In addition to aspirin, clopidogrel or prasugrel is recommended to administer with percutaneous coronary intervention (PCI) or fibrinolytic therapy for acute coronary syndrome (ACS) patients with ST elevation myocardial infarction (STEMI) (Wiviott et al. 2007; Levine et al. 2011). Both clopidogrel and prasugrel are administered as a prodrug that is metabolized to an active metabolite by cytochrome P450 (CYP) isoenzymes. Prasugrel is a third-generation thienopyridine with a more potent antiplatelet agent and features faster onset and less dependency on CYP2C19 in converting to its active metabolite than clopidogrel does. Prasugrel maintains a higher concentration (~70%) of active metabolites with renal excretion (Mousa et al. 2010). The effects of clopidogrel are variable in patients and its resistance has been issued with related to genetic polymorphisms (Serebruany et al. 2005).

Several CYP isoenzymes mediate clopidogrel metabolism and among them CYP2C19 plays a major role in two oxidative steps of the biotransformation process (Kazui et al. 2010). More importantly, genetic variants of CYP2C19 contribute to variations in pharmacokinetics of clopidogrel. Several CYP2C19 alleles are known to lead to different metabolic phenotypes and can be classified into four types: ultra-rapid, extensive, intermediate, and poor metabolizers. For example, the conversion of clopidogrel to its active metabolite is extensively reduced in patients who carry the CYP2C19*2 allele, which leads to a poor metabolizer. Allele frequencies of these CYP2C19 alleles vary considerably by ethnicity. For example, the CYP2C19*2 allele has 27% frequency in Asians, 14% in Caucasians, and 18% in African Americans (Collet et al. 2009; Lee et al. 2009; Simon et al. 2009; Beitelshees et al. 2011). Frequency of the CYP2C19*3 allele, another poor-metabolizer type, is 9% in Asians but this type of allele is nonexistent in both Caucasians and African Americans. CYP2C19 genetic polymorphisms are also associated with pharmacodynamics and clinical outcomes. For example, the CYP2C19*2 allele is associated with decreased response to clopidogrel treatment and increased occurrence of adverse cardiovascular outcomes when treated with clopidogrel (Mega et al. 2009a).

The FDA approved a label for clopidogrel with a boxed warning regarding the reduced effectiveness of clopideogrel with less enzyme activity of CYP2C19 to convert the drug into its active metabolite. Genetic testing might be considered to identify a patient at high risk for poor clinical outcome (Levine et al. 2011). When a patient predisposed to inadequate platelet inhibition with clopidogrel is identified by genetic testing, treatment with an alternate P2Y12 inhibitor (e.g., prasugrel or ticagrelor) might be considered (Holmes et al. 2010). Different from clopidogrel which shows different responses based on the genotype, prasugrel does not show different responses according to the genotype.

Although current and emerging evidence suggest that better targeted and more effective treatment strategy utilizing pharmacogenomic knowledge have the potential to produce significant gains in patients’ health, it might be difficult to manage patients based on these scientific findings due to the budget constraint and other issues in the healthcare system. As the FDA label includes the boxed warning about the relationship between the efficacy and pharmacogenetic characteristics of clopidogrel, it becomes necessary to evaluate whether genotype testing for all the patients would be a cost-effective strategy. Especially, evaluating the efficiency of utilizing genotype information should be performed for each ethnic group because the distribution of genotypes varies by ethnicity. In case of being cost effective, the result of economic evaluation could serve as a ground for establishing the reimbursement guideline.

Several cost-effectiveness studies have been published in various countries (Reese et al. 2012; Deiman et al. 2016; Fu et al. 2020; Narasimhalu et al. 2020; AlMukdad et al. 2021; Claassens et al. 2022). The objective of this study was to evaluate the cost-effectiveness of a CYP2C19 genotype-guided antiplatelet therapy reflecting the characteristics of Asians when compared with using clopidogrel or prasugrel without considering genotypes in patients with acute coronary syndrome scheduled to undergo percutaneous coronary intervention in Korea.

