Construction and Validation of a CYP2C19-Related Genetic Marker-Based Risk Prediction Model for Recurrent Angina After Percutaneous Coronary Intervention in Elderly Patients with STEMI

JIA Gaopeng¹, CHEN Qiuyu²*

¹Geriatric Medical Center, Affiliated Hospital of Inner Mongolia Medical University, Hohhot 010050, China
²Department of Hematology, Affiliated Hospital of Inner Mongolia Medical University, Hohhot 010050, China

Corresponding author: CHEN Qiuyu, Attending physician; E-mail: 624774855@qq.com

Abstract

Background: Acute ST-segment elevation myocardial infarction (STEMI) has high mortality and disability rates. Percutaneous coronary intervention (PCI) is an important revascularization method that can improve prognosis. However, some patients experience recurrent angina after PCI, which affects their quality of life and long-term outcomes. Drug-metabolizing enzyme gene polymorphisms influence drug efficacy and adverse reactions. Cytochrome P450 2C19 (CYP2C19) is involved in the metabolism of multiple drugs, and its gene polymorphisms can alter enzyme activity and affect drug metabolism. The correlation between different CYP2C19 metabolic levels and recurrent angina after PCI in STEMI patients warrants investigation.

Objective: To investigate the correlation between different CYP2C19 metabolic levels and recurrent angina after PCI in STEMI patients.

Methods: A total of 128 patients who underwent emergency PCI for acute coronary occlusion at the Chest Pain Center of the First Affiliated Hospital of Inner Mongolia Medical University in 2022 were selected as study subjects. Patient medical records and CYP2C19 gene test results were collected. Follow-up was conducted via telephone or outpatient visits at 1, 3, 6, and 12 months after PCI, ending on December 31, 2023. The endpoint event was angina attack. Lasso regression analysis was used to screen variables related to angina attacks, followed by construction of a predictive model using multivariate logistic regression analysis and development of a nomogram. Bootstrap resampling was used for internal model validation. The training and validation sets were evaluated using receiver operating characteristic (ROC) curves, goodness-of-fit tests, calibration curves, and decision curve analysis (DCA) to construct a risk prediction model for recurrent angina after PCI in elderly STEMI patients.

Results: A total of 128 patients were included, with 92 males (71.9%) and 36 females (28.1%), and a median age of 63.5 (61.0, 66.0) years. During follow-up, 45 patients (35.2%) experienced recurrent angina, while 83 patients (64.8%) did not. There were statistically significant differences in gender, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and CYP2C19 genotype between patients with and without recurrent angina (P<0.05). Lasso regression analysis identified 7 independent predictive variables, including gender, LDL-C, HDL-C, homocysteine (Hcy), apolipoprotein B (ApoB), D-dimer, and CYP2C19 genotype. Multivariate logistic regression analysis showed that female gender (OR=3.4929, 95%CI=1.2888-15.0662), elevated LDL-C (OR=3.1237, 95%CI=1.6859-6.3484), and elevated Hcy (OR=1.0614, 95%CI=1.0288-1.1036) were risk factors for recurrent angina after STEMI intervention, while elevated HDL-C (OR=0.0167, 95%CI=0.0009-0.2091), intermediate CYP2C19 metabolism (OR=0.2734, 95%CI=0.0747-0.9237), and normal CYP2C19 metabolism (OR=0.0867, 95%CI=0.0255-0.2561) were protective factors. The model was internally validated using Bootstrap resampling with 1,000 replications, and the Hosmer-Lemeshow calibration curve showed good model fit. ROC curves were plotted for the training and validation sets, with areas under the ROC curve (AUC) of 0.869 (95%CI=0.796-0.943) and 0.789 (95%CI=0.701-0.877), respectively, indicating good discrimination in both populations. Further DCA showed that the model had good clinical utility.

