Boston, MA - Whether or not the benefits of the new antiplatelet drug prasugrel (Effient, Lilly/Daiichi Sankyo) suggested in the TRITON-TIMI 38 trial are clinically worthwhile has been the subject of recent discussion. The trial, first reported in November 2007, showed fewer ischemic events with prasugrel than with clopidogrel (Plavix, Sanofi-Aventis/Bristol-Myers Squibb), but at the expense of an increase in major bleeding [1]. Mortality did not differ significantly between the two groups.

The major benefit seen with prasugrel in TRITON-TIMI 38 was a reduction in nonfatal MI. There was also a large reduction in stent thrombosis. Several observers have asked for more information about these events to determine whether the benefit of prasugrel is worth the increased bleeding risk.

In the most recent exchange, which takes place in the correspondence column of the March 20, 2008 issue of the New England Journal of Medicine, Dr Victor Serebruany (Johns Hopkins University, Baltimore, MD) notes that the efficacy benefit of prasugrel in TRITON is driven by nonfatal MI but that "it is not clear whether the numbers . . . reported represent real clinical scenarios or just the increase in cardiac ischemic biomarkers that is so common during successful reperfusion" [2].

In a second letter, a group of Italian doctors make a similar point, saying: "The use of prasugrel might involve trading a reduction in the risk of small MIs for an increase in the risk of major bleeding" [3].

Serebruany commented to heartwire that "there has been much discussion about the risk/benefit ratio of prasugrel, but we can't understand this until we know exactly what the benefit is."

TRITON investigators Drs Stephen Wiviott, Eugene Braunwald, and Elliott Antman (Brigham and Women's Hospital, Boston, MA) reply that the risks of both procedural and nonprocedural (spontaneous) MIs were significantly lower in the prasugrel group, as was the risk of new ST-elevation MI [4]. "Even if data for procedural MI were excluded from the analysis, both the primary end point and the secondary end point of a net clinical benefit would still significantly favor prasugrel," they write.

Serebruany raised similar concerns about stent thrombosis, pointing out that the reduction reported with prasugrel was in the combination of "definite and probable" stent thrombosis. "The TRITON investigators report a large reduction in stent thrombosis, but if this is real, I wonder why this did not translate into a reduction in mortality," he commented to heartwire.

The TRITON investigators responded that the risk of definite stent thrombosis was significantly reduced by the use of prasugrel. Antman added to heartwire that both overall cardiovascular mortality in the entire trial and cardiovascular mortality after stent thrombosis were numerically reduced with prasugrel. As previously reported, cardiovascular mortality following a subsequent MI was significantly reduced, which is relevant because stent thrombosis causes subsequent MIs, he explained.

Commenting on these issues for heartwire, Dr Deepak Bhatt (currently at Cleveland Clinic, OH, but moving to Brigham and Women's Hospital later this year), who wrote the editorial [5] accompanying the original TRITON paper, said that he thought the questions about what type of MI and stent thrombosis were reduced were legitimate and that other doctors had raised these concerns, but he also thought that these questions had been answered quite well in the TRITON investigators' response.

"I think this was an important exchange of letters, in that the TRITON investigators have now stated that nonperiprocedural and definite stent thrombosis were both significantly reduced with prasugrel and that the net clinical benefit still holds up when periprocedural MIs are excluded. It would be nice to see the numbers on this, but I think these will be presented at meetings later this year. I think these questions have been worrying others, but they seem to have been addressed in the authors' response—as long as we see the numbers in the not-too-distant future."

Others were not quite as easily placated by the TRITON investigators' response. Dr Sanjay Kaul (Cedars-Sinai Medical Center, Los Angeles, CA) pointed out that it was essential to see the actual numbers on the types of MI prevented before reaching any judgment on the risk/benefit ratio of prasugrel.

"This trial is very large, and even if a reduction in nonprocedural MI is statistically significant, it may still be quite small. I would want to know what proportion of the MIs prevented by prasugrel was nonprocedural before deciding whether the benefit is worthwhile. This drug has a major trade-off between efficacy and bleeding; I would want to know exactly what type of MI and exactly how many of each type are being prevented before deciding whether it's worth the bleeding risk," Kaul commented to heartwire.

The TRITON-TIMI 38 investigators indicated that details of the findings regarding MI will be appearing in peer-reviewed publications in the future but summarized the principal observations by saying that "significant reductions in MIs with prasugrel were seen both early and late in the trial. The absolute and relative reductions in spontaneously occurring MIs were greater than those for periprocedural MIs. The largest reductions (both absolute and relative) in MI occurred in those subjects who had the largest infarcts, assessed by biomarker release."

