Article ID: CJ-20-0806
Background: This observational study validated Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria and the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score in patients with ST-elevation myocardial infarction (STEMI) undergoing percutaneous coronary intervention.
Methods and Results: Risk clusters of 939 STEMI patients with traceable 1-year outcomes were assessed according to ARC-HBR criteria and PRECISE-DAPT score. The diagnostic accuracy and first-year probability of bleeding events, defined as Bleeding Academic Research Consortium (BARC) 3 or 5, according to risk cluster were assessed. Of all patients, 42.9% and 46.8% were classified as HBR (ARC-HBR criteria) and at high risk (PRECISE-DAPT score), respectively, and bleeding events were observed in 13.7% and 16.2% of these patients. The C-statistic for ARC-HBR criteria and the PRECISEDAPT score was 0.60 and 0.69, respectively (P<0.01). Patients with mechanical hemodynamic support devices had high bleeding rates, even in the non-HBR group (22.6%), and excluding these patients improved the C-statistics, making them equivalent between the 2 models (0.72 vs. 0.74; P=0.53). Bleeding event probabilities (95% confidence intervals) were equivalent in high-risk patients in the 2 models (0.12 [0.09–0.16] vs. 0.12 [0.08–0.16]).
Conclusions: After exclusion of patients with mechanical devices, who had high bleeding event rates regardless of risk cluster, both ARC-HBR criteria and the PRECISE-DAPT score had high predictive ability.
In the mature drug-eluting stent (DES) era, the dilemma between preventing bleeding events and stent thrombosis (ST) tilts towards bleeding events. Numerous clinical studies, such as the Randomized Evaluation of Sirolimus-eluting versus Everolimus-eluting stent Trial (RESET) and Optimized Duration of Clopidogrel Therapy Following Treatment with the Zotarolimus-Eluting Stent in Real-World Clinical Practice (OPTIMIZE) trials, have investigated the optimal period of dual-antiplatelet therapy (DAPT).1,2 Because short DAPT is the main strategy to avoid adverse bleeding events, risk stratification is important to determine the appropriate duration of DAPT.
To establish universal criteria for bleeding endpoints in cardiovascular clinical trials, the Academic Research Consortium (ARC) proposed the Bleeding ARC (BARC) definition of bleeding in 2011,3 followed by the ARC high bleeding risk (HBR) criteria to stratify bleeding risk in 2019.4 These scales are standard tools for the stratification of bleeding risk in patients who undergo percutaneous coronary intervention (PCI).
Because patients with ST-elevation myocardial infarction (STEMI) have a higher risk of both bleeding and ST, risk stratification in this group is clinically important. However, all risk assessment models, including ARC-HBR, are based primarily on elective PCI data. The diagnostic ability of ARC-HBR in STEMI patients undergoing primary PCI has been poorly explored.
The purpose of the present study was to assess the predictive ability of ARC-HBR criteria and the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score for bleeding events during the first year after primary PCI in STEMI patients.
To assess the predictive ability of ARC-HBR criteria and the PRECISE-DAPT score for bleeding events during the first year after primary PCI in STEMI patients, this study assessed the relationship between these predictive models and bleeding events.
Study Design and PopulationThis study was designed as an observational study, and reviewed the medical records of suitable patients, namely those with STEMI who were admitted to the Tokai University School of Medicine with acute symptom onset between January 2006 and January 2019 and underwent primary PCI within 24 h of symptom onset and who had traceable 1-year outcomes. The study cohort comprised 939 consecutive patients who were classified into risk clusters according to the 2 risk models, namely ARC-HBR criteria (non-HBR and HBR groups) and PRECISE-DAPT score (very low-, low-, moderate-, and high-risk groups).5 The performance of these risk models was evaluated using patients’ medical records.
Outcomes and DefinitionsSTEMI was defined according to the Fourth Universal Definition of Myocardial Infarction.6 The primary outcome of the study was bleeding events during the first year after primary PCI. Bleeding events were defined as BARC 3 or 5.3
Statistical AnalysisNumerical factors with a normal distribution are given as the mean±SD and were compared between 2 groups using Student’s t-test. Fisher’s test was used to determine the significance of differences in categorical variables between groups.
Receiver operating characteristic (ROC) analysis and C-statistics were used to quantify the ability of ARC-HBR criteria and the PRECISE-DAPT score to predict bleeding events. The STATA (StataCorp, College Station, TX, USA) command roccomp was used to test the equality of the C-statistics for the 2 models.
