatomy and physiology. The mechanisms of insulin secretion and the blood supply to the pancreas are also similar in both species. It should also be noted that the pig is also used as a research model in other studies on metabolic disorders [24,25,26].
2. Results
2.1. Hyperglycemia
All pigs that were used in the present investigation developed diabetes within approximately 7 days. The main criterion for the development of diabetes was significant hyperglycemia. The mean glucose level before the injection of streptozotocin in both groups remained at the physiological level (5 mmol/L) (Figure 2A). From the first week of the experiment, a significant increase in glucose concentration was observed in all pigs in the experimental group (Figure 2B). The highest glucose concentration was observed in the second week of the experiment. Then, in the animals of the experimental group, the average concentration of glucose in the blood remained above 20 mmol/L (Figure 2C–G). Such a high concentration was maintained until the end of the experiment. It should be underlined that although blood glucose level in STZ-treated animals was notably higher than in controls, all pigs with hyperglycemia survived the duration of the experiment in good condition, and none of the pigs required exogenous insulin supplementation.
2.2. Immunofluorescence
Double-labeling immunohistochemistry revealed that 5-HT was present in the investigated parts of the gut. Their presence was detected in both submucosal plexuses as well as in the myenteric plexus. The number of neurons that express serotonin differed between particular segments of the intestine as well as between individual plexuses.
2.2.1. Duodenum
The population of neurons synthesizing 5-HT in the control animals located in the myenteric plexus was established at the level of 8.34 % (±1.67) (Figure 3 and Figure 4A). The 6-week hyperglycemia led to statistically significant changes in the number of 5-HT-immunoreactive neurons (5-HT-IR) in the myenteric plexus. We have noted an increase in the total number of 5-HT neurons to the level of 14.78 % (±1.90) (Figure 3 and Figure 4B). Within the inner and outer submucosal plexus in the control group population of 5-HT-positive neurons constituted 2.98 % (±0.23) and 4.43 % (±0.56), respectively (Figure 3). Treatment with streptozotocin enhanced the population of neurons expressing 5-HT in both submucosal plexuses. In the case of the inner submucosal plexus, an increase has reached the level of 7.56 % (±1.56) (Figure 3 and Figure 4C,D), while in the outer submucosal plexus, the increase in 5-HT positive neurons was slightly higher (9.48 % (±2.06)) (Figure 3 and Figure 4E,F).
2.2.2. Jejunum
In these parts of the small intestine, 5-HT positive neurons were located in all investigated plexuses. In control animals, the most numerous population was observed in the myenteric plexus, 10.33% (±1.98) (Figure 5 and Figure 6A). In the submucosal plexuses, the population of 5-HT-IR neurons was less numerous. In the inner submucosal plexus, 4.66% (±1.33) of Hu C/D-positive neurons expressed 5-HT simultaneously, while the outer plexus contained only 1.95 % (±0.44) of neurons immunopositive to 5-HT (Figure 5 and Figure 6C,D). Diabetes increased the number of 5-HT positive neurons in the myenteric and inner submucosal plexuses, while in the outer submucosal plexus, changes in the number of 5-HT-IR neurons were not observed (Figure 5 and Figure 6B,D,F). In relation to the myenteric plexus, the population of 5-HT positive neurons increased to 15.69 % (±2.34), while in the inner submucosal plexus, the level of 5-HT immunopositive neurons was estimated at 7.89 % (±1.87).
2.2.3. Ileum
In the control group, neurons containing serotonin were present in the myenteric plexus at 3.87% (±0.56) and inner submucosal plexus at 2.04% (±0.23) (Figure 7 and Figure 8A,C), while in the outer submucosal plexus, 5-HT-positive neurons were not observed (Figure 7 and Figure 8E,F). High blood glucose level led to an increase in the number of 5-HT-IR neurons in the myenteric plexus to 5.77% (±0.41) as well as de novo synthesis of 5-HT in neurons forming the outer submucosal plexuses (2.29% (±0.77)) and lack of changes in serotonin expression in the inner submucosal plexus (Figure 7 and Figure 8B–D,F).
2.2.4. Nerve Fibers
Nerve fibers immunoreactive for serotonin were found in all investigated parts of the small intestine. They were present both in the muscle layer as well as in the submucosal area (Figure 9 and Figure 10). In the duodenum muscle layer, the nerve fibers were located between the longitudinal and circular muscle layers. They created long bunds (++), and their density did not change under hyperglycemia conditions (Figure 9A,B). In turn, in the jejunum only single (+) nerve fibers have been observed. Very similar density (+) of serotonin-positive nerve fibers were observed in the ileum muscle layer. In both intestinal areas, differences between control and experimental group were not observed (Figure 9C–F). Microscopic analysis of nerve fibers in the duodenum submucosal layer showed the most numerous population of nerve fibers in the entire area of the small intestines (+++). Their density was the same in the control and experimental group (Figure 10A,B). A slightly less dens network was observed in the jejunum submucosal layer (++) (Figure 10C,D). In turn, the least numerous nerve fibers were noted in the ileum, where we observed only single fibers in both groups (Figure 10E,F). Moreover, in the course of nerve fibers, numerous varicosities were present.
3. Discussion
The role of 5-HT in the gastrointestinal tract has been a subject of intense research over many years [15,27]. While its physiological function in secretomotor processes is well understood, its significance in the course of gastrointestinal tract pathological disorders is still not completely known. This study conducted tests aimed at understanding the quantitative changes in the 5-HT positive neurons that are part of the enteric neurons of porcine small intestines in the course of experimentally induced diabetes. The nosological entity that provided the basis for the tests conducted in the study is currently recognized as a condition with an exceptionally dynamic growth trend. As the incidence of diabetes increases, the number of complications often accompanying the disease increases as well [22]. The gastrointestinal tract, due to its complex innervation, is a structure whose proper functioning is impaired in the course of diabetes [21,22,23,28]. These changes are usually caused by disturbances in the normal gastrointestinal tract motility, which converts into the abnormal passage of the gastric content and impairs the resorption and secretion processes within the gastrointestinal tract [29,30,31]. The results obtained in the current study indicate that 5-HT-positive neurons that are part of the small intestinal enteric neurons may be involved in pathological processes of the gastrointestinal tract in the course of diabetes. During the study, a statistically significant increase was observed in the population of 5-HT-synthesising neurons in the myenteric plexuses within all sections of the small intestine. The neurons of this plexus primarily control the intestinal contractile activity, which is clearly impaired in the course of diabetes. The role of 5-HT in this process is multidirectional. It stimulates both the intramural excitatory and inhibitory neurons and is able to induce an effect both increasing and decreasing the intestinal motor activity [5,6,11]. The first effect is achieved by the activation of cholinergic neurons, which increases the release of acetylcholine and causes the contraction of intestinal smooth muscles [32]. The inhibitory effect results from the stimulation of nitrergic neurons, which increases the release of nitric oxide, which, in turn, causes smooth muscle relaxation [32]. Moreover, when the gastric contents reach the intestines, the pressure inside the intestinal lumen increases. This results in an increased release of 5-HT, which then stimulates efferent fibers of the vagus nerve and initiates the intestinal peristaltic reflex [33]. An increased number of 5-HT-positive neurons in the enteric neurons in the course of diabetes may translate into an increase in its concentration in both the intestinal wall and the intestinal lumen, which leads to the intensification of the processes described above. These changes consequently impair motility, which is clinically manifested as alternately occurring diarrhea and constipation episodes that are among the main symptoms of diabetic gastroenteropathy [25]. What is also important is that 5-HT is responsible for the development of the migrating motor complex (MMC) and for normal postprandial intestinal contractions that are often impaired in the course of diabetes [34,35,36]. One of the major disorders in the course of diabetes is the impairment of sensory and secretory functions within the gastrointestinal tract. Serotonin is one of the signaling molecules involved in the transmission of sensory information in the intestinal mucous membrane and a powerful stimulus for increased intestinal juice release [37,38]. The current study demonstrated an increase in the number of neurons in the submucosal ganglia of the small intestine, particularly in the jejunum, and the emergence of a de novo population of 5-HT-positive neurons in the outer submucosal plexus of the ileum. The submucosal plexuses, which are primarily involved in the resorptive and secretory processes through an increase in the 5-HT-positive neuron population, are also involved in changes in adapting the enteric neurons to hyperglycemia. The 5-HT synthesized and released by these changes may modify the intestinal resorption processes. The mechanism of electrolyte secretion regulation by 5-HT is a result of its interaction with the 5-HT2 receptor [39]. The above data indicate that one of the intestinal neuronal responses to hyperglycemia is an increase in the amount of 5-HT, reflected in changes in the number of neurons synthesizing it. Undoubtedly, this results in a significant effect of this amine on shaping both the motor as well as the secretory and sensory processes in the course of diabetes and the often-accompanying gastroenteropathy.
One of the possible reasons for changes in the expression of 5-HT-positive neurons may be an inflammatory process that often develops in the course of diabetes [40]. This process is of particular importance in the context of the development of complications affecting the nervous system, including the autonomic nervous system, which includes the enteric neurons [21,25,40,41]. The main mechanism responsible for the development of inflammation is an increase in the concentration of protein glycation end products, the activation of the receptor for glycation products, the development of oxidative stress, and an increase in the oxygen radical concentration, which results in the activation of the nuclear factor NF-kappaβ and an increase in pro-inflammatory cytokine synthesis [41,42]. The research conducted so far has demonstrated that 5-HT is a molecule whose concentration increases in the course of inflammatory conditions that affect the gastrointestinal tract [15]. The involvement of 5-HT was confirmed in the course of inflammatory processes within the gastrointestinal mucous membrane in the course of inflammatory bowel disease, ulcerative colitis, and Crohn’s disease [14,15,29]. Experimental inflammatory intestinal processes induced by different doses of bisphenol A resulted in an increase in the amount of serotonin in the gastrointestinal tract wall [43,44]. To date, relatively few studies have been dedicated to changes in the expression of 5-HT in the gastrointestinal tract in the course of diabetes. A study conducted on rats with streptozotocin-induced diabetes lasting for 3 and 8 weeks, respectively, showed an increase in the amount of 5-HT in the duodenum and the ileal myenteric plexus [21]. The results presented above confirm the results obtained in the current study.
In conclusion, the current study demonstrated that hyperglycemia lasting for six weeks significantly affects the number of 5-HT-positive neurons within the porcine small intestine. The double immunohistochemical staining method applied in the study enabled the precise determination of quantitative changes within the individual ganglia of the enteric nervous system. In addition, the results of the current study confirm the previous observations concerning the involvement of 5-HT in inflammatory processes, where it acts as an immunomodulatory substance. Moreover, an increase in the number of serotonin-positive neurons may result in an increase in the 5-HT concentration in the gastrointestinal tract wall and affect the motor and secretory processes, which are particularly intense in the small intestines. Further research using the agonists and antagonists of individual 5-HT receptor types may contribute to a better understanding of the precise mechanisms of 5-HT action in the gastrointestinal tract, particularly in the course of diabetic gastroenteropathy.
4. Materials and Methods
In total, 10 juvenile pigs of the White Large Polish breed were used in this study. The experiments had been approved by the Local Ethical Committee in Olsztyn (Poland) (decision number 13/2015/DTN) and according to the Act for the Protection of Animals for Scientific or Educational Purposes of 15 January 2015 (Official Gazette 2015, No. 266), corresponding in the Republic of Poland with special attention paid to minimizing any pain and stress reaction. After one week acclimatization period, pigs were randomly divided into two groups: control and experimental with chemically induced diabetes (5 animals in each). Hyperglycemia was induced by a single intravenous injection of streptozotocin under premedication induced by atropine (0.05 mg/kg body weight /BW, given intramuscularly; Atropinum sulf. WZF, Warszawskie Zakłady Farmaceutyczne Polfa S.A., Poland), azaperone (2 mg/kg BW, given intramuscularly Stresnil, Janssen Pharmaceutica, Beerse, Belgium), and streptozotocin (STZ, (Sigma-Aldrich, St. Louis, MO, USA, 0130; 150 mg/kg). Directly before injections, STZ was dissolved in disodium citrate buffer solution (pH = 4.23, 1 g streptozotocin/10 mL solution). The needle was inserted into the ear venous, and the STZ infusion time was about 5 min. In order to eliminate nausea and vomiting after streptozotocin infusion, pigs were fasted for 18 h before the experiment, and the control animals were injected with equal amounts of vehicle (citrate buffer). After diabetes induction, animals were kept in cages tailored to pigs. Animals from both groups receive a standard swine diet (rapeseed meal, 6.0%; soybean meal, 9.0%; wheat, 54.0%; barley, 28.5%; others, 2.5%), and water ad libitum. Blood glucose level was measured before STZ injection, 48 h after induction of diabetes. Next measurements were made in each week of the experiment. Blood glucose concentrations were measured in plasma using an Accent-200 (Cormay, Warsaw, Poland) biochemical analyzer, with the colorimetric measurement at a wavelength of 510 nm/670 nm.
Six weeks after diabetes induction, animals in both groups were deeply anesthetized via intravenous administration of pentobarbital 60 mg/kg body weight (Vetbutal, Biowet, Poland). Afterward, pigs were perfused, and the gastrointestinal tract was prepared as previously described [28]. Next, 2 cm long fragments of the small intestine from the place where nerves from inferior mesenteric ganglia supply the intestine were collected. The samples were postfixed by immersion by the 4% paraformaldehyde for 1 h, rinsed several times with phosphate buffer (PB), and finally transferred to 30% sucrose solution and stored at 4 °C until sectioning. The tissue blocks were cut in frontal or sagittal planes using a Microm HM 560 cryostat (Carl Zeiss, Germany) at a thickness of 12 μm and mounted on gelatinized glass. Subsequently, sections were double-stained by first incubating with primary antisera overnight. Following antibodies were used marker Hu C/D proteins (dilution 1:1000; host rabbit; Invitrogen, Waltham, MA, USA; code A-21271) and serotonin (dilution 1:1000; host rabbit; Zymed Laboratories, San Francisco, CA, USA, code 30778610) were used. After overnight incubation with bovine serum albumin (BSA), the sections were incubated with secondary antibodies Alexa Fluor 488 (dilution 1:1000; host donkey; Invitrogen, USA; code A21202) and Alexa Fluor 546 (dilution 1:1000; host donkey; Invitrogen, USA; code A10040). Both in primary and secondary antibodies, slides were incubated at room temperature. The slides were viewed and photographed using an Olympus BX51 microscope equipped with epifluorescence and appropriate filter sets, coupled with a digital monochromatic camera (Olympus XM 10) connected to a PC and analyzed with Cell Dimension software (Olympus, Tokyo, Japan). Standard controls, i.e., pre-absorption for the serotonin antisera 20 μg per 1 mL of antibody at working dilution. Additionally, omission and replacement of the respective primary antiserum with the corresponding non-immune serum completely abolished immunofluorescence and eliminated specific staining (Figure 11).
Counting of the Nerve Structures and Statistical Evaluation
To evaluate the percentage of examined neurons, at least 700 Hu C/D-labeled cell bodies in a particular plexus (MP, OSP, and ISP) of each studied animal were examined. Only neurons with well-visible nuclei were counted. To prevent double counting of Hu C/D immunoreactive neurons, the sections were located at least 100 μm apart. The data pooled from all animal groups were statistically analyzed using Statistica 13 software (StatSoft Inc., Tulsa, OK, USA) and expressed as a mean ± standard error (SEM) of mean. Significant differences were evaluated using Student’s t-test for independent samples (* p < 0.05, ** p < 0.01, and *** p < 0.001). Moreover, for semiquantitative evaluation of the density of nerve fibers immunoreactive to each substance studied, an arbitrary scale was used, where (−)—absence of nerve fibers; (+)—single nerve fibers; (++)—rare nerve fibers; (+++) dense nerve fibers.
Author Contributions
M.B., Conceptualization, Methodology, Formal Analysis, Investigation, Writing—Original Draft Preparation, Writing—Review and Editing; K.P., Investigation, Formal Analysis, Writing—Review and Editing; J.C., Writing—Review and Editing; J.K., Formal Analysis, Writing—Review and Editing; M.N., Formal Analysis, Writing—Review and Editing. All authors have read and agreed to the published version of the manuscript.
Funding
Project financially supported by the Minister of Education and Science under the program entitled “Regional Initiative of Excellence” for the years 2019-2022, Project No. 010/RID/2018/19, amount of funding 12.000.000 PLN.
Institutional Review Board Statement
The animal study protocol was approved by the Institutional Review Board. Local Ethical Committee in Olsztyn (Poland) (decision number 13/2015/DTN) and ac-cording to the Act for the Protection of Animals for Scientific or Educational Purposes of 15 January 2015 (Official Gazette 2015, No. 266), corresponding in the Republic of Poland with special attention paid to minimizing any pain and stress reaction.
Informed Consent Statement
Not applicable.
Data Availability Statement
MDPI Research Data Policies at https://www.mdpi.com/ethics (accessed on 6 March 2022).
Conflicts of Interest
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
Figure 1 Organization of the enteric neurons in the pig intestine wall demonstrated by the labeling with Hu C/D—used as a pan neuronal marker. Elements of the enteric nervous system: MP—myenteric plexus; OSP—intestinal outer submucosal plexus; ISP—intestinal inner submucosal plexus.
Figure 2 Serum glucose levels after citrate buffer (light blue bars) or streptozotocin (dark blue bars) administration (A) at the beginning of the experiment, (B) after 1 week, (C) after 2 weeks, (D) after 3 weeks, (E) after 4 weeks, (F) after 5 weeks, and (G) after 6 weeks.
Figure 3 Schematic diagram of the proportion of perikarya immunoreactive to 5-HT of the control (blue bars) and experimental group (grey bars) in the particular parts of duodenum. OSP—the outer submucosal plexus; ISP—the inner submucosal plexus; MP—the myenteric plexus. * p < 0.05 and *** p < 0.001 indicate differences between all groups for the same neuronal population.
Figure 4 Immunofluorescent microphotographs showing serotonin immunoreactive perikarya in duodenum in the myenteric plexus of the control (A) and in experimental group (B); inner submucosal plexus of the control (C) and in experimental group (D); outer submucosal plexus of the control (E) and in the experimental group (F). (C)—control group; (DM)—diabetes mellitus. Photographs in the right column were created by digital superimposition of two-color channels; Hu C/D (green)- and serotonin-positive (red). The arrows indicated studied cells’ bodies.
Figure 5 Schematic diagram of the proportion of perikarya immunoreactive to 5-HT of the control (blue bars) and experimental group (grey bars) in the particular parts of jejunum. MP—the myenteric plexus; OSP—the outer submucosal plexus; ISP—the inner submucosal plexus. ** p < 0.01 indicates differences between all groups for the same neuronal population.
Figure 6 Immunofluorescent microphotographs showing serotonin immunoreactive perikarya in jejunum in the myenteric plexus of the control (A) and in experimental group (B); inner submucosal plexus of the control (C) and in experimental group (D); outer submucosal plexus of the control (E) and in the experimental group (F). (C)—control group; (DM)—diabetes mellitus. Photographs in the right column were created by digital superimposition of two-color channels; Hu C/D (green)- and serotonin-positive (red). The arrows indicated studied cells’ bodies.
Figure 7 Schematic diagram of the proportion of perikarya immunoreactive to 5-HT of the control (blue bar) and experimental group (grey bars) in the particular parts of ileum. OSP—the outer submucosal plexus; ISP—the inner submucosal plexus; MP—the myenteric plexus. ** p < 0.01 indicates differences between all groups for the same neuronal population.
Figure 8 Immunofluorescent microphotographs showing serotonin immunoreactive perikarya in ileum in the myenteric plexus of the control (A) and in experimental group (B); inner submucosal plexus of the control (C) and in experimental group (D); outer submucosal plexus of the control (E) and in the experimental group (F). (C)—control group; (DM)—diabetes m mellitus. Photographs in the right column were created by digital superimposition of two-color channels; Hu C/D- (green) and serotonin-positive (red). The arrows indicated studied cells’ bodies.
Figure 9 Serotonin-immunoreactive nerve fibers in various parts of mucosal layer in the porcine small intestine. (A) The duodenum in control animals (+++); (B) the duodenum in experimental group (+++); (C) the jejunum in control animals (++); (D) the jejunum in experimental group (++); (E) the ileum in control animals (+); (F) the ileum in experimental group (+). (+)—single nerve fibers; (++)—rare nerve fibers; (+++)—very dense nerve fibers.
Figure 10 Serotonin-immunoreactive nerve fibers in various parts of submucosal layer in the porcine small intestine. (A) The duodenum in control animals (+++); (B) the duodenum in experimental group (+++); (C) the jejunum in control animals (++); (D) the jejunum in experimental group (++); (E) the ileum in control animals (+); (F) the ileum in experimental group (+). (+)—single nerve fibers; (++)—rare nerve fibers; (+++)—very dense nerve fibers.
Figure 11 Negative control for 5-HT.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Hamlin K.E. Fischer F.E. The synthesis of 5-hydroxytryptamine J. Am. Chem. Soc. 1951 73 5007 10.1021/ja01154a551
2. Erspamer V. Asero B. Identification of enteramine, the specific hormone of the enterochromaffin cell system, as 5-hydroxytryptamine Nature 1951 169 800 10.1038/169800b0 14941051
3. Gershon M.D. Drakontides A.B. Ross L.L. Serotonin: Synthesis and release from the myenteric plexus of the mouse intestine Science 1965 149 197 199 10.1126/science.149.3680.197 14305120
4. Gershon M.D. Tack J. The serotonin signaling system: From basic understanding to drug development for functional GI disorders Gastroenterology 2007 132 397 414 10.1053/j.gastro.2006.11.002 17241888
5. Kim D.Y. Camilleri M. Serotonin: A mediator of the brain-gut connection Am. J. Gastroenterol. 2000 95 2698 2709 10.1111/j.1572-0241.2000.03177.x 11051338
6. Banskota S. Ghia J.E. Khan W.I. Serotonin in the gut: Blessing or a curse Biochimie 2019 161 56 64 10.1016/j.biochi.2018.06.008 29909048
7. Koopman N. Katsavelis D. Hove A.S.T. Brul S. de Jonge W.J. Seppen J. The Multifaceted Role of Serotonin in Intestinal Homeostasis Int. J. Mol. Sci. 2021 22 9487 10.3390/ijms22179487 34502396
8. Liu N. Sun S. Wang P. Sun Y. Hu Q. Wang X. The Mechanism of Secretion and Metabolism of Gut-Derived 5-Hydroxytryptamine Int. J. Mol. Sci. 2021 22 7931 10.3390/ijms22157931 34360695
9. Gershon M.D. Serotonin receptors and transporters-roles in normal and abnormal gastrointestinal motility Aliment. Pharmacol. Ther. 2004 20 3 14 10.1111/j.1365-2036.2004.02180.x
10. Hoyer D. Hannon J.P. Martin G.R. Molecular, pharmacological and functional diversity of 5-HT receptors Pharmacol. Biochem. Behav. 2002 71 533 554 10.1016/S0091-3057(01)00746-8 11888546
11. Annaházi A. Berger T.E. Demir I.E. Zeller F. Müller M. Anneser M. Skerra A. Michel K. Schemann M. Metabotropic 5-HT receptor-mediated effects in the human submucous plexus Neurogastroenterol Motil. 2022 19 e14380 10.1111/nmo.14380 35438222
12. Jonnakuty C. Gragnoli C. What do we know about serotonin? J. Cell. Physiol. 2008 217 301 306 10.1002/jcp.21533 18651566
13. Seuwen K. Pouyssegur J. Serotonin as a growth factor Biochem. Pharmacol. 1999 39 985 990 10.1016/0006-2952(90)90276-Q
14. Fanburg B.L. Lee S.L. A new role for an old molecule: Serotonin as a mitogen Am. J. Physiol. 1997 272 L795 L806 10.1152/ajplung.1997.272.5.L795 9176241
15. Coates M.D. Tekin I. Vrana K.E. Mawe G.M. Review article: The many potential roles of intestinal serotonin (5-hydroxytryptamine, 5-HT) signalling in inflammatory bowel disease Aliment. Pharmacol. Ther. 2017 46 569 580 10.1111/apt.14226 28737264
16. Furness J.B. Callaghan B.P. Rivera L.R. Cho H.J. The enteric nervous system and gastrointestinal innervation: Integrated local and central control Microbial Endocrinology: The Microbiota-Gut-Brain Axis in Health and Disease Springer Berlin, Germany 2014 Volume 817 39 71
17. Furness J.B. The organisation of the autonomic nervous system: Peripheral connections Auton. Neurosci. 2006 130 1 5 10.1016/j.autneu.2006.05.003 16798102
18. Schemann M. Neunlist M. The human enteric nervous system Neurogastroenterol. Motil. 2004 16 55 59 10.1111/j.1743-3150.2004.00476.x 15066006
19. Destexhe A. Marder E. Plasticity in single neuron and circuit computations Nature 2004 431 789 795 10.1038/nature03011 15483600
20. Forbes J.M. Cooper M.E. Mechanisms of diabetic complications Physiol. Rev. 2013 93 137 188 10.1152/physrev.00045.2011 23303908
21. Chandrasekharan B. Srinivasan S. Diabetes and the enteric nervous system Neurogastroenterol. Motil. 2007 19 951 960 10.1111/j.1365-2982.2007.01023.x 17971027
22. Bennet H. Balhuizen A. Medina A. Nitert M.D. Laakso E.O. Essen S. Spegel P. Storm P. Krus U. Wierup N. Altered serotonin (5-HT) 1D and 2A receptor expression may contribute to defective insulin and glucagon secretion in human type 2 diabetes Peptides 2015 71 113 120 10.1016/j.peptides.2015.07.008 26206285
23. De Deurwaerdère P. Di Giovanni G. Serotonin in Health and Disease Int. J. Mol. Sci. 2020 21 3500 10.3390/ijms21103500 32429111
24. Callaghan B.C. Cheng H.T. Stables C.L. Smith A.L. Feldman E.L. Diabetic neuropathy: Clinical manifestations and current treatments Lancet Neurol. 2012 11 521 534 10.1016/S1474-4422(12)70065-0 22608666
25. Rodrigues M.L.C. Motta M.E.F.A. Mechanisms and factors associated with gastrointestinal symptoms in patients with diabetes mellitus J. Pediatr. 2012 88 17 24 10.2223/JPED.2153 22344626
26. Wolf E. Braun-Reichhart C. Streckel E. Renner S. Genetically engineered pig models for diabetes research Transgenic Res. 2014 23 27 38 10.1007/s11248-013-9755-y 24065178
27. Larsen M.O. Rolin B. Use of the Goettingen Minipig as a Model of Diabetes with Special Focus on Type 1 Diabetes Research ILAR 2004 45 303 313 10.1093/ilar.45.3.303 15229377
28. Bulc M. Palus K. Zielonka Ł. Gajecka M. Całka J. Changes in expression of inhibitory substances in the intramural neurons of the stomach following streptozotocin- induced diabetes in the pig World J. Gastroenterol. 2017 23 6088 6099 10.3748/wjg.v23.i33.6088 28970724
29. Atkinson W. Lockhart S. Whorwell P. Keevil B. Houghton L. Altered 5-hydroxytryptamine signaling in patients with constipation- and diarrhea-predominant irritable bowel syndrome Gastroenterology 2006 130 34 43 10.1053/j.gastro.2005.09.031 16401466
30. Xie Z. Chang C. Zhou Z. Molecular mechanisms in autoimmune type 1 diabetes: A critical review Clin. Rev. Allergy Immunol. 2014 47 174 192 10.1007/s12016-014-8422-2 24752371
31. Takahara H. Fujimura M. Taniguchi S. Hayashi N. Nakamura T. Fujimiya M. Changes in serotonin levels and 5-HT receptor activity in duodenum of streptozotocin-diabetic rats Am. J. Physiol. Liver Physiol. 2001 281 G798 G808 10.1152/ajpgi.2001.281.3.G798
32. Hansen M.B. Witte A.B. The role of serotonin in intestinal luminal sensing and secretion Acta Physiol. 2008 193 311 323 10.1111/j.1748-1716.2008.01870.x 18462271
33. Suzuki A. Naruse S. Kitagawa M. Ishiguro H. Yoshikawa T. Ko S.B. Yamamoto A. Hamada H. Hayakawa T. 5-hydroxytryptamine strongly inhibits fluid secretion in Guinea pig pancreatic duct cells J. Clin. Investig. 2001 108 749 756 10.1172/JCI12312 11544281
34. Jin J.G. E Foxx-Orenstein A. Grider J.R. Propulsion in Guinea pig colon induced by 5-hydroxytryptamine (HT) via 5-HT4 and 5-HT3 receptors J. Pharmacol. Exp. Ther. 1999 288 93 97 9862758
35. Kendig D.M. Grider J.R. Serotonin and colonic motility Neuro Gastroenterol. Motil. 2015 27 899 905 10.1111/nmo.12617
36. Israelyan N. Del Colle A. Li Z. Park Y. Xing A. Jacobsen J.P.R. Luna R.A. Jensen D.D. Madra M. Saurman V. Effects of Serotonin and Slow-Release 5-Hydroxytryptophan on Gastrointestinal Motility in a Mouse Model of Depression Gastroenterology 2019 157 507 521 10.1053/j.gastro.2019.04.022 31071306
37. Lesurtel M. Soll C. Graf R. Clavien P.A. Role of serotonin in the hepato-gastrointestinal tract: An old molecule for new perspectives Cell Mol. Life. Sci. 2008 65 940 952 10.1007/s00018-007-7377-3 18080089
38. Jones L.A. Sun E.W. Martin A.M. Keating D.J. The ever-changing roles of serotonin Int. J. Biochem. Cell Biol. 2020 125 105776 10.1016/j.biocel.2020.105776 32479926
39. Siriwardena A. Kellum J.M. A 5-HT2 receptor mediates serotonin-induced electrolyte transport in rat left colon J. Surg. Res. 1993 55 323 329 10.1006/jsre.1993.1149 8412118
40. Gatopoulou A. Papanas N. Maltezos E. Diabetic gastrointestinal autonomic neuropathy: Current status and new achievements for everyday clinical practice Eur. J. Intern. Med. 2012 23 499 505 10.1016/j.ejim.2012.03.001 22863425
41. Gonkowski S. Bisphenol A (BPA)-Induced Changes in the Number of Serotonin-Positive Cells in the Mucosal Layer of Porcine Small Intestine-the Preliminary Studies Int. J. Mol. Sci. 2020 21 1079 10.3390/ijms21031079
42. Singh R. Barden A. Mori T. Beilin L. Advanced glycation end-products: A review Diabetologia 2001 44 129 146 10.1007/s001250051591 11270668
43. Yan S.F. Ramasamy R. Schmidt A.M. Mechanisms of disease: Advanced glycation end-products and their receptor in inflammation and diabetes complications Nat. Clin. Pract. Endocrinol. Metab. 2008 4 285 293 10.1038/ncpendmet0786 18332897
44. Matsuda S. Matsuzawa D. Ishii D. Tomizawa H. Sajiki J. Shimizu E. Perinatal exposure to bisphenol A enhances contextual fear memory and acts the serotoninergic system in juvenile female mice Horm. Behav. 2013 63 709 716 10.1016/j.yhbeh.2013.03.016 23567477
|
PMC009xxxxxx/PMC9099900.txt |
==== Front
J Clin Med
J Clin Med
jcm
Journal of Clinical Medicine
2077-0383
MDPI
10.3390/jcm11092567
jcm-11-02567
Article
The Utility of Pentraxin and Modified Prognostic Scales in Predicting Outcomes of Patients with End-Stage Heart Failure
https://orcid.org/0000-0001-9736-422X
Szczurek-Wasilewicz Wioletta 1*
https://orcid.org/0000-0003-3410-7407
Skrzypek Michał 2
https://orcid.org/0000-0002-2555-8102
Romuk Ewa 3
Gąsior Mariusz 4
Szyguła-Jurkiewicz Bożena 4
Triposkiadis Filippos Academic Editor
Iacoviello Massimo Academic Editor
1 Silesian Center for Heart Diseases in Zabrze, 41-800 Zabrze, Poland
2 Department of Biostatistics, School of Public Health in Bytom, Medical University of Silesia, 40-055 Katowice, Poland; [email protected]
3 Department of Biochemistry, School of Medical Sciences in Zabrze, Medical University of Silesia, 40-055 Katowice, Poland; [email protected]
4 3rd Department of Cardiology, School of Medical Sciences in Zabrze, Medical University of Silesia, 40-055 Katowice, Poland; [email protected] (M.G.); [email protected] (B.S.-J.)
* Correspondence: [email protected]; Tel.: +48-694138970
04 5 2022
5 2022
11 9 256703 4 2022
29 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Risk stratification is an important element of management in patients with heart failure (HF). We aimed to determine factors associated with predicting outcomes in end-stage HF patients listed for heart transplantation (HT), with particular emphasis placed on pentraxin-3 (PXT-3). In addition, we investigated whether the combination of PTX-3 with the Heart Failure Survival Score (HFSS), the Seattle Heart Failure Model (SHFM), or the Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) improved the prognostic strength of these scales in the study population. We conducted a prospective analysis of 343 outpatients with end-stage HF who accepted the HT waiting list between 2015 and 2018. HFSS, SHFM, and MAGGIC scores were calculated for all patients. PTX3 was measured by sandwich enzyme-linked immunosorbent assay with a commercially available kit. The endpoints were death, left ventricular assist device implantation, and HT during the one-year follow-up. The median age was 56 (50–60) years, and 86.6% were male. During the follow-up period, 173 patients reached the endpoint. Independent risk factors associated with outcomes were ischemic etiology of HF [HR 1.731 (1.227–2.441), p = 0.0018], mean arterial pressure (MAP) [1.026 (1.010–1.042), p = 0.0011], body mass index (BMI) [1.055 (1.014–1.098), p = 0.0083], sodium [1.056 [(1.007–1.109), p = 0.0244] PTX-3 [1.187 (1.126–1.251, p < 0.0001) and N-terminal pro-brain natriuretic peptide (NT-proBNP) [HR 1.004 (1.000–1.008), p = 0.0259]. The HFSS-PTX-3, SHFM-PTX-3 and MAGGIC-PTX-3 scores had significantly higher predictive power [AUC = 0.951, AUC = 0.973; AUC = 0.956, respectively] than original scores [AUC for HFSS = 0.8481, AUC for SHFM = 0.7976, AUC for MAGGIC = 0.7491]. Higher PTX-3 and NT-proBNP concentrations, lower sodium concentrations, lower MAP and BMI levels, and ischemic etiology of HF are associated with worse outcomes in patients with end-stage HF. The modified SHFM-PTX-3, HFSS-PTX-3, and MAGGIC-PTX-3 scores provide effective methods of assessing the outcomes in the analyzed group.
pentraxin-3
heart failure
scales
risk stratification
the Medical University of SilesiaPCN-1-015/N/1/K This work was supported by a grant (PCN-1-015/N/1/K) from the Medical University of Silesia.
==== Body
pmc1. Introduction
Despite major drug and device therapy advances, the number of patients with end-stage heart failure (HF) requiring a heart transplant (HT) is gradually increasing [1,2]. The limited supply of donor hearts requires the need to search for new and simple tools that will facilitate patient allocation to the waiting list for HT. Over the years, many different prognostic models have been proposed to improve mortality risk assessment, optimize treatment and promote more effective use of therapy, each with its own set of advantages and limitations [3,4,5,6]. The widely used prognostic scale is the Heart Failure Survival Score (HFSS), which includes clinical variables and maximal oxygen uptake [3]. The Seattle Heart Failure Model (SHFM) offers a comprehensive risk assessment, including device therapy [4]. The comorbid conditions have been included in the Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) score, showing their significant utility in predicting patient outcomes [5]. Some biomarkers have also been shown to improve the diagnostic accuracy of preexisting models [7,8,9]. Despite these prognostic tools, there might still be room for further refinement of risk stratification through the use of new simple biomarkers and scales related to HF. Biomarkers that reflect the grade of inflammation and myocardial remodeling, such as pentraxin (PTX-3), may provide important information on risk stratification and monitoring HF therapy [10].
PTX-3 is a member of a superfamily of multimeric pattern-recognition proteins that play important roles at the interface of the innate immune response, inflammation, and extracellular matrix remodeling [11,12,13,14]. PTX-3 is produced by different vascular and inflammatory cells in response to primary inflammatory stimuli and might reflect local inflammatory status in the cardiovascular system [11,12,14]. The presence of PTX3 was detected in the myocardium in various pathological conditions, which was parallel to the observation of increased levels of PTX3 in plasma in patients with cardiovascular disorders [13]. However, the clinical significance of the plasma PTX3 levels in end-stage HF referred for HT has not been fully established.
Because of the potential relationship between PTX-3 and the pathogenesis of HF, we aimed to determine factors associated with predicting outcomes in end-stage HF patients listed for HT, with particular emphasis placed on PTX-3. In addition, we investigated whether the combination of PTX-3 with SHFM, HFSS, or MAGGIC score improves the prognostic strength of these scales in our study population.
2. Materials and Methods
We conducted a prospective analysis of 383 outpatients with end-stage HF who were referred to our center and underwent qualification for HT between 2015 and 2018. Patients with contraindications to HT were excluded from the study (n = 40). In the analyzed patients, medical history, anthropometric measurements, physical examinations, transthoracic echocardiographic measurements, ergospirometric exercise tests, right heart catheterization, and a panel of laboratory tests were performed. All patients received standard medical treatment, including angiotensin-converting-enzyme inhibitors/angiotensin II receptor blockers, mineralocorticoid receptor antagonists, and beta-blockers, at the maximum tolerated doses for at least 3 months prior to study inclusion.
The glomerular filtration rate was estimated according to the Modification of Diet in Renal Disease equation.
The study was approved by the Bioethical Committee of the Medical University of Silesia (specific ethics code—KNW/0022/KB1/88/15, date of approval: 7 July 2015). The study conformed to the principles outlined in the Declaration of Helsinki on the ethical principles for medical research involving human subjects. Written informed consent was obtained from all included patients.
2.1. Laboratory Measurements
Venous blood samples were obtained under stable and fasting conditions to measure the serum levels of the laboratory test panel. The complete blood count and hematologic parameters were determined using automated blood cell counters (Sysmex XS1000i and XE2100, Sysmex Corporation, Kobe, Japan). Liver and kidney function parameters, as well as cholesterol and albumin plasma levels, were measured with a COBAS Integra 800 analyzer (Roche Instrument Center AG, Rotkreuz, Switzerland). A highly sensitive latex-based immunoassay was used to detect plasma C-reactive protein (CRP) with a Cobas Integra 70 analyzer (Roche Diagnostics, Ltd., Rotkreuz, Switzerland). The plasma concentration of fibrinogen was measured using an STA Compact analyzer (Roche). The plasma concentration of N-terminal prohormone of brain natriuretic peptide (NT-proBNP) was measured with a commercially available kit from Roche Diagnostics (Mannheim, Germany) on an Elecsys 2010 analyzer. Human PTX3 was measured by sandwich enzyme-linked immunosorbent assay (ELISA) with a commercially available kit (Human PTX3 ELISA Kit, SunRedBio Technology Co, Ltd., Shanghai, China). The concentration of PTX3 was expressed as ng/mL. The sensitivity of the assay was 0.051 ng/mL. The assay range was 0.08 ng/mL–20 ng/mL. This ELISA test was performed using a BioTek Elx50 reader (BioTek Instruments Inc., Tecan Group, Männedorf, Switzerland).
2.2. Scales
Three HF prognostic scores were analyzed in the entire cohort:
- The HFSS score was calculated based on the following equation incorporating seven variables: ([0.0216 × resting heart rhythm] + [−0.0255 × mean arterial blood pressure (MAP)] + [−0.0464 × left ventricular ejection fraction (LVEF)] + [−0.0470 × serum sodium] + [−0.0546 × peak VO2 ] + [0.6083 × presence (1) or absence (0) of interventricular conduction defect (QRS duration ≥ 0.12 due to any cause)] + [0.6931 × presence (1) or absence (0) of ischemic cardiomyopathy]), as described previously [3].
SHFM was derived on the basis of the original risk factor coefficients as described by Levy et al. The SHFM includes 10 continuous variables (age, LVEF, New York Heart Association (NYHA) class, systolic blood pressure, diuretic dose adjusted for weight, lymphocyte count, hemoglobin, serum sodium, total cholesterol, and uric acid) and 10 categorical variables (gender, ischemic cardiomyopathy, use of device therapy (implantable cardioverter-defibrillator, cardiac resynchronization therapy), use of beta-blockers, angiotensin-converting enzyme inhibitor, angiotensin receptor blockers, potassium-sparing diuretic, statins, and allopurinol) in an equation that provides a continuous risk score for each patient [4].
- The MAGGIC score [5] was developed from 13 routinely available patient characteristics: (1) age, (2) sex, (3) body mass index (BMI), (4) systolic blood pressure, (5) creatinine concentration, (6) presence or absence of diabetes mellitus and (7) chronic obstructive pulmonary disease, (8) HF diagnosed in the last 18 months, (9) NYHA class, (10) LVEF, (11) current smoking status, (12) b-blockers, and (13) angiotensin-converting enzyme inhibitors or angiotensin receptor blockers. From 18 September 2013, the integer score increased by 2 if HF was diagnosed >18 months ago, which is reflected in our analysis.
To assess the ability of PTX-3 to improve the prognostic values of the scales, new combined scores were created. The scores for HFSS and PTX-3, MAGGIC and PTX-3, as well as for SHFM and PTX-3, were included in the Cox regression model as continuous variables, and each variable was multiplied by its corresponding β-coefficient. The final scores for new scales were calculated based on the following formulas:HFSS-PTX-3 = 0.1595 ∗ PTX3 − 0.9743 ∗ HFSS
MAGGIC-PTX-3 = 0.2025 ∗ PTX + 0.0943 ∗ MAGGIC
SHFM-PTX-3 = 0.2014 ∗ PTX + 0.3757 ∗ SHFM
The raw score for HFSS-PTX-3 was multiplied by (−1) to achieve a positive value and facilitate the interpretation of the results.
2.3. Outcome Data
The composite outcome was represented by death, lvad left ventricular assist device (LVAD), implantation, and HT during the one-year follow-up. Follow-up was performed according to the local HF program, with regular physician’s office visits (every 6 months). The information about the one-year mortality was based on the data obtained from the national health care provider.
2.4. Statistical Analysis
The statistical analysis was performed using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA). Categorical variables are presented as counts and percentages. Continuous variables were evaluated for normal distribution assumption using the Kolmogorov–Smirnov and Shapiro–Wilk tests and were reported as the mean plus standard deviation in brackets or the median with lower and upper quartiles. Differences between the study groups were assessed using Student’s t-test, the Mann–Whitney test, or the χ2 test. The prognostic utility of each score was quantified by the area under the ROC curve (AUC), sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), negative likelihood ratio (LR-), positive likelihood ratio (LR+) and accuracy. Comparison between areas under the curve (AUCs) was achieved with the method used by DeLong et al. The differences between the AUC values were tested using the Hanley and McNeil method. The Spearman rank correlation test was used for correlation analysis. The tolerance and variance inflation factor were used to assess the correlation between explanatory variables and to assess multicollinearity. Schoenfeld residuals were used to check the proportional hazards assumption. Cox proportional hazard regression analysis was used to determine which variables were significantly related to the composite endpoint. Only variables with p values less than 0.20 in the univariable Cox regression analysis were entered into the multivariable Cox regression analysis. A p value ≤ 0.05 was considered statistically significant.
3. Results
The final study group consisted of 343 patients with end-stage HF awaiting HT classified into NYHA functional classes III and IV (87.2% and 12.8%, respectively) and profiles 4 to 6 according to the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) classification. The demographic and clinical characteristics of patients in the pooled population and divided into event and nonevent groups are presented in Table 1.
During the one-year follow-up, 109 (31.8%) deaths occurred, 35 (20.2%) patients underwent HTx, and 29 (8.5%) received LVAD.
The ROC curves for each score, NT-proBNP and PTX-3 are shown in Figure 1 and Figure 2.
A summary of the ROC curves analysis is shown in Table 2.
A comparison of the area under the ROC curves for the combined scales and their components is presented in Table 3.
The univariable and multivariable Cox proportional hazards analyses to predict composite endpoints for PTX3 and other variables are shown in Table 4. With the multivariable Cox proportional hazard analysis, ischemic etiology of HF, lower levels of mean arterial pressure MAP, BMI, and sodium, as well as higher levels of PTX-3 and NT-proBNP were independent predictors of the composite endpoint.
4. Discussion
This single-center study revealed an independent association between serum PTX-3 and worse outcomes in patients with advanced HF awaiting HT. PTX-3 serum concentrations allow for the accurate risk stratification of one-year outcomes in the analyzed group of patients. Previous studies have also confirmed the prognostic utility of PTX-3 in the assessment of outcomes in patients with chronic HF with reduced ejection fraction [12,15,16]. Kotooka et al. showed that higher plasma PTX3 levels are associated with a high risk of cardiac events in patients with HF. In addition, they also reported that PTX3 expression was higher in myocardial biopsy samples from HF patients compared to the control group [15]. In turn, Suzuki et al. demonstrated that plasma PTX3 concentration was increased in patients with HF compared to the control group and was also an independent predictor of cardiac events in HF patients [16]. Latili et al. also showed that baseline PTX3 levels and three-month changes in PTX3 levels were independently associated with worse outcomes in patients with chronic and stable HF. However, the authors demonstrated that after the addition of NT-proBNP to the prognostic model with PTX-3, only changes in PTX-3 concentrations were associated with outcomes [16].
HF is a systemic disorder that is associated with the activation of the inflammatory and immune systems [12,17,18]. Previous evidence has shown that activation of the inflammatory and immune systems may play an important role in HF [17,18,19]. From the pathophysiological point of view, PTX-3 can be associated with the development and progression of HF, especially due to its important regulatory role in inflammation, extracellular matrix organization, and remodeling [20,21,22]. PTX3 is produced by a variety of cell types, including monocytes/macrophages, vascular endothelial cells, vascular smooth muscle cells, adipocytes, fibroblasts, and dendritic cells [20,21,22]. Unlike CRP, which is synthesized in the liver and reflects systemic inflammation, PTX-3 is partially synthesized at the site of inflammation and released into the circulation, thus reflecting local inflammation in the cardiovascular system [16]. The main inducers of PTX-3 production are primary proinflammatory signals, such as interleukin-1β (IL-1β), TNFα, or bacterial molecules that engage Toll-like receptors (TLRs) [20]. In turn, elevated concentrations of these cytokines are observed in patients with HF in plasma and circulating leukocytes, as well as in the failed myocardium itself [17,21]. In addition, IL1β and TNFα are mediators involved in processes that lead to the remodeling of the heart, such as fibrosis and apoptosis [21]. The activation of the inflammatory system plays an important role in the pathogenesis of HF and is associated with an increase in plasma inflammatory cytokine levels, which stimulate the production of PTX-3. In turn, PTX-3 modulates inflammation in several cells, including endothelial cells, smooth muscle cells, and fibroblasts, and enhances further remodeling, which contributes to the intensification of the unfavorable cascade of changes in the heart muscle [12,15,20,21,22]. Another important property of PTX-3 is its ability to activate the classical complement activation pathway by binding the complement component C1q [23]. In turn, complement activation affects many processes related to the development and progression of HF, such as promotion of endothelial cell activation, monocyte infiltration into the extracellular matrix, and stimulation of cytokine release [24]. PTX3, by binding to the gamma Fc receptor, also influences the activation of MAP kinases, ERK1/2 and NF-κB proteins, which play an important role in heart remodeling [25,26].
Another interesting finding of our study is that PTX-3 may improve the prognostic value of recognized prognostic scales in patients with HF. The present study is the first to demonstrate that the modified SHFM-PTX-3, HFSS-PTX-3, and MAGGIC-PTX-3 scores provide effective methods of assessing the outcomes in patients with advanced HF awaiting HT. An improvement in prognostic power was observed in the SHFM-PTX-3 score relative to those of individual components. In turn, HFSS-PTX-3 and MAGGIC-PTX-3 had a significant improvement in prognostic power compared to the original scales.
This is an important finding because accurate risk stratification in patients with HF can prevent delays in the appropriate treatment of high-risk patients or the overtreatment of patients with low risk [27]. Sometimes there are difficulties in estimating the risk of death because of the multiplicity of risk factors related to HF and personal beliefs [28]. Therefore, objective risk scales are needed to assess the prognosis of patients with HF. There are many risk scales available; however, only two scales–the HFSS and SHFM–are included in the ISHLT guidelines for HT as prognostic tools in groups of patients with end-stage HF awaiting HT [28]. A relatively new scale, the MAGGIC, was originally developed in 2012 by Pocock et al. from a cohort of 39,372 patients with HF [5] and was confirmed in several external studies to have a discriminatory power ranging from 0.67–0.80 [5,7,29]. Our study showed for the first time that the prognostic power of the SHFM, HFSS, and MAGGIC scores were significantly improved when PTX-3 was added to the models. However, only in the case of SHFM-PTX-3 was better predictive power compared to both components observed. Some studies also showed that the prognostic power of HFSS, SHFM, and MAGGIC scores could be improved by adding other significant parameters associated with worse prognosis in patients with HF [5,7,9,30,31,32]. It seems that the modified risk scores may better stratify the outcomes by considering important risk factors in the current population of patients with HF and facilitate appropriate decisions regarding HF therapy.
