Randomized Trials and Registries: all smoke and mirrors or clear enlightenment?

Transcription

1 Randomized Trials and Registries: all smoke and mirrors or clear enlightenment? Laura Mauri, MD, MSc Brigham and Women s Hospital Associate Professor of Medicine, Harvard Medical School ADVANCED CARDIOVASCULAR INTERVENTION 2012

2 Disclosures Research support to Harvard Clinical Research Institute from: Abbott, Boston Scientific Corporation, Cordis Corporation, Medtronic, Inc., Bristol-Myers Squibb/Sanofi-Aventis Pharmaceuticals Partnership, Eli Lilly and Company and Daiichi Sankyo Company Limited Consulting: Abbott, Cordis Corporation, Medtronic

3 The Information Age In Cardiology Journals 2797 Clinical Reports 399 Randomized Trials

4 The Information Age From paper chart review to electronic health records From few labor-intensive studies to simple analysis of large comprehensive databases How does medical knowledge benefit from the availability of data? How do we judge the reliability of these sources particularly when results are contradictory?

5 The Information Age Can this new wealth of data improve medical care? Yes! Observational data fully represent the patients we treat and must be used to compare treatments No! Observational data have misled us time and time again. Randomized trials are the only way to go

6 6 Observational Studies vs Randomized Trials Randomized trials more likely to exclude children, elderly, and women based on inclusion criteria. (Van Spall HG. JAMA. 2007;;297(11): ). Randomized trials, even when large, often exclude patients with the most grave clinical presentations and the highest mortality. Maasland L. Stroke Aug;;40(8): Bahit MC. Am Heart J Jan;;145(1):109-17). Even when inclusion criteria are very broad ( Pragmatic, real world or all-comers trials), randomized trials can exclude ~50% of patients. (De Boer SP. Eur Heart J Sep;;32(17):2161-7).

7 December 2006 Advisory Panel to FDA Day 1: Randomized trials Day 2: Off-label and observational data September 2006

8 Pooled Randomized Trials of DES Target Vessel Failure Stone Stone, et al. NEJM NEJM ;; 356.

9 Pooled Randomized Trials of DES Definite and Probable Stent Thrombosis at 4 years 1.7 % 1.5 % 1.8 % 1.4 % Mauri, L. N Engl J Med 2007;;356:

10 SCAAR Cumulative Risk of Death Death Drug-eluting stent RR: 1.03 ( ) Bare-metal stent RR: 1.32 ( ) No. at Risk Years BMS 12,880 12,473 12,354 12,228 9,298 5,966 3,199 DES 5,770 5,605 5,541 5,471 3, , Lagerqvist B et al. N Engl J Med 2007;;356:

11 Ontario Registry DES vs BMS 5.5% 7.8% Tu J et al. NEJM 2007;;357:

12 Drug-Eluting and Bare Metal Stenting in Massachusetts 2-Year Outcome in Matched Patients Cumulative Incidence 30% 20% 10% 0% 0 Mortality Time after Initial Procedure (days) DES n=5,441 BMS n=5,441 Log Rank P value % 9.4% Days DES 5,441 5,338 5,219 5,123 BMS 5,441 5,279 5,105 4,995 Mauri et al. Circulation 2008 October 28.

13 . N Engl J Med 2008;;359:

14 Horizons Trial: MACE 10 TAXUS DES (n=2257) Safety MACE (%) EXPRESS BMS (n=749) Diff [95%CI] = 0.1% [-2.1, 2.4] HR [95%CI] = 1.02 [0.76, 1.36] P NI =0.01 P Sup = % 8.0% Number at risk TAXUS DES EXPRESS BMS Time in Months * Safety MACE = death, reinfarction, stroke, or stent thrombosis Stone, G. NEJM

15 Percentage standardized differences in clinical and procedural characteristics: DES vs BMS before and after matching on 63 pretreatment variables. Mauri L et al. Circulation 2008;;118: Copyright American Heart Association

