1. Pragmatic TRIALS WWW.UCDENVER.EDU/IMPLEMENTATION Efficacy (Explanatory) versus Pragmatic Trials Consideration on Trial Design William R. Hiatt, MD Professor of Medicine, Division of Cardiology University of Colorado School of Medicine President, Colorado Prevention Center

2. WWW.UCDENVER.EDU/IMPLEMENTATION William R. Hiatt Conflicts Peripheral artery disease research grants: AstraZeneca, CSI, DNAVEC, Kyushu University, NIH, Pluristem, ReNeuron, Rigel Obesity drugs: None Diabetes drugs: None Lipid drugs: None

3. WWW.UCDENVER.EDU/IMPLEMENTATION Efficacy versus Effectiveness (Pragmatic) Trials 1=very explanatory (efficacy clinical trial) 2=rather explanatory 3=equally pragmatic – explanatory 4=rather pragmatic 5=very pragmatic (effectiveness trial)

4. WWW.UCDENVER.EDU/IMPLEMENTATION How Do Efficacy Trials Evaluate Benefit and Risk?  Phase 3 trials for FDA regulatory approval require a frequentist approach to determine benefit (e.g. p < 0.05 or less)  Phase 3 trials require a rigorous, highly defined study population  Study drug and dose and background therapy tightly controlled  Safety evaluation more Bayesian and descriptive with use of ‘exclusion of risk’ to define thresholds for unacceptable risk

5. WWW.UCDENVER.EDU/IMPLEMENTATION FDA Guidance Evaluating CV Risk Trials Developing New Antidiabetic Rx Trials can use HbA1c as an acceptable surrogate of glycemic control for approval Guidance on how to demonstrate a new antidiabetic therapy is not associated with unacceptable CV risk Individual trial designs typically under-powered to demonstrate CV benefit or harm Use of the exclusion of risk approach to rule out an unacceptable level of safety concern CDER Guidance Dec 2008

6. WWW.UCDENVER.EDU/IMPLEMENTATION FDA Guidance Evaluating CV Risk  Establish independent CV endpoints committee  Adjudicate all CV events from all phase 2 and 3 trials  Meta-analysis and prospective analysis plan  Pre-approval, upper boundary of 2-sided 95% CI of risk must be < 1.8. If between 1.3-1.8 then:  Post marketing a CVOT must demonstrate < 1.3.  New development programs may be subject to greater scrutiny CDER Guidance Dec 2008

7. WWW.UCDENVER.EDU/IMPLEMENTATION Rosiglitazone Controversy Nissen - NEJM 2007;356:2457-71 Rosiglitazone approved in 1999 Risk concerns: increased LDL cholesterol level, anemia, fluid retention and heart failure

8. WWW.UCDENVER.EDU/IMPLEMENTATION 2010 Updated Rosiglitazone Meta-analyses Nissen/WolskiFDA Number of trials5652 Type analysisStudy levelPatient level MIOR 1.28 (1.01-1.62)1.80 (1.03-3.25) CV deathOR 1.03 (0.78-1.36)1.46 (0.60-3.77) MACE1.44 (0.95-2.20) Arch Intern Med 2010;170:1191-1201 FDA briefing document 2010 and 2013 2013 FDA update on meta analysis separated placebo from active controls and supports 2010 observations

9. WWW.UCDENVER.EDU/IMPLEMENTATION RECORD Trial Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycaemia in Diabetes RCT in 4447 patients with diabetes Rosiglitazone plus metformin or rosiglitazone plus sulfonylurea versus combination MET/SU (flexibility on delivery) Limitations: Open label, non-inferiority design Primary endpoint CV hospitalization or CV death Active control not established as safe Low adherence, high crossover (flexibility on adherence) Imbalance between groups in statin and diuretic use (flexibility on delivery) Lancet 2009;373:2125-35

10. WWW.UCDENVER.EDU/IMPLEMENTATION RECORD Results Lancet 2009;373:2125-35

11. WWW.UCDENVER.EDU/IMPLEMENTATION FDA Decisions on Rosiglitazone 2010 Janet Woodcock acknowledged “multiple and conflicting signals of CV risk associated with rosiglitazone” and: 1.Questioned the integrity of the Record trial 2.Put rosiglitazone on restricted distribution 3.Stopped the definitive TIDE trial to assess CV risk -11,680 event-driven trial evaluating MACE 4.Ordered a re-adjudication of RECORD by DCRI

12. WWW.UCDENVER.EDU/IMPLEMENTATION FDA 2013 EMDAC Meeting DCRI Re-adjudication of Record RSG N=2220 MET/SU N=2227 Hazard ratio (95% CI) CV death, MI, CVA181 (8.3%)188 (8.4%)0.95 (0.78-1.17) CV death88 (4.0%)96 (4.3%)0.90 (0.68-1.21) MI68 (3.1%)60 (2.7%)1.13 (0.80-1.59) Stroke50 (2.3%)63 (2.8%)0.79 (0.54-1.14) All cause mortality139 (6.3%)160 (7.2%)0.86 (0.68-1.08) DCRI found more events but overall results unchanged DCRI did not find any trial misconduct or data integrity concerns Limitations: Reliance on original database and source docs Retrospective Additional follow up on vital status with limited information on MI or stroke

13. WWW.UCDENVER.EDU/IMPLEMENTATION Rosiglitazone – the final outcome A somewhat pragmatic trial (Record) was ultimately vindicated and determined to be informative on drug safety In 2014 FDA released the restrictions on Rosiglitazone but it was too late, as other diabetes drug alternatives have emerged Does this experience inform the pragmatic trialist? Trial design and execution really determines trial integrity Be explicit on design issues that may either confound the outcome or impact sample size

14. WWW.UCDENVER.EDU/IMPLEMENTATION Pragmatic Trial Perspective (from a pure efficacy trials perspective) Pragmatic trials are critically important to perform to assess the utility of a therapy that ‘works’ in an idealized phase 3 clinical trial environment in the general population The principles of trial design developed for efficacy trials are equally valid for pragmatic trials, e.g.: Many pragmatic trial decisions may decrease power so be honest on a realistic sample size Power for population heterogeneity and subgroup analyses Control the intervention as much as possible