Classification of metastatic non-small cell lung cancer (NSCLC) based on biomarker analysis

DR LOVE: Welcome to this special CME program focusing on a recent survey conducted of 25 prominent US clinical investigators in lung cancer attempting to define how they sequence systemic treatments of metastatic non-small cell lung cancer. 

Drs Gregory J Riely, MD, PhD and Marc Ladanyi, MD co-chaired the project with me and helped select the participants and design the questionnaire, which asked respondents to describe their typical approach based on histology being squamous or nonsquamous, TPS PD-1 assay of greater or less than 50% and the presence or absence of 7 key genomic alterations: EGFR mutations, ALK and ROS1 alterations, BRAF and HER2 mutations, MET exon 14 and RET alterations. These biomarkers define an algorithm with 20 subsets of non-small cell lung cancer similarly to the way breast cancer systemic treatment is driven by the 4 major ER and HER2 subsets, so to begin I asked Dr Riely about the new taxonomy of non-small cell lung cancer.

DR RIELY: We’ve obviously proposed subsetting of lung cancer for a decade, really, starting with EGFR and really driving these molecular subsets, showing the pie chart.

But the challenge over the last year has been how we overlay the PD-L1 story onto it. And my gut instinct when you say we’re going to make an analogy to breast cancer is, oh, no, no. We’re different. But, in fact, we are the same now. And I think the fact that we’re using a genetic marker along with an IHC marker, and the vagaries of both those, I think it’s a perfect analogy to breast cancer. And now, unfortunately, we have a few more subsets than the breast folks do, which elevates the complexity. But I think focus on delineating what you have to know. And then how you bend patients based on those factors, it’s a great schema for categorizing patients.

DR LOVE: So I want to actually go through with you a little bit about each of these different subsets and what we saw in the investigators in terms of how they approach it.

I guess maybe we can start out with the histology. And maybe you can comment a little bit about the issue — it seems simple, squamous or nonsquamous, but maybe it’s not quite that simple. There are issues about adequacy, about adequacy of the specimen, to determine this, et cetera. But in general, how straightforward do you think it is to determine histology, squamous or nonsquamous?

DR RIELY: This is an estimate, but I would guess that 85% to 90% of cases, what you see in terms of histology on first blush is a great representation of the entire tumor. Most tumors are — they have their moments of heterogeneity, but they’re in general very similar. I think there are cases, though — and it’s not a tiny percentage, but there are cases where you get a sample that’s just not quite representative.

And so for an adenosquamous tumor, you get a squamous cell section. And that turns people off and turns people to think that this entire tumor is squamous and all of the things that we’ve learned about nonsquamous tumors don’t apply to this patient, when I think that that’s often jumping just a little bit too far. So I think histology is clearly important. And it’s a great starting point, but we can’t put too much stock in it.

Optimal testing platforms for patients with metastatic NSCLC

DR LOVE: And what about testing, particularly genomic testing, mutational testing, squamous versus nonsquamous? In terms of nonsquamous, in the past people would say, “We get EGFR, ALK and ROS1. If they’re negative, we,” quote, “send off NGS or multiplex testing. And we consider that in second line.”

But when we thought about it after having talked to so many people, it’s fine to send off — maybe it’s more economic to send off NGS and get, I don't know, hundreds of assays. But from a practical point of view, there are really only 7 things that really were going to change what you’re going to do. And, theoretically, if you had enough tissue, you could just assay the 7 alterations and not do a very multicentric assay. Any thoughts about that?

DR RIELY: Yes. I think you’ve hit upon a key thing, which is that these assays — we test for many more things that we need to know. And I think that’s actually a source of criticism by rational people, that we’re testing for more things than we know what to do with. So if we had 400 genes in an assay and we knew what to do with 9, what are we going to do with the other 400 or the other 391? And so I think that’s a reasonable concern.

I think the rationale for doing NGS is tissue efficiency. We make sure that we get as many tests done on 1 sample as we can. And that’s what we’re aiming for.

The challenge is that developing NGS assays is not the most straightforward process. And when folks develop these assays, they want to develop it for a broad range of tumors and they want to be able to run the same assay on breast cancer, colon cancer, lung cancer, melanoma, all these things, all in one fell swoop. And I think that’s the key to why the NGS assays are what they are today.

And our job as oncologists is to separate the wheat from the chaff, realize that p53 mutations aren’t something that we can make a decision based on today and recognize that when a provider recommends an mTOR inhibitor because of some mutation you’ve never heard of, we have to realize that it’s not the right next step for the patient. We have to recognize the mutations that we’ve heard of, the ones that we know have targeted therapies with proven efficacy and focus on using those.

Molecular profiling to detect targetable alterations in patients with newly diagnosed metastatic nonsquamous NSCLC

DR LOVE: So the other thing was the I-PASS issue, so to speak, of when do you think about — and we’ll go through these with the specific subset, but this kind of ties in a little bit, particularly, obviously, for the nonsquamous, where it’s more common — the issue of what do you really know in a stable patient who doesn’t need a treatment, like, right away this instant, what do you need to know to make a first-line decision?

In the past, we had — really, it was really EGFR, ALK and ROS1. But when we really started to ask people and we showed it here, we saw that BRAF and MET exon 14 seemed to be moving into the first line and that those were 2 other assays that ideally you’d want to know. Because we see a lot of investigators using first-line targeted therapy. Is that your take? And is that what you do?

DR RIELY: Yes, absolutely. I think the driving factor for including EGFR, ALK and ROS as our main things we’re looking for is really about FDA approvals. And those were the targets for which FDA-approved agents were available and specific approvals for those indications were available. So I think that’s what drove that.

And as you’ve said, BRAF has moved into that space, because it’s a newly approved target in non-small cell lung cancer. And I think we’re seeing MET exon 14 move there. And I have to admit, I’m impressed by the use of crizotinib for MET exon 14 in the first line in the survey, because the data is relatively modest that we’ve seen out there. And I think that probably speaks as much to the comfort of docs using crizotinib as it does to the strength of the data around MET exon 14. Because one can imagine that the MET exon 14 approval would happen even in the next year. So we’re all jumping to this sooner than then.

Choice of first-line therapy for patients with metastatic squamous NSCLC and no targetable mutations

DR LOVE: So I want to go through also the specifics in terms of these various subsets, in terms of at least what the investigators are doing right now. And I thought we could start at the top of this graphic and kind of work our way down, beginning with the majority of the patients who do not have a targetable mutation, so the so-called wild type.

And there what we saw was a couple of interesting things, and I’m curious what your thoughts are about it.

The first is, in terms of squamous, it’s kind of straightforward, really based on TPS status. If it’s over 50%, they’re going to give a checkpoint inhibitor, pembrolizumab. If it’s not, they’re going to give chemotherapy. Any comments about that and also the issue of genomic testing in patients that you feel pretty comfortable you have adequate tissue — they’re squamous, pure squamous. Maybe they’re a smoker. For a doc in practice, is there really any mutation testing, if a patient’s not going to go in the trial, that would be considered standard in a patient with squamous?

DR RIELY: I think it might push the definition of “standard,” but I think it’s a very reasonable choice to test all patients who have any histology of lung cancer, to test all patients with a mutation platform that identifies all of the mutations of interest. And I think this is based on really the fact that almost all patients with metastatic disease are diagnosed based on a biopsy. And the biopsy is really — while it may be representative, we don’t know.

So perhaps the exception would be a patient who was biopsied or diagnosed based upon a large resection, and then this was a recurrent disease site. And in that large resection specimen, there was no evidence of adenocarcinoma — it was a complete squamous cell carcinoma. That patient, the rare patient like that I probably wouldn’t do mutation testing. But for every other patient with squamous cell carcinoma, I would do mutation testing.

Now, of course, I acknowledge that the identification of mutations is going to be relatively uncommon for these patients. And so I wouldn’t push to hold up first-line therapy for these patients, but I would want to know this information when it came time for second-line therapy.

Ongoing investigation of immune checkpoint inhibitors with chemotherapy as first-line treatment for metastatic squamous NSCLC with no targetable mutation

DR LOVE: Another thing, obviously, that’s very different about squamous versus nonsquamous is, in nonsquamous we have an indication for chemotherapy and checkpoint inhibitor. In squamous, we don’t. But I’m curious. Do you think that’s coming? Do we have any data out there right now on that strategy in squamous? And what do you think we’ll see over the next couple of years?

DR RIELY: I think this is easily the hottest area in lung cancer. There are a considerable number of trials going on right now that are looking at the combination of chemotherapy plus checkpoint inhibitors. And we’re going to see readouts on data, the larger Phase III trial with pembrolizumab, this year. We’re going to see data with atezolizumab. We’re going to see data with durvalumab. All of these agents will have combination trials that we will see results of over the coming year.

So I think we’re going to learn a lot more about these in general, and particularly for squamous cell lung cancer. Most of the trials have included an arm that gives patients with squamous cell lung cancer a combination of chemotherapy plus immunotherapy. Now, whether this approach pans out we’re going to see, but we are going to have lots of data to inform the decision, I think, over the course of the coming year.

DR LOVE: What about the issue of first-line therapy? Again, it looks pretty straightforward, mainly based on the FDA requirement of a PD-L1 assay over 50%. But from the data that you’ve seen up to this point, could you envision situations clinically where you might want to use a checkpoint inhibitor up front in a patient who has a TPS score less than 50%? For example, a patient who has minimal disease burden, is asymptomatic, maybe an older patient that you want to get away from chemotherapy? Or is that 50% such a watershed that once you start getting below it, it really doesn’t match up to chemotherapy?

