Expert Insights in the Treatment Landscape of Oncogenic Driver-Positive NSCLC

Justin F. Gainor, MD

TARGETED ONCOLOGY: In terms of clinical presentation and prognosis, what are the key differences between patients with and without driver mutations?

Gainor: Over the past 15 years, we’ve come to recognize that advanced lung cancer can really be divided and stratified based on patients who have genetic alterations and oncogenic drivers. The first targetable oncogenic driver that was identified was the epidermal growth factor receptor, which was identified in 2004. We’ve since learned that these genetic alterations are found in about 10% to 15% of Western patients, and these patients tend to have characteristic clinical and pathologic charac- teristics. Most notably, we know that these genetic alterations confer sensitivity to treatment with targeted therapies or tyrosine kinase inhibitors [TKIs].

Since the initial discovery ofEGFRmutations in lung cancer, we’ve come to recognize this as a paradigm of trying to identify oncogenic drivers that could be targeted with TKIs. Collectively, we know that in the United States, there are 4 different genetic alterations where we have FDA-approved targeted therapies. These areEGFR,ALK,ROS1, andBRAF. And, in general, patients with these specific genetic alterations tend to have pretty characteristic clinical and pathologic characteristics. For example, patients withEGFR,ALK, andROS1—these tend to occur in patients who have a never or a very light smoking history. The same can be said for emerging oncogenic drivers such asRETrearrangements, which, again, tend to be enriched for patients with a never or light smoking history. One of the outliers of this, though, isBRAF.BRAFmutations are found in 4% of patients with lung cancer. And in contradistinction to the other targetable alterations,BRAFmutations can also be seen in patients who have a heavy smoking history.

TARGETED ONCOLOGY: Are you seeing adequate early testing in patients referred from the community setting? Is the testing strategy changing?

Gainor: The current standard of care in the United States is to perform genetic testing on newly diagnosed patients with advanced non—small cell lung cancer. We know that among patients withEGFRmutations [and]ALKandROS1rearrangements, that these patients derive substantial benefits to treatment with the corresponding targeted therapies. So, one would expect that, based upon these benefits, testing would be 100%. Unfortunately, that still isn’t the case in the United states, and that could be for a variety of reasons. In speaking to my colleagues, many times patients who come to academic centers don’t have comprehensive genetic profiling and for a variety of reasons. Sometimes it can be because there’s no tissue available. Biopsies or fine needle aspirates were performed as the biopsy procedure, and there’s limited tissue available to perform the corresponding genetic analysis.

Sometimes patients don’t have a disease that’s accessible to a core needle biopsy for a variety of reasons. Nonetheless, I do think it should be our goal, and it should be the standard to perform genetic analyses on all newly diagnosed patients and to do so expeditiously. For patients who don’t have tissue available to perform genetic analyses, I think one can approach this in 2 different ways. The first is to actually repeat a biopsy. I think if one has sufficient suspicion that a patient may have a targetable genetic alteration, that certainly makes sense, especially if a lesion is amenable to a repeat biopsy.

We know that in patients who are never-smokers, more than 50% of those patients are going to have a genetic altera- tion that can dramatically transform how those patients are managed. For patients who don’t have a disease that’s amenable to a repeat biopsy, be it because of comorbidities or the location of their tumors, an emerging tool is actually to perform a liquid biopsy—so, to look at circulating-tumor DNA as a means to identify targetable genetic alterations. It should be emphasized, though, that liquid biopsies are not a substitute for baseline genetic sequencing if you actually have the tumor material available. Liquid biopsies can be a helpful complement, especially if the baseline tissue has been exhausted. The reason I say this is because circulating-tumor DNA assays are not nearly as sensitive as doing genetic sequencing on a baseline tumor biopsy. A negative CT-DNA assay isn’t truly a negative test; it could just be that a patient isn’t shedding. And so, negative results on a CT-DNA assay are really nondiagnostic, and you should keep an open mind that that patient may indeed actually still have a targetable genetic alteration.

TARGETED ONCOLOGY: What are the emerging predictive driver mutations and how has this evolving identification changed the landscape of mutational testing?

