We Should Not Tailor Antiplatelet Therapy Based on Platelet Function Testing and Genotyping CRT 2011, Washington DC Sanjay Kaul, MD Division of Cardiology Cedars-Sinai Heart Institute Professor, Cedars-Sinai Medical Center & Geffen School of Medicine at UCLA Los Angeles, California
Disclosure Sanjay Kaul, MD No conflicts to disclose Division of Cardiology Cedars-Sinai Heart Institute Professor, Cedars-Sinai Medical Center & Geffen School of Medicine at UCLA Los Angeles, California No conflicts to disclose
Clopidogrel Poor Metabolizers FDA Statement FDA, March 2010 (clopidogrel boxed warning) “Clopidogrel may be less effective in people who are unable to metabolize the drug because of low CYP2C19 activity” “Be aware that tests are available to determine CYP2C19 genotype” “Consider use of other antiplatelet meds or alternative dosing strategies for clopidogrel” No outcome study informs this recommendation! Is it a rush to judgment? Is the Warning Actionable? 3
“The evidence base is insufficient to recommend either genetic testing or platelet function testing at the present time” Holmes, Dehmer, Kaul et al. JACC 2010
Platelet Phenotype and Genotype Evaluation RCT Post hoc/RCT COGENT GRAVITAS Platelet Phenotype and Genotype Evaluation Post hoc/RCT Meta-analysis TRITON PLATO CREDO CURE ACTIVE-A CHARISMA OASIS-7 Sofi et al. Hulot et al. Mega et al. Observational, & PK/PD/Genotype (Numerous)
Hierarchy of Evidence in Medicine Randomized Controlled Trial Meta-analysis - Clinical homogeneity - Statistical homogeneity Observational Trial - Case-control - Cohort Observational dataset within RCT - Candidate gene studies, GWAS Laboratory study - PK/PD - Genotyping
Impact of PPI on Cardiovascular Events in Clopidogrel-Treated Subjects Results of COGENT Trial 1.00 0.98 Placebo = 20/1885 (1.06%) Treated = 23/1876 (1.23%) MACE HR = 1.16 95% CI = 0.64; 2.10 P=0.63 HR = 0.99 95% CI = 0.68; 1.44 P=0.96 0.96 Survival Probability 0.94 Placebo = 54/1885 (5.7%) 0.92 Treated = 55/1876 (4.9%) Adjustment through Cox Proportional Hazards Model Adjusted to Positive NSAID Use and Positive H. Pylori Status 0.90 30 60 90 120 150 180 210 240 270 300 330 360 390 Days Bhatt et al, NEJM 2010 7
GRAVITAS Patient Flow 5429 patients screened with VerifyNow P2Y12 12-24 hours post-PCI 2214 (41%) with high residual platelet reactivity (PRU ≥ 230) 3215 (59%) without high residual platelet reactivity (PRU < 230) We screened 5,429 patients with the VerifyNow test -- the largest cohort of patients to ever undergo platelet function testing in a single study. 41% of patients were categorized as having high residual reactivity, and these 2,214 patients were randomly assigned to study drug. Random selection Clopidogrel Standard Dose N=586 Clopidogrel High Dose N=1109 Clopidogrel Standard Dose N=1105 Non-Randomized Comparison 8
GRAVITAS Secondary Comparison High vs GRAVITAS Secondary Comparison High vs. Not High Reactivity Treated with Clopidogrel 75-mg daily Here are the results of this non-randomized comparison. The observed rate of the composite of cardiovascular death, MI, or stent thrombosis was 2.3% in patients with high reactivity compared with 1.4% in patients without high reactivity, representing a hazard ratio of 1.68. The difference in event rates did not reach significance; note that the lower boundary of the 95% confidence interval is less than one. The point estimate of this large hazard ratio appears consistent with previous studies documenting an association between residual reactivity and cardiovascular events after PCI. Observed event rates are listed. P value by log-rank test. 9
