The Development of TRK Inhibitors in Cancer

How a single patient and serendipity led to a novel FDA approval


Presented by Robert Doebele

About the Speaker

Robert Doebele is Associate Professor of Medicine in the Division of Medical Oncology at the University of Colorado School of Medicine. Dr. Doebele earned his degree in molecular biology from Princeton University and earned his MD/PhD from the University of Pennsylvania School of Medicine. He was trained in medical oncology at the University of Chicago. Dr. Doebele's laboratory studies oncogene-addicted cancers to identify novel therapeutic approaches as well as intrinsic and acquired mechanisms of resistance to targeted therapies. Dr. Doebele was the recipient of the 2013 V Scholar Award from the V foundation for Cancer Research and was elected to the American Society of Clinical Investigation in 2018. He is currently the Director of the Thoracic Oncology Research Initiative at the University of Colorado Cancer Center. As an NTRK expert who has been part of the Trk testing and targeted therapy since its inception, Dr. Doebele delved into the history and development of Trk inhibitors in cancer in this webinar.

Webinar notes  


Recent advance of targeted therapy - The targeted therapies have evolved in cancer over the decades. I think that we've obviously had a large number of successes in various oncogene-driven cancers, for example, trastuzumab for breast cancer, erlotinib for EGFR-mutated NSCLC, imatinib for CML and BRAF and MEK inhibitors for melanoma. All the above drugs were developed in a tumor or histology-specific manner. And we had thought of many of these oncogenes as belonging to one type of tumor. Once we thought of EGFR mutations, we associated lung adenocarcinoma. Once we thought of HER2 amplification, we associated breast cancer, although obviously that is changing now. Once we thought of ABL fusions, we associated CML. Once we thought of BRAF V600E, we associated melanoma. The experience had shaped tumor-specific approval mechanism that drugs only worked in one type of setting.

Further to that, we started this journey where oncogenes in different tumor types may not behave the same based on early data. It was around the time that we started this Trk inhibitor development story. We found that BRAF V600E mutations in melanoma had a dramatic objective response rate as well as durability, especially when combined with a MEK inhibitor response rates approaching nearly 70%.[1] However, BRAF 600E mutations, the same oncogenic mutation, in a different cancer type, colorectal cancer, had a response rate of only 5%.[2] In this sense, a tumor agnostic approach may not be a successful way to go. The tumor agnostic approach was a breakthrough under such background in the past decade. 

The tumors used to have a therapeutic algorithm, lung, breast, thyroid, etc. Many of our tumor specific therapies have been based on a number of factors, such as anatomic location empiric data over the years in terms of which chemotherapies or other therapies worked, also based on biology. There are some biological signatures or predictive markers that predict a cancer is going to be addicted to a certain oncogene. The tumor agnostic approach came from biology that eventually drove the development of these Trk inhibitors. We've developed these therapies for various specific tumor types.

The origin - Before we reached the advance today, the story started with a single patient. This was a patient back in 2012. She was a 46-year old woman, a never smoker like many of lung adenocarcinomas are, having multiple prior lines of therapy including all standard therapies at that time.  When she reached Dr. Doebele, her tumor had tested negative for EGFR. ALK, ROS1, which were druggable as well as some of the newer ones that were being developed at the time, such as RET, BRAF, KRAS, HER2 and others.  It was frustrating an oncologist speaking of a never smoker. Then Dr. Doebele figured there may still yet be oncogenes that were unidentified in non- small cell lung cancer. He looked at a cohort of patients, who were pan-negative for known oncogene in lung adenocarcinoma, presuming that this would be the right type of setting to find new oncogenes. Together with Vaishnavi at Dana-Farber, Cancer institute, they collected a total of 35 cases that were negative for most of these known oncogenes, and submitted them in collaboration with Foundation Medicine, to potentially identify new oncogenes that were targetable. They found two putative oncogenes involved in NTRK1 gene fusion.[3]

NTRK genes are a family. NTRK1 gene has two other family members, NTRK2/3. These three NTRK genes encode receptor tyrosine kinase unlike EGFR or HER2 or others and those are called TrkA, TrkB, and TrkC. They are predominantly expressed in the nervous system. They typically require neurotrophins to activate them and their normal role is for neural outgrowth, pain and sensory modulation. For example, mutation loss or function mutations of NTRK1 cause congenital insensitivity to pain with anhidrosis patients who cannot sense pain, which is quite dangerous.  The NTRK2 regulates obesity and simulate cerebral development. These physiological features are associated with side effects of Trk inhibitors.  

