I actually think the consistency among triple negetive, high GPNMB, triple negative and high GPNMB groups is positive. I don't know where you get the information ORR rarely seem to scale with the degree of overexpression. The boundary between high and low expression is always going to be a cutoff - that's what a dignostic is all about - people can always say it is arbitrary. Even if you think so, triple negative fills in nicely to me.
I think SGEN has noted similar data, but it's not at my fingertips right now.
As for the PFS, I guess what struck me is that despite their claim of triple negative and high GPNMB worsening prognosis (none of which I doubt), the control PFS curves in those groups don't look all that different to me... medians around 1.5 months and, sadly, pretty much all progressed by 3 months. I guess it would have been nice to see the control group curves scale a little bit with prognosis.
But I'll leave it there. I think I'm trying to read too much from the trial, given the number of patients enrolled.