GRAVITAS: VerifyNow and High Platelet Reactivity How Good is the Test? HRPR Positive Negative MACE Present Absent 1055 578 25 8 33 1633 TOTAL 1080 586 Here are the results of this non-randomized comparison. The observed rate of the composite of cardiovascular death, MI, or stent thrombosis was 2.3% in patients with high reactivity compared with 1.4% in patients without high reactivity, representing a hazard ratio of 1.68. The difference in event rates did not reach significance; note that the lower boundary of the 95% confidence interval is less than one. The point estimate of this large hazard ratio appears consistent with previous studies documenting an association between residual reactivity and cardiovascular events after PCI. 10
GRAVITAS: VerifyNow and High Platelet Reactivity How Good is the Test? 0.2 0.4 0.6 0.8 1 Pre-Test Probability Post-Test Probability Test + Test - Sensitivity = 76% Specificity = 35% PPV = 2% NPV = 99% LR+ = 1.17 LR- = 0.69 AUC = 0.59 Here are the results of this non-randomized comparison. The observed rate of the composite of cardiovascular death, MI, or stent thrombosis was 2.3% in patients with high reactivity compared with 1.4% in patients without high reactivity, representing a hazard ratio of 1.68. The difference in event rates did not reach significance; note that the lower boundary of the 95% confidence interval is less than one. The point estimate of this large hazard ratio appears consistent with previous studies documenting an association between residual reactivity and cardiovascular events after PCI. Impact on MACE risk prediction is tiny! 11
Predictive Performance of Platelet Function Testing and Genotyping
Shear-dependent tests The POPular Study (Primary Endpoint) Effect AUC Cutoff OR (95% CI) P value Aggregation tests LTA 5 μmol/L ADP 0.63 42.9% 2.09 (1.34-3.25) 0.009 LTA 20 μmol/L ADP 0.62 64.5% 2.05 (1.32-3.19) 0.001 VerifyNow® P2Y12 236 PRU 2.53 (1.63-3.91) <0.001 Plateletworks® 0.61 80.5% 2.22 (1.25-3.93) 0.005 Shear-dependent tests IMPACT-R 0.56 8.4% SC 1.34 (0.84-2.14) 0.21 IMPACT-R ADP 0.53 3.0% SC 1.11 (0.69-1.78) 0.68 PFA-100 COL/ADP 0.50 116s 0.77 (0.47-1.28) 0.31 INNOVANCE® PFA P2Y* 299s 1.59 (0.85-2.94) 0.15 Breet NJ et al. JAMA 2010;303:754-762 13
Predictive Performance of Platelet Function Test Effect Sn Sp PPV NPV LR+ LR- AUC LTA 20 μmol/L ADP 0.53 0.64 12% 94% 1.49 0.73 0.62 VerifyNow® P2Y12 0.59 0.63 13% 1.62 Plateletworks® 0.61 1.47 0.66 VASP 0.93 0.50 99% 1.86 0.14 0.85 Impact on risk prediction is tiny! Breet NJ et al. JAMA 2010;303:754-762
The POPular Study Predictive Model Logistic regression modeling Determine predictive value of addition novel risk factor: HPR Identify independent correlates primary endpoint Model predicting the primary endpoint Model 1 Classic Risk factors Age, Gender, Hypertension, Hypercholesterolemia, Diabetes Mellitus, Current smoking, Family Hx of CAD, LVEF<45%, Renal failure, Prior CABG AUC = 0.64 Model 2 Model 1 + Procedural Risk Factors Total stent length, no. of lesions, no. of stents, LAD stenting, graft stenting, bifurcation lesion AUC = 0.64 AUC = 0.72, p=0.004 Model 3 Model 2 + HPR AUC = 0.72 AUC = 0.73-0.78, p<0.01
