Hi hankmanhub…
The problem as I see it with predicting what might be the blended median OS right now is that it does not reflect the 100 STILL ALIVE patients.
Are those 100 patients some of the longest living patients in the trial? If so, then their data isn’t shown in whatever might be the median OS at this point in time. And since it’s fairly likely the majority of the 100 alive are treatment patients, than the ratio of 1 to 2 would not be at all reflected in the blended median OS as it stands right now.
But, I’ll take a stab at it based on averages, and what the protocol had hypothesized.
The protocol indicated that it had assumed median survival times of 17 and 34 for placebo and treatment cohorts respectively.
So if we were to work with an average of everyone eventing on average at those median survival times… and assume everyone had passed… it would look like this.
Control patients - 110 x 17 months = 1870 months
Treatment patients - 220 x 34 months = 7480 months
1870 + 7480 = 9350 total blended months / 331 patients = 28 months blended median OS
But that is with everyone passing… which we know hasn’t happened. 100 are STILL ALIVE.
So it may currently look more like this still using averages and the hypothesized numbers.
Control patients - 105 patients x 17 months = 1785 months
Treatment patients - 126 patients x 34 months = 4284 months
1785 + 4284 = 6069 total blended months / 231 patients OS evented = 26 months blended median OS
So I would think 28 months blended median would be what you want to see if everyone had passed. But since they haven’t, I’d guess maybe somewhere around 26 months blended based on what the protocol had indicated, and those 100 still alive not being included in the mOS right now.
Of course, this is all based on averages, and it's important to note that establishing a median really doesn't work that way.
I think everything ultimately depends on how the alpha sorts out.
What type of separation needs to be seen between the two arms if they are not working with an alpha of 5, but instead an alpha of 3 for OS? If this is the case, t's likely they were originally planning on having a full 5 for OS for this trial... and it's possible they may not.
Hence... the prolonged trial while they wait, we wait, and patients wait for the arms to separate more. It's kind of a curse of relevancy.
And while Avii and Ex think there is no alpha left for OS, I'll have to, once again, and until proven otherwise, disagree with them.
Besides the fact that they've earlier indicated they are using only 2 for PFS - which should leave 3 - I really have to think NWBO was smart enough to see what happened with DNDN. They unfortunately used all 5% of its alpha on PFS (which failed) and subsequently, had nothing left over for OS in their 2007 ruling.
So I doubt NWBO did not repeat that DNDN mistake when they set this DCVax trial back up in 2011. And the fact that they have indicated they were using just 2 for PFS, strongly indicates there is still more to be applied with OS.
