Effectiveness and safety of P2Y₁₂ inhibitors in patients with ST-segment elevation myocardial infarction undergoing percutaneous coronary intervention: a nationwide registry-based study 

Abstract

Aims

To compare the effectiveness and safety of clopidogrel, ticagrelor, and prasugrel in patients with ST-segment elevation myocardial infarction (STEMI) undergoing percutaneous coronary intervention (PCI).

Methods and results

Nationwide, registry-based study of STEMI patients treated with primary PCI (2011–17) and subsequently with aspirin and a P2Y₁₂ inhibitor. The effectiveness outcome was major adverse cardiovascular events (MACE) defined as a composite of recurrent myocardial infarction, repeat revascularization, stroke, or cardiovascular death at 12 months. The safety outcome was bleeding requiring hospitalization at 12 months. Multivariable logistic regression with average treatment effect modeling was used to calculate absolute and relative risks for outcomes standardized to the distributions of demographic characteristics of all included subjects. We included 10 832 patients; 1 697 were treated with clopidogrel, 7 508 with ticagrelor, and 1,627 with prasugrel. Median ages were 66, 63, and 59 years (P < 0.001). Standardized relative risks of MACE were 0.75 for ticagrelor vs. clopidogrel (95% confidence interval [CI], 0.64–0.83), 0.84 for prasugrel vs. clopidogrel (95% CI, 0.73–0.94), and 1.12 for prasugrel vs. ticagrelor (95% CI, 1.00–1.24). Standardized relative risks of bleeding were 0.77 for ticagrelor vs. clopidogrel (95% CI, 0.59–0.93), 0.89 for prasugrel vs. clopidogrel (95% CI, 0.64–1.15), and 1.17 for prasugrel vs. ticagrelor (95% CI, 0.89–1.45).

Conclusion

Ticagrelor and prasugrel were associated with lower risks of MACE after STEMI than clopidogrel, and ticagrelor was associated with a marginal reduction compared with prasugrel. The risk of bleeding was lower with ticagrelor compared with clopidogrel, but did not significantly differ between ticagrelor and prasugrel.

Graphical Abstract

Summary of major findings in ‘effectiveness and safety of P2Y12 inhibitors in patients with ST-segment elevation myocardial infarction undergoing percutaneous coronary intervention: a nationwide register-based study’. Abbreviations: C, clopidogrel; T, ticagrelor; P, prasugrel; MACEs, major adverse cardiovascular events.

Open in new tabDownload slide

Introduction

Dual antiplatelet therapy (DAPT) with a combination of aspirin and a P2Y₁₂ inhibitor reduces the risk of recurrent ischaemic events in patients with acute coronary syndromes (ACS).^1,2 For long-term secondary prevention after ST-segment elevation myocardial infarction (STEMI), both the American College of Cardiology/American Heart Association (ACC/AHA) and the European Society of Cardiology (ESC) generally recommend DAPT for at least 12 months, followed by lifelong aspirin.^3,4 At present, three oral P2Y₁₂ inhibitors are approved for clinical use in Denmark, i.e. clopidogrel, ticagrelor, and prasugrel.

The optimal P2Y₁₂ inhibitor choice, however, remains debated.^5,6 Ticagrelor and prasugrel are more potent platelet inhibitors than clopidogrel and have been associated with a reduction in recurrent ischaemic events in specific subsets of ACS, albeit at the expense of an increased risk of bleeding.^7,8 Moreover, in the recent Intracoronary Stenting and Antithrombotic Regimen: Rapid Early Action for Coronary Treatment (ISAR-REACT) 5 open-label trial among patients with ACS with a planned invasive strategy, a prasugrel-based strategy after percutaneous coronary intervention (PCI) was superior to a ticagrelor-based pre-treatment strategy in preventing death, myocardial infarction, or stroke, without an increased risk of bleeding.⁹ Therefore, the aim of this study was to compare the effectiveness and safety of clopidogrel, ticagrelor, and prasugrel in a Danish nationwide STEMI population treated with PCI.

Methods

Study design and population

A registry-based, nationwide study with 12 months of follow-up. We included patients who (i) were hospitalized for a first-time STEMI from 1 January 2011 to 31 December 2017; (ii) underwent PCI within 7 days of admission; (iii) were alive at discharge; and (iv) claimed a prescription for clopidogrel, prasugrel, or ticagrelor within 30 days of discharge. It is common practice in Denmark to provide the first few days of prescription medication at discharge. In case the patients receive their medication through multi-dose drug dispensing, hospital staff may pack medication for even a few weeks. Therefore, we allowed for a 30-day delay in claiming the prescriptions from the pharmacy. The index date was defined as the day of the first claimed prescription for any of the three P2Y₁₂ inhibitors. Patients with atrial fibrillation or flutter (AF), or concomitant anticoagulant therapy were excluded. Patients who claimed a prescription for more than one type of P2Y₁₂ inhibitor (clopidogrel, ticagrelor or prasugrel) on the same day were also excluded.

