Bleeding risk and P2Y12 inhibitors in all-comer patients with ST-segment elevation myocardial infarction treated with percutaneous coronary intervention: a single-centre cohort study

Abstract

Aims

To characterize and follow patients with ST-segment elevation myocardial infarction (STEMI) at high bleeding risk (HBR) according to the predicting bleeding complications in patients undergoing stent implantation and subsequent dual antiplatelet therapy (PRECISE-DAPT) score, and to examine the use of P2Y12 inhibitors and the subsequent risk of major adverse cardiovascular events (MACE) and bleeding.

Methods and results

This single-centre cohort study included 6179 consecutive STEMI patients who underwent percutaneous coronary intervention (PCI) at Copenhagen University Hospital, Rigshospitalet, between 2009 and 2016. Individual linkage to nationwide registries was conducted to obtain information on diagnoses, claimed drugs, and vital status. Of the 5532 (89.5%) patients with available PRECISE-DAPT scores, 33.0% were at HBR and more often elderly and female with more comorbidities than non-HBR patients. One-year cumulative incidence rates per 100 person-years were 8.7 and 2.1 for major bleeding and 36.8 and 8.3 for MACE in HBR and non-HBR patients, respectively. Among the 4749 (85.8%) patients who survived and collected a P2Y12 inhibitor ≤7 days from discharge, 68.2% of HBR patients were treated with ticagrelor or prasugrel and 31.8% with clopidogrel, while 18.2% non-HBR patients were treated with clopidogrel. Adherence was high for all (>75% days coverage). The risk of MACE was lower in ticagrelor- and prasugrel-treated patients than in clopidogrel-treated patients without differences in major bleeding.

Conclusion

One-third of PCI-treated all-comer patients with STEMI were at HBR according to the PRECISE-DAPT score and were more often treated with potent P2Y12 inhibitors instead of clopidogrel. Thus, ischaemic risk may be weighted over bleeding risk in STEMI patients at HBR.

Graphical Abstract

PCI-treated STEMI patients at high bleeding risk and the use of P2Y12 inhibitors according to bleeding risk. PCI, percutaneous coronary intervention; PRECISE-DAPT, predicting bleeding complications in patients undergoing stent implantation and subsequent dual antiplatelet therapy; STEMI, ST-segment elevation myocardial infarction.

Open in new tabDownload slide

Introduction

Bleeding complications of dual antiplatelet therapy (DAPT) have an important prognostic impact in patients with acute coronary syndromes and are associated with short- and long-term mortality.^1–4 Several risk scores facilitate the identification of patients at high bleeding risk (HBR), including the predicting bleeding complications in patients undergoing stent implantation and subsequent DAPT (PRECISE-DAPT) score.⁵ Importantly, stratification of bleeding risk to guide DAPT is not performed routinely in clinical practice, and the focus is often to reduce the ischaemic risk. Hence, most patients with acute coronary syndromes are treated with the more potent P2Y12 inhibitors, prasugrel and ticagrelor, instead of clopidogrel for the standard duration of 12 months following percutaneous coronary intervention (PCI), irrespective of the bleeding risk.

Optimal DAPT after a ST-segment elevation myocardial infarction (STEMI) in HBR patients represents a common but unmet challenge in clinical practice. Contemporary data are lacking on the magnitude of the problem, as the numbers of unselected STEMI patients at HBR and the related use of P2Y12 inhibitors are vastly unknown.^6,7 The first aim of this study was to describe the proportion of HBR patients according to the PRECISE-DAPT score in an all-comer STEMI population treated with PCI and the incidence rates of major bleeding and major adverse cardiovascular events (MACE) in HBR and HBR patients. Secondly, we aimed to describe the use (initiation, switch, and adherence) of different P2Y12 inhibitors in patients who survived until 7 days after discharge according to their a priori bleeding risk and determine the risk of major bleeding and MACE.

