|Home | About | Journals | Submit | Contact Us | Français|
CYP2C19 is a principal enzyme involved in the bioactivation of the antiplatelet prodrug clopidogrel and common CYP2C19 loss-of-function alleles are associated with adverse cardiovascular events. To assess the impact of the CYP2C19*17 increased activity allele in the Ashkenazi Jewish (AJ) and Sephardi Jewish (SJ) populations and to determine the frequencies of additional variant alleles, 250 AJ and 135 SJ individuals were genotyped for CYP2C19*2–*10, *12–*17, *22 and P-glycoprotein (ABCB1) c.3435C>T. Importantly, CYP2C19*4, a loss-of-function allele, was identified in linkage disequilibrium with *17. This novel haplotype, designated CYP2C19*4B, significantly alters the interpretation of CYP2C19 genotyping when testing *17. Moreover, genotyping CYP2C19*17 changed the frequency of extensive metabolizers from ~70 to ~40%, reclassifying ~30% as ultrarapid metabolizers. Combining CYP2C19 and ABCB1 identified ~1 in 3 AJ and ~1 in 2 SJ individuals at increased risk for adverse responses to clopidogrel. These data underscore the importance of including *4B and *17 when clinically genotyping CYP2C19.
The human cytochrome P450 2C (CYP2C) subfamily of enzymes is comprised of CYP2C8, CYP2C9, CYP2C18 and CYP2C19, which together are involved in the hepatic metabolism of ~25% of prescription and over-the-counter drugs.1 Like many other CYP450 superfamily members, the CYP2C19 gene is highly polymorphic with over 25 known variant alleles.2,3 For example, CYP2C19*2 (c.681G>A; rs4244285) and *3 (p.W212X; rs4986893) are the most commonly studied CYP2C19 loss-of-function alleles and their frequencies vary among racial groups (average multiethnic allele frequencies of ~18 and ~2%, respectively).4–6 In contrast to these and other variant CYP2C19 alleles with decreased enzyme activity (*4–*8), the *17 promoter variant allele (c.−806C>T; rs12248560) is unique in that it results in increased activity as a consequence of enhanced transcription.7,8 Based on identified CYP2C19 genotypes, individuals can be categorized into four CYP2C19 metabolizer phenotypes:9–11 ultrarapid (*1/*17 or *17/*17), extensive (*1/*1), intermediate (deficient allele heterozygote) and poor (deficient allele compound heterozygote or homozygote).
CYP2C19 is involved in the metabolism of various clinically important drugs, including some antiulcer medications, antidepressants, β-adrenoceptor blockers and the anticonvulsant drug S-mephenytoin.4,12 However, recent attention has focused on its role in the bioactivation of the antiplatelet agent clopidogrel,13,14 which is commonly prescribed in patients with acute coronary syndrome and those undergoing percutaneous coronary intervention. Importantly, carriers of CYP2C19 loss-of-function alleles have reduced responsiveness to clopidogrel with significantly increased risks of stent thrombosis, stroke, myocardial infarction and death following treatment.11,15–19 However, these adverse effects may be reduced among patients with certain indications such as non-ST-segment elevation acute coronary syndrome and atrial fibrillation.20 The growing body of literature implicating CYP2C19 loss-offunction alleles in adverse clopidogrel responses prompted the United States Food and Drug Administration to implement a recent boxed warning on the clopidogrel label describing the relationship between CYP2C19 pharmacogenetics and drug response, particularly noting the diminished effectiveness in poor metabolizers.21 Additionally, these studies offer a compelling rationale for pharmacogenetic-guided anti-platelet drug selection and/or dosing in this patient population.22–26
In contrast to the deficient activity of CYP2C19 alleles, CYP2C19*17 recently has been associated with an enhanced platelet response to clopidogrel with an accompanying slight increase in bleeding risk during treatment.27–30 CYP2C19*17 appears to be common in Caucasians and less so in Asian populations (allele frequencies of~20 and~5%, respectively);31 however, its frequency has not been adequately established in other racial and ethnic groups. Moreover, recent studies have suggested that previously reported CYP2C19 allele, genotype and metabolizer phenotype frequencies should be reassessed in light of the recent identification of the *17 allele.32
