 It’s important to remember that we built a paradigm in molecular oncology to look for mechanisms of acquired resistance, to learn from that and try to guide the next line of treatment based on that understanding. I can’t say that that’s fully established for 

, but I don’t see any reason why we wouldn’t want to continue to learn. It’s good to retest the patient for the development of secondary mutations. We’ve learned there are some very specific 

 mutations that can potentially match patients for potentially active second-line therapy, such as lorlatinib, which is a highly potent next-generation drug for the management of both 

 translocation-positive patients. I tend to use the same paradigm that we use for 

-positive patients in terms of ctDNA [circulating tumor DNA] with tissue-based retesting and learn from that experience. Fortunately, we have second-line agents to possibly be used. There are others under development, such as repotrectinib, which is a very potent and highly promising agent as well.

 In terms of mechanisms of resistance to ROS1 inhibition, there have been a lot of exciting developments. As we do for many oncogene-driven lung cancers, I really look at mechanisms of resistance in 2 categories. One is the on-target resistance mechanisms, for example, within 

 that eventually cause changes that prevent the drug from binding. Then there are bypass tracks and off-target resistance mechanisms that signal around 

, bypassing track signals that allow the tumor to grow despite the inhibition of the target with the drug.

There are a number of secondary on-target resistance mutations and bypass tracks. In terms of on-target mutations, there are mutations such as G2032R, in which response to lorlatinib occurred about over 20% of the time. There are solvent front mutations as well. Some of these mutations looking at structural modeling that causes interference with binding of these ALK and ROS1 TKIs [tyrosine kinase inhibitors]; some of these mutations—lorlatinib, crizotinib, and entrectinib—may be more sensitive to certain ROS1 inhibitors than others.

 lung cancer, there is some type of a heat map you could use, where you can look at what you would expect the relative inhibitory concentrations to be and then potentially pick a ROS1 inhibitor that may have efficacy. A tougher one to target is G2032R. With that, there is a drug called repotrectinib that’s now being looked at. You could look to potentially match those up in clinical trials. There are also bypass tracks, for example, as a resistance mechanism to lorlatinib; MET amplification has been noted as well.

-resistance mechanism, and that’s a major bypass track that signals around the drug. I divide it into these on-target mutations, which may differ a bit among crizotinib and then lorlatinib and entrectinib, and off-target mutations. Plus, molecular aberrations such as MET amplification, where going back to crizotinib—crizotinib is also a MET inhibitor—may have some additional activities. We’ve certainly seen that in 

-rearranged non–small cell lung cancer. That could apply to 

Videos

Future Directions for ROS1 rearranged mNSCLC

 Entrectinib approval was certainly welcome because it gives our patients another option. Entrectinib approval was based on the combined analysis of 3 trials, which are called ALKA-372-001, STARTRK-1, and STARTRK-2. It is a very effective drug, and a response rate of 78% is excellent. In addition, entrectinib has confirmed CNS (central nervous system) efficacy, which is important for our patients with oncogenic-driven cancers.

-positive non–small cell lung cancer that's metastatic, there's been the development of several exciting therapies. As mentioned, crizotinib has been around for several years and does have good activity. Entrectinib has recently been approved as well. The approval was based on a pooled analysis of 3 studies with entrectinib, as it's also an NTRK and ALK inhibitor and was studied in a cross section of these patients.

 in the ALKA-372-001, STARTRK-1 and STARTRK-2 studies, overall response rates were over 70%, and median duration of response came out to be 24.6 months. The major advantage of entrectinib is that it does have additional CNS activity and penetration. There are potential adverse effects: weight gain; neutropenia; some nervous system disorders, such as paresthesias; GI (gastrointestinal) upset with nausea; myalgia; dysgeusia; and attention disturbance.

Overall, it's an excellent option for patients with 

 rearranged non–small cell lung cancer. In terms of the pharmacodynamics and pharmacokinetics of entrectinib compared with crizotinib and ceritinib, I think the big advantage is CNS activity, and we've seen that. We've seen CNS activity play a role, for example, with alectinib and brigatinib versus crizotinib in 

 rearranged non–small cell lung cancer, as well as osimertinib compared with erlotinib and gefitinib in 

 mutated lung cancer, where those have become first-line treatments.

One could look at entrectinib in a similar way. It is a selected ATP (adenosine triphosphate)-competitive inhibitor of TRK

ROS1, and ALK and does have great potency in terms of its pharmacokinetics and pharmacodynamics, with maximal concentration reaching 2 to 4 hours. Half-life is about 21 hours in the fed state, so it can be taken with food. It is a good option for these patients.

 Coming now to entrectinib, a recently developed and very potent NTRK and ROS1 inhibitor, we’ve seen great success with this agent in both 

-fusion positive cancers, including a fair number of lung cancers. This is an agent that’s also very selective. It doesn’t have a lot of off-target adverse effects. It’s a very safe molecule. It’s also designed to be CNS-penetrant, which is very important in the treatment of these patients.

