Mutational analysis clopidogrel resistance and platelet function in patients scheduled for coronary artery bypass grafting

Abstract

Clopidogrel is an oral antiplatelet pro-drug prescribed to 40 million patients worldwide who are at risk for thrombotic events or receiving percutaneous coronary intervention (PCI). However about a fifth of patients treated with clopidogrel do not respond adequately to the drug. From a cohort of 105 patients on whom we had functional data on clopidogrel response, we used ultra-high throughput sequencing to assay mutations in CYP2C19 and ABCB1, the two genes genetically linked to respond. Testing for mutations in CYP2C19, as recommended by the FDA, only correctly predicted if a patient would respond to clopidogrel 52.4% of the time. Similarly, testing of the ABCB1 gene only correctly foretold response in 51 (48.6%) patients. These results are clinically relevant and suggest that until additional genetic factors are discovered that predict response more completely, functional assays are more appropriate for clinical use.

1. Introduction

Clopidogrel is an oral antiplatelet pro-drug prescribed to 40 million patients worldwide who are at risk for thrombotic events or receiving percutaneous coronary intervention (PCI) [1]. Many studies have demonstrated the efficacy of its use in addition to aspirin to prevent death, myocardial infarction, and stroke in patients with coronary syndromes. The FDA currently recommends clopidogrel as a therapy administered once daily for 365 days for patients undergoing PCI [1,2]. Clopidogrel has an irreversible antagonistic effect on the adenosine diphosphate (ADP) P2Y12 receptor and has been shown to significantly improve clinical outcomes in patients who respond to the medication [2–5]. However, the response to clopidogrel varies. About a fifth of patients treated with clopidogrel do not respond adequately due to decreased absorption, inadequate conversion to its active metabolite, or other reasons such as increased body mass index, diabetes mellitus, and acute coronary syndrome [6,7]. Interactions with other medications, such as the proton pump inhibitor omeprazole, have been also identified as responsible for the lack of effectiveness of clopidogrel [1].

Clopidogrel is administered orally, absorbed through the intestine, and then metabolized to its active form in the liver by cytochrome P450 enzymes in a two-step process [6,7]. Two genes, ABCB1 and CYP2C19, have been identified as important players in this process. ABCB1 encodes the P-glycoprotein efflux transporter, which is responsible for absorbing clopidogrel from the intestine [6–8]. CYP2C19 (Cytochrome P450 2C19) encodes a member of the cytochrome P450 superfamily of enzymes which catalyze many reactions involved in drug metabolism and synthesis of cholesterol, steroids and other lipids.

The CYP2C19*2 (G681A) and ABCB1 (C3435T) variants are the most common loss-of-function mutations described in the literature [1,9,10]. Patients with these variants are classified as low- or non-responders to clopidogrel and tend to have higher platelet reactivity, which places them at risk for adverse cardiovascular events [6,8,10–12]. However, the Genotype Information and Functional Testing (GIFT) study has cited CYP2C19*2, not ABCB1 or other variants, as the crucial gene variant that significantly influences the reactivity and pharmacodynamics of clopidogrel in patients who have undergone PCI [10]. Nevertheless, Price and colleague note that this piece of genetic data provides only limited utility in patients who exhibit the potential for high adverse reactivity to clopidogrel [10].

A patient’s likely functional response to clopidogrel can be assessed by pre-administration genetic testing. There are two bedside genetic tests currently used, but they require long wait times and are tested only for specific allele variations; the CYP2C19 genotype has been shown to account for just 12% of the variation of clopidogrel response [9]. For these reasons, a functional platelet inhibition test (PIT) has become a more practical choice in clinical practice. The VerifyNow P2Y12 assay (Accumetrics, San Diego, CA) uses whole blood to measure aggregation of platelets to fibrogen-coated beads in response to exposure to prostaglandin E1 and ADP [7,10,13]. Patients with VerifyNow scores >240 P2Y12 reaction units (PRU) have a higher risk for cardiovascular-related death, stent thrombosis, and non-fatal myocardial infarction [3–5,14]. This >240 PRU cut-off is comparable to the established thresholds for other PITs such as light transmittance aggregometry, Plateletworks, and Innovance PFA P2Y assays (Breet, van Werkum et al., 2010). However, there is only a modest correlation between these tests and cardiovascular (CV)-related death, stent thrombosis, or non-fatal MI [3,5].