16 Observational Studies of DES and BMS Total N DES patients DES (%) Follow up at publication Sweden* % 3y Western Denmark % 15m Ontario % 2y Massachusetts % 2y DESCover % 1y NHLBI NA 96% 1y Emilia Romagna % 100% 2y 16 Mauri L and Normand SL. Circulation. April 23, *Lagerqvist B et al. N Engl J Med 2007;;356:

17 SCAAR Update N= Cumulative risk of death BMS DES RR: 0.95 (0.83, 1.1) Time (years) BMS DES James, S. ESC 2007;; NEJM. May 2009.

18 The Information Age Metaanalysis Nested observational studies Cluster randomization Randomized trial with administrative data collection Instrumental variables

19 19 Observational vs Randomized Studies Randomized trials Every patient has a non-zero probability of receiving either the control or the experimental therapy Eliminate selection bias/ confounding by indication Require patients to satisfy inclusion and exclusion criteria Require patient consent Represent a narrower patient population than an observational study Standards for ensuring validity and minimizing bias: blinding, prospective statistical plan, endpoint definition, single vs multicenter, completeness of follow-up, ascertainment, adjudication

20 20 Observational Studies are Diverse Include any non-randomized study of treatments Include metaanalysis of randomized trial data May represent a broader spectrum of patients with greater generalizability and power Are varied in design (retrospective vs prospective, single center vs multicenter, population-based, unselected vs restricted by inclusion criteria) Also require rigor to ensure validity and minimize bias: (endpoint definition, completeness of follow-up, ascertainment, adjudication)

21 21 Observational Studies are Diverse Varied objectives Descriptive or hypothesis driven Description of treated patient population, of mode of utilization, concurrent treatments Prediction of and monitoring of adverse events Evaluation of treatment effectiveness Evaluation of adverse treatment effects

22 22 Methods to Reduce Selection Bias Regression Propensity Score (Matching simulates what would occur in a randomized trial) Instrumental Variables (a variable that is associated with the treatment of interest, but is not expected to be associated with the outcome apart from its relationship with that treatment. (e.g. time after introduction of DES)

23 23 Observational and RCT of DES What we knew from RCT DES prevent restenosis What we learned from registries - DES are not associated with increased mortality, and are associated with prevention of repeat procedures Still unknown: when is it safe to stop dual antiplatelet therapy before or after one year.

24 Duration of Dual Antiplatelet Therapy after DES There is broad variation in practice regarding actual duration of therapy worldwide, and among physicians Until recently, all data regarding duration of therapy were observational, not randomized

25 Prescription of DAPT after 12 m is highly variable across regions TIMI 38 C R Percent Continuing Thienopyridine Time from Index ACS Bonaca M. ACC 2011

26 Duke Registry: comparison of patients treated with and without clopidogrel beyond 12m 24-Month Events With Without Clopidogrel Clopidogrel Difference (95% CI) P Value Drug-Eluting Stent (DES) Patients (n) Death 2.0% 5.3% -3.3% (-6.3% to -0.3%) 0.03 Death or MI 3.1% 7.2% -4.1% (-7.6% to -0.6%) 0.02 Bare-Metal Stent (BMS) Patients (n) 417 1,976 Death 3.7% 4.5% -0.7% (-2.9% to 1.4%) 0.50 Death or MI 5.5% 6.0% -0.5% (-3.2% to 2.2%) 0.70 *Exclusions: DES group (Death 62, nonfatal MI 18, revasc 76, meds not reported 129 (total 285/1502) BMS group (Death 123, nonfatal MI 94, revasc 289, meds not reported 266 (total 772/3165) Eisenstein EL et al. JAMA. 2007;;297:

27 Comparing Durations of Antiplatelet Therapy Goal is to define the balance of risk of ischemic vs bleeding risk Stent thrombosis is a rare event, that requires data review and adjudication, but has profound clinical significance, not available in most registries The presence of a small confounder could reverse the findings of an observational study on this rare event DAPT Study designed as a pragmatic and inclusive randomized study

