Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Dec 1.
Published in final edited form as: Circ Genom Precis Med. 2021 Oct 21;14(6):e003353. doi: 10.1161/CIRCGEN.121.003353

Genotype-Guided P2Y12 Inhibitor Therapy After Percutaneous Coronary Intervention: A Bayesian Analysis

Vibhu Parcha 1, Brittain F Heindl 1, Peng Li 2, Rajat Kalra 3, Nita A Limdi 4,5, Naveen L Pereira 6,7,8, Garima Arora 1, Pankaj Arora 1,9
PMCID: PMC8692353  NIHMSID: NIHMS1746411  PMID: 34670396

Abstract

Background:

Among patients receiving percutaneous coronary intervention (PCI), the role of a genotype-guided approach for antiplatelet therapy compared with usual care is unclear. We conducted a Bayesian analysis of the entire TAILOR-PCI randomized clinical trial (RCT) population to evaluate the effect of the genotype-guided antiplatelet therapy post-PCI compared with the usual care on the risk of major adverse cardiovascular events (MACE).

Methods:

The primary outcome for our study was the composite of MACE (myocardial infarction [MI], stroke, and cardiovascular death). Secondary outcomes included cardiovascular death, stroke, MI, stent thrombosis, and major/minor bleeding. Bayesian modeling was used to estimate the probability of clinical benefit of genotype-guided therapy using 1) non-informative priors (i.e., analyzing the TAILOR-PCI trial) and 2) informative priors derived from the ADAPT, POPular Genetics, IAC-PCI, and PHARMCLO trials (i.e., analyzing TAILOR-PCI trial in the context of prior evidence). Risk ratio (RR: ratio of cumulative outcome incidence between genotype-guided and conventional therapy group) and 95% Credible Interval (CrI) were estimated for the study outcomes, and probability estimates for RR<1 were computed.

Results:

Using non-informative priors, in TAILOR-PCI the RR for MACE was 0.78 (95%CrI:0.55–1.07) in genotype-guided therapy after PCI and the probability of RR<1 was 94%. Using non-informative priors, the probability of RR<1 for cardiovascular death (RR:0.95 [95%CrI: 0.52–1.74]), stroke (RR:0.68 [95%CrI:0.44–1.06]), MI (RR:0.84 [95%CrI:0.37–1.89]), stent thrombosis (RR:0.75 [95%CrI:0.37–1.45]), and major or minor bleeding (RR:1.22 [95%CrI: 0.84–1.77]) were 57%, 96%, 67%, 94%, and 15%, respectively. Using informative priors, the posterior probability of RR<1 for MACE, from genotype-guided therapy, was 99% (RR:0.69 [95%CrI:0.57–0.84]). Using informative priors, the posterior probability of RR<1 for cardiovascular death (RR:0.86 [95%CrI:0.61–1.19]), stroke (RR: 0.69 [95% CrI:0.48–0.99]), MI (RR:0.56 [95%CrI:0.40–0.78]), stent thrombosis (RR:0.59 [95% CrI:0.38–0.94]), and major or minor bleeding (RR:0.84 [95%CrI:0.70–0.99]) were 81%, 99%, 99%, 99%, and 99%, respectively.

Conclusions:

Bayesian analysis of the TAILOR-PCI trial provides clinically meaningful data on the posterior probability of reducing MACE using genotype-guided P2Y12 inhibitor therapy after PCI.

Keywords: Bayesian Analysis, Myocardial Infarction, Percutaneous Coronary Intervention, Pharmacogenomics, Precision Medicine

Introduction

The cytochrome P450, family 2, subgroup C, polypeptide 19 (CYP2C19) hepatic enzyme converts the pro-drug clopidogrel to an active form.1, 2 Retrospective, and nonrandomized studies have described an association between CYP2C19 loss-of-function (LOF) alleles (most common being CYP2C19*2) and increased rates of adverse cardiovascular (CV) events after percutaneous coronary intervention (PCI), while being treated with clopidogrel.3–9 Hence, assessing patient genotype at the time of PCI and subsequently allocating antiplatelet therapy accordingly may reduce adverse CV events post PCI. However, the prospective randomized clinical trials of genotype-guided antiplatelet therapy approach have been inconclusive, and the major cardiology society guidelines have not recommended routine CYP2C19 testing.10 Randomized trials comparing genotype-guided antiplatelet therapy with usual care, had numerous limitations. These limitations included mid-study design changes due to updated antiplatelet guidelines,11 uncertified genotyping equipment,12 as well as design limitations such as using atypical treatment strategies13 or lacking a standard treatment protocol.14

The TAILOR-PCI trial was designed to overcome these limitations and to determine if there is a benefit for genotype-guided antiplatelet therapy based on CYP2C19-genotype testing at the time of PCI.15 The trial randomized 5,276 patients after PCI to either a CYP2C19 genotype-guided therapy group (if LOF alleles were present, patients were placed on ticagrelor, rather than clopidogrel) or a conventional therapy group (clopidogrel only) for 12 months of follow-up.15 The primary composite endpoint (myocardial infarction, stroke, stent thrombosis, and severe recurrent ischemia) occurred in 4.0% of the genotype-guided group and 5.9% of the conventional group (HR, 0.66; P=0.06), but this difference did not meet the predetermined 50% reduction in cardiovascular events necessary to reach statistical ‘significance.’ In multiple events-per-person analysis, a ‘significant’ reduction in CV events was noted in those with CYP2C19 LOF variants treated with ticagrelor compared with clopidogrel. Notably, the TAILOR-PCI primary analysis was performed in ~35% of the randomized cohort and was limited to individuals who had CYP2C19 LOF variants in the genotype-guided therapy and conventional therapy arms (i.e., did not include those without CYP2C19 LOF variants). The primary analysis did not compare the genotype-guided approach with the conventional therapy in the whole cohort. Furthermore, the escalation and de-escalation of antiplatelet therapy based on the availability of CYP2C19 LOF variant status information, the clinical risk-profile, and cost of therapy is becoming increasingly common in clinical care.16, 17

