Annual adverse event profiles after clopidogrel, prasugrel, and ticagrelor in the Food and Drug Administration Adverse Event Reporting System

Introduction

The randomized trials with oral P₂Y12 platelet inhibitors such as clopidogrel, prasugrel, and ticagrelor are focused predominantly on efficacy outcomes, while adverse events are frequently not under the scope and are commonly underreported beyond bleeding risks.^1–3 Since post-trial safety data are usually not systematic and are hard—if even possible—to be adequately assessed, the adverse event profiles of antiplatelet agents are largely unknown, especially in ‘real-life’ clinical scenarios. Importantly, while excess bleeding is expected after antiplatelet therapy, other types of complications represent a mystery, especially considering that safety antithrombotic trial data are commonly overoptimistic due to missed adverse event counts secondary to massive discontinuations and double-digit incomplete follow-ups.⁴ Thus, evidence from large, uniform, government-mandated data sets are helpful to identify frequencies of complications and to inform prescribing physicians.

The Food and Drug Administration Adverse Event Reporting System (FAERS, formerly AERS) is a database that contains information on adverse event and medication error reports submitted to the Agency. Food and Drug Administration Adverse Event Reporting System is a passive surveillance system that relies on voluntary reporting by health care professionals and consumers and required but mandatory reporting by pharmaceutical manufacturers. Food and Drug Administration Adverse Event Reporting System includes spontaneous reports from US sources, serious and unlabelled spontaneous reports from non-US sources, and serious, unlabelled, and attributable post-marketing clinical trial reports from all sources.⁵ We recently reported the mortality differences among oral P₂Y12 platelet inhibitors⁶ and worldwide distribution of ticagrelor deaths⁷ in FAERS. Here, we expand our analyses by pooling adverse events into clinically relevant categories. All FAERS events were dichotomized into the following types: cardiac, thrombotic, haemorrhagic, all-cause death, and other.

Methods

Food and Drug Administration Adverse Event Reporting System reports originating in 2015 qualified, meaning that the initial report was received by the FDA (FDA Receipt Date) in 2015. There may have been follow-up reports associated with certain cases that occurred after 2015, but the initial report was received in 2015. Any repeated reports were omitted. The pooled FAERS database was searched using the terms ‘clopidogrel’, ‘prasugrel’, ‘ticagrelor’, ‘Plavix’, ‘Iscover’, ‘Zyllt’, ‘Effient’, ‘Effient’, ‘Brilinta’, ‘Brilique’ which were reported with an adverse event. Cardiac-related terms included all Medical Dictionary for Regulatory Activities (MedDRA) preferred terms associated with any of the following High-Level Group Terms (HLTs), and included ‘cardiac arrhythmias’, ‘cardiac disorder signs and symptoms’, ‘cardiac valve disorders’, ‘coronary artery disorders’, ‘endocardial disorders’, ‘heart failures’, ‘myocardial disorders’, and ‘pericardial disorders’. Haemorrhage-related terms included all MedDRA preferred terms associated with any of the following HLTs: ‘bruising, ecchymosis and purpura’, ‘gastrointestinal haemorrhages’, ‘haemorrhages NEC’, ‘nervous system haemorrhagic disorders’, ‘ocular haemorrhagic disorders’, ‘renal haemorrhagic disorders’, and ‘reproductive system haemorrhages’. Thrombosis-related terms included all MedDRA terms associated with any of the following HLTs: ‘aortic embolism and thrombosis’, ‘cerebrovascular embolism and thrombosis’, ‘gastrointestinal embolism and thrombosis’, ‘hepatic and portal embolism and thrombosis’, ‘non-site specific embolism and thrombosis’, ‘peripheral embolism and thrombosis’, ‘pulmonary embolism and thrombosis’, ‘renal embolism and thrombosis’, ‘retinal embolism and thrombosis’, ‘site specific embolism and thrombosis’, ‘vena cava embolism and thrombosis’. All-cause fatalities were counted when the term ‘death’ was co-reported with an antiplatelet agent. Annual adverse events that did not fit into the aforementioned categories were classified as ‘other’ and included mostly ‘dizziness’, ‘diarrhoea’, and ‘nausea’. To avoid bias, data mining and statistics were performed by the independent researchers at FDAble, LLC (Glastonbury, CT, USA www.fdable.com), a for-profit group that specializes in FAERS database analyses.

