Incidence, Prediction, and Outcomes of Major Bleeding After Percutaneous Coronary Intervention in Chinese Patients

Background The patterns of late major bleeding (MB) after percutaneous coronary intervention (PCI) remain unknown in Chinese patients. Objectives This study sought to determine the incidence, prediction, and long-term outcomes of late MB in Chinese patients. Methods This was a retrospective cohort study from 14 hospitals in Hong Kong. Participants were patients undergoing first-time PCI without MB within 30 days or death within 1 year. Patients were stratified by the presence of late MB, defined as MB between 30 and 365 days. The primary endpoint was all-cause mortality. The secondary endpoints were major adverse cardiac events (MACE). Results A total of 32,057 patients were analyzed. After adjustment for baseline characteristics, periprocedural characteristics, and medications on discharge, the risks of all-cause mortality at 5 years were significantly higher with late MB (HR: 2.15; 95% CI: 1.92-2.41; P < 0.001). Late MB was also associated with a higher risk of MACE (HR: 1.57; 95% CI: 1.03-1.50; P < 0.001), myocardial infarction (HR: 1.25; 95% CI: 1.04-1.52; P = 0.02), and stroke (HR: 1.38; 95% CI: 1.09-1.73; P = 0.006). The CARDIAC (anti-Coagulation therapy, Age, Renal insufficiency, Drop In hemoglobin, baseline Anemia in Chinese patients) score had a good discriminating power for prediction of MB within 365 days (area under the receiver-operating characteristic curve: 0.76). Conclusions Late MB was independently associated with a higher risk of mortality, MACE, myocardial infarction, and stroke in patients undergoing PCI. The CARDIAC score is a simple model that can predict MB after PCI. Prevention of MB represents an important strategy to optimize cardiovascular outcomes for patients undergoing PCI.

intensification or extension of antithrombotic therapy after PCI were counterbalanced by bleeding-related harm. 6,7 Known as the East Asian paradox, East Asians have different thrombotic and bleeding profiles from White patients, with a higher vulnerability to bleeding and lower susceptibility to ischemic events. [8][9][10][11] Unfortunately, even though East Asians are the largest ethnic group (>1.6 billion in population) and half of those undergoing PCI are considered at high bleeding risk, 12,13 the impacts of MB were derived from studies that East Asians were largely underrepresented. Therefore, association between MB and subsequent long-term outcomes in East Asians remains unexplored.
Prevent of bleeding during a late period (approximately 1-12 months after PCI) has been utilized as a therapeutic window to improve clinical outcomes, as antithrombotic therapy can be safely curtailed. Examples include shorter dual antiplatelet therapy (DAPT), [14][15][16] de-escalation to a less potent P2Y 12 inhibitor, 17 and omission of aspirin in those receiving oral anticoagulation therapy. 18 These strategies are contingent upon associations between late MB and adverse outcomes, as well as reliability of bleeding risk score predication. Because such data on East Asian populations were sparse, we sorted to determine the incidence, prediction, and long-term outcomes of late MB in a territory wide cohort of patients receiving PCI in Hong Kong. To assess for any residual confounding by treatment selection, we performed falsification testing with new diagnosis of cancer and hip fracture after PCI. Cancer and hip fracture were selected based on their association with mortality but were biologically unlikely to be causally related to late MB. 21,22 In the primary analysis, the complete case method was adopted to address missing data. To test the robustness of our results, the regression analysis was repeated with the entire cohort using the technique of multiple imputations by chained equations to account for missing data.

METHODS
EXPLORATORY ANALYSES. We explored the timevarying effects of late MB on all-cause mortality with a landmark analysis. The outcome were examined separately between 1 to 3 years, 3 to 5 years, 1 to 5 years, and 1 year to last follow-up after PCI using the same regression model in the primary analysis.
We used our Chinese cohort to externally validate the performance of the ADAPT (Asian Dual Anti-Platelet Therapy) bleeding risk score, developed in a Korean cohort, to predict MB within 3 years after PCI. 23 Next, we developed a risk score to predict the   Table 2). EXPLORATORY ANALYSES. In landmark analysis, the excess mortality risk associated with late MB was strongest between 1 and 3 years, became less strong between 3 and 5 years, but remained significant even after 5 years ( Table 3).
External validation of the ADAPT bleeding risk score with our cohort showed moderate discriminating power for prediction of MB at 3 years in our cohort (AUC: 0.71; 95% CI: 0.70-0.72). The sensitivity and specificity were 59% and 72%, respectively, in our cohort at the recommended cutoff of >3 points. 23 We developed a new risk score for prediction of MB between hospital discharge and 1 year. Five variables were included into the final logistic regression model: anticoagulation therapy on discharge, age, eGFR, drop in hemoglobin after PCI, and baseline anemia (Supplemental Table 3). In the development cohort with an overall MB risk of 5.4%, the discriminating power of the new risk score was good (AUC: 0.76; 95% CI: 0.74-0.77; P < 0.001). The optimal cutoff for prediction of MB was $5 points, conferring a sensitivity of 63% and specificity of 75% (Figure 3). In the validation cohort with an overall MB risk of 5.4%, the risk score retained its discriminating power (AUC: 0.74; 95% CI: 0.72-0.76; P < 0.001; sensitivity of 61% and specificity of 77% at optimal cutoff) ( Figure 3).
The risk score is referred to as the CARDIAC (anti-

Coagulation therapy, Age, Renal insufficiency, Drop
In hemoglobin, baseline Anemia in Chinese patients) score. The score calculation is detailed in Table 4. The calibration plot and absolute risk of MB are detailed in Supplemental Figure 1 and Supplemental Table 4.

DISCUSSION
In this territory-wide PCI registry with Chinese patients exclusively, the estimated incidence rate of MB after PCI was approximately 5% in the first year and 2% thereafter. Late MB was significant associated with long-term mortality up to 5 years after PCI, and the association remained consistent after adjustment for potential confounding factors and multiple sensitivity analyses. A simple prediction model was able to predict MB within the first year after PCI.
The incidence rate of MB in our cohort was higher than other Western cohorts (approximately 1%-3% within first year) 13 but was similar to other East Asian registries. 12,24,25 This was probably related to the East Asian paradox and a high proportion of patients with HBR in East Asian countries. 8,10,12 The bleeding-mortality relationship in our study is similar to previous findings in Western populations.
Early post-procedural bleeding events are associated with 3-fold increase in long-term mortality. 26 The excess risk persists up to 6 years and monotonically accrues over time. 27,28 Post-discharge (ie, nonprocedure related) bleeding events have also been The unadjusted probability of all-cause mortality was higher in the late major bleeding (MB) group (log-rank P < 0.001). PCI ¼ percutaneous coronary intervention. inhibitors, and ulcer prophylaxis with proton pump inhibitors. [14][15][16][17]30 The relationship between late MB and long-term outcomes was mostly derived from European and American cohorts. 2,3 However, East Asians are known to have different thrombotic and bleeding profiles than White patients. [8][9][10][11] This may lead to a differential   Total score 0-16 in our cohort The optimum cutoff was $5.