DR RIELY: I think we have to recognize that PD-L1 staining, like any immunohistochemistry, is very much an art and not a science. Unlike with the mutation tests, where it’s positive or negative, PD-L1 testing is based upon intensity of tumor staining. And again, it’s a sample. So you’re looking at one little section of one very large tumor. And you have a number you come up with.

So I think we have to take all of this with a grain of salt. And I think that if you have a tumor that’s PD-L1 30%, it’s not irrational to think that a different biopsy site would get you a PD-L1 of 60%. So I think bearing that in mind, I think when I see something that’s a little bit lower than 50%, when I see something that’s in the 30% to 40% range and I see a patient who, as you suggest, might have a harder time with chemotherapy — just last week in clinic I met a gentleman who was 86 years old, had recurrent lung cancer and had no interest in chemotherapy. His PD-L1 was 40%. It’s hard to imagine that I really need to try chemotherapy. And he’s asymptomatic. And it’s perfectly reasonable to give a try with immunotherapy at this point. So I think I’m exploring that with this patient, trying to see if various approving insurance companies will agree.

Duration and level of response to immune checkpoint inhibitors based on PD-L1 tumor proportion score (TPS)

DR LOVE: And I’m kind of curious. When you see a patient like that or a patient who has 50% or 60%, in your own mind — and maybe even if the patient asks you — of course, the thing we’re really looking for is a prolonged period of disease-free progression, whether there’s a response or not. But what’s so fascinating about immunotherapy is, you see patients who go a lot longer than you might see with other forms of systemic therapy, like chemotherapy. I don't know how long we’re talking about, maybe a couple of years, or maybe even longer.

When you see a patient like your patient and you think about that, in your mind, what do you think the chance is that they're going to have a prolonged period without having to switch therapy or maybe even go off therapy?

DR RIELY: Yes. I think for the patient with a high PD-L1 staining, I think they have a very high chance. So I’m going to guess it’s 50% to 60%. This is all early data. We don’t have good answers for this, but that’s my suspicion and that’s my experience thus far with first-line immune checkpoint inhibitors.

I think when we move the cut point lower, it gets more complicated. And I think one thing that’s important to remind ourselves is that there have been trials using nivolumab that used a cut point of 5% in the front-line setting. And that trial didn’t show superiority of immunotherapy. Now, granted, the outcomes were very similar between immunotherapy and chemotherapy, but it didn’t show superiority.

And, similarly, we’ve seen the top-line data from the MYSTIC trial, which compared durvalumab to chemotherapy. And they used a cutoff of 25%, and again, that was not a positive trial. That was not a superior thing, to give immunotherapy in that setting.

So we have trials of 5% and 25% saying immunotherapy was not better than chemotherapy. And now the only trial that’s positive is one that had a 50% cutoff. So we should respect that 50% cutoff but understand the data upon which it’s based.

Use of immune checkpoint inhibitors in combination with chemotherapy as front-line therapy for metastatic nonsquamous NSCLC without a targetable mutation

DR LOVE: And maybe moving over to nonsquamous, again, patients without targetable mutations. Of course, because of the presence of the FDA approval and the option of chemo plus checkpoint inhibitors, that became a little bit different. And what we saw there is that for patients who had TPS scores less than 50%, there was kind of a split. About two thirds of the investigators in general are going with chemotherapy. And about a third are going with chemotherapy, carbo/pem and pembrolizumab.

But interestingly, when we turned — and, of course, this is just their general practice pattern. We know that people vary based on what the individual situation is. But when we said the TPS was over 50%, the vast majority of investigators would use a checkpoint inhibitor alone and not use the combination.

So I’m curious, first of all, how that lines up with your practice and what you think the content forces are that are determining it.

DR RIELY: So I think first, I’ll say that when I looked through the entire survey, the thing that jumped out at me most was how people have taken up the combination of carboplatin/pemetrexed and pembrolizumab. I personally had not taken that into my practice. And I think when I view use of immunotherapy and chemotherapy, I view very much that we’re going to consider this for all patients. And is it a — it’s a sequential approach, where you give it chemotherapy first, followed by immunotherapy? Or is it a combined approach, where you combine them all together?

And I think time and again, in clinical trials that we’ve done in patients with non-small cell lung cancer, we’ve seen that if you combine 3 agents, you always get a higher response rate. You always get a better progression-free survival. But it comes with the cost of excess toxicity as compared to just doing a sequential approach.

A good solid one third of my colleagues have already adopted the combined approach up front. Clearly, we can imagine a setting in which the combination is the best thing. And I think some of the survey data showing that in those patients who are highly symptomatic, where you feel like you don’t have a lot of opportunities to try multiple therapies, that doing that combination up front really would be the hot place for these patients.

DR LOVE: I’m curious about kind of what you hear about the rationale for those people who are using the triplet. One thing that I could envision is the question of is the patient going to get to second-line therapy? I don't know in your own practice or what numbers we have about people who just don’t get to second-line therapy. What we’ve seen is that they usually do.

But if you say to yourself the thing that’s really going to make the difference to a patient is whether or not they have a prolonged period of disease-free benefit and the only thing that’s going to get that to them is a checkpoint inhibitor and from an FDA approval you can’t use it first line, does that kind of thinking kind of work for you?

DR RIELY: Yes. I think it makes sense. And I think that we do want to get patients that great response. But we do sometimes see those great responses with chemotherapy. And I think one of the other important things for me is, I think about how to incorporate this combination data into my practice — is, I’ve got a lot of patients who respond. And then when they respond, if they respond to the triplet, what are we going to continue? Are we going to continue chemotherapy? Are we going to continue immunotherapy? How long do we continue any component of that?

And I think that that gets much more complicated. And that’s why the combined approach is more challenging, because you don’t know what they responded to. If you took a sequential approach and you saw that they responded beautifully to carboplatin/pemetrexed and then you were able to continue that for a year or two, then you’ve done really great things for that patient.

But if you give them the triplet and they respond and you’re not sure what they respond to, then you continue with the carboplatin/pemetrexed and pembrolizumab for a little bit, and then at the maintenance point you try to figure out: Do you continue pemetrexed and pembrolizumab? Then they’re continuing a drug that may have immunotherapy side effects when they’re not necessarily deriving the benefit.

And you can make the contrasting story as well. When somebody’s clearly going to benefit from immunotherapy but, for whatever reason, their PD-L1 was low, wouldn’t it be better if they were just getting immunotherapy and not getting chemotherapy as well?

Selection of EGFR tyrosine kinase inhibitors (TKIs) as first-line therapy for EGFR-mutated NSCLC

DR LOVE: So let’s move on now and talk about the subsets defined in terms of the presence of genomic alterations. Let’s talk about what we found in these genomic subsets, beginning with EGFR.

And, of course, there we saw first-line therapy. At this point that’s pretty straightforward, I think, in terms of generally being targeted therapy. Interesting, though, that these data that we were collecting were also in light of TPS scores. And we’ll get into a little bit what we saw here in general in terms of the use of immunotherapy in patients with targetable lesions.

But now we’re looking at, even with high TPS scores, that people are going with anti-EGFR TKIs, most commonly erlotinib followed by afatinib and very little but some gefitinib.

Can you comment a little bit about the first-line decision, the choice of agent based on the type of mutation and whether or not the TPS score really means anything in this situation?

DR RIELY: So as you approach the patient with EGFR-mutant lung cancer, we have a broad variety of data supporting the use of EGFR TKIs. And in general, the data seems to be strong, independent of which type you pick, whether it’s erlotinib, gefitinib or afatinib. So I think that’s what we walk into this with.

When we think about first-line use of pembrolizumab, say, for patients with high TPS, we know that those patients were actually excluded from the first-line trials. So patients with EGFR mutations were not allowed in the first-line trial. And so we kind of have to set that group of patients aside as a front-line therapy.

Now, we do know a little bit about how these patients respond to immunotherapies, though, because we’ve looked at how they’ve done in the second-line trials. And so in trials where immunotherapies were randomized against docetaxel, patients who had EGFR mutations routinely were on the side of things favoring docetaxel in every single trial, whether it was the nivolumab trial, the pembrolizumab trial, the atezolizumab trial. In each of those trials, patients who received docetaxel did a little bit better than those patients who received the immune checkpoint inhibitor.

So that tells us that these patients with EGFR mutations are a little bit less likely to respond to immunotherapy. And that colors our decisions, whether it’s first line or second line, of whether we want to use immunotherapy or conventional therapy, such as EGFR TKIs or chemotherapy. So I think that really sets up a paradigm where we start with an EGFR TKI for these patients.

We look for that T790M at the time of resistance. If they’ve got that, then we move on to osimertinib, a second-line EGFR TKI. And if they don’t have T790M, then we have that decision point of immunotherapy versus chemotherapy. But based on the absence of data to compare, we all generally lean toward chemotherapy in that setting.

Types of EGFR mutations and activity of EGFR TKIs

DR LOVE: So I want to tease that out a little bit more. But first, can you comment a little bit more though on the type of EGFR mutation and choice of first-line EGFR TKI and also what we know about osimertinib in the first-line setting in the patient without a T790 mutation and the upcoming FLAURA trial that I know will be presented very soon?

