Clustering of ABCB1 and CYP2C19 Genetic Variants Predicts Risk of Major Bleeding and Thrombotic Events in Elderly Patients with Acute Coronary Syndrome Receiving Dual Antiplatelet Therapy with Aspirin and Clopidogrel.

OBJECTIVE: The clinical efficacy of clopidogrel in secondary prevention of vascular events is hampered by marked inter-patient variability in drug response, which partially depends on genetic make-up. The aim of this pilot prospective study was to evaluate 12-month cardiovascular outcomes in elderly patients with acute coronary syndrome (ACS) receiving dual antiplatelet therapy (aspirin and clopidogrel) according to the clustering of CYP2C19 and ABCB1 genetic variants.
METHODS: Participants were 100 consecutive ACS patients who were genotyped for CYP2C19 (G681A and C-806T) and ABCB1 (C3435T) polymorphisms, which affect clopidogrel metabolism and bioavailability, using PCR-restriction fragment length polymorphism. They were then grouped as poor, extensive and ultra-rapid metabolisers based on the combination of CYP2C19 loss-of-function (CYP2C19*2) and gain-of-function (CYP2C19*17) alleles and ABCB1 alleles. The predictive value of each phenotype for acute vascular events was estimated based on 12-month cardiovascular outcomes.
RESULTS: The poor metabolisers were at an increased risk of thrombotic events (OR 1.26; 95% CI 1.099-1.45; χ2 = 5.676; p = 0.027), whereas the ultra-rapid metabolisers had a 1.31-fold increased risk of bleeding events compared with the poor and extensive metabolisers (OR 1.31; 95% CI 1.033-1.67; χ2 = 5.676; p = 0.048). Logistic regression model, including age, sex, BMI and smoking habit, confirmed the differential risk of major events in low and ultra-rapid metabolisers.
CONCLUSIONS: Our findings suggest that ACS patients classified as 'poor or ultra-rapid' metabolisers based on CYP2C19 and ABCB1 genotypes should receive alternative antiplatelet therapies to clopidogrel.

Key Points

Introduction

Clopidogrel, a platelet activation and aggregation inhibitor, acts through irreversible binding of its active metabolite to P2Y12, an adenosine diphosphate (ADP) receptor involved in platelet glycoprotein GPIIb/IIIa complex activation. It is a safe and effective medication for secondary prevention of cardiovascular events (CVE) [1]. A number of trials have documented the benefit of adding clopidogrel to aspirin in treating patients with acute coronary syndrome (ACS), and the drug has emerged as the most common antiplatelet agent combined with aspirin [2-4]. Pharmacoeconomic data from several countries consider their combination for up to 12 months as a cost-effective antiplatelet therapy compared with aspirin alone in ACS patients [5]. Despite the introduction of two new P2Y12 inhibitors, prasugrel and ticagrelor, clopidogrel is still the mainstay of antiplatelet therapy, especially in elderly and old patients, and one of the most commonly prescribed drugs worldwide [6, 7]. However, the variability of inter-patient response to the drug is an outstanding issue that requires further investigation. It has been estimated that antiplatelet response is inadequate in over 30% of patients treated with clopidogrel, who are at increased risk of developing CVE or bleeding [8].

Clopidogrel is a pro-drug that requires intestinal absorption and biotransformation to an active metabolite, mediated by multiple cytochrome P450 (CYP) enzymes coding for the CYP2C19, CYP3A, CYP2B6 and CYP1A2 genes. CYP2C19 converts clopidogrel into its active metabolite. On the contrary, the esterases pathway leads to hydrolysis of clopidogrel into an inactive carboxylic acid derivative (85% of circulating metabolites).

Differences in the extent of biotransformation are believed to account for the variability found in the inter-individual response to the drug, and there is mounting evidence that such variability is mainly associated with CYP2C19 gene polymorphisms [9]. The CYP2C19*1 allele is associated with a fully functional metabolism, whereas CYP2C19*2 and *3 are associated with loss of function (LOF). CYP2C19 LOF allele carriers convert less clopidogrel into its active metabolite, which results in diminished antiplatelet response and higher CVE rates [10, 11]. Since the effect is especially marked among patients undergoing percutaneous coronary intervention (PCI), CYP2C19 allele screening, performed to guide in the prescription of clopidogrel to ACS patients who are likely to undergo coronary stenting, currently focuses on LOF alleles (CYP2C19*2 and *3) [12, 13].