The Bayesian approach allows for the integration of prior information into newly obtained data to obtain a final quantitative summary (informed posterior probability).18–22 A Bayesian analysis of TAILOR-PCI, in the context of the previous clinical trials (priors), may offer beneficial information, in terms of probability estimate of the effect sizes, to practicing clinicians. We present the results of a Bayesian analysis of the TAILOR-PCI trial, first using non-informative priors and subsequently with priors derived from previous randomized controlled trials, to determine the informed posterior probability of benefit for the combined endpoint of MACE (CV death, stroke, and myocardial infarction [MI]), as well as individually for each component of MACE, stent thrombosis, and major or minor bleeding. We compared these outcomes among all participants of the TAILOR-PCI study to provide an updated summary of the RCT evidence on the genotype-guided antiplatelet therapy.

Methods

All aggregate data used in this study are derived from the respective study publications and are available publicly.12–15 The current study was approved by the University of Alabama at Birmingham Institutional Review Board. The full study methods are available as Supplementary Methods.

Results

Bayesian Analysis of the Study Outcomes Using Non-informative Priors (Beta [1,1])

In the TAILOR-PCI all randomized cohort, 62 patients in the genotype-guided therapy group developed MACE (2.3%), compared with 80 patients in the conventional therapy group (3.0%) (Supplementary Table I). Using non-informative priors, the risk difference for MACE was −0.7% (95% CrI, −1.6 to 0.2) in the genotype-guided therapy group compared with the conventional therapy group (Table 1). Using non-informative priors, the risk ratio for MACE was 0.78 (95% CrI, 0.56–1.07) between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 for MACE, in the genotype-guided therapy arm, was 94% (Supplementary Table II).

Table 1.

Risk Ratio and Risk Difference for Study Outcomes Determined Using Non-informative Priors

Study Outcomes Beta (1,1) Beta (0,0) Beta (10,10)
Risk Ratio (95% CrI) Risk Ratio (95% CrI) Risk Ratio (95% CrI)
MACE 0.78 (0.56–1.07) 0.77 (0.56–1.07) 0.80 (0.59–1.08)
 CV Death 0.95 (0.52–1.74) 0.95 (0.51–1.76) 0.96 (0.59–1.60)
 Stroke 0.68 (0.43–1.06) 0.68 (0.43–1.06) 0.73 (0.49–1.09)
 MI 0.84 (0.37–1.89) 0.83 (0.35–1.94) 0.90 (0.49–1.67)
Stent Thrombosis 0.75 (0.37–1.45) 0.73 (0.36–1.46) 0.82 (0.48–1.42)
Bleeding * 1.22 (0.84–1.77) 1.22 (0.84–1.77) 1.18 (0.84–1.67)
Study Outcomes Beta (1,1) Beta (0,0) Beta (10,10)
Risk Difference % (95% CrI) Risk Difference % (95% CrI) Risk Difference % (95% CrI)
MACE −0.7 (−1.6 to 0.2) −0.7 (−1.6 to 0.2) −0.7 (−1.6 to 0.2)
 CV Death −0.0 (−0.5 to 0.4) −0.0 (−0.5 to 0.4) −0.0 (−0.6 to 0.5)
 Stroke −0.6 (−1.2 to 0.1) −0.6 (−1.2 to 0.1) −0.6 (−1.3 to 0.2)
 MI −0.1 (−0.4 to 0.3) −0.1 (−0.4 to 0.3) −0.1 (−0.6 to 0.4)
Stent Thrombosis −0.2 (−0.6 to 0.2) −0.2 (−0.6 to 0.2) −0.2 (−0.7 to 0.3)
Bleeding * 0.4 (−0.4 to 1.2) 0.4 (−0.4 to 1.2) 0.4 (−0.4 to 1.3)
Open in a new tab

Abbreviations: risk difference, absolute risk reduction; CrI, credible interval; CV, cardiovascular; MACE, composite of cardiovascular death, nonfatal stroke, and nonfatal myocardial infarction; MI, myocardial infarction.

Note: Risk ratio is defined as the ratio of cumulative incidence of the outcome in the genotype-guided treatment group to the cumulative incidence in the conventional therapy group. Risk difference is defined as the absolute difference in the cumulative incidence of outcome between the genotype-guided treatment group and the conventional therapy group. Beta(1,1), Beta(0,0) and Beta(10,10) are different non-informative priors. Beta(0,0) is a non-informative prior known as Haldane’s prior.

*

Bleeding represents both major and minor bleeding.

