Roy S. Herbst, MD, PhD: Immunotherapy has an important role in all aspects of lung cancer, driver or nondriver disease. Certainly, in the patients with no known driver, immunotherapy offers hope that we haven’t seen in many years. We can use this drug in a certain percentage of frontline therapies that are PD-L1-positive—pretty much everyone is getting it now as a second-line therapy. It’s one of the drugs where we’re seeing activity: 20% or more response rates. Certainly, there are populations that benefit more. They’re the patients who are living disease-free for 5 to 6 years. We need to figure out what it is about those patients, and also what is it about those patients who receive immunotherapy for 6 weeks and are already progressing. That’s part of our research at places like Yale—we have a lung support grant that’s trying to figure that out.
There’s also a big role for immunotherapy in patients with driver mutations, because I’ve treated patients, for example, with EGFR inhibitors since the first clinical trials we did with gefitinib. We called it ZD1839 back then, in 1998, and no one was cured because patients became resistant to these targeted therapies. So, there has got to be a role for these immunotherapies in the patients who become resistant. It has been a bit challenging to incorporate them in, because of combined toxicities of small molecule EGFR inhibitors—especially the newest generation of drugs and these antibodies. But those trials are ongoing, and I think now that we understand how to better manage the toxicities. We will see that in place. There’s a role for immunotherapy across the whole spectrum of advanced non–small cell lung cancer.
PD-L1 expression is used in a number of ways. Certainly, if you’re going to use an agent that targets PD-1 or PD-L1 upfront in an untreated lung cancer patient, one would want to use a selected population that’s PD-L1-positive. That’s quite clear based on the KEYNOTE-024 trial showing that, in the patients that are PD-L1-positive at 50%, there’s a survival advantage: a hazard ratio 0.5; a PFS benefit 0.5; and a hazard ratio 0.6 for using pembrolizumab. Interestingly, with nivolumab using a cutoff much lower, that wasn’t seen. So, I think PD-L1 clearly has a role in choosing frontline treatment.
As far as using the drug in the second-line setting, given that PD-L1 is a dynamic marker and there’s some heterogeneity, it’s a little less clear. Certainly, in the KEYNOTE-010 trial on which I was the PI, we actually did see that, between the 1% and the 50% cutoffs, there was a clear improvement in the benefit of the pembrolizumab versus docetaxel.
You can use PD-L1 to tell you who’s more likely to benefit, but from the trials with nivolumab and atezolizumab—either one looked at the biomarker, with nivolumab, they looked at it more informally—there did appear to be benefit in the patients who were PD-L1-negative. I think that there are clearly patients who would benefit who are PD-L1-negative, either because it’s a false negative or because there are other mechanisms at play that we don’t understand yet, whether it be mutational load, neoantigen load, gene signatures—meaning what’s going on in the immune microenvironment— how many T cells there are, or what the character of T cells is. So, it’s still a bit of a moving target, but my feeling is that one can use these drugs in all settings.