Biomarker Assessment in Metastatic NSCLC: Predictors of Response to Targeted Therapy and Immunotherapy

DR GOVINDAN: I’m going to discuss the level of biomarker assessment in metastatic non-small cell lung cancer today, with a focus on, of course, targeted therapies and immunotherapies.

As we know, lung adenocarcinoma is a disease driven by all patients in the RAS RTK pathway. And we have established that through extensive genomic studies done by the Cancer Genomics Project and other projects.

And what is obvious is that nearly 75% to 80% of lung adenocarcinomas have critical alterations in a key member of the RAS RTK pathways. And, of course, only about a third of them are currently targetable with available and approved drugs. And hopefully this number will grow as we move along.

In a paper that we published a few years ago, we showed clearly that these alterations are mutually exclusive in the X axis of the samples or the Y axis of the alterations that you see, and a number of alterations were added from the initial TCGA paper back in 2014 to this paper in 2016. In our ongoing work we are trying to expand this population and whittle down the so-called oncogene negative to about 10% or so, and that is currently under review.

The point of this slide is to show that if you use diligent methods, we are able to identify critical alterations that drive these tumors. And of course, it is up to us to convert them into potentially actionable targets by working with our industry colleagues and doing more functional studies. And in clinical practice today, of course, routinely we test these tumors for EGFR, ALK, ROS, RET alterations to really use them in the front-line setting in most situations. And, of course, with a more extensive panel we are finding alterations in HER2, MET and other gene alterations as well.

And so the question really is, how do we approach that? There are 2 major questions, in my view, with regard to the targeted therapy biomarker space. One is what kind of a tissue resource we use. Is it the tumor tissue or is it the plasma from these patients? And the second is what platform you would use, a targeted platform analysis looking at a few genes, particularly in those critical locations that we know they are altered, like exon 19 deletion, EGFR or L858R in EGFR. Or are we going to do the comprehensive analysis using next-generation sequencing? In some instances we can use FISH or IHC to identify these alterations.

In the interest of time, I would just say some general principles. And we use, in the clinical setting, assays ranging from RT PCR to NGS to FISH to IHC.

In the next slide, we are going to focus on the role of plasma testing, and I’ll come back to the tissue-based testing in a minute. And there’s a lot of interest in doing this plasma-based genotyping assay to deliver personalized therapy. Obviously the advantages are we don’t have to do repeat biopsies. The results come back these days within 3 to 5 days. It’s very convenient for both the providers and for the patients.

And how good are these tests? And I just want to point to a couple of studies. A number of groups have been looking at this — International Association for the Study of Lung Cancer has a nice paper published in JTO of 2019. I would refer you to this. And a nice paper from UPenn group — one in JAMA Oncology in 2019 that you see, and there is 1 more in the e-print. I had a publication in Clinical Cancer Research. Both drive home the point that the plasma-based genotyping is pretty useful in personalized therapies for metastatic lung adenocarcinomas.

The important caveats here are, in about 20% to 30% of the times the plasma-based testing may give you false-negative results. In other words, patients may have actionable mutations in the tumor tissues, but the plasma-based assays may not pick them up. And so that’s something important to keep in mind. They are very good when these known actionable mutations are present, but if they are negative it’s important for us to do tissue-based testing to really make sure that these patients do not have an actionable mutation.

And also, the other important point to make is that when you do both plasma- and tissue-based assays, the JAMA 2019 paper that I’m citing here from Dr Agarwal clearly shows the concordance rate is about 80%. That’s what we’ve seen before in other studies as well.

And the third important point is that if you have extensive disease, for example, liver metastasis, you will find these actionable mutations almost invariably in the plasma. Whereas if it is predominantly intrathoracic disease, as you can see here, you may not always find that in the tissue-based assays — I mean the plasma-based assays — and you have to go to the tissue-based assay. That’s important to keep in mind.

And so it’s okay to do plasma-based assays as long as you are able to really keep in mind that if it is negative we may have to pursue a tissue-based assay as we go along. And it’s an important point to keep in mind.

In terms of tissue-based assay, in the past we were doing targeted assays looking at EGFR, ALK or ROS, and then we went on to the next-generation sequencing using tumor tissues, and now the tissue-based assays, next-generation assays, they take about 2 to 3 weeks. And sometimes we need this information sooner and earlier. In that case we have to do a PCR-based assay to get the results very quickly if we’re going to rely on tissue-based testing.

This is where I personally think in my practice, I do the plasma-based assay. If they show EGFR mutation or ALK rearrangement, I go with that. And if it is not really showing any actionable mutation, I pursue tissue-based testing either using next-generation sequencing or targeted testing. It is important to keep that in mind.

And I want to spend the rest of the time focusing on the immunotherapy biomarkers. And a very quick primer here is that the immune checkpoint inhibitors, the PD-1 and the PD-L1 inhibitors, work on the premise that the tumor cells have expressed neoantigen, and then the T cells are exhausted using this pathway. And in the right situation, if you inhibit this PD-1/PD-L1 pathway you end up activating the T cells against the neoantigen bearing or the tumor cells. And clearly it is contingent upon many factors. One, the T-cell exhaustion’s driven primarily by the PD-1/PD-L1 axis and enough neoantigens plus some of the tumor cell’s surface. And then the immune response is not inhibited by other factors, like, for example, the T regulatory cells or myeloid-derived suppressive T cells.

Of course there is also the presumption that the tumor cells have adequate neoantigens, and the class 1 apparatus is well intact. And so the PD-L1 expression is only 1 part of the equation, so that’s important to keep in mind as we go along. And clearly there’s a lot of interest in testing PD-L1 as a biomarker in the tumor cells. And a number of studies have looked at this.

Again, in the interest of time, this slide from Fred Hirsch, presented a few months ago at the targeted therapy meeting, clearly summarizes the approach that others have shown in various meetings. And what I would say the take-home point is that — and in my opinion is less controversial — is that in non-small cell lung cancer, particularly in tumors that have 50% or over PD-L1 expression, single-agent pembrolizumab appears to be an appropriate choice of therapy for these patients.

Of course there have been a lot of discussions in that population: Do these patients do better with pembrolizumab and chemotherapy or pembrolizumab alone? We don’t have final answers to those questions. But I think it’s fair to say that in practice if you have somebody with minimal symptoms, low-volume disease, where you do have time for the immunotherapy to act, if the tumor cells are 50% or more expression, single-agent pembrolizumab is perfectly appropriate. On the other hand, if somebody has a lot of symptoms, bulky disease, you want to have a quick cytoreduction, a combination of pembrolizumab with chemotherapy is certainly appropriate for this population. In my own personal practice, those with less than 50% expression I tend to use chemotherapy and immunotherapy combination in both squamous cell and nonsquamous cell non-small cell lung cancer.

That’s kind of a 30,000 feet high altitude summary of the PD-L1 expression level. I don’t find the lack of PD-L1 expression by itself is a negative biomarker for not giving immunotherapy at all. And these patients appear to benefit — chemotherapy and immunotherapy combination regimen.

But the real challenge is, can we improve upon this? Is there a way to really use other biomarkers? And I think it’s worth talking about a few biomarkers. And one of them is, of course, tumor mutation burden. And the tumor mutation burden has been of great interest, because the presumption is that the more tumor mutation burden you have, it’s likely that those tumors have high neoantigen expression.

In fact, in the TCGA analysis we went back and asked the question, what is the average number, or what percentage of tumor cells, or the tumors, express 5 or more neoantigens in lung adenocarcinomas? It turns out that more than 70% of the lung adenocarcinomas, especially among smokers, have at least 5 or more neoantigens. And we don’t know exactly what the cutoff point is, and of course the quality of neoantigens might also matter. The HLA cross-type also would matter.

There’s a lot of work to do in this area, but the point is that a substantial number of lung adenocarcinomas, particularly from smokers, tend to have higher numbers of neoantigens. And tumor mutation burden is a little easier to assess compared to the neoantigen prediction.

By the way, the neoantigen prediction algorithms use some of the assumptions that have historically been accumulated from a lot of other studies, using the type of peptide and also the HLA class genotype as well. And is more a prediction. It is not very clear how good these predictions are. There have been various estimates. And functional studies are being launched here, as well as in other places, will hopefully sort out how good are the neoantigen prediction algorithms. But it’s fair to say that it’s too complicated to use in real life. Tumor mutation burden has become a shorthand to really identify the neoantigen burden in those patients.

Intuitively it makes a lot of sense to use tumor mutation burden as one of the predictive markers of the other side of the equation compared to the PD-L1 on the T-cell, tumor mutation burden on the tumor side. And it’s really useful to consider this.

How good are the data? We’ll talk about this as we go along. But also, it’s important to keep in mind, as I said initially, it’s not just the tumor mutation burden and the PD-L1 expression but also the other factors in the tumor marker environment. For example, the presence and absence of T regulatory cells and the myeloid-derived suppressive cells. Both can clamp down the immune response. And even if you have a high mutation burden and a high PD-L expression, presence of these molecules may determine response to immune therapy. And, of course, that’s only what we know. A lot of things that we don’t know.

Given all these things, it is not a surprising fact that a number of groups are interested in looking at tumor mutation burden not only in the tissues but also in the peripheral blood. And peripheral blood, of course, is even more attractive, because you don’t need to get tumor tissue and repeat biopsies. And what have we really found out? The early studies came from the work done by Charlie Swanton and Rizvi showing that in a group of patients who are receiving immune checkpoint inhibitors, using whole-exome analysis, those who have a high tumor burden did better compared to those who did not have a high tumor burden. And that, as I said, presumption on the neoantigen expression prediction. The assumption is that those with a high mutation burden have neoantigen burden.

And in the clinical trial setting, what is clear is that the tumor mutation burden may be useful when we are giving a single-agent immune checkpoint inhibitor with nivolumab or pembrolizumab here. And, in fact, in the CheckMate 026 study, the single-agent nivolumab was better than chemotherapy in those with a high tumor mutation burden, as you can see here. And the same with KEYNOTE-42 trial, single-agent pembro versus chemotherapy.

If you look at the tumor mutation burden, looking at the number of mutations for the entire exome, 175 or more mutation, the median overall survival was significantly better with pembrolizumab compared to chemotherapy. And in the less than 175, that’s a low mutation burden, in the pembrolizumab/chemotherapy the difference was not there. And this is actually quite interesting and very intriguing.

But again, if you notice one thing, what is the high tumor mutation burden? It is not an absolute cutoff marker like EGFR mutation present or absent. And the numbers have really differed based on either the mutation per megabase or the total mutations for the whole exome. And also the cutoff, even if you use the mutation per megabase, it has been variably cut off at 10, 20, at the median level. That’s all over the place. And also, you introduce a different variable with the use of plasma mutation burden as well.

But this tumor mutation burden difference appears to disappear to some extent if you add chemotherapy to the equation. Here is an example from the KEYNOTE-407 study from squamous cell lung cancer, pembro/chemo versus chemo. If you look at this, there is not an appreciable difference whether the tumor mutation burden is high or low in terms of the differences between chemo/pembro versus chemo alone.

But there was a lot of interest — the same was seen in the KEYNOTE study with chemotherapy and pembrolizumab versus chemotherapy alone in nonsquamous non-small cell lung cancer. Again, there is not much of a difference here in general. And there was some initial interest in looking at nivo/ipi. It appears that it is not very reliable, and the initial enthusiasm for looking at tumor mutation burden for nivo/ipi has waned quite a bit, in my opinion.

And there is some interest in looking at the plasma-based, or the blood-based TMB, in this population. And my summary is that while these studies have been presented in the meetings last year, and ongoing reports are there looking at the harmonization for looking at the tumor mutation burden for various different platforms, I would say that at the moment these are somewhat intriguing, and they need to be studied further. And I do not use that in the clinical setting at the moment. I only use PD-L1 tumor testing to identify patients who will get single-agent pembrolizumab in the high expressors of 50% or more. Otherwise, I don’t find that to be useful in the clinical setting in practice in 2020, at the moment.

But I do want to point out that a number of different groups are looking at various different — this is an exhaustive list looking at the T-cell infiltration, the type of T cells, and the various other mechanisms that are known to be associated with resistance, including the presence or absence of subtle alterations.

For example, inactivating mutations in STK11 and PTEN mutations have been associated with resistance to immune therapy, and in melanoma increasing data to suggest that beta-catenin mutations or Wnt signaling activation has been associated with resistance to immune therapy. Of course you have known that the presence of EGFR mutation or ALK rearrangement has been associated with lack of response to immune checkpoint inhibitors. On the other hand, KRAS mutations have been associated with increased response to immune therapy.

So we can be looking at the tumor genomic alterations to predict immune responses. And, in fact, there are some studies combining the tumor mutation burden, PD-L1 expression, and the presence of these mutations or absence of them, to correlate for response to immune therapy. Again, these are very interesting, but I don’t think we can use that in real life to decide to give immune checkpoint inhibitors or not. I would say in the context of clinical trials. As intriguing as the data are with regard to STK11 mutation, the lack of immune response, I would strongly suggest that we enroll them in appropriate clinical trials than off-hand decide in clinical practice to decline or defer immune therapy for this population.

There are also fascinating tools coming down the pike looking at the multiplex immunofluorescence to look for colocalization of PD-1/PD-L1, presence of PD-L1 in macrophages to see whether they can predict or help us identify the group of patients who would or would not respond to immune checkpoint inhibitors. And these are really exciting new areas of research currently ongoing.

