Peripheral & Central Nervous System Advisory Committee
Public Meeting
(April 28, 1999)
Statistical Review of Aggrenox (Dipyridamole/Aspirin) capsule
Mushfiqur Rashid, Ph.D.
Mathematical Statistician
Division of Biometrics II
CDER/FDA
1.
Introduction
2.
Study design and sample size3.
Efficacy results4.
Statistical Issues5.
Reviewer’s comments
Introduction
Drug:
Aggrenox (extended release Dipyridamole 200 mg/Aspirin 25 mg b.i.d.) CapsuleIndication:
Prevention of :
(2) Death (any cause)
Primary endpoints:
c) Stroke or Death (composite endpoint)
Only (a) and (b) specified in the protocol.
Study Design
A Randomized, Double Blind, Multi-center, Parallel Group Design with four treatment arms:
(4) Placebo
Primary Comparisons of Interest for the Combination Drug:
(1) DP/ASA vs. DP alone
and
(2) DP/ASA vs. ASA alone
Both tests require statistical significance in favor of the combination drug.
Sample Size Estimation
3. Based on the interim analysis(about 4000 patients) results, sample size increased to 7000 patients.
4. In general, increasing sample size affects the Type I error rate.
Efficacy Analyses
Risk reduction (RR)
Survival rate increase (SRI)
Efficacy Results
- Stroke
- Death
- Composite (Death or Stroke)
Stroke
Table 1: Drug Comparisons
Comparison |
p-value (G-W) |
K-M stroke free rate |
RR% |
SRI % |
DP/ASA Vs DP DP/ASA Vs ASA ASA Vs Placebo DP Vs Placebo |
.002 .008
.009
.036 |
89.9 vs. 86.7 89.9 vs. 87.1 87.1 vs. 84.1 86.7 vs. 84.1
|
24.7
23.1
18.3
16.3 |
3.7
3.2
3.6
3.1 |
Figure 1: K-M Curves: Stroke Free Rate for DP/ASA, DP and Placebo
Figure 2: K-M Curves: Stroke Free Rate for DP/ASA, ASA and Placebo
Deaths
Table 2: Drug comparisons for deaths
comparison |
p-value (G-W ) |
K-M survival |
RR% |
SRI% |
DP/ASA Vs. DP DP/ASA Vs. ASA ASA Vs. Placebo DP Vs. Placebo |
.79
.74
.16
.42 |
88.7 vs. 88.5 88.7 vs. 88.9 88.9 vs. 87.6 88.5 vs. 87.6 |
1.3
-2.7
10.9
7.3 |
.2
-.2
1.5
1.0 |
Figure 3: Survival Curves: Death for DP+ASA, DP and Placebo
Figure 4: Survival Curves: Death for DP+ASA, ASA and Placebo
Stroke or Death:
Table 3: Drug Comparisons
Comparisons |
p-value (G-W) |
K-M stroke/death free rate |
RR% |
SRI% |
DP/ASA vs. DP DP/ASA vs. ASA DP vs. Placebo ASA vs. Placebo |
.079
.084
.012
.009 |
82.4 vs. 80.3
82.4 vs. 79.9
80.3 vs. 76.7
79.9 vs. 76.7 |
10.3
12.1
15.7
13.9 |
2.6
3.1 4.7
4.2 |
Table 4: Multiple Endpoint Adjustments: Hochberg - 3 endpoints
Stroke or Death |
DP/ASA Vs. DP |
DP/ASA Vs. ASA |
||
P-value (G-W) |
Unadj. |
Adj.
|
Unadj. |
Adj. |
.079 .157 |
.084 .167 |
Figure 5: Survival Curves of Stroke or Death for DP/ASA vs. DP
Figure 6: Survival Curves of Stroke or Death for DP/ASA vs. ASA
Internal Consistency:
Results for Stroke or Death by region
Table 5: Comparisons by Region (Stroke or Death)
Region |
DP/ASA Vs. DP |
DP/ASA Vs. ASA |
DP vs. PL
|
ASA vs. PL |
Scandinavia .51 .47 .32 .36 (n=1665) N. Europe .88 .32 .02 .12 (n=2425) S. Europe .02 .05 .51 .27 (n=1075) U. K. .81 .43 .22 .03 and Ireland (n=1437)
|
Summary of Issues:
Adjustment needed for multiple endpoints
2. Sample size is increased at the interim analysis:
Adjustments needed for the increase in sample size
Reviewer’s Comments:
From a statistical perspective:
1. The study has shown that the combination drug product is effective only in stroke (fatal and non-fatal). It is not clear that there is any benefit for fatal stroke.
2. Significance efficacy results are not demonstrated for either mortality or the composite endpoint.