Estimation of the chemical-induced eye injury using a weight-of-evidence (WoE) battery of 21 artificial neural network (ANN) c-QSAR models (QSAR-21): part I: irritation potential.

Evaluation of potential chemical-induced eye injury through irritation and corrosion is required to ensure occupational and consumer safety for industrial, household and cosmetic ingredient chemicals. The historical method for evaluating eye irritant and corrosion potential of chemicals is the rabbit Draize test. However, the Draize test is controversial and its use is diminishing - the EU 7th Amendment to the Cosmetic Directive (76/768/EEC) and recast Regulation now bans marketing of new cosmetics having animal testing of their ingredients and requires non-animal alternative tests for safety assessments. Thus, in silico and/or in vitro tests are advocated. QSAR models for eye irritation have been reported for several small (congeneric) data sets; however, large global models have not been described. This report describes FDA/CFSAN's development of 21 ANN c-QSAR models (QSAR-21) to predict eye irritation using the ADMET Predictor program and a diverse training data set of 2928 chemicals. The 21 models had external (20% test set) and internal validation and average training/verification/test set statistics were: 88/88/85(%) sensitivity and 82/82/82(%) specificity, respectively. The new method utilized multiple artificial neural network (ANN) molecular descriptor selection functionalities to maximize the applicability domain of the battery. The eye irritation models will be used to provide information to fill the critical data gaps for the safety assessment of cosmetic ingredient chemicals.

Estimation of the chemical-induced eye injury using a Weight-of-Evidence (WoE) battery of 21 artificial neural network (ANN) c-QSAR models (QSAR-21): part II: corrosion potential.

This is part II of an in silico investigation of chemical-induced eye injury that was conducted at FDA's CFSAN. Serious eye damage caused by chemical (eye corrosion) is assessed using the rabbit Draize test, and this endpoint is an essential part of hazard identification and labeling of industrial and consumer products to ensure occupational and consumer safety. There is an urgent need to develop an alternative to the Draize test because EU's 7th amendment to the Cosmetic Directive (EC, 2003; 76/768/EEC) and recast Regulation now bans animal testing on all cosmetic product ingredients and EU's REACH Program limits animal testing for chemicals in commerce. Although in silico methods have been reported for eye irritation (reversible damage), QSARs specific for eye corrosion (irreversible damage) have not been published. This report describes the development of 21 ANN c-QSAR models (QSAR-21) for assessing eye corrosion potential of chemicals using a large and diverse CFSAN data set of 504 chemicals, ADMET Predictor's three sensitivity analyses and ANNE classification functionalities with 20% test set selection from seven different methods. QSAR-21 models were internally and externally validated and exhibited high predictive performance: average statistics for the training, verification, and external test sets of these models were 96/96/94% sensitivity and 91/91/90% specificity.

Prediction of drug-related cardiac adverse effects in humans--B: use of QSAR programs for early detection of drug-induced cardiac toxicities.

This report describes the use of three quantitative structure-activity relationship (QSAR) programs to predict drug-related cardiac adverse effects (AEs), BioEpisteme, MC4PC, and Leadscope Predictive Data Miner. QSAR models were constructed for 9 cardiac AE clusters affecting Purkinje nerve fibers (arrhythmia, bradycardia, conduction disorder, electrocardiogram, palpitations, QT prolongation, rate rhythm composite, tachycardia, and Torsades de pointes) and 5 clusters affecting the heart muscle (coronary artery disorders, heart failure, myocardial disorders, myocardial infarction, and valve disorders). The models were based on a database of post-marketing AEs linked to 1632 chemical structures, and identical training data sets were configured for three QSAR programs. Model performance was optimized and shown to be affected by the ratio of the number of active to inactive drugs. Results revealed that the three programs were complementary and predictive performances using any single positive, consensus two positives, or consensus three positives were as follows, respectively: 70.7%, 91.7%, and 98.0% specificity; 74.7%, 47.2%, and 21.0% sensitivity; and 138.2, 206.3, and 144.2 chi(2). In addition, a prospective study using AE data from the U.S. Food and Drug Administration's (FDA's) MedWatch Program showed 82.4% specificity and 94.3% sensitivity. Furthermore, an external validation study of 18 drugs with serious cardiotoxicity not considered in the models had 88.9% sensitivity.