The US Food and Drug Administration has recently incorporated CYP2C19 genetic information in the updated clopidogrel label as a black box warning stating that LOF allele carriers may have a reduced response to standard doses [14, 15].

In contrast, CYP2C19 allele *17 is responsible for gain of function (GOF), and has recently been associated with an increased risk of bleeding events [15].

CYP2C19*2 allele explains 12% of the variability. However, additional variants in this gene could explain a high percentage of variation [16], as well as other candidate genes, including the ABCB1 gene. Among the key proteins involved in thienopyridine absorption, the ATP-dependent efflux pump P-glycoprotein (P-gp) is encoded by the ATP-binding cassette, sub-family B, member 1 (ABCB1 gene) [17]. P-gp transports various molecules across extra- and intracellular membranes. Among other sites, it is expressed on intestinal epithelial cells, where its overexpression or increased function has the potential to alter drug bioavailability.

Literature data suggest that the levels of the active clopidogrel metabolite are lower in individuals with ABCB1 gene variants, specifically those who are TT homozygous for the C3435T variant, and that this may result in higher rates of adverse clinical outcomes [17]. Only a few large studies have investigated the effect of CYP2C19 LOF and GOF alleles and ABCB1 C3435T alleles [18].

This pilot prospective study assessed 12-month cardiovascular outcomes in elderly ACS patients receiving dual antiplatelet therapy (aspirin and clopidogrel) and grouped into three phenotypes based on the clustering of CYP2C19 and ABCB1 genetic variants.

Materials and Methods

Statistical Analysis

Hardy–Weinberg equilibrium and linkage disequilibrium between polymorphisms were evaluated. The Hardy–Weinberg equilibrium (HWE) was tested using the exact test proposed by Wigginton et al (2005) [19]. Pairwise measures of linkage disequilibrium (LD) between the analysed loci were calculated with the Haploview 4.2 [20].

Continuous variables are presented as mean ± standard deviation (SD), categorical variables as count and percentage. The independent sample t test was used to compare continuous variables, the chi-square test to compare categorical variables. Differences between groups were analysed by one-way analysis of variance (ANOVA) for independent groups. Odds ratios (OR) and 95% confidence intervals (CI) were reported.

As this was a multivariate analysis, a logistic regression model was applied. Statistical significance was defined as a two-tailed p value < 0.05. Data analysis was carried out with the SPSS/Win program version 18.0 (SPSS, Chicago, IL, USA).

Patients

Participants were 100 consecutive Caucasian subjects referred to the Coronary Care Unit (CCU) of INRCA, Ancona, Italy, for ACS from January to December 2015. Patients were prescribed dual antiplatelet therapy with aspirin and clopidogrel according to the current best clinical practice criteria applied at our centre. Those receiving clopidogrel (a single daily dose of 75 mg initiated with a 300-mg loading dose) combined with acetylsalicylic acid (100 mg/day initiated with a 300-mg loading dose) and proton pump inhibitor (PPI), were subjected to genetic testing for CYP2C19 and ABCB1 variants. All participants received the same loading and maintenance dose of clopidogrel and aspirin for the entire year of follow-up. Since the elderly patients enrolled for the study were at risk of gastric bleeding, they were on PPI inhibitors. To minimise the interaction between PPI inhibitors and clopidogrel, the two drugs were administered in the morning and in the evening, respectively.

Informed consent was obtained from all participants included in the study. Patients events were classified as previously described [21].

Genotyping

Blood samples were collected in tubes containing K-EDTA (potassium ethylenediaminetetraacetic acid). Genomic DNA was extracted using a commercially available DNA isolation kit (QIAGEN, DNA isolation kit) according to the manufacturer’s instructions. The presence of DNA was confirmed by running DNA on 0.8% agarose gel.

