Toxicity predictions, criticism

Recently, Isshi et al. demonstrated that high levels of orotate phosphorybosyl transferase (OPRT) may be associated with increased sensitivity to 5-FU based chemotherapy (79). OPRT catalyzes the reduction of FUDP to the actively TS-inhibiting metabolite FdUMP, indicating a role for chemosensitivity to 5-FU. A recent study by Ichikawa et al. indicated that a newly identified SNP of OPRT exon 3 (G-to-A substitution) may be critical to predict toxicity to 5-FU based chemotherapy (80). 

The ECVAM models were designed to facilitate the blinded categorization of a broad array of structurally unrelated agents into three categories of embryotox-icity. Thus, the correct embryotoxicity prediction is critical. We do not use the tests in that way instead, we use them to rank-order the relative developmental toxicity risk within or across chemical series, to identify the agent with the least inherent risk- irrespective of the ECVAM-predicted embryotoxicity class. We are not alone in this approach, as the chick embryo retinal culture is also used for ranking (G. Daston (daston.gp pg.com), personal communication). 

Computational modeling is a powerful tool to predict toxicity of drugs and environmental toxins. However, all the in silico models, from the chemical structure-related QSAR method to the systemic PBPK models, would beneht from a second system to improve and validate their predictions. The accuracy of PBPK modeling, for example, depends on precise description of physiological mechanisms and kinetic parameters applied to the model. The PBPK method has primary limitations that it can only predict responses based on assumed mechanisms, without considerations on secondary and unexpected effects. Incomplete understanding of the biological mechanism and inappropriate simplification of the model can easily introduce errors into the PBPK predictions. In addition values of parameters required for the model are often unavailable, especially those for new drugs and environmental toxins. Thus a second validation system is critical to complement computational simulations and to provide a rational basis to improve mathematical models. 

Brown M A and DeVito S C, Predicting azo dye toxicity . Critical Reviews in Environmental Science and Technology, 1993,23(3), 249-324. 

In general, the complex and changing nature of mixtures of naphthenic acids make it difficult to predict toxicity. By determining the critical mechanism of toxicity of naphthenic acids, it might be possible to develop more effective predictive relationships to account for the toxic effects observed in living organisms exposed to naphthenic acids. 

Botham, RA. et al., Skin sensitization—a critical review of predictive test methods in animals and man, Fd. Chem. Toxic., 29, 275, 1991. 

Product quality, purity and consistency are critically important in the pharmaceutical sector, applying to all stages of the supply chain and final dosed product. The human body is an exceptionally complex system and the full effect of a pharmaceutical product, consisting of the API, impurities and formulation components, is impossible to predict from first principles. The industry relies on rigorous clinical trials to assess drug efficacy, toxicity and side effect profiles. 

An a priori classification of these various reactions as either toxification or detoxification is simply impossible, since each product from these various pathways may be toxic or not depending on its chemical properties and own products. Furthermore, the biological context plays a critical role [154], yet this role, best viewed as the influence of biological factors on the relative importance of competitive routes of metabolism, is often underplayed by those who venture to make predictions of metabolic outcome. Indeed, in the cascade of intertwined metabolic routes exemplified by haloalkenes, a small difference in pathway selectivity at an early metabolic crossroad may be amplified downstream, giving rise to major differences in relative levels of metabolites and overall toxicity. 

It is generally accepted that the expression of toxicity in a mammalian system is dependent on a sequence of key events taking place, each of which is critical to the manifestation of the toxic endpoint. If at least one of these key events identified in experimental animals does not occur in humans, then it could be concluded that the toxic endpoint would not be observed in humans, and thus this effect is of limited relevance for the prediction of effects in humans. 

Critical for predictivity in a recent comprehensive study was the number and choice of parameters measured [4]. Early, sublethal effects on cell proliferation, cell morphology and mitochondria occurred consistently and ubiquitously with toxicity and when used collectively were most diagnostic. It is noteworthy that the toxicity of many drugs is attributable to various mitochondrial targets, including oxidative phosphorylation, fatty acid oxidation, Krebs cycling, membrane transport, permeability transition pore, proliferation and oxidative stress (Table 14.4). 

---