Data sources

Citizens of Denmark have free, tax-based access to the healthcare system, including general practitioners, hospital admissions, and outpatient specialists.¹⁰ Furthermore, expenses for prescription medications are partially reimbursed by the public sector.¹⁰ All legal residents of Denmark are allocated a unique Civil Personal Registry (CPR) number upon birth or immigration, enabling direct linkage between registries.¹¹ Social, demographic, and healthcare data are gathered by the Danish authorities and stored in various databases sorted by the CPR number.

This study included data from: (i) The Registry of Causes of Death: holds information on death since 1970;¹² (ii) The Danish National Prescription Registry: contains information on all filled prescriptions since 1995 (sorted by the Anatomical Therapeutic Chemical codes);¹³ and (iii) The Danish National Patient Registry (DNPR): contains information on all hospital admissions, discharge diagnoses, and procedure codes using the International Classification of Diseases system since 1978 and outpatient contacts since 1995.¹⁴ The databases were linked on an individual level by encrypted CPR numbers on a Statistics Denmark server.

The Danish registries have previously been described in detail.¹⁰ The non-self-reported nature of data collection enhances the quality of these databases.¹⁵ The Danish National Patient Registry is characterized by a high degree of completeness and very high positive predictive values for most conditions included in the Charlson’s comorbidity index.¹⁶ For example, the positive predictive value of a diagnosis of first-time myocardial infarction (MI) in the DNPR is 97%.¹⁷

Comorbidities and antithrombotic therapy

Comorbidities were tracked 10 years prior to STEMI admission using DNPR discharge and outpatient diagnoses, with three exceptions: (i) diabetes mellitus was identified either by a DNPR diagnosis registration or through claimed prescriptions for an antidiabetic medication no more than 180 days prior; (ii) hypertension was tracked as either a DNPR diagnosis or redemption of ≥2 successive prescriptions of ≥2 antihypertensive drugs within two consecutive quarters no more than 5 years prior to STEMI admission;^18,19 and (iii) chronic obstructive pulmonary disease (COPD) was tracked as DNPR diagnoses and through claimed prescriptions for either an anticholinergic drug or an adrenergic/anticholinergic combination drug within 180 days prior to STEMI admission.

The following comorbidities were included: hypertension, diabetes mellitus, chronic kidney disease, COPD, known CAD, peripheral artery disease, ischaemic stroke, heart failure, bleeding episodes requiring hospital admission (gastrointestinal, urogenital, and intracerebral bleeding), and any cancer. Bleeding episodes within 180 days prior to STEMI admission were included.

Claimed prescriptions for platelet inhibitors (aspirin and P2Y₁₂ inhibitors), nonsteroidal anti-inflammatory drugs (NSAIDs), and proton pump inhibitors (PPIs) were also tracked within 180 days of STEMI admission.

Outcomes

The primary effectiveness outcome was major adverse cardiovascular events (MACE), defined as a composite of recurrent MI, repeat revascularization (PCI or coronary artery bypass grafting), stroke, or cardiovascular death at 12 months. The primary safety outcome was a composite of gastrointestinal, urogenital, or intracerebral bleeding requiring hospitalization at 12 months.

Secondary outcomes included stroke, a composite of recurrent MI or repeat revascularization, cardiovascular death, all-cause death, a MACE outcome including all-cause death instead of cardiovascular death, individual bleeding outcomes, and net adverse clinical events (NACE) as a composite of all-cause death, stroke, MI, or a bleeding episode.

A blanking period of 30 days after STEMI admission was included for all outcomes except stroke to ensure truly independent outcome events. This also ensured that staged PCI procedures occurring within the first 30 days were not included in the outcome assessment. The concept of staged PCI gained attention during our study period.^20–23 While the optimal timing may still be debated, staged procedures in Denmark generally occur within the first few weeks of an MI.

Statistical analysis

Continuous variables were presented as medians and 25th–75th percentiles (1st–3rd quartiles, Q1–Q3) and compared across groups using the Kruskal-Wallis test. Due to the regulations of Statistics Denmark, absolute numbers between 1 and 3 must be reported as NA. Categorical variables were shown as counts and percentages and compared across groups using Pearson’s chi-squared test.

