Diabetes Influences Relative Effects of Ticagrelor, Prasugrel in ACS: ISAR-REACT 5

For patients with ACS in whom an invasive strategy is planned, diabetes status appears to affect the relative safety and efficacy of ticagrelor versus prasugrel, a prespecified analysis of the ISAR-REACT 5 trial shows.

Consistent with the surprising finding from the main trial, the composite endpoint of death, MI, or stroke at 12 months was higher among nondiabetic patients treated with ticagrelor versus prasugrel (8.6% vs 5.2%: HR 1.70; 95% CI 1.29-2.24). But no significant difference was seen between groups among the patients with diabetes (11.2% vs 13.0%; HR 0.84; 95% CI 0.58-1.24). The interaction with diabetes status was significant (P = 0.0035 for interaction).

Bleeding did not differ between the ticagrelor and prasugrel arms of the trial regardless of diabetes status (P = 0.732 for interaction), lead author Gjin Ndrepepa, MD (German Heart Center Munich and Technical University of Munich, Germany), and colleagues report in the October 12, 2020, issue of JACC: Cardiovascular Interventions.

Asked which agent should be preferred for patients with ACS and diabetes considering data from this and other analyses, study co-author Adnan Kastrati, MD (German Heart Center Munich and Technical University of Munich), said via email, “The choice of P2Y12 antagonist in patients with ACS intended for an invasive treatment strategy should rely on the main efficacy and safety results of the ISAR-REACT 5 trial in favor of prasugrel. The fact that the subgroup analysis focused on diabetics did not show a significant disadvantage of prasugrel as compared with ticagrelor offers no support for treating them differently from nondiabetic ACS patients.”

The ISAR-REACT 5 trial took everybody by surprise, however. That includes the investigators, who set out with the hypothesis that ticagrelor would fare better than prasugrel. Outside commentators have raised concerns about the trial.

In an editorial accompanying the current diabetes analysis, for instance, John Bittl, MD (AdventHealth, Ocala, FL), and colleagues “propose that the findings in ISAR-REACT 5 and the subgroup analysis were statistical ‘false positives.’”

Bittl, speaking with TCTMD, said, “When an investigator gets an unexpected result in a trial, like the original ISAR-REACT 5 trial, we can rule out bias, but we can’t rule out a false positive. It’s analogous to performing stress tests or even checking COVID antibodies in a low-risk population. False-positive results are just as likely or even more likely than true-positive results, and the same kind of thinking applies to trials.” That uncertainty would then get magnified in any secondary analyses, he said.

ACS and Diabetes

According to Kastrati, the rationale for taking a deep dive into the impact of diabetes on treatment in an ACS population is this: “Diabetic patients present with platelet hyperactivity as compared with nondiabetic patients. Plasma levels of the active metabolite of prasugrel may be reduced in diabetics. Ticagrelor is an active drug that does not need any metabolization step. This and its pleiotropic effects might be of particular benefit in diabetic patients.”

This prespecified analysis of ISAR-REACT 5 included 892 patients with diabetes and 3,124 without diabetes. The ticagrelor and prasugrel arms were generally well matched in both the diabetes and nondiabetes cohorts, except that a higher proportion of prasugrel-treated patients had prior CABG in the diabetes group.

Looking at the individual components of the primary composite outcome, there were no differences between the ticagrelor and prasugrel groups among patients with diabetes. The same held true for stent thrombosis. In the nondiabetic cohort however, prasugrel provided a numerical advantage over ticagrelor for death (3.0% vs 4.1%; P = 0.077), along with significantly lower rates of MI (2.1% vs 4.6%; P < 0.001) and definite stent thrombosis (0.3% vs 1.0%; P = 0.020).

The safety endpoint was BARC type 3 to 5 bleeding at 12 months, and this did not differ between the ticagrelor and prasugrel arms either in patients with diabetes (6.9% vs 5.5%; P = 0.425) or in the nondiabetes cohort (5.2% vs 4.6%; P = 0.500).

“Although ticagrelor and prasugrel showed comparable efficacy and safety in the present study [in the diabetic group], specifically designed randomized and powered studies are needed to establish the most optimal antithrombotic therapy in patients with diabetes mellitus presenting with ACS and undergoing invasive therapy,” the investigators conclude.

A ‘False-Positive’ Finding?

In their editorial, Bittl et al perform a series of statistical calculations to support their contention that the overall and diabetes results from ISAR-REACT 5 are false positives. But Kastrati, along with study co-author Alexander Hapfelmeier, MSc (Institute of General Practice and Health Services Research, Technical University of Munich), the trial’s statistician, pushed back on that assertion.