Our study also confirms the importance of conventional HF risk factors. Lower sodium concentrations, higher NT-proBNP concentrations, lower BMI levels, lower MAP, and ischemic etiology of HF were also associated with an increased risk of worse outcomes during a one-year follow-up in the analyzed group of patients.
Lower serum sodium is an important and well-known factor of worse prognosis in patients with HF [4,33,34]. Moreover, sodium concentrations are also a component of some prognostic models in patients with HF [8,35]. Another factor related to a worse prognosis in our population, NT-proBNP, is one of the most widely researched and used biomarkers in everyday practice [8,34,36]. Many studies have confirmed the importance of NT-proBNP as an indicator of mortality and morbidity in various HF patient populations [8,36]. The inverse relationship between BMI and prognosis in HF is well known as the “obesity paradox”, in which a lower BMI is a factor in worse outcomes in HF [35,37]. Furthermore, BMI is one of the parameters of the MAGGIC score [5]. Similar paradoxical relationships are observed for blood pressure because lower SBP and MAP levels are associated with worse outcomes in HF [3,38,39]. Our results are completely in line with this. In addition, the MAP value is an important factor in the HFSS score, and SBP is included in the SHFM score [4]. The ischemic etiology of HF is also a well-known predictive factor of worse outcomes in patients with HF and has been widely discussed in the literature [4,40,41]. Moreover, ischemic etiology is a component of both the HFSS and the SHFM [3,4].
Limitations
This single-center study has several limitations. Our study analyzed only the baseline PTX3 concentration at the time of inclusion in the study, while serial measurements of PTX-3 concentration over time might be more useful for evaluating one-year outcomes in ambulatory patients with HF. Furthermore, there was a lack of an independent validation cohort that could support our results. It is likely that if an independent validation cohort were used, the AUC for PTX-3 and other analyzed parameters would be lower. Although the size of the study group is relatively large for a single-center study, it can be considered small on the epidemiological scale. Considering the intrinsic limitations related to a single-center study and a small sample size, further multicenter studies with a large study population are necessary to confirm the clinical significance of PTX3 in the population of patients with HF.
5. Conclusions
In summary, our study showed that PTX-3 is a strong independent predictor of worse outcomes in end-stage HF patients awaiting HT. PTX-3 serum concentrations with excellent predictive power, sensitivity, and specificity allow for the accurate risk stratification of one-year outcomes in the analyzed group of patients. Furthermore, PTX-3 may improve the prognostic value of recognized scales, and the modified SHFM-PTX-3, HFSS-PTX-3, and MAGGIC-PTX-3 scores provide effective methods of assessing the outcomes in patients with advanced HF awaiting HT. Our study also confirms an independent association between conventional HF risk factors: higher NT-proBNP concentrations, lower sodium concentrations, lower MAP, lower BMI levels, ischemic etiology of HF, and worse outcomes in analyzed populations.
Author Contributions
Conceptualization, W.S.-W. and B.S.-J.; data curation, W.S.-W., M.S., E.R., M.G. and B.S.-J.; formal analysis, W.S.-W. and M.S.; funding acquisition, M.G. and B.S.-J.; investigation, W.S.-W., M.S., E.R., M.G. and B.S.-J.; methodology, W.S.-W. and B.S.-J.; project administration, W.S.-W. and B.S.-J.; software, M.S.; supervision, B.S.-J.; visualization, W.S.-W.; writing—original draft, W.S.-W. and B.S.-J.; writing—review & editing, W.S.-W., M.G. and B.S.-J. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of the Medical University of Silesia in Katowice (protocol code: KNW/0022/KB1/88/15 and date of approval: 7 July 2015).
Informed Consent Statement
Informed consent was obtained from all subjects involved in the study.
Data Availability Statement
The data presented in this study are available on request from the corresponding author.
Conflicts of Interest
The authors declare no conflict of interest.
Figure 1 The ROC curves for HFSS (A), SHFM (B), MAGGIC (C), PTX-3 (D) and NT-proBNP (E). Abbreviations: AUC, are under the curve; HFSS— Heart Failure Survival Score; MAGGIC—Meta-Analysis Global Group in Chronic Heart Failure; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; PTX-3, pentraxin-3; SHFM, Seattle Heart Failure Model; SHFM and MAGGIC scores revealed acceptable discrimination ability at 1 year of observation (AUC between 0.7 and 0.8), whereas the HFSS score showed good discrimination (AUC 0.85). PTX-3 displayed superior discriminative power against HFSS, MAGGIC, and SHFM scores for the composite endpoint.
Figure 2 The ROC curves for HFSS + PTX-3 (A), SHFM + PTX-3 (B), MAGGIC + PTX-3 (C). Abbreviations: AUC, are under the curve; HFSS—Heart Failure Survival Score; MAGGIC—Meta-Analysis Global Group in Chronic Heart Failure; PTX-3, pentraxin-3; SHFM, Seattle Heart Failure Model. The HFSS-PTX-3, SHFM-PTX-3, and MAGGIC-PTX-3 scores generated excellent power to predict the composite endpoint (AUC > 0.90, p < 0.001). It is worth mentioning that combined scales reached high sensitivity, specificity, PPV, NPV, and accuracy and generated good results in terms of likelihood ratios. An improvement in AUC and p values for the composite endpoint was observed in the SHFM-PTX-3 score relative to those of individual components. In turn, HFSS-PTX-3 and MAGGIC-PTX-3 had a significant improvement in AUCs compared to the original scales. However, the prognostic power of HFSS-PTX-3 and MAGGIC-PTX-3 was comparable to that of PTX3.
jcm-11-02567-t001_Table 1 Table 1 Baseline population characteristics and comparison between alive and events group.
Overall Population
N = 343 # Patients without Events
N = 170 Patients with Events
N = 173 p
Baseline data
Age, years 56 (50–60) 56 (49–61) 56 (50–60) 0.7533
Male, n (%) 297 (86.6) 150 (88.2) 147 (85) 0.3751
Ischemic etiology of HF, n (%) 199 (58) 73 (42.9) 126 (72.8) <0.0001 *
SBP, mmHg 102.00 (92.00–116.00) 113.00 (100.00–120.00) 98.00 (90.00–105.00) <0.0001 *
MAP, mmHg 76.67 (71.67–85.33) 81.33 (76.00–90.00) 73.33 (68.67–78.33) <0.0001 *
BMI, kg/m2 26.93 (23.85–30.08) 27.47 (24.49–31.21) 26.15 (23.25–29.05) 0.0002 *
NYHA III, n (%) 299 (87.2) 163 (95.9) 136 (78.6) <0.0001 *
NYHA IV, n (%) 44 (12.8) 7 (4.1) 37 (21.4)
Comorbidities
Hypertension, n (%) 168 (49) 82 (48.2) 86 (49.7) 0.7846
Type 2 diabetes, n (%) 177 (51.6) 80 (47.1) 97 (56.1) 0.095
Persistent FA, n (%) 160 (46.6) 85 (50) 75 (43.4) 0.2173
COPD, n (%) 42 (12.2) 20 (11.8) 22 (12.7) 0.788
Laboratory parameters
WBC, ×109/L 7.18 (6.02–8.46) 6.96 (5.84–8.27) 7.33 (6.21–8.72) 0.1256
Lymphocytes, % 24.00 (17.70–30.06) 22.75 (17.80–28.60) 25.10 (17.70–32.50) 0.1179
Hemoglobin, mmol/L 8.80 (8.20–9.60) 8.80 (8.20–9.50) 8.90 (8.20–9.70) 0.4866
Creatinine, µmol/L 108.00 (93.00–126.00) 103.00 (88.00–121.00) 113.00 (102.00–134.00) <0.0001 *
GFR, mL/min/1.73 m2 61.78 (51.63–75.73) 68.11 (55.49–81.65) 56.81 (50.15–68.78) <0.0001 *
Platelets, ×109/L 197.00 (172.00–228.00) 193.00 (171.00–220.00) 206.00 (175.00–237.00) 0.0317 *
Total bilirubin, µmol/L 18.40 (1220–24.10) 17.35 (11.30–21.90) 20.00 (13.40–25.90) 0.0021 *
Albumin, g/L 44.00 (41.00–46.00) 44.00 (42.00–46.00) 43.00 (41.00–46.00) 0.0445 *
Uric acid, µmol/L 441.00 (371.00–526.00) 403.50 (339.00–483.00) 470.00 (403.00–565.00) <0.0001 *
Urea, µmol/L 8.10 (5.90–12.60) 7.35 (5.60–10.30) 8.90 (6.20–13.80) 0.0059 *
Sodium, mmol/L 139.00 (136.00–140.00) 140.00 (139.00–141.00) 137.00 (135.00–139.00) <0.0001 *
Fibrinogen 379.00 (312.00–443.00) 363.50 (296.00–424.00) 396.00 (330.50–483.50) 0.001 *
AST, U/L 26.00 (20.00–31.00) 26.00 (20.00–31.00) 25.00 (20.00–33.00) 0.6419
ALT, U/L 22.00 (16.00–32.00) 23.00 (17.00–33.00) 21.00 (16.00–30.00) 0.1087
ALP, U/L 78.00 (62.00–102.00) 75.50 (58.00–101.00) 81.00 (65.00–102.00) 0.0642
GGTP, U/L 73.00 (35.00–125.00) 64.50 (34.00–111.00) 84.00 (36.00–137.00) 0.0256 *
Cholesterol, mmol/L 4.54 (4.16–5.00) 4.43 (4.02–4.86) 4.62 (4.22–5.15) 0.0019 *
LDL, mmol/L 2.05 (1.58–2.83) 2.04 (1.55–2.73) 2.06 (1.61–2.93) 0.43
hs-CRP, mg/L 3.40 (1.64–8.75) 2.75 (1.50–5.42) 4.52 (2.06–10.75) 0.0003 *
HBA1c, % 5.80 (5.40–6.30) 5.80 (5.40–6.30) 5.80 (5.30–6.30) 0.2926
NT-proBNP, pg/mL 4334.00 (1965.00–7102.00) 3023.00 (1743.00–6101.00) 5539.00 (2916.00–8310.00) <0.0001 *
Pentraxin-3, ng/mL 3.65 (2.57–6.23) 2.58 (2.10–3.25) 6.22 (5.12–8.66) <0.0001 *
Haemodynamic parameters
mPAP, mmHg 23.00 (17.00–30.00) 23.50 (17.00–30.00) 22.00 (18.00–30.00) 0.7619
CI, l/min/m2 1.84 (1.72–1.94) 1.83 (1.70–1.94) 1.84 (1.72–1.95) 0.4959
TPG, mmHg 8.00 (6.00–10.00) 8.00 (6.00–10.00) 8.00 (6.00–10.50) 0.9387
PVR, Wood units 1.97 (1.47–2.36) 1.99 (1.56–2.35) 1.95 (1.41–2.35) 0.6186
Echocardiographic parameters
LA, mm 52.00 (48.00–56.00) 52.00 (46.00–56.00) 53.00 (49.00–57.00) 0.0708
RVEDd, mm 34.00 (30.00–40.00) 33.00 (30.00–40.00) 34.00 (31.00–41.00) 0.065
LVEDd, mm 73.00 (68.00–80.00) 73.00 (68.00–80.00) 73.00 (69.00–81.00) 0.2354
LVEF, % 18.00 (15.00–20.00) 19.00 (16.00–21.00) 17.00 (15.00–20.00) <0.0001 *
Treatment
B-blockers, n (%) 320 (93.3) 161 (94.7) 159 (91.9) 0.7176
ACEI, n (%) 244 (71.1) 127 (74.7) 117 (67.6) 0.1482
ARB, n (%) 74 (21.6) 31 (18.2) 43 (24.9) 0.1361
Loop diuretics, n (%) 343 (100) 170 (100) 173 (100)
MRA, n (%) 322 (93.9) 159 (93.5) 163 (94.2) 0.7898
Digoxin, n (%) 102 (29.7) 51 (30) 33 (19.1) 0.9161
Ivabradine, n (%) 63 (18.4) 30 (17.8) 33 (19.1) 0.7522
Statin, n (%) 261 (76.1) 135 (79.4) 126 (72.8) 0.1532
Coumarin derivatives, n (%) 186 (54.2) 93 (54.7) 93 (53.8) 0.86
Acetylsalicylic acid, n (%) 124 (36.2) 59 (34.7) 65 (37.6) 0.5806
Allopurinol, (n%) 163 (47.5) 76 (44.7) 87 (50.3) 0.3006
ICD n (%) 201 (58.6) 97 (57.1) 104 (60.1) 0.5655
CRT-D n (%) 142 (41.4) 73 (42.9) 69 (39.9)
Other parameter
VO2 max, mL/kg/min 10.80 (10.00–11.50) 10.90 (10.20–11.60) 10.70 (9.70–11.30) 0.0216 *
Current smoker, % 40 (11.7) 13 (7.6) 27 (15.6) 0.0217 *
QRS > 0.12 s 136 (39.7) 52 (30.6) 84 (48.6) 0.0007 *
Scales
MAGGIC score 26.00 (24.00–29.00) 25.00 (23.00–27.00) 28.00 (26.00–30.00) <0.0001 *
SHFM score 0.43 (−0.004–0.907) 0.10 (−0.25–0.46) 0.73 (0.39–1.26) <0.0001 *
HFSS score 7.65 (7.22–8.20) 8.08 (0.52) 7.31 (0.54) <0.0001 *
# Data are presented as medians (25th–75th percentile) or numbers (percentage) of patients. * p < 0.05 (statistically significant). Abbreviations: ACEI, angiotensin-converting-enzyme inhibitor; ALP, alkaline phosphatase; ALT, alanine aminotransferase; ARB, angiotensin II receptor blocker; AST, aspartate aminotransferase; BMI, body mass index; CI, cardiac index; COPD—chronic obstructive pulmonary disease; CRT-D, cardiac resynchronization therapy-defibrillator; FA, atrial fibrillation; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; GFR, glomerular filtration rate; GGTP, gamma-glutamyl transpeptidase; HBA1c, glycated hemoglobin; HF, heart failure; ; HFSS—Heart Failure Survival Score; HR, heart-rhythm; hs-CRP, high-sensitivity C-reactive protein; ICD, implantable cardioverter-defibrillator; LA, left atrium; LDL, low density lipoprotein; ; LVEDd, left ventricular end-diastolic dimension; LVEF, left ventricular ejection fraction; MAGGIC—Meta-Analysis Global Group in Chronic Heart Failure; MAP, mean arterial pressure; mPAP, mean pulmonary artery pressure; MRA, mineralocorticoid receptor antagonists; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; PH, pulmonary hypertension; PVR, pulmonary vascular resistance; RVEDd, right ventricular end-diastolic dimension; SBP, systolic blood pressure; SHFM, Seattle Heart Failure Model; sPAP, systolic pulmonary artery pressure; TPG, transpulmonary gradient; Vo2 max, maximal oxygen uptake; WBC, white blood cells.
jcm-11-02567-t002_Table 2 Table 2 A summary of ROC curves analysis for analyzed parameters.
AUC
[±95 CI] p Cut-off Sensitivity
[±95 CI] Specificity
[±95 CI] PPV
[±95 CI] NPV
[±95 CI] LR+
[±95 CI] LR-
[±95 CI] Accuracy
HFSS 0.8481
[0.8079–0.8883] <0.0001 <7.86 0.88
[0.82–0.92] 0.66
[0.59–0.74] 0.73
[0.66–0.79] 0.84
[0.77–0.90] 2.62
[2.04–3.20] 0.18
[0.11–0.26] 0.77
[0.72–0.82]
SHFM 0.7976
[0.7510–0.8442] <0.0001 ≥0.299 0.80
[0.73–0.85] 0.66
[0.58–0.73] 0.70
[0.63–0.77] 0.76
[0.68–0.83] 2.34
[1.82–2.86] 0.31
[0.21–0.40] 0.73
[0.68–0.78]
MAGGIC 0.7491
[0.6979–0.8003] <0.0001 ≥27 0.69
[0.62–0.76] 0.70
[0.63–0.77] 0.70
[0.63–0.77] 0.69
[0.62–0.76] 2.31
[1.73–2.89] 0.44
[0.33–0.55] 0.70
[0.65–0.75]
PTX-3 0.9558
[0.9345–0.9772] <0.0001 ≥3.926 0.88
[0.83–0.93] 0.95
[0.91–0.98] 0.95
[0.91–0.98] 0.89
[0.84–0.93] 18.79
[5.95–31.63] 0.12
[0.07–0.17] 0.92
[0.88–0.94]
NT-proBNP 0.6598
[0.6024–0.7171] <0.0001 ≥3136 0.73
[0.66–0.79] 0.52
[0.44–0.59] 0.61
[0.54–0.67] 0.65
[0.56–0.73] 1.51
[1.24–1.78] 0.52
[0.37–0.67] 0.62
[0.57–0.68]
HFSS+
PTX-3 0.9508
[0.9277–0.9838] <0.0001 <6.772 0.89
[0.83–0.93] 0.91
[0.85–0.95] 0.91
[0.85–0.94] 0.89
[0.83–0.93] 9.46
[4.99–0.1393] 0.12
[0.07–0.17] 0.90
[0.86–0.93]
SHFM+
PTX3 0.9727
[0.9588–0.9867] <0.0001 ≥1.062 0.89
[0.83–0.93] 0.95
[0.90–0.98] 0.94
[0.90–0.97] 0.89
[0.84–0.94] 16.81
[6.01–27.61] 0.12
[0.07–0.17] 0.92
[0.88–0.95]
MAGGIC+
PTX-3 0.9562
[0.9354–0.9770] <0.0001 ≥3.388 0.86
[0.79–0.90] 0.96
[0.92–0.98] 0.95
[0.91–0.98] 0.87
[0.81–0.91] 20.78
[5.55–36.00] 0.15
[0.10–0.21] 0.91
[0.87–0.94]
Abbreviations: see Table 1, AUC, area under the curve; LR−, negative likelihood ratio; LR+, positive likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.
jcm-11-02567-t003_Table 3 Table 3 A comparison of the area under the ROC curves for the combined scales and their components.
HFSS-PTX-3, AUC [±95 CI] 1 p
HFSS, AUC [±95 CI] 0.1027 [0.0754–0.1299] <0.0001
PTX-3, AUC [±95 CI] −0.0051[−0.0285–0.0184] 0.6720
SHFM-PTX-3, AUC [±95 CI] p
SHFM, AUC [±95 CI] 0.1751 [0.1320–0.2183] <0.0001
PTX-3, AUC [±95 CI] 0.0169 [0.0014–0.0324] 0.0330
MAGGIC-PTX-3, AUC [±95 CI] p
MAGGIC, AUC [±95 CI] 0.2071 [0.1637–0.2505] <0.0001
PTX-3, AUC [±95 CI] 0.0004 [−0.0186–0.0194] 0.9693
Abbreviations: see Table 1; AUC, area under the curve. 1 The difference between AUCs.
jcm-11-02567-t004_Table 4 Table 4 Univariable and multivariable Cox proportional hazard analysis of factors associated with events.
Parameter Univariable Data Multivariable Data
Etiology of HF 2.075 [1.477–2.914] <0.0001 1.731 [1.227–2.441] 0.0018
MAP (−) 1.044 [1.030–1.059] <0.0001 1.026 [1.010–1.042] 0.0011
BMI (−) 1.073 [1.034–1.114] 0.0002 1.055 [1.014–1.098] 0.0083
NYHA IV (+) 1.990 [1.354–2.924] 0.0005
Creatinine (+) 1.012 [1.006–1.018] 0.0001
Bilirubin (+) 1.016 [1.005–1.027] 0.0030
Uric acid (+) 1.002 [1.001–1.003] 0.0029
hs-CRP (+) 1.040 [1.019–1.061] 0.0001
Na (−) 1.157 [1.107–1.209] <0.0001 1.056 [1.007–1.109] 0.0244
GGTP (+) 1.002 [1.000–1.004] 0.0382
Cholesterol (+) 1.304 [1.087–1.059] <0.0001
PTX-3 (+) 1.268 [1.212–1.326] <0.0001 1.187 [1.126–1.251] <0.0001
NT-proBNP (a) 1.007 [1.004–1.010] <0.0001 1.004 [1.000–1.008] 0.0259
(+) per one unit increase; (−) per one unit decrease; a per 100 units increase. Abbreviations: see Table 1, HR, hazard ratio; CI, confidence intervals.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Freitas P. Aguiar C. Ferreira A. Tralhão A. Ventosa A. Mendes M. Comparative analysis of four scores to stratify patients with heart failure and reduced ejection fraction Am. J. Cardiol. 2017 120 443 449 10.1016/j.amjcard.2017.04.047 28629552
2. Westerdahl D.E. Kobashigawa J.A. Heart transplantation for advanced heart failure Card. Intensive Care 2019 504 524.e2 10.1016/B978-0-323-52993-8.00048-5
3. Aaronson K.D. Schwartz J.S. Chen T.-M. Wong K.-L. Goin J.E. Mancini D.M. Development and prospective validation of a clinical index to predict survival in ambulatory patients referred for cardiac transplant evaluation Circulation 1997 95 2660 2667 10.1161/01.CIR.95.12.2660 9193435
4. Levy W.C. Mozaffarian D. Linker D.T. Sutradhar S.C. Anker S.D. Cropp A.B. Anand I. Maggioni A. Burton P. Sullivan M.D. The seattle heart failure model: Prediction of survival in heart failure Circulation 2006 113 1424 1433 10.1161/CIRCULATIONAHA.105.584102 16534009
5. Pocock S.J. Ariti C.A. McMurray J.J.V. Maggioni A. Køber L. Squire I.B. Swedberg K. Dobson J. Poppe K.K. Whalley G.A. Predicting survival in heart failure: A risk score based on 39 372 patients from 30 studies Eur. Heart J. 2013 34 1404 1413 10.1093/eurheartj/ehs337 23095984
6. Di Tanna G.L. Wirtz H. Burrows K.L. Globe G. Evaluating risk prediction models for adults with heart failure: A systematic literature review PLoS ONE 2020 15 e0224135 10.1371/journal.pone.0224135 31940350
7. Szczurek W. Gąsior M. Skrzypek M. Szyguła-Jurkiewicz B. Apelin improves prognostic value of HFSS (Heart Failure Survival Score) and MAGGIC (Meta-Analysis Global Group in Chronic Heart Failure) scales in ambulatory patients with end-stage heart failure J. Clin. Med. 2020 9 2300 10.3390/jcm9072300
8. Nadar S.K. Shaikh M.M. Biomarkers in routine heart failure clinical care Card. Fail. Rev. 2019 5 50 56 10.15420/cfr.2018.27.2 30847246
9. Cabassi A. de Champlain J. Maggiore U. Parenti E. Coghi P. Vicini V. Tedeschi S. Cremaschi E. Binno S. Rocco R. Prealbumin improves death risk prediction of BNP-added Seattle Heart Failure Model: Results from a pilot study in elderly chronic heart failure patients Int. J. Cardiol. 2013 168 3334 3339 10.1016/j.ijcard.2013.04.039 23623341
10. Braunwald E. Biomarkers in heart failure N. Engl. J. Med. 2008 358 2148 2159 10.1056/NEJMra0800239 18480207
11. Matsubara J. Sugiyama S. Nozaki T. Sugamura K. Konishi M. Ohba K. Matsuzawa Y. Akiyama E. Yamamoto E. Sakamoto K. Pentraxin 3 is a new inflammatory marker correlated with left ventricular diastolic dysfunction and heart failure with normal ejection fraction J. Am. Coll. Cardiol. 2011 57 861 869 10.1016/j.jacc.2010.10.018 21310324
12. Suzuki S. Takeishi Y. Niizeki T. Koyama Y. Kitahara T. Sasaki T. Sagara M. Kubota I. Pentraxin 3, a new marker for vascular inflammation, predicts adverse clinical outcomes in patients with heart failure Am. Heart J. 2008 155 75 81 10.1016/j.ahj.2007.08.013 18082493
13. Norata G.D. Garlanda C. Catapano A.L. The long pentraxin PTX3: A modulator of the immunoinflammatory response in atherosclerosis and cardiovascular diseases Trends Cardiovasc. Med. 2010 20 35 40 10.1016/j.tcm.2010.03.005 20656213
14. Kaess B.M. Vasan R.S. Heart failure: Pentraxin 3—A marker of diastolic dysfunction and HF? Nat. Rev. Cardiol. 2011 8 246 248 10.1038/nrcardio.2011.50 21448138
15. Kotooka N. Inoue T. Aoki S. Anan M. Komoda H. Node K. Prognostic value of pentraxin 3 in patients with chronic heart failure Int. J. Cardiol. 2008 130 19 22 10.1016/j.ijcard.2007.07.168 18045709
16. Latini R. Gullestad L. Masson S. Nymo S.H. Ueland T. Cuccovillo I. Vårdal M. Bottazzi B. Mantovani A. Lucci D. Investigators of the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA) and GISSI-Heart Failure (GISSI-HF) trials. Pentraxin-3 in chronic heart failure: The CORONA and GISSI-HF trials Eur. J. Heart Fail. 2012 14 992 999 10.1093/eurjhf/hfs092 22740508
17. Pan J.-P. Liu T.-Y. Chiang S.-C. Lin Y.-K. Chou C.-Y. Chan W.-L. Lai S.-T. The value of plasma levels of tumor necrosis factor-alpha and interleukin-6 in predicting the severity and prognosis in patients with congestive heart failure J. Chin. Med Assoc. 2004 67 222 228 15357108
18. Zhang Y. Bauersachs J. Langer H.F. Immune mechanisms in heart failure Eur. J. Heart Fail. 2017 19 1379 1389 10.1002/ejhf.942 28891154
19. Sharma R. Coats A.J. Anker S.D. The role of inflammatory mediators in chronic heart failure: Cytokines, nitric oxide, and endothelin-1 Int. J. Cardiol. 2000 72 175 186 10.1016/S0167-5273(99)00186-2 10646959
20. Ristagno G. Fumagalli F. Bottazzi B. Mantovani A. Olivari D. Novelli D. Latini R. Pentraxin 3 in cardiovascular disease. front immunol Front. Immunol. 2019 10 823 10.3389/fimmu.2019.00823 31057548
21. Gullestad L. Ueland T. Vinge L.E. Finsen A. Yndestad A. Aukrust P. Inflammatory cytokines in heart failure: Mediators and markers Cardiology 2012 122 23 35 10.1159/000338166 22699305
22. Garlanda C. Bottazzi B. Magrini E. Inforzato A. Mantovani A. PTX3, a humoral pattern recognition molecule, in innate immunity, tissue repair, and cancer Physiol. Rev. 2018 98 623 639 10.1152/physrev.00016.2017 29412047
23. Nauta A.J. Bottazzi B. Mantovani A. Salvatori G. Kishore U. Schwaeble W.J. Gingras A.R. Tzima S. Vivanco F. Egido J. Biochemical and functional characterization of the interaction between pentraxin 3 and C1q Eur. J. Immunol. 2003 33 465 473 10.1002/immu.200310022 12645945
24. Carter A.M. Complement activation: An emerging player in the pathogenesis of cardiovascular disease Scientifica (Cairo) 2012 2012 402783 10.6064/2012/402783 24278688
25. Souza D.G. Amaral F.A. Fagundes C.T. Coelho F.M. Arantes R.M. Sousa L.P. Matzuk M.M. Garlanda C. Mantovani A. Dias A.A. The long pentraxin PTX3 is crucial for tissue inflammation after intestinal ischemia and reperfusion in mice Am. J. Pathol. 2009 174 1309 1318 10.2353/ajpath.2009.080240 19286566
26. Suzuki S. Shishido T. Funayama A. Netsu S. Ishino M. Kitahara T. Sasaki T. Katoh S. Otaki Y. Watanabe T. Long pentraxin PTX3 exacerbates pressure overload-induced left ventricular dysfunction PLoS ONE 2013 8 e53133 10.1371/journal.pone.0053133 23372656
27. Alba A.C. Agoritsas T. Jankowski M. Courvoisier D. Walter S.D. Guyatt G.H. Ross H.J. Risk prediction models for mortality in ambulatory patients with heart failure: A systematic review Circ. Heart Fail. 2013 6 881 889 10.1161/CIRCHEARTFAILURE.112.000043 23888045
28. Mehra M.R. Canter C.E. Hannan M.M. Semigran M.J. Uber P.A. Baran D.A. Danziger-Isakov L. Kirklin J.K. Kirk R. Kushwaha S.S. The 2016 International Society for Heart Lung Transplantation listing criteria for heart transplantation: A 10-year update J. Heart Lung Transplant. 2016 35 1 23 10.1016/j.healun.2015.10.023 26776864
29. Khanam S.S. Choi E. Son J.-W. Lee J.-W. Youn Y.J. Yoon J. Lee S.-H. Kim J.-Y. Ahn S.G. Ahn M.-S. Validation of the MAGGIC (Meta-Analysis Global Group in Chronic Heart Failure) heart failure risk score and the effect of adding natriuretic peptide for predicting mortality after discharge in hospitalized patients with heart failure PLoS ONE 2018 13 e0206380 10.1371/journal.pone.0206380 30485284
30. Goda A. Williams P. Mancini D. Lund L.H. Selecting patients for heart transplantation: Comparison of the heart failure survival score (HFSS) and the seattle heart failure model (SHFM) J. Heart Lung Transplant. 2011 30 1236 1243 10.1016/j.healun.2011.05.012 21764604
31. AbouEzzeddine O.F. French B. Mirzoyev S.A. Jaffe A.S. Levy W.C. Fang J.C. Sweitzer N.K. Cappola T.P. Redfield M.M. From statistical significance to clinical relevance: A simple algorithm to integrate brain natriuretic peptide and the Seattle Heart Failure Model for risk stratification in heart failure J. Heart Lung Transplant. 2016 35 714 721 10.1016/j.healun.2016.01.016 27021278
32. Chirinos J.A. Orlenko A. Zhao L. Basso M.D. Cvijic M.E. Li Z. Spires T.E. Yarde M. Wang Z. Seiffert D.A. Multiple plasma biomarkers for risk stratification in patients with heart failure and preserved ejection fraction J. Am. Coll. Cardiol. 2020 75 1281 1295 10.1016/j.jacc.2019.12.069 32192654
33. Lupón J. de Antonio M. Vila J. Peñafiel J. Galán A. Zamora E. Urrutia A. Bayes-Genis A. Development of a novel heart failure risk tool: The barcelona bio-heart failure risk calculator (BCN bio-HF calculator) PLoS ONE 2014 9 e85466 10.1371/journal.pone.0085466 24454874
34. Ali K. Workicho A. Gudina E.K. Hyponatremia in patients hospitalized with heart failure: A condition often overlooked in low-income settings Int. J. Gen. Med. 2016 9 267 273 10.2147/IJGM.S110872 27536157
35. Kenchaiah S. Pocock S.J. Wang D. Finn P.V. Zornoff L.A. Skali H. Pfeffer M.A. Yusuf S. Swedberg K. Michelson E.L. Body mass index and prognosis in patients with chronic heart failure: Insights from the Candesartan in Heart failure: Assessment of Reduction in Mortality and morbidity (CHARM) program Circulation 2007 116 627 636 10.1161/CIRCULATIONAHA.106.679779 17638930
36. MacGowan G.A. Neely D. Peaston R. Wrightson N. Parry G. Evaluation of NT-proBNP to predict outcomes in advanced heart failure Int. J. Clin. Pract. 2010 64 892 899 10.1111/j.1742-1241.2010.02388.x 20584222
37. Lavie C.J. Alpert M.A. Arena R. Mehra M.R. Milani R.V. Ventura H.O. Impact of obesity and the obesity paradox on prevalence and prognosis in heart failure JACC Heart Fail. 2013 1 93 102 10.1016/j.jchf.2013.01.006 24621833
38. Kalantar-Zadeh K. Block G. Horwich T. Fonarow G.C. Reverse epidemiology of conventional cardiovascular risk factors in patients with chronic heart failure J. Am. Coll. Cardiol. 2004 43 1439 1444 10.1016/j.jacc.2003.11.039 15093881
39. Torres D. Cuttitta F. Paterna S. Garofano A. Conti G. Pinto A. Parrinello G. Bed-side inferior vena cava diameter and mean arterial pressure predict long-term mortality in hospitalized patients with heart failure: 36 months of follow-up Eur. J. Intern. Med. 2016 28 80 84 10.1016/j.ejim.2015.11.029 26751720
40. Remme W.J. Overview of the relationship between ischemia and congestive heart failure Clin. Cardiol. 2000 23 4 8 10.1002/clc.4960230703 10680023
41. Frazier C.G. Alexander K.P. Newby L.K. Anderson S. Iverson E. Packer M. Cohn J. Goldstein S. Douglas P.S. Associations of gender and etiology with outcomes in heart failure with systolic dysfunction: A pooled analysis of 5 randomized control trials J. Am. Coll. Cardiol. 2007 49 1450 1458 10.1016/j.jacc.2006.11.041 17397674
|
PMC009xxxxxx/PMC9099901.txt |
==== Front
Animals (Basel)
Animals (Basel)
animals
Animals : an Open Access Journal from MDPI
2076-2615
MDPI
10.3390/ani12091135
animals-12-01135
Article
Pilot Study of Attitudes of Taiwanese Veterinarians and Undergraduate Veterinary Students toward Animal Abuse and Interpersonal Violence
Chen Yi-Hsuan 12
https://orcid.org/0000-0002-0098-0278
Huang Wei-Hsiang 123*
Farnworth Mark J. Academic Editor
1 Department of Veterinary Medicine, School of Veterinary Medicine, National Taiwan University, Taipei City 10617, Taiwan; [email protected]
2 National Taiwan University Veterinary Hospital, College of Bioresources and Agriculture, National Taiwan University, Taipei City 10672, Taiwan
3 Graduate Institute of Molecular and Comparative Pathobiology, School of Veterinary Medicine, National Taiwan University, Taipei City 10617, Taiwan
* Correspondence: [email protected]; Tel.: +886-2-33663760
28 4 2022
5 2022
12 9 113517 3 2022
26 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Simple Summary
There is a strong association between animal abuse and interpersonal violence; therefore, veterinarians may encounter both. Dealing with animal abuse cases is beneficial for advancing animal welfare and the overall public health. Veterinarians play an important professional role in identifying and responding to this relationship. This is the first study in Taiwan to investigate the current situation of animal abuse from the perspective of veterinarians. Our results established that the incidence of suspected physical animal abuse encountered by veterinarians in Taiwan was 0.16 cases per 100 patients, and 63.9% of our respondents had seen suspected of animal physical abuse in the past five years. Approximately 10% of animal abuse cases were likely concurrent with human abuse. Most respondents (about 80%) agreed that animal abuse and human abuse frequently co-occur. However, it did not affect their willingness to assist or report abuse. In total, 88.7% of respondents supported mandatory reporting of animal abuse. The results of this study underscore the urgent advancement of education and the importance of the crucial veterinary role in both animal and interpersonal abuse cases.
Abstract
There is a strong association between animal abuse and interpersonal violence; therefore, veterinarians may encounter both. Dealing with animal abuse cases is beneficial for advancing animal welfare and the overall public health. Veterinarians play an important role in identifying and responding to this relationship. This study estimated the incidence of animal abuse encountered by veterinarians, examined veterinarians’ awareness of the relationship between animal abuse and human abuse, examined veterinarians’ attitudes towards how they deal with abuse cases, and related demographic characteristics to their attitudes of intervention and the frequency of encountering abuse cases. An anonymous self-administered questionnaire was designed and distributed through social media. Our results show that respondents’ motivation to interfere for animal abuse cases was positively related to their moral or legal responsibility, willingness to assist, and agreement of mandatory reporting. Our results indicated that respondents who believed they had been provided with adequate training were more willing to deal with animal abuse, more capable of distinguishing abuse cases, and did not believe that dealing with abuse cases was beyond their ability. However, more than 60% of our respondents self-evaluated that the animal cruelty awareness training courses were insufficient. Hence, in addition to the traditional role of veterinarians, identifying and responding to animal cruelty should be enhanced through education.
animal abuse
animal cruelty
interpersonal violence
veterinarians
education
animal welfare
forensic science
This research received no external funding.
==== Body
pmc1. Introduction
There is a strong association between animal abuse and interpersonal violence; therefore, veterinarians may encounter both animal abuse and interpersonal violence clinically [1]. Traditionally, veterinarians care for animal health. However, animal abuse and interpersonal violence will affect each other and sometimes cannot be separated. Therefore, dealing with animal abuse cases by veterinarians positively contributes to the overall public health. Veterinarians play a crucial role in identifying and responding to this link. There is still a lack of research on the association between animal abuse and interpersonal violence from a veterinary perspective in Taiwan. The current situation, i.e., the incidence of animal abuse cases faced by clinical veterinarians in Taiwan, is unknown. Other information, such as the veterinarians’ attitudes towards abuse cases, the difficulties they may encounter, the extent of background knowledge (the knowledge of animal abuse and the understanding of legality), and awareness of the correlation between animal abuse and interpersonal violence, is lacking. Therefore, it is necessary to collect and analyze these data.
Animal abuse is an unkind behavior towards animals and an intentional act that harms animals [2]. In Taiwan, the Animal Protection Act (since 1998, last amended in 2021) states that animal abuse refers to “harming an animal or making it unable to function properly with violence, drug diversion, physical objects, acts of omission, or other means, beyond what is necessary to rear, tend or dispose of an animal” [3]. There are various other definitions of animal abuse; one of the most cited is “a socially unacceptable behavior that intentionally causes useless pain, suffering or anguish to the animal, and/or its death” [4]. Animal abuse can be categorized based on the type of abuse and intent to abuse. Companion animal abuse is classified into physical and mental types, and both include intentional and non-intentional maltreatment [5].
Animal abuse is a part of the spectrum of family and community violence and should be viewed as a leading public health problem worldwide [6]. Extensive research has linked animal abuse with interpersonal violence and even used it as an indicator and/or predictor of ongoing crimes of interpersonal violence and public health problems [7]. The earliest article in which this concept was proposed could be traced to McDonald’s triad of behaviors in children that appeared to be predictive of violence: enuresis, fire setting, and cruelty to animals [8]. In 1987, the American Psychiatric Association first included “physical aggression to people and animals” in the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) as one of the earliest and most severe symptoms of a Conduct Disorder.
A review of human and animal service professionals stressed strong associations between domestic violence, child abuse, and animal abuse, proposing that it is the responsibility of both animal service and human service professionals to be aware of such occurrences, understand the significance, and promote appropriate professional and policy responses [9]. As empirical research demonstrates a close relationship between animal abuse and crimes against humans, the legal consequences of animal abuse have been strengthened. By 2014, all 50 states in the United States regarded animal cruelty as a felony. In 2016, the US Federal Bureau of Investigation began tracking animal cruelty crimes using the National Incident-Based Reporting System (NIBRS) [10].
Veterinarians play an important role in early identifying animal abuse and even interpersonal violence. However, most veterinarians are not trained to intervene in cases of animal abuse and interpersonal violence. Therefore, veterinary medicine schools should increase the amount of time devoted to animal cruelty issues in their curricula and offer online and on-site postgraduate and continuing education to build confidence in reporting animal cruelty [9].
In addition to education, policy protection should be provided for assisting veterinarians in solving ethical dilemmas. When veterinarians encounter such animal abuse cases, it is difficult for them to balance economic, safety, confidentiality, legal, and management concerns with ethical principles, personal beliefs, and professional standards [11]. While veterinarians remain divided on whether to report suspected abuse, Randour et al. suggested that all states in the United States should mandate that veterinarians report animal cruelty and offer immunity for good faith reporting [9]. The American Veterinary Medical Association considers it the responsibility of the veterinarians to report acts of animal cruelty and educate clients regarding humane care and treatment of animals. Currently, there is no mandatory reporting expected of veterinarians about animal abuse in the United Kingdom. In the United States, 20 states place a mandatory duty upon state-licensed veterinarians to report suspected animal cruelty; other states do not require veterinarians to report but rather allow veterinary professionals to act at their discretion. About 14 states have no laws that allow or require reporting [12]. In Taiwan’s Animal Protection Act or Veterinarian Act, there is no requirement to report any suspected or confirmed cases of animal abuse by veterinarians.
Considering the veterinarian’s awareness of the link between animal abuse and interpersonal violence, according to Monsalve et al., 2017, only 4.2% of the 96 published articles discussing this link were from Asia, i.e., China, India, Japan, and Malaysia [13]. In total, 79.2% of the studies were concentrated in North America, suggesting that this “link” remains unknown in various countries. In addition, only 7.3% of the articles were published in the field of veterinary medicine. Since Taiwan’s Animal Protection Act has been amended several times in recent years, we wish to further enhance the awareness of animal protection in veterinary medicine. In Taiwan, there are currently no published data regarding the incidence of veterinarians identifying human or animal abuse. Therefore, we designed a questionnaire, which is a modified version of the survey in New Zealand [14], aimed at investigating animal abuse cases encountered by veterinarians and the attitudes of veterinarians on issues of suspected animal abuse, interpersonal violence, and the link between these factors in Taiwan.
This study aims to (1) estimate the incidence of suspected animal abuse encountered by veterinarians, (2) study veterinarians’ awareness of the relationship between animal abuse and human abuse, (3) explore the attitudes of veterinarians towards dealing with abuse cases, and (4) relate demographic characteristics to their attitudes of intervention and their frequency of encountering abuse cases.
2. Materials and Methods
2.1. The Questionnaire and Data Collection
The questionnaire applied in this research was a modified version of the survey in New Zealand [14], which was based on similar research in Australia [15]. We revised this questionnaire to accommodate Taiwanese culture and language.
Before data collection, this questionnaire was first reviewed by 15 selected veterinarians to rectify inappropriate sentences and avoid meaningless answers.
An anonymous Google form link of our questionnaire was then created and distributed through the School of Veterinary Medicine, National Taiwan University’s Facebook page, and the Taiwanese Veterinary Medical Association’s website.
The questionnaire consisted of five sections (see Appendix A).
Section 1: Demographics. Demographic information was collected, including gender, age group, year of graduation, type of clinical practice, and number of monthly visiting clinical cases. Multiple choices were allowed to be selected for “type of clinical practice.”
Section 2: Attitudes. This section focused on the attitudes of veterinarians in facing abuse cases. In this section, the respondents would answer “how they would deal with abuse cases”, “what is their perception of the relationship between the perpetrator and other interpersonal violence”, and “whether veterinary education provides relevant knowledge.” The respondents were asked to rate their answers of 10 questions using the 6-point Likert scale from 1 to 6, with 1 = strongly disagree, 2 = disagree, 3 = slightly disagree, 4 = slightly agree, 5 = agree, and 6 = strongly agree.
Section 3: Managing Abuse. In this section, the questions included “if they believed they could identify abuse or not”, “whether they would choose to report and/or to assist when they encounter suspicious animal and/or human abuse cases”, and “What are the consequences they worry about when dealing with abuse cases?” In addition, opinions regarding “it should be mandatory to report animal abuse and human abuse” were asked.
Sections 4 and 5: Physical and mental animal abuse. These two sections focused on physical and mental animal abuse, respectively, and were limited to veterinarians with clinical experience. In the beginning, the respondents were asked again, “if they believe they have the knowledge to identify suspected animal abuse”. Other questions included “the frequency of suspicious animal abuse cases they had encountered within five years”, “if there was any interpersonal violence involved”, “who brought the animal to the clinic?”, “why it was suspected to be caused by abuse?”, and “what was the species of animal?”. Multiple choices were allowed in “what was the species of the animal?”.
Data were collected from 15 October 2020 to 30 November 2020. A total of 247 responses were obtained during this period. This questionnaire was conducted with absolute anonymity, and it was ensured that all the data including personal identifiable information collected and analyzed in the study complied with the country’s privacy and security laws. All responses were collected with the consent of respondents.
2.2. Statistical Methods
Descriptive statistics and frequency distributions were computed using SPSS (SPSS Statistics, IBM Corp., Somers, NY, USA). Likert-scale questions with ranking data were analyzed using Spearman’s rho between questions to interpret their correlation. A correlation coefficient between 0 and 0.3 is defined as low, a correlation coefficient between 0.3 and 0.7 is moderate, and a correlation coefficient between 0.7 and 1 is high. One-way analysis of variance (ANOVA) was used to compare the mean ratings of the Likert-scale questions with categorical variables. Chi-square tests were used to compare categorical variables.
3. Results
3.1. Section 1: Demographics
Details of the demographic and practice characteristics of the respondents are summarized in Table 1.
A total of 247 respondents completed this questionnaire, of which 53.8% (n = 133) were practicing veterinarians. The definition of practicing veterinarians excludes those who will graduate after 2021 and those who chose “none” in the number of visiting cases per month in this study. Sections 4 and 5 of the questionnaire were only filled by practicing veterinarians. Among these practicing veterinarians, most were female (59.4%, n = 79), while a minority were male (40.6%, n = 54). In 2020, there were 5486 veterinarian practice licenses in Taiwan [16]. The practicing veterinarians in this questionnaire are estimated to account for 2.42% of veterinarians in Taiwan.
The largest proportion of respondents (65.2%, n = 161) comprised those aged 18–29 years. Veterinary student respondents, whose graduation year was after 2021, accounted for 40.5% (n = 100). Among veterinarian respondents, the largest proportion had graduated between 2016 and 2020 (24.3%, n = 60).
In terms of clinical practice type, most veterinarians worked with dogs and cats (74.1%, n = 183). The second largest group of veterinarians worked with small animals, such as rabbits and guinea pigs (21.5%, n = 53). Other optional responses filled in by respondents included veterinary pathologist (n = 3), laboratory animal veterinarian (n = 2), horse veterinarian (n = 1), and animal protection officer (n = 1).
3.2. Section 2: Attitudes
3.2.1. Comparison among Items
The attitudes of veterinarians when facing abuse cases are shown in Table 2. Items 1 to 3 were questions about attitudes related to interventions for animal abuse, and there was a noticeable trend that the higher the score, the greater the proportion among the three items. For item 1, 42.3% strongly agreed that the veterinarian has a moral or legal responsibility to intervene in suspected animal abuse, and the mean score was 5.05. For item 3, 45.6% strongly agreed with, “In practice, when presented with an animal abuse case, I would assist in preventing it from happening again,” and the mean score was 5.17. However, in item 2 “In practice, when presented with a suspected animal abuse case, I know my legal rights and responsibilities”, although most veterinarians were aware of their legal rights and responsibilities, only 28.6% of them strongly agreed this; scores in item 2 appeared to be more evenly distributed than item 1 and item 3, with a mean score of 4.28.
We analyzed the results of these three items using Spearman’s rho to determine whether a correlation between the respondents’ choices existed. The analysis showed that they were all moderately correlated with each other. The correlation coefficient between items 1 and 2 was 0.507 (p < 0.01). Likewise, the correlation coefficient between items 1 and 3 was 0.585 (p < 0.01), and the correlation coefficient between those who agreed in items 2 and 3 was 0.455 (p < 0.01).
Item 7 evaluated the attitude of veterinarians regarding the relationship between “training animals using punitive methods” and “animal cruelty.” The highest rate (28.2%) of responses was for point 4 (slightly agree) on the scale, with a mean score of 4.14.
Items 4 to 6 were questions about whether the perpetrators of animal abuse were more likely to be involved in other crimes. Surprisingly, the answers to these three questions were highly consistent: more than 40% of the respondents strongly agreed that people who abuse animals are more likely to commit other types of crimes, including interpersonal violence. The highest percentage (47.6%) of respondents who strongly agreed was in item 4 regarding child abuse, with a mean score of 4.92. The correlation coefficient between items 4 and 5 was very high (0.876, p < 0.01).
Items 9 and 10 were questions about attitude toward intervention in interpersonal domestic violence. Item 9 explored, “when domestic violence is suspected, the veterinarian has a moral responsibility to intervene.” Although the mean score was 3.85, the majority of respondents (27.4%) selected point 4 (slightly agree). In addition, the score was significantly lower than that for intervention for animal abuse (item 1, mean score = 5.05). Item 10 explored, “when presented with child or spousal abuse, I feel I should provide assistance to the client.” Compared to item 9, the mean score of item 10 was higher (mean score = 4.23), and the largest proportion of respondents (25%) selected point 5 (agree).
The correlation coefficient between item 10 (“in practice, when presented with child or spousal abuse, I feel I should provide assistance to the client”) and item 3 (“in practice, when presented with an animal abuse case, I would assist in preventing it from happening again”) was 0.420 (p < 0.01), indicating a moderate correlation. The mean score for item 10 was lower than that for item 3.
Furthermore, the correlation coefficient between item 10 (“in practice, when presented with child or spousal abuse, I feel I should provide assistance to the client”) and item 4 (“people who abuse animals are more likely to abuse their children”) was 0.294 (p < 0.01). The correlation coefficient between item 10 (“In practice, when presented with child or spousal abuse, I feel I should provide assistance to the client”) and item 5 (“People who abuse animals are more likely to abuse their spouse”) was 0.303 (p < 0.01). These results revealed low correlations.
Finally, item 8 explored, “during veterinary training, I was provided with adequate information and training to identify and prevent animal abuse”. The mean score was 3.07, and the majority of respondents (26.2%) selected point 2 (disagree). The correlation coefficient between item 8 (“during veterinary training, I was provided with adequate information and training to identify and prevent animal abuse”) and item 2 (“In practice, when presented with a suspected animal abuse case, I know my legal rights and responsibilities”) was 0.412 (p < 0.01), indicating a moderate correlation.