Model overview

We performed a cost-effectiveness analysis utilizing a decision-analytic model with one-year timeframe. Societal perspective was employed to examine the impact of adopting treatment strategy on our society. Target population was Korean patients with acute coronary syndrome (ACS) who receive antiplatelet therapy following a scheduled percutaneous coronary intervention (PCI). A hypothetical cohort of 10,000 patients was assigned to each of the three treatment alternatives: (1) drug selection of antiplatelet therapy (i.e., clopidogrel or prasugrel) based on the genotype information of CYP2C19; (2) clopidogrel for all patients without using genotype information; (3) prasugrel for all patients without using genotype information. The pharmacogenetic-based strategy involved genotyping for the CYP2C19*1, CYP2C19*2, CYP2C19*3, CYP2C19*4, CYP2C19*5, CYP2C19*6, CYP2C19*7, CYP2C19*8, and CYP2C19*17 alleles. Under the treatment alternative 1, simulated patients determined to be carriers of reduced function CYP2C19 allele according to genetic test were prescribed prasugrel. Noncarriers of reduced function CYP2C19 allele were prescribed clopidogrel. Based on recent findings, it was assumed that genetic testing is able to identify genotypes correctly (that is, sensitivity and specificity are 100%) (Saracini et al. 2012). The decision tree was developed to simulate the natural history of patients with ACS receiving antiplatelet therapy following PCI (Fig. 1). Patients might experience one of the five cardiovascular events during the simulation period of one year: nonfatal myocardial infarction, nonfatal stroke, bleeding defined as the major or minor Thrombolysis In Myocardial Infarction (TIMI) bleeding, all-cause death including cardiovascular and non-cardiovascular death, or no event.

Figure 1. Decision tree to evaluate cost-effectiveness of using genotype information for selecting antiplatelet therapy.

Model parameters

1) Probability of clinical event

The decision model requires various probability parameters. First, proportion of Korean ACS patients who are carriers of reduced function CYP2C19 allele was obtained from two Korean studies (Lee et al. 2009; Oh et al. 2012). Second, probabilities of each cardiovascular event were derived from published literature, separately for each type of ACS patients: those taking clopidogrel or prasugrel, respectively, regardless of having genetic variants); those without genetic variants and taking clopidogrel); and those carrying genetic variants and taking prasugrel. Probability of stroke and TIMI major or minor bleeding in patients carrying variant allele and receiving prasugrel were estimated using the hazard ratios of 1.73 and 1.01 for variant allele versus nonvariant, respectively (Lee et al. 2009; Mega et al. 2009a; 2009b; Jang et al. 2012; Oh et al. 2012). We extracted the parameters from several published studies based on following criteria in order (Table 1): (1) a study with meta-analysis and systematic review in Korean or Asian patients (Jang et al. 2012); (2) a large-scale clinical study with Korean or Asian patients (Lee et al. 2009; Choi et al. 2011; Oh et al. 2012); (3) a study with meta-analysis and systematic review with Caucasian patients (Mega et al. 2010); and (4) a large-scale clinical study with Caucasians (Mega et al. 2009a; 2009b).

Table 1 . Model input parameters.

