When 2 clinical trials compare 2 new drugs using the same comparator and in similar patients, it is tempting to compare the results to decide which of the new drugs is better.

We recently came across such a comparison during our teaching activities in connection with the results from the TRITON-TIMI 38 (TRial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet InhibitioN with Prasugrel–Thrombolysis In Myocardial Infarction),1 which tested the efficacy of prasugrel vs clopidogrel in patients with acute coronary syndrome (ACS) treated with percutaneous coronary intervention (PCI) compared with those of the Platelet Inhibition and Patient Outcomes (PLATO) trial,2 which evaluated the efficacy of ticagrelor also in comparison with clopidogrel in patients with ACS. All 3 drugs are inhibitors of the platelet P2Y12 receptor and are used in combination with acetylsalicylic acid and sometimes with glycoprotein IIb/IIIa inhibitors in patients with ACS.

Although a study indirectly comparing the efficacy of prasugrel and ticagrelor was recently published,3 we believe that the comparison was inappropriate due to methodological differences between the studies, differences in the characteristics of the patients included and, to a lesser extent, the length of time between the studies.

There was an interval of 2 years between the 2 studies, which may have led to small differences in clinical practice that the studies did not measure. Although both studies used the same main event used to measure efficacy, differences in the design, the duration of follow-up, the inclusion criteria, and the dose of the drugs being tested limited the comparability of results. By protocol, in TRITON all patients were assigned to coronary catheterization without initial dual antiplatelet therapy; candidates for angioplasty were randomized to prasugrel or clopidogrel. As a result, all patients in this study underwent angioplasty. In contrast, patients in the PLATO study were randomized to ticagrelor or clopidogrel on admission to hospital regardless of whether they were to receive catheterization and/or coronary angioplasty. Thus, there were differences between the 2 studies not only in terms of patient characteristics, but also in the antiplatelet therapies used.

We compared the baseline characteristics of patients included in the 2 trials (Table A), and the incidence of events to 12 months in patients treated with clopidogrel (Table B). PLATO included patients with more severe acute events than TRITON (Table). For example, the PLATO trial included about 38% of patients with acute myocardial infarction and ST-segment elevation (STEMI) and 65% with PCI, while the TRITON study included approximately 26% of patients with STEMI, all with PCI. Antiplatelet therapy differed in the 2 studies: in PLATO, 46.1% received clopidogrel before randomization and 19.6% received a loading dose of clopidogrel≥600mg, while in TRITON no patient received prior clopidogrel and the loading dose was always <600mg. Furthermore, in TRITON, 55% of patients received triple antiplatelet therapy with glycoprotein IIb/IIIa inhibitors vs 26.8% in PLATO.

In terms of events, the median follow-up in the PLATO trial was 9.1 months compared to 14.5 months in the TRITON study. A comparison of the incidence of cardiovascular events in the control group (treated with clopidogrel and acetylsalicylic acid) in the 2 clinical trials showed that the rate was higher in the PLATO control group (Table B). These data support the idea that there were differences in the clinical risk profiles of patients in the 2 studies.

We believe that any comparison between prasugrel and ticagrelor using the published results of these 2 studies is not valid. A conclusion might be possible if participants’ individualized data were available, by selecting subgroups, and/or by adjusting exposure to the study drugs by relevant covariates. Clearly, though, as the recent National Institute for Clinical Excellence guidelines for the use of ticagrelor indicate,4 a new clinical trial with sufficient statistical power to compare the 2 drugs is needed. Indirect comparisons in which patient characteristics differ are inappropriate, should be discouraged, and the scientific community should actively avoid this type of analysis.