Absolute and relative risks for outcomes standardized to the distributions of age, sex, known CAD, PAD, hypertension, diabetes mellitus, heart failure, chronic kidney disease, cancer, prior bleeding, and year of MI as well as concomitant pharmacotherapy including aspirin, PPI, NSAID and glucocorticoid across all included study participants were calculated through multivariable logistic regression with average treatment effect modeling (G-formula). This standardization method ensures equal distributions of factors with impact on effectiveness and safety outcomes across the three treatment arms to minimize confounding by indication.²⁴ Accordingly, the G-formula approach is used in observational (non-randomized) studies in an attempt to mimic some of the particular characteristics of a randomized controlled trial. In this study, the marginal differences were computed based on multivariable logistic regression assuming that the entire population had similar age, sex, selected comorbidity, and pharmacotherapy distributions.

SAS 9.4 (SAS institute, Inc., Cary, NC) was used for data management and RStudio, version 4.0.3 (https://www.r-project.org/) for statistical analysis.²⁵

Ethics

We were permitted full access to anonymized data from nationwide registries by Statistics Denmark. In Denmark, registry studies performed on encrypted data do not require ethical approval. An umbrella approval from the data responsible unit in the Capital Region of Denmark had been obtained previously (P-2019-403).

Results

Patients and characteristics

Between 2011 and 2017, 11 732 patients were hospitalized with STEMI, treated with PCI, and claimed a prescription for a P2Y₁₂ inhibitor within 30 days of discharge. Of these, 714 patients on concomitant anticoagulant therapy, 50 patients who claimed a prescription for more than one type of P2Y₁₂ inhibitor in the same day, and 136 with AF were excluded, leaving a final study population of 10 832 individuals. One-thousand six-hundred and ninety-seven patients claimed a prescription for clopidogrel, 7 508 for ticagrelor, and 1 627 for prasugrel.

Median age distributions were 66 (Q1–Q3: 57–77), 63 (Q1–Q3: 54–72), and 59 (Q1–Q3: 51–66) years for patients treated with clopidogrel, ticagrelor, and prasugrel, respectively (P < 0.001 for between-group difference). Women comprised 31.2% of the clopidogrel group, 25.3% of the ticagrelor group, and 20.4% of the prasugrel group (P < 0.001).

Patients treated with clopidogrel were more likely to have been diagnosed with hypertension, peripheral artery disease, COPD, chronic kidney disease, cancer, and stroke (P < 0.05 for all). Patients experiencing a STEMI between 2011 and 12 were most likely to be treated with clopidogrel, while those admitted between 2013 and 14 were most often prescribed prasugrel. Finally, patients treated for a STEMI between 2014 and 17 were mostly discharged on ticagrelor. Population characteristics are summarized in Table 1. Patients treated with clopidogrel were more likely to have been treated with platelet inhibitors within 180 days prior to STEMI admission. Prescription medications claimed within 180 days leading up to the index event are shown in Table 2. Generally, the majority filled a successive, second prescription for a P2Y₁₂ inhibitor: 95.1% of the clopidogrel group, 98.4% of ticagrelor group and 98.8% of the prasugrel group filled a second prescription for any P2Y₁₂ inhibitor. Ninety-four percent of the clopidogrel group filled a second, successive prescription for clopidogrel, 93% of the ticagrelor group filled a second prescription for ticagrelor, and 97% of the prasugrel group filled a second prescription for prasugrel. Finally, patients in the clopidogrel group were more likely to be discharged on a PPI while on DAPT. Concomitant medications at discharge after the index event are displayed in Table 3.

Primary effectiveness outcome

Standardized absolute risks of MACE at 12 months were 13.0% for clopidogrel (95% confidence interval [CI], 11.7–14.3), 9.7% for ticagrelor (95% CI, 9.1–10.3), and 10.9% for prasugrel (95% CI, 9.8–11.9). Corresponding standardized relative risks were 0.75 for ticagrelor vs. clopidogrel (95% CI, 0.66–0.83, P < 0.001), 0.84 for prasugrel vs. clopidogrel (95% CI, 0.73–0.94, P = 0.001), and 1.12 for prasugrel vs. ticagrelor (95% CI, 1.00–1.24, P = 0.045). Standardized absolute risks of MACE are shown in Figure 1.