The CYP2C19 gene was analysed for CYP2C19*2 (G681A) and CYP2C19*17 (C-806T) polymorphisms and the ABCB1 gene for the C3435T polymorphism.

Amplification was performed using the commercial BIOAESIS line-100 CLOPIDOGREL oligo mix kit (BIOAESIS srl) according to the manufacturer’s protocol.

Primers 5′_ACAACCAGAGCTTGGCATATT_3′ and 5′_TGTCCATCGATTCTTGGTGT_3′ were used to amplify the CYP2C19 gene sequence containing the single nucleotide polymorphism (SNP) G681A. The CYP2C19 gene sequence containing the SNP (C-806T) was amplified using primers 5′_CATCTCTGGGGCTGTTTTCCTTA_3′ and 5′_GCGCATTATCTCTTACATCAGGGAT_3′.

Primers 5′_CAAAGTGTGCTGGTCCTGAA_3′ and 5′_TGCTCCCAGGCTGTTTATTT_3′ were used to amplify the ABCB1 gene sequence containing the SNP C3435T.

Polymerase chain reaction (PCR) amplification steps included an initial denaturation step at 95 °C for 2 min, followed by 40 cycles of denaturation at 95 °C for 20 s, an annealing step at 56 °C for 30 s and an extension step at 72 °C for 30 s. The resulting amplicons of CYP2C19*2 (203 bp), CYP2C19*17 (209 bp) and ABCB1 (287 bp) were digested with SmaI, BtscI, and MboI restriction enzymes (CLOPIDOGREL enzymes kit, BIOAESIS srl), respectively.

CYP2C19*2, CYP2C19*17 and ABCB1 PCR products were digested overnight at 25, 50 and 37 °C, respectively. Enzyme deactivation was done at 65 °C for 20 min; the resulting PCR-restriction fragment length polymorphism (RFLP) products were analysed by 2.5% agarose gel electrophoresis and stained with GelRed. We re-genotyped 10% randomly selected samples, obtaining a concordance rate of 100%. Moreover, the genotypic status of 10 samples was further validated by direct sequencing of the target region.

Patient Classification by Phenotype

Patients were divided into three metaboliser phenotypes using the established common-consensus ‘star allele’ nomenclature [22]. The three drug bioavailability classes were defined according to Parè et al., with some modifications to include the ABCB1 polymorphism [23]. Since we analysed a small sample of patients, in order to reduce the number of groups to compare, patients previously classified as ‘poor’ and ‘intermediate’ were grouped as ‘poor’, since they have a reduced CYP2C19 activity, whereas patients previously classified as ‘rapid’ and ‘ultra-rapid’ were grouped as ‘ultra-rapid’.

The three identified groups were as follows:

Ultra-rapid: heterozygosity or homozygosity for the CYP2C19*17, absence of variant CYP2C19*2 and CT heterozygosity or CC homozygosity for ABCB1 3435C > T;

Extensive: absence or heterozygosity of variant CYP2C19*2 and absence of CYP2C19*17 and CT heterozygosity or CC homozygosity for ABCB1 3435C > T;

Poor: heterozygosity or homozygosity for the CYP2C19*2, absence of variant CYP2C19*17 and TT homozygosity for ABCB1 3435C > T.

Genotype combinations of ABCB1 3435C > T and CYP2C19 are reported in Supplementary Table 1 [see electronic supplementary material (ESM)].

Patients were re-evaluated after 12 months of treatment for the following clinical outcomes: acute myocardial infarction, ischaemic stroke, stent thrombosis and major bleeding.

Results

The 100 consecutive patients enrolled in the study included 56 men and 44 women who had a mean age of 79.7 ± 8.5 years (range 68–95); 21 patients had unstable angina (UA), 38 patients had ST-elevation myocardial infarction (STEMI) and 41 patients had non-ST-elevation myocardial infarction (NSTEMI).