In a previous study, we demonstrated the benefit of using a PIT (without genetic testing) to guide the timing of coronary surgery for patients receiving clopidogrel [15]. Since then, genetic testing has become widely available and endorsed by the FDA [16]. With this current study, our goal was to validate in our surgical series of patients a genetic screen with a functional PIT (VerifyNow) to ascertain the true yield of these new assays against a functional test.

2. Results

Of the 105 patients analyzed, 71 (67.62%) were deemed non-responders by the VerifyNow Assay. Non-responders were more likely to female, older, not current smokers, diabetics, with a history of renal failure and taking beta blockers and ADP inhibitors (Table 1).

Table 1.

Patient demographics.

	All (n = 105)		Responder (n = 34)		Non-responder (n = 71)		p-Value
Male	81	77.4%	33	97.1%	48	67.6%	0.001
Caucasian	100	95.2%	33	97.1%	67	94.4%	0.544
Age	68.3	11.3	62.7	8.8	71.0	11.3	0.0001
Weight (kg)	85.2	17.3	86.6	20.7	84.6	15.5	0.577
Height (cm)	172.4	9.9	176.3	8.8	170.6	10.0	0.006
Body surface area	2.01	0.24	2.05	0.28	2.00	0.22	0.323
Body mass index	28.62	4.95	27.77	5.81	29.03	4.47	0.224
History of Smoking	59	56.2%	22	64.7%	37	52.1%	0.224
Currently smoking	21	20.0%	13	38.2%	8	11.3%	0.001
Present comorbidity
Diabetes	41	39.1%	9	26.5%	32	45.1%	0.052
Dyslipidemia	98	93.3%	33	97.1%	65	91.6%	0.290
Creatine	1.08	0.5	0.98	0.2	1.14	0.6	0.138
Renal failure	14	13.3%	1	2.9%	13	18.3%	0.024
Dialysis	2	1.9%	0	0.0%	2	2.8%	0.323
Hypertension	96	91.4%	32	94.1%	64	90.1%	0.496
Endocarditis	0	0.0%	0	0.0%	0	0.0%	1.000
Chronic lung disease		9.5%	3	8.8%	7	9.9%	0.785
Immuno-suppressed	2	1.9%	1	2.9%	1	1.4%	0.591
Peripheral vascular disease	11	10.5%	3	8.8%	8	11.3%	0.702
Pacemaker	2	1.9%	1	2.9%	1	1.4%	0.591
Cardiac History
Myocardial infarction	55	52.4%	21	61.8%	34	47.9%	0.183
Cerebrovascular accident	3	2.9%	1	2.9%	2	2.8%	0.971
Cardiovascular disease	11	10.5%	3	8.8%	8	11.3%	0.702
Percutaneous coronary intervention	35	33.3%	14	41.2%	21	29.6%	0.238
Chronic heart failure	15	14.3%	2	5.9%	13	18.3%	0.089
Angina	88	83.8%	28	82.4%	60	84.5%	0.779
Arrhythmia	11	10.5%	3	8.8%	8	11.3%	0.702
Medications
Beta blockers	98	93.3%	30	88.2%	68	95.8%	0.039
ACE inhibitors	53	50.5%	19	55.9%	34	47.9%	0.433
Statins	91	86.7%	29	85.3%	62	87.3%	0.229
Nitrates	6	5.7%	3	8.8%	3	4.2%	0.342
Heparin	44	41.9%	16	47.1%	28	39.4%	0.532
Steroids	1	1.0%	0	0.0%	1	1.4%	0.487
Aspirin	82	78.1%	26	76.5%	56	78.9%	0781
ADP inhibitors	21	20.0%	3	8.8%	18	25.4%	0.048
Glycoprotein IIbIIIa inhibitor	2	1.9%	1	2.9%	1	1.4%	0.591
# of diseased vessels	2.9	0.4	2.9	0.4	2.9	0.3	0.789
Left main disease	56	53.3%	16	47.1%	40	56.3%	0.372
Ejection fraction	52.7	11.4	53.5	10.5	52.3	11.9	0.624
30-day mortality	1	0.95%	0	0.00%	1	1.41%	0.487
Late mortality	8	8.6%	1	2.9%	7	11.3%	0.211
Total mortality	9	9.5%	1	2.9%	8	12.5%	0.154