Dr. Doebele was not taking credit for discovering these oncogenes.  A Nature paper, dated back to 1982, found this onc D oncogene and that was one of the earliest detections of an NTRK gene fusion. It also involved a portion of the triple myosin genes along with this NTRK1. The researchers did not know what they were looking at. It looked like a receptor kinase, but it had sequences that look like it was from another gene. The same paper also discovered onc B in these tumors which turned out to be KRAS. [5] This gave us an idea how old these genes are.

What are oncogene fusions? They follow the paradigm like BCR-ABL, ALK, ROS1 fusions. Take NTRK1 gene fusion as an example, it always includes the kinase domain which is the fixed end of the molecule. The five-prime end of the gene is far more variable. The two NTRK1 genes, identified by Dr. Doebele, had different five-prime gene partners. One was MPRIP, and the other was CD74. [3] He leveraged next generation sequencing sequence to cross the junction of these novel gene fusions and initially identified them.

Next, Dr Doebele talked about carcinogenesis of NTRK genes and he again used TrkA as an example.  In the normal tissue, once the growth factor ligand binds, the transmembrane receptor dimerizes, followed by a signaling cascade and eventually neuron development as an outcome. These are the same pathways that cancer cells like to use. In contrast, the cancerous tissue loses that ligand dependence and these genes become turned on permanently such that they always are signaling, resulting in formation of cancer.

The first step was to verify whether these fusions were oncogenic. In brief, Dr Doebele showed that these novel genes could be expressed, auto activated, in the absence of ligand.  They caused cell proliferation when they were put into cells, and they could cause tumors in mice. [3] They were comparable to some of the other gene fusions that we had studied previously such as EML4, ALK, or ROS1 fusions. Taken together, the evidences suggested the NTRK genes be indeed oncogenic. From our history of being able to treat ALK, ROS1 and ABL fusions, NTRK fusion would be a potentially druggable target.

The next step was to identify a drug that could work on NTRK genes. In fact, this was a real-time exercise for this female lung adenocarcinoma patient. She had exhausted most of our other therapies. Dr Doebele was looking desperately to find a drug that might have activity against TrkA for this patient. In the end, unfortunately, the patient was on crizotinib, which was a very weak Trk inhibitor. Crizotinib was not recommended to be a Trk therapy. That time was very early on before we could hunt extensively across the world for a Trk inhibitor. Another drug, CEP-701, unfortunately had just failed its AML trial. It also was a multi-kinase inhibitor with Trk activity. Here’s where the serendipity came in. Curtis Davis, the talented post doc now faculty member in Dr Doebele’s lab, turned out to do prior post doc work at Array Biopharma before. Array, located right in Boulder Colorado, was working on Trk program mainly slated for arthritis at that time because TrkA regulates pain. Thus, Dr Doebele did not have to look very far across the world.  Array had a number of Trk inhibitors including ARRY-470, which is now better known as LOXO-101 or Larotrectinib. This drug had very high potency against all three Trk receptors, Trk A/B/C. It was very highly selected, meaning that it might not cause many side effects. Even 1000-fold above where we inhibited track A/B /C, it was not significantly hitting other kinase targets.[6] Again, larotrectinib was very potent against Trk but not against other receptor tyrosine kinases or other kinases.