Results

From 7 732 656 screened FAERS cases, we omitted 7 713 888 and qualified 18 768 events, which included 5398 cardiac, 4594 haemorrhagic, 634 thrombotic, 1689 all-cause death, and 6453 other adverse events. Of the 13 234 cases qualified for clopidogrel, 3839 (29.0%) were cardiac, 2777 (21.0%) were haemorrhagic, 386 (2.9%) were thrombotic, 1156 (8.7%) were all-cause death, and 5076 (38.4%) were other (Table 1). Of the 2927 cases qualified for prasugrel, 505 (17.3%) were cardiac, 1312 (44.8%) were haemorrhagic, 61 (2.1%) were thrombotic, 151 (5.2%) were all-cause death, and 898 (30.6%) were kept as other. Of the 2607 cases qualified for ticagrelor, 1054 (40.4%) were cardiac, 505 (19.4%) were haemorrhagic, 187 (7.2%) were thrombotic, 382 (14.7%) were all-cause death, and 479 (18.3%) were other (Table 1). Comparing ticagrelor to clopidogrel, reported cardiac (relative odds ratio (ROR) = 1.66, proportional reporting ratio (PRR) = 1.39, $χ2$=132.55, P = NS), haemorrhagic (ROR = 0.91, PRR = 0.92, $χ2$=3.35, P = NS), were not significantly different, while all-cause death events (ROR = 1.79, PRR = 1.68, $χ2$=86.30, P = 1.52e⁻²⁰) and thrombotic events (ROR = 2.57, PRR = 2.46, $χ2$=112.00, P = 3.66 × 10⁻²⁶) were significantly worse for ticagrelor compared to clopidogrel (Table 2). Comparing ticagrelor to prasugrel, reported cardiac (ROR = 3.25, PRR = 2.34, $χ2$=364.8, P = 2.43 × 10⁻⁸¹), thrombotic (ROR = 3.63, PRR = 3.44, $χ2$=82.20, P = 1.21 × 10⁻¹⁹), and all-cause death events (ROR = 3.16, PRR = 2.84, $χ2$=141.7, P = 1.13e⁻³²) were significantly worse for ticagrelor, while reported haemorrhagic events (ROR = 0.30, PRR = 0.43, $χ2$=403.9, P = 7.71 × 10⁻⁹⁰) were significantly lower for ticagrelor. Comparing clopidogrel to prasugrel, reported cardiac (ROR = 1.96, PRR = 1.68, $χ2$=167.90, P = NS), thrombotic (ROR = 1.41, PRR = 1.40, $χ2$=5.87, P = NS), and all-cause death (ROR = 1.76, PRR = 1.69, $χ2$=40.76, P = NS) were not significantly different, while reported haemorrhagic events (ROR = 0.33, PRR = 0.47, $χ2$=719.50, P < 0.0001) significantly favoured clopidogrel.

Table 2

Open in new tab

Statistical comparison of clopidogrel, prasugrel, and ticagrelor and co-reported adverse reactions

Cells highlighted in red or green reach the threshold criteria for a significant signal specified by Evans et al.⁹ Cells highlighted in tan have a PRR that approaches, but does not reach, the threshold for a significant signal, and thus represent a trend towards significance.

NS, not significant; PRR, proportional reporting ratio; ROR, relative odds ratio.

Table 2

Open in new tab

Statistical comparison of clopidogrel, prasugrel, and ticagrelor and co-reported adverse reactions

Cells highlighted in red or green reach the threshold criteria for a significant signal specified by Evans et al.⁹ Cells highlighted in tan have a PRR that approaches, but does not reach, the threshold for a significant signal, and thus represent a trend towards significance.

NS, not significant; PRR, proportional reporting ratio; ROR, relative odds ratio.