I’ll just spend the last few minutes — I’ll say a few things. One of them is, can we use these biomarkers to predict who would relapse following curative therapies? And in the interest of time, I want to talk about the early-stage non-small cell lung cancer following surgical resection, but I do think there is a fantastic opportunity in patients with Stage III non-small cell lung cancer who get chemotherapy/radiation followed by immune therapy.

And this is a work from my colleague Aadel Chaudhuri published in Cancer Discovery a couple of years ago showing that after chemoradiation, at 4 months, there is minimal residual disease using ctDNA. If you find persistent ctDNA after chemoradiation, those patients did very poorly compared to those with no ctDNA after chemoradiation.

As you can imagine, these studies were done before the durvalumab era. It will be interesting to see whether this population, the population of patients who have persistent ctDNA following chemoradiation are the ones who benefit from durvalumab or other populations that we can identify who would not benefit from durvalumab after 3 or 4 months of therapy are actually fascinating questions to answer as we go along. But I do think there’s a lot of potential to look for minimal residual disease in Stage III population.

And I would say in closing, clearly we know exactly what group of patients respond to targeted therapies, EGFR, ALK and ROS and RET, using appropriate testing, using PCR, next-gen or FISH — ALK — IHC-based assay, either in the tissue for IHC and FISH or the sequencing studies done in blood or plasma with all the caveats I highlighted.

For immune therapy, the only one ready for prime time is tumor-based PD-L1 expression. There are a lot of other markers that are coming up, and they will help us identify patients who would benefit or not benefit from checkpoint inhibitors. I also think you should watch this space for identifying patients who are at high risk for relapse following curative therapies. To begin with Stage III non-small cell lung cancer following chemoradiation with or without durvalumab and also in early-stage non-small cell lung cancer. That’s the accumulative data to identify patients who would relapse following adjuvant therapy.

Thank you.

First-line treatment for patients with metastatic NSCLC without a targetable tumor mutation

DR HORN: So, I’m going to be covering first-line therapy for patients with metastatic non-small cell lung cancer who do not have driver mutations, choosing between checkpoint inhibitor therapy alone or therapy in combination with chemotherapy.

And really, in the time that I’ve been practicing, we’ve seen this evolution of therapy for patients with lung cancer, where we’ve gone from a one-size fits all model, to looking at treatment based on non-small cell and small cell, looking at non-small cell histologies, their patients’ molecular profile, and now we also look at PD-L1 expression level in selecting therapies. And tumor mutation burden to me is still — is it a useful biomarker question?

We know that many cancers in which checkpoint inhibitors are approved are those that have a high frequency of mutation based on data from the TCGA. On the right-hand side you have melanoma, but right next to it we have non-small cell lung cancer, both adenocarcinoma as well as patients with squamous cell histology.

PD-L1 has been looked at as our most reliable potential predictive biomarker of ALK cancer patients with non-small cell lung cancer who were treated with a checkpoint inhibitor. But the ROC curves that have been generated for non-small cell lung cancer did not appear as good as those curves that we see for patients with melanoma. It shows that PD-L1 is variable. It doesn’t allow us for a sufficient binary classification for patients who will benefit versus patients who will not benefit. And we also know from data from the second-line studies that patients who are PD-L1-negative do derive benefit from checkpoint inhibitor therapy, while those who are positive not necessarily all get that benefit we’re hoping from treatment with these novel agents.

We’ve also seen from prior data from the CheckMate studies that tumor mutation burden and PD-L1 are 2 distinct populations of non-small cell lung cancer patients.

On the left-hand side we have on the X-axis, PD-L1 expression and on the Y-axis, TMB. And what you can see is, those patients who have low PD-L1 expression may have high TMB. And conversely, those patients who have high TMB may still be patients who have low PD-L1 expression.

On the right-hand side, we see a breakdown of TMB greater than 10 mutations per megabase, and less than 10 mutations of megabase by PD-L1 expression, and we can see that it's even. That those patients who are TMG greater than or equal to 10, around 71% of those patients are PD-L1 greater than 1%. And conversely, those patients who are TMB less than 10, approximately 71% of those patients are PD-L1 expression greater than 1%.

So, where’s the data for checkpoint inhibitor therapy for patients with non-small cell lung cancer. Now, KEYNOTE 024 was really a paradigm shift in the first-line therapy for patients with advanced non-small cell lung cancer. This trial enrolled patients whose PD-L1 expression was greater than 50% by the 22C3 assay. Patients had to be EGFR and ALK wild-type to enroll on study. One could argue that there also should be ROS1 and RET and NTRK fusion negative. And patients were randomized to either pembrolizumab alone or platinum doublet chemotherapy. Patients who were on the platinum doublet chemotherapy were able to crossover to pembrolizumab at the time of progression.

This trial showed us the treatment with pembrolizumab alone in this cohort of patients had a superior response rate, progression-free survival, and more importantly, there was an improvement in median overall survival — 30 months for patients on the pembrolizumab arm compared to 14.2 months for patients who were treated with chemotherapy. We can also see at the landmark 2-year analysis, that there was a 15% improvement in survival, 35% for patients on chemotherapy compared to 51% for patients who are receiving pembrolizumab.

We also saw recent data at SITC and at ESMO last year looking at atezolizumab, the PD-L1 inhibitor, in patients with metastatic non-small cell lung cancer. This trial opened a very similar patient population. This trial though did include patients who were PD-L1 greater than 1%. And this is the first analysis in the patients who were TC3 or IC3 in this cohort of patients. Now, TC3 means that those patients’ PD-L1 expression was greater than 50%. And IC3 was looking at the immune infiltrate. And in those patients that percentage had to be 15% or higher.

This study also demonstrated a significant improvement in overall survival, with a median overall survival of 20 months for patients treated with atezolizumab, compared to 13.1 months for patients who received chemotherapy.

We also saw the data from this trial looking at patients, including an entire cohort, the 1, 2 and 3. And what we see is that the data is less impressive. The median overall survival is 17.5 months for patients treated with atezolizumab compared to 14.1 months for patients treated with chemotherapy.

I actually think this data looks very, very similar to the data that we’ve seen with pembrolizumab alone in the KEYNOTE 042 study.

Now the 042 study enrolled patients whose PD-L1 expression was greater than or equal than 1%, randomized patients to pembrolizumab alone or a platinum-based chemotherapy. On the right-hand side you see the overall survival in all patients, the median overall survival of 16.7 months, almost identical numerically to the atezolizumab data, and median overall survival of 12.1 months for patients with chemotherapy. And when you look at the patients whose PD-L1 expression was greater than 50%, you see a median overall survival of 22 months with pembrolizumab compared to 12.2 months for patients treated with chemotherapy.

Now a lot of people, when they’re looking at some of these other studies that we have with checkpoint inhibitors in the greater than 50% population, they’re comparing it to the 024 data. And that data shows us a median overall survival of 30 months. I think that what we’re seeing in 042 is a more realistic lung cancer patient population. It was a larger study. It wasn’t at such specialized sites. And I think that when I’m talking to my patients who are PD-L1 greater than 50% or higher, the median overall survival that I’m talking about to my patients is more in the range of 2 years as opposed to the almost 3 years that we see with KEYNOTE 024.

We also saw data last year at ESMO, looking at nivolumab with or without ipilimumab compared to chemotherapy in patients enrolled in the CheckMate 227 study. And what this study showed us was that for patients treated with nivolumab and ipilimumab, PD-L expression greater than 1%, we see very numerically similar numbers to what we see in both the IMpower study, as well as the KEYNOTE study, a median overall survival of 17.1 months for nivo/ipi. The chemotherapy arm here did even better with a median survival of 15 months. But nevertheless, the hazard ratio is 0.79 and it does have a positive p-value.

What was interesting in this study was also looking at nivo/ipi and nivo/chemo versus chemo in patients enrolled in CheckMate 227. And this was in the patients whose PD-L1 expression was less than 1%. And to me, this was interesting in 2 ways: One, the nivo/ipi arm appeared to do exceedingly well. A median overall survival of 17.2 months in a PD-L1-negative patient population. And it was also surprising that here the overall survival of nivo and chemotherapy was not as good as what we’ve seen with other checkpoint inhibitors and chemotherapy, even in the lower expressing patients.

From this data, we also saw that PD-L1 greater than 50% patients, those treated with nivo/ipi compared to nivo compared to chemo in that cohort of patients, again, nivolumab and ipilimumab appeared to be superior to either nivolumab alone or chemotherapy alone, both in progression-free survival and overall survival. Here, again, the nivo/ipi overall survival in the greater than 50% of patients was 21.2 months, which compares numerically to the numbers that we’ve seen from KEYNOTE 042 and from the IMpower study. While nivolumab alone didn’t do quite as well at 18.1 months and for chemotherapy, a median overall survival of 14 months.

We’ve also seen data form the MYSTIC trial, looking at durvalumab with or without tremelimumab in patients with PD-L1 greater than 25%. And this trial did not show an improvement in overall survival for patients treated with durvalumab and tremelimumab compared to chemotherapy. Here, the median overall survival was 11 months with durva/treme versus chemotherapy. A really low number, not all we expect from a combination checkpoint inhibitor therapy. And when you look at the durvalumab and chemotherapy numbers on the left-hand side, the median overall survival was positive at 16.3 months compared to, again, 12.9 months for chemotherapy.

Now, we’ve seen a lot of studies looking at combination checkpoint inhibitor therapy with chemotherapy in an unselected patient population. And one could argue is, is this a step forward or is this a step backwards. In some ways, to me, it's a step backwards. These studies were all conducted with pick your chemo, pick your histology, and pick your checkpoint inhibitor, all demonstrating a significant improvement in survival for patients treated with combination chemotherapy. The reason I think of it as a step backwards is because it sort of makes me think about the TALENT and TRIBUTE trials that were done a long time ago in an unselected patient population, given the EGFR inhibitors gefitinib or erlotinib in combination with chemotherapy.

Now those trials were negative and granted, these trials are positive, but those trials were done in patients where we didn’t know about the EGFR inhibitor. And when I think about checkpoint inhibitors, while PD-L1 is the best biomarker that we have right now, I don’t think that it's a biomarker that we're going to be using 5 years from now and helping us select who should get checkpoint inhibitor therapy. And I think, once we have a better idea of what that biomarker is, we’ll be better able to select patients for appropriate checkpoint inhibitor therapy.

I’m also very curious to see the 5-year survival data with all of these trials. Because we’ve seen 5-year survival data now with KEYNOTE 001. We’ve seen 5-year survival data now with CheckMate 003. Both of those studies showed that in an all patient, all comer patient population, the percentage of overall survival at 5 years is 16 to 20%. In the KEYNOTE 001 study, the 5-year survival in the greater than 50% of patients was 30%. And what I want to see from these combination chemotherapy arms, are we still going to have just 15% to 20% of these patients alive at 5 years.

So, is TMB a better biomarker? There has been data, both in melanoma and additionally, from Naiyer Rizvi, looking at non-small cell lung cancer patients, as well as small-cell lung cancer patients from the CheckMate 012 study, suggesting using tumor mutation burden at different cutoffs, using different platforms, may be a better predictor of benefit for treatment outcomes with checkpoint inhibitors.

Now, we’ve also had seen data that tumor mutation burden is a fairly consistent marker across different disease types. And there’s been data looking at response rate correlation with the number of somatic mutations per megabase in patients with non-small cell lung cancer, small-cell lung cancer, colorectal, multiple other tumor types. And we’ve seen that there’s a higher response rate as you increase the median number of mutations per megabase. But we also know that response rate does not always correlate with progression-free survival or overall survival for treatment with checkpoint inhibitor therapy.

We’ve also seen data from the MYSTIC trial, looking at tumor mutation burden in blood and how that correlates with responses. And also showing that there’s a fairly good correlation between tumor mutation burden in blood by the Guardant assay and the tumor mutation burden in tissue that's done by the FoundationOne assay. We also have to remember that when you’re comparing blood from Guardant and tissue from FoundationOne, you’re sort of comparing apples to oranges. The Guardant platform does a much smaller analysis. There are only about 100 genes included in that analysis. The FoundationOne platform has over 300 genes. What Guardant is calling a high mutation burden is not necessarily what FoundationOne is calling a high mutation burden. And so, we really have to take some of the analysis that we see with a grain of salt.

When they looked at this in the MYSTIC data, when they looked at blood tumor mutation burden, they found that as you got to a higher tumor mutation burden it was a better predictor of outcome with durvalumab and tremelimumab, which was actually a negative arm in the study of the MSYTIC — durva and treme was not superior to chemotherapy alone. And they ended up deciding that greater than 20 mutations per megabase was the best cap point to look at for outcomes for patients treated with durvalumab and tremelimumab.

Now, this was also looked at in the CheckMate 227 data for patients treated with nivolumab and ipilimumab. And in this data, they used the FoundationOne platform and found that tumor mutation burden was not a good outcome to predict survival for patients treated with nivolumab and ipilimumab. And both patients with a high mutation burden and a low mutation burden appear to derive benefit when treated with nivolumab and ipilimumab. That's in contrast to when, just above that, we have PD-L1 expression where we can see that greater than 50% was a clear marker of benefit with nivolumab and ipilimumab, while the 1 to 49% patients did not appear to have the same benefit. And those patients appear to do just as well with chemotherapy.

Now, the tumor mutation burden was also looked at as a potential predictor of outcomes for chemo/IO therapy in the KEYNOTE 189 study. And so, this was presented at World Lung last year by Marina Garassino. And what they found is that for patients who were treated with platinum/pemetrexed with or without pembrolizumab, the tumor mutation burden — and they had different cutoffs — did not predict for survival with checkpoint inhibitor therapy and chemotherapy compared to chemotherapy alone.