DR RIELY: Sure. So when we think of up-front therapy for patients with EGFR mutations, we have a broad array of choices, erlotinib, gefitinib and afatinib. And we also know that there are a couple of different types of EGFR mutations. The 2 most common mutations that represent about 90% altogether are EGFR L858R and EGFR exon 19 deletion. The exon 19 deletions, those patients tend to do a bit better with EGFR TKIs in general than those patients who have EGFR L858R.

There have been explorations of whether a particular TKI is better for that group of patients. And there was a big splash a couple of years ago with the EGFR exon 19 deletion analysis, where it was combined from a couple of different clinical trials. And when we compared those patients or took the patients with EGFR exon 19 deletions and compared those who got EGFR TKI to those who got chemotherapy, this is actually the first setting in which we’d seen an overall survival advantage for EGFR TKI versus chemotherapy.

So it was, I think, appealing to a lot of folks. I think that was probably a universal phenomena associated with benefit of EGFR TKI in patients with exon 19 deletions and not something specific to afatinib. But that rationale is why a lot of folks feel like for patients with EGFR exon 19 deletions we should consider afatinib as a superior agent. I think the data is out. And there hasn’t been a direct comparison of erlotinib and afatinib.

There has been a direct comparison of afatinib and gefitinib, and there were some suggestions overall that afatinib was better in terms of response rate and progression-free survival, but with an overall survival endpoint there wasn’t a difference between those TKIs. And so I think it remains unclear whether there’s a superior front-line EGFR TKI.

And now you introduce into that the osimertinib story. So osimertinib is what’s often called a third-generation EGFR TKI. And this was specifically developed to target the most common pathway of resistance, the T790M, but it also has activity as a first-line drug.

Now, there’s an ongoing trial, a randomized trial called FLAURA, where first-line TKIs, such as erlotinib/gefitinib, are compared to first-line use of osimertinib at the time of diagnosis. And I think this is a trial that we’re all really excited to hear the results of. We know the top line is that it was a positive trial, but we don’t know any of the details. But I think we’re all excited about it.

Efficacy and tolerability of the EGFR TKI osimertinib as first-line therapy

DR RIELY: What we know from a subset of an earlier trial is that it’s likely that osimertinib has very good activity in the first line. And we have data suggesting 1-year progression-free survival of around 70%, which is significantly better than what we’d expect to see for something like erlotinib, gefitinib or afatinib. But that was a small trial, a select group of patients. It’ll be interesting to see how this pans out in a much larger study. And it sounds like we’re going to have these data in the next month or so.

DR LOVE: Yes. It’ll be really exciting to see that. It looks like it is going to be, quote, better. And in kind of reflecting back on the alectinib story in ALK first line, do you think this is going to be better because it’s, quote, a better EGFR inhibitor biologically or that it’s better in the brain, or both?

DR RIELY: I think it’s going to be better biologically, and one of the great features of osimertinib is that they’ve dialed out a lot of the wild-type EGFR effect, and so that’s a lot less rash, a lot less diarrhea. And my suspicion is that if it’s better — and one of the mechanisms it’ll be better is that you’re just able to get better sustained treatment for more patients. And so more patients at higher dose levels are probably going to do better than if you had to dose reduce for wild-type toxicities.

Another clear path to benefit for patients is that when we treat with erlotinib, gefitinib or afatinib and patients become resistant and they have progressive disease, we biopsy. We look for T790M. And we probably find T790M in about half of those patients. And then they go on to osimertinib.

If you start up front with osimertinib, you avoid that point where you have to find the T790M. And you avoid the half of patients already having to move off to a chemotherapy or immunotherapy. And so there’s a potential benefit overall for that approach.

Mechanisms of acquired resistance to osimertinib

DR LOVE: So we’ll talk about the current algorithm in terms of T790 testing, but I’m curious, if the FLAURA trial is so positive that people end up utilizing it first line, does that mean on disease progression they won’t do T790 testing?

DR RIELY: I think we’re going to have to learn a lot more about that. We know relatively little at this time about mechanisms of resistance to osimertinib. We’ve actually, in the biopsies that have been reported, we’ve seen a lot of patients lose T790M in the resistance setting. And we’ve seen emergency of a mutation called C797S, which is a mutation at the binding site of osimertinib. And so these mutations are of interest. And I think we have a lot to learn to really teach us how to take care of these patients if that is our first-line approach going forward.

DR LOVE: I’m curious, also — and I guess specifically in this situation, even the second-line situation, T790-positive, I’ve heard it said that in some cases maybe the mechanism of resistance is development of, like, HER2 or MET exon 14. Is that true?

DR RIELY: So not MET exon 14, but HER2 and MET amplification.

DR LOVE: MET amplification.

DR RIELY: So amplification of wild-type MET.

DR LOVE: Okay. Right.

DR RIELY: And so when you look at patients with resistance to erlotinib, gefitinib or afatinib, we see HER2 amplification and MET amplification as clear evidence of resistance. But their frequency is relatively small. So for MET amplification, it’s probably around 5% to 10%. For HER2 amplification it’s similar, around 5% to 10%. And if we go forward with osimertinib as a first-line approach, you can imagine that we’re going to enrich for these non-T790M mechanisms of resistance. And potentially MET amplification/HER2 amplification will be targets for resistance to osimertinib.

Response to EGFR TKIs in patients with CNS metastases

DR LOVE: I want to just elaborate a little bit more, though, which is the relative efficacy in the brain of osimertinib versus the first-generation agents, like erlotinib/afatinib, and how often CNS progression is the reason for recurrence, for example, in contrast to ALK.

DR RIELY: Right. I think when you look at ALK-positive lung cancer, the frequency of patients with brain metastases is high, like 20%. And that’s been a real focus of drug development and a focus of how we take care of patients, is identifying those brain metastases and surveilling for new brain metastases in patients with ALK. That’s very much a novel thing for patients with ALK-positive lung cancer.

Strangely, even though EGFR mutations are 4 times as common as ALK gene rearrangements, we have a lot less data in this space. What we do know is that patients with EGFR-mutant lung cancer do respond. Patients who have brain metastases with EGFR-mutant lung cancer do respond, and relatively well, to EGFR TKIs, even the first-generation ones, erlotinib, gefitinib, even afatinib. And more recently we’re seeing a lot of data supporting the use of osimertinib in brain metastases as well as in patients with leptomeningeal disease.

I think the quality of the data to support that erlotinib, gefitinib and afatinib are effective in the brain is very modest. It hasn’t been a focus of clinical trials. And so it’s harder to tell us how good these drugs are there. My clinical experience, though, is that these drugs are very good in brain metastases. And we can see complete responses with erlotinib, gefitinib and afatinib. And so I think these drugs have excellent efficacy, but at the same time, there’s preclinical data that suggests that osimertinib is very effective in the brain, that the preclinical data suggests that it concentrates in brain tissue and so it may be even more effective than some of these other agents. Right now it’s provocative, but there’s no solid data to tell us one way or the other.

Plasma versus tissue genotyping for detection of T790M mutations

DR LOVE: So another issue with the biomarker-driven algorithm that we are looking at here is the issue of disease progression on a first-generation TKI and the issue of T790 testing. And, of course, it’s no surprise that T790 testing in general is absolutely standard, no question about it. Whether it actually happens in practice, I think, is another question. But certainly it’s happening with investigators.

But we actually saw a little bit of a refinement on just the T790 question, which is how you do it. And what we actually saw in the survey is that at this point, pretty much all the investigators are doing plasma testing on progression. And if it’s negative, they’re going to get tissue. The only difference we saw is, interestingly, there’s some docs who do both right from the beginning. So they go after the tissue and they send off the plasma. Others send the plasma, and then they wait. If it’s negative, I just thought it was kind of interesting there are people who do it simultaneously.

First of all, any question about this issue of it’s standard and there’s really — you should start out with at least plasma and the issue of do you wait or just go ahead and do both together?

DR RIELY: So I think you’re absolutely right that EGFR T790M testing is absolutely the standard for patients who have EGFR-mutant lung cancer and have progressed on erlotinib, gefitinib or afatinib. This is one of the great advances in understanding how we treat lung cancer. And if we don’t look for T790M, docs are making a big mistake. So I think definitely looking is key.

Now how we look is, as you said, it’s a new era. We for the longest time have biopsied patients looking for this T790M trying to understand mechanisms. And now we have a drug that we can target T790M with, and it’s very exciting. But we have this new technology that we can identify T790M in plasma. And, of course, that’s great, because then you can avoid the process of biopsy. And biopsy followed by DNA extraction followed by sequencing takes a while. And if you compare that to just doing plasma DNA, where you can get DNA out faster, it’s actually a much faster process.

So I think we have these two competing platforms. Our sense is that the specificity of both is near perfect, so that if we identify a T790M in plasma or if we identify T790M in tissue, we should take it and treat the patient based on that. Now, it’s harder when you have a negative result from plasma. The question is, do you believe a negative result is reflective of the tumor?

And I think what we’re beginning to learn about testing for plasma DNA is that some tumors shed DNA into plasma, some don’t. We think we might have some hints that pleural-based disease doesn’t shed so much DNA, bone disease maybe more, but we don’t have solid answers for that. And so we can’t take a negative plasma DNA result and stop there. I think that from a practical perspective, we can test plasma and then test tissue, or we can start the process for both of those at the same time and just see which one comes out positive first.