We previously reported the CYP2C19*2–*8 allele frequencies for the Ashkenazi Jewish (AJ) population and found that they were similar to those in other European Caucasian populations.33 Notably, the *4 (c.1A>G; rs28399504) null mutation frequency was found to be ~2%, slightly higher than its reported frequencies in Caucasian and Hong Kong Chinese cohorts (~0.6%).34,35 To assess the impact of including *17 in AJ testing panels and to determine the population structure of additional variant CYP2C19 alleles in both the AJ and Sephardi Jewish (SJ) populations, we regenotyped the AJ individuals previously tested for CYP2C19*2–*8, for CYP2C19*2–*10, *12–*17 and *22, and compared these results with a large SJ cohort. Additionally, the ATP-binding cassette, subfamily B (MDR/TAP), member 1 (ABCB1) c.3435C>T (rs1045642) polymorphism recently associated with impaired clopidogrel absorption and adverse clinical response17,36–38 was interrogated in both populations. By employing an expanded CYP2C19 genotyping panel coupled with allele-specific cloning and sequencing, our study identified a novel haplotype, designated CYP2C19*4B, that has important pharmacogenetic implications for drug metabolism, particularly when clinically assessing CYP2C19 for clopidogrel responsiveness among patients initiating antiplatelet therapy. Moreover, genotyping CYP2C19*17 in the AJ and SJ populations significantly changed the frequency of extensive metabolizers by reclassifying ~30% as ultrarapid metabolizers. Importantly, when combining all of the studied CYP2C19 and ABCB1 variants, ~1 in 3 AJ and ~1 in 2 SJ individuals may be at increased risk for adverse cardiovascular responses to clopidogrel.
Peripheral blood samples were obtained with informed consent from self-reported unrelated AJ (n=250) and SJ (n=135) individuals from the greater New York metropolitan area undergoing routine carrier screening for Jewish genetic diseases.33,39,40 All personal identifiers were removed, and the samples were tested anonymously. Genomic DNA was isolated using the Puregene DNA Purification kit (Qiagen, Valencia, CA, USA) as per the manufacturer’s instructions.
The designations of all CYP2C19 alleles refer to those defined by the Cytochrome P450 Allele Nomenclature Committee (http://www.cypalleles.ki.se/cyp2c19.htm).3 Eleven variant CYP2C19 alleles (*2–*10, *13 and *17) were genotyped using the eSensor 2C19 Test (GenMark Diagnostics, Carlsbad, CA, USA) as per the manufacturer’s instructions. The wild-type CYP2C19*1 allele was assigned in the absence of other detectable variant alleles. Representative positive control samples for identified alleles were confirmed by bidirectional sequencing (see below).
Genotyping of CYP2C19*12, *14, *15, *16, *22 and ABCB1 c.3435C>T was performed using a multiplexed SNaPshot single base extension assay (Applied Biosystems, Carlsbad, CA, USA), whereby CYP2C19 exons 1, 4 and 9 were initially PCR amplified with exon 27 of ABCB1. Reactions were performed in 10 μl containing ~50 ng of DNA, 1× PCR buffer (Invitrogen, Carlsbad, CA, USA), 3.0mM MgCl2, 0.2mM of each dNTP, forward and reverse primers (CYP2C19 exon 1: 0.4 μM; exon 4: 0.5 μM; exon 9: 0.2 μM; ABCB1 exon 27: 0.4 μM; Supplementary Table S1) and 1.0 unit of Platinum Taq DNA Polymerase (Invitrogen). Amplification consisted of an initial denaturation step at 94 °C for 5 min followed by 38 amplification cycles (94 °C for 30 s, 58 °C for 30 s and 72 °C for 1 min) and a final incubation at 72 °C for 5 min. Amplicons were digested with 3.0 units of shrimp alkaline phosphatase and 2.0 units of Exonuclease I (both from USB Corporation, Cleveland, OH, USA). SNaPshot primer extension reactions were performed in 10 μl containing 1× SNaPshot Reaction Mix (Applied Biosystems), 0.2 μM of each allele-specific primer (Supplementary Table S1) and 4.0 μl of PCR product. Following the recommended thermal cycling, samples were treated with 1.0 unit of shrimp alkaline phosphatase, electrophoresed on an ABI Prism 310 Genetic Analyzer, and analyzed using GeneMarker software v1.85 (SoftGenetics, State College, PA, USA). Representative positive control samples for identified alleles were confirmed by bidirectional sequencing (see below).
