A National Center for Toxicological Research liver cancer database (NCTRlcdb) containing 999 chemicals has been developed based on liver-specific carcinogenicity. This database was constructed with the assumption that organ-specific carcinogenicity is a cleaner endpoint for building a biological predictive model. The NCTRlcdb benefits FDA reviewers by:
- Facilitating the construction of cleaner and better carcinogenicity models by FDA and other organizations thus providing a means to rapidly predict carcinogenicity to aid in evaluating new chemicals submitted for approval.
- Serving as a prototype for other organ-specific carcinogenicity databases for better Quantitative Structure-Activity Relationships (QSAR) and toxicogenomics model prediction.
In the NCTRlcdb, each chemical was assigned one of the following toxicity classifications based on studies of male and female mice and rats:
- liver carcinogen (273)
- other carcinogen (293)
- non-carcinogen (304)
- other (129)
Note: Running NCTRlcdb requires an Internet connection and Java Runtime Environment (JRE) 1.6 or greater installed. To install or upgrade Java please visit the Oracle Java website. Please e-mail NCTR Bioinformatics Support with any questions or problems running NCTRlcdb.
The Carcinogenic Potency Database (CPDB) of Gold and colleagues was used as the basis for the chemicals contained in the NCTRlcdb. The CPDB classifies carcinogens based on multi-organ toxicity data, summarized as a single record for each chemical. However, it is very hard for any sophisticated classification algorithms to build a carcinogenicity-predictive model based on the classification of combined multi-organ carcinogenicity. Therefore, we created the NCTRlcdb to support more reliable predictive models based on liver-specific carcinogenicity.
NCTRlcdb Development Process
- Information from the National Toxicology Program (NTP) and scientific literature was compiled through the use of a hierarchical scheme to reduce multiple records representing each gender, species, route of administration, and liver-specific toxicity into a single record for each chemical.
- A structural cleanup was conducted that involved the removal of all molecules that would interfere with the generation of SAR type descriptors from commercial software programs, i.e., inorganic compounds, mixtures, organometallics, and counterions.
- Once the information was extracted, toxicological reviewers classified each chemical as a "liver carcinogen," "other carcinogen," or "non-carcinogen" for male and female mice and rats. Some chemicals were unable to be classified and were categorized as "other."
The NCTRlcdb will facilitate the construction of cleaner and better carcinogenicity models by FDA and other organizations, thereby providing a means to rapidly predict carcinogenicity. This will aid FDA reviewers evaluate new chemicals submitted for approval.
Beger, R.D., Young, J.F., and Hong Fang. “Discriminant Function Analyses of Liver-Specific Carcinogens.” Journal of Chemical Information and Computer Science, 44:1107–1110, 2004. Abstract
Young, J.F., Tong, W., Fang, H., Xie, Q., Pearce, B., Hashemi, R., Beger,R.D., Cheeseman, M.A., Chen, J.J., Chang, Y.I., Kodell, R.L. "NCTRlcdb is described in Building an Organ Specific Carcinogenic Database for SAR Analyses." Journal of Toxicology and Environmental Health, Part A, 67:1363–1389, 2004. Abstract