Prediction of drug-related cardiac adverse effects in humans--A: creation of a database of effects and identification of factors affecting their occurrence.

This is the first of two reports that describes the compilation of a database of drug-related cardiac adverse effects (AEs) that was used to construct quantitative structure-activity relationship (QSAR) models to predict these AEs, to identify properties of pharmaceuticals correlated with the AEs, and to identify plausible mechanisms of action (MOAs) causing the AEs. This database of 396,985 cardiac AE reports was linked to 1632 approved drugs and their chemical structures, 1851 clinical indications (CIs), 997 therapeutic targets (TTs), 432 pharmacological MOAs, and 21,180 affinity coefficients (ACs) for the MOA receptors. AEs were obtained from the Food and Drug Administration's (FDA's) Spontaneous Reporting System (SRS) and Adverse Event Reporting System (AERS) and publicly available medical literature. Drug TTs were obtained from Integrity; drug MOAs and ACs were predicted by BioEpisteme. Significant cardiac AEs and patient exposures were estimated based on the proportional reporting ratios (PRRs) for each drug and each AE endpoint as a percentage of the total AEs. Cardiac AE endpoints were bundled based on toxicological mechanism and concordance of drug-related findings. Results revealed that significant cardiac AEs formed 9 clusters affecting Purkinje nerve fibers (arrhythmia, bradycardia, conduction disorder, electrocardiogram, palpitations, QT prolongation, rate rhythm composite, tachycardia, and Torsades de pointes), and 5 clusters affecting the heart muscle (coronary artery disorders, heart failure, myocardial disorders, myocardial infarction, and valve disorders). Based on the observation that each drug had one TT and up to 9 off-target MOAs, cardiac AEs were highly correlated with drugs affecting cardiovascular and cardioneurological functions and certain MOAs (e.g., alpha- and beta-adeno, dopamine, and hydroxytryptomine receptors).

Identification of structure-activity relationships for adverse effects of pharmaceuticals in humans. Part A: use of FDA post-market reports to create a database of hepatobiliary and urinary tract toxicities.

The Informatics and Computational Safety Analysis Staff at the US FDA's Center for Drug Evaluation and Research has created a database of pharmaceutical adverse effects (AEs) linked to pharmaceutical chemical structures and estimated population exposures. The database is being used to develop quantitative structure-activity relationship (QSAR) models for the prediction of drug-induced liver and renal injury, as well as to identify relationships among AEs. The post-market observations contained in the database were obtained from FDA's Spontaneous Reporting System (SRS) and the Adverse Event Reporting System (AERS) accessed through Elsevier PharmaPendium software. The database contains approximately 3100 unique pharmaceutical compounds and 9685 AE endpoints. To account for variations in AE reports due to different patient populations and exposures for each drug, a proportional reporting ratio (PRR) was used. The PRR was applied to all AEs to identify chemicals that could be scored as positive in the training datasets of QSAR models. Additionally, toxicologically similar AEs were grouped into clusters based upon both biological effects and statistical correlation. This clustering created a weight of evidence paradigm for the identification of compounds most likely to cause human harm based upon findings in multiple related AE endpoints.

Identification of structure-activity relationships for adverse effects of pharmaceuticals in humans: Part B. Use of (Q)SAR systems for early detection of drug-induced hepatobiliary and urinary tract toxicities.

This report describes the development of quantitative structure-activity relationship (QSAR) models for predicting rare drug-induced liver and urinary tract injury in humans based upon a database of post-marketing adverse effects (AEs) linked to approximately 1600 chemical structures. The models are based upon estimated population exposure using AE proportional reporting ratios. Models were constructed for 5 types of liver injury (liver enzyme disorders, cytotoxic injury, cholestasis and jaundice, bile duct disorders, gall bladder disorders) and 6 types of urinary tract injury (acute renal disorders, nephropathies, bladder disorders, kidney function tests, blood in urine, urolithiases). Identical training data sets were configured for 4 QSAR programs (MC4PC, MDL-QSAR, BioEpisteme, and Predictive Data Miner). Model performance was optimized and was shown to be affected by the AE scoring method and the ratio of the number of active to inactive drugs. The best QSAR models exhibited an overall average 92.4% coverage, 86.5% specificity and 39.3% sensitivity. The 4 QSAR programs were demonstrated to be complementary and enhanced performance was obtained by combining predictions from 2 programs (average 78.4% specificity, 56.2% sensitivity). Consensus predictions resulted in better performance as judged by both internal and external validation experiments.

