We don't know. I would estimate anytime between now and July will see completion of enrollment. HALO guides completion of enrollment in 2H 15 but they enrolled about 146 subjects in the first half of 202 between 4/13 and 4/14. The second half, which will enroll about 114 subjects, started enrolling summer '14 and should go more quickly. Very soon, I'd guess.
The best clue about results was in the pre-halt cohort look in January. Very good. I would expect the post halt numbers to be even better since pre-halt included some subjects who did not receive the full PEG dosing.
I have made mention on the YMB that the main worry regarding efficacy (but not a great worry for me) is the determination of high and low HA status in the tumor. It isn't clear that the new test will replicate the old method. Moreover, it appears the HA determination had been done after results had started to be known. Since the very good results on ORR completely depend upon high HA rating, one has to ask themselves how independent these evaluations are in an open label study.
FWIW, there are many reasons to trust HALO on this and I don't have the time to argue both sides of this right now - so let this be the cautionary post. I have debated with myself for more than a month if I wanted to wade into this territory since I have decent rebuttals for my concerns. If anyone gets overly alarmed, eg: pentawhatever on YMB, I will put on my white hat and get further into it. but I think I'd still like to put this out there for discussion.
To continue to play the devil's advocate (the genesis of my NdP) it wouldn't take too many non-responders on that ORR slide #66 to be classified as "HA rating not available" to make excellent results out of poor results. (The data on the next slide on PFS has similar, if lesser, problems.)
Notice that of the 74 subjects in the PAG arm, there were only 35 who had HA classification. This, despite the fact that HA classification of biopsied material is supposedly a criteria for enrollment. Also note that 12/17 High HA responded while only 9/18 Low HA responded to PAG. This would be above reproach if there weren't only 25/74 responders in the whole PAG treatment group.
Think about it; you randomize 74 subjects to PAG and 25 respond. Twenty one of those responders are in the group they have HA id on. Only four responders are in the group that, for some reason, they have no HA id on. It doesn't take much to see how data could be manipulated here. If it were truly randomized, the percentage of responders in the PAG arm without HA status id should be consistent with the cohort with HA identification. It should be about 60% (9+12 /35), not 10% (4/39).
This is why I believe the FDA will demand efficacy data in Ph3 on a cohort prospectively identified as High HA by a reproducible test. HALO has the test, now to study design.
The talk by fezziwig on the YMB about Breakthrough status this year is premature. You can't hang your hat on these "breakthrough worthy" efficacy numbers with so much of the methodology unclear.
IMO, if the second, uninterrupted cohort of 202 is studied with more rigor around the HA status determination, and they can show independent correlation between the old pathologist HA determination method and the new HA test, we might be able to talk about early breakthrough status before Ph3 is finished. Otherwise, I think we'll need to see the Ph3 trial to the end. If so, let's hope Torely can negotiate an endpoint like PFS instead of OS.
