Notes from Piper Jaffray. Caveats:
a) These are notes from my perspective. They are far from comprehensive - instead they are the things that were either new to me or the first time I had seen something definitive on the topic.
b) They are terse. I might never get around to making them more readable, but decided to post them anyway.
c) I am not a stenographer. I am sure that I made errors. Hopefully none of substance, but no guarantees.
d) The italic text is my commentary (as if that wasn't obvious -g-).
And now for the notes:
1) Trisonox has trials going on and new data from independent trials inre other blood cancers.
2) CT retains US and European rights to all their drugs.
3) Per a graph the last point in Stellar 3 is about 295 in Jan. (note that the graphs are all given as a calendar date now)
4) End point statistic will be log rank p value.
5) Said they finally got number of events needed but “will take 8-12 weeks to lock (not 'unlock') the database and give top line results” ( The quote clearly explains the discrepency between item number 3 showing only about 295 and yet having gotten the number of events needed. Previously the company has said that they know the number of events about 4 to 6 weeks ahead of it getting entered into the database. So the plot is based on the database, and now they just need to get those last 16 deaths into the database.)
6) Stellar 3 95% confidence interval on the blended median survival exceeds any of the reported literature.
7) Stellar 4 latest number of cycles was as of 11/04 and was 23% for 4 cycle, 7/5, 20/6. (Note: this is exactly the same as the data reported in the Wachovia conference which was purported to be from an unlocked database of 9/2/04. Given the 4-6 week nature of getting stuff into the database and the fact that the cycles take 3 weeks each and the fact that the trial finished enrolling in June 04, I suspect that this data is wrong. The numbers have to have changed. If not, something is very wrong since it means the lower dose xyotax patients, who were all enrolled at the end, did not get to more cycles.)
8) FACT-LCS poor and equal across geographic areas. (This and the next point are important. They are clearly addressing the concern about non US sites enrolling healthier patients and driving up the blended survival. Does this data absolutely put it to bed? No, but within the limits of what they can know, it certainly significantly lowers the probability with some semi independent measures of related items.)
9) Weight loss>5% percentage equal across geographic areas.
10) Stellar 4 median blended survival greater than all pac doublet medians. Same for 1 yr survival.
11) Stellar 4 is about 250 events of 311 needed (not clear whether this is 'in database only' or also including reports)
12) Note on curves shown: Stellar 4 curves show Jun enrollment additions to trial (perhaps 15?). Deaths really really rolling off – Jan showing total between 300 and 305.
13) Maintenance pac (3 mo vs 12 mo) has shown benefit to Breast Cancer (21 vs 28 mo survival)
14) Will target to reach enrollment target of Pix by end of year, and interim analysis by mid year.
Feel free to correct or make comments.
Clark |
