Median OS, Belated Subgroups, and the Moonshot:
Nothing new in what I am saying here, but I think still worth saying:
What is the median OS improvement for the CI's when they are showing efficacy for only 25% of the patient population? Isn't it zilch by definition? The efficacy is great for the patients where it helps, the mean OS improvement is substantial, but the improvement in median OS is zilch. Right? Regardless of the powering of the trial, the median OS improvement would be 0. Yet the FDA approved them, in some cases even based on phase 2 data. And I don't think the 25% subgroup was identified in advance. I don't know, but I don't think so. Not sure they yet know in advance who will respond.
What do the above considerations mean for DCVax-L? What if it only shows efficacy for 20% of the patients, but for those, shows very high efficacy? The median OS improvement would be 0, but clearly that should not be a stopper for approval, yet it would by traditional rules. One goal of the Moonshot should be to plow through any such nonsensical barriers.
At the same time the statistical taboo in making up subgroups after the fact has a very real basis. It is a difficult topic but may need to evolve to make sense for modern therapies, or perhaps modern analysis that allows resolving efficacy to subgroups.
If DCVax-L proves to have high efficacy for a small subgroup that can be identified prior to surgery then it is in everybody's interest for it to be approved, but for only that subgroup. The reimbursement overall might be the same as if you spread the efficacy, and approved over the entire patient population... but you would be preventing many patients that it will not work for from trying another therapy that might work, and you would be adding manufacturing burden unnecessarily on NWBO. NWBO would make more money with only the proper subgroup(s) approved, and the patients would be better off. This would not be true if DCVax-L was nearly cost-free to manufacture as are many of the existing drugs. In that case approving for everyone when it only works for a small set is bad for patients but not bad for $ for the company.
