The Combinations of the Patterns of Non-adherence to Anti-platelet Regimen in Stented Patients (PARIS) and Coronary Revascularization Demonstrating Outcome Study in Kyoto (CREDO-Kyoto) Thrombotic and Bleeding Risk Scores on Clinical Outcomes in Patients with Acute Myocardial Infarction

Abstract

Objective

The Patterns of Non-adherence to Anti-platelet Regimen in Stented Patients (PARIS) and Coronary Revascularization Demonstrating Outcome Study in Kyoto (CREDO-Kyoto) thrombotic and bleeding risk scores were established to predict ischemic and bleeding events in patients undergoing percutaneous coronary intervention (PCI). However, whether or not the combination of these risk scores is predictive of clinical outcomes is unclear.

Methods

This bicenter registry included a total of 1,098 patients with acute myocardial infarction (MI) undergoing primary PCI. Patients were divided into three groups according to the PARIS and CREDO-Kyoto thrombotic and bleeding risk scores. The study endpoints included the rates of both ischemic (cardiovascular death, recurrent MI, and ischemic stroke) and major bleeding (Bleeding Academic Research Consortium type 3 or 5) events at two years.

Results

Two years after primary PCI, ischemic and major bleeding events occurred in 17.3% and 10.2% of patients, respectively. The higher-risk categories of PARIS and CREDO-Kyoto scores were associated with increased risks of ischemic and bleeding events. The rates of ischemic and major bleeding events progressively increased with the increase in risk categories in the two risk scoring systems. In the receiver operating characteristic curve analysis, the addition of CREDO-Kyoto thrombotic and bleeding risk scores to PARIS scores significantly improved diagnostic ability in predicting ischemic (area under the curve: 0.59 vs. 0.63, p=0.01) and bleeding (area under the curve: 0.65 vs. 0.68, p=0.01) events.

Conclusion

The combinations of the PARIS and CREDO-Kyoto risk scores might be useful for evaluating ischemic and bleeding risks in patients with acute MI undergoing primary PCI.

Introduction

In patients suffering from acute myocardial infarction (MI), primary percutaneous coronary intervention (PCI) is currently a standard-of-care procedure to reduce subsequent cardiac events and mortality (1). With advances in medical therapy and primary PCI, the prognosis of acute MI has considerably improved. However, acute MI remains a leading cause of mortality worldwide (2).

Previous reports concerning patients undergoing PCI showed that both ischemic and bleeding events after the index procedure had similar prognostic impacts on the mortality (3-5). Thus, risk assessments for both ischemic and bleeding events are recommended in recent guidelines (6,7). In this context, the Patterns of Non-adherence to Anti-platelet Regimen in Stented Patients (PARIS) and Coronary Revascularization Demonstrating Outcome Study in Kyoto (CREDO-Kyoto) thrombotic and bleeding risk scores were developed from Western and Japanese PCI populations respectively to evaluate ischemic and bleeding risks individually (8,9).

Although we recently reported that the PARIS and CREDO-Kyoto risk scores were both predictive for ischemic and bleeding events in patients with acute MI (10), the two risk scoring systems often determined patient risks differently.

We therefore evaluated whether or not the combination of the PARIS and CREDO-Kyoto risk scores provides additive predictability for ischemic and bleeding outcomes in patients with acute MI undergoing primary PCI.

Materials and Methods

Study design and population

This was a retrospective, bicenter, observational study. From January 2012 to March 2020, a total of 1,102 patients with acute MI underwent primary PCI at 2 tertiary-care hospitals: Chiba University Hospital and Eastern Chiba Medical Center.

Study details have been described previously (10-12). In brief, acute MI was defined according to the fourth universal definition of MI (13). In the present study, patients with both ST-segment elevation and non-ST-segment elevation MI were included. The term “primary PCI” represents the procedures for not only ST-elevation MI but also non-ST-elevation MI. All patients underwent PCI procedures within 48 h of their admission in this study. All PCI procedures were performed per local standard practice and guideline recommendations (1,6,7), including dual antiplatelet therapy, intracoronary imaging, and contemporary drug-eluting stents (14-17). Patients undergoing plain balloon angioplasty followed by coronary bypass grafting (n=4) were excluded. Thus, a total of 1,098 patients with acute MI undergoing primary PCI were included in the present analysis.

The studies complied with the Declaration of Helsinki. Informed consent for the present study was obtained in the form of opt-out, and the ethics committee of Chiba University Hospital and Eastern Chiba Medical Center approved this study.

