Patient characteristics and safety outcomes in new users of ticagrelor and clopidogrel—An observational cohort study in Sweden

We aimed to describe characteristics of new users of ticagrelor or clopidogrel following a recent coronary event, and to compare incidences of selected safety outcomes.


Key Points
• In an observational cohort study, we identified new users of ticagrelor or clopidogrel following a recent coronary event, using data from national Swedish registers to analyze and describe comorbidities, concomitant medications, and safety outcomes.
• Clopidogrel users were older and had a higher prevalence of concomitant medications than ticagrelor users.
• Ticagrelor users had an elevated risk of dyspnea and respiratory bleeding (mostly epistaxis) compared with clopidogrel users.
• Adjusted analyses also suggested a higher risk of gout and acute renal failure among ticagrelor users than clopidogrel users.

| INTRODUCTION
Antagonists of the platelet adenosine diphosphate P2Y 12  Antiplatelet agents with the potential to overcome this clinical challenge have therefore been developed. Ticagrelor, a reversible and direct-acting antagonist of the P2Y 12 receptor, has demonstrated improved platelet inhibition over that achieved with clopidogrel, 7,8 and was shown to be superior to clopidogrel for the prevention of vascular events in a large randomized trial of patients with ACS, the Study of PLATelet Inhibition and Patient Outcomes (PLATO). 9 Consistency between randomized trial results and real-world data was shown in a study based on a national Swedish quality register. 10,11 As the latest entrant in its class, ticagrelor might have initially been prescribed to patients deemed less likely to achieve adequate inhibition of platelet aggregation with other antiplatelet drugs, or to individuals with a particular cardiovascular comorbidity pattern. If this were the case, differences in patient characteristics due to channeling 12 would hamper direct comparisons of ticagrelor with other antiplatelet agents in early studies of safety outcomes. Improved knowledge of the ticagrelor patient population in terms of comorbidities and concomitant medication use would thus be helpful when interpreting data from sources such as adverse event reporting systems, and would provide information on ticagrelor treatment patterns in clinical practice.
A post-authorization safety study (PASS) 13 of ticagrelor was initiated in 2011 as part of the European Union Risk Management Plan of an approved medicinal product. The goal of the PASS was to assess patient characteristics, drug utilization patterns, and incidence of selected outcomes in new users of P2Y 12 receptor antagonists.
Predefined outcomes were selected based on the results of PLATO and other information available at the time of the PASS. Adverse event rates observed with ticagrelor in PLATO raised no major safety concerns. 9 As would be expected for an antiplatelet drug, bleeding was the primary safety issue. Dyspnea was also commonly reported with ticagrelor, leading to discontinuation of the drug by 1% of patients. Serum creatinine and uric acid levels increased slightly more during treatment with ticagrelor than with clopidogrel; however, this was not associated with a rise in clinically meaningful adverse renal outcomes. Ticagrelor was shown to increase the frequency of Holterdetected ventricular pauses, but with no increase in clinically relevant events. 9 A few cases of severe hepatotoxicity potentially related to clopidogrel treatment had been reported in the literature, 14 leading to an evaluation of hepatotoxicity potential with ticagrelor as part of the PASS. At the time, there was also limited knowledge of treating patients with renal impairment with ticagrelor.
The present study was an extension of the original PASS. Its purpose was to describe the characteristics of patients in whom ticagrelor or clopidogrel treatment was initiated for the first time following a recent coronary event, to assess comorbidities and concomitant medication use and to compare incidences of selected safety outcomes. It was limited to patients with a likely indication of a coronary event and had a longer study period than the original PASS. It also included adjusted analyses.

| Study design and cohorts
A cohort study was performed using data for patients taking P2Y 12 receptor antagonists identified from the PDR.  Table S1). The use of patient register data for diagnoses has shown acceptable validity. [17][18][19][20] All individuals were followed up from the index date until one of the following events occurred: a selected outcome, the patient's 85th birthday, emigration, death, or 31 December 2014. For each outcome of interest a separate followup was performed. Patients with a history of the specific study outcome before the index date in each respective follow-up were excluded. Two cohorts were identified: ticagrelor initiators and clopidogrel initiators; patients could contribute data to both cohorts if separate treatment episodes occurred. The first continuous use period was defined as starting at the index date and ending at the first gap in drug supply (i.e., the start of the first recent use period) or the end of follow-up, whichever came first. Note that the first continuous use period included a grace period of 30 days as defined above, and was right censored at the end of the study.