Gainor: The success of targetingEGFR,ALK,andROS1has spurred additional efforts to identify additional oncogenic drivers that can be targeted therapeutically. Over the past several years, there have been several notable examples with very promising clinical data that we’ve seen. Important examples of this includeRETrearrangements,METexon 14 skipping, andNTRKrearrangements. So, takingRETrearrangements as an example: RET rearrangements were first identified in lung cancer in late 2012, early 2013. These were found in about 1% to 2% of patients, and they tended to have characteristic clinical and pathologic characteristics. Like other oncogenic rearrangements,RETrearrangements resulted inRETfusions, and these tended to occur in patients with a light- or never-smoking history.

At the time these were first identified, we didn’t really have selectiveRETinhibitors in the clinic, and so investigators relied on multikinase inhibitors that targeted multiple kinases, not justRET. And so early on, the response rates that were seen with these multikinase inhibitors were relatively modest compared with other oncogenic drivers in lung cancer. So, we observed response rates in the 20% to 30% range. More recently, there have been 2 selectiveRETinhibitors that have entered clinical testing, and both have shown high response rates, confirming our initial suspicion thatRETis a bona fide oncogenic driver mutation in non—small cell lung cancer.

Like the experience withRET, another emerging oncogenic driver in lung cancer has beenMETexon 14 skipping mutations. Here, the pathophysiology is quite distinct, so exon 14 encodes the CBL-binding region inMET, which is important in that it targets the protein for ubiquitin-medi- ated proteasomal degradation. In the setting of an absence of exon 14, you get decreased degradation of MET, which results in increased accumulation ofMETon the cell surface. We found thatMETexon 14 skipping is found in about 3% to 5% of patients. And like other oncogenic drivers, it does appear thatMETexon 14 skipping mutations confer sensitivity to treatment withMET-targeted therapies. So, this does appear to be another driver mutation that may be important to test for moving forward.

As the number of targetable genetic alterations increases, it just becomes infeasible to perform single-gene tests one after the other. It really speaks to the need to compre- hensively evaluate the genetic landscape of the tumor and do so while simultaneously using the least amount of tumor tissue available. And to that end, I do think that next-generation sequencing using multiplexed assays is becoming the standard of care—if not is already the standard of care at most academic institutions—which really does speak to the need to rapidly assess the genetic landscape of the tumor and do so while preserving tissue for future clinical trials.

TARGETED ONCOLOGY: What is your treatment approach for patients with a new diagnosis? Can you discuss the trial design, results, and practical implications of the FLAURA trial?

Gainor: The approach to treatment of patients with advancedEGFR-mutant lung cancer has certainly changed over the past several years. Historically, patients would be treated with either a first- or second-generation EGFR inhibitor. We know that the median progression-free survival with that approach was about a year for first-generation EGFR inhibitors. And at the time of development of resistance, about 50% to 60% of patients would have developed a single gatekeeper resistance mutation,EGFRT790M. Among those patients who developed T790M, we’ve recently developed data showing that use of a mutant-selectiveT790M-specific inhibitor, such as osimertinib, could induce responses in the vast majority of patients after progression on a first- or second-generation inhibitor.

Over the last year or so, though, that paradigm of start- ing with a first- or second-generation inhibitor and then using a third-generation inhibitor among those who developedT790M has changed, and it has changed in response to the FLAURA data. So, FLAURA was a randomized phase III study that compared using a first-generation inhibitor, gefitinib or erlotinib, versus osimertinib in newly diagnosedEGFR-mutant patients. In FLAURA, osimertinib led to significant improvements in progression-free survival compared with the first-generation inhibitors. And the median PFS in FLAURA was approximately 19 months for patients receiving osimertinib. Importantly, patients who received osimertinib also had improved control of central nervous system [CNS] metastases. In my clinical experience, I’ve also found that osimertinib tends to be quite well tolerated, with a favorable adverse effect profile. And as a result, I believe that osimertinib should be the standard of care for newly diagnosed first-lineEGFR-mutant patients.

