This transcript has been edited for clarity.
Jacob Sands, MD: Hello. I'm Dr Jacob Sands, thoracic medical oncologist at Dana-Farber Cancer Institute. Welcome to Medscape's InDiscussion series on lung cancer. Today, we'll be discussing immunotherapy and lung cancer treatment. First, let me introduce my guest, Dr Jared Weiss. Dr Weiss is a professor of medicine at UNC's Lineberger Comprehensive Cancer Center, where he runs a joint lung and head and neck cancer section. Dr Weiss, welcome to InDiscussion.
Jared Weiss, MD: Thanks for having me.
Sands: Before we start diving into the topic of immunotherapy in lung cancer, I want to start out with your career path. What was it that led you to this point? I'm interested in some of the details of your journey.
Weiss: If you had asked me in medical school what I thought my career would be, at least in first year, I would have said private practice neurology, probably somewhere in suburban New Jersey. That's obviously not at all what happened. I made most of my decisions in life very analytically. I've made very few decisions intuitively, and they've been the most important ones I've ever made. The first intuitive decision was to become a cancer doctor. My plans to become a neurologist, probably a pediatric neurologist, were derailed with the first cancer patients that I met in medical school. They felt like my people … who I wanted to serve. Then it took a further twist. I thought I was going to be a melanoma doctor, and that came out of an interest in immunotherapy. As you know, when I was in training, that was the only cancer, maybe aside from renal, where immunotherapy was really a thing in solid tumors. The twist came when I met my first lung and head and neck cancer patients in fellowship and, again, very intuitively changed my mind. These were heavily stigmatized cancers but it felt like someone needed to care about these people. I was very intuitively drawn to wanting to serve them. At that point, I had this disconnect between my scientific interests and my human clinical passions, and that led me to be an efficacy-based clinical trialist for the first years of my career until happily, PD-1 [programmed death-1] became a thing in solid tumors; thus, our conversation today. And I became immediately involved in those efforts from phase 1 through phase 3. In recent years, I have been inspired by collaborators in the development of the science to focus on what I might loosely call personalized immunotherapy … the idea of making an individual product for each patient based on individual tumor biology, rather than selecting an off-the-shelf product based on tumor biology. Examples include therapeutic neoantigen vaccination, TCR [T-cell receptor], and CAR [chimeric antigen receptor] T cells.
Sands: It has been amazing what has happened in the last decade in lung cancer treatment. A significant portion of that has been in targeted therapies, which I think we need to get into a little bit before we discuss other things. Then, of course, immunotherapies: in the last 6 to 7 years, we saw checkpoint inhibitors initially in the second-line setting and now well established in the first-line setting. Before we dive into the first-line treatments, can you just discuss a little bit about the importance of genomic testing prior to decision making in the first-line setting and maybe include some of the specifics around that genomic testing? Do you do PCR? Do you do targeted [testing]? Do you do NGS [next-generation sequencing]? Give us a bit of insight into what needs to happen before we start looking at first-line treatment and immunotherapy considerations?
Weiss: We now have three basic classes of therapeutics: targeted therapy, immunotherapy, and chemotherapy; targeted therapy, in most cases, is our preferred option. There are details and nuance to this, but they tend to be the least toxic. They tend to have the highest response rates. They tend to have the greatest durability. And they demedicalize the life of the patient. Genomic testing is absolutely critical before initiating front-line [treatment]. The first reason is because of how wonderful targeted therapy is. But it actually goes further than that. Number two is that multiple of our genomic markers are not only positive predictors for targeted therapy efficacy, but powerful negative predictors against immunotherapy efficacy, particularly EGFR [epidermal growth factor receptor], ALK [anaplastic large-cell lymphoma kinase], and RET ["rearranged during transfection"]. But the data are evolving on other targets as well. The final reason why it's so critical is that in many cases, a TKI [tyrosine kinase inhibitor] is more toxic if given right after immunotherapy. So, for all of these reasons, it's absolutely critical that the first step in decision-making be molecular, even though the results for that come back after the PD-L1 [results]. I'm also very passionate about the kinds of testing. In CME, folks like you and me and the guidelines love to say "broad and sensitive" testing. And that's fine. But I think it leaves a little to be desired for those who don't think about lung cancer nonstop all day, as you and I do; I think that the standard really needs to be a DNA next-generation panel with an RNA fusion panel to have both optimal sensitivity and specificity. There are multiple commercial vendors that offer that. Those are the characteristics that we need to make sure the panel has.
