QSAR modeling with the electrotopological state indices: predicting the toxicity of organic chemicals

A quantitative structure-activity relationship model, based on the atom-type electrotopological state (E-state) indices, for the prediction of toxicity to fathead minnow for a diverse set of 140 organic chemicals is presented. Multiple linear regression and artificial neural network techniques were employed in the modeling of experimental toxicity (-logLC(50)) values ranging from 0.85 to 6.09. For the training set of 130 organic compounds a linear regression model with r(2)=0.84 and s=0.36 was obtained with 14 atom-type E-state indices. For the test set of 10 compounds, the corresponding statistics were r(2)=0.83 and s=0.47, respectively. Neural networks gave a significant improvement using the same set of parameters, and the standard deviations were s=0.31 for the training set and s=0.30 for the test set when an artificial neural network with five neurons in the hidden layer was used. The results clearly show that accurate models can be rapidly calculated for the prediction of toxicity for a diverse set of organic chemicals using easily calculated parameters.     

MOA-based linear and nonlinear QSAR models for predicting the toxicity of organic chemicals to Vibrio fischeri

Environmental Science and Pollution Research - Risk assessment of pollutants to humans and ecosystems requires much toxicological data. However, experimental testing of compounds expends a large...     

In silico prediction of Tetrahymena pyriformis toxicity for diverse industrial chemicals with substructure pattern recognition and machine learning methods

There is an increasing need for the rapid safety assessment of chemicals by both industries and regulatory agencies throughout the world. In silico techniques are practical alternatives in the environmental hazard assessment. It is especially true to address the persistence, bioaccumulative and toxicity potentials of organic chemicals. Tetrahymena pyriformis toxicity is often used as a toxic endpoint. In this study, 1571 diverse unique chemicals were collected from the literature and composed of the largest diverse data set for T. pyriformis toxicity. Classification predictive models of T. pyriformis toxicity were developed by substructure pattern recognition and different machine learning methods, including support vector machine (SVM), C4.5 decision tree, k-nearest neighbors and random forest. The results of a 5-fold cross-validation showed that the SVM method performed better than other algorithms. The overall predictive accuracies of the SVM classification model with radial basis functions kernel was 92.2% for the 5-fold cross-validation and 92.6% for the external validation set, respectively. Furthermore, several representative substructure patterns for characterizing T. pyriformis toxicity were also identified via the information gain analysis methods.     

The efficiency of two standardized procedures for extracting mutagenic chemicals from soils

A major technical limitation to monitoring the fate of mutagenic compounds in soil is the lack of an established extraction procedure. This study was conducted to evaluate the efficiency of the blender extraction procedure for extracting benzo(a)pyrene (BAP) or 2-nitrofluorene (2NF) from a Weswood silt loam or a Bastrop clay loam soil. Extracted residues were evaluated using the Salmonella/microsome mutagenicity assay and high performance liquid chromatography (HPLC) to quantify recovery of the two chemicals from soil. In addition, a limited study using only the mutagenicity assay was conducted to compare the efficiency of the Blender and Soxhlet extraction procedures for recovering organic mutagens from both soils amended with a wood preserving bottom sediment waste.Extracted residues were dissolved in dimethylsulfoxide to concentrations that could be detected in the bioassay. Over all treatment levels and for both soils, the extraction efficiency of the blender procedure for BAP was greater than 80% for 12 of 16 treatments as measured using both bioassay and HPLC analysis; while, for 2NF the extraction efficiency was greater than 90% for all 15 treatments as measured by HPLC, and greater than 80% for twelve of 15 treatments as measured by bioassay. These results indicate that blender extraction provided efficient recovery of the pure compounds tested. When the efficiency of the blender and Soxhlet procedure for extracting the wood-preserving bottom sediment and waste amended soil were compared, significantly greater levels of mutagenic activity were detected in the fractions extracted using the blender extraction than were detected using the Soxhlet extraction.     

Near-field dispersion modeling for regulatory applications

This paper evaluates the application of dispersion models to estimate near-field pollutant concentrations in two case studies. The Industrial Source Complex Short-Term Model (ISCST3) was evaluated with hexavalent chromium measurements collected within 100 m of two facilities in Barrio Logan, San Diego, CA. ISCST3 provided reasonable estimates for higher pollutant concentrations but underestimated lower concentrations. To understand the observed distribution of concentrations in Barrio Logan, a recently conducted tracer experiment was analyzed. The tracer, sulfur hexafluoride, was released at ambient temperature from an urban facility at the University of California at Riverside, and concentrations were measured within 20 m of the source. Modeling results indicated that Industrial Source Complex-Plume Rise Model Enhancement and American Meteorological Society/U.S. Environmental Protection Agency Regulatory Model-Plume Rise Model Enhancement overestimated high concentrations and underestimated low concentrations. A diagnostic study with a simple Gaussian dispersion model that incorporated site-specific meteorology was used to evaluate model results. This study found that incorporating lateral meandering for nonbuoyant urban plumes in Gaussian dispersion models could improve concentration estimates even when downwash is not considered. Incorporating a meandering component in ISCST3 resulted in improvements in estimating hexavalent chromium concentrations in Barrio Logan. Credible near-source concentration estimates depend on accurate characterization of emissions, onsite micrometeorology, and a method to account for lateral meandering in the near field.     

