There is currently a lack of data among the high-bleeding risk (HBR) population on the use of dual antiplatelet loading (LD) with a diagnosis of ACS and undergone PCI, with concerns that the potential bleeding risk may outweigh any benefit in terms of cardiovascular event reduction.

Of 21,654 evaluable HBR—ACS patients who underwent PCI and were included in the CCC-ACS registry in China (2014–2019), 66.2% were treated with LD.

After adjustment for baseline differences, LD was associated with a significantly increased risk of in-hospital mortality [1.89 vs. 1.02%; hazard ratio (HR): 1.71 (95% confidence interval 1.12, 2.42); p < 0.001] and in-hospital bleeding [3.89 vs. 3.3%; HR: 1.25 (1.03, 1.53); p = 0.03].

This analysis of data—excluding HBR patients who received antiplatelet therapy without getting with the guidelines or with a chronic use of antiplatelet therapy—provides evidence that LD may indeed increase in-hospital mortality and bleeding after PCI for HBR-ACS, after adjustment for key baseline characteristics.

The results of this analysis provide the basis for further investigation on the use of LD in HBR-ACS patients, calling for bleeding risk stratification ahead of LD decisions in daily practice.

We used data from the nationwide, real-world registry of CCC-ACS (Improving Care for Cardiovascular Disease in China). Launched in 2014, the CCC-ACS project (clinicaltrial.gov NCT 02306616) is a prospective registry including 159 tertiary and 82 secondary hospitals across China. The registry is a common program of the AHA and the Chinese Society of Cardiology (CSC), recruiting patients with ST-segment elevation myocardial infarction (STEMI) and non-ST elevation acute coronary syndromes (NSTE-ACS). Detailed information has been reported previously (6). Briefly, eligible cases in each participated site are reported by trained data abstractors every month, while the acute myocardial infarction cases have reporting priority.

Baseline characteristics, procedural data, in-hospital treatment and outcomes are recorded via an electronic data capture platform by trained data abstractors. Standardized definitions are utilized across all hospitals for variable collection (7). Third-party clinical research associates were in charge of quality assurance. Audit reports were fed back to each center regularly for quality control. The study was approved by the institutional review board at Beijing Anzhen Hospital, Capital Medical University.

As of 31th December 2019, on the basis of principal discharge diagnosis, 113,650 patients with ACS have been registered. Of these, 21,654 patients with HBR who received a known dose of both aspirin and P2Y₁₂ inhibitor within 24 h of first medical contact and received PCI were included in this study. Patients who received antiplatelet therapy without getting with the guideline recommendations (e.g., only aspirin loading, only P2Y₁₂ loading or loading with both ticagrelor and clopidogrel), received aspirin or P2Y₁₂ inhibitor within 2 weeks before admission were excluded from the analysis. Patients admitted at a situation that could not be treated with oral dual loading (e.g., active bleeding, out of hospital cardiac arrest) were also excluded.

On the basis of data availability, some of the ARC-HBR criteria were not available or need to be modified. The comparison of definitions between ARC-HBR criteria and the currents study were listed in the Supplementary Table 1. ARC-HBR criteria applied in this study are as follows: Age ≥ 75 years (minor), eGFR 30–59 mL/min (minor); Hemoglobin 11–12.9 g/dL for men and 11–11.9 g/dL for women (minor); bleeding tendencies before admission (minor); long-term use of aspirin before admission (minor); prior ischemic stroke or TIA (minor); Warfarin usage before admission (major); eGFR < 30 mL/min (major); Hemoglobin < 11 g/dL (major); platelet count < 100 × 10⁹/L (major); prior hemorrhagic or ischemic stroke (major); medical history of surgery or tooth extraction (major). Patients were classified to HBR patients if at least one major or two minor ARC-HBR criteria were met.

Patients were considered as LD if they received dual loading with both aspirin and P2Y₁₂ inhibitor (clopidogrel or ticagrelor) within 24 h of first medical contact. The administration of aspirin and clopidogrel or ticagrelor with a dose of over 100 and 300 mg or 180 mg, respectively, was defined as LD. Patients who received loading with neither aspirin nor P2Y₁₂ inhibitor were defined as No LD. The information of the type and dose of antiplatelet therapy received within 24 h of first medical contact were specifically entered into the registry.