Sands: To specifically round that out, you mentioned, for example, targeted therapies like osimertinib, which we've seen in combination with checkpoint inhibitors, PD1, PD-L1 [programmed death ligand-1] inhibitors, but that combination leads to a surprisingly high rate of pneumonitis, and with these checkpoint inhibitors with a long half-life, if patients start on this and then later switch over, they're essentially getting a dual therapy that has high risk. So, I absolutely agree. I start with this point because I think it's so important. To add to that, when I see patients and we need to do genomic testing, we're still doing the work-up, they're having symptoms and we really need to start treatment urgently, then I will do chemo alone as we finish out that genomic testing. So, this whole discussion we're going to have about immunotherapy is very exciting. It is a new era. This has really altered outcomes, but it needs to be couched within the context of that genomic testing having been done. There are nuances to the genomic testing and there are some genomic alterations that show up, where immunotherapy is still something we'd expect to have a benefit from. This is not just broadly "all genomics or no genomics," but rather "get genomics and look at the specifics." And then for those patients who are appropriate, now we're going to move forward with the discussion about immunotherapy, which has been truly revolutionary. About 6 or 7 years ago, we started with checkpoint inhibitors in the second-line setting. And this has now moved very squarely into the first-line setting. Why don't we start with pembro [pembrolizumab] as a single agent, going from KEYNOTE-024, which was PD-L1 >50% [expression], and then also KEYNOTE-042, which is PD-L1 at any expression? Can you go over those two studies in general and share your thoughts around single-agent pembro to start with?
Weiss: Those were the two studies that got pembro approved for the PD-L1–positive population. To my mind, the "who" is the most important question here. The FDA approval for pembrolizumab, as a consequence of KEYNOTE-042, is for any PD-L1–positive patient, but that comes as a result of a predefined and agreed-upon statistical analysis that I like to call "Russian nested doll" analysis, where you have the ≥50% enriched population. But then instead of looking at [patients with] 1%-49% [expression] discretely, you look at 1 to 100%, right? And if you take apart these Russian nested dolls and you look at KEYNOTE-042, it's absolutely clear that in the 50%-plus population, pembro is clearly superior for survival to platinum-based doublet therapy. It's less clear that that's true in the 1%-49% population. The survival there is fairly similar for monotherapy pembro and platinum-based doublet therapy. That might be fine in a world that had only pembro alone or platinum-based doublet. As a result of this Russian nested doll–type analysis, we have FDA approval in the 1%-plus population, but it's not clear that it's appropriate therapy for the 1%-49%. If you look at those survival curves, pembro and chemo seem roughly comparable for survival. And that would be fine if we lived in a world that only had pembro monotherapy or platinum-based doublet therapy. But as I'm sure you'll get to in a few minutes, we live in a world that has platinum-based triplet therapy that is probably superior for those folks. What KEYNOTE-042 told me, together with KEYNOTE-024, was that for PD-L1 50%-plus, it is better to give pembro monotherapy than platinum-based chemotherapy. It left unanswered the question of triplet versus pembro. And what really was left unanswered is whether 50%-99% is a homogeneous group in terms of benefit over platinum doublet or whether we really need to torture the data further into confessing a difference between the different levels of high expression. We had data that came out subsequently — I think the best data set was with cemiplimab — showing that the "very highs" may be driving most of that benefit; then, an FDA analysis (that I'm a little more circumspect about) at ASCO further strengthened that point. So, what I do in clinical practice with this not-totally-clear data in the 50%-100% population is ask myself how scared I am of progression in those patients with high but not very high [levels of PD-L1]? In my 50%-89% population, if I'm scared … if the patient has bulky central disease starting to compress an airway, becoming symptomatic, growing rapidly … that's where I'll give the patient triplet therapy, even though they have a PD-L1 that is at least 50%. The reverse can be said, by the way, in the 1%-49% group. If I have a patient with an unusually low fear factor for what progression is going to mean … say that patient who has a bunch of dots in their lungs proving metastatic disease, perhaps by a wedge biopsy, but really very low-bulk disease … I don't think it's wrong to consider pembro monotherapy alone in that patient. Now, there's a high chance that your first scan will show progressive disease, but it's also very likely, in that patient, that's going to be a radiologic event, not a human event.
