 Let’s wrap up our whole discussion. What a great discussion, and everything that you all have brought to this will be valuable to all who listen in. Tony, give me a sense of what you think are the challenges going forward for 

 The biggest challenge is the obvious. These are rare, incredibly rare, and the rarity becomes even more problematic when you’re not consistent in looking for them. This means for something that’s 1 in 100 or 1 in 1000, if you just randomly pick every other patient that comes to your clinic and send for testing, then you are 1 in 100,000 just by randomness alone.

One thing we have to keep in mind is that if you a have a fusion, you’re going to have a great response, for a patient who may not do as well on chemotherapy. For some patients, this is 1 of the longest reported responses so far. There are many that are ongoing. It is now almost 4 years. We know that if patients respond, they can respond pretty well. So we want to find them.

This is the same thing with MSI [microsatellite instability] high and the same thing, frankly, with a lot of these alterations that are at the level of 1% or 2%. We have to be consistent. We have to look for them consistently—that’s the biggest challenge.

Luis was talking about how these next-generation agents that are looking, essentially, at patients who fail because of the development of the mutation. Now we’re looking at even a smaller subgroup of patients and how to salvage them and make a difference in those patients. That also brings the question of those who are intrinsically resistant because they have the core mutation, and how do we actually put all these together? Frankly, it’s not just multidisciplinary collaboration, it’s also multispecialty collaboration.

We can’t do it in the GI [gastrointestinal] space or in just the lung or saliva. So how do we bring everyone together to essentially work on this? That’s 1 of the challenges. The other challenge is, of course, testing.

We heard a little about IHC [immunohistochemistry], FISH [fluorescence in situ hybridization], RNA for fusions. DNA-based testing is problematic because it may miss 25% or 30%, and we know that the most commonly used 1 in clinic is actually DNA-based genomic testing. You’re missing on some of these patients—30% of a very small percentage is problematic, so the DNA-RNA hybrid assay is probably the best, but at the same time it’s the most costly and takes a long time.

IHC and FISH are useful, but again, how many patients are you going to test for IHC and FISH to find 1 patient? They’re not costly. IHC is highly sensitive and specific and requires 1 unstained slide, but you’re wasting a slide, and you may be wasting time when you need to find a lot of other alterations.

If we can find a better way to do this DNA-RNA, have sequencing more efficient or faster so we can give the answer to our patients, and rely more on these liquid-based biopsies to find them better, that would actually be the best bet to continue to move these things forward and eliminate some of these barriers that are prohibiting us from finding these patients, because they do better.

 Jyoti, maybe you could comment for me. I hear so many comments from community oncologists and the like who are frustrated by the concept of raw molecular profiling, and they’re like, “I never find anything.” They get tired of looking because they don’t see much value in it. What’s your counsel to somebody who’s having that feeling?

 Sure. These drugs and these targets certainly feel like unicorns, but if you don’t look, you’ll never find them. They radically change the outcome for these patients, their ability to get back to life, the improvements in performance status, and the rapid diminution of symptoms. It’s really remarkable.

Certainly just at the beginning of this COVID-19 [coronavirus disease 2019] crisis, we believe we just had a patient with what looked like horrific disease, and we found out she had an 

 fusion before we all went in to hunker down. The entire team was ecstatic and so happy for her. She’s already feeling a lot better.

The point is, even if it’s 1% in common cancers, that adds up to a lot of people every year. These uncommon cancers in which the fusions are pathognomonic, certainly that’s easy, that’s knee-jerk. But taking the time to individualize for each patient, developing the right plan, takes discipline, as you mentioned, but it also takes moving away from what we think a patient with an 

 fusion should look like. It can occur sporadically in anyone with colon cancer, pancreatic cancer, or lung cancer. You just need to look.

As the cost of testing goes down, that barrier is changing. The ability for us to look across multiple fusions and multiple mutations is now backed by better and better evidence. Again, that barrier should be less, and we should really push our systems to make that cleaner, to rely on less tissue, and to get answers faster for our patients.

 Yeah, and I know this whole wave of precision medicine has got it to where most insurance is covering these tests with CMS [Centers for Medicare & Medicaid Services] support. I think about how many Chem-20 tests and how many screening CT [computed tomography] scans I’ve done over my career where I got nothing back. It’s normal, or it’s negative, yet that hasn’t slowed me down from ordering those Chem-20s and the CT scans. I wonder if that’s the way we should be thinking about this.

Shubham, why don’t you give us your thoughts on this issue of testing and impact on patients?

 As Tony and Jyoti said, if we don’t test them, we won’t find them. So testing is important, especially where you can really find it. Right now, at least for these fusions, you really want to find these fusions, otherwise you might miss them. You might get false negatives. So RNA-Seq is a very important factor in this. As Tony was saying, we don’t catch all these on DNA or RNA-Seq.

It’s still the missing part that can really make it robust. This is because the sensitivity and specificity of RNA-Seq is far better, and we can really find that. When you’re looking at a needle in the haystack, you want to find that needle; you don’t want to miss it.

The second thing is, once you pick out in these patients, especially with these rare tumors—as I said, the responses and everything are very gratifying, obviously for patients but also for the treating physicians when you see these patients do better for a long period of time.