Methods

Study design

This was a single-centre cohort study utilizing data from the Eastern Danish Heart Registry, which comprises detailed clinical, angiographic, and procedural information on consecutive patients admitted to the Copenhagen University Hospital, Rigshospitalet, Denmark, between 2009 and 2016. In Denmark, all citizens have a unique civil registration number, enabling individual linkage to data from an electronic laboratory database and three Danish nationwide registries. The Danish National Prescription Registry was used to obtain information on claimed prescriptions from all pharmacies according to the Anatomical Therapeutic Chemical System codes, including strengths, quantities, and dates of dispensing (Supplementary material online, Table S1). The Danish National Patient Registry was utilized to obtain information on every hospital admission and discharge, including diagnosis codes according to the International Classification of Disease, Tenth Revision codes (Supplementary material online, Table S2). From the Civil Registration System, information on the day of death was retrieved. Comorbidities were defined based on diagnosis codes from inpatient and outpatient contacts before PCI. To prevent underestimation, hypertension, hypercholesterolaemia, and diabetes were also defined based on relevant prescriptions collected 1 year before PCI, as these conditions are often treated in the primary healthcare system and not recorded in the National Patient Registry.⁸ Lastly, redeemed prescriptions 1 year before PCI and ≤7 days from discharge were assessed.

Study population

Patients who underwent PCI for STEMI were eligible. If the patients had multiple admissions during the study period, only their first admission was included. For the first objective (characterization of patients according to bleeding risk), the PRECISE-DAPT score was calculated using the web-based calculator, including age, prior bleeding event, haemoglobin, leucocytes, and creatinine clearance.⁹ A prior bleeding event was defined as a bleeding diagnosis from inpatient and outpatient contacts (Supplementary material online, Table S2). Levels of haemoglobin, leucocytes, and creatinine clearance were considered preferably before PCI (minimum 30 days before hospitalization and maximum on the day of discharge). Creatinine clearance was determined with the Cockcroft–Gault equation.¹⁰ A PRECISE-DAPT score ≥25 was considered as HBR and a score <25 as non-HBR. To account for the high number of missing blood parameters (mainly leucocytes), a categorized PRECISE-DAPT classification was introduced. The maximum and minimum values of 1 or more of the missing parameters in the score were calculated and used to determine the highest and lowest possible PRECISE-DAPT scores. If the highest and lowest possible scores were both ≥25 or both <25, then patients were categorized as HBR and non-HBR, respectively. Patients with the highest possible PRECISE-DAPT score of ≥25 and the lowest possible score of <25 remained missing.

For the second objective (DAPT according to bleeding risk), patients alive and collecting a P2Y12 inhibitor within 7 days from discharge were considered and categorized according to the first claimed P2Y12 inhibitor.

Outcomes

For the first objective, the follow-up period started on the date of PCI. For the second objective, a blanking period of 7 days from discharge was needed to allow the patient time to claim a P2Y12 inhibitor and avoid immortal time bias. Follow-up lasted until the outcome of interest, death, or emigration occurred within the first year after PCI. The outcomes were defined based on diagnosis codes from all inpatient contacts and included major bleeding (composite of bleeding leading to hospitalization) and MACE [composite of all-cause mortality, recurrent myocardial infarction (MI), and stroke]. Recurrent MI was considered as a hospital admission with MI minimum of 28 days after the index MI.^11,12 The individual components of the outcomes were also evaluated.

Statistics

Differences between continuous variables [median with interquartile range (IQR) or mean with standard deviations (SD)] were assessed with Wilcoxon rank-sum or Student’s t-test, respectively, and for categorical variables (frequency) with Chi-squared test for the first objective (HBR vs. non-HBR) and second objective (prasugrel vs. clopidogrel and ticagrelor vs. clopidogrel according to bleeding risk). Any switch from the first claimed P2Y12 inhibitor to another within the first year after PCI was reported and treatment adherence with the P2Y12 inhibitor was calculated as mean proportion of days covered, as previously described.¹³ Incidence rates (IRs) per 100 person-years of the outcomes were calculated. Cumulative incidence of major bleeding and MACE were plotted for HBR and non-HBR patients. Multivariate Cox proportional-hazard models were performed to assess the association between the outcomes and use of potent P2Y12 inhibitors (ticagrelor and prasugrel) vs. clopidogrel in HBR and non-HBR patients. Clinically relevant covariates considered predictive of the outcomes were included in these models. A sub-analysis excluding HBR patients with concomitant use of oral anticoagulants was performed. All models were tested for interactions and proportion of hazards and found valid. Statistical analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC, USA) and R.¹⁴

Ethical approval

The Danish Data Protection Agency has approved the study (2007-58-0015/GEH-2014-014 and I-suite number: 02732). Danish register-based studies do not require ethical approval. The civil registration number was encrypted to avoid identification.