Preclinical data - To show proof of concept, Dr Doebele demonstrated in immunoblot analysis of lung cancer cell line that larotrectinib or ARRY-470, at very low doses, inhibited both the Trk kinase domain and downstream canonical signaling pathways like MAP kinase and AKT. It could inhibit cells expressing NTRK fusion at very low IC50, less than 10 nM. CEP-701, a very selective track inhibitor, also had very high potency. The other TKI, gefitinib, which was mainly an EGFR inhibitor, had very low potency. [3] The early data suggested NTRK gene was a very druggable target. Most of those were shown in engineered cells. Thanks to gracious donation of tumor specimens from his patient, Dr Doebele was fortunate to derive CUTO3 cell lines, which expressed an MPRIP and NTRK1 fusion. It was a human derived cell line from pleural effusion originated from lung adenocarcinoma. It also was inhibited at very low doses, <100 nM doses with LOXO-101 or larotrectinib. Importantly, this drug was not very potent against other cell lines that had ROS or other alterations. Even at 1000 nM it did not cause inhibition of cell growth to any degree in these other cell lines. This was an important feature that it was not indiscriminately killing cells.[7] Finally, the xenograft mice with human-derived cell line was fed with LOXO-101 or Larotrectinib. It demonstrated significant tumor regression at either dose of 60 mg/kg or 200 mg/kg. There was no obvious toxicity in mouse models. Afterwards when Dr Doebele started treating patients and noticing some side effects, he did go back and noticed that there was some anecdotal weight gain in the mice. These were very early, promising data in this lung cancer model that led to develop drugs in patients.

The tumor agnostic approach - Up to this point, only NTRK1 fusion has been discussed. There are two other family members, NTRK2/3. Those are encoded protein kinase, Trk A/B/C. The intracellular kinase domains across gene family are highly homologous. The residues interact with either ATP or Trk inhibitors. It suggests if Trk inhibitors worked against TrkA, they would also work against TrkB/C fusions. The NTR2 fusions were quite rare, according to early data. Although the researchers were not obtaining human cell lines that expressed TrkB, in an engineered cell LOXO-101 was nearly equally potent against NTRK2 compared to NTRK1 fusion. [8] That stood true in cell-free kinase and these cell-based assay. The mice data also showed there was good tumor growth inhibition or tumor regression in NTRK3 expressing model. [7] The key was whether the tumor harbored NTRK1/2/3 fusion, regardless of the variety of five prime gene partners. In addition, the xenograft mice of lung adenocarcinoma, colorectal cancer, acute myeloid leukemia, once harboring the NTRK fusion, showed comparable results. Based on these preclinical data, we should combine all these rare patients in one cohort. Be lumpers rather than splitters. The tumor agnostic approach therefore was guiding the clinical development later.

Let’s go back to treating BRAF V600E mutations, the early data on target therapy may be misleading. First of all, the responses was improved when combined with an EGFR monoclonal antibody.  The paradigm had gotten a recent FDA approval. If we looked around BRAF V600E mutations further across other tumor types, the target therapy worked well in general. The activity was good in lung cancer [9], even hairy cell leukemia [10], anaplastic thyroid cancer [11] and Erdheim–Chester disease [12]. The tumor agnostic approach was working more often than it was not. The colorectal cancer was an exception. There were still many potentials for tumor agnostic development. People were excited about KRAS G12C mutations and looked forward to the target therapy. We are anticipating more approvals like this in the coming years.

The clinical development of Trk inhibitors - The larotrectinib program launched a Phase I clinical trial. At the dose escalation stage, there was biomarker unselected. The NTRK fusion patients or other NTRK alterations were hoped to enroll as to showing proof of concept. Any tumor histology was allowed. The pharmacokinetics of drug looked quite good. The Cohort 3, the 100 mg twice daily, was selected as the recommended Phase II dose. Cmax was high, 2 mM or 2100 nM. Dr Doebele treated his second NTRK patient in the study. It was by random chance. The patient happened to be a 46-year old female. She had an undifferentiated sarcoma of the left thigh. Her demographics were similar to those of first NTRK patient. She underwent resection in her left thigh and then rapidly progressed with metastasis to the lung. She was not responding Adriamycin and had rapid progression. During this time, the resection specimen was sent off for next generation sequencing and NTRK1 fusion was identified. The patient, who had NTRK1 fusion, was the very first patient of the Phase I trial. She was not from Colorado but traveled to Colorado with her family to enroll the trial. She was at the recommended Phase II dose. In this case, NTRK fusion was not restricted by cancer type.