 Yeah, and to add to what you just said, RNA-Seq is important. It’s important for you to understand where your test is going, where your sample is going, and that your shop is doing the appropriate testing in support of that. Is that fair enough?

Videos

Challenges for TRK Fusion–Targeting Agents

 Luis, just to let you have the last word on this, as with many of these agents, resistance is emerging in 

 fusion-positive cancers. Could you talk a little about what’s going on in that space to deal with those people who are progressing on these drugs?

 This is a really fascinating story, because as Mark mentioned before, we have broken several paradigms in drug development.

Normally we develop 1 drug, then we wait 3 or 4 years, and then we develop the next 1. This was amazing, and that’s why in 2016 it was not a surprise that they already published the first 2 resistant mutations. For me, it’s very easy to remember, because there’s a road 595 nearby. There is G595, and G667 looked very similar arrangement to a lot of the drugs in development.

But the thing is, the reason we have to give a lot of credit to Alexander Drilon and all these people involved, is that in 2016 we already knew the resistance. To go back a little, when you develop resistance to a drug, the resistance can be on target and it can be off target. In the case of 

, the resistance is so far on target, meaning that it is a mutation in the 

 tyrosine kinase that is caused by amino acid substitution. That originated in the binding of larotrectinib and entrectinib. This mutation essentially it doesn’t happen anymore.

In 2016 they discovered the first 2 resistance mutations in vitro. In 2017 Alex Drilon had a chance to put the first 2 patients who had failed larotrectinib in treatment with the drug LOXO-195, which is now called selitrectinib.

We already have 2 patients treated, 1 with colon cancer who responded for many months and a patient with a salivary gland who was unfortunately a baby. She responded for, I think, 6 to 8 months, and finally developed resistance again. But since that time, LOXO-195, or selitrectinib, is accruing patients who have resistant mutations, and there are several other resistant mutations that have developed.

This is a very similar story with what we have seen through 

 fusions in developing resistant mutations. We are also really lucky that we have other the drugs. We have 3 compounds more that are in vitro. They have shown that they can rescue patients who have failed. The second is repotrectinib. Repotrectinib is the 1 that was called TPX-0005.

Repotrectinib is effective against ROS2—that’s what we are accruing now nationwide for ROS or NTRK failures. But repotrectinib already has a preliminary report of patients who have responded after they have failed larotrectinib. The other drug is called DS-6051b, from Japan. That’s a drug that in vitro shows benefit, but in vivo there is at least 1 report already—it was a year ago—that a patient with thyroid cancer already has responded.

The last drug is a drug from Ono Pharmaceutical Co, Ltd, that in vitro seems to have some efficacy, but in the clinical arena we haven’t seen any publications yet.

 It is great to know that these things are on the way and being developed. I was just thinking as you were describing all that, that it requires us to collaborate. No one center can do these centers because of the frequency or the rareness of this mutation and these fusions. It really is encouraging to see everybody working together to get that.

Acquired Resistance in TRK Fusion-Positive Cancers

: Jyoti, I couldn’t believe it when it happened, but there’s a second drug approved, entrectinib, and it’s basically the same space. Maybe you could give us that big picture about if it is a different mechanism, or why would you decide one versus the other?

:Sure. Larotrectinib is so well tolerated that we see patients getting back to function, and we’ve seen long duration of responses. When entrectinib was approved, it added another option for our patients. Entrectinib is a little bit different in that it has activity against a number of kinases. We know that there is also 

 activity, and so it’s been actually approved for both kinases.

For entrectinib, the earlier trials were based on a number of basket trials as well as a pediatric study, and the number of patients with CNS [central nervous system] metastases was a little bit higher. Perhaps there are more data there, particularly for those of us that deal with lung cancer, in which we see patients with 

 disease and translocations who have a proclivity for brain metastases. That figures into our rubric of putting these drugs together.

I would say that the clinical activity profile, to me, is functionally pretty similar. The total number of patients that we see who have been treated across these trials is very small and limited. Suffice to say, there are more patients with CNS metastases treated with entrectinib, so perhaps a little greater experience there.

The toxicity might be a little bit of a different story, a little bit more fatigue with entrectinib, and some weight gain, at least in my experience. Patients have had a little bit harder time with some of these low-grade toxicities. But again, for someone with known CNS metastases, that may be a way to tease these out. But again, these are small-number studies looking at different pockets of patients, 1 larger basket as opposed to subgroup analysis of some smaller ones.

:Yeah, that’s very insightful. Shubham, have you used this drug at all, entrectinib, and is it equally active across all of the 3 subtypes?

:Yes. So I think it does the same thing, just as Jyoti said, so it does the 

 for rearrangements. I think it’s fairly similar. Most of my experience has been more with larotrectinib than entrectinib, but I think…the data seem to be fairly similar. Just as Jyoti pointed out, brain metastases and everything, they looked at that, so we have larotrectinib data with brain metastases also.

Approaching Treatment for TRK Fusion–Positive Cancers 

:Tony, our friend Jordan Berlin published some work at ASCO GI [American Society of Clinical Oncology/The Gastrointestinal Cancers Symposium] or has an abstract for this ASCO about 

 fusions in GI [gastrointestinal] cancers. Have you had a chance to take a peek at that, or have any sense of what that’s going to show?