Non–small cell lung cancer has a great predilection for CNS metastases, in particular 

-positive cancers have a great likelihood of ultimately metastasizing to the CNS. Having targeting agents with CNS penetrants is critical, and entrectinib is a molecule that fits the profile and has shown great efficacy in the combined analysis of the phase 1 and phase 2 studies, showing a PFS (progression-free survival) of 19 to 20 months. It’s shown a great benefit in the overall group of patients,specifically patients with CNS disease, with a response rate surpassing 50% in the CNS and a substantial duration of response, maybe a year or so in the CNS as well. This is a highly promising agent for the management of 

 translocation-positive non–small cell lung cancers.

I think we’re fortunate to be able to witness the treatment of patients with this targeted agent. 

-positive cancers stand out as having great benefits from 

-targeting drugs. Crizotinib is sort of the historical molecule that we’ve used for these patients with great success. In fact, some of the longest responders among advanced lung cancers tend to be patients with 

translocation-positive disease. Entrectinib seems to have a profile analogous to crizotinib in terms of overall benefit and has the advantage of the CNS penetration. That’s significant for this subset of patients, so it’s a favorable asset to entrectinib and offers a great alternative for our patients with 

fusion-positive non–small cell lung cancer.

 The overall response rate to entrectinib in the entrectinib trials was 78%. Responses were durable, with median duration of response of almost 25 months. It's important that responses were seen in patients with and without brain metastases. The median progression-free survival was 19 months. We also saw that entrectinib appears to have a very good CNS penetration, with intracranial overall response rate of 55%. Entrectinib is well tolerated.

Even though nearly all the patients in that trial experienced adverse events, the majority of them were grade 1 or 2. There are certain adverse events that are expected for NTRK inhibitors, and those are called on-target treatment-emergent adverse events, which we think are due to inhibition of TRK. Those are dizziness, sometimes perioral numbness. Interestingly enough, TRK TKIs (tyrosine kinase inhibitors) also cause weight gain in our patients due to inhibition of the TRK pathway.

Despite the fact some of the patients had adverse events, discontinuation was very rare in that trial. Only 3% of the patients discontinued the drug due to adverse events. About one-third of the patients required dose interruption, and 20% required dose reduction. Evidence of serious adverse events was also very low, about 9%. Overall, entrectinib is well tolerated by our patients.

I have experience with entrectinib, and I have given it for both 

 rearranged cancers. My experience is very similar to what has been described. Dizziness is something we commonly see, as well as perioral numbness. Another interesting adverse effect that we see with NTRK inhibitors is if you hold the drug for one reason or another, your patient will experience pain. It's important to be aware of withdrawal symptoms and manage that pain with any kind of pain medication.

 Intracranial response is important, particularly in patients with CNS disease or in oncogene-driven lung cancers that have a propensity to develop CNS disease. It can be a sanctuary site for recurrence as well. It's a very good option for patients with 

 rearranged non–small cell lung cancer, particularly regarding the CNS activity as a potential sanctuary site.

Like I said with alectinib and brigatinib in 

 rearranged lung cancer, and osimertinib in 

 mutated lung cancer, the longer you can forestall metastases to the brain, the better off you are. The other thing is for small asymptomatic brain metastases: If you can forestall radiation to the brain, these are patients who will sometimes live 4 to 6 years or longer. If you do brain radiation upfront, particularly whole-brain radiation, you're risking some serious neurocognitive toxicities down the road. If you give a TKI that has CNS activity, you'd be able to forestall that and potentially save the patient from some life-impactful toxicities down the road.

 Entrectinib is a molecule, a drug that’s easy for patients to take. It has a simple administration schedule, 600 mg once daily, if I remember correctly. It can be taken with or without food. The adverse effect profile is very favorable. Whenever we look at studies of these targeted compounds, it’s nice to see the number of patients who needed to dose reduce due to adverse effects. It’s also important to look at the number of patients who had to stop the drug due to adverse events. Those percentages are very low with entrectinib. It’s a very well-tolerated compound.

It’s important to remember that it also blocks the NTRK pathway, and that’s an important pathway in the CNS. That explains some of the unique adverse effects of the molecule, as well as any molecule blocking the TRK pathway. There can be CNS adverse effects, such as dizziness, and neuromuscular adverse effects, such as neuropathies, polyphagia, and potentially weight gain. These patients have a very good appetite, which is nice on a certain level, but it can sometimes yield to significant weight gains.

ALKA-372-001, STARTRK-1, and STARTRK-2 Trials

-mutated translocated lung cancers is very important. It’s a small percentage of our patients, maybe 0.5% to 1%. This particular subset derives tremendous benefit from targeted therapy. We have different generations of agents to offer, so identifying these patients is absolutely critical. To integrate 

testing into your platform is an absolute must and cannot be forgotten.