The primary efficacy endpoint was in-hospital mortality. The primary safety endpoint was in-hospital bleeding. In detail, stent thrombosis was defined as an acute/subacute stent occlusion after procedure. The event “myocardial infarction (MI)” corresponds in our study to a “reinfarction” within the hospitalization following the index MI. The stroke was defined as a new neurological deficit during hospitalization. Major bleeding was defined as any of the following events: fatal bleeding, intracranial bleeding, retroperitoneal bleeding, and drop in hemoglobin ≥ 40 g/L during hospitalization, transfusion with overt bleeding, or bleeding requiring surgical intervention. Fatal bleeding was defined as any death within 7 days following a major bleeding. All of these endpoints were reported by clinical doctors and documented in medical records during hospitalization.

Baseline characteristics and procedural details of ACS patients were described. Continuous variables are presented as a mean and standard deviation (SD) and categorical variables are presented as frequency and percentage. T-test for continuous variables and Chi-square test for categorical variables were used to test the statistical differences between groups. Kaplan-Meier methods were used to estimate 30-days event rates for each endpoint, and comparisons between study groups were performed using log-rank test.

To consolidate the findings, we also carried out inverse probability of treatment weighting (IPTW) using the propensity score method in study cohort and compared difference among two groups. A logistic regression was used to estimate propensity score, adjusting following variables: age, gender, previous disease history (MI, PCI, CABG, hypertension, dyslipidemia, diabetes, renal failure history, heart failure history, peripheral artery disease, ischemic stroke, hemorrhagic stroke), current smoker, cardiogenic shock, Killip class, systolic blood pressure, heart rate, elevated troponin (Tn)T or TnI, hemoglobin, diagnosis of STEMI, Global Registry of Acute Coronary Events (GRACE) score, preadmission of aspirin and P2Y₁₂ inhibitor, DAPT at arrival, glycoprotein IIb/IIIa inhibitors (GPI) at arrival, ACEIs/ARBs or β-blockers at admission, PCI access and drug-eluting stent. IPTW was calculated by each individual base on his/her propensity score (PS). Each case of LD group was given a weight of Pt/PS, and each case from the no LD group was given a weight (1-Pt)/(1-PS), where Pt refers to the percentage of patients receiving any LD among the whole cohort. By this way we could get a stabilized weight for each case of the study cohort, avoiding any extreme values that may result in unreliable outcomes.

Cox proportional hazard models were used to analyze the effects of LD on each endpoint, and Hazard ratios (HR) and 95% confidence intervals (95% CI) were reported. Two models were applied to explore the association between LD and endpoints: (1) an unadjusted model; and (2) an adjusted model controlling for prespecified variables with impact of in-hospital outcomes (gender, medical history, current smoker, cardiogenic shock, Killip class, elevated TnT or TnI, anticoagulant during hospitalization, year of enrolment, and hospital type). Subgroup analysis among thirteen subsets was performed to assess the consistency of treatment effects on primary efficacy endpoint and safety endpoint.

Imputation was performed for variables with missing data with the sequential regression multiple imputation method by IVEware software version 0.2 (Survey Research Center, University of Michigan, Ann Arbor, MI, USA). No LD was used as reference in all analyses. All tests were two-sided, and a P-value of <0.05 was considered to indicate statistical significance. All analyses were performed using SAS version 9.3 (SAS Institute Inc.; Cary, NC) statistical software.

This research was done without patient involvement.