Parameter	Base value	Reference
Probabilities
Proportion of carriers of genetic variants of CYP2C19 in Korean patients with coronary artery disease	0.450	Lee et al. (2009)
		Oh et al. (2012)
Occurrence of CV events in patients without genetic variants of CYP2C19 after taking clopidogrel
- Nonfatal MI	0.048	Jang et al. (2012)
- Nonfatal stroke	0.002	Mega et al. (2009a), Mega et al. (2010)
- Bleeding (TIMI major or minor)	0.028	Mega et al. (2009a), Mega et al. (2010)
- All-cause death	0.007	Jang et al. (2012)
Occurrence of CV events in patients with genetic variants of CYP2C19 after taking prasugrel
- Nonfatal MI	0.066	Mega et al. (2009b)
- Nonfatal stroke	0.001	Mega et al. (2009b)
- Bleeding (TIMI major or minor)	0.045	Mega et al. (2009b)
- All-cause death	0.010	Mega et al. (2009b)
Occurrence of CV events in patients regardless of genetic variants after taking clopidogrel
- Nonfatal MI	0.049	Jang et al. (2012)
- Nonfatal stroke	0.004	Mega et al. (2009a)
- Bleeding (TIMI major or minor)	0.028	Mega et al. (2009a)
- All-cause death	0.010	Jang et al. (2012)
Occurrence of CV events in patients regardless of genetic variants after taking prasugrel
- Nonfatal MI	0.077	Mega et al. (2009b)
- Nonfatal stroke	0.009	Mega et al. (2009b)
- Bleeding (TIMI major or minor)	0.038	Mega et al. (2009b)
- Cardiovascular death	0.014	Mega et al. (2009b)
Costs (KRW)
Genotyping CYP2C19	63,000	Severance Hospital (2009)
Clopidogrel 75 mg/day for one year (unit price: 1,164 KRW)	424,860	The Health Insurance Review & Assessment Service (2020)
Prasugrel 10 mg/day for one year (unit price: 2,298 KRW)	838,770	The Health Insurance Review & Assessment Service (2020)
PCI for one time	3,272,820	The National Evidence-based Healthcare Collaborating Agency (2010)
Annual cost for nonfatal MI	4,942,074
- Direct medical costs	3,767,000	The National Health Insurance Corporation; Health Insurance Review & Assessment Service (2021)
- Direct non-medical costs	436,000	The Korea National Statistical Office (2020a, 2020b, 2020c)
- Indirect costs (productivity loss)	739,074	The Korea Center for Disease Control and Prevention (2005)
Annual cost for nonfatal stroke	5,151,000
- Direct medical costs	2,230,000
- Direct non-medical costs	1,290,000
- Indirect costs (productivity loss)	1,631,000
Annual cost for nonfatal bleeding per year	5,073,265	Ewen et al. (2009)
Annual cost for cardiovascular death	2,778,374
- Direct medical costs	870,699
- Direct non-medical costs	369,409
- Indirect costs (productivity loss)	1,538,265

CV, cardiovascular; MI, myocardial infarction; TIMI, Thrombolysis in Myocardial Infarction; PCI, percutaneous coronary intervention; KRW, Korean won. All costs are presented in 2020 currency Korean won (one US dollar approximately equals to 1,100 KRW)..

2) Resource use and costs

Three types of cost parameters were incorporated into the model: costs of receiving antiplatelet therapy, costs of undergoing PCI, and costs of treating each of the clinical events included in the model (Table 1). All costs consisted of direct medical and non-medical costs, and indirect costs. Direct medical costs consisted of insurance-covered and uncovered medical costs. Direct non-medical costs included transportation costs to visit hospitals and caregiver costs. Indirect costs consisted of loss of productivity due to absence from work during hospitalization and premature death. All costs were presented in 2020 Korean won (KRW; one US dollar approximately equals to 1,100 KRW) in accordance with the medical and transportation component of the Consumer Price Index and average daily wage provided by National Statistical Office (The Korean Statistical Information Service 2020a; 2020b; 2020c).

(1) Costs of antiplatelet therapy

Unlike the treatment alternatives of 2 and 3, individuals under the alternative 1 bears cost for genotyping in addition to drug therapy with clopidogrel or prasugrel. Since the genotyping test related to antiplatelet therapy is not widely adopted in Korea, we depended the cost information for the test upon a single institution, which is one of the largest tertiary hospitals in Korea (Severance Hospital 2009). The costs of genotyping for CYP2C19 are reimbursed and taking the genotyping test does not cause a noticeable discomfort to the patients. Thus, we did not consider the discomfort occurred in patients due to this genotyping test. Annual drug therapy costs were estimated by summing up drug acquisition and dispensing costs, and direct and indirect costs of outpatient visits occurred to receive prescription. Drug acquisition costs for clopidogrel and prasugrel were based on the list drug prices by the National Health Insurance (The Health Insurance Review & Assessment Service 2020). We chose the mode price for clopidogrel 75 mg among 57 marketed products including both branded and generic products. For prasugrel 10 mg, there was only one branded drug. We assumed patients would take clopidogrel 75 mg daily or prasugrel 10 mg daily for one year.

(2) Costs of PCI

Costs of percutaneous coronary intervention were extracted from a Governmental organization report (The National Evidence-based Healthcare Collaborating Agency 2010). We included unit cost for procedure, material costs, hospitalization costs, and costs for several tests needed for the procedure which were estimated from the national reimbursement database of Health Insurance Review and Assessment (HIRA). We only considered one-time procedural costs and did not consider revascularization costs.