Conclusion: Intermediate and normal CYP2C19 metabolic types are protective factors against recurrent angina after STEMI intervention. This study established a risk prediction model for recurrent angina that includes five clinical indicators: female gender, LDL-C, Hcy, HDL-C, and CYP2C19. The model can be used to predict the risk of recurrent angina in patients for early screening and has good fit, discrimination, and clinical application value.

Keywords: ST elevation myocardial infarction; Angina pectoris; CYP2C19; Percutaneous coronary intervention; Prediction model; Nomogram

Introduction

Acute myocardial infarction (AMI) is a major cardiovascular disease with high mortality and disability rates worldwide. In China, the incidence of AMI has been rising annually with the acceleration of population aging. Although AMI onset is gradually becoming younger, the formulation of prevention and postoperative personalized plans is particularly important. Despite percutaneous coronary intervention (PCI) significantly improving prognosis in patients with ST-segment elevation myocardial infarction (STEMI), recurrent angina after PCI remains a key issue affecting long-term survival and quality of life.

Recent advances in medical technology and data analysis have made the construction and validation of genetic marker-based risk prediction models a research hotspot. These models identify key prognostic factors by analyzing clinical, laboratory, and imaging data, thereby reducing adverse events. The role of the CYP2C19 gene in clopidogrel metabolism has been extensively studied, as CYP2C19 gene polymorphisms affect clopidogrel metabolism speed and consequently its antiplatelet effects. Fast metabolizers have normal clopidogrel metabolism speed, maintaining stable levels of active metabolites and achieving expected antiplatelet effects. Intermediate metabolizers have slower metabolism, reduced active metabolite generation, weakened antiplatelet effects, and potentially increased cardiovascular risk. Slow metabolizers have very slow metabolism, minimal active metabolite generation, nearly lost antiplatelet effects, and significantly increased cardiovascular risk. Different metabolic levels lead to significant variations in patient response to clopidogrel, thereby affecting prognosis after PCI in AMI patients.

Existing prediction models for recurrent angina after PCI have limitations in genetic considerations, personalization, prediction accuracy, and dynamic assessment. Therefore, it is necessary to construct more comprehensive and precise risk prediction models to better guide clinical decision-making and improve long-term patient outcomes. This study proposes a risk prediction model based on CYP2C19-related genetic markers to address these limitations. By retrospectively analyzing clinical data from STEMI patients, screening independent predictors, and establishing a Lasso-Logistic regression model, we aim to precisely assess patients' risk of recurrent angina. This study compensates for deficiencies in existing research regarding personalized risk assessment combining genetic markers and clinical indicators, providing new insights and tools for long-term prognosis management in elderly STEMI patients with important clinical practice value.

Methods

1.1 Clinical Data

This retrospective study selected 128 patients who underwent emergency PCI for acute coronary occlusion at the Chest Pain Center of the First Affiliated Hospital of Inner Mongolia Medical University in 2022. The study was approved by the Ethics Committee of the Affiliated Hospital of Inner Mongolia Medical University (approval number: YKD2024021019), and informed consent was waived for this retrospective study.

Inclusion criteria: (1) Admission diagnosis of STEMI, age >60 years; (2) Coronary angiography showing acute single-vessel occlusion; (3) Ability to adhere to oral antiplatelet medications such as aspirin and ticagrelor; (4) Post-PCI culprit vessel flow restored to TIMI grade 3.

Exclusion criteria: (1) Coronary lesions: non-native large vessel disease, left main disease, chronic total occlusion, severe calcification, graft vessel disease, or multi-vessel disease; (2) Previous history of coronary heart disease; (3) History of coronary artery bypass grafting or previous intervention in the target vessel; (4) >50% stenosis in coronary arteries other than the culprit vessel; (5) Expected survival <1 year; (6) Procedure failure; (7) Poor compliance with follow-up.

1.2 Research Methods

1.2.1 Perioperative PCI Management: Procedures were performed by operators with coronary intervention training certification. All STEMI patients received oral aspirin 300 mg loading dose, clopidogrel 300 mg loading dose (or ticagrelor 180 mg loading dose), followed by maintenance oral therapy, and atorvastatin 20 mg/d. Preoperative examinations were completed urgently, including blood routine, urine routine, stool routine, coagulation function, liver and kidney function, electrocardiogram, and echocardiography. Patients and their families were informed of procedural risks and complications, and informed consent was obtained.