28 Study Design Eligible for Enrollment after PCI Any PCI with DES or BMS >18 years of age No contradictions to dual antiplatelet therapy Able and willing to provide written informed consent Eligible for Randomization at 12 m Stratified by DES v BMS, drug type, and complexity (ACS or lesion-based) Not Eligible for Randomization at 12 m Death MI or repeat PCI at > 6 weeks CABG Stroke Major Bleed Total 33 month follow-up 12 m DAPT Arm Aspirin + blinded placebo 30 m DAPT Arm Aspirin + blinded thienopyridine Study treatment period m Study observation period 30-33m Total 33 month follow-up Mauri, Kereiakes et al AHJ 2010;; 160(6): Primary analysis of DES treated subjects, 12-33m Secondary analysis of propensity matched BMS to DES subjects 0-33m 2 co-primary endpoints: stent thrombosis and MACCE (death, myocardial infarction or stroke) Powered safety endpoint: major bleeding (GUSTO) 28

29 Randomized Antiplatelet Rx Duration Trials REAL+ZEST LATE EXCELLENT PRODIGY Inclusion Group, N month event free 1443 Non-STEMI month event free DAPT Duration ~12 vs 24 All DES 6 vs 12 SES or EES 6 vs 24 DES Type 1º Endpoint 2º Endpoint DES and BMS ITALIC vs 12 EES ISAR-SAFE OPTIMIZE month event free 3120 non-stemi DAPT 20, month event free 6 vs 12 All DES 3 vs 12 ZES 12 vs 30 1.DES 2.BMS 2-year cardiac death/mi 1-year cardiac death/mi/tvr 2-year death/mi Presented ARC ACC ST, 2010 bleeding Death/MI/CVA/ Presented ST/major ACC 2011 bleeding Presented ARC ESC ST, 2011 bleeding 1-year death/mi/repeat urgent Enrolling revasc/stroke/majorbleeding Death/MI/stroke/ TIMI major bleed at 15 months 1-year death/mi/ stroke/bleed 1. Death/MI/stroke at 33 months 2. Def/prob ST at 33 months Individual component Enrolling endpoints Enrolling ARC ST Enrollment Major bleeding Complete PES = paclitaxel-eluting stent ZES = zotarolimus-eluting stent SES = siroliumus-eluting stent EES = everolimus-eluting stent

30 REAL-LATE/ZEST-LATE: 2-Year Endpoints 2701 patients with DES from two trials HR, 1.65 ( ) HR, 1.73 ( ) Cardiac death or MI(%) P = DAPT 1.8 ASA Park SJ, et al. N Engl J Med. 2010;;362: Primary Endpoint Number at Risk Baseline 1 year 2 year DAPT ASA Lack of difference not interpretable because of insufficient power and follow-up: <1/4 reached 2y follow up

31 REAL-LATE/ZEST-LATE: Timing of Randomization Characteristic Time to randomization Clopidogrel + Aspirin (n = 1357) Aspirin Alone (n = 1344) P value mo. 18 mo. after procedure 18 mo. 24 mo. after procedure >24 months after procedure Median (interquartile range) ( ) 12.8 ( ) Time of Randomization varied from m post PCI Park SJ, et al. N Engl J Med. 2010;;362:

32 PRODIGY Study: 6 vs 24m DAPT after DES or BMS, randomized at 30 days 2,013 randomly allocated to recieve one of the four study stent types 499 randomized to and received EES 498 randomized to and received PES 500 randomized to and received ZES 502 randomized to and received BMS (1497 DES) 1,970 DES and BMS randomized at 30 days Months DAPT Months DAPT year follow- up year follow- up Valgimigli ESC 2011.

