Search

Interpretation of endpoints (e.g. overall response rate) in clinical trials depends on the accurate and reliable measurement and identification of tumors. Regulatory agencies recommend blinded reviews of imaging data by independent review committees (IRCs). Differences in response outcomes that arise between IRCs and site investigators raise regulatory/sponsor concerns. Here, we evaluate discrepant tumor response assessments by the IRC and unblinded investigators (complete versus partial response, respectively) occurring in 52 (13% of 393 IRC-assessed responders) of 447 enrolled patients with treatment-naïve non-Hodgkin lymphoma from a randomized study. The IRC and investigators were ‘likely correct’ in 73% and 25% of cases, respectively (p < .001). Investigators were more likely to make errors by misinterpreting lymph node data and not utilizing PET results. Thispost hocfinding suggests a possible role for post-training site evaluation/audit, with retraining as needed, and a specialized consensus committee for concurrent blinded review of site/central data. [ABSTRACT FROM PUBLISHER]