I think to some extent, it depends on the patient sitting in front of you. Are you looking solely at scan-based progression that you could even watch for a little while longer? For that patient I’ll definitely order a T790M. And if we don’t get the result of a T790M-positive and change to osimertinib, if we don’t get that then we’ll just continue to follow them on erlotinib, gefitinib or afatinib. And I think that that’s the approach that I take.

Now, I do typically try to start the process for both plasma testing and biopsy at the same time. That said, I’m watching for the plasma results. And typically at my institution, the plasma results are back before the patient has gone for the biopsy. And so it’s a little bit of the best of both worlds.

Response to osimertinib in patients with T790M-negative metastatic NSCLC

DR LOVE: What about the situation — and I’m not sure how often it occurs — where maybe the patient’s had a really good response, now they have progression. But where they have disease is very difficult to access for a biopsy, maybe a bone in a difficult area. Maybe it’s an older patient. Are there any situations where you can depend upon the plasma testing? I’ve heard people talk about, “If the EGFR mutation is picked up in the plasma but not the T790, that’s more likely that it really is a negative.” And what about the possibility — I don’t even know if you can get it paid for — for empiric osimertinib, where you really just can’t get tissue?

DR RIELY: Right. So I’ll start with the empiric osimertinib. I think it’s an appealing notion that if 50% or 60% of patients have T790M, why not try this? Unfortunately, the clinical trial data don’t really bear that out. If we look at the clinical trial data in the early-phase trials of osimertinib, they treated patients who did not have T790M.

And what they found was that the response rate was low. It was about 16%. And the duration of response, or progression-free survival, was short. I think around 2 months was the median progression-free survival. When you compare T790M-negatives to those patients who have T790M, there was a much longer median progression-free survival. So I’m resistant to do empiric osimertinib, but I think, like you said, sometimes your back is against the wall. And if you can get it paid for, it sometimes makes sense as a reasonable approach.

DR LOVE: What about this issue of looking at the plasma and, if you see the mutation but not T790, feeling more comfortable that it probably is negative?

DR RIELY: I think that’s a very seductive idea that’s probably wrong.

DR LOVE: Hmm.

DR RIELY: And the reason it’s seductive is, you say to yourself, “If we can detect the exon 19 deletion,” for instance, “then we should be able to detect the T790M.” The 2 factors that are at play here are, (1), whenever we do a biopsy and look at the relative ratios of 19 deletion to T790M or L858R to T790M, the T790M is always a lot less frequent, on the order of less than 5% to 10% as frequent as the underlying EGFR mutation. So number one, they’re always a different ratio.

And the second is, back to the idea of the tumor biology and what dictates plasma DNA shedding, we see differences in resistance patterns. And it may be that the troublesome lesion that is growing and causing trouble is not shedding the T790M, but it has T790M. And the other sites of disease are shedding the EGFR exon 19 deletion just fine. But that one site, the one that’s going to cause you trouble isn’t shedding.

DR LOVE: So it was also interesting to look at how people manage these patients after progression.

Incidentally, you talked about T790 being standard on progression on a TKI. Does it matter whether they responded or not? Of course, most people do respond. But sometimes you see people, I guess, who progress through, but even in that situation, T790 testing?

DR RIELY: That’s a good point. For people who progress right through an EGFR TKI, I don’t look for T790M. It’s part of when you test an initial tumor. So you know whether or not they have it at baseline. But if they didn’t have it at baseline and they aren’t benefiting from an EGFR TKI, I think there’s something about their tumor that is resistant to this approach. And whether or not you find the T790M, I think it’s less likely that that patient will benefit. So I don’t typically push for it as the next step. I might come back to it down the road, but not right up front. I’ll quickly move to chemotherapy for most of those patients.

Therapeutic options for patients with T790M-negative metastatic NSCLC

DR LOVE: So when you look at how people manage patients based on the T790 results, I think it was kind of interesting.

If they’re T790-positive, it’s pretty straightforward: Everybody uses osimertinib. But what’s interesting has been the T790-negative. And in spite of the fact that in general what we saw with this survey was, as you were talking about before, kind of a delay in using or sequencing a checkpoint inhibitor in patients, particularly with the 5 driver mutations that really are getting benefit from targeted therapy, the idea of less mutational load, et cetera, et cetera.

But, interestingly, in the T790 situation, there’s a little bit of a difference about how they approach treatment based on the TPS score. So when we said, “If it was 10%,” or on the low side, most people were just, “Chemotherapy plus or minus a biologic.”

But when we bumped up the TPS score to over 50%, still most people were using chemo and a biologic, but now we’re seeing about a quarter of the docs start to bring in a checkpoint inhibitor, specifically in combination with chemotherapy. Any thoughts about that?

DR RIELY: I think to see them use chemotherapy in combination with a checkpoint inhibitor in those patients who have a high TPS really — and in the patients with high TPS who don’t have EGFR mutations, they never recommend the combination, right? If you look at the data for patients who have TPS of 60% without a driver, 100%, virtually, give pembrolizumab alone.

In contrast, for those patients, TPS 60% with EGFR mutations, nobody’s still recommending pembrolizumab alone, but some are giving chemo plus pembrolizumab. They’re hedging their bet.

DR LOVE: Well, I should say 8% actually give a checkpoint inhibitor alone, so much less than the combination, but not zero.

DR RIELY: Okay. Right. Right. And I think it’s indicative that people really aren’t having a high degree of faith that patients are going to respond to immunotherapy. But that said, I think when we have patients in front of us, if they have a very minimally symptomatic or asymptomatic progression and you feel like you need to change therapy and the patient really wants to try immunotherapy, sometimes it’s hard to convince them otherwise. And one of the best ways to help everybody agree is to try a short course and do 6 weeks’ worth of immunotherapy and see how the patient does.

As long as the patient doesn’t have a lot of symptoms and you have a little room, I think, particularly for patients with EGFR-mutant lung cancer, they’re often a bit slower growing. It’s reasonable to give them a try on immunotherapy.

Choosing between targeted therapy and immunotherapy for patients with metastatic nonsquamous NSCLC and actionable mutations

DR LOVE: So I want to go through the other subsets, particularly, specifically the ones with genomic alterations. And rather than get into specifically a lot about the specific choice of agent — we’ll talk about that — but more about the type of therapy that people are sequencing.

So the next thing we looked at was actually ALK, but also ROS1 really fell into the same paradigm, which was, regardless of TPS score, the targeted therapy was used up front. Any comment about this and, again, not so much the issue — I think people have kind of come to terms with the issue of targeted therapy versus chemo.

But I wonder if everybody in practice has really come to terms with the issue of targeted therapy versus checkpoint inhibitor. Now, as you mentioned, the indication specifies ALK-negative. I’m not sure about ROS1, actually. Do you know?

DR RIELY: That’s right. ROS1 was not a specific exclusion factor for that trial.

DR LOVE: And as you go down the list, BRAF, MET, RET, HER2, none of those were addressed in the trial, just ALK and EGFR. But in any event, the investigators seem to go a straightforward route in terms of first-line therapy, again, regardless of TPS status, targeted therapy.

Any comment, first of all, generically, about this approach to these two alterations?
And then maybe you can comment a little bit about choice of agent, particularly as it relates to ALK.

DR RIELY: So with EGFR, we have these relatively suggestive data that’s very uniform across a variety of second-line trials suggesting that for patients with EGFR, chemotherapy is probably better than immunotherapy. And again, those patients weren’t included in the first-line trials.

Now we’re taking the next step, most investigators, to say that ALK, ROS, BRAF, they’re probably the same. And we should probably treat those patients with a TKI first. Now, in truth, there’s even less data to support that approach. And I think we’re beginning to learn a bit more about this, but the data is very minimal. And I think we all have patterns that we approach this with, but I think none of us can be absolutely certain that putting immunotherapy first for a patient with high TPS and ROS1 or high TPS and MET exon 14, it’s not clear.

Management of lung cancer in patients with ALK and ROS1 genomic alterations

DR LOVE: So can you comment briefly on the issue of choice of first-line and second-line therapy in ALK and ROS1?

DR RIELY: So beginning with ALK, choosing an ALK inhibitor got a lot more complicated with the results of the recent global ALEX trial. Now, we had, a year ago, the results of the Japanese ALEX trial that compared alectinib to crizotinib for patients with newly diagnosed ALK-positive lung cancer. In the Japanese ALEX trial, there was a significant improvement in progression-free survival for alectinib versus crizotinib. We all wondered whether that might have been a Japan-specific phenomena. And we awaited the global ALEX results.

We saw the global ALEX results at ASCO 2017. And the results were just as impressive. We saw a more than doubling in median progression-free survival. The numbers are very impressive. And just as impressive as the overall numbers are the numbers for those patients with brain metastases from ALK-positive lung cancer — for those patients, again, a dramatic improvement in progression-free survival.

I should acknowledge that there’s no difference in overall survival when you compare crizotinib to alectinib, but nonetheless, the alectinib patients had a much better response rate, much better progression-free survival.