Multivariate toxicity profiles and QSAR modeling of non-dioxin-like PCBs--an investigation of in vitro screening data from ultra-pure congeners

The non-dioxin-like PCBs (NDL-PCBs) found in food and human samples have a complex spectrum of adverse effects, but lack a detailed risk assessment. The toxicity profiles of 21 carefully selected PCBs (19 NDL-PCBs) were identified by in vitro screening in 17 different assays on specific endpoints related to neurotoxicity, endocrine disruption and tumor promotion. To ensure that the test results were not affected by polychlorinated dioxins, dibenzofurans or DL-PCB contaminants, the NDL-PCB congeners were thoroughly purified before testing. Principal component analysis (PCA) was used to derive general toxicity profiles from the in vitro screening data. The toxicity profiles indicated different structure-activity relationships (SAR) and distinct mechanisms of action. The analysis also indicated that the NDL-PCBs could be divided into two groups. The first group included generally smaller, ortho-substituted congeners, comprising PCB 28, 47, 51, 52, 53, 95, 100, 101, 104 and 136, with PCB 95, 101 and 136 as generally being most active. The second group comprising PCB 19, 74, 118, 122, 128, 138, 153, 170, 180 and 190 had lower biological activity in many of the assays, except for three endocrine-related assays. The most abundant congeners, PCB 138, 153, 170, 180 and 190, cluster in the second group, and thereby show similar SAR. Two quantitative structure-activity relationship (QSAR) models could be developed that added information to the SAR and could aid in risk assessments of NDL-PCBs. The QSAR models predicted a number of congeners as active and among these e.g., PCB 18, 25, 45 and 49 have been found in food or human samples.     

Mechanism of concentration addition toxicity: they are different for nonpolar narcotic chemicals, polar narcotic chemicals and reactive chemicals

According to the toxicity mechanism of the individual chemicals, the concentration addition toxicity mechanism is revealed for nonpolar-narcotic-chemical mixtures, polar-narcotic-chemical mixtures and reactive-chemical mixtures, respectively. For nonpolar-narcotic-chemical mixtures, the partitioning of individual chemicals from water to biophase was determined, and the result shows that their concentration additive effect results from no competitive partitioning among individual chemicals. For polar-narcotic-chemical mixtures, their toxicity are contributed by two factors (the total baseline toxicity and the hydrogen bond donor activity of individual chemicals), and it is the concentration additive effect for either of these two factors that leads to their concentration addition toxicity. In addition, the interactions between the reactive chemicals and the biological macromolecules are discussed thoroughly. The results suggest that the net effect of these interactions is zero, and it is this zero net effect that leads to the concentration addition toxicity mechanism for reactive-chemical mixtures.     

Data ranges in aquatic toxicity of chemicals

A significant problem for effect assessment of aquatic ecosystems arises from the large ranges of toxicity data, which can be found in different databases and literature. Here, ranges are given for the aquatic toxicity of 27 high production volume chemicals. Based on these illustrative examples and on the current literature on uncertainty in aquatic effect assessment, toxicity ranges are discussed for their possible causes (variation in experimental condition, species, endpoint, time) and ways to handle them (safety factors). Implications and recommendations on the procedure of risk analysis of chemical substances are drawn. Two main requirements for a comprehensive risk assessment are identified, which often play a minor role in current practice (as they are often neglected) as well as in scientific discussion (as they are meant to be trivial). First, data quality must be checked critically before applying any result of a toxicity test. Secondly, experimental data should take into account different species and acute as well as chronic data. If these aspects are considered in risk analysis, which is common practice in ecotoxicology but not always in the context of practical applications in risk engineering, a more comprehensive picture of the environmental toxicity of a chemical substance can be obtained.     

A new hybrid system of QSAR models for predicting bioconcentration factors (BCF)

The aim was to develop a reliable and practical quantitative structure–activity relationship (QSAR) model validated by strict conditions for predicting bioconcentration factors (BCF). We built up several QSAR models starting from a large data set of 473 heterogeneous chemicals, based on multiple linear regression (MLR), radial basis function neural network (RBFNN) and support vector machine (SVM) methods. To improve the results, we also applied a hybrid model, which gave better prediction than single models. All models were statistically analysed using strict criteria, including an external test set. The outliers were also examined to understand better in which cases large errors were to be expected and to improve the predictive models. The models offer more robust tools for regulatory purposes, on the basis of the statistical results and the quality check on the input data.     

A new hybrid system of QSAR models for predicting bioconcentration factors (BCF)

The aim was to develop a reliable and practical quantitative structure-activity relationship (QSAR) model validated by strict conditions for predicting bioconcentration factors (BCF). We built up several QSAR models starting from a large data set of 473 heterogeneous chemicals, based on multiple linear regression (MLR), radial basis function neural network (RBFNN) and support vector machine (SVM) methods. To improve the results, we also applied a hybrid model, which gave better prediction than single models. All models were statistically analysed using strict criteria, including an external test set. The outliers were also examined to understand better in which cases large errors were to be expected and to improve the predictive models. The models offer more robust tools for regulatory purposes, on the basis of the statistical results and the quality check on the input data.     