An alternative approach that has been put forward after FLAURA has been that clinicians should still continue us- ing a first- or a second-generation inhibitor first followed by a third-generation inhibitor such as osimertinib. In my mind, that approach is flawed, and the reason that that’s the case is that among patients who receive first- or second- generation inhibitors, just 50% to 60% will actually developT790Mand thus be candidates for receiving osimertinib in the second-line setting. Furthermore, we also know that from various clinical trials that not every single patient makes it to the next therapy. Even when patients are eligible for crossover in clinical trials, the numbers of crossover don’t reach 100%. And so, I feel that we should use our very best therapy first in order to try to maximize efficacy while also maintaining a favorable safety profile.

TARGETED ONCOLOGY: How have the results and implications of the ALEX trial influenced the sequencing of therapies in ALK-rearranged NSCLC?

Gainor: The global ALEX study was a randomized phase III study for patients with newly diagnosedALK-positive lung cancer, and patients were randomized to receive either crizotinib, the standard ALK inhibitor, prior to the ALEX study, or alectinib, a second-generation ALK inhibitor. And in the ALEX study, alectinib produced significant improvements in progression-free survival compared with crizotinib. Indeed, the median progression-free survival with first-line alectinib was extremely impressive. With the latest ASCO [American Society of Clinical Oncology] update, the median progression-free survival was nearly 3 years, which is quite impressive for newly diagnosed pa- tients withALK-positive lung cancer.

Importantly, compared with crizotinib, alectinib also resulted in superior control of CNS metastases, with far fewer cumulative incidence of CNS metastases, compared with patients treated with crizotinib. And so, based upon the results of the ALEX study—which I should also mention confirmed the results of a parallel Japanese study called the J-ALEX study, which had a very comparable design—together, those 2 phase III studies have really confirmed that alectinib should be the first-line standard of care for patients with newly diagnosedALK-rearranged lung cancer.

In July 2018, we had 4 approved ALK inhibitors in the United States: the first-generation inhibitor crizotinib and 3 second-generation ALK inhibitors—ceritinib, alectinib, and brigatinib. The historical paradigm was to start with crizotinib, which induced high response rates, followed by use of a second-generation ALK inhibitor such as ceritinib, alectinib, or brigatinib. The second-generation inhibitors are generally characterized by increased potency compared with crizotinib, increased selectivity, and better CNS penetration. All 3 of those second-generation inhibitors are currently approved in the United States for use after crizotinib.

As I mentioned earlier, alectinib is also approved for first- line use based upon the ALEX study, and ceritinib is also approved for first-line use. Together, though, I would consider that alectinib is really the first-line standard of care, putting together the data from the ALEX study with impressive activity combined with CNS penetrability and a favorable safety profile. Nonetheless, there are a number of patients who are currently still on crizotinib in the United States or receiving crizotinib in other areas of the world. And so, it is very much a question of, if patients have started with crizotinib, what should be the preferred next second-generation ALK inhibitor? There are currently no head-to-head comparisons across the second-generation ALK inhibitors, so we’re generally left with cross-trial comparisons.

The most recent drug that was approved in this space is brigatinib, another second-generation inhibitor that has shown very high response rates and, in the crizotinib-resistant-setting, has shown impressive progression-free survival—around 15 to 16 months in the latest update. And so, based upon that impressive data, brigatinib is now being explored in the first-line space with a head-to-head comparison against crizotinib, and I think we’ll eagerly await that data.

The the question is, what do we do for patients who are pro- gressing on first-line alectinib, which I argued was the new standard of care? The agent that has the most data in that set- ting is still an experimental agent. This is a so-called third- generation ALK inhibitor, lorlatinib, which was a granted breakthrough therapy designation by the FDA. And that was based upon encouraging response rates among patients who were treated with 1, 2, and even 3 prior ALK inhibitors. And so, that’s an agent that has shown activity in the postalectinib setting and, if approved, would be an option in that setting.

TARGETED ONCOLOGY: Can you review the use of combination BRAF- and MEK-targeted agents?