Onset Maintenance Offset The ONSET-OFFSET Study Last Maintenance Dose Loading Dose Time (hours) Onset Maintenance Offset 100 90 80 70 60 50 40 30 20 10 IPA % Ticagrelor (n=54) Clopidogrel (n=50) * ‡ † 20 µM ADP- Final Extent IPA at 2hrs after first dose (loading): 88% ticagrelor vs. 38% clopidogrel, p<0.0001 0 .5 1 2 4 8 24 6 weeks 0 2 4 8 24 48 72 120 168 240
All ACS Cohort in PLATO CVD/MI/Stroke: Time Course Ticagrelor Clopidogrel Pinteraction = 0.313 RR 0.88 (0.77-1.00) RR 0.80 (0.70-0.91) CVD/MI/Stroke (%) ARD=0.6% ARD=1.3% NNT=166 NNT=77 Randomization to 30d >30d to end of follow-up 50% of total events (945/1878) occurred within 30 days In the STEMI cohort, outcomes favored clopidogrel in day 0-30 FDA Briefing Document, July 28, 2010 17
Major Fatal/Life-Threatening Bleeding by Days from Last Dose of Treatment to CABG 100% Ticagrelor 80% Clopidogrel 60% % Patients with Bleeding post-CABG 40% 20% 0% 1 2 3 4 5 6 7 >8 Days Bleeding differences favor ticagrelor >5 days post discontinuation
Metabolism of Clopidogrel Topol et al, Nature Medicine 2011;17;40-41
How Much Does Carrier Status Matter? Sensitivity 45% Specificity 75% Hochholzer, et al. J Am Coll Cardiol 2010
Clopidogrel Response Variability Genetic Factors Polymorphisms of CYP Polymorphisms of PON1 Polymorphisms of ABCB1 Polymorphisms of P2Y12 Clopidogrel Response Variability Clinical Factors Failure to prescribe/poor compliance Under-dosing Poor absorption Drug-drug interactions involving CYP3A4 Acute coronary syndrome/PCI Diabetes mellitus/insulin resistance Elevated body mass index Cellular Factors Accelerated platelet turnover Reduced CYP3A metabolic activity Increased ADP exposure Up-regulation of the P2Y12 pathway Up-regulation of the P2Y1 pathway Up-regulation of P2Y–independent pathways (collagen, epinephrine, TXA2, thrombin) Adapted from Angiolillo DJ et al. J Am Coll Cardiol. 2007; 49: 1505-1516 . 21
EXCELSIOR: CYP2C19 and High Platelet Reactivity How Good is the Test? 0.2 0.4 0.6 0.8 1 Pre-Test Probability Post-Test Probability Sn = 45% Sp = 75% LR+ = 1.8 LR- = 0.7 AUC = 0.65 Test + Test - 44% 24% 1.80 Hochholzer, et al. J Am Coll Cardiol 2010
CURE – Loss-of-Function Carrier Status No heterogeneity for the first primary (P=0.84), second primary (P=0.87) or safety (P=0.74) endpoint
ACTIVE – Loss-of-Function Carrier Status No heterogeneity for the primary (P=0.73) or safety (P=0.16) endpoints.
Primary Endpoint in the Clopidogrel Group in Relation to Any CYP2C19 LOF Allele 12 Clopidogrel LOF 11.2 10 10.0 p=0.25 8 Clopidogrel No LOF K-M estimate (%) 5.7 6 4 3.8 p=0.028 2 0 60 120 180 240 300 360 Days from randomization No. at risk Clopidogrel LOF Clopidogrel No LOF 1,388 1,275 1,259 1,226 1,027 801 658 3,516 3,321 3,256 3,186 2,691 2,123 1,757 Wallentin L et al Lancet 2010: 376, Online Aug 29, 2010
Predictive Performance of Platelet Genotyping Effect Sn Sp PPV NPV LR+ LR- AUC CYP2C19 for MACE CURE 0.23 0.74 8% 91% 0.87 1.05 0.47 ACTIVE A 0.26 0.75 21% 80% 0.99 0.51 TRITON 0.36 12% 92% 1.36 0.57 PLATO 0.31 0.72 11% 1.11 0.96 0.52 Hulot meta-analysis 0.33 10% 1.17 0.93 0.54 Mega meta-analysis 1.16 0.94 ABCB1 for MACE PLATO (CC, high) 0.27 0.73 90% 1.00 0.50 TRITON (TT, low) 0.39 13% 1.46 0.83 0.59
Predictive Performance of Platelet Genotyping Effect Sn Sp PPV NPV LR+ LR- AUC CYP2C19 for bleeding CURE 0.21 0.74 3% 96% 0.80 1.07 0.45 ACTIVE A 0.44 0.76 10% 1.85 0.65 PLATO 0.57 0.49 11% 91% 1.12 0.88 0.54 ABCB1 for bleeding TRITON 0.37 0.72 4% 98% 1.32 Risk prediction for bleeding: little to none!
