1 Can One Evaluate An Outcomes Claim Based On An Active Controlled Study? Pfizer Response Cardiovascular and Renal Drugs Advisory Committee Rockville, Maryland June 15, 2005

2 Sponsor Representatives And Introduction Introduction and Objective Lance Berman, MD Medical Director US Team Leader Pfizer Inc Methodology and Analysis Overview Michael Gaffney, PhD Senior Director Statistical Research and Consulting Pfizer Inc

3 Introduction  Placebo-controlled CV outcome studies in hypertensive patients are no longer ethical  ALLHAT can be considered a well-conducted, large, randomized, double-blind, active-controlled study that provides CV outcomes for amlodipine besylate*  Using a non-inferiority analysis of the ALLHAT data, Pfizer submitted an sNDA for inclusion of the results in the amlodipine label  Invited by FDA to present this methodology as an illustrative example of how an active controlled study that showed no superiority between treatment arms might be used to support an outcomes claim. * From here on, all mention of amlodipine besylate will be referred to as amlodipine

4 Brief Summary of ALLHAT  ALLHAT, an NHLBI study  Designed to determine if newer antihypertensive agents (amlodipine, lisinopril and doxazosin) were superior to first-line treatment with a diuretic (chlorthalidone)  Began in 1994  First major trial to assess the long-term CV effects of amlodipine in a large hypertensive population

5 Brief Summary of ALLHAT  ALLHAT enrolled an ethnically diverse patient population across North America:  N=42,418, age > 55 years  Mild-to-moderate hypertension + at least 1 other CHD risk factor. Median follow-up was 4.9 years  Subjects randomized to treatment initiated with amlodipine, lisinopril, doxazosin or chlorthalidone with add-on therapy to achieve BP<140/90 mm Hg  Primary endpoint (CHD death and nonfatal MI):  Amlodipine-based treatment vs chlorthalidone-based treatment RR=0.98; 95% CI,  ALLHAT showed that long-term amlodipine-based treatment was not superior to chlorthalidone-based treatment with respect to cardiovascular outcomes

6 Outline Of Approach To Determining The Benefit Of Amlodipine A vs C Post-hoc analysis was done to demonstrate that amlodipine (A) was “not inferior” to chlorthalidone (C) treatment in ALLHAT ALLHAT results demonstrated that amlodipine-based therapy was not superior to chlorthalidone based therapy

7 Outline Of Approach To Determining The Benefit Of Amlodipine A vs C C vs placebo Effect of C is consistent and reproducible Post-hoc analysis was done to demonstrate that amlodipine (A) was “not inferior” to chlorthalidone (C) treatment in ALLHAT Need first to establish the efficacy and consistency of the standard treatment (chlorthalidone) vs placebo in reducing CV events ALLHAT results demonstrated that amlodipine-based therapy was not superior to chlorthalidone based therapy

8 Outline Of Approach To Determining The Benefit Of Amlodipine A vs C C vs placebo Effect of C is consistent and reproducible Non-inferiority A relative to C Estimate the benefit A relative to placebo Post-hoc analysis was done to demonstrate that amlodipine (A) was “not inferior” to chlorthalidone (C) treatment in ALLHAT Then show that the new treatment (amlodipine) preserves a substantial portion of this effect Estimate the effects of amlodipine relative to placebo Need first to establish the efficacy and consistency of the standard treatment (chlorthalidone) vs placebo in reducing CV events ALLHAT results demonstrated that amlodipine-based therapy was not superior to chlorthalidone based therapy

9 Establishing The Benefit And Consistency Of Chlorthalidone Treatment  SHEP study Placebo-controlled trial in a population with isolated-systolic hypertension. Chlorthalidone- based treatment was shown to reduce the risks of fatal coronary events, non-fatal MI and stroke  Meta-analysis Identify all randomized placebo-controlled hypertension studies that used low-dose diuretics to evaluate CV risk reduction