Sands: That's a lot of ground to cover. And so maybe I'll just mix in KEYNOTE-189 and KEYNOTE-407, these being the chemo plus pembro studies for non-squamous and squamous [cell lung cancer], respectively, and share a lot of what you're describing. In the greater than 50% group, I think there's sufficient evidence of a benefit to single-agent pembro and a high enough likelihood, in that population, that it is a very reasonable first-line treatment and probably the preferred treatment in my mind, just focusing on pembro, in particular greater than 50% pembro. I also agree in the less than 50% [group], there's enough from that study to say pembro is a reasonable option if you have particular concerns about chemotherapy in select populations. But clearly, although it is approved with any PD-L1 expression, I don't have a lot of enthusiasm about pembro alone if patients can tolerate chemotherapy. I also want to highlight your point that if patients have greater than 50% PD-L1 expression, but they're having symptoms and you want more response, I think it's reasonable to add in chemotherapy, and I've done that in that population. Essentially, what you're describing, and what I'm saying I agree with, is not necessarily being beholden to one specific, rigid regimen in any scenario, but rather recognizing the specifics of that patient's situation and the situation of the cancer and that you have the option of either chemo plus pembro or pembro alone, depending on some of the specifics. But more broadly, greater than 50% [PD-L1]: pembro alone; less than 50%: chemo plus pembro. Is that a fair summary, underlining what you've already stated?
Weiss: Absolutely. What we have right now are noncomparative data sets. What we're doing here is cross-trial comparison. That's where the art of medicine comes in, in cross-trial comparisons to try to put together patient values, patient physical status, and the data to adjudicate a decision for each patient. I strongly agree that until we have head-to-head data on each of these questions in narrow situations, it's the art of medicine, not a defined board-testable answer.
Sands: It is exactly what we're doing with every individual that we're treating. We're having to look at different studies and we essentially do cross-trial comparison, not only broadly, but actually for that individual in front of us. So, it's even more complicated and difficult.
Weiss: Yes. You can have equipoise on a podcast. You can't have equipoise in an exam room, right? You certainly can talk about more than one. But at some point, with imperfect data, we have to come down on an opinion for an individual patient.
Sands: That's right. So, from the IMpower data, we've seen the atezolizumab combination treatment for first-line non-small-cell lung cancer. Can you walk us through a bit of that and some of your thoughts about atezolizumab and the atezo regimens?
Weiss: I think atezo is a fine drug as monotherapy for the high expressers in combination; for the low [expressers], it is very similar to pembro. The major meaning of that data is that it reinforces the basic story about PD access in access inhibition, and I think that a clinician who prefers one agent or the other can pretty comfortably use either and see each of the stories as reinforcing each other. The one place that I find atezo to have a unique data set is the quadruplet regimen of carbo, Taxol, bev, atezo. There was subgroup analysis on that trial. They allowed patients with EGFR or ALK mutations who had progressed on all available targeted therapy to go on that trial, and their subgroup analysis was mostly EGFR patients, with just a few ALK patients in there. But the subgroup analysis suggested that these patients seemed to uniquely benefit from immunotherapy with that quadruplet. And so, the places where I'm considering that regimen are in my patients with EGFR mutations who've gotten all possible targeted therapies and I do mean all. I'll leave that out of the scope of our podcast, but I push it hard, I'll say, for this context. The other place that I find that interesting is front line for my exon 20 insertion EGFR patients, where some platinum-based regimen is going to be used in the front line anyway. The biology of exon 20 insertion is really very similar to the classical actionable mutations. It's just different in druggability. And I extrapolate and consider it, in my practice, a preferred front-line regimen for those patients.