1.2.2 CYP2C19 Gene Detection: (1) Sample collection: EDTA-anticoagulated whole blood (purple tube) 2 mL was collected for genomic DNA extraction and testing. Samples should be sent for testing as soon as possible to ensure DNA quality and integrity. (2) Detection sites: including CYP2C192 (c.681G>A), CYP2C193 (c.636G>A), and CYP2C1917 (c.-806C>T). (3) Result interpretation: Based on melting curve Tm value changes or gene chip hybridization signal strength, CYP2C19 gene polymorphism types can be determined. CYP2C191/1 is normal metabolizer; CYP2C192/2 or CYP2C192/3 is poor metabolizer, indicating slowed metabolism; CYP2C191/2 or CYP2C191/*3 is intermediate metabolizer, indicating metabolism speed between normal and slow.

1.3 Data Collection

Modeling data included basic patient information, laboratory tests, and examination indicators. Basic information included age, gender, marital status, smoking (defined as regular smoking within the past year), alcohol consumption (defined as >3 times/week within the past year, average amount >50 mL/time), and history of hypertension and diabetes. Laboratory indicators included blood lipids, homocysteine (Hcy), apolipoproteins (Apo), cardiac enzymes, troponin, D-dimer, international normalized ratio (INR), N-terminal pro-brain natriuretic peptide (NT-proBNP), and liver function indicators. Examination indicators included echocardiography and angiography results such as target lesion location and lesion length. Data were entered into R 4.4.1 after collation. Missing values <10% were handled using multiple imputation based on the mice package, while data with missing values >10% were excluded. Bootstrap resampling was used for validation. CYP2C19 metabolic levels were categorized as normal, intermediate, and poor metabolizers for comparison, with normal vs. intermediate recorded as 1, intermediate vs. poor as 2, and normal vs. poor as 3, which were then included as multicategorical variables in the dataset for statistics.

1.4 Follow-up

Patients were followed up via telephone or outpatient visits at 1, 3, 6, and 12 months after PCI, ending on December 31, 2023. The endpoint event was angina attack, characterized by compressive pain or chest tightness in the precordial area or behind the sternum, possibly accompanied by palpitations, fatigue, and profuse sweating. Pain was mainly located behind the sternum and could radiate to the precordial area, left upper limb, neck, and jaw. Angina typically occurred during exertion or emotional excitement, lasting several minutes, with symptoms rapidly relieved by rest or nitrates.

1.5 Statistical Methods

Data analysis was performed using SPSS 26.0 and R 4.4.1 software. Normally distributed continuous data were expressed as (x̄±s) and compared between groups using independent samples t-test; skewed distributed continuous data were expressed as M(P25, P75) and compared using Wilcoxon test; categorical data were expressed as n(%) and compared using χ² test or Fisher's exact test. Lasso regression analysis was used to screen variables related to angina attack events, followed by multivariate logistic regression analysis to construct a prediction model and draw a nomogram. Bootstrap was used for internal model validation. Training and validation set models were evaluated using receiver operating characteristic (ROC) curves, goodness-of-fit tests, calibration curves, and decision curve analysis (DCA). P<0.05 was considered statistically significant.

Results

2.1 Patient Baseline Data

A total of 128 patients were included, with 92 males (71.9%) and 36 females (28.1%), and a median age of 63.5 (61.0, 66.0) years. During follow-up, 45 patients (35.2%) experienced recurrent angina, while 83 patients (64.8%) did not. There were statistically significant differences in gender, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and CYP2C19 genotype between patients with and without recurrent angina (P<0.05). No statistically significant differences were found between groups in age, smoking, alcohol consumption, marital status, lesion location, lesion length, total cholesterol, triglycerides, Hcy, complement C1q, ApoA, ApoB, creatine kinase-MB, creatinine, D-dimer, hypersensitive troponin T, NT-proBNP, INR, fasting glucose, uric acid, aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transferase, left ventricular ejection fraction, PCI methods, or medication history (P>0.05), see Table 1 [TABLE:1].

