Liu Lends Insight on Acquired Resistance in Oncogene-Driven NSCLC

Stephen Liu, MD, associate professor of oncology and urology at Johns Hopkins Medicine

Stephen Liu, MD

The development of next-generation tyrosine kinase inhibitors (TKIs) has led to deep and durable responses in patients with EGFR-mutant non—small cell lung cancer, but these agents have added to the complexity of acquired resistance mechanisms, said Stephen Liu, MD.

“[We thought] that with better drugs, maybe resistance for EGFR would get simpler, but it is going to get more complex down the line,” Liu said. “One thing we know for sure is that it is going to get better.”

Following the results of the phase III FLAURA trial, the FDA approved osimertinib (Tagrisso) for the frontline treatment of patients with NSCLC who harbor exon 19 deletions or exon 21 L858R substitution mutations in EGFR. The FLAURA data demonstrated a 54% reduction in the risk of progression or death with frontline osimertinib versus first-generation TKIs erlotinib (Tarceva) or gefitinib (Iressa). Median progression-free survival was 18.9 months (95% CI, 15.2-21.4) and 10.2 months (95% CI, 9.6-11.1) with osimertinib and standard therapy, respectively (HR, 0.46; 95% CI, 0.37-0.57; P <.0001).

Osimertinib was initially approved by the FDA in November 2015 for the treatment of patients with advanced EGFR T790M-positive disease following progression on a prior EGFR TKI.

However, the decision of whether to begin with a first-generation TKI and reserve osimertinib for the second-line setting or vice versa is unclear. How resistance manifests is dependent on the sequence of therapy, and an increased understanding may enable physicians to predict the path of resistance from the beginning of a disease course, explained Liu, an associate professor of Medicine, Division of Hematology and Oncology, Georgetown University Medical Center.

In ALK-rearranged NSCLC, Liu added, there are also resistance mechanisms and questions surrounding the optimal sequence following the frontline approval of alectinib (Alecensa) in November 2017.

OncLive: What are the updates in EGFR-mutant and ALK-rearranged disease beyond frontline treatment?

In an interview during the 2018 OncLive^® State of the Science Summit™ on Advanced Non—Small Cell Lung Cancer, Liu provided an enhanced understanding of oncogenic drivers such as EGFR and ALK in patients with NSCLC, sequencing strategies, and the potential application of immunotherapy in later lines of therapy.Liu: We're familiar with EGFR and ALK as actionable subsets within NSCLC. We've become much more comfortable with using targeted therapies in those settings. This year, we saw the publication of 2 large randomized phase III trials that showed the newer-generation drugs are more effective than the older ones, which gives us preferred and more effective treatments. What we are struggling with now is resistance and what to do when those drugs stop working.

In the EGFR setting, targeted agents have revolutionized care. These are responses that can be rapid, they can be deep, and they can be durable. This is a setting in lung cancer where we're not really hoping for response, we are expecting it. We're also expecting and anticipating resistance. Unfortunately, it remains something that we're going towards.

When we look at treatment for EGFR-mutant lung cancer with first- and second-generation drugs, resistance was surprisingly straightforward. The majority of patients would develop a point mutation in exon 20 at T790M. Those patients would then move on to osimertinib, which retains its activity in that setting. Following osimertinib, it's become extremely complex. Some of the recent work has pointed out just how complex and heterogeneous resistance is.

In a recent study looking at patients who had progressed on second-line osimertinib, we saw that those patients who all had T790M starting on osimertinib can lose T790M. About one-third of patients retained T790M at the time of progression, and about two-thirds lost it. In looking at those subsets, we can see that the biology is fundamentally distinct. Those who retain T790M seem to develop another acquired mutation. Approximately half of those patients will develop C797S. That acquired mutation prevents osimertinib from binding from the ATP cleft. [C797S] acts in a very similar way to T790M in the [face of the] first- and second-generation drugs.

However, in tumors that lose T790M, we see a very broad range of different resistance mechanisms. Some of the mechanisms we're familiar with, such as MET amplification and PI3K mutations. We see a lot of surprising characters [too]. These are familiar, but are in unexpected places, such as KRAS mutations, fusions in RET, BRAF, FGFR3, and some other series fusions in ALK, NTRK, and ROS1. These catch our attention because we have drugs that have some activity against these targets. Can we use [them] in those settings? It is very intriguing but work still needs to be done.

What's also different is the time it takes to develop that resistance. In tumors that retain T790, the time to treatment discontinuation exceeded 1 year. Patients who lost T790M came off osimertinib much quicker, suggesting that those competing mechanisms of resistance—those clones that had those other alternations&mdash;were already present in those cases where T790M was detected.

How does your choice of first-line therapy impact second-line treatment?