Results

We identified 6179 STEMI patients who underwent PCI at Copenhagen University Hospital between 2009 and 2016 (Figure 1). The numeric PRECISE-DAPT score could be calculated in 4443 patients (71.9%), of whom 1545 (34.8%) were at HBR due to missing on blood parameters (data not shown). When applying the categorized PRECISE-DAPT score, 5532 (89.5%) patients were considered (Figure 1).

Characterization of patients according to bleeding risk

Of the 5532 patients, 1826 (33.0%) patients were at HBR according to the PRECISE-DAPT score (Figure 1). As expected, the variables of the PRECISE-DAPT score (age, prior bleeding, haemoglobin, leucocytes, and creatinine clearance) differed significantly between patients with and without HBR. HBR patients were more often female (38.2% vs. 18.2%), had more comorbidities (e.g. diabetes, heart failure, cardiac arrhythmia), and higher use of drugs related to bleeding risk upon admission (e.g. oral anticoagulants, oral steroids, and proton-pump inhibitors) than non-HBR patients (Table 1). They also had a longer delay in time from symptoms to PCI and more often ventricular fibrillation or pulseless ventricular tachycardia before PCI (14.1% vs. 10.4%, respectively) than non-HBR patients (Table 1). No differences were found in terms of culprit lesion, however, HBR patients were more often treated with balloon angioplasty without subsequent stent implantation compared with non-HBR patients (22.6% vs. 18.0%, respectively) (Table 1). One-year IR and cumulative incidence curves of the outcomes are provided in Table 2 and Supplementary material online, Figures S1 and S2, respectively.

DAPT according to bleeding risk

Out of 5532 STEMI patients with categorized PRECISE-DAPT scores, 4749 (85.8%) patients were alive and collected a P2Y12 inhibitor within 7 days from discharge (Figure 1). A total of 1394 (29.4%) patients were at HBR, of whom 68.2% were treated with ticagrelor or prasugrel and 31.8% with clopidogrel; for non-HBR patients, the corresponding numbers were 81.8% and 18.2%, respectively (Figure 2). Temporal trends in the use of P2Y12 inhibitors for HBR and non-HBR patients are illustrated in Supplementary material online, Figures S3 and S4. Characteristics of P2Y12 inhibitor users at HBR are presented in Table 3. Clopidogrel-treated patients were older [78 years (IQR 70–84)] than ticagrelor [76 years (IQR 68–81)] and prasugrel users [68 years (IQR 61–75)] and more often female (41.1% vs. 32.0%) compared with prasugrel-treated patients. Clopidogrel-treated patients had more comorbidities and were more often treated with anticoagulants. Implantation of a drug-eluting stent was more frequent in patients receiving prasugrel (81.0%) or ticagrelor (76.2%) than clopidogrel (57.1%), but without differences in culprit lesions.

The mean proportion of days covered in the first year after PCI, as a measurement of treatment adherence, was numerically lower in HBR patients than non-HBR patients for all P2Y12 inhibitors (Figure 3). A total of 175 (12.6%) HBR patients switched to another P2Y12 inhibitor in the first year after PCI, mainly from a potent P2Y12 inhibitor (ticagrelor or prasugrel) (Figure 4). One-year IR of the outcomes and multivariate analyses are presented in Supplementary material online, Table S3 and Figure S5, respectively. Compared with clopidogrel, ticagrelor and prasugrel treatment were associated with reduced risk of MACE and all-cause mortality without differences in risk of major bleeding or ischaemic events. Similar results were found in a sub-analysis excluding HBR patients treated with anticoagulants (Supplementary material online, Table S4).