This data has been fairly consistent with some small-cell cancer that we’ve seen. Because we saw with the IMpower 133 study, we also saw with the CASPIAN study, that tumor mutation burden did not predict for outcomes for small-cell lung cancer patients treated upfront with chemotherapy and a PD-l1 inhibitor compared to chemotherapy alone.

So, what about moving on to different biomarkers? And this is data that's come out of Dr Skoulidis lab, where he’s looked at STK11 and LKB1 alterations, which are thought to predict for a cold tumor microenvironment. On the left-hand side, we see the percentage of patients who may have either LKB1 loss of mutation compared to those tumors that are LKB1 proficient. And what Dr Skoulidis found was that in patients’ tumors that were LKB1 mutant, compared to those patients who were wild type, the amount of CD8 infiltration into those tumor cells were less.

He also looked at outcome in tumor cells treated with an anti-PD-1 inhibitor in those tumor samples that were wild type compared those to that were not. And he found that in tumors that were LKB1 wild-type, treated with anti-PD-1 antibody, that those tumors appear to respond to checkpoint inhibitor therapy, while if you knock out an LKB1, and those tumors are now considered mutant, that those tumors actually do not respond to anti-PD-1 therapy. This also appeared to predict for survival in initial retrospective analysis that he looked at in tumor cells that were KRAS wild type as well as LKB1 wild-type, compared to those that were negative and showing that those patients who were wild type appeared to have a benefit with checkpoint inhibitor therapy compared to those that were knocked out.

He looked at this data in patients who enrolled in the 189 study, and this again is the study where patients were treated with chemotherapy with or without pembrolizumab. He looked at the patients who had the STK11 mutation and he looked at whether or not those patients had a response to checkpoint inhibitor therapy.

And what he found was similar to his preclinical data, suggesting that in those patients who either had STK11 or KEAP mutations, that those patients, who are in red, had a decreased benefit in terms of progression-free survival as well as overall survival when treated with chemotherapy with carboplatin and pemetrexed and pembrolizumab, compared to those patients who were wild-type and who were treated with carboplatin and pemetrexed and pembrolizumab. And he found that the overall survival, the hazard ratio was 2.0 and a p-value of 0.01 and also progression-free survival was worse for these patients.

I will say that this is something that I’m seeing in clinic as well. I’ve had patients where I get back their molecular testing reports. I find out that they have KRAS and they have STK11. I’m treating with checkpoint inhibitor and chemotherapy because that is really is our only front-line option now, and I’m 4 for 4 with my patients having a really quick progression on combination chemotherapy with checkpoint inhibitor therapy. And that also goes along with some of the data suggesting that STK11 and KEAP-1 are actually poor prognostic indicators to begin with in non-small cell cancer patients.

When he looked at STK11 and KEAP and how that goes with TMB, what you can see is that even in those patients who are TMB high, if they have an STK11 mutation or a KEAP1 mutation, that’s the purple line, that those patients appear to do worse compared to those patients who are STK11 and KEAP1 wild-type with high TMB. So, the presence of those mutations seem to outweigh the TMB profile in these patients who enrolled in KEYNOTE 189.

This is also being looked at in the durvalumab data, so adding to our confusion, while we think that that, okay, that makes sense. That's crystal clear. They looked at the role of STK11 alone in patients who got single-agent durvalumab rather than the durvalumab and chemotherapy or the pembro and chemotherapy. And what they found from the MYSTIC study is that STK11 didn’t have as clear of a predictor role. ON the left-hand side you have STK11 mutant and you have the STK11 wild type. And the hazard ratio for those patients who were wild type. So, the mutant patients appear to do fine with durvalumab, a median overall survival of 10.3 months, and those patients who are wild type did not do any better with durvalumab compared to chemotherapy alone.

So, what do we know at this time for patients treated with checkpoint inhibitors in the first-line setting? We know that checkpoint inhibitors have become our standard first-line therapy for patients, either in combination therapy or in chemotherapy, for patients who have no driver mutations. We know that PD-L1 can be used to select patients for single-agent pembrolizumab, and in my opinion, patients who had a PD-L1 expression of greater than 50% should be selected for single-agent pembrolizumab.

The role of TMB in selecting therapy remains unknown. Still not ready for prime time.

Combination checkpoint inhibitor therapy really does lack a good biomarker. And it would be so much more helpful if we could figure out which patients should get that combination therapy and which patients probably shouldn’t even be getting a checkpoint inhibitor to begin with.

And STK11 and KEAP are associated with poorer clinical outcomes with chemotherapy and immunotherapy. However, that data with immunotherapy alone does not seem to be there. And so, I think that's a piece of the puzzle that still needs to be answered.

Second- and later-line treatment for patients with metastatic NSCLC without a targetable tumor mutation who have experienced disease progression on a checkpoint inhibitor alone or combined with chemotherapy

DR GARON: I appreciate the opportunity to speak with you today about second- and later-line treatment for patients with metastatic non-small cell lung cancer who do not have a targetable tumor mutation and have experienced disease progression on either a checkpoint inhibitor alone, or a checkpoint inhibitor combined with chemotherapy.

Before I get onto systemic therapies, first, I would like to address oligoprogression. Oligoprogression is a particularly difficult entity to study because for each person, what they define as oligoprogression may be different than other people. Anecdotally, nonetheless, all of us in our clinics have seen patients who were initially doing well with an immune checkpoint inhibitor, then stopped doing well and had, for instance, a single lesion that was growing. And in many of those cases, we have radiated or ablated those. And, although it is difficult, as I say, to rigorously study, anecdotally we have often found that to be an effective therapy. And, certainly in my own practice, when I do have a single lesion, for instance, that is growing, particularly a single lesion in light of other lesions, I generally have radiated or ablated that lesion rather than changing systemic therapy.

But what I’m going to focus on today in broad strokes are the systemic therapies that would be giving. I think for this talk, I need to break the groupings down into 2 groupings: one would be a group of patients who received a PD-1 inhibitor alone and the other would be patients who received both a PD-1 or PD-L1 inhibitor plus chemotherapy.

So, although, I have a bunch of things listed here, you will see that there are a few things that I am not going to talk about.

One, in patients who have received a PD-1 or PD-L1 inhibitor alone, is that I will not talk about switching to chemotherapy versus keeping the PD-1 or PD-L1 inhibitor going and adding chemotherapy. The reason is I just don’t think there’s very high-quality data to answer what I think is a very important question, which is should a PD-1 or PD-L1 inhibitor be continued under these circumstances? Again, I think we don’t know.

The other thing that I will not talk about is whether or not immunotherapy should be continued in patients who receive chemotherapy. I think that, although there is a great deal of data looking at chemo/immunotherapy regimens, I think there is very little looking at this in patients who have previously received chemotherapy plus immunotherapy.

So, the other thing that I won’t talk about are the multitude of clinical trials with agents that are not available for practitioners. I think that that is a very interesting topic. On the other hand, if a practitioner can’t prescribe it, I don’t think it's of great value today.

So, in broad strokes again, if a PD-1 or PD-L1 inhibitor was used as monotherapy, one option would be to reinitiate immunotherapy. And I think that whether or not to do this I think really depends on a variety of factors.

One factor, of course, is why was the immunotherapy stopped? So, there was a study out of France in which patients underwent rechallenge with nivolumab. Now, some of these people may have stopped for toxicity, whereas others may have had alternate reasons for stopping their immunotherapy.

What was shown in this study was that patients who resumed their immunotherapy often did quite well. This was particularly true among patients who had been on immunotherapy for at least 6 months. And in many respects, I think that this is encouraging. On the other hand, I think an important thing to note here is that the metric was overall survival. And so, somebody who has done well on their therapy over a period of time may be anticipated to do better in terms of overall survival. And whether or not this means that resuming the immune checkpoint inhibitor is the appropriate thing to do, I think is difficult to know.

So, the conclusion of that was that about 30% of patients who had discontinued their first immunotherapy did well. The survival outcomes were promising, but, again, I think that we need to get further data on this.

For me, one sobering note on the study that I was involved with, the CheckMate 153 study. In this study, there were over 1,000 patients who were followed and patients who were continuing to do well at 1 year were randomized to either continue nivolumab or stop nivolumab.

As you can see, there are a little over 150 patients that ended up having response or stable disease at the time of randomization — one thing unique to this study is that it actually did allow patients who were deriving benefit in the eyes of the clinician but who were progressors by RECIST criteria.

What you can see here is that here was a striking difference in terms of progression-free survival, based on patients who continued therapy, in blue, versus those patients who stopped therapy after a year.

And one thing that I think is particularly sobering is looking at the retreatment in 1-year arm, what happened to those patients who had stopped nivolumab?

And, what you see here is the spider plot amongst those patients. And what you can see is that really, it didn’t look very good. When you had patients, who had not progressed but had stopped nivolumab then progressed subsequently, you can see that the great majority of patients had continued growth of their lesions. Which, again, I think leads to some concern. It doesn’t specifically address whether you should reinitiate, but has, perhaps, more implications on duration of therapy.

So, now, what if, instead of a PD-1 inhibitor monotherapy, a patient received a PD-1 or PD-L1 inhibitor plus chemotherapy? So, chemotherapy with or without angiogenic therapy is one approach.

It’s, for many of us, a little frustrating to have to go back to presenting this slide on the efficacy of single-agent docetaxel, but this, of course, is one of our options. The response rate was 7% and the overall survival was better than what was seen with best supportive care, but clearly disappointing.

The other agent in nonsquamous non-small cell lung cancer that can be used in this setting is pemetrexed. So, for pemetrexed, what you can see is it was approved based on a study that showed probably favorable toxicity but noninferiority as compared to single-agent docetaxel.

Now, there have, of course, have been newer studies, the REVEL study, one particularly dear to me, in which ramucirumab was added to docetaxel. The control arm had placebo added to docetaxel.

And there was an improvement that was clearly significant statistically with respect to progression-free survival as there was for survival.

This study has generated some controversy certainly because, although statistically it had been very clear that there is a survival and progression-free survival advantage with this study, many practitioners have been disappointed with the duration of benefit that was added when ramucirumab was added to docetaxel.

There are a host of small studies that have looked at this combination after patients have received immunotherapy. I show one here, but there are many other small studies. The approval for ramucirumab and docetaxel is for patients who received prior platinum-based therapies. So, it would not exclude patients who received chemotherapy plus an immunotherapy regimen. However, we do not have high-quality data assessing the efficacy of that regimen in patients who have already received immunotherapy.

This study was looking at docetaxel plus ramucirumab after a patient would have received an immune checkpoint inhibitor, in this case nivolumab. This was a retrospective analysis and what was shown was that the response rate was very high. The response rate for docetaxel/ramucirumab in the REVEL study was over 20%. But here the response rate was 60%. So, from this study, the data would look very encouraging. However, this is probably the most positive of the studies, when I looked through this data. There are many counterpoint studies that, unfortunately, showed what would be expected from docetaxel/ramucirumab, and did not show any enhanced benefit when it was given after an immune checkpoint inhibitor compared as what would be expected based on the REVEL study.

For completeness sake, I would like to mention the LUME-Lung 1 study. This is a study of docetaxel with nintedanib versus placebo and docetaxel.

This study is not of particular relevance in the United States, but for people who are watching outside of the United States, in Europe there is approval for this combination based on an advantage for progression-free survival, that is shown here, and then there is also a survival benefit in patients with adenocarcinoma. The drug was not approved, that being nintedanib, in the United States based on a step wise analysis for survival that was basically added after the progression-free survival data was there. But for patients in Europe, this is another option. Again, very little data after chemoimmunotherapy.

Last, I will very briefly address immunotherapy combinations. And, although we are all familiar with immunotherapy combination data from the CheckMate 227 study, there is very little data that looks at dual checkpoint inhibitor, which, in this case, would generally be PD-1 inhibitors plus CTLA4 inhibitors. There was a small study that I was involved with, where we looked at the combination of durvalumab and tremelimumab in patients who had received a prior PD-1 or PD-L1 inhibitor, and with that combination the response rates overall were quite disappointing.

So, certainly not the definitive data, but at least now particularly encouraging that you can add a CTLA4 inhibitor to a PD-1 or PD-L1 inhibitor on a progressing patient and have the efficacy that one might hope for.

So, in conclusion, I think that in patients with oligometastasis, even though we don’t have what I would describe as high-level data, I certainly would encourage practitioners to look at local therapies to control the specific site that has escaped the immune surveillance of the immune checkpoint inhibitor and the patients’ own immune system.

I think that the role of rechallenge in patients who stopped a PD-1 or PD-L1 inhibitor is still ill-defined. Salvage chemotherapy can be effective, although whether or not to keep the PD-1 or PD-L1 inhibitor in that setting I think is something that is an important question, but one that we hopefully be able to eventually answer with good quality data. And I would also say that there is no good data at this point for boosting the immunotherapy at least by adding a CTLA4 inhibitor. Of course, there are a whole host of studies that are looking at adding some agent to a PD-1 or PD-L1 inhibitor in this setting.

Anti-angiogenic agents in NSCLC: Where we are, where we are going

DR HEYMACH: It's my pleasure to talk today about a topic that's always been near and dear to my heart, which is antiangiogenic therapy for lung cancer.

 And it's been such an exciting time now for lung cancer treatment with the emergence of immunotherapy used in a variety of different ways, targeted agents now for more and more different subgroups.

So, now with all the new targeted agents and different treatment options for lung cancer, it's easy to forget that anti-angiogenic agents were the first targeted type agents for lung cancer that actually prolonged survival, and that they still may have an important role in lung cancer therapy. And, in fact, the role for antiangiogenic therapy may be growing as we talk about ways that it can act as a partner with other types of treatments.