And what’s really amazing, Neil, is, we’ve been doing this a long time. To see the median progression-free survivals not being reached and approaching 2 years, it’s really an amazing thing for patients with ALK-positive lung cancer when you compare them to how things were just 10 years ago for those patients. Specifically, though, in terms of ALK, I mean, a lot of people already had come on board with alectinib. You can imagine a lot more. But then the question is, what’s next?

DR RIELY: Yes. So I think that’s really the challenge here. We’ve kind of blown the old paradigm around. Where we once had “everybody gets crizotinib” and then we try to figure out the sequence of second- and third-line ALK TKIs, now we’re going to move alectinib into the first-line setting. And the knowledge about mechanisms of resistance is modest.

The data that we have today suggests that a particular mutation, G1202R, is somewhat more frequent in patients with ALK, but we have to find the right drugs for these. There are some hints that lorlatinib is an effective therapy for those patients. There may be some value with brigatinib. But we don’t have the clinical trial data that we were really sitting on just 6 months ago. We knew what to do in the second line.

Now we have a lot less information. And I think there may be an attempt for folks to try to sequence some of these ALK inhibitors in a way that maybe doesn’t make sense. So if you give alectinib, and then giving ceritinib, and then giving brigatinib, there may not be a lot of value in that approach, because there just may not be much efficacy. And I think we need to build the data set to help guide how we treat these patients.

DR LOVE: For practical purposes, though, in general, what is your next therapy after alectinib today?

DR RIELY: For today, I think chemotherapy is your best approach, because when I think about the response rate for, say, carboplatin/pemetrexed with or without bevacizumab, the response rate’s likely higher than the response rate for, say, ceritinib, for brigatinib. And so I would move to chemotherapy next.

DR LOVE: What about ROS1? Of course, crizotinib is standard first line. Again, chemo second line?

DR RIELY: I think that’s the right answer. I think there are available kinase inhibitors with activity. We’ve seen single-agent data for ceritinib. We’ve seen single-agent data for cabozantinib. And there’s a lot of suggestive data for patients with ROS1 but nothing that looks as good as a platinum doublet to me.

Selection of up-front therapy for BRAF mutation-positive metastatic NSCLC

DR LOVE: So moving on to BRAF V600E mutations, one thing we saw — and we’ve kind of seen this before, but it was just kind of interesting to see it playing out here — is that the vast majority of docs do utilize targeted therapy, specifically dabrafenib and trametinib. But again, we were able to tease out that there’s a subset of the investigators, a minority who do take into consideration TPS score when they make the decision.

So for a low TPS score, almost everybody uses targeted therapy up front. But for a high TPS score, about two thirds of the docs will use targeted therapy up front. And actually about a quarter would use — let me make sure I’ve got this right — a quarter would use a checkpoint inhibitor, and then 8% would use a checkpoint inhibitor and chemotherapy. Very few people, actually only 4%, are actually using chemotherapy up front, and again, only in these high TPS scores.

Any thoughts about first-line therapy in these situations and the idea of using a checkpoint inhibitor with or without chemotherapy instead of targeted therapy?

DR RIELY: Sure. So let me focus on the use of dabrafenib/trametinib as a first-line combination. It’s what I do. It’s what, as you’ve seen, a lot of folks recommend.

But when we also think about the role of immunotherapy, I think we need to realize that patients with BRAF mutations are actually a little bit different than some of the more common oncogene-driven cancers. So if you think of patients with EGFR, patients with ALK, we have this stereotype in our minds of a young never smoker. And that’s actually not a very good stereotype for patients with BRAF mutations. Patients with BRAF mutations are more likely to have a history of cigarette smoking. And so their tumors may be more complicated.

We don’t have as much information about, say, the tumor mutation burden or anything like that. But I think we’re beginning to understand that. And this may be a subset of oncogenically driven cancers where it’s reasonable to consider immunotherapy. Whether that’s based on TPS or something else, it’s not clear.

MET exon 14 alterations and implications for treatment

DR LOVE: So in terms of the patterns of how people approach first-line therapy, you can look at the colors, and you see how they match up. So, for example, ALK and ROS1 kind of look the same. But, interestingly, MET exon 14 also looks similar to BRAF, in that most people do use targeted therapy up front. If the TPS score is higher, now you start to see, again, a minority, but maybe a quarter or a third using chemotherapy or a checkpoint inhibitor alone.

Can you talk about how you think through this situation? And also, you were talking about MET amplification versus this mutation. If you could, distinguish that clinically, also.

DR RIELY: Yes. I think this is a really important thing to highlight. As we’ve learned more about MET exon 14 mutations, people have begun to realize how effective a biomarker it is for response to MET inhibitors, particularly crizotinib.

Now, we’ve been discussing MET amplification for years. MET amplification has been talked about for at least 10 years as a mechanism of resistance to EGFR TKIs, as a unique thing on its own, without EGFR. And in some of the earliest data — a couple of years ago, we saw data presented at ASCO saying that crizotinib might have value for patients with high-level MET amplification.

Now, when you look at large series of data, though, what you find is that real high-level MET amplification is rare. It’s probably much less than 1%. And there is oftentimes an overlap of high-level MET amplification with this new thing, MET exon 14 mutation. Now, to be clear, MET exon 14 mutations are a specific subset of MET. And when these MET exon 14 mutations happen, it leads to a splice variant that stabilizes the protein and makes the protein signal more and more. And so it’s clearly an oncogenic event, whereas MET amplification is a lot less clear that that’s actually what’s going on.

So with MET exon 14 mutations, that’s the group of patients that we have the best data saying that crizotinib is an effective MET inhibitor for those patients. So that’s one group. And then the MET amplification is a little bit of a different group, a lot less common and, I think, a lot less important for us to focus on now.

DR LOVE: You kind of get the feeling, just looking at these data in terms of this paradigm of targeted therapy ver — checkpoint inhibitor in patients with genomic alterations, that with MET and BRAF, people are not quite as heavily into the EGFR/ALK of delaying checkpoint inhibitors. And there’s less confidence that checkpoint inhibitors might not be as effective. What do you think about that? I know your group presented some data at ASCO looking at MET exon 14 in terms of checkpoint inhibitors.

DR RIELY: Yes. So I think at ASCO, we looked at the relative overlap of MET exon 14 and high TPS and found that there were many patients, in fact, with MET exon 14 that had high TPS. But then we looked at a very small number of patients — I believe it was 12 patients — who had MET exon 14 and had received immunotherapy and looked at their outcomes based upon tumor proportion score. And we were surprised, I think, to see that even the patients with high TPS didn’t seem to benefit.

But I will issue the general caveat: This was very small numbers, fewer than a dozen patients. And so as a consequence, I’m hesitant to universalize this. I will also highlight that patients with MET exon 14, like those with BRAF mutations, do have a bit higher frequency of cigarette smoking in the past. And patients who have MET exon 14 are much older than a lot of our other patients. So there may be a rationale for using immunotherapy in these patients.

Sequencing therapy for patients with RET rearrangements

DR LOVE: So the last two genomic subsets that we looked at again seemed to group together in terms of functionally how people sequence therapies, RET rearrangements and HER2 mutations.

And there we saw really much more chemotherapy being used up front, relatively little targeted therapy as up-front treatment. And in terms of TPS correlation — so within RET, we actually saw mainly the most common choice was chemotherapy up front or chemotherapy and a biologic.

But in high TPS scores, we saw a lot of people using — almost half using checkpoint inhibitors alone. Some people use combination, but mainly by itself, so more in terms of immunotherapy versus targeted therapy, almost like the wild type, at least in first line, and kind of the same thing, very similar numbers, with HER2.

Again, what do you think about what we saw here? Does it reflect your practice?

DR RIELY: I think the reason you’re seeing that is because of the relative values of the drugs for those two oncogenes. So RET rearrangements, they’re clearly a target. But the problem is that the drugs that are available aren’t so good. So cabozantinib, in the best data, has a response rate in the 30% to 40% range. When you compare this to, say, EGFR, where the response rates are 50% to 60%, it’s much different. And it’s much more equivalent to chemotherapy.

And then also the drugs that are available, particularly cabozantinib for RET, it’s got a side-effect profile that isn’t a lot better than chemotherapy. And so I think when we put all this stuff into the calculus of making a decision, there’s a real temptation to ignore the targeted inhibitors because they’re just not that great in this particular setting.

And I think that this really tells us that we need to develop better RET inhibitors, better HER2 inhibitors, so that we can target these oncogenes. But in the meantime, investigators and practicing docs see that these inhibitors aren’t as great as some of the other ones. And they’re moving towards an approach that’s based more like wild-type disease.

DR LOVE: And yes, in terms of using checkpoint inhibitors, this paradigm of targetable mutations and lack of response, I guess, for a lot of reasons, as you just mentioned. Again, it’s more of a wild-type paradigm.

Efficacy of HER2-targeted therapy for patients with metastatic NSCLC and HER2 alterations

DR LOVE: Where are we today in terms of anti-HER therapy? It kind of seems like, as you say — I mean, we haven’t seen great response rates. Seems like people are struggling a little bit. But I am curious about a couple of things I heard about recently. One was, your group and another presented some data on T-DM1. I’m curious what you saw there. I’m also curious about the drug neratinib, that was just approved as postadjuvant therapy of breast cancer, but now it is available clinically to use in lung cancer patients with HER2 mutations.

