Response to Letter Regarding Article, "Greater Clinical Benefit of More Intensive Oral Antiplatelet Therapy With Prasugrel in Patients With Diabetes Mellitus in the Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel: Thrombolysis in Myocardial Infarction 38"
We appreciate the letter by Drs Rosenstein and Parra about the diabetes mellitus (DM) analysis of the Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel: Thrombolysis in Myocardial Infarction (TRITON-TIMI) 38. 1 We agree with the general principle that the overall results of a clinical trial should be considered to apply to those patients who were included in the clinical trial and that subgroup analyses should be interpreted with caution. The
... with caution. The results should be considered to be hypothesis generating. Clinical trials in general are, and TRITON-TIMI 38 in specific was, not powered for efficacy end points in all individual subgroups. It is impractical to repeat an adequately powered clinical trial in each individual subgroup. We believe that the altered biology of platelet function in DM, higher rates of poor response to clopidogrel in patients with DM, 2 and previous experience showing greater benefit of intensive antiplatelet therapy with glycoprotein IIb/IIIa receptor blockers in patients with DM 3 all make the hypothesis-generating results of our subgroup analysis biologically plausible and clinically compelling. We too look forward to future antiplatelet therapies being tested prospectively and directly in patients with DM. The authors of the letter state that statistical significance (presumably of the interaction term, although this is not specifically noted in the letter) should be corrected for the number of subgroups tested and interpreted in this context. Their commentary applies these criteria selectively, however. The authors challenge the results of the diabetes analysis that showed a statistically significant benefit in patients with and without DM but with a statistically greater relative benefit among diabetic patients (P ϭ 0.05 for the outcome of death/ myocardial infarction/stroke/major bleed) while highlighting the "concern" raised by subgroups without a statistical interaction for the same composite outcome (women, P interaction ϭ 0.36; age Յ65, P ϭ 0.16; creatinine clearance Ͻ60 mL/min, P interaction ϭ 0.50). Further, the authors state that the major benefit of prasugrel was associated with the loading dose only and was observed in the first 3 days, referencing a speculative editorial. 4 Our previously published data have addressed this question directly in the overall TRITON-TIMI 38 population showing a similar and significant net benefit of prasugrel both within 3 days (hazard ratio [HR] ϭ 0.85, P ϭ 0.02) and beyond 3 days (HR ϭ 0.87, P ϭ 0.03), demonstrating benefit after loading and maintenance dosing. 5 For the DM population, a similar net benefit of prasugrel was found both within 3 days (HR ϭ 0.74, P ϭ 0.06) and beyond 3 days (HR ϭ 0.73, P ϭ 0.005) and also within 30 days (HR ϭ 0.68, P ϭ 0.003) and beyond 30 days (HR ϭ 0.76, P ϭ 0.03).