So, here are what I think are important questions for angiogenesis inhibitors, particularly in regard to lung cancer therapy. First is what’s their role in the era of targeted agents and immunotherapy? Secondly, for immunotherapy refractory patients, which, sooner or later, is going to include virtually all non-small cell lung cancer patients, can it make them responsive? Is this something that adds in that space? And, finally, something that is an area that people have always been asking about, and remains an important question, do we have biomarkers for identifying who is more likely to benefit from antiangiogenic therapy, or is this an example of a targeted agent for which we’re using in an untargeted manner?

And just to highlight the relevance of this area, an old mentor of mine, Bill Kaelin, at Dana-Farber, along with Greg Semenza and Peter Ratcliffe, actually got the Nobel Prize this past year for investigating the HIF pathway. The HIF pathway is probably one of the most important regulators of angiogenesis. So it's turned on by hypoxia and tumors co-opt this HIF pathway to drive angiogenesis in tumors.

So really, they did the seminal work for uncovering how angiogenesis is regulated in tumors and in the setting of hypoxia.

So, let me start with these questions: Should antiangiogenic therapy be integrated with targeted agents? And do we have an idea of which subgroups are likely to benefit?

And the first answer is something that was a bit paradoxical to us when this really first emerged — here, we’re looking at what’s called the BeTa study. This was a Phase III study, led by Roy Herbst and colleagues, and it compared bevacizumab and erlotinib, erlotinib of course being an EGFR inhibitor, versus erlotinib alone. And they looked at who received the most benefit from treatment. And this is when EGFR mutations were first emerging. And surprisingly, the group that got the most benefit from a VEGF inhibitor was actually patients with EGFR mutation. In fact, I’ll summarize a lot of work by saying that the best biomarker we have right now for who benefits from VEGF inhibitors is actually EGFR mutation. And so, this struck people initially as a bit surprising.

Now, this led to some clinical studies. This was one by Seto and colleagues, where they did in just EGFR mutation patients, erlotinib versus erlotinib with bevacizumab.

And really here, if you look at the PFS, the benefit was quite striking. So,16-month PFS versus 9.7-month PFS, a hazard ratio of 0.54. Now, keep in mind, this is dramatically greater PFS benefit than was seen for the trials of VEGF inhibitors, for example, with chemotherapy, in the ECOG 4599 setting, and, in fact, in just about every other setting.

So, this really suggested that EGFR mutant patients may be getting a lot of benefit from VEGF inhibitors. And below that, we’ve got the updated PFS data here as well.

There was a second clinical study that was recently published by Saito and colleagues in Lancet, that showed a similar degree of benefit here, a little different population, and the benefit overall was slightly less. But it supported that VEGF inhibitors seem to work particularly well in patients with EGFR mutation.

Now, the treatment of EGFR mutant disease has pretty quickly in the US evolved from first-generation therapy, like erlotinib and gefitinib, now the third-generation inhibitors, like osimertinib. So, the question is, what about osimertinib plus VEGF inhibitors like bevacizumab?

And this is an early study published by Helena Yu at Memorial Sloan Kettering and colleagues. And it showed that osimertinib with bevacizumab certainly seems to be a highly active compound. And, based on this, there’s now a randomized study that's going as well.

So, there’s no reason biologically, to think that if VEGF inhibitors don’t help with first generation — if they help in first-generation inhibitors, they shouldn’t help just a much with third-generation inhibitors, as well.

And there’s also the RELAY study, this is a randomized Phase III study that was recently reported. Similar design as what I just showed you for the Saito study, erlotinib versus erlotinib with ramucirumab. Now I’ll remind everyone that bevacizumab is antibody against VEGF, the ligand. Ramucirumab is an antibody against VEGF receptor 2, the key receptor. But, overall, a similar concept.

And once again a really striking benefit in terms of progression-free survival, 7-month improvement in PFS. A hazard ratio 0.59. And this, again, is for a properly powered, randomized Phase III — 7-month improvement in PFS. Now keep in mind, things like immunotherapy, we saw nowhere near this degree of benefit in terms of PFS. And for VEGF inhibitors with chemo we saw only a 2- or 3-month improvement in PFS. So this really is an enormous improvement in PFS.

Now on the basis of this 7-month improvement in PFS, and I’ll remind everyone that PFS was actually the primary endpoint of this randomized Phase III clinical study. So, this was a positive clinical study that was recently published. The FDA ODAC committee voted in favor of this combination of ramucirumab with erlotinib.

So, based on their recommendation, usually the FDA will approve regimens. And we’ll wait and see if the formal FDA approval for this regimen does come shortly.

It's worth pointing out that the benefit for adding ramucirumab here extended, not only to exon 19 deletion, which, overall does better, but even to the tougher to inhibitor L858R subgroup. And you see a comparable hazard ratio in both subgroups.

Now the L858R subgroup of EGFR is one that commonly think about adding chemotherapy or other therapies on top of because they have a worse outcome overall. But you can see here, with the addition of ramucirumab, the got every bit as much benefit.

So this may be a regimen, erlotinib plus ramucirumab, that will be FDA-approved in the near future. And, if so, that’ll offer another option for patients. Now I mentioned that that was erlotinib plus ramucirumab and the median PFS was 19.4 months.

In the meanwhile, while that study was ongoing, much of the world has gone from erlotinib to osimertinib as their first-line treatment. So, do you get the same benefit for osimertinib with ramucirumab as you do for erlotinib with ramucirumab?

So, this is a randomized Phase II study that's being led by Xiuning Le at MD Anderson, comparing osimertinib plus ramucirumab versus osimertinib. This study’s enrolling. There’s a similar study that's enrolling well in Japan as well. So, between the osimertinib and bevacizumab and osimertinib and ramucirumab studies, I think there’s a real likelihood that a third-generation inhibitor, like osimertinib, with a VEGF inhibitor could become the standard of care in the near-future for EGFR mutant disease.

All right, now, what about immunotherapy combinations? Does this make sense? And are there any relevant interactions between angiogenesis and immune pathways?

And something that's been known for a number of years, but I think often gets ignored, is that VEGF is not only a key factor of angiogenesis, but VEGF plays a key role as an immunosuppressive agent. So VEGF is known to play an immunosuppressive role in the microenvironment in several different ways. For one thing, it's involved in dendritic cell maturation. It promotes myeloid differentiation towards what we call these MDSCs, the myeloid derived suppressor cells, that are thought to be more protumorigenic. It discourages T-cells from infiltrating in tumors and can inhibit T-cell function. And it can promote more of these T-regulatory cells as opposed to the T-effector cells. So, through all these different mechanisms, VEGF actually can play an immunosuppressive role. So, this suggests the blocking VEGF may enhance anti-tumor immunity.

And, in fact, in preclinical models, in fact this is a study from Gabriele Berger’s lab at UCSF, when you put together DC101, which is D and PD-1 blockade, you actually see a synergistic effect in terms of getting more T-cells into the tumor, activating dendritic cells, and improving the function of cytotoxic T-lymphocytes.

So, all this is saying there’s a good preclinical rationale for putting PD-1 inhibitors together with VEGF pathway blockade.

Now, the original study that led to bevacizumab’s approval in lung cancer is called ECOG 4599. And that was combining bevacizumab with chemotherapy, CP or PC chemotherapy, carboplatin and paclitaxel. And you can see the benefit here was a little bit more than 2 months overall.

Recently the IMpower 150 study asked if you took that BCP regimen, bevacizumab/carboplatin/paclitaxel, and added atezolizumab, the PD-L1 inhibitor, was there a benefit? And the answer was yes. When you add a PD-L1 blockade on top of the BCP regimen, there was an improvement in both PFS and in overall survival as well. So the addition of atezolizumab to bevacizumab improved both PFS and overall survival. And this regimen is now FDA-approved.

It’s worth noting that this trial also included patients with EGFR mutation.

Now that's not the only study looking at these combinations. The study looked at lenvatinib. This is a small molecule oral multikinase inhibitor with pembrolizumab. And this really showed impressive activity for the two together. Overall response rate of 33%. You can see the median PFS and duration of response. And the baseline response rate for pembrolizumab by itself in this setting is about 15 to 20%. So this certainly looked promising.

Another VEGFR TKI called sitravatinib works as a similar mechanism, multitargeted TKI.

And that's been studied in combination with nivolumab in checkpoint-refractory non-small cell lung cancer. So, keep in mind, these are patients who’ve already progressed after PD-1 inhibitors. And if you look at that waterfall plot, you can see the majority of patients are having some degree of shrinkage. So, really, for a group of patients that are refractory, this really does suggest this may be reactivating tumors to immunotherapy.

And based on these results, a Phase III study called SAPPHIRE is actually enrolling patients. Now, ramucirumab, it’s a VEGF receptor-2 inhibitor, is — is also being tested in combination with immunotherapy. There are these 5 different studies and others that are in development as well.

And finally, there’s one being studied in China, sintilimab, in combination with anlotinib, which is a TKI, as first-line therapy for advanced non-small cell lung cancer. And the dataset is small here, but it does show a promising response rate there as well.

So, the bottom line for angiogenesis inhibitors in 2020, first is that EGFR mutant patients appear to be more VEGF-dependent and an EGFR TKI plus VEGF inhibitors, whether it's bevacizumab or ramucirumab, is in progress but shows really promising early data. The RELAY study of course was a randomized Phase III study that was positive, it did achieve its primary endpoint and has now received a positive vote from the ODAC committee.

Resistance to angiogenesis inhibitors is associated with increased myeloid recruitment. I didn’t show all the data supporting that, but VEGF blockade plus this multitargeted TKIs, with PD-1 inhibitors, may reactivate the tumors through multiple mechanisms. And, as before, we do need biomarkers for selecting who’s going to benefit from antiangiogenic therapy? But we do know EGFR mutant tumors appear to be particularly sensitive. And angiogenesis inhibitor combinations with immunotherapy look like they may be particularly effective as well.

So, really, the way I think now of angiogenesis inhibitors, they’re not standing by themselves as therapy, but I think they’re ready to make a comeback, not just in combination with chemotherapy as before, but as immunotherapy’s best sidekick, also as the partner for EGFR TKIs, potentially other DNA damaging agents like PARP inhibitors and so forth.

So, I like to think of this antiangiogenic agents as great sidekicks looking for the right partner, sort of like Robin to Batman here as well.

Recognition and management of autoimmune toxicities with immune checkpoint inhibitors in patients with metastatic NSCLC

DR BRAHMER: Today I’ll be talking about recognizing and managing autoimmune toxicities with immune checkpoint inhibitors in patients with metastatic non-small cell lung cancer.

Immune-related toxicities come in various different flavors. When we’re seeing patients in clinic, we always suspect autoimmune toxicities in our patients being treated with checkpoint inhibitors — of course, we have to rule out competing diagnoses such as infections or progression of their cancer, but certainly evaluating the toxicity, identifying exactly what it is, such as, is it just diarrhea or is it colitis, which is basically increased bowel movements plus abdominal pain, as well as grading the toxicities. And I’ve outlined this here: Grade 1 is basically supportive care, considering drug hold. Grade 2, holding the immune therapy, considering starting corticosteroids. And Grade 3 and 4 where we’re stopping the drug and giving high-dose corticosteroids.

Grading the toxicities will help you manage the patient and decide what to do with immune checkpoint inhibitors as well as what treatment should be performed. Certainly there’s various different toxicity recommendations or guidelines. As I’ve outlined here, ESMO has an oncology expert consensus. SITC has a multidisciplinary expert consensus, as well as ASCO and NCCN — both have separate immune-related toxicity guidelines — that are both driven by multidisciplinary specialties, as well as looking at expert opinion and also including other collaborations within oncology, including nursing.

Now, what organ systems are affected by immune therapy? If we have a patient on immune therapy, what should we expect? It's all fair game, as we can see here — all the way from skin rashes to neurologic, musculoskeletal abnormalities that can occur, and we have to keep all these organ systems in mind.

When we talk about PD-1 or CTLA4 antibodies, or PD-L1 antibodies as well, one of the most common toxicities is the skin toxicity. First and most frequent is the skin toxicities, and this typically occurs within the first couple of weeks, but these types of rashes can be very diverse. One of the most common is maculopapular rashes. And typically these cases are generally easy to manage, and less than 5% of cases lead to treatment discontinuation.

Grading and severity is based on the amount of skin that's involved. And the thought is again that most of this type of skin rash occurs early on and certainly is much more common in patients on combination with PD-1 or PD-L1 with a CTLA4 inhibitor.

Certainly there are different mechanisms of action that is thought. Certainly certain HLA subtypes are more common to have rashes occur, and certainly different cytokine panels may actually predict for cutaneous side effects from this type of combination.

Colitis is another combination. Certainly classically seen with a CTLA4 inhibitor. But certainly as we’re combining these PD-1 or CTLA4 inhibitors or antibodies together, we’re seeing more colitis. But it can also occur with single-agent PD-1 or PD-L1 therapy. Certainly it's very important to rule out alternative diagnoses, particularly in our patients who are immune compromised. C. diff can still be seen. Other GI infections — I just had my first norovirus diagnosed in the setting of immune therapy. And then also distinguishing between diarrhea and colitis.

Invasive testing can be used in patients, certainly colonoscopy can be used, particularly if patient’s diarrhea is refractory to steroids. It is helpful to decide what type of therapy to go to next.

We typically have a low threshold for starting corticosteroids. If, typically, patients have — some patients will have an irritable bowel type of syndrome, even going onto immunotherapy, and typically there’s not been a benefit seen for patients being treated beforehand to try to prevent colitis with oral budesonide.