In the TAILOR-PCI all randomized cohort, there were 20 CV deaths in the genotype-guided therapy group (0.7%), compared with 21 patients in the conventional therapy group (0.8%). Using non-informative priors risk difference and risk ratio for CV death were 0.0% (95% CrI, −0.5 to 0.4) and 0.95 (95% CrI, 0.52–1.74), respectively between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 in favor of genotype-guided therapy was 56.5%.

In the TAILOR-PCI trial all randomized cohort, 32 patients had a stroke in the genotype-guided therapy group (1.2%), compared with 47 patients in the conventional therapy group (1.7%). The non-informed risk difference and risk ratio for incident stroke were −0.6% (95% CrI, −1.2 to 0.1%) and 0.68 (95% CrI, 0.43–1.06), respectively between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 for the genotype-guided therapy in regards to incident stroke was 95.6%.

Among patients from TAILOR-PCI all randomized cohort, there were 10 patients who had an MI in the genotype-guided therapy group (0.4%), compared with 12 patients in the conventional therapy group (0.5%). The non-informed risk difference and risk ratio for MI were −0.1% (95% CrI, −0.4 to 0.3) and 0.84 (95% CrI, 0.37 to 1.89) respectively between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 for the outcome of MI was 66.5%.

There were 14 patients that had a stent thrombosis in the genotype-guided therapy group (0.5%), compared with 19 patients in the conventional therapy group (0.7%) in the TAILOR-PCI all randomized cohort. The non-informed risk difference and risk ratio for stent thrombosis was −0.2% (95% CrI, −0.6 to 0.2) and 0.75 (95% CrI, 0.37–1.45), respectively between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 for the outcome of stent thrombosis was 88.7%.

In the TAILOR-PCI trial all randomized cohort, 61 patients had a major or minor bleed in the genotype-guided therapy group (2.3%), compared with 50 patients in the conventional therapy group (1.9%). The non-informed risk difference and risk ratio for major or minor bleeding was 0.4% (95% CrI, −0.4 to 1.2%) and 1.22 (95% CrI, 0.84–1.77), respectively, between groups. The non-informed posterior probability of risk difference<0% and risk ratio<1 for the outcome of major or minor bleeding was 14.9%.

Sensitivity Analyses Using Different Non-informative Priors

The sensitivity analyses using different non-informative prior distributions demonstrated a largely unchanged non-informative posterior probability of risk difference and risk ratio for the study outcomes (Table 1).

Identification of Priors for Outcome Analysis Using Informed Priors

Our prespecified search strategy yielded 1072 original records, from which 4 independent randomized controlled trials comparing genotype-guided therapy with a comparison control group underwent a full-text review. These were the POPular Genetics (Cost-effectiveness of Genotype Guided Treatment with Antiplatelet Drugs in STEMI Patients: Optimization of Treatment),11 PHARMCLO (Pharmacogenetics of Clopidogrel in Acute Coronary Syndromes),12 IAC-PCI (Individual Applications of Clopidogrel after Percutaneous Coronary Intervention),13 and ADAPT (Assessment of Prospective CYP2C19 Genotype Guided Dosing of Anti-Platelet Therapy in Percutaneous Coronary Intervention)14 trials. There were a total of 4480 patients identified, with 2240 patients each in the genotype-guided and conventional therapy arms. The number of events in each arm for the endpoints of MACE, CV death, stroke, MI, stent thrombosis, and major or minor bleeding for each trial are presented in Supplementary Table I.11–14

Bayesian Analysis of the Study Outcomes Using Informative Priors

The POPular Genetics, PHRAMCLO, IAC-PCI, and ADAPT trials reported 102 MACE events in the combined genotype-guided treatment groups (4.6%, N=2,240) and 156 MACE events in the combined comparator groups (6.9%, N=2,240). Using these combined results as the prior, and the likelihood of MACE from TAILOR-PCI results (all randomized cohort), the informed risk difference for MACE was lower by −0.8 (95% CrI, −1.6 to −0.1) in the genotype-guided therapy group. The informed risk ratio for MACE was 0.69 (95% CrI, 0.57–0.84) between groups. The informed posterior probability of risk difference<0% and risk ratio<1 for MACE was 96.7% and 99.9%, respectively (Figure 1–2). Table 2–3 depicts the probability of benefit in terms of MACE at various risk ratio and risk difference thresholds.

Figure 1.

Figure 1.

Open in a new tab

Posterior Probability Distribution for Risk Difference of the Study Outcomes for Genotype-Guided P2Y12 Inhibitor Therapy After Percutaneous Coronary Intervention

Figure 2.

Figure 2.

Open in a new tab

Posterior Probability Distribution for Risk Ratio of the Study Outcomes for Genotype-Guided P2Y12 Inhibitor Therapy After Percutaneous Coronary Intervention

Table 2.

Risk Ratio and Posterior Probability Determined Using Informative Priors

Outcome Risk Ratio (95% CrI) Posterior Probability
Prior TAILOR-PCI Likelihood Posterior RR<1 RR<0.9 RR<0.8
MACE 0.65 (0.51–0.82) 0.77 (0.56–1.07) 0.69 (0.57–0.84) 99.99% 99.88% 94.29%
 CV Death 0.82 (0.55–1.22) 0.95 (0.51–1.76) 0.86 (0.61–1.19) 81.28% 58.52% 32.34%
 Stroke 0.73 (0.38–1.38) 0.68 (0.43–1.06) 0.69 (0.48–0.99) 98.69% 92.89% 76.22%
 MI 0.52 (0.33–0.74) 0.83 (0.35–1.94) 0.56 (0.40–0.78) 99.99% 99.97% 99.18%
Stent Thrombosis 0.50 (0.27–0.92) 0.73 (0.36–1.46) 0.59 (0.38–0.94) 99.64% 97.71% 90.45%
Bleeding 0.75 (0.62–0.92) 1.22 (0.84–1.77) 0.84 (0.70–0.99) 98.55% 79.66% 29.99%
Open in a new tab

Abbreviations: CrI, credible interval; CV, cardiovascular; MACE, composite of cardiovascular death, nonfatal stroke, and nonfatal myocardial infarction; MI, myocardial infarction; RR, risk ratio.