DR RIELY: So T-DM1 is really a great story that we saw play out at ASCO 2017. There was a clinical science symposium where people used T-DM1. And there were 2 presentations. One group, the Memorial group, looked at T-DM1 in patients with HER2-mutant lung cancer. And the UNC, Tom Stinchcombe, presenting on behalf of a larger group of investigations, looked at T-DM1 in patients who had HER2 overexpression, whether HER2 2+ or HER2 3+.

And it’s really interesting to look at the data in that context, because the overexpressed patients weren’t necessarily mutant. A couple probably were, in the grand series. And the HER2 mutants didn’t necessarily overexpress. So there’s a Venn diagram that shows some overlap of these 2 populations.

And what we see from the data, from the efficacy data, is that in HER2-mutant patients there was a high response rate, about 40% response rate, for T-DM1 in HER2-mutant lung cancer. Perhaps the only disappointing thing was that the median progression-free survival was short. It was just 4 months, but clearly a responsive molecular entity.

In contrast, the group that presented HER2 overexpression showed that if you looked at the group of patients that were HER2 2+ there was a 0% response rate to T-DM1. But if you took the somewhat higher bar of HER2 3+, those patients who overexpressed, they had a response rate that was a bit higher, at 20%. When they dug into the responders in that group of patients, two of the patients had HER2 mutations. And so those were the groups that seemed to benefit.

So I think when we take away this data, I think it tells us a couple of things. One, T-DM1 probably is effective for patients with HER2-mutant lung cancer. HER2 mutations is probably a better biomarker of HER2 dependency than is HER2 overexpression, and that, I think, IHC remains a difficult biomarker to use in general. The only place we’ve ever found a place for it in lung cancer is in the TPS.

DR LOVE: What about neratinib?

DR RIELY: Neratinib is a drug that has been explored for a long time in HER2-mutant lung cancer. There was a large clinical trial led by Leena Gandhi that was completed a couple of years ago. And in a single-agent neratinib arm, there were very few responses. I think there was 1 response. And they looked at also a combination of neratinib plus an mTOR inhibitor and saw a little bit more of a hint there. But the take-away from those trials really was that neratinib probably wasn’t an effective drug for HER2-mutant lung cancer.

Analysis of PD-L1 expression and variation over time

DR LOVE: So obviously, PD-L1 testing has become a critical part of management of non-small cell. Can you talk about some of the assays that have been utilized in this regard?

DR RIELY: So I think over the past few years, we’ve seen all of these immunotherapies developed. We’ve seen each company develop a drug and develop their own companion biomarker with a different PD-L1 assay. I mean, here on this slide, you see the available biomarker assays. So nivolumab had the 28-8 assay. Pembrolizumab had the 22C3 assay. Atezolizumab used SP142.

And so you see that they’re all different antibodies. And they also used different scoring systems. And the data to date have been in different patients. So it’s all been kind of a mishmash of things. And what people have really been looking for is an opportunity to compare these different antibodies and these different approaches. And there’s been some recent data looking at the comparison of these.

So I think as we decide which patients are PD-L1-low, -medium or -high or no PD-L1, some expression or high PD-L1 expressers, we have to recognize that there may be some variety in patients, whether you’re biopsying multiple different sites. Is there heterogeneity?

And does this change over time, so that if you have PD-L1-positive at baseline, is it always PD-L1-positive? Or in converse, does it go from PD-L1-negative at diagnosis, they go through a couple of lines of chemotherapy, do they become PD-L1-positive?

So we have only a little bit of data looking at this type of question. And what you see here is 2 different sets of data. One is on the left, looking at 2 different time points, so primary tumor and metastatic tumor. And you see that for the majority of patients, 6 of the ones described here, they were both negative. But you also see some patients that went from positive to negative and some that went from positive to positive. And so there is a possibility that these numbers will change over time.

Similarly, on the right you see patients that are metachronous — so these are 2 different pieces of tissue obtained at different time points — and you see that they’re generally the same, though there’s some slight changes. And then the final little bit at the bottom here, looking at a total of 57 tumors — and this is particularly in patients with EGFR mutations pre- and post-EGFR TKI. You see that 72% didn’t have a change in their PD-L1 status, though there were some patients that went from negative to positive and, similarly, some that went from positive to negative.

So I think this remains a Wild West in terms of understanding this. I think the general takeaway from all of the PD-L1 data is that it’s probably stable and there’s not a lot of role for rebiopsy for PD-L1 status on a routine basis, though we can sometimes consider it.

This data set looked at whether you need to get fresh tissue just before you treat a patient. And it’s not particularly easy to sort out all this data here, but I would say that when we look at using archival tissue and the available clinical trial data for the second line, it doesn’t seem to make a big difference in terms of what proportion of patients are tumor proportion score high or low, whether you got it from just before starting second line or at diagnosis. And it also doesn’t seem to make a difference in terms of response rate, as to whether you got the positive or negative results, and when you got that positive or negative result.

So this is, again, a further description of the different PD-L1 assays that are available, looking at 28-8, 22C3, SP142 and SP263. And this is from a paper by David Rimm and colleagues. David Rimm is at Yale. And he did this paper in JAMA Oncology, where he sat down with a whole bunch of lung cancer tumors and these 4 different immunohistochemistry assays for PD-L1 and decided to compare them head to head.

And this is an example of what the PD-L1 staining looked like for the same piece of tumor, looking at the 4 different antibodies here. And you can see — even though we’re not pathologists, you can see that the 22C3, 28-8 and E1L3N all looked very similar. In this particular example SP142 was negative, where all the other 3 were positive.

And then there’s a broader assay here. And this looks at the mean tumor proportion score for all the antibodies in all the tumors.

And what you see here is that 3 of them really cluster together relatively nicely. The 22C3, the 28-8 and the E1L3N all cluster together nicely. It’s just this one, SP142, that seems to be the outlier.

And similarly, if we look at it in this slide, at the cut points, you see that using the cut point of 50% or 1%, that the SP142 is the clear outlier in terms of identifying tumors that are PD-L1-positive at the 1% cutoff. And similarly for the 50% cutoff, it’s just a lot lower hit rate for the SP142 assay.

DR LOVE: Is that most likely because of the antibody?

DR RIELY: It’s probably because of the antibody and maybe some other characteristic of the test, but I think my takeaway message from this is, if a lab is setting up a PD-L1 assay, it’s probably not a lot of value to set up the SP142 assay. And probably, most labs are setting up the 22C3 antibody.

DR LOVE: I guess for practical purposes right now, it’s the TPS that really is affecting clinical decision-making.

DR RIELY: That’s right. And the first-line data that’s mostly driving clinical decision-making is all about pembrolizumab and the 22C3 antibody test.

Molecular pathology of lung cancer

DR LADANYI: The distinction between squamous and adeno can be difficult, especially on small biopsies. So there will be cases in which either you won’t be able to get that distinction, they’ll be signed out by the pathologist as non-small cell not otherwise specified, or you’ll have a situation where the biopsy was read as squamous, but when the tumor comes out at surgery we realize it’s an adenosquamous. And we’ve learned that adenosquamous cancers actually have the mutational profile more typical of adenocarcinoma.

So there are situations where if you get a non-small cell NOS or squamous on a biopsy, you still have to have in the back of your mind the possibility that this patient could have the adenocarcinoma alterations that could be targeted.

Genomic testing for patients with newly diagnosed metastatic NSCLC

DR LOVE: We also looked into the kind of assays that these investigators are utilizing. And, of course, most of them are using some type of NGS. But in terms of defining first-line therapy, it seemed that now that’s becoming a little bit of a moving target. In the past, it was kind of EGFR, ALK, ROS1. But now we’re starting to see investigators talking about using first-line targeted therapy with the BRAF and MET. Any thoughts about the issue of getting results back of assays like this in time to make a decision for first-line treatment of metastatic disease?

DR LADANYI: Yes. So what you’re bringing up is the concern or the limitation of some of the more complex NGS assays in terms of turnaround time. So they often might take 2 or 3 weeks to get results. And people don’t typically like waiting that long.

That’s one of the reasons why in our practice we also have quick assays for EGFR and ALK alterations, so that we can get an answer back within a few days. But I think as more targets move into first line, into FDA approval as first-line therapies, I think we’re going to see kind of more pressure to have broader quick assays that will give information quickly on these critical, already-approved targets and then later, as a second test, a more comprehensive NGS panel. So I think that’s the way things are going.

Actionable alterations in patients with adenocarcinoma of the lung

DR LOVE: Can you talk about your view on how the molecular pathology of lung cancer has evolved?

DR LADANYI: I think lung cancer’s been really an amazing story, because starting in 2004 with the discovery of EGFR mutations, we’ve seen this steady kind of definition of further subsets of lung adenocarcinoma, most of them targetable with available agents, either in clinical trials or already FDA approved. So it’s really become the poster child for targeted therapies in solid cancers.

At the same time, more recently it’s become one of the main major cancers in which immunotherapies are quite attractive. And the subset of lung adenocarcinoma in which immunotherapies are most effective is the subset that mostly was lacking targetable alterations. So it’s almost like complementary in a very nice way.

So one of the things to remember about these targetable genetic alterations in lung adenocarcinoma is that they tend to be quite mutually exclusive.

So you typically see these pie charts that show the proportions of the different alterations. And the experience is, the numbers might be a little bit different from center to center.

