AstraZeneca PLC’s demonstration that high-dose aspirin was linked to reduced efficacy for Brilinta (ticagrelor) through a variety of analyses convinced senior FDA review staff that the platelet inhibitor was worthy of approval, even though U.S. patients in the pivotal trial fared worse than those in the rest of the world.

The consistency of results across various methods for characterizing the aspirin maintenance dose of patients in the Phase III PLATO trial proved to be the persuasive evidence needed to overcome FDA’s initial reluctance to approve ticagrelor without a new clinical trial.

The multiple analyses based upon aspirin dosage calculations helped save ticagrelor because agency officials remained unconvinced by the sponsor’s proffered biological explanation for the high-dose aspirin interaction.

The multiple analyses, which were requested by FDA in a “complete response” letter, gave ticagrelor a regulatory lifeline because agency officials remained unconvinced by the sponsor’s proffered biological explanation for the aspirin/ticagrelor interaction.

The disparate U.S. efficacy results in PLATO came close to derailing AstraZeneca’s bid for approval of the oral P2Y12 receptor, despite having shown clear superiority to Bristol-Myers Squibb Co./Sanofi’s Plavix (clopidogrel) across the 18,000-patient study as a whole.

It was no secret that FDA concerns about AstraZeneca’s “aspirin hypothesis” were the key stumbling block in the ticagrelor NDA review, leading to the December 2010 “complete response” letter. However, review documents indicate the breadth of the new analyses provided by AstraZeneca and the consistency of those results gave agency officials enough comfort that they deemed no further studies, either pre- or post-approval, were needed to explore the aspirin/ticagrelor interaction further in acute coronary syndrome patients.

Review documents also reflect an unusually harsh critique of the sponsor and its pivotal data by the Cardiovascular and Renal Products Division cross-discipline review team leader for the ticagrelor NDA – criticism that drew a sharply worded rebuke from the division director (see related story, Also see "Brilinta Review Sparks War Of Words Between FDA Cardio-Renal Division’s Marciniak, Stockbridge" - Pink Sheet, 1 January, 2012.).

Strong Efficacy, Except In The U.S.

FDA approved ticagrelor, a selective and reversible P2Y12 ADP-receptor antagonist, on July 20, 2011 to reduce the rate of thrombotic cardiovascular events in patients with ACS, including unstable angina, non-ST elevation myocardial infarction, and ST elevation MI.

Approval was based on PLATO, a double-blind, randomized, multi-regional study comparing the effects of clopidogrel with ticagrelor in a broad ACS population. Aspirin was a required concomitant treatment, but the dose was chosen by the investigator.

Ticagrelor was associated with a statistically significant 16% reduced risk of a composite endpoint comprising cardiovascular death, non-fatal MI and non-fatal stroke across the study as a whole. The effect on the primary endpoint was driven by significant risk reductions in non-fatal MI (16%) and CV death (21%), with no difference in stroke. Ticagrelor also demonstrated a statistically significant benefit on all-cause mortality compared to clopidogrel, with a 22% reduced risk.

Despite the robust overall results, PLATO presented a big problem for AstraZeneca. As a subgroup, North American patients fared worse than those in the rest of the world, a poor showing driven by the U.S. cohort. While non-U.S. patients across the trial saw a 19% reduced risk of a primary endpoint event, U.S. patients had a non-significant 27% increased risk of experiencing a major CV event.

Failure to show that efficacy in U.S. patients is similar to those outside the country is not necessarily fatal to a drug’s chances of securing FDA approval. The agency has previously granted approval when the benefits in a U.S. subgroup fall shy of those seen in a global clinical trial population as a whole.

A case in point is another AstraZeneca drug, the beta blocker Toprol-XL (metoprolol). A 14-country pivotal trial supporting the heart failure claim was terminated early due to a statistically significant reduction in all-cause mortality. However, the drug appeared to have less effect on overall mortality and CV mortality in the subgroup of U.S. patients.

The tricky, and to FDA concerning, part about ticagrelor was that the U.S. results trended in the wrong direction on all adjudicated endpoints, favoring the comparator clopidogrel. At a July 2010 advisory committee review, FDA Center for Drug Evaluation and Research Deputy Director for Clinical Science and Office of Drug Evaluation I Director Robert Temple called the negative U.S. trend “a little unusual and … a little worse than some of the cases where you just don’t see effect” Also see "Toprol-XL Eyed As Precedent For Brilinta Labeling On U.S. Efficacy Data" - Pink Sheet, 9 August, 2010..

AstraZeneca Has A Theory …

AstraZeneca’s investigation into the subpar U.S. performance offered up the maintenance dose of aspirin as the reason behind the disparity. U.S. patients in the trial were more likely to have been treated with high-dose aspirin (>300 mg/day), while patients outside the U.S. primarily received low-dose aspirin (<100>

AstraZeneca took great pains to explain its theory, both from a clinical and mechanistic standpoint.