10 Meta-Analysis Study Characteristics Treatment Comparison NPopulation Primary Endpoint Follow- up SHEP- Pilot Chlorthalidone 25 mg/day vs placebo 551 ≥60 years, ISH Events collected 2.8 years SHEP Chlorthalidone mg/d +/- atenolol or reserpine vs placebo 4736 ≥60 years, BP /<90, no signs of major CVD Stroke 4.5 years MRC-O Amiloride mg/ HCTZ mg vs atenolol vs placebo years, BP /<115 All-cause mortality, stroke, coronary events 5.8 years EWPHE HCTZ/triamterene +/- methyldopa vs placebo 840 >60 years, BP / Death and non- fatal CV, renal events 4.7 years Psaty BM, Smith NL, Siscovick DS, etal. Health outcomes associated with antihypertensive therapies used as first line agents. JAMA 1997; 277 (9):739-45

11 Results Of The Meta-Analysis For CHD Death And Nonfatal MI The relative risks were combined across trials by the Mantel–Haenszel method. P=.66 homogeneity of RR Relative Risk (95% CI) ( ) ( ) ( ) ( ) ( ) Diuretic Events/N (%) Placebo Events/N (%) 15/443 (3.4)4/108 (3.7) 104/2365 (4.4)141/2371 (6.0) 48/1081 (4.4)159/2213 (7.2) 48/416 (11.5)59/424 (13.9) Trial SHEP-Pilot SHEP MRC-O EWPHE Combined

12 Chlorthalidone Benefit On Fatal/ Non-Fatal MI Preserved By Amlodipine Chlorthalidone (n=15255) Amlodipine (n=9048)A/C One-Sided 97.5 Upper CL A/C CHD Death/ Non-Fatal MI 1362 (8.93)798 (8.82)

13 Chlorthalidone Benefit On Fatal/ Non-Fatal MI Preserved By Amlodipine Benefit of Chlorthalidone to Placebo Estimated by: Imputed Placebo RR % of Chl Effect Preserved Point Estimate1/0.72 =1.391−0.07/0.39= 82% Upper CL1/0.85 = 1.181−0.07/0.18= 60% Chlorthalidone (n=15255) Amlodipine (n=9048)A/C One-Sided 97.5 Upper CL A/C CHD Death/ Non-Fatal MI 1362 (8.93)798 (8.82)

14 Estimating The Benefit Of Amlodipine Relative To Placebo Meta-analysisALLHAT chlorthalidone x amlodipine=amlodipine placebochlorthalidoneplacebo

15 Estimated Relative Risk Of CHD Death/ Non-Fatal MI Of Amlodipine To Placebo And 95% Confidence Limit Chlorthalidone/PlaceboAmlodipine/Chlorthalidone (Low-Dose Diuretic Studies)(ALLHAT) RRLn RRSERRLn RRSE Amlodipine/Placebo Ln RRSERR95% CI (0.58, 0.85)

16 Estimated Relative Risk Of Stroke Of Amlodipine To Placebo Chlorthalidone/PlaceboAmlodipine/Chlorthalidone (Low-Dose Diuretic Studies)(ALLHAT) RRLn RRSERRLn RRSE Amlodipine/Placebo Ln RRSERR95% CI (0.49, 0.76)

17 Estimated Relative Risk Of Hospitalized/ Fatal CHF For Amlodipine vs Placebo Chlorthalidone/PlaceboAmlodipine/Chlorthalidone (Low-Dose Diuretic Studies)(ALLHAT) RRLn RRSERRLn RRSE Amlodipine/Placebo Ln RRSERR95% CI (0.58, 1.04)

18 Points To Consider  Consistency of effect of active control  Meta-analysis  Extrapolation of chlorthalidone benefit to ALLHAT  population of ALLHAT vs populations of the trials in the meta-analysis  Conduct of the ALLHAT study  Other secondary ALLHAT outcomes  Statistical considerations  Adjusting for multiplicity  Post-hoc nature

19 Summary ALLHAT was a large active-controlled study that showed no superiority between the newer agent (amlodipine) and the standard agent (chlorthalidone) In the meta-analysis, the standard therapy (chlorthalidone) was found to have a consistent and reproducible effect (0.72 [ ]) The new agent (amlodipine) was found to be non-inferior to the standard therapy (chlorthalidone) (82% of chlorthalidone effect was preserved) The new agent (amlodipine) was estimated to reduce the risk of CHD death and non-fatal MI by 29% relative to placebo

20 Conclusion This non-inferiority analysis provides an illustrative example of how one can use an active controlled trial to support an outcomes claim.