Questions surrounding a subanalysis of PLATO data showing that AstraZeneca's experimental antiplatelet drug Brilinta (ticagrelor) apparently didn't help North American acute coronary syndrome patients center not on whether the observation is valid, but on what FDA's reaction will be.

The consensus seems to be that FDA will look hard at the data but that the drug's overwhelming efficacy and safety profile will win out in the end, and the agency will chalk up the anomaly to chance. Alternatively, regulators could require a confirmatory trial in North American patients.

"I think this is a classic subgroup analysis which is throwing up something by chance," Marcus Flather, head of the Clinical Trials and Evaluation Unit at the Royal Brompton Hospital, London, said during a Sept. 1 call hosted by Morgan Stanley to discuss trial results, including PLATO, that were released at the European Society of Cardiology congress in Barcelona Aug. 30.

Flather said he doubts FDA will be convinced the data are representative "because the strength of the evidence in the trial overall, particularly the mortality benefit," will persuade people ticagrelor is really worthwhile.

"The Totality Of The Data"

Cowan analysts agree. In an Aug. 31 report they called the North America data "a potential regulatory issue," but said "FDA should look at the totality of the data."

The 18,624-patient PLATO study (the Study of Platelet Inhibition and Patient Outcomes) compared ticagrelor plus aspirin head-to-head with Bristol-Myers-Squibb/Sanofi-Aventis' Plavix plus aspirin. Brilinta was associated with a 16 percent relative risk reduction with regard to the primary endpoint - a composite of death from cardiovascular causes, myocardial infarction and stroke (p<0.001). the="" incidence="" of="" major="" bleeding="" was="" 11.6="" percent="" versus="" 11.2="" percent="" (p="">

The event rate reduction was driven by a reduction in cardiovascular deaths (4.0% vs. 5.1%, p=0.001) and myocardial infarction (5.8% vs. 6.9%, p=0.005), with no impact on stroke, a finding that is old news, for developers of antiplatelet drugs.

Ticagrelor also was associated with a surprising overall mortality benefit. The rate of death from any cause was reduced by 5.9 percent with Brilinta compared to 4.5 percent with clopidogrel (p=<0.001). indeed,="" the="" strength="" of="" the="" data="" is="" already="" prompting="" speculation="" that="" the="" product="" might="" crowd="" lilly/daiichi="" sankyo's="" recently="" approved=""> Effient (prasugrel) out of the market (see ¹ Also see "AstraZeneca's Ticagrelor Brilliant At ECS Congress; Will It Outshine Effient?" - Pink Sheet, 7 September, 2009.).

At issue is puzzling data from a subgroup analysis of patients in North America that not only showed no treatment advantage for ticagrelor, but also indicated patients on Brilinta actually had a greater percent chance of having a cardiovascular event than those on Plavix.

While analysts coped with how FDA might perceive the finding, AstraZeneca said it is investigating the inconsistency but it won't impede plans to file an NDA in the fourth quarter.

"Given the large number of tests performed, 33 efficacy and 33 safety, these differences may have been due to chance," the company said in an e-mail.

"It is important to remember that this is a subgroup analysis. Geographical differences between patient populations or practice patterns may have influenced the effects of the randomized treatments, although no apparent explanations have been found to date."

In an Aug. 31 note, Credit Suisse analyst Cathrine Arnold predicted FDA will ask for a detailed analysis, causing a regulatory delay in the U.S. She put the chance of a 2010 Brilinta launch at 60 percent.

Problems With Doing Trials In North America

According to Arnold, the lead investigator on the trial has noted that North America was characterized by a large number of small study sites (<5 patients).="" "given="" that="" errors="" are="" more="" commonly="" made="" in="" the="" first="" few="" patients="" of="" any="" study,="" this="" could="" provide="" an="" explanation,"="" she="">

Among the subgroups analyzed, patients who were not taking a statin on the day of randomization also showed no treatment advantage. Neither did patients who weighed less than the median for their sex.

Another point of concern: the North American results involve 1,814 pateints, only about 10 percent of the whole study cohort.

The small number of patients enrolled at North American centers reflects the difficulty of enrolling U.S. patients in large-scale clinical trials, Gabriel Steg, Hospital Bichat-Claude Bernard, Paris, a primary investigator on the PLATO trial, said during the Morgan Stanley call.

The vast majority of patients in the trial, almost 14,000, were enrolled in Europe, the Middle East and Africa, he said. Hence, the relative risk of having the primary composite endpoint is about 0.8 to 0.9 in Asia, South America, Central America, Europe and Africa, while in North America the chance is "substantial," he said.

The estimate is 1.25 with a hazard ratio of 0.93 to 1.7 and a 95 percent confidence interval. The p value for the interaction is 0.05, "so, borderline, and it reflects indeed the fact that only 1,800 patients were enrolled in North America," Steg explained.

Steg said PLATO investigators have considered a number of possible reasons for the conflicting readout, including the interaction between weight and benefit because U.S. patients tended to be heavier.

But the overall finding of the study is in favor of heavier patients deriving benefit, he said. Another suspected cause was the dose of aspirin used in the U.S., which is somewhat higher than in the rest of the world, but that didn't pan out either, Steg said.

"When you have an interaction p value of 0.05 and you're looking at approximately 20 subgroups, you can pretty reasonably make a case that this is fortuitous and related to the multiplicity of analyses in various subgroups," he concluded. "It doesn't seem to fit with any other biological explanation that seems plausible."

Of course, he said, "I cannot guarantee that this is the way the regulators are going to look at it and whether they are going to request additional studies."

- Shirley Haley ( ² s.haley@elsevier.com )

Brilinta's North America Anomaly Stymies AstraZeneca, Worries Analysts

More from Archive

More from Pink Sheet