We found a significant deviation from HWE for the three analysed SNPs (p < 0.05). Moreover, we assessed linkage disequilibrium (LD) between CYP2C19*2 (G681A) and CYP2C19*17 (C-806T) polymorphisms, both located on the same gene on chromosome 10. Although recent evidence suggested that the impact of the CYP2C19*17 variant is primarily being driven by CYP2C19*2, our finding showed absence of LD between these two polymorphisms (D′ = 0.526; r2 = 0.025) indicating that their effects are independent.

Clustering of CYP2C19 and ABCB1 genetic variants enabled patients to be divided into 22 ultra-rapid metabolisers, 56 poor metabolisers, and 22 extensive (normal) metabolisers [Supplementary Table 1 (see ESM)]. The three groups did not exhibit significant differences in terms of chemical–clinical parameters or ACS diagnosis (UA, NSTEMI, STEMI) (Table 1).

Table 1

Chemical-clinical parameters according with the clustering of CYP2C19 and ABCB1 genetic variants

Characteristics of study population	Extensive	Poor	Ultra-rapid	p-value*
Age (year)	79.2 ± 9.0	78.8 ± 8.0	81.5 ± 7.6	0.507
BMI	26.3 ± 10.0	27.5 ± 9.8	24.1 ± 3.2	0.414
Creatinine (mg/dL)	1.8 ± 1.3	2.0 ± 1.8	2.6 ± 1.6	0.085
Total cholesterol (mg/dL)	184.8 ± 38.8	176.8 ± 40.2	166.1 ± 38.5	0.150
HDL (mg/dL)	51.0 ± 13.3	47.3 ± 9.4	45.8 ± 10.5	0.179
Triglycerides (mg/dL)	146.2 ± 46.1	146.2 ± 43.9	158.1 ± 38.7	0.543
Glucose (mg/dL)	102.6 ± 12.6	101.6 ± 13.6	100.1 ± 11.4	0.736
hs-TnT (pg/mL)	221.3 ± 274.1	215.1 ± 298.8	259.5 ± 283.2	0.839
NT-proBNP (pg/mL)	889.0 ± 1481.5	752.1 ± 746.0	1222.3 ± 1536.3	0.494
hs-PCR (mg/dL)	0.69 ± 0.88	1.25 ± 1.48	1.29 ± 1.57	0.067
ACS diagnosis (UA/STEMI-NSTEMI)	4/18	12/44	5/17	0.929

*ANOVA

ACS acute coronary syndrome, BMI body mass index, HDL high-density lipoprotein, hs-PCR high-sensitivity C reactive protein, hs-TnT high-sensitivity troponin T; NSTEMI non-ST-elevation myocardial infarction, STEMI ST-elevation myocardial infarction, UA unstable angina, NT-proBNP N-terminal pro b-type natriuretic peptide

The number of thrombotic events and major bleeding events in the three patient groups are shown in Table 2. The rates of major bleeding events, thrombotic events and ‘other outcomes’ (i.e. re-hospitalisation and death) are reported in Fig. 1. These rates differ significantly among the groups (Fisher’s Exact Test = 20.640; p < 0.001), whereas there were no significant differences in relation to ACS diagnosis at admission (data not shown).

Table 2

Thrombotic events and major bleeding events in the ACS patient groups divided based on the clustering of CYP2C19 and ABCB1 genetic variants

Class	Genotype clustering criteria	Total N	Thrombotic eventsN (%)	Major bleedingN (%)	Other outcomesaN (%)
Extensive	Genotype combinations: Absence or heterozygosity of CYP2C19*2 Absence of CYP2C19*17 CT heterozygosity or CC homozygosity for ABCB1 3435C>T	22	0	0	2 (9.1)
Poor	Genotype combinations: Heterozygosity or homozygosity for CYP2C19*2 Absence of CYP2C19*17 TT homozygosity for ABCB1 3435C>T	56	11 (19.6)	0	8 (14.3)
Ultra-rapid	Genotype combinations: Heterozygosity or homozygosity for the CYP2C19*17 Absence of variant CYP2C19*2 – CT heterozygosity or CC homozygosity for ABCB1 3435C>T	22	0	5 (22.7)	2 (9.1)