The genome sequencing mutation detection results are presented in Table 2. Overall, 17 patients (16.2%) had no mutations in either ABCB1 or CYP2C19. Of these, 7 responded to the drug and 10 were non-responders. Mutations were common in the ABCB1 gene, with 76.2% of patients having the C3435T variation. About a third of patients (32.4%) had a loss-of-function mutation in CYP2C19, with a majority of these showing the*2 variant (G636A) (29.5%) compared to the 1.9% and 1.0% of *4 (C1297T) and *8 (T358C) variants, respectively. 23 patients had mutations in both genes. We were unable to determine frequency of *17 variants (T-808C) as the mutation occurs in an intron and our experimental design assayed only the exons.

Table 2.

Correlating genetic and VerifyNow results.

Table	All		Responder		Non-responder
# of patients	105		34	32.38%	71	67.62%
Wild type	17	16.19%	7	20.59%	10	14.08%
Any mutation	88	83.81%	27	79.41%	61	85.92%
ABCB1	80	76.19%	25	73.53%	55	77.46%
+/+	25	23.81%	9	26.47%	16	22.54%
+/−	54	51.43%	17	50.00%	37	52.11%
−/−	26	24.76%	8	23.53%	18	25.35%
CYP2C19	34	32.38%	10	29.41%	24	33.80%
CYP2C19*2
+/+	74	70.48%	26	76.47%	48	67.61%
+/−	28	26.67%	7	20.59%	21	29.58%
−/−	3	2.86%	1	2.94%	2	2.82%
CYP2C19*4
+/+	103	98.10%	33	97.06%	70	98.59%
+/−	2	1.90%	1	2.94%	1	1.41%
−/−	0	0.00%	0	0.00%	0	0.00%
CYP2C19*8
+/+	104	99.05%	33	97.06%	71	100.00%
+/−	1	0.95%	1	2.94%	0	0.00%
−/−	0	0.00%	0	0.00%	0	0.00%
Both mutations	23	21.90%	8	23.53%	15	21.13%
Avg Plavix dose	233.6	219.5	308.8	244.8	197.5	198.1
Avg PRU	275.7	71.2	197.4	31.3	313.2	51.3

Twenty-four of the 34 patients with a CYP2C19 loss-of-function mutation were non-responders (p = 0.414). Similarly, of the 80 patients with a loss-of-function mutation in ABCB1, 55 were non-responders (p = 0.483). CYP2C19 mutations were found in 10 (29.41%) of responders and in 24 (33.80%) of non-responders. The C3435T ABCB1 variant was found in 25 (73.53%) responders and in 55 (77.46%) non-responders. A chi-square analysis showed non-significance (p = 0.781).

3. Discussion

There has been great interest in surveying genetic variants to predict potential response to clopidogrel. A number of commercial tests have been developed that focus on two genes, CYP2C19 and ABCB1, which have been strongly linked to respond to the drug. We determined the germline genetic sequence of these genes in 105 patients using ultra-high throughput sequencing technologies that provide an unprecedented depth of coverage of the entire gene, allowing us to accurately assess each of the purported informative variants. To our surprise, we found little correlation between mutational status in these genes and clopidogrel responsiveness, suggesting that there may be a number of other genetic factors that contribute to determining the response. Our results clearly suggest genetic testing for clopidogrel resistance has low sensitivity. Testing for mutations in CYP2C19, as recommended by the FDA, only correctly predicted if a patient would respond to clopidogrel 52.4% of the time. Similarly, testing of the ABCB1 gene only correctly foretold response in 51 (48.6%) patients.

