Reporting of cancer clinical trial adverse drug events: A RADAR report.

Abstract

6090 Background: The timely identification of unknown toxicities of cancer drugs is an important, unsolved problem. Because of their cytotoxic effects, distinguishing cancer drug toxicity from disease progression or comorbidity can be difficult. Also, because of the need for novel cancer therapies, regulatory approval is typically based on fewer than 3000 patients. The adverse event detection failure rate during cancer clinical trials has not previously been determined and there has been relatively little evaluation of the utility of applying validated methods for ADE abstraction in this setting. Here, we compare routine adverse event reporting to structured case abstraction (SCA) of the research record and medical record. Outcomes of interest included detection, description, severity grading, and causality attribution. Methods: We obtained all adverse event data (2001-2008) reported to the Northwestern University Institutional Review Board (IRB) for all six clinical trials involving bevacizumab or oxaliplatin. We compared routine IRB reports to SCA for ability to detect an adverse event and for utility of assessing causality attribution with the 10-item Naranjo Scale. Results: Of the 205 adverse events found in the research record and medical record, 182 events (75%) were not reported; six of the 30 (20%) events with a severity grade of “severe,” “life-threatening,” or “fatal” and a causality assessment of “possible”, “probable”, or “definite” were unreported. The mean number of Naranjo items for which there was useful information was higher with SCA than with the standard IRB report (6.0 vs. 2.4 items, p<0.0001). Conclusions: One-fifth of “reportable” adverse events were unreported to the IRB via routine adverse event reporting. SCA also provided more useful information as to whether the adverse event was caused by the cancer drug exposure. Our results suggest that SCA has potential to improve adverse event detection and to improve the accuracy of causality attribution during cancer drug clinical trials. Use of SCA may reduce delays in recognition of novel cancer drug safety signals. There is an urgent need to prospectively evaluate use of structured case abstraction during cancer drug clinical trials.