And here also they don't really have much choice. CM012 showed that Nivo+Ipi worked in PD-L1 positive (1%+) but fell short in PD-L1 negative. So, I think we should read the PR language literally "PD-L1 expressing" = any expression. Of course, they could play it safe and set it at 5 or 10%, but even BMY isn't doing this (after getting burned on CM024).