PARIS and CREDO-Kyoto risk scores

Thrombotic and bleeding risks were evaluated by the PARIS and CREDO-Kyoto risk scores, as previously reported (Supplementary material 1) (8,9). The PARIS thrombotic and bleeding risk scores comprise six factors in each, including diabetes, acute coronary syndrome presentation, current smoking, renal impairment, prior PCI, and a history of coronary artery bypass grafting for the thrombotic risk score and age, body mass index, current smoking, anemia, renal impairment, and triple therapy on discharge for the bleeding risk score (8). In contrast, the CREDO-Kyoto thrombotic and bleeding risk scores consist of eight and seven components, respectively, including renal impairment, atrial fibrillation, peripheral artery disease, anemia, age, heart failure, diabetes, and chronic total occlusion for the thrombotic risk score and low platelet counts, renal impairment, peripheral artery disease, heart failure, prior MI, malignancy, and atrial fibrillation for the bleeding risk score (9). Patients were divided into low, intermediate, and high thrombotic and bleeding risk groups based on the thresholds (Supplementary material 1).

Outcomes and statistical analyses

Follow-up data were obtained from medical records at Chiba University Hospital and Eastern Chiba Medical Center. The primary endpoint of the present study was the rate of ischemic (cardiovascular death, recurrent MI, and ischemic stroke) and major bleeding (Bleeding Academic Research Consortium type 3 or 5) events at two years, adjudicated according to the consensus documents (13,18,19).

Statistical analyses were performed using the JMP Pro software program, ver. 15.0.0 (SAS Institute, Cary, USA). All data are expressed as the mean±standard deviation or frequency (%). Time-to-event outcomes were evaluated with Kaplan-Meier methods, and the log-rank test was used for between-group comparisons. The event rates at two years were estimated by Kaplan-Meier methods. A receiver operating characteristic (ROC) curve analysis was performed based on ischemic and bleeding events. The area under the curve (AUC) of the ROC curve was compared using the Delong method. Beyond the risk scores, a Cox proportional hazards model was employed to identify factors associated with ischemic and bleeding events in the present patient population, with unadjusted and adjusted hazard ratios and corresponding 95% confidence intervals. The variables were selected based on an association with ischemic and bleeding events using an inclusion criterion of p＜0.10 on univariable analyses. A p value ＜0.05 was considered statistically significant.

Results

Table 1 and Supplementary material 2 list baseline characteristics and medications at discharge. Approximately 20% of patients presented with Killip class IV, and intravascular ultrasound and contemporary drug-eluting stents were used in most cases. The distributions of the PARIS and CREDO-Kyoto low, intermediate, and high thrombotic and bleeding risk groups are shown in Supplementary materials 3 and 4 showing that patients were often categorized in discordant risk groups. For instance, 127 (11.6%) patients were determined to have high PARIS and low CREDO-Kyoto thrombotic risks (Supplementary material 3). Similarly, 175 (15.9%) had high PARIS and low CREDO-Kyoto bleeding risks (Supplementary material 4). Concordant risk categories between the PARIS and CREDO-Kyoto risk scores were found in 373 (34.0%) and 367 (33.4%) patients in thrombotic and bleeding risk scores, respectively.

Table 1.

Baseline Characteristics.

Variable	All (n=1,098)
Age (years)	67.4±12.1
Men	843 (76.8%)
Body mass index (kg/m2)	24.2±3.9
Hypertension	743 (67.7%)
Diabetes mellitus	409 (37.2%)
Dyslipidemia	673 (61.3%)
Current smoker	365 (33.2%)
Prior MI	73 (6.6%)
Prior PCI	107 (9.7%)
Prior CABG	19 (1.7%)
Prior heart failure	29 (2.6%)
Atrial fibrillation	71 (6.5%)
Peripheral artery disease	30 (2.7%)
Malignancy	60 (5.5%)
eGFR (mL/min/1.73 m2)	63.2±24.3
Hemoglobin (g/dL)	13.7±2.2
Platelet count (×109/L)	22.2±7.6
Total cholesterol (mg/dL)	193.1±49.5
HDL cholesterol (mg/dL)	50.2±14.6
LDL cholesterol (mg/dL)	119.1±41.6
Non-fasting triglyceride (mg/dL)	147.4±141.4
STEMI presentation	745 (67.9%)
Killip class IV on admission	198 (18.0%)
Cardiac arrest on admission	134 (12.2%)
Culprit lesion in the LMT/LAD	556 (50.6%)
Chronic total occlusion	28 (2.6%)
Intravascular ultrasound	1,062 (96.7%)
Mode of intervention
Drug-eluting stent	999 (91.0%)
Bare metal stent	40 (3.6%)
Plain old balloon angioplasty	47 (4.3%)
Drug-coated balloon	4 (0.4%)
Aspiration only	7 (0.6%)