:I think it was presented at ASCO GI. There may be an update coming at ASCO [American Society for Clinical Oncology Annual Meeting] 2020 as well. There were very few patients with GI cancers, primarily colorectal cancer. There were a couple of cholangiocarcinoma, pancreas, and appendix cases. I think there was 1 liver, as well. Overall, what was seen in the general population both in terms of efficacy as well as toxicity data were some remarkable responses.

You see some durable responses, again, very similar. The median overall survival was around 2.5 years or so. But with a small sample size it’s very difficult to have a very heterogenous group and a very small subgroup of patients. It’s very difficult to say whether it’s standard, individually, for the different tumors. Those patients respond, and they respond deep. The responses tend to be durable.

I think that’s the main message. So when you separate the gastrointestinal cancers from the overall group, you still see about the same. It is possible that there is a differential response, if you have salivary gland or maybe a lung cancer, or perhaps a less pretreated tumor, that you may actually see a deeper and more prolonged response. Of course, I think for the experience, as Luis put it, it’s too early to tell at this point of time whether there may be variations dependent on the site of origin.

But overall, I think for the GI cancers, we’re seeing very similar trends in terms of the responses, the durability, and their depth.

:Mark, maybe let’s drill down on the all-important toxicity. We really haven’t focused on that at all. I think you mentioned earlier in the program about some of the specific toxicities. Can you give us a high-level look? Then I’ll ask others to chime in about toxicity.

With these very targeted agents, we’re all getting used to looking at the grade 3 or 4 columns with the 0s or 1s, which is great. When you look at this particular agent, actually the most common toxicity—all of which was grade 3 or less—was an alteration in either ALT [alanine aminotransferase] or AST [aspartate aminotransferase], with most of them being grade 1, to be honest. Fatigue, some GI toxicities, as well as dizziness, are common. But surprisingly there can also be some anemia, so some myelosuppression there.

Other GI toxicities include diarrhea, which we can see with many of the TKIs [tyrosine kinase inhibitors]. There can also be a low rate of fever, or some arthralgias. Weight gain was another one that stood out to me, as well as headache. But again, if you start to look at the grade 3 or 4 columns, it’s very uncommon to be either severe or life-threatening. Most of these end up in the grade 1 column and are manageable.

:The one I come back to all the time is neurotoxicity, and I think you mentioned that as well. When you think about chronic therapy, cumulative toxicities, have others experienced patients with the neurotoxicity or have any issues that you think should be brought out during this discussion?

: Normally, just as Mark said, these patients tolerate this drug well, even with chronic therapy. They’re not like some of the other drugs that, when you go to chronic therapy, grade 2 diarrhea is not grade 2 diarrhea. If you’re getting consistent grade 2 diarrhea, that’s not good—that’s impeding the quality of life. Grade 2 fatigue is impeding of quality of life. It’s pretty bad.

So what I’m saying is we’re trying to really look at these chronic toxicities, but as far as it comes to these drugs, in my experience they have been mostly fairly mild. So for toxicity, just as Mark said, there can be some fatigue. But normally for even the chronic therapy patients, they have done fairly well.

Some diarrhea has been easily controlled. The neurotoxicities are very interesting. It’s actually dizziness. In most of the studies we’ve done a neuro exam and everything, but that’s what we see. It’s more a loss of balance than the classic neuropathy. Some patients do experience dizziness, and it could be sometimes bad enough that they might need to use something to walk, such as a cane or something, especially if they’re older.

TRK Inhibitor in GI Cancers: Adverse Effect Profile

 Luis, let me shift to you a second and drill down more about the tumor types. You mentioned them earlier, but let’s go through them again. Were there certain tumors that were more prevalent? Were there certain tumor types where there was more benefit seen?

 Yes. In 2018, when Alexander Drilon published in the 

—Mark just mentioned the paper—there were only 55 patients and 17 tumors. At that time, I remember, I was lucky to be part of that publication. We didn’t have a specific list of the response rate of the tumor, but it probably didn’t matter because there were very few tumors. But now that David Hong has published in 

 this year, we have expanded the number of patients. Now we have 155 patients, so we can at least start to see what the response is for certain tumors. For example, these are tumors that are very enriched with 

I’d never seen a fibrous sarcoma—these are tumors in kids. For soft-tissue sarcomas, thyroid cancers, we have responses that average out to 80% to 100%. What is interesting to me in our business is that the responses for lung cancer and breast cancer are a little lower, maybe 75%, but also the number of patients are very old. There are only 10 to 15 patients with lung and breast cancer, and for pancreatic cancer it’s the lowest response, around 43%. But there are only 2 or 3 patients. I’m not sure we are ready to say the real response based on tumor type. It probably doesn’t matter because we are very lucky that the average response is high among the tumors.

But with time, as more patients are accruing and we have more of this information, we are going to be certain of what is the real chance of response for all these tumors.

 Yeah. I always think these tumors that are messy or that have multiple drivers respond less than the ones that are more driven by, perhaps, 1 gene. Shub, why don’t you take us through quickly? I want to go through a few studies with everybody, including the pooled analysis that was published in 

 Great. That’s what Luis was talking about. This is 1 of my colleagues, David Hong. We enrolled a number of patients on the trial as a group. Essentially the pooled analysis was for the phase 1 trial, which was an adult population. Then there was a phase 1/2, which was a pediatric population. And a phase 2 trial, which was an adolescent and adult trial. There were 3 trials that were pooled together, as Luis said—about 156 patients.