Cloning and allele-specific sequencing of a 1.2-kb fragment encompassing CYP2C19*17 (c.−806C>T) and *4 (c.1A>G) was performed on samples that harbored both alleles as determined by targeted genotyping. An initial PCR was performed in 25 μl containing ~100 ng of DNA, 1× Pfu DNA polymerase reaction buffer (Agilent Technologies, La Jolla, CA, USA), 0.2mM of each dNTP, 0.4 μM of forward and reverse primers (Supplementary Table S1) and 2.5 units of PfuTurbo Hotstart DNA polymerase (Agilent Technologies). Amplification consisted of an initial denaturation step at 95 °C for 5 min followed by 35 amplification cycles (95 °C for 30 s, 58 °C for 30 s and 72 °C for 2 min) and a final incubation at 72 °C for 10 min. A 3′-deoxyadenosine over-hang was added to blunt-ended amplicons by incubation with 1.0 unit of Platinum Taq DNA Polymerase (Invitrogen) at 72 °C for 10 min. These products were purified and cloned into the pCR2.1-TOPO vector using the TOPO TA Cloning Kit (Invitrogen) as per the manufacturer’s instructions. For each sample, 6 to 10 colonies were propagated and bidirectionally sequenced using M13 and T7 vector-specific primers. All plasmid sequence data were analyzed using Mutation Surveyor software v3.30 (SoftGenetics).
CYP2C19 exon sequencing was performed by an initial PCR in 25 μl containing ~100 ng of DNA, 1× PCR buffer (Invitrogen), 1.5mM MgCl2, 0.2mM of each dNTP, 0.4 μM of forward and reverse primers and 1.0 unit of Platinum Taq DNA Polymerase (Invitrogen). Amplification consisted of an initial denaturation step at 95 °C for 5 min followed by 35 amplification cycles (95 °C for 30 s, Ta (Supplementary Table S1) for 30 s and 72 °C for 1 min) and a final incubation at 72 °C for 10 min. Amplicons were digested with 2.5 units of shrimp alkaline phosphatase and 10.0 units of Exonuclease I (both from USB Corporation), and bidirectionally sequenced using amplification primers and/or nested sequencing primers as noted in Supplementary Table S1. In addition, an upstream CYP2C19 promoter fragment encompassing the c.−3402C>T (rs11188072) allele common to *17 was sequenced in selected samples as described above. All bidirectional sequence data were analyzed using Mutation Surveyor software v3.30 (SoftGenetics).
To assess the impact of including *17 in AJ testing panels and to determine the population structure of variant CYP2C19 alleles in the AJ population, 250 AJ individuals previously tested for CYP2C19*2–*8 were re-genotyped for CYP2C19*2–*10, *12, *13, *16, *17 and *22. The initial CYP2C19*1, *2, *4 and *17 allele frequencies detected among the 250 AJ individuals were 0.619, 0.145, 0.020 and 0.216, respectively; CYP2C19*3, *5–*10, *12, *13, *16 and *22 were not detected among the 500 AJ alleles tested. Of note, complete concordance for *2 and *4 was observed between the previously reported genotyping platform33 and that used in the current study. However, interestingly, four subjects were identified with three alleles each: two with *2/ *4/*17 and two with *4/*17/*17, inappropriately resulting in 504 total alleles among the 250 individuals. Moreover, the genotype frequencies deviated from Hardy–Weinberg equilibrium as no AJ individuals were identified with a *1/*4 genotype, even though the *1 and *4 allele frequencies predicted that 2.5% of the population (approximately six individuals in 250) should have the genotype. In fact, all other *4 heterozygotes (n=6) were also heterozygous for *17, resulting in a *4/*17 genotype frequency (2.4%) greater than what was predicted (0.9%). Taken together, these results suggested that CYP2C19*4, a loss-of-function poor metabolizer allele,34 occurred on a *17 background in the AJ population.