A variety of tumors have been reported to harbor NTRK fusions, such as cholangiocarcinoma, colorectal, leukemias and others. The frequency varied across tumor types. These NTRK fusions were relatively rare in the common tumors, like lung and colorectal cancer, but they were very common in some rare tumors. The secretary breast carcinoma typically affected adolescents. The mammary analogue secretary carcinoma (MASC) was a rare salivary gland tumor. The NTRK fusions were prevailing in 100% of these two tumors.  Some pediatric tumors like congenital fibrosarcoma are associated with NTRK fusions too. [13]

The hope was that this therapy would work across all these different tumor types, which had same oncogene. Finding these patients might be a challenge except for these rare tumors appeared to be pathognomonic. The second patient had undifferentiated sarcoma with LMNA-NTRK1 gene fusion. Her CT scan at baseline showed multiple lung metastases. The radiologist commented that they were too enumerable, and the patient had significant respiratory difficulty. Dr Doebele was pleasantly surprised as she moved along therapy. Her first scan at two months showed a remarkable reduction in the tumor burden in the lungs. At C9D1, she's approached a near complete response. She agreed to do patient story, patient advocacy.  Being able to treat the patient was very fortunate especially some of the work that we've done in the lab turned out to apply this to a patient. This patient started treatment in March of 2015. Those scans only went out through C9D1. [7] Afterwards, the yearly updates were received. She's now five years disease-free on larotrectinib. Dr Doebele was sad to miss her last year. She was in Denver for a patient advocacy conference and she stopped by clinic and left a very kind note on his desk. This patient had proven the concept that this drug was going to be a very useful one. There were six NTRK fusion positive patient enrolled in the Phase I trial. There was very significant activity with objective response rate of 100%.

Dr Doebele ‘s second patient was still on therapy and approaching five years. She had rapid onset to respond. The first scan was at C3D1, which was approximately two months.  The time to first response was 1.9 months, showing that this was a very rapid onset. To manage a patient with a heavy disease burden, it is important to have these tests in hand and believe the target therapy for these patients work very quickly. There were durable responses. In a small number of patients, the duration of response was not reached. The confidence intervals ranged from eight to 33 months. [14] These are early hints that the treatment was durable, even based on the first few patients in Phase I trial. There were five different tumor types, sarcoma, lung and others. Again, the early clinical data proved that the tumor agnostic strategy was going to work because we were seeing activity across different tumor histology and different types of NTRK fusions, even from the first few patients.

Following success of Phase I trial, larotrectinib moved rapidly into Phase II trial at the dose of 100 mg BID. Based on Phase I data, it was meant to be biomarker selected for NTRK1/2/3 gene fusions by CLIA-certified laboratory tests. There was still no FDA approved companion diagnostic, but that did not mean that there was no good test to detect these alterations. Any tumor histology was allowed here, as well as pediatric patients. The registration data set published in the New England Journal of Medicine. [15] Recently, the study update was published as well. The 14 different tumor types and 17 different histology showed tumor regressions with the objective response rate of 79% by investigator review. The complete response was 16%. There was good disease control overall. The response to latrotrectinib was durable where PFS was 28.3 months. [16] The durability was consistent with the observations in Dr Doebele’s many patients in the same study. Of note, NTRK2 fusions were relatively rare among the clinical patients. They were more difficult to detect because the genes were so large that caused some issues around coverage on assays of whether they were well covered or not. As predicted in preclinical data, NTRK2 patients were responding similarly compared with NTRK1/3. The findings further supported that the three NTRK genes should be seen as a single-family target for Trk inhibitors. The variable five prime partner genes were less relevant to prognosis, according to post hoc analysis. [13][15] In other words, NTRK genes themselves were essential and we just had to make sure the in-frame encoded kinase domains were included in the assay report.

Dr Doebele have had a few patients on larotrectinib for four or five years. The larotrectinib was incredibly tolerated, promising for long-term consumption. The dizziness and increased body weight were noted.[15] TrkB was related to regulating obesity, although there's controversy about how it worked and that whether it was a metabolic effect.  While some patients gained weight, some did not. We did not fully understand the mechanism behind this. The weight gain was thought to be an on-target effect of inhibiting TrkB. The patients also can get dizziness or imbalance, which was variably documented across different clinical trials. TrkB was known to play a role in cerebral development. From Doebele’s studies long ago, Trk was thought to affect dizziness. The arthralgias were frequent upon dose interruptions or dose holds. [15] If stopping the treatment abruptly, the patients might experience kind of a rebound pain of arthralgias sometimes described as myalgias in various body areas while the large joints like the knees were most affected. Some patients went to emergency room. When dose hold is prescribed, it is recommended to warn in patients of these pains and take p preventative measures in advance. The pain went away within a couple days following the half-life of the drug. It was not a permanent side effect. The pains possibly came back if re-challenging the drug, either at the same dose or a lower dose. It was not worrisome, however.