Ultrasensitive Troponin (Sn = 88%, Sp = 85%) Predictive Performance of Prognostic Tests Football, Banana, or a Carrot Ultrasensitive Troponin (Sn = 88%, Sp = 85%) 0.2 0.4 0.6 0.8 1 Pre-Test Probability VASP (Sn = 93%, Sp = 50%) 0.2 0.4 0.6 0.8 1 CYP2C19 (PAPI) (Sn = 47%, Sp = 73%) Pre-Test Probability 1 0.8 Test + Test - 0.6 Post-Test Probability 0.4 0.2 0.2 0.4 0.6 0.8 1 Pre-Test Probability Football >> banana > carrot
Bouman et al, Nature Medicine 2011;17;110-116 PON1 Polymorphism and Impact on Nonfatal Stent Thrombosis A Case Cohort Elective PCI Study (41 cases, 71 noncases out of a cohort of 7719) 0.2 0.4 0.6 0.8 1 Sensitivity = 66% Specificity = 65% LR+ = 1.87 LR- = 0.53 PPV = 52% NPV = 77% AUC = 0.70 R2 = 0.725 Test + Test - 52% Post-Test Probability 23% PON1 QQ vs QR/RR 27/52 (52%) vs 14/60 (23%) OR 3.6 (1.6, 7.8); P=0.003 Pre-Test Probability Bouman et al, Nature Medicine 2011;17;110-116
Bouman et al, Nature Medicine 2011;17;110-116 PON1 Polymorphism and Impact on Nonfatal or Fatal Stent Thrombosis A Prospective PCI in ACS Study (N=1982) 1 Sensitivity = 70% Specificity = 60% LR+ = 1.76 LR- = 0.49 PPV = 4% NPV = 99% AUC = 0.70 Test + Test - 0.8 0.6 Post-Test Probability 0.4 0.2 PON1 QQ vs QR/RR 31/808 (3.8%) vs 13/1174 (1.1%) OR 3.6 (1.9, 6.9); P<0.001 4% 1% 0.2 0.4 0.6 0.8 1 Pre-Test Probability Bouman et al, Nature Medicine 2011;17;110-116
Predictive Performance of Prognostic Tests Impact of CYP2C19 Genotype: A Tale of Two Trials Shuldiner et al CYP2C19 (MACE) 14/67 vs 16/160; RR 2.4, p=0.02 (Sn = 47%, Sp = 73%) Bouman et al CYP2C19 (Stent thrombosis) 26/75 vs 15/37: RR 0.86, p=0.55 (Sn = 63%, Sp = 31%) 1 0.2 0.4 0.6 0.8 1 PPV =21% NPV =90% LR+ =1.73 LR- =0.73 AUC = 0.64 R2 =0.12 PPV =35% NPV =59% LR+ =0.92 LR- =1.18 AUC = 0.46 0.8 0.6 Post-Test Probability Post-Test Probability 41 0.4 21 35 Test + Test - Test + Test - 0.2 10 0.2 0.4 0.6 0.8 1 Pre-Test Probability Pre-Test Probability Shuldiner et al, JAMA 2009;302:849-858 Bouman et al, Nature Medicine 2011:1:110-116
Is Routine Testing of Platelet Genotype or Phenotype Reasonable? Requirements of A Screening Tool The test must be practical in the clinical setting. The test must accurately characterize low- and high-risk patients. Identification of at-risk individuals should lead to a treatment that improves outcome. The process should be cost-effective. Adapted from Miller et al, J Am Coll Cardiol 2006;48:761-4
Does Genetic Testing or Platelet Function Testing Fulfill the Criteria For a Screening Tool? Practical clinical use (POC) Risk prediction accuracy Improved outcomes Cost effective CYP2C19 No PPV = 10-21% NPV = 90-92% ? LTA PPV = 12% NPV = 94% VerifyNow Yes PPV = 13% ?/No VASP PPV = 10-12% NPV = 96-99% 33