The interesting thing was that criteria, if you really look at it as you’re pooling, was 1 month or older. The cutoff was 1 month, so you had to be greater than 1 month old for enrollment on the pediatric trial. These patients had received some type of standard therapy that was previously available, so they were treated as if it was previously available. The response rate was 79%, the CR [complete response] rate was about 16%, and 63% of patients had a partial response.

Interestingly, as we said, the median duration of response was about 35.2 months, so that’s almost 3 years, and the progression-free survival was about 28 months. The amazing thing was the median overall for these tumors, which were advanced, was about 44 months. It’s amazing—that’s almost 4 years for these patients.

As I said, I’ve treated a number of these patients, and it’s very gratifying when they get the response. First, it improves their performance status. Quality of life is improved, and obviously their improvement of overall survival. They are very interesting pooled data.

 Yeah. Jyoti, Dr Patel, please give us a little review of what’s going on in brain metastases. Do these drugs work in that setting?

 Sure. In the larotrectinib trials, a smaller number of patients were enrolled. If we look across the gamut, I think most patients who were enrolled were those with CNS [central nervous system] primary tumors. There were a minority of other cancers, such as lung cancer, which have a proclivity for the CNS metastases. But in those patients, 14 patients total, the response rate was good. It was about 55% and varied a little between histologic type.

This is 1 of the situations where CNS response is a little more difficult to quantitate. It’s probably similar to pancreatic cancer response, for example, where we don’t see those really robust response rates primarily because of prior treatment there or lesions that can be target lesions but had received prior treatment. So it’s a multitude of factors.

In this trial, what we see is that patients do respond. But more important, there was freedom from CNS progression and long duration of response. You’re preventing other disease, and this is really where you see the benefit. Certainly larotrectinib has an active CNS profile.

Efficacy of TRK Inhibitors

 We’re going to shift gears to a very specific set of data around the clinical experience with the TRK inhibitors. Tony, we’ve tagged you to start with a fairly obscure abstract that just came out around biliary cancer. I know it’s a disease you like discussing. Talk about the recent ASCO GI [American Society for Clinical Oncology Gastrointestinal Cancers Symposium] meeting, and what that showed in biliary tract cancers.

 To focus on the gist of it, this was primarily focused on IHC [immunohistochemistry], but it also had NGS [next-generation sequencing] confirmation of the IFC [integrated fluidic circuit]. It just shows that there is a relative concordance between the 2, the IHC and the NGS. Although, as we’ve heard, we all favor an expanded genomic sequencing because we’re looking for a lot more than just the 1, at least in specific tumors such as biliary cancer, where 

 is relevant. We just had an FGFR agent approved for 

 fusions. We have IDH alterations, we have 

 [human epidermal growth factor receptor 2] amplifications.

We want to have the full gamut, not just 

 fusions. What this study mostly shows is that these 

 fusions in biliary cancer are present in about 0.7% of all patients, and that aligns with the rarity of these fusions, which is pretty consistent for most cancers. We know some cancers have these fusions at a much higher rate, but in GI [gastrointestinal] malignancies, and others as well, the 0.7% seems to be consistent, and that’s pretty much the gist of this. This confirms the rarity of these fusions in a rare tumor like this.

Biliary tract for us is a target-rich place, but it’s about the same as we’re seeing in other cancers as well.

 Yes. If you find it, you’re going to go after it, because these tend to be driver mutations, and the responses from the study look pretty interesting.

 Mark, we’ve tagged you with setting the stage for larotrectinib, along with its pathway and eventual approval. Set the stage for us about the trials, the inclusion criteria, and how that went.

 Yeah. The original trial that led to the FDA approval was a trial that selected prospectively patients that were 

-fusion positive. It was actually a combination of 3 different studies, a phase 1 study in the pediatric population, a phase 1/2 in the adult population, and then a phase 2 study that involved both adolescents and adults with 

There were 17 different cancers that were included in this. There were only 4 patients with lung cancer. Talking about the bile tract, I think there were a couple of patients with biliary cancer in this. It was fascinating that patients ranged in age from 4 months to 76 years. The interesting thing is that the response rate objective by independent radiological review was 75%, so the vast majority of patients responded. They had a couple of patients who were early progressors, and it’s not clear they were true positives. There were some issues about the testing that was done.

They also had a patient that progressed early that had received a prior 

 drug and had a resistance mutation, so that patient would not have been expected to respond in this particular setting. It was remarkable in the sense that no patient had the drug discontinued because of toxicity, so it was very well tolerated. There were very few grade 3 or 4 adverse events. Mostly they were grade 1 or 2, so it was the best of both situations. In the most recent update of this—correct me if I’m wrong, but I’m not quite sure the median duration of response has been reached at this point.

It’s the second drug after the pembrolizumab MSI [microsatellite instability]–high indication that was tumor agnostic but linked to a molecular alteration regardless of pathology. It’s an important step in how we think about cancer and get drugs approved.

