FDA panel review of ticagrelor on July 28th, 2010

Editor-In-Chief: C. Michael Gibson, M.S., M.D. [1]

FDA Panel Review of Ticagrelor on July 28th, 2010

While the administration of Ticagrelor was associated with a reduction in adverse events in the PLATO trial as a whole , there was a lack of benefit, and in fact an excess risk of adverse events in patients treated with Ticagrelor in North America. The confidence interval for the hazard in the United States did not overlap with the confidence interval for benefit in the ex-US population, and the p-value for the interaction by region was 0.045. The regional analysis was not pre-specified. There were few patients enrolled from Canada, and the hazard was apparent when the analysis was confined to those patients from the United States of America (H.R for the primary endpoint was 1.27 for US, and 1.17 for Canada). The hazard ratio observed in the United States for MI was 1.38 (95% CI 0.95-2.01), for CV death was 1.26 (95% CI 0.69-2.31) and for [[stroke] was 1.75. There were three countries with a hazard ration > 1.27 and these were the United States, Australia and Taiwan. As part of a sensitivity analysis, when the United States (an outlier in terms of hazard) was excluded as were Turkey, Hungary, and Poland (outliers in terms of benefit), the overall results of the study were similar.

Much of the panel discussion centered around the basis for the hazard associated with Ticagrelor administration observed in the United States. A key finding was that the aspirin dose in United States patients was higher than that in the rest of the world. The PLATO study recommended, but did not mandate that low-dose aspirin be used. As a result, 92% of patients were treated with low dose aspirin. It should be noted that the current ACC / AHA guidelines recommend high dose aspirin in those patients undergoing intracoronary stent placement, and this may explain in part the reason why some patients, and particularly those in the United States, were treated with high dose (325 mg dialy) aspirin. The median aspirin dose in the United States was 325 mg and the average dose was 217 mg; ex-US the median aspirin dose was 100 mg, and the average dose was 99 mg.

The interaction term regarding the impact of aspirin dose on outcomes was highly statistically significant (p=0.00006, Chi2 of 16). Even if multiple exploratory comparisons were made, this p-value would remain statistically significant. The interaction with aspirin dose was observed for the dose chosen for chronic administration and not for the loading dose chosen for acute administration. Ex-US data was used to model the relationship between the aspirin dose and clinical outcomes in the PLATO trial. The projected event rates associated with different aspirin doses based upon ex-US data successfully predicted the event rates observed in the United States. It should be noted though, that there were very few patients outside the US who received high dose aspirin that were used to create this model. Inverting the clinical outcomes in just 20 patients (i.e. stating that a patient with an event did not have an event) significantly altered the model, so much so that no hazard was observed at high doses in this sensitivity analysis. While the interaction with clinical outcomes was explained almost exclusively by aspirin dose, and not by the country the patient came from, the presenters did vary in their interpretation as to whether other factors may have explained the effect of aspirin dose: aspirin was associated with PCI performance, stent placement and glycoprotein IIb IIIa inhibitor administration. Patients in the United States were heavier, had more diabetes, had a greater incidence of prior MI, underwent PCI more often, more often had a stent placed, more often had a drug eluting stent placed rather than bare metal, underwent CABG more often, had a greater incidence of NSTEMI, less unstable angina and they were less compliant with study drug than the rest of the world. It was also speculated by the FDA that higher aspirin dose may be a surrogate for higher patient risk in general in ways that could not be easily quantified.

While aspirin dose was associated with efficacy, it was not associated with differences in the rate of bleeding. The rate of serious bleeding was similar between the two arms of the PLATO study. However, hemorrhagic stroke was doubled in Ticagrelor patients. Among patients with hemorrhagic stroke, 1/2 died while being treated with Ticagrelor while 1/6 died while treated with clopidogrel.

It should be noted that there are no randomized trials that evaluate the optimal dose of aspirin as part of chronic pharmacotherapy. Meta-analyses suggest that low-dose aspirin may be as if not more effective than full dose aspirin (325 mg daily). The CURRENT / OASIS 7 trial evaluated the relative efficacy of high and low dose aspirin over the course of the first 30 days in patients with acute coronary syndromes. High dose aspirin was not of benefit over the course of 30 days in this randomized study, but this study is not informative with respect to safety and efficacy of high vs low dose aspirin after 30 days. Panel members as well as FDA officials also commented on the fact that aspirin monotherapy has not been compared with thienopyridine monotherapy in the setting of ACS. There has been a supposition that a thienopyridine can be substituted for aspirin in the patient with aspirin intolerance, but this has not been studied.

Several theories were offered regarding the underlying pathobiology of adverse outcomes associated with higher doses of aspirin in the context of ticagrelor therapy. At low doses, aspirin inhibits thromboxane A2 and thereby inhibits platelet aggregation. At high doses, aspirin begins to inhibit prostaglandins which are vasodilators, and may therefore cause vasoconstriction. The balance between platelet inhibition and vasoconstriction may be shifted to different degrees in the presence of a thienopyridine such as clopidogrel or a thienopyridine-like agent such as Ticagrelor.

With respect to the hazard observed in patients from the United States, the FDA concluded that the mechanism seemed unlikely to be due to a difference in aspirin dose alone, that the data regarding the model of aspirin dose was not robust, that the role of chance although low could not be excluded, that there was no clear explanation, that the reversal in direction is quite odd and hard to explain, and that the region-specific analysis was not pre-specified.

There was no benefit observed associated with Ticagrelor administration among patients who were troponin negative and among those patients classified as having unstable angina at discharge. Among the 10,000 patients who were enrolled in a substudy of pharmacogenomics, 28% of patients were identified as carriers of the Cyp 2c19 allele (associated with impaired generation of the active metabolite of clopidogrel), and the benefits were similar in this carrier population as they were in non-carriers. There was no difference in angiographic outcomes between the Ticagrelor vs clopidogrel treated patients. 48% of ticagrelor patients had been previously treated with clopidogrel, and the results were more favorable in those patients pre-treated with clopidogrel. The HR was 0.82 if no clopidogrel had been administered. The FDA pointed out that there was little to no benefit at day 1, and that the benefit began to emerge by day 2-3.

The FDA briefing document indicated that Ticagrelor reduced atorvastatin metabolism (degradation) by 36%. 90% of the patients in PLATO were on a lipid lowering agent, and there was no difference between the two treatment arms and the achieved LDL. It therefore does not appear that an off target LDL-lowering effect of Ticagrelor explains the late benefits that were observed, although the impact on other lipid elements has not been closely examined.

The rate of drug discontinuation was 2% higher in the ticagrelor arm, and half of the cases of discontinuation were due to dyspnea. Most case of dyspnea lasted only 20 days, and 2/3rds resolved spontaneously.