>>Albuferon achieved an SVR of 80% vs 85% for
the Pegasys arm; the 5% difference was less than the 12%
non-inferiority margin in the SPA and hence the trial is
considered a success. The Albuferon and Pegasys arms
were similar in safety and tolerability.<<
I guess I'm confused, Dew - probably because I haven't gotten around to listening to the 12/8/08 cc yet. I thought HGS just made it by the skin of their teeth because the 12% non-inferiority range actually translated to the measured difference between Albuferon's SVR24 and Pegasys' SVR24 being less than ~5% in either direction. Read this from Stump and see if you agree or not.
(2/27/07 4Q06 cc)
>>DAVID STUMP: The key aspects of the Phase III trial are the noninferiority boundary, which is agreed at 12%. The overall power for the noninferiority hypothesis is well above 90%, conventional 90% power. And that includes accounting for the multiple testing that we're doing with two doses in the study. If we achieve our non inferiority hypothesis, there is an allowed comparison for superiority between a dose that is already deemed to be noninferior. The power will be less, I would say will be closer to 80%, which would be nice conventional Phase II power for a superiority hypothesis, but certainly enough to make the test interesting if we get far enough to be allowed to do it.
We hear from treating physicians that in fact their boundary of interest is a 5% absolute difference in point estimate and we will probably not be adequately powered to see that difference, but this field actually responds more to observed point estimates that actual statistics in the end. It is important to remember that in our noninferiority test, if we are more than 5% absolute worse for Albuferon, we will not achieve our non-inferiority boundary. The two come together nicely despite the testing approach. Is that clear?
MEG MALLOY: I want to clarify that. When you talk about 12%, that's plus or minus 5%?
DAVID STUMP: Yeah. That's the boundary, your noninferiority boundary. Your confidence interval can't drift beyond that. Your starting assumption is in a non-inferiority trial is that your point estimates will actually be observed to be the same.
MEG MALLOY: Okay. So plus or minus, say 6% in terms of SVR-24?
DAVID STUMP: Yes.
MEG MALLOY: Percent?
DAVID STUMP: So if we're within 5% on SVR -- final SVR, we'll meet our noninferiority hypothesis. Now obviously, if you're going to have point estimate drift, you would prefer to drift in a favorable direction.<<
(7/30/08, 2Q08 cc)
>>DAVID STUMP: This was a point of very specific discussion in our end-of-Phase II discussion with FDA where we set the boundary to be excluded of 12%. That roughly would translate in to the statistical outcome being met if Albuferon was no more than 5% worse – observed worse than Pegasys.<<
If HGS did make it only by the skin of their teeth with Achieve 2/3, then it is still very unclear what the result will be after Achieve 1, and that could be why the HGSI stock price gyrated wildly after the results came out, as you noted. A quick read of the PR indicates the trial was successful; upon closer examination, the Albuferon results were inferior - in a statistically meaningless way, but in a way which could easily become statistically meaningful after Achieve 1.
micro
News 
Market Data 
Discover 
AI