 As another member of the panel who graduated from fellowship in a year beginning in 

, I remember being scolded for suggesting that we wouldn’t need phase 3 randomized trials if the benefit was obvious and that it would transform drug development. Even though I got beat up pretty bad for saying, “No, you’re wrong,” back then, I’m so pleased this is the way it’s going. Comment a little about how this has changed our whole perspectives of cancer and drug design.

 It’s been fascinating. John, you may be the only 1 on the call who remembers a gentleman named Joel Tepper, a radiation oncologist at the University of North Carolina. He would say, “I’d like to do a phase 3 trial that required only 50 patients.”

It was so big that he didn’t need 1000 patients to get the benefit. But again, the larotrectinib indication is a growing list of drugs that should be approved for these molecular subsets of patients that can be clearly defined. In lung cancer, we have the EGFR and ALK experience, where we actually went through a number of randomized trials to show that when you have a target that’s targetable with a targeted therapy, either an EGFR or an ALK drug, it’s better than chemotherapy.

 mutation? Do we have to do it for every single molecular category that we have to prove that it’s better than platinum? Why are we in love with platinum-based therapy in this setting if you have a target where it’s driving the cancer? With the targeted therapy, all the response rates are north of 50%, and the durability is 3 to 4 times longer than we see with chemotherapy.

 This wouldn’t be feasible with something so rare to randomize it against something.

 To me, the wisdom and the flexibility of the FDA has been terrific in this space.

Early Development of TRK Inhibitors

Let me say in closing for this session, I hear a lot of people complain about their molecular reports that come back, they don't understand how to incorporate that and interpret it. A favorite study we ran at Medstar Georgetown University Hospital was an eye-tracking experiment, because academic oncologists kept saying, “Don't tell me what drugs are related.” They wanted to see the genes or the fusions, and they'll do the reporting. Whereas community oncologists traditionally have said, “I don't want to know the genes. Just tell me what drug to give.”

When we did eye-tracking experiments we showed academic and community oncologists the report. Guess where everyone's eyes went as soon as they showed it? It was right to the drugs, and only then people would go back and reconfirm with that. I think that put our partner companies that do molecular profiling on the hot seat for some of the recommendations they were making with therapies. It was interesting to watch that tighten up over time. Tony, give me a sense of what your favorite type of report is. Do you want the genes, do you want to know the drugs, or you want both?

I primarily want to know the genes. I frankly find the drug listing a bit random sometimes for most of the drugs. We're looking for the drug genes that are relevant to the disease. If they're relevant, then we know we have to target them. I don't care what's on the listing.

On the other hand, a lot of the other background noise can actually end up detrimental in some ways. This is because I've seen patients with a random mutation, there was a case report that links it to everolimus, and then the physician wrote everolimus before even sending the patient to consider a clinical trial that may or may not be relevant for that tissue.

I find that very problematic. Linking it to clinical trials is very useful, and that does help. It helps especially when we have driver mutations or driver alterations that are targetable. By the way, I look at the variant of nonsignificance, as well, and what time they graduate to normal significance, and they do. Essentially, some of the 

 alterations of unknown significance have, in my clinic, initiated with some dramatic responses to FOLFIRI [leucovorin calcium, 5-fluorouracil, and irinotecan] or FOLFOX [leucovorin calcium, 5-fluorouracil and oxaliplatin].

 Yes, sometimes that little tiny pulmonary nodule is actually a met [metastasis]. Mark, any thoughts?

 I struggle a lot with some of the reports that we see. I agree with Tony, I'd rather see which genes are altered and how they're altered because I know what to do once you tell me what's going on with the DNA or the RNA. We had a recent case at our center where a patient was sent for a second opinion on a Guardant Health assay. He had a 

 mutation, and not a fusion. But he got started on entrectinib because the report said 

entrectinib, not making the distinction between the point mutation and the fusion.

The payer from Blue Cross Blue Shield missed it.

Yes, or wherever he was from, but I agree with Tony. Sometimes the reports are very damaging, so I think we have to be aware of that and educate our community doctors about these types of things. I get several calls a month about, “How do you interpret this report, or that report, what should I do?” It's good that they should be calling us to get the information.

You have to remember, though, that we are very focused, and it would be difficult if we had to know everything for every disease.

The nice thing about our center, too, is we do have a twice-a-month molecular tumor board. I'm sure others have it, too. We discuss a lot of these things in the tumor board, so that's very helpful as well.

I think the main point I want to leave us with is it's OK not to know. We're all learning. If you're looking at something that you're not sure of, you should check on it. There are ways to do that. Any other comments?

 One last comment. Looking at the drugs, a lot of the time there are 2 things out there. Let's say we’re talking about a 

 G13 mutation for pancreatic cancer. For the longest time, 1 of the NGS [next-generation sequencing] companies that we use all over has listed a MEK inhibitor for it. You've seen it, I've seen it, and we've seen it. We know what the response is. From our side it's an educational perspective, exactly as Mark said.

Then the second thing is the co-mutation when it comes up, not with 

 mutation and a KRAS, or something that is a co-mutation in which you could give a more of an antibody-targeted therapy, not a small molecule antibody target, it's not going to work in that. So I think they also list out, these are the 5 mutations, these are the 5 drugs you can give them. Sometimes that leads to the thought of what if I combine A, B, C, and D? We've seen that, too.