Colitis that is slow to improve or refractory to steroids we typically recommend treating with anti-TNF drugs such as infliximab. This is typically given at a 5-mg/kg dose once or twice every 14 days. And if it's still refractory, then we do consider vedolizumab, which is another drug, if, again, the colitis or diarrhea is refractory to initial therapies.

Now, lessons to try to manage colitis, we do take some lessons from our GI colleagues from the irritable bowel or inflammatory bowel disease. Stool lactoferrin is certainly a stool biomarker for colitis. It typically is known to have a 90% concordance with histologic inflammation and a 70% sensitivity for endoscopic abnormalities. Stool calprotectin can also be used to evaluate these patients. And if it's positive, it's typically associated with presence of ulcers on endoscopy. Certainly there are some high-risk features on endoscopy that our GI colleagues look for, and they can use this to decide whether or not to go onto TNF inhibition or the need for hospitalization.

Vedolizumab is another monoclonal antibody that is typically used for IBD, but in those patients with refractory colitis it can be used. And our GI colleagues follow stool calprotectin to see if this is being helpful or not, in addition to monitoring the actual grade of colitis or diarrhea.

Endocrine toxicities are another common toxicity that we see. We typically see thyroid disease in our patients, certainly hypothyroidism is one of the most common endocrine side effects. But we can also see hypophysitis, particularly with the anti-CTLA4 antibodies and in combination. There are some rare Grade 1 or Type 1 diabetes seen, and it is important to differentiate from our Type 2 diabetes.

Adrenal insufficiency can be seen as well but typically is actually hypophysitis first causing then adrenal insufficiency downstream. But I do have a few patients that have primary adrenal insufficiency. And then rarely hypoparathyroidism or hypogonadism.

The diagnostic workup. Typically in my case would be done in concert with my favorite endocrinologist, but certainly we do use laboratory monitoring for hormone function. And then if a patient is suspected to have hypophysitis, particularly if they’re having headaches, an MRI would be done. And sometimes the pituitary can be seen and is inflamed.

Certainly endocrine is helpful when managing some of these more complex autoimmune endocrine abnormalities, and hormone replacement typically is needed. Sometimes we will use steroids for pituitary dysfunction, particularly if they’re having headaches or symptoms, this will help. But typically long-term endocrine hormone replacement is required.

Again, hypophysitis is a classic endocrine toxicity of CTLA4 inhibition. These symptoms include even vague fatigue and weakness but, also can come into the ER with acute onset of headache and confusion. Typically if the pituitary is abnormal on MRI, about 75% of the time, and their thyroid function can also be abnormal.

We know that CTLA4 itself is expressed in the pituitary and antibodies can bind to the pituitary cells after exposure to the drug and deposition of complement and pathway activation can be seen, thus causing damage to the pituitary. And by the ipilimumab can directly activate complement itself. There’s various different ways that the pituitary can be damaged when patients are treated with an anti-CTLA4 antibody.

Now, for lung cancer patients, those patients that I typically take care of, pneumonitis can be seen, but this can occur in any of our patients treated with PD-1 or PD-L1 inhibition, as well as with combination with anti-CTLA-4. Pneumonitis can come in various different flavors. As you can see, it can be nodule. It can be diffuse. The timing is typically unpredictable, as seen in one of these cohorts. That has been published. Typically it would be seen in the first few months but can occur even beyond a year while being treated.

Histology, as you can see here, adenocarcinoma tends to be seen more often — or associated more often with pneumonitis, but again, that can be seen also in nonsquamous histology as well as squamous histology. The grading of pneumonitis is helpful. A lot of times we will see Grade 1 pneumonitis because we are doing routine scans on these patients. If we’re concerned, you can consider holding immunotherapy while you’re figuring out what is going on. But certainly if they have asymptomatic pneumonitis you can continue to treat through while you’re evaluating. We do monitor these patients very closely with pulse oximetry and then consider evaluation with pulmonary. Certainly with Grade 2, now this is symptomatic, where we’re holding immunotherapy and starting low-dose oral prednisone, 1 to 2 mg/kg, and monitoring every few days to make sure that they’re recovering.

And then those patients with high-grade, or Grade 3 or 4 pneumonitis, these patients typically we’ll admit, consider whether or not to do a bronchoscopy if we have a wide differential causing these changes on CT or if there are symptoms, and give them 1 to 2 mg/kg a day of methylprednisolone, and then continue to monitor these patients very closely. If no improvement after 48 hours, then we’ll consider either doing various different things — infliximab, IVIG or mycophenolate have been used as well. Again, only if these patients are not improving after 48 hours. Typically, though, pneumonitis will rapidly improve with the institution of corticosteroids.

Now, the question is, what are the risk factors and T-cell mediated mechanisms for pneumonitis? You can see here. Certainly we’ve seen patients who have a smoking history have a higher risk of developing pneumonitis. And looking at various different cytokines, as well as the milieu between T-cells and myeloid immune cells that may tip off the immune inflammatory cascade causing pneumonitis.

Certainly we’re trying to figure out how best to care for these patients, since, at least form a lung cancer standpoint, pneumonitis can be complicated. Typically, again, these patients will rapidly improve on corticosteroids, but those patients who don’t improve within 48 hours, we are considering clinical trials to figure out how best to treat these patients in that situation. And this is one particular trial that is open across the cooperative groups, looking at taking patients with steroid-refractory pneumonitis and randomizing them to either infliximab or IVIG. This is led by Dr Naidoo, who’s part of my group.

Rheumatologic immune side effects are, I think, underevaluated and under-reported in a lot of clinical trials. We are working with rheumatology to evaluate this further. A lot of times, particularly in our patients with lung cancer, they’re older, have a lot of underlying osteoarthritis that we may see flare. And not so much flare from an osteoarthritis standpoint, but then becomes more inflammatory. And some of this, if you’re looking across the clinical trials that have been done in lung cancer specifically with PD-1 or PD-L1 inhibitors, it is very hard to tease out how often this is truly seen.

There’s really a lack of standardized reporting of arthralgias, arthritis — they’re reported as joint pain or myalgias, joint effusions. But we can see rheumatologic disorders in a wide array, all the way from sicca syndrome to polymyalgia rheumatica to myositis, even vasculitis or psoriatic arthritis or just even more of a rheumatologic abnormality.

And you can see here in the clinical phenotypes of rheumatologic disorders that we can see. The management depends on the clinical phenotype — seronegative spondyloarthropathies versus reactive arthritis versus this more classic rheumatoid arthritis. Typically, these patients are managed with oral steroids but can also be managed with intra-articular steroids or even steroid-sparing nonbiologic disease modifying agents. Sometimes anti-TNF agents can be used as well. Interestingly, a lot of these arthritis type of syndromes may not respond well to oral steroids.

Now, there are some rare immune-related toxicities that I wanted to make sure and go over. Now, these are uncommon but can be devasting. Certainly the cardiac toxicities — myocarditis — has gotten a lot of press. It is uncommon. It can be more commonly seen in tandem with myositis, which I’ll get into in a bit. Certainly there can be ocular immune-related side effects, a nephritis type of picture. Pancreatitis picture. Certainly on clinical trials and in our patients, we may see a lot of asymptomatic elevations of amylase and lipase, and I don’t typically follow this regularly when these patients are on immunotherapy, unless on a clinical trial. However, we have some patients that do develop symptomatic pancreatitis. Evaluating them to try to figure out what exactly is the cause, is it just a standard stone, biliary tree stone that's causing the pancreatitis or drinking or smoking, or is this truly immune related.

There are, also, neurologic side effects that can occur in patients on immune-related type of drugs, such as myasthenia gravis, encephalitis and neuropathies. We can also see Guillain-Barré syndrome. But, again, these are much less common. There are some rare hematologic side effects, including classic anemias as well, or even ITP can be seen as well.

For neurologic immune-related side effects, this is really diverse. We can see neuromuscular phenotype, which is more likely compared to just a neurologic immune-related toxicity. The time of onset can be unpredictable, but typically the onset can be very rapid and, in some patients, very life threatening. Treatments may include steroids, IVIG, plasma exchange or even rituximab. And certainly in these cases we get our neurology colleagues involved.

Now, myocarditis is more commonly seen when we’re using PD-1 or PD-L1 in combination with CTLA4, but still, these fatal cases are extremely rare and just the occurrence — myocarditis is very uncommon. But when these patients develop myocarditis, there is a high rate of fatality.

And we can see actually T-cell infiltrates in the heart, and this can also be seen with — T-cell infiltrates at the same time in skeletal muscle. And high-dose steroids are recommended right away instead of low-dose steroids. We start with the high dose. We admit these patients. And if no improvement, you can try ATG.

I just went over a lot of these patients, particularly with PD-1 and CTLA-4 combinations, can develop multisystem IRAEs or overlap syndromes. This was one trial that was published that looked at patients with metastatic skin cancers treated with immune ICIs, and myocarditis was one of the most frequent neuromuscular IRAEs and in about 30% of the time was concomitantly diagnosed with myocarditis, as well as myositis at the same time, as you can see here. Myasthenia gravis can be also an overlap syndrome in these cases as well.

At Johns Hopkins, we have immune-related toxicity teams, so I am quite spoiled, but certainly having folks within each of these disciplines that you can call on to have them help you work up these patients is very helpful. For our group, we’ve centralized the discussion, particularly of complex immune-related side effects, and we’ve discussed and refined our guidelines quite frequently. We do have order sets for these immune-related toxicities so that we can diagnose and manage these quite quickly. And certainly within our group, there’s a lot of education as well as research that's ongoing.

Dr Naidoo and Dr Cappelli — Dr Naidoo is from oncology and Dr Cappelli is from rheumatology, and they are our co-directors.

Certainly we’ve noticed over time there’s a frequency of reporting, so we have a centralized email where questions about patients are sent in and we’re available to connect these providers with the appropriate subspeciality, and this is a way that we can get our patients in quickly to be seen by these subspecialties or get them consulted on if they’re on an in-patient unit.

Immune-related side effects, certainly I’ve tried to simply the management. I think across the board, if patients have Grade 1 toxicities, these are mild or no symptoms, we consider delaying our immune treatments, have frequent reassessment and then, if they’re symptomatic or develop symptoms, then we start treating with them with steroids.

Patients with moderate symptoms, or Grade 2, we delay the ICA and then evaluate early and frequently, and we consider steroids, typically dosed at 1 mg/kg. Once symptoms improve, steroids are tapered slowly, over at least 4 weeks, and sometimes it may take longer to taper these patients off. If patients have moderate to severe symptoms, we say Grade 3 or 4, we discontinue the immune therapy, have very early assessment — typically these patients are admitted. They’re given higher-dose steroids, and if no improvement, depending on the immune-related side effect, consider adding other immune suppressants.

One thing to note when putting patients on steroids, patients should really receive stomach protection, either PPI or H2 blockers, and consider prophylactic antibiotics. During the steroid taper, patients should be evaluated frequently and for an extended period of time. During these tapers, sometimes their side effects can flare.

Again, going back to important, you’re not alone out there. We have the ability to — these guidelines, across the board, and certainly trying to find a guideline that seems to work well with you and your staff, either the NCCN, ASCO, SITC or ESMO. These are all expert guidelines that are available for reference and are relatively easy to follow.

Certainly this is becoming the mainstay of therapy. Immunotherapy is here to stay. We definitely do need to know how to take care of our patients while being treated with these types of drugs. I think this will only continue to expand its use, hopefully into even earlier-stage disease. And certainly seeing these immune-related side effects, while in general are uncommon, that they may need treatment, we do need to be able to assess, reassess, treat and evaluate these patients routinely.

Management of metastatic NSCLC in special populations

DR LEVY: I’m going to be talking today about the management of non-small cell lung cancer in special populations.

And I think special populations has many different definitions when it comes to lung cancer patients. And for the sake of the presentation today, we’ll limit that population definition to one, rapid progression on first-line therapy. Two, long-term responders to immunotherapy — and certainly a lot of work looking at this group of patients, importantly — and then finally, pre-existing autoimmune disorder.

So, we’ll go through the data we have for each of these patient populations and the relevance that the data has for clinical application.

We’ll start with rapid progression on first-line therapy.

I think that the data that we have from this comes from the REVEL trial. As a reminder, the REVEL trial was one of the largest Phase III studies in the second line, evaluating the addition of ramucirumab to docetaxel compared to docetaxel alone in patients with advanced non-small cell lung cancer who’d received and had progressed on 1 prior platinum-based chemotherapy, with or without bevacizumab. Importantly, the primary endpoint was overall survival. This study did include all histologies and prior bevacizumab was permitted, another anti-angiogenesis drug. Stratification was based on performance status, prior maintenance and location, in terms of East Asia versus the rest of the world.

And the baseline characteristics in this trial are fairly indicative of what we see for our lung cancer patients. The average age, of course, was slightly lower than we usually see, around 62, 61. But importantly, this patient population was both adeno and squamous cell, roughly 25% of patients in this trial were squamous cell. And, unfortunately, we don’t know much about the mutational analysis for these patients — close to two thirds of patients in each of the arms didn’t have tissue analysis done for molecular interrogation.

As a reminder, in terms of the intention to treat analysis, the addition of ramucirumab to docetaxel did improve survival, did improve progression-free survival, did improve response rate, when compared to docetaxel alone.

These benefits were modest, nevertheless, it did lead to an approval of ramucirumab with the combination to docetaxel for patients who had received 1 prior line of platinum chemotherapy.