Fisher’s Exact Test = 20.640, p < 0.001

ACS acute coronary syndrome

aOther outcomes = re-hospitalisation and death

Fig. 1

Percentages of major bleeding events, thrombotic events, re-hospitalisation and death in studied patients according with the clustering of CYP2C19 and ABCB1 genetic variants

A greater percentage of patients with bleeding was found among ultra-rapid metabolisers, and a greater rate of thrombotic events was found among poor metabolisers (Table 3). The ultra-rapid metabolisers showed a 1.31-fold increased risk of bleeding compared with extensive and poor metabolisers (OR 1.31; 95% CI 1.033–1.67; χ2 = 5.676; p = 0.048), and the poor metabolisers showed an increased risk of thrombotic events compared with the other two groups (OR 1.26; 95% CI 1.099–1.45; χ2 = 5.676; p = 0.027).

Table 3

Numbers and percentages of ACS patients according with the 12-month follow-up outcomes and genotypic characteristics

	Metaboliser
	Extensive	Poor	Ultra-rapid	p valuea
Major bleeding
No	22 (100%)	56 (100%)	17 (77.3%)	0.001
Yes	0 (0%)	0 (0%)	5 (22.7%)
Thrombotic events
No	22 (100%)	45 (80.4%)	22 (100%)	0.006
Yes	0 (0%)	11 (19.6%)	0 (0%)

ACS acute coronary syndrome

aFisher’s Exact Test

Logistic regression analysis, including major bleeding, thrombotic events, re-hospitalisation and death as ‘combined event’, and age, sex, body mass index (BMI) and smoking habit as confounding variables, confirmed the different risk of combined events for the ultra-rapid and poor metabolisers compared with the extensive metabolisers (Table 4).

Table 4

Logistic regression model

	p value	OR	95% CI
			Low	High
Poor vs extensive	0.034	10.256	1.195	88.010
Ultra-rapid vs extensive	0.039	10.695	1.125	101.630
Sex (F/M)	0.820	1.133	0.387	3.315
Age	0.427	0.971	0.904	1.044
BMI	0.118	1.083	0.980	1.197
Smoke (Y/N)	0.065	4.155	0.914	18.877
Constant	0.250	0.021

Major bleeding, thrombotic events, re-hospitalisation and death were considered as ‘combined event’

BMI body mass index, CI confidence intervals, OR odds ratio

Discussion

Dual antiplatelet therapy with aspirin and platelet P2Y12 ADP receptor antagonist reduces recurrent major adverse cardiovascular events in patients with ACS. A considerable percentage of patients have recurrent cardiovascular events despite the clopidogrel regimen [24]. Therefore, early recognition of impaired responders to clopidogrel is imperative. Findings from a recent study by Legrand et al. showed that anaemia, BMI and diabetes mellitus, which together define the STIB score, act as independent variables with similar weight in predicting the risk of clopidogrel resistance [25]. Legrand was the first to develop a simple clinical indicator (STIB score) in order to predict impaired response to clopidogrel at the bedside. Clinicians demand technologies that effectively help them to select the most appropriate treatment options for their patients by providing information quickly and in an easily interpretable form; in this setting, Legrand used the platelet function test (PFT) to asses platelet inhibition due to clopidogrel and we know that, despite many efforts undertaken, this method has huge discrepancies between laboratories and between laboratory methods [26, 27]. On the other hand, we have compact and portable point-of-care genotyping instruments for evaluating clopidogrel metabolism, and it is well established that CYP2C19 polymorphisms are associated with impaired response to clopidogrel [28, 29]. In particular, CYP2C19*2 and *3 (LOF) alleles are associated with a reduced response to standard doses [13, 30, 31]. Notably, a recent study of a Chinese population suggests that CYP2C19*2 alleles are related to the occurrence and recurrence of cerebral ischaemic stroke [32]. The major finding of our study is that the combination of three polymorphisms (CYP2C19 LOF and GOF alleles and ABCB1 C3435T allele) can identify patients at increased risk of developing not only thrombotic events but also major bleeding. These data suggest that the combination of CYP2C19 and ABCB1 polymorphisms is more informative than a single-gene polymorphism. The observation that the above polymorphisms showed a significant deviation from HWE is not surprising as we analysed a selected group of patients. Indeed, several authors proposed to incorporate deviation from HWE as a measure of correlation within the frame of genetic association studies [33, 34]. Thus, from this point of view, the deviation from HWE further confirms our findings.

