@OP, I always appreciate your careful, insightful analyses.
Yes, we can discuss whether it's better to analyze ORR vs OS. For understanding this complex illness, survival at 100 days is very straightforward. We can discuss whether binomial distribution or log rank or chi-square gives a better statistical picture... and we can get lost in the forest for all the statistical trees.
Forget statistics and primary vs secondary endpoints for a moment. Everyone understands 14 deaths out of 54 children. And everyone with a calculator can understand that's better than 12 deaths out of 30... but worse than 0 deaths out of 54. SR-aGvHD is a complicated illness. Multiple organs may be impacted, multiple ancillary medications may be used, standard of care is evolving ie "the goalposts" are shifting, and multiple other factors can play on morbidity. In other words there are many "confounding variables" that make it hard to match up cases and analyze results with certainty. The best way to proceed is to randomize patients prospectively and use blinded controls.
That being said I think everyone can understand 0 deaths out of 54 would have left no doubt. Zero deaths out of 54 is like rescuing people on the top of a burning high-rise by using a helicopter. There's no doubt as to effectiveness, no binomial distribution required. Zero deaths out of 54 are the results one might achieve say... by putting new onset type 1 diabetics on insulin. There's just no doubt as to insulin's effectiveness compared to no insulin. In fact, in an illness with 100% 100 day mortality, even lowering the (imputed) death rate to 30% (higher than rem-L's 26%) in just 30 patients leaves no doubt as to effectiveness. That's essentially what Lantidra did.
For rem-L, somewhere between zero deaths out of 54 and 14 deaths out of 54, the FDA might have been swayed. I gave a rough estimate of where the cutoff could have been, where "exceptional statistical significance" ended and "promising but not proven" began. The key point is that running study 001 was not a futile mistake. No one knew in advance how the results would play out.
But pushing forward after that... trying to work out these complex issues by advancing an unproven product in children without benefit of randomized, blinded controls... that was a mistake
AS A LEAD INDICATION for this therapeutic. The company should have put childhood GvHD on the back burner and proceeded with RCT's elsewhere.
And it appears at least one person agrees with me.
Silviu Itescu, who announced on August 2 that he is switching over to testing in adults... like he could have done years ago. Better late than never as the saying goes.
(20min delay)