What is the proper test? Well, there are different tests we can perform. The most commonly used platforms are either FISH (fluorescence 

 hybridization), looking for very specific translocations, or next-generation sequencing (NGS) technology. The latter allows us to do more…testing, allowing complete molecular testing. IHC (immunohistochemistry) testing is also under development. It’s not as good for 

, we’re not so confident yet. Usually, it’s FISH or NGS testing that we want to integrate into the molecular testing paradigms, with NGS providing a one-stop shop for molecular testing with 1 single test integrating all the molecular needs.

You can have translocations that do not yield a functional protein, but that’s unusual. When you see a 

translocation, you have to think this might be an actionable alteration, and the largest majority of those are. Testing companies will do their best to make sure that the information is relayed in an appropriate fashion in regard to the reliability of detection. Was the translocation detected in both directions? Most of the time, when we see a 

 fusion detected, that’s an actionable abnormality. It’s most commonly seen in lung cancers but can be seen rarely in other cancers and cholangiocarcinomas, as well as certain CNS (central nervous system) malignancies.

, there are certain subsets of patients where these abnormalities are enriched, usually younger patients, nonsmokers. With 

 mutations, there’s a gender predilection for females. That’s not the case for 

. The gender balance seems to be about 50/50. At the same time, it tends to be younger patients with adenocarcinoma histology. That being said, there’s always a certain percentage of patients who are smokers, who are elderly, sometimes with nonadenocarcinoma histology. The best is not to use the phenotype to drive testing, but to make sure that all of our patients with advanced non–small cell lung cancer get appropriate testing at the end of the day.

 non–small cell lung cancer, my first-line treatment is an ROS1 TKI (tyrosine kinase inhibitor). That could either be crizotinib, which has a median progression-free survival of over 19 months and a response rate of about 70% in the pivotal trial, or entrectinib, which was recently approved for 

 non–small cell lung cancer and has additional CNS activity. That's a consideration, particularly in patients with

 rearranged non–small cell lung cancer who have CNS disease. Its duration of response was in excess of 20 months in the pooled clinical trials with entrectinib, looking specifically at 

 rearranged non–small cell lung cancer. Those are the preferred agents. Ceritinib has also been looked at and has some activity, but it’s unclear if it's any better than crizotinib or entrectinib. I don't think it's been studied as extensively in 

Lorlatinib in subsequent lines of therapy appear to have some activity, and there are clinical trials with additional agents, such as repotrectinib, that are particularly of interest. I would also like to mention that beyond ROS1 TKIs, pemetrexed-based therapy, like in 

 non–small-cell lung cancer, does appear preferentially sensitive to pemetrexed-based therapy in terms of chemotherapy combinations in the future. My initial approach is a ROS1 TKI first, based upon the extraordinary PFS (progression-free survival) benefit observed with crizotinib and entrectinib.

 As I mentioned, it is best to hold your treatment until you know that your patient has 

 rearranged non–small cell lung cancer. There are several agents available. The data behind efficacy of immunotherapy in that population are quite limited. The only thing we have is a retrospective registry, which is called IMMUNOTARGET. In that registry, patients with 

 rearranged lung cancer were given immunotherapy, and the response rate was only 16%.

Eighty-three percent of the patients with 

 rearranged lung cancer progressed on immunotherapy. We also know that PD-L1 expression in patients with oncogenic-driven tumors can be falsely elevated and does not necessarily predict efficacy of immunotherapy in that population. That's why I would wait for the molecular testing results. I would like to treat patients with 

 rearranged lung cancer with oncogenic-driven therapies and use target therapy instead of immunotherapy.

Current Management of ROS1 rearranged NSCLC

 There are a couple of things to keep in mind when looking at the prevalence and natural history of 

 rearranged non–small cell lung cancer. The prevalence of 

 is about 1.5% to 2% of non–small cell lung cancer adenocarcinoma histology. It is predominantly adenocarcinoma histology and occurs in patients who have never smoked or have a light smoking history, whose age tends to skew younger. They tend to be metastatic, and like with many oncogene-driven lung cancers, there can be a propensity for CNS (central nervous system) metastatic disease.

The prognosis for these patients is generally better than the overall population because we have good targeted therapies we can bring to bear. Crizotinib has been the most extensively studied. It is also an ALK inhibitor, as well as an ROS1inhibitor. It was initially developed as a MET inhibitor. In the most known clinical trial, which was published in 

, response rates were about 70%, with median progression-free survival a bit over 19 months, so it was highly clinically active.