And importantly, there has been additional data that has come out now about rapidly progressing patients. Patients on this trial who had rapid progression on first-line therapy prior to the enrollment on the study. Now, REVEL was not powered for subgroup analysis, but this effort did have a sensitivity analysis on other subgroup of patients with aggressive or rapidly progressing disease from the intention-to-treat population, including patients with all histologies or only adenocarcinoma histology who remained on first-line therapy for less than 4, less than 8, and less than 12 weeks of initiation of front-line therapy. And essentially, what this subset was trying to answer is, does the addition of ramucirumab to docetaxel improve outcomes specifically for these patients who do develop rapid progression on first line.

And, if we look at the data, I think importantly, if we look at those patients who had received duration of first-line therapy less than 4, less than 8 and less than 12 weeks of time on first-line therapy, we saw that the addition of ramucirumab to docetaxel did salvage these patients, at least did lead to an improvement in survival when compared to docetaxel alone. And this was not a small group of patients from the trial. This was close to 30% of patients on the trial who, at best, had progressive disease on front-line chemotherapy prior to going on this study. I think importantly, we saw that the addition of ramucirumab to docetaxel did improve outcomes.

And I would say anecdotally, in somebody who sees patients like this, these patients are in trouble. These are patients that you give first-line chemotherapy to. You’re hoping for the best. But then, if they develop rapid progression, we’ve had very little data to understand what to do with these patients.

So I think this is a data set that we need to think about.

Now, this data was generated prior to the advent of immunotherapy being frontline and whether the same rules would apply for patients who had progressed on chemo/IO or IO alone we don’t know. But nevertheless, this is some of the data that has been generated out of a very large Phase III study that may be able to inform treatment decisions for these patients. And I will say that patients who are progressing on the clinical, on platinum/immunotherapy, I will consider using ramucirumab in combination with docetaxel, extrapolating from this subset analysis from REVEL.

The outcomes, in terms of safety, there weren’t additional safety signals that we saw from the refractory subgroup. We see that treatment related adverse events were no different between the 2 groups in terms of ramucirumab versus placebo. So that there wasn’t additional toxicity in this subgroup.

And I think importantly, again, we think that this may help us inform treatment decisions for patients who have rapidly progressing disease and we don’t have a lot of other data on what to do with these patients.

So, I think it's important to remember, even though is a subset, it did show improved outcomes for this group of patients who very little treatment options exist.

The next set of special population patients to discuss is the long-term responders to immunotherapy. There’s been a tremendous amount of work with this group of patients in terms of longitudinally following these patients to see what their outcomes are, looking at the tail of the curve, which is so important, but also looking at biological or clinical features that may better define who these patients are.

If we look at a pooled analysis from the Phase I, II, and III trials with nivolumab, and looking at long-term outcomes for these patients, we see that at the landmark analysis at 48 months showed a survival of 14%, that's pretty impressive. I will tell you for CheckMate 003, just the Phase I experience, the 5-year survival rate was 16%. And that's just something we haven’t seen before prior to the advent of immunotherapy. And I think it's certainly telling us or starting to weave a story that you can achieve long-term outcomes with some of these patients.

Now, most of these patients, from at least CheckMate 003, had received at least 3 lines of therapy. So, clearly, more work needs to be done to better define who these patients are. But we are beginning to see a tail of the curve.

And if we look at KEYNOTE 001, which was the Phase I experience with pembrolizumab, and looking at long-term outcome, the 5-year overall survival — again, many of these patients had been previously treated. This is the graph on the right — and these, the 5-year survival rate was 15%. We look at the treatment-naïve patients because some of these patients hadn’t received a front-line treatment, some of these patients were treatment-naïve. The 5-year survival rate was 23%. Again, similar data, perhaps a little more encouraging from the nivolumab experience, but certainly in the previously treated patients, very similar numbers — roughly 15% of patients who were highly pretreated are still alive at 5 years.

If we break this down, this same group of patients from the same trial, based on PD-L1, not surprisingly PD-L1 seems to be a predictive biomarker for outcome for these patients, both for treatment-naïve patients, patients who had a PD-L1 greater than 50%, had a 5-year survival rate of 30% incredibly. And in the previously treated patients, patients who had a PD-L1 greater than 50%, again a 5-year survival rate of 25%.

So I think PD-L1 certainly is one of these factors to look at to say who’s going to be still living or who’s going to be able to achieve that tail of the curve. PD-L1 is certainly a factor to consider.

And then, looking at the pooled analysis from CheckMate 017 and 057, these are the 2 trials comparing nivolumab to docetaxel in second line for squamous cell and adenocarcinoma patients, respectively. We looked at the pooled analysis again and stratified, based on PD-L1 — or looked specifically at the PD-L1 less than 1 and the PD-L1 greater than 1, we see for the less than 1 long-term outcome for these patients, the landmark 48-month analysis showing 9% of patients alive treated with nivolumab. And again, PD-L1 greater than 1, we see a little bit better outcome with nivolumab, 20% alive at 48 months.

So, PD-L1 may be one, again a biomarker beginning — we’ve already learned about this biomarker, but it certainly may be correlated with long-term outcome with these patients.

And then finally, looking at this same dataset but looking at patients in the nivolumab arm who achieved either a complete response or partial response compared to those who had stable disease or progressive disease, not surprisingly the chances of being alive at 48 months was much higher for those patients who achieved a CR/PR, 58%, that's pretty remarkable compared to those that had stable disease or progressive disease. And we see similar trends with docetaxel, but nevertheless, these number are much less.

There’s a lot of questions that I brought up when this discussing this data and we’re still trying to learn is there a clinical phenotype for long-term responders? What is the tumor biology for long-term responders? Is there some sort of shared genetic or immunophenotypic features that all long-term responders have? And then, how long should patients be treated? You have disease stability on immunotherapy. This is a question that continues to pop up at many meetings and committees and venues about what is the optimal duration of immunotherapy? Many of these trials stopped after 2 years. And there’s still questions about whether patients should continue after 2 years or not, specifically for these patients who have disease stability and are doing well.

I put this dataset in here because I’m a huge advocate of circulating tumor DNA, number one, and number two is, this data is relatively new and very thought provoking. And it certainly tries to tackle or leverage ctDNA use in terms of looking at long-term responders. And I think we’ve always viewed or more recently viewed ctDNA through the prism of genotyping lung cancer. This is a platform that can be used when trying to understand the genetic alterations of a tumor when tissue is insufficient. And we’ve learned that layering in a liquid biopsy or plasma test for patients may enhance your ability to capture genetic alterations that are relevant and actionable and can be treated with TKIs. But there are other uses of ctDNA — certainly to evaluate for minimal residual disease, as well as kinetics of ctDNA, looking at changes over time and correlating that with response that we see on scan.

This was a little bit different. This was looking at ctDNA as a surrogate for long-term responders, and this was a retrospective analysis out of Memorial Sloan Kettering that essentially identified many of their patients had gone onto clinical trials, tried to identify those patients who had had long-term benefit from immunotherapy as defined as a PFS greater than 12 months. And they looked at roughly 300 patients and then found roughly 31 of these patients had some long-term benefit from immunotherapy. And they utilized in these patients next-generation sequencing-based, liquid-based platform to try to identify ctDNA in plasma during or after treatment. And the collection time point was a median of 25.7 months after the initiation of treatment.

So, these are 31 patients, who were doing very well, on immunotherapy. Had disease stability. And at some time point the average being roughly 26 months, they drew a ctDNA analysis. And intuitively, but interestingly, those patients who were negative for ctDNA at that time point had very little chance of further progression. But if those patients had identified positive ctDNA — again, these were all long-term responders — but at that point, if you identified ctDNA in their blood, there was a good chance that they would subsequently progress. These are small numbers — 27 patients were negative, and they did very well. Four patients ended up being positive and essentially, in that group, they were already long-term responders. If you then subsequently identified ctDNA in their blood, 4 patients, but most of these patients progressed.

So, this may have some utility for our long-term responders. And this is thought provoking and certainly hypothesis generating, and we’ll need more data. You can imagine a scenario for patients who’ve been on immunotherapy for quite a long time, whether you can use ctDNA to say, if they’re positive maybe this is an opportunity to either continue the immunotherapy or to layer in additional drugs. And if they're negative, maybe this is the time that you can stop and be rest assured that they’re going to do fine moving forward.

So, certainly thought provoking. And I await to see more data like this because I think we’ve just begun to understand how to leverage ctDNA in our lung cancer patients.

Finally, a special population, the pre-existing autoimmune disease patient population. This is not a small group of patients with lung cancer. There’s been some estimates that up to 15% of all lung cancer patients do have pre-existing autoimmune diseases.

The question really is, why do immune-related adverse events occur in the first place? The precise mechanism is not known. We know that CTLA4 and PD-1 and PD-L1 access is important in self-tolerance, and this is evidenced by CTLA4 knock out mice that die from lymphoproliferative disease. And PD-1, like PD-1 knock out mice that can develop lupus-like autoimmune diseases. So, we know that this is important.

And probably the best data we have for patients with pre-existing autoimmune diseases with lung cancer, is a retrospective analysis. It's 56 patients with non-small cell lung cancer and a pre-existing autoimmune disease treated with immunotherapy. In this group, roughly 80% were asymptomatic. They did not require treatment for their autoimmune disease. Twenty percent did.

I think this effort tried to answer 2 questions: one is, are there flares of pre-existing autoimmune diseases when patients get immunotherapy? And then two is, are there immune-related adverse events unrelated to their pre-existing autoimmune disease that occur more frequently in patients treated with immunotherapy?

And the first question, really 25% of patients had some sort of exacerbation of their autoimmune disease. The majority of these were Grade 1 and 2. The majority of these did not require any treatment, although 4 did require systemic steroids. So, I think that's important.

The second question answered, in terms of immune-related adverse events that are unrelated to autoimmune phenomenon, and there was about 38% experienced in immune-related adverse event. But again, most of these were mild, three-quarters of these were Grade 1 or 2 and a quarter of them were Grade 3 or higher. The AEs were generally manageable. They infrequently led to permanent discontinuation of immunotherapy. And then finally, the response rates to immunotherapy was no different than other lung cancer patients, around 22% — that's single-agent immunotherapy and eliciting a response rate in this patient population — around 22% which is right in line with what we see in the general population for lung cancer.

And so, this is just a table, going over this again — 56 patients, most of these patients were asymptomatic. Of the 56, 13 developed some sort of flare of their underlying autoimmune disorder. You were more likely to develop flare if you had autoimmune symptoms at baseline. The exacerbations were usually Grade 1 or 2, and most patients didn’t need to permanently discontinue immunotherapy.

I think one of the caveats of this study is that — they studied a wide variety of autoimmune diseases in this group of 56 patients that range from rheumatoid arthritis flare, derma psoriasis, grades, ulcerative colitis, myasthenia gravis, multiple sclerosis.

And I would just make the point that treatment decisions for these patients has to be individualized. I’m not sure I would feel comfortable giving immunotherapy to a patient with multiple sclerosis or even myasthenia gravis.

So, I think every autoimmune disease has to be viewed through the prism of an individualized treatment approach. And I think you can’t just gloss over and say, well, they have an autoimmune disease, there’s safety in published literature so we do it. I think you have to be very careful and look at each one specifically as it relates to that particular autoimmune disease the patient may have.

So, the conclusion about immunotherapy in these patients with pre-existing autoimmune disease is, is that flares and immune-related adverse events in these patients can be managed by discontinuing therapy. They generally are mild, Grade 1 or 2. We have to be vigilant in monitoring the development of flares, but I think if we jump on top of this early, we can address these.

And again, we all need to make individualized treatment approaches when it comes to patients with autoimmune disorders who are treated with immunotherapy.

Long-term care for patients with metastatic NSCLC with EGFR tumor mutations: Resistance mechanisms and related management approaches

DR BORGHAEI: Today we are going to discuss the care of patients with EGFR mutated non-small cell lung cancer and some of the more recent data and management questions that have come up as a result of some of the advancements that we have made but also as a result of the fact that the field is advancing. And now we have newer information that we need to react to and keep in mind when we care for our patients with this particular subset of non-small cell lung cancer.

I would say that probably the EGFR pathway is one of the best-studied oncogenic pathways in most malignancies. There’s been a lot of interest in this particular pathway in a number of diseases, but I have to say that, at least in my view, and I know that I’m biased in the world of lung cancer, we’ve had a lot of success targeting this particular pathway to the benefit of our patients. We know that there are many different subtypes of EGFR mutations. The one that we have concentrated on the most have been the exon 19 deletions or the L858R mutations. These were traditionally called activating mutations of the EGFR, and there was interest in them because we learned that some of the earlier tyrosine kinase inhibitors that we were using had preferential activity when there was a mutation in these particular sites.

We have come to learn that there are many other forms of EGFR mutations, some uncommon mutations, that could not be appropriately targeted with some of the first-generation drugs. We now know that about 10% to 12% of our patients have these exon 20 insertions that are of particular interest for drug development, and we’re going to be covering some of that data.

Over the last decade or so, we’ve learned a lot about the EGFR mutations and how to use these mutations, in terms of drug development, to figure out which one of our patients should get what kind of drug. In general, when we think about oral tyrosine kinase inhibitors that target this pathway, we think about several different classes of drugs. We call them first-, second- and third-generation drugs.

The first-generation drugs are represented by gefitinib and erlotinib, one of the earliest drugs we used in this particular pathway. These were traditionally reversible inhibitors, and they had activity against the common mutations but also had quite a lot of activity against wild-type EGFR, which is why we had to deal with side effects such as rash and diarrhea.