(3) Costs of treating cardiovascular events

We estimated annual insurance-covered medical costs per patient for hospitalization, ambulatory visit, and drug therapy associated with nonfatal myocardial infarction, nonfatal stroke, and cardiovascular death using the National Health Insurance (NHI) Statistics data (The National Health Insurance Corporation; the Health Insurance Review & Assessment Service 2021). We used the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) billing codes of I21-22 to identify myocardial infarction and I60-69 & G45 for stroke. For cardiovascular death, we assumed these patients used healthcare services related with hypertension (ICD-10 code: I10-15), ischemic heart disease (ICD-10 code: I20-25), and stroke (ICD-10 code: I60-69 and G45) before death. For nonfatal bleeding costs, we used data from a peer-reviewed publication (Ewen 2009) due to absence of data in the NHI Statistics. We applied purchasing power parity to accommodate the difference in price level between the United States of America and Korea (Organisation for Economic Co-operation and Development 2020).

Direct non-medical costs included transportation and caregiver’s costs related to inpatient and outpatient care. We estimated transportation costs by multiplying the average round-trip transportation cost (2,896 and 891 KRW for admission and outpatient visits, respectively) by the total annual number of admissions and ambulatory visits per patient having each event (The Korea Center for Disease Control and Prevention 2005; The National Health Insurance Corporation; Health Insurance Review & Assessment Service 2021). Reflecting severity of disease and family-oriented care culture in Korea, transportation costs for a caregiver were also considered for all hospital admissions and ambulatory visits for the cases of ischemic heart disease and stroke. In addition, assuming that patients would be taken care by one of the family members during hospitalization, caregiver’s cost was calculated as a product of the average daily wage rate and the annual total length-of-stay of hospitalizations per patient for each event based on the NHI data (The National Health Insurance Corporation; Health Insurance Review & Assessment Service 2021; The Korean Statistical Information Service 2020c).

Indirect costs consisted of loss of productivity due to absence from work during hospitalization and premature death. We estimated the costs of loss of productivity based on human capital approach from the age of 20 to 69 years who are the age groups involved in economic activities (Drummond et al. 2015). Productivity loss costs of absenteeism were computed as a product of the total length-of-stay for admissions per year and average ten-year age-specific daily wage rates. We assumed patients would lose one day for a work for each of an ambulatory visit. The length-of-stay for hospitalizations and number of outpatient visits were based on the NHI data for each event. Premature death costs were calculated as the expected value of an individual’s future earnings up to 69 years-old if the patient did not die due to cardiovascular disease. The average ten-year age-specific yearly wage rate, the average increase rate of wage, the rate of participation in economic activities, the unemployment rate, and the case-fatality rate were considered (The Korean Statistical Information Service 2020c; The National Health Insurance Corporation; Health Insurance Review & Assessment Service 2021). We assumed patients would die in the middle age of ten-year age intervals.

Data analysis

The primary outcome was the incremental cost per effectiveness ratio (ICER) which was calculated as the difference in costs divided by the difference in effectiveness between treatment alternatives. We presented the ICER with 95% confidence interval obtained by the Monte-Carlo simulation (Gold et al. 1996). The effectiveness measures were a death avoided and cardiovascular event avoided. The cardiovascular event was defined as any occurrence of nonfatal myocardial infarction, nonfatal stroke, and bleeding. The secondary outcome was the incidence of each event for each treatment alternative. We also calculated the number needed to treat (NNT) with a specific alternative compared to another alternative to examine the number of patients to treat to avoid the occurrence of one event among death, myocardial infarction, stroke, and bleeding. To calculate the NNT, the number of total events was estimated for each alternative. The NNT was the inverse of the differences of total events between alternatives divided by the number of simulated patients.

A univariate sensitivity analysis was performed, and results were shown in a tornado diagram. The lower limit of each parameter used in sensitivity analysis was –20% of the base-case value. To be parallel with this, the upper limit of each parameter was +20% of the base-case value for each parameter. The Microsoft Excel^Ⓡ and TreeAge^Ⓡ software were used for the analysis.

