Thursday, September 13, 2007 12:27:50 PM
SPPI/GPCB Satraplatin OS results.
In GPC's FDA briefing document for the Straplatin AC meeting they have the interrum analysis K/M curves for survival on page 40. It doesn't look all that impressive, so they present a curve for the subset of patients followed for at least 1 year on the next page. They suggest/speculate that the data in the final analysis might look more like this "mature" subset data.
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4309b1-03-GPC.pdf
But this doesn't seam to hold up (even as a speculative argument). The problem is that the "mature" subset curves seperate quite well by 26 weeks, while the ITT curves have no seperation through 1 year.
First off, the ITT curve is "final" at about 26 weeks, so the set of patients excluded from the "mature" subgroup is clearly not doing well at the start.
Second, if the subgroup data accurately relects the total ITT, should not the curves be about the same despite the censored data points?
So is the difference chance or real?
If chance, then wouldn't the best guess of the final HR be 90% (the present HR for the overall group)? And if so doesn't that (probably) make the p value come in over the required .044?
If not chance, then things are even worse as Satra. is doing worse off in the data set that will be providing more data points as we go on (and furthmore, questions about the change in SOC come into play).
I do realise that anything can happen, it's just I don't buy the company line of looking at the "mature" subgroup.
If I'm really screwed up on these graphs I would love for somebody to tell me so (hopefully with a decent explanation)
In GPC's FDA briefing document for the Straplatin AC meeting they have the interrum analysis K/M curves for survival on page 40. It doesn't look all that impressive, so they present a curve for the subset of patients followed for at least 1 year on the next page. They suggest/speculate that the data in the final analysis might look more like this "mature" subset data.
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4309b1-03-GPC.pdf
But this doesn't seam to hold up (even as a speculative argument). The problem is that the "mature" subset curves seperate quite well by 26 weeks, while the ITT curves have no seperation through 1 year.
First off, the ITT curve is "final" at about 26 weeks, so the set of patients excluded from the "mature" subgroup is clearly not doing well at the start.
Second, if the subgroup data accurately relects the total ITT, should not the curves be about the same despite the censored data points?
So is the difference chance or real?
If chance, then wouldn't the best guess of the final HR be 90% (the present HR for the overall group)? And if so doesn't that (probably) make the p value come in over the required .044?
If not chance, then things are even worse as Satra. is doing worse off in the data set that will be providing more data points as we go on (and furthmore, questions about the change in SOC come into play).
I do realise that anything can happen, it's just I don't buy the company line of looking at the "mature" subgroup.
If I'm really screwed up on these graphs I would love for somebody to tell me so (hopefully with a decent explanation)
Join the InvestorsHub Community
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