A logit model was used to assess the effects of ARC-HBR criteria and the PRECISE-DAPT score on bleeding events. The effect of the risk cluster, defined according to both risk models, on the bleeding event was assessed by marginal effects with 95% confidence intervals (CIs). The marginal effect of a predictor in a categorical response model estimates how much the probability of a response level changes as the risk cluster changes, and the mean marginal effect was calculated for each observation in the data and then averaged. Mean marginal effects for the probability of the bleeding event by risk model were plotted and are shown.
P<0.05 were considered significant. All statistical calculations were performed using STATA statistical software version 14.0 (StataCorp).
The present study assessed the ability of ARC-HBR criteria and the PRECISE-DAPT score to predict bleeding events during the first year after primary PCI.
Patients’ baseline characteristics and use of antiplatelet therapy are summarized in Table 1 and Table 2. The proportion of patients with HBR according to ARC-HBR criteria was 42.9%. The number of patients with major and minor criteria was as follows: 327 had no major and no minor criteria, 209 had no major criteria and 1 minor criterion, 164 had no major and ≥2 minor criteria, 183 had 1 major criterion, 40 had 2 major criteria, and 16 had ≥3 major criteria.
| Overall (n=939) |
Non-HBR (n=536) |
HBR (n=403) |
P value | |
|---|---|---|---|---|
| Age (years) | 66.3±12.4 | 61.4±10.6 | 72.9±11.5 | <0.01 |
| ≥75 yearsA | 261 (27.8) | 44 (8.2) | 217 (53.9) | <0.01 |
| Male sex | 744 (79.2) | 445 (83.0) | 299 (74.2) | <0.01 |
| Hypertension | 699 (74.4) | 388 (72.4) | 311 (77.2) | 0.10 |
| Dyslipidemia | 652 (69.4) | 419 (78.2) | 233 (57.8) | <0.01 |
| Diabetes | 329 (35.0) | 169 (31.5) | 160 (39.7) | <0.01 |
| Insulin | 47 (5.0) | 20 (3.7) | 27 (6.7) | 0.04 |
| Current smoker | 334 (35.6) | 242 (45.2) | 92 (22.8) | <0.01 |
| Any ischemic strokeA | 82 (8.7) | 10 (1.9) | 72 (17.9) | <0.01 |
| Moderate to severe ischemic stroke within the past 6 monthsB | 2 (0.2) | 0 | 2 (0.5) | 0.18 |
| Previous spontaneous ICHB | 15 (1.6) | 0 | 15 (3.7) | <0.01 |
| Previous traumatic ICH within the past 12 monthsB | 18 (1.9) | 0 | 18 (4.5) | <0.01 |
| Prior spontaneous bleedingC | 33 (3.5) | 3 (0.6) | 30 (7.4) | <0.01 |
| Requiring hospitalization or transfusion in the past 6 monthsB |
5 (0.5) | 0 | 5 (1.2) | 0.01 |
| Requiring hospitalization or transfusion in the past 12 monthsA |
2 (0.2) | 0 | 2 (0.5) | 0.18 |
| Brain arteriovenous malformationsB | 1 (0.1) | 0 | 1 (0.3) | 0.43 |
| Chronic bleeding diathesisB | 0 | 0 | 0 | N.S. |
| Liver cirrhosis with portal hypertensionB | 2 (0.2) | 0 | 2 (0.5) | 0.18 |
| Active cancer in the past 12 monthsB | 21 (2.2) | 0 | 21 (5.2) | <0.01 |
| Non-deferrable major surgery on DAPTB | 1 (0.1) | 0 | 1 (0.3) | 0.43 |
| Recent major surgery or trauma within 30 daysB | 4 (0.4) | 0 | 4 (1.0) | 0.03 |
| Oral NSAIDs or steroidsA | 25 (2.7) | 6 (1.1) | 19 (4.7) | <0.01 |
| White blood cells (×103/μL) | 10.8±3.8 | 11.0±3.7 | 10.4±3.9 | 0.02 |
| Hb (g/dL) | 14.2±2.3 | 15.1±1.6 | 13.0±2.5 | <0.01 |
