Viewpoint: Central adjudication of myocardial infarction in outcome-driven clinical trials--common patterns in TRITON, RECORD, and PLATO?

Abstract

Central adjudication in randomised controlled outcome-driven trials represents a traditional approach to maintain data integrity by applying uniformed rules for assessment of clinical events. It was the purpose of this investigation to determine the patterns of myocardial infarction (MI) adjudication in the TRITON, RECORD, and PLATO trials. We were matching centrally-adjudicated MI's (CAMI's) from the official trial publication with the site-reported MI (SRMI's) count from the Food and Drug Administration's secondary analyses for the investigational compounds prasugrel (TRITON), rosiglitazone (RECORD), and ticagrelor (PLATO). CAMI numbers showed a remarkable discrepancy to SRMI's by more than a doubling of the difference: from 72 to 145 events in TRITON favoring prasugrel (from a hazard ratio [HR]=0.76, p=0.08; to a HR=0.76, p<0.001), and from 44 to 89 events in favour of ticagrelor in PLATO (from a HR=0.94, p=0.095; to a HR=0.84, p<0.001). In contrast, in the RECORD trial, the CAMI count was less than the SRMI count (from 24 to 8 events, from a HR=1.42, p=0.93; to a HR=1.14, p=0.96), in this case diminishing cardiovascular hazards in favour of rosiglitazone. In conclusion, central adjudication in the TRITON, the RECORD, and the PLATO trial turned out to have a critical impact on study outcomes. Trial publications should in the future include site-reported major efficacy and safety endpoints to preserve data integrity. The regulatory authorities should consider independent audits when there is a major disagreement between centrally adjudicated and site reported events influencing the results of a major clinical trial.