HYGIENIC SUBSTANTIATION OF CALCULATING MODELS FOR PREDICTING TOXICITY OF DIFFERENT CLASSES INSECTICIDES

Hygienic substantiation of calculating models for predicting toxicity of different classes insecticides (first part). Antonenko A.M., Vavrinevych O.P., Omelchuk S.T., Shpak B.I. This work is the first part of our study to develop alternative experimental mathematic models for predicting toxicity of insecticides. In the first stage, calculations will be carried out and the most reliable models will be proposed. In the second – a statistical analysis and comparative estimation of the toxicometric parameters obtained experimentally and calculated according to the proposed equations. The purpose of the research is the scientific substantiation of the calculation models for predicting toxicity of insecticides of different classes. Data on the physico-chemical properties and toxicometry parameters of fungicides are taken from the PPDB pesticides database. Insecticides of such chemical classes as derivatives of tetram and tetronic acids, benzoylureas, carbamates, neonicotinoids, pyrethroids, organophosphorus compounds, avermectins were selected for analysis. It has been established that there is a significant positive correlation between NO(A)EL in the chronic experiment of all insecticides, the median lethal doses at oral administration (LD50 per os) of pyrithoids and neonicotinoids, and the molecular weight (at p<0.05). There is a significant negative correlation between the toxicometry parameters of all insecticides and their individual groups (pyrithoids, neonicotinoids, organophosphorus compounds) and melting temperature and the octanol-water partition coefficient, log Po/w (at p<0.05). It is proved that the proposed calculation models for predicting insecticide hazards are adequate according to Fisher's criterion, and the coefficients of regression equations are reliable according to Student's criterion (p<0,05). Реферат. Гігієнічне обґрунтування розрахункових моделей прогнозування токсичності інсектицидів різних класів (перша частина). Антоненко А.М., Вавріневич О.П., Омельчук С.Т., Шпак Б.І. Ця робота є першою частиною нашого дослідження з розробки альтернативних експериментальних математичних моделей для прогнозування токсичності інсектицидів. На першому етапі будуть проведені розрахунки й запропоновані найбільш надійні моделі. На другому – буде проведено статистичний аналіз і порівняльна оцінка токсикометричних параметрів, отриманих експериментально і розрахованих за пропонованими рівняннями. Метою дослідження було наукове обґрунтування розрахункових моделей прогнозування токсичності інсектицидів різних класів. Дані про фізико-хімічні властивості та токсикометричні параметри фунгіцидів взяті з бази даних пестицидів PPDB. Для аналізу були обрані інсектициди таких хімічних класів: похідні тетрамової і тетронової кислот, бензоїлсечовини, карбамати, неонікотиноїди, піретроїди, фосфорорганічні сполуки, авермектини. Встановлено, що існує достовірний прямий кореляційний зв'язок між NO(A)EL у хронічному експерименті всіх інсектицидів, середньосмертельними дозами при пероральному надходженні (ЛД50 per os) піритоїдів і неонікотиноїдів і молекулярною масою. Існує достовірний негативний кореляційний МЕДИЧНІ ПЕРСПЕКТИВИ / MEDICNI PERSPEKTIVI 107 19/ Vol. XXIV / 3 зв'язок між параметрами токсикометрії всіх інсектицидів та їх окремих груп (піритоїди, неонікотиноїди, фосфорорганічні сполуки) і температурою плавлення та коефіцієнтом розподілу октанол-вода, log Po/w. Доведено, що запропоновані розрахункові моделі для прогнозування небезпеки інсектицидів є адекватними за критерієм Фішера, а коефіцієнти рівнянь регресії є надійними за критерієм Стьюдента. Today's pesticides are a very effective in increasing yields, but at the same time a serious risk factor for human health because of contamination of environmental objects [6, 7]. Methods for determining the toxicological parameters of pesticides are long-term, labor-intensive and require significant financial and resource costs [9, 14]. In solving this problem, the important role is played by methods of mathematical modeling and predicting of toxicity of xenobiotics, the results of which can be used both for substantiation of toxicological parameters, and at the stage of experiment planning, this will reduce the probability of error and reduce the duration of the study [9]. Foreign laboratories and institutes have been using modeling dependence of toxicometric parameters on the physico-chemical properties of xenobiotics for a long time [5, 11]. In Ukraine today, there are such models for fungicides and herbicides proposed by the specialists of the Hygiene and Ecology Institute of Bogomolets National Medical University [10, 15], as for insecticides, only attempts were made to develop hazard prediction models for prognosis of the risk of neonicotinoid insecticides [2]. This work is the first part of our study to develop alternative experimental mathematic models for predicting toxicity of insecticides. At the first stage, calculations will be carried out and the most reliable models will be proposed. At the second – a statistical analysis and comparative estimation of the toxicometric parameters obtained experimentally and calculated according to the proposed equations. The purpose of the research is the scientific substantiation of the calculation models for predicting toxicity of insecticides of different classes. MATERIALS AND METHODS OF RESEARCH For the development and substantiation of the calculation models in the hygienic assessment of the studied insecticides hazard, an array of experimentally established values of LD50 (median lethal dose) after oral and percutaneous administration, LC50 (median lethal concentration) at inhalation and NO(A)EL (no observed (adverse) effect level) [12], physical and chemical properties (molecular mass, water solubility, vapor pressure, melting point, octanol-water partition coefficient (LogP Ko/w), surface tension) was taken. Data on the physicochemical properties of fungicides are taken from the PPDB pesticides database [12]. For analysis, we selected insecticides of chemical classes the most widely used in the world [4, 8]: 3 compounds of derivatives of tetram and tetronic acids class (spiromesifen, spirodiclofen and spirotetramat); 3 benzoyl-ureas (diflubenzuron, navalurone, teflubenzuron); 4 compounds of the carbamates class (carbosulfan, methomyl, carbaryl, phenoxycarb); 5 substances of the neonicotinoid class (acetamiprid, thiacloprid, imidacloprid, thiamethoxam, clothianidin); 12 active ingridients – pyrethroids (zeta-cypermethrin, lambda-cyhalothrin, beta-cyflutrin, tau-fluvalinate, gamma-cyhalothrin, tefluthrin, bifentrine, esfenvalerate, beta-cypermethrin, cypermethrin, deltamethrin, alpha-cypermethrin); 6 organophosphorus compounds (chlorpyrifos-methyl, dimethoate, malathion, fosalon, pyrimifos-methyl, phenythrothion); 2 avermectins (abamectin and emamectin benzoate). Statistical processing of the results was performed using the package of licensed statistical software IBM SPSS StatisticsBase v.22 and MS Exсel. Correlation and regression analyses were carried out, taking into account the determination coefficient, which most closely approximates the connection between the selected toxicological parameters and physical-chemical properties. The significance of the obtained regression equations was checked by Fisher’s F-criterion, the reliability of individual coefficients in the regression equation (a, b) – according to the Student's t-criterion. RESULTS AND DISCUSSION Initially, by the Pirson method we analyzed the correlation relations between the physical and chemical properties of the derivatives of tetram and tetronic acids, benzoylureas, carbamates, neonicotinoids, pyrethroids, organophosphorus compounds, avermectins and their parameters of toxicometry. Statistically significant results are given in Table 1. The results of the correlation-regression analysis revealed a reverse (negative) correlation between the “octanol-water” partition coefficient and LD50 percut. and NO(A)EL of all insecticides and LD50 per os of neonicotinoids (r= -0.47; -0.39 and -0.83, respectively, at p<0.05) on the array of 35, 28 and 5 active ingredients of the corresponding classes. Using the determination coefficient (R), the proportion of the effect of the investigated factor on the parameters of toxicometry was determined and it was established that the share of influence of the