| Covariates
Comorbidities related to the indication of clopidogrel and ticagrelor were included: coronary events (MI, ACS) and coronary interventions, atrial fibrillation and flutter, and stroke (Table S1). Further additions were comorbidities corresponding to the 13 selected safety outcomes (see below) and additional disease groups specified by Charlson et al., 21,22 with modified definitions for chronic obstructive pulmonary disease, moderate or severe liver disease, and acquired immune deficiency syndrome/human immunodeficiency virus infection (Table S2).  Table 1).
In all covariate definitions, a time window of 1 year before the index date was used.

| Outcomes
The 13 selected safety outcomes were intracranial bleeding, gastrointestinal bleeding, respiratory bleeding, other bleeding, pacemaker insertion, bradyarrhythmias, cardiac arrest, heart failure, acute renal failure, acute liver injury, dyspnea, syncope, and gout. Diagnoses were identified from the NPR using ICD-10 codes (Table S3) from main diagnoses from inpatient and outpatient care.

| Statistical analysis
Numbers and proportions were used to describe categorical variables, and means with standard deviations and/or medians with interquartile ranges were used to describe continuous variables. Ticagrelor and clopidogrel cohorts were described in terms of age and sex distribution, and prevalence of concomitant treatments and recorded comorbidities.
Crude incidences (with 95% confidence intervals [CIs]) were estimated as the ratio of the number of cases with the outcome of interest divided by the number of person-years among users of the study drug grouped by exposure category (current, recent, or past use). An analysis comparing ticagrelor and clopidogrel was also carried out for the selected safety outcomes using Cox regression, adjusting for sex, age, income, education, use of concomitant medication (ASA, cardiac therapy, diuretics, beta blocking agents, agents acting on the reninangiotensin system, and lipid-modifying agents), heart failure, PCI, ACS, stroke, atrial fibrillation, MI, and cerebrovascular disease. This analysis was not specified in the original PASS protocol.

| RESULTS
In total, 45 987 unique naïve users of ticagrelor or clopidogrel were included in this study, with 30 492 contributing to the clopidogrel cohort and 15 607 contributing to the ticagrelor cohort (i.e., 112 patients contributed to both cohorts) ( Figure 1).

| Patient characteristics
The majority of the naïve ticagrelor or clopidogrel users were men (71%), and the mean age was 68 years ( dose ASA in the year before the index date. Other common concomitant medications were beta-blocking agents, agents acting on the renin-angiotensin system, and lipid-modifying agents ( Table 1). The clopidogrel cohort had a numerically higher proportion of use of these concomitant medications than the ticagrelor cohort (51%-62% vs. 39%-44%, respectively).  Table 2 and  Abbreviations: CI, confidence interval; GI, gastrointestinal; HR, hazard ratio; NA, not applicable. a The model was adjusted for sex, age, income, education, acetylsalicylic acid use, cardiac therapy, diuretic use, beta blocking agent use, use of agents acting on the renin-angiotensin system, use of lipidmodifying agents, heart failure, percutaneous coronary intervention, acute coronary syndrome, stroke, atrial fibrillation, myocardial infarction, and cerebrovascular disease. For clopidogrel, incidences of all outcomes except pacemaker insertion, acute liver injury, and gout showed a declining pattern when moving from current to recent to past use (Figure 3). For ticagrelor, all outcomes except other bleeding, pacemaker insertion, acute liver injury, and gout showed a declining incidence pattern moving from current to recent to past use.

| Survival analysis of safety outcomes
Comparison of the current use of ticagrelor and clopidogrel showed a statistically significantly increased risk of respiratory bleeding, dyspnea, acute renal failure, and gout with ticagrelor ( Table 2). When considering recent use, there were significantly increased risks of gastrointestinal and respiratory bleeding with ticagrelor compared with clopidogrel. No significant differences in risks between the two agents were observed with past use.