Osimertinib is targeting those sensitive ones, and you have another resistant clone ready to assume the ranks. That would also suggest that if you detected those early on, that you could tailor your frontline treatment to eliminate those clones.If you start with a first- or second-generation drug, about 60% of patients will develop T790M as the mechanism of resistance. This specific point mutation alters the binding of those drugs, changes the configuration, and provides more steric hindrance. The first-generation drugs don't outcompete ATP for binding, whereas osimertinib—which has a different binding site&mdash;will retain its activity. If you're starting with osimertinib, you wouldn't expect T790M to develop as a mechanism of resistance. The C797S mutation affects the binding site of osimertinib. It covalently binds to the ATP cleft that you would expect as a mechanism of resistance for osimertinib.

If you start with osimertinib, you may expect C797S alone; whereas, if you start with a first-generation drug and then move on to osimertinib, you may have a combination of T790M and C797S. That has some important implications because C797S prevents osimertinib from binding but has no implications on first-generation drugs.

Is there a way to predict what combinations to use?

Though KEYNOTE-189 excluded patients with EGFR mutations and ALK rearrangements, does immunotherapy have potential in patients with oncogenic drivers?

Agents such as erlotinib and gefitinib would, in theory, still retain activity. If you have a compound mutation of T790M with C797S, it depends on the configuration. If those are in separate alleles in the so-called transfiguration, perhaps a combination of a first- and third-generation drug would be effective. If it is on the same allele, unfortunately, those combinations would be ineffective. C797S is difficult to overcome. Right now, those patients would be treated with chemotherapy.What is tantalizing is the suggestion that all factors that lead to resistance are present at the very beginning. As our techniques get more sensitive to detect small changes, will mutations in Rb and p53 eventually predict small cell transformation? Will we be able to detect rare RET or BRAF fusions at the time of diagnosis, and eventually predict how those tumors will succeed in growing beyond osimertinib? Those clues may be there for us. Some of them may take time to emerge, but some may be present at the beginning. We're still trying to piece together how to pick out and use that information.In KEYNOTE-189, we saw a combination of carboplatin, pemetrexed, and pembrolizumab (Keytruda) outperform chemotherapy alone. This was independent of PD-L1 expression, which set a new standard for NSCLC of nonsquamous histology. What is important to remember is that those studies excluded patients with EGFR mutations and ALK rearrangements. While it has changed our standard of care for most patients with NSCLC, targeted therapy remains the preferred initial treatment for patients with a driver mutation and the preferred salvage treatment as well.

What role do you envision for immunotherapy in patients with EGFR and ALK rearrangements?

Immunotherapy has changed our management of NSCLC, but the outcomes have been very disappointing in EGFR and ALK rearrangements. In the second-line setting, response rates are under 5%. In the frontline setting, we saw no responses in a recent study looking at pembrolizumab for EGFR-mutant patients with PD-L1 expression. A concern there is that subsequent use of TKIs, which are normally relatively safe drugs, may be influenced by that prior treatment. The sequence may have implications for the duration of that patient's treatment. Immunotherapy can still play a role for those patients. We've seen that some of the responders in the initial studies of pembrolizumab were patients with EGFR mutations. In the original phase I salvage study with nivolumab (Opdivo), one of the long-term survivors was an EGFR-mutant patient. Responses are durable and possible, they're just not likely. Those patients are better served by using a TKI upfront.

Is sequencing equally as complex in ALK-positive lung cancer?

In the recent IMpower150 trial, patients with EGFR and ALK alterations were included. There was some suggestion that those patients may benefit from that [therapy]. That may be a combination of PD-L1 and VEGF inhibitors. Immunotherapy is not off the table. In fact, it holds a lot of promise. Today, targeted therapy is the preferred treatment. The proper sequence as of now should be targeted therapy first and reserve immunotherapy not just for the second-line setting, but beyond it.In ALK-positive disease, we started with crizotinib (Xalkori) and we now have 3 approved drugs after that drug. Moving alectinib into the frontline setting resets everything. Everything we learned about resistance we have to redo because the resistance patterns and mechanisms were very different [with crizotinib].

There is some early mostly retrospective data looking at modest but greater than 0 response rates to brigatinib (Alunbrig) and ceritinib (Zykadia) following alectinib. Lorlatinib probably holds the most promise. In the early studies we saw response rates upwards of 40% for patients who had received 2 or more ALK TKIs.

What ongoing trials are you excited about?

As new agents develop, have continued central nervous system penetration, and maintain their activity in the face of resistance, we will have more empiric therapies to use in the salvage setting. We have to believe that in the very near future, we will be using mutation profiling to rationally select which drugs to use in the salvage setting.In September, we'll see the results from the first-line brigatinib study. It’s very similar to the ALEX trial, but the study compared brigatinib with crizotinib as the control. We have seen that those results were very promising by the press release, but we want to see the data. It's not a head-to-head comparison against alectinib, so we'll need to sort of use our own judgement as to which would be the preferred treatment in that setting. Certainly, having more options and more active drugs in that setting is only a good thing.

Ramalingam S, Reungwetwattana T, Chewaskulyong B, et al. Osimertinib vs standard of care (SoC) EGFR-TKI as first-line therapy in patients (pts) with EGFRm advanced NSCLC: FLAURA. In: Proceedings from the 2017 ESMO Congress; September 9-12, 2017; Madrid, Spain. Abstract LBA2_PR.