But overall, what we’re seeing for the percentage of cases, for instance, that have the alterations that you were asking about in the survey, namely EGFR, ALK, ROS fusions, BRAF V600E, ERBB2 mutations and MET exon 14 mutations, we’re seeing about 35% of adenocarcinoma patients overall with one of these targetable mutations. And I think the experience elsewhere, for instance, Dana-Farber, is very similar, around 30% to 35%.

DR LOVE: Can you talk a little bit about your vision for what a driver mutation is and how you determine what a driver mutation is?

DR LADANYI: Yes. The term “driver mutation” can be used kind of loosely, basically as any cancer mutation, or it can be used in a more kind of stringent fashion, in the sense of a mitogenic driver. So if you look at the alterations in lung adenocarcinoma, the real targetable drivers are alterations that activate the MAP kinase signaling pathway. And the thinking is that you need to activate that pathway at 1 place in the pathway. And once you’ve done that, you don’t need to activate it at other points in the pathway.

So this is kind of the approach to using the term “driver” in lung adenocarcinoma. If a particular alteration that activates the MAP kinase pathway is mutually exclusive with EGFR and KRAS and so on, then it really tells you that that is a mitogenic driver that defines a molecular subset of lung adenocarcinoma.

Targeting KRAS mutation-positive NSCLC

DR LOVE: So obviously, KRAS mutations are a very common finding. Can you talk a little bit about why it’s been so difficult to find therapies for people with KRAS-mutant disease?

DR LADANYI: KRAS is more difficult in terms of drug design. That’s my understanding. I’m not a pharmacologist, but it has been more difficult in terms of drug design. There are now some agents, some compounds that specifically target certain mutant forms of KRAS. There are subsets of KRAS-mutated lung adenos that do respond to MEK inhibition. We don’t exactly understand which subset of KRAS that is.

So there are a lot of efforts, obviously, to develop therapies for KRAS-mutated tumors. And I think it’s only a matter of time until we see more and more of these patients eligible for targeted therapies, although at the same time, I should mention that a big slice of KRAS-mutated lung adenocarcinoma is smoking associated. A lot of the major KRAS mutations have the mutational signature of tobacco carcinogens. And, as such, they tend to have higher mutational loads, which also makes them better candidates for immunotherapies. So there might be competing approaches for this subset of lung adenocarcinoma patients.

DR LOVE: That's interesting. Of course, there’s tons of data on mutational load. But is there any correlation between KRAS status and response to immune checkpoint inhibitors?

DR LADANYI: I wouldn’t be able to quote you a study in particular, but I would not be surprised that patients with the smoking-associated KRAS mutations respond better, because they also have a higher number of other mutations as well.

Assays to detect genomic alterations in patients with lung cancer

DR LOVE: Let me ask a more practical question, clinically. We were only really able to identify right now, today, 7 genomic alterations where we have targeted therapy that we see people utilizing targeted therapy. Yet you hear a lot of emphasis on multiplex testing. A lot of people reflex into next-gen sequencing, where you get tons and tons of information. But really from a practical perspective, aside from research, you want to know about these 7 alterations. Would it make more sense to have a panel of these 7 alterations?

DR LADANYI: I think one could make an argument for that, for sure. I think you might want to have an up-front quick assay, which would be a multiplex NGS assay by necessity, because already this is a large number of genes and exons and so on. But you could have such a kind of small, quick panel up front and then do a broader panel as the follow-up test. But it really depends on, I guess, a number of factors.

If your patient population really is not going to have access to clinical trials, then you could reason that it’s really only useful to test for the currently targetable alterations. But there are other alterations, which may be targetable, that are being targeted in the context of clinical trials. And I think it’s worthwhile to do that additional testing in case the patient can get connected to and eligible for something that’s being targeted in a clinical trial elsewhere.

So with ALK, RET and ROS fusions, one of the challenges for molecular testing is that each of these kinases can rearrange with a whole number of other different partners to be turned on. So although EML4 is the main partner for ALK, there are also many other partners that have been described. ROS1 is even more varied in terms of its fusion partners. And what this means is that we can’t have just individual assays for EML4, ALK specifically and some other fusion of ALK and so on.

So our approaches, testing approaches, are designed to pick up any fusion involving one of these 3 genes. And historically this was done by FISH, just looking for rearrangement of those genes, but now we’re using different NGS methods, both DNA based and RNA based, to pick up any fusion involving these 3 kinases.

DR LOVE: Could you explain at a medical student level how FISH assays are done?

DR LADANYI: With FISH assays — and the most common form of FISH assay for these fusions is a so-called break-apart FISH assay. So basically you have 1 FISH probe that binds at the beginning of the gene, and the other one binds at the end of the gene. And normally, those 2 signals, which are different colors, should be next to each other. If you have a rearrangement and a fusion with a gene somewhere else in the genome, those 2 signals then are separated. And you can see that they are no longer contiguous. And you can diagnose a rearrangement that way. So that is certainly historically the most common way to pick up these rearrangements.

Biology of MET exon 14 alterations in NSCLC

DR LOVE: Let’s talk about targetable tumor alterations, and one of the most recent in lung cancer has been MET exon 14 skipping alterations. Can you comment on what we know about these tumors?

DR LADANYI: MET exon 14 skipping alterations are a very interesting story, because it’s something that was detected originally many years ago, but the exact significance and prevalence was clear until quite recently. And one thing I think people should be aware of is that when you look at lung adenocarcinomas very comprehensively with NGS assays, you realize that MET exon 14 alterations are probably the third most common targetable alteration in lung adenocarcinoma behind EGFR and ALK. And they’re actually, in our experience, more common than RET and ROS1 fusions.

One of the challenges with MET exon 14 alterations is that the exon 14 skipping can be caused by a variety of mutations that affect the splice sites of exon 14. And what happens when exon 14 is skipped is that the exon 14 contains a region that is recognized by the cellular machinery that turns over the MET receptor tyrosine kinase. So when that exon is missing from MET, the receptor tyrosine kinase is not degraded and accumulates at the cell membrane. And there’s this excessive signaling through the MET receptor tyrosine kinase down to the MAP kinase pathway. So it’s an alteration that was underappreciated previously but has really emerged as one of the significant targetable subsets of lung adenocarcinoma.

DR LOVE: Can you differentiate that? I know in the past I’ve heard about MET amplification. What’s the difference between that and these skipping alterations?

DR LADANYI: MET amplification, which means just increase in the number of copies of the MET gene per cell, is something that was the object of many studies previously. Once we figured out MET exon 14 alterations, we realized that the two often occur together. So the gene can undergo this MET exon 14 skipping and then get amplified.

So some of the earlier studies that just looked at MET amplification, they had a combination of cases that included MET exon 14 cases and MET cases without mutations that were just amplified. Now that we can tease those two apart, we can also recognize that there is a small subset of lung adenocarcinomas that just have increased copies of the MET gene without any mutation of the MET gene. And those may also be potentially targetable, if the amplification level is high enough. That’s a subset, I think, that probably still needs a little bit more clarification.

Incidence of RET fusions in patients with lung adenocarcinoma; response to targeted therapy

DR LOVE: Maybe you can comment a little bit on RET.

DR LADANYI: So RET fusions are seen, in our experience, in about 1% to 2% of lung adenocarcinomas.

They do respond to RET inhibitors, but they do stand out a little bit among these targetable subsets of lung adenocarcinoma, in that the response rate is lower than your typical response rate in EGFR and ALK and ROS1.

So you can see in this paper that was led by my colleague Alex Drilon, that the overall response rate was 28%, in contrast to the 60% to 80% response rates that you kind of expect when you have a targetable alteration and you treat it with a drug that hits the target.

So there’s been a lot of work, and there’s a lot of work going on to figure out why these response rates for RET fusion lung cancers tend to be lower and how to improve them. I think that’s still a work in progress.

DR LOVE: Any hypotheses that you find appealing?

DR LADANYI: There’s a hypothesis that we haven’t been using really good RET inhibitors yet, that some of the RET inhibitors that have been used are kind of dirty, multitargeted kinase inhibitors. There’s also been some evidence that RET fusion lung cancers have a very active bypass signaling pathway available so that when you block RET, the growth signaling via another pathway, such as EGFR, becomes active and compensates for the RET inhibition. So these are things people are working on right now, actually.

Activity of NTRK inhibitors in patients with NTRK fusions

DR LOVE: How about NTRK fusions?

DR LADANYI: Yes, NTRK fusions. As I think most of us saw at ASCO recently, there are dramatic responses to NTRK inhibition in patients with NTRK fusions. NTRK fusions are I wouldn’t say unique, but they do have this particularity of occurring in a variety of different cancers. And in each individual cancer, except for a few very rare subtypes, they are seen in only a very, very small percentage of that cancer. So that’s the situation in lung cancer, where I don’t have a percentage, but it’s probably less than 1%.

But if you do identify these patients, they have really remarkable responses to the currently available NTRK inhibitors. So it’s one of those situations where yes, it’s difficult to identify them unless you’re doing this kind of broad NGS screening up front for every cancer, but it pays off when you get these amazing responses.

DR LOVE: What are some of the other cancers where the NTRK fusions are seen? And I’ve heard people make analogies to the MSI-high situation that’s out there right now. You have, actually, a checkpoint inhibitor approved across tumor lines based on MSI. I’ve heard people say maybe NTRK’s going to go the same way. What other cancers do you see it in? And do you think that people are going to start to view these separately?