Identification of structure-activity relationships for adverse effects of pharmaceuticals in humans: Part C: use of QSAR and an expert system for the estimation of the mechanism of action of drug-induced hepatobiliary and urinary tract toxicities.

This report describes an in silico methodology to predict off-target pharmacologic activities and plausible mechanisms of action (MOAs) associated with serious and unexpected hepatobiliary and urinary tract adverse effects in human patients. The investigation used a database of 8,316,673 adverse event (AE) reports observed after drugs had been marketed and AEs noted in the published literature that were linked to 2124 chemical structures and 1851 approved clinical indications. The Integrity database of drug patent and literature studies was used to find pharmacologic targets and proposed clinical indications. BioEpisteme QSAR software was used to predict possible molecular targets of drug molecules and Derek for Windows expert system software to predict chemical structural alerts and plausible MOAs for the AEs. AEs were clustered into five types of liver injury: liver enzyme disorders, cytotoxic injury, cholestasis and jaundice, bile duct disorders, and gall bladder disorders, and six types of urinary tract injury: acute renal disorders, nephropathies, bladder disorders, kidney function tests, blood in urine, and urolithiasis. Results showed that drug-related AEs were highly correlated with: (1) known drug class warnings, (2) predicted off-target activities of the drugs, and (3) a specific subset of clinical indications for which the drug may or may not have been prescribed.

Combined Use of MC4PC, MDL-QSAR, BioEpisteme, Leadscope PDM, and Derek for Windows Software to Achieve High-Performance, High-Confidence, Mode of Action-Based Predictions of Chemical Carcinogenesis in Rodents.

ABSTRACT This report describes a coordinated use of four quantitative structure-activity relationship (QSAR) programs and an expert knowledge base system to predict the occurrence and the mode of action of chemical carcinogenesis in rodents. QSAR models were based upon a weight-of-evidence paradigm of carcinogenic activity that was linked to chemical structures (n = 1,572). Identical training data sets were configured for four QSAR programs (MC4PC, MDL-QSAR, BioEpisteme, and Leadscope PDM), and QSAR models were constructed for the male rat, female rat, composite rat, male mouse, female mouse, composite mouse, and rodent composite endpoints. Model predictions were adjusted to favor high specificity (>80%). Performance was shown to be affected by the method used to score carcinogenicity study findings and the ratio of the number of active to inactive chemicals in the QSAR training data set. Results demonstrated that the four QSAR programs were complementary, each detecting different profiles of carcinogens. Accepting any positive prediction from two programs showed better overall performance than either of the single programs alone; specificity, sensitivity, and Chi-square values were 72.9%, 65.9%, and 223, respectively, compared to 84.5%, 45.8%, and 151. Accepting only consensus-positive predictions using any two programs had the best overall performance and higher confidence; specificity, sensitivity, and Chi-square values were 85.3%, 57.5%, and 287, respectively. Specific examples are provided to demonstrate that consensus-positive predictions of carcinogenicity by two QSAR programs identified both genotoxic and nongenotoxic carcinogens and that they detected 98.7% of the carcinogens linked in this study to Derek for Windows defined modes of action.

In Silico Screening of Chemicals for Genetic Toxicity Using MDL-QSAR, Nonparametric Discriminant Analysis, E-State, Connectivity, and Molecular Property Descriptors.