To confirm that CYP2C19*4 and *17 were in cis among the *4 carriers in our AJ cohort, a 1.2-kb promoter/exon 1 amplicon encompassing the *4 and *17 variants was PCR amplified from each informative carrier and cloned into the pCR2.1-TOPO vector. Six to 10 colonies from the eight *4 and *17 heterozygotes were propagated and bidirectionally sequenced, and in all individuals, the *4 and *17 variants were on the same CYP2C19 allele. To further define the coding sequence of this novel allele, all nine CYP2C19 exons and exon/intron boundaries were bidirectionally sequenced, which identified two additional coding variants (c.99C>T (rs17885098) and c.991A>G (rs3758581)) consistent with the previously defined *4 haplotype.3,4,34
Detailed sequencing results of all identified single-nucleotide polymorphisms, including intronic variants, among the *4 carriers are summarized in Table 1. The identified haplotype of the novel allele (NG_008384.1:c.[−3402T; −806T; 1G; 99T; 991G]) was submitted for review by the Cytochrome P450 Allele Nomenclature Committee2,3 and subsequently named CYP2C19*4B. All sequencing and genotyping data were correlated with known CYP2C19 haplotypes3 and the identified diplotypes among the *4B carriers, including their coding region variants, are illustrated in Figure 1.
Interestingly, one *1/*4B AJ individual was found by sequencing to be heterozygous for the CYP2C19*15 (p.I19L; rs17882687) allele, suggesting a *4B/*15 genotype (Subject: AJ 8; Table 1; Figure 1). The unanticipated identification of *15 prompted its subsequent inclusion into the multiplexed SNaPshot assay, in addition to the neighboring CYP2C19*14 (p.L17P; rs55752064) allele (see Materials and methods).
After accounting for the novel CYP2C19*4B allele and incorporating *14 and *15 into the genotyping panel, the revised AJ CYP2C19 allele and genotype frequencies were in Hardy–Weinberg equilibrium and are summarized in Tables 2 and and3.3. The CYP2C19*1, *2, *4B, *15 and *17 allele frequencies among the 250 AJ individuals were 0.632, 0.146, 0.020, 0.004 and 0.198, respectively. Based on their genotypes, the reassigned metabolizer phenotypes in the AJ population were distributed as ultrarapid (29%), extensive (42%), intermediate (18%) and poor (4%) metabolizers.
In addition to the AJ cohort, 135 individuals of SJ descent were genotyped for all 16 variant CYP2C19 alleles. Interestingly, in contrast to the AJ population, both the original CYP2C19*4A allele (without c.−806C>T) and the novel *4B allele (with c.−806C>T), as well as the *8 (p.W120R; rs41291556) loss-of-function allele, were detected in the SJ population. Like the AJ CYP2C19*4B carriers, all informative SJ carriers were confirmed by allele-specific cloning and sequencing (Table 1; Figure 1). The CYP2C19*1, *2, *4A, *4B, *8, *15 and *17 allele frequencies among the SJ individuals were 0.722, 0.096, 0.004, 0.011, 0.004, 0.004 and 0.159, respectively (Table 2). Although additional variant alleles were detected in the SJ compared with the AJ, the CYP2C19 allele and genotype frequencies did not reach statistical significance between the two populations (P=0.07 and 0.16, respectively). Based on their genotypes, the assigned metabolizer phenotypes in the SJ population were distributed as ultrarapid (27%), extensive (53%), intermediate (16%) and poor (3%) metabolizers (Table 3).
Several studies have found that homozygous carriers of the ABCB1 c.3435C>T (p.I1145I) synonymous variant have higher rates of adverse cardiovascular events than c.3435C homozygotes during clopidogrel therapy, which was independent from and compounded by CYP2C19 loss-of-function alleles.17,36–38 However, conflicting data have been reported regarding which allele (c.3435C or c.3435T) is associated with the increased risk.41 Supplementary Tables S2 and S3 summarize the identified ABCB1 c.3435C>T allele and genotype frequencies, which were statistically different between the AJ and SJ populations (P=0.0002 and 0.0005, respectively). Of note, ~1 in 13 (8.4%) AJ and ~1 in 5 (19.3%) SJ individuals were c.3435T homozygotes.
Figure 2 and Supplementary Table S4 summarize the combined CYP2C19 metabolizer phenotype and ABCB1 genotype frequency profiles for the AJ and SJ, which were statistically different between the two populations (P<0.0001). Of note, when combining the CYP2C19 and ABCB1 frequency data, ~60–65% of AJ and SJ individuals carried a CYP2C19 allele (*2, *4A, *4B, *8, *17) and/or ABCB1 genotype (c.3435T/T) that could influence their response to clopidogrel. However, increased risk for clopidogrel non-responsiveness and adverse cardiovascular events has largely been reported among individuals carrying a CYP2C19 loss-of-function allele and/or those with an ABCB1 c.3435T/T genotype.17,38 Categorizing the AJ and SJ subjects using these specific criteria indicated that 34% of AJ (~1 in 3) and 43% of SJ (~1 in 2) individuals carried CYP2C19 and ABCB1 genotypes that conferred an increased risk for clopidogrel nonresponsiveness and/or adverse effects (P<0.0001; Figure 3; Supplementary Table S5).