There's a lot of education needed on that. I really think the focus should be on the genes, and especially what Tony is saying because it doesn't mean that therapies don't exist, it just means that they really haven't looked at it. It's just not been that well described in the database.

There are a lot of little points in the report, and the reports are around 30 pages. Sometimes that can lead to things getting lost. That's the other thing with some of these NGS reports—they just go pages and pages. How many people look at more than the first 2 pages? Because some of the other things, nuances maybe hidden deep inside; that's the other issue.

 I want to say that there is no perfect report, because for us, we only practice 1 tumor, so it's very easy to know. Actually, I'm very excited with the genes because we have been discovering new genes, things like that. But I will mention I passed my boards, and in CML [chronic myeloid leukemia], I just discovered it's 4 or 5 lines now of TKIs [tyrosine kinase inhibitors], with a lot of names and resistant mutations.

I guess it's very hard for our community oncologists or general oncologists to know all of this. There is 1 platform where they mention the drug, but when you click there, you go to a link that includes the original study. You can not only educate yourself, but there is also corroboration that this is not a fluke. So if it's something very important or more serious that you can evaluate and take a more important suggestion in that case.

That's one of the complexities of the simple gene predicting for a response that ends up being the gist of the report. For example, as Shubham just mentioned, in colorectal cancer and I'm assuming in all cancers, if you have a 

 amplification, 2 things happen. One, you're not going to respond to the EGFR inhibitor. Two, you're not going to respond as well to the 

-targeted strategy that's essentially in the checkbox. That's unfortunate, because then at the end of the day it's contraindicated, meaning it's not going to work for the patient. You may have to go through a million hoops to get the patient on the study. I also mentioned the rare alterations that have a case report here and there. I'm not disagreeing with the fact that it's OK to have some recommendations that are there.

The problem is that many of those are very random, and they're probably generated by a computer listing that instructs, “If you have this, you go to this drug,” without looking at the global picture. That picture gets much more complicated now that we find a lot of these co-occurrences, some driver mutations, some mutations that actually drive resistance when you have a fusion or an amplification present.

I think the story is so complex now, it's much more complex, that I really want to have access to the gene. If I want to have access to understanding what all these co-alterations are doing together, I want a much more comprehensive type of report that tells me, “Wait a second, you shouldn't use this. Although the amplification is there, watch out for the mutation that's telling you that this will not work.”

We're all very used to if we get a funny radiology report, CT scan report, we'll either call up or go down to talk to the radiologist and say, “What do you mean with this?” I think we should treat these types of reports the same way and go to the source. We should apply good clinical judgment to the decision-making on the therapeutics.

John, I have one last point. Also these reports don’t make a distinction between tumor types. So as we know, if you target the 

 G12C mutation in lung cancer, you will see that it is very different than for colorectal cancer.

We know these differences, so it's really more of a push toward comprehensive NGS reports to get better on their algorithms they use to report this out. This is because it's very tumor-specific as well. So it's more of a challenge toward them. Obviously, it's very hard in the community to make sense of all this.

I think if they get their act together it might also help out a little bit more with the community oncologists getting the right therapeutics to the right patient. That's what it's about.

OK, we'll let that be the last word for that discussion.

Advice for Interpreting NGS Reports

Dr Pant, I want you to take us through this issue of when more than 1 target is found. As we talked about this, this is more common in the GI [gastrointestinal] cancer space. But how do you prioritize this? I'll be very interested to get some feedback from our lung cancer colleagues after your comments. How do you decide when you hit the gold mine, and you have a bunch of different targets to play with?

First, I wanted to add onto Tony's comment about when we are looking at somatic mutations. My patients say, “Of all these mutations, why can't you target it? I've got this thing called 

Some of them are drivers, but then you get on these NGS [next-generation sequencing] reports, and you get a whole host of passenger mutations, which are not the driver mutations. That's exactly how I explain it to my patients. You have the driver and you have the passengers. If you take out the passengers, that's not going to stop the bus. That's one of the ways to look at how to interpret these NGS reports because we get a lot of referrals about 

 mutations and everything in the early days of NTRK inhibitors when we had it in clinical trials.

 mutations, and there was nothing we could do about it. Some of the folks used to send us patients who were upset about it because the physician had talked it up on their 1 patient who had an 

 fusions, when staining studies have been done in pathology, about 25% of patients could be PD-L1 positive with 

 fusions, and a smaller amount with MSI [microsatellite instability] or MSI-high. I think the 

 fusion is such an oncogenic driver. I've seen in the early days some patients with sarcoma, who couldn't get out of bed with a PS [performance status] of 3. If you give them this NTRK inhibitor, the PS improves. It's fairly dramatic, there’s a difference in a couple of weeks, and I think Mark said this before. The PS improves for these patients, and they are dramatically improved. I think it just depends on the response patients get.

I have a patient who still comes to me from New Jersey, one of the first patients that I put on one of the NTRK inhibitors and had multiple lung lesions. The lesions disappeared, they just melted away like magic. This has been sustained now for 3+ years. I think in this context of 

 fusion, I'm going to go after it first with an NTRK inhibitor, because I think it's a very powerful oncogenic driver.

 fusion first, with an NTRK inhibitor first, yes.