Genotype-guided therapy was expected to have the lowest probability of total events (5.06%) when compared with nongenotype-guided therapy of clopidogrel or prasugrel (9.13% and 13.7%, respectively) (Table 2). During the one-year period following PCI, using genotype information to select antiplatelet therapy was predicted to lower total events of 389 cases from the simulated cohort of 10,000 patients as compared to using clopidogrel for all patients. The NNT of genotype-guided therapy versus clopidogrel therapy for all patients was 26, which means 26 patients need to undergo genotyping to avoid the occurrence of one cardiovascular event. Genotype-guided therapy versus prasugrel for all patients produced approximately 850 fewer total events and NNT of 12.

Table 2 . Results of cost-effectiveness analysis and number needed to treat (NNT) of genotyping to avoid a clinical event.

Treatment alternative	Probability (%)					Annual cost per patient in KRW
	MI	Stroke	Bleeding	Death	Total
Using genotype information to select antiplatelet therapy (①)	2.79	0.28	1.76	0.42	5.24	2,709,114
Clopidogrel regardless of genotypes (②)	4.91	0.04	2.82	0.98	9.13	4,132,063
Prasugrel regardless of genotypes (③)	7.67	0.87	3.79	1.41	13.7	4,766,918
NNT
① versus ②	47	312	94	179	25
① versus ③	21	129	49	101	12
ICERa
① versus ②	−36,610,303
① versus ③	−24,202,531

MI, myocardial infarction; KRW, Korean currency won in 2020 value (1,100 KRW = 1 US$); ICER, incremental cost-effectiveness ratio; NNT, number needed to treat..

^aICER defined as cost difference divided by total events avoided..

When we examined results for each event, the probabilities of nonfatal myocardial infarction (2.79%), nonfatal bleeding (1.76%), and death (0.42%) were the lowest with the treatment strategy of using genotype information among three alternatives. On the other hand, the probability of stroke was the lowest when using clopidogrel for all patients (0.28%).

The expected annual cost per patient was the lowest with the treatment alternative of using genotype information to select antiplatelet therapy (2,709,114 KRW) followed by using clopidogrel (4,132,063 KRW) and prasugrel (4,766,918 KRW) for all patients regardless of genotype information. This leads to annual cost saving of 34.4% and 43.2% if we replace nongenotype-guided antiplatelet therapy with genotype-guided therapy.

The incremental cost per cardiovascular event avoided of genotype-guided therapy versus using clopidogrel without considering genotype information was −36,610,303 KRW. The ICER of genotype-guided therapy versus using prasugrel regardless of genotypes was −24,202,531 (Table 2).

As shown in Fig. 2, the probability of myocardial infarction after taking clopidogrel and the cost of PCI were the most influential parameter to determine the cost-effectiveness of genotype-guided therapy versus using clopidogrel without taking into account for genotypes. The ranges of ICER change from the base-case ICER value were −18% to 31% and −20% to 20%, respectively. The probability of bleeding after taking clopidogrel and the probability of noncarrier with reduced function of CYP2C19 allele were the third and fourth influential factor in determining the ICER of genotype-guided therapy versus using clopidogrel regardless of genotypes. However, the ICER values were still negative when the lower or upper limits of each parameter were used. The lower and upper limits of the rest of parameters changed the base-case ICER with less than 5%. Similar patterns were found when comparing genotype-guided therapy with using prasugrel without considering genotype information.

Figure 2. Tornado diagrams depicting results of sensitivity analyses of parameter impact on incremental cost per event avoided. MI, myocardial infarction; PCI, percutaneous coronary intervention. Dotted line represents the base-case result. KRW, Korean won (one US dollar approximately equals to 1,100 KRW). Alternative 1: selecting antiplatelet therapy (i.e., clopidogrel or prasugrel) based on genotype information of CYP2C19. Alternative 2: clopidogrel therapy without using genotype information. Alternative 3: prasugrel therapy without using genotype information.

We performed the cost-effectiveness analysis of the CYP2C19 genotype-guided antiplatelet therapy compared with using clopidogrel or prasugrel without considering genotypes in patients with acute coronary syndrome undergoing percutaneous coronary intervention in Korea with societal perspective. The estimated incremental cost-effectiveness ratios of the treatment strategy of selecting antiplatelet therapy utilizing the genotype information were −36,610,303 KRW (−33,282 USD) and −24,202,531 KRW (−22,002 USD) when compared with using clopidogrel and prasugrel for all patients without considering the genotype, respectively. The probability of myocardial infarction after taking clopidogrel or prasugrel played the key role in the cost-effectiveness results followed by the cost of percutaneous coronary intervention. Using genotype information to select antiplatelet therapy was less costly and more effective when compared with using clopidogrel or prasugrel for all patients. Genotype-guided therapy was a dominant treatment strategy versus other two alternatives.