3.2.2. Differences among Demographic Categories
Table 3 shows a comparison of mean scores of selected items (items 3, 4, 5, 8, and 10) among demographic categories, including gender, age, and year of graduation.
The comparison of scores for items 4 and 5 (the veterinarian’s attitude toward the recognition that animal abusers are more likely to abuse a child or spouse) showed only a borderline significant difference (p = 0.052) for age group in item 5. There were no significant differences for gender, age, or year of graduation.
The comparison of scores for items 3 and 10 showed only a borderline significant difference (p = 0.064) for age group in item 3. There were no significant differences for gender, age, or year of graduation. In terms of whether veterinarians will choose to intervene for animal abuse cases (item 3), the 18–29 age group was the most willing to assist, while the 40–49 age group was the least willing to assist.
The comparison of scores for item 8 showed significant differences among graduation year groups (p < 0.01). Veterinary students who will graduate after 2021 agreed more with “adequate information and training”. By contrast, veterinarians who had graduated in 2011–2015 agreed less with this statement.
We further analyzed the opinions of veterinary students expected to graduate after 2021 and veterinarians who had graduated before 2020 on these issues and found significant differences, mostly relating to items 3 and 8 (p < 0.01 for both). It was verified again that students who had not yet graduated agreed more with “adequate information and training”, and the responses further indicated that they were more willing to provide assistance for animal abuse.
3.3. Section 3: Managing Abuse
3.3.1. Willingness to Report Abuse Cases
In this section, respondents were asked to indicate their willingness to report suspected animal and human abuse. Of the 247 respondents, 122 (49.4%) respondents chose to report all cases, 114 (46.2%) chose to report only severe cases, and 11 (4.5%) chose not to report when they encountered suspected animal abuse cases. In suspected human abuse, 115 respondents (46.6%) chose to report all cases, 107 (43.3%) chose to report only severe cases, and 25 (10.1%) chose not to report.
Table 4 shows the mean scores of selected items in Section 2 of the questionnaire in Appendix A, among groups for the willingness to report suspected animal and human abuse.
For item 1, concerning the attitude toward the veterinarian’s responsibility to intervene in suspected animal abuse, respondents who chose not to report animal abuse had a lower mean score (4.00 ± 0.77 for animal abuse and 4.44 ± 1.04 for human abuse). By contrast, respondents who chose to report all cases had a higher mean score (5.41 ± 0.81 for animal abuse and 5.28 ± 1.01 for human abuse). The p values among these were all <0.01, indicating that the more likely a veterinarian was to report abuse cases, the higher the moral or legal responsibility they felt toward animals.
With respect to item 9, concerning whether to intervene in suspected interpersonal domestic violence, the mean scores of those who chose not to report were very low (2.45 ± 1.29 for animal abuse and 2.44 ± 1.16 for human abuse). Even the respondents who chose to report all cases had relatively lower mean scores (4.20 ± 1.41 for animal abuse and 4.40 ± 1.36 for human abuse) compared with item 1. There was a similar trend for item 1, and the p values among these were all < 0.01.
Similar trends were observed for items 3 and 10. Irrespective of animal abuse or child/spousal abuse, the more likely the respondents were to report, the more they agreed to assist in abuse cases.
For items 4 and 5, all mean scores were greater than 4; however, there was no significant difference between whether they chose to report abuse or not.
3.3.2. Concerns and Obligations Associated with Reporting Abuse Cases
Respondents were then asked to indicate how they would respond to potential abuse cases in the clinic. The results are shown in Figure 1. Cross-analyses were performed based on the results. Table 5 shows the mean score of item 8 of Section 2 of the questionnaire in Appendix A, i.e., “during veterinary training, I was provided with adequate information and training to identify and prevent animal abuse”, among the groups concerning the reporting of suspected animal and human abuse. Table 6 shows the cross table between each concern and obligation (i.e., whether it should be mandatory to report abuse cases). Table 7 shows a cross table between obligation and demographic categories.
When asked, “I can distinguish animal abuse features”, 81.8% of the respondents agreed (Figure 1). Interestingly, in item 8 of Section 2, the majority of respondents stated that they had not acquired sufficient knowledge in veterinary training. The mean score of item 8 for respondents who answered that they could distinguish abuse features (3.17 ± 1.62 for animal abuse and 3.27 ± 1.62 for human abuse) was significantly higher than that of those who answered that they could not (2.62 ± 1.50 for animal abuse and 2.61 ± 1.49 for human abuse) (Table 5).
Approximately 45.7% of the respondents stated that dealing with animal abuse cases was beyond their profession or competence, and 76.1% of the respondents stated that they were unable to deal with human abuse. The results of the cross-analysis with item 8 also indicated that the more the respondents believed they had been provided with adequate training, the less they agreed that dealing with abuse cases was beyond their ability (Table 5).
Most respondents (78.9% for animal abuse and 80.2% for human abuse) were worried that dealing with abuse cases would irritate clients (Table 6). However, this issue was not significantly associated with item 8 of Section 2, “during veterinary training, I was provided with adequate information and training to identify and prevent animal abuse”, and only a borderline significant association was observed when dealing with animal abuse (p = 0.077) (Table 5). Moreover, there was no statistically significant difference between this issue and their choice of mandatory reporting of abuse cases (Table 6).
Moreover, most respondents (91.1% for animal abuse and 87.9% for human abuse) stated that reporting abuse cases was a civic responsibility (Table 6).
As shown in Table 6, those who stated that reporting animal abuse cases was their civic responsibility were significantly associated with those who stated that it should be mandatory to report animal abuse. Among those who agreed that reporting animal abuse should be mandatory, 94.5% stated that it was their civic responsibility to report animal abuse cases; even among those who did not agree to report animal abuse, 64.3% thought it was their civic responsibility to report abuse.
The comparison between the mandatory reporting of human abuse and whether the reporting of human abuse and animal abuse is a civic responsibility both showed significant associations (p < 0.01 and p = 0.029, respectively), as listed in Table 6. Among the respondents who agreed that reporting human abuse should be mandatory, the proportion of those who stated that reporting animal abuse is a civic responsibility (93.9%) and the proportion of those who stated that reporting human abuse is a civic responsibility (94.5%) was similar.
Respondents who stated that it should be mandatory to report animal abuse and that reporting animal abuse is a civic responsibility accounted for 83.8% of all respondents (207 out of 247), while respondents who stated that it should be mandatory to report human abuse and that reporting human abuse is a civic responsibility accounted for 66.4% of all respondents (164 out of 247).
3.3.3. Willingness and Obligation Associated with Reporting Abuse Cases
With respect to the association between the willingness and the obligation to report animal abuse, among those who agreed with the mandatory reporting of animal abuse, there was a significantly higher percentage who would report animal abuse cases (p < 0.01) (Table 6). Among those who stated that it should not be mandatory to report animal abuse cases, the largest proportion chose to report only severe animal abuse cases (64.3%), and a few respondents chose not to report at all (14.3%). Although there was no significant association between whether it should be mandatory to report animal abuse cases and the willingness to report human abuse cases, the trend was similar.
In addition, among those who agreed that it should be mandatory to report human abuse, there was a significantly higher percentage who would report both animal abuse and human abuse (p = 0.032 and p < 0.01, respectively) (Table 6). Among those who stated that it should be mandatory to report human abuse cases, the largest proportion chose to report all abuse cases (54.9% for animal abuse and 56.7% for human abuse). Among those who stated that it should not be mandatory to report human abuse cases, the largest proportion chose to report only severe abuse cases (40.2% for animal abuse and 37.2% for human abuse).
3.3.4. Demographic Characteristics and Respondents’ Management of Abuse Cases
With respect to the correlation between demographic characteristics and respondents’ attitudes toward mandatory reporting of abuse cases, there were no significant differences for gender, age, and year of graduation (Table 7).
In addition, we compared the differences in the thinking of students (after 2021) and veterinarians (before 2021) in this section. The results indicated that there were no significant differences between the two graduation status groups for these questions, except for the willingness to report both animal and human abuse cases (Table 8).
3.4. Sections 4 and 5: Physical and Mental Animal Abuse
Sections 4 and 5 were filled in by practicing veterinarians with clinical experience. In the raw data, some respondents claimed they did not have the knowledge to identify abuse cases, and therefore they had encountered no abuse cases clinically. Because this might mean that these respondents were unfamiliar with the definition of physical and mental animal abuse, we excluded these respondents. Comparisons of the raw and recalculated data are listed in Table 9.
Of the 119 veterinarians who claimed they had the knowledge to identify physical abuse, or who had encountered abuse cases, only 36.1% (43 out of 119) had not seen suspected physical abuse in the past five years. We obtained similar results (41 out of 113, 36.3%) for those who said they had encountered animal mental abuse cases in the past five years. Among these respondents, 110 believed they had the knowledge to identify both physical and mental abuse.
For these 119 veterinarians, we analyzed the correlation between their belief about whether they had the knowledge to identify physical abuse and the physical animal abuse cases they had suspected in the past five years. The results showed a significant correlation (p = 0.013). Conversely, in the analysis of 113 veterinarians who believed they had the knowledge to identify mental abuse, the result was insignificant (p = 0.429).
Our results showed male veterinarians were more likely to encounter physical abuse cases than females (p = 0.016). There were no significant differences between encountering physical animal abuse and other examined demographic categories, including age and year of graduation. If we combined these data with the average number of visiting cases per month, as asked in Section 1, the incidence of physical animal abuse encountered by the veterinarians was 0.19 cases per 100 patients.
Details of the abuse cases are listed in Table 10. With respect to, “who brought the abused animal to the veterinarian”, the most likely person was the owner, followed by other people. Dogs were the most common species being abused, followed by cats. Approximately 10% of the animal abuse cases co-occurred with suspected human abuse.
In the two sections, the reasons for suspecting animal abuse were also explored. The results for physical abuse are summarized in Table 11, while the results for mental abuse are summarized in Table 12.
The responses regarding the injured body parts and types of injury are presented in Table 13. Among these cases, the most commonly injured body parts were limbs and/or head, and the most common injury types were emaciation, malnutrition, or poor fur condition caused by varying degrees of neglect. Three cases in which the animals were evidently dead or had been eaten by other animals are noteworthy. There was also a mention by a carcass inspection veterinarian of inhumane slaughter of economic animals that were improperly stunned and exsanguinated.
We further analyzed whether physical abuse cases had been encountered in the past five years and veterinarians’ willingness to report them. The results are shown in Table 14, which shows that there was a significant difference among groups (p = 0.013). We found that veterinarians who chose to report all abuse cases had encountered fewer suspected cases of physical abuse in the past five years; by contrast, veterinarians who only reported severe cases had encountered many more physical abuse cases.
4. Discussion
4.1. Animal Abuse Cases Encountered by Veterinarians
In this study, the incidence of suspected physical animal abuse encountered by veterinarians in Taiwan was 0.16 cases per 100 patients. In the past five years, 63.9% of our respondents had seen suspected physical abuse. In the United States, the incidence of suspected animal abuse cases was reported as 0.56 cases per 100 patients [1]. In Australia, the incidence was 0.12 cases per 100 patients, and 92% of respondents had seen suspected abuse [15]. In New Zealand, 63% of respondents had seen suspected abuse [14], while 87% of respondents had seen suspected abuse in the United States [10]. In South Korea, 86.5% of respondents had seen suspected abuse [17]. Therefore, compared with those in other countries, Taiwanese veterinarians showed a very low probability of encountering suspected animal abuse cases. However, different methods of recording the frequency of animal abuse and different definitions of animal abuse mean that making meaningful comparisons between studies is difficult.
In addition, education level and personal cognition, resulting in inconsistent individual judgment standards, may cause data deviation. Therefore, the incidence of animal abuse may not reflect the actual situation in Taiwan. A previous study, which specifically underscored the need for education and training in a group of board-certified veterinary pathologists, pointed out the current situation of lacking associated knowledge for handling veterinary forensic cases [18]. In this study, respondents who believed they knew how to distinguish animal abuse features were correlated with the encounter of animal abuse. The fact that veterinarians in Taiwan generally evaluated veterinary forensic training courses as insufficient cannot be overlooked. One reason for veterinarians believing that they knew how to identify animal abuse might be that they had learned through experience; however, they might not have had a systematic understanding of the topic. Although empirical learning is indeed important, the lack of curricula in Taiwan’s veterinary training on how to identify animal abuse may be responsible for the low incidence of animal abuse.
In addition, it is worth noting that approximately 10% of the animal abuse cases co-occurred with suspected human abuse. Co-occurrence rates in other studies were 23.7% in Australia [15] and 56% in New Zealand [14]. Due to different definitions of abuse, the reported co-occurrence rate was 9 to 88%, according to one study in the United States [19]. Veterinarians are not professionally trained to detect human abuse; therefore, the co-occurrence rate may be underestimated.
In addition, we found that the frequency of encountering physical animal abuse was greater for male veterinarians than for females. There was no difference for age or year of graduation. In other studies, females [15,17] and younger [1,15,17] veterinarians had a higher frequency of encountering animal abuse. However, this trend was not observed in our results.
Similar across all surveys, dogs were the most commonly abused species encountered by veterinarians, followed by cats. This could be explained by the fact that as companion animals, humans have easy access to them. In addition, because they are relatively small, humans can easily control them. However, this result was also affected by the composition of veterinary practice types of the respondents.
In line with the results of other studies, the most common injury types in this study were emaciation, malnutrition, or poor fur condition caused by varying degrees of neglect (n = 49, 41.2%). The second most common injury type was bruises or hemorrhage (n = 41, 34.5%), followed by abrasions or lacerations. This result is consistent with those of previous studies [15]. Among the responses we received, one summarized the current situation in Taiwan for us: “The abuses often seen in animal hospitals are negligence: starvation, extreme dehydration, long-term exposure to the sun without shelter, dirty and covered in maggots, space constraints, performing obsessive-compulsive behavior on animals. Because the owner did not think these were abusive acts, the animals would be sent to the veterinarian only if there was an imminent danger. Too many cases were attributed to a long-term lack of quality of life. Most cases of obvious acts of violence would not be sent to veterinarians by the owner. Only when violent behavior causes death may animals be sent for forensic examination.” In Section 3, this respondent stated that he/she would report only severe cases when encountering animal abuse cases. This result reveals that veterinarians exposed to a large number of animal abuse cases might not report all of them. Most practicing veterinarians exposed to many cases of animal abuse chose to report only the severe cases. We also found that compared to veterinarians who had graduated, students who had yet to graduate were more likely to report all abuse cases. These changes in thinking may be related to the veterinarian’s practice environment, the impact of education, policy support, and/or practical considerations.
4.2. Attitudes toward Animal and Interpersonal Abuse
In this study, veterinarians generally agreed that animal abuse should be prevented and differed only with respect to the degree of personal willingness. Veterinarians’ motivation to interfere in animal abuse cases was positively related to their moral or legal responsibility, willingness to assist, and agreement with mandatory reporting. However, compared to animal abuse, veterinarians were significantly less willing to interfere in interpersonal violence. In addition, dealing with human abuse was considered to be beyond their ability and was not considered as much a civic responsibility as the former. We also found that approximately 80% of the respondents agreed that animal abuse and domestic abuse frequently co-occur, but this did not affect their willingness to report it. This is probably related to the traditional role of veterinarians in serving to promote animal health.
Veterinarians play a crucial role in animal welfare and public health. By dealing with animal abuse cases, veterinarians contribute to people’s overall health. However, a significant barrier faced by veterinarians is the fact that current legal and police systems are unable to ensure the safety and welfare of victims [17]. Other factors that are often discussed include insufficient training, damage to veterinarian–client relationships, loss of economic support, and a possible breach of personal safety.
Unexpectedly, 88.7% of our respondents thought it should be mandatory for veterinarians to report animal abuse cases, but only 66.5% felt the same about human abuse cases. Furthermore, those who were more willing to report animal abuse cases were more likely to agree that it should be mandatory to report animal abuse cases. A recent study [10] in the United States on whether reporting should be mandatory showed that respondent veterinarians strongly supported (24.0%) or supported (42.1%) these laws, with 19.5% having no opinion, 11.2% opposing, and 3.2% being strongly opposed to such laws. These data imply that veterinarians in Taiwan have a considerably high level of awareness regarding animal protection. We hope that these data lead to a revision of related laws. Minimal protection for veterinarians and animals could be initiated by modifying the law that immunizes veterinarians from civil and criminal liability. We also suggest putting in place an animal abuse tracking system to actively intervene and prevent animal abuse cases.
Our results also suggest that education may influence veterinarians’ attitudes when they are faced with abuse cases. Respondents who considered they had been provided with adequate training were more willing to deal with animal abuse and more capable of distinguishing abuse cases and did not feel that dealing with abuse cases was beyond their ability. These results were similar to those of a previous study [20], which established that specialized training equips veterinarians with the skills and confidence to report abuse. However, a large proportion of our respondents believed that the training provided was insufficient. Enhanced education and training could increase veterinarians’ confidence in communicating with clients and improve their ability to identify abuse cases and understand the notification process and other assistive resources that can be applied when encountering abuse cases.
4.3. Limitations
There were some limitations to this study. First, the process we used to collect data from the questionnaires made it impossible to guarantee that there were no repeat respondents or respondents who were not veterinarians. Second, the respondents’ definition of physical or mental abuse could not be assessed. Therefore, the differences in defining animal abuse could have led to response bias. Third, the low response rate and the fact that 40% of respondents are undergraduate veterinary students could influence the results. Last, the voluntary nature of the survey and the low possibility of the abused animal being sent to veterinary hospitals might cause statistical bias.
5. Conclusions
This is the first study in Taiwan to examine the current situation of animal abuse from the perspective of veterinarians and also the first study to collect the attitudes of veterinarians when dealing with abuse cases. Our results revealed that the incidence of suspected physical animal abuse encountered by veterinarians in Taiwan was 0.16 cases per 100 patients. Approximately 10% of animal abuse cases co-occur with suspected human abuse. This study and other research indicated that animal abuse and domestic abuse co-occur frequently. Hence, the traditional role of veterinarians should be enhanced through education for identifying and responding to animal cruelty. Increasing the sensitivity of veterinarians to animal abuse will enable early detection and tracking of animal abuse and further reduce the incidence of domestic abuse, thereby promoting animal welfare and public health. The fact that 88.7% of our respondents thought that veterinarians should mandatorily report animal abuse cases must not be ignored. However, mandatory reporting may cause pressure on veterinarians. Therefore, it is hoped and recommended that the relevant authorities modify the law to provide immunity to veterinarians from civil and criminal liability when reporting animal abuse cases.
Acknowledgments
The authors thank Williams et al. for providing their questionnaire [14] and Ya-Wen Yang for contributions to the design of our questionnaire.
Author Contributions
Conceptualization, Y.-H.C. and W.-H.H.; methodology, data collection, curation and analyses, Y.-H.C.; writing—original draft preparation, Y.-H.C.; writing—review and editing; W.-H.H., supervision. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
The study was conducted according to the guidelines of the Declaration of Helsinki. This questionnaire was conducted with absolute anonymity. It was ensured that all the data including personal identifiable information collected and analyzed in the study complied with the country’s privacy and security laws. In Taiwan, according to the “Human Subjects Research Act”, the definition of Human subject research refers to “research involving obtaining, investigating, analyzing, or using human specimens or an individual person’s biological behavior, physiological, psychological, genetic or medical information.” Only the human subject research mandatorily applies for approval, and the current questionnaire was not involved in any biological behavior, physiological, psychological, genetic, or medical information of the human subject. Therefore, according to our Regulation, the application for Ethical review and approval was not mandatory for the anonymous questionnaire regarding only veterinarians’ attitudes. Ethical review and approval were not applied. This questionnaire was initially a class assignment for the purpose of a class presentation, rather than for the purpose of research. The application for approval from the National Taiwan University Behavioral and Social Sciences Research Ethics Committee was therefore not considered initially, and there were no sufficient time and funding to apply for the approval.
Informed Consent Statement
The informed consent statement was added at the beginning of the anonymous questionnaire. All respondents were informed that “continuing filling questionnaire constitutes your informed consent to act as a participant in this survey”.
Data Availability Statement
The data presented in this study are available on request from the corresponding author.
Conflicts of Interest
The authors declare no conflict of interest.
Appendix A. Questionnaire
(The questionnaire in Chinese is also available online at https://forms.gle/iB7BRfNGrwea8Dj4A (accessed on 17 March 2022)).
Section 1: Demographics and practice characteristicsGenderMale
Female
Age (years)18–29
30–39
40–49
50–64
≥65
Year of graduationAfter 2021
2016–2020
2011–2015
2001–2010
1991–2000
1981–1990
1971–1980
Before 1970
Type of clinical practice (Multiple choices were allowed.)Dogs and cats
Small animals (except dogs and cats)
Economic animals
Exotic animals
Researchers
Government veterinarians
Others
Average number of visiting cases per month0
1–30
31–60
61–100
100–200
>200
Section 2: Attitudes (6-point Likert scale from 1 to 6, 1 = strongly disagree, 2 = disagree, 3 = slightly disagree, 4 = slightly agree, 5 = agree, and 6 = strongly agree)Item 1. When animal abuse is suspected, the veterinarian has a moral or legal responsibility to intervene.
Item 2. In practice, when presented with a suspected animal abuse case, I know my legal rights and responsibilities.
Item 3. In practice, when presented with an animal abuse case, I would assist in preventing it from happening again.
Item 4. People who abuse animals are more likely to abuse their children.
Item 5. People who abuse animals are more likely to abuse their spouse.
Item 6. People who abuse animals are more likely to commit other types of crime.
Item 7. People who train animals using punitive methods (physical correction, choker leash, electric collar, etc.) are more likely to abuse animals.
Item 8. During veterinary training, I was provided with adequate information and training to identify and prevent animal abuse.
Item 9. In practice, when domestic violence is suspected, the veterinarian has a moral responsibility to intervene.
Item 10. In practice, when presented with child or spousal abuse, I feel I should provide assistance to the client.
Section 3: Management of abuse Q1. Indicate your willingness to report suspected animal and human abuse.Target of abuse: Animal🙢 Not report at all
🙢 Report severe cases
🙢 Report all cases
Target of abuse: Human🙢 Not report at all
🙢 Report severe cases
🙢 Report all cases
Q2. It is my civic responsibility to report abuse cases.Target of abuse: Animal🙢 Yes
🙢 No
Target of abuse: Human🙢 Yes
🙢 No
Q3. I can distinguish abuse features.Target of abuse: Animal🙢 Yes
🙢 No
Target of abuse: Human🙢 Yes
🙢 No
Q4. Dealing with abuse cases is beyond my profession or competence. Target of abuse: Animal🙢 Yes
🙢 No
Target of abuse: Human🙢 Yes
🙢 No
Q5. I am worried that dealing with abuse cases will irritate clients (i.e., the owner). Target of abuse: Animal🙢 Yes
🙢 No
Target of abuse: Human🙢 Yes
🙢 No
Q6. It should be mandatory to report abuse cases when veterinarians encounter deliberate abuse. Target of abuse: Animal🙢 Yes
🙢 No
Target of abuse: Human🙢 Yes
🙢 No
Section 4: Physical animal abuse (limited to veterinary practitioners with clinical experience) Q1. I believe I have the knowledge to identify suspected animal abuse.Yes
No
Q2. Frequency of suspected animal abuse cases I have encountered within the past five years No abuse case was found
≤1 time
2–3 times
4–11 times
>11 times
Q3. Who brought the abused animals to you? (Multiple choices were allowed.)Owner
Others
Public institutions (government veterinarians, animal protection officers, police, etc.)
Private groups
Q4. Reasons for suspecting the case to be physical animal abuse (Multiple choices were allowed. Open-ended question.)Nature of injury
Exposed or seen by witnesses
Owner’s behavior
Inconsistent medical history
Repeated presentation of injuries
Neglect
Q5. What was the species of the abused animal? (Multiple choices were allowed. Open-ended question.)Dog
Cat
Small animals (except dogs and cats)
Economic animal
Exotic animal
Q6. Injured body part caused by physical abuse. (Multiple choices were allowed. Open-ended question.)Head injury
Ocular injury or visual impairment
Ear and auricular impairment
Teeth injury
Sternum, ribs, or vertebra injury
Limb injury
Genital injury
Tail injury
Q7. Type of injury caused by physical abuse. (Multiple choices were allowed. Open-ended question.)Poisoning
Thermal burn (chemical or heat, etc.)
Friction burn
Asphyxia
Sharp force injuries (incised or stab wounds, etc.)
Bruises or hemorrhage
Abrasions or lacerations
Animal fighting (dogfighting, etc.)
Emaciation, malnutrition, or poor fur condition
Q8. Was suspected human abuse also involved in the animal abuse case?No abuse
Suspected abuse
Known abuse
No idea
Section 5: Mental animal abuse (limited to veterinary practitioners who had the clinical experience) Q1. I believe I have the knowledge to identify suspected animal abuse.Yes
No
Q2. Frequency of suspected animal abuse cases I have encountered within the past five yearsNo abuse case was found
≤ 1 time
2–3 times
4–11 times
> 11 times
Q3. Who brought the abused animals to you? (Multiple choices were allowed.)Owner
Others
Public institutions (government veterinarians, animal protection officers, police, etc.)
Private groups
Q4. Reasons for suspecting the case to be physical animal abuse (Multiple choices were allowed. Open-ended question.)Trembling, curl up
Anxiety
Aggressive
Alert, easily frightened
Q5. What was the species of the abused animal? (Multiple choices were allowed. Open-ended question.)Dog
Cat
Small animals (except dogs and cats)
Economic animal
Exotic animal
Q6. Was suspected human abuse also involved in the animal abuse case?No abuse
Suspected abuse
Known abuse
No idea
Figure 1 Percentages of respondents who responded to statements about how they would deal with abuse cases.
animals-12-01135-t001_Table 1 Table 1 Demographic and practice characteristics. (n = 247).
Demographic Characteristics Number (%)
Gender
Male
Female 86
161 34.8
65.2
Age (years)
18–29
30–39
40–49
50–64
≥65 161
52
18
16
0 65.2
21.1
7.3
6.4
0
Year of graduation
After 2021
2016–2020
2011–2015
2001–2010
1991–2000
1981–1990
1971–1980
Before 1970 100
60
32
24
24
5
2
0 40.5
24.3
13.0
9.7
9.7
2.0
0.8
0
Type of clinical practice *
Dogs and cats
Small animals (except dogs and cats)
Economic animals
Exotic animals
Researchers
Government veterinarians
Others 183
53
44
50
28
8
7 74.1
21.5
17.8
20.2
11.0
3.2
2.8
* Multiple choices were allowed.
animals-12-01135-t002_Table 2 Table 2 Attitude of veterinarians when faced with abuse cases.
Statement Mean Score ± SD Likert Scale
1 2 3 4 5 6
Percentage of Responses
Item 1
When animal abuse is suspected, the veterinarian has a moral or legal responsibility to intervene. 5.05 ± 1.04 0.4 2.0 5.2 19.4 30.6 42.3
Item 2
In practice, when presented with a suspected animal abuse case, I know my legal rights and responsibilities. 4.28 ± 1.56 6.5 10.9 10.9 20.6 22.6 28.6
Item 3
In practice, when presented with an animal abuse case, I would assist in preventing it from happening again. 5.17 ± 0.97 0.4 1.2 4.8 13.3 34.7 45.6
Item 4
People who abuse animals are more likely to abuse their children. 4.92 ± 1.28 2.0 3.2 8.5 21.4 17.3 47.6
Item 5
People who abuse animals are more likely to abuse their spouse. 4.74 ± 1.35 2.8 4.0 11.3 21.4 19.8 40.7
Item 6
People who abuse animals are more likely to commit other types of crime. 4.75 ± 1.43 2.4 6.5 12.1 17.3 16.1 45.6
Item 7
People who train animals using punitive methods (physical correction, choker leash, electric collar, etc.) are more likely to abuse animals. 4.14 ± 1.38 3.6 10.9 14.9 28.2 23.4 19.0
Item 8
During veterinary training, I was provided with adequate information and training to identify and prevent animal abuse. 3.07 ± 1.61 18.1 26.2 18.1 15.7 10.5 11.3
Item 9
In practice, when domestic violence is suspected, the veterinarian has a moral responsibility to intervene. 3.85 ± 1.47 6.5 14.9 16.1 27.4 18.5 16.5
Item 10
In practice, when presented with child or spousal abuse, I feel I should provide assistance to the client. 4.23 ± 1.41 3.2 10.5 15.7 23.0 25.0 22.6
Note: Mean scores and percentage of 6-point Likert scale from responses in Section 2 of the questionnaire in Appendix A (n = 247); 1 = strongly disagree, 2 = disagree, 3 = slightly disagree, 4 = slightly agree, 5 = agree, and 6 = strongly agree.
animals-12-01135-t003_Table 3 Table 3 Respondents’ demographic characteristics and their responses to the relationship between animal abuse and interpersonal violence and the intention to provide client assistance.
Mean Score ± SD
Item 4 Item 5 Item 3 Item 10 Item 8
Statement Animal abusers are more likely to… I need to assist when there’s… I was provided with adequate information in the veterinary training course.
abuse children. abuse spouse. animal abuse. child/spousal abuse.
Gender
Male
Female 4.83 ± 1.44
4.97 ± 1.19 4.67 ± 1.50
4.77 ± 1.27 5.09 ± 1.09
5.22 ± 0.89 4.37 ± 1.46
4.16 ± 1.37 3.24 ± 1.69
2.98 ± 1.56
Age (years)
18–29
30–39
40–49
50–64 4.80 ± 1.26
5.17 ± 1.26
4.83 ± 1.73
4.92 ±1.28 4.58 ± 1.34 a
5.08 ± 1.30 a
5.25 ± 0.93 a
4.74 ± 1.35 a 5.28 ± 0.88 b
5.02 ± 1.02 b
4.72 ± 1.27 b
5.17 ± 0.97 b 4.25 ± 1.35
4.06 ± 1.56
4.50 ± 1.34
4.38 ± 1.50 3.22 ± 1.57
2.77 ± 1.81
2.44 ± 1.10
3.25 ± 1.57
Graduation year
After 2021
2016–2020
2011–2015
2001–2010
1991–2000
1981–1990
1971–1980 4.85 ± 1.28
4.72 ± 1.25
5.13 ± 1.23
5.33 ± 0.87
5.04 ± 1.57
4.80 ± 2.17
5.00 ± 0.00 4.62 ± 1.30
4.61 ± 1.29
4.90 ± 1.60
5.25 ± 0.90
4.79 ± 1.59
4.80 ± 2.17
5.00 ± 0.00 5.38 ± 0.91
5.00 ± 0.84
5.10 ± 1.08
5.21 ± 0.88
4.83 ± 1.34
5.20 ± 0.84
5.00 ± 0.00 4.26 ± 1.49
4.31 ± 1.18
4.10 ± 1.45
4.21 ± 1.53
4.33 ± 1.40
4.20 ± 1.64
2.00 ± 0.00 3.54 ± 1.60 c
2.90 ± 1.61 c
2.42 ± 1.54 c
2.71 ± 1.49 c
2.83 ± 1.37 c
3.20 ± 1.92 c
2.00 ± 0.00 c
Graduation year
After 2021
Before 2020 4.85 ± 1.28
4.97 ± 1.28 4.62 ± 1.30
4.82 ± 1.38 5.38 ± 0.91 d
5.03 ± 0.98 d 4.26 ± 1.49
4.22 ± 1.35 3.54 ± 1.60 e
2.76 ± 1.53 e
In this table, except for the borderline significant differences in groups a and b (group a, p = 0.052; group b, p = 0.064) and significant differences in groups c, d, and e (group c, p = 0.008; group d, p = 0.005; group e, p < 0.001), there were no significant differences in the other groups.
animals-12-01135-t004_Table 4 Table 4 A comparative analysis of veterinarians’ willingness to report abuse cases and veterinarians’ attitudes toward abuse cases.
Mean Score ± SD p Value
Statement Target of Abuse Not Report at All Report Severe Cases Report All Cases
Item 1
When animal abuse is suspected, the veterinarian has a moral or legal responsibility to intervene. Animal
Human 4.00 ± 0.77
4.44 ± 1.04 4.76 ± 1.12
4.94 ± 1.00 5.41 ± 0.81
5.28 ± 1.01 <0.01
<0.01
Item 9
In practice, when domestic violence is suspected, the veterinarian has a moral responsibility to intervene. Animal
Human 2.45 ± 1.29
2.44 ± 1.16 3.51 ± 1.38
3.60 ± 1.37 4.20 ± 1.41
4.40 ± 1.36 <0.01
<0.01
Item 3
In practice, when presented with an animal abuse case, I would assist in preventing it from happening again. Animal
Human 3.91 ± 1.38
4.24 ± 1.01 4.95 ± 0.91
5.06 ± 0.96 5.50 ± 0.82
5.49 ± 0.80 <0.01
<0.01
Item 10
In practice, when presented with child or spousal abuse, I feel I should provide assistance to the client. Animal
Human 3.36 ± 1.75
2.64 ± 1.29 3.92 ± 1.34
3.99 ± 1.29 4.61 ± 1.34
4.81 ± 1.19 <0.01
<0.01
Item 4
People who abuse animals are more likely to abuse their children. Animal
Human 4.73 ± 1.95
4.80 ± 1.55 4.96 ± 1.27
4.90 ± 1.30 4.89 ± 1.23
4.97 ± 1.21 0.804
0.822
Item 5
People who abuse animals are more likely to abuse their spouse. Animal
Human 4.73 ± 1.95
4.32 ± 1.73 4.71 ± 1.41
4.72 ± 1.37 4.76 ± 1.25
4.84 ± 1.24 0.958
0.212
p Values < 0.05 were labeled in bold.
animals-12-01135-t005_Table 5 Table 5 Comparisons of mean scores of item 8 of Section 2 of the questionnaire in Appendix A (“During veterinary training, I was provided with adequate information and training to identify and prevent animal abuse”) among groups segregated by statement and target of abuse.
Mean Score ± SD p Value
Statement Target of Abuse No Yes
I can distinguish abuse features. Animal
Human 2.62 ± 1.50
2.61 ± 1.49 3.17 ± 1.62
3.27 ± 1.62 0.037
<0.01
Dealing with abuse cases is beyond my profession or competence. Animal
Human 3.33 ± 1.54
3.64 ± 1.65 2.77 ± 1.64
2.89 ± 1.55 <0.01
<0.01
I am worried that dealing with abuse cases will irritate clients. Animal
Human 3.42 ± 1.43
3.14 ± 1.46 2.98 ± 1.64
3.06 ± 1.64 0.077
0.734
p Values < 0.05 were labeled in bold.
animals-12-01135-t006_Table 6 Table 6 Comparisons between respondents who stated that it should be mandatory to report abuse cases and the consequences that they worried about and their attitude toward reporting abuse cases.
It Should Be Mandatory to Report
Animal Abuse.
Statement Target of Abuse Belief (n, %) No (n, %) Yes (n, %) p Value
Dealing with abuse cases is beyond my profession or competence. Animal No (134, 54.3%)
Yes (113, 45.7%) 17 (60.7%)
11 (39.3%) 117 (53.4%)
102 (46.6%) 0.466
Human No (59, 23.9%)
Yes (188, 76.1%) 6 (21.4%)
22 (78.6%) 53 (24.2%)
166 (75.8%) 0.746
It should be mandatory to report
human abuse.
Dealing with abuse cases is beyond my profession or competence. Animal No (134, 54.3%)
Yes (113, 45.7%) 51 (61.4%)
32 (38.6%) 83 (50.6%)
81 (49.4%) 0.106
Human No (59, 23.9%)
Yes (188, 76.1%) 11 (13.3%)
72 (86.7%) 48 (29.3%)
116 (70.7%) <0.01
It should be mandatory to report
animal abuse.
I am worried that dealing with abuse cases will irritate clients. Animal No (52, 21.1%)
Yes (195, 78.9%) 6 (21.4%)
22 (78.6%) 46 (21.0%)
173 (79.0%) 0.959
Human No (49, 19.8%)
Yes (198, 80.2%) 6 (21.4%)
22 (78.6%) 43 (19.6%)
176 (80.4%) 0.823
It should be mandatory to report
human abuse.
I am worried that dealing with abuse cases will irritate clients. Animal No (52, 21.1%)
Yes (195, 78.9%) 20 (24.1%)
63 (75.9%) 32 (19.5%)
132 (80.5%) 0.404
Human No (49, 19.8%)
Yes (198, 80.2%) 14 (16.9%)
69 (83.1%) 35 (21.3%)
129 (78.7%) 0.405
It should be mandatory to report
animal abuse.
It is my civic responsibility to report abuse cases. Animal No (22, 8.9%)
Yes (225, 91.1%) 10 (35.7%)
18 (64.3%) 12 (5.5%)
207 (94.5%) <0.01
Human No (30, 12.1%)
Yes (217, 87.9%) 7 (25.0%)
21 (75.0%) 23 (10.5%)
196 (89.5%) 0.057
It should be mandatory to report
human abuse.
It is my civic responsibility to report abuse cases. Animal No (22, 8.9%)
Yes (225, 91.1%) 12 (14.5%)
71 (85.5%) 10 (6.1%)
154 (93.9%) 0.029
Human No (30, 12.1%)
Yes (217, 87.9%) 21 (25.3%)
62 (74.7%) 9 (5.5%)
164 (94.5%) <0.01
It should be mandatory to report animal abuse.
Willingness to report abuse cases Animal Not report at all (11, 4.5%)
Report severe cases (114, 46.2%)
Report all cases (122, 49.4%) 4 (14.3%)
18 (64.3%)
6 (21.4%) 7 (3.2%)
96 (43.8%)
116 (53.0%) <0.01
Human Not report at all (25, 10.1%)
Report severe cases (107, 43.3%)
Report all cases (115, 46.6%) 6 (21.4%)
12 (42.9%)
10 (35.7%) 19 (8.7%)
95 (43.4%)
105 (47.9%) 0.135
It should be mandatory to report human abuse.
Willingness to report abuse cases Animal Not report at all (11, 4.5%)
Report severe cases (114, 46.2%)
Report all cases (122, 49.4%) 3 (3.6%)
48 (57.8%)
32 (38.6%) 8 (4.9%)
66 (40.2%)
90 (54.9%) 0.032
Human Not report at all (25, 10.1%)
Report severe cases (107, 43.3%)
Report all cases (115, 46.6%) 15 (18.1%)
46 (55.4%)
22 (26.5%) 10 (6.1%)
61 (37.2%)
93 (56.7%) <0.01
p Values < 0.05 were labeled in bold.
animals-12-01135-t007_Table 7 Table 7 Respondents’ demographic characteristics and their attitude toward whether it should be mandatory to report abuse cases.
Statement It Should Be Mandatory to Report Abuse Cases When Veterinarians Encounter Deliberate abuse.
Target of Abuse: Animal Target Of Abuse: Human
Characteristics (n, %) No (n, %) Yes (n, %) p Value No (n, %) Yes (n, %) p Value
Gender 0.343 0.977
Male (86, 34.8%)
Female (161, 65.2%) 12 (14.0%)
16 (9.9%) 74 (86.0%)
145 (90.1%) 29 (33.7%)
54 (33.5%) 57 (66.3%)
107 (66.4%)
Age (years) 0.740 0.526
18–29 (161, 65.2%)
30–39 (52, 21.1%)
40–49 (18, 7.3%)
50–64 (16, 6.5%) 17 (10.6%)
7 (13.5%)
3 (16.7%)
1 (6.3%) 144 (89.4%)
45 (86.5%)
15 (83.3%)
15 (93.8%) 57 (35.4%)
16 (30.8%)
7 (38.9%)
3 (18.8%) 104 (64.6%)
36 (69.2%)
11 (61.1%)
13 (81.3%)
Graduation year 0.836 0.412
After 2021 (100, 40.5%)
2016–2020 (61, 24.7%)
2011–2015 (31, 12.6%)
2001–2010 (24, 9.7%)
1991–2000 (24, 9.7%)
1981–1990 (5, 2.0%)
1971–1980 (2, 0.8%) 10 (10.0%)
8 (13.1%)
3 (9.7%)
4 (16.7%)
3 (12.5%)
0 (0.0%)
0 (0.0%) 90 (90.0%)
53 (86.9%)
28 (90.3%)
20 (83.3%)
21 (87.5%)
5 (100.0%)
2 (100.0%) 34 (34.0%)
21 (34.4%)
10 (32.3%)
9 (37.5.6%)
9 (37.5%)
0 (0.0%)
0 (0.0%) 66 (66.0%)
40 (65.6%)
21 (67.7%)
15 (62.5%)
15 (62.5%)
5 (100.0%)
2 (100.0%)
animals-12-01135-t008_Table 8 Table 8 A comparative analysis of management of abuse and veterinarians’ graduation status.
Veterinarians’ Graduation Status
Statement Target of Abuse Belief (n, %) After 2021 (n, %) Before 2020 (n, %) p Value
I can distinguish abuse features. Animal No (45, 18.2%)
Yes (202, 81.8%) 22 (22.0%)
78 (78.0%) 23 (15.6%)
124 (84.4%) 0.204
Human No (75, 30.4%)
Yes (172, 69.6%) 26 (26.0%)
74 (74.0%) 49 (33.3%)
98 (66.7%) 0.219
Dealing with abuse cases is beyond my profession or competence. Animal No (134, 54.3%)
Yes (113, 45.7%) 59 (59.0%)
41 (41.0%) 75 (51.0%)
72 (49.0%) 0.217
Human No (59, 23.9%)
Yes (188, 76.1%) 28 (28.0%)
72 (72.0%) 31 (21.1%)
116 (78.9%) 0.211
I am worried that dealing with abuse cases will irritate clients. Animal No (52, 21.1%)
Yes (195, 78.9%) 25 (25.0%)
75 (75.0%) 27 (18.4%)
120 (81.6%) 0.209
Human No (49, 19.8%)
Yes (198, 80.2%) 24 (24.0%)
76 (76.0%) 25 (17.0%)
122 (83.0%) 0.176
It is my civic responsibility to report abuse cases. Animal No (22, 8.9%)
Yes (225, 91.1%) 7 (7.0%)
93 (93.0%) 15 (10.2%)
132 (89.8%) 0.386
Human No (30, 12.1%)
Yes (217, 87.9%) 12 (12.0%)
88 (88.0%) 18 (12.2%)
129 (87.8%) 0.954
It should be mandatory to report abuse cases when veterinarians encounter deliberate abuse. Animal No (28, 11.3%)
Yes (219, 88.7%) 10 (10.0%)
90 (90.0%) 18 (12.2%)
129 (87.8%) 0.585
Human No (83, 33.6%)
Yes (164, 66.4%) 34 (34.0%)
66 (66.0%) 49 (33.3%)
98 (66.7%) 0.913
Willingness (n, %)
Willingness to report abuse cases Animal Not report at all (11, 4.5%)
Report severe cases (114, 46.2%)
Report all cases (122, 49.4%) 2 (2.0%)
34 (34.0%)
64 (64.0%) 9 (6.1%)
80 (54.4%)
58 (39.5%) <0.01
Human Not report at all (25, 10.1%)
Report severe cases (107, 43.3%)
Report all cases (115, 46.6%) 7 (7.0%)
36 (36.0%)
57 (57.0%) 18 (12.2%)
71 (48.3%)
58 (39.5%) 0.022
p Values < 0.05 were labeled in bold.
animals-12-01135-t009_Table 9 Table 9 Frequency of suspected physical and mental animal abuse cases encountered by veterinarians in the past five years in Taiwan.
Average Frequency of Encounters in a Year Raw Number (%) Number Possessing Knowledge (%)
Physical Abuse Mental Abuse Physical Abuse Mental Abuse
No abuse case was found 57 (42.9%) 61 (45.9%) 43 (36.1%) 41 (36.3%)
≤1 time 31 (23.0%) 32 (24.1%) 31 (26.1%) 32 (28.3%)
2–3 times 33 (24.8%) 32 (24.1%) 33 (27.7%) 32 (28.3%)
4–11 times 7 (5.3%) 5 (3.8%) 7 (5.9%) 5 (4.4%)
>11 times 5 (4.0%) 3 (2.2%) 5 (4.2%) 3 (2.7%)
Total 133 133 119 113
Note: Raw number indicates that practicing veterinarians (n = 133) believed they had encountered animal abuse case(s); the number possessing knowledge indicates among veterinarians who considered they had the knowledge to identify physical abuse cases or who had encountered abuse case(s).
animals-12-01135-t010_Table 10 Table 10 Who brought physically abused animals to the veterinarian, what species of abused animal was involved, and whether interpersonal violence was also involved in the animal abuse case. The results were calculated only from the answers of 119 veterinarians who claimed they had the knowledge to identify physical abuse, or who had encountered abuse cases.
Statement Number (%)
Who brought the abused animal to the veterinarian? *
Owner
Others
Public institutions (government veterinarians, animal protection officers, police, etc.) 61 (51.3%)
34 (28.6%)
15 (12.6%)
Private groups 19 (16.0%)
What species was the abused animal? *
Dog
Cat
Small animals (except dogs and cats)
Economic animal
Exotic animal
Laboratory animal ^ 61 (51.3%)
53 (44.5%)
12 (10.1%)
7 (5.9%)
6 (5.0%)
1 (0.8%)
Was suspected human abuse also involved in the animal abuse case?
No abuse
Suspected abuse
Known abuse
No idea 72 (60.5%)
9 (7.6%)
1 (0.8%)
37 (31.1%)
* Multiple choices were allowed. ^ Options added by respondents.
animals-12-01135-t011_Table 11 Table 11 Reasons for suspecting the case to be physical animal abuse. Multiple choices were allowed for this question. This was an open-ended question, and the respondents could add options themselves. The results were calculated only from the answers of 119 veterinarians who claimed they had the knowledge to identify physical abuse, or who had encountered abuse cases.
Statement Number (%)
Nature of injury 56 (47.1%)
Neglect 52 (43.7%)
Owner’s behavior 36 (30.3%)
Repeated presentation of injuries 21 (17.6%)
Inconsistent medical history 20 (16.8%)
Exposed or seen by witnesses 18 (15.1%)
Witness at clinic ^ 1 (0.8%)
Inhumane slaughter 1 (0.8%)
^ Options added by respondents.
animals-12-01135-t012_Table 12 Table 12 Reasons for suspecting the case to be mental animal abuse. Multiple choices were allowed for this question. This was an open-ended question, and the respondents could add options themselves. The results were calculated only from the answers of 113 veterinarians who claimed they had the knowledge to identify mental abuse, or who had encountered abuse cases.
Statement Number (%)
Alert, easily frightened 42 (37.2%)
Trembling, curl up 35 (31.0%)
Anxiety 29 (25.7%)
Aggressive 27 (23.0%)
Self-harm and compulsive behavior ^ 1 (0.9%)
^ Options added by respondents. We listed the mental animal abuse table here because it could reflect the interaction situation when animals were abused; however, we did not perform further statistical analyses based on these data.
animals-12-01135-t013_Table 13 Table 13 The injured body part and type of injury caused by physical abuse. Multiple choices were allowed for these questions. These were open-ended questions, and the respondents could add options themselves.
Statement Number (%)
Injured body part
Limb injury
Head injury
Sternum, ribs, or vertebra injury
Ocular injury or visual impairment
Tail injury
Genital injury
Teeth injury
Ear and auricular impairment
Abdominal injuries (abdominal trauma or organ hemorrhage) ^
Pelvic fracture ^
Improper stun and exsanguination ^ 50 (42.0%)
31 (26.1%)
26 (21.8%)
16 (13.4%)
10 (8.4%)
6 (5.0%)
5 (4.2%)
2 (1.7%)
2 (1.7%)
1 (0.8%)
1 (0.8%)
Type of injury
Emaciation, malnutrition, or poor fur condition
Bruises or hemorrhage
Abrasions or lacerations
Sharp force injuries (incised or stab wounds, etc.)
Poisoning
Thermal burn (chemical or heat, etc.)
Animal fighting (dogfighting, etc.)
Friction burn
Asphyxia
Fracture or dislocation ^
Neurological signs (carrying) ^
Clamped by traps or tied with rubber bands ^
Firearm injuries ^ 49 (41.2%)
41 (34.5%)
29 (24.4%)
28 (23.5%)
16 (13.4%)
13 (10.9%)
13 (10.9%)
9 (7.6%)
5 (4.2%)
4 (3.4%)
1 (0.8%)
1 (0.8%)
1 (0.8%)
^ Options added by respondents.
animals-12-01135-t014_Table 14 Table 14 Comparisons between the suspected physical animal abuse cases encountered by veterinarians and their attitude toward reporting suspected animal abuse.
Average Frequency of Encounter in a Year Not Report at All Report Severe Cases Report All Cases
No abuse case was recorded 1 (14.3%) 15 (24.2%) 25 (52.1%)
≤1 time 2 (28.6%) 21 (33.9%) 8 (16.7%)
2–3 times 1 (14.3%) 21 (33.9%) 11 (22.9%)
4–11 times 2 (28.6%) 4 (6.5%) 1 (2.1%)
>11 times 1 (14.3%) 1 (1.6%) 3 (6.3%)
Total 7 62 48
Note: There was a significant difference (p = 0.013).