Of 113,650 diagnosed ACS patients enrolled into the CCC-ACS database between November 2014 and December 2019, 21,654 patients with HBR and PCI were analyzed as the final population (Figure 1). Of them, 7,332 were classified in the no LD group and 14,322 in the LD group. Baseline characteristics and treatment of the study population were presented in Table 1. Patients receiving a LD were on average older, less likely to have heart failure history and prior ischemic stroke. They presented with worse cardiac function and higher hemoglobin level. They were less likely to receive in-hospital ACEIs/ARBs or β-blockers. But they were prescribed more frequently of GPI and anticoagulant during hospitalization. Among LD patients, PCI was more frequently performed through radical access with drug-eluting stent. The use of potent P2Y₁₂ inhibitor, ticagrelor, was less frequent at 40.7% in the LD patients. After adjusting with IPTW method, baseline characteristics were well balanced (Table 1).

Table 2 and Figure 2 showed the in-hospital outcomes between two groups. Mortality before discharge was higher in patients who were with LD than those without LD (1.89 vs. 1.02%, p < 0.001), with an adjusted HR (95% CI) of 1.71 (1.12, 2.42), (Table 2; Figure 2A). In the multivariate-adjusted analysis, LD was not associated with a higher risk of in-hospital ischemic stroke, with an adjusted HR of 2.01 (0.86, 4.69). The estimated risk did not differ between the two groups after adjustment in other in-hospital events, such as the MI (0.25 vs. 0.25%, P = 0.93) and stent thrombosis (0.11 vs. 0.05%; P = 0.19).

Higher rates of bleeding, major bleeding and fatal bleeding were observed in LD group than in the No LD group in the whole study population (3.89 vs. 3.30%, P = 0.03; 2.54 vs. 2.48%, P = 0.79; 0.16 vs. 0.08%, P = 0.13, respectively). After adjusting patients' characters, antithrombotic treatment and hospital effect with cox model, LD was associated with a higher rate of bleeding, with an adjusted HR of 1.25 (1.03, 1.53); (Table 2; Figure 2B). However, the estimated risk did not differ between the two groups either for major bleeding [adjusted HR (95% CI): 1.07 (0.84, 1.36)] or fatal bleeding [adjusted HR (95% CI): 2.01 (0.63, 6.47)] (Table 2). The location and intervention of bleeding were showed in Table 3. Gastrointestinal bleeding, subcutaneous hemorrhage and access site bleeding were reported more in LD group (0.98 vs. 0.67%, P = 0.02; 0.51 vs. 0.15%, P < 0.001; 0.52 vs. 0.25%, P = 0.02, respectively). There was no statistically difference between groups in intracranial or retroperitoneal bleeding, as well as in the surgical intervention and blood transfusion.

Subgroup analysis was conducted in thirteen subgroups of the whole study population (Figure 3). The results showed LD was associated with an increased risk of mortality with increasing risk of bleeding, which were consistent with the main results. In these analyses, there were no significant interactions with baseline or procedural variables, except advanced age, female, diabetes, smoking, renal insufficiency and anticoagulation during hospitalization.

For patients treated with clopidogrel, the rate of mortality was 1.93% for individuals with LD and 1.02% for those without LD, with an adjusted HR (95% CI) of 1.70 (1.18, 2.43). For patients treated with ticagrelor, the rate of the all-cause death was 1.84% for individuals with LD and 1.02% for those without LD, with an adjusted HR (95% CI) of 1.57 (1.00, 2.47) (Table 4). The higher risk of bleeding of LD remained consistent across the P2Y₁₂ inhibitor types, especially in ticagrelor [adjusted HR (95% CI): 1.31 (1.02, 1.69)]. The use of LD was not associated with increased risk of MI, stent thrombosis and major bleeding for patients treated with either clopidogrel or ticagrelor.

Several limitations of this observational study should be noted. First, selection bias and residual confounding exist in our analyses as all observational studies. Nevertheless, this study provides real-world data from a large cohort with complete coverage of ACS patients in China. Second, we did not adjust for the switch in P2Y₁₂ inhibitors, which is happened frequently among patients treated with PCI. However, chronic, or incorrect use of antiplatelet drugs, such as loading with both ticagrelor and clopidogrel, were excluded in order to minimize their confounding effect. The CCC-ACS does not gather all information for ARC-HBR classification, and we used modified criteria instead. Some factors related to the patient's adjunctive treatment regimens changed throughout the study period. We accounted these changes as a covariate in the statistical models.