ABSTRACT Genetic toxicity testing is a critical parameter in the safety assessment of pharmaceuticals, food constituents, and environmental and industrial chemicals. Quantitative structure-activity relationship (QSAR) software offers a rapid, cost-effective means of prioritizing the genotoxic potential of chemicals. Our goal is to develop and validate a complete battery of complementary QSAR models for genetic toxicity. We previously reported the development of MDL-QSAR models for the prediction of mutations in Salmonella typhimurium and Escherichia coli ( Contrera et al. 2005b ); this report describes the development of eight additional models for mutagenicity, clastogenicity, and DNA damage. The models were created using MDL-QSAR atom-type E-state, simple connectivity and molecular property descriptor categories, and nonparametric discriminant analysis. In 10% leave-group-out internal validation studies, the specificity of the models ranged from 63% for the mouse lymphoma (L5178Y-tk) model to 88% for chromosome aberrations in vivo. Sensitivity ranged from a high of 74% for the mouse lymphoma model to a low of 39% for the unscheduled DNA synthesis model. The receiver operator characteristic (ROC) was >/=2.00, a value indicative of good predictive performance. The predictive performance of MDL-QSAR models was also shown to compare favorably to the results of MultiCase MC4PC ( Matthews et al. 2006b ) genotoxicity models prepared with the same training data sets. MDL-QSAR software models exhibit good specificity, sensitivity, and coverage and they can provide rapid and cost-effective large-scale screening of compounds for genotoxic potential by the chemical and pharmaceutical industry and for regulatory decision support applications.

Comparison of MC4PC and MDL-QSAR rodent carcinogenicity predictions and the enhancement of predictive performance by combining QSAR models.

This report presents a comparison of the predictive performance of MC4PC and MDL-QSAR software as well as a method for combining the predictions from both programs to increase overall accuracy. The conclusions are based on 10 x 10% leave-many-out internal cross-validation studies using 1540 training set compounds with 2-year rodent carcinogenicity findings. The models were generated using the same weight of evidence scoring method previously developed [Matthews, E.J., Contrera, J.F., 1998. A new highly specific method for predicting the carcinogenic potential of pharmaceuticals in rodents using enhanced MCASE QSAR-ES software. Regul. Toxicol. Pharmacol. 28, 242-264.]. Although MC4PC and MDL-QSAR use different algorithms, their overall predictive performance was remarkably similar. Respectively, the sensitivity of MC4PC and MDL-QSAR was 61 and 63%, specificity was 71 and 75%, and concordance was 66 and 69%. Coverage for both programs was over 95% and receiver operator characteristic (ROC) intercept statistic values were above 2.00. The software programs had complimentary coverage with none of the 1540 compounds being uncovered by both MC4PC and MDL-QSAR. Merging MC4PC and MDL-QSAR predictions improved the overall predictive performance. Consensus sensitivity increased to 67%, specificity to 84%, concordance to 76%, and ROC to 4.31. Consensus rules can be tuned to reflect the priorities of the user, so that greater emphasis may be placed on predictions with high sensitivity/low false negative rates or high specificity/low false positive rates. Sensitivity was optimized to 75% by reclassifying all compounds predicted to be positive in MC4PC or MDL-QSAR as positive, and specificity was optimized to 89% by reclassifying all compounds predicted negative in MC4PC or MDL-QSAR as negative.

In silico approaches to explore toxicity end points: issues and concerns for estimating human health effects.

The European Chemicals Bureau and the Organisation for Economic Cooperation and Development are currently compiling a sanctioned list of quantitative structure-activity relationship (QSAR) risk assessment models and data sets to predict the physiological properties, environmental fate, ecological effects and human health effects of new and existing chemicals in commerce in the European Union. This action implements the technical requirements of the European Commission's Registration, Evaluation and Authorisation of Chemicals legislation. The goal is to identify a battery of QSARs that can furnish rapid, reliable and cost-effective decision support information for regulatory decisions that can substitute for results from animal studies. This report discusses issues and concerns that need to be addressed when selecting QSARs to predict human health effect end points.

Progress in QSAR toxicity screening of pharmaceutical impurities and other FDA regulated products.

Active ingredients in pharmaceutical products undergo extensive testing to ensure their safety before being made available to the American public. A consideration during the regulatory review process is the safety of pharmaceutical contaminants and degradents which may be present in the drug product at low levels. Several published guidances are available that outline the criteria for further testing of these impurities to assess their toxic potential, where further testing is in the form of a battery of toxicology assays and the identification of known structural alerts. However, recent advances in the development of computational methods have made available additional resources for safety assessment such as structure similarity searching and quantitative structure-activity relationship (QSAR) models. These methods offer a rapid and cost-effective first-pass screening capability to assess toxicity when conventional toxicology data are limited or lacking, with the potential to identify compounds that would be appropriate for further testing. This article discusses some of the considerations when using computational toxicology methods for regulatory decision support and gives examples of how the technology is currently being applied at the US Food and Drug Administration.