Computer-based QSARs for predicting mixture toxicity of benzene and its derivatives

During the past decades, the Quantitative structure-activity relationships (QSARs) have been proven to be reliable tools when little or no empirical data are available in medicinal chemistry, biochemistry, toxicology, and environmental sciences. However, only few studies that quantitatively predict mixtures toxicity have been reported. In this study, the QASR models for the binary mixtures toxicity of 12 benzene and its derivatives, including eight non-polar-narcotic compounds and four polar narcotic compounds were developed, without reference to exact toxicity mechanisms of single compounds. All parameters for the QSAR studies were defined on the basis of quantum mechanical calculations and these parameters were selected by the stepwise procedure. The results of this study provided a simple means of predicting the binary mixtures toxicity from the chemical structure.     

Solvophobic approach for predicting sorption of hydrophobic organic chemicals on synthetic sorbents and soils

The application of a solvophobic approach for predicting the sorption of hydrophobic organic compounds (HOC) was evaluated with data collected using synthetic sorbents and soils. The experimental data consisted of batch equilibrium sorption coefficients (K_D), as well as soil-TLC and reversed-phase liquid chromatographic (RPLC) retention factors (κ′). All data were collected using aqueous solutions and binary or ternary solvent mixtures of water, methanol, acetone, and acetonitrile. As predicted by the theory, the chromatographic retention factors and sorption coefficients for HOC decreased log-linearly with increasing fraction of organic cosolvent in binary solvents. Model parameters estimated from the binary solvent data could be used to predict sorption (or retention) from ternary solvents. Reasonable agreement was found between model parameters reported in the literature and those estimated using the data from batch sorption, soil-TLC, and RPLC studies.     

化学物质对发光菌的联合毒性评价方法

毒性单位法(TU)的理论基础来源于剂量加和模型(DA),目前仅在二元联合毒性评价中广泛应用。为了确定TU模型适合评价的混合物类型,实验选取5种剂量效应曲线类型不同的物质,采用微板光度计测试了一元、二元混合物对发光菌青海弧菌-Q67(Vibrio-qinghaiensis sp.-Q67)的急性毒性。根据物质的剂量效应曲线形状将物质分为A、B、C 3类,利用毒性单位法(TU)和联合作用定义法分别对AA类、AB类、AC类、BC类混合物进行分析。结果表明,TU法仅适合于由剂量效应曲线接近直线的物质组成的混合物进行联合毒性的评价。以效应为基准、TU模型为框架建立了TU’模型,该模型可以满足对任何类型已知成分的混合物或者未知成分的实际水样之间的多元联合作用的评价。     

Statistical evaluation of chronic toxicity data on aquatic organisms for the hazard identification: the chemicals toxicity distribution approach

Presently, in the Globally Harmonised System of Classification and Labelling of Chemicals the classification of substances for long-term effects to aquatic life is based on acute toxicity in combination with degradation and/or bioaccumulation potential. Recently an OECD Working Group was created to develop the classification scheme to accommodate chronic toxicity data related to aquatic organisms for assigning a chronic hazard category. This study focuses on a new approach for setting chronic toxicity cut-off values based on Chemicals Toxicity Distributions (CTDs). A CTD is obtained through statistical fitting of the data used by regulatory bodies for setting hazard-based classifications. The CTDs were made using the lowest aquatic NOEC value of each chemical. A review of different toxicological sources reporting acute aquatic toxicities was carried out. Initially, the data were arranged according to the specific source and distributions for key taxonomic groups (i.e. fishes, crustaceans and algae) were evaluated separately. In most cases, no significant departures from normality were observed. Thereafter, a compiled database containing >900 values was developed and the CTDs were constructed for each taxonomic group. Significant deviation from normality (P < 0.05) was observed in the fishes and crustaceans' CTDs. However, this deviation was apparently produced by the presence of only seven values with NOECs <1 x 10(-5) mg l(-1), while high correlation between the data and the normal scores (r-values>or= 0.989) indicated that the data were samples from normal distributions. From these observations, potential cut-off values would allow quantitative estimations of the percentage of chemicals falling into each specific category. This approach results in a simple classification hazard scheme where most chemicals are covered in one of the categories, allowing a clear distribution of the chemicals among three categories for chronic toxicity.     

A fragment constant QSAR model for evaluating the EC50 values of organic chemicals to Daphnia magna.

The quantitative relationship between the median effective concentration (EC50) of organic chemicals to Daphnia magna and the number of molecular fragments was investigated based on experimental EC50 values for 217 chemicals derived from the literature. A fragment constant model was developed based on a multivariate linear regression between the number of fragments and the logarithmically transformed reciprocal values of EC50. Functional correction factors were introduced into the model. The model was verified using an independent set of randomly selected data. The mean residual of the final model was 0.4 log-units. The robustness of the model was discussed based on the results of three jackknife tests.