Second-generation drugs improved to some extent on the first-generation drugs in that they were irreversible inhibitors, but they still had some activity against wild-type EGFR. And then finally we ended up with osimertinib as our only third-generation drug because the other drugs, unfortunately, did not show clinical activity in the setting that we were investigating as part of clinical drugs, but osimertinib did.

This drug is another irreversible inhibitor of the EGFR pathway, and it is active both in the first line and second line. However, it also comes with the added benefit of having less activity against wild-type EGFR and also has some intracranial activity. Over the years, as a result of clinical trials, we have gradually moved osimertinib into our front-line treatment strategy.

The study that really established osimertinib as our front-line treatment is FLAURA. The design — as shown in this particular slide — there is nothing different about the design of this trial as opposed to many others in this category. In essence, patients with good performance status, with activating mutations in exon 19 or 21 of the EGFR, were randomly assigned to receive standard dosing of osimertinib, 80 mg once a day versus one of the first-generation drugs, either gefitinib or erlotinib, and then they were followed with a primary endpoint of overall survival.

At the time of initial reporting, these are the graphs that we had. On the left-hand side you can see the progression-free survival curve. What is shown in blue are the patients who were assigned to osimertinib. The red is the standard first-generation EGFR TKIs. You can see that the curves separate very early on, and clearly there is a benefit in terms of progression-free survival for the use of osimertinib. And the numbers are given on the slide with a median progression-free survival of roughly 19 months versus 10 months in favor of osimertinib.

Now, on the right-hand side you can see the preliminary overall survival, but if we go to the next slide, this represents the final overall survival analysis as was presented at ESMO in 2019. Again, here the yellow graph represents the patients who were treated with osimertinib, and the green on the bottom are patients who were treated with the first-generation oral TKIs.

Here we get a median overall survival of almost 39 months with osimertinib versus about 32 months with the first-generation EGFR inhibitors, with a good hazard ratio of 0.79, so establishing at least in parts of the world where osimertinib is available this drug as our first-line treatment of choice or the preferred treatment of choice for patients with the usual EGFR mutations.

Unfortunately, what we know is that we’ve not been able to cure our patients with this particular mutation — resistant mechanisms have been known to occur when we were using first-generation drugs. One of the more common mutations were this T790M mutation. Osimertinib has activity against this particular mutation, and that is why initially we were using this drug in a second-line setting in those patients that had T790M. But then, as a result of FLAURA, as we just discussed, we recognize that the drug has activity in the first-line setting. However, again, unfortunately, patients progress.

There’s been a lot of interest in identifying potential bypass mechanisms or resistant mechanisms, and some of those are shown in this particular slide. I’d like to bring your attention to the right-hand side of the slide, where up to about 15% of these patients can actually have a MET amplification as a potential way to bypass the EGFR pathway. As you can see, other mutations in the EGFR pathway, such as the C797X, or alterations in the HER2 pathway have also been documented, in addition to alterations in PIK3CA and some other minor pathways as shown on this particular slide. The reason we have interest in identifying these pathways is because we have the ability to try to address these resistant mechanisms with other drugs, perhaps in conjunction with osimertinib, to see if we can improve on the efficacy of osimertinib.

You can think about the process, as depicted on this particular slide — so if you start with treatment-naïve tumor cells, they seem to be sensitive to a particular treatment, let’s say in this case osimertinib. We get to a state of what we can refer to as minimal residual disease, meaning that the majority of the cancer cells are responding, but there are an occasional cells or groups of cells or family of cells that become resistant to the treatment. Perhaps it is in this setting that approaches such as local radiation therapy, for instance, could be effective. However, we do get to a state where there’s actual drug resistance, and there is for the growth of these tumors. If we can identify these potential resistant mechanisms and address them in a systemic manner, we might be able to extend the efficacy of a drug that, by and large, is well tolerated, quite effective against particular cancers.

How do we address these resistant mechanisms? Over the years we’ve devised a number of different strategies to see if we can take advantage of some potential resistant mechanisms or additive benefit from other drugs to overcome resistance. One of the ways that we have favored doing that, particularly for patients with EGFR mutation, is the use of anti-angiogenic agents.

Bevacizumab is a drug that most of us are very familiar with. We know that it has activity in non-small cell lung cancer. And based on preclinical studies, it was postulated that the combination of a VEGF inhibitor plus an EGFR TKI could be effective in terms of overcoming some resistant mechanism or, perhaps, extending the clinical efficacy of oral EGFR TKIs.

As you can see on this slide, we started with first-generation drugs, using drugs such as erlotinib plus bevacizumab. In this particular trial from our Japanese colleagues, we see that there was an improvement in PFS. However, there really wasn’t a significant improvement in the overall survival of patients. And this study has been published and discussed at various meetings.

Another study, also from Japan, with a very similar design — patients were receiving either erlotinib or erlotinib plus bevacizumab — arrived at a very similar conclusion in that PFS was improved with the addition of erlotinib to standard dose of bevacizumab. Again, overall survival here is not really available.

Where the availability of ramucirumab, a different VEGF inhibitor, this one, again, as many of you are familiar, is a receptor inhibitor as opposed to bevacizumab. We tried erlotinib with the combination of ramucirumab in the RELAY study. We were expecting another update for the RELAY study at upcoming meetings, which, unfortunately for now, have been delayed. And again, here we find an improvement in PFS. We’re waiting for the results in terms of discussing the overall survival.

This was a multinational study. And more recently, again, in the US there was an attempt to investigate the combination of erlotinib plus bevacizumab. And now I’ll show you another slide that we have, a rather large effort combining osimertinib plus bevacizumab to investigate the potential clinical efficacy of this sort of approach, in terms of a VEGF inhibitor plus an oral TKI, to see if, in fact, we can improve on the clinical efficacy for this particular group of patients.

This is the study that I was talking about. This is an ECOG study. Again, patients with untreated EGFR-positive non-small cell lung cancer without any prior treatment are randomized to standard dosing of osimertinib versus osimertinib plus bevacizumab. About 300 patients overall. Standard imaging criteria just to see if the overall survival can be impacted. That's a secondary endpoint, as you can see on this slide, and the primary endpoint remains progression-free survival. This study is now active.

What else have we been able to do? We have known that chemotherapy can be very effective for this particular subset of non-small cell lung cancer. We also know that the combination of platinum plus pemetrexed is (A) relatively well tolerated and (B) is highly effective in patients with EGFR mutations. Therefore, there was interest to see if a combination of an oral TKI plus chemotherapy is feasible and can lead to better clinical efficacy. Feasibility was established in smaller studies. But now we have actually 2 very large randomized, Phase III studies, 1 from Japan, 1 from India, that have clearly shown that a combination of a first-generation oral TKI, in this case gefitinib, plus standard platinum doublet chemotherapy, can not only lead to an improvement in PFS, as you can see here, but also, these studies seem to suggest that there is an improvement in overall survival of these patients.

For those of you who might be curious as to why gefitinib, you have to realize that the drug is available in the rest of the world, whereas osimertinib might not be available at this point and everywhere. And it also represents a drug that is easily available, has been around for quite some time and might be a little bit more economic in terms of accessibility to the drug, so it made sense to have a trial where a first-generation drug was used. Obviously we’re interested in learning whether a combination of osimertinib plus chemotherapy can achieve similar a type of result, and I think those are studies that are under planning at this point and I would not be surprised if we hear about them soon in one of our meetings. This appears to be another approach where one can think about improving on the efficacy of some of the available oral TKIs.

For patients that have specific mutations, there might be a combination of combining 2 targeted drugs. This particular study, osimertinib plus selumetinib — selumetinib is a MEK inhibitor — as part of the TATTON study, this combination was investigated in patients who had prior exposure to either first- or second-EGFR targeted therapies or had prior osimertinib. Now, at the time of enrollment on this study, most of the patients who received osimertinib had had first- or second-generation drugs and then had developed T790M mutations and had received targeted therapy.

I realize that there are small numbers enrolled in each category here, but this study shows the feasibility of combining 2 targeted therapies and also suggests that for a specific group of patients, it is possible that a combination like this can, again, improve on the clinical efficacy. Now, here we only have response rate. And the real question is, as PFS and OS data can become available, but these are the kind of data that we’re waiting for, with subsequent analysis of these particular groups of patients.

We’ve talked about MET being a potential resistant mechanism. Again, osimertinib plus savolitinib, a MET inhibitor, was investigated in a similar manner. These waterfall plots show that patients can have reasonable responses. I think the overall response here, if you look at it, it’s about 25%, with another roughly 44% having stable disease, but again suggests that a dual inhibition sort of strategy for patients that have these specific mutations might be a reasonable approach. Obviously we’re hoping that we can expand on this data and, again, improve on our knowledge about the combination here.

Another approach that has been advocated is total EGFR blockade, if you want to think about it like that. This is a study of combining osimertinib with necitumumab — obviously necitumumab is an anti-EGFR monoclonal antibody that was investigated in several Phase III studies in combination with chemotherapy. Here, the investigators decided to have this total EGFR inhibition. As you can see on the waterfall plots, many different subgroups of patients were able actually to have decent responses to this combination. I’ve highlighted the fact that even patients with EGFR exon 20 insertions seem to have responses. Again, the numbers are very, very small, but it does suggest that there are strategies that can be employed to either improve or overcome some resistant mechanisms. The real question here becomes, what are the toxicities? Do we have biomarkers for this patient population? And how do you best select patients for various treatments? And I think some of these are being worked out as part of various clinical trials.

These are some of the key ongoing combinatorial studies that we were able to come up with in combination with osimertinib. Again, trials with chemotherapy, with targeted therapy and a number of other approaches have been entertained and are being investigated. Many of these are in the first-line setting. A few of these studies are being conducted in the second-line setting. And again, we look forward to future meetings where we can hear about the results of some of these trials to, again, decide if there is a best approach going forward.

We said a lot about exon 20 insertions. The reason is shown on this pie chart, that this does represent a good chunk of patients with the EGFR mutation. Depending on the study that you see, I mean, anywhere from 10%, sometimes 15% of patients could actually have an exon 20 insertion. These are usually not responsive to the usual EGFR targeted therapies.

There are a number of drugs that are being investigated. Because of the limitations of time and space, I’ve only selected 1 example, but again, there are many other examples of these drugs. This is an investigation of a drug called TAK-788. This is a complicated schema for a Phase I/II clinical trial that’s ongoing. There are multiple cohorts as part of this study, as you can see, but what they have in common is that patients either have an EGFR exon 20 mutation or HER2 alteration.

And the data that we have so far shows not only fairly decent responses in this patient population but also what I would consider to be very interesting durability of responses. As you can see on this swimmers plot, many of these, as depicted by the arrow at the end, seem to have durable responses that are ongoing, at least at the time of the reporting of this particular set of data that we have with Dr Jannë at ASCO last year.

Some of these patients had CNS metastases — again, does make it a very interesting compound. And more investigation is underway to, again, determine a role and a place for this particular group of drugs.

Again, being a new class of drug, I think it’s important to look at the safety. This is again a safety summary from the same presentation. You notice that diarrhea happens to be the more common side effect. About 18% had at least Grade 3 or higher diarrhea. These drugs, even though they’re effective, even though I think there’s a role for them in a clinical practice, come at the expense of some toxicities that I think we have to learn to manage and be aware of so that patients can have a decent quality of life and also have appropriate treatment for their malignancies.

At least in my view, osimertinib is the preferred first-line treatment in the US for patients with activating EGFR mutations. I think the combinatorial approaches that could potentially delay or address resistant mechanisms are very important, whether it’s front line with the addition of chemotherapy, maybe even with the addition of a VEGF inhibitor, or in the second-line setting in combination with other targeted drugs for the select patients with specific mutations are really important and can address a significant need in this area.

I would still consider local therapy, for instance, with stereotactic radiation for isolated progression while a patient is on osimertinib an acceptable approach and one that I use in my own clinic. And I would emphasize that there is a significant role for repeat biopsies in this particular patient population, because we are able to identify potential resistant mechanisms. And now I hope I have shown you that we have studies that could potentially address these particular resistant mechanisms that can be quite useful down the road in terms of management of our patients.

Selection and sequencing of treatment for patients with metastatic NSCLC with alterations in ALK, ROS1, RET and NTRK

DR JOHNSON: Let’s just jump right into selection and sequencing of treatment for ALK, ROS1, RET and NTRK alterations. I’m Melissa Johnson from Sarah Cannon.

This slide is a summary of the common genomic alterations that we’re used to discussing in lung cancer. Although ALK, RET, ROS1 and NTRK are not as commonly encountered as EGFR and KRAS mutation, shown at the top of the screen, they still encompass thousands of patients in the United States each year when you consider that over 200,000 patients are diagnosed with lung cancer in the US alone.

You see that what ALK, RET, ROS and NTRK have in common is that they occur because of a genetic alteration called a rearrangement. So you can see that illustrated in the bottom right corner of the screen. I explain this to my patients as just 2 pieces of chromosome. Think about wet spaghetti. They get tangles around themselves, and when they get detangled, as the cells replicate and divide, little bits of DNA material have become connected to other bits of DNA that weren’t connected previously. That creates a kinase mutation that’s constitutively turned on connected to a new partner.

This slide shows the landscape of ALK inhibitors that are approved for clinical use in the United States. We’ll march through many of them during the course of this presentation. What I want to point out, though, is that while many of these drugs were initially developed to treat ALK rearrangements, you see that there’s significant homology between ALK, ROS1 and NTRK. And so many of these tyrosine kinase inhibitors also will inhibit ROS1.