All right. Any of the lung cancer specialists want to chime in? Do you guys ever see multiple targets at the same time? Mark?

It's very uncommon to rare. In the original lung cancer mutation consortium study that had about 800 adenocarcinomas in it, I believe it was 2% or 3% that had more than 1 mutation. Most of the co-mutations were more of PI3-kinase, 

, and so forth that at the time were not particularly targetable. So it's not a common occurrence in lung cancer.

Although, that definitely changes after treatment. It's an acquired resistance that develops. That's really where these co-mutations now have become more interesting, right, as we're starting to develop trials around them, but not at diagnosis.

Yes. My comments were restricted to the patients with driver mutations. Shub made this point, that you start looking at the average patient with lung cancer who has a 50-pack-year history of smoking, you send off to somewhere like Foundation Medicine for genomic testing, and 14 different alterations come back. It's a lot of noise on the report. Everyone doesn't know how to interpret it. At the end of the day you find a lot of mutations, I guess you'd call those co-mutations and such, but we just don't know what to do with it. It's like a genomic mess, so to speak. But in the driver populations, you tend to see the 1 driver that dominates.

I think it’s important because we're learning, as you said at the beginning, but I think until now we're assuming there is only 1 driver. That is why, for example in the lung community, if I remember, there is only a very small number of patients in Japan. They tested more than 3000 people, and it's a very small cohort of patients who may have 

 at the same time. I think the researchers want to give more drugs…but until now we are assuming there is 1 driver. When there is resistance, the driver may switch to another driver, but it's always 1.

John, I want to add in there that my colleagues and I, in the phase 1 unit at MD Anderson Cancer Center, have treated a number of these patients with sarcoma, secretory cancers, or some others such as one of the rare tumors, or the common ones. But as I said, the response is so dramatic with NTRK inhibitors when you have the 

 fusion. You see the tumor change and the patients’ PS is dramatically improved. If you inhibit that, it's really powerful.

Driver Mutations in NTRK Fusion-Positive Cancers

Dr Patel, Jyoti, tell me, for general oncologists, what do you want to do in lung cancer patients?

I think you really need multidisciplinary buy-in. You need good conversation with the pathologists. They need to understand the clinical relevance of these oncogenic drivers and why we prioritize them. One reason so many of us medical oncologists have gravitated toward blood-based testing is we control it. We can send a nurse to someone’s house to get the blood. The results come straight to us. We can look at the time flow and make it happen according to our needs.

With pathology it can be more cumbersome. Some institutions, such as ours, have been able to do reflex testing for all adenocarcinomas at diagnosis or all nonsquamous cancers at diagnosis and then special one-offs. That takes shared effort and decisions about what you can do in house and what needs to be sent out.

Initially there’s a new diagnosis at tumor board, and what does the general oncologist do? Talk to the pathologist. How soon can you get this tissue? This is because you may ask for a send-out lab, but it may sit in the lab—particularly during these crazy times—for a week before it even gets sent to the outside laboratory. Some type of workflow within your institution that makes sure these happen quickly is absolutely essential.

You work at Northwestern University, right? Are the pathologists giving you trouble there?

No, it’s been great. We’ve been able to tie in a shared model of reflex testing, and they help us control tissue. We have shown, along with multiple others, that if you have that, it shortens the time to treatment initiation to getting results.

I have a comment to complement what Jyoti is saying. You’re talking about our institutions where we have certain authority to ask the pathologists that can be your friend, “Please send the tissue.” But remember, all of us are at referral centers. A lot of the patients that come to our centers, because they’re very large centers, come from a hospital from some city around you. You don’t have any authority when you request to have the tissue delivered where you want to send the test.

That’s the main problem—the referral hospital wants to send the tissue 3 weeks from now. There is no way around, to call them and ask a favor to do it. That’s a major problem that we have, and that’s why in plasma it’s very helpful sometimes because we can send it right away.

 I agree with Luis. That’s part of the reason why there’s a particular vendor that will go directly to the source of where the tissue is. I call them the golden retriever of molecular testing vendors. You just throw the tennis ball to that hospital, they go get the tennis ball, and it just takes us out of the loop. They still have trouble getting it quickly, but at least they take us out of the loop.

Yeah, there was an interesting time when I sent a test on a patient, and he was a physician. It was his own institution where his tissue was, and his pathologist wouldn’t release the tissue. I had to call the pathologist, and they sent the block to me. We don’t want to get in the weeds of that. It really is a barrier to getting access and getting it sent out.

Tony, let me ask you a hard question about this, because you love pancreatic cancer—or hate it, I should say. It’s not uncommon for us to have a very small amount, like the lung cancer folks get. What’s it worth doing to get tissue enough to send for broad testing? Do you go for a repeat biopsy? Do you try to use circulating markers? Tony, what are you doing in pancreatic cancer when you don’t have enough tissue?

 This is a very particular challenge in this type of cancer. The problem is, the liquid piece is even more challenging with pancreatic cancer than a lot of cancers. These tend to be low-volume types of cancer, and they shed very low DNA. You do find them, but they shed very low DNA. The return is actually not as powerful. That said, we’ve used it, and many times we find what you would expect—no alterations because you can’t find the DNA.