2.2 Variable Screening by Lasso Regression

Lasso regression analysis was performed on selected variables. Vertical lines were drawn at the minimum criteria and 1 standard error of the minimum criteria at optimal values. Based on 10-fold cross-validation, the optimal model was at λ1se=0.0710, log(λ)=-2.6450, screening out 7 independent predictor variables including gender, LDL-C, HDL-C, Hcy, ApoB, D-dimer, and CYP2C19 genotype, see Figure 1 [FIGURE:1] and Figure 2 [FIGURE:2].

2.3 Prediction Model Construction

Using recurrent angina (assignment: no=0, yes=1) as the dependent variable and the 7 predictor variables screened by Lasso regression analysis as independent variables, multivariate logistic regression analysis identified 5 predictive variables. Female gender (OR=3.4929, 95%CI=1.2888-15.0662), elevated LDL-C (OR=3.1237, 95%CI=1.6859-6.3484), and elevated Hcy (OR=1.0614, 95%CI=1.0288-1.1036) were risk factors for recurrent angina after STEMI intervention, while elevated HDL-C (OR=0.0167, 95%CI=0.0009-0.2091), intermediate CYP2C19 metabolism (OR=0.2734, 95%CI=0.0747-0.9237), and normal CYP2C19 metabolism (OR=0.0867, 95%CI=0.0255-0.2561) were protective factors, see Table 2 [TABLE:2].

2.4 Nomogram Construction

A dynamic nomogram was constructed to display the influence of Hcy, LDL-C, gender, HDL-C, and CYP2C19 on total points. The horizontal axis represents point values for each variable, while the vertical axis represents cumulative total points. Red dots mark specific thresholds (0.493), with total points exceeding this threshold considered high risk. Hcy point values range from 0-100, showing its large influence range; LDL-C variable point values range from 0-50, showing moderate contribution; gender variable box plots show point distribution differences between males and females with outliers marked; HDL-C variable point values show concentrated distribution with small variation range; CYP2C19 variable box plots show point distribution across different genotypes with red dots marking significant effects of specific genotypes. The total points axis at the chart bottom shows overall point distribution after considering all variables, see Figure 3 [FIGURE:3].

2.5 Prediction Model Performance Evaluation

The Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI) indices were used to evaluate the new model (including CYP2C19 genotype) compared with the baseline model (without CYP2C19 genotype). For categorical variables, NRI=0.1759 (95%CI=0.0075-0.3443, P=0.041), indicating significantly improved classification performance; for continuous variables, NRI=0.7550 (95%CI=0.4175-1.0926, P<0.001), indicating significantly improved prediction performance; IDI=0.1458 (95%CI=0.0793-0.2124, P<0.001), indicating significantly improved discrimination ability.

Bootstrap resampling with 1,000 replications was used for internal validation. The Hosmer-Lemeshow calibration curve showed good model fit, see Figure 4 [FIGURE:4]. ROC curves were plotted for training and validation sets with AUCs of 0.869 (95%CI=0.796-0.943) and 0.789 (95%CI=0.701-0.877), respectively, indicating good discrimination in both populations, see Figure 5 [FIGURE:5]. Further DCA showed good clinical utility, see Figure 6 [FIGURE:6].