Discussion

Data from this large, uniform, US government-run international registry suggest a consistent disproportional excess of mortality, cardiac, and thrombotic adverse events co-reported with ticagrelor compared with clopidogrel and especially prasugrel. Indeed, increased bleeding and residual recurrent thrombosis are expected in ACS patients treated with antiplatelet regimens. This trade-off between bleeding and secondary vascular events was heavily used to justify the development of novel antiplatelet agents.⁸ Clearly, the 2015 FAERS adverse event profile for prasugrel suggests superiority to both clopidogrel and ticagrelor in regard to better vascular safety at the expense of excess bleeding. One would expect that the ticagrelor profile would look similar to prasugrel when considering the impressive results of the PLATO trial. However, the distribution of all-cause death, cardiac, and thrombotic events co-reported with ticagrelor are entirely worse than those co-reported with prasugrel. Although these differences are startling, there are several important issues to consider. First, ticagrelor FDA approval process was quite controversial, and four internal Agency reviewers recommended a ‘no approval’ decision. Ticagrelor is the only drug among the observed oral antiplatelet agents exhibiting numerically large event counts, a PRR consistently greater than two, and a large three-digit $χ2$ suggesting definite signal.⁹ There are obvious strengths in our approach with this study. This analysis was conducted within the frame of a government database, requiring mandatory serious event reporting. We hired independent specialists specifically focused in FAERS data mining to apply advanced statistics in an unbiased manner. The sample sizes for the adverse event reports were sufficient to make reasonable comparisons among the antiplatelet agents, and the differences in adverse events were large. Our study also has some clear limitations. We did not analyse the comparative population data, which might be relevant. Further digging into ‘other’ event category less prominent with ticagrelor is also warranted. Another disadvantage of this study is that it does not include reports from the entire lifespans of the drugs but rather focuses on a single snapshot in time, limiting our work to the most recent full year—2015—for which the FAERS data are available. Nonetheless, focusing on the most recent full year allows an even platform for the three antiplatelet agents since prasugrel (2009) and ticagrelor (2011) were FDA approved many years ago after clopidogrel (1997). Regardless, the consistent and alarming magnitude of excess all-cause deaths, cardiac, and thrombotic events after ticagrelor suggests that the index data are realistic, and should not be ignored. However, these preliminary findings need to be interpreted with caution. This is especially true since the possible different clinical features of patients treated with ticagrelor may vary in light of the published guidelines, as compared to prasugrel, and especially after clopidogrel. Indeed, advanced age, lower body weight, impaired renal function, or previous stroke which could at least in part justify the current results. Since FAERS poorly describe these important variables, the index data should be considered as hypothesis-generated. Importantly, the reporting quality and initial sources differ among oral P₂Y12 platelet inhibitors in FAERS. The ticagrelor surveillance in 2015 was inadequate when compared to clopidogrel and prasugrel. Patients filed most AEs for clopidogrel and prasugrel, while physicians originated most ticagrelor complains. These differences justify stricter compliance control for ticagrelor manufacturer, and may be attributed to the confusion of treating physicians with unexpected fatal, cardiac, and thrombotic AEs linked to ticagrelor.¹⁰ Further analyses of large ‘real-life’ registries will be mandatory for better understanding of comparative safety profiles among antiplatelet agents.

Acknowledgements

Special thanks to Paul A. Danese, PhD from FDAble, LLC for statistical skills and detailed FAERS mining.

Funding

Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI); Ministry of Health & Welfare, Republic of Korea (Grant Number HI14C1731); Ministry of Education to Kim. ‘Brain Pool’ program funded by the Korean Ministry of Science and Technology to Serebruany. HeartDrug Research LLC (Wilmington, Del) supported FAERS data mining and statistical work. Special thanks to Paul Danese, PhD, for final event count.

Conflict of interest: HeartDrug™ Research LLC (Wilmington, DE, USA) owned by V.L.S supported FAERS data mining and statistical work. The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Dr Serebruany received research grants from clopidogrel and prasugrel manufacturers, lecture fees from clopidogrel manufacturer, and consultant fees from clopidogrel and ticagrelor manufacturers. He sold a patent on prasugrel to Lilly. Others have nothing to declare.

References