Gainor:BRAFmutations are found in about 4% of patients with non—small cell lung cancer. One important point to draw contrast with melanoma, whereBRAFmutations are found in about 50% of patients, is that not allBRAFmutations in lung cancer are V600E. In fact, only about 50% of lung cancers withBRAFmutations will actually have the canonical V600E mutation. So, that really brings down the frequency to about 2% of patients who will haveBRAFV600E. In general, these patients tend to have clinical characteristics in contrast withEGFR,ALK, andROS1. We do observeBRAFmutations in patients with a history of smoking. But we also know thatBRAFmutations do confer sensitivity to treatment with targeted therapies—specifically, BRAF inhibitors, as well as MEK inhibitors.

We have learned early on from our colleagues treating melanoma, thatBRAFinhibitors can induce responses to tumors harboring V600E mutations—most notably in mela- noma and non—small cell lung cancer. Actually, the combination of BRAF plus MEK inhibition can result in improved activity and tolerability. We know from single-arm studies that the combination of BRAF plus MEK [inhibition] leads to high response rates in the 60% range and favorable progression- free survivals.

We’ve seen in non—small cell lung cancer, as well as in melanoma, that the combination of BRAF plus MEK inhibition results in better activity, as well as better tolerability. And this results from blocking downstream of the mutantBRAFwith a MEK inhibitor. So, in the United States, the combination of dabrafenib and trametinib is currently approved for patients withBRAFV600E non—small cell lung cancer. I should note that this is different from the experience in colon cancer. I thinkBRAFis a helpful lesson for targeted therapies more broadly because we know that the same genetic alteration can behave very differently depending upon the tumor in which it occurs in. So,BRAFV600E in melanoma and in non—small cell lung cancer is very sensitive to treatment with BRAF plus MEK. By contrast, the same mutation in colorectal cancer is far less sensitive to targeted therapies. And so, I do think that it’s an important reminder that we need to actually do clinical trials of genetic alterations across different diseases, because they may behave quite differently.

For non-V600E mutations, currently, the standard of care would be the default: chemotherapy, immunotherapy, or a combination of the 2. There are actually many different types of non-V600E mutations. They generally fall into 3 different classes, and the approaches to each of those classes are different. But right now, there’s no currently FDA-approved therapy for non-V600E mutations. And so, if I were to encounter one of those patients, I would actually direct them to a clinical trial, but the standard of care to start would be as if I were approaching that type of patient.

TARGETED ONCOLOGY: What is the evolving role of immuno-oncology (I-O) therapy for patients with newly diagnosed NSCLC?

Gainor: Over the past 6 months, we’ve had some really exciting data emerge on the role of chemotherapy-plus- immunotherapy combinations. And we’ve seen now 3 randomized controlled studies showing overall survival improvements with the addition of PD-1 or PD-L1 inhibition to standard chemotherapy in non—small cell lung cancer. I think the most important of these studies was KEYNOTE-189. This was a randomized phase III study that compared platinum/pemetrexed versus platinum/ pemetrexed plus pembrolizumab. And in the study, the triplet combination resulted in improvement in both progression-free survival as well as overall survival. And an important part of that study is that with respect to overall survival, there was an improvement in every strata of PD-L1 expression, so greater than 50%, 1% to 49%, as well as in less than 1%.

We’ve recently had data from the IMpower150 study. This also focused on nonsquamous patients, and here it is a bit of a more complicated study in that there are 3 different arms to the study: arms A, B, and C. Control group was arm C. This was carboplatin/paclitaxel/bevaci- zumab, and this was compared against either carboplatin/ Taxol, or carboplatin/paclitaxel, and bevacizumab plus atezolizumab or carboplatin/paclitaxel/atezolizumab, so removing the bevacizumab from that arm. At both AACR and ASCO, we saw data from this study, and it did look like the quadruplet regimen of carboplatin/paclitaxel/ bevacizumab/atezolizumab resulted in improvements in both progression-free survival as well as overall survival compared [with] the control arm of carboplatin/paclitaxel/ bevacizumab. So, overall, we have another positive study for patients with nonsquamous histology.