A comprehensive model for reproductive and developmental toxicity hazard identification: I. Development of a weight of evidence QSAR database.

A weight of evidence (WOE) reproductive and developmental toxicology (reprotox) database was constructed that is suitable for quantitative structure-activity relationship (QSAR) modeling and human hazard identification of untested chemicals. The database was derived from multiple publicly available reprotox databases and consists of more than 10,000 individual rat, mouse, or rabbit reprotox tests linked to 2134 different organic chemical structures. The reprotox data were classified into seven general classes (male reproductive toxicity, female reproductive toxicity, fetal dysmorphogenesis, functional toxicity, mortality, growth, and newborn behavioral toxicity), and 90 specific categories as defined in the source reprotox databases. Each specific category contained over 500 chemicals, but the percentage of active chemicals was low, generally only 0.1-10%. The mathematical WOE model placed greater significance on confirmatory observations from repeat experiments, chemicals with multiple findings within a category, and the categorical relatedness of the findings. Using the weighted activity scores, statistical analyses were performed for specific data sets to identify clusters of categories that were correlated, containing similar profiles of active and inactive chemicals. The analysis revealed clusters of specific categories that contained chemicals that were active in two or more mammalian species (trans-species). Such chemicals are considered to have the highest potential risk to humans. In contrast, some specific categories exhibited only single species-specific activities. Results also showed that the rat and mouse were more susceptible to dysmorphogenesis than rabbits (6.1- and 3.6-fold, respectively).

A comprehensive model for reproductive and developmental toxicity hazard identification: II. Construction of QSAR models to predict activities of untested chemicals.

This report describes the construction, optimization and validation of a battery of quantitative structure-activity relationship (QSAR) models to predict reproductive and developmental (reprotox) hazards of untested chemicals. These models run with MC4PC software to predict seven general reprotox classes: male and female reproductive toxicity, fetal dysmorphogenesis, functional toxicity, mortality, growth, and newborn behavioral toxicity. The reprotox QSARs incorporate a weight of evidence paradigm using rats, mice, and rabbit reprotox study data and are designed to identify trans-species reprotoxicants. The majority of the reprotox QSARs exhibit good predictive performance properties: high specificity (>80%), low false positives (<20%), significant receiver operating characteristic (ROC) values (>2.00), and high coverage (>80%) in 10% leave-many-out validation experiments. The QSARs are based on 627-2023 chemicals and exhibited a wide applicability domain for FDA regulated organic chemicals for which they were designed. Experiments were also performed using the MC4PC multiple module prediction technology, and ROC statistics, and adjustments to the ratio of active to inactive (A/I ratio) chemicals in training data sets were made to optimize the predictive performance of QSAR models. Results revealed that an A/I ratio of approximately 40% was optimal for MC4PC. We discuss specific recommendations for the application of the reprotox QSAR battery.

An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: I. Identification of carcinogens using surrogate endpoints.

A retrospective analysis of standard genetic toxicity (genetox) tests, reproductive and developmental toxicity (reprotox) studies, and rodent carcinogenicity bioassays (rcbioassay) was performed to identify the genetox and reprotox endpoints whose results best correlate with rcbioassay observations. A database of 7205 chemicals with genetox (n = 4961), reprotox (n = 2173), and rcbioassay (n = 1442) toxicity data was constructed; 1112 of the chemicals have both genetox and rcbioassay data and 721 chemicals have both reprotox and rcbioassay data. This study differed from previous studies by using conservative weight of evidence criteria to classify chemical carcinogens, data from 63 genetox and reprotox toxicological endpoints, and a new statistical parameter of correlation indicator (CI, the average of specificity and positive predictivity) to identify good surrogate endpoints for predicting carcinogenicity. Among 63 endpoints, results revealed that carcinogenicity was well correlated with certain tests for gene mutation (n = 8), in vivo clastogenicity (n = 2), unscheduled DNA synthesis assay (n = 1), and reprotox (n = 3). The current FDA regulatory battery of four genetox tests used to predict carcinogenicity includes two tests with good correlation (gene mutation in Salmonella and in vivo micronucleus) and two tests with poor correlation (mouse lymphoma gene mutation and in vitro chromosome aberrations) by our criteria.