Sands: And to that point, I would love to see more transparent reporting of the subtyping of EGFR mutations included in these trials. Was that a bunch of sensitizing mutations … EGFR mutations? That's a different kind of a scenario. It's great to hear you specify some of that because I think I've seen people more broadly speaking to EGFR mutations as if it's sensitizing mutations in that study that were enrolled. I encourage anyone running these trials and reporting data to be very clear about some of the subtyping of EGFR within these patient populations, recognizing that those really are different scenarios. Although we're discussing immunotherapy, you did mention bevacizumab and, to me, this is a standout part of this regimen, and I think we need to discuss that. Is bevacizumab (bev) something that you utilize? Is it something you want to utilize? Does that factor into your decision on this atezolizumab regimen?
Weiss: It's not an agent that I have particular passion for using or not using, in the generic sense. There is some historic data that patients with EGFR mutations may benefit more from the addition of bev to cytotoxic regimens than other patients, so I certainly don't mind it in these populations. In the context of Impower, though, in the modern data, you really do seem to need the bev. If you gave just carbo, Taxol, and atezo, those patients didn't seem to benefit from the addition of the immunotherapy. I'm not even positive that clinical signal is really true. But to the best of the available shaky data I have, I do think you need the bev for now to get that benefit. And so, when you get to subgroup analyses in this way, and you dice the data thinner and thinner … we all demand these analyses and then we criticize how poorly powered they are. At some point, you just have to keep it simple as a clinician and do the thing that the trial is suggesting until you have better data. And so that's what I'm doing. I'm restricting quadruplet.
Sands: That leads us to the next topic of discussion … nivolumab/ipilimumab or nivo/ipi. We have the CheckMate 227 trial of nivo plus ipi, as well as CheckMate 9La, being nivo/ipi plus two cycles of chemotherapy as the caveat. The idea being now you get fewer cycles of chemotherapy — as opposed to four, now two. So broadly speaking, what are your thoughts on nivo/ipi and how does that add to your armamentarium of treatment options?
Weiss: Based on the predefined and agreed statistical plan, nivo/ipi is approved in the PD-L1–positive population, and two cycles of chemo with nivo/ipi are approved regardless of PD-L1 [level]. And I want to disclose directly that I'm about to speak about off-label use that has nothing to do with the FDA approval. These were the late entrants to the market. By the time these regimens were approved, we already had pembro monotherapy for the PD-L1 high [expressers] and triplet therapy for everyone else. So, the question isn't only "did these beat platinum-based chemo?" — they did. The question is also (and as a clinician consumer, I asked the question as well), "when does this new regimen look better than what I already have? Where am I tempted to change my practice?" Because, of course, I'm not alone … doctors are creatures of habit. And the place where I see the clearest answer here are the PD-L1–negative squamous cell cancers, where either of these regimens seems to me, in the omnipresent but forbidding cross-trial comparison, superior to what I already have. In comparing ipi/nivo to ipi/nivo plus chemo, beyond the difference in approval, I think it's important to reflect on what the chemo is buying us. The idea here in adding two cycles of chemo is to protect the left side of the curve. With every immunotherapy curve, there's a lot of progression at the first scan and the idea is that the two cycles of chemo would protect patients from that while waiting for the immunotherapy to kick in. I'm relatively unimpressed by the difference in the left side of the curve for PFS [progression-free survival] here, relative to the added toxicity from two cycles of chemo. So, to me, CheckMate 9La is a niche regimen that I've used exactly once since approval. In my practice, I consider off-label ipi/nivo for the PD-L1–negative squamous [cell cancers].
Sands: To fully round this out, of course, we now have cemiplimab. What are your thoughts on the cemiplimab data? You've referenced that earlier in the discussion, particularly around PD-L1 expression, but generally from the data that you see in the study, what are your thoughts on use of cemiplimab and of the data itself?