Our results are not consistent with those of the GIFT study [10]. This may reflect the differences in design between the two studies. The GIFT study researchers used a prospective, multinational, multicenter, double-blind, randomized, and active-controlled approach, showing that a strong relationship between CYP2C19 variants in PCI patients and on-treatment reactivity (OTR) to clopidogrel exists. However, to their own admission, the researchers stated that they selected for patients with higher OTR levels [10]. Ours uses an observational approach that used a single, non-manipulated, surgical series of patients from a single institution whom had been on clopidogrel prior to CABG, reflecting a situation more in line with what practitioners face on a daily basis.

Current American College of Cardiologist (ACC)/American Heart Association (AHA) guidelines recommend a 300–600 mg loading dose of clopidogrel before PCI and a daily 75 mg dose for the following 12 months [2]. In addition, clopidogrel should be discontinued 5 days before major surgery [2]. CYP2C19*2 heterozygotes require a daily dose of 225 mg to maintain “normal” platelet inhibition response, which is triple the FDA recommended dose of 75 mg [17]. The *2/*2 homozygous patients do not reach normal range even with daily doses of 300 mg [17] (Table 3).

Table 3.

Responder status of patients taking 75 mg clopidogrel daily.

	All on 75 daily		Responder		Non-responder
Total	63		16	25.40%	47	74.60%
PRU	282.9	66.2	199.3	29.6	311.4	48.6
Wild type	8	q	2	12.50%	6	12.77%
Mutation	55	87.3%	14	87.50%	41	87.23%
CYP2C19	22	34.9%	6	37.50%	16	34.04%
CYP2C19*2	20	31.7%	4	25.00%	16	34.04%
+/−	18	28.6%	3	18.75%	15	31.91%
−/−	2	3.2%	1	6.25%	1	2.13%
CYP2C19*4	1	1.6%	1	6.25%	0	0.00%
CYP2C19*8	1	1.6%	1	6.25%	0	0.00%
ABCB1	47	74.60%	12	75.00%	35	74.47%
+/−	30	47.6%	7	43.75%	23	48.94%
−/−	17	27.0%	5	31.25%	12	25.53%
Both mutations	12	19.00%	4	25.00%	8	17.02%

In early 2009, the FDA first warned the public about the effect of certain CYP2C19 variants on the effectiveness of clopidogrel, citing two studies that found an association between CYP2C19 polymorphisms and adverse clinical outcomes but did not expand further on the topic [11,12,18]. Later that year, the FDA suggested physicians avoid prescribing clopidogrel to patients with known CYP2C19 polymorphisms [19]. The FDA revised the warning label in 2010 to indicate that physicians “consider alternative treatment or treatment strategies in patients identified as CYP2C19 poor metabolizers” [16]. This label emphasizes the results of Gladding and colleagues’ study of the pharmacokinetic and antiplatelet response to clopidogrel. In this study, 40 patients were put into four groups based upon responsiveness to clopidogrel (poor, intermediate, extensive, and ultrarapid) based upon CYP2C19 variant status [20]. Like our study, they used the results of a VerifyNow assay to determine effectiveness of clopidogrel. They found that doubling the dose (from 75 mg daily to 150 mg) of clopidogrel in patients with CYP2C19*2 mutations reduced platelet reactivity by 26 ± 38 PRU. While the FDA has recently reaffirmed its pro-genetic testing stance, it is imperative to point out that the FDA does not mandate genetic testing.

The American College of Chest Physicians, ACC, and AHA currently do not support routine platelet function or genetic testing [21]. According to the ACC/AHA’s response to the 2010 label update, this is based on the lack of strong “evidence-based data upon which to develop specific recommendations on the role of genetic testing” [1]. They point out that CYP2C19 polymorphisms are associated with just 12% of the variability in platelet response to clopidogrel [9]. Furthermore, studies have shown CYP2C19 mutations to accurately predict adverse clinical events in just 12–20% of patients receiving PCI [1]. The ACC/AHA also expresses its concerns regarding the clinical impact of different polymorphisms and the effect of increased doses of clopidogrel [1]. In addition, a recent meta-analysis of 32 studies on this topic showed no significant correlation between genotype and cardiovascular events [22].