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Sharpe M.S. Wittum T.E. Veterinarian involvement in the prevention and intervention of human violence and animal abuse: A survey of small animal practitioners Anthrozoös 1999 12 97 104 10.2752/089279399787000309
2. McMillan F.D. Duffy D.L. Zawistowski S.L. Serpell J.A. Behavioral and psychological characteristics of canine victims of abuse J. Appl. Anim. Welf. Sci. 2015 18 92 111 10.1080/10888705.2014.962230 25257564
3. The Animal Protection Act Available online: https://law.moj.gov.tw/ENG/LawClass/LawAll.aspx?pcode=M0060027 (accessed on 10 October 2020)
4. Ascione F.R. Children who are cruel to animals: A review of research and implications for developmental psychopathology Anthrozoös 1993 6 226 247 10.2752/089279393787002105
5. Vermeulen H. Odendaal J.S. Proposed typology of companion animal abuse Anthrozoös 1993 6 248 257 10.2752/089279393787002178
6. Gullone E. Animal Cruelty, Antisocial Behaviour, and Aggression: More than a Link Springer Berlin/Heidelberg, Germany 2012
7. Lockwood R. Arkow P. Animal abuse and interpersonal violence: The cruelty connection and its implications for veterinary pathology Vet. Pathol. 2016 53 910 918 10.1177/0300985815626575 26936222
8. Hellman D.S. Blackman N. Enuresis, firesetting and cruelty to animals: A triad predictive of adult crime Am. J. Psychiatry 1966 122 1431 1435 10.1176/ajp.122.12.1431 5929498
9. Randour M.L. Smith-Blackmore M. Blaney N. DeSousa D. Guyony A.-A. Animal abuse as a type of trauma: Lessons for human and animal service professionals Trauma Violence Abus. 2021 22 277 288 10.1177/1524838019843197 31043145
10. Kogan L.R. Schoenfeld-Tacher R.M. Hellyer P.W. Rishniw M. Ruch-Gallie R.A. Survey of attitudes toward and experiences with animal abuse encounters in a convenience sample of US veterinarians J. Am. Vet. Med. Assoc. 2017 250 688 696 10.2460/javma.250.6.688 28263111
11. Arkow P. Recognizing and responding to cases of suspected animal cruelty, abuse, and neglect: What the veterinarian needs to know Vet. Med. Res. Rep. 2015 6 349 10.2147/VMRR.S87198 30101120
12. Wisch R.F. Table of Veterinary Reporting Requirement and Immunity Laws Available online: https://www.animallaw.info/topic/table-veterinary-reporting-requirement-and-immunity-laws (accessed on 13 October 2020)
13. Monsalve S. Ferreira F. Garcia R. The connection between animal abuse and interpersonal violence: A review from the veterinary perspective Res. Vet. Sci. 2017 114 18 26 10.1016/j.rvsc.2017.02.025 28279899
14. Williams V.M. Dale A. Clarke N. Garrett N. Animal abuse and family violence: Survey on the recognition of animal abuse by veterinarians in New Zealand and their understanding of the correlation between animal abuse and human violence N. Z. Vet. J. 2008 56 21 28 10.1080/00480169.2008.36800 18322556
15. Green P. Gullone E. Knowledge and attitudes of Australian veterinarians to animal abuse and human interpersonal violence Aust. Vet. J. 2005 83 619 625 10.1111/j.1751-0813.2005.tb13275.x 16255286
16. BAPHIQ Statistical Annual Report Available online: https://www.baphiq.gov.tw/ws.php?id=21384 (accessed on 13 October 2020)
17. Joo S. Jung Y. Chun M.-S. An Analysis of Veterinary Practitioners’ Intention to Intervene in Animal Abuse Cases in South Korea Animals 2020 10 802 10.3390/ani10050802
18. McEwen B.J. McDonough S.P. A Survey of Attitudes of Board-Certified Veterinary Pathologists to Forensic Veterinary Pathology Vet. Pathol. 2016 53 1099 1102 10.1177/0300985815625756 26926083
19. Arkow P. The relationships between animal abuse and other forms of family violence Fam. Violence Sex. Assault Bull. 1996 12 29 34
20. Alleyne E. Sienauskaite O. Ford J. To report, or not to report, animal abuse: The role of perceived self-efficacy in veterinarians’ decision-making Vet. Rec. 2019 185 538 10.1136/vr.105077 31659115
|
PMC009xxxxxx/PMC9099902.txt |
==== Front
Molecules
Molecules
molecules
Molecules
1420-3049
MDPI
10.3390/molecules27092882
molecules-27-02882
Communication
Dihydroisatropolone C from Streptomyces and Its Implication in Tropolone-Ring Construction for Isatropolone Biosynthesis
Liu Jiachang
Liu Xiaoyan
Fu Jie
Jiang Bingya
Li Shufen *
Wu Linzhuan *
Comte Gilles Academic Editor
NHC Key Laboratory of Biotechnology of Antibiotics, CAMS Key Laboratory of Synthetic Biology for Drug Innovation, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences & Peking Union Medical College, Tiantan Xili, Beijing 100050, China; [email protected] (J.L.); [email protected] (X.L.); [email protected] (J.F.); [email protected] (B.J.)
* Correspondence: [email protected] (S.L.); [email protected] (L.W.)
30 4 2022
5 2022
27 9 288204 4 2022
28 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Isatropolones/isarubrolones are actinomycete secondary metabolites featuring a tropolone-ring in their structures. From the isatropolone/isarubrolone producer Streptomyces sp. CPCC 204095, 7,12-dihydroisatropolone C (H2ITC) is discovered and identified as a mixture of two interchangeable diastereomers differing in the C-6 configuration. As a major metabolite in the mycelial growth period of Streptomyces sp. CPCC 204095, H2ITC can be oxidized spontaneously to isatropolone C (ITC), suggesting H2ITC is the physiological precursor of ITC. Characterization of H2ITC makes us propose dihydrotropolone-ring construction in the biosynthesis of isatropolones.
Streptomyces
7,12-dihydroisatropolone C
spontaneous oxidation
National Key Research and Development Program of China2018YFA0902000 National Natural Science Foundation of China81903530 CAMS Innovation Fund for Medical Sciences2021-I2M-1-055 National Infrastructure of Microbial ResourcesNIMR-2018-3 This work was supported by National Key Research and Development Program of China (2018YFA0902000), National Natural Science Foundation of China (81903530), CAMS Innovation Fund for Medical Sciences (CIFMS, 2021-I2M-1-055) and National Infrastructure of Microbial Resources (No. NIMR-2018-3).
==== Body
pmc1. Introduction
Isatropolones/isarubrolones are a complex of actinomycete secondary metabolites featuring a tropolone-ring in their structures, with isarubrolones being the non-enzymatic conjugation products of isatropolones with amines or amino acids [1]. Rubrolones and rubterolones are also actinomycete secondary metabolites featuring a tropolone-ring in their structures [2,3,4]. These metabolites share very similar structures and biosynthetic pathways [1,5,6]. In particular, the tropolone-rings in these metabolites are all constructed by enzyme-catalyzed oxidative rearrangement of poly-β-ketoacyl intermediates from type-II polyketide synthase (PKS) pathways [5]. However, tropolone-ring construction in the biosynthesis of these secondary metabolites has not been chemically confirmed, possibly because intermediates appearing in the tropolone-ring construction process are rather unstable.
We are interested in novel secondary metabolites, together with their biosynthesis, from actinomycetes [7,8,9]. Previously, we identified an isatropolone/isarubrolone producer Streptomyces sp. CPCC 204095 and discovered autophagic activity for the isatropolones/isarubrolones characterized from the strain. In a recent study of isatropolones/isarubrolones and their production by Streptomyces sp. CPCC 204095, we identified 7,12-dihydroisatropolone C (H2ITC) as a novel component of isatropolones, and determined it as the precursor of isatropolone C (ITC). In particular, characterization of H2ITC led us to present a revised tropolone-ring construction process in isatropolone (and rubrolone) biosynthesis that had been proposed before [1,5].
2. Results and Discussion
2.1. Discovery of 7,12-Dihydroisatropolone C (H2ITC) from Streptomyces sp. CPCC 204095
Previously, we identified the isatropolone/isarubrolone producer Streptomyces sp. CPCC 204095, with isatropolone C (Figure 1a) as a major component [10]. In the exploration of more metabolites with a tropolone-ring from the producer, we observed one HPLC pair peaks with identical but novel UV–visible absorption from an acidified ethyl acetate (EtOAc-5% acetic acid) extract of fermentation culture of the producer (Figures S1, S4 and S5). In addition, the acidified EtOAc extract produced a lower isatropolone C (ITC) peak than the EtOAc extract that gave a dominant ITC peak (Figure S2), suggesting a relationship between the pair peaks and the ITC peak. The pair peaks aroused our interest in their identities.
2.2. Structural Elucidation of 7,12-Dihydroisatropolone C (H2ITC)
The pair peaks exhibited the same molecular mass (m/z at 475 for [M+H]+) and MS2 fragmentation pattern in LC-MS analysis (Figure S6), indicating two isomeric molecules in the pair peaks. When the pair peaks were separated by HPLC, each peak would change back instantly to the former pair peaks, and the latter peak was always higher than the former peak. Therefore, compounds in the pair peaks were purified as a whole (mixture of epimers) for structure elucidation (Figure S3).
A procedure of silica gel column chromatography, ODS column chromatography and reverse phase HPLC was used to purify the compound (1ab) in the pair peaks. Specifically, elutes from column chromatography and HPLC were stored at low temperature (0 or −20 °C), as 1ab was able to change slowly to ITC at room temperature. The purified 1ab sample was kept at −20 °C, and its NMR assay was conducted at −4 °C.
Compound 1ab was obtained as light-yellow amorphous powder. HRESIMS established its molecular formula C24H26O10, two hydrogen atoms more than ITC (Figure S7). Its NMR spectra indicated a major set of signals for 1ab, together with an expected minor set of signals for ITC (Figure S8). Luckily, these minor signals did not cause much difficulty in recognizing the major signals. A close examination of major signals revealed that they were all paired ones, indicating that 1ab was a mixture of two diastereomers.
The 1H and 13C NMR spectra of 1ab were very similar to ITC, except that a pair methine signals [δC (40.48; 40.37), δH (2.91, d; 2.74, 2H, d)] and a pair methene signals [δC (42.33; 42.24), δH (3.65;3.56, 2H, dd)] replaced the two sp2 carbon signals [δC (134), δC (135), δH (7.12, s), δH (7.77, s)] in ITC. The structure of 1ab was established as 7,12-dihydrogenated ITC (Figure 1a) based on detailed analysis of NMR spectra, especially the 1H-1H COSY correlations of H-7/H-12 and the HMBC correlations from H-7 to C-8 and C-9 (Figure 2). Additionally, compared with ITC, the C-6, C-8, C-11 and C-9 resonances of 1ab were shielded by ΔδC −2.4, −1.6, −2.8 and −0.5 ppm, respectively, which further supported 1ab as 7,12- dihydroisatropolone C (H2ITC). The NMR data of H2ITC (1ab) were assigned as in Table 1 (Figures S9–S21).
2.3. Keto-Enol Tautomerization of 7,12-Dihydroisatropolone C (H2ITC, 1ab)
Keto-enol tautomerizations have been reported for rubrolones and rubterolones [5,6]. A keto-enol tautomerization at C-6 of H2ITC (1ab) may result in a mixture of two diastereomers (1a and 1b) differing in the chiral C-12 configuration (Figure 1b). The one with C-12 S configuration was designated as 1a, and the other one with C-12 R configuration as 1b. The ΔG calculated by Multiwfn at ωB97XD/TZVP level revealed only a small energy difference for 1a and 1b [11]. Population distribution was 66.29% for 1a and 33.71% for 1b when they reached equilibrium, approximately agreeing with peak area ratio of the pair peaks (Figure 1b). Thus, 1a was assigned to the higher one, and 1b to the lower one, of the pair peaks.
2.4. Spontaneous Oxidation of 7,12-Dihydroisatropolone C (H2ITC, 1ab) to Isatropolone C (ITC)
The mechanism for spontaneous oxidation of H2ITC to ITC was also proposed as in Figure 1b, in which keto-enol exchange occurred at both C-6 and C-8 to generate H2ITC double-enol form. Like phenols/hydroquinones [12], H2ITC double-enol form is sensitive to air (O2) oxidation, becoming ITC upon abstraction of two hydrogen atoms. The oxidation process was sped up at high temperature and pH, or slowed down at low temperature and pH. A careful examination of the 1H-NMR spectrum of H2ITC revealed two low-field signals at δH 11.5 and 11.4 for active hydrogen atoms that were further proved by deuterium exchange (Figure S22), thus confirming the existence of the H2ITC double-enol form in H2ITC.
Spontaneous oxidation of H2ITC to ITC was then compared at: (a) 20 °C plus pH7.0, (b) −20 °C plus pH7.0 and (c) 20 °C plus pH8.0. A proportion of ca. 23% H2ITC was oxidized to ITC in 60 h at 20 °C plus pH7.0, and all H2ITC was oxidized to ITC in 40 min at 20 °C plus pH8.0, while H2ITC remained nearly unchanged for 60 h at −20 °C plus pH7.0 (Figures S25–S27). In addition, a time-course monitoring of Streptomyces sp. CPCC 204095 indicated a slightly higher titer of H2ITC than ITC in the mycelial growth period (26–32 h) of the strain, then a higher titer of ITC than H2ITC afterwards due to H2ITC oxidation to ITC and its accumulation (Figure 3; ITC titer declined after 45 h due to ITC conjugating amines and amino acids for isarubrolone production). These results indicate H2ITC is the physiological precursor of ITC.
Isatropolones are able to conjugate amines to form isarubrolones. H2ITC was explored for its conjugation with NH3 for 7,12-dihydroisarubrolone C formation. However, 7,12-dihydroisarubrolone C was not identified from H2ITC with NH3, while isarubrolone C was produced from the reaction. A possible reason for this may be that 7,12-dihydroisarubrolone C is much more sensitive to oxidation than H2ITC, so it is oxidized immediately to isarubrolone C after its formation (Figure S28).
It is very interesting that H2ITC can be spontaneously oxidized to ITC. A similar case has been reported for kibdelone B, a heterocyclic polyketide from Kibdelosporangium. Keto-enol tautomerizations of kibdelone B lead to aromatization of its ring C. Then, the aromatized intermediate, as a hydroquinone, undergoes spontaneous oxidation to convert kibdelone B with a C-C single bond in ring C to kibdelone A with a corresponding C-C double bond [13].
2.5. Dihydrotropolone-Ring Construction in Isatropolone Biosynthesis Implicated by H2ITC
As the most interesting and intriguing part in the biosynthesis of some actinomycete secondary metabolites with a tropolone-ring, Yan et al. proposed a tropolone-ring construction for rubrolones based on feeding with [13C]-acetate [5]. Specifically, a mono-cyclic/aromatic intermediate derived from a poly-β-ketoacyl chain underwent complex oxidative rearrangement to construct the tropolone-ring (route 1, Scheme S1), in which two oxidations at C-11 and C-12 occurred. A similar tropolone-ring construction using a bi-cyclic/aromatic intermediate was proposed for isatropolone biosynthesis by Cai et al. (route 2, Scheme S1) [1]. However, neither tropolone-ring constructions is chemically proved.
The discovery of H2ITC from Streptomyces sp. CPCC 204095 and its spontaneous oxidation to ITC made us propose a dihydrotropolone-ring construction in isatropolone biosynthesis (Scheme 1). The construction involved only one oxidation at C-11 in the oxidative rearrangement. The dihydrotropolone-ring in H2ITC could be converted to the tropolone-ring in ITC by spontaneous oxidation.
Genetic studies have demonstrated that two oxygenase genes are essential for the oxidative rearrangement of tropolone-ring construction in rubrolone biosynthesis [14]. However, the biochemical roles of the two oxygenases are still not clear, as substrates (the mono- or bi-cyclic/aromatic intermediates with polyketoacyl chains in Scheme 1) for the two oxygenases are very unstable and difficult to prepare, which prevents in vitro characterization of the biochemical reactions they catalyze.
Cai et al. conducted a heterologous expression of istG-R from an isatropolone gene cluster for aglycone biosynthesis in Streptomyces lividans, which resulted in the identification of two aglycones (compound 10 and its reduced derivative compound 9) [1]. Recently, Yijun Yan et al. reported multifunctional and non-stereoselective oxidoreductase RubE7/IstO for oxidation and reduction of these aglycones [15]. Our discovery of H2ITC seems to suggest that the primary aglycone in isatropolone biosynthesis may be a dihydrogenated 10. Its glycosylation (and hydroxylation) in Streptomyces sp. CPCC 204095 leads to the production of H2ITC, whose spontaneous oxidation generates ITC (Scheme 1, Figure S29).
3. Materials and Methods
3.1. General Experimental Procedures
HPLC was conducted on an Agilent system with a 1260 Quat-Pump and DAD detector. For analytical HPLC, a reverse-phase C18 column (YMC-Pack ODS-A column: 250 mm × 4.6 mm, S-5 μm, 12 nm) was used with a gradient solvent system from 15% to 70% CH3CN-H2O (0.1% HAc, v/v), 1.0 mL/min. For semipreparative HPLC, a reverse-phase C18 column (YMC-Pack ODS-A column: 250 mm × 10 mm, S-5 μm, 12 nm) was used with an isocratic solvent system for 30% CH3CN-H2O (0.1% HAc, v/v), 1.5 mL/min. UV spectra were acquired with a Thermo Scientific (New York, NY, USA) Evolution 201 UV–visible spectrophotometer. NMR data were collected using an ADVANCE HD 800 MHz and a Bruker Avance III HD 700 MHz spectrometer, where chemical shifts (δ) were reported in ppm and referenced to the acetone-d6 solvent signal (δH 2.05 and δC 29.84, 206.33). ESIMS and MS2 were conducted on a 1100-6410 Triple Quad from Agilent (Santa Clara, CA, USA). HRESIMS were conducted on a Thermo (New York, NY, USA) LTQ Orbitrap XL.
3.2. Fermentation of Streptomyces sp. CPCC 204095
Frozen stock spores of Streptomyces sp. CPCC 204095 were thawed, inoculated on culture medium (soluble starch 2.0%, yeast extract 0.4%, malt extract 1.0%, glucose 0.4% and agar 1.5%), and incubated at 28 °C for 7 days for sporulation. Fresh spores were collected and spread on fermentation medium (yeast extract 0.6%, malt extract 2.0%, glucose 1.5%, soybean cake 0.6% and agar 1.5%) plates and incubated at 28 °C for 32–36 h for 1ab (H2ITC) production, or for 20–60 h for monitoring 1ab (H2ITC) and ITC titers.
3.3. Extraction and Isolation of Compound 1ab
Fermentation culture (5 L) of Streptomyces sp. CPCC 204095 was extracted with an equal volume of EtOAc (5% HAc, v/v) two times. Specifically, each extraction took no longer than a few hours. The combined organic layer was vacuum dried below 30 °C, yielding a dark brown residue (4.85 g). The residue was loaded onto a preparative silica column for fractionation with CH2Cl2-MeOH (0% MeOH, 20 min; 1% MeOH, 20 min; 2% MeOH, 30 min; 3% MeOH, 60 min, v/v) at a constant flow rate of 35 mL/min, which yielded 14 fractions from F1-1 to F1-14. Each fraction was analyzed by TLC and HPLC. Fractions from F1-2 to F1-5 were found to contain compound 1ab.
Fractions from F1-2 to F1-5 were combined and concentrated under reduced pressure below 30 °C, yielding a dark brown residue. The residue was then loaded onto a preparative ODS column for fractionation with MeOH-H2O (20% MeOH-H2O, 30 min; 25% MeOH-H2O, 40 min; 30% MeOH-H2O, 15 min; 35% MeOH-H2O, 300 min and finally 100% MeOH, 20 min; H2O contained 0.1% HAc, v/v) at a constant flow rate of 25 mL/min, which yielded 72 fractions from F2-1 to F2-72. Each fraction was analyzed by HPLC. Fractions from F2-25 to F2-44 were found to contain compound 1ab, so they were combined. A part of the preparation was used for semipreparative HPLC (30% MeCN-H2O, 1.5 mL/min, tR = 37 min) to obtain the 1ab sample (2.0 mg) for NMR.
3.4. Quantitative Assay of H2ITC (1ab) and Isatropolone C (ITC)
A freshly prepared 1ab solution was divided into two parts. One part was vacuum-dried to obtain the quantity of 1ab in the solution. The other part was serially diluted for analytical HPLC, establishing a linear relationship of 1ab quantity with its pair peaks area. The linear relationship was used for HPLC assay of 1ab, or 1ab titer in Streptomyces sp. CPCC 204095. Quantitative assay of isatropolone C (ITC) was also conducted by HPLC in a way similar to H2ITC, according to Liu Xiaoyan et al. (Figures S23–S24) [16].
3.5. Spontaneous Oxidation of H2ITC (1ab) to Isatropolone C
(1) H2ITC (1ab) was dissolved in 3.0 mL 30% MeOH/H2O (pH7.0) at a concentration of 0.45 mg/mL. It was equally separated into two 2.0 mL glass vials. One vial was kept at 20 °C, and the other at −20°C. The two vials were then analyzed by HPLC for 1ab and isatropolone C derived from spontaneous oxidation of 1ab in 20 and 60 h, respectively. (2) H2ITC (1ab) was dissolved in 1.0 mL 30% MeOH/H2O plus 0.5 mL phosphate buffer (0.1 mol/L, pH8.0) at a concentration of 0.45 mg/mL within a 2.0 mL glass vial. It was kept at 20 °C, and then analyzed by HPLC for 1ab and isatropolone C derived from spontaneous oxidation of 1ab in 1.0 and 40.0 min, respectively.
Acknowledgments
We would like to thank Li Li from the Institute of Materia Medica, CAMS & PUMC, for performing the Multiwfn calculation. NMR and MS analyses were performed at the Nuclear Magnetic Resonance Center of the Institute of Materia Medica, CAMS & PUMC.
Supplementary Materials
The following supporting information can be downloaded at: www.mdpi.com/article/10.3390/molecules27092882/s1. Figure S1. Three HPLC chromatograms of EtOAc extract (with an extraction time of 2 h) of Streptomyces sp. CPCC 204095. Figure S2. A parallel HPLC analysis of Streptomyces sp. CPCC 204095 revealing a relationship of the pair peaks with isatropolone C peak. Figure S3. HPLC separation of the pair peaks and confirmation of their exchange. Figure S4. HPLC of freshly prepared compound 1ab. Figure S5. UV–visible absorption of compound 1ab. Figure S6. LC-MS of compound 1ab sample containing a small amount of isatropolone C. Figure S7. HRESIMS of compound 1ab. Figure S8. Alignment of 13C NMR spectra of 7,12-dihydroisatropolone C (1ab) and isatropolone C. Figure S9. 1H NMR spectrum (700 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S10. 13C NMR spectrum (700 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S11. DEPT spectrum (700 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S12. 1H-1H COSY spectrum (800 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S13. HSQC spectrum (800 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S14. HMBC spectrum (800 MHz) of 7,12-dihydroisatropolone C (1ab) in acetone-d6. Figure S15. 13C-NMR spectrum (0–30 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S16. 13C-NMR spectrum (30–60 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S17. 13C-NMR spectrum (60–90 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S18. 13C-NMR spectrum (90–120 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S19. 13C-NMR spectrum (120–150 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S20. 13C-NMR spectrum (150–180 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S21. 13C-NMR spectrum (180–210 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S22. Active hydrogen atom signals in the 1H-NMR spectrum (10–12 ppm) of 7,12-dihydroisatropolone C (1ab). Figure S23. HPLC for an identical amount of H2ITC and ITC. Figure S24. Analytical HPLC of various amounts of H2ITC. Figure S25. HPLC of 7,12-dihydroisatropolone C (in 30% MeOH) changing to isatropolone C at pH7.0 plus 20 °C. Figure S26. HPLC of 7,12-dihydroisatropolone C (in 30% MeOH) changing to isatropolone C at pH8.0 plus 20 °C. Figure S27. HPLC of 7,12-dihydroisatropolone C (in 30% MeOH) changing to isatropolone C at pH7.0 plus −20 °C. Figure S28. H2ITC conjugates NH3 for 7,12-dihydroisarubrolone C production. Scheme S1. Biosynthesis of rubrulone A focusing on oxidative rearrangement for tropolone-ring construction from mono-cyclic/aromatic intermediate proposed by Yan et al. (route 1), and biosynthesis of isatropolone C (ITC) focusing on oxidative rearrangement for tropolone-ring construction from bi-cyclic/aromatic intermediate proposed by Cai et al. (route 2). Figure S29. Two compounds 9–10 characterized from S. lividans heterologously expressing istG-R for the aglycone biosynthesis of isatropolone (reported by Cai et al.).
Click here for additional data file.
Author Contributions
S.L. and L.W. conceived the study. J.L., X.L. and J.F. performed culture and fermentation; J.L. performed chemical analysis and compound isolation; J.L. and B.J. performed NMR structure determination. J.L. and L.W. drafted the manuscript; J.L., S.L. and L.W. reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Not applicable.
Conflicts of Interest
The authors declare no conflict of interest.
Figures, Scheme and Table
Figure 1 The structure of 7,12-dihydroisatropolone C (H2ITC, 1ab) and its oxidation to isatropolone C (ITC). (a) 7,12-dihydroisatropolone C and isatropolone C. (b) Keto-enol tautomerization of 1a and 1b in H2ITC, and their spontaneous oxidation to isatropolone C. The HPLC trace showed the pair peaks of H2ITC (1ab) and a small isatropolone C peak.
Figure 2 Key COSY and HMBC correlations for 7,12-dihydroisatropolone C (H2ITC, 1ab).
Figure 3 H2ITC and ITC titers of Streptomyces sp. CPCC 204095.
molecules-27-02882-sch001_Scheme 1 Scheme 1 Biosynthesis of isatropolone C (ITC) focusing on oxidative rearrangement for dihydrotropolone-ring construction from mono-cyclic/aromatic intermediate (route 1) or bi-cyclic/aromatic intermediate (route 2).
molecules-27-02882-t001_Table 1 Table 1 NMR data of 7,12-dihydroisatropolone C (1ab) and isatropolone C in acetone-d6.
7,12-Dihydroisatropolone C (1ab) Isatropolone C
Position δC, Type δH, Mult,
(J in Hz) HMBC δC, Type δH, Mult, (J in Hz)
1 10.20, 10.11, CH3 0.92, t (7.3) C-2, C-3 10.11, CH3 0.95, t
2 28.30, 28.01, CH2 1.35, overlap C-3, C-4 28.18, CH2 1.39, m
3 76.43, 76.34, CH 4.78, m; 4.70, m * (3.5, 7.5) C-1, C-4 76.62, CH 4.84, m *
3-OH
4 199.06, 198.99, C 199.77, C
6 188.12, 187.41, C 190.21, C
7 42.44, 42.33, CH2 2.91, d; 2.74, d (15.1) C-8, C-9, C-12 134.35, CH2 7.12, s
8 184.18, 182.76, C 185.09, C
9 131.68, 131.43, C 132.02, C
10 158.96, 156.82, C 159.47, C
11 142.08, 141.98, C 144.83, C
12 40.48, 40.37, CH 3.65, dd; 3.56, dd (3.0, 15.1) C-7 135.54, CH
13 122.58, 119.85, C 120.98, C
14 162.97, 162.89, C 162.58, C
15 106.95, 106.90, C 7.70, s; 7.68, s C-14, C-16, C-17 107.57, C 7.77, s
16 168.41, 168.39, C 170.24, C
17 20.77, 20.74, CH3 2.58, s C-15, C-16 21.17, CH3 2.64, s
1′ 110.14, 110.08, CH 5.69, s; 5.63, s C-3′, C-5′ 109.89, CH 5.78, s
2′ 81.54, 81.37, C 82.04, C
2′-OH
3′ 69.20, 68.91, CH 4.81, d; 4.62, d * 68.79, CH 5.00, brs *
3′-OH
4′ 80.75, 80.39, CH 3.33, m; 3.32, m C-3′, C-5′, C-7′ 79.96, CH 3.34, brs
5′ 67.93, 66.65, CH 4.24, m; 4.08, m 65.98, CH 4.15, m
6′ 18.31, 18.26, CH3 1.29, d; 1.28, d (6.4) C-4′, C-5′, 18.75, CH3 1.31, d
7′ 57.47, 57.33, CH3 3.37, s C-4′ 57.90, CH3 3.34, s
* Exchangeable.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Cai X. Shi Y.M. Pohlmann N. Revermann O. Bahner I. Pidot S.J. Wesche F. Lackner H. Buchel C. Kaiser M. Structure and Biosynthesis of Isatropolones, Bioactive Amine-Scavenging Fluorescent Natural Products from Streptomyces Gö66 Angew. Chem. Int. Ed. Engl. 2017 56 4945 4949 10.1002/anie.201701223 28371116
2. Guo H. Benndorf R. Konig S. Leichnitz D. Weigel C. Peschel G. Berthel P. Kaiser M. Steinbeck C. Werz O. Expanding the Rubterolone Family: Intrinsic Reactivity and Directed Diversification of PKS-derived Pyrans Chem. Eur. J. 2018 24 11319 11324 10.1002/chem.201802066 29846024
3. Moureu S. Caradec T. Trivelli X. Drobecq H. Beury D. Bouquet P. Caboche S. Desmecht E. Maurier F. Muharram G. Rubrolone production by Dactylosporangium vinaceum: Biosynthesis, modulation and possible biological function Appl. Microbiol. Biotechnol. 2021 105 5541 5551 10.1007/s00253-021-11404-w 34189614
4. Palleroni N.J. Reichelt K.E. Mueller D. Epps R. Tabenkin B. Bull D.N. Schüep W. Berger J. Production of a novel red pigment, rubrolone, by Streptomyces echinoruber sp. nov. I. Taxonomy, fermentation and partial purification J. Antibiot. 1978 31 1218 1225 10.7164/antibiotics.31.1218 738965
5. Yan Y. Ma Y.T. Yang J. Horsman G.P. Luo D. Ji X. Huang S.X. Tropolone Ring Construction in the Biosynthesis of Rubrolone B, a Cationic Tropolone Alkaloid from Endophytic Streptomyces Org. Lett. 2016 18 1254 1257 10.1021/acs.orglett.6b00074 26926531
6. Guo H. Benndorf R. Leichnitz D. Klassen J.L. Vollmers J. Gorls H. Steinacker M. Weigel C. Dahse H.M. Kaster A.K. Isolation, Biosynthesis and Chemical Modifications of Rubterolones A-F: Rare Tropolone Alkaloids from Actinomadura sp. 5-2 Chem. Eur. J. 2017 23 9338 9345 10.1002/chem.201701005 28463423
7. Wang J. Hu X. Sun G. Li L. Jiang B. Li S. Bai L. Liu H. Yu L. Wu L. Genome-Guided Discovery of Pretilactam from Actinosynnema pretiosum ATCC 31565 Molecules 2019 24 2281 10.3390/molecules24122281 31248172
8. Hu X. Hu X. Hu X. Li S. Li L. Yu L. Liu H. You X. Wang Z. Li L. Cytotoxic and Antibacterial Cervinomycins B1-4 from a Streptomyces Species J. Nat. Prod. 2019 82 2337 2342 10.1021/acs.jnatprod.9b00198 31381332
9. Hu X. Wang Y. Zhao C. Li S. Hu X. You X. Shen J. Wang Z. Hong B. Jiang B. Mintaimycins, a Group of Novel Peptide Metabolites from Micromonospora sp. C-3509 Molecules 2022 27 1150 10.3390/molecules27041150 35208940
10. Li L. Li S. Jiang B. Zhang M. Zhang J. Yang B. Li L. Yu L. Liu H. You X. Isarubrolones Containing a Pyridooxazinium Unit from Streptomyces as Autophagy Activators J. Nat. Prod. 2019 82 1149 1154 10.1021/acs.jnatprod.8b00857 31070914
11. Lu T. Chen F. Multiwfn: A multifunctional wavefunction analyzer J. Comput. Chem. 2012 33 580 592 10.1002/jcc.22885 22162017
12. Guo W. Siegel D. Ross D. Stability of the Hsp90 inhibitor 17AAG hydroquinone and prevention of metal-catalyzed oxidation J. Pharm. Sci. 2008 97 5147 5157 10.1002/jps.21394 18393300
13. Ratnayake R. Lacey E. Tennant S. Gill J.H. Capon R.J. Kibdelones: Novel anticancer polyketides from a rare Australian actinomycete Chem. Eur. J. 2007 13 1610 1619 10.1002/chem.200601236 17091523
14. Yan Y. Yang J. Yu Z. Yu M. Ma Y.T. Wang L. Su C. Luo J. Horsman G.P. Huang S.X. Non-enzymatic pyridine ring formation in the biosynthesis of the rubrolone tropolone alkaloids Nat. Commun. 2016 7 13083 10.1038/ncomms13083 27713400
15. Yan Y. Yu Z. Zhong W. Hou X. Tao Q. Cao M. Wang L. Cai X. Rao Y. Huang S.X. Characterization of Multifunctional and Non-stereoselective Oxidoreductase RubE7/IstO, Expanding the Functional Diversity of the Flavoenzyme Superfamily Angew. Chem. Int. Ed. Engl. 2022 61 e202200189 10.1002/anie.202200189 35191152
16. Liu X. Li S. Li L. Liu J. Jiang B. Wu L.Z. Solid state fermentation of Streptomyces sp. CPCC 204095 and preparation of isatropolone C Chin. Med. Biotechnol. 2021 16 302 306 (In Chinese)
|
PMC009xxxxxx/PMC9099903.txt |
==== Front
Sensors (Basel)
Sensors (Basel)
sensors
Sensors (Basel, Switzerland)
1424-8220
MDPI
10.3390/s22093269
sensors-22-03269
Article
Mapping Tree Canopy in Urban Environments Using Point Clouds from Airborne Laser Scanning and Street Level Imagery
https://orcid.org/0000-0002-4844-1759
Rodríguez-Puerta Francisco 1*
Barrera Carlos 2
García Borja 2
https://orcid.org/0000-0001-7825-7824
Pérez-Rodríguez Fernando 2
https://orcid.org/0000-0002-6848-481X
García-Pedrero Angel M. 34
Narayanan Ram M. Academic Editor
1 EiFAB—iuFOR, Campus Duques de Soria s/n, Universidad de Valladolid, 42004 Soria, Spain
2 Föra Forest Technologies sll, Campus Duques de Soria s/n, 42004 Soria, Spain; [email protected] (C.B.); [email protected] (B.G.); [email protected] (F.P.-R.)
3 Department of Computer Architecture and Technology, Universidad Politécnica de Madrid, 28660 Madrid, Spain; [email protected]
4 Center for Biomedical Technology, Universidad Politécnica de Madrid, 28223 Madrid, Spain
* Correspondence: [email protected]
24 4 2022
5 2022
22 9 326911 3 2022
21 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Resilient cities incorporate a social, ecological, and technological systems perspective through their trees, both in urban and peri-urban forests and linear street trees, and help promote and understand the concept of ecosystem resilience. Urban tree inventories usually involve the collection of field data on the location, genus, species, crown shape and volume, diameter, height, and health status of these trees. In this work, we have developed a multi-stage methodology to update urban tree inventories in a fully automatic way, and we have applied it in the city of Pamplona (Spain). We have compared and combined two of the most common data sources for updating urban tree inventories: Airborne Laser Scanning (ALS) point clouds combined with aerial orthophotographs, and street-level imagery from Google Street View (GSV). Depending on the data source, different methodologies were used to identify the trees. In the first stage, the use of individual tree detection techniques in ALS point clouds was compared with the detection of objects (trees) on street level images using computer vision (CV) techniques. In both cases, a high success rate or recall (number of true positive with respect to all detectable trees) was obtained, where between 85.07% and 86.42% of the trees were well-identified, although many false positives (FPs) or trees that did not exist or that had been confused with other objects were always identified. In order to reduce these errors or FPs, a second stage was designed, where FP debugging was performed through two methodologies: (a) based on the automatic checking of all possible trees with street level images, and (b) through a machine learning binary classification model trained with spectral data from orthophotographs. After this second stage, the recall decreased to about 75% (between 71.43 and 78.18 depending on the procedure used) but most of the false positives were eliminated. The results obtained with both data sources were robust and accurate. We can conclude that the results obtained with the different methodologies are very similar, where the main difference resides in the access to the starting information. While the use of street-level images only allows for the detection of trees growing in trafficable streets and is a source of information that is usually paid for, the use of ALS and aerial orthophotographs allows for the location of trees anywhere in the city, including public and private parks and gardens, and in many countries, these data are freely available.
street level imagery
deep learning
LiDAR
urban trees
==== Body
pmc1. Introduction
In September 2015, the United Nations adopted the 2030 Agenda for Sustainable Development, which includes 17 Sustainable Development Goals (SDGs) that are achieved through 169 targets. In this 2030 Agenda, environmental sustainability is a key component that depends on the sustainable management of the earth’s natural resources. On the other hand, it is unclear how the SDG targets relate to urban ecosystems. Maes et al. [1] define what changes in urban ecosystem management are needed and describe how urban ecosystem management can reinforce or undermine action to achieve the 169 Agenda 2030 targets. Resilient cities incorporate a social, ecological, and technological systems perspective through their trees, both in urban and peri-urban forests and linear street trees, and help promote and understand the concept of ecosystem resilience [2,3].
In the public landscape of cities, trees have been used in two main areas. First, they have been used in spaces for public activities, such as recreational areas, pedestrian walkways, and plazas or parks. Secondly, trees have been used as extensions of the private garden, and more often as street trees in front of houses [4]. Street trees are public resources that complement urban forests and provide numerous benefits to people. However, the value of these urban trees to wildlife is not well understood, which is a gap in our knowledge of urban ecosystem conservation [5,6]. One of the most important and most studied effects of urban trees is their ability to sequester carbon and reduce house cooling energy consumption, due to the shade generated by these trees [7,8]. In short, urban trees play an essential role in making our cities more sustainable, livable, and resilient to climate change. To maximize the benefits of urban trees, city managers need to know where these trees are located and how the different species are distributed in our cities [9].
Urban tree inventories usually involve the collection of field data on the location, genus, species, crown shape and volume, diameter, height, and health status of these trees [10]. Nielsen et al. [11] identify four main ways of acquiring and updating urban tree inventories: satellite-based methods, aircraft-supported methods, digital field inventory through photography or laser scanning, and finally, field surveys. On the other hand, the two most current trends for large-area, low-cost urban tree inventories are [12]: first, the use of Convolutional Neural Networks (CNN) for abstract feature and object extraction in imagery [13], and second, the use of increasingly available, low-cost, and detailed street-level imagery [14], such as Google Street View (GSV) imagery. Moreover, Light Detection and Ranging (LiDAR), aerial photography, and multispectral and hyperspectral imaging have become widely used for earth observation and large-scale analysis of forest ecosystems. Such new remote sensing technologies, in conjunction with novel computer vision (CV) algorithms, allow for the semi-automated identification of urban trees and the automatic identification of their main metrics, such as crown width or total height [15,16,17,18,19,20], which can result in being more time-efficient and less costly when compared to field inventory. These methods have already made it possible to analyze forests at different temporal and geographic scales, progressing from the stand level to the plot level and down to the individual tree level [21,22,23]. In that sense, both active and passive remote sensing are robust alternatives for estimating forestry variables and can also be used for urban trees. Optical data is useful for providing spectral information on species and tree condition [16], while active remote sensing technologies, such as Airborne Laser Scanning (ALS), provide very accurate estimation of individual tree height [24], allowing precise canopy height modeling (CHM) and, therefore, individual tree detection (ITD).
Originally, ITD was performed by using photogrammetric point clouds, whereas ALS is now the main technology for the 3D mapping of trees [25]. Hence, numerous methods for individual tree detection developed for optical imagery have been expanded to LiDAR data. Algorithms for ITD can be divided into those using CHM raster data and those using LiDAR point cloud directly [26]. Most of these algorithms for individual tree detection are based on tree canopies representing the highest part of the landscape and therefore find local maxima (LM) of height within the data at individual tree canopies [17,27]. In addition, given some spurious local maxima might be generated by individual tree branches, smoothing filters are often applied to remove them [28]. As a result, the parameterization of LM algorithms is centered on two parameters: a Smoothing Window Size (SWS) and a Tree Window Size (TWS), which defines a fixed boundary within which the algorithm searches for treetops [28]. Following this treetop detection, segmentation is performed to delineate the canopy boundary of individual trees, which is most commonly based on region growth [29,30], watershed delineation [6], or clustering [31,32,33,34].
A data source that has recently received much attention from the urban forest research community due to its low cost and global coverage is general street-level imagery, where the best-known service is GSV [14]. In addition to GSV, other digital platforms have launched street-level panoramic imagery products, such as Apple Look Around (some US and international cities), Microsoft Bing StreetSide (US and some European cities), Baidu Total View and Tencent Street View (Chinese cities), Kakao/Daum Road View and Naver Street View (South Korea), and Yandex (Russia and some Eastern European countries), as well as the corporate crowdsourcing platforms KartaView (formerly OpenStreetCam) and Mapillary (acquired by Facebook) [35]. These street-level images have great potential for researchers as a large repository of panoramic images as a source of urban big data [36]. GSV has been used successfully to assess urban trees on streets and highways [14] and even to assess the state of tree health [37]. GSV is a geospatial platform with extensive worldwide coverage that provides standardized, geocoded street-level imagery in different formats and resolutions at a relatively low cost [9]. GSV Street-level imagery is collected through a panoramic camera, which records single snapshots in time covering a 360-degree range of view, spaced every 15 m, meaning that one tree can be seen in multiple images [38]. GSV data can be accessed online through an official API. In addition to street-level data, another interesting source of data, usually freely available on government geospatial portals, are RGB and near-infrared (NIR) orthoimages. There is a strong correlation between RGB and NIR values of tree pixels and certain parameters such as their leaf area, biomass, or phenotypic activity [39] that is usually addressed through spectral signature. This approach can be further improved by merging NIR and RGB data with other sources of information such as ground-level images or ALS data and applying machine learning (ML) algorithms to them [34].
Computer vision is the field that deals with the development of techniques that allow computers to evaluate and analyze images or videos (sequences of images). The most common tasks in computer vision of images include object detection and object classification [40,41]. Deep learning (DL) is a subset of machine learning based on neural networks of multiples layers that attempt to emulate how the brain perceives and understands multimodal information. The multiple processing layers of DL methods are able to learn and represent data at multiple levels of abstraction, thus capturing the intricate structures of large-scale data [42]. Advances in the combination of computer vision [43] and deep learning are enabling automation in street-level data collection in urban environments [44]. These computer vision-based algorithms have been applied to assess urban change [45], building types [46], and urban morphology [47]. With respect to urban tree inventory, street-level imagery in combination with CV has been successfully applied in three key areas: (1) urban tree mapping [4,38], (2) quantification of perceived urban canopy cover [5,12,48,49,50], and (3) estimation of shade provision [51,52,53]. In this work, we have combined (and compared) three of the most common data sources: we have used ALS information, RGB and NIR orthomosaics, and GSV street-level imagery. This information has been used to solve the main objective of an urban tree inventory, which is to locate all trees accurately and inexpensively, based on remote data and without the need for field work. The developed method is novel because it compares (and also combines) a methodology based on CV on GSV images, with another methodology based on ITD in ALS data. Finally, in both cases a filtering of the results is performed through a ML algorithm trained with RGB and NIR orthomosaics.
2. Methodology
2.1. Study Area and Validation of Results
This work was carried out in the city of Pamplona (Spain). Pamplona is located in northern Spain and is the capital of the region of Navarra (Figure 1). It has a population of 203,944 inhabitants, spread over an area of 25,098 km2. It also has 63,962 trees according to its official tree inventory. This city has been selected because it has the three sources of data contrasted in this study: (i) high density ALS data (14 points per square meter), (ii) different complete coverages of GSV images (from 2009 to the present), (iii) several coverages of RGB and NIR orthophotos (from 2005 to the present), and (iv) a complete collection of thematic cartography of the municipality (Figure 2). Another important reason is that Pamplona has a free access database with all the urban trees, where the trees are geolocated with high precision and where their main attributes (genus, species, health status, etc.) are included. This database substantially reduced the field work to collect “ground truth” in the city. Finally, as Pamplona is a large city, it was decided to reduce the study area to different random circles of 300 m radius in which to evaluate the different methodologies proposed. To randomize these eight points, the QGIS software research tool [54] “Vector -> Research tools -> Random points inside polygons” was used to create eight random points separated by a minimum distance of 300 m so that there would be no intersection between their circles. To create the circles, the “Vector -> Geoprocessing -> Buffer” tool of the QGIS software [54] with a radius of 150 m was used. These circles were classified according to three zone typologies: (i) most of the trees are in streets (suitable for vehicles), (ii) most of the trees are in parks (pedestrian only), and (iii) a mix between trees in streets and in parks. Figure 1 shows the 8 selected circles.
2.2. Remote Data Sources
In this work, we have used three data sources: (i) Street-level images, (ii) ALS LiDAR cloud-point data, and (iii) RGB and NIR digital orthophotos. Regarding street-level images, these can be downloaded from Google Street View (GSV), OpenStreetMap, Bing Maps, or we can obtain them ourselves. These images are georeferenced, and we know their heading, pitch, and field of view (FOV). In our case, we have used Google Street View with its corresponding API. Since not all areas were covered at the same time, we had to use images from different years. In total, we used 77 images from 2015, 37 images from 2017, 1100 images from 2018, and 1888 images from 2019. ALS data came from the National Geographic Institute of Spain (IGN). In this case, data were acquired between September and November 2017, with a LEICA SPL100 sensor, obtaining an average point density of 14 first returns per square meter, and with an XY precision of 20 cm and a Z precision of 15 cm. Finally, the orthophoto coverage was also obtained from the IGN. Two coverages were used, one only in RGB carried out in 2014 and another in RGB and NIR captured in 2017. Both coverages provided a pixel size of 0.25 m, a planimetric accuracy in XY of less than 50 cm, an ETRS89 geodetic reference system, and a TIFF file format with its corresponding georeferencing TFW file. In this way, the three data sources were acquired on reasonably similar dates. Figure 2 shows the data sources used in this study.
2.3. Geolocation of Urban Trees
Regardless of the database used to geolocate all urban trees, we have designed a methodology based on two stages: (1) detect all possible trees, even knowing that there may be many false positives, and (2) debug those false positives from the previous step. After each stage, we always performed a merging of the trees that were too close to each other (distance less than 4 m), thus eliminating artifacts caused by branches or mispositioning of LiDAR and GSV images. This merging was performed based on the methodology proposed by Picos et al. [55]. In the first stage, two methodologies have been contrasted: (1A) individual tree detection (ITD) through computer vision (CV) on Google Street View (GSV) images, and (1B) ITD from LiDAR point clouds (ALS). In the second stage, two other techniques were again used: (2A) false positive debugging through CV using GSV imagery, and (2B) false positive debugging through ML using RGB and NIR orthophotos. Therefore, four combinations (two in each of the first two stages) were performed and compared to evaluate the accuracy and efficiency of the different results. Results were also included using only the first stage where false positives are not filtered out, as is usual in other investigations consulted [56,57,58,59,60]. It is important to note that the GSV-based techniques are only used to analyze transited street trees (circles 1, 2, 3, 6, and 8), while the LiDAR and orthophoto-based techniques are used to analyze all urban trees. The results were compared with the official data on urban trees, which are freely available on the Pamplona City Council website (https://www.pamplona.es/la-ciudad/geopamplona/descargas, last access on 13 March 2022). Figure 3 shows a scheme of the procedure followed.
2.3.1. ITD through Computer Vision, Using GSV Images (Stage 1A)
Each image captured from GSV is associated with an identifier called PanoID. Each PanoID has a 360° panoramic image associated with it. These images are georeferenced, and we know their heading, pitch, and field of view (FOV). In two-way streets, images are available in both directions of the street. Google API [61] was used to download the images. The acquisition of GSV images consisted of 3 steps: (1) detecting the areas where GSV images are available, (2) selecting the most suitable images based on date and image parameters, and (3) downloading the images through the GSV API. To detect the presence of trees in each of the images, a model based on MASK R-CNN convolutional neural network (available in the DETECTRON2 library [62]) was retrained. The model was pre-trained with the public database IMAGENET [52]. From this pre-trained model, the model was improved to be able to distinguish between the tree stem and the tree crown. We performed a fine-tuning of the same model using 2000 images manually segmented with LABELME software [53] from other areas of the city of Pamplona. This model was then applied to all the downloaded images and the possible trees were identified. Through the image parameters and based on trigonometric rules we were able to estimate the position of the tree (azimuth and distance to the point where the image was taken). Each detected tree was assigned a unique Id and geolocated based on the azimuth and distance from the origin of the image. Usually, the same tree is detected in more than one image, so its positioning from one or another point should be the same or very close. These possible duplicate trees (if they were less than 4 m away) were eliminated (Section 2.3.5). This methodology, being based on GSV street-level imagery, was only possible on vehicle-traversable streets. Figure 4 shows a schematic of the process followed.
