1. Participants and clinical characteristics
Hospitalized patients with CAD were consecutively recruited from August 2018 to May 2022 in the Department of Cardiology, General Hospital of the Chinese People’s Liberation Army. Patients were eligible for inclusion in the current study if they were older than 65 years and were discharged with more than 2 prescriptions. Exclusion criteria included patients with severe diseases and expected life expectancy < 12 months, severe hepatic and renal insufficiency (Child-Pugh class C liver disease or K/DOQI Stage 5 chronic kidney disease), history of organ transplant, invasive solid tumours, or hematologic malignancies. Patient demographic information and medical history were collected. Comorbidities were further subdivided into cardiovascular and non-cardiovascular conditions. Pre-existing cardiovascular multimorbidity includes heart failure, hypertension, diabetes mellitus, stroke, and atrial fibrillation. Pre-existing non-cardiovascular multimorbidity includes chronic obstructive pulmonary disease, hepatic and renal insufficiency, peptic ulcer disease, connective tissue disease, dementia and any canner. Baseline use of drugs for cardiovascular conditions was collected. This included antiplatelet agents, oral anticoagulation agents, lipid-lowering agents, beta-blockers, calcium channel blockers, diuretics, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and oral antidiabetic agents. Additionally, the use of proton pump inhibitors and psychotropic agents was collected at baseline, but no information on other drugs was available. Participants were followed up for incident events for at least 1 year. The study complied with the Declaration of Helsinki and was approved by the ethics committee of the People’s Liberation Army General Hospital. All patients signed informed consent forms.
2. Identification of actionable genetic variants
Participants were genotyped using the MassARRY platform (Agena Bioscience, San Diego, CA, USA), with the inclusion of the following actionable pharmacogenetic variants including CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), CYP2C19*2 (rs4244285), CYP2C19*3 (rs4986893), CYP2C19*17 (rs12248560), CYP2D6*10 (rs1065852), CYP3A5*3 (rs776746), and SLCO1B1*5 (rs4149056), all of which had the clinical annotation from Pharmacogenetics and Pharmacogenomics Knowledge Base (PharmGKB) and/or implementation guidelines from the Clinical Pharmacogenetics Implementation Consortium (CPIC) for the cardiovascular drugs [15] [16]. Allele frequency was calculated and compared with the frequency in the East Asian population reported in the 1,000 Genomes Project [17].
3. Assessment of drug-gene interactions and drug-drug interactions
Drug-gene interaction (DGI) is defined as an individual’s genetic phenotype that affects the patient’s ability to clear a drug [18]. For pharmacokinetic DGIs, substrate drugs were available in the clinical annotations from the PharmGKB for the above genotyped pharmacogenes (Table S1). we identified and counted the total number of unique DGIs, and the numbers of interactions per patient and gene were determined. Information about DGIs of clinical recommendations was collected from corresponding CPIC [16] or Dutch Pharmacogenetics Working Group (DPWG) guidelines [19]. DGIs not mentioned in pharmacogenomics guidelines were considered to need to be applied under clinician monitoring.
Drug-drug interactions (DDIs) were defined as the individual’s regimen affecting that individual’s ability to clear another drug [18]. For pharmacokinetic DDIs, substrates, inhibitors and inducers were available from Flockhart Table [20]. The drugs involved in this study are summarized in Table S2.
4. Clinical outcomes
The primary outcome was defined as all-cause mortality. The secondary outcome was defined as the major adverse cardiovascular events (MACEs) including a composite of cardiovascular mortality, nonfatal myocardial infarction, stent thrombosis, nonfatal stroke, and unplanned revascularization. Drug response phenotypes defined as “drug-stop” or “dose-decrease” were also evaluated. “Drug-stop” after only one prescription was considered to avoid ADRs, lack of therapeutic efficacy, or both, while “dose-decrease” similarly indicated ADRs-related intolerance or extreme efficacy. There may be other reasons why people stop their drugs that may not reflect drug efficacy or tolerance, such as being unable to obtain the medication prescription from local pharmacies or not following medical advice. We excluded these conditions when counting the drug response phenotypes.
5. Statistical analysis
Continuous variables with a normal distribution were summarized as the mean ± standard deviation. Categorical clinical variables were presented as group numbers (percentages). Integer count variables were presented as median [interquartile range (IQR) /range] for numbers of comorbidities, drugs and DGIs. Multivariable Cox regression was applied to test whether the number of DGIs was associated with efficacy outcomes, and drug response phenotypes during follow-ups. Efficacy outcomes based on time to the first event were evaluated by comparing Kaplan‒Meier-based cumulative incidence rates with the log-rank test. In sensitivity analyses, the relationship between the number of DGIs and clinical outcomes within the follow-up interval of 1 year, 2 years, 3 years and all relevant analyses were analyzed. All multivariate analyses were adjusted for age, sex, number of comorbidities, coexisting diseases (myocardial infarction, heart failure, hepatic and renal insufficiency), number of drugs, and number of DDIs. All data analyses were conducted with SPSS 26.0 (IBM, Armonk, NY, USA). The P value was used to test for the significance threshold (P < 0.05).