TARGETED ONCOLOGY: Can you describe the PFS of the patients who hadEGFR-mutated orALK-rearranged NSCLC? What is the potential explanation for this finding? For which patients would this regimen be most beneficial?

Gainor: One key distinction between KEYNOTE-189 and IMpower150 is that KEYNOTE-189 excluded patients with sensitizingEGFRmutations andALKrearrangements. By contrast, those patients were allowed into IMpower150 provided they received the relevant targeted therapy. It was a bit surprising to all of us to see that the benefits to adding atezolizumab to carboplatin/paclitaxel/bevacizumab even extended to theEGFR- andALK-positive subgroups compared with the control arm. The reason this was surprising is because in earlier studies of just using single-agent checkpoint inhibitors, the benefits amongEGFR- andALK- positive patients were quite minimal with just single-agent checkpoint inhibitors. We saw response rates amongEGFRpatients were 10% or less, and in meta-analyses, there was no improvement in survival compared [with] single-agent docetaxel. And amongALK-positive patients, really the re- sponse rates have been 0 in the limited studies to date.

And so, in IMpower150, we saw that now combining chemotherapy withVEGFinhibition and PD-L1 inhibition, it did look like there was an improvement in theEGFR-andALK-positive patients. So, this really represented the first study where we have an improvement with checkpoint inhibition in those molecular subgroups. And I think that’s one of the key take-homes from the IMpower150 data.

What [makes] IMpower150 different [from] other studies of single-agent checkpoint inhibitors amongEGFR- andALK-positive patients? One potential explanation is that chemotherapy may have led to tumor antigen release, and this may have spurred or resulted in increased sensitivity to the checkpoint inhibition. An alternative hypothesis is that there may have been an interplay between the VEGF inhibition with bevacizumab and PD-L1 inhibition with atezolizumab. That interplay may have been particularly important in patients withEGFR- andALK-positive lung cancer. Still, a third hypothesis is that there may have been something unusual about this particular subgroup of patients.

In total, there were a little over 100 patients withEGFR- orALK-positive non—small cell lung cancer who were treated as part of the IMpower150 study, but there were several things that were a bit unusual about that group. In particular, there were about 6 patients who had bothEGFRmutations andALKrearrangements. While that’s been reported, it’s typically quite rare. It did seem like a disproportionate number of patients in that study had both, which calls into question the molecular testing that was performed and the quality of testing. [This] raises the question [whether] all of these patients [were] trulyEGFRmutation-positive orALK-positive. Perhaps that may have explained some of the benefit.

TARGETED ONCOLOGY: Should patients with oncogenic—driver positive NSCLC be considered for immunotherapy as first-line therapy? Combination I-O therapy? Could immunotherapy be used after progression in these patients?

Gainor: Over the past 15 years, there have been 2 major new paradigms that have transformed the management of non— small cell lung cancer: targeted therapy and immunotherapy. I can say that today, they still have very important roles in the management of our patients. Each has been guided by a different biomarker, and the importance of those biomarkers, I would say, are still very important today. So, patients with newly diagnosed advanced non–small cell lung cancer should still have comprehensive genetic profiling as well as PD-L1 expression scores. I think those are just baseline what we should expect to see in all of our patients.

The question, though, is, what are the roles of immunotherapies in certain genetic subsets of patients with non—small cell lung cancer? Early on, there was enthusi- asm for exploring the role of immunotherapies in patients with defined oncogenic drivers. We recognized that the activity of immunotherapy as a single agent really pales in comparison with targeted therapies in patients withEGFRmutations andALKrearrangements.

From a retrospective single-institution series performed at my own institution, we’ve seen thatEGFR- andALK-positive patients have about a 3% response rate to single-agent checkpoint inhibitors. More recent meta-analyses have looked at the randomized studies, generally in the second-line setting comparing single-agent checkpoint inhibitors versus docetaxel, and specifically amongEGFR-mutant patients, really seeing that single-agent checkpoint inhibitors did not result in an improvement in survival compared with docetaxel.