I actually tried everything possible to have a tissue biopsy on these patients. First, we trained our gastroenterologists. We said most of these patients may come with the tissue already, or whatever tissue they’ve got. But for our patients, even in the locally advanced where you typically have endoscopic ultrasound, you have very low tissue. We actually have trained them to try to get a larger core, like a 19-gauge needle rather than a 22-gauge or a 25-gauge—whatever they use—to give us enough to be able to run a test.

For those with liver lesions—again, most of those patients with metastatic disease will have liver lesions if they just had an endoscopic ultrasound with scant tissue—I’m doing a biopsy on the liver. I think it’s justifiable. One may argue with pancreatic cancer that most of these patients will have nothing. But for the 10% to 15% that will have something, it could be quite significant. It could be an 

 fusion, it could be MSI [microsatellite] high—those are 0.5% to 1.0%. But you know what? We find them, and they have remarkable responses.

 [human epidermal growth factor 2], we’re finding 

 fusions, which certainly in the lung world as well are becoming relevant in our trials, as well as RAS B12 and then 

. So these are little pieces—a needle here, a needle there. Even if you test germline in the negative, you still have a 50/50 chance if they have a 

 will be somatic, not germline, and those actually respond equally well to the PARP inhibitors.

If you put all this together, there’s about 10% to 15%, so it’s not a majority. But for a disease that does incredibly bad, these are important because they help us and direct us in terms of treatment. I’m very aggressive trying to get a biopsy on those patients in whichever way I can.

 It comes back to Jyoti’s comment about this team multidisciplinary approach. I bet every one of us would nod that we have a deal with interventional radiology. If they’re down there for a tissue, not just for a diagnosis but for profiling, it’s really changed the SOP [standard operating procedure] in many ways, referrals for tissue acquisition. I think it’s just critical for us all to do that.

Genomic Testing for NTRK Fusions in Lung and GI Cancers

Luis, Dr Raez, let me let you take the lead on this in terms of the mechanics of testing. We’ve all hinted at it—some do it internally, some are sending it out. Can you give us a sense of how one can get 

We have several options, as somebody else mentioned. I think it was Jyoti. For example, we can do RT-PCR [reverse transcription polymerase chain reaction]. We can do RT-PCR to find the fusions. We can do FISH [fluorescence in situ hybridization] testing to find the fusions. We can do DNA sequencing, and lately we’re doing RNA sequencing. So now we have 4 options. How we deal with this, and what is the best approach? It actually depends. ESMO [European Society for Medical Oncology] has a recommendation published last year about this, but these are common sense.

For example, if this is a tumor that is very uncommon—lung cancer, colon cancer, breast cancer—they have a very low prevalence of 

 fusions. The best thing we can do is to do the whole sequence, the whole DNA sequence. We also have discovered in the last 3 or 4 years doing that for 

 that the DNA sequence may not be as complete or comprehensive as the transcriptome sequencing, the RNA sequencing.

That’s why we have platforms at our center that run both, the DNA sequencing and then the RNA sequencing, to touch any possibility or any fusion that has not been discovered. That’s for uncommon tumors. When you have 1 of these tumors that are very rich in 

, like mammalian—salivary cancer, secretory carcinoma of the saliva, secretory cancers of the breast, congenital lymphomas in kids or fibrosarcomas—you know that the prevalence of 

 is very high, maybe 90%. So you can shoot straight, you can do RT-PCR, you can do FISH. For example, the salivary cancers are very high in 

, so you can go straight and try to find that genetic aberration. That way maybe you don’t have to do a whole sequencing.

The last option we have—option No. 3, which not very common—we used to do before. We used to love FISH testing. Oncologists love FISH testing for breast cancer, but when we started to do lung cancer, we learned about that FISH is good for 

. Then we discovered that maybe FISH is good for ROS. But it’s too cumbersome; there are too many FISH testings now. We have 

That’s why the developing of immunohistochemistry is a very good option, because the problem is the amount of use of the tissue, that there is not enough tissue to look for other genetic aberrations. But some people have been successful publishing data; that antibody for 

 can probably catch more than 90% of genetic aberrations. Maybe that’s an option for a center that doesn’t have the technology for RNA sequencing.

Yeah, but what you just described is complicated. I can barely keep track of how to get FOLFOX [folinic acid, 5-flourouracil, oxaliplatin].

Approaching Genomic Testing for NTRK Fusions

The lung cancer specialists have been consistent, right from the beginning they want to know everything. Shub, from a GI [gastrointestinal] cancer world, we’ve gotten a little slower with this. What’s your current practice on when you’re doing testing, such as next-generation sequencing, plasma samples, and so forth, for 

I think since I was in fellowship. Of course not as long as Mark.

 Sorry, Mark. At that time it was still the carbo-taxol [carboplatin-paclitaxel] days, but this pie chart was coming out. In all the GI meetings we go to, we always look at the lung cancer pie chart. We used to look at breast cancer when I was in fellowship. That changed to the lung pie chart, and we look at it longingly to see if we can bring the same things to patients with colorectal cancer and with pancreatic cancer. But it’s growing. Especially in the colorectal space where we have now with the 

Mark A. Socinski, MD, AdventHealth Cancer Institute

Articles