DR LADANYI: We’ve seen it in rare colorectal cancers, rare pancreatic cancer, rare sarcomas. So it’s a rare event but very targetable. And right now, there’s no pathologic feature that would tell you that this lung cancer is likely to have an NTRK fusion. There are, as I mentioned previously, very rare subsets of breast cancer and sarcoma that have NTRK fusions consistently. So secretory breast carcinoma, a very rare subtype, consistently is basically defined by presence of ETV6-NTRK3 fusion. And some congenital fibrosarcomas are also defined by ETV6-NTRK3 fusions. But aside from those very rare tumors, in common cancers it’s quite a rare event.

MSK-IMPACT^™: Next-generation sequencing assay to detect genomic alterations and inform therapeutic decision-making

DR LOVE: How about this MSK IMPACT initiative?

DR LADANYI: So MSK IMPACT is a broad NGS assay that we’ve been running since 2014. Currently it’s a panel of 468 genes that allows us to detect mutations, copy number changes, as well as rearrangements in certain genes. And it has really been extremely powerful at our center to identify comprehensively the targetable alterations. And basically I enroll as many patients as possible into the trials of targeted therapies at our center.

DR LOVE: Maybe we can talk a little bit about what you’ve seen up to this point.

DR LADANYI: So we recently reported our experience with MSK IMPACT sequencing in the first 860 lung adenocarcinoma patients that were studied. And it gives a very comprehensive picture of targetable alterations. As I mentioned previously, for the alterations that you included in the survey, we’re seeing about 35% of patients present with one of these alterations, so it’s definitely a very important population of lung adenocarcinoma patients who stand to benefit from this kind of profiling.

So there’s 35% of patients who can benefit from a drug that’s now FDA approved. And then there’s probably another 15% that are eligible for drugs that are currently in clinical trials at our institution. So altogether, this makes about 50% of lung adenocarcinoma patients who stand to benefit in one way or another from the MSK IMPACT sequencing.

And I should mention another important thing that emerges from the MSK IMPACT sequencing is, we get a sense of the tumor mutation burden, which has also emerged as an important predictor of response to immunotherapies. So even if we don’t identify a targetable alteration, we are providing information on tumor mutation burden, which can be used to prioritize patients for immunotherapy instead.

Mechanisms of resistance to EGFR TKIs

DR LOVE: Let’s talk about tumors with EGFRF mutations and particularly what happens when patients receive an EGFR TKI and then have disease progression.

DR LADANYI: So for EGFR-mutated lung cancer, as you know, the major resistance mechanism to erlotinib and other EGFR kinase inhibitors is the development of a second mutation, T790M. And we’re able to detect this in somewhere between 50% to 60% of patients who develop resistance to first-line EGFR TKIs.

But we’re also able to detect other resistance mechanisms. And the major other resistance mechanisms are MET amplification and ERBB2 amplification. And both of these potentially could be treated/targeted.

So this is useful information. We’re also still discovering new resistance mechanisms. And that work is in progress and is partly based on the MSK IMPACT sequencing.

DR LOVE: So just to clarify — because I don’t recall hearing that before — you’re saying that one of the mechanisms of resistance that develops with patients on TKIs is not only T790 but also MET and HER2?

DR LADANYI: Yes. So those are what we would call target-independent resistance mechanisms or bypass mechanisms. So you can develop resistance either by having a subclone of cells with this new T790M mutation grow out, which makes the TKI no longer active against the mutant EGFR molecule, or you can develop another alteration, not an EGFR, but in another signaling molecule. So like I said, the most common is MET amplification and ERBB2 amplification.

And they’re seen in probably 5% of these resistant samples. And in those cases, there’s often — there’s no T790M. The cells have become resistant because they’ve turned on. They’ve amplified either a MET or ERBB2 and have kind of compensated for the inhibition of EGFR by doing that and increasing the mitogenic MAP kinase signaling back to the level that was there before the EGFR TKI.

T790M mutation testing for patients who develop resistance to EGFR TKIs

DR LOVE: How do you manage patients with EGFR mutations with disease progression on a first-line TKI?

DR LADANYI: So the approach at our place, at least, is to do the T790M testing first on a patient who’s had a recurrence on the EGFR TKI, do the T790M first either in plasma DNA or from a biopsy. And then if that’s negative, then go on to MSK IMPACT comprehensive testing of the biopsy.

DR LOVE: What about the issue of resistance because of transformation to small cell?

DR LADANYI: Small cell transformation is maybe, I would say, less than 5% of these resistance mechanisms. And it is the kind of mechanism that would get missed if you’re just doing a T790M testing on plasma DNA. In terms of histology, it’s a very characteristic finding. So if there’s a biopsy and there’s small cell transformation, I think it will get picked up that way.

DR LOVE: Can you provide your perspective on T790 testing nowadays, both in terms of serum as well as tissue?

DR LADANYI: I think T790M testing in plasma DNA is probably the most widely used plasma DNA molecular assay. And I believe it’s becoming pretty standard at most centers. It can give you a quick answer on whether a patient is progressing due to T790M.

And one of the advantages of plasma DNA testing is that it may give you information on every site of disease. So if a patient is progressing at more than 1 site, the plasma DNA is presumably derived from every site and, therefore, you would know that at least 1 of these sites is a recurrence due to T790M. But in terms of looking at other mechanisms, you still need to get a biopsy to get a good understanding of other mechanisms of resistance in that patient.

Acquired EGFR C797S mutation as a mechanism of resistance to osimertinib

DR LADANYI: So one of the interesting things that’s been observed with patients who receive osimertinib, which is an EGFR TKI that is active in patients who have the T790M mutation that has made them resistant to erlotinib, for instance, these patients, in a subset of cases, develop progression on osimertinib, because they’ve acquired yet another mutation that now inhibits the binding of osimertinib to the mutant EGFR. And that mutation is C797S.

And we saw one of the first patients like that. And this was a patient who had developed a T790M on erlotinib, was switched to osimertinib and then developed the C797S and progressed on osimertinib.

And in the sequencing data, you can see that originally her EGFR gene did not have either the T790M or C797S. It did have an EGFR exon 19 deletion.

And then in 2011 when she progressed on erlotinib, she had the T790M mutation. And by 2014, when she progressed on osimertinib, she had an additional mutation in addition to the T790M mutation. She had that C797S mutation, which abolished the binding of osimertinib.

So it does make you feel like these kinase inhibitors can be a moving target with resistance mechanisms developing almost inevitably. But they do, of course, provide extra years of good-quality survival.

DR LOVE: Another question about — I’ve heard people discuss the situation where a patient maybe has a bone met that’s hard to get to, some other site that’s difficult to get tissue on, a patient who responded and now progressed again on EGFR TKI, trying to figure out how hard to look for tissue in a patient who’s T790-negative on serum. And I’ve heard people say that they look to see whether the original, let’s say, exon 14 or whatever mutation is there, and if it is there and you don’t see T790, then, quote, you can believe the serum. Can you talk about that sort of thinking?

DR LADANYI: I think interpreting a negative result on plasma DNA is very tricky, because it can either mean there’s no T790M or it can mean that, for whatever reason, the tumor cells did not shed a sufficient amount of DNA into the blood to be detectable. So I think a negative plasma DNA result is not something you can really act on.

DR LOVE: Another question about T790 is, I’ve heard that there are actually people with T790 germline mutations. Do you know anything about that?

DR LADANYI: Yes. It was discovered several years ago that the T790M can occur as a germline mutation and that it’s associated with an increased risk of lung adenocarcinoma. So these are families in which there’s a high rate of lung cancer unrelated to smoking. And so I would say it’s probably less than 1% of patients that have this germline T790M, but it’s something that could explain, in a patient, why there’s T790M detected at diagnosis, before any exposure to TKIs. And that patient who has a germline T790M, has developed lung adenocarcinoma, would show, essentially, primary resistance to EGFR TKI.

There’s also another very small subset of patients that have an acquired T790M, not germline, acquired, but it’s present at diagnosis, before any TKI. And we don’t know exactly why that is, but it’s also been observed, again, maybe 1% of patients or less.

Detection of ALK fusions in NSCLC

DR LOVE: Can you talk about ALK fusions and challenges in making the diagnosis?

DR LADANYI: Detecting ALK fusions can be done by several different approaches, starting with immunohistochemistry. So it turns out that ALK is not normally expressed in most tissues at a level that could be detected by immunohistochemistry. But when it’s involved in a fusion in lung cancer, for instance, the expression is ramped up to such a degree that you can pick it up with the antibody. And it’s really a very robust biomarker.

Then there’s FISH, as I mentioned, which tells you that there’s been a rearrangement of the ALK gene, using a break-apart FISH assay. Then there are a variety of NGS assays, either DNA based or RNA based, that can detect ALK fusions. And they can be designed so that you don’t need to know the partner that is fused to ALK. So you’re able to pick up an ALK fusion regardless of what the other gene that ALK is fused to is.

DR LOVE: This concludes our program, and we should note that subsequent to recording this discussion, results from the Phase III FLAURA study were presented at ESMO demonstrating that osimertinib significantly improved progression-free survival compared to standard-of-care EGFR TKIs in patients with EGFR-mutated non-small cell lung cancer.

Special thanks to our faculty, and thank you for listening. This is Dr Neil Love for this special CME program focusing on a survey of clinical investigators in lung cancer.