The specifics of the VerifyNow test have been extensively defined [13,23]. In short, platelet reactivity can be quickly tested with low volumes of whole blood. The test measures the degree of ADP-induced inhibition of adenylyl cyclase, which is directly affected by clopidogrel binding to the P2Y12 receptor [14]. The assay yields absolute PRU values and calculates a percent P2Y12 inhibition [14,23]. The results of the VerifyNow assay are comparable to the light transmittance aggregometry, Plateletworks, and Innovance PFA P2Y assays [3–5,13,23]. VerifyNow received 501(k) clearance from the FDA to be used clinically. The mean coefficient of variation of test precision is <8% in volunteers and 3.2% in patients with coronary artery disease [13]. Multiple studies have used the VerifyNow assay as a method of measuring platelet function and have found a PRU >240 to correlate with increased risk for CV-related death, stent thrombus, and non-fatal MI with a sensitivity of 78% and a specificity of 70% [14,23]. The assay is limited by the lack of a pre-clopidogrel baseline, creating the space for potential inaccuracies when reporting the degree of platelet inhibition [13]. VerifyNow is less able to discriminate between strong and weak P2Y12 inhibition signals than other PITs due to its smaller dynamic window [4]. Nonetheless, the VerifyNow assay is a trusted method of determining platelet inhibition levels.

In view of the data provided by the GIFT trial and our study as well as others we can conclude the following: 1) in controlled situations, particular gene variants influence clopidogrel reactivity, 2) while this relationship may not be as strong in actuality, knowledge of this association and how it can work in a non-manipulated setting can go far in formulating and modifying recommendations on appropriate clopidogrel use in clinical and surgical settings. We highly recommend that these genetic findings be used in tandem with other clinical tools to determine the best course of treatment for these patients.

For subpopulations that are dependent on clopidogrel and communities that have limited resources, it is imperative that pre- and post-administration diagnostic tests be conducted to assure the medical treatment is effective to reduce the chance for cardiovascular-related readmissions and adverse events. While the use of clopidogrel in clinical practice is not likely to end soon, the recent warnings by the FDA underscore the danger of administering this drug without full knowledge of its effects in non-responders. Employing clinically useful and relevant screening tools will surely remedy the shortcomings of any treatment.

3.1. Limitations

Our study used retrospective data but analyzed the information in a prospective way. Furthermore, this study was conducted at a medium-sized hospital. Results may have manifested differently in larger or smaller institutions, but steps were taken to minimize bias without compromising the goal or integrity of the study. Because we sequenced only the exonic portions of the CYP2C19 and ABCB1 genes, we were not able to determine frequency of *17 variants (T-808C) as the mutation occurs in an intron.

4. Materials and methods

4.1. Patient identification

From June 2007 to November 2011, 1269 patients underwent isolated coronary artery bypass grafting (CABG) surgery. Of these, 105 patients were administered clopidogrel before surgery, were previously included in the CV Blood and Tissue Bank at The Valley Hospital (IRB approval # 11.0009 and to the Helsinki declaration) and had the results of a VerifyNow P2Y12 assay on file (Table 1). The patients gave consent to gather blood samples for later use at the time of CABG. The mean age was 68.3 years old with a body mass index of 28.6. The cohort favored Caucasians (95.2%) and males (77.1%). This cohort of patients was not artificially manipulated or selected based on predetermined characteristics. Instead, it reflected a realistic sampling of the patient population that an institution like ours would encounter.

4.2. Sample collection

At the time of surgery, patient’s blood was drawn into a single sterile 8 cm³ BD vacutainer cell preparation tube (CPT) with sodium citrate and labeled accordingly. Immediately after blood draw, the tube was carefully inverted 8–10 times to mix the anticoagulant agent with the sample. Within 120 min, the tube was centrifuged at room temperature for a minimum of 20 min at 1500–1800 RCF.