| 11–12.9 g/dL for men, 11–11.9 g/dL for womenA | 30 (3.2) | 10 (1.9) | 20 (5.0) | <0.01 |
| <11 g/dLB | 77 (8.2) | 0 | 77 (19.1) | <0.01 |
| Platelets (×104/μL) | 21.4±6.6 | 22.4±6.3 | 20.1±6.6 | <0.01 |
| <100×104/μLB | 14 (1.5) | 0 | 14 (3.5) | <0.01 |
| Serum creatinine (mg/dL) | 1.1±1.2 | 0.9±0.2 | 1.5±1.6 | <0.01 |
| eGFR (mL/min/1.73 m2) | 62.7±23.0 | 70.8±18.2 | 52.1±24.5 | <0.01 |
| 30–59A | 418 (44.5) | 139 (25.9) | 279 (69.2) | <0.01 |
| <30B | 66 (7.0) | 0 | 66 (16.4) | <0.01 |
| Hemodialysis | 18 (1.9) | 0 | 18 (4.5) | <0.01 |
| LVEF (%) | 52.5±13.0 | 53.5±12.4 | 51.2±13.8 | <0.01 |
| Approach site | ||||
| Transradial | 733 (78.1) | 458 (85.5) | 275 (68.2) | <0.01 |
| Transfemoral | 191 (20.3) | 71 (13.3) | 120 (29.8) | |
| Other | 15 (1.6) | 7 (1.3) | 8 (2.0) | |
| IABP | 240 (25.6) | 123 (23.0) | 117 (29.0) | 0.02 |
| ECLS | 58 (6.2) | 35 (6.5) | 23 (5.7) | 0.35 |
| Stent | ||||
| None | 121 (12.9) | 50 (9.3) | 71 (17.6) | <0.01 |
| BMS | 474 (50.5) | 280 (52.2) | 194 (48.1) | |
| 1 st-generation DES | 58 (6.2) | 34 (6.3) | 24 (6.0) | |
| 2nd-generation DES | 339 (36.1) | 204 (38.1) | 135 (33.5) | |
| GRACE risk score | 168.9±50.1 | 152.0±43.4 | 191.6±49.6 | <0.01 |
| PRECISE-DAPT | ||||
| Score | 25.2±12.1 | 19.4±7.3 | 33.1±12.8 | <0.01 |
| Risk cluster | ||||
| High | 439 (46.8) | 128 (23.9) | 311 (77.2) | <0.01 |
| Moderate | 245 (26.1) | 191 (35.6) | 54 (13.4) | |
| Low | 185 (19.7) | 158 (29.5) | 27 (6.7) | |
| Very low | 70 (7.5) | 59 (11.0) | 11 (2.7) | |
| Bleeding | 91 (9.7) | 36 (6.7) | 55 (13.7) | <0.01 |
| Bleeding or transfusion | 171 (18.2) | 60 (11.2) | 111 (27.5) | <0.01 |
| All-cause death | 238 (25.4) | 78 (14.6) | 160 (39.7) | <0.01 |
| Myocardial infarction | 42 (4.5) | 27 (5.0) | 15 (3.7) | 0.34 |
| Stroke | 66 (7.0) | 27 (5.0) | 39 (9.7) | <0.01 |
Unless indicated otherwise, data are given as the mean±SD or as n (%). AMinor criterion for ARC-HBR. BMajor criterion for ARC-HBR. CCriterion for the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score. BMS, bare-metal stent; DAPT, dual antiplatelet therapy; DES, drug-eluting stent; ECLS, extracorporeal life support; eGFR, estimated glomerular filtration rate; Hb, hemoglobin; IABP, intra-aortic balloon pump; ICH, intracranial hemorrhage; LVEF, left ventricular ejection fraction; NSAIDs, non-steroidal anti-inflammatory drugs.
| Overall (n=939) |
Non-HBR (n=536) |
HBR (n=403) |
P value | |
|---|---|---|---|---|
| No. antiplatelet agents | ||||
| None | 19 (2.0) | 2 (0.4) | 17 (4.2) | <0.01 |
| Single | 67 (7.1) | 14 (2.6) | 53 (13.2) | |
| Two | 839 (89.4) | 514 (95.9) | 325 (80.7) | |
| Three | 14 (1.5) | 6 (1.1) | 8 (2.0) | |
| Duration of DAPT (days) | 336.2±465.0 | 377.1±502.3 | 271.5±390.8 | <0.01 |
| Type of antiplatelet agents | ||||
| Aspirin | 899 (95.7) | 531 (99.1) | 368 (91.3) | <0.01 |
| Clopidogrel | 555 (59.1) | 313 (58.4) | 242 (60.1) | 0.61 |
| Prasugrel | 168 (17.9) | 116 (21.6) | 52 (12.9) | <0.01 |
| Ticlopidine | 142 (15.1) | 87 (16.2) | 55 (13.7) | 0.27 |
| Cilostazol | 23 (2.5) | 13 (2.4) | 10 (2.5) | 0.96 |
| Oral anticoagulation | 99 (10.5) | 0 | 99 (24.6) | <0.01 |
Unless indicated otherwise, data are given as the mean±SD or as n (%). DAPT, dual antiplatelet therapy.