The entrectinib was the other Trk inhibitors now FDA approved. The entrectinib had showed very similar activity in tumor types such as cholangiocarcinoma, thyroid long, etc. The objective response rate was 57%. The study looked specifically at NTRK1 versus NTRK3 and found the response rates were nearly identical. There was only one NTRK2 patient in this cohort. A small subset of patients were not responding. These Trk inhibitors were not perfect, but they did have very similar activity across multiple types of fusions. The progression free survival was 11 months. [16] It was not possible to do cross trial comparisons especially looking at more than a dozen different tumor types with different proportions. This study, in contrast to the larotrectinib study did not include pediatric patients.  The pediatric population was studied in the separate study. The entrectinib was designed as a CNS penetrating drug. More than a fifth of the patients enrolled in this cohort had CNS or brain metastases.

Brain metastasis and primary CNS tumor - Larotrectinib has been shown to cross the blood–brain barrier. [20][21] Larotrectinib was active within the CNS based on reported intracranial responses.[20][21][22] The post-hoc exploratory analyses were performed against subsets of brain metastases, or primary CNS tumors. Out of 13 brain metastatic patients with lung, thyroid, breast, melanoma tumors, the overall response was 75%, which was comparable to outcome of those without brain metastases before enrollment. [16] An ORR of 36% was achieved with larotrectinib in patients with primary brain tumors where 14 patients were evaluable for response. After a median follow up of 4.4 months, median PFS for primary brain tumor patients (n=18) was 11.0 months.

Since brain metastases have represented a source of morbidity and mortality in fusion-positive NSCLC patients, it was clinically meaningful to develop therapeutics which was predictive of CNS activity. The ALK inhibitor, alectinib, was CNS penetrating while crizotinib was not. Let’s look at progression free survival comparing these two drugs. The poorer PFS of crizotinib was probably attributed to its higher cumulative incidence of CNS progression than alectinib. [18] The treatment outcome of osimertinib verses erotinib was alike. These examples alluded CNS penetrant drugs could delay progression due to suppression or treatment of brain metastases. The CNS-penetrating entrectinib was designed as not to be a Pgp substrate. Dr. Farrago at Massachusetts General Hospital published a case report. A patient with stage IV lung adenocrcinoma with NTRK1 fusion was treated with entrectinib. Restaging CT scans demonstrated a RECIST partial response (PR) and complete resolution of all brain metastases. [19]

The integrated analysis of Phase I/II studies of entrectinib showed the intracranial objective response rate of entrectinib was 55% in relatively small cohort of patients, very similar to the systemic response. This was one of the endpoints of the clinical trial because entrectinib was designed to be a CNS-penetrating inhibitor. [17]

Pediatric NTRK patients- Aside from adult tumors, the larotrectinib had an objective response rate of 92% in NTRK altered pediatric cancers. [16] Some non- NTRK fusion positive cancers were included but they were not responding.[15] This was a very biomarker driven indication. All the larotrectinib responsive pediatric patient had showed tumor reduction. They were primarily infantile fibrosarcoma, soft tissue sarcomas, Congenital mesoblastic nephroma. [16]

Entrectinib also initiated a pediatric study which encompassed ALK/ROS1/ NTRK alterations. The objective response rate for the NTRK fusions was 100%, although two of those were unconfirmed at the time of this presentation. Similarly, the pediatric tumors without presenting a fusion did not respond. [23] We could not emphasize more about the importance of identifying an NTRK fusion gene for Trk inhibiting therapy.

To wrap up, Dr Doebele’s work on NTRK started around 2012-2013, and he first published the NTRK fusion in lung cancer. Then ta number of new cancers were identified to harbor NTRK. The Phase I trial started in late 2014. He treated the first patient in 2015 and saw a response. Only a few years later, FDA granted approval to larotrectinib. It was a relatively rapid drug development. If we looked back from long historical perspective, onc D or NTRK1 fusions were first identified back in 1982 and they just sat there waiting until 2013.  In terms of drug development, it was incredibly slow.