In other studies with crizotinib, progression-free survival was a bit less but still highly active for well over a year. There is entrectinib, which is an ALK and NTRK inhibitor but recently approved for 

. Its duration of response in an analysis of patients with 

 rearrangement treated in multiple clinical trials with entrectinib was over 20 months. There is also a drug, ceritinib, which is approved as an ALK inhibitor. It's not quite as active as alectinib or brigatinib, and it does have more of a toxicity profile. It does have some activity in 

as well, although it is unclear whether it has any benefit over crizotinib or entrectinib and may have less activity in 

. There are newer drugs that have been studied with activities, such as lorlatinib and repotrectinib. For practicing clinicians, a common question I get is: Can you use alectinib for 

 because it's a potent ALK inhibitor? The answer is no.

Alectinib is an ALK inhibitor, and it has some RET inhibitory activity but has no ROS1 inhibitory activity. That's a question I get sometimes. There is a spectrum of drugs in development for this uncommon mutation, but one that we detect frequently in patients who have never smoked or have a light smoking history.

ROS1 Rearrangements in mNSCLC

: In terms of best approaches for molecular testing for patients with metastatic non–small cell lung cancer, there are different methodologies that could be used with either plasma or tissue. My general approach is to look at broad molecular profiling because we have 7 approved targeted therapies for 

 mutated lung cancer matching the TKI (tyrosine kinase inhibitor) to a great targeted therapy. We have osimertinib for 

, among others; alectinib and brigatinib for 

 fusions; larotrectinib and entrectinib for really rare 

 mutations; and dabrafenib and trametinib for 

To capture all of those, I look at broad-based genomic profiling, next-generation sequencing (NGS). You could look at tissue. You could look at plasma. There are pros and cons for each one. If I need something quickly, I will often send plasma since you don't need to set up a biopsy if you need additional tissue, and you get the results in a week. If there's a big surgical specimen sitting in the bank, I'll often use tissue. Where I don't have tissue, I'll first try with plasma as I'm setting up the biopsy.

There are methodologies, such as IHC (immunohistochemistry) and FISH (fluorescence 

 For IHC, there is an FDA-approved test, and those can come back quickly. There are certain situations where, if I really suspect there's an 

 fusion and I need to start treatment quickly, I will occasionally send that as part of a targeted panel to get the results quicker. Generally, I send next-generation sequencing for broad genomic profiling.

: My personal approach is to use both liquid and tissue next-generation sequencing at the time of the diagnosis. I believe that those tests are complementary. An alternative approach would be to do next-generation sequencing on tissue and save a plasma NGS, or so-called liquid biopsy, for patients in whom tissue was insufficient. I wait until I have all molecular testing before I initiate therapy. In general, most of your patients with metastatic non–small cell lung cancer will also have PD-L1 testing done.

I expect that PD-L1 testing will come first before I have the results of the molecular testing. I always wait for the molecular testing results before I start chemotherapy or any kind of treatment. Occasionally, I have to start therapy before NGS results are back because the patient is highly symptomatic and rapidly declining. In this case, I would recommend not giving immunotherapy with the first cycle. The reason to hold immunotherapy is to avoid emerging adverse effects, such as pneumonitis, if immunotherapy was given before tyrosine kinase inhibitors.

If you think about testing for our patients with metastatic non–small cell lung cancer, you need to remember to test. There are several ways of testing for molecular alterations. In some instances, immunohistochemistry or FISH are appropriate. I use next-generation sequencing, which has both a DNA and RNA capacity. I know that some institutions will have a rapid immunohistochemical screen for 

, and others. It doesn't really matter what system you use as long as you use the one that assures that your patients with metastatic non–small cell lung cancer get tested for all the abnormalities that have FDA-approved drugs.

 could be a part of NGS. Alternatively, you can order those genes separately. The most important thing when deciding to test and what company to use is to check your platform, if you're using an NGS platform, and make sure that the company you're using is testing for 

 is a new addition; we weren't testing for 

 fusions 3, 4 years ago. In my practice, I prefer NGS over single-gene testing for 2 reasons: (1) I believe NGS testing saves tissue, and (2) it's likely to save costs because now we have to test for 7 molecular abnormalities. I think the cost will probably add up if you use single-gene testing.

If you identified a molecular abnormality that has an associated targeted therapy, you should go ahead and treat. For example, if you use the single-gene testing and find that you have an 

 mutation, that's enough for me to start treatment. Our patients with oncogenic drivers are unlikely to have 2 separate oncogenic drivers in the same patient. Those cases have been described, and at this point, we do not know if they should go for 

 first, for example, for patients who have both of the alterations.

It is not necessary to complete the NGS testing if you already know that your patient has an 

 mutation. NGS testing does give you some additional information. We know that patients who have, for example, a 

 mutation have a worse prognosis than patients who don't have either. They're also more likely to transform into small cell lung cancer. Even though I'm not making any treatment decision based on that knowledge, I do find that it does give me some prognostic information. I might also look at that patient a bit more closely.

