Computational Toxicology

Research at NCTR

National Center for Toxicological Research (NCTR) scientists have developed a predictive model for dioxin isomer toxicity using Quantitative Spectral Data Activity Relationships (QSDAR), Pat. 6,898,533. Different Quantitative Structure Activity Relationship (QSAR) models have been used for years to predict properties of chemicals including toxicity and to design more effective drugs.

QSDAR takes advantage of the fact that computer programs can easily and accurately predict the carbon-13 NMR spectra of chemicals, and these spectra are direct measures of the structure’s quantum mechanical properties. By comparison, QSAR models rely on a series of classical mechanical calculations or laboratory measurements that approximate the quantum mechanical properties, thereby consuming resources and accumulating errors with each approximation.

The QSDAR model predicted high toxicity for four dioxin isomers not thought to be important. Dr. Stephen Safe, Texas A&M, was able to provide authentic samples of two isomers, and Office of Regulatory Activities (ORA) scientists at the Arkansas Regional Laboratory demonstrated:

1. the purity of each authentic isomer
2. each isomer had commensurately high values in the AH-receptor assay, a traditional model for evaluating potential dioxin isomer toxicity
3. mass spectrometry peaks for these same two isomers were indeed present in the historical dioxin assay program database. (Accepted for publication: Wilkes et al., Toxicological Sciences, Advance Access published December 7, 2007; DOI: 10.1093/toxsci/kfm294)

QSDAR has been used previously to create accurate prediction models for estrogenicity, corticosteroid binding, and aromatase binding. Future projects will evaluate the effectiveness of QSDAR models measuring the biological action of a series of drugs within a structural class.

The NCTR Computational Chemistry Group has a continuing collaborative effort with the University of Arkansas for Medical Sciences for development of noninvasive breast cancer detection methods and brain disease diagnostic markers. In addition, QSDAR models of predictive toxicity have been developed and experimentally validated for two dioxins previously believed to be nontoxic.

To meet such a need, a quick screening method for species-level identification was developed by the Chemistry staff and is currently being commercialized to the food industry. Sensitivity of the analysis is to the single cell in the volume of rinsate or concentrate assayed: 100 to 200 L. Each assay takes from 1.5 to 2.5 minutes. The methods have been refined so that there are no false-positives or false-negatives. The detection technology has been demonstrated in food matrices as diversified (and relevant to public health) as raw or cooked chicken products, peanut butter, ice cream, and raw spinach. Litmus, LLC, an Arkansas-based company, is commercializing the Chemistry staff’s rapid bacterial identification methods, food quality indicators, and the Spectrometric Data Activity Relationship (SDAR) methods (a series of techniques that can drastically reduce the time and cost of pharmaceutical development and enable rapid toxicological screening).