| DISCUSSION
This observational cohort study used data from national Swedish registers to assess comorbidities, concomitant medication use, and inci- and who are not at heightened bleeding risk. [25][26][27] As could be expected in a population with cardiovascular disease, the age distribution of ticagrelor and clopidogrel users was skewed towards older age, with a higher proportion of men than women. Not all patients had a recorded MI and/or ACS diagnosis, and some were thus included because of the coronary intervention procedure (PCI or CABG) only. The absence of a diagnosis code of coronary disease may be explained by the use of data from the NPR, which does not cover primary care, although the ACS definition used in this study includes angina pectoris diagnoses in inpatient and outpatient specialized care.
At the time the study was conducted clopidogrel had a broader indication than ticagrelor, including stroke, peripheral artery disease, and prevention of thromboembolic events in atrial fibrillation. During the study both drugs were recommended for use in combination with low-dose ASA, but this differed by indication for clopidogrel. 28 Concomitant medications used in this study were dominated by antithrombotic agents, low-dose ASA, beta blocking agents, agents acting on the renin-angiotensin system, and lipid-modifying agents for both study drugs. For clopidogrel, these concomitant medications were purchased by about 50%-60% of users within 1 year before the index date; for ticagrelor, these proportions were around 40%-45%.
When considering safety outcomes, we found a declining pattern in the crude incidence of heart failure over current, recent, and past use for both clopidogrel and ticagrelor, and incidences were similar for the two drugs. The crude incidence of dyspnea was higher during current use of ticagrelor than with current use of clopidogrel, and adjusted analyses showed an increased risk of dyspnea with ticagrelor compared with clopidogrel for current use. Dyspnea is a well-known adverse event that was commonly reported in patients receiving ticagrelor in PLATO. 9 Ticagrelor-related dyspnea in PLATO was mostly transient, mild, or moderate in intensity. 29 No differences in pulmonary function parameters were observed between the ticagrelor and clopidogrel groups in the PLATO pulmonary function sub-study. 30 Our study found no clear difference in the crude incidence of intracranial bleeding, gastrointestinal bleeding, and other bleeding between clopidogrel and ticagrelor groups. The crude incidence of respiratory bleeding was higher with ticagrelor than with clopidogrel. ticagrelor. 31 Moreover, ticagrelor is metabolized through the kidneys, which may influence renal function and the uptake/secretion of uric acid leading to gout. 32 For the treating physician, the observed safety signals may be considered before deciding between ticagrelor and clopidogrel in patients with elevated risk for these outcomes. Randomized trials imply no need for update of clinical guidelines, for example, in PLATO, the slight increase in serum levels of creatinine and uric acid for ticagrelor as compared with clopidogrel resolved after the end of treatment. 9 Reports of gout did not differ between treatment groups in PLATO. 9 In two other large randomized trials, gout occurred more commonly with ticagrelor than with placebo, but there was no noticeable difference between ticagrelor and placebo in rates of renal adverse events or renal impairment. 33,34 Future observational research on antiplatelet agents should include longer followup and more patients and focus on a wide range of adverse events.
The research should be planned to be able to confirm or refute the findings of the current study.

| Strengths and limitations
The major strength of this study is the use of Swedish national regis-  Table S5). Furthermore, the exact date of treatment discontinuation was not known, and the duration of therapy was estimated based on the amount of drug dispensed due to nonavailable information on prescribed dose. The defined daily dose was in this study chosen for calculation of treatment duration since on average it is correct, with few patients deviating from the once/ twice daily recommendation. Also, adherence was unknown, but we believe that it does not differ between the two study drugs. Moreover, treatment episodes continuing beyond the end of follow-up were cut short. Finally, residual confounding may occur in the adjusted analyses presented.

| CONCLUSIONS
Patients taking clopidogrel were older and had a higher prevalence of concomitant medication use than those taking ticagrelor. Our study showed an elevated risk of dyspnea for current ticagrelor users compared with clopidogrel users. We also found an elevated risk of respiratory bleeding (mainly epistaxis) with current and recent use of ticagrelor. Furthermore, the results of adjusted analyses suggest a higher risk of gout and acute renal failure among current ticagrelor users compared with clopidogrel users.

ETHICS STATEMENT
The study was approved by the Ethical Review Board at Karolinska Institutet, Stockholm (reference number 2013/1:11).

AUTHOR CONTRIBUTIONS
Marie Linder made substantial contributions to the conception and design of the study, the acquisition of data, and the interpretation of data, as well as carrying out the statistical analysis and drafting the paper. Morten Andersen contributed to the design of the study, the interpretation of the results, and writing and revision of the paper.