33 Primary Endpoint Overall Death, MI or CVA mo DAPT 6 mo DAPT P=0.91 % 4 Hazard Ratio: 0.98 ( ) No. at Risk 24-Month Clopidogrel Valgimigli 6-Month ESC Clopidogrel

34 Type II, III or V BARC bleeding mo DAPT 6 mo DAPT P= % Hazard Ratio: 0.46 ( ) No. at Risk 24-Month Clopidogrel Valgimigli 6-Month ESC Clopidogrel

35 Bleeding endpoint in PRODIGY Investigators changed bleeding endpoint from TIMI major to Bleeding Academic Research Consortium* (BARC) type II, III, IV before final analysis BARC definitions provide potential for uniform definition across trials and are hierarchical BARC type II includes any bleeding that triggers testing or treatment, even if no change in hemoglobin, blood transfusion or hemodynamic sequelae are present. *Mehran et al. Circulation Jun 14;;123(23):

36 EXCELLENT Trial: Stent Thrombosis (Definite or probable stent thrombosis by ARC definition) Cumulative incidence rate (%) P=0.327 HR = 2.00 (95% CI ) 6-mo DAT 12-mo DAT 0.8% 0.4% Months after initial procedure Patient Number at Risk 6-month month Circulation Jan 24;;125(3): Gwon H-C. ACC 2011

37 EXCELLENT Trial: TIMI Major Bleeding (Overt clinical bleeding with a drop of Hb > 5 g/dl or Hct > 15%) Cumulative incidence rate (%) P=0.419 HR = 0.50 (95% CI ) 6-mo DAT 12-mo DAT 0.6% 0.3% Months after initial procedure Patient Number at Risk 6-month month Circulation Jan 24;;125(3): Gwon H-C. ACC 2011

38 Recent Studies of DAPT Duration in Context Questions continue regarding benefit vs risk of longer thienopyridine therapy on MACCE Recent study results have not been definitive Not powered to determine differences in stent thrombosis Variable treatment durations Not blinded Yet each of these studies highlights the remaining clinical question regarding DAPT: Is there a benefit (stent thrombosis or MACCE prevention) that outweighs the risk (bleeding) or cost

39 NCT United Kingdom Germany Poland Romania France United States Hungary Czech Republic New Zealand Australia Principal Investigators: PI: Laura Mauri, MD, MSc, Brigham and Women s Hospital, Boston, MA, USA Co- PI: Dean Kereiakes, MD, Christ Hospital, Cincinnati, OH, USA National Coordinating Investigators: Anthony Gershlick, MD, University Hospitals of Leicester, United Kingdom Andrzej Hoffman, MD, Wielospecj Szpital Miedjski im.dr. E Warminsigo SPZOZ, Poland Ian Meredith, MD, Monash Cardiovascular Research Centre, Australia John Ormiston, MD, Mercy Angiography, New Zealand Wolfgang Rutsch, MD, Charite Univeitaetsmedizin Berlin, Germany P.Gabriel Steg, MD, Hospital Bichat, France

40 Total Enrollment Complete August July 1, , ,645 Actual Projected Projected

41 Total Subject Enrollment Medtronic EDUCATE Don Cutlip, Harold Dauerman 2274 Cordis CYPRESS Daniel Simon, David Kandzari Boston Scientific Liberte PAS David Lee, Kirk Garratt DES n = 23,212 Abbott Xience V USA James Hermiller, Mitch Krucoff 2998 HCRI DAPT-DES Laura Mauri, Dean Kereiakes HCRI DAPT-BMS Laura Mauri, Dean Kereiakes BMS n = 2,986

42 Patient Characteristics Age DES N=23,212 BMS N=2,986 All Patients N=26,198 Mean ± SD 62.1 ± ± ± 10.7 Sex (Female) 28.2% 25.9% 27.9% Diabetes Mellitus 33.0% 24.0% 32.0% Previous PCI 34.9% 20.4% 33.3% Previous CABG 13.9% 7.6% 13.2%