Primary safety outcome

A total of 843 patients were admitted with a bleeding episode within 12 months of follow-up. Bleeding episodes requiring hospitalization were predominantly gastrointestinal (Table 4). Standardized absolute bleeding risks were 6.1% for clopidogrel (95% CI, 5.0–7.2), 4.7% for ticagrelor (95% CI, 4.2–5.2), and 5.5% for prasugrel (95% CI, 4.3–6.7). Standardized relative risks were 0.77 for ticagrelor vs. clopidogrel (95% CI, 0.60–0.94, P = 0.007), 0.90 for prasugrel vs. clopidogrel (95% CI, 0.65–1.15, P = 0.42), and 1.17 for prasugrel vs. ticagrelor (95% CI, 0.89–1.45, P = 0.23). Standardized absolute risks of bleeding are presented in Figure 1.

Secondary outcomes

A forest plot of the standardized risks of the primary and secondary outcomes is presented in Figure 2. Pertinent findings included a higher risk of repeat revascularization or recurrent MI for prasugrel vs. ticagrelor and lower risks of cardiovascular and all-cause deaths with both prasugrel and ticagrelor vs. clopidogrel. The results for the NACE and modified MACE outcomes mimicked those observed for MACE (Figure 2). No between-group differences were observed for stroke.

Discussion

In this nationwide registry-based study of patients with STEMI who underwent PCI, survived to discharge, and filled a prescription for a P2Y₁₂ inhibitor within 30 days, we found that use of ticagrelor or prasugrel was associated with lower risks of MACE, NACE, cardiovascular death, and all-cause death compared with use of clopidogrel. Ticagrelor was also associated with a marginal reduction in MACE and NACE compared with prasugrel. The risk of bleeding requiring hospitalization was lower in ticagrelor vs. clopidogrel users, but there were no difference in bleeding risk between clopidogrel and prasugrel, and between ticagrelor and prasugrel.

Prior randomized, controlled trials of patients with ACS have suggested similar benefits of ticagrelor and prasugrel compared with clopidogrel.^7,8 Curiously, ticagrelor is associated with a more pronounced inhibition of platelet reactivity than prasugrel,²⁶ but prasugrel after PCI vs. ticagrelor pretreatment significantly reduced the risk of death, myocardial infarction, or stroke compared with ticagrelor in the open-label ISAR-REACT 5, with similar effect sizes in the STEMI and non-ST-segment elevation ACS (NSTE-ACS) populations.^9,27 Although the loading strategies of these two antiplatelet agents differ, the results were driven mainly by a reduction in recurrent myocardial infarction. Therefore, prasugrel may truly have stronger antiplatelet effects, or adherence to this once daily drug with a much longer half-life may have been better in the open-label setting of ISAR-REACT 5, or pretreament, a strategy not supported by RCTs or the ESC guidelines, may be harmful due to bleeding risks.^4,28–30 While we found a marginal benefit of ticagrelor vs. prasugrel in terms of MACE, the older, smaller Comparison of Prasugrel and Ticagrelor in the Treatment of Acute Myocardial Infarction trial found no difference in the efficacy and safety between use of these two P2Y₁₂ inhibitors and was terminated early for futility.³¹

The results for the effectiveness of clopidogrel use were expected given the lower degree of platelet inhibition associated with this drug and the results from the Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition with Prasugrel-Thrombolysis in Myocardial Infarction 38 and the Study of Platelet Inhibition and Patient Outcomes.^7,8 Both studies included all-comer patients with ACS, but the safety profiles of ticagrelor and prasugrel in the STEMI subpopulations appeared more favourable than for the overall study populations.^32,33 While the Ticagrelor or Prasugrel Versus Clopidogrel in Elderly Patients With an Acute Coronary Syndrome and a High Bleeding Risk: Optimization of Antiplatelet Treatment in High-risk Elderly trial showed a lower risk of bleeding without an increased risk of ischaemic events among those aged ≥70 years, only patients with NSTE-ACS were included.³⁴ These patients generally have lower platelet reactivity than patients with STEMI,³⁵ and the results are not directly comparable to our study.

Contrary to our results, a Canadian study investigating cardiovascular outcomes and drug adherence in ACS patients treated with either clopidogrel or ticagrelor found no between-drug difference with respect to MACE while patients treated with ticagrelor had a higher risk of major bleeding events.³⁶ The findings are discordant with ours. However, the Canadian study considered a broad, heterogenous group of ACS patients treated with PCI, while our population was comprised of STEMI patients alone. The higher platelet reactivity of STEMI patients—and as a result, a potential greater benefit from more potent platelet inhibition—might help explain the difference. Furthermore, the investigators included data on stent type, and it is possible that newer generation drug-eluting stents attenuate the relative benefit of stronger platelet inhibition with ticagrelor.