The growing body of literature on clopidogrel pharmacogenetics and the paucity of frequency data for variant CYP2C19 alleles beyond the commonly studied *2 and *3 loss-of-function alleles prompted our population structure investigation of CYP2C19 and ABCB1 in the AJ and SJ populations. Importantly, our study identified a novel CYP2C19 allele, designated CYP2C19*4B, in both Jewish ethnicities that has significant pharmacogenetic implications for drug metabolism, particularly when clinically assessing CYP2C19 for clopidogrel responsiveness. The haplotype of the CYP2C19*4B allele (NG_008384.1: c.[−3402T; −806T; 1G; 99T; 991G]) is a combination of two previously identified alleles, *4 (NG_008384.1:c.[1G; 99T; 991G]) and *17 (NG_008384.1:c.[−3402T; −806T; 99T; 991G]), each with opposing phenotypic consequences. CYP2C19*4 is a loss-of-function ‘poor metabolizer’ allele and *17 is an increased activity ‘ultrarapid metabolizer’ allele.7,8,34
The c.1A>G mutation of CYP2C19*4 abolishes the ATG initiation codon and was originally identified by sequencing CYP2C19 in Caucasians who were poor mephenytoin metabolizers.34 The defective activity of CYP2C19*4 was confirmed in vitro by its lack of recombinant protein expression in yeast and failure to translate CYP2C19 peptides in a coupled transcription/translation assay. However, CYP2C19*4 was able to transcribe mRNA in vitro, indicating that the failure of CYP2C19 protein expression was at the level of translation.34 Consequently, CYP2C19*4B would also fail to produce CYP2C19 protein, regardless of the increased transcriptional capacity mediated by the upstream c.–806C>T promoter variant.
Although the frequency of CYP2C19*4B was only ~2% in the AJ and SJ populations, its inclusion in genotyping panels is important, particularly given the high frequency of CYP2C19*17 in Caucasian populations and the increasing interest in adding *17 to panels that include the commonly tested *2 and *3 alleles. For example, testing for *2, *3 and *17 without *4B would have misclassified ~1 in 25 AJ individuals, including eight intermediate metabolizers who would have been incorrectly classified as ultrarapid metabolizers and two poor metabolizers who would have been incorrectly classified as intermediate metabolizers. Testing for *2, *3 and *17 without *4B in the SJ would have misclassified ~1 in 45 individuals, including one intermediate metabolizer who would have been incorrectly classified as an ultrarapid metabolizer and two poor metabolizers who would have been incorrectly classified as intermediate metabolizers.
Recently, CYP2C19*17 has been reported to be in linkage disequilibrium with the neighboring CYP2C8*1 and CYP2C9*1 wild-type alleles among Nordic individuals.42 The identified linkage disequilibrium between *17 and *4 (comprising the *4B allele) in the AJ and SJ populations was not observed in the Nordic cohort nor was *4B identified in a recent CYP2C19 sequencing study of Han Chinese individuals,43 together suggesting that *4B may be specific to Jewish subpopulations. However, preliminary studies in our laboratory have identified the *4B allele in both Caucasians and Hispanics.
The inclusion of CYP2C19*17 alone also significantly altered the frequencies of the predicted metabolizer phenotypes. For example, CYP2C19*17 changed the frequency of extensive metabolizers (*1/*1) in both populations from ~70 to ~40%, with ~30% of individuals reclassified as ultrarapid metabolizers (*1/*17 or *17/*17). Very recently, Sibbing et al.29 assessed the impact of CYP2C19*2 and *17 on clopidogrel responsiveness among patients undergoing clopidogrel maintenance therapy. They determined that individuals with a *1/*17 or *17/*17 genotype had lower residual ADP-induced platelet aggregation compared with wild-type individuals, suggesting that the *17 allele resulted in the enhanced bioactivation of clopidogrel.29 In addition, they reported that individuals with a *2/*17 genotype had higher residual platelet aggregation compared with wildtype individuals, but less than those carrying *2 without *17, suggesting that the ultrarapid *17 allele could not completely compensate for a heterozygous null allele. Although this implies that individuals with a *2/*17 genotype (or other loss-of-function allele/*17 compound heterozygotes) are intermediate metabolizers, in the absence of independent validation these genotypes were provisionally classified as having an ‘unknown’ metabolizer phenotype in our current study.