Characteristics of users of P2Y12 inhibitors without HBR are provided in Supplementary material online, Table S5. In short, clopidogrel-treated patients were older and more often female compared with those treated with prasugrel. Users of clopidogrel also had more comorbidities, a higher use of oral anticoagulants, and were less often treated with drug-eluting stents compared with both ticagrelor and prasugrel users. The mean proportion of days covered was ≥80% for all non-HBR users of P2Y12 inhibitors (Figure 3). Within the first year after PCI, 297 (8.9%) non-HBR patients switched to another P2Y12 inhibitor (Supplementary material online, Figure S6). One-year IR and multivariate analyses of the outcomes are provided in Supplementary material online, Table S3 and Figure S5, respectively.

Discussion

In this study of >5500 consecutive STEMI patients treated with PCI, several important findings stand out. First, 33.0% of all PCI-treated all-comer patients and 29.4% of the patients alive and collecting a P2Y12 inhibitor within 7 days from discharge were at HBR according to the PRECISE-DAPT score. These numbers are higher than previous register-based studies on consecutive patients with acute coronary syndromes, of whom 20.4–21.0% were at HBR,^15,16 but similar to an observational study on consecutive patients with STEMI in which 29.1% of patients were at HBR according to the PRECISE-DAPT score.¹⁷ Including patients on oral anticoagulants in our study may have overestimated the proportion of HBR patients, as these patients are often elderly. However, HBR patients still accounted for 31.8% of patients when excluding patients on anticoagulants, and we argue that it is more accurate to include these patients in reporting the proportion of all-comer STEMI patients at HBR.

The PRECISE-DAPT score was derived from pooled data on patients with coronary artery disease treated with stent implantation and DAPT and was validated prospectively in two different cohorts, where it performed moderately with C-indices for out-of-hospital TIMI major or minor bleeding of 0.70 and 0.66.⁵ The Academic Research Consortium (ARC)-HBR evaluator is another tool to assess bleeding risk in PCI-treated patients. It is a consensus-based rather than a data-based evaluator containing 12 dichotomized variables defined as major or minor bleeding criteria, and patients meeting one major or two minor criteria are considered at HBR.¹⁸ It can be argued that the main purpose of the ARC-HBR evaluator has been to align HBR definitions in clinical trials, whereas the PRECISE-DAPT score may be a more feasible 5-item score for clinical use. Both have performed moderately in predicting bleeding events in a STEMI cohort.⁷ In the current study, the PRECISE-DAPT score seemed able to differentiate HBR and non-HBR patients, as HBR patients were older, had more relevant comorbidities (cancer, peptic ulcer, etc.), and a higher use of oral anticoagulants and steroids, exposing them to a greater bleeding risk. Ischaemic risk factors were also higher in HBR, such as relevant comorbidities (diabetes, stroke, etc.), and the PCI operator more often abstained from stent implantation than in non-HBR patients. This emphasizes the dilemma of whether to reduce DAPT to lower bleeding complications in patients with STEMI.¹⁹

P2Y12-inhibitor use differed between HBR and non-HBR patients who survived until 7 days after discharge. Clopidogrel was more often prescribed to HBR patients than non-HBR patients, although most HBR patients received potent P2Y12 inhibitors (prasugrel and ticagrelor). This suggests that even in patients at HBR, the interventional cardiologists focus on ischaemic risk when determining the DAPT regimen. Also, cardiologists do not face most bleeding events, as these patients are usually admitted to non-cardiological departments. In this perspective, bleeding risk is most likely underestimated compared with ischaemic risk. However, the PRECISE-DAPT score was not designed to guide de-escalation of DAPT intensity, but rather to shorten the duration of DAPT in HBR patients.⁵ Additionally, the emphasis on bleeding was formally recognized with the publication of the European-focused update on DAPT in 2017, thus after the present study period, which may explain why most patients in this study received potent DAPT for 12 months according to the guidelines at that time.²⁰

Treatment adherence in the first year after STEMI was high among all P2Y12-inihibitor users, consistent with previous reports of >85% adherence in PCT-treated high-risk MI patients.²¹ The proportion of days covered, as a measurement of treatment adherence, was numerically lower in HBR than non-HBR patients, and more HBR than non-HBR patients had a shorter treatment duration of ≤6 months with the first claimed P2Y12 inhibitor (18.1% vs. 11.7%, respectively). This suggests that the interventional cardiologists may have chosen a shorter DAPT duration for HBR patients, or that discontinuation or switch in DAPT during the first year was more common in HBR patients due to side effects.²² To support this argument, 12.6% of HBR patients compared with 8.9% of non-HBR patients switched P2Y12 inhibitors in the first year after PCI.