Larotrectinib was the first oncogene targeted therapy to achieve a tumor agnostic FDA approval. Although pembrolizumab was the first drug to get a tumor agnostic approval, larotrectinib had many first’s- the first oncogene-driven approval, the first drug to have an indication across multiple oncogenes, and the first drug to have pediatric and adult indication. As a landmark, larotrectinib was making the lives of NTRK fusion patient better.


NTRK testing - For detecting NTRK fusion genes, Dr Doebele was fortunate to consult Dr. Aisner, the expert molecular pathologist and pathology lab at University of Colorado. There were many ways to potentially detect NTRK oncogenes. The capture based NGS was based on DNA.[3][7] Foundation Medicine was doing this. The anchored multiplex PCR(AMP) was RNA-based. [19] An example was the Archer Fusion Plex Assay. The fluorescence in situ hybridization or FISH was possible but was tending to fall out of favorite just because it detected typically one gene at a time while the oncologists preferred to look for different oncogene simultaneously. The early study using FISH technique had found irregularities particularly false positives, including copy number alterations and false positive rearrangements. [3][7] The RT-PCR was possible but that was biased towards the prior knowledge of five prime fusion partners. It was challenging as novel five prime partner were being identified over time. The tests usually lagged the current knowledge of partners.[3][7] Finally, immunohistochemistry (IHC) has been recommended by some experts as a screening tool, at least as a first step. It however was not a diagnostic but a screening tool. Since the protein expression was not very specific, IHC required a reflex validation test, which could be NGS mostly recommended. Overall, the NGS assay regardless of DNA or RNA were recommended for NTRK detection.

Target NTRK fusions - There were many types of gene alterations and these were confusing on genetic reports from diagnostic companies to companies. Apart from gene fusions, there were also point mutations, small insertions or deletions, gene amplifications, low or high copy number, etc. The early study had verified the activity of Trk inhibitors against a variety of gene altered patients. It turned out only gene fusion was something to aim at. Both larotrectinib and entrectinib were reported to work on gene fusions. [13]

More recently, Dr Hong presented additional data on the Phase I study including patients who had NTRK mutations. Although there was no response noted in mutated patients, they continued to explore. The hematologic malignant patient with NTRK mutations was reported to respond. For gene amplification, again, Dr. Hong was aware of one partial response that was very short-lived in a patient with gene amplification. [25] The FDA approved indication of Trk inhibitors was only NTRK1/2/3 gene fusions. We would not recommend other alterations outside of a clinical trial.

The acquired resistance following Trk inhibitors- Lovely and Shaw, demonstrated the multitude of ways that cancer cells became resistant to targets. [26]

The target modification was well studies by ctDNA assays or by tumor sequencing. Relatively straightforward, the therapeutic approach could overcome that. The drugs were designed to hit those specific kinase domain mutations. Dr Doebele anticipated this when he started treating patients with Larotrectinib. Consequently, he did some preclinical work, trying to identify the putative mutations, using mutagenesis strategies with larotrectinib. Using in-vitro model, he had predicted a lot of mutations in patients with NTRK fusions. [27] [28]

The findings were nearly at the same time when the initial New England Journal of Medicine report came out. Those mutations predicted in-vitro turned out to occur in clinical tumor samples out of clinical trials or via ctDNA. [15]

Similar with entrectinib, the mutation rate was at about a third. For mutations in ALK or other fusions, the rate was comparable. In the entrectinib study, ten out of the 29 NTRK fused patients, evaluated by ctDNA, had shown resistance mutations. They were all solvent front mutations. But there was also bypass signaling resistance. [29]

Fortunately, there were two drugs being in development, TPX-005, (repotrectinib) and LOXO- 195(selitrectinib). Both drugs tackled with solvent front mutations of NTRK3 and NTRK1. [30][31] Then Dr. Hyman presented a larger cohort of patients with acquired resistance to larotrectinib. LOXO-195 showed good activity against the solvent front mutations, some activity against gatekeeper and other mutations. They identified a bypass signaling pathway, RAS or elsewhere where LOXO-195 did not have an effect. [31] Those having a secondary mutation mainly benefited from the drug. It was biomarker driven all the way. At the start, we supposedly treated fusion patients with these Trk inhibitors. Upon resistance, a secondary resistance mutation in the kinase domain, such as a solvent front mutation, would be amenable to enrollment in a clinical trial of either repotretinib or selitrectinib.