One patient received none of above interventional procedures (guidewire crossing only). CABG: coronary artery bypass grafting, eGFR: estimated glomerular filtration rate, HDL: high-density lipoprotein, LAD: left anterior descending artery, LDL: low-density lipoprotein, LMT: left main trunk, MI: myocardial infarction, PCI: percutaneous coronary intervention, STEMI: ST-segment elevation myocardial infarction

During the mean follow-up period of 647 days, 185 (16.8%) and 109 (9.9%) patients experienced ischemic and major bleeding events, respectively (Table 2). At 2 years after primary PCI, ischemic and major bleeding events occurred in 17.3% and 10.2% of patients, respectively. The incidence of ischemic and bleeding events at two years across the low, intermediate, and high PARIS and CREDO-Kyoto thrombotic and bleeding risk groups are displayed in Figure 1, 2. The rates of ischemic and major bleeding events progressively increased with the increase in risk categories in the two risk scoring systems (Fig. 1, 2). The PARIS (AUC: 0.59, p＜0.001) and CREDO-Kyoto (AUC: 0.63, p＜0.001) thrombotic risk scores significantly predicted ischemic events, and the PARIS (AUC: 0.65, p＜0.001) and CREDO-Kyoto (AUC: 0.63, p＜0.001) bleeding risk scores were also predictive for bleeding events. In the ROC curve analysis, the addition of CREDO-Kyoto thrombotic and bleeding risk scores to the PARIS scores significantly improved the diagnostic ability in predicting both ischemic (AUC: 0.63, p=0.01) and bleeding events (AUC: 0.68, p=0.01). In contrast, however, the addition of the PARIS thrombotic risk score to the CREDO-Kyoto score did not significantly improve the diagnostic ability in predicting ischemic events (p=0.60), and the addition of the PARIS bleeding risk score was associated with better diagnostic ability in bleeding outcomes with marginal statistical significance than the CREDO-Kyoto risk score alone (p=0.055).

Table 2.

Clinical Outcomes.

Variable	All (n=1,098)
Ischemic events	185 (16.8%)
Cardiovascular death	123 (11.2%)
Recurrent MI	43 (3.9%)
Ischemic stroke	37 (3.4%)
Major bleeding events	109 (9.9%)
BARC 3	99 (9.0%)
BARC 5	10 (0.9%)

BARC: Bleeding Academic Research Consortium, MI: myocardial infarction

Figure 1.

Incidence of ischemic events at two years. Ischemic events included cardiovascular death, recurrent myocardial infarction, and ischemic stroke. The event rates were estimated using Kaplan-Meier methods.

Figure 2.

Incidence of major bleeding events at two years. Bleeding events included Bleeding Academic Research Consortium type 3 or 5. The event rates were estimated using Kaplan-Meier methods.

Beyond the two risk scores, Cox proportional-hazard analyses identified an impaired renal function, low hemoglobin level and platelet count, dyslipidemia, Killip class IV, and culprit at the left main trunk or left anterior descending artery as factors associated with ischemic events (Supplementary material 5) and an impaired renal function, low hemoglobin level, and Killip class IV as factors associated with bleeding events (Supplementary material 6) in the present study.

Discussion

The present study tested the PARIS and CREDO-Kyoto risk scores in patients with acute MI undergoing primary PCI, both of which were developed from PCI populations to evaluate ischemic and bleeding risks individually. The two risk scoring systems were predictive for clinical outcomes but were often discordant. Approximately only one-third of patients were determined to have concordant risk categories. The risks of ischemic and major bleeding events progressively increased with the increase in risk categories in the PARIS and CREDO-Kyoto risk scores. Furthermore, combinations of the two risk scores had better diagnostic ability than either one alone. The present study confirmed that a risk prediction model does not necessarily fit different populations and that multiple risk assessments may be useful in stratifying patient risks.