HBR all-comer patients had high IRs of major bleeding and MACE, indicating a higher morbidity. Importantly, all-cause mortality was the main driver of MACE, which also included patients who died during PCI. These patients are more likely to be categorized as HBR, as age and impaired renal function predict both bleeding risk and mortality.²³ Nonetheless, a meta-analysis of PCI-treated patients suggests that a PRECISE-DAPT score ≥25 is associated with an increased risk of ischaemic outcomes.²⁴ Balancing the ischaemic and bleeding risks is key, and deviating results exist in terms of whether the predictive risk of ischaemia or bleeding is greater. Perhaps bleeding risk does not depend on which P2Y12 inhibitor is prescribed, but rather on having, e.g. a high PRECISE-DAPT score. This argument finds support in a recent analysis of 14 450 consecutive patients with acute coronary syndromes, in which the transition from clopidogrel to ticagrelor did not lead to more bleeding events.²⁵ This indicates that treatment duration may be of more importance. Currently, an increasing number of studies focus on reduced DAPT in HBR patients and have shown that de-escalation of DAPT intensity and shortening of DAPT duration are non-inferior in terms of net adverse clinical events and superior in terms of bleedings.^26,27 Still, most patients are treated with potent P2Y12 inhibitors for 12 months, irrespective of the bleeding risk. There is a need for randomized controlled trials to personalize DAPT in patients with high ischaemic and bleeding risk, such as the ongoing DAN-DAPT trial (NCT05262803).

This study has important limitations. Due to its observational nature, conclusions cannot be made regarding optimal DAPT in HBR patients. However, when comparing outcomes between different P2Y12 inhibitors in HBR patients, we found a higher risk of MACE and all-cause mortality in clopidogrel users than users of prasugrel and ticagrelor, without differences in major bleeding or ischaemic events. This may be due to lower antithrombotic efficacy of clopidogrel, differences in treatment with P2Y12 inhibitors over the study period, or, more likely, confounding by indication due to higher age and comorbidity burden in clopidogrel-users. Despite relevant statistical adjustments, the results should also be interpreted considering the high risk of residual confounding. Treatment adherence may be overestimated as the proportion of days covered by the claimed prescription was calculated and not the actual intake. Lastly, major bleeding was defined as bleeding leading to hospitalization based on diagnosis codes from inpatient contacts, which could have led to underestimation.

Conclusions

In all-comer STEMI patients treated with PCI, every third patient was at HBR according to the PRECISE-DAPT score, and these patients were more often treated with potent P2Y12 inhibitors (prasugrel or ticagrelor) than with clopidogrel.

Funding

This work was supported by the Novo Nordisk Foundation (grant number 0071947) and Gangstedfonden (grant number R625-A41879).

Conflict of interest. T.E. reports personal fees from Abbott, Bayer, and Novo Nordisk; L.K. reports personal fees from Novartis, Novo Nordisk, and Boehringer; C.T.-P. reports grants from Bayer and Novo Nordisk outside the submitted work; M.M. is supported by a grant from the Novo Nordisk Foundation, has received institutional research grants from Bayer and Novo Nordisk, and has received lecture and/or advisory board fees from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, and Novo Nordisk; E.L.G. has received speaker honoraria or consultancy fees from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, Pfizer, MSD, Novo Nordisk, and Lundbeck Pharma. He is an investigator in studies sponsored by AstraZeneca, Idorsia, and Bayer and has received unrestricted research grants from Boehringer Ingelheim. R.S. is supported by grants from the Novo Nordisk Foundation and Gangstedfonden and has received speaker fees from Boehringer Ingelheim and Amgen. The other authors have nothing to declare.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References