“At the time the ISAR-REACT 5 trial was launched, there was still no direct comparison but an equipoise between ticagrelor and prasugrel, as reflected by the similar strength of guideline recommendation regarding them in patients with ACS,” Kastrati and Hapfelmeier said in an emailed response. “The equipoise informs us to use 1:1 odds in the Bayesian approach rather than the arbitrary 1:9 values used by the editorialists. Using the 1:1 odds, the main ISAR-REACT 5 finding has a 94% chance of being true.”

Commenting for TCTMD, Robert Storey, DM (University of Sheffield, England), who was not involved in the study, said “a false positive is certainly a credible option,” citing the unexpected imbalance in noncardiovascular deaths favoring prasugrel.

But the fact that MI rates did not differ between the two drug arms in the diabetes group but were lower with prasugrel versus ticagrelor in the nondiabetes group, in addition to higher-than-expected stent thrombosis and MI rates in the ticagrelor-treated patients without diabetes, points to another explanation for the discrepant results in Storey’s view: worse adherence to treatment in the ticagrelor arm.

Adherence would be less of a problem in patients with diabetes, he explained, because they’re used to taking medication to control chronic conditions and reduce cardiovascular risk. In nondiabetics, on the other hand, the ACS event might be the first time they’re being asked to take a daily medication, and thus they might be less trained to do so, Storey suggested. “I think poor adherence in the nondiabetes patients is probably a more likely explanation for the findings than a false positive.”

It could be that dyspnea, a well-known side effect of ticagrelor, was responsible for many of the discontinuations, Storey said. “If patients were receiving ticagrelor shortly before being sent home, it may be that they haven’t had sufficient education of the importance of adhering to their medication and not stopping it without seeking medical attention if they got an adverse effect.”

Debating the Optimal Antiplatelet Therapy

Regarding treatment decisions, Bittl said clinicians have “basically a choice of three reasonably good agents—ticagrelor, prasugrel, and even clopidogrel, for diabetic patients.” But he laid out an argument for why clopidogrel might be preferred over the others, pointing out that the pivotal trials establishing ticagrelor and prasugrel in the setting of ACS—PLATO and TRITON-TIMI 38, respectively—were completed more than a decade ago, when most patients received BMS or first-generation DES that are not used anymore. Since then, four trials of antiplatelet therapy escalation based on platelet function testing have been negative, as has the TAILOR-PCI trial evaluating genotype-based P2Y12 prescribing after PCI. Also, the POPULAR AGE trial showed that ticagrelor increased bleeding without reducing thrombotic events versus clopidogrel in older NSTE ACS patients.

Tipping the scales toward clopidogrel and away from ticagrelor and prasugrel is patient preference, Bittl said, noting that he treats mostly older patients, who “do not like the more-potent agents.” In particular, they dislike the dyspnea side effect, the twice-daily dosing, the cost, and the bleeding risk that come along with ticagrelor, he said. Younger patients, too, prefer clopidogrel, he added, because of the lower cost.

“All three drugs, in terms of efficacy, I think they’re equivalent,” Bittl said. “In terms of safety, I think clopidogrel is the winner. And in terms of patient tolerability, nothing can compare with clopidogrel.”

Kastrati rebutted: “Patient and physician preferences are not a good substitute for evidence. The evidence is strong in favor of the superiority of prasugrel and ticagrelor over clopidogrel in ACS patients, and [that’s] embraced by expert guidelines.”

Storey, who was a member of the executive committee for the PLATO trial, also disagreed with the stance that clopidogrel has an advantage over the more-potent P2Y12 inhibitors, although he said it would be understandable from the US perspective for patients to prefer the less-expensive clopidogrel.

“If patients were asked to choose what they prefer after education about the efficacy of the drugs, then it may well be that they would prefer ticagrelor on the basis that they would be better protected from stent thrombosis and other thrombotic events with that strategy,” he commented.

Moreover, “from a pharmacological point of view, we know that clopidogrel is unreliable in all patients, but particularly in diabetes patients, who have a tendency towards lesser platelet inhibition with clopidogrel compared to nondiabetes patients. Those patients have a higher atherothrombotic risk and so really need higher levels of platelet inhibition,” Storey advised. “I think there’s quite a danger to arguing for first-line use of clopidogrel in diabetes patients with acute coronary syndromes given the overwhelming evidence we’ve shown for higher efficacy of the more-potent agents, ticagrelor and prasugrel, particularly in diabetes patients.”