2.3.2. ITD Using ALS Data (Stage 1B)
Algorithms for ITD can be divided into those that use raster data from the vegetation canopy model (CHM) and those that use the ALS point cloud directly [26]. Most of these algorithms are based on finding local maxima (LM) of height both in the point cloud and in the MDAV [17,27]. Additionally, since individual tree branches can generate some false local maxima, smoothing filters are often applied to remove them [63]. As a result, the parameterization of LM algorithms focuses on two parameters: a smoothing window size (SWS) and a tree window size (TWS), which defines a fixed limit within which the algorithm searches for the tops of the trees [28,64]. Although there are many algorithms to perform ITD, the most common in the forestry field are the packages used from the R software [65], such as FORESTTOOLS [66], LIDR [67] and RLIDAR [68,69], as well as the algorithms integrated in FUSION/LDV [70], such as TREESEG and CANOPYMAXIMA [71]. In this work, a pre-selection was realized and finally the FORESTTOOLS package was chosen, based on a smoothed CHM developed with FUSION [70]. In this way, the LIDAR point cloud data was first downloaded from the IGN website. The smoothed CHM was then generated using FUSION software (CANOPYMODEL procedure, cellsize = 0.25 and smooth = 3). This CHM was clipped with the vector layer of buildings that can be downloaded from the Pamplona City Council website. From this clipped CHM, treetops were found with the R package FORESTTOOLS. Different parameterizations were evaluated, and it was considered that the one that offered the best results was the following: “minHeight = 2; winFun = 0.12 x + 0.5; maxWinDiameter = NULL; minWinNeib = queen”. Each of the treetops had a unique Id and incorporated the height measured over the CHM. Figure 5 shows a schematic of the process followed.
2.3.3. False Positive Debugging through CV Using GSV Imagery (Stage 2A)
In stage 1A, we selected all GSV images every 10–15 m and we applied the object detection algorithm on all of them, identifying in each of them each tree (stem and crown) with a unique Id. In stage 2A, the procedure was the other way around. First, we started from the trees identified in stage 1, then we selected the three closest images to each of the trees and downloaded them through the Google API. For each of the trees, we analyzed whether that tree was detected in the three closest images. Then, we positioned it using trigonometry and checked if all three positions obtained were less than 4 m away from each other. When this happened, we considered it to be a single tree and positioned it at the centroid of the three positions. Figure 6 shows a schematic of the process followed.
2.3.4. False Positive Debugging through ML Using RGB and NIR Orthophotos (Stage 2B)
For each tree detected in stage 1, a buffer of 50 cm radius was generated and the zonal statistics (mean and standard deviation) of each of the bands of the different orthophotos (RGB-2017, RGB-2014, and NIR-2017) were calculated. A ML algorithm was then trained using a ground truth of 15,766 points sampled from orthophotos to determine whether points detected corresponded to the classes TREE or NOT TREE. To reduce the training processing time, a variable selection was performed using the VSURF procedure [72]. The four most used algorithms in ML for this type of training were evaluated [73]; ANN, SVML, SVMR, and RF, executing the NNET, SVMLINEAR, SVMRADIAL, and RF methods using the CARET package in R software [74]. Finally, a cross-validation was performed using three replicates to control for overfitting. As in the previous method, trees that were less than 4 m away from each other were grouped.
2.3.5. Merging Nearby Trees
The method is based on the methodology proposed by Picos et al. [55] to perform ITD in Eucalyptus. This false positive debugging starts by creating a 2D buffer around the detected and projected treetop. The width of the buffer should be above the X-Y point spacing and below the tree spacing. As the spacing between urban trees is usually larger than in the forest, we tested higher distances than Picos et al. [55], starting at 2 m and ending at 5 m, obtaining the best results for a distance of 4 m. This distance of 4 m coincides with the distance threshold selected by Wegner et al. [38] for considering a tree as a TP. As a result, the point cloud was transformed into a polygon cloud. The intersecting treetops were then combined into a single polygon. These centroids approximate the geospatial position of each individual tree.
2.3.6. Accuracy Evaluation
To validate our results, we have calculated the distance from the suggested point with respect to the closest ground truth point applying the following criteria: (i) if the suggested point is less than 5 m away and there is no other suggested point closer to the ground truth point, we considered it a true positive (TP), (ii) if the suggested point is less than 5 m away from the reference point, but there is another TP point closer, or if the suggested point is more than 5 m away, we considered it a false positive (FP), while (iii) if the ground truth point has no suggested point less than 5 m away, we considered it a false negative (FN).
To explore the influence of the different methods used, an evaluation of the performance in terms of relative error rate was carried out to evaluate the precision of the proposed method three statistics were used: recall (r), precision (p), and F1 score. These statistics are widely used to assess the detection error of individual trees [63,75,76]. Recall gives us a measure of trees detected and is inversely related to error of omission, precision implies a measure of trees correctly detected and is inversely related to error of commission, and the F1 score allows us to combine precision and recall in a single value through a modification of its mean. The formulation of these statistics is shown below: (1) r=TP/(TP+FN)
(2) p=TP/(TP+FP)
(3) F1=(2×r×p)/(r+p)
Precision is a great measure when the data are symmetric (similar number of FPs and FNs), and where both errors have the same influence. In our case, the FPs have less influence since in stage 2, debugging, and our goal is to minimize them, so it is better to use F1. On the other hand, recall refers to the number of TPs with respect to all detectable trees (n), and it is the most important statistic when your aim is to have the maximum number of TPs, even if the FPs are also numerous. For this work, we have assessed the trade-off between the three statistics.
3. Results
Table 1 shows the results obtained in each of the combinations of stages. In order to compare methodologies, we focused only on the areas where GSV imagery is available (areas trafficable by vehicles). The results obtained in the first stage are very satisfactory, identifying more than 86% of urban trees. Once the FP debugging is performed (second stage), recall decreases to 78%, given that this process removes some true positives, while eliminating a large number of FPs. Figure 7 shows an example of four of the six methodologies tested in one of the areas of the city. The left column shows the first stage (ITD through computer vision (A) or ALS (C)), while the right column shows the combination of stages (ITD and false positive debugging).
If we focus on recall, a higher value indicates that we have more true positives (regardless of the false positives we found). We have obtained very similar results in the ITD with both ALS and GSV. Regarding FPs in this first stage, we generally found more when using GSV. Therefore, if we are not conditioned by FPs, we always detect more TPs using only the first stage than by combining stages. While the results obtained with both GSV and ALS are similar, they are slightly higher when using ALS.
Considering the second stage to debug FPs, GSV- and ML-based methods are both equally valid. The combination of all stages is somewhat better when starting from an ITD performed with ALS, although differences are not significant. In general, the best result is obtained by performing the first stage with ALS and the second with GSV. Even so, the advantage of the ALS + ML method is that it works for any place where ALS and orthophoto data exist, which allows the inventory to be performed also in public and private parks and gardens and in pedestrian areas.
4. Discussion
The use of street-level imagery through CV techniques has recently been employed for mapping urban trees [4,38]. Berland and Lange [14] used GSV and obtained 93% of recall on urban trees and discovered that it was possible to assess genus, species, location, diameter at breast height, and tree health. Rousselett et al. [37] were capable of identifying trees affected by pine processionary with a 96% success rate. However, these studies and many others were not automated, so they were limited by costly manual effort. In addition, Li et al. [5] estimated a factor to quantify tree shade provision and assessed the percentage of vegetation on streets by measuring the number of green pixels observed in a GSV image. As for Seiferling et al. [48], they quantified urban tree canopy cover using GSV and ML. These methodologies were the origin of the Green Vision Index [5]. Wegner et al. [38] designed a workflow for automatic detection and geolocation of street trees from GSV and Google Maps images, based on the convolutional neural network model Faster R-CNN. They obtain a recall of 0.706 in tree detection, but also perform a Tree species classification with an average recall of 0.79 (varying as a function of the species classified). This study is more complete than ours, since it identifies the species, but it is the most comparable to ours, in terms of methodology and results.
Methodologies based on ALS data for urban tree detection are less abundant but have also been implemented and automated in some major cities. Tanhuanpää et al. [59] were able to detect 88.8% of urban trees using an automated mapping procedure in the city of Helsinki (Finland). They also measured their height, obtaining a Root Mean Squared Error (RMSE) of 1.27 m, and the diameter at breast height (RMSE = 6.9 cm). Holopainen et al. [60] used a non-automated methodology and found that Vehicular LIDAR (VLS) obtained higher recall than ALS (79.22% versus 68.04%, respectively) on a sample of 438 trees located in parks and urban forests of the city of Helsinki (Finland). After automating their methodology, recall dropped significantly for VLS (26.94%) but not so much for ALS (65.53%). Matasci et al. [77] evaluated the urban tree inventory in Vancouver with ALS on a sample of 22,211 trees, obtaining a recall of 76.6%. Furthermore, they estimated their respective heights (RMSE = 2.6 m) and crown diameters (RMSE = 3.85 m) on a subsample of trees. In Munich (Germany), Wu et al. [78] compared VLS and ALS, obtaining a better percentage of detected trees (83.36%) with ALS, compared to VLS (77.2%). Finally, Hanssen et al. [79] performed a comprehensive analysis of urban tree canopy cover in Oslo (Norway) using ALS, obtaining a recall of 73.6%. Finally, although the use of orthophotography is not common in urban tree mapping, some authors report its potential. For instance, Juel et al. [80] combined RGB and NIR orthoimages with data acquired by ALS to train a Random Forest algorithm that was used to map semi-natural coastal vegetation.
In our case, using a fully automated methodology, we located over 86% of the trees (results of the first stage of the methodology, not debugging false positives). After removing false positives during the remaining stages, our recall decreased to 78%, which is comparable to that obtained by other researchers. Moreover, we obtained quite a balanced proportion of TPs, FNs, and FPs, regardless of the data source used during the first stage. We believe that the most important challenge in this study is to achieve a fully automated methodology that allows us to perform an urban tree inventory with the minimum cost of error correction, either through photointerpretation or tree identification at street level. If we included a third phase of photointerpretation in our study, we would certainly be able to clean up almost all the FNs and include almost all the TNs (originally omitted trees).
Furthermore, we have observed that GSV-based methods perform worse in streets where we find different parallel rows of trees, as they cover each other, hence increasing the error rate. This allows for the use of different combinations of methods to perform low-cost and automated urban inventories depending on the available data source (ALS, GSV and/or orthophotos). It should be noted that GSV-based methods only work in areas passable by vehicles, while ALS-based methods work for all areas with a similar error rate, including parks and gardens. In many parts of the world there are open access ALS coverages that can allow for the implementation of this methodology. On the other hand, GSV is available (under license) in almost the whole world.
Something that has not been evaluated in this work is the measurement of tree crown metrics. If we use ALS data during the first stage of our method, we can segment the tree crown boundary through the usual algorithms [6,29,30,31,32,33], while if we use GSV we can perform tree crown measurements applying semantic segmentation and trigonometry [53,81,82,83]. The difference between the two methods is that when relying on ALS data we can perform measurements on the horizontal axis (crown width), whereas when relying on GSV data we can include measurements from both axes (crown width and crown length).
5. Conclusions
This study evaluates, in a cost-effective and accurate way, different methods to detect and geolocate trees in urban environments using several data sources (airborne LiDAR point clouds, street level imagery, and digital orthophotos). In many countries, these sources are freely available, so that all the combinations of methodologies evaluated in this study can be carried out, allowing the inventory of both public and private areas as well as pedestrian areas. Thanks to this inventory, public administrations can have much more precise data on the amount of vegetation in a city, as well as the benefits it generates (e.g., their ability to sequester carbon and reduce house cooling energy consumption, due to the shade generated by these trees).
In this research, we use an innovative approach by prioritizing the detection of the maximum possible number of trees (TPs), even if this also implies a higher number of FPs, since we refine these FPs in a second stage. In these two stages, we combine data and techniques based on object detection, such as the use of street-level imagery, and unsupervised classification data and techniques, which are more common in remote sensing. Although GSV images have been used in this study, any street-level image is valid to replicate the proposed methodology. The same is valid for ALS data, although ALS captured from an aircraft has been used, point clouds captured with unmanned aerial vehicle (UAV) LiDAR systems or even point clouds generated by UAV-photogrammetry integrated with structure from motion (SfM) techniques can be equally valid.
Finally, street-level images can allow us to identify some qualitative (species, genus, health status) and quantitative (height, crown width, etc.) characteristics of trees, and this should be our next challenge, as well as the use of semantic segmentation techniques.
Acknowledgments
Airborne LiDAR data and orthophotos were provided by CNIG-PNOA. The Google Street View images were provided by Google. The database of urban trees in Pamplona was provided by the City Council of Pamplona.
Author Contributions
Conceptualization, F.R.-P. and A.M.G.-P.; methodology, C.B., A.M.G.-P., F.R.-P., and F.P.-R.; validation, C.B. and B.G.; formal analysis, C.B.; data curation, C.B.; writing—original draft preparation, F.R.-P.; writing—review and editing, all authors; funding acquisition, F.R.-P. and F.P.-R. All authors have read and agreed to the published version of the manuscript.
Funding
This research was funded by the Government of Spain, through the Ministry of Economy and Enterprise, grant number TSI-100909-2019-62.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Not applicable.
Conflicts of Interest
The authors declare no conflict of interest.
Figure 1 Location of the methodology analysis circles within the city of Pamplona. Circles 1, 2, 3, 6, and 8 were considered fully suitable for vehicles. The rest of the circles (4, 5 and 7) were either pedestrian areas or were residential areas with private gardens.
Figure 2 Remote data used in this study.
Figure 3 Flowchart of the different stages of the proposed methodology.
Figure 4 Graphical summary of stage 1A to perform individual tree detection based on the use of computer vision on Google Street View images.
Figure 5 Graphical summary of stage 1B to perform individual tree detection based on the use of ForestTools package on LiDAR point cloud.
Figure 6 Graphical summary of stage 1A to perform individual tree detection based on the use of computer vision on Google Street View images.
Figure 7 Example of the methodology in one of the areas of the city. (A) ITD through GSV imagery and (C) ITD through ALS. In this first stage, the white dots indicate the trees from the municipality’s database, while the red dots indicate the trees detected by the methodology. The images on the right show the combination of stages. (B) ITD through GSV and FP debugging through GSV, (D) ITD through ALS and FP debugging through ML. Green dots indicate well-identified trees (TPs) and red dots indicate incorrectly detected trees (FPs).
sensors-22-03269-t001_Table 1 Table 1 Results obtained in the urban tree inventory for the different combinations of stages in the circles (only trafficable by vehicles) of the city of Pamplona (n is the official number of trees, TP is the true positives detected, FP is the false positives detected, FN is the false negatives detected, p is the precision, r is the recall, and F1 is the overall precision). The highest rated combination of methods is identified in bold.
Zone Method n TP FP FN p (%) r (%) F1 (%)
1 GSV 700 635 911 51 41.07 92.57 56.90
ALS 608 655 64 48.14 90.48 62.84
GSV + GSV 581 271 95 68.19 85.95 76.05
GSV + ML 581 304 95 65.65 85.95 74.44
ALS + GSV 542 146 141 78.78 79.36 79.07
ALS + ML 509 51 186 90.89 73.24 81.12
2 GSV 215 112 351 73 24.19 60.54 34.57
ALS 164 205 36 44.44 82.00 57.64
GSV + GSV 91 104 104 46.67 46.67 46.67
GSV + ML 98 120 92 44.95 51.58 48.04
ALS + GSV 102 66 103 60.71 49.76 54.69
ALS + ML 141 68 66 67.46 68.12 67.79
3 GSV 652 461 420 177 52.33 72.26 60.70
ALS 490 367 132 57.18 78.78 66.26
GSV + GSV 409 137 227 74.91 64.31 69.20
GSV + ML 421 142 221 74.78 65.58 69.88
ALS + GSV 421 87 219 82.87 65.78 73.34
ALS + ML 393 55 251 87.72 61.02 71.98
6 GSV 366 306 345 44 47.00 87.43 61.14
ALS 341 314 20 52.06 94.46 67.13
GSV + GSV 282 97 75 74.41 78.99 76.63
GSV + ML 278 102 77 73.16 78.31 75.65
ALS + GSV 308 82 52 78.97 85.56 82.13
ALS + ML 298 76 61 79.68 83.01 81.31
8 GSV 839 759 1228 54 38.20 93.36 54.21
ALS 772 718 50 51.81 93.92 66.78
GSV + GSV 707 358 105 66.38 87.07 75.33
GSV + ML 711 389 98 64.64 87.89 74.49
ALS + GSV 707 270 113 72.36 86.22 78.69
ALS + ML 612 207 217 74.73 73.82 74.27
mean zones GSV 2.772 2.273 3.255 399 41.12 85.07 55.44
ALS 2.375 1.541 252 50.99 86.42 64.13
GSV + GSV 2.070 967 606 68.16 77.35 72.47
GSV + ML 2.089 1.057 583 66.40 78.18 71.81
ALS + GSV 2.080 651 628 76.16 76.81 76.48
ALS + ML 1.953 457 781 81.04 71.43 75.93
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Maes M.J.A. Jones K.E. Toledano M.B. Milligan B. Mapping Synergies and Trade-Offs between Urban Ecosystems and the Sustainable Development Goals Environ. Sci. Policy 2019 93 181 188 10.1016/j.envsci.2018.12.010
2. Hamstead Z.A. Iwaniec D.M. McPhearson T. Berbés-Blázquez M. Cook E.M. Muñoz-Erickson T.A. Resilient Urban Futures Springer International Publishing Cham, Switzerland 2021 978-3-030-63130-7
3. Iwaniec D.M. Grimm N.B. McPhearson T. Berbés-Blázquez M. Cook E.M. Muñoz-Erickson T.A. A Framework for Resilient Urban Futures Resilient Urban Futures Springer International Publishing Cham, Switzerland 2021
4. Branson S. Wegner J.D. Hall D. Lang N. Schindler K. Perona P. From Google Maps to a Fine-Grained Catalog of Street Trees ISPRS J. Photogramm. Remote Sens. 2018 135 13 30 10.1016/j.isprsjprs.2017.11.008
5. Cai B.Y. Li X. Seiferling I. Ratti C. Treepedia 2.0: Applying Deep Learning for Large-Scale Quantification of Urban Tree Cover Proceedings of the 2018 IEEE International Congress on Big Data, Big Data Congress 2018 San Francisco, CA, USA 2–7 July 2018
6. Chen Q. Baldocchi D. Gong P. Kelly M. Isolating Individual Trees in a Savanna Woodland Using Small Footprint Lidar Data Photogramm. Eng. Remote Sens. 2006 72 923 932 10.14358/PERS.72.8.923
7. McHale M.R. Burke I.C. Lefsky M.A. Peper P.J. McPherson E.G. Urban Forest Biomass Estimates: Is It Important to Use Allometric Relationships Developed Specifically for Urban Trees? Urban Ecosyst. 2009 12 95 113 10.1007/s11252-009-0081-3
8. Pearlmutter D. Calfapietra C. Samson R. O’Brien L. Krajter Ostoić S. Sanesi G. Alonso del Amo R. The Urban Forest: Cultivating Green Infrastructure for People and the Environment Springer International Publishing Cham, Switzerland 2017 Volume 7 978-3-319-50279-3
9. Lumnitz S. Devisscher T. Mayaud J.R. Radic V. Coops N.C. Griess V.C. Mapping Trees along Urban Street Networks with Deep Learning and Street-Level Imagery ISPRS J. Photogramm. Remote Sens. 2021 175 144 157 10.1016/j.isprsjprs.2021.01.016
10. Keller J.K.K. Konijnendijk C.C. Van Den Bosch C. Keller and Konijnendijk: A Comparative Analysis of Municipal Urban Tree Inventories Short Communication: A Comparative Analysis of Municipal Urban Tree Inventories of Selected Major Cities in North America and Europe Arboric. Urban For. 2012 38 24 30 10.48044/jauf.2012.005
11. Nielsen A.B. Östberg J. Delshammar T. Review of Urban Tree Inventory Methods Used to Collect Data at Single-Tree Level Arboric. Urban For. 2014 40 96 111 10.48044/jauf.2014.011
12. Stubbings P. Peskett J. Rowe F. Arribas-Bel D. A Hierarchical Urban Forest Index Using Street-Level Imagery and Deep Learning Remote Sens. 2019 11 1395 10.3390/rs11121395
13. Ma L. Liu Y. Zhang X. Ye Y. Yin G. Johnson B.A. Deep Learning in Remote Sensing Applications: A Meta-Analysis and Review ISPRS J. Photogramm. Remote Sens. 2019 152 166 177 10.1016/j.isprsjprs.2019.04.015
14. Berland A. Lange D.A. Google Street View Shows Promise for Virtual Street Tree Surveys Urban For. Urban Green. 2017 21 11 15 10.1016/j.ufug.2016.11.006
15. Brandtberg T. Walter F. Automated Delineation of Individual Tree Crowns in High Spatial Resolution Aerial Images by Multiple-Scale Analysis Mach. Vis. Appl. 1998 11 64 73 10.1007/s001380050091
16. Wang L. Gong P. Biging G.S. Individual Tree-Crown Delineation and Treetop Detection in High-Spatial-Resolution Aerial Imagery Photogramm. Eng. Remote Sens. 2004 70 351 357 10.14358/PERS.70.3.351
17. Ke Y. Quackenbush L.J. A Review of Methods for Automatic Individual Tree-Crown Detection and Delineation from Passive Remote Sensing Int. J. Remote Sens. 2011 32 4725 4747 10.1080/01431161.2010.494184
18. Ferraz A. Saatchi S. Mallet C. Meyer V. Lidar Detection of Individual Tree Size in Tropical Forests Remote Sens. Environ. 2016 183 318 333 10.1016/j.rse.2016.05.028
19. Gomes M.F. Maillard P. Deng H. Individual Tree Crown Detection in Sub-Meter Satellite Imagery Using Marked Point Processes and a Geometrical-Optical Model Remote Sens. Environ. 2018 211 184 195 10.1016/j.rse.2018.04.002
20. Kansanen K. Vauhkonen J. Lähivaara T. Seppänen A. Maltamo M. Mehtätalo L. Estimating Forest Stand Density and Structure Using Bayesian Individual Tree Detection, Stochastic Geometry, and Distribution Matching ISPRS J. Photogramm. Remote Sens. 2019 152 66 78 10.1016/j.isprsjprs.2019.04.007
21. Hansen E. Gobakken T. Bollandsås O. Zahabu E. Næsset E. Modeling Aboveground Biomass in Dense Tropical Submontane Rainforest Using Airborne Laser Scanner Data Remote Sens. 2015 7 788 807 10.3390/rs70100788
22. Yu X. Hyyppä J. Karjalainen M. Nurminen K. Karila K. Vastaranta M. Kankare V. Kaartinen H. Holopainen M. Honkavaara E. Comparison of Laser and Stereo Optical, SAR and InSAR Point Clouds from Air- and Space-Borne Sources in the Retrieval of Forest Inventory Attributes Remote Sens. 2015 7 15933 15954 10.3390/rs71215809
23. Zhang Z. Papaik M.J. Wang X. Hao Z. Ye J. Lin F. Yuan Z. The Effect of Tree Size, Neighborhood Competition and Environment on Tree Growth in an Old-Growth Temperate Forest J. Plant. Ecol. 2016 10 970 980 10.1093/jpe/rtw126
24. Wulder M. Niemann K.O. Goodenough D.G. Local Maximum Filtering for the Extraction of Tree Locations and Basal Area from High Spatial Resolution Imagery Remote Sens. Environ. 2000 73 103 114 10.1016/S0034-4257(00)00101-2
25. Zaforemska A. Xiao W. Gaulton R. Individual Tree Detection from Uav LiDAR Data in a Mixed Species Woodland Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2019 XLII-2/W13 657 663 10.5194/isprs-archives-XLII-2-W13-657-2019
26. Valbuena-Rabadán M.-Á. Santamaría-Peña J. Sanz-Adán F. Estimation of Diameter and Height of Individual Trees for Pinus Sylvestris, L. Based on the Individualising of Crowns Using Airborne LiDAR and the National Forestry Inventory Data For. Syst. 2016 25 e046 10.5424/fs/2016251-05790
27. Wallace L. Lucieer A. Watson C.S. Evaluating Tree Detection and Segmentation Routines on Very High Resolution UAV LiDAR Data IEEE Trans. Geosci. Remote Sens. 2014 52 7619 7628 10.1109/TGRS.2014.2315649
28. Mohan M. Silva C. Klauberg C. Jat P. Catts G. Cardil A. Hudak A. Dia M. Individual Tree Detection from Unmanned Aerial Vehicle (UAV) Derived Canopy Height Model in an Open Canopy Mixed Conifer Forest Forests 2017 8 340 10.3390/f8090340
29. Solberg S. Naesset E. Bollandsas O.M. Single Tree Segmentation Using Airborne Laser Scanner Data in a Structurally Heterogeneous Spruce Forest Photogramm. Eng. Remote Sens. 2006 72 1369 1378 10.14358/PERS.72.12.1369
30. Dalponte M. Coomes D.A. Tree-centric Mapping of Forest Carbon Density from Airborne Laser Scanning and Hyperspectral Data Methods Ecol. Evol. 2016 7 1236 1245 10.1111/2041-210X.12575 28008347
31. Gupta S. Weinacker H. Koch B. Comparative Analysis of Clustering-Based Approaches for 3-D Single Tree Detection Using Airborne Fullwave Lidar Data Remote Sens. 2010 2 968 989 10.3390/rs2040968
32. Ferraz A. Bretar F. Jacquemoud S. Gonçalves G. Pereira L. Tomé M. Soares P. 3-D Mapping of a Multi-Layered Mediterranean Forest Using ALS Data Remote Sens. Environ. 2012 121 210 223 10.1016/j.rse.2012.01.020
33. Lindberg E. Eysn L. Hollaus M. Holmgren J. Pfeifer N. Delineation of Tree Crowns and Tree Species Classification from Full-Waveform Airborne Laser Scanning Data Using 3-D Ellipsoidal Clustering IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2014 7 3174 3181 10.1109/JSTARS.2014.2331276
34. Xiao W. Xu S. Elberink S.O. Vosselman G. Individual Tree Crown Modeling and Change Detection from Airborne Lidar Data IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2016 9 3467 3477 10.1109/JSTARS.2016.2541780
35. Cinnamon J. Jahiu L. Panoramic Street-Level Imagery in Data-Driven Urban Research: A Comprehensive Global Review of Applications, Techniques, and Practical Considerations ISPRS Int. J. Geo-Inf. 2021 10 471 10.3390/ijgi10070471
36. Shapiro A. Street-Level: Google Street View’s Abstraction by Datafication New Media Soc. 2018 20 1201 1219 10.1177/1461444816687293
37. Rousselet J. Imbert C.-E. Dekri A. Garcia J. Goussard F. Vincent B. Denux O. Robinet C. Dorkeld F. Roques A. Assessing Species Distribution Using Google Street View: A Pilot Study with the Pine Processionary Moth PLoS ONE 2013 8 e74918 24130675
38. Wegner J.D. Branson S. Hall D. Schindler K. Perona P. Cataloging Public Objects Using Aerial and Street-Level Images-Urban Trees Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Las Vegas, NV, USA 27–30 June 2016 6014 6023
39. Myneni R.B. Hall F.G. Sellers P.J. Marshak A.L. Interpretation of Spectral Vegetation Indexes IEEE Trans. Geosci. Remote Sens. 1995 33 481 486 10.1109/TGRS.1995.8746029
40. Forsyth D. Ponce J. Computer Vision: A Modern Approach Prentice Hall Hoboken, NJ, USA 2011 013608592X
41. Gonzalez R.C. Woods R.E. Digital Image Processing 2nd ed. Addison-Wesley Pub Boston, MA, USA 2002 Volume 455
42. Voulodimos A. Doulamis N. Doulamis A. Protopapadakis E. Deep Learning for Computer Vision: A Brief Review Comput. Intell. Neurosci. 2018 2018 7068349 10.1155/2018/7068349 29487619
43. Lin T.-Y. Maire M. Belongie S. Bourdev L. Girshick R. Hays J. Perona P. Ramanan D. Zitnick C.L. Dollár P. Microsoft COCO: Common Objects in Context arXiv 2015 1405.0312
44. He K. Gkioxari G. Dollár P. Girshick R. Mask R-Cnn arXiv 2017 1703.06870
45. Naik N. Kominers S.D. Raskar R. Glaeser E.L. Hidalgo C. Do People Shape Cities, or Do Cities Shape People? The Co-Evolution of Physical, Social, and Economic Change in Five Major U.S. Cities SSRN Electron. J. 2015 10.2139/ssrn.2698292
46. Kang J. Körner M. Wang Y. Taubenböck H. Zhu X.X. Building Instance Classification Using Street View Images ISPRS J. Photogramm. Remote Sens. 2018 145 44 59 10.1016/j.isprsjprs.2018.02.006
47. Middel A. Lukasczyk J. Zakrzewski S. Arnold M. Maciejewski R. Urban Form and Composition of Street Canyons: A Human-Centric Big Data and Deep Learning Approach Landsc. Urban Plan. 2019 183 122 132 10.1016/j.landurbplan.2018.12.001
48. Seiferling I. Naik N. Ratti C. Proulx R. Green Streets—Quantifying and Mapping Urban Trees with Street-Level Imagery and Computer Vision Landsc. Urban Plan. 2017 165 93 101 10.1016/j.landurbplan.2017.05.010
49. Li X. Zhang C. Li W. Ricard R. Meng Q. Zhang W. Assessing Street-Level Urban Greenery Using Google Street View and a Modified Green View Index Urban For. Urban Green. 2015 14 675 685 10.1016/j.ufug.2015.06.006
50. Duarte F. Ratti C. What Big Data Tell Us about Trees and the Sky in the Cities Humanizing Digital Reality Springer Berlin/Heidelberg, Germany 2018 59 62
51. Li X. Ratti C. Using Google Street View for Street-Level Urban Form Analysis, a Case Study in Cambridge, Massachusetts Modeling and Simulation in Science, Engineering and Technology Springer Berlin/Heidelberg, Germany 2019
52. Li X. Ratti C. Seiferling I. Mapping Urban Landscapes along Streets Using Google Street View Lecture Notes in Geoinformation and Cartography Springer Berlin/Heidelberg, Germany 2017
53. Li X. Ratti C. Seiferling I. Quantifying the Shade Provision of Street Trees in Urban Landscape: A Case Study in Boston, USA, Using Google Street View Landsc. Urban Plan. 2018 169 81 91 10.1016/j.landurbplan.2017.08.011
54. Graser A. Learning Qgis Packt Publishing Ltd. Birmingham, UK 2016 1785888153
55. Picos J. Bastos G. Míguez D. Alonso L. Armesto J. Individual Tree Detection in a Eucalyptus Plantation Using Unmanned Aerial Vehicle (UAV)-LiDAR Remote Sens. 2020 12 885 10.3390/rs12050885
56. Zhang C. Zhou Y. Qiu F. Individual Tree Segmentation from LiDAR Point Clouds for Urban Forest Inventory Remote Sens. 2015 7 7892 7913 10.3390/rs70607892
57. Timilsina S. Aryal J. Kirkpatrick J.B. Mapping Urban Tree Cover Changes Using Object-Based Convolution Neural Network (OB-CNN) Remote Sens. 2020 12 3017 10.3390/rs12183017
58. Hanssen F. Barton D.N. Venter Z.S. Nowell M.S. Cimburova Z. Utilizing LiDAR Data to Map Tree Canopy for Urban Ecosystem Extent and Condition Accounts in Oslo Ecol. Indic. 2021 130 108007 10.1016/j.ecolind.2021.108007
59. Tanhuanpää T. Vastaranta M. Kankare V. Holopainen M. Hyyppä J. Hyyppä H. Alho P. Raisio J. Mapping of Urban Roadside Trees—A Case Study in the Tree Register Update Process in Helsinki City Urban For. Urban Green. 2014 13 562 570 10.1016/j.ufug.2014.03.005
60. Holopainen M. Vastaranta M. Kankare V. Hyyppä H. Vaaja M. Hyyppä J. Liang X. Litkey P. Yu X. Kaartinen H. The Use of ALS, TLS and VLS Measurements in Mapping and Monitoring Urban Trees Proceedings of the 2011 Joint Urban Remote Sensing Event Munich, Germany 10–13 April 2011 29 32
61. Google Google Street View Image API Google Mountain View, CA, USA 2012
62. Wu Y. Kirillov A. Massa F. Lo W.-Y. Girshick R. Wu U. Kirillov A. Massa F. Wan-YenLo Girshick R. Detectron2 Available online: https://github.com/facebookresearch/detectron2 (accessed on 18 January 2022)
63. Mohan M. Mendonça B.A.F. de Silva C.A. Klauberg C. de Saboya Ribeiro A.S. Araújo E.J.G. de Monte M.A. Cardil A. Optimizing Individual Tree Detection Accuracy and Measuring Forest Uniformity in Coconut (Cocos nucifera L.) Plantations Using Airborne Laser Scanning Ecol. Model. 2019 409 108736 10.1016/j.ecolmodel.2019.108736
64. Mohan M. Leite R.V. Broadbent E.N. Jaafar W.S.W.M. Srinivasan S. Bajaj S. Corte A.P.D. Amaral C.H. do Gopan G. Saad S.N.M. Individual Tree Detection Using UAV-Lidar and UAV-SfM Data: A Tutorial for Beginners Open Geosci. 2021 13 1028 1039 10.1515/geo-2020-0290
65. Hornik K. The Comprehensive R Archive Network Wiley Interdiscip. Rev. Comput. Stat. 2012 4 394 398 10.1002/wics.1212
66. Plowright A. R Package ‘ForestTools’ CRAN Available online: https://github.com/andrew-plowright/ForestTools (accessed on 18 January 2022)
67. Roussel J.-R. Auty D. Coops N.C. Tompalski P. Goodbody T.R.H. Meador A.S. Bourdon J.-F. de Boissieu F. Achim A. LidR: An R Package for Analysis of Airborne Laser Scanning (ALS) Data Remote Sens. Environ. 2020 251 112061 10.1016/j.rse.2020.112061
68. Silva C.A. Crookston N.L. Hudak A.T. Vierling L.A. Klauberg C. Cardil A. RLiDAR: LiDAR Data Processing and Visualization, R Package Version 0.1.5 2017 Available online: https://cran.r-project.org/web/packages/rLiDAR/rLiDAR.pdf (accessed on 10 March 2022)
69. Silva C.A. Klauberg C. Mohan M.M. Bright B.C. LiDAR Analysis in R and RLiDAR for Forestry Applications NR 404/504 Lidar Remote Sens. Environ. Monit. Universyty of Idaho 2018 1 90 Available online: https://www.researchgate.net/profile/Carlos-Silva-109/publication/324437694_LiDAR_Analysis_in_R_and_rLiDAR_for_Forestry_Applications/links/5acd932faca2723a333fc1b2/LiDAR-Analysis-in-R-and-rLiDAR-for-Forestry-Applications.pdf?origin=publication_detail (accessed on 10 March 2022)
70. McGaughey R.J. FUSION/LDV: Software for LiDAR Data Analysis and Visualization Version 3.01; US Department of Agriculture, Forest Service, Pacific Northwest Research Station University of Washington Seattle, WA, USA 2012
71. Popescu S.C. Wynne R.H. Nelson R.F. Estimating Plot-Level Tree Heights with Lidar: Local Filtering with a Canopy-Height Based Variable Window Size Comput. Electron. Agric. 2003 37 71 95 10.1016/S0168-1699(02)00121-7
72. Genuer R. Poggi J.-M. Tuleau-Malot C. Vsurf: An R Package for Variable Selection Using Random Forests R J. 2015 7 19 33 10.32614/RJ-2015-018
73. Rodríguez-Puerta F. Ponce R.A. Pérez-Rodríguez F. Águeda B. Martín-García S. Martínez-Rodrigo R. Lizarralde I. Comparison of Machine Learning Algorithms for Wildland-Urban Interface Fuelbreak Planning Integrating Als and Uav-Borne Lidar Data and Multispectral Images Drones 2020 4 21 10.3390/drones4020021
74. Kuhn M. Building Predictive Models in R Using the Caret Package J. Stat. Softw 2008 28 1 26 10.18637/jss.v028.i05 27774042
75. Goutte C. Gaussier E. A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation Losada D.E. Fernández-Luna J.M. Springer Berlin/Heidelberg, Germany 2005 345 359
76. Sokolova M. Japkowicz N. Szpakowicz S. Beyond Accuracy, F-Score and ROC: A Family of Discriminant Measures for Performance Evaluation Sattar A. Kang B. Springer Berlin/Heidelberg, Germany 2006 1015 1021
77. Matasci G. Coops N.C. Williams D.A.R. Page N. Mapping Tree Canopies in Urban Environments Using Airborne Laser Scanning (ALS): A Vancouver Case Study For. Ecosyst. 2018 5 31 10.1186/s40663-018-0146-y
78. Wu J. Yao W. Polewski P. Mapping Individual Tree Species and Vitality along Urban Road Corridors with LiDAR and Imaging Sensors: Point Density versus View Perspective Remote Sens. 2018 10 1403 10.3390/rs10091403
79. Hanssen F. Barton D.N. Nowell M. Cimburova Z. Mapping Urban Tree Canopy Cover Using Airborne Laser Scanning. Applications to Urban Ecosystem Accounting for Oslo Norsk Institutt for Naturforskning (NINA) Trondheim, Norway 2019
80. Juel A. Groom G.B. Svenning J.C. Ejrnæs R. Spatial Application of Random Forest Models for Fine-Scale Coastal Vegetation Classification Using Object Based Analysis of Aerial Orthophoto and DEM Data Int. J. Appl. Earth Obs. Geoinf. 2015 42 106 114 10.1016/j.jag.2015.05.008
81. Yin L. Wang Z. Measuring Visual Enclosure for Street Walkability: Using Machine Learning Algorithms and Google Street View Imagery Appl. Geogr. 2016 76 147 153 10.1016/j.apgeog.2016.09.024
82. Wang W. Xiao L. Zhang J. Yang Y. Tian P. Wang H. He X. Potential of Internet Street-View Images for Measuring Tree Sizes in Roadside Forests Urban For. Urban Green. 2018 35 211 220 10.1016/j.ufug.2018.09.008
83. Keralis J.M. Javanmardi M. Khanna S. Dwivedi P. Huang D. Tasdizen T. Nguyen Q.C. Health and the Built Environment in United States Cities: Measuring Associations Using Google Street View-Derived Indicators of the Built Environment BMC Public Health 2020 20 215 10.1186/s12889-020-8300-1 32050938
|
PMC009xxxxxx/PMC9099904.txt |
==== Front
Int J Mol Sci
Int J Mol Sci
ijms
International Journal of Molecular Sciences
1422-0067
MDPI
10.3390/ijms23094755
ijms-23-04755
Review
On the Base Composition of Transposable Elements
https://orcid.org/0000-0002-8760-1284
Boissinot Stéphane
Symonova Radka Academic Editor
Center for Genomics and Systems Biology, New York University Abu Dhabi, Saadiyat Island, Abu Dhabi P.O. Box 129188, United Arab Emirates; [email protected]
26 4 2022
5 2022
23 9 475524 2 2022
23 4 2022
© 2022 by the author.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Transposable elements exhibit a base composition that is often different from the genomic average and from hosts’ genes. The most common compositional bias is towards Adenosine and Thymine, although this bias is not universal, and elements with drastically different base composition can coexist within the same genome. The AT-richness of transposable elements is apparently maladaptive because it results in poor transcription and sub-optimal translation of proteins encoded by the elements. The cause(s) of this unusual base composition remain unclear and have yet to be investigated. Here, I review what is known about the nucleotide content of transposable elements and how this content can affect the genome of their host as well as their own replication. The compositional bias of transposable elements could result from several non-exclusive processes including horizontal transfer, mutational bias, and selection. It appears that mutation alone cannot explain the high AT-content of transposons and that selection plays a major role in the evolution of the compositional bias. The reason why selection would favor a maladaptive nucleotide content remains however unexplained and is an area of investigation that clearly deserves attention.
transposable elements
GC content
base composition
codon bias
==== Body
pmc1. Introduction
The base composition is one of the most fundamental properties of a genome or of a DNA sequence. Although all DNA sequences consist of 4 nucleotides, the relative proportion of Guanine/Cytosine (GC%) and Adenosine/Thymine (AT%) can differ considerably among organisms and among genomic regions. For instance, mammalian and avian genomes are highly heterogenous in base content, with gene-rich GC-rich compartments embedded in AT-rich intergenic regions, while reptiles, amphibians and fish are generally homogenous in base composition [1,2]. At a smaller scale, the GC% may vary among genes, among regions of a gene but also among codon positions. These differences in base composition can in turn affect a number of fundamental biological processes including transcription efficacy [3,4], the secondary structure of RNA molecules, translation efficacy and accuracy [5,6,7,8], the amino acid composition of proteins, and epigenetic modifications of DNA.
Transposable elements (TEs) are major components of genomes and have a profound impact on the size, structure, and function of their hosts’ genomes (Reviewed in [9]). Although most TE insertions are neutral or deleterious, TEs can also be a source of new genes or of regulatory motifs [9,10,11,12]. An aspect that has received little attention is the impact TEs can have on their host’s genome in terms of base composition. In many organisms, the base composition of TEs differ drastically from the genomic average and from hosts’ genes [13,14,15,16,17], to the point that the unusual base composition of TEs can be used to detect them in genomes [18]. This compositional bias, which is most commonly an AT-bias, may thus impact the structure and function of the genome in a number of ways. For instance, the accumulation of a type of TEs in specific genomic regions can potentially affect the GC genomic landscape, which in turn can affect other biological properties such as chromatin structure. Another interesting aspect is the effect the base composition of TEs can have on their own replication. In a number of organisms, the high AT% exhibited by TEs results in poor transcription and sub-optimal translation of TE-encoded proteins and thus seems maladaptive. Nevertheless, the AT-richness of TEs is widespread, and the persistence of such an unusual base composition across many categories of TEs remains a puzzle.
Here, I will review the state of our knowledge on the evolution of base composition in TEs, as well as the numerous questions that remain unanswered on this topic. After a short introduction on the biology of TEs, I will review what is known about their base composition, and in particular, I will emphasize that the high AT content observed for many TEs in many organisms is, in fact, not universal. I will then describe the consequences the unusual base composition of TEs may have on their hosts but also on their own replication. I will finally explore the evolutionary processes that are potentially driving the base composition of TEs towards nucleotide contents that appear, at first, maladaptive.
2. A Primer on Transposable Elements
Transposable elements constitute a diverse group of sequences that have in common the ability to move from one location in the genome of their host to another location [19,20]. They are typically classified based on their mode of mobility [21,22]. Elements that move using an RNA intermediate are called class I elements, and those that do not are called class II (Figure 1). Each of these classes contains a myriad of subsets. Class I elements are further divided into LTR-retrotransposons, which are flanked by Long Terminal Repeats (LTRs) and include LTR-retrotransposons sensus stricto, endogenous retroviruses and DIRS elements, non-LTR retrotransposons (also called Long Interspersed Nuclear Elements or LINEs), and Penelope Elements. All these elements have in common the use of a reverse-transcriptase for their replication [23]. The replicative machinery of class I elements can also act on other transcripts and is responsible for the amplification of non-autonomous retroelements (such as Short INterspersed Elements or SINEs), which can far outnumber their autonomous progenitors [24,25,26]. Class II elements constitute a very disparate group of elements [27], which only have in common the fact that their replication does not require an RNA intermediate. They consist of four subgroups: DDE transposons that mobilized by a cut-and-paste mechanism mediated by a transposase, Cryptons that use a tyrosine recombinase for their transposition, Helitrons that use a rolling-circle mode of replication [28,29], and Mavericks that are mobilized by a self-synthetizing process mediated by a protein-primed polymerase B [30,31]. Class II elements can also mediate the transposition of non-autonomous copies, which can outnumber autonomous copies [32,33,34].
Class I and class II elements also differ by their long-term evolutionary dynamics within their host. Most LINEs in vertebrates are transmitted vertically over extended periods of evolutionary time and are thus long-term residents of these genomes. For instance, L1 retrotransposons have persisted in the genome of mammals since the origin of this vertebrate class, and mammalian genomes contain a near complete record of the successive waves of L1 amplification they have experienced since their origin [35,36]. The investigation of L1 in mammals revealed that a very small number of lineages, often only one, persisted over long periods of time [35,37,38,39,40], which could reflect an arms race between L1 and the repression machinery of the host [41,42]. In contrast, class II elements tend to invade the genome of their hosts by horizontal transfer, then amplify to large number but eventually get extinct [43,44,45]. Consequently, they rarely persist for long periods of time and are typically transient residents of genomes.
The number of TE copies and the diversity of elements differ considerably among genomes and depends on a number of parameters including the rate of transposition, the rate of fixation and the rate of DNA loss caused by deletions [46]. The rate of transposition depends on the number of progenitor copies and on the location of these progenitors in the genome (transcriptionally active vs. inactive genomic regions). The rate of transposition will also be affected by host-encoded repression processes. Since TE activity can be deleterious, a number of defense mechanisms have evolved to protect the integrity of the genome [42,47,48], DNA methylation being the best-known mechanism of defense against TE activity [49,50]. The rate of fixation will depend on the combined effect of purifying selection and genetic drift (reviewed in [46]) as well as linked selection [51]. Since the majority of new insertions is either deleterious or neutral, most of them are not expected to remain in the population and to be lost by chance (in the case of a neutral insertion) or to be eliminated by purifying selection (if the insertion is deleterious) [52,53,54,55]. However, in small populations, genetic drift can counteract the effect of selection and deleterious insertions can reach fixation [51,56,57,58,59]. Thus, one can expect that the overall rate of fixation and the accumulation of new copies will be higher in small populations than in large populations [60]. Finally, the number of TE derived sequence in a genome will depend on the rate of decay of these elements resulting from the rate of DNA loss by large deletions, which was shown to differ among organisms [33,61,62] and may be correlated to the number of copies (i.e., the accordion model of evolution) [63].
3. Variation in the Base Composition of Transposable Elements
Early analyses of base composition in transposable elements were focused on the composition of the ORFs, in the context of codon bias. Multiple codons encode for the same amino acid, yet there is often a bias in the use of synonymous codons (i.e., codons that encode for the same amino acid), where some codons are preferred over others. This is a common phenomenon in eukaryotes that is referred to as “codon bias” and has been the subject of extensive attention by evolutionary biologists [64]. Codon bias may result from selection in favor of codons that are optimal in terms of translation accuracy [5,8] or efficiency, which is supported by the observation that strongly expressed genes exhibit a stronger codon bias than weakly expressed genes [6,7,65,66]. Because of this relation between codon bias and expression, it has been proposed that the analysis of codon bias in TEs could inform on the nature of the interactions between TEs and their hosts [14]. Early studies in Drosophila found that the codon usage of TEs differed from the codon usage of the host, with a bias in favor of codons ending in A or T [16]. Further studies in model organisms (Arabidopsis thaliana, Caenorhabditis elegans, Saccharomyces cerevisiae, Drosophila melanogaster and Homo sapiens) revealed a general AT-richness of TE’s ORFs for both class I and class II elements compared with host genes [14] and a codon bias in TEs in favor of AT-ending codons, independently of the host. In most species, A-ending codons are preferred, except in the plants A. thaliana and Oryza sativa, where codons ending in T are preferred [17]. In general, the codon usage of TEs is different from the codon usage of host’s genes but tends to be similar to that of weakly expressed genes, at least in some species [14,17]. These observations suggest two things. First, a general mechanism, common to all TEs and independent of the host, may be responsible for the AT-richness of TE’s ORFs. Second, there is no tendency in TEs for codon optimization that would enhance the translation efficiency of the proteins they encode.
Although the general trend of an AT-richness and an AT-preference at the third position of codons seems to hold true for most TEs, more detailed analyses of a larger diversity of elements and the analysis of TEs in non-model organisms suggest a more nuanced and complex picture [15,67,68]. The analysis of base composition of non-LTR retrotransposons in vertebrates revealed large differences among clades of non-TR retrotransposons within the same genomes as well as large differences for the same clade among organisms [15]. For instance, in the lizard Anolis carolinensis, ORF2 (i.e., the ORF encoding for the reverse transcriptase) of elements belonging to the L1 clade are enriched in AT (~67%) relative to host genes (~52% AT) while L2 elements are GC-rich (~55% GC). In fish, L1 and L2 elements are AT-rich (~64% and ~58% AT, respectively) while elements of the Rex1 clade (~52% GC) have a base composition close to the one of hosts genes (~50% AT). Although L1 elements are universally AT-rich, there is a strong A bias on the positive strand in mammals and lizard (~41% and 43% A, respectively), a smaller bias in frogs (~33% A), and no bias in fish where A and T are equally represented. Interestingly, the base composition of the different clades is evolutionarily conserved and has persisted over long periods of evolutionary time within the same genome [15]. At the codon level, AT-rich codons are typically favored but there is no significant synonymous bias since the base frequency at the third position of codons fits the expectations given the overall nucleotide content of the sequences [15]. However, the codon usage of TEs tends to be closer to the codon usage of the host than expected given their base composition, which suggests a certain level of codon adaptation. Although TEs can be classified in AT-rich or GC-rich elements, some TEs show a highly unusual base composition. Such is the case of L2 in the frog Xenopus tropicalis which is enriched in C (34%) and T (30%) on the positive strand [15]. These observations are not limited to vertebrates and similarly large differences in the GC% among class I elements were detected in the insect Anopheles gambiae [69].
Variation in base composition is not limited to class I elements. In a recent survey of GC content of TEs in fish, large variation in the base composition of class II elements was reported [68]. For example, class II elements in zebrafish are 36.9% GC while they are 44.1% GC in the pufferfish Takifugu rubripes. Interestingly, this study identified a positive correlation between the genomic GC content and the TE GC content suggesting an effect of the overall genomic environment on the base composition of TEs.