Objective scoring systems are essential for evaluating ischemic and bleeding risks in patients undergoing PCI, as recommended in the current guidelines (6,7). In this context, several risk-predicting models have been developed, among which the PARIS and CREDO-Kyoto risk scores are unique in assessing ischemic and bleeding risks individually. As shown in our previous report and the present study (10), the PARIS and CREDO-Kyoto thrombotic and bleeding risk scores were predictive for ischemic and bleeding events. Although both risk models were derived from large-scale PCI cohorts, only a few similar items were included. In the thrombotic risk scores, only diabetes and renal impairment were concordant in the two risk scores among six and eight items of these thrombotic risk scores. Similarly, renal impairment was the only factor included in both the PARIS and CREDO-Kyoto bleeding risk scores. Thus, it is conceivable that the two scoring systems evaluated patient risks differently.

Given the modest to moderate diagnostic ability of the PARIS and CREDO-Kyoto risk scores for ischemic and bleeding events (i.e. AUC: 0.59-0.65) and the fact that the risk assessment using these two scores was often discordant, risk stratification by a single scoring model may not be sufficient in clinical practice. For instance, the rate of ischemic events at 2 years in patients with high PARIS and high CREDO-Kyoto thrombotic risk scores was 37.6%, while ischemic events occurred in 9.7% in those with low risk scores in both systems. Even in the high PARIS thrombotic risk group, the rates of ischemic events at 2 years progressively increased from low to intermediate to high CREDO-Kyoto thrombotic risk groups (17.4% vs. 28.1% vs. 37.6%), reinforcing the possibility that patients in a risk category determined by a scoring system can be further stratified by another risk-predicting model.

Recently, patient-tailored therapeutic strategies have been of clinical interest in patients undergoing PCI (e.g. antithrombotic therapy) (20). Therefore, a combination of risk scoring models may be useful in stratifying patient risks in some cases. In patients with extremely high ischemic and/or bleeding risks, dedicated PCI techniques and tailored antithrombotic regimens potentially improve clinical outcomes (20-22).

In the multivariable analyses in the present study, an impaired renal function and low hemoglobin level were identified as predictors of ischemic events, as listed in the PARIS and CREDO-Kyoto thrombotic risk scores, while a low platelet count, Killip class IV, and culprit at left main trunk or left anterior descending artery were also suggested. A previous study indicated that an abnormal platelet count influenced ischemic outcomes after PCI in patients with acute MI (23), and Killip class IV, which may be closely associated with culprit at the left main trunk or left anterior descending artery, was known to be one of the strongest predictors of ischemic events after acute MI (24).

In terms of bleeding events, an impaired renal function and low hemoglobin level were identified as predictors in the present study, as listed in the PARIS and CREDO-Kyoto bleeding risk scores. Killip class IV was also indicated as a factor associated with major bleeding, as shown in a previous report (25). Taken together, these findings confirm that there are numerous variables that may be incorporated to predict clinical outcomes, so a comprehensive risk prediction assessment rather than a single simple risk stratification model may be warranted.

In daily practice, the use of such scoring systems may be complex and time-consuming, even with a single risk score. Thus, the combined use of two scoring systems may be challenging. However, among patients with an intermediate risk according to a scoring system, for example, another risk score would aid in risk stratification. A dedicated smartphone application and/or effective artificial intelligence systems may be a future direction for risk stratification for the sake of clinical use (26).

Study limitations

Several limitations associated with the present study warrant mention. This study was performed in a retrospective manner, and decisions concerning medical treatment, including antithrombotic therapy, were left to the treating physicians. Because the PARIS and CREDO-Kyoto risk scores are able to evaluate ischemic and bleeding risks individually, we tested the scoring systems in the present study. However, the impact of different risk prediction models (e.g. Academic Research Consortium criteria for high bleeding risk) is unknown (27,28).

Conclusions

The combination of the PARIS and CREDO-Kyoto thrombotic and bleeding risk scores was associated with improved discrimination in ischemic and bleeding risks in patients with acute MI undergoing primary PCI, suggesting that further risk stratification using multiple scoring system may be useful in some cases.

The authors state that they have no Conflict of Interest (COI).

Supplementary Material

Table S1

PARIS and CREDO-Kyoto thrombotic and bleeding risk scores

Table S2

Medications at discharge

Table S3

Distribution of PARIS and CREDO-Kyoto thrombotic risk score categories

Table S4

Distribution of PARIS and CREDO-Kyoto bleeding risk score categories

Table S5

Multivariable analysis for ischemic events

Table S6

Multivariable analysis for bleeding events