In addition to that, we’ve now gotten some prospective data, and I think the best prospective data with single-agent checkpoint inhibitors came from the ATLANTIC study. And this was evaluating durvalumab in specific subgroups, and that included patients withEGFRmutations andALKrearrangements. Notably, even when that study tried to enrich for responders by focusing on PD-L1 expression—in here, it was 25% or greater—the response rate among EGFR patients with that degree of PD-L1 expression was only 12%. And there were no responses observed in the ALK subgroup, de- spite having high PD-L1 expression. And I think this speaks to a point that PD-L1 expression is a less reliable predictive biomarker in patients withEGFR- andALK-positive disease. This is for several reasons, [the] first [being] that these tend to be low mutation tumors.EGFRandALKtend to occur in never- or light-smokers, and so we would expect them to have a low TMB.

The second [reason] is [that] work from our institution shows that there are multiple different ways to induce PD- L1 expression on the surface of tumor cells. Classically, this would be from infiltrating immune cells, leading to interferon gamma secretion and driving PD-L1 expression on the cell surface. We think that [this] is likely to be the most immunogenic, in that you have preexisting immune cells there together with PD-L1 expression. By contrast, oncogenic signaling [itself] viaEGFRorALKcan lead to, can result in PD-L1 expression independent [of] the presence of immune cells. The responsiveness to PD-1 inhibition in that setting is likely to be very different because, again, it’s not presupposing that there are T cells in the local microenvironment.

I think putting this all together speaks for patients withEGFRandALK. We should really be prioritizing targeted therapies for those patients and really maximizing targeted therapies wherever possible for [them]. Unfortunately, we know that patients will eventually develop resistance to [targeted] therapy and, at some point, need to move on to other forms of therapy. I think the IMpower150 data tells us that one path forward may be exploring chemotherapy plus PD-1 combinations in those patients instead of just single-agent checkpoint inhibitors, which I think right now are not the best [way] forward for those particular patients.

Moving forward, I think we need to also look at other oncogenic drivers and how they respond to immune checkpoint inhibitors. We have to keep in mind that smoking status has been a pretty good surrogate. Patients with oncogenic drivers that are classically defined with those who have minimal smoking history, are probably unlikely to derive significant benefit from single-agent checkpoint inhibitor. A final point is that, even within a specific driver subgroup, there are likely to be co-occurring genetic mutations that may also affect sensitivity to checkpoint inhibitors. Here, the best example would be KRAS. We know thatKRASis the most common oncogenic driver mutation in non—small cell lung cancer. But there are important co-occurring mutations withKRAS, and I think the most important co-occurring mutations are tumor protein 53 [TP53], liver kinase B1 [LKB1], and kelch-like ECH-associated protein 1 [KEAP1].

Recently, we participated in a Stand Up to Cancer Lung Cancer Dream Team project and demonstrated that the presence of an LKB1 mutation withKRASresulted in insensitivity to treatment with checkpoint inhibitors. Response rates to single-agent checkpoint inhibitors amongKRASplus LKB1 were quite low, on the order of 10%. By contrast, amongKRASplus TP53, response rates were much higher in the 40% range. [Therefore], I think the primary driver by itself has to also be taken into account with the co-occurring mutations, andKRASis one such example. I think we’re just right now starting to scratch the surface of that.

TARGETED ONCOLOGY: Is there a potential role for sequencing or combining targeted therapy and immunotherapy in NSCLC with actionable mutations?

Gainor: The hope had always been to combine targeted therapies—taking an advantage of the high response rates of targeted therapies—with immunotherapies, with the hopes for durability of responses with a way to get both high re- sponse rates and durable responses. Unfortunately, the approaches of combining targeted therapies with immunotherapies have been fraught with difficulty in non—small cell lung cancer. And the principal challenge has been toxicity, and it hasn’t been the same toxicity. It’s actually varied depending on which combinations have been explored. So, for example, combining an EGFR inhibitor such as osimertinib with checkpoint blockade, in this case durvalumab in the TATTON trial, resulted in very high rates of pneumonitis, prompting discontinuation of that approach.

Similarly, the combination of crizotinib plus nivolumab for patients withALK