Note: Prior: This is derived from the previous clinical trials. Posterior: Final effect size computed by combining the prior with data from TAILOR-PCI. Posterior Probability: Probability of effect size below a given threshold.

Table 3.

Risk Difference and Posterior Probability Determined Using Informative Priors

Outcome Risk Difference (95% CrI) Posterior Probability
Prior TAILOR-PCI Likelihood Posterior Risk Difference<0% Risk Difference<−1%
MACE −2.1 (−4.3 to 0.001) −0.7 (−1.6 to 0.2) −0.8 (−1.6 to −0.1) 96.69% 31.75%
 CV Death −0.03 (−1.5 to 0.8) −0.0 (−0.5 to 0.4) −0.1 (−0.5 to 0.4) 64.79% <0.01
 Stroke −0.3 (−0.8 to 0.3) −0.6 (−1.2 to 0.1) −0.4 (−0.8 to −0.03) 96.30% 0.10%
 MI −1.7% (−3.9 to 0.4) −0.1 (−0.4 to 0.3) −0.1 (−0.5 to 0.2) 75.00% <0.01%
Stent Thrombosis −0.8 (−1.6 to 0.1) −0.2 (−0.6 to 0.2) −0.3 (−0.7 to 0.1) 93.80% 0.20%
Bleeding −2.2 (−3.6 to −0.7) 0.4 (−0.4 to 1.2) −0.2 (−0.9 to 0.5) 70.73% 9.45%
Open in a new tab

Abbreviations: risk difference, Absolute Risk Reduction; CrI, Credible Interval; CV, cardiovascular; MACE, composite of cardiovascular death, nonfatal stroke, and nonfatal myocardial infarction; MI, myocardial infarction.

Note: Prior: This is derived from the previous clinical trials. Posterior: Final effect size computed by combining the prior with data from TAILOR-PCI. Posterior Probability: Probability of effect size below a given threshold.

The POPular Genetics, PHRAMCLO, IAC-PCI, and ADAPT trials reported 43 CV deaths in the combined genotype-guided treatment groups (1.9%), compared with 52 in the combined comparator groups (2.3%). The informed risk difference and risk ratio CV death were −0.1% (95% CrI, −0.5 to 0.4) and 0.86 (95% CrI, 0.62 to 1.19) respectively between groups. The informed posterior probability of risk difference<0% and risk ratio<1 was 64.8% and 81.3%, respectively.

The 4 trials reported 16 strokes in the combined genotype-guided treatment groups (0.7%), compared with 22 in the combined comparator groups (0.9%). Using these combined results as the prior, the informed risk difference and risk ratio for stroke were −0.4% (95% CrI, −0.8 to −0.03) and 0.69 (95% CrI, 0.48–0.99), respectively between groups. The informed posterior probabilities of risk difference<0% and risk ratio<1, for the outcome of stroke, were 96.3% and 98.7%, respectively.

The 4 trials reported 43 MIs in the combined genotype-guided treatment groups (1.9%) compared with 82 in the combined comparator groups (3.7%). The informed risk difference and risk ratio for MI was −0.1% (95% CrI, −0.5 to 0.2) and 0.56 (95% CrI, 0.40–0.78), respectively between groups. The informed posterior probabilities of risk difference<0 % and risk ratio<1, for the outcome of MI, were 75% and 99%, respectively.

The 4 trials reported 15 stent thromboses in the combined genotype-guided treatment groups (0.7%) compared with 30 cases in the combined comparator groups (1.3%). The informed risk difference and risk ratio for stent thrombosis were −0.3% (95% CrI, −0.7 to 0.1) and 0.59 (95% CrI, 0.37 to 0.94), respectively between groups respectively. The informed posterior probabilities of risk difference<0% and risk ratio<1, for the outcome of stent thrombosis, were 93.8% and 99.6%, respectively.

The 4 trials reported 160 major or minor bleeding events in the combined genotype-guided treatment groups (7.1%) compared with 213 in the combined comparator groups (9.5%). The informed risk difference and risk ratio for major and minor bleeding were −0.2% (95% CrI, −0.9 to 0.5) and 0.84 (95% CrI, 0.70–0.99), respectively, between groups. The informed posterior probabilities of risk difference<0% and risk ratio<1, for the outcome of major and minor bleeding, was 70.7% and 98.6%, respectively.

Sensitivity Analyses of TAILOR-PCI Study Outcomes

The additional analyses of the primary outcome (composite of CV death, MI, stroke, severe recurrent ischemia, and stent thrombosis) of the TAILOR-PCI trial (all randomized cohort) using the non-informative prior (Beta [1,1]) demonstrated a 93% probability of risk difference<0% and risk ratio<1 for the primary composite endpoint in those undergoing genotype-guided therapy after PCI. The risk ratio and risk difference for the primary outcome of the TAILOR-PCI using non-informative (Beta [1,1]) prior was 0.84 (95% CrI, 0.66–1.06) and −0.8% (95% CrI, −2.0 to 0.003), respectively.