43 Complexity DES N=23,212 BMS N=2,986 All Patients N=26,198 Any Clinical Complexity Factors 35.2% 66.6% 38.7% Any Anatomic Complexity Factors 32.3% 38.5% 33.0% Complexity (any clinical or anatomic) 53.6% 73.2% 55.8% Clinical Complexity= acute coronary syndrome (including STEMI), renal insufficiency, or EF<30% Anatomic Complexity= 3 vessels stented, in-stent restenosis of DES, prior brachytherapy, unprotected left main, > 2 lesions stented per vessel, lesion length 30m, bifurcation lesion with sidebranch > 2.5mm, vein bypass graft, or thrombuscontaining lesions

44 Thienopyridine DAPT Trial Enrolled Subjects 44 All Subjects N=26,198

45 Stent Type DAPT Trial Enrolled Subjects 45 All Subjects N=26,198 DES Type* Cypher (n=3056) 13.2% DES n = 23,212 BMS n = 2,986 Endeavor (n=3458) TAXUS (n=5216) Xience/ PROMUS (n=11752) 14.9% 22.5% 50.6% *Some patients have received more than one DES type

46 Can observational data be used to compare safety of different DES? SCAAR 2012 BMS old DES new DES <50% 12m followup and <10% 24m follow up for new DES group trend to improved outcomes over time in all groups Sarno et al. EHJ

47 Drug-eluting Stent Components are Varied Metal alloy (stainless steel, cobalt chromium, platinum) Polymer durable (PEVA, PBMA, SIBS, phosphorylcholine), bioerodable (PLGA, PLA) Drug sirolimus, paclitaxel, zotarolimus, everolimus Yet the optimal constituents are not well-understood Materials science, polymer chemistry Pharmaco-kinetics Bench and animal studies not yet closely tied to adverse clinical outcomes (restenosis, stent thrombosis)

48 PROTECT - STUDY DESIGN Prospective, international, multi-center, randomized trial of Zotarolimus-eluting vs Sirolimus-eluting stent Patients: Single and Multiple Coronary Artery Lesions Endeavor Stent N=4400 8,800 patients 200 sites Cypher Stent N=4400 Procedure 30d 6mo 1yr 1½yr 2yrs 2½yr 3yrs 4yrs 5yrs Clinical FU Primary Endpoint: stent thrombosis (ARC definite and probable) at 3y Main Secondary Endpoint: composite endpoint of death or MI at 3y Dual Anti-platelet therapy for 3 months minimum, extended use prespecified according to operator discretion Camnezind, Wins, Mauri AHJ 2009.

49 ENDEAVOR IV Trial 5-Year Final Clinical Results 30% p = 0.06 p = 0.22 p = 0.70 p = 0.69 p = p = 0.75 p = % 21.3% Incidence Rate (%) 20% 15% 10% 17.3% 12.7% 15% 7.8% 8.4% 5% 0% 4.3% 3.8% TVF TVR TLR Cardiac Death Non-Q MI Q-wave MI ARC Stent Thrombosis 2.1% 5.3% 0.6% 0.7% 1.4% 1.9 % Endeavor TM Stent (n = 722) TAXUS Express Stent (n = 718) TVF=Cardiac Death, MI and TVR. Kandzari D TCT 2011.

50 SPIRIT IV Trial 3-Year Clinical Results p=0.02 p=0.06 p=0.02 p=0.31 p=0.03 p= % Incidence Rate (%) 10% 5% 0% 9.2% 11.7% 6.2% 7.8% 3.2% 5.1% 1.4% 1.9% 2.7% 4.0% 0.6% TLF ID TLR All Death Cardiac Death TV MI ARC ST (def/prob) 1.6% XIENCE V ä (PROMUS ä ) Stent (n = 2458) TAXUS Express Stent (n = 1229) *n=patients at risk. Rates (%) are from Kaplan Meier estimates TVF=Cardiac Death, MI, or Ischemia Driven TVR. TLF=Cardiac Death, Target Vessel MI or Ischemia-driven TLR. Stone G. TCT 2011.