  • NTRK1/2/3 (TRKA/B/C) fusions are oncogenic and can be considered a single oncogene family

  • In vitro and in vivo NTRK fusion models demonstrate pre-clinical activity of TRK inhibitors

  • Significant clinical benefit using TRK inhibitors for patients with NTRK gene fusion positive tumors

  • TRK inhibitors are effective regardless of tumor histology Drug resistance can be mediated by kinase domain mutations in vitro and in patients and can be overcome by next generation inhibitors

  • Larotrectinib is the first oncogene-targeted therapy to achieve tumor agnostic FDA approval

Questions and Discussion

Are we generally still seeing the strongest responses of NTRK fusions in H&N and lung?  good but not as durable responses in CRC and GBM tumors? Any trends here?

I was alluding that the Trk inhibitors worked regardless of tumor histology. One of challenges was that the number of individual tumor types is relatively small. There have been signals that colorectal and perhaps glioblastomas may not have remarkable effect. Remember the example of target therapy against BRAF V600E mutation. Not all the responses were created equal. Thus, the colorectal cancer was a standout as a non-responder while other cancers responded well. The hypothesis stood that colorectal cancers were more associated with bypass signaling pathways that interrupted the activity. My recommendation still would be to treat these patients and perhaps watch them more closely. To treat GBM, the treatment should demonstrate blood brain barrier penetration. We will eagerly await additional data on these. I was not familiar with the head and neck squamous cell cancers that were responding to Trk inhibitors. We're going to need larger patient numbers to see whether one out of three was a holds-up or it was just by chance. Investigating further the individual tumor was important just in the same way we looked at 90-100% responses in pediatric tumors. There may be some subtle differences that emerge over the next coming years with accumulated dataset of patients.

You mentioned the good response in pediatric tumors. Do you feel that testing for fusions and targeted therapies may be the key for pediatric?  I know that's not your specialty, but I know that you work with colleagues who do.

When it comes down to certain subsets of cancer, it's critically important to be use broad comprehensive next generation sequencing. The pediatric cancers are one example. Apart from NTRK, the entrectinib was targeting ROS1/ALK in pediatric cancers. The lung cancer, especially lung adenocarcinoma, was the other place where a number of alterations made nearly all the patients eligible for potential target therapy, whether FDA approved or off-label. Back to pediatric cancers, fusion specific testing is important. There were other types of fusions like MET, RET. The panel testing enables the patients to have best available therapy.

There have been a few cases reporting that NTRK fusions co-exist with other real driver mutations. Regardless, does the general rule still hold to look for NTRK fusions where there are no other driver mutations?

The exclusivity has been long observed in lung adenocarcinoma. Some people thought that ALK fusions were just 100%, mutually exclusive from other alterations, and same with EGFR mutations. In fact, we were one of the first to describe a patient with a dual KRAS and ALK. And that patient did very poorly on an ALK inhibitor, I think the rule still holds that they tend to be mutually exclusive. It makes sense to treat patients with dual drivers by an inhibitor but watch that patient more closely. Fortunately, the majority were found exclusive mutated. If one oncogene was known in a cancer patient, we need not keep looking. I think that is likely true for other tumor types as well. Now, the confusion comes around as to what an oncogene is. There are lots of mutations in cancers. I adhere to a very strict definition of a driver oncogene. This is an area full of debate and argument. For example, TP53 is not an oncogene. A lot of cancers that harbor NTRK fusions, EGFR mutations, or ALK are also going to harbor TP53 mutation. TP53 should not stop you from treating that patient with the appropriate oncogene directed therapy. The genetic reports are sometimes confusing. There will be an NTRK fusion coupled with six, eight or a dozen other mutations, but rarely do those other mutations impact therapy. TP53 is the most common one, but it should not stop you from treating these patients with TrK inhibitors.