Today's pesticides are a very effective in increasing yields, but at the same time a serious risk factor for human health because of contamination of environmental objects [6,7]. Methods for determining the toxicological parameters of pesticides are long-term, labor-intensive and require significant financial and resource costs [9,14]. In solving this problem, the important role is played by methods of mathematical modeling and predicting of toxicity of xenobiotics, the results of which can be used both for substantiation of toxicological parameters, and at the stage of experiment planning, this will reduce the probability of error and reduce the duration of the study [9].
Foreign laboratories and institutes have been using modeling dependence of toxicometric parameters on the physico-chemical properties of xenobiotics for a long time [5,11].
In Ukraine today, there are such models for fungicides and herbicides proposed by the specialists of the Hygiene and Ecology Institute of Bogomolets National Medical University [10, 15], as for insecticides, only attempts were made to develop hazard prediction models for prognosis of the risk of neonicotinoid insecticides [2].
This work is the first part of our study to develop alternative experimental mathematic models for predicting toxicity of insecticides. At the first stage, calculations will be carried out and the most reliable models will be proposed. At the second -a statistical analysis and comparative estimation of the toxicometric parameters obtained experimentally and calculated according to the proposed equations.
The purpose of the research is the scientific substantiation of the calculation models for predicting toxicity of insecticides of different classes.

MATERIALS AND METHODS OF RESEARCH
For the development and substantiation of the calculation models in the hygienic assessment of the studied insecticides hazard, an array of experimentally established values of LD 50 (median lethal dose) after oral and percutaneous administration, LC 50 (median lethal concentration) at inhalation and NO(A)EL (no observed (adverse) effect level) [12], physical and chemical properties (molecular mass, water solubility, vapor pressure, melting point, octanol-water partition coefficient (LogP K o/w ), surface tension) was taken. Data on the physicochemical properties of fungicides are taken from the PPDB pesticides database [12].
Statistical processing of the results was performed using the package of licensed statistical software IBM SPSS StatisticsBase v.22 and MS Exсel. Correlation and regression analyses were carried out, taking into account the determination coefficient, which most closely approximates the connection between the selected toxicological parameters and physical-chemical properties. The significance of the obtained regression equations was checked by Fisher's F-criterion, the reliability of individual coefficients in the regression equation (a, b)according to the Student's t-criterion.