Discussion

Major adverse cardiovascular events (MACE) remain serious adverse outcomes in AMI patients, with previous studies reporting MACE incidence of approximately 4.2%-51%. In recent years, AMI case numbers have increased annually with a trend toward younger onset, while elderly patients have more comorbidities and higher PCI treatment risks, with higher postoperative recurrent angina risk. WANG et al. conducted a comparative study of myocardial infarction in young and elderly patients, including 114 young AMI patients (≤42 years) and 179 elderly AMI patients (≥60 years), analyzing baseline data, clinical manifestations, and coronary angiography results. The study showed young AMI patients had higher male proportion than elderly patients (94.7% vs. 64.2%, P<0.05), with higher rates of smoking history and positive family history but lower rates of hypertension and diabetes. Elderly AMI patients were more likely to have various clinical manifestations and multi-branch lesions, while young patients had fewer symptoms and more limited tissue lesions. Clinical manifestations of AMI differ between young and elderly patients. This study constructed a risk model for recurrent angina after PCI in elderly AMI patients based on multiple indicators and validated it. Results showed that female gender, LDL-C, Hcy, HDL-C, and CYP2C19 genotype had high value in identifying recurrent angina after STEMI.

Female patients had higher recurrent angina risk than males, possibly due to reduced estrogen (a vascular protective factor) after menopause, which requires further clinical research confirmation. LDL-C was an independent risk factor and HDL-C an independent protective factor for recurrent angina, consistent with conventional clinical understanding. The core component of coronary plaque formation is lipid plaque. The 2019 European Society of Cardiology (ESC)/European Atherosclerosis Society (EAS) guidelines emphasize that coronary heart disease progression risk is positively correlated with blood lipid levels, especially LDL-C. HDL-C can carry cholesterol from peripheral tissues and convert it to bile acids or directly excrete it through the intestine, reducing total cholesterol levels and readmission risk. LDL-C has the opposite effect. In early atherosclerosis, endothelial dysfunction and monocyte-macrophage phagocytosis of oxidized LDL-C to form foam cells are core aspects of atherosclerosis. WU et al. conducted a nested case-control study on blood lipids and coronary heart disease including 3,438 cases (1,719 pairs) of CHD patients, comprising 1,084 cases (542 pairs) of white Americans, 1,244 cases (622 pairs) of black Americans, and 1,110 cases (555 pairs) of Chinese adults. The study emphasized that several lipoprotein biomarkers, including ApoB/ApoA1, ApoB, and LDL-C cholesterol, were closely related to CHD occurrence. Hcy is a predictor of cardiovascular and cerebrovascular diseases and can also reflect early cardiac damage. WU et al. conducted a study on the diagnostic value of Hcy and lipoprotein-associated phospholipase A2 (Lp-PLA2) levels in coronary heart disease, including 232 patients in a retrospective study divided into AMI, unstable angina, and stable angina groups, with 75 healthy adults as controls. Univariate and multivariate logistic regression analysis showed that Hcy, Lp-PLA2, hypertension, and hyperlipidemia were important risk factors for coronary heart disease.

CYP2C19 enzyme plays an important role in drug metabolism, particularly for clopidogrel. Clopidogrel is a prodrug requiring CYP2C19 enzyme conversion to active metabolites to exert antiplatelet effects. Therefore, CYP2C19 gene polymorphisms significantly affect clopidogrel efficacy. In contrast, ticagrelor is a new antiplatelet drug whose activity does not depend on CYP2C19 metabolism and is thus unaffected by CYP2C19 gene polymorphisms.

The CYP2C19 gene encodes cytochrome P450 family proteins, primarily present in the liver, involved in metabolism of various important drugs including proton pump inhibitors, antidepressants, antiepileptics, and antiplatelet drugs such as clopidogrel. CYP2C19 gene variants can significantly affect enzyme activity, leading to different metabolic phenotypes. CYP2C19 gene typing differences can influence drug efficacy and safety, holding important significance in personalized medicine by allowing adjustment of drug treatment plans based on individual genetic backgrounds. This study showed that patients with high metabolic level CYP2C19 genotypes had reduced recurrent angina risk. ZHANG et al. conducted a study on the relationship between CYP2C192/CYP2C193 polymorphisms and coronary heart disease development, evaluating their impact on adverse clinical events. The study included 231 PCI patients with genotyping for CYP2C192 and CYP2C193, followed up for 14 months. Results showed CYP2C192 carriers had significantly higher cardiovascular event incidence than non-carriers (21.6% vs. 6.3%, P=0.019). Cox proportional hazard model indicated CYP2C192 was an independent predictor of MACE (OR=3.65, 95%CI=1.09-12.25, P=0.036). CYP2C192 polymorphism increased coronary heart disease and MACE risk, while CYP2C193 did not show the same effect in Chinese Han population.