Discussion

In this Bayesian reanalysis of the TAILOR-PCI trial, we observed that there is a 99.9% probability of risk ratio<1 for MACE in patients treated with genotype-guided P2Y12 therapy after PCI using informative priors from previous clinical trials. The informed posterior probability that the risk ratio is less than 1 in the genotype-guided therapy patients was 81.3% for CV death, 98.7% for stroke, 99.9% for MI, and 99.6% for stent thrombosis. In TAILOR-PCI, bleeding was higher in the genotype-guided therapy group, but it was lower in previous trials, resulting in an informed posterior probability of risk ratio<1 being 98.6%. The cumulative evidence indicates the benefit of genotype-guided antiplatelet therapy following PCI. However, the benefit of the genotype-guided approach may be smaller in terms of risk difference for ischemic events.

The original TAILOR-PCI trial results suggested an absolute risk difference for the primary composite endpoint of 1.8% (HR, 0.66; 95% CI, 0.43–1.02; P = .06), which did not meet the prespecified level of statistical significance for superiority.15 This result was contrary to those from multiple prior retrospectives and prospective studies, which suggested a strong benefit from genotype-guided antiplatelet therapy.5, 7, 8, 12, 13 The primary reason for the failure to meet statistical ‘significance’ was that the trial was powered for an anticipated 50% reduction in ischemic events in the treatment and control groups. This power estimate was chosen to ensure that genotype-guided therapy exceeded the benefit achieved from ticagrelor alone. Consistent with contemporary PCI trials, the number of CV events in the trial was low.23 A prespecified secondary analysis, allowing for the inclusion of multiple events, did show a ‘significant’ reduction in CV events in those with CYP2C19 LOF variants treated with ticagrelor compared with clopidogrel (HR, 0.60; 95% CI, 0.41–0.89; P = .01). Given these signals of benefit, the authors caution that the non-significant P-value inaccurately characterized the trial results. An adequately powered study with a higher event rate may have yielded a ‘significant’ result. Given the relatively short 12 month follow-up period in TAILOR-PCI, a follow-up study has been funded by the National Heart, Lung, and Blood Institute to accrue additional events for the assessment of long term events in the trial populations.24 Meanwhile, clinicians remain uncertain regarding advantage of genotyping for CYP2C19 LOF variants in patients undergoing PCI. Our study utilizes Bayesian analysis to provide a quantitative probability estimate of the effect size of benefit for multiple clinical endpoints between genotype-guided and conventional therapy. Our analyses found that while there was a strong relative risk reduction for MACE and its components with genotype-guided therapy, the absolute risk reduction was relatively low.

The availability of data from randomized clinical trials prior to TAILOR-PCI allowed us to incorporate previously acquired information to assess new clinical evidence on the benefit of a genotype-guided approach for post-PCI P2Y12 inhibitor therapy.20, 25, 26 The POPular Genetics Trial randomized patients with an ST-elevation MI to a genotype-guided group (CYP2C19*2 or *3 LOF allele testing, with de-escalation to clopidogrel if patients were LOF negative), or to usual care group (ticagrelor)(Supplementary Table III).11 The PHARMCLO Trial12 randomized patients hospitalized with acute coronary syndrome to a pharmacogenomics arm (ABCB1, CYP2C19*2 and *17 testing with the transition to prasugrel or ticagrelor based on a detailed algorithm) and a usual care arm (therapy chosen by clinician preference). The IAC-PCI trial randomized patients undergoing PCI to either personalized antiplatelet therapy, based on CYP2C19 phenotype (patients with 1 LOF allele received clopidogrel 150mg daily; cilostazol 100 mg twice daily was added for those with 2 LOF alleles), or conventional antiplatelet therapy (clopidogrel 75 mg daily).13 The ADAPT trial randomized participants at the time of PCI to point-of-care genotyping of CYP2C19*2, *3, and *17 alleles.14 For each of these 4 trials, the primary outcomes were substantially better for the genotype-guided therapy group than for the respective control groups. There may be heterogeneity in bleeding ascertainment across the trials used to derive the priors. Additionally, the bleeding events may have been higher due to the increased use of antiplatelet therapy irrespective of genotyping.

The original frequentist analyses prevented the derivation of probability estimates for smaller relative-risk effect sizes between the treatment arms. The traditional frequentist analyses are advantageous because of fewer subjective assumptions but allow for only a dichotomous positive or negative reporting of the trial results. Bayesian methods have several advantages by interpreting the trial data with probability estimates for the various effect sizes in the outcomes of interest. The probability estimates for the clinical outcomes may be updated as more real-world data accumulates. Notwithstanding its numerous advantages, Bayesian analyses are limited by the subjective ascertainment of the prior distribution and slow acceptance by medical researchers. This study highlights the value of including a prespecified Bayesian analysis in the design of randomized clinical trials to allow clinicians to derive probability estimates for outcomes of interest and improve patient-provider discussions of treatment benefits and risks.