Previous data show that carriers of the CYP2C19*17 variant are more responsive to clopidogrel than non-carriers, but are at an increased risk of bleeding [35, 36]. However, a recent meta-analysis of studies of the impact of the CYP2C19 polymorphisms on the risk of adverse clinical events showed that in clopidogrel-treated patients the polymorphisms were significantly associated with adverse clinical events, but not with bleeding [37].

Balancing the ischaemic and haemorrhagic risk is a complex effort. Genotype testing for clopidogrel resistance should be considered ‘reasonable’ if the results strongly suggest a change in management for these patients and if drugs alternative to clopidogrel are available. Two new antithrombotic agents, prasugrel and ticagrelor, have recently become available. Both are faster acting and more potent than clopidogrel. Ticagrelor is a reversible, direct-acting P2Y12 inhibitor, whereas prasugrel, like clopidogrel, is a pro-drug but has a faster onset of action and a more consistent inhibitory effect on platelet aggregation. Antiplatelet therapy is commonly prescribed to ACS patients and to those undergoing primary PCI. The introduction of ticagrelor and prasugrel and the inevitable addition of a generic form of the ubiquitous clopidogrel have complicated the decision-making process of antiplatelet therapy.

Regulations encouraging the use of drugs, such as clopidogrel, whose patent expiration makes them less expensive in preference to drugs under patent such as prasugrel and ticagrelor, have been adopted in several countries [38]. Therefore, even though recent data suggest a greater antithrombotic efficacy of the combination of new oral antiplatelet agents with aspirin, instead of clopidogrel with aspirin, financial considerations involve restrictions in the use of the new drugs [39, 40]. The present findings and those of other studies indicate that clopidogrel administration without genetic testing may on the one hand deprive patients of an effective antiplatelet therapy, and on the other hand increase the risk of bleeding events.

The limitations of clopidogrel are even more important if one considers the large number of patients with coronary artery disease (CAD) who have aspirin allergy or intolerance, and are therefore candidates for clopidogrel treatment [41-43].

Even if there are some limitations to the current study, including the small sample size and the analysis of only one ethnicity, our results reinforce the hypothesis that genetic testing should be encouraged: this would both ensure that all suitable patients receive an antiplatelet therapy with proven effectiveness and at the same time help meet healthcare cost reduction goals. The ability to identify polymorphisms related to drug response provides useful tools for personalised medicine. The response to single agents or drug combinations can be optimised based on each patient’s unique genetic make-up. In this framework, the use of genetic screening to predict drug response will not only improve patient quality of life, but also contribute to healthcare cost reduction, avoiding administration of an ineffective drug and the cost of treating its potential adverse effects.

Conclusion

The present findings highlight the need for introducing routine genetic testing at the bedside to guide personalised antiplatelet therapy with clopidogrel, and suggest that ACS patients who are ultra-rapid or poor metabolisers based on CYP2C19 and ABCB1 genotyping should be treated with antiplatelet agents other than clopidogrel, such as prasugrel or ticagrelor.

Below is the link to the electronic supplementary material.

Supplementary Table 1 Phenotype metaboliser status according to CYP2C19 and ABCB1 genotypes (DOCX 18 kb)

The study aims to evaluate the correlation between the CYP2C19 and ABCB1 genotypes and the cardiovascular events in elderly patients with acute coronary syndrome on dual antiplatelet therapy.

The study showed that patients with a genotype associated with ‘poor’ metabolism of clopidogrel are at greater risk of thrombotic events whereas those with genotypes characterised by an ultra-rapid metabolism are at greater risk of bleeding.

It has been confirmed that the risk of cardiovascular events in elderly patients on dual antiplatelet therapy is significantly associated with ABCB1 and CYP2C19 genetic polymorphisms.