I found an interesting thing that in the EV-302 study the proportion of the high PD-L one patients is nearly about 58%, but I think in many studies, the proportion is nearly 20% or 30%.  As we all know, the high PD-1 expression will lead to a more effective result. So, why the PD-L1 expression in EV-302 study is so high?

The PD-L1 story, the PD-L1 biomarker story is not one biomarker. Each drug, and I was involved in the development of those. I think all three of them, maybe two or three ,each drug, develops its own biomarkers. We talked about collaboration earlier. This is a good example of poor collaboration.

The first group is atezolizumab group that I was involved with. We only measured PD-L1 on immune cells. We didn't measure PD-L1 on tumour cells. And we used the SP142 antibody. And actually, when you look at it, about 50% of immune cells express PD-L1 with SP142.

The pembrolizumab folks have already had success in lung cancer, but they had done in measuring tumor cells and immune cells. So they said we're not gonna do the same as Roche, we will do on our way,  we will repeat what we did in lung cancer, that's what we did in EV-302, about 65% of patients are positive for immune cell or tumor cell staining.

And then finally, the nivolumab folks, and I wasn't involved in this group. Pursued exclusively tumor cells staining, only 25% of bladder cancers expressed PD-L1 on tumor cells.

That explained the variability in some trials being 25% positive, such as CheckMate 274  adjuvant nivolumab, and other trials such as EV+Pembro being 65%  positive because they are measuring tumor cells and immune cells staining.