Benjamin Levy, MD: Not to make things more complicated, but we not only have to follow comprehensive testing and driver mutations, but also integrate PD-L1 [programmed cell death protein 1] and TMB [tumor mutational burden] potentially into a treatment decision. How do you carry this out? The PD-L1 comes back first. Talk to us about your decision making when you’re getting all this information back. What do you start the patient on while we’re waiting for the tissue to come back and maybe a liquid biopsy hasn’t been done, and we’re trying to make a treatment decision for a symptomatic patient?

Zofia Piotrowska, MD: I think almost always the PD-L1 does come back first, and often we have that information and then have to figure out how to approach that, while waiting for the rest of the information from the molecular testing. To a certain extent, I think this does depend on patient characteristics, although I will say, in one of the discussions today someone made the point, that a word of caution, that it’s not all about smoking history anymore, that some of these targets are being seen in patients who have a significant smoking history. I think we have to be a little bit cautious about applying that, OK, for the young never-smoker, I’m going to wait for the molecular testing, but for everyone else I’m not. I think there may be some risk there.

For most of my patients, even if they come back PD-L1 high, I do think it’s still really important if patients can wait, to wait for the molecular testing to come back. We have good data to suggest that, for example, if a patient has EGFR and is PD-L1 high, they clearly still do better with TKIs [tyrosine kinase inhibitors]. And in fact, starting immunotherapy [I-O] as a first-line therapy may have some risk in terms of the subsequent toxicities when you introduce the TKI. I think there is some risk in jumping and acting on a PD-L1, even if it’s 90%, until you have all the information.

For those patients where we need to start treatment, there are some where they’re just too symptomatic, you feel that they’re going to decline, you can’t wait, my approach has been to say, well, chemotherapy is probably universally a reasonable option to give. I may give them a cycle, for example, let’s say it’s an adenocarcinoma, of carboplatin and pemetrexed. Even if they’re PD-L1 high, maybe withhold the immunotherapy until the molecular testing comes back, and then you can always add it with cycle 2. I think that’s one way to mitigate the risks of the immunotherapy and the switch to TKI if need be.

D. Ross Camidge, MD: I think that’s really good. There is a wonderful expression, if we start everybody on chemoimmunotherapy, we’re poisoning the well for some people if they turn up with an EGFR mutation next week.

Zofia Piotrowska, MD: That’s right.

Benjamin Levy, MD: I think we’ve learned a lot about toxicity, I think we’ve also learned about the lack of meaningful responses with at least single-agent immunotherapy for EGFR-positive patients.

Zofia Piotrowska, MD: Yes. Although I always think if it was just the lack of efficacy, I could say all right, well, let’s try it, if it doesn’t work we could always switch. But it’s the risk. And we’ve seen this not just with EGFR TKIs but with crizotinib. We see increased risk of hepatitis. This seems to be a recurrent theme that prior immunotherapy can poison the well for a subsequent TKI treatment.

Joshua Bauml, MD: There was an abstract I believe at ASCO [American Society of Clinical Oncology] Annual Meeting 2019 that showed that even in BRAF, with BRAF inhibitors that there was also increased toxicity when that was used after immunotherapy. I think the exact mechanism by which these 2 don’t play nicely together is not clear at this point to me, but something is happening. I feel like when we’re using these drugs, we need to be careful if we’re going to be using a tyrosine kinase inhibitor as well.

Benjamin Levy, MD: I’ll just throw this out to anybody. TMB. You know we’ve had a lot of interesting data on TMB as a predictive biomarker in tissue. We’ve had more recently some data from the MYSTIC trial looking at blood TMB as a biomarker for potentially dual checkpoint blockade. Is this something that has traction?

D. Ross Camidge, MD: Should we do a vote?

Zofia Piotrowska, MD: Yes, really.

D. Ross Camidge, MD: Who on the table routinely does TMB?

Benjamin Levy, MD: Not by choice. Well, I’m at an institution that rides on the immunotherapy coattails. They really have led a lot of the way with some of the immunotherapy drugs. Our NGS [next-generation sequencing] panel reflexively does TMB. Do I use it is another question. And if that, I would not have raised my hand.

D. Ross Camidge, MD: No one from Johns Hopkins Sidney Kimmel Cancer Center should watch this.

Benjamin Levy, MD: I don’t use it.

Joshua Bauml, MD: It’s a surrogate biomarker. It’s orthogonal to PD-L1. It says different information, but it’s a surrogate of a surrogate. Because really what we’re talking about is that the gold standard be TMB by whole exome sequencing, and that is a surrogate of neo-antigenicity, which is associated with possibly responding. What I’m finding is that it doesn’t add a whole heck of a lot. We don’t have any treatments that are FDA approved for which our treatment decision is guided by TMB, so I don’t order it. If someone comes and they have an NGS and it happens to have a TMB score on it, well, there are lots of things I can ignore on a piece of paper, right? Like I don’t pay attention to that number at all. It doesn’t affect anything I do.

Robert Doebele, MD, PhD: I think the critical question for the future is does it identify a patient population that’s not PD-L1 high that may have significant benefit from monotherapy? And that’s the real question that we need to answer in the future because it may identify another set of patients where we can de-escalate combination therapy plus I-O.

Zofia Piotrowska, MD: Or conversely, does it identify a group of patients where immunotherapy just really doesn’t work, and we shouldn’t be exposing them to the toxicities? I think that’s really the promise....

D. Ross Camidge, MD: The truth is that’s such a low hurdle that we choose to get over. If you’ve got a little smidge of benefit, we’re just going to add it in.

Zofia Piotrowska, MD: I think we need to learn more about that because we know these immunotherapy drugs have toxicities, and ultimately if it’s a biomarker that helps us select in either direction, I think it could be useful. But today I agree, it’s not.

Joshua Bauml, MD: What I would strongly prefer, rather than taking these 2 surrogate biomarkers where neither of them are very good, is to identify, try to figure out 1 that is working. When someone has an EGFR mutation and you give them a tyrosine kinase inhibitor, you can say to them, “You’re going to take this, this is going to work.” Like you’re very confident. When somebody has PD-L1 overexpression, you give them pembrolizumab monotherapy, response rate is less than 50%. It’s a coin flip. That’s not a great biomarker.

Transcript Edited for Clarity