Novel Predictive Biomarkers for Non-Small Cell Lung Cancer - Episode 5

Evaluating the Role of Molecular Analysis in NSCLC


Jared Weiss, MD: Let’s be honest about it, laboratory testing in non—small cell lung cancer is confusing. We have multiple axes that we’re looking to test and, at least within the molecular realm, there’s a lot to understand. There are a lot of companies marketing different tests, and this space is confusing. There is a natural desire to minimize this testing when possible and simplify our clinical practice. The question is, do the data support it? The first answer comes from the molecular realm. While response rates for patients with non-smoking mutations such as EGFR and ALK are not 0, they are dramatically lower with immunotherapy than patients without these changes, dramatically lower. And so, it is inappropriate to use immunotherapy agents as an early option for such patients, at least on their own. I believe that there is an absolute consensus among expert physicians in this realm that if you have an actionable mutation like EGFR or ALK, there’s an imperative to treat to that targeted therapy axis as long as possible.

The technical reasons are that it works better. You have higher response rates there, you have better PFS [progression-free survival] there, and guaranteed exposure is very important for maximizing total survival. But there’s a human element to this as well. The patients feel better. They feel better because you’re controlling their cancer-related symptoms because the drugs work better. They feel better because there are fewer adverse effects to targeted therapy than to cytotoxic chemotherapy, and you de-medicalize their life. These are pills instead of IV [intravenous therapy]. Between fewer and less severe adverse effects and the lack of need to be in an infusion room every 2 or 3 weeks, I think that there is a dramatic human imperative to offer targeted therapy and to offer it for as long possible. Starting with the molecular axis, it is absolutely mandatory to do that testing. There is nothing about any of the new data sets that will let us out of that confusing box.

One question that remains after accepting that is whether the results of KEYNOTE-189 allow us out of the box of PD-L1 [programmed-cell death ligand 1] testing. I would say in a way that this question matters a little bit less. The human imperatives matter a little bit less. KEYNOTE-189 competes with multiple other orthogonal data sets that all give reasonable options of what to do. I would criticize any expert who stood in front of a camera and said they know the absolute answer. But to not totally dodge the question, I do think that if we look to the results of KEYNOTE-024 and KEYNOTE-042, and the patient experience with single-agent immunotherapy, there is a human driver to still consider PD-L1.

When you get a PD-L1 expression of 50% or higher, we know from KEYNOTE-024 and now from KEYNOTE-042 that the patient has a superior survival with single-agent PD-1 axis inhibition, in this case pembrolizumab, compared with chemotherapy. What we don’t know is whether that patient with 50% or higher PD-L1 expression will do better with immunochemotherapy or immunotherapy alone. The studies are too different, and the populations accrued too differently to do the cross-trial comparison that we all say we’re not going to do, but end up doing. They really are too different to actually do that.

We just don’t know. In the absence of hard data to answer that question, I think we come back to the human drivers of our care. We want our patients to live for as long as possible—there’s no answer as to which approach will do that—we want our patients to feel well, and we want them to feel well for as long as possible. This is an opinion, not a fact that’s totally data-based, but to me, single-agent immunotherapy fulfills these human goals of my care better than immunochemotherapy. Until we have comparative data, the default in my practice is to offer a patient with 50% or higher PD-L1 expression single-agent pembrolizumab. That creates, to me, a strong human driver to test.

Transcript Edited for Clarity