Structure toxicity relationships--how useful are they in predicting toxicities of new drugs?

This chapter provides just a few newer examples of structural moieties found in drugs that have been associated with reactive metabolite formation and toxicities. For a discussion of several other structures in drugs that undergo metabolic activation to reactive intermediates, the reader is directed to previous volumes in this series and other chapters in this book, as well as a previous condensed review (Nelson, 1982). Since that review, some new knowledge allows us to better predict that some structural moieties are more likely than others to form drug reactive metabolites that may be involved in causing toxic effects in humans. For example, most aniline-, thiophene-, and nitroaromatic-containing drugs have had a relatively high incidence of adverse effects, and it would be prudent in the drug discovery process to avoid these substructures if possible. However, as illustrated by the case of olanzapine, these structures may be important for potent activity, and could therefore be beneficial in some cases. The glitazones represent a new class of drugs with a unique thiazolidinedione structure. This raises two important points. First, it demonstrates how limited our knowledge base is in regard to structure toxicity relationships when new structures are introduced. Our approaches must be very empirical and are far from quantitative for the reasons outlined in the introduction. Secondly, the glitazones point out the importance of benefit/risk considerations. This was a new structural class of drugs with a unique spectrum of action that is very beneficial in the treatment of a major disease. Despite some suspected risk of toxicity, based on early trials, troglitazone was approved for use with careful monitoring. This author believes that was the right decision, as was the decision to withdraw the drug when the risk became unacceptable, especially with the introduction of safer alternatives. If this were just another NSAID (e.g., bromfenac), there would be little reason for approval. In summary, as I pointed out previously (Nelson, 1982), with our limited knowledge of structure toxicity relationships, we can only make reasonable judgments as to risk assessment of a new drug in humans, and hope that we neither release a dangerous chemical entity nor, as importantly, abort an effective one.