“Statistically significant and highly clinically meaningful” perhaps, but not so much that they can say what that actually means. My understanding is that of all cancer drugs approved by the FDA between 2000-2016 (around 90 drugs), the median life extension was just over 2 months...
An exciting, underrated possibility: therapies in conjunction with each other, even if the therapies are initially studied in isolation. Single-agent chemo doesn't work for many cancer types, but multi-agent chemo often does. The same may be true of the small molecules, antibody drug conjugates (ADCs), and monoclonal or bispecific antibodies being tested now. I have squamous cell carcinoma initially of the tongue, and now of the head, neck, and lungs (https://jakeseliger.com/2023/07/22/i-am-dying-of-squamous-ce...), and I'm on a clinical trial monotherapy at UCSD called MCLA-158 / petosemtamab. But, if I live long enough, which isn't super likely but isn't impossible, there's a decent shot that the FDA will finally approve Moderna's mRNA-4157 personalized vaccine for melanoma—hopefully in 2025. If I can get mRNA-4157 off label and combine it with pembrolizumab (Keytruda) and petosemtamab (assuming the latter is approved, too), that combo may be much more potent than any of the three in isolation, and I've already failed pembro as a monotherapy and pembro + carboplatin + paclitaxel.
And if that combo doesn't work, or has some unexpected negative effect (including death), well, I'm going to die anyway, and the risk seems worthwhile.
As another commenter observed, the median can be skewed by the number of people who don't respond; among those who respond to a given drug, some don't respond, but some live surprisingly long.