A bleeding event occurred in 9.7% of patients. The bleeding event rates by risk model cluster were as follows: 6.7% and 13.7% in the non-HBR and HBR groups, respectively (ARC-HBR criteria); and 4.3%, 1.1%, 6.1%, and 16.2% in the very low-, low-, moderate-, and high-risk groups based on PRECISE-DAPT scores, respectively.
Comparison of Diagnostic AccuracyThe diagnostic accuracy of the 2 risk models was assessed by C-statistics after ROC analysis, as shown in Figure 1. The C-statistic was significantly smaller with the ARC-HBR criteria than with the PRECISE-DAPT score (0.60 vs. 0.69, respectively; P<0.01).
Receiver operating characteristic (ROC) analysis between Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria and the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score. The C-statistics for ARC-HBR criteria and the PRECISE-DAPT score are 0.60 and 0.69, respectively (P<0.01).
There were 245 patients (26.1%) who used mechanical hemodynamic support devices, intra-aortic balloon pumps (IABPs), and/or extracorporeal life support (ECLS) during primary PCI. Bleeding event rates were high overall (19.6%; n=48), even in the non-HBR group (22.6%), although the rate of bleeding events in patients in the non-HBR group without mechanical devices was as low as 1.9% (Figure 2).
Bleeding events in patients with and without mechanical devices stratified using Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria into HBR and non-HBR groups. Among patients without mechanical devices, the rate of bleeding events was 1.9% in the non-HBR group (n=8/412) and 12.4% in the HBR group (n=35/282). Among patients with mechanical devices, the rate of bleeding events was 22.6% in the non-HBR group (n=28/124) and 16.5% in the HBR group (n=20/121).
ROC curves excluding patients who used mechanical hemodynamic support devices are shown in Figure 3. In this analysis, the C-statistics were statistically equivalent between the ARC-HBR criteria and the PRECISE-DAPT score (0.72 vs. 0.74, respectively; P=0.53).
Receiver operating characteristic (ROC) analysis between Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria and the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score in patients without mechanical devices. The C-statistics for ARC-HBR criteria and the PRECISE-DAPT score are 0.72 and 0.74, respectively (P=0.53).
To compare predictive ability by risk cluster between the ARC-HBR criteria and PRECISE-DAPT scores in patients without mechanical support devices, the probability of a bleeding event by risk cluster was determined using marginal effects (Figure 4). The probability of HBR with ARC-HBR criteria was almost equivalent to that of being in the high-risk group based on the PRECISE-DAPT score (0.12 [95% CI 0.09–0.16] vs. 0.12 (95% CI 0.08–0.16], respectively). In addition, the probability of non-HBR with the ARC-HBR criteria was 0.02 (95% CI 0.01–0.03), whereas the probabilities of very low, low, and moderate risk with the PRECISE-DAPT score were 0.00 (95% CI 0.00–0.01), 0.01 (95% CI 0.00–0.02), and 0.03 (95% CI 0.01–0.05), respectively.
Bleeding event probability by risk cluster according to (A) Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria and (B) the Predicting Bleeding Complication in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score, demonstrated using marginal effects. (A) Bleeding event probability (95% confidence interval) according to ARC-HBR criteria was 0.02 (0.01–0.03) in the non-HBR group and 0.12 (0.09–0.16) in the HBR group. (B) Bleeding event probabilities (95% confidence intervals) in the very low-, low-, moderate-, and high-risk groups according to PRECISE-DAPT score were 0.00 (0.00–0.01), 0.01 (0.00–0.02), 0.03 (0.01–0.05), and 0.12 (0.08–0.16), respectively.
The present study assessed the predictive ability of ARC-HBR criteria and the PRECISE-DAPT score for bleeding events in the first year after primary PCI in STEMI patients. Patients with HBR according to ARC-HBR criteria had a high rate of bleeding events. The diagnostic ability of ARC-HBR criteria was lower than that of the PRECISE-DAPT score in the raw cohort. Patients with mechanical support devices had a high rate of bleeding events regardless of their risk cluster. Excluding these patients improved the diagnostic ability of both risk models, making them statistically equivalent.
