Iconic,
“It is quite possible that Phase 2 had a small sample problem. So RVX may have jumped the gun on assuming that it was Crestor that really did it...when in fact, it could be spurious correlation”
I totally agree with this, but our resident biochemist (and the RVX researchers) believes that it is possible there is a relatively large statin-dependent difference in response to apabetalone so let’s go with it for the moment….
“Second, if you are assuming a RRR of 30% for Phase 3, then it is statistically nearly impossible for half the dosed population (on Lipitor) to deliver sub par results on any metric”
I’m glad you put the ‘nearly’ in there. Let’s try this on for size. RVX was expecting a 10% decrease in MACE in the Lipitor group and a 50% decrease in the Crestor group. Median dosing period was 27 months (2.25 yrs), 1000 placebo patients stayed in trial, 500 each Crestor and Lipitor patients stayed in trial. Expected placebo event rate of 0.07 was observed. If this scenario played out, then:
Placebo: 1000 x 2.25 x 0.07 = 158 MACE
Lipitor: 500 x 2.25 x 0.063 = 71 MACE
Crestor: 500 x 2.25 x 0.035 = 39 MACE
Placebo MACE: 158 Apabetalone MACE: 110 Total MACE = 268 (pretty close to observed)
RRR = 48/158 = 0.31 or 31%
Do I think it is likely this is what really happened? No, it is a very optimistic scenario. But it is possible.
Like you, I am sceptical of the earlier observations on the statins (small sample size effects). Most likely possibilities if the RVX team is still positive about the CVD potential of this drug (an assumption, we don’t know if this is true but it appears so):
The overall reduction in MACE (even on Crestor) was less than expected
-
The expected overall reduction in MACE was observed but took place in the later stages of the trial, so that time-to-event differences between the placebo and apabetalone arms were minimized
Jupe