This study has several limitations, including single-center design, retrospective methodology, and relatively small sample size, which may introduce confounding factors. Additionally, this study lacks external validation. To further confirm the reliability of conclusions, future prospective, multi-center studies are needed. In summary, female gender, LDL-C, Hcy, HDL-C, and CYP2C19 genotype are key factors affecting 1-year recurrent angina after PCI in STEMI patients. The prediction model constructed based on these factors shows high predictive efficacy and may assist in clinical decision-making and early intervention.

Author Contributions

JIA Gaopeng was responsible for data collection and manuscript writing; CHEN Qiuyu was responsible for statistical analysis guidance and manuscript writing guidance.

Conflicts of Interest: None declared.

References

[1] POUDEL I, TEJPAL C, RASHID H, et al. Major adverse cardiovascular events: an inevitable outcome of ST-elevation myocardial infarction? a literature review[J]. Cureus, 2019, 11(7): e5280. DOI: 10.7759/cureus.5280.

[2] WANG J, LI L, MA N, et al. Clinical investigation of acute myocardial infarction according to age subsets[J]. Exp Ther Med, 2020, 20(5): 120. DOI: 10.3892/etm.2020.9248.

[3] 苏崇弘, 赵铁夫, 马涵英, 等. 女性冠心病危险因素的研究进展[J]. 心肺血管病杂志, 2024, 43(4): 424-428.

[4] 王家琦, 陈晓敏. 血脂亚组分检测在动脉粥样硬化性心血管疾病诊疗中的研究进展[J]. 心电与循环, 2024(1): 88-93.

[5] VOGT A, WEINGÄRTNER O. Management of dyslipidaemias: the new 2019 ESC/EAS-guideline[J]. Dtsch Med Wochenschr, 2021, 146(2): 75-84. DOI: 10.1055/a-1199-8193.

[6] AJALA O N, DEMLER O V, LIU Y Y, et al. Anti-inflammatory HDL function, incident cardiovascular events, and mortality: a secondary analysis of the JUPITER randomized clinical trial[J]. J Am Heart Assoc, 2020, 9(17): e016507. DOI: 10.1161/JAHA.119.016507.

[7] LIU Y X, DU L Q, LI L, et al. Development and validation of a machine learning-based readmission risk prediction model for non-ST elevation myocardial infarction patients after percutaneous coronary intervention[J]. Sci Rep, 2024, 14(1): 13393. DOI: 10.1038/s41598-024-64048-x.

[8] WU D F, YIN R X, DENG J L. Homocysteine, hyperhomocysteinemia, and H-type hypertension[J]. Eur J Prev Cardiol, 2024, 31(9): 1092-1103. DOI: 10.1093/eurjpc/zwae022.

[9] WU L L, SHAO P, GAO Z Y, et al. Homocysteine and lp-PLA2 levels: diagnostic value in coronary heart disease[J]. Medicine (Baltimore), 2023, 102(46): e35982. DOI: 10.1097/MD.0000000000035982.

[10] LEE C R, LUZUM J A, SANGKUHL K, et al. Clinical pharmacogenetics implementation consortium guideline for CYP2C19 genotype and clopidogrel therapy: 2022 update[J]. Clin Pharmacol Ther, 2022, 112(5): 959-967. DOI: 10.1002/cpt.2526.

[11] ZHANG Y Y, ZHOU X, JI W J, et al. Association between CYP2C192/3 polymorphisms and coronary heart disease[J]. Curr Med Sci, 2019, 39(1): 44-51. DOI: 10.1007/s11596-019-2010-1.

(Received date: 2025-02-14; Revised date: 2025-06-03)
(Editor: ZOU Lin)