Abstract 988: Do High-Risk Characteristics (History of Stroke or TIA, Age >75 years, Weight <60 Kg) Reliably Predict Bleeding in TRITON-TIMI 38?
Background: The TRITON-TIMI 38 investigators reported that “with exclusion of patients with prior stroke/TIA and dose reduction in the elderly (>75y) and those with low body weight (<60kg), the bleeding risk with prasugrel will be minimized”.
Objective: To critically examine whether the three high-risk characteristics (history of stroke or TIA, age >75 years, weight <60 kg) reliably identify patients at higher risk of bleeding associated with prasugrel compared with clopidogrel.
Methods: Benefit (all-cause death, nonfatal MI, nonfatal stroke) and risk (non-CABG TIMI Major bleeding) outcomes reported in TRITON-TIMI 38 were assessed separately and as combined net clinical benefit in the high-risk and non-high-risk subgroups as hazard ratios (95% CI). Between-group comparisons were assessed via interaction test.
Results (Table⇓): Examination of the data reveals that while the high-risk subgroup was numerically at a higher risk for bleeding with prasugrel (42% vs 24% relative increase in non-high-risk), the between-group difference was not statistically significant (interaction P=0.64). In contrast, the between-group difference in ischemic events (26% decrease vs 2% increase in high-risk, interaction P=0.008) and net clinical benefit (20% improvement vs 7% worsening in high-risk; interaction P=0.006) was significantly in favor of prasugrel in the non-high-risk subgroup.
Conclusion: In TRITON-TIMI 38, the high-risk characteristics identify patients with improved benefit, not reduced bleeding risk, associated with prasugrel, thereby directly controverting the claims of the TRITON investigators. Combining the efficacy endpoint (benefit) with the safety endpoint (risk) into a “net clinical benefit” endpoint appears to obscure this subtle, but important, distinction. This has important implications for regulatory approval and clinical practice.