Toxicological dataset

An alternative interpretive approach is to leave the human biomonitoring result as is but develop applied dose-biomarker relationships in animals. That requires obtaining animal PK data to support PBPK modeling or the collection of animal biomarker information in study designs that mimic key toxicology datasets. 

Roberts, D.W, Padewicz, G., Kern, P.S., Gerberick, F., Kimber, 1., Dearman, R.J., Ryan, C.A., Basketter, D.A. and Aptula, A.O. (2007) Mechanistic applicability domain dassification of a local lymph node assay dataset for skin sensitization. Chemical Research in Toxicology, 20, 1019-1030. 

In the theory, it is possible to create new QSAR models with almost all datasets of compounds with known biological/toxicological activity. But practically it is a question of the quality and predictivity of a QSAR model to be applied in prediction of biological/toxicological activity. For this reason evaluation of each QSAR model is extremely important. The evaluation of a QSAR model can be preformed either by internal validation (cross validation) or external validation (use of a test-set). External validation is preferred, but is not always possible, e.g. because of the small size of a dataset (Dearden, 2003). 

Clinical Pharmacokinetics (PK) and Toxicological (Tox) Datasets 255 Tablell.l Experimental ADME/Tox data for model development. 

It should be stressed that not every experimental data set gives rise to a coherent SAR model. Failure to construct a model may be caused by the fact that the experimental data are invalid or that they do not reflect a specific toxicological phenomenon. Additionally, the phenomenon under investigation may be so complex or be the result of so many different mechanisms that the experimental database is not sufficiently large to describe it. With this in mind, it should be stressed that the predic-tivity of the SAR model will be a reflection of the eomplexity of the phenomenon, the size of the database (i.e., the number of chemicals for which experimental data are available), and the ratio of actives/inactives in the dataset (3,22),... 

R. B. Conolly, J. S. Kimbell, D. Janszen, P. M. Schlosser, D. Kalisak, J. Preston and F. J. Miller, Human Respiratory Tract Cancer Risks of Inhaled Formaldehyde Dose-Response Predictions Derived from Biologically-Motivated Computational Modeling of a Combined Rodent and Human Dataset, Toxicological Sciences, 82, 279-296. 

Public databases allow the scientihc community to publish, share, and compare the data obtained from toxicology and toxicogenomics experiments (see Chapter 6). They are a resource for data mining, and for the discovery of novel genes/proteins through their coexpression with known molecules. They also help to identify and minimize the use of experimental practices that introduce undesirable variability into toxicogenomics datasets. 

It has to be emphasized that comprehensive, high-quality datasets for more complex toxicological endpoints like organ toxicities (e.g., liver, kidney), cardiac safety, and teratogenicity are still not really available. Extraction of all relevant data from different sources and structured storage to enable automated data mining and analysis must be the first step. This would be crucial for any further progress in the field of in silico toxicity predictions. 

It is unlikely that any one program will ever become a complete computational toxicology toolbox. Rather, it is more likely that certain predictive software will be used for specific applications, while more general applications will be used to create relevant datasets for multiple uses. Systems biology pathway software, in concert with metabolite prediction, pharmacokinetics, and ADME software will be critical in reducing cost and time necessary for the development of new pharmaceuticals [49],... 

Since the time of this publication, literally thousands of datasets have been correlated the database of the equations generated by the Pomona group (BioLoom) contains more than six thousand equations. In addition, such studies led to the commercialization of products. Most recently, the European Community in its Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) program states that for each chemical circulating in the European territory, a complete dossier on physico-chemical, biological, and toxicological properties must be compiled. For this purpose, validated QSAR methods are accepted. ... 

Central to the success of proteomics applied to toxicology will proteome reference libraries. These datasets will permit comparisons between drugs, clarify species-dependent effects, allow predictions to be made about the potential for toxicity and provide a diagnostic resource for consultation. 

In a second design stage, the chemical library was evaluated with in silico health hazard estimation methods developed by lUCT and state of the art eco-tox expert systems. Prediction of the health hazard and eco-toxicological profiles and physico-chemical properties with a sufficient degree of confidence allowed selection of those molecular structures exhibiting lower levels of intrinsic hazard. Figure 10.4 shows the calculated health hazard distribution of the SOLVSAFE dataset. 

---