Weiss: I think it's great data. To me, it's very confirmatory of the data from pembrolizumab and atezolizumab. And what I think it really adds to our literature is the idea that not all PD-L1 highs are the same, that it is different having a PD-L1 of 50% or a PD-L1 of 90%. So, I think some cemiplimab is a fine drug. I have very little passion, though, for the question of use of cemiplimab or use of pembrolizumab for a particular patient.
Sands: What are your thoughts on TMB [tumor mutation burden]? Is there a place for it? Are super-high TMBs worth considering? If there's low PD-L1 expression but a high TMB, does that impact your expectations in any way?
Weiss: My take on TMB, as currently measured in commercial assays, is that it's not quite ready for prime time. There are some data suggesting you can use it when you're on the fence, but I am not doing so in my practice at this time. I'm pretty much ignoring it. I think you have a warmer take on it, right?
Sands: Well, I'll say that it's not something that I am using clinically. It's something that I've felt for a long time that there was just too much confusion around. And frankly, I think the point you are making is that different assays have different cutoffs. There isn't even a broad acceptance of what represents high TMB or where cutoffs might have clinical utility. We've seen differences in different data sets, but that nivo data I found particularly interesting with a low PD-L1 expression and high TMB … we actually had a poster at ASCO looking at a broad data set. It was broad data taking Flatiron data and Foundation Medicine data looking at TMB expression. With the particularly high TMB cutoffs, it looks like there may be an indicator of higher likelihood of durable responses. You've touched on this, and I love the way you described the curves as three sections of what we see within these checkpoint inhibitors, the first part of that is a drop-off. That really represents, as you stated, that these checkpoint inhibitors don't work for everybody. And if they don't work, we see rapid progression because the treatment is not effective. Similarly, you celebrated that right side of the curve that is inspirational in those really durable responses. For immunotherapy, I use a baseball analogy, where immunotherapy is like swinging for the fences. If you connect, you can get these amazing durable responses. But for those that don't connect, it can be a strikeout. And so defining which patients are going to benefit from these drugs becomes particularly important. I think we all have patients in our clinics now that are 5 years or more out from when they started their treatment, and they continue to come in and have good scans. Obviously, this is a small number of individuals, broadly speaking. But those really are the most fun to see. And the question then becomes, are these patients actually cured of their incurable diagnosis? I can't say one way or the other whether they are or are not. And the fact that that's a question itself is pretty exciting.
Weiss: I'm convinced we cure some patients. I have patients from phase 1 who haven't gotten a drug for more than 5 years who are alive, thriving, and without evidence of cancer. Lung cancer is an awful cancer. One of the only nice things that I can say about an awful aggressive cancer is that if it doesn't come back quickly, that's because there's nothing there to come back. If you're 5 years without drug with non-small-cell lung cancer or, for that matter, head and neck cancer, and you're still NED [no evidence of disease], thriving, and alive, you're cured.
Sands: I think that's likely true. It'll be good to see when we have 10 years of data, and I do expect that to happen. It comes down to choosing the patients that go on that treatment. Let's do a quick transition. We'll briefly discuss small cell lung cancer, because by that analogy — that whole swinging for the fences — we do see from the IMpower 133 trial and the CASPIAN trial, the inclusion of atezolizumab and/or durvalumab based upon those two trials is now the standard of care, with an impressive tail to those curves, although the median has not statistically significantly changed … I'd say not clinically substantially different. Can you highlight some of what has made oncologists celebrate those outcomes despite the small median changes.
Weiss: Low expectations? Small cell is a very frustrating disease. You and I share a passion for these patients and for improving their care. But unfortunately, almost all of our trials in the last decades have been negative, at least in terms of improvement in survival. It's so frustrating. All the time, you get these dramatic, even clinically complete responses. You know you're close to a cure and you also know that it's going to come roaring back. I think that we've been teased by the ability to cure people with small cell for a long time. And the idea that we can achieve durable control, even in a small number of patients, is very exciting. I like to make an analogy here to pizza. I love pizza. If I'm hungry and you give me a slice of pizza, I'm actually hungrier for the second slice after I'm done with the first slice, not less hungry. That's what immunotherapy in small cell has done for me. It has shown me what's possible, but now I'm hungry for a few more slices. I think we all want to cure more people.