Using a thin pipette, the plasma layer (approximately 4 cm³) was carefully removed and stored in a new sterile storage tube and frozen at −70 °C. Using a second pipette, the white blood cell (WBC) layer (approximately 1.5 cm³) was transferred into a sterile 15 cm³ test tube. The remaining sample containing the erythrocytes and granulocytes was discarded as per hospital policy.

Phosphate-buffered saline solution was added to the WBCs to bring the volume up to 15 mL and the solution was mixed by gentle inversion. This mixture was then centrifuged for 15 min at 300 RCF to allow complete separation of the WBC and the buffer; the supernatant was then discarded. Phosphate-buffered saline solution was then added to the WBC pellet to bring the volume to 2 mL.

The WBC solution was divided equally into two 1.5 mL sterile collection tubes. The tubes were centrifuged for 10 min at 300 RCF and then the supernatant was again decanted. The WBC pellets were labeled and stored in a −70 °C freezer. Pellets were put on dry ice, placed in a secure container, and shipped to Dana-Farber Cancer Institute for genetic testing.

4.3. Genetic testing

DNA was extracted from the frozen the pellets using the Qiagen DNeasy kit and prepared for targeted sequencing of ABCB1 and CYP2C19. A panel of DNA primer pairs was designed to each of the exons in the target genes for use with the Fluidigm AccessArray system. Individual primers consist of a 30 bp universal segment and a unique 20 bp segment. Each pair of primers is designed to hybridize immediately upstream and downstream of a specific genomic target. Supplementary Table 1 provides the genomic coordinates for the regions targeted by our panel.

The sequence capture and library generation process involved by a two-step PCR process where the first PCR reaction captures an amplicon from the region of interest, and a second reaction uses the universal primer segment to append sample barcodes and Illumina sequencing adapters. Bi-directional amplicon sequencing was performed on the Illumina HiSeq 2000, and configured to generate 100 base-pair single-end reads.

Sequencing reads were demultiplexed using the standard Illumina pipeline. Reads from each sample were processed using a workflow developed specifically for this project. First, each read was mapped to one of the 103 PCR primer pairs and trimmed to the initial portion of the read based on the size (between 20 and 30 nucleotides in length) and sequence of the primer. Second, reads that contain “empty” probes (consisting of two primers ligated together without their target sequence) were identified and eliminated. Sequence-similarity searches were performed using megablast from the NCBI BLAST package.

“Filtered” and trimmed reads were then aligned to the reference Human genome (build hg19, GRCh37), using the open-source program Bowtie [24] with the “-k” alignment mode with a seed size of 20 and up to 5 allowable mismatches. After alignment, we called and filtered SNPs using the Samtools suite of programs [25]. We chose not to use the edge and strand bias filtering options as these were designed for random, whole-genome sequencing and our targeted sequencing data has both of these features by design. Supplementary Table 2 provides the specific version number and parameter settings for the tools used in processing. The mutational calls for all patients are provided in Supplementary Table 3.

4.4. Platelet inhibition testing

Platelet inhibition was assessed using the VerifyNow P2Y12 assay. The details of this test have been previously described elsewhere [13,23]. In short, the assay uses a combination of prostaglandin E1 and ADP to increase platelet cAMP, thereby increasing specificity and sensitivity of the ADP-mediated effects of the P2Y12 receptor. The assay uses light transmittance as a measure of ADP-induced platelet aggregation. An algorithm interprets the data and reports the results in PRU. A high PRU indicates greater platelet agglutination. The previously established cut-off of 240 PRU was used to determine responder status.

4.5. Data analysis

Continuous data are displayed as means with standard deviation. Categorical data are expressed as proportions. Categorical variables and differences in rates between groups were analyzed using the chi-squared test. A value of p < 0.05was used to determine the statistical significance of all tests used. Analyses were performed using the SPSS statistical software package version 19.0 (IBM/SPSS Inc., Chicago, IL).