After late ST was recognized as a major problem with first-generation DESs,7,8 progress in DESs and novel P2Y12 inhibitors has shifted medical opinion from lifelong DAPT to short-term administration.9,10 The DAPT Study showed that therapy extension beyond 12 months with clopidogrel or prasugrel reduced ischemic events and ST, but had no benefit on mortality and even increased bleeding events.11 In order to define patients at high risk of bleeding to avoid an unreasonably long DAPT, several bleeding risk scoring models have been reported, such as ACUITY,12 GUSTO,13 ISTH,14 PLATO,10 and TIMI.15 The STOPDAPT-2 trial demonstrated that 1 month of DAPT followed by clopidogrel monotherapy resulted in a significantly lower rate of the composite of cardiovascular and bleeding events compared with 12 months of DAPT.16 Current European Society of Cardiology guidelines regarding the antiplatelet regimen after PCI published in 2018 are divided into stable coronary artery disease and acute coronary syndrome.17 For the former condition, the antiplatelet regimen is generally recommended to be 6 months of DAPT, but 1 or 3 months in patients at HBR. For acute coronary syndrome, the antiplatelet regimen is typically recommended to be 12 months, but 6 months in patients at HBR. Hence, although the definition of HBR is important to determine the duration of DAPT, a unified definition was still required. The ARC-HBR was established to detect patients with HBR who underwent PCI and to develop universal criteria for bleeding endpoints for cardiovascular clinical trials. In its updated 2020 guidelines, the Japanese Circulation Society for patients with coronary artery disease recommends using ARC-HBR criteria to assess bleeding risk, the first such national society to do so.18
The present study has several novel aspects. First, the ARC-HBR criteria, originally developed to assess bleeding risk in patients after elective PCI, were applied to STEMI patients. Second, the risk assessment was performed using data from the acute phase immediately after arrival at hospital. Third, the study used real-world data that included many critical patients who received IABP or ECLS. Finally, the data were from Japanese patients. The results of the study indicate good diagnostic ability of the models and suggest that both models can be used to assess STEMI patients during the acute phase.
Patients with HBR who satisfied ARC-HBR criteria accounted for 43% of the study cohort. Almost 40% of patients who undergo PCI are assumed to have HBR taking into consideration HBR rates of 39% in the Bern PCI Registry19 and 43% in the CREDO-Kyoto Registry Cohort-2.20 The proportion of patients with a high-risk of PRECISE-DAPT score and the probability of bleeding events in this group (47% and 0.12, respectively) were equivalent to those of patients with an ARC-HBR classification. Conversely, the probability of a bleeding event in the non-HBR group was in the mid-range between the low- and moderate-risk groups according to the PRECISE-DAPT score. Using these predictive models, patients categorized as “high risk” should receive the minimum dose and shortest duration of antiplatelet therapy. Moreover, PCI operators should aim to use the minimum number and shortest length of stents to avoid bleeding events.
Patients with mechanical support devices were at high risk of bleeding, even when judged to be low risk; bleeding events in these patients were independent of the risk stratification according to the predictive models. We consider these patients to be inappropriate for risk stratification using the predictive models because the diagnostic accuracy of the models was improved when these patients were excluded from the analysis. To reduce the risk of bleeding, antiplatelet therapy should be minimized in these patients. Hence, it is important to minimize the number and length of stents in primary PCI. The bleeding event rate was 6.7% even in the non-HBR group. This rate was higher than that stated in the original report on the ARC-HBR definition (4%). One reason to explain this observation may be that the study population consisted of STEMI patients, including a high proportion of patients with mechanical support devices. The high rate of bleeding in Japanese patients is also evident in real-world clinical practice.
The present study has several limitations. First, the bleeding event rate was as high as 9.7%. Many high-risk patients received mechanical support devices: 25.6% received IABPs and 6.2% received ECLS. Moreover, although there were many patients whose 1-year outcome could not be tracked and so were excluded from the study, these patients were mostly event free. Hence, the exclusion of these patients may have actually increased the event rate. Second, this study was an observational in study. The observation period was short in some patients due to their loss to follow-up. Third, the data are derived from a single center. Fourth, devices and medications, such as the types of stents or antiplatelet agents used, changed over the period of patient registration from 2006 to 2019. Fifth, there is a possibility that the ARC-HBR criteria were underestimated because the present study was an observational study, and the clinical component of the criteria had not been diagnosed during hospitalization.
In conclusion, the ARC-HBR criteria have sufficient diagnostic ability to predict bleeding events in STEMI patients after primary PCI, except in patients with mechanical hemodynamic support devices.
All authors have contributed to the study in one or more of the following ways: conception, design, conduct, analysis, interpretation of data, and reporting of the work described in the article.
None.
Y.I. is a member of Circulation Journal’s Editorial Team. T.F. has no conflicts of interest to declare.
This study was approved by the Institutional Review Board for Clinical Research of Tokai University Hospital (20R135).
The deidentified participant data will not be shared.