You see crizotinib was initially developed as a MET inhibitor. It’s actually a better ALK and ROS1 inhibitor. Down at the bottom of the screen you see that entrectinib was also developed as an ALK inhibitor but is actually a better ROS1 and NTRK inhibitor. Finally, you’ll see that lorlatinib, once again developed to address resistant mutations in ALK, also is being used for patients with crizotinib-resistant ROS1-altered lung cancer.

So let’s begin at the beginning. It’s a good place to start. Of course, we know that crizotinib was the first tyrosine kinase inhibitor to be approved for use in patients with ALK-rearranged lung cancer based on its ability to prolong PFS and overall survival as compared to chemotherapy alone for these patients. Forever after crizotinib then became the comparator arm.

Here you see in the ALEX trial alectinib versus crizotinib for frontline ALK-positive patients. You see that treatment with alectinib prolonged progression-free survival, as well as on the right-hand side, cumulative incidents of CNS progression was significantly lower for patients treated with alectinib as compared to crizotinib. This ALEX trial also showed for the first time that these TKIs, both targeting ALK, can have different side effect profiles. You see very small in the middle of the screen that the most common side effects that we associate with crizotinib: nausea, vomiting, diarrhea, transaminase elevation, much, much, much less common for patients treated with alectinib. Instead weight gain, myalgias, even anemia associated with alectinib more commonly than crizotinib.

Then, we see a second-generation ALK inhibitor, brigatinib, once again compared to crizotinib for patients with newly diagnosed ALK-rearranged lung cancer. In this trial once again we see that progression-free survival was improved for patients treated with brigatinib over crizotinib. This also translated into improved intracranial PFS for patients with brain metastasis at baseline, that you can see in the bottom right corner of the screen, for patients treated with brigatinib versus crizotinib.

Brigatinib is an interesting drug. Once again, with a little bit unique side effect profile compared to crizotinib. Patients treated with brigatinib very rarely can develop pulmonary toxicity. This typically happens within the first few days of starting brigatinib. So patients are started on a lower dose, 90 mg daily, and then after a week, provided no increase in cough, shortness of breath, fevers, the dose is escalated to 180.

So both of these drugs, brigatinib and alectinib, have been approved for frontline use for patients with ALK-rearranged lung cancer.

But we know from our work treating patients with crizotinib, alectinib and brigatinib that all patients with ALK-rearranged lung cancer will developed resistance over a period of months after diagnosis. We know a little bit more about the molecular mechanisms of resistance than we once did. Resistance can develop in tumors based on amplification of ALK, based on mutations in ALK that you see on the left-hand side of the screen. Resistance can also develop if the tumor figures out how to become dependent on a bypass track or off ALK-target pathway.

It turns out that when patients are treated with frontline crizotinib the most common mechanism of ALK resistance is L1196M. It’s a gatekeeper mutation akin to T790M for patients treated with first-line EGFR inhibitors, erlotinib and gefitinib, for EGFR mutations. But interestingly, more common, for patients treated with second-generation ALK inhibitors alectinib and brigatinib, the ALK-resistance mutation that develops is more commonly, 50% of the time, the G1202R solvent front mutation. And so these are important mechanisms to know as we begin to develop new targeted inhibitors that will address resistance to these front-line agents.

Enter lorlatinib. Lorlatinib was a drug developed in particular to address patients’ resistance to ALK-directed acquired resistance. In the initial trial, Phase I/Phase II trial, that led to the accelerated approval of lorlatinib, patients with ROS1 rearrangements were also included.

Here you see some of the waterfall plots summarizing the results from patients with ALK rearrangements. The top left waterfall plot, you can see patients who had been treated previously with crizotinib who then rolled to lorlatinib. For patients in the bottom left-hand waterfall plot you can see patients that were treated with a prior second-generation TKI, i.e. brigatinib or alectinib. In both cases, having really impressive response to lorlatinib in this acquired-resistance setting.

Perhaps most important of all, the patients with 2 or more prior ALK inhibitors. We enrolled many patients here at Sarah Cannon to this cohort to receive lorlatinib as their third-plus ALK-TKI. You can see the response rate’s still very impressive, in the 40% range. And most notable the intracranial objective response, 50%. So this is a drug with much more potent brain penetration, which is a well-known sanctuary site for cancer to develop for patients who have been treated with first- and second-generation inhibitors.

We’ll come back and talk about lorlatinib in the ROS1 section in a few slides, I do want to say a little bit more about ALK resistance because it will be how we ultimately define the optimal sequence for patients treated with serial ALK inhibitors, which we know is possible. On the left you see a table that summarizes, at the far left, the most common both gatekeeper and solvent front mutations that we have come to associate with ALK-rearranged lung cancer. And across the top you see the sensitivity or relative resistance of ALK inhibitors to these particular resistance mutations.

So here you see very nicely that lorlatinib is capable of addressing almost all of these known resistant mutations and in particular, the most recalcitrant, the G1202R resistance mutations. With a nice waterfall plot on the right-hand side showing patients who were shown to have this G1202R mutation in their ALK within the ALK gene driving their cancer that had convincing and durable responses to lorlatinib.

So is the knowledge of the acquired resistance mutation really so important? We talk about the importance of asking patients at the time of acquired resistance to their primary or their secondary ALK-directed TKI to undergo either a tissue biopsy or a liquid biopsy looking for these mutations because we want to assign the very best therapy in the second- or third-line setting. But does it really matter in terms of how long patients will benefit from their next line of therapy? The answer is we don’t know yet.

And this ALK Master Protocol is one of the ways that we’re going to figure out whether progression-free survival, if one is assigned to lorlatinib for a G1202R resistance mutation, for example, is the very best therapy in this case. Exciting trial, it will enroll a large number of ALK-rearranged lung cancer patients, and this is available at many, many centers across the country.

Now I’m going to switch gears a little bit from ALK rearrangements to ROS1 rearrangements. This slide encompasses most of what I intend to talk about ROS1-directed therapies. Although we know quite a lot, once again, because of the similar homology between the ALK and the ROS1 proteins.

Thankfully we have a number of agents with proven efficacy for patients with ROS1 rearrangements, mostly based on the fact that ROS1 and ALK, once again, have significant homology. And so lessons that we’ve learned about ALK rearrangements have also been applicable for patients with ROS1 rearrangements.

Here across the top you see the median PFS for patients who are TKI naïve who are treated with a ROS1 rearrangement. In most cases, about a year and a half, which is pretty remarkable considering that we’re talking about metastatic lung cancer. You see once again, for the most part, these drugs are potent. They cross through the blood-brain barrier, where they achieve intracranial response rate in about 50% of patients treated.

I’ll skip to the safety considerations where you see, although these drugs target ALK and ROS, they have different safety profiles. And so may, in some ways, dictate or help you to select which ROS1 TKI would be best for your patient. Visual impairments, peripheral edema, GI toxicity for crizotinib. Very potent nausea, vomiting, diarrhea, constipation that can be ameliorated with dose reduction for ceritinib. Lorlatinib: peripheral neuropathy, some cognitive adverse events, a little bit different for patients treated with lorlatinib. Entrectinib: weight gain has been described, as well as dizziness.

Finally, efficacy in pretreated patients. This is an important row because it is going to help us define the second and beyond TKIs for patients with ROS1 rearrangements. You see here that yes, lorlatinib has efficacy for patients who are previously treated with ROS1 rearrangement. And so does experimental therapy repotrectinib. This is a star in the making that I think we’ll be talking about in years to come for these meetings.

Now we’re going to move on to RET. RET alterations, once again, happen commonly in lung cancer, 1% to 2%. They also happen for patients with both sporadic and hereditary thyroid cancers in pretty common frequencies. We note that while RET fusions have been known about as lung cancer drivers for several years now, to date there is no specific selective RET inhibitor that has received regulatory approval. In the slides that follow, I’m going to show you 2 stars in the making that we’ll be talking about for sure for years to come.

But first of all, why don’t we have a RET-selective inhibitor yet approved for our patients’ use? And this slide summarizes the real problem with RET inhibitors is that it’s hard to hit just RET. Here you see a number of multikinase-targeted therapies that have been approved for other indications, not for lung cancer per se, but that hit RET. And so you see that vandetanib, cabozantinib, lenvatinib, alectinib all inhibit RET among other kinases. And you see the kinomes that are shown on the right-hand side of the screen show you just how many different kinases these drugs are hitting.

This illustration, to me, shows very nicely why these drugs are difficult to give. They have a wide variety of associated toxicities: hypertension, hand-foot syndrome, mucositis, LFT abnormalities, that also make these drugs hard to give at sufficient doses to prolong disease response.

By contrast, then, we have 2 selective RET inhibitors, shown at the bottom of the screen, LOXO-292 and actually right door BLU-667. Its label got left off. Both of these drugs now have names. LOXO-292 is selpercatinib. BLU-667 is pralsetinib. And these drugs have been formulated to hit RET and RET alone. You see first of all that the ramification of this is a higher response rate, in the 60% to 80% range. And a more durable response, with PFS as long as 18 months.

So in the next 2 slides that follow, then we see LOXO-292. This summarizes the Phase I and II trials that have led to the regulatory package being submitted for selpercatinib. You see that this trial enrolled patients that had RET fusion-positive lung cancer, thyroid cancer and others. Patients could have been treated with prior platinum, or they could have been TKI resistant.

At the bottom, you see the waterfall plots. On the left, that was the pretreated patient population, we treated a lot of patients on this trial at Sarah Cannon, with response rates as high as 70% range. Really exciting and well-tolerated agent. On the right, you see the efficacy of selpercatinib in a treatment naïve patient population, even higher, 80%, 85% of patients having a response.

So based on these results, the company filed a priority review for selpercatinib in early January of 2020. And we are eagerly awaiting a decision about this agent.

Almost in parallel, then, BLU-667 is a second selective RET inhibitor also with activity in RET-rearranged lung cancer. You see a summary on the left of the patients that were included in this development plan, Phase I and Phase II trials, both prior platinum treated, as well as platinum naïve. You see that the response rate for patients treated, prior platinum shown in light blue, and TKI naïve in dark blue, with a response rate of 60%. And at the bottom you see a swimmer’s plot showing once again how durable these responses are, which speaks very nicely to the tolerability of these 2 drugs. Based on these results, a submission to the FDA, also in January of 2020. And we eagerly await the results of this, as well. How exciting to have 2 new therapies on the horizon.

But if you need a RET inhibitor in your clinic today, this is what the NCCN guidelines will recommend that you do: cabozantinib, vandetanib, some of the other multikinase inhibitors. We would expect this to be revised at the time of what we anticipate will be an FDA approval of 2 new agents coming soon.

And now, let’s switch gears 1 more time and talk about NTRK rearrangements. These are rearrangements involving NTRK1, 2 and 3, which are a family of neurotransmitter receptors with the function shown at the right. You see that unlike the other 3 rearrangements that we have discussed that happen most commonly in adenocarcinoma in light or never smokers, NTRK fusions occur irrespective of smoking history, irrespective of age and irrespective of tumor histology.

NTRK rearrangements happen at low frequencies in common tumor types, such as lung cancer. But again, as we discussed at the beginning, there are so many patients with lung cancer that we still will find these in 0.2% of patients, which equals a couple thousand. We do find NTRK rearrangements more commonly in rare tumors such as secretory breast carcinoma, infantile fibrosy — sarcoma, congenital mesoblastic nephroma and salivary cancers such as mammary analog secretory cancers.

We have 2 NTRK inhibitors that have now achieved FDA approval. The table at right shows that even these 2 drugs are a little different in what they target. Entrectinib was developed as an ALK inhibitor, but once again inhibits ROS1 and NTRK more effectively, while larotrectinib is purely a TRK inhibitor. And we’ll see that borne out in the side effects profiles that are slightly different between…

Here’s the Phase I/II experience of larotrectinib, a pooled analysis of several Phase I and II trials. This is a notable group, right? It’s not just lung cancer by any means. It’s multiple, multiple tumor types. The objective response rate 80% in this group of patients, and a median duration of 35% months. It was based on this pooled analysis that larotrectinib became the second tissue agnostic drug to receive FDA approval in late 2018, followed very quickly by entrectinib became the third tissue agnostic FDA approval based on this data. Once again, a pooling of both pediatric and adult clinical trials that led to this drug’s approval in early 2019.

We have data now to show that yes larotrectinib is as active in TRK fusion lung cancer, as it is across the wider spectrum of tumor types. You see with the response rate here, 75%, and durability of many, many months on the waterfall plot, although in small numbers as of yet.

And entrectinib response rate of 70%. Really fantastic results for this TRK inhibitor specific for lung cancer. So while it is easy in our clinical practice to be a little dismissive because these mutations happen so infrequently, it is important to look for them for the patients who will benefit the implications are so very important and significant.

This is my favorite slide in the deck because it shows that NTRK resistance mutations are paralogous to the mutations that mediate resistance in ALK and ROS1. It means that there is a commonality that exists in how these tyrosine kinase inhibitors are developed to inhibit the ATP binding pocket of these kinases. And so both the gatekeeper mutations, as well as the solvent front mutations, tend to occur in the same place in these different proteins. So you see the gatekeeper mutations, we’re finding them in TRK-rearranged patients just like in ALK and ROS, as well as in RET just recently in the Journal of Thoracic Oncology.

So in summary, when choosing a frontline therapy for patients with ALK, ROS1, RET and NTRK rearrangements, right now your choices are pretty straightforward. They’re listed here. You can find them on the NCCN guidelines. It’s important to note that while there’re other agents being developed, the optimal sequence has not yet been defined. And both tumor biopsies and liquid biopsies are helping us understand and one day may help us select therapies.

And finally, resistance is inevitable, but we’re learning from it, both how to predict what drugs to use next, but also how to develop drugs that will silence the resistance one day.