51 ARC Stent Thrombosis to 1 year % (n) R-Zotarolimus stent n = 1119 Everolimus stent n = 1126 P Definite ST Acute: (0 1 day) 0.4% (4) 0.1% (1) NS Sub-Acute: (2 30 days) 0.4% (5)* 0.0% (0) 0.03 Late: (31 days 360 days) 0.4% (5)* 0.2% (2) NS All: (0 days 360 days) 1.2% (13) 0.3% (3) 0.01 Definite/Probable ST Acute: (0 1 day) 0.4% (5) 0.2% (2) NS Sub-Acute: (2 30 days) 0.7% (8) * 0.4% (4) NS Late: (31 days 360 days) 0.6% (7)* 0.2% (2) NS All: (0 days 360 days) 1.6% (18) 0.7% (8) NS *One patient had a definite ST at day 4 and 31 One patient had a probable ST on day 0 and a definite ST on day 5 Serruys et al. PCR NEJM 2010.

52 Stent Thrombosis (ARC Definite/Probable) Through 2 Years Cumulative Incidence of ARC def/prob ST 5% 4% 3% 2% 1% 0% R-ZES EES Pooled patient level data, 5 Trials* Adjusted P = Hazard Ratio = % CI (0.66,2.71) % 1.0% Time After Initial Procedure (days) Outcomes remain consistent when adjusted for duration of DAPT Days No. at risk %CI No. at risk % CI Mauri et al. PCR 2011.

53 Stent Thrombosis in Patients With and Without DAPT Interruption Through 2 Years N = 11,219 Xience V pts 3.0 ST through 2 years (%) % N ST events = 58 N at risk = 7,139 Never interrupted DAPT through the 2-year study period (or until a ST event) Unadjusted comparison 0.63% N ST events = 25 N at risk = 4,080 One or more DAPT interruptions through the 2-year study period Gregg W. Stone, MD, TCT 2011.

54 Randomized Antiplatelet Rx Duration Trials REAL+ZEST LATE EXCELLENT PRODIGY Inclusion Group, N month event free 1443 Non-STEMI month event free DAPT Duration ~12 vs 24 All DES 6 vs 12 SES or EES 6 vs 24 DES Type 1º Endpoint 2º Endpoint DES and BMS ITALIC vs 12 EES ISAR-SAFE OPTIMIZE month event free 3120 non-stemi DAPT 20, month event free 6 vs 12 All DES 3 vs 12 ZES 12 vs 30 1.DES 2.BMS 2-year cardiac death/mi 1-year cardiac death/mi/tvr 2-year death/mi ARC ST, bleeding Death/MI/CVA/ ST/major bleeding ARC ST, bleeding 1-year death/mi/repeat urgent revasc/stroke/majorbleeding Death/MI/stroke/ TIMI major bleed at 15 months 1-year death/mi/ stroke/bleed 1. Death/MI/stroke at 33 months 2. Def/prob ST at 33 months Individual component endpoints ARC ST Major bleeding PES = paclitaxel-eluting stent ZES = zotarolimus-eluting stent SES = siroliumus-eluting stent EES = everolimus-eluting stent

55 SYNERGY Stent Abluminal Bioabsorbable Polymer Polymer Mass Remaining (%) Polymer resorption is complete within 4 months Bioabsorbable polymer (PLGA) + Everolimus Applied only to the abluminal surface (rollcoat) Thin strut ( ) PtCr Stent

56 EVOLVE First in Human Trial Late Loss at 6 Months Target Lesion Failure at 6 Months P=0.19* P=0.56* 10 8 Late loss, mm PROMUS Element SYNERGY SYNERGY ½ Dose Patients, % PE SYN SYNERGY ½ PROMUS Element SYNERGY ½ Dose P value for noninferiority <0.001, both comparisons Meredith 2011 TCT