In some unicellular organisms, the pattern of base composition is drastically different. For instance, in the choanoflagellate Salpingoeca rosetta [70], all TEs exhibit a preference for GC-ending codons and for translationally optimal codons, thus suggesting selection for translational efficiency. Similarly, in the stramenopile genus Phytophthora [71], LTR retrotransposons show preference for GC-ending codons that mirrors host genes. Although additional analyses of unicellular eukaryotes will be necessary, this observation suggests some differences between unicellular and multicellular organisms, perhaps related to different effective population size between these categories [60,70].
Different regions of TEs can also differ considerably in base composition and even ORFs from the same elements can exhibit different base composition. This is exemplified in the mammalian L1 retrotransposons which have 5′UTRs (57.2 GC%) and 3′UTRs (46.3% GC) that are richer in GC than the two ORFs (39.1% for ORF1 and 37.9% for ORF2) [13,15]. The GC-richness of the L1 promoter is consistent with the nucleotide content found around transcription initiation sites in vertebrates, which is in part due to the abundance of CpG dinucleotides [72]. The AT-richness of ORF1 in vertebrate L1 is always higher than ORF2, which is consistent with the fact that much more ORF1 protein is produced than ORF2 protein [73,74]. The difference between ORFs is even more striking for elements of the L2 clade. In lizard, L2 elements have a GC rich ORF2 (55% GC) but an AT rich ORF1 (54% AT) [15].
Finally, the base composition of non-autonomous elements is extremely variable and is not necessarily related to the base composition of the autonomous elements responsible for their mobility. This is exemplified for SINE elements that are mobilized by LINE elements. The Alu element in primates, which is mobilized by the AT-rich L1 element, is GC-rich (63.3% GC for AluY) while the SINE elements in mouse are either AT-rich (e.g., B2 elements; 52.2% AT) or GC-rich (e.g., B1 elements; 59.9% GC).
4. Consequences of the Unusual Base Composition of Transposable Elements
The unusual base composition of TEs has a number of consequences for the mobility of TEs but also for the genome of their hosts. First the AT-richness of TEs, which is prevalent in multicellular organisms, is suboptimal for the transposition process, both at the transcriptional and at the translational level. In mammalian L1 elements, the A-richness of the positive strand results in poor transcription because A-rich L1 sequences constitute a poor substrate for transcription elongation (either because of a slower rate of elongation, stalling of the RNA polymerase complex or premature dissociation) and because of the presence of A-rich premature poly-adenylation signals that are causing early transcription termination [4,75,76]. It should be noted however that the number of canonical poly-adenylation signals differs among non-LTR retrotransposons and is not directly related to the AT-richness since the number of predicted poly-adenylation signals could vary more than two folds among elements with the same base composition [15]. Although experimental data are lacking for most TEs, it is likely that the AT richness of most TEs impedes their efficient transcription, but the prediction that elements devoid of AT bias exhibit a more efficient transcription remains to be tested. The second potentially negative consequence of a high AT content is at the translational level. The prevalence of AT-ending codons in most TEs makes codon usage of their ORFs poorly adapted for efficient translation, which is supported by the similarity between the codon usage of TEs and weakly expressed host genes [14]. From the point of view of TEs, their AT-richness may appear maladaptive since it negatively affects their transcription and the translation of their proteins.
The negative effect of the biased base composition of TEs is not limited to the TEs but can also impact the expression of host genes in a number of ways. For instance, an AT-rich element inserted within a host gene could decrease the transcription of the gene either by reducing the efficiency of transcription or by producing prematurely terminated transcripts [77,78]. This is one of the reasons AT-rich elements are rarely found in introns, and when they are, they tend to be oriented in the direction that is the least negative to gene expression [79,80,81]. This is exemplified in mammals where L1 elements, which are AT-rich and have a strong A-bias on the positive strand, are extremely rare in introns, and the ones that have reached fixation are found in the opposite orientation to the host gene [79]. Another means TEs will affect the expression of host’s genes is via epigenetic regulation [82]. Repression by DNA methylation at CpG sites constitutes the main means of defense against transposon activity in many organisms [49,50]. Although AT-rich elements will by definition contain few CpG sites, elements that are enriched in GC can contain a number of CpG sites that will be the target of methylation. The repressive mark can spread to the flanking sequences of the transposons and occasionally affect the expression of neighboring genes. The fact that methylated TEs are on average found further away from genes than unmethylated TEs [83,84] and tend to be at lower frequency in populations [85,86] is consistent with a negative effect of TE repression on their neighboring genes.
The base composition of TEs will also affect the overall genomic composition as well as the structure and function of the genome. It is well known that the abundance of TEs is the main determinant of the haploid genome size and TE amplification can cause rapid genome expansion [87,88,89], yet the impact of TEs on the base composition of the host has been underappreciated. In a recent study in fish, a positive correlation between the genomic base composition and the base composition of TEs was found [68]. Since most TEs are AT-rich, small genomes that contain few TEs tend to have a higher GC content than genomes that have experienced large TE amplifications. This observation suggests that the GC content of TEs will drive the GC content of the genomes in which they amplify. This observation is not limited to fish, and the amplification of AT-rich TEs in fungi can cause rapid changes in the genomic base composition. This is exemplified in the genus Leptosphaeria where strains that have experienced TE amplification have genomes with a lower GC content (45% GC) than strains that have not (51% GC) [90].
The GC content can differ among genomic regions (e.g., in birds, mammals and gars) or can be relatively homogenous (e.g., reptiles, amphibians and teleost fish) [1,2,91,92]. The cause of GC heterogeneity in birds and mammals has been the subject of extensive research. It is believed that the main driver of base content heterogeneity is a process called GC-biased gene conversion [93], which causes a fixation bias of G:C alleles over A:T alleles by a recombination-dependent process. Thus, regions of high recombination tend to be GC-rich while regions of low recombination tend to be GC-poor. An aspect that has received little attention is the contribution of TEs to GC heterogeneity. In mammals, the differential accumulation of TEs that differ in base composition contributes to the GC heterogeneity of the genome. AT-rich L1 retrotransposons accumulate in regions of low recombination, presumably because they are eliminated by purifying selection from high recombining regions due to their ability to mediate ectopic recombination [54,94]. They will thus contribute to the higher AT content of regions of low recombination. In contrast, their non-autonomous counterpart, the Alu SINE, is GC-rich and tends to accumulate in genic regions with a high recombination rate [79], thus contributing to the evolution of these GC-rich genomic compartments. Although TEs certainly have an effect on regional base composition, this effect remains to be quantified and the recent development of tools that jointly analyze base composition and TE distribution will contribute to solving this gap in our knowledge [95].
5. Why Do Transposable Elements Have Such Unusual Base Composition?
Two main questions emerge from the analyses of the base composition of TEs. First, why do some TEs exhibit a nucleotide content that is so different from the genome average or from host genes? And second, why do some TEs from the same genome have drastically different composition? The unusual base composition of TEs could result from a number of non-exclusive factors that fall into three broad categories: horizontal transfer, mutational bias, and selective pressure.
TEs that have recently invaded a genome by horizontal transfer will exhibit a base composition that does not reflect processes which have taken place within the genome they occupy. In this case, we do not expect those TEs to show evidence of adaptation to the base composition of their new host. For this reason, a bias in the codon usage of TEs (which is related to the base composition) has been used as evidence of horizontal transfer [96,97,98,99], although many vertically transmitted TEs exhibit a similar bias [15,100,101]. Horizontal transfer can also explain differences in base composition between elements within the same genome. For instance, the genome of the medaka fish Oryzias latipes contains three families of RTE retrotransposons that differ substantially in base composition, but since RTE is prone to horizontal transfer [102,103], it is likely that these differences are caused by the independent transfer of RTEs from different sources [15].
The previous explanation does not apply to elements that are strictly (or mostly) vertically inherited. Many elements have persisted in genomes for very long periods of evolutionary time and have thus had time to evolve within the context of their host, and in this case, the evolution of their nucleotide content can be affected by mutational bias and/or selective pressure. Analyses of the pattern of mutation of recent copies of non-LTR retrotransposons, which are in majority AT-rich, revealed that mutations from C to T and G to A are the most abundant ones and that this mutational bias affects all elements, independently of their clade or base composition [15]. Although the overall mutational bias towards A and T is consistent with the general AT-richness of TEs, it fails to explain the strand bias observed for some elements such as L1. The cause of this mutational bias remains unclear but could result from a number of processes. The use of an error-prone reverse transcriptase by class I elements is unlikely to play a major role because the misincorporation of dATP by reverse transcriptase is exceedingly rare ([104], although this has only been tested on retroviral reverse transcriptase) and because some class I elements, like L2, use a reverse transcriptase for their transposition, yet they can have high GC content. In addition, this does not explain the high AT% of class II elements, which do not rely on a reverse transcriptase for their transposition. Another possibility comes from the action of DNA editing enzymes of the APOBEC family which are part of the defense system against viruses and retrotransposons. APOBEC3 proteins cause G to A mutations on the positive strand and could thus contribute to the A richness of some LINEs [105], although the signature of such editing was detected in a very small fraction of L1 elements in humans. Interestingly, it was shown that APOBEC affects differently the AT-rich L1 and the GC-rich L2 elements in Anolis carolinensis; L1 exhibited a signature of APOBEC editing while L2 did not show any [106], a pattern consistent with the different base composition of these two clades of LINEs. More research is needed to assess the effect of editing on a broader range of organisms and to quantify the impact of APOBEC enzymes on base composition in a variety of contexts. Finally, the genomic environment in which elements are inserted could affect the type of mutations they experience. GC-biased gene conversion will affect differently elements inserted in regions with different recombination rate. Elements that accumulate in low recombining regions would be less subject to GC-biased gene conversion than elements that reside in regions of high recombination and will thus diverge in terms of nucleotide content. This process could be exacerbated by TE-specific insertion bias in favor of genomic regions with high or low recombination rate [107,108,109]. This hypothesis could be tested by comparing the mutation spectrum of TEs residing in different genomic compartments. It should be noted that the bias towards AT is not general, and for instance, a mutation bias from AT to GC was detected in the choanoflagelate Salpingoeca rosetta [70]. Interestingly, TEs in this species do not exhibit the AT-richness found in other organisms.
Another observation suggestive of a mutation-driven evolution of base composition comes from the correlation between the GC% of TEs and the GC% of non-repeated genomic DNA in fish [68]. This positive correlation suggests that the genomic context could be having an effect on the base composition of TEs, but it is unclear how the genomic GC% drives the GC% of TEs. A possibility is that, for their replication, TEs have to use the pool of nucleotides available in the genome of their host, which thus constrains the base composition of TEs. Consequently, it is plausible that the composition of the pool of nucleotides will drive the evolution of the base composition of the TEs closer to the genomic average. This hypothesis remains to be tested, and the variation in base composition reported in teleost fish suggests that this group constitutes a good model.
A strictly neutralist mutational process is however unlikely to fully account for the unusual base composition of TEs, and a number of observations suggest that selection acts on base composition. First, in vertically transmitted TEs, such as LINEs, the base composition remains constant over long periods of evolutionary time suggesting selective pressure or functional constraint on elements [13,15]. This is exemplified in mammals where the L1 retrotransposon has maintained its AT-richness and an A bias on the positive strand since the origin of this vertebrate class. Second, elements can differ in base composition within the same genome (e.g., L1 and L2 elements in Anolis carolinensis), although they are experiencing a similar pattern of mutation [15]. Third, different regions of the same element can exhibit a drastically different base composition. This is exemplified by the difference in base composition between the first and second ORFs of L1 and L2 in mammals and lizard [13,15]. Fourth, some TEs have retained a GC-rich or CT-rich content despite a mutational pressure towards A and T [15]. Fifth, in some species the codon usage is more similar to the host than expected given the base composition of elements, indicating a certain level of codon adaptation.
It is relatively easy to explain the base composition of TEs that exhibit GC% similar to their host’s genes and those cases exemplify adaptation of TEs to their host [70]. It is much harder to provide a selectionist explanation for the apparently maladaptive base composition reported in many TEs. The first possibility is that the unusual nucleotide composition of TEs is not detrimental but is in fact adaptive and is responsible for the fine-tuning of TEs’ transcription and translation. In this context, it is plausible that the high AT content is a mechanism of self-regulation reflecting a trade-off between the efficiency of transposition and the negative impact of transposition on the host [75]. For instance, selection could favor an inefficient transcription of TEs because an element that would be transcribed too efficiently could transpose at a level that would be detrimental to the host [76]. Selection at the transcriptional level seems more likely than selection at the translational level since the AT-richness is often observed at the three codon positions [15]. It is not to say that selection acts only to reduce the efficacy of transposition. In some species, the codon usage tends to be more similar to the host than expected given the composition of the element, which is evidence in favor of adaptation to the translational context of the host genome. This could also explain why the base composition may differ among regions of TEs. For instance, the ORF1 of LINEs has usually a less biased base composition than ORF2, possibly because ORF1p needs to be produced in a much larger amount than ORF2p for successful transposition [73,74]. Another source of selection is that the high AT content is a means of escaping repression by the host. TE expression is regulated by methylation of cytosine at CpG, and a low GC content could constitute a defense of TEs against such inactivating mechanisms. The general under-representation of the CpG dinucleotide in TEs is consistent with this possibility [15,110].
6. Conclusions
Understanding the causes of the unusual base composition of TEs and the respective role of mutational bias and selective pressure remains an understudied aspect of TE biology. It can however teach us a lot about the nature of the interactions between TEs and their hosts. The long-term persistence of a suboptimal base composition could support a model of coexistence between TEs and host, whereby elements evolve towards a reduced transposition rate that minimizes the negative impact they can have on their host. It has been proposed that such model may be more prevalent than the arms-race model that has prevailed until recently [111], yet additional studies will be necessary to confirm that the biased composition of TEs is in fact adaptive. A second unexplored aspect of TE biology that could be related to base composition is how different TEs coexist and possibly compete in their genomic environment. By analogy with the field of ecology, it has been proposed that TEs are comparable with organisms that are sharing a genomic habitat within which they interact [112,113]. The coexistence within the same genomes of TEs with different base composition could inform on the nature of these interactions. For instance, it is possible that elements with different GC content do not use the same resources (either tRNA or amino acids) and could thus coexist since they are not using the same “genomic niche”.
A better understanding of the compositional bias will require further studies. In particular, comparing the timing and intensity of expression of TEs that differ in base composition, within the same genome, could potentially inform on the functionality and possible adaptive value of a biased base composition. Experimental approaches that synthetically modify the base composition of elements by optimizing or de-optimizing codon usage could also prove informative. Finally, we should not underestimate how biased in favor of model organisms our knowledge of TEs is. Recent investigations on unicellular organisms [70] for instance have challenged the common belief that all TEs are AT rich, and we can expect that studies on more non-model organisms will also bring their share of surprises.
Acknowledgments
I thank Dareen Almojil, Sebastian Kirchhof, and two anonymous reviewers for their helpful comments on the manuscript.
Funding
This research was funded by New York University Abu Dhabi (NYUAD) research funds AD180 (to S.B.).
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Not applicable.
Conflicts of Interest
The author declares no conflict of interest.
Figure 1 Schematic representation of the main categories of autonomous transposable elements. The elements are not drawn to scale. The following abbreviations are used: APE, apurinic endonuclease; RT, reverse transcriptase; ORF1, open-reading frame 1; EN, GIY-YIG endonuclease; gag, gag gene; PR, proteinase; IN, integrase; RH, RNase H domain; TR, transposase; YR, tyrosine recombinase; RPA, replication protein A; Rep, replication initiation domain; Hel, helicase; PRO, cysteine protease; Pol, protein-primed type B DNA polymerase; ATP, ATPase. The boxed arrows represent terminal repeats.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Costantini M. Cammarano R. Bernardi G. The evolution of isochore patterns in vertebrate genomes BMC Genom. 2009 10 146 10.1186/1471-2164-10-146
2. Bernardi G. Bernardi G. Compositional patterns in the nuclear genome of cold-blooded vertebrates J. Mol. Evol. 1990 31 265 281 10.1007/BF02101122 2124275
3. Zhou Z. Dang Y. Zhou M. Li L. Yu C.-H. Fu J. Chen S. Liu Y. Codon usage is an important determinant of gene expression levels largely through its effects on transcription Proc. Natl. Acad. Sci. USA 2016 113 E6117 E6125 10.1073/pnas.1606724113 27671647
4. Han J.S. Szak S.T. Boeke J.D. Transcriptional disruption by the L1 retrotransposon and implications for mammalian transcriptomes Nature 2004 429 268 274 10.1038/nature02536 15152245
5. Akashi H. Synonymous codon usage in Drosophila melanogaster: Natural selection and translational accuracy Genetics 1994 136 927 935 10.1093/genetics/136.3.927 8005445
6. Duret L. Evolution of synonymous codon usage in metazoans Curr. Opin. Genet. Dev. 2002 12 640 649 10.1016/S0959-437X(02)00353-2 12433576
7. Duret L. Mouchiroud D. Expression pattern and, surprisingly, gene length shape codon usage in Caenorhabditis, Drosophila, and Arabidopsis Proc. Natl. Acad. Sci. USA 1999 96 4482 4487 10.1073/pnas.96.8.4482 10200288
8. Stoletzki N. Eyre-Walker A. Synonymous codon usage in Escherichia coli: Selection for translational accuracy Mol. Biol. Evol. 2007 24 374 381 10.1093/molbev/msl166 17101719
9. Almojil D. Bourgeois Y. Falis M. Hariyani I. Wilcox J. Boissinot S. The structural, functional and evolutionary impact of transposable elements in eukaryotes Genes 2021 12 918 10.3390/genes12060918 34203645
10. Capy P. Taming, Domestication and Exaptation: Trajectories of Transposable Elements in Genomes Cells 2021 10 3590 10.3390/cells10123590 34944100
11. Bourque G. Burns K.H. Gehring M. Gorbunova V. Seluanov A. Hammell M. Imbeault M. Izsvák Z. Levin H.L. Macfarlan T.S. Ten things you should know about transposable elements Genome Biol. 2018 19 199 10.1186/s13059-018-1577-z 30454069
12. Sinzelle L. Izsvak Z. Ivics Z. Molecular domestication of transposable elements: From detrimental parasites to useful host genes Cell. Mol. Life Sci. 2009 66 1073 1093 10.1007/s00018-009-8376-3 19132291
13. Boissinot S. Sookdeo A. The evolution of LINE-1 in vertebrates Genome Biol. Evol. 2016 8 3485 3507 10.1093/gbe/evw247 28175298
14. Lerat E. Capy P. Biemont C. Codon usage by transposable elements and their host genes in five species J. Mol. Evol. 2002 54 625 637 10.1007/s00239-001-0059-0 11965435
15. Ruggiero R.P. Boissinot S. Variation in base composition underlies functional and evolutionary divergence in non-LTR retrotransposons Mob. DNA 2020 11 14 18 10.1186/s13100-020-00209-9 32280379
16. Shields D.C. Sharp P.M. Evidence that mutation patterns vary among Drosophila transposable elements J. Mol. Biol. 1989 207 843 846 10.1016/0022-2836(89)90252-0 2547975
17. Jia J. Xue Q. Codon usage biases of transposable elements and host nuclear genes in Arabidopsis thaliana and Oryza sativa Genom. Proteom. Bioinform. 2009 7 175 184 10.1016/S1672-0229(08)60047-9
18. Andrieu O. Fiston A.-S. Anxolabéhère D. Quesneville H. Detection of transposable elements by their compositional bias BMC Bioinform. 2004 5 94 10.1186/1471-2105-5-94
19. Tollis M. Boissinot S. The evolutionary dynamics of transposable elements in eukaryote genomes Repetitive DNA 2012 7 68 91
20. Wells J.N. Feschotte C. A Field Guide to Eukaryotic Transposable Elements Annu. Rev. Genet. 2020 54 539 561 10.1146/annurev-genet-040620-022145 32955944
21. Kojima K.K. Structural and sequence diversity of eukaryotic transposable elements Genes Genet. Syst. 2018 94 233 252 10.1266/ggs.18-00024 30416149
22. Wicker T. Sabot F. Hua-Van A. Bennetzen J.L. Capy P. Chalhoub B. Flavell A. Leroy P. Morgante M. Panaud O. A unified classification system for eukaryotic transposable elements Nat. Rev. Genet. 2007 8 973 982 10.1038/nrg2165 17984973
23. Eickbush T.H. Malik H.S. Origins and evolution of retrotransposons Mobile DNA II American Society of Microbiology Washington, DC, USA 2002 1111 1144
24. Dewannieux M. Esnault C. Heidmann T. LINE-mediated retrotransposition of marked Alu sequences Nat. Genet. 2003 35 41 48 10.1038/ng1223 12897783
25. Dewannieux M. Heidmann T. LINEs, SINEs and processed pseudogenes: Parasitic strategies for genome modeling Cytogenet. Genome Res. 2005 110 35 48 10.1159/000084936 16093656
26. Ohshima K. Okada N. SINEs and LINEs: Symbionts of eukaryotic genomes with a common tail Cytogenet. Genome Res. 2005 110 475 490 10.1159/000084981 16093701
27. Feschotte C. Pritham E.J. DNA transposons and the evolution of eukaryotic genomes Annu. Rev. Genet. 2007 41 331 368 10.1146/annurev.genet.40.110405.090448 18076328
28. Kapitonov V.V. Jurka J. Rolling-circle transposons in eukaryotes Proc. Natl. Acad. Sci. USA 2001 98 8714 8719 10.1073/pnas.151269298 11447285
29. Kapitonov V.V. Jurka J. Helitrons on a roll: Eukaryotic rolling-circle transposons Trends Genet. 2007 23 521 529 10.1016/j.tig.2007.08.004 17850916
30. Kapitonov V.V. Jurka J. Self-synthesizing DNA transposons in eukaryotes Proc. Natl. Acad. Sci. USA 2006 103 4540 4545 10.1073/pnas.0600833103 16537396
31. Pritham E.J. Putliwala T. Feschotte C. Mavericks, a novel class of giant transposable elements widespread in eukaryotes and related to DNA viruses Gene 2007 390 3 17 10.1016/j.gene.2006.08.008 17034960
32. Hartl D. Lozovskaya E. Lawrence J. Nonautonomous transposable elements in prokaryotes and eukaryotes Genetica 1992 86 47 53 10.1007/BF00133710 1334917
33. Novick P.A. Smith J.D. Floumanhaft M. Ray D.A. Boissinot S. The evolution and diversity of DNA transposons in the genome of the lizard Anolis carolinensis Genome Biol. Evol. 2011 3 1 14 10.1093/gbe/evq080 21127169
34. Yang G. Nagel D.H. Feschotte C. Hancock C.N. Wessler S.R. Tuned for transposition: Molecular determinants underlying the hyperactivity of a Stowaway MITE Science 2009 325 1391 1394 10.1126/science.1175688 19745152
35. Smit A.F. Tóth G. Riggs A.D. Jurka J. Ancestral, mammalian-wide subfamilies of LINE-1 repetitive sequences J. Mol. Biol. 1995 246 401 417 10.1006/jmbi.1994.0095 7877164
36. Furano A.V. The biological properties and evolutionary dynamics of mammalian LINE-1 retrotransposons Prog. Nucleic Acid Res. Mole. Biol. 2000 64 255 294
37. Furano A.V. Duvernell D.D. Boissinot S. L1 (LINE-1) retrotransposon diversity differs dramatically between mammals and fish Trends Genet. 2004 20 9 14 10.1016/j.tig.2003.11.006 14698614
38. Khan H. Smit A. Boissinot S. Molecular evolution and tempo of amplification of human LINE-1 retrotransposons since the origin of primates Genome Res. 2006 16 78 87 10.1101/gr.4001406 16344559
39. Sookdeo A. Hepp C.M. Boissinot S. Contrasted patterns of evolution of the LINE-1 retrotransposon in perissodactyls: The history of a LINE-1 extinction Mob. DNA 2018 9 12 10.1186/s13100-018-0117-4 29610583
40. Sookdeo A. Hepp C.M. McClure M.A. Boissinot S. Revisiting the evolution of mouse LINE-1 in the genomic era Mob. DNA 2013 4 3 10.1186/1759-8753-4-3 23286374
41. Boissinot S. Furano A.V. Adaptive evolution in LINE-1 retrotransposons Mol. Biol. Evol. 2001 18 2186 2194 10.1093/oxfordjournals.molbev.a003765 11719568
42. Jacobs F.M. Greenberg D. Nguyen N. Haeussler M. Ewing A.D. Katzman S. Paten B. Salama S.R. Haussler D. An evolutionary arms race between KRAB zinc-finger genes ZNF91/93 and SVA/L1 retrotransposons Nature 2014 516 242 245 10.1038/nature13760 25274305
43. Blumenstiel J.P. Birth, school, work, death, and resurrection: The life stages and dynamics of transposable element proliferation Genes 2019 10 336 10.3390/genes10050336 31058854
44. Schaack S. Gilbert C. Feschotte C. Promiscuous DNA: Horizontal transfer of transposable elements and why it matters for eukaryotic evolution Trends Ecol. Evol. 2010 25 537 546 10.1016/j.tree.2010.06.001 20591532
45. Gilbert C. Feschotte C. Horizontal acquisition of transposable elements and viral sequences: Patterns and consequences Curr. Opin. Genet. Dev. 2018 49 15 24 10.1016/j.gde.2018.02.007 29505963
46. Bourgeois Y. Boissinot S. On the population dynamics of junk: A review on the population genomics of transposable elements Genes 2019 10 419 10.3390/genes10060419
47. Goodier J.L. Restricting retrotransposons: A review Mob. DNA 2016 7 16 10.1186/s13100-016-0070-z 27525044
48. Czech B. Hannon G.J. One Loop to Rule Them All: The Ping-Pong Cycle and piRNA-Guided Silencing Trends Biochem. Sci. 2016 41 324 337 10.1016/j.tibs.2015.12.008 26810602
49. Deniz O. Frost J.M. Branco M.R. Regulation of transposable elements by DNA modifications Nat. Rev. Genet. 2019 20 417 431 10.1038/s41576-019-0106-6 30867571
50. Yoder J.A. Walsh C.P. Bestor T.H. Cytosine methylation and the ecology of intragenomic parasites Trends Genet. 1997 13 335 340 10.1016/S0168-9525(97)01181-5 9260521
51. Bourgeois Y. Ruggiero R. Hariyani I. Boissinot S. Disentangling the determinants of transposable elements dynamics in vertebrate genomes using empirical evidences and simulations PLoS Genet. 2020 16 e1009082 10.1371/journal.pgen.1009082 33017388
52. Ruggiero R.P. Bourgeois Y. Boissinot S. LINE insertion polymorphisms are abundant but at low frequencies across populations of Anolis carolinensis Front. Genet. 2017 8 44 10.3389/fgene.2017.00044 28450881
53. Biémont C. Lemeunier F. Guerreiro M.G. Brookfield J. Gautier C. Aulard S. Pasyukova E. Population dynamics of the copia, mdg1, mdg3, gypsy, and P transposable elements in a natural population of Drosophila melanogaster Genet. Res. 1994 63 197 212 10.1017/S0016672300032353 8082837
54. Boissinot S. Entezam A. Furano A.V. Selection against deleterious LINE-1-containing loci in the human lineage Mol. Biol. Evol. 2001 18 926 935 10.1093/oxfordjournals.molbev.a003893 11371580
55. Charlesworth B. Langley C.H. The population genetics of Drosophila transposable elements Annu. Rev. Genet. 1989 23 251 287 10.1146/annurev.ge.23.120189.001343 2559652
56. Xue A.T. Ruggiero R.P. Hickerson M.J. Boissinot S. Differential effect of selection against LINE retrotransposons among vertebrates inferred from whole-genome data and demographic modeling Genome Biol. Evol. 2018 10 1265 1281 10.1093/gbe/evy083 29688421
57. Gonzalez J. Macpherson J.M. Messer P.W. Petrov D.A. Inferring the strength of selection in Drosophila under complex demographic models Mol. Biol. Evol. 2009 26 513 526 10.1093/molbev/msn270 19033258
58. Lockton S. Ross-Ibarra J. Gaut B.S. Demography and weak selection drive patterns of transposable element diversity in natural populations of Arabidopsis lyrata Proc. Natl. Acad. Sci. USA 2008 105 13965 13970 10.1073/pnas.0804671105 18772373
59. Garcia Guerreiro M.P. Chavez-Sandoval B.E. Balanya J. Serra L. Fontdevila A. Distribution of the transposable elements bilbo and gypsy in original and colonizing populations of Drosophila subobscura BMC Evol. Biol. 2008 8 234 10.1186/1471-2148-8-234 18702820
60. Lynch M. Conery J.S. The origins of genome complexity Science 2003 302 1401 1404 10.1126/science.1089370 14631042
61. Blass E. Bell M. Boissinot S. Accumulation and rapid decay of non-LTR retrotransposons in the genome of the three-spine stickleback Genome Biol. Evol. 2012 4 687 702 10.1093/gbe/evs044 22534163
62. Petrov D.A. Sangster T.A. Johnston J.S. Hartl D.L. Shaw K.L. Evidence for DNA loss as a determinant of genome size Science 2000 287 1060 1062 10.1126/science.287.5455.1060 10669421
63. Kapusta A. Suh A. Feschotte C. Dynamics of genome size evolution in birds and mammals Proc. Natl. Acad. Sci. USA 2017 114 E1460 E1469 10.1073/pnas.1616702114 28179571
64. Hershberg R. Petrov D.A. Selection on codon bias Annu. Rev. Genet. 2008 42 287 299 10.1146/annurev.genet.42.110807.091442 18983258
65. Ingvarsson P.K. Gene expression and protein length influence codon usage and rates of sequence evolution in Populus tremula Mol. Biol. Evol. 2007 24 836 844 10.1093/molbev/msl212 17204548
66. Shields D.C. Sharp P.M. Higgins D.G. Wright F. “Silent” sites in Drosophila genes are not neutral: Evidence of selection among synonymous codons Mol. Biol. Evol. 1988 5 704 716 3146682
67. Gaffaroglu M. Majtanova Z. Symonova R. Pelikanova S. Unal S. Lajbner Z. Rab P. Present and Future Salmonid Cytogenetics Genes 2020 11 1462 10.3390/genes11121462
68. Symonova R. Suh A. Nucleotide composition of transposable elements likely contributes to AT/GC compositional homogeneity of teleost fish genomes Mob. DNA 2019 10 49 10.1186/s13100-019-0195-y 31857829
69. Besansky N. Codon usage patterns in chromosomal and retrotransposon genes of the mosquito Anopheles gambiae Insect Mol. Biol. 1993 1 171 178 10.1111/j.1365-2583.1993.tb00089.x 8269095
70. Southworth J. Grace C.A. Marron A.O. Fatima N. Carr M. A genomic survey of transposable elements in the choanoflagellate Salpingoeca rosetta reveals selection on codon usage Mob. DNA 2019 10 44 10.1186/s13100-019-0189-9 31788034
71. Jiang R.H. Govers F. Nonneutral GC3 and retroelement codon mimicry in Phytophthora J. Mol. Evol. 2006 63 458 472 10.1007/s00239-005-0211-3 16955239
72. Aerts S. Thijs G. Dabrowski M. Moreau Y. De Moor B. Comprehensive analysis of the base composition around the transcription start site in Metazoa BMC Genom. 2004 5 34 10.1186/1471-2164-5-34 15171795
73. Basame S. Li P.W.-l. Howard G. Branciforte D. Keller D. Martin S.L. Spatial assembly and RNA binding stoichiometry of a LINE-1 protein essential for retrotransposition J. Mol. Biol. 2006 357 351 357 10.1016/j.jmb.2005.12.063 16434051
74. Doucet A.J. Hulme A.E. Sahinovic E. Kulpa D.A. Moldovan J.B. Kopera H.C. Athanikar J.N. Hasnaoui M. Bucheton A. Moran J.V. Characterization of LINE-1 ribonucleoprotein particles PLoS Genet. 2010 6 e1001150 10.1371/journal.pgen.1001150 20949108
75. Perepelitsa-Belancio V. Deininger P. RNA truncation by premature polyadenylation attenuates human mobile element activity Nat. Genet. 2003 35 363 366 10.1038/ng1269 14625551
76. Han J.S. Boeke J.D. A highly active synthetic mammalian retrotransposon Nature 2004 429 314 318 10.1038/nature02535 15152256
77. Han J.S. Boeke J.D. LINE-1 retrotransposons: Modulators of quantity and quality of mammalian gene expression? Bioessays 2005 27 775 784 10.1002/bies.20257 16015595
78. Roy-Engel A. El-Sawy M. Farooq L. Odom G. Perepelitsa-Belancio V. Bruch H. Oyeniran O. Deininger P. Human retroelements may introduce intragenic polyadenylation signals Cytogenet. Genome Res. 2005 110 365 371 10.1159/000084968 16093688
79. Medstrand P. van de Lagemaat L.N. Mager D.L. Retroelement distributions in the human genome: Variations associated with age and proximity to genes Genome Res. 2002 12 1483 1495 10.1101/gr.388902 12368240
80. Cutter A.D. Good J.M. Pappas C.T. Saunders M.A. Starrett D.M. Wheeler T.J. Transposable element orientation bias in the Drosophila melanogaster genome J. Mol. Evol. 2005 61 733 741 10.1007/s00239-004-0243-0 16315105
81. van de Lagemaat L.N. Landry J.R. Mager D.L. Medstrand P. Transposable elements in mammals promote regulatory variation and diversification of genes with specialized functions Trends Genet. 2003 19 530 536 10.1016/j.tig.2003.08.004 14550626
82. Slotkin R.K. Martienssen R. Transposable elements and the epigenetic regulation of the genome Nat. Rev. Genet. 2007 8 272 285 10.1038/nrg2072 17363976
83. Kinoshita Y. Saze H. Kinoshita T. Miura A. Soppe W.J. Koornneef M. Kakutani T. Control of FWA gene silencing in Arabidopsis thaliana by SINE-related direct repeats Plant J. 2007 49 38 45 10.1111/j.1365-313X.2006.02936.x 17144899
84. Parinov S. Sundaresan V. Functional genomics in Arabidopsis: Large-scale insertional mutagenesis complements the genome sequencing project Curr. Opin. Biotechnol. 2000 11 157 161 10.1016/S0958-1669(00)00075-6 10753770
85. Hollister J.D. Gaut B.S. Epigenetic silencing of transposable elements: A trade-off between reduced transposition and deleterious effects on neighboring gene expression Genome Res. 2009 19 1419 1428 10.1101/gr.091678.109 19478138
86. Lee Y.C.G. Karpen G.H. Pervasive epigenetic effects of Drosophila euchromatic transposable elements impact their evolution eLife 2017 6 e25762 10.7554/eLife.25762 28695823
87. Vitte C. Panaud O. LTR retrotransposons and flowering plant genome size: Emergence of the increase/decrease model Cytogenet. Genome Res. 2005 110 91 107 10.1159/000084941 16093661
88. Hawkins J.S. Kim H. Nason J.D. Wing R.A. Wendel J.F. Differential lineage-specific amplification of transposable elements is responsible for genome size variation in Gossypium Genome Res. 2006 16 1252 1261 10.1101/gr.5282906 16954538
89. Piegu B. Guyot R. Picault N. Roulin A. Sanyal A. Kim H. Collura K. Brar D.S. Jackson S. Wing R.A. Doubling genome size without polyploidization: Dynamics of retrotransposition-driven genomic expansions in Oryza australiensis, a wild relative of rice Genome Res. 2006 16 1262 1269 10.1101/gr.5290206 16963705
90. Grandaubert J. Lowe R.G. Soyer J.L. Schoch C.L. Van de Wouw A.P. Fudal I. Robbertse B. Lapalu N. Links M.G. Ollivier B. Transposable element-assisted evolution and adaptation to host plant within the Leptosphaeria maculans-Leptosphaeria biglobosa species complex of fungal pathogens BMC Genom. 2014 15 891 10.1186/1471-2164-15-891
91. Symonova R. Majtanova Z. Arias-Rodriguez L. Morkovsky L. Korinkova T. Cavin L. Pokorna M.J. Dolezalkova M. Flajshans M. Normandeau E. Genome Compositional Organization in Gars Shows More Similarities to Mammals than to Other Ray-Finned Fish J. Exp. Zool. B Mol. Dev. Evol. 2017 328 607 619 10.1002/jez.b.22719 28035749
92. Costantini M. Auletta F. Bernardi G. Isochore patterns and gene distributions in fish genomes Genomics 2007 90 364 371 10.1016/j.ygeno.2007.05.006 17590311
93. Mugal C.F. Weber C.C. Ellegren H. GC-biased gene conversion links the recombination landscape and demography to genomic base composition: GC-biased gene conversion drives genomic base composition across a wide range of species Bioessays 2015 37 1317 1326 10.1002/bies.201500058 26445215
94. Song M. Boissinot S. Selection against LINE-1 retrotransposons results principally from their ability to mediate ectopic recombination Gene 2007 390 206 213 10.1016/j.gene.2006.09.033 17134851
95. Matoulek D. Borůvková V. Ocalewicz K. Symonová R. GC and repeats profiling along chromosomes—The future of fish compositional cytogenomics Genes 2021 12 50 10.3390/genes12010050 33396302
96. Rödelsperger C. Sommer R.J. Computational archaeology of the Pristionchus pacificus genome reveals evidence of horizontal gene transfers from insects BMC Evol. Biol. 2011 11 239 10.1186/1471-2148-11-239 21843315
97. McHale M.T. Roberts I.N. Noble S.M. Beaumont C. Whitehead M.P. Seth D. Oliver R.P. CfT-I: An LTR-retrotransposon in Cladosporium fulvum, a fungal pathogen of tomato Mol. Gen. Genet. MGG 1992 233 337 347 10.1007/BF00265429 1377773
98. Wallau G.L. Ortiz M.F. Loreto E.L.S. Horizontal transposon transfer in eukarya: Detection, bias, and perspectives Genome Biol. Evol. 2012 4 801 811 10.1093/gbe/evs055
99. Powell J.R. Gleason J.M. Codon usage and the origin of P elements Mol. Biol. Evol. 1996 13 278 279 10.1093/oxfordjournals.molbev.a025564 8583900
100. Springer M.S. Tusneem N.A. Davidson E.H. Britten R.J. Phylogeny, rates of evolution, and patterns of codon usage among sea urchin retroviral-like elements, with implications for the recognition of horizontal transfer Mol. Biol. Evol. 1995 12 219 230 7535375
101. Lerat E. Biémont C. Capy P. Codon usage and the origin of P elements Mol. Biol. Evol. 2000 17 467 468 10.1093/oxfordjournals.molbev.a026326 10723747
102. Kordis D. Gubensek F. Unusual horizontal transfer of a long interspersed nuclear element between distant vertebrate classes Proc. Natl. Acad. Sci. USA 1998 95 10704 10709 10.1073/pnas.95.18.10704 9724768
103. Walsh A.M. Kortschak R.D. Gardner M.G. Bertozzi T. Adelson D.L. Widespread horizontal transfer of retrotransposons Proc. Natl. Acad. Sci. USA 2013 110 1012 1016 10.1073/pnas.1205856110 23277587
104. Preston B.D. Poiesz B.J. Loeb L.A. Fidelity of HIV-1 reverse transcriptase Science 1988 242 1168 1171 10.1126/science.2460924 2460924
105. Carmi S. Church G.M. Levanon E.Y. Large-scale DNA editing of retrotransposons accelerates mammalian genome evolution Nat. Commun. 2011 2 519 10.1038/ncomms1525 22044998
106. Lindič N. Budič M. Petan T. Knisbacher B.A. Levanon E.Y. Lovšin N. Differential inhibition of LINE1 and LINE2 retrotransposition by vertebrate AID/APOBEC proteins Retrovirology 2013 10 156 10.1186/1742-4690-10-156 24344916
107. Duret L. Marais G. Biémont C. Transposons but not retrotransposons are located preferentially in regions of high recombination rate in Caenorhabditis elegans Genetics 2000 156 1661 1669 10.1093/genetics/156.4.1661 11102365
108. Kawakami T. Mugal C.F. Suh A. Nater A. Burri R. Smeds L. Ellegren H. Whole-genome patterns of linkage disequilibrium across flycatcher populations clarify the causes and consequences of fine-scale recombination rate variation in birds Mol. Ecol. 2017 26 4158 4172 10.1111/mec.14197 28597534
109. Myers S. Freeman C. Auton A. Donnelly P. McVean G. A common sequence motif associated with recombination hot spots and genome instability in humans Nat. Genet. 2008 40 1124 10.1038/ng.213 19165926
110. Lerat E. Capy P. Biémont C. The relative abundance of dinucleotides in transposable elements in five species Mol. Biol. Evol. 2002 19 964 967 10.1093/oxfordjournals.molbev.a004154 12032253
111. Cosby R.L. Chang N.-C. Feschotte C. Host–transposon interactions: Conflict, cooperation, and cooption Genes Dev. 2019 33 1098 1116 10.1101/gad.327312.119 31481535
112. Venner S. Feschotte C. Biémont C. Dynamics of transposable elements: Towards a community ecology of the genome Trends Genet. 2009 25 317 323 10.1016/j.tig.2009.05.003 19540613
113. Brookfield J.F. The ecology of the genome—Mobile DNA elements and their hosts Nat. Rev. Genet. 2005 6 128 136 10.1038/nrg1524 15640810
|
PMC009xxxxxx/PMC9099905.txt |
==== Front
Cells
Cells
cells
Cells
2073-4409
MDPI
10.3390/cells11091518
cells-11-01518
Article
Focused Ultrasound Treatment of a Spheroid In Vitro Tumour Model
Landgraf Lisa 1
https://orcid.org/0000-0002-1346-2365
Kozlowski Adam 12
Zhang Xinrui 1
https://orcid.org/0000-0002-0690-9384
Fournelle Marc 3
Becker Franz-Josef 3
https://orcid.org/0000-0002-5710-0825
Tretbar Steffen 3
Melzer Andreas 1*
Raghunath Michael Academic Editor
1 Innovation Center Computer Assisted Surgery, Institute at the Faculty of Medicine, Leipzig University, 04103 Leipzig, Germany; [email protected] (L.L.); [email protected] (A.K.); [email protected] (X.Z.)
2 Department of Anatomy, School of Medicine, Collegium Medicum, University of Warmia and Mazury, 10-082 Olsztyn, Poland
3 Fraunhofer Institute for Biomedical Engineering (IBMT), 66280 Sulzbach, Germany; [email protected] (M.F.); [email protected] (F.-J.B.); [email protected] (S.T.)
* Correspondence: [email protected]
30 4 2022
5 2022
11 9 151815 3 2022
26 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Simple Summary
Ultrasound waves can be applied for diagnostic and therapeutic purposes. Focused ultrasound is approved for tissue ablation, e.g., in the treatment of uterine fibroids or essential tremors. Besides the non-invasive image-guided surgical intervention at temperatures above 55 °C, FUS is investigated in other fields like blood-brain barrier opening, hyperthermia, and neuromodulation. FUS offers potential as an adjuvant therapy in cancer treatment. Therefore, analysis of FUS effects on cancer cells is necessary. We performed studies on two human cancer cell line spheroids using a newly developed high-throughput in vitro FUS applicator with 32 individual transducers. This study aimed to perform basic experiments with a new in vitro FUS device on a 3D tumour model to acquire insight into the effects of FUS at the cellular level. These experiments may contribute to a better understanding and predictions of cancer treatment efficacy.
Abstract
Focused ultrasound (FUS) is a non-invasive technique producing a variety of biological effects by either thermal or mechanical mechanisms of ultrasound interaction with the targeted tissue. FUS could bring benefits, e.g., tumour sensitisation, immune stimulation, and targeted drug delivery, but investigation of FUS effects at the cellular level is still missing. New techniques are commonly tested in vitro on two-dimensional (2D) monolayer cancer cell culture models. The 3D tumour model—spheroid—is mainly utilised to mimic solid tumours from an architectural standpoint. It is a promising method to simulate the characteristics of tumours in vitro and their various responses to therapeutic alternatives. This study aimed to evaluate the effects of FUS on human prostate and glioblastoma cancer tumour spheroids in vitro. The experimental follow-up enclosed the measurements of spheroid integrity and growth kinetics, DNA damage, and cellular metabolic activity by measuring intracellular ATP content in the spheroids. Our results showed that pulsed FUS treatment induced molecular effects in 3D tumour models. With the disruption of the spheroid integrity, we observed an increase in DNA double-strand breaks, leading to damage in the cancer cells depending on the cancer cell type.
focused ultrasound
FUS
spheroid
in vitro experiments
prostate cancer
glioblastoma
German Federal Ministry of Education and Research (BMBF)03Z1L511 This research was funded by the German Federal Ministry of Education and Research (BMBF) under grant no. 03Z1L511 (SONO-RAY project).
==== Body
pmc1. Introduction
In the management of cancerous disease, modern minimal or non-surgical strategies like radiofrequency, microwave ablation, or focused ultrasound (FUS) are used to ablate the tissue for the direct destruction of cancer cells. FUS can be applied to sensitise them by heating (41–47 °C) as an adjuvant therapy [1] to other treatment modalities, including chemo and/or radiation therapy. In this context, FUS brings value with its non-invasive, incision-free, and ionising-free treatment characteristics, controllability via real-time MR image guidance, and the capacity to activate the immune system [2].
In developing new cancer treatment strategies or new drugs, the process routinely starts with cancer cells in standardised culture conditions in vitro using a two-dimensional (2D), homogeneous model [3]. The method is advantageous in terms of its wide availability, ease of use, and low costs. In such a scenario, a singular cell suspension is distributed on a flat plate surface, where further investigative manipulations are performed to analyse the potential therapeutic agent effect. This given design inadequately reflects the in-vivo cellular arrangement and structure, which might also risk being expressed when translating into the clinical environment, producing a lack of success in achieving desired and adequate results. A more realistic 3D cell model is an initial step required to create more lifelike and complex in-vitro models. This not entirely eliminates but significantly improves the in vitro model scope and limitations. The artificial format restrains the reproduction of tissue and systems-level cellular interactions, preventing and maintaining accurate physiological processes as found occurring in a tumour’s natural in-vivo habitat.
In our study, spherical (3D) homogenous cancer aggregates were an entry point in establishing in-vivo imitating tumour systems. Three-dimensional cell cultures reveal features that cell monolayer models lack, such as an advanced network of the cell-cell complexity with the development of pH, oxygen, and metabolic gradients—stratifying mature spheroids to develop a secondary necrotic core and proliferation zone [4], resembling the avascular stage of solid tumours and actively circulating cancer cells (micro-metastases). Investigation of such 3D model behaviour in response to various treatment modalities is superior to the previously mentioned 2D setting and serves as a better scientific approach. Peripheral cells of such a model resemble the situation of actively circulating tumour cells being adjacent to capillaries in the in vivo state. In contrast, the innermost ones eventually die due to hypoxic conditions via apoptosis or necrosis, forming a secondary necrotic core [5]. Oxygenic stress is an important aspect of many physiological processes embraced in numerous human diseases, including cancer. Notably, hypoxic areas in tumours often lead to lower efficiency of radio-therapies [6]. The investigation of hypoxia is challenging to establish due to the low availability of sensitive fluorescent dyes [7], explicitly measuring the activity of hypoxic cells.
Many novel 3D organotypic models have been introduced in recent years. These sophisticated cancer cell models resemble tissue structure, function, and even disease progression. Furthermore, 3D multicellular tumour spheroid co-cultures, combining different, e.g., immune cells with cancer cells, are being created [4], resembling one of the more realistic approaches in tumour modelling. 3D tumor models have been recently noted for efficient evaluating the cancer ability or therapeutic efficiency [8] such as proliferation [9], alternation [10], invasion [11], morphology [12], or drug resistance [13]. These systems enabled the promotion of the development of new drug candidates or novel therapeutic effect [8]. However, there are few papers on the application of FUS on tumor spheroids mainly in the context of nanoparticle formulations [14,15].
Furthermore, DNA double-strand breaks (DSBs) are conducive to both genomic instability and cancer treatment. Monitoring DNA damage in a cell by detecting immunolabelled γH2A.X is useful to track cancer progression and treatment effects [16]. Flow cytometry, to rapidly quantify γH2A.X, allows for attaining the profile of tumour behaviour under therapeutic stress. Another classical investigative endpoint between treated and untreated spheroids relies on observing the cell survival and spheroid growth delay, including integrity and volume loss [17].
Ultrasound is a safe, inexpensive, and widespread diagnostic tool capable of producing real-time, non-invasive images without evoking significant biological effects. In addition to ultrasound imaging, ultrasound waves can be focused at higher energies and sound pressures as therapeutic tools. Ultrasound can be generated by piezoelectric crystals, driven to vibrate by a specific fluctuating voltage. The devices containing these piezoelectric crystals and some electronics are called transducers since they convert electrical to mechanical energy and vice versa. In the case of focused ultrasound, the transducer creates ultrasound beams focused to a single focal zone, whereby the acoustic energy increases near and in the focus (Figure 1a).
FUS is a platform technology that produces biological responses through thermal or mechanical effects that act therapeutically on the target. These effects depend on the tissue composition (e.g., muscle vs. bone) and the ultrasound parameters (power, duration, mode—continuous vs. pulsed). The most pronounced effects caused by FUS on tissue are thermal ablation and mechanical tissue destruction (cavitation). The first one is a consequence of heating the tissue that denatures proteins and leads to the death of all cells, regardless of whether they are normal or abnormal. The dose required to produce irreversible damage and coagulative necrosis depends on the cell type, temperature, and duration of exposure—from 1 s at 56 °C to 240 min at 43 °C. Mechanical tissue destruction by FUS appears when the disruption of cells occurs through purely mechanical effects (no heating). The effect, called cavitation, occurs when gas bubbles oscillate in an ultrasonic field [18]. When these structures collapse, it is known as inertial cavitation, where enough force is accumulated to allow for targeted, localised tissue destruction.