An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: II. Identification of genotoxicants, reprotoxicants, and carcinogens using in silico methods.

This study examined a novel method to identify carcinogens that employed expanded data sets composed of in silico data pooled with actual experimental genetic toxicity (genetox) and reproductive and developmental toxicity (reprotox) data. We constructed 21 modules using the MC4PC program including 13 of 14 (11 genetox and 3 reprotox) tests that we found correlated with results of rodent carcinogenicity bioassays (rcbioassays) [Matthews, E.J., Kruhlak, N.L., Cimino, M.C., Benz, R.D., Contrera, J.F., 2005b. An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: I. Identification of carcinogens using surrogate endpoints. Regul. Toxicol. Pharmacol.]. Each of the 21 modules was evaluated by cross-validation experiments and those with high specificity (SP) and positive predictivity (PPV) were used to predict activities of the 1442 chemicals tested for carcinogenicity for which actual genetox or reprotox data were missing. The expanded data sets had approximately 70% in silico data pooled with approximately 30% experimental data. Based upon SP and PPV, the expanded data sets showed good correlation with carcinogenicity testing results and had correlation indicator (CI, the average of SP and PPV) values of 75.5-88.7%. Conversely, expanded data sets for 9 non-correlated test endpoints were shown not to correlate with carcinogenicity results (CI values <75%). Results also showed that when Salmonella mutagenic carcinogens were removed from the 12 correlated, expanded data sets, only 7 endpoints showed added value by detecting significantly more additional carcinogens than non-carcinogens.

In silico screening of chemicals for bacterial mutagenicity using electrotopological E-state indices and MDL QSAR software.

Quantitative structure-activity relationship (QSAR) software offers a rapid, cost effective means of prioritizing the mutagenic potential of chemicals. MDL QSAR models were developed using atom-type E-state indices and non-parametric discriminant analysis. Models were developed for Salmonella typhimurium gene mutation, combining results from strains TA97, TA98, TA100, TA1535, TA1536, TA1537, and TA1538 (n=3228), and Escherichia coli gene mutation tests WP2, WP100, and polA (n=472). Composite microbial mutation models (n=3338) were developed combining all Salmonella, E. coli, and the Bacillus subtilis rec spot test study results. The datasets contained 74% non-pharmaceuticals and 26% pharmaceuticals. Salmonella and microbial mutagenesis external validation studies included a total of 1444 and 1485 compounds, respectively. The average specificity, sensitivity, positive predictivity, concordance, and coverage of Salmonella models was 76, 81, 73, 78, and 98%, respectively, with similar performance for the microbial mutagenesis models. MDL QSAR and discriminant analysis provides rapid and highly automated mutagenicity screening software with good specificity, sensitivity, and coverage that is simpler and requires less user intervention than other similar software. MDL QSAR modules for microbial mutagenicity can provide efficient and cost effective large scale screening of compounds for mutagenic potential for the chemical and pharmaceutical industry.

Estimating the safe starting dose in phase I clinical trials and no observed effect level based on QSAR modeling of the human maximum recommended daily dose.