Our study has several public health implications. The TAILOR-PCI trial was the largest study assessing genotype-guided post-PCI treatment in a rigorous manner, and it may serve as the principal trial to inform current guidelines. Our analysis of the TAILOR-PCI trial data, separately and in conjunction with the only prior randomized controlled trials, provides physicians and hospitals with the needed perspective on the value of CYP2C19 LOF allele assessment and subsequent implications of the choice of antiplatelet therapy. The current results support prior observational and clinical trial evidence that patients benefit from integrating a genomics-driven precision medicine approach into the clinical care of those undergoing PCI.26, 27 Our data indicates that adoption of the genotype-guided approach upstream at the time of PCI helps in reducing the risk of MACE post PCI. Contemporary cumulative evidence indicates that those with CYP2C19 LOF allele have a lower risk of ischemic events if treated with ticagrelor or prasugrel compared with clopidogrel.26 The TRANSLATE-ACS registry data indicate that up to 30% of patients undergo modification in antiplatelet therapy within a year of having acute coronary syndrome.28 Hence, the availability of the CYP2C19 variant status information may also aid in the safe escalation or de-escalation of P2Y12inhibitors. The results from our study align with the trend in clinical practice of switching P2Y12 inhibitors based on the patient’s risk-profile, medication cost, medical insurance coverage, and the CYP2C19 variant status.16, 17, 27 Given the high incidence and prevalence of ischemic heart disease in the US, the widespread adoption of the genotype-guided approach may ensure the best use of cardiovascular therapeutics through informed patient selection.

Our study had several limitations. This analysis was based on aggregate data derived from respective study publications and did not include patient-level data. The estimation of the probability of risk ratio<1 for the MACE outcome did not account for the time-to-event dimension of the endpoint and treated it as a composite endpoint. The study designs for the 4 prior randomized controlled trials, used to inform the prior distribution, were inconsistent. Each used different primary endpoints, different LOF alleles (though all included the most common CYP2C19*2), different treatment protocols for the intervention group, and different usual care treatments for the control group. The primary outcome of our study was not the same as the primary outcome for the TAILOR-PCI trial because data for the endpoint of severe recurrent ischemia was not available for each of the 4 prior studies. The primary outcome for our study was chosen as MACE, which is commonly used in cardiovascular clinical trials, to allow for a comparison across prior clinical trials. Due to lack of access to patient-level data, we were unable to assess the outcomes stratified by the coronary artery disease status (stable vs. acute coronary syndrome) before undergoing PCI. The TAILOR-PCI protocol required the conventional therapy group to receive clopidogrel, which may have impacted the comparison of study outcomes between the two arms. The previously conducted trials were completed in different racial populations and had differing study designs, contributing to heterogeneity in prior evidence. These racial differences may contribute to differences in the CYP2C19 LOF variant frequency, platelet reactivity, response to therapy, and access to advanced cardiovascular care. Hence, the risk ratio and risk difference for the study outcomes have been reported using both non-informative and informative priors to allow adequate interpretation of the study results. Despite these limitations, our analyses produced similar results as the reported in the original trial and allowed for a comprehensive accumulation of the evidence regarding genotype-guided antiplatelet therapy following PCI.

Conclusion

In this post-hoc Bayesian reanalysis of the TAILOR-PCI trial, the posterior probability of the risk ratio<1 for MACE among those receiving genotype-guided P2Y12 inhibitor therapy following PCI was 94% using a non-informative uniform prior distribution of Beta (1,1). Using informative priors from 4 previous clinical trials, the posterior probability of risk ratio<1 for MACE in the genotype-guided P2Y12 inhibitor therapy arm was 99%.

Supplementary Material

003353 - Supplemental Material

Acknowledgments:

We would like to thank the investigators and the participants of the TAILOR-PCI trial.

Source of Funding:

This work was supported by the National Institutes of Health Mentored Patient-Oriented Research Award [5K23HL146887-03] to Dr. Pankaj Arora. Dr. Pereira is supported by the National Institutes of Health Award 3U01HL128606-03S1 and U01 HL128606.

Non-Standard Abbreviations and Acronyms

CrI

Credible Interval

CYP2C19

Cytochrome P450, Family 2, Subgroup C, Polypeptide 19

LOF

Loss of function

MACE

Major adverse cardiovascular events

MI

Myocardial infarction

PCI

Percutaneous Coronary Intervention

TAILOR-PCI

Tailored Antiplatelet Initiation to Lessen Outcomes Due to Decreased Clopidogrel Response after Percutaneous Coronary Intervention

Footnotes

Conflicts of Interest: None of the authors had any conflicts of interest or financial disclosures to declare.