Sands: I do have some patients that are now years out from first-line treatment, particularly the ones that had enrolled in the clinical trials. They now have more years out, in some cases, off therapy ongoing with disease control, with what I like to call "high-five" visits where they come in, their scans look good, and I say, "great, see you next time." And to that point, there are now different trials that are adding drugs to some of these in the first-line setting, as well as later lines, now with some bispecific, trispecific T-cell engagers. What are your thoughts on some of what we're seeing down the line that might be that next slice of pizza, as you described it?
Weiss: If you'll allow me yet one more analogy, I sometimes think about immunotherapy as a dramatic immune response … like a bonfire. And I think about our existing PD-1 and PD-L1 agents like gasoline to amplify an existing fire. The problem is, if you pour gasoline on an unlit log, all you have is a wet, smelly log. And so, I find the new crop of approaches that try to light that fire to be very exciting. You mentioned tarlatamab and I find this to be a really exciting agent. It's a personalized immunotherapy. It's going after a neoantigen that's particularly common in small cell [lung cancer]. Unlike some of the more unwieldy approaches that I spend my time on, it's off the shelf and we have real clinical data behind it. That bispecific, in addition to the later-line studies you mentioned, is also being studied in the front line, and I'm very excited about that. My institution has voted with our feet to try to open that trial.
Sands: And with that, Dr Weiss, it was wonderful having you on. You always describe things so well, and I really like these analogies. I'm going to bring them forward with me into practice. So, to summarize our discussion with immunotherapy in non-small-cell lung cancer, we started out highlighting the importance of doing genomic testing and had a bit of a discussion about what that genomic testing should look like. And then all discussion after is relevant, particularly in the population that should not start with initial targeted therapy. Going forward, we discussed PD-L1 expression and how that impacts first-line decision-making, with KEYNOTE-024 and KEYNOTE-042 showing that pembrolizumab as monotherapy was particularly beneficial in those with greater than 50% PD-L1 expression. We both expressed a preference for combination chemo plus pembro based upon KEYNOTE-189 and KEYNOTE-407 in non-squamous and squamous cell non-small-cell lung cancer, respectively. We discussed the regimens with atezolizumab from the IMpower studies, which also included bevacizumab. We then went on to discuss nivo/ipi based upon CheckMate 227 and CheckMate 9La, both of which are FDA-approved options. I mentioned particularly in these studies looking at low PD-L1 expression in [patients with] high TMB, which I found the most interesting data from those studies, which led into some of the biomarker discussion that I'll summarize in a moment. And then of course, we discussed cemiplimab, which is also now an approved option for first-line non-small-cell lung cancer. And this is a study that really highlighted increasing PD-L1 expression as something that indicated a higher likelihood of response to cemiplimab. And of course, some other data sets have shown higher PD-L1 expression leading to a higher likelihood of responses, showing that PD-L1 really is a spectrum in itself. I think we both agree that TMB is not something that we'd use in clinical practice at this time. I did mention that I do find it a compelling area of further study, and it may be that higher TMB cutoffs could be an indicator for higher likelihood of durable responses, which is really the homerun in the analogy that that I used. We then discussed briefly small cell lung cancer and, in the first-line setting, the use of atezolizumab with chemotherapy or durvalumab with chemotherapy is now FDA approved. And I think we both consider that standard of care in the first line, with a portion of patients having durable responses in this very difficult to treat diagnosis. And that led into a discussion of later lines of therapy in some of the studies, particularly in small cell lung cancer, highlighting tarlatamab, or AMG 757, as a drug that has really shown some compelling results. And with that, that wraps up our discussion. I thank you so much for tuning in. This is Dr Jacob Sands for InDiscussion.
Pembrolizumab Versus Chemotherapy for Previously Untreated, PD-L1-Expressing, Locally Advanced or Metastatic Non-Small-Cell Lung Cancer (KEYNOTE-042): A Randomised, Open-Label, Controlled, Phase 3 Trial
Medscape © 2022 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Immunotherapy and Lung Cancer Treatment - Medscape - Sep 15, 2022.