Another tip that I have for next-generation sequencing is if you have a patient who is never a smoker, or is a former light smoker, who has a higher incidence of molecular abnormalities… If you did not find it on your DNA-based next-generation sequencing, you could consider RNA-based testing, which would be more sensitive to fusion.

 It is great to discuss the proper molecular testing paradigms nowadays for our patients with non–small cell lung cancer, especially when we face situations where the tissue is either somewhat limited or very limited. Fortunately, we have a lot of different ways to facilitate that effort, to enhance the completion of molecular testing, which is critical to put our patients on the best treatment path and achieve the best long-term outcome. We want to make sure that we understand the patient’s tumor the best upfront. I complete biomarker testing.

Now, how do we deal with this situation of having potentially limited tissue? Issue number 1 is to make sure that we have good communication between members of the treatment team. This is where a multidisciplinary tumor board is so important to educate each of our colleagues. For example, with our pulmonary colleagues, we know that every time they do a biopsy, their biopsy might be needed for molecular testing. They do their absolute best to do a number of different passes, build a specimen that’s…enough to allow molecular testing, even from a cytology specimen. It’s the same way for interventional radiologists. We know that they need to take a couple different cores, as long as they’re safe from a particular patient, again, to allow the need for molecular and biomarker testing through the patient’s treatment continuum.

Still we’ll face situations when what’s obtained is limited due to safety concerns. Lymph nodes or nodules might not allow much of a specimen to be obtained. The next issue is having the pathology department ensure there’s appropriate tissue stewardship, that the tissue is prioritized for testing that’s necessary, as opposed to potentially wasting tissue for immunohistochemical stains that might not be absolutely necessary. Both coordination and stewardship are very important.

Lastly, we’ve been greatly assisted through the development of ctDNA (circulating tumor DNA) technology, liquid biopsy technology, where you have a patient with very limited tissue where we cannot perform all the molecular tests. We can utilize this platform, ctDNA testing, to complement the tissue testing we’ve been able to do. It is dependent upon the tumor burden of the patient, the stage of the patient, the yield of ctDNA testing, but in the metastatic lung cancer case, 70% to 80% yield is expected. A large majority of the patients will have informative ctDNA results, and they are strongly reliable enough to guide treatment. For example, if you detect an 

, we have strong data to say that the patient should benefit from the targeted treatment just as much as if we had detected the same mutation abnormality from tissue-based testing approaches.

Molecular Testing for NSCLC

 Lung cancer has enjoyed tremendous progress in understanding oncogenic-driven subsets. We used to pay attention only to histology in treatment selection. Now we have 7 oncogenic drivers, all associated with FDA-approved targeted therapies. Those are the EGFR, 

In addition, we have 3 other abnormalities such as 

 [human epidermal growth factor receptor 2] exon 20 insertions, which have compounds in clinical trials with preliminary evidence of efficacy. With novel targeted therapies, we see unprecedented survival with some of the drivers. Take, for example, recent update of the ALEX trial for 

-rearranged lung cancer. The alectinib arm in this trial showed a 5-year survival of 67%, and this data set is only 37% mature.

The median survival has not been reached, and the median follow-up is already 48 months. We also have learned in the past 10 years that the CNS [central nervous system] is currently a battleground for many of oncogenic-driven patients. In response to that, all novel targeted therapies are being developed with CNS coverage in mind.

-rearranged non–small cell lung cancer has evolved recently, there have been a lot of exciting developments. For the past several years, crizotinib, which is both an ALK and ROS1 inhibitor in addition to be initially developed as a MET inhibitor, has had potent activity in ALK and ROS1.

That was the first-generation ALK inhibitor, and it was compared with next-generation ALK inhibitors, alectinib and brigatinib, in the ALEX and J-ALEX trials for alectinib and in the ALTA-1L trial with brigatinib as first-line therapy. There was a substantial improvement in progression-free survival with bringing these next-generation ALK inhibitors, such as alectinib and brigatinib, up front. Alectinib had a median progression-free survival of about 3 years, and brigatinib median progression-free survival was about 2 years.

This really changed the landscape of how we treat ALK-positive non–small cell lung cancer. When we talk about 

 non–small cell lung cancer, crizotinib is a really good ROS1 inhibitor. Its approval was based on a study of about 50 patients. It was published in the 

 in 2014. The median progression-free survival was a bit over 19 months, so really effective response rates at around 70%. Lately, there’s a new approved ROS1 inhibitor, entrectinib, which also appears quite active and has CNS penetration as well. It’s a bit better than crizotinib, which is also an advantage.