References

  • 1.Holmes DR Jr, Dehmer GJ, Kaul S, Leifer D, O’Gara PT and Stein CM. ACCF/AHA clopidogrel clinical alert: approaches to the FDA “Boxed Warning” a report of the american college of cardiology foundation task force on clinical expert consensus documents and the american heart association. Circulation. 2010;122:537–557. [DOI] [PubMed] [Google Scholar]
  • 2.Pereira NL and Weinshilboum RM. Cardiovascular pharmacogenomics and individualized drug therapy. Nat Rev Cardiol. 2009;6:632–638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Mega JL, Simon T, Collet J-P, Anderson JL, Antman EM, Bliden K, Cannon CP, Danchin N, Giusti B and Gurbel P. Reduced-function CYP2C19 genotype and risk of adverse clinical outcomes among patients treated with clopidogrel predominantly for PCI: a meta-analysis. JAMA. 2010;304:1821–1830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mega JL, Close SL, Wiviott SD, Shen L, Walker JR, Simon T, Antman EM, Braunwald E and Sabatine MS. Genetic variants in ABCB1 and CYP2C19 and cardiovascular outcomes after treatment with clopidogrel and prasugrel in the TRITON–TIMI 38 trial: a pharmacogenetic analysis. The Lancet. 2010;376:1312–1319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Collet J-P, Hulot J-S, Pena A, Villard E, Esteve J-B, Silvain J, Payot L, Brugier D, Cayla G and Beygui F. Cytochrome P450 2C19 polymorphism in young patients treated with clopidogrel after myocardial infarction: a cohort study. The Lancet. 2009;373:309–317. [DOI] [PubMed] [Google Scholar]
  • 6.Harmsze AM, van Werkum JW, ten Berg JM, Zwart B, Bouman HJ, Breet NJ, van’t Hof AW, Ruven HJ, Hackeng CM and Klungel OH. CYP2C19* 2 and CYP2C9* 3 alleles are associated with stent thrombosis: a case–control study. Eur Heart Jor. 2010;31:3046–3053. [DOI] [PubMed] [Google Scholar]
  • 7.Shuldiner AR, O’Connell JR, Bliden KP, Gandhi A, Ryan K, Horenstein RB, Damcott CM, Pakyz R, Tantry US and Gibson Q. Association of cytochrome P450 2C19 genotype with the antiplatelet effect and clinical efficacy of clopidogrel therapy. JAMA. 2009;302:849–857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Williams AK, Klein MD, Martin J, Weck KE, Rossi JS, Stouffer GA and Lee CR. CYP2C19 genotype-guided antiplatelet therapy and 30-day outcomes after percutaneous coronary intervention. Circ Genom Precis Med. 2019;12:e002441. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Simon T, Verstuyft C, Mary-Krause M, Quteineh L, Drouet E, Méneveau N, Steg PG, Ferrières J, Danchin N and Becquemont L. Genetic determinants of response to clopidogrel and cardiovascular events. N Engl J Med. 2009;360:363–375. [DOI] [PubMed] [Google Scholar]
  • 10.Sibbing D, Aradi D, Alexopoulos D, Ten Berg J, Bhatt DL, Bonello L, Collet J-P, Cuisset T, Franchi F and Gross L. Updated expert consensus statement on platelet function and genetic testing for guiding P2Y12 receptor inhibitor treatment in percutaneous coronary intervention. JACC Cardiovas Interv. 2019;12:1521–1537. [DOI] [PubMed] [Google Scholar]
  • 11.Claassens DM, Vos GJ, Bergmeijer TO, Hermanides RS, Van’t Hof AW, Van Der Harst P, Barbato E, Morisco C, Tjon Joe Gin RM and Asselbergs FW. A genotype-guided strategy for oral P2Y12 inhibitors in primary PCI. N Engl J Med. 2019;381:1621–1631. [DOI] [PubMed] [Google Scholar]
  • 12.Notarangelo FM, Maglietta G, Bevilacqua P, Cereda M, Merlini PA, Villani GQ, Moruzzi P, Patrizi G, Tagliazucchi GM and Crocamo A. Pharmacogenomic approach to selecting antiplatelet therapy in patients with acute coronary syndromes: the PHARMCLO trial. J Am Coll Cardiol. 2018;71:1869–1877. [DOI] [PubMed] [Google Scholar]
  • 13.Xie X, Ma Y-T, Yang Y-N, Li X-M, Zheng Y-Y, Ma X, Fu Z-Y, Li Y, Yu Z-X and Chen Y. Personalized antiplatelet therapy according to CYP2C19 genotype after percutaneous coronary intervention: a randomized control trial. Int J Cardiol. 2013;168:3736–3740. [DOI] [PubMed] [Google Scholar]
  • 14.Tuteja S, Glick H, Matthai W, Nachamkin I, Nathan A, Monono K, Carcuffe C, Maslowski K, Chang G and Kobayashi T. Prospective CYP2C19 genotyping to guide antiplatelet therapy following percutaneous coronary intervention: a pragmatic randomized clinical trial. Circ Genom Precis Med. 2020;13:e002640. [DOI] [PubMed] [Google Scholar]
  • 15.Pereira NL, Farkouh ME, So D, Lennon R, Geller N, Mathew V, Bell M, Bae J-H, Jeong MH and Chavez I. Effect of genotype-guided oral P2Y12 inhibitor selection vs conventional clopidogrel therapy on ischemic outcomes after percutaneous coronary intervention: the TAILOR-PCI randomized clinical trial. JAMA. 2020;324:761–771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Martin J, Williams AK, Klein MD, Sriramoju VB, Madan S, Rossi JS, Clarke M, Cicci JD, Cavallari LH, Weck KE, et al. Frequency and clinical outcomes of CYP2C19 genotype-guided escalation and de-escalation of antiplatelet therapy in a real-world clinical setting. Genet Med. 2020;22:160–169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Cavallari LH and Lee CR. A case for genotype-guided de-escalation of antiplatelet therapy after percutaneous coronary angioplasty. Future Cardiol. 2019;15:251–254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Quintana M, Viele K and Lewis RJ. Bayesian analysis: using prior information to interpret the results of clinical trials. JAMA. 2017;318:1605–1606. [DOI] [PubMed] [Google Scholar]