Many other effects can be induced by FUS, such as sonoporation, vaso-dilation/constriction, substance delivery vehicles, or increasing vascular permeability, the clinical potentials of which are currently being investigated in, i.e., drug delivery, neuro-/immune-modulation, or radiation sensitisation.
For that reason, translation of FUS experiments performed on 3D tumour spheroids is necessary. It may also serve as a prudent act in avoiding animal experiments (replacement), limiting the number of animals (reduction) and their suffering (refinement) in tests to an absolute minimum [19]. The 3D tumour model (spheroid) development is essential to achieve the system’s benefits of mimicking avascular solid tumours. It provides more physiologically information when compared with 2D systems [20].
2. Materials and Methods
This study used a newly developed high-throughput in vitro FUS system to treat 3D spheroids in vitro.
2.1. Tumor Cell Lines and Cell Culture
The human prostate cancer cell line PC-3 was purchased from the ECACC (Salisbury, UK) and grown in Roswell Park Memorial Institute 1640 Medium (RPMI, gibco® by Thermo Fisher Scientific (Waltham, MA, USA)). Human glioblastoma cell line U87 was obtained from the Department of Radiation Oncology (University of Leipzig, Leipzig, Germany). It was cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, gibco® by Thermo Fisher Scientific (Waltham, MA, USA)). All cell culture mediums were supplemented with 10% (v/v) fetal bovine serum (FBS, Sigma-Aldrich (St. Louis, MO, USA)) and 100 U/mL penicillin, and 100 mg/mL streptomycin (Biochrom GmbH (Berlin, Germany)) for culturing of the cell lines at 37 °C in a humidified incubator supplemented with 5% (v/v) CO2. The cells were routinely washed with phosphate-buffered saline without Ca+, Mg+, and phenol red (PBS BioWhittaker®, Lonza Group Ltd., (Basel, Switzerland)) and detached using trypsin/EDTA (Lonza Group Ltd., (Basel, Switzerland)).
2.2. Generation of PC-3 and U87 Spheroids
Following cell detachment, the dissociation enzyme was neutralised with a dedicated medium and the cells were centrifuged at 300× g for 5 min. The supernatant was removed, and the cell pellet was resuspended again in a complete growth medium. Cells were then counted with a Neubauer counting chamber (Paul Marienfeld GmbH & Co. KG (Lauda-Königshofen, Germany)) to achieve 5000 cells/200 μL for PC-3 and 1000 cells/200 μL for U87, respectively. A liquid overlay spheroid formation technique was chosen, adapted by Froehlich et al. [21], and a dedicated cancer cell line suspension of 200 μL was added to each well of ultra-low attachment (ULA) 96-well CELLSTAR® round-bottom plates (Greiner Bio-One GmbH (Frickenhausen, Germany)). The cells were cultivated for four days in a humidified incubator (Figure 1c).
2.3. Characterisation of the In Vitro Applicator
The output was characterised before using the cell applicator for FUS treatment experiments on cell cultures. For this purpose, three sets of characterisation experiments were performed. First, the sound field of an individual transducer element was assessed to ensure that the sound fields from adjacent transducers did not overlap. Second, the output of all transducers was evaluated to characterise the homogeneity of the applicator. Finally, the acoustic output as a function of the applied power settings was measured for a single transducer element.
For assessment of the transducer performance, sound field measurements were performed in a water tank. The generated acoustic signals were acquired by a calibrated hydrophone (Type S, RP Acoustics (Leutenbach, Germany)) and analysed offline using Matlab (The MathWorks (Natick, MA, USA)). The hydrophone was moved in 2D to acquire the pressure distribution fields using an in-house-developed sound field scanning system.
An XY scan was performed in front of the whole applicator using the hydrophone in a second step. For data analysis, the surface corresponding to one well cross-section was automatically segmented in front of each transducer, and the intensity (ISPTA) was averaged within the segmented area.
2.4. Establishment of FUS Treatment of Tumour Cell Spheroids
The FUS in vitro system (Figure 1b) consists of a newly developed customised ultrasound cell applicator (Fraunhofer IBMT, (St. Ingbert, Germany)) designed for delivering acoustic energy to a 96-well cell culture plate. The applicator includes 32 cylindrically focusing transducers working at a frequency of 1.1 MHz. The 32 individual transducers are driven by a high power generator/amplifier (AG 1016, T&C Power Conversion Systems (Rochester, NY, USA)). The transducer can either be driven in parallel or in two subgroups of 16 transducers (each consisting of two rows of 8 transducers for sonication of one row of the well plate). An impedance matching circuit (T1K-7A, Power Conversion Systems) is connected between the applicator and the generator for improved efficiency. The transducers are water-cooled to prevent heat damage with an external pump (WK 16-1 DS, Colora Messtechnik GmbH, (Lorch, Germany)). An acoustically transparent membrane seals the cooling circuit. A water stand-off achieves acoustic coupling between the membrane and the well plate. A 3D-printed well-plate adapter was used to hold the well plate in the correct lateral and axial position. For control of the cell applicator, the customised software tool “Cell Therapy Planning Tool” (Fraunhofer IBMT (St. Ingbert, Germany)) was used. The software allows programming of the sonication time, the duty cycle, the burst repetition rate, and the applied power level, either for the whole applicator or for one of the two subgroups of 16 transducers.
It was essential for the experiment to prevent cells and the system from excessive heating and achieve merely mechanical acoustic effects. The real-time temperature in the wells during FUS treatment was monitored using an infrared thermal camera PI450 (Optris GmbH (Berlin, Germany)) and imaging software (PI Connect v2.16, https://www.optris.global/optris-pix-connect (accessed on 15 March 2020)).
2.5. Experimental Protocol for In Vitro Focused Ultrasound Treatment
Spheroids for each cancer cell line were formed as previously described. The spheroid selection was conducted to choose the most appropriate ones for the experiment (exclusion: externally-sourced cotton fibres and any other disintegrating factor) or for substitution in case of mishandling (pre-treatment stage only). For the execution of in vitro FUS treatment, the spheroids for each cancer cell were recruited. The remaining ones were devoted to negative (untreated group) and positive control (+5% DMSO) groups creation. The 96-well ultrasound penetrable μclear plate wells were filled with 150 μL of the dedicated cancer cell line medium and prepared + control solution medium (dedicated culture medium +5% DMSO). Then, 150 μL medium was removed from the spheroid-containing U-bottom plates, and the content (50 μL medium + spheroids) was transferred to the 96-well μclear plates. Before sonication, the 96-well μclear plates were sealed with paraffin titer top films (Electron Microscopy Sciences (Hatfield, PA, USA)) to prevent contamination of the wells and air bubble formation during FUS treatment. Water was placed on top of the foil above the transducers for coupling. The sealed well plate was slid into the yellow drawer until the aligned position with the transducer was reached, and where knobs ensured its ideal position. A visual inspection was performed to ensure that there is no air bubble between the well plate bottom and the membrane. The experiment proceeded with software settings for the desired parameters: 1st treatment parameter—sonication time: 90 s, power: 20% (ISPTA=2.95 W/cm2), signal repetition: 5 Hz (20 ms), duty cycle: 10%; 2nd treatment parameter—sonication time: 90 s, power: 40% (max) (ISPTA=~5.9 W/cm2), signal repetition: 5 Hz (20 ms), duty cycle: 10%. After sonication, the entire content of each well of 96-well μclear plate was transferred to corresponding wells of the U-bottom plate volume for further spheroid cultivation and follow-up. Cell culture medium with 5% DMSO of the positive control group was removed and refilled with ~200 μL/well fresh medium. The spheroids were then incubated for a further 48 and 96 h.
2.6. Microscopical Analysis of Spheroids
Morphological characteristics of formed spheroids were evaluated and recorded using microscopy (ZEISS Axio Observer, Carl Zeiss microscopy GmbH (Jena, Germany)) before and immediately after FUS treatment, and at 48 and 96 h after treatment. Pictures were taken with an AxioCam camera using ZEN v3.1 (blue edition) https://www.zeiss.com/microscopy/int/products/microscope-software/zen.html (accessed on 18 March 2020). Microscopy-acquired images of the spheroids were further processed and analysed with Java-based open-source software project ImageJ v1.53, https://imagej.nih.gov/ij/ (accessed on 10 May 2020), and MATLAB-based (© The MathWorks, Inc. (Natick, MA, USA)) open-source software AnaSP v1.4, https://sourceforge.net/projects/anasp/ (accessed on 8 May 2020) and ReViSP v2.2, https://sourceforge.net/projects/revisp/ (accessed on 8 May 2020).
2.7. Measurement of Cellular Metabolic Activity
A viability assay was performed at 48 and 96 h post-treatment to determine the metabolic activity of the cell spheroids. According to the manufacturers’ instructions, 150 μL of the medium was removed from each well of the 96-well clear F-bottom black plate (Greiner Bio-One GmbH (Frickenhausen, Germany)) and 50 μL of CellTiter-Glo® 3D Reagent (Promega GmbH (Madison, WI, USA)) was added. Spheroids were then incubated at room temperature for 30 min in the dark to stabilise the bioluminescent signal, which was then measured using the Synergy H1™ Hybrid Multi-mode plate reader (BioTek Instruments®, Inc. (Winooski, VE, USA)).
2.8. Determination of DNA Double-Strand Breaks (DSBs) in PC-3 Spheroids
DSBs determination was performed by the γH2A.X assay 1 and 24 h post-treatment. Cell culture medium was aspirated from each well, and PC-3 cancer cell line spheroids were washed twice with PBS prior to disaggregation into single-cell suspension with trypsin (Lonza Group Ltd. (Basel, Switzerland)). Disaggregated spheroids were pooled in falcon tubes (TPP Techno Plastic Products AG (Trasadingen, Switzerland)), neutralising the enzyme by adding a dedicated cell culture medium and centrifuging. After medium aspiration, cells were fixed with 4% formaldehyde at 37 °C for 10 min and chilled on ice for 1 min. Afterwards, the fixative was removed, and cells were washed twice with PBS. Cells were permeabilised with 90% methanol on ice for 30 min, and again washed twice with a non-specific antibody binding blocking agent—0.5% bovine serum albumin (BSA, Cell Signalling Technology (Danvers, USA), 100 μL/well) in PBS. After removing the block solution (0.5% BSA in PBS), cells were incubated with phospho-histone H2A.X (Ser 139) rabbit primary monoclonal antibody (Cell Signalling Technology (Danvers, MA, USA)) at a concentration of 1:500 diluted with block solution at room temperature for 1 h. Cells were washed twice with 0.5% BSA in PBS solution and incubated with secondary antibody (anti-rabbit IgG conjugated with Alexa Fluor® 488 fluorescent dye; Cell Signalling Technology (Danvers, MA, USA)) at a concentration of 1:1000 diluted with block solution at room temperature for 30 min in the dark. Cells were then washed twice with block solution, and cells were incubated with RNAse A (Sigma-Aldrich (St. Louis, MO, USA)) at 37 °C for 20 min. Finally, propidium iodide (Invitrogen by Thermo Fisher Scientific (Waltham, MA, USA)) was added and DSBs measurements were performed using Attune™ NxT flow cytometer (Thermo Fisher Scientific (Waltham, MA, USA)).
2.9. Statistical Analysis
Statistical analysis was performed using the statistical program Origin (Origin v6.0, https://www.originlab.com/origin (accessed on 21 June 2020) (Northampton, MA, USA)). All data of spheroid morphology analysis, cellular metabolic activity, spheroid cell hypoxia, and DNA double-strand breaks (γH2A.X) are expressed as means ± standard deviation of three independent experiments with three replicates, respectively. A one-way ANOVA test assessed the significance of the difference between the two mean values. A p-value ≤ 0.05 was considered to be statistically significant.
3. Results
3.1. Characteristics of the FUS In Vitro System
The obtained 2D and 1D pressure distribution fields are shown in Figure 2, in which the Peak to Peak pressure (scaled in dB) is plotted. The −6 dB focal widths in the x- and y-directions are 1.4 mm and 5.5 mm, respectively. This is below the pitch of the well plate (in both dimensions), confirming that there is little influence on individual wells by neighbouring transducers.
The corresponding average intensity for each well is shown in Figure 3a. In this experiment, the generator was driven with a power of only 1% so that the measurement can only be taken for the relative comparison of the intensity output from well to well and not for the absolute maximum intensity. A histogram of the average ISPTA is given in Figure 3b. Finally, for one transducer element (E3), the ISPTA in the focus was measured as a function of the power level defined in the “Cell Therapy Planning Tool”. In this analysis, a duty cycle (DC) of 100% was assumed such that the ISPTA equals the ISPPA.
3.2. Mitigation of Spheroid Growth
Microscopy analysis of the spheroid morphology showed decomposition of the PC-3 spheroids immediately after FUS treatment at 5.9 W/cm2. Interestingly, the spheroid reassembles 48 h post-treatment. Depending on time and thus an absence of nourishment, the PC-3 spheroids lost their integrity at 96 h in the untreated control group (Figure 4a). In contrast, the U87 spheroids had a more tightened spheroid structure, and growth of the spheroid structure was detected in the untreated group. Morphologically, there was no loss of spheroid structure apparent in the U87 cell line in all treatment groups. However, significant changes (p ≤ 0.05) of the measured spheroid area in U87 cancer cell line spheroid were noticed 48 and 96 h after treatment, with a reduction of the area to 246,387 µm2 and 219,976 µm2 after 5.9 W/cm2, respectively, as compared with the negative control. A slight swelling of the spheroid size immediately after sonication was determined in U87 spheroids (Figure 4b).
3.3. Diminished Spheroid Cell Viability
To evaluate the effects of FUS treatment on spheroids, cellular metabolic activity was checked 48 and 96 h post-treatment. The FUS treatment reduced the cell metabolic activity in both cell lines in an intensity-dependent manner (Figure 5). This reduction below 80% of viability was pronounced 48 and 96 h after the treatment compared with the untreated control (Figure 5). The glioblastoma cell line U87 showed a higher sensitivity with a statistically significant loss in ATP metabolism after treatment at 5.9 W/cm2 (Figure 5b) to 1.61 ± 2.45% (48 h) and 0.70 ± 0.94% (96 h) in U87 cells (Figure 5b).
3.4. FUS Enhanced Spheroid Cell DNA Damage
The γH2A.X assay was performed 1 and 24 h post-treatment to explore the potential of FUS to affect DNA repair. Flow cytometer analysis of dissociated PC-3 spheroids revealed a significant increase of fluorescence intensity (p ≤ 0.05) 24 h post-treatment with FUS at an intensity of ~5.9 W/cm2 and 5% DMSO groups, 22.1 ± 0.18% and 22.62 ± 1.49%, respectively (Figure 6). No significant increase of DNA double-strand breaks after FUS treatment was observed 1 h after treatment.
4. Discussion
While most of the studies evaluating the effect of FUS treatment in vitro utilise the 2D model, with the cancer cells distributed as a monolayer, our research aimed to test a new developed FUS in vitro system for high-throughput sonication as a starting point to establish the effect of FUS on the tumour entity, using 3D tumour culture. The difference in sphere-forming potential of the selected cancer cell lines was revealed with the predisposition of GBM (U87) spheroid to be regular in shape constituting tight cell-cell adhesions, while prostate cancer (PC-3) resembled more of a roundly-arranged, still irregular, tumour cell aggregate. This peculiar characteristic might be the reason behind the noticeable, but statistically insignificant, reduction of the PC-3 spheroid size after FUS treatment in all experiments. However, it significantly manifested in the case of U87 spheroid, suggesting that the response of the specific cancer cell’s type cluster to the acoustic constrain might be determined by its biophysical properties. This phenomenon was described in previous studies [22], indicating heightened resistance of spheroids over the standard flask cultures to radiation and chemotherapeutics, an aspect articulating the necessity to modify the research design with various cancer types in analogous future experiments.
Since our newly developed FUS in vitro system (2nd gen. cell applicator) was applied and no existing studies for FUS on spheroids were available, the system’s setup and FUS parameters needed to be established first. For acoustic characterisation of the FUS applicator, various measurements were performed by our partner, Fraunhofer IBMT, the manufacturer of the system. The results stated that the intensity distribution was relatively homogeneous within distinct well outliners, while ISPTA varied with the increase of the power setting to its maximum (40%) starting from the power at the level of 12%, with the need of its (ISPTA) extrapolation. It resulted in the approximation of the 2nd treatment parameter to ~5.9 W/cm2, which may not reflect the identical acoustic wave intensity distribution across all wells, which is the greatest limitation of this FUS setup. Adaption of the FUS intensity parameter sets (~3–6 W/cm2) was cerebrated depending on a recently conducted study [23], revealing that using a low-intensity pulsed ultrasound (LIPUS) with acoustic intensity similar to that of diagnostic levels, pulse duration of greater than 10 ms induces cyto-disruption. Intriguingly, what was noticed in this study was the expansion in the size of U87 spheroids right after the FUS treatment. It may be questioned whether the acoustic exposure on the spheroids evoked dismantling of its cell-cell architecture leading to its cyto-disruptive collapse and whether the same phenomenon may occur in more solid conditions, where tumour cells are embedded within surrounding tissue. Indeed, future in vivo studies are needed to test this hypothesis.
CellTiter-Glo® 3D Cell Viability Assay [17] using ATP-dependent luminescence signal quantification to determine the number of viable cells in 3D cell culture was the best method in this study to determine the metabolic activity in physical treatments. This ATP detection assay is currently the most sensitive and has minimal interferences [24]. While FUS treatment induced a significant reduction of viable cells in both tumour spheroids over time with plate reader estimation, specialised techniques of spatial-temporal quantification [25] might bring more relevant data. The significant reduction of the spheroid area in U87 spheroids is congruent with the significant loss in cellular ATP metabolism showing the collapse of U87 cells. Furthermore, the short loss of the spheroid structure of PC-3 immediately after FUS seems to have no impact on the ATP metabolism, only on the binding, e.g., tight junctions, between the cells. The detected decrease of ATP content in PC-3 was not reflected in the observed reduction of the spheroid area within the 96 h time frame.
Measurement of the DNA double-strand breaks, which can be used as a marker for treatment-induced cell death in tumours [16], showed an increased fluorescence intensity of immunolabelled γH2A.X in cells of PC-3 spheroids after FUS treatment with the intensity of ~5.9 W/cm2, revealing a damaging effect of pulsed FUS, like the cytotoxic agent 5% DMSO used in the positive control group. Our study proved, in 24 h post-treatment, that the FUS treatment accelerated notably the DNA damage in PC-3 spheroids, leading to cell death. On the other hand, U87 spheroids could not be completely dissociated with the methods (trypsinisation) used in the case of PC-3 spheroids.
5. Conclusions
The present work highlighted the potential of FUS application for the treatment of tumour spheroids. It was demonstrated that low-intensity pulsed focused ultrasound reduced spheroid growth metabolic activity and increased DNA double-strand breaks in particular cancer cell lines used in this study. The results suggest that FUS treatment in LIPUS mode harm cancer cells and the modality itself has great potential to be further investigated in vivo.
Acknowledgments
We wish to acknowledge the team of Saxonian Incubator for Clinical Translation (SIKT), Leipzig University, Leipzig, Germany for making the laboratory rooms and infrastructure available to us. Many thanks to Ina Patties and Annegret Glasow for donation of the cell line U87.
Author Contributions
Conceptualisation, L.L.; methodology, A.K.; cell culture: X.Z.; concept and design of FUS device: S.T. and F.-J.B.; device building: F.-J.B.; characterisation: M.F. and F.-J.B.; statistics: A.K.; writing—original draft preparation, L.L.; writing—review and editing, A.M.; supervision, A.M. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Not available.
Informed Consent Statement
Not applicable.
Data Availability Statement
Not applicable.
Conflicts of Interest
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
Figure 1 FUS treatment system and generation process of tumour spheroids. (a) Scheme of the in vitro FUS transducer and spheroid treatment. (b) The FUS in vitro system was specially designed for 96-well cell culture plates, with 32 transducers at a frequency of 1.1 MHz and water cooling. (c) The layer of the spheroid formation process for dedicated cancer cell lines. The prostate cancer cell line PC-3 and glioblastoma U87 were cultured in a dedicated medium using a liquid overlay technique to generate tumour spheroids.
Figure 2 Assessment of the single-transducer pressure distribution in 2D and 1D. (a,b) show two 2D sound fields (Peak-to-Peak pressure in dB is plotted) acquired in a water tank measurement to assess the extent of the focal area. (c) Shows a lateral cross-section through the depth of highest pressure.
Figure 3 Assessment of the applicator performance. An XY sound field scan was performed in front of the cell applicator. For (a), the acoustic intensity ISPTA was averaged over the surface of one well of the 96-well plate in front of each transducer element. (b) A histogram allowing assessing the homogeneity of the intensity output. (a,b) were performed with a power setting of 1% and were made to compare the relative performance of the different wells, not the absolute intensities. (c) The ISPTA as a function of the power setting that can be user-defined using the “Cell Therapy Planning Tool”.
Figure 4 FUS reduced spheroid size and led to a loss of integrity. (a,b) Representative microscopy images showing alterations in spheroid morphology and (c,d) brightfield images of FUS-treated spheroids; the corresponding 3D reconstructions were obtained using ReViSP, http://sourceforge.net/p/revisp/ (accessed on 8 May 2020). (e/f) Bar chart representation of changes in the spheroid area before, immediately, 48, and 96 h after FUS treatment at an intensity of 2.95 and 5.9 W/cm2, + control: +5% DMSO. Data analysis was carried out by one-way ANOVA. * Significantly different from the untreated group. (p ≤ 0.05). # 4th day spheroid formation; * in the ‘untreated’ group only, well-to-well (re-)transfer of spheroid was carried out to equalise the impact. PC-3 cancer cell line: Scale bar = 200 μm. U87 cancer cell line: Scale bar = 100 μm. n = 9.
Figure 5 FUS diminished spheroid metabolic activity. ATP content of PC-3 (a) and U87 (b) spheroids was assessed using CellTiter-Glo® 3D Cell Viability assay 48 and 96 h after treatment, showing the reduction of cell metabolic activity. Data sets were normalised to the untreated control group (100%), while data analysis was carried out by one-way ANOVA. * Significantly different from the untreated group (p ≤ 0.05). n = 9.
Figure 6 FUS treatment enhanced the number of DNA double-strand breaks after 24 h. (a) Representation of γH2A.X percentages as a function of cell count determined by gating histograms derived from dissociated PC-3 spheroids with flow cytometric analysis 1 and 24 h post-treatment. (b) Overlayed flow cytometry images (c) and quantified results show an increasing number of γH2A.X positive cells 24 h after 5.9 W/cm2. Data analysis was carried out by one-way ANOVA. * Significantly different from negative control (p ≤ 0.05).
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Carina V. Costa V. Sartori M. Bellavia D. De Luca A. Raimondi L. Fini M. Giavaresi G. Adjuvant Biophysical Therapies in Osteosarcoma Cancers 2019 11 348 10.3390/cancers11030348 30871044
2. Zhang X. Landgraf L. Bailis N. Unger M. Jochimsen T.H. Melzer A. Image-guided High-Intensity Focused Ultrasound, A Novel Application for Interventional Nuclear Medicine? J. Nucl. Med. 2021 62 1181 1188 10.2967/jnumed.120.256230 34088775
3. Imamura Y. Mukohara T. Shimono Y. Funakoshi Y. Chayahara N. Toyoda M. Kiyota N. Takao S. Kono S. Nakatsura T. Comparison of 2D- and 3D-culture models as drug-testing platforms in breast cancer Oncol. Rep. 2015 33 1837 1843 10.3892/or.2015.3767 25634491
4. Hirschhaeuser F. Menne H. Dittfeld C. West J. Mueller-Klieser W. Kunz-Schughart L.A. Multicellular tumor spheroids: An underestimated tool is catching up again J. Biotechnol. 2010 148 3 15 10.1016/j.jbiotec.2010.01.012 20097238
5. Friedrich J. Ebner R. Kunz-Schughart L.A. Experimental anti-tumor therapy in 3-D: Spheroids--old hat or new challenge? Int. J. Radiat. Biol. 2007 83 849 871 10.1080/09553000701727531 18058370
6. Harada H. How can we overcome tumor hypoxia in radiation therapy? J. Radiat. Res. 2011 52 545 556 10.1269/jrr.11056 21952313
7. Mandavilli B. Chen A. Hu Y. Hypoxia measurements in live and fixed cells using fluorescence microscopy and high content imaging Molecular Biology of the Cell American Society for Cell Biology Bethesda, MD, USA 2017
8. Nii T. Makino K. Tabata Y. Three-Dimensional Culture System of Cancer Cells Combined with Biomaterials for Drug Screening Cancers 2020 12 2754 10.3390/cancers12102754
9. Loessner D. Rockstroh A. Shokoohmand A. Holzapfel B.M. Wagner F. Baldwin J. Boxberg M. Schmalfeldt B. Lengyel E. Clements J.A. A 3D tumor microenvironment regulates cell proliferation, peritoneal growth and expression patterns Biomaterials 2019 190–191 63 75 10.1016/j.biomaterials.2018.10.014
10. Asadian S. Piryaei A. Gheibi N. Aziz Kalantari B. Reza Davarpanah M. Azad M. Kapustina V. Alikhani M. Moghbeli Nejad S. Keshavarz Alikhani H. Rhenium Perrhenate (188ReO4) Induced Apoptosis and Reduced Cancerous Phenotype in Liver Cancer Cells Cells 2022 11 305 10.3390/cells11020305 35053421
11. Nii T. Makino K. Tabata Y. A Cancer Invasion Model Combined with Cancer-Associated Fibroblasts Aggregates Incorporating Gelatin Hydrogel Microspheres Containing a p53 Inhibitor Tissue Eng. Part C Methods 2019 25 711 720 10.1089/ten.tec.2019.0189 31621504
12. Wang P. Dreger M. Madrazo E. Williams C.J. Samaniego R. Hodson N.W. Monroy F. Baena E. Sanchez-Mateos P. Hurlstone A. WDR5 modulates cell motility and morphology and controls nuclear changes induced by a 3D environment Proc. Natl. Acad. Sci. USA 2018 115 8581 8586 10.1073/pnas.1719405115 29987046
13. Fontoura J.C. Viezzer C. Dos Santos F.G. Ligabue R.A. Weinlich R. Puga R.D. Antonow D. Severino P. Bonorino C. Comparison of 2D and 3D cell culture models for cell growth, gene expression and drug resistance Mater. Sci. Eng. C Mater. Biol. Appl. 2020 107 110264 10.1016/j.msec.2019.110264 31761183
14. Ho V.H. Smith M.J. Slater N.K. Effect of magnetite nanoparticle agglomerates on the destruction of tumor spheroids using high intensity focused ultrasound Ultrasound Med. Biol. 2011 37 169 175 10.1016/j.ultrasmedbio.2010.09.007 21084159
15. Yildirim A. Shi D. Roy S. Blum N.T. Chattaraj R. Cha J.N. Goodwin A.P. Nanoparticle-Mediated Acoustic Cavitation Enables High Intensity Focused Ultrasound Ablation Without Tissue Heating ACS Appl. Mater. Interfaces 2018 10 36786 36795 10.1021/acsami.8b15368 30339360
16. Bonner W.M. Redon C.E. Dickey J.S. Nakamura A.J. Sedelnikova O.A. Solier S. Pommier Y. γH2AX and cancer Nat. Rev. Cancer 2008 8 957 967 10.1038/nrc2523 19005492
17. Zanoni M. Piccinini F. Arienti C. Zamagni A. Santi S. Polico R. Bevilacqua A. Tesei A. 3D tumor spheroid models for in vitro therapeutic screening: A systematic approach to enhance the biological relevance of data obtained Sci. Rep. 2016 6 19103 10.1038/srep19103 26752500
18. Prentice P. Cuschieri A. Dholakia K. Prausnitz M. Campbell P. Membrane disruption by optically controlled microbubble cavitation Nat. Phys. 2005 1 107 110 10.1038/nphys148
19. Russell W. Burch R.L. The Principles of Humane Experimental Technique Methuen London, UK 1959
20. Kijanska M. Kelm J. In vitro 3D Spheroids and Microtissues: ATP-based Cell Viability and Toxicity Assays Assay Guidance Manual Markossian S. Grossman A. Brimacombe K. Arkin M. Auld D. Austin C.P. Baell J. Chung T.D.Y. Coussens N.P. Dahlin J.L. Eli Lilly & Company and the National Center for Advancing Translational Science Bethesda, MD, USA 2004
21. Froehlich K. Haeger J.D. Heger J. Pastuschek J. Photini S.M. Yan Y. Lupp A. Pfarrer C. Mrowka R. Schleussner E. Generation of Multicellular Breast Cancer Tumor Spheroids: Comparison of Different Protocols J. Mammary Gland Biol. Neoplasia 2016 21 89 98 10.1007/s10911-016-9359-2 27518775
22. Curran S. Vantangoli M.M. Boekelheide K. Morgan J.R. Architecture of Chimeric Spheroids Controls Drug Transport Cancer Microenviron. 2015 8 101 109 10.1007/s12307-015-0171-0 26239082
23. Mittelstein D.R. Ye J. Schibber E.F. Roychoudhury A. Martinez L.T. Fekrazad M.H. Ortiz M. Lee P.P. Shapiro M.G. Gharib M. Selective ablation of cancer cells with low intensity pulsed ultrasound Appl. Phys. Lett. 2020 116 013701 10.1063/1.5128627
24. Riss T.L. Moravec R.A. Niles A.L. Duellman S. Benink H.A. Worzella T.J. Minor L. Cell Viability Assays Assay Guidance Manual Markossian S. Grossman A. Brimacombe K. Arkin M. Auld D. Austin C.P. Baell J. Chung T.D.Y. Coussens N.P. Dahlin J.L. Eli Lilly & Company and the National Center for Advancing Translational Sciences Bethesda, MD, USA 2004
25. Hou Y. Qin Z. Fan K. Xu Y. Huang X. Combination therapeutic effects of high intensity focused ultrasound and Metformin for the treatment of adenomyosis Exp. Ther. Med. 2018 15 2104 2108 10.3892/etm.2017.5601 29434812
|
PMC009xxxxxx/PMC9099906.txt |
==== Front
Int J Mol Sci
Int J Mol Sci
ijms
International Journal of Molecular Sciences
1422-0067
MDPI
10.3390/ijms23094760
ijms-23-04760
Review
Insight into the Role of Psychological Factors in Oral Mucosa Diseases
https://orcid.org/0000-0001-6529-6226
Guo Yuexin 1
Wang Boya 2
Gao Han 3
He Chengwei 3
Hua Rongxuan 4
Gao Lei 5
Du Yixuan 1
Xu Jingdong 3*
Luparello Claudio Academic Editor
Ferreira Rita Academic Editor
1 Department of Oral Medicine, Basic Medical College, Capital Medical University, Beijing 100069, China; [email protected] (Y.G.); [email protected] (Y.D.)
2 Department of Clinical Medicine, Peking University Health Science Center, Beijing 100081, China; [email protected]
3 Department of Physiology and Pathophysiology, Basic Medical College, Capital Medical University, Beijing 100069, China; [email protected] (H.G.); [email protected] (C.H.)
4 Department of Clinical Medicine, Basic Medical College, Capital Medical University, Beijing 100069, China; [email protected]
5 Department of Bioinformatics, College of Bioengineering, Capital Medical University, Beijing 100069, China; [email protected]
* Correspondence: [email protected]; Tel./Fax: +86-10-8391-1469
26 4 2022
5 2022
23 9 476003 4 2022
23 4 2022
© 2022 by the authors.
2022
https://creativecommons.org/licenses/by/4.0/ Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
With the development of psychology and medicine, more and more diseases have found their psychological origins and associations, especially ulceration and other mucosal injuries, within the digestive system. However, the association of psychological factors with lesions of the oral mucosa, including oral squamous cell carcinoma (OSCC), burning mouth syndrome (BMS), and recurrent aphthous stomatitis (RAS), have not been fully characterized. In this review, after introducing the association between psychological and nervous factors and diseases, we provide detailed descriptions of the psychology and nerve fibers involved in the pathology of OSCC, BMS, and RAS, pointing out the underlying mechanisms and suggesting the clinical indications.
psychology
oral mucosa
disease
interactions
pathology
==== Body
pmc1. Introduction
The oral cavity is the entrance of our digestive system, and accounts for the generation of our taste [1]. Despite its small size, the tissue structure and nerve innervations of the oral cavity are complex and form close interactions with the brain, as well as the whole body [2]. The food we eat can exert an impact on our emotions, and lesions in the oral mucosa can seriously affect our daily life. Although several diseases of the oral mucosa are common in a wide number of people of different ages and areas, their detailed pathologies are largely unknown [3]. This might limit the process of drug development, as well as the treatment effects. Based on our current understanding, several factors such as dietary habits and food stimulation might account for disease genesis and pathology. Other factors including immune reactions and psychological conditions have also been raised, although knowledge of them is largely limited. In the meantime, there are distinctions between clinicians and patients considering the definition of treatment success. Most patients consider it as the relief of symptoms and the recovery of their functions, while clinicians may put more emphasis on tissue repair and the radical cure of disease course with the assistance of histological and imaging technology [4]. This difference may increase the difficulty in permanent cures and the long-term maintenance of treatment which has been a growing quandary for many clinicians.
Psychological factors are significant mediators of several diseases’ pathologies (as shown in Table 1, Table 2 and Table 3 for their relations, respectively) and form close interactions with multiple tissues and systems. Although they are most recognized by psychologists and utilized by means of mental consultation, their biological foundations are also recognized by biologists and doctors, promoting their use in drug development and clinical treatment. Known as the bio–psycho–social medical model first put forward by George Engel in 1977 [5], psychology has received more and more attention in the genesis, progression, and treatment of the medical process. As a complex system regulated by various areas in the human brain, the psychology and mental health of people are in direct contact with the nervous system. Considering the broad functions of neurotransmitters (NT) and the wide distribution of their receptors, not only could they form close interactions with body immunity by the activation or inhibition of immune pathways, but also regulate the functions of normal tissues with the initiation of physiological reactions [6]. In the meantime, they are also essential in the regulation of disease pathology, as an increasing number of diseases have been found to be associated with psychological origins or manifestations, such as gastritis [7] and breast cancer [8]. However, despite the broad and deep investigations of interactions between the nervous system and other organs, those of the oral mucosa are relatively scarce with no comprehensive analysis.
ijms-23-04760-t001_Table 1 Table 1 Oral symptoms involved in psychological problems.
Psychology Problems Oral Symptoms Refs.
Olanzapine-induced anticholinergic toxicity Dry oral mucosa [9]
Alzheimer’s disease (AD) Tau protein in oral mucosa [10]
Bulimia and anorexia nervosa Abrasion of teeth enamel [11]
Epidermoid cysts of the central nervous system Similar to symptoms in oral mucosa [12]
Neurocutaneous syndromes (and other diseases associated with DNA repair) Teeth and oral mucosa lesions [13]
Structural changes in innervations in oral cavity Loss of sense [14]
ijms-23-04760-t002_Table 2 Table 2 Oral diseases with psychological abnormalities.
Oral Diseases Psychological Abnormalities Oral Manifestations Refs.
Oral squamous cell carcinoma (OSCC) Higher α1 adrenergic receptors Oral ulcers and lumps, pain feeling [15]
Primary Sjögren’s syndrome Depression and anxiety More frequent oral lesions; negative impact on life quality [16]
Burning mouth syndrome (BMS) Structural and functional deficits within the nervous system Burning feeling without an obvious cause [17]
Herpes simplex encephalitis Fatal disease of the central nervous system Painful blisters or open sores (ulcers) [18]
Oral mucosa cancers Antitumor drugs improve psychological symptoms Oral pumps and lesions [19]
Primary burning mouth syndrome PNS involvement Burning feeling in mouth [20]
Recurrent aphthous stomatitis Hypofunction of the sympathetic nervous system Painful round shallow ulcers [21]
Inflammatory stimulation of the oral mucosa Activation of microglial cells Oral mucosal inflammation [22]
Poliovirus Affects the anterior horn motor neurons of the spinal cord causing paralysis [23]
Xerostomia Autonomic nervous system imbalance Dryness of the oral mucosa [24]
Burning mouth syndrome Decreased or modified steroid synthesis Burning feeling in mouth [25]
Lingual conical papillae Alterations of different kinds of neurons Bumps and rough tongue [26]
Lichen planus and lichenoid reactions Loss of PNS fibers Asymptomatic white reticular striae to painful erythema and erosions [27]
Heat stimulation Large primary neurons responding to high-threshold noxious heat are abundant in the tooth pulp Altered pain [28]
Oral mucosa continuous remodeling Sensory nervous apparatus involvement Leaky epithelial barrier, a fibrotic lamina propria, the release of inflammatory mediators, and the recruitment of immune infiltrate [29]
Oral dysesthesia (OD) Soft tissue grafts Merkel cells and permanent dysesthesia in the oral mucosa [30]
ijms-23-04760-t003_Table 3 Table 3 Diseases with both the psychological problems and oral manifestations.
Diseases Psychological Problems Oral Manifestations Refs.
Congenital herpes simplex CNS infection Oral infection [31]
Enterovirus A71 (EV-A71) Viral antigens/RNA in the CNS Viral antigens/RNA in the squamous epithelia of the oral cavity [32]
Verrucous lesions Malocclusions in CNS Verrucous lesions affect the oral mucosa (rare) [33]
Cryptococcosis Presentations in the CNS Excision of nodules in the oral mucosa assists recovery [34]
Paracoccidioidomycosis (PCM) Involvement of CNS Common infected symptoms [35]
Herpes simplex virus type 1 (HSV-1) Clinical diseases in CNS Vesicular lesions of the oral mucosa [36]
Cowden syndrome Similar CNS symptoms with that in the oral mucosa Multiple hamartomatous neoplasms of the oral mucosa [37]
Lipoid proteinosis (LP) Involvement of CNS Yellow-white plaques on oral mucosa [38]
Enterovirus 71 (EV71) (hand-foot-and-mouth disease (HFMD)) brainstem encephalitis Vesicular lesions on oral mucosa [39]
Wilson disease (WD) multi-organ manifestations involve the nervous system Repeated oral candidiasis [40]
HSV-1 Transmission in the CNS Infection in the oral mucosa [41]
Tuberous sclerosis Hamartoma formation in the nervous system Hyperpigmented and hypopigmented macules affecting the oral mucosa [42]
Bacillary angiomatosis CNS BA lesions in the oral mucosa [43]
Cowden disease (CD) CNS manifestations Normal oral involvement [44]
Dettol liquid CNS symptoms Oral involvement [45]
Sweet’s diseases [46]
HSV-1 infection Vagus nerve transmission Oral manifestations [47]
Adamantiades–Behçet disease Lesions of ulcerating systemic vasculitis in the CNS Oral manifestations [48]
In this review, after briefly introducing the broad effect of psychological molecules in the whole body, we put our emphasis on the roles of psychological factors in the generation and progression of oral mucosa injuries, especially those of OSCC, BMS, and RAS. Besides this, we highlight the similarities in their roles in pathogenesis, summarizing the related factors and providing the general interactive patterns. Furthermore, we analyze the current understanding of the interactions of oral diseases and psychological factors and put forward the directions for future research. These would be of great use considering the broad roles of psychological factors in the whole body, and would provide novel and practical suggestions for clinical treatment.
It has been recognized that receptors for adrenalin and its derivatives are widely distributed in a variety of tissues and organs, and employ multiple functions all over the body [49]. Their basic roles lie in the activation of neurons and the initiation of neuro-excitement. However, they are also responsible for muscle contraction and gland secretion [50,51]. Further investigations have also shown their roles in immune cell activation and proliferation, and the transmission of typical feelings [52]. Common neurotransmitters are responsible for these functions, including adrenalin and several amino acids. These regulations form a close regulatory network in our body which is only partly understood due to the limitations of our current knowledge and techniques.
2. Roles of Psychological Factors in Oral Mucosal Diseases
2.1. Psychological Factors in the Progress of OSCC
OSCC is one of the most malignant tumors around the world, with an overall 5-year survival rate of approximate 50–60% [53]. Common symptoms of OSCC include vomiting and swallowing abnormalities, and seriously affect patients’ normal lives [54]. Although smoking has been confirmed as the most dangerous factor associated with OSCC [55], multiple factors have been corroborated in association with OSCC pathology and a significantly higher level of distress thermometer (DT) score has been found in OSCC patients [56]. Several biomarkers could be utilized for the evaluation of OSCC progression and malignancy, some of which are morphological characteristics of cells such as tumor-infiltrating lymphocytes (TILs), as well as immune checkpoints such as PD-L1, FKBP51, B7-H4, B7-H6, ALHD1, IDO1, and B7-H3 [57]. Other studies have shown the potential of osteopontin and fractal dimension in predicting diagnoses and choosing the proper treatment in OSCC [58]. Further analysis has found an increase in the expression of α1 adrenergic receptors (α1-ARs) in serum and saliva lesions in oral mucosa [15], mainly accounting for carcinoma cell proliferation and translocation. Concomitant with this, exposure to nicotine has been found to induce OSCC proliferation [59], which has also been corroborated in other carcinoma models [60]. In the meantime, various studies have also shown the stress-related upregulation of α1-ARs in both human beings and animal models [61,62]. Multiple pathways could be activated via increasing the expression of α1-ARs such as the mitogen-activated protein kinase (MAPK) pathway, cAMP metabolism, and phospholipase D (PLD) and A2 in different cells, and Nishioka et al. confirmed the activation of the EFGR/ERK/AKT pathway in the course of the disease [59]. However, further analysis is also required concerning the potential roles of other nicotine-related pathways, and their long-term detection and examination are also needed.
Another significant difference involving the nervous system is the increase in salivary cortisol level, indicating the disturbance of the hypothalamus pituitary axis [63]. This is also related to the norepinephrine level, as Lee et al. found a higher cortisol level to be induced in children with more variability after stress and following the secretion of norepinephrine [64]. In the meantime, both α1 adrenergic receptors and cortisol account for the significant anti-inflammation roles of NSAIDS [65], and the induction of hydrocortisone results in alterations of the cardiac adrenergic receptor density which are associated with the functional outcome in rats [66]. Further analysis has confirmed a similar effect of hydrocortisone in increasing the sensitivity to α1 adrenergic receptors in humans after hemorrhagic shock [67]. However, an induction of 0.1 mM hydrocortisone prohibited the decrease in 3H-clonidine binding sites with no impact on 3H-yohimbine or the 3H-prazosin binding sites in rat organ culture [68], and Prazosin pretreatment did not affect the basal or peak plasma cortisol level during hypoglycemic stress in human beings [69]. These studies show the complex interactions between α1 adrenergic receptors and the cortisol level which could be affected in dose- or tissue-dependent ways. Further investigations would be of great help in uncovering the detailed interactions between them and their potential use in clinical treatment.
Intensive investigations into neurotransmitters and other neurokinins have found that neurokinin B (NKB) can stimulate the proliferation of OSCC, probably via the pAkt/pmTOR signaling pathway [70]. Additionally, this is concomitant with the different expression of the neurokinin 3 receptor (NK-3R) in both the central and peripheral nervous systems in OSCC [71]. In the meantime, Obata et al. found a higher expression of tachykinin 3 (TAC3) in the invasion front of oral squamous cell carcinoma in bone matrix; TAC3 (Tachykinin-3) is probably released by the peripheral sensory nerves and contributes to tumor progression [71]. Gamma-amino butyric acid (GABA), a negative neurotransmitter in the nervous system, is also found to promote the proliferation of OSCC via the activation of the p38 MAPK and inhibition of the JNK/MAPK signaling pathways [72]. Concomitant with this, glutamate acid decarboxylase 1 is reported to promote the metastasis of human oral cancer by β-catenin translocation and MMP7 activation [73]. PrPC (cellular prion protein), well known for its roles in neurodegenerative diseases, has also been corroborated for the ability to resist tumor necrosis factor α (TNFα) apoptosis in OSCC, colonic, and the renal adenocarcinoma ACHN [74]. Say et al. further showed this enhancement in the expression of PrPC resulted from stable snRNA knockdown and could lead to the inhibition of glycosylation [75]. Semaphorin7A, a chemotactic factor in neurogenesis as well as a significant immunomodulator, has been found to promote tumoral growth and metastasis in human oral cancer by activating the G1 cell cycle and matrix metalloproteases [76]. Natriuretic peptide systems, which are important regulators of the nervous system and in controlling the secretion of saliva, also show abnormalities in OSCC probably related to the disturbance of the water-salt imbalance [77]. Taken as a whole, these molecules may work together and form a series of cascade reactions that contribute to perineural invasion (PNI), which might facilitate the formation of a tumor microenvironment (TME) and tumor metastasis [78,79].
With the development of gene research, more investigations have reported the significant roles of ncRNA in regulating the expression of genes encoding proteins associated with nerve-related tumor progression. LncRNA could induce the activation of the NF-κB/STAT3 pathway and the secretion of proinflammatory cytokines [80], and miRNAs are suggested as markers in OSCC, as the overexpression of miR-181a and miR-181b may increase lymph-node metastasis and vascular invasion and is associated with a poor prognosis in OSCC patients, while the downregulation of miR-125b has been found in OSCC cell lines [81]. Despite the lack of genetic level studies nowadays, more investigations would be of great help in understanding the mechanistic interactions with the development of molecular biology.
2.2. Burning Mouth Syndrome (BMS)
Known as a common disease affecting almost every 1 in 10,000 people, BMS often occurs with no typical symptoms. However, many patients show a loss of/reduced taste perceptions during the disease course which is often considered to be in close connection with psychological factors, such as depression and anxiety [82]. Simultaneously, the severity of its syndrome is also closely correlated with patients’ mental condition [83]. Meta-analysis also shows lower cold detection and pain thresholds, as well as higher warm detection and pain thresholds at the tongue and lip of BMS patients compared with healthy participants. However, no significant differences in mechanical detection and pain thresholds were found [84]. Superficial biopsies of the lateral aspect of the anterior two-thirds of the tongue showed a significantly lower density of epithelial nerve fibers with diffusing morphological changes that reflect axonal degeneration [85]. In the meantime, the expression of the transient receptor potential vanilloid 1 (TRPV1) and P2 × 3 receptor are upregulated and account for the generation of pain feeling [86]. Personality analysis also confirmed a tendency of BMS patients to be more introversive and unstable compared to the healthy groups [87]. In addition, their plasma norepinephrine level was also higher which might account for the fiercer reactions after stimuli. BMS is also known as a disease affecting various parts of the body and is witnessed as a manifestation of diabetes. However, despite the small distinction of several prognostic indexes between the placebo spray (PS) and artificial saliva (AS), no significant difference was found according to statistical methods (the indexes include salivary flow rate, antioxidant capacity of saliva, and ultrasound variables) [88].
Simultaneously, studies focusing on abnormalities of the central nervous system have also shown an altered structural connectivity of pain-related brain networks using graph analyses of probabilistic tractography [89]. Further research has shown that six out of eight regions of the gray were affected (anterior and posterior cingulate gyrus, lobules of the cerebellum, insula/frontal operculum, inferior temporal area, primary motor cortex, dorsolateral pre-frontal cortex (DLPFC)). In the anterior cingulate gyrus, the lobules of the cerebellum, the inferior temporal lobe, and the DLPFC, pain intensity is correlated with gray matter concentration [90]. Concomitant with the reduced volume of gray matter, cerebral blood flow (CBF) also decreased which is related to the pain severity [91]. Apart from the alterations mentioned above, Jääskeläinen et al. found a hypofunction of dopaminergic neurons in the basal ganglia in a subgroup of people. This suggests the complexity of this regulatory network and the need for more detailed examination before clinical practice [92]. Further investigations have shown a more comprehensive regulation of pain feeling via the dopaminergic nervous system, as dopamine can act on striatal dopamine D2/D3 receptors and serotonin on cortical 5-HT1A and 5-HT2A receptors and affect top-down pain regulation in humans [93].