It’s been great that we’ve had these new ALK and ROS1 inhibitors to bring to the clinic. Also in development is lorlatinib, which is approved for ALK non–small cell lung cancer after progression on a prior ALK TKI [tyrosine kinase inhibitor] and also has activity in 

 non–small cell lung cancer, although it’s not approved for that indication.

 fusions are oncogenic and lead to downstream activation of signaling pathways. In lung cancer, 

 fusions are found in approximately 1% of patients. The incidence is slightly higher in a subset of lung cancer called lung neuroendocrine carcinomas. We have 2 FDA-approved histology-agnostic 

-fusion agents, such as larotrectinib and entrectinib. Those drugs have remarkable activity with a response rate of approximately 60% to 70%. Responses appear to be durable, and CNS activity has been reported with both larotrectinib and entrectinib. We definitely need to be testing for 

Overview of Rare Driver Alterations in NSCLC

There are various KRAS G12C direct inhibitors beyond AMG 510. MRTX849 is another one of these. It was explored in a phase 1 trial that looked at the safety and activity of this drug, and while we've seen fewer patients with lung cancer with 

 G12C mutations, we've certainly seen proof of concept that the drug is active. We've seen responses in 3 out of 5 patients in the last presentation for an objective response rate of 60%, recognizing that we would need to see more patients treated with this drug moving forward, but certainly, it's very exciting. It also shows that the strategy of direct inhibition of a specific mutated KRAS protein works across different molecules that target this mutated protein.

In terms of MRTX849, it's being studied in several clinical trials, including the KRYSTAL-1 trial, which is a phase 1/2 study with expansion cohorts looking at the direct KRAS G12C inhibitor MRTX849 stratified by histology, whether colorectal cancer, non–small cell lung cancer, or pancreatic adenocarcinoma. It's also looking at combination studies in certain histologic subsets that may have activity in combination, such as the PD-1 antibody pembrolizumab in non–small cell lung cancer, the EGFR TKI [tyrosine kinase inhibitor] afatinib in non–small cell lung cancer, and cetuximab in colorectal cancer. MRTX849 is a covalent G12C inhibitor that binds to the GDP downstate of 

 and in this trial there were some adverse events but, generally, it appeared to be well tolerated. Adverse events included mostly grade 1 toxicities, the most common being diarrhea and nausea. Two patients had grade 3 or higher toxicities in the initial clinical trial, including fatigue and decreased appetite. Overall, the drug was well tolerated with a low incidence of grade 3 and 4 toxicities.

KRAS G12C Inhibition in Advanced NSCLC: MRTX849

Thankfully, we now have a variety of different selective direct inhibitors of 

 G12C. As a reminder, it's a specific mutation that you're looking for. One of these drugs is AMG 510, which right now is named sotorasib, and it is in clinical trials for patients who have these G12C mutations, such as lung cancer, and other cancers that might harbor G12C, such as colorectal cancers. We've seen very good activity in 

 G12C-mutated lung cancers, and we’re seeing response rates that approach 50%. That’s a dramatic difference from the older strategies that target 

. This drug has been very exciting for the field in general. The design of these drugs is unique and has been made possible because of the particular configuration of the KRAS G12C mutant protein that’s unique relative to other proteins such as G12D or G12V. In a particular, there is an IL [interleukin] group on the cysteine of 

 G12C that can be targeted with these covalent inhibitors. In general, one thing we haven't mentioned is that these direct inhibitors target the protein and lock it in the inactive state, noting thatthe KRAS mutant protein, shuttles between these active and inactive states for it to trigger continuous downstream signaling.

 Obviously, in an early phase 1/2 study, the goal is to determine the dose, safety, and early signs of efficacy. Those early signs of efficacy are generally, first, objective response rate. And then later, duration of response, progression-free survival, and overall survival.

Most molecular therapies have objective response rates as single agents of 50% or more and median progression-free survival of 9 to 10 months or more. The FDA has determined from other molecular drivers that chemotherapy and other standard treatments in the past have had response rates lower than 50%. The median progression-free survival is in the 4- or 5-month range. They have given accelerated approval to agents that have response rates over 50% and a median progression-free survival of 9 months or more.

These 2 agents do seem to have response rates that are appearing to be in the 50% or higher range. The median progression-free survival is too early to determine, but certainly, could be in the 9 month or more range. The safety profiles have been really good for these agents so far. The rates of grade 3 and 4 serious adverse events is quite low. So far, these agents seem to be as safe or safer than some of the other molecular therapies that we have. That's because normal cells don’t have 

 G12C mutations. These therapies seem to be highly specific and therefore, not causing serious adverse effects. For the grade 2 anemias and so on that have been reported, it's not clear whether those are due to the disease or due to the treatment at this point.

The CodeBreak 100 study is a phase 1/2 trial of the selective KRAS G12C inhibitor AMG 510, or sotorasib, for patients with 

mutated lung cancers and other cancers. In particular, we're looking for that G12C mutation. The goals of the study were to show the safety and activity of this molecule in this molecularly enriched population, and we certainly saw that the drug was not only active, but also safe. I think that's critical to know because 

 G12C doesn't exist in the normal human body, and so there is not a 

 G12C molecule that is harbored by normal tissues. Conceptually, in the design of these drugs you would have a good therapeutic window where you're able to hit the target hard, but not actually hit normal tissues to a substantial extent. You would see a high frequency of drug-related toxicities.

