The Study: This meta-analysis of randomized phase 2/3 trials comparing inhibitors of the PD-1 axis to cytotoxic chemotherapy across a whole host of disease sites shows some things we kinda already knew. It included > 4K patients from 8 trials = 4 CheckMates (testing nivolumab) + 2 KEYNOTEs (pembrolizumab) + 2 OAK and POPLAR (atezolizumab). They defined PD-L1(+) cancers as any with ≥1% of cells expressing PD-L1. As expected, across the board, patients with PD-L1(+) tumors saw the greatest survival benefit with PD-1 axis targeting (HR 0.66). However, patients with PD-L1(-) tumors also saw a clear benefit (HR 0.80), though significantly less than their PD-L1(+) counterparts, supporting KEYNOTE-189 that reported a spectrum of benefit with PD-L1 status. Plus the many many forest plots of subgroup analyses fall almost exclusively in the PD-1 camp.
Bottom Line: The authors don’t go so far as to recommend PD-1 axis targeting for everything, but, if we want to reliably rule out people who won't benefit, we probably need a better surrogate than PD-L1 staining. | Shen, BMJ 2018