Estimating the maximum recommended starting dose (MRSD) of a pharmaceutical for phase I human clinical trials and the no observed effect level (NOEL) for non-pharmaceuticals is currently based exclusively on an extrapolation of the results of animal toxicity studies. This process is inexact and requires the results of toxicity studies in multiple species (rat, dog, and monkey) to identify the no observed adverse effect level (NOAEL) and most sensitive test species. Multiple uncertainty (safety) factors are also necessary to compensate for incompatibility and uncertainty underlying the extrapolation of animal toxicity to humans. The maximum recommended daily dose for pharmaceuticals (MRDD) is empirically derived from human clinical trials. The MRDD is an estimated upper dose limit beyond which a drug's efficacy is not increased and/or undesirable adverse effects begin to outweigh beneficial effects. The MRDD is essentially equivalent to the NOAEL in humans, a dose beyond which adverse (toxicological) or undesirable pharmacological effects are observed. The NOAEL in test animals is currently used to estimate the safe starting dose in human clinical trials. MDL QSAR predictive modeling of the human MRDD may provide a better, simpler and more relevant estimation of the MRSD for pharmaceuticals and the toxic dose threshold of chemicals in humans than current animal extrapolation based risk assessment models and may be a useful addition to current methods. A database of the MRDD for over 1300 pharmaceuticals was compiled and modeled using MDL QSAR software and E-state and connectivity topological descriptors. MDL QSAR MRDD models were found to have good predictive performance with 74-78% of predicted MRDD values for 120 internal and 160 external validation compounds falling within a range of +/-10-fold the actual MRDD value. The predicted MRDD can be used to estimate the MRSD for pharmaceuticals in phase I clinical trials with the addition of a 10-fold safety factor. For non-pharmaceutical chemicals any compound-related effect can be considered an undesirable and adverse toxicological effect and the predicted MRDD can be used to estimate the NOEL with the addition of an appropriate safety factor.

Assessment of the health effects of chemicals in humans: I. QSAR estimation of the maximum recommended therapeutic dose (MRTD) and no effect level (NOEL) of organic chemicals based on clinical trial data.

The primary objective of this investigation was to develop a QSAR model to estimate the no effect level (NOEL) of chemicals in humans using data derived from pharmaceutical clinical trials and the MCASE software program. We believe that a NOEL model derived from human data provides a more specific estimate of the toxic dose threshold of chemicals in humans compared to current risk assessment models which extrapolate from animals to humans employing multiple uncertainty safety factors. A database of the maximum recommended therapeutic dose (MRTD) of marketed pharmaceuticals was compiled. Chemicals with low MRTDs were classified as high-toxicity compounds; chemicals with high MRTDs were classified as low-toxicity compounds. Two separate training data sets were constructed to identify specific structural alerts associated with high and low toxicity chemicals. A total of 134 decision alerts correlated with toxicity in humans were identified from 1309 training data set chemicals. An internal validation experiment showed that predictions for high- and low-toxicity chemicals were good (positive predictivity >92%) and differences between experimental and predicted MRTDs were small (0.27-0.70 log-fold). Furthermore, the model exhibited good coverage (89.9-93.6%) for three classes of chemicals (pharmaceuticals, direct food additives, and food contact substances). An additional investigation demonstrated that the maximum tolerated dose (MTD) of chemicals in rodents was poorly correlated with MRTD values in humans (R2 = 0.2005, n = 326). Finally, this report discusses experimental factors which influence the accuracy of test chemical predictions, potential applications of the model, and the advantages of this model over those that rely only on results of animal toxicology studies.

Assessment of the health effects of chemicals in humans: II. Construction of an adverse effects database for QSAR modeling.

The FDA's Spontaneous Reporting System (SRS) database contains over 1.5 million adverse drug reaction (ADR) reports for 8620 drugs/biologics that are listed for 1191 Coding Symbols for Thesaurus of Adverse Reaction (COSTAR) terms of adverse effects. We have linked the trade names of the drugs to 1861 generic names and retrieved molecular structures for each chemical to obtain a set of 1515 organic chemicals that are suitable for modeling with commercially available QSAR software packages. ADR report data for 631 of these compounds were extracted and pooled for the first five years that each drug was marketed. Patient exposure was estimated during this period using pharmaceutical shipping units obtained from IMS Health. Significant drug effects were identified using a Reporting Index (RI), where RI = (# ADR reports / # shipping units) x 1,000,000. MCASE/MC4PC software was used to identify the optimal conditions for defining a significant adverse effect finding. Results suggest that a significant effect in our database is characterized by > or = 4 ADR reports and > or = 20,000 shipping units during five years of marketing, and an RI > or = 4.0. Furthermore, for a test chemical to be evaluated as active it must contain a statistically significant molecular structural alert, called a decision alert, in two or more toxicologically related endpoints. We also report the use of a composite module, which pools observations from two or more toxicologically related COSTAR term endpoints to provide signal enhancement for detecting adverse effects.