Prasugrel and Ticagrelor vs. Clopidogrel Bleeding Outcomes Trial Active Rx Control RR, 95% CI Major Bleeding (Study Criteria) TRITON 2.4% 1.8% 1.32 (1.03 to 1.68) PLATO 11.6% 11.2% 1.04 (0.95 to 1.13) CABG TIMI Major bleeding TRITON 13.4% 3.2% 4.74 (1.9 to 11.81) PLATO 5.3% 5.8% 0.94 (0.82 to 1.07) Non-CABG TIMI Major bleeding TRITON 2.4% 1.8% 1.32 (1.03 to 1.68) PLATO 2.8% 2.2% 1.25 (1.03 to 1.53) TIMI Major or Minor Bleeding TRITON 5.0% 3.8% 1.32 (1.11 to 1.56) PLATO 11.4% 10.9% 1.05 (0.96 to 1.15) Bleeding requiring transfusions TRITON 4.0% 3.0% 1.34 (1.11 to 1.63) PLATO 8.9% 8.9% 1.01 (0.91 to 1.11) Fatal Bleeding TRITON 0.4% 0.1% 4.19 (1.58 to 11.12) PLATO 0.3% 0.3% 0.87 (0.48 to 1.59) 1 2 3 4 5 Relative Risk 34
Platelet Genotyping in Practice Challenges and Uncertainties CYP2C19 is one of many polymorphisms Point of care (POC) assay not currently available (3-7 days turnaround time) Expensive ($500-$600) Not reimbursed by insurance companies Ischemic events not predicted consistently, low PPV Bleeding not predicted Holmes, Dehmer, Kaul et al. JACC 2010
Platelet Genotyping in Practice Challenges and Uncertainties No prospective data for how to respond to genetic or functional testing - Dose (double dose, ? higher dose) - Additional agent (cilastozol) - Alternative agent (prasugrel, ticagrelor) No prospective evidence that outcomes are improved Remains a useful investigation tool at the current time Numerous RCTs underway for both genotyping and phenotyping strategies to guide treatment Holmes, Dehmer, Kaul et al. JACC 2010
Role of Platelet Function & Genetic Testing A Case of Enthusiasm Exceeding the Evidence Enthusiasm Evidence
Tailoring Antiplatelet Therapy The Science of Medicine Phenotype/ Genotype Benefit Risk Modest prediction for ischemic outcomes Little or no prediction for bleeding risk Unfavorable logistics/cost Insufficient evidence to recommend either platelet function or genetic testing
Tailoring Antiplatelet Therapy The Art of Medicine Phenotype/ Genotype Benefit Risk Modest prediction for ischemic outcomes Little or no prediction for bleeding risk Unfavorable logistics/cost May consider in patients with recurrent ischemic events on clopidogrel May consider in high-risk intervention (complex lesion, diabetics, etc.)
Tailoring Antiplatelet Therapy Based on Platelet Function Testing and Genotyping Paul Gurbel, August 2010 "The bottom line is we have no prospective studies at this time that alteration of therapy based on genotype or phenotype really affects patient outcomes” White Paper, JACC 2010 (Bonello/Gurbel et al) “However, until the results of large scale trials of personalized antiplatelet therapy are available, the routine use of platelet function measurements in the care of patients with cardiovascular disease cannot be recommended” 2010 ACCF/ACG/AHA Expert Consensus Document on the Concomitant Use of PPIs and Thienopyridines "The role of either pharmacogenomic testing or platelet-function testing in managing therapy with thienopyridines and PPIs has not yet been established"