Poor P2Y₁₂ inhibitor adherence in patients with MI undergoing primary PCI has previously been associated with increased mortality, hospitalization, and stent-thrombosis.³⁷ The Canadian study found a greater risk of MACE in non-adherent patients, emphasizing the importance of proper drug adherence. However, in our study, most patients filled a second prescription for a P2Y₁₂ inhibitor, i.e. discontinuation was unlikely to play a major role in our findings.

Major bleeding episodes are among the most serious non-cardiac adverse events in ACS patients and independently associated with increased mortality.⁶ Treatment with a PPI in select patients reduces the risk of gastrointestinal bleedings during DAPT.^38,39 In the present study, concomitant treatment with a PPI most often occurred in the clopidogrel group. Given the overlap between risk factors related to thrombosis and bleeding,^40,41 the greater association with bleeding episodes among patients on clopidogrel compared with patients on ticagrelor might be due to an overall greater vulnerability,^42,43 e.g. higher prevalences of almost all comorbidities and use of NSAID and glucocorticoids. It seems reasonable to assume that patients with a higher baseline bleeding risk—and possibly higher ischaemic risk as well—would have preferentially been prescribed clopidogrel and thus skewing the results in favour of ticagrelor and prasugrel.

The 2020 ESC NSTE-ACS guidelines recommend consideration of prasugrel over ticagrelor for NSTE-ACS patients who proceed to PCI (class IIa recommendation).⁴⁴ While the major cardiovascular societies have not published updated STEMI guidelines, the most recent Danish national treatment guidelines recommend prasugrel as the first choice P2Y₁₂ inhibitor for all patients with ACS undergoing PCI.⁴⁵ A Danish single-centre study found similar effectiveness of ticagrelor and prasugrel, albeit with a more traditional statistical method that to a lesser extent takes into account confounding by indication.⁴⁶ Adding to that our results, the clear preference of one agent over the other is questionable. Finally, it should be noted that while ISAR-REACT 5 tested traditional ticagrelor- and prasugrel-based strategies, i.e. ticagrelor preloading irrespective of ACS type and prasugrel preloading only in patients with STEMI, the current consensus in Denmark is to defer loading of all P2Y₁₂ inhibitors until angiography has been performed. It is possible that such deferred loading decreases the risk of bleeding associated with ticagrelor, but this approach may only be feasible in settings with short first medical contact to balloon times.^47,48

Limitations

Due to the retrospective nature of this study, selection of the antiplatelet agent may have been subject to confounding by indication, particularly given the clear between-group differences in baseline characteristics. Average treatment effect models standardizing the risks were applied to counteract this, but cannot account for unmeasured confounders. Second, we did not possess information on negative predictive values of the registries, i.e. patients having a condition but not registered. Comorbidities were mostly obtained through hospital records, potentially missing diagnoses made by general practitioners. Some patients categorized with hypertension may also have been misclassified as patients with heart failure also receive a two-or-more-drug antihypertensive regimen. Nevertheless, this assessment of hypertension based on medications has previously been validated.¹⁵ In addition to being a prescription medication, low-dose aspirin can be purchased as an over-the-counter medication in Denmark which likely accounts for the low uptake that we reported. Patients with a STEMI who were discharged with a diagnosis of ‘myocardial infarction, not otherwise specified’ would not have been included in the present study, leading to an underestimation of the incidence of STEMI, though this is unlikely to have affected between-group differences. The national guidelines for STEMI patients changed multiple times during the study period, reflecting the temporal variation in the preferred P2Y₁₂ inhibitor, although with some delay.⁴⁵ Per these guidelines, clopidogrel was the P2Y₁₂ inhibitor of choice until 2012, ticagrelor from 2012 until 2020, and prasugrel from 2020. Despite our attempt to counteract the effect of time variation by standardizing to the distributions of year of MI, there may have been regional preferences in the use of P2Y₁₂ inhibitors which may in itself have affected outcomes. Taken together, it will be important to examine outcomes related to the different P2Y₁₂ inhibitors after 2020, keeping in mind that the pandemic itself may have affected outcomes.^49,50 Our primary composite effectiveness outcome included repeat revascularization. The current convention in ACS patients with multivessel disease is to treat the culprit lesion and postpone remaining treatable lesions. Although we did include a blanking period of 30 days after STEMI admission, it is possible that some revascularization events were planned staged procedures related to the index hospitalization. Finally, information on body weight, symptom onset, true drug adherence, and, more importantly, stent type and first medical contact to balloon time was not available.

Conclusion

In this large observational study of STEMI patients treated with PCI during 2011–2017, ticagrelor and prasugrel were both associated with a lower risk of MACE than clopidogrel. The risk of bleeding was lower with ticagrelor compared with clopidogrel, but did not significantly differ between ticagrelor and prasugrel.

References

Author notes