At the advisory committee meeting, the company asserted that ticagrelor is a more potent P2Y12 inhibitor than clopidogrel. Aspirin adds antiplatelet benefit predominantly when combined with suboptimal P2Y12 inhibition but will have only minor positive impact when combined with optimal P2Y12 inhibition. Aspirin also reduces prostacyclin levels dose-dependently, and high aspirin dose could negatively affect endothelial function, potentially facilitate platelet activation and ultimately contribute to increased thrombotic risk, AstraZeneca said ([A#00720320005 ]).

The committee ultimately voted 7-1 in favor of approval, but panelists rejected AstraZeneca’s aspirin hypothesis as the sole reason for the poor U.S. efficacy results. The panel said multiple factors may have contributed to the divergent results, including chance ([A#14100728006 ]).

The advisory committee’s endorsement notwithstanding, FDA’s primary efficacy, safety and statistical reviewers recommended against a first-cycle approval, calling for more data to confirm the aspirin hypothesis.

The committee’s endorsement for approval notwithstanding, FDA’s primary efficacy, safety and statistical reviewers recommended against a first-cycle approval, suggesting instead that further data were needed to confirm the aspirin hypothesis and address lingering uncertainties about efficacy in U.S. patients.

Cross-Discipline Team Leader Thomas Marciniak supported approval for ACS, except in STEMI patients undergoing early percutaneous coronary intervention. Marciniak said he also would support complete non-approval until a confirmatory study was conducted or approval for medical management alone, but not unrestricted approval in all ACS patients.

In an Oct. 7, 2010 memo, Cardiovascular and Renal Products Division Director Norman Stockbridge said he favored a “complete response” action pending development of evidence that ticagrelor provides benefit likely to be realized in U.S. practice. “Ideally, that would be an outcome study in the U.S., but it could be independent support for the aspirin hypothesis as an outcome study anywhere and in ACS or some closely related condition,” Stockbridge wrote. “It might also be possible to support the hypothesis that higher degrees of P2Y12 inhibition adversely affect the response to high-dose aspirin using measures short of outcomes.”

In a Dec. 16, 2010 memo, Temple pointed to uncertainty surrounding how the aspirin dose was defined in various analyses that suggested a favorable effect with low-dose aspirin in and outside the U.S., but not with high-dose aspirin.

“It seems clear that we must understand exactly how ASA dose was determined for these analyses, and how much the analyses changed with various decisions,” Temple wrote. “I believe … that if the sponsor can show the ASA dose explanation is robust to differences in methods of determining what dose patients took, ticagrelor should be approved to provide patients with the first P2Y12 antagonist that has had a clear survival effect.”

… But FDA Wants More Data On Dose Calculations

In its “complete response” letter, FDA requested more information about the methods AstraZeneca used to determine the aspirin dose for each subject, up to the time of an endpoint event or censoring. “To enable us to understand the basis for the aspirin categorizations used in these analyses, please provide the specific raw dataset(s), detailed algorithm and corresponding program used to derive the daily aspirin dose for each subject,” the letter said.

The letter acknowledged that AstraZeneca already had employed a number of methods to categorize the aspirin dose for each study subject. However, the agency requested additional analyses using other methods for characterizing the dose.

FDA requested 13 new analyses of the possible interaction between aspirin and ticagrelor using the following methods and definitions for calculating aspirin dose:

Median and mean of daily aspirin doses taken in the last 5 and 10 days prior to the primary event or censoring date

Last aspirin dose taken within 30 days prior to the primary event or censoring date

Median and mean of daily aspirin doses taken in the last month prior to the primary event or censoring date

For analyses of events that occurred within 30 days of randomization, the aspirin dose can be defined as:

Mean of the daily aspirin doses in the first 30 days
Median of the daily aspirin doses in the first 30 days
Maximum of the daily aspirin doses in the first 30 days

For analyses of events that occurred after 30 days from randomization, aspirin dose can be defined as:

Median of the daily aspirin doses throughout the trial excluding the first 30 days
Median of the daily aspirin doses throughout the trial excluding the first day loading dose
Last daily aspirin dose prior to the primary event or censoring date

AstraZeneca performed a quick turnaround on FDA’s information request, submitting its response a little over a month later.

In an April 28, 2011 review of the new analyses, FDA biostatistician Jialu Zhang said that methods for imputing aspirin dose on subjects with incomplete data, and inclusion or exclusion of the first-day loading dose of aspirin, seemed to have the most impact on suggesting whether there was a significant treatment interaction between ticagrelor and aspirin.

“Various ASA definitions appear to demonstrate some degrees of consistency in analyses,” Zhang said. But, she added, “these analyses are still limited by the fact that there were only a small number of high ASA dose subjects in OUS.”