The CYP2C19*15 variant allele was originally identified in an African population;44 however, its phenotypic consequence is unknown. Therefore, the detected *1/*15, *2/*15 and *4B/*15 individuals in our study were also classified as having an ‘unknown’ metabolizer phenotype (Table 3). Although missense alterations at the amino-terminal region of CYP450 enzymes are often considered benign,44 aberrant splicing, protein misfolding and/or expression alterations mediated by the CYP2C19*15 allele could not be ruled out. As more ethnic subpopulation-specific CYP2C19 alleles are identified and commercial genotyping panels are expanded, appropriate clinical assessment of unique genotype combinations will be critical to assign the appropriate metabolizer phenotypes and for the clinical application of expanded CYP2C19 pharmacogenetic panels, particularly for clopidogrel responsiveness testing.
The ABCB1 gene encodes the P-glycoprotein membrane efflux transporter, which is involved in the intestinal absorption and bioavailability of clopidogrel. Previously, the c.3435C>T allele was associated with duodenal protein expression and lower bioavailability of established P-glycoprotein substrates,45,46 suggesting that ABCB1 c.3435C>T might influence clopidogrel efflux and drug bioavailability. 36 However, given the conflicting data available on c.3435C>T and P-glycoprotein expression,45–47 expanded studies are warranted to identify ABCB1 haplotypes and assess their relationship to gene expression. Despite this discrepancy, some large clinical studies found that c.3435T/ T patients had a higher rate of adverse cardiovascular events than c.3435C homozygotes during clopidogrel therapy, which was independent from and compounded by CYP2C19 loss-of-function alleles.17,38 Our study identified a high frequency of c.3435C>T homozygotes in both Jewish populations (~10–20%), and when combined with the CYP2C19 data, the majority (~60–65%) of AJ and SJ individuals harbored a CYP2C19 and/or ABCB1 genotype that could influence their response to clopidogrel.
In conclusion, our study identified a novel allele in the AJ and SJ populations, designated CYP2C19*4B, that is a variant of the *4 loss-of-function allele occurring on a *17 increased transcriptional background. The high frequency of *17 without the *4 variant in our cohorts suggests that the c.−806C>T (*17) promoter variant predates the initial occurrence of the c.1A>G (*4) allele. Three other important conclusions can be made from our population structure study: CYP2C19*4B would significantly alter the interpretation of CYP2C19 genotyping when testing for *17 without the *4 mutation; inclusion of CYP2C19*17 in our genotyping panel significantly changed the frequency of extensive metabolizers in the AJ and SJ cohorts; and, when combining CYP2C19 and ABCB1 genotypes, ~1 in 3 AJ and ~1 in 2 SJ individuals could have an increased risk for an adverse response to the commonly prescribed antiplatelet prodrug clopidogrel. Taken together, these data underscore the importance of including both *4B and *17, in addition to the commonly tested *2 and *3 alleles, when clinically assessing CYP2C19 for pharmacogenetic-guided dosing, particularly when testing for clopidogrel responsiveness among patients initiating antiplatelet therapy. Moreover, these data suggest that additional prospective clinical studies on clopidogrel response are warranted that include the common ABCB1 c.3435C>T polymorphism in addition to variant CYP2C19 alleles.
We thank Dr Joseph M Sweeny, Mount Sinai Medical Center, New York, and Dr Sarah Sim, Karolinska Institutet, Stockholm, Sweden, for critical reading of the manuscript. This research was supported in part by a research training grant (5 T32 GM082773) and a grant (UL1RR029887) for the Mount Sinai Institutes for Clinical and Translational Sciences from the National Center for Research Resources, National Institutes of Health. eSensor 2C19 Test reagents for this study were supplied by GenMark Diagnostics (Carlsbad, CA, USA).
Conflict of interest
The authors declare no conflict of interest.