57 EVOLVE Short DAPT Trial All patients treated with SYNERGY Stent (PtCr-PLGA-everolimus) Except subjects requring coumadin or 12 m DAPT Randomized 1:1 4m DAPT 12m DAPT Double-blind, placebo controlled, noninferiority design Endpoints: cardiac death and myocardial infarction;; definite and probable stent thrombosis;; major bleeding Mauri ACI 2011

58 Treatment Strategy: High Impact Cardiovascular Intervention Studies Can we improve coronary intervention Interaction of coronary devices with antiplatelet therapy Extension vs reduction in the need for medications Valvular heart disease intervention Multivessel coronary intervention Comparison to standards of care (medical, surgical)

59 COURAGE Patients with stable angina randomized to Optimal Medical Therapy (OMT) PCI + OMT Survival Free of Death and MI PCI+OMT Hazard Ratio 1.05, 95% CI , P=0.62 OMT Years Criticisms -Not representative of daily practice, angiography after randomization - Low risk population - High level of cross over to PCI - Many non-significant stenoses stented - Many significant stenoses left alone - > 95% BMS, POBA Unrealistic hypothesis reduction of death and MI in stable, low risk patients, already disproven in POBA era

60 FAME II Independent Data Safety Monitoring Board Recommends St. Jude Medical s FAME II Clinical Trial Stop Enrollment Following Positive Interim Analysis Follow-up trial to landmark FAME trial examines benefits of using FFR-guided assessment for stent procedures with optimal medical treatment compared to optimal medical treatment alone ST. PAUL, Minn. Jan. 18, 2012 St. Jude Medical, Inc. (NYSE:STJ), a global medical device company, announced today that an interim analysis of the FAME II trial has found a highly statistically significant reduction in the need for hospital readmission and urgent revascularization when Fractional Flow Reserve (FFR)-guided assessment was used to direct treatment in patients with coronary artery disease. As a result of the positive interim analysis, the FAME II independent Data Safety Monitoring Board (DSMB) has recommended investigators stop patient enrollment in this trial as the DSMB considers it unethical to continue to randomize patients to optimal medical therapy (OMT) alone.

61 Courtesy Bernard Debruyne, FAME 2 PI. FAME II FAME 2 Trial

62 FAME II Courtesy Bernard Debruyne, FAME 2 PI. COURAGE FAME II No/limited Ischemia 71% Ischemia 29% Ischemia 100% Shaw L et al JACC 2010;;55:A /1388 (29%) Multiple vascular territories or 10% ischemia

63 Cumulative Event Rate (%) Repeat Revasc CABG (N=897) SYNTAX Repeat Revascularization at 12 Months CABG Months Serruys PW et al. NEJM 2009;;360(10): PCI PCI 4.7% 11.4% CABG 1.3% 2.8% PCI (N=903) 13.7% P< * 5.9%

64 Cumulative Event Rate (%) p= % 3.5% CABG, PCI Similar Safety Months CABG (N=897) PCI (N=903) 20 p= % 3.2% Death Myocardial Infarction Serruys PW et al. NEJM 2009;;360(10):

65 CABG vs PCI (POBA or BMS) Metaanalysis, N=7812, 10 RCT 5-year event rate (%) CABG PCI p value Death Death or MI Death, MI, or repeat revascularization < Hlatky MA et al. Lancet 2009;; 373:

66 Treatment Strategy Trials How to understand PCI vs CABG for multivessel disease beyond Syntax and observational data favoring CABG Consider change in strategy to FFR-guided PCI Consider change in endpoints repeat revascularization at 1 year not the same as 30d death and MI Consider patient centered outcomes quality of life

67 Conclusions We know what makes a good RCT We need to apply equally rigorous standards to observational data particularly when used to determine treatment effects or safety Observational designs may be complementary, yet fundamental treatment strategy questions continue to require randomized trials to define new paradigms

68 Conclusions Sharing of data from administrative and randomized trial sources will allow medical care to benefit from increasing access to information Powerful studies of treatment strategy may result from collaboration of diverse stakeholders to answer treatment strategy questions