Several cost-effectiveness and risk-benefit analyses were performed to examine the economic value of CYP2C19 genotype-guided antiplatelet therapy in patients with acute coronary syndrome in different countries but not in Korea (Reese et al. 2012; Sorich et al. 2013; Kazi et al. 2014; Deiman et al. 2016; Wang et al. 2018; Fu et al. 2020; Limdi et al. 2020; Narasimhalu et al. 2020; AlMukdad et al. 2021; Kim et al. 2021; Claassens et al. 2022). Although there are several studies already done evaluating the cost-effectiveness, it is important to have economic evaluation for each country reflecting its own healthcare system and most importantly reflecting racial characteristics. We found similar results with previous evidence as genotype-guided therapy was cost-effective compared to the universal use of antiplatelet therapies in patients with acute coronary syndrome. Also, the drug costs and the efficacy of antiplatelet therapies were found to be common influential factors of the cost-effectiveness analyses. Because of the economic value of genotype-guided antiplatelet therapy in patient with acute coronary syndrome, widening the use of this strategy to different patient population such as patients with peripheral arterial disease is under investigation as a randomized controlled trial with the study name of GENotype-guided antithrombotic treatment versus conventional clopidogrel therapy in Peripheral Arterial Disease (GENPAD) (Kranendonk et al. 2022).

Interestingly, we found that the genotype-guided antiplatelet therapy in Korea was dominant whereas this strategy was cost-effective in other countries because of the lower incremental cost-effectiveness ratio than the willingness-to-pay threshold. This might be because numerous numbers of generic versions of clopidogrel with lower prices than the price level of prasugrel are currently on the market in Korea. In addition, genotyping for antiplatelet therapy is getting more accessible to patients and widely adopted in many hospitals.

To our knowledge, this study is the first study to evaluate the cost-effectiveness of the CYP2C19 genotype-guided antiplatelet therapy compared with using clopidogrel or prasugrel without considering genotypes in patients with acute coronary syndrome undergoing percutaneous coronary intervention in Korea using the real-world data. However, the results of this study should be interpreted with caution because of several limitations. First, we did not include the treatment strategy of using ticagrelor because it was not widely used at the time when we performed the analysis. Further studies are needed to include ticagrelor as the treatment strategy for antiplatelet therapy in patients with acute coronary syndrome undergoing percutaneous coronary intervention in Korea. Second, we employed a relatively shorter time horizon but not a life-time span. This was because shorter time horizon was sufficient to evaluate the cost-effectiveness of genotype-guided antiplatelet therapy in patients with acute coronary syndrome. In this way, we lessened the uncertainty in efficacy measures by not incorporating extrapolation. Third, we performed a cost-effectiveness analysis but not a cost-utility analysis. Cost-utility analysis is regarded as a gold standard of economic evaluation because it includes quality of life measures and enables decision making using a willingness-to-pay threshold derived for incremental costs per quality adjusted life years gained. However, since we had a dominant result for a genotype-guided antiplatelet therapy compared with treatment strategy not considering genotypes in patients, decision can be made. Last but not the least, we used the bleeding costs incurred in patients who underwent percutaneous coronary intervention in another country due to the absence of Korean data.

Among patients with acute coronary syndrome undergoing percutaneous coronary intervention, the genotype-guided antiplatelet therapy compared with using clopidogrel or prasugrel without considering genotypes was found to be a dominant strategy and a cost-effective strategy. Using patient-specific factors CYP2C19 genotype offers promise for actively adopting a personalized medicine approach to antiplatelet treatment regimens not only in clinical aspect but in economic aspect.

The authors declare that they have no conflict of interest.

This research was supported by a grant (12182KFDA686) from the Korea Food and Drug Administration in 2012 and a grant (21153MFDS601) from Ministry of Food and Drug Safety in 2023. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Korean Food and Drug Administration.