  • 19.Wijeysundera DN, Austin PC, Hux JE, Beattie WS and Laupacis A. Bayesian statistical inference enhances the interpretation of contemporary randomized controlled trials. J Clin Epidemiol. 2009;62:13–21. e5. [DOI] [PubMed] [Google Scholar]
  • 20.Goligher EC, Tomlinson G, Hajage D, Wijeysundera DN, Fan E, Jüni P, Brodie D, Slutsky AS and Combes A. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc Bayesian analysis of a randomized clinical trial. JAMA. 2018;320:2251–2259. [DOI] [PubMed] [Google Scholar]
  • 21.Brophy JM. Bayesian Interpretation of the EXCEL Trial and Other Randomized Clinical Trials of Left Main Coronary Artery Revascularization. JAMA Intern Med. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Lewis RJ and Angus DC. Time for Clinicians to Embrace Their Inner Bayesian?: Reanalysis of Results of a Clinical Trial of Extracorporeal Membrane Oxygenation. JAMA. 2018;320:2208–2210. [DOI] [PubMed] [Google Scholar]
  • 23.Moliterno DJ, Smyth SS and Abdel-Latif A. CYP2C19 Genotyping to Guide Antiplatelet Therapy After Percutaneous Coronary Interventions: One Size Rarely Fits All. JAMA. 2020;324:747–749. [DOI] [PubMed] [Google Scholar]
  • 24.Pereira NL, Avram R, So DY, Iturriaga E, Byrne J, Lennon RJ, Murthy V, Geller N, Goodman SG and Rihal C. Rationale and Design of the TAILOR-PCI Digital Study: Transitioning a Randomized Controlled Trial to a Digital Registry. Am Heart Jor. 2021; 232:84–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Freedman L Bayesian statistical methods. BMJ 1996;313:569. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Pereira NL, Rihal C, Lennon R, Marcus G, Shrivastava S, Bell MR, So D, Geller N, Goodman SG, Hasan A, et al. Effect of CYP2C19 Genotype on Ischemic Outcomes During Oral P2Y12 Inhibitor Therapy. JACC Cardiovasc Interv 2021;14(7):739–750. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Cavallari LH, Lee CR, Beitelshees AL, Cooper-DeHoff RM, Duarte JD, Voora D, Kimmel SE, McDonough CW, Gong Y, Dave CV, et al. Multisite Investigation of Outcomes With Implementation of CYP2C19 Genotype-Guided Antiplatelet Therapy After Percutaneous Coronary Intervention. JACC Cardiovasc Interv. 2018;11:181–191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Zettler ME, Peterson ED, McCoy LA, Effron MB, Anstrom KJ, Henry TD, Baker BA, Messenger JC, Cohen DJ, Wang TY et al. Switching of adenosine diphosphate receptor inhibitor after hospital discharge among myocardial infarction patients: Insights from the Treatment with Adenosine Diphosphate Receptor Inhibitors: Longitudinal Assessment of Treatment Patterns and Events after Acute Coronary Syndrome (TRANSLATE-ACS) observational study. Am Heart J. 2017;183:62–68. [DOI] [PubMed] [Google Scholar]
  • 29.Kalra R, Arora P, Morgan C, Hage FG, Iskandrian AE and Bajaj NS. Conducting and interpreting high-quality systematic reviews and meta-analyses. J Nucl Cardiol. 2017;24:471–481. [DOI] [PubMed] [Google Scholar]
  • 30.Viechtbauer W Conducting Meta-Analyses in R with the metafor Package. J Stat Soft. 2010;36(3):1–48. [Google Scholar]
  • 31.Bittl JA and He Y. Bayesian Analysis: A Practical Approach to Interpret Clinical Trials and Create Clinical Practice Guidelines. Circ Cardiovasc Qual Outcomes. 2017;10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Zampieri FG, Casey JD, Shankar-Hari M, Harrell FE Jr., and Harhay MO. Using Bayesian Methods to Augment the Interpretation of Critical Care Trials. An Overview of Theory and Example Reanalysis of the Alveolar Recruitment for Acute Respiratory Distress Syndrome Trial. Am J Respir Crit Care Med. 2021;203:543–552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Jacobs AK, Kushner FG, Ettinger SM, Guyton RA, Anderson JL, Ohman EM, Albert NM, Antman EM, Arnett DK, Bertolet M, et al. ACCF/AHA clinical practice guideline methodology summit report: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation. 2013;127:268–310. [DOI] [PubMed] [Google Scholar]
  • 34.Kerman J Neutral noninformative and informative conjugate beta and gamma prior distributions. Electron J Stat. 2011;5:1450–1470. [Google Scholar]
  • 35.Gelman A, Carlin J, Stern H, Dunson D, Vehtari A and Rubin B. Bayesian Data Analysis. 3rd edition. 2013.
  • 36.Bernardo JM. Reference posterior distributions for Bayesian inference. J R Stat Soc. 1979;41:113–128. [Google Scholar]
  • 37.GitHub. TAILOR PCI Bayesian Analysis Code. Available at: https://github.com/torsades-de-pointes/TAILOR_PCI_Bayesian/tree/v1.0. Accessed on: August 30, 2021.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

003353 - Supplemental Material
NIHMS1746411-supplement-003353_-_Supplemental_Material.pdf (214.2KB, pdf)

RESOURCES