Curiously, in the pre-specified subgroup with PD-L1>=50%, Keytruda’s DFS compared to placebo had a non-statsig HR=0.82—a worse result than in the overall trial.
The HR for OS, a secondary endpoint that is not yet mature, was 0.87.
All told, this is not a great dataset, IMO, but it may be enough for FDA approval. These data may put Keytruda at a competitive disadvantage to Tecentriq and Opdivo in early-stage NSCLC.