In terms of the development of AMG 510 in KRAS G12C inhibition, I do think it has the potential to be practice changing. In non–small cell lung cancer, overall response rates are about half, about 50%. I think we need to know more about duration of response and progression-free survival among a larger group of patients to see if it stacks up to osimertinib in 

-positive non–small cell lung cancer. We need those data to mature and get the result of larger trials. I'm hopeful it will be practice changing, even if it seems too early to say. I would note that there appears to be some differential activity based on histology. For example, activity in non–small cell lung cancer seems to be better than colorectal cancer where response rates are lower. That's where I think we need to look at the effects of histology and other comutations and how they may impact the activity of KRAS G12C inhibitors.

In terms of the practical implications of AMG 510 as your direct KRAS inhibitor, I do think it has the potential to be practice changing. I also think it's too early to say without further data, without the progression-free survival, duration of response, and having more data with that among a larger group of patients, to say how practice changing it will be and what those practical implications are.

 For these agents, toxicity has been pretty minimal. These have been well tolerated. There have been a couple of cases of diarrhea and anemia, and these have been well controlled and have not been dose-limiting. In fact, when it comes to anemia, it's not totally clear whether it's due to the drug or the disease, but diarrhea and anemia have been quite infrequent and easily managed and have not required discontinuation of the drug.

KRAS G12C Inhibition in Advanced NSCLC: AMG 510

 has long been known to be "undruggable" for a variety of factors. The KRAS protein itself was perceived to lack a good drug binding pocket outside of the nucleotide binding pocket of the RAS molecule. In addition, RAS does complex itself with GTP, or bind to GTP. In fact, it has a very high affinity for GTP, making it difficult for us to design GTP-competitive inhibitors, to outcompete the GTP molecule for that particular pocket. We've tried many different strategies in the past, and to name a few of them, there have been farnesyl transferase inhibitors that inhibit the localization of RAS to the membrane; those didn't work. We tried to leverage the fact that RAS activates downstream pathways, such as the MEK and ERK pathways, and have given either single-agent MEK or ERK inhibitors or combinations that hit these downstream pathways. While we've seen isolated responses in patients here and there, we haven't really seen a single series result in a very large number of patients who benefit from therapy.

KRAS G12C direct inhibitors have been an exciting development and potentially a transformational development in the treatment of non–small cell lung cancer. Several years ago, Kevan Shokat’s, PhD, laboratory and his colleagues identified small molecules that can irreversibly bind to the mutant reactive cysteine at residue 12. They identified a previously unappreciated binding pocket near the KRAS effector region. Small molecules binding to this pocket can inhibit KRAS by locking the protein in a GDP balance in inactive state. That development has spurred these new small molecule, direct KRAS G12C inhibitors that preliminarily looked to be able to induce responses and to be efficacious, particularly in non–small cell lung cancer. That's been an exciting development, that we're now able to take these small molecules to the clinic and see responses.

In terms of the rationale for identifying that cysteine at residue 12, it can target the binding pocket near the KRAS effector region where it locks the protein and its GDP bound in an active state. That underlies its activity as a direct KRAS inhibitor. I think once again that's an exciting development to be able to target that binding pocket.

 in that situation because the G12C substitutes a glycine for a cysteine. Potentially, EGFR inhibition may have some impact on wild-type 

 and that interaction may potentiate activity. That's being studied in future clinical trials.

In terms of tethering approach to drug design, that's a methodology to screen ligands that bind to targeted site proteins via a disulfide tether, where a native cysteine or an introduced cysteine can capture the ligands. That's a way to screen for these direct KRAS inhibitors of G12C. It's the drug development approach that was initially taken to identify these small molecule inhibitors of KRAS G12C.

, there are some covalent drugs that have been developed. These drugs covalently bind to the receptor and prevent GTP binding activation. These G12C agents that bind covalently are called AMG 510 and MRTX849. These are the first KRAS-specific drugs that have entered a clinical trial. Early clinical trial results have established safety, dose, and some efficacy of both of these agents. There's a limited number of patients, less than 30 for both drugs, so we don't actually have anywhere near enough data to have these drugs approved yet. We don't know the median progression-free survival, the overall survival, what it will be with these drugs. But we do know that these drugs can cause responses in 50% or more of 

Development of KRAS G12C-Targeted Therapy in NSCLC

Unfortunately, to date, we have not seen regulatory approval of a specific targeted therapeutic for 

-mutant lung cancers. We certainly hope that will change in the future, but for now, the systemic therapy options that are available to these patients include chemotherapy, immunotherapy, or a combination of both in the upfront setting, and the other second- and third-line options that are approved for non–small cell lung cancers in general.