RESULTS AND DISCUSSION
Initially, by the Pirson method we analyzed the correlation relations between the physical and chemical properties of the derivatives of tetram and tetronic acids, benzoylureas, carbamates, neonicotinoids, pyrethroids, organophosphorus compounds, avermectins and their parameters of toxicometry. Statistically significant results are given in Table 1.
The results of the correlation-regression analysis revealed a reverse (negative) correlation between the "octanol-water" partition coefficient and LD 50 percut. and NO(A)EL of all insecticides and LD 50 per os of neonicotinoids (r= -0.47; -0.39 and -0.83, respectively, at p<0.05) on the array of 35, 28 and 5 active ingredients of the corresponding classes. Using the determination coefficient (R 2 ), the proportion of the effect of the investigated factor on the parameters of toxicometry was determined and it was established that the share of influence of the Licensed under CC BY 4.0 «octanol-water» distribution coefficients was 22.1; 15.0 and 69.0%, respectively.
The obtained results can be explained by the fact that fat-soluble compounds are slowly metabolized and eliminated from the body, tend to accumulate [13], this increases their toxicity and reduces NO(A)EL and LD 50 . This dependence is typical, since log P o/w is an indicator of the bioavailability of toxic substances [1,3]. At the same time, for herbicides, the opposite dependence was found: their toxic manifestations decreased with increasing log P o/w , which can be explained by the presence of alternative routes of herbicide penetration into the organism due to slightly lower molecular weight and higher solubility in water [10].
There is an inverse (negative) correlation between the melting point and LC 50 in inhalation and NO(A)EL of organophosphorus compounds, as well as LD 50 per os of pyrethroids (r= -0.88; -0.92 and -0.62, respectively at p<0,05) on an array of 5 and 12 active substances of the corresponding classes. Percentage of impact of the melting temperature on the parameters of toxicometry is quite high and makes 78.0; 84.0 and 38.8%, respectively. A similar correlation of the melting point with the parameters of toxicometry (with an increase in the melting point LD 50 per os and inhal., NO(A)EL, values decrease, it means that toxicity of these compounds increases) was also found in the chemical class of herbicides pyrazoles-triketones [10]. One of the reasons for this effect may be the need for a higher temperature for the degradation of the mentioned compounds. At body temperature they retain their initial physical and chemical properties longer and, accordingly, have more time to implement toxic manifestations.
Positive (direct) correlation was found between molecular weight and LD 50 per os of pyrethroids (n=11) and neonicotinoids (n=5); NO(A)EL of all insecticides without phosphorus compounds (n=27) (r=0.72; 0.85 and 0.51, respectively, at p<0.05). The proportion of the influence of the molecular weight, as well as the melting point, is quite high and amounts to 51.3; 71.6 and 25.8%, respectively. 19/ Vol. XXIV / 3 Detected dependence is due to the fact that compounds with very high molecular weight poorly penetrate into the body, in contrast to low molecular weight compounds that easily pass through histohemic barriers [1,3,13]. It should be noted that for fungicides, the relationship between molecular weight and toxicometry parameters is reversible, since most of them, in contrast to insecticides, form isomers that greatly increase the specificity of their action and toxicity [15]. The next step was to carry out a regression analysis with the verification of the obtained equations (tabl. 2, 3) according to Fisher's F statistics, and separate coefficients in the regression equations (a, b) -according to Student's t-criterion. Notes: «*» -statistical significant results; «**» -at р=0.05 and k 1 =1, k 2 =n-2); n -number of observations; X 1 -octanol-water partition coefficient, log P o/w ; X 2 -molecular weight; X 3 -melting temperature, ºС.

Model adequacy indices Coefficients reliability indices
The results of the regression analysis given in Table 2 indicate that the 4 derived linear regression equations N 1-14 are adequate according to Fisher's criterion (p<0.05).
The assessment of coefficients «a» and «b» reliability showed that in all regression equations, they are significant according to Student's t-criterion (p<0,05), except for the equations N 8 and 9. In this equations, the coefficients «a» and / or «b» were unreliable, since the absolute value of the criterion t is less than t cr , which may be due to a small number of observations (n=5).
We also used the exponential, logarithmic, polynomial and step functions in addition to the linear one. In Table 3  Notes: n -number of observations; X 1 -log P o/w ; X 2 -molecular weight; X 3 -melting temperature, ºС.

CONCLUSIONS
1. It has been established that there is a significant positive correlation between NO(A)EL in the chronic experiment of all insecticides, the median lethal doses in oral admission (LD50 per os) of pyrithoids and neonicotinoids, and the molecular weight (at p<0.05). 19/ Vol. XXIV / 3 2. There is a significant negative correlation between the toxicometry parameters of all insecticides and their individual groups (pyrithoids, neonicotinoids, organophosphorus compounds) and the melting temperature and the octanol-water partition coefficient, log Po/w (at p<0.05).
3. It is proved that the proposed calculation models for predicting insecticide hazards are adequate according to Fisher's criterion, and the coefficients of regression equations are reliable according to Student's criterion (p<0,05).