Braunwald also commented on another benefit of prasugrel noted in TRITON-TIMI 38: "After the first nonfatal element of the primary end point, patients who were assigned to prasugrel had significantly fewer subsequent events—both fatal and nonfatal. Thus, the total number of events (initial and subsequent) over the trial were markedly reduced with prasugrel."

Kaul pointed out that for composite end points, such as those used in TRITON, to be valid, three prerequisites are necessary: components should be equally frequent; components should be of equal or comparable clinical importance; and components should be equally or comparably responsive to therapy.

"These criteria were not met in TRITON. There was a large gradient in the prevalence (nonfatal MI >> CV death > stroke), clinical importance (CV death > stroke > nonfatal MI), and treatment effect (nonfatal MI >> CV death > nonfatal stroke) across the components. Therefore, Serebruany is correct in stating that the benefit of prasugrel in TRITON was driven by nonfatal MI—the most prevalent but the least important component of the composite end point."

Kaul pointed out that similar patterns were observed in the CURE trial, where the treatment benefit was driven in favor of clopidogrel primarily by the impact on nonfatal MI. However, unlike CURE, where two-thirds of MI benefit was due to the impact on Q-wave MIs, "the relative proportion of benefit across the different types of MI is not readily apparent from the TRITON data presented so far," he added.

Antman told heartwire that more data on stent thrombosis in TRITON would be presented at the American College of Cardiology (ACC) meeting later this month, and an abstract on the MI data had been submitted to the European Society of Cardiology (ESC) meeting taking place in the summer.

Regarding the primary end point of the trial, Antman emphasized that "while discussions of the merits of composite end points go far beyond the scope of the design of TRITON-TIMI 38, the vast majority of clinical trials in acute coronary syndromes [ACS] and heart failure employ a composite end point. Also, the TRITON-TIMI 38 composite end point was in compliance with guidance documents from regulatory agencies and was approved by the FDA and [European Medicines Agency] EMEA before the trial began."

Serebruany also raises these and other issues on TRITON in an editorial published online March 7, 2008 in the American Journal of Cardiology [6], in which he suggests that the definition of MI used in TRITON was too liberal, reflected by the higher rate of nonfatal MIs in this trial than in other trials with clopidogrel and in ACS patients. He writes: "If a more liberal definition of MI is applied and the diagnostic window for MI is expanded, it may explain the increased rate of nonfatal MI, suggesting that some of the qualifying episodes represent transitory enzymatic 'bombs' rather than real clinical events." "It is like playing hockey with larger nets and a smaller puck and then trying to match the results with an NHL game," he commented to heartwire.

He added that the TRITON trial "may create an illusion that, for future drug development, it is fine to increase the dose of an antiplatelet agent, ignore alarming bleeding risks, and compare it with the suboptimal regimen of standard of care."

Responding to this, Antman commented to heartwire that the definition of MI, which was described in the TRITON design paper published in the American Heart Journal in 2006 [7], was actually more stringent than the universal definition referred to by Serebruany. "The definition of periprocedural MI in TRITON-TIMI 38 required either that CK-MB be more than three times the upper limit of normal [ULN] on two separate occasions or five times the ULN on a single measurement. This definition is in fact more stringent than the 'Universal definition of myocardial infarction' published jointly by the ESC, the ACC Foundation, the American Heart Association, and the World Heart Federation in 2007, which requires biomarker elevation of more than three times the ULN," he said.

Antman argued that MI rates from different trials should not be compared because populations enrolled will differ. But he added: "The ACUITY trial had the closest population to TRITON-TIMI 38. In TRITON-TIMI 38, at 30 days the rate of nonfatal MI in the clopidogrel group was 6.3% and similar to the MI rates (6%-7%) observed in the PCI cohort of ACUITY."

Antman added that a full response from the TRITON-TIMI 38 investigators to Serebruany's editorial will appear shortly in the American Journal of Cardiology [8].

Other experts contacted agreed that indirect cross-trial comparisons should generally be avoided because there are too many other variables to make comparisons valid.

Most were also not overly concerned about the definition of MI used in TRITON. Bhatt commented to heartwire that "different definitions of MI are often used in different studies, and this is not really a big deal as long as the definition is prespecified. For me, the main limitation of TRITON is that it involved a pretty defined patient population—ACS patients who were known to be going to PCI. But within that population, there does appear to be a clear-cut reduction in events."

Dr Robert Harrington (Duke Clinical Research Institute, Durham, NC) voiced a similar opinion, noting that the MI definition used in TRITON was consistent with that used in other trials in both ACS and PCI and, in his view, it was entirely appropriate to include periprocedural MIs because they had been correlated with later mortality.