In terms of EGFR TKIs [tyrosine kinase inhibitors] in 

-mutant non–small cell lung cancer, that's not really a practice that's used anymore. Afatinib is a second-generation EGFR TKI that's approved after platinum-based chemotherapy in squamous non-small cell lung cancer, based upon a modest benefit that was observed in lung adenocarcinoma. Erlotinib used to be approved in lung adenocarcinoma after progression on platinum-based chemotherapy, even without 

 mutation. Sometimes that was used, but it's really not used anymore and it doesn't particularly work in 

There are some interesting data, and I think what's more relevant now is ERBB pathway inhibition with EGFR and HER2 on impacting GDP and GTP loading, in terms of shifting KRAS and the GDP downstate that may help synergize in certain settings with direct KRAS G12C inhibitors. That's where I see the future heading, with EGFR TKIs and 

-mutant non–small cell lung cancer as part of a combinatorial approach, with direct KRAS inhibition or other combination treatments targeting the RAS/RAF/MAP kinase pathway, rather than as a single agent in 

mutant non–small cell lung cancer. With the changing landscape of much more effective therapies, its role has been substantially diminished.

 For many molecular drivers, the immunotherapies are less effective. If you look at 

, the immunotherapies don't seem to work so well. For 

 mutations, the immunotherapies tend to work at about the same frequency in the 

 mutations as in tumors that don't have any actionable driver.

Immunotherapy is a good option for patients who have 

mutations. If they're those, the selection would primarily be based on the PD-L1 expression. Whereas high PD-L1 patients with 

 mutations could receive a checkpoint inhibitor alone, and those with lower PD-L1 expression might receive checkpoint inhibitors plus chemotherapy.

There are some co-occurring molecular changes in 

 such as STK11 or PEAK1, where the chance that immunotherapy might work tends to be lower. The best thing to do in most cases is unknown, but the likelihood that immunotherapy will work in the presence of a 

 mutation plus a PEAK1 mutation, or an LKB1 mutation, may be lower, but we don't have a specific thing to do for them at the moment.

Systemic Therapy for KRAS+ Advanced NSCLC

 mutation on a patient’s molecular report after testing for lung cancer, it’s important to underscore that 

 mutations are really a heterogeneous group of different mutation types, meaning there are different amino acid changes that involve different residues along the 

 G12C mutation is the most common mutation. But in nonsmokers you see other mutations, such as the 

 G12D mutation, that are much more common. Beyond that, it can involve other amino acids, such as the amino acid 13 of the 

 protein. The difference is that we now have drugs for particular mutations or substitutions, such as 

 G12C, versus the other mutation types where we might employ different strategies altogether. The specific mutation that’s identified in a patient’s cancer may dictate the likelihood of us having a more exciting clinical trial for a patient, versus just more standard options for these patients on a targeted-therapy front.

 is found? The discussion we typically have is that previously, it was thought to be a poor prognostic feature, but with the evolution of immunotherapy and also these direct KRAS inhibitors, it’s really unclear if that’s true anymore. In terms of up-front treatment outside a clinical trial, the 

 mutation, I explain to patients that it’s the most common mutation we encounter in their type of lung cancer, and that immunotherapy can still be effective. For example, unlike the 

 lung cancer patients—especially the ones associated with smoking, such as G12C—can still derive good benefit from immunotherapy. But for 

, there’s really limited demonstration of effectiveness of immunotherapy, such as pembrolizumab and atezolizumab, especially as single agents.

I counsel them that they could still respond to immunotherapy or chemoimmunotherapy, stratified by what their PD-L1 status is. For example, PD-L1 high, they get pembrolizumab or chemotherapy with pembrolizumab. I also counsel them that we may have clinical trials for them. If they have 

 G12C, the direct inhibitor trials are very exciting and are in development. We also have trials targeting certain computations of 

. I emphasize that we have potentially excellent treatments for them, whether they are immunotherapy, chemoimmunotherapy, direct RAS inhibitors, or clinical trials with combination therapies.

Counseling Patients With NSCLC When KRAS is Detected

 In general, in the beginning, if the driving alteration was 

, it’s unlikely that the tumor cells would have 

. On the other hand, sometimes in the presence of an 

 drug, some of the cells will become resistant by developing 

 fusions, some cells, minimally, may become resistant by developing 

 mutation. If you have a patient with an 

 mutation, it will make sense in the future when we consider treating the patient with the 

For all these molecular alterations, 1 mechanism of resistance is they develop mutations that will prevent binding of the drug. In another way, it becomes resistant to development activation of another path, which may be 

. In some instances, there are other activating pathways such as the 

Jonathan Riess, MD, MS, UC Davis Comprehensive Cancer Center

Articles