ESMO 2025 | CD8 T cell and PD-L1 expression linked to survival in first-line NSCLC

Immunodriver 2 study is a collaborative study that UCLA did with Tempest using the Tempest Lens database. And in this, in Immunodriver 2 study, we looked at response, median overall survival rates for patients that receive first-line chemo IO or IO alone in the metastatic setting. And what we were correlating was the co-expression of PD-L1 expression with CD8 T cells. PD-L1 expression needs to be interpreted in the context of CD8 T cells, but there is the concept of constitutive PD-L1 and inducible PD-L1. Inducible PD-L1 is where when a CD8 T cell comes into the tumor microenvironment, the tumor cell upregulates PD-L1 to shut down the CD8 T cell activity. And this is why this upregulation of PD-L1 is called inducible PD-L1 and this is most predictive of checkpoint inhibitor efficacy because coming in with a PD-L1, PD-L1 inhibitor will release that break in that setting. In contrast, constituent of PD-L1 is where there’s a baseline expression of PD-L1 independent of CD8 T cells. There could be no CD8 T cells or low-level CD8 T cells, and there’s constitutive levels of PD-L1 expression. And this constitutive PD-L1 expression does not predict immune checkpoint inhibitor efficacy. So in the current phase three registrational trials, we don’t look at the CD8 component when we interpret PD-L1 to be high at 50% or 1 to 49 and less than 1. We only talk about PD-L1 expression outside of the context of CD8 T cells. And I think this is fundamentally one of the problems of why we’re seeing heterogeneity and readouts of trials where PD-L1 has not been performing consistently because it’s by random chance how much inducible versus constituent of PD-L1 is present. So immunodriver 2 looks at the co-expression of the two together. And we separated patients by PD-L1 high, 1 to 49, PD-L1 negative. And then we looked at CD8 T-cells at a median cutoff. And if it was above, we grouped them as CD8 T cell high. If they were below the median T cell counts, then it was CD8 T cell low. And what we found is that if your PD-L1 expression is high or progressively increasing from 1% all the way up to 50% and higher, and they were also co-correlated with CD8 T-cell high, those patients performed the best. And the CD8 T-cell high and PD-L1 high didn’t reach its median overall survival. It was at 22 months and it was still not reached. Whereas on the flip side, if you’re CD8 T-cell low and PD-L1 negative, these patients have the poorest overall survival, median overall survival at 11 months. But all of the other groupings between the two analyses, all of the CD8 T-cell high groups all outperformed the PD-L1 varying levels, especially if they’re CD8 T-cell low. So the take-home message, and this was over 1,300 patients in the first-line setting with chemo, IO, or IO alone, is that maybe in future trials, we should revisit and go back to the fundamentals of enriching the patients who are most likely to respond, and that’s going to be the PD-L1-positive patients with CD8 T-cells. And also, in the implication, the early stages, following resection, when you have the entire specimen, whether it’s in the adjuvant setting or post-chemo IO, and you have the resection specimen, the tumor microenvironment should be interrogated for, are there CD8 T cells present? And that should be a useful clinical parameter. And we should validate this in phase three trials of whether those are the responsive patients with more additional immunotherapy management side.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.