Molecular Testing's Role in Non-Small Cell Lung Cancer

Transcript:

Vassiliki Papadimitrakopoulou, MD: The use of PD-L1 testing in the frontline setting still remains very important for reasons of choice of therapy because we know that certain patients with high PD-L1 expression will benefit from single-agent pembrolizumab, but also because we need to learn more about what the benefit is in different subsets of patients depending on their expression. Therefore, I don’t see an abandonment of the use of PD-L1 expression as a frontline biomarker. It will be supplemented by additional biomarkers, such as tumor mutation burden and possibly signatures that predict for outcome, but at this point it remains a standard biomarker.

Corey J. Langer, MD: PD-L1 is a crucial predictive biomarker. The results of KEYNOTE-024 and now KEYNOTE-042 have shown a major survival advantage: first in those patients who have 50% or higher expression for pembrolizumab compared with standard chemotherapy, and now in the KEYNOTE-042 trial for those with 1% or higher expression. The key question, however, is whether that advantage is actually realized in the 1% to 49% expression group. In KEYNOTE-042, it’s a positive result with apparent OS benefit, which is roughly about 8 months better, 12 months versus 20 months. Is that driven mostly by the results in the 50% or higher group, or is it shared equally with the 1% to 49% group?

For those with no PD-L1 expression, I would not use a checkpoint inhibitor alone, single agent, up front. There, we have data from KEYNOTE-189 that again show a significant survival advantage. That may be the group that will ultimately benefit from some combination immunotherapy. Data from CheckMate-227 have looked at TMB, tumor mutation burden, of more than 10 Mb and have shown that group in particular, which is about 40% of our overall population, has a significant PFS benefit versus chemotherapy. I would personally have equipoise in either the 0% expression group or the 1% to 49% expression group, in those with high TMB, comparing the immunotherapy combination of ipilimumab/nivolumab—a CTLA4 inhibitor with a PD-1 inhibitor—versus the KEYNOTE-189 regimen, paclitaxel/carboplatin/pembrolizumab. Again, this is in the nonsquamous population. I’m not sure what I would do at this point in the squamous population.

That is a critical question I think will ultimately need to be addressed. Right now, it’s moot. As of early June 2018, the I-O combination does not have FDA approval. That may change if the CheckMate-227 data actually show survival benefit. I strongly suspect that I-O combination will be approved in individuals with high TMB, and then we’ll be left with this dilemma. What do we give? That’s actually a good dilemma, when we have so many different options.

My particular bias at this point, given the probable emergence of TMB as a very informative biomarker, is that we should get in the habit of doing next-generation sequencing in all our patients. Certainly, in our nonsquamous patients, we’re already doing that. Beyond EGFR, ALK, and ROS1, we need to look for less common mutations such as BRAF, HER2, or c-MET. But TMB can be generated only by NGS. You cannot do this on the basis of an immunohistochemical stain as you would with PD-L1. In the squamous population, every now and then we find an oncogenic driver. For that reason alone, you can make an argument for doing NGS in that population. With the likely emergence of TMB, I would get in the habit of doing it. You can make an argument, perhaps, not to do NGS in the squamous population. I think that’s going to change if TMB finally does emerge. It’s an arbiter of our therapeutic decisions.

Transcript Edited for Clarity