However, she said concern remains as to whether aspirin dose “is truly the only factor that might affect the ticagrelor effect” since there appeared to be no such similar interaction with Eli Lilly & Co./Daiichi Sankyo Co. Ltd.’s antiplatelet agent Effient (prasugrel) in the pivotal TRITON clinical trial (see related story, Also see "Shades Of Prasugrel: Brilinta Review Felt The Effects Of Effient’s Experience" - Pink Sheet, 1 January, 2012.).

In a May 15, 2011 review, Marciniak, who was assigned as the efficacy reviewer for the company’s CRL response, asserted that the ticagrelor/aspirin interaction had not been confirmed. There were more compelling interactions deserving of attention, he said. These included a negative interaction between ticagrelor and early PCI, and favorable interactions between ticagrelor and statin use, and diabetes and aspirin dose.

“All of the focus upon ASA dosage, while ignoring the other interactions, has impeded the understanding of ticagrelor activity,” Marciniak writes. He recommended against approval, saying, “There are many PLATO deficiencies, as well as critical outstanding issues, such that a confirmatory trial is mandatory.”

Stockbridge, Temple Are Persuaded On Approvability

Marciniak’s views aside, the additional aspirin dosing analyses supplied by AstraZeneca appeared to ease review division leaders’ uncertainty about approval, even though they found little support for the biologic explanation for such an interaction.

During an April 20, 2011 mid-cycle review meeting, Temple told AstraZeneca personnel “that no one has been persuaded that a biologically plausible mechanism for an interaction between aspirin dose and ticagrelor has been identified,” according to the meeting minutes.

In a July 8, 2011 review memo, Stockbridge briefly discussed the additional in vitro and animal data submitted by AstraZeneca to support its hypothesis about high-dose aspirin.

Stockbridge said the non-clinical findings were consistent with the absence of further benefit from use of aspirin with complete P2Y12 receptor blockade but provide “little support for the hypothesis that high-dose aspirin leads to ‘harm’ of increased vascular resistance in the presence of P2Y12 blockade.”

Turning to the additional aspirin dose analyses supplied by AstraZeneca, Stockbridge wrote: “I and the statistical review team rate the aspirin hypothesis in PLATO as being not highly dependent on the analysis performed, but also only moderately robust. At least some of the p-values for the interaction with aspirin are smaller than the p-value for the interaction by region.”

Bob Temple’s Five Reasons Why The PLATO Aspirin Dose Explanation Is Credible

The aspirin dose difference entirely accounts for the striking US/OUS difference in efficacy.
Two main components of the study endpoint – CV death and non-fatal MI – are similarly affected by region and aspirin dose, which argues strongly against the theory of chance effects.
The aspirin effect is present for a variety of ways of counting aspirin maintenance dose; uncertainty about aspirin dose should work against the aspirin dose finding.
Diligent examination of many other US/OUS differences showed none that at all explained the outcome difference.
Aspirin dose-related difference in the overall study, and in each region and for both components of the primary endpoint is impressive, even if there are relatively few patients in some subsets.

Office Director Decisional Memo, July 20, 2011

The division director said he did not find the aspirin hypothesis highly persuasive in explaining the U.S./ex-U.S. differences, either by mechanism or analyses as a factor in the study outcomes.

Nevertheless, “unless one can identify a factor intrinsic to U.S. patients or their care that explains the regional disparity in outcomes and then cannot circumvent the effect of that factor, then, despite residual doubts, one is forced to conclude that the evidence favoring true regional differences is far less compelling than is the overall study result.”

The consistency of the findings was especially convincing.

In his July 20, 2011 decisional memo, Temple wrote that the various analyses using a number of maintenance dose estimates “almost all uniformly support the aspirin effect, generally at p<0.01, sometimes="" considerably="">

Acknowledging Stockbridge’s view that evidence supporting the aspirin dose interaction is only moderately robust, Temple says that while robustness depends upon the particular analysis, “it seems greatly strengthened by the very similar findings for both regions (US/OUS) and for both CV mortality and NFMI.”

Temple admits that he, too, was troubled by the lack of a good mechanistic explanation. However, he concluded that ticagrelor should be approved, regardless of whether the U.S./ex-U.S. difference resulted from the play of chance or differences in aspirin dose, and he cited five reasons why evidence supporting the aspirin hypothesis provides a “powerful” basis for approval (see chart).

Ultimately, the agency approved ticagrelor with a boxed warning that states aspirin maintenance doses above 100 mg reduce efficacy and should be avoided. The label’s Clinical Trials section includes a forest plot reflecting the primary endpoint results by maintenance aspirin dose in and outside the U.S.

Limiting daily aspirin dose to 100 mg or less is also one of the key messages of the patient medication guide and a communication plan, both of which are part of a Risk Evaluation and Mitigation Strategy.

Multiple Aspirin Dose Analyses Reassured FDA On Brilinta’s U.S. Efficacy

More from Archive

More from Pink Sheet