Evolution of predictive risk factor analysis for chemotherapy-related toxicity.

Abstract

The causes of variation in toxicity to the same treatment regimen among seemingly similar patients remain largely unknown. There was tremendous optimism that the patient's germline genome would be strongly predictive of treatment-related toxicity and could be used to personalize treatment and improve therapeutic outcomes. However, there has been limited success in discovering robust pharmacogenetic predictors of treatment-related toxicity and even less progress in translating the few validated predictors into clinical practice. It is apparent that identification of toxicity predictors that can be used to predict and prevent treatment-related toxicity will require thinking beyond germline genomics. To that end, we propose an integrated biomarker discovery approach that recognizes that a patient's toxicity risk is determined by the cumulative effects of a broad range of "omic" and non-omic factors. This commentary describes the limited success in discovering and translating clinical and pharmacogenetic toxicity predictors into clinical practice. We illustrate the evolution of cancer toxicity biomarker discovery and translation through studies of taxane-induced peripheral neuropathy, which is one of the most common and debilitating side effects of cancer treatment. We then discuss the opportunities for discovering non-genomic (e.g., metabolomic, lipidomic, transcriptomic, proteomic, microbiomic, medical, behavioral, environmental) and integrated biomarkers that may be more strongly predictive of toxicity risk and the potential challenges with translating integrated biomarkers into clinical practice. This integrated biomarker discovery approach may circumvent some of the major limitations in toxicity biomarker science and move precision oncology treatment forward so that patients receive maximum treatment benefit with minimal toxicity.