Besides the manifestations within the central nervous system, research has also revealed that alterations in the oral mucosa are in accordance with several changes in the central nervous system. The loss of peripheral nervous fibers is concomitant with the lower threshold of pain detection which has suggested the roles of neuroinflammation in oral nerve defects [94] and is in line with the positive effect of benzodiazepines in BMS patients. Based on the analysis above, although the detailed and comprehensive mechanisms underlying BMS have not been fully characterized, the roles of several typical molecules have been confirmed. This has laid the foundation for drug development for conditions such as idiopathic glossodynia which could exert an altered excitability in the trigeminal nociceptive pathway in the peripheral and/or central nervous systems, and enhance GABA concentration while decreasing glutamate functions at postsynaptic sites. Clinical uses of several such drugs, for example, topiramate, have been evaluated and received good feedback in some patients [95]. Additionally, neurophysiological evaluation as well as psychological tests for BMS found that sufferers were characterized by a mild sensory and autonomic small fiber neuropathy with concomitant central disorders [96]. A frequency analysis of the heart rate variability (HRV) revealed autonomic instability and the tracking of these changes corrected with stellate ganglion near-infrared irradiation (SGR) also provided a novel evaluation for follow-up survey, as well as a therapeutic measurement for BMS patients [97]. Simultaneously, based on the findings that chronic anxiety could result in a dysregulation of the adrenal cortex’s production of steroids, the decreased production of some major precursors for neuroactive steroid synthesis and the resulting brisk alteration of neuroactive steroid production might contribute to the neurodegenerative alterations of small nerves fibers of the oral mucosa and/or brain areas involved in oral somatic sensations [25]. These neuropathic changes could probably become irreversible with disease progression and result in burning pain, dysgeusia, and xerostomia associated with stomatodynia from thin nerve fibers. Considering the possible complications of BMS, Varvet et al. also suggested the potential roles of the dopaminergic nervous system in a case of Lewy body [98]. Another research study has also reported the defects of dopaminergic neuron in the central nervous system in BMS. This offers the potential for interventions targeting pathophysiological mechanisms and related molecules [92]. Moreover, researchers also made the comparation between BMS and Parkinson’s disease due to the similar intensity of autonomic nervous system dysfunction. A significant impairment of both the sympathetic and parasympathetic nervous systems was found with the maintenance of sympathetic/parasympathetic balance in BMS patients [99]. This indicates the probable existence of common regulatory mechanisms between them which are waiting for further elucidation. Another case of a diabetic patient revealed the roles of trigeminal nociceptive pathways in BMS pathology [100]. As an essential regulator of maxillofacial sensation, trigeminal sensory fibers have received much attention and are worth the research considering the wide innervated regions and broad functions. Studies concerning the roles of oral immunity in BMS have also shown the wide involvement of the immunoendocrine system which could mainly and specifically account for depression in BMS. Although the activation of the hypothalamic–pituitary–adrenal axis and the sympathetic nervous system are predominantly due to psychological stress and are not specific to BMS, the immunoendocrine mechanism in BMS is worth attention regarding its broad roles in a variety of diseases [101].
Tricyclic antidepressants (nortriptyline and amitriptyline), serotonin-noradrenaline reuptake inhibitors (SNRIs) (duloxetine and milnacipran), and antiepileptic drugs including potential-dependent calcium channel α2δ subunit ligands (gabapentin and pregabalin) which are regarded as the first-choice drugs for neuropathic pain under current standards. However, their effects are not the same for different patients in various pathological conditions. Additionally, pregabalin might be a novel option for BMS patients who are resistant or not responsive to SNRIs due to its different activating mechanism [102]. Despite the use above, the general effectiveness and safety of psychological treatment in BMS have not been fully confirmed, as Eccleston et al. found no significant variation between the control group and those treated with psychological interventions [103]. Other research works focusing on neuroprotective steroids are still in the trial period with no unified conclusions [104].
Apart from direct disorders of the nervous system, other oral manifestations may also be associated with neuro disorders and provide potential clinical indications such as the change in microbial amounts and components. This is concomitant with findings that BMS is not limited to oral diseases, but has close interactions with general body health [105]. According to the human oral microbiome database (HOMD) [106], an approximation of 700 prokaryote species are localized in the oral cavity [106]. These microbes are closely related with the body health condition of human beings and show relative alterations in times of disease [107]. This alteration might be achieved by microbial metabolic products which could be utilized as neurotransmitters or antibodies, and initiate a serious of host actions [107]. These interactions might also account for the different pathologies and outcomes of comorbidities, as the amount of bacteria in tongue mucosa is significantly reduced in benign migratory glossitis compared with atrophic glossitis and BMS. The change in mucosal bacteria is associated with morphological alterations, making the oral environment more acceptable for H. pylori (HP) colonization and facilitating oro-oral transmission [108]. In the meantime, HP could also alter the pathogenesis of BMS, as a majority of BMS patients recovered after HP elimination [109]. The potential roles of HP in the generation and progression of oral pain have also suggested its probable relation with BMS [110].
2.3. Recurrent Aphthous Stomatitis (RAS)
RAS, characterized by rounded shallow painful ulcers with a yellowish gray pseudomembranous center and a well-defined erythematous rim, is the most common ulcerative disease of the oral mucosa in clinical practice [111]. Similar to other ulcers including ulcerative colitis (UC) and gastric ulcers, RAS occurs in approximately 2% to 10% of Caucasian populations and seriously affects their life quality and work efficiency [112]. The relationship between RAS and psychology has also received much investigation and shown its roles in RAS pathology [113]. Intriguingly, despite the close relationship of depression and salivary cortisol level [114], no significant difference was found between the RAS patients and the control group [115,116]. However, when the p value was set as 0.01 rather than 0.05 to indicate the difference, a distinction was found in an experiment with a similar design [117]. Other studies have also reported a higher level of cortisol in both the saliva and serum of RAS patients [118]. Catecholamine concentration shows a similar trend [119]. This is similar to that of OSCC and BMS, and suggests a similar mechanism underlying them. Further investigations have shown the hypofunction of the sympathetic nervous system in RAS patients [21] with a significantly higher level of salivary α-amylase enzyme (sAAE) in RAS patients [120]. As a common biomarker in the nervous system, sAAE is associated with stress, anxiety, and defects of the nervous system [121]. In the meantime, sAAE is also associated with alterations of microbial components and diversity, as Zulfiqar et al. found that F. nucleatum cells inhibited the enzymatic activity of salivary α-amylase in a dose-dependent manner [122]. sAAE is even utilized as a bio-marker for RAS patients, especially when evaluating their stress level [121]. Stress in RAS patients might also account for the alteration of the promoter region of the serotonin transporter (5-HTT) and result in decreased transcriptive activities for serotonin expression [118,119,123].
In addition to their roles in RAS pathology, several psychological factors are also important in the prognosis of RAS [124]. However, related research works are scarce and the detailed mechanisms underlying them still require further investigation.
All these findings suggest the essential roles of bacteria in the pathogenesis of oral mucosal diseases. Here, we discuss three typical and serious roles as representative and apart from these, other oral symptoms involved in psychological problems are summarized in Table 1. From the analysis above, several similarities for psychological factors in the pathology of oral diseases could be extracted. To begin with, the pathologies of many oral diseases are associated with psychological factors, and many oral manifestations are accompanied by nerve dysfunctions and/or degeneration alterations. This lays the foundation for their interactions, and is found to be relative to changes in central nervous system and/or the peripheral nervous system. Moreover, changes in the nervous system are accompanied by alterations of the oral bacteria. This increases the relation between oral diseases and overall body health, and might also account for the comorbidities of certain diseases. In addition, psychological factors, especially small molecules such as neurotransmitters, are of potential use in clinical treatment. These lay the basis for psychological interventions by means of drug therapy and mental consultation. However, further investigation is required before effective use due to the complex regulatory mechanisms and potential dose-dependent regulatory pathways.
3. Psychological Factors Affected by Oral Diseases and Dietary Habits
Studies have corroborated that the food we eat every day is closely associated with the feelings we have. Simultaneously, they might also become the origin of several diseases, as some oral diseases are accompanied with psychological abnormalities as shown in Table 2. Studies have corroborated that these impacts are achieved via several pathways similar to the ones accounting for several pathological processes [125]. Additionally, the key regulatory molecules also show much resemblance. The well-known TRPV channel conducting the spicy sense depends on the activation of CB1 and CB2 receptors on the neuronal synapse in the brain. OTOP1 receptors are responsible for the sour taste, and SEMA3A and SEMA7A are shown to be significant transmitters in bitter and sweet taste neurons for taste-receptor cells (TRCs) [126]. The activation of these key factors in oral diseases might serve as the initiators of alterations in the nervous system and further result in psychological changes. Some typical preferences are characteristic of certain diseases; for example, most patients of ulcerative colitis often feel stressed and fond of spicy food [127], although detailed relations remain unknown. Other common preferences for food taste, including those for the most common sweet and sour food, have been summarized [128] and even utilized for treatment [129]. The effects of these foods varies a lot among people in both age and sex, as studies found a higher preference for sweet for women, which is also related to typical physical status including menstrual period and estrogen level [130]. Adolescents show a similar tendency for sweets with a stronger desire especially in times of being lured by the outer environment [131]. Heavy smokers are taught to have some sweets in times of craving hits, and those with diabetes are found to have a decrease in food taste and sweet perception [132]. All these phenomena are of potential use in clinical treatment and would bring bright prospects for patients, considering the increase in life quality and treatment outcomes.
The interactions among psychological factors are also worth attention, as the occurrence is often a mixture of several bad feelings which could work together and strengthen the neuro feedback. This exacerbation might aggravate the pathology of oral mucosa and result in a bad prognosis. These reactions could be mediated by both local tissues and signals from the central nervous system, and could exert an impact on multiple systems. Moreover, they are also likely to be of great benefit for patients concerning their replacement of drugs with everyday food.
4. Information Is Analyzed in the Brain and Sent Back to Oral Mucosa
Despite the large effort in understanding how information is analyzed and comprehended within the brain, clear and convictive mechanisms remain unknown. However, the existence of neurotransmitters in oral tissue and receptors on oral glands and epithelial cells have been confirmed, and several typical feedback mechanisms between oral tissue and the nervous system have been elucidated. The sensitivity and amount of these receptors are related to the extent of the stress and psychological conditions of individuals. This regulation is mediated via both the local tissues and interactions within the brain. Significant roles of endocannabinoids (ECS) have been reported in the mediation of this information transportation, with PPAR-α and TRPV being their co-workers [133]. Current knowledge of these interactions is mainly limited in different neurotransmitters in different part of the brain and types of neurons, respectively, while those elaborate innervations are largely unveiled. More research is worthwhile concerning its potential uses in understanding the pathology of and providing treatment for diseases of both the oral mucosa and psychological well-being. In fact, many diseases come with both the psychological problems and oral manifestations (as shown in Table 3), and are worth more research considering the potential treatments it might bring forward.
5. Conclusions
Psychological factors are of great importance in clinical treatment under our current knowledge. They are also of great use in the treatment of oral mucosal diseases and the development of related drugs. However, the current understanding of their detailed mechanisms is largely limited to the level of key molecules. In this review, after briefly introducing the general interactions between psychological and nervous system factors, we have provided detailed information about the three most common diseases in oral mucosa. We have also put forward relative disease pathologies based on current knowledge and analyzed the similarities among them. Besides this, we have summarized the type of oral lesions associated with psychological and nervous system problems, pointing out the potential clinical prospects for clinical diagnosis and treatment. The knowledge of these key molecules is also important for drug development, and offers great help in relieving patients’ symptoms and increasing their life quality. Despite our efforts to generalize the common roles of psychological factors in the progression of oral mucosal diseases, there might also be some limitations. For example, whether similar drugs could be utilized in different disease models and the specific dosages of these is waiting to be evaluated. Moreover, psychological treatment is mainly limited to mental consultations, while chemical drugs are relatively scarce. This may be due to the lack of molecular-targeted investigations on drug development of psychological diseases. However, with the development of psychology and medical technology, more genetic methods would be of great help in understanding the underlying mechanisms in disease pathologies and offering psychological methods for the treatment of oral mucosal diseases.
Author Contributions
Conceptualization, Y.G. and J.X.; writing—original draft preparation, Y.G., B.W., H.G. and C.H.; writing—review and editing, Y.G., B.W. and R.H.; visualization, H.G., C.H., L.G. and Y.D.; supervision, J.X.; project administration; funding acquisition, J.X. All authors have read and agreed to the published version of the manuscript.
Funding
This work was funded by the National Natural Science Foundation of China Grant (No. 82174056, 81673671).
Institutional Review Board Statement
This article does not involve confidentiality or plagiarism.
Informed Consent Statement
As clinical sample acquisition is not involved in this paper, informed consent is not required.
Data Availability Statement
Not applicable.
Conflicts of Interest
The authors declare no conflict of interest.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References
1. Shibahara T. Oral cancer—Diagnosis and therapy Clin. Calcium 2017 27 1427 1433 28947694
2. Gao L. Xu T. Huang G. Jiang S. Gu Y. Chen F. Oral microbiomes: More and more importance in oral cavity and whole body Protein Cell 2018 9 488 500 10.1007/s13238-018-0548-1 29736705
3. Costacurta M. Benavoli D. Arcudi G. Docimo R. Oral and dental signs of child abuse and neglect ORAL Implantol. 2015 8 68 73 10.11138/orl/2015.8.2.068 27555907
4. Wolff K.D. Follmann M. Nast A. The diagnosis and treatment of oral cavity cancer Dtsch. Arztebl. Int. 2012 109 829 835 10.3238/arztebl.2012.0829 23248713
5. Shrivastava A. Boylan J. Bureau Y. Sousa A.D. Shah N. Biological Trajectory for Psychosocial Risk Factors in Psychiatric Disorders—A Concept Based Review Open J. Psychiatry 2015 5 7 18 10.4236/ojpsych.2015.51002
6. Wu H. Denna T.H. Storkersen J.N. Gerriets V.A. Beyond a neurotransmitter: The role of serotonin in inflammation and immunity Pharmacol. Res. 2019 140 100 114 10.1016/j.phrs.2018.06.015 29953943
7. Guo L.W. Zhang S.K. Liu S.Z. Yang F.N. Zheng L.Y. Chen Q. Cao X.Q. Sun X.B. Zhang J.G. Analysis of endoscopic screening compliance and related factors among high risk population of upper gastrointestinal cancer in urban areas of Henan Province from 2013 to 2017 Zhonghua Yu Fang Yi Xue Za Zhi (Chin. J. Prev. Med.) 2020 54 523 528 10.3760/cma.j.cn112150-20200304-00238
8. Jelvehzadeh F. Dogaheh E.R. Bernstein C. Shakiba S. Ranjbar H. The effect of a group cognitive behavioral therapy on the quality of life and emotional disturbance of women with breast cancer Support. Care Cancer 2021 30 305 312 10.1007/s00520-021-06421-4 34278530
9. Serrano W.C. Maldonado J. The Use of Physostigmine in the Diagnosis and Treatment of Anticholinergic Toxicity after Olanzapine Overdose: Literature Review and Case Report J. Acad. Consult. Liaison Psychiatry 2021 62 285 297 10.1016/j.jaclp.2020.12.013 34102130
10. Arredondo L.F. Aranda-Romo S. Rodríguez-Leyva I. Chi-Ahumada E. Saikaly S.K. Portales-Pérez D.P. González-Amaro R. Salgado-Bustamante M. Enriquez-Macias L. Eng W. Tau Protein in Oral Mucosa and Cognitive State: A Cross-Sectional Study Front. Neurol. 2017 8 554 10.3389/fneur.2017.00554 29081764
11. Lourenço M. Azevedo Á. Brandão I. Gomes P.S. Orofacial manifestations in outpatients with anorexia nervosa and bulimia nervosa focusing on the vomiting behavior Clin. Oral Investig. 2018 22 1915 1922 10.1007/s00784-017-2284-y
12. Nakasu S. Matsumura K. Nioka H. Handa J. Lectin histochemistry of dermoid and epidermoid cysts of the central nervous system Noshuyo Byori (Brain Tumor Pathol.) 1993 10 11 17
13. Giannetti L. Bambini F. Consolo U. Teeth and oral mucosa in neurocutaneous syndromes, metabolic diseases and in diseases with defects of DNA repair Minerva Stomatol. 2003 52 175 180 12874525
14. Gemonov V.V. Kutvitskaia S.A. Cherep O.E. Chebotareva T.L. The reactivity characteristics of the organs of the oral cavity in an with different types of nervous systems Stomatologiia 1996 75 12 14
15. Gholizadeh N. Mohebbi A.H. Mirzaii-Dizgah I. Sheykhbahaei N. α1 adrenergic receptors in serum and saliva of patients with oral squamous cell carcinoma Clin. Transl. Oncol. 2021 23 1705 1710 10.1007/s12094-021-02571-3 33644847
16. Fernández Castro M. López-Pintor R.M. Serrano J. Ramírez L. Sanz M. Andreu J.L. Muñoz Fernández S. Protocolised odontological assessment of patients with primary Sjögren’s syndrome Reumatol. Clin. 2021 17 25 31 10.1016/j.reuma.2018.12.005 30745279
17. Ritchie A. Kramer J.M. Recent Advances in the Etiology and Treatment of Burning Mouth Syndrome J. Dent. Res. 2018 97 1193 1199 10.1177/0022034518782462 29913093
18. Mancini M. Vidal S.M. Insights into the pathogenesis of herpes simplex encephalitis from mouse models Mamm. Genome 2018 29 425 445 10.1007/s00335-018-9772-5 30167845
19. Baig S. Rubab Z. Farooq W. Molecular Pathogenesis of Chewable Tobacco J. Coll. Phys. Surg. Pak. 2018 28 381 385 10.29271/jcpsp.2018.05.381
20. Puhakka A. Forssell H. Soinila S. Virtanen A. Röyttä M. Laine M. Tenovuo O. Teerijoki-Oksa T. Jääskeläinen S.K. Peripheral nervous system involvement in primary burning mouth syndrome—Results of a pilot study Oral Dis. 2016 22 338 344 10.1111/odi.12454 26847146
21. Present S.I. Check J.H. Hypofunction of the Sympathetic Nervous System as a Possible Etiologic Cause of Recurrent Aphthous Stomatitis Compend. Contin. Educ. Dent. 2016 37 381 385 27517475
22. Huang H.C. Nakatsuka M. Iwai Y. Activation of microglial cells in the trigeminal subnucleus caudalis evoked by inflammatory stimulation of the oral mucosa Okajimas Folia Anat. Jpn. 2013 89 137 145 10.2535/ofaj.89.137 23614987
23. Fernández-Cruz Pérez E. Rodríguez-Sainz C. Poliovirus immunology: Vaccines, problems for the prevention/eradication and future interventions Rev. Esp. Salud Publica 2013 87 443 454 10.4321/S1135-57272013000500003 24322281
24. Takahashi K. Xerostomia and dysphagia Clin. Calcium 2012 22 59 65 22201100
25. Woda A. Dao T. Gremeau-Richard C. Steroid dysregulation and stomatodynia (burning mouth syndrome) J. Orofac. Pain 2009 23 202 210 19639097
26. Nakatsuka M. Iwai Y. Expression of TRPV4 in the stimulated rat oral mucous membrane—Nociceptive mechanisms of lingual conical papillae Okajimas Folia Anat. Jpn. 2009 86 45 54 10.2535/ofaj.86.45 19877445
27. Nissalo S. Hietanen J. Malmström M. Hukkanen M. Polak J. Konttinen Y.T. Disorder-specific changes in innervation in oral lichen planus and lichenoid reactions J. Oral Pathol. Med. 2000 29 361 369 10.1034/j.1600-0714.2000.290801.x 10972344
28. Ichikawa H. Sugimoto T. Vanilloid receptor 1-like receptor-immunoreactive primary sensory neurons in the rat trigeminal nervous system Neuroscience 2000 101 719 725 10.1016/S0306-4522(00)00427-9 11113320
29. Verzé L. Paraninfo A. Ramieri G. Viglietti-Panzica C. Panzica G.C. Immunocytochemical evidence of plasticity in the nervous structures of the rat lower lip Cell Tissue Res. 1999 297 203 211 10.1007/s004410051348 10470490
30. Aimetti M. Romano F. Cricenti L. Perotto S. Gotti S. Panzica G. Graziano A. Merkel cells and permanent disesthesia in the oral mucosa after soft tissue grafts J. Cell. Physiol. 2010 224 205 209 10.1002/jcp.22115 20333649
31. Fernandes N.D. Arya K. Ward R. Congenital Herpes Simplex StatPearls StatPearls Publishing LLC. Treasure Island, FL, USA 2021
32. Phyu W.K. Ong K.C. Wong K.T. Modelling person-to-person transmission in an Enterovirus A71 orally infected hamster model of hand-foot-and-mouth disease and encephalomyelitis Emerg. Microbes Infect. 2017 6 e62 10.1038/emi.2017.49 28698666
33. Colletti G. Allevi F. Moneghini L. Rabbiosi D. Bertossi D. Frau I. Biglioli F. Tadini G. Epidermal nevus and ameloblastoma: A rare association Oral Surg. Oral Med. Oral Pathol. Oral Radiol. 2014 117 e275 e279 10.1016/j.oooo.2013.06.003 23969333
34. Pennisi M.G. Hartmann K. Lloret A. Ferrer L. Addie D. Belák S. Boucraut-Baralon C. Egberink H. Frymus T. Gruffydd-Jones T. Cryptococcosis in cats: ABCD guidelines on prevention and management J. Feline Med. Surg. 2013 15 611 618 10.1177/1098612X13489224 23813826
35. Bellissimo-Rodrigues F. Bollela V.R. Da Fonseca B.A. Martinez R. Endemic paracoccidioidomycosis: Relationship between clinical presentation and patients’ demographic features Med. Mycol. 2013 51 313 318 10.3109/13693786.2012.714529 22928923
36. O’Connor M.B. Phelan M.J. The impact of HSV for inflammatory arthropathy patients Rheumatol. Int. 2012 32 489 490 10.1007/s00296-010-1758-x 21243495
37. Ha M. Chung J.W. Hahm K.B. Kim Y.J. Lee W. An J. Kim D.K. Kim M.G. A case of Cowden syndrome diagnosed from multiple gastric polyposis World J. Gastroenterol. 2012 18 861 864 10.3748/wjg.v18.i8.861 22371648
38. Gündüz O. Sahiner N. Atasoy P. Senyücel C. Acitretin treatment for lipoid proteinosis Case Rep. Dermatol. Med. 2012 2012 324506 10.1155/2012/324506 23259080
39. Zhang Y. Cui W. Liu L. Wang J. Zhao H. Liao Y. Na R. Dong C. Wang L. Xie Z. Pathogenesis study of enterovirus 71 infection in rhesus monkeys Lab. Investig. J. Tech. Methods Pathol. 2011 91 1337 1350 10.1038/labinvest.2011.82
40. Tovaru S. Parlatescu I. Dumitriu A.S. Bucur A. Kaplan I. Oral complications associated with D-penicillamine treatment for Wilson disease: A clinicopathologic report J. Periodontol. 2010 81 1231 1236 10.1902/jop.2010.090736 20384464
41. Kastrukoff L.F. Lau A.S. Takei F. Smyth M.J. Jones C.M. Clarke S.R. Carbone F.R. Redundancy in the immune system restricts the spread of HSV-1 in the central nervous system (CNS) of C57BL/6 mice Virology 2010 400 248 258 10.1016/j.virol.2010.02.013 20199790
42. Binitha M.P. Thomas D. Asha L.K. Tuberous sclerosis complex associated with dyschromatosis universalis hereditaria Indian J. Dermatol. Venereol. Leprol. 2006 72 300 302 10.4103/0378-6323.26729 16880579
43. López de Blanc S. Sambuelli R. Femopase F. Luna N. Gravotta M. David D. Bistoni A. Criscuolo M.I. Bacillary angiomatosis affecting the oral cavity. Report of two cases and review J. Oral Pathol. Med. 2000 29 91 96 10.1034/j.1600-0714.2000.290207.x 10718405
44. Nelen M.R. Padberg G.W. Peeters E.A. Lin A.Y. van den Helm B. Frants R.R. Coulon V. Goldstein A.M. van Reen M.M. Easton D.F. Localization of the gene for Cowden disease to chromosome 10q22-23 Nat. Genet. 1996 13 114 116 10.1038/ng0596-114 8673088
45. Chan T.Y. Critchley J.A. Pulmonary aspiration following Dettol poisoning: The scope for prevention Hum. Exp. Toxicol. 1996 15 843 846 10.1177/096032719601501009 8906435
46. von den Driesch P. Sweet’s syndrome (acute febrile neutrophilic dermatosis) J. Am. Acad. Dermatol. 1994 31 535 560 535–556; quiz 557–560 10.1016/S0190-9622(94)70215-2 8089280
47. Gesser R.M. Valyi-Nagy T. Altschuler S.M. Fraser N.W. Oral-oesophageal inoculation of mice with herpes simplex virus type 1 causes latent infection of the vagal sensory ganglia (nodose ganglia) J. Gen. Virol. 1994 75 Pt 9 2379 2386 10.1099/0022-1317-75-9-2379 8077936
48. Zouboulis C.C. Kurz K. Bratzke B. Orfanos C.E. Adamantiades-Behçet disease: Necrotizing systemic vasculitis with a fatal outcome Hautarzt Z. Dermatol. Venerol. Verwandte Geb. 1991 42 451 454
49. Molinoff P.B. α- and β-adrenergic receptor subtypes properties, distribution and regulation Drugs 1984 28 (Suppl. S2) 1 15 10.2165/00003495-198400282-00002 6098436
50. Moretti-Rojas I. Ezrailson E.G. Birnbaumer L. Entman M.L. Garber A.J. Serotonergic and adrenergic regulation of skeletal muscle metabolism in the rat. II. The use of [125I]iodolysergic acid diethylamide and [125I]iodopindolol as probes of sarcolemmal receptor function and specificity J. Biol. Chem. 1983 258 12499 12508 10.1016/S0021-9258(17)44204-9 6313663
51. Lüthy I.A. Bruzzone A. Piñero C.P. Castillo L.F. Chiesa I.J. Vázquez S.M. Sarappa M.G. Adrenoceptors: Non conventional target for breast cancer? Curr. Med. Chem. 2009 16 1850 1862 10.2174/092986709788186048 19442150
52. van der Heijden C. Groh L. Keating S.T. Kaffa C. Noz M.P. Kersten S. van Herwaarden A.E. Hoischen A. Joosten L.A.B. Timmers H. Catecholamines Induce Trained Immunity in Monocytes In Vitro and In Vivo Circ. Res. 2020 127 269 283 10.1161/CIRCRESAHA.119.315800 32241223
53. Zhou J. Chen C. Chen X. Fei Y. Jiang L. Wang G. Vitamin C Promotes Apoptosis and Cell Cycle Arrest in Oral Squamous Cell Carcinoma Front. Oncol. 2020 10 976 10.3389/fonc.2020.00976 32587830
54. Gondivkar S.M. Gadbail A.R. Sarode S.C. Hedaoo A. Dasgupta S. Sharma B. Sharma A. Yuwanati M. Gondivkar R.S. Gaikwad R.N. Oral Psychosomatic Disorders in Family Caregivers of Oral Squamous Cell Carcinoma Patients Asian Pac. J. Cancer Prev. APJCP 2021 22 477 483 10.31557/APJCP.2021.22.2.477 33639663
55. Morse D.E. Psoter W.J. Cleveland D. Cohen D. Mohit-Tabatabai M. Kosis D.L. Eisenberg E. Smoking and drinking in relation to oral cancer and oral epithelial dysplasia Cancer Causes Contr. 2007 18 919 929 10.1007/s10552-007-9026-4
56. Schell J.T. Petermann-Meyer A. Kloss-Brandstätter A. Bartella A.K. Kamal M. Hölzle F. Lethaus B. Teichmann J. Distress thermometer for preoperative screening of patients with oral squamous cell carcinoma J. Cranio-Maxillo-Facial Surg. 2018 46 1111 1116 10.1016/j.jcms.2018.04.022
57. Almangush A. Leivo I. Mäkitie A.A. Biomarkers for Immunotherapy of Oral Squamous Cell Carcinoma: Current Status and Challenges Front. Oncol. 2021 11 616629 10.3389/fonc.2021.616629 33763354
58. D’Addazio G. Artese L. Traini T. Rubini C. Caputi S. Sinjari B. Immunohistochemical study of osteopontin in oral squamous cell carcinoma allied to fractal dimension J. Biol. Regul. Homeost. Agents 2018 32 1033 1038 30043590
59. Nishioka T. Tada H. Ibaragi S. Chen C. Sasano T. Nicotine exposure induces the proliferation of oral cancer cells through the α7 subunit of the nicotinic acetylcholine receptor Biochem. Biophys. Res. Commun. 2019 509 514 520 10.1016/j.bbrc.2018.12.154 30598264
60. Catassi A. Servent D. Paleari L. Cesario A. Russo P. Multiple roles of nicotine on cell proliferation and inhibition of apoptosis: Implications on lung carcinogenesis Mutat. Res. Rev. Mutat. Res. 2008 659 221 231 10.1016/j.mrrev.2008.04.002
61. Hartwig C.L. Sprick J.D. Jeong J. Hu Y. Morison D.G. Stein C.M. Paranjape S. Park J. Increased vascular α1-adrenergic receptor sensitivity in older adults with posttraumatic stress disorder Am. J. Physiol. Regul. Integr. Comp. Physiol. 2020 319 R611 R616 10.1152/ajpregu.00155.2020 32966119
62. Aykac A. Şehirli A. Gören M.Z. Evaluation of the Effect of Prazosin Treatment on α-2c Adrenoceptor and Apoptosis Protein Levels in the Predator Scent-Induced Rat Model of Post-Traumatic Stress Disorder J. Mol. Neurosci. 2020 70 1120 1129 10.1007/s12031-020-01518-7 32133592
63. Sharma P. Sandhu S.V. Bhandari R. Verma I. Bhullar R.K. Khangura R.K. Estimation of cortisol levels in patients with premalignant disorders and oral squamous cell carcinoma J. Oral Maxillofac. Pathol. JOMFP 2018 22 27 34 10.4103/jomfp.JOMFP_181_16 29731553
64. Lee S.H. Shin D.-W. Stein M.A. Increased Cortisol after Stress is Associated with Variability in Response Time in ADHD Children Yonsei Med. J. 2010 51 206 211 10.3349/ymj.2010.51.2.206 20191011
65. Suleyman H. Halici Z. Cadirci E. Hacimuftuoglu A. Bilen H. Indirect role of beta2-adrenergic receptors in the mechanism of anti-inflammatory action of NSAIDS J. Physiol. Pharmacol. 2008 59 661 672 19212002
66. Myslivecek J. Rícný J. Kolár F. Tucek S. The effects of hydrocortisone on rat heart muscarinic and adrenergic alpha 1, beta 1 and beta 2 receptors, propranolol-resistant binding sites and on some subsequent steps in intracellular signalling Naunyn-Schmiedebergs Arch. Pharmacol. 2003 368 366 376 10.1007/s00210-003-0825-1 14564448
67. Hoen S. Mazoit J.X. Asehnoune K. Brailly-Tabard S. Benhamou D. Moine P. Edouard A.R. Hydrocortisone increases the sensitivity to alpha1-adrenoceptor stimulation in humans following hemorrhagic shock Crit. Care Med. 2005 33 2737 2743 10.1097/01.CCM.0000189743.55352.0E 16352953
68. Maeda H. Watanabe Y. Lai R.T. Yoshida H. Effect of glucocorticoids on alpha-2 adrenoceptors in vas deferens of reserpinized rat in organ culture Life Sci. 1983 33 39 46 10.1016/0024-3205(83)90709-9 6306373
69. Cuneo R.C. Livesey J.H. Nicholls M.G. Espiner E.A. Donald R.A. Effects of alpha-1 adrenergic blockade on the hormonal response to hypoglycaemic stress in normal man Clin. Endocrinol. 1987 26 1 8 10.1111/j.1365-2265.1987.tb03632.x
70. Obata K. Shimo T. Okui T. Matsumoto K. Takada H. Takabatake K. Kunisada Y. Ibaragi S. Yoshioka N. Kishimoto K. Role of Neurokinin 3 Receptor Signaling in Oral Squamous Cell Carcinoma Anticancer Res. 2017 37 6119 6123 10.21873/anticanres.12060 29061792
71. Obata K. Shimo T. Okui T. Matsumoto K. Takada H. Takabatake K. Kunisada Y. Ibaragi S. Nagatsuka H. Sasaki A. Tachykinin Receptor 3 Distribution in Human Oral Squamous Cell Carcinoma Anticancer Res. 2016 36 6335 6341 10.21873/anticanres.11230 27919954
72. Ma J. Zhang Y. Wang J. Zhao T. Ji P. Song J. Zhang H. Luo W. Proliferative effects of gamma-amino butyric acid on oral squamous cell carcinoma cells are associated with mitogen—Activated protein kinase signaling pathways Int. J. Mol. Med. 2016 38 305 311 10.3892/ijmm.2016.2597 27222045
73. Kimura R. Kasamatsu A. Koyama T. Fukumoto C. Kouzu Y. Higo M. Endo-Sakamoto Y. Ogawara K. Shiiba M. Tanzawa H. Glutamate acid decarboxylase 1 promotes metastasis of human oral cancer by β-catenin translocation and MMP7 activation BMC Cancer 2013 13 555 10.1186/1471-2407-13-555 24261884
74. Yap Y.H. Say Y.H. Resistance against tumour necrosis factor α apoptosis by the cellular prion protein is cell-specific for oral, colon and kidney cancer cell lines Cell Biol. Int. 2012 36 273 277 10.1042/CBI20110088 21980981
75. Yap Y.H. Say Y.H. Resistance against apoptosis by the cellular prion protein is dependent on its glycosylation status in oral HSC-2 and colon LS 174T cancer cells Cancer Lett. 2011 306 111 119 10.1016/j.canlet.2011.02.040 21439722
76. Saito T. Kasamatsu A. Ogawara K. Miyamoto I. Saito K. Iyoda M. Suzuki T. Endo-Sakamoto Y. Shiiba M. Tanzawa H. Semaphorin7A Promotion of Tumoral Growth and Metastasis in Human Oral Cancer by Regulation of G1 Cell Cycle and Matrix Metalloproteases: Possible Contribution to Tumoral Angiogenesis PLoS ONE 2015 10 e0137923 10.1371/journal.pone.0137923 26378920
77. Ahmed A. Gulino A. Amayo S. Arancio W. Florena A.M. Belmonte B. Jurjus A. Leone A. Miletich I. Natriuretic peptide system expression in murine and human submandibular salivary glands: A study of the spatial localisation of ANB, BNP, CNP and their receptors J. Mol. Histol. 2020 51 3 13 10.1007/s10735-019-09849-5 31722080
78. Binmadi N.O. Basile J.R. Perineural invasion in oral squamous cell carcinoma: A discussion of significance and review of the literature Oral Oncol. 2011 47 1005 1010 10.1016/j.oraloncology.2011.08.002 21865078
79. Ide F. Ito Y. Matsuoka K. Muramatsu T. Saito I. Re-excision perineural invasion in oral squamous cell carcinoma Oral Dis. 2014 20 219 220 10.1111/odi.12156 23848996
80. Zhang K. Qiu W. Wu B. Fang F. Long non-coding RNAs are novel players in oral inflammatory disorders, potentially premalignant oral epithelial lesions and oral squamous cell carcinoma (Review) Int. J. Mol. Med. 2020 46 535 545 10.3892/ijmm.2020.4628 32626947
81. Pereira C.M. Sehnem D. da Fonseca E.O. Barboza H.F.G. de Carvalho A.C.P. DaSilva A.F.M. Moura-Neto V. DosSantos M.F. miRNAs: Important Targets for Oral Cancer Pain Research BioMed Res. Int. 2017 2017 4043516 10.1155/2017/4043516 29214166
82. Xie S.F. Li C.C. He H. Xia X.Y. Wang X.T. Current advances in research of burning mouth syndrome: Biological nerve and social psychology patterns Zhonghua Kou Qiang Yi Xue Za Zhi 2020 55 123 128 10.3760/cma.j.issn.1002-0098.2020.02.011 32074676
83. Galli F. Lodi G. Sardella A. Vegni E. Role of psychological factors in burning mouth syndrome: A systematic review and meta-analysis Ceph. Int. J. Headache 2017 37 265 277 10.1177/0333102416646769
84. Mo X. Zhang J. Fan Y. Svensson P. Wang K. Thermal and mechanical quantitative sensory testing in Chinese patients with burning mouth syndrome—A probable neuropathic pain condition? J. Headache Pain 2015 16 84 10.1186/s10194-015-0565-x 26399417
85. Lauria G. Majorana A. Borgna M. Lombardi R. Penza P. Padovani A. Sapelli P. Trigeminal small-fiber sensory neuropathy causes burning mouth syndrome Pain 2005 115 332 337 10.1016/j.pain.2005.03.028 15911160
86. Feller L. Fourie J. Bouckaert M. Khammissa R.A.G. Ballyram R. Lemmer J. Burning Mouth Syndrome: Aetiopathogenesis and Principles of Management Pain Res. Manag. 2017 2017 1926269 10.1155/2017/1926269 29180911
87. Lin M. Li B. Gu F. Yue Y. Huang Y. Chen Q. Zeng G. Xia J. Study on psychoprophylaxis and monoamines neurotransmitter of patients with burning mouth syndrome Hua Xi Yi Ke Da Xue Xue Bao 2001 32 576 578 12528553
88. Sinjari B. Feragalli B. Cornelli U. Belcaro G. Vitacolonna E. Santilli M. Rexhepi I. D’Addazio G. Zuccari F. Caputi S. Artificial Saliva in Diabetic Xerostomia (ASDIX): Double Blind Trial of Aldiamed((R)) Versus Placebo J. Clin. Med. 2020 9 2196 10.3390/jcm9072196
89. Wada A. Shizukuishi T. Kikuta J. Yamada H. Watanabe Y. Imamura Y. Shinozaki T. Dezawa K. Haradome H. Abe O. Altered structural connectivity of pain-related brain network in burning mouth syndrome-investigation by graph analysis of probabilistic tractography Neuroradiology 2017 59 525 532 10.1007/s00234-017-1830-2 28361345
90. Sinding C. Gransjøen A.M. Schlumberger G. Grushka M. Frasnelli J. Singh P.B. Grey matter changes of the pain matrix in patients with burning mouth syndrome Eur. J. Neurosci. 2016 43 997 1005 10.1111/ejn.13156 26741696
91. Lee Y.C. Jahng G.H. Ryu C.W. Byun J.Y. Change in gray matter volume and cerebral blood flow in patients with burning mouth syndrome J. Oral Pathol. Med. 2019 48 335 342 10.1111/jop.12838 30735586
92. Jääskeläinen S.K. Pathophysiology of primary burning mouth syndrome Clin. Neurophysiol. 2012 123 71 77 10.1016/j.clinph.2011.07.054 22030140
93. Martikainen I.K. Hagelberg N. Jääskeläinen S.K. Hietala J. Pertovaara A. Dopaminergic and serotonergic mechanisms in the modulation of pain: In vivo studies in human brain Eur. J. Pharmacol. 2018 834 337 345 10.1016/j.ejphar.2018.07.038 30036531
94. Guarneri F. Guarneri C. Marini H. Contribution of neuroinflammation in burning mouth syndrome: Indications from benzodiazepine use Dermatol. Ther. 2008 21 S21 S24 10.1111/j.1529-8019.2008.00228.x
95. Siniscalchi A. Gallelli L. Marigliano N.M. Orlando P. De Sarro G. Use of topiramate for glossodynia Pain Med. 2007 8 531 534 10.1111/j.1526-4637.2006.00152.x 17716328
96. Mendak-Ziółko M. Konopka T. Bogucki Z.A. Evaluation of select neurophysiological, clinical and psychological tests for burning mouth syndrome Oral Surg. Oral Med. Oral Pathol. Oral Radiol. 2012 114 325 332 10.1016/j.oooo.2012.04.004 22862972
97. Momota Y. Takano H. Kani K. Matsumoto F. Motegi K. Aota K. Yamamura Y. Omori M. Tomioka S. Azuma M. Frequency analysis of heart rate variability: A useful assessment tool of linearly polarized near-infrared irradiation to stellate ganglion area for burning mouth syndrome Pain Med. 2013 14 351 357 10.1111/pme.12008 23240990
98. Varvat J. Thomas-Anterion C. Decousus M. Perret-Liaudet A. Laurent B. Atypical Lewy body disease revealed by burning mouth syndrome and a pseudo-psychiatric syndrome Rev. Neurol. 2010 166 547 549 10.1016/j.neurol.2009.10.018 20045161
99. Koszewicz M. Mendak M. Konopka T. Koziorowska-Gawron E. Budrewicz S. The characteristics of autonomic nervous system disorders in burning mouth syndrome and Parkinson disease J. Orofac. Pain 2012 26 315 320 23110271
100. Mignogna M.D. Adamo D. Schiavone V. Ravel M.G. Fortuna G. Burning mouth syndrome responsive to duloxetine: A case report Pain Med. 2011 12 466 469 10.1111/j.1526-4637.2010.01035.x 21223496
101. Koike K. Shinozaki T. Hara K. Noma N. Okada-Ogawa A. Asano M. Shinoda M. Eliav E. Gracely R.H. Iwata K. Immune and endocrine function in patients with burning mouth syndrome Clin. J. Pain 2014 30 168 173 10.1097/AJP.0b013e31828c4bf1 24398391
102. Ito M. Tokura T. Yoshida K. Nagashima W. Kimura H. Umemura E. Tachibana M. Miyauchi T. Kobayashi Y. Arao M. Five Patients with Burning Mouth Syndrome in Whom an Antidepressant (Serotonin-Noradrenaline Reuptake Inhibitor) Was Not Effective, but Pregabalin Markedly Relieved Pain Clin. Neuropharmacol. 2015 38 158 161 10.1097/WNF.0000000000000093 26166242
103. Eccleston C. Hearn L. Williams A.C. Psychological therapies for the management of chronic neuropathic pain in adults Cochrane Database Syst. Rev. 2015 2015 CD011259 10.1002/14651858.CD011259.pub2
104. Imamura Y. Okada-Ogawa A. Noma N. Shinozaki T. Watanabe K. Kohashi R. Shinoda M. Wada A. Abe O. Iwata K. A perspective from experimental studies of burning mouth syndrome J. Oral Sci. 2020 62 165 169 10.2334/josnusd.19-0459 32161235
105. de Pedro M. López-Pintor R.M. Casañas E. Hernández G. General health status of a sample of patients with burning mouth syndrome: A case-control study Oral Dis. 2020 26 1020 1031 10.1111/odi.13327 32153093
106. Adler C.J. Dobney K. Weyrich L.S. Kaidonis J. Walker A.W. Haak W. Bradshaw C.J. Townsend G. Sołtysiak A. Alt K.W. Sequencing ancient calcified dental plaque shows changes in oral microbiota with dietary shifts of the Neolithic and Industrial revolutions Nat. Genet. 2013 45 450 455 10.1038/ng.2536 23416520
107. Baker J.L. Bor B. Agnello M. Shi W. He X. Ecology of the Oral Microbiome: Beyond Bacteria Trends Microbiol. 2017 25 362 374 10.1016/j.tim.2016.12.012 28089325
108. Gall-Troselj K. Mravak-Stipetić M. Jurak I. Ragland W.L. Pavelić J. Helicobacter pylori colonization of tongue mucosa—Increased incidence in atrophic glossitis and burning mouth syndrome (BMS) J. Oral Pathol. Med. 2001 30 560 563 10.1034/j.1600-0714.2001.300909.x 11555160
109. Brailo V. Vuéiaeeviae-Boras V. Alajbeg I.Z. Alajbeg I. Lukenda J. Aeurkoviae M. Oral burning symptoms and burning mouth syndrome-significance of different variables in 150 patients Med. Oral Patol. Oral Y Cir. Bucal 2006 11 E252 E255
110. Chung M.K. Wang S. Oh S.L. Kim Y.S. Acute and Chronic Pain from Facial Skin and Oral Mucosa: Unique Neurobiology and Challenging Treatment Int. J. Mol. Sci. 2021 22 5810 10.3390/ijms22115810 34071720
111. Porter S.R. Scully C. Pedersen A. Recurrent Aphthous Stomatitis Crit. Rev. Oral Biol. Med. 1998 9 306 321 10.1177/10454411980090030401 9715368
112. Altenburg A. El-Haj N. Micheli C. Puttkammer M. Abdel-Naser M.B. Zouboulis C.C. The treatment of chronic recurrent oral aphthous ulcers Dtsch. Arztebl. Int. 2014 111 665 673 10.3238/arztebl.2014.0665 25346356
113. Dhopte A. Naidu G. Singh-Makkad R. Nagi R. Bagde H. Jain S. Psychometric analysis of stress, anxiety and depression in patients with recurrent aphthous Stomatitis—A cross-sectional survey based study J. Clin. Exp. Dent. 2018 10 e1109 e1114 10.4317/jced.55012 30607229
114. Nadendla L.K. Meduri V. Paramkusam G. Pachava K.R. Relationship of salivary cortisol and anxiety in recurrent aphthous stomatitis Indian J. Endocrinol. Metab. 2015 19 56 59 10.4103/2230-8210.131768 25593827
115. Rezaei F. Aminian M. Raygani A.V. Evaluation of Salivary Cortisol Changes and Psychological Profiles in Patients with Recurrent Aphthous Stomatitis Contemp. Clin. Dent. 2017 8 259 263 10.4103/ccd.ccd_165_17 28839413
116. Eguia-del Valle A. Martínez-Conde-Llamosas R. López-Vicente J. Uribarri-Etxebarria A. Aguirre-Urizar J.M. Salivary cortisol determination in patients from the Basque Country with recurrent aphthous stomatitis. A pilot study Med. Oral Patol. Oral Y Cir. Bucal 2013 18 e207 e211 10.4317/medoral.18110
117. McCartan B.E. Lamey P.J. Wallace A.M. Salivary cortisol and anxiety in recurrent aphthous stomatitis J. Oral Pathol. Med. 1996 25 357 359 10.1111/j.1600-0714.1996.tb00278.x 8890048
118. Albanidou-Farmaki E. Poulopoulos A.K. Epivatianos A. Farmakis K. Karamouzis M. Antoniades D. Increased anxiety level and high salivary and serum cortisol concentrations in patients with recurrent aphthous stomatitis Tohoku J. Exp. Med. 2008 214 291 296 10.1620/tjem.214.291 18441503
119. Hong C. Wang W.M. Wang X. Tang W. Assessment of mental health status and plasma catecholamine levels in patients with recurrent aphthous ulcer Shanghai Kou Qiang Yi Xue (Shanghai J. Stomatol.) 2018 27 48 51
120. Cardoso J.A. Dos Santos Junior A.A. Nunes M.L. de Figueiredo M.A. Cherubini K. Salum F.G. Salivary Alpha-Amylase Enzyme, Psychological Disorders, and Life Quality in Patients with Recurrent Aphthous Stomatitis Int. J. Dent. 2017 2017 5269856 10.1155/2017/5269856 28408928
121. Broche N. Jr. Takeshita R.S.C. Mouri K. Bercovitch F.B. Huffman M.A. Salivary alpha-amylase enzyme is a non-invasive biomarker of acute stress in Japanese macaques (Macaca fuscata) Primates J. Primatol. 2019 60 547 558 10.1007/s10329-019-00757-6
122. Zulfiqar M. Yamaguchi T. Sato S. Oho T. Oral Fusobacterium nucleatum subsp. polymorphum binds to human salivary α-amylase Mol. Oral Microbiol. 2013 28 425 434 10.1111/omi.12036 23906425
123. Victoria J.M. Correia-Silva Jde F. Pimenta F.J. Kalapothakis E. Gomez R.S. Serotonin transporter gene polymorphism (5-HTTLPR) in patients with recurrent aphthous stomatitis J. Oral Pathol. Med. 2005 34 494 497 10.1111/j.1600-0714.2005.00344.x 16091117
124. Ni Riordain R. Meaney S. McCreary C. Impact of chronic oral mucosal disease on daily life: Preliminary observations from a qualitative study Oral Dis. 2011 17 265 269 10.1111/j.1601-0825.2010.01734.x 20860762
125. Frid A.H. Kreugel G. Grassi G. Halimi S. Hicks D. Hirsch L.J. Smith M.J. Wellhoener R. Bode B.W. Hirsch I.B. New Insulin Delivery Recommendations Mayo Clin. Proc. 2016 91 1231 1255 10.1016/j.mayocp.2016.06.010 27594187
126. Lee H. Macpherson L.J. Parada C.A. Zuker C.S. Ryba N.J.P. Rewiring the taste system Nature 2017 548 330 333 10.1038/nature23299 28792937
127. Wang Y.F. Ou-Yang Q. Xia B. Liu L.N. Gu F. Zhou K.F. Mei Q. Shi R.H. Ran Z.H. Wang X.D. Multicenter case-control study of the risk factors for ulcerative colitis in China World J. Gastroenterol. 2013 19 1827 1833 10.3748/wjg.v19.i11.1827 23555172
128. Motoki K. Suzuki S. Extrinsic Factors Underlying Food Valuation in the Human Brain Front. Behav. Neurosci. 2020 14 131 10.3389/fnbeh.2020.00131 32848655
129. Winsor A.M. Ihle M. Taylor L.A. Methods for independently manipulating palatability and color in small insect prey PLoS ONE 2020 15 e0231205 10.1371/journal.pone.0231205 32255810
130. Bowen D.J. Grunberg N.E. Variations in food preference and consumption across the menstrual cycle Physiol. Behav. 1990 47 287 291 10.1016/0031-9384(90)90144-S 2333344
131. Potvin Kent M. Pauzé E. Roy E.A. de Billy N. Czoli C. Children and adolescents’ exposure to food and beverage marketing in social media apps Pediatr. Obes. 2019 14 e12508 10.1111/ijpo.12508 30690924
132. Jensterle M. DeVries J.H. Battelino T. Battelino S. Yildiz B. Janez A. Glucagon-like peptide-1, a matter of taste? Rev. Endocr. Metab. Disord. 2021 22 763 775 10.1007/s11154-020-09609-x 33123893
133. Ghonghadze M. Pachkoria K. Okujava M. Antelava N. Gongadze N. Endocannabinoids receptors mediated central and peripheral effects (review) Georgian Med. News 2020 298 137 143
|