So then to follow up along those same lines of confusion, do you find that there's still confusion between fusions and mutations? And where do you find that and how do we mitigate that?

I do. Having worked in this field for a while, I commonly came across questions from colleagues and emails about this. When we were excited to see NTRK on the report, we had to pay very close attention. Most of the commercial vendors that were doing this testing and reporting were very cautious to note that a mutation was classified as a variant of unknown significance, whereas a fusion should have then led you to a potential treatment page either with the FDA approved therapies or for clinical trials. Dr. Hong presented the data that none of mutations were amenable to therapy. That did not mean that there would never be one who was responding we might find eventually. But currently, there's really no data supporting the treatment of mutations with Trk inhibitors. The most commercial testing vendors are ought to help people interpret confusing report.

I wanted to shift to resistance. Do you think there will be any role for combination therapy approach to prolong the time to resistance with Trk inhibitors?

This has been a relatively slow development process when we got to understand other types of resistance may occur through bypass signaling more than a decade ago in lung cancer and other cancers. Further, the development process for combination therapies has always been a bit slower. This is a place where the tumor agnostic may break down. The drugs that you need to co-inhibit with NTRK and colorectal cancer may be different from lung or pancreatic cancer, etc. There is a role for looking at combination therapies up front. The combination with chemotherapy was empirical but the progress was slow. I would highlight big success stories of adding of MEK inhibitors to vemurafenib and dabrafenib to treat BRAF V600E mutations. These were positive examples where addition of a second drug prolonged time to resistance meaningfully and also improved the upfront response rate. There are a lot of studies going investigating combinations among MEK inhibitor, SHIP2 inhibitors, other therapies. We are waiting which evolves to be the best strategy.

Are NTRK patients immunologically cold generally?

The response often varies by tumor type. In lung cancer, these gene altered patients tended to be never smokers and respond very poorly in general to immune therapies, which were characterized by cold tumors. In colorectal cancer, the fusion patients were MSI high, which as indicated for immunotherapy too. Another advantage of doing comprehensive next generation sequencing is evaluating the tumor mutation burden, which is however controversial in terms of how predictive it is.

Speaking of all of these advances in targeted therapy and molecular screening, I was wondering if you could speak just a little bit about the oncologist-pathologists relationship, at least from your perspective? How did it evolve with all of these changes, and if there are any challenges that remain?

There has been always challenge, I think now more than ever. We're very fortunate at University of Colorado to have a world-class molecular pathologist, Dr. Aisner. Given that I do lung cancers, we're highly reliant on those molecular reports. I'm on the phone with our molecular pathologist or email constantly. All tests are not created equal. When it comes down to next generation sequencing assay, the options could be a 20 gene panel, 50 gene panel or 400 gene panel, ctDNA assays. We did not talk about ctDNA, which lacks NTRK gene fusions and it also has lower sensitivity. It is important to talk to your molecular pathologist, or even the commercial vendors, which typically have very good customer support in terms of what a negative test means what a positive test means. Whenever I have questions, I pick up the phone, I email, because they are so important to correctly identify these alterations for treatment that you really need to have a good communication stream. The other challenges is that we all use different assays, in-house test, Foundation Medicine, Caris, Guardian or other tests. They all have differential capabilities. One of the mantras that I learned from the molecular pathologists is that no test is perfect. It is reasonable to consider an alternate testing method if you suspect that you may have missed something. Our in-house testing is great, but it's not perfect. We know where the holes are, too. If we don't find something, we order orthogonal tests or different type of tests to look for this.




[1] Chapman, PB et al. N Engl J Med. 2011 Jun 30;364(26):2507-16.

[2] Kopetz, S et al. Journal of Clinical Oncology 2010 28:15_suppl, 3534-3534

[3] Vaishnavi, A et al. Nat Med. 2013; 19(11): 1469–1472.

[4] Hondermarck H. Cytokine Growth Factor Rev. 2012;23(6):357-65.

[5] Pulciani S  et al.Nature.1982;300(5892):539-42.

[6] Hong, DS et al. AACR 2016. Abstract#CT008.

[7] Doebele,RC et al. Cancer Discov. 2015; 5(10): 1049–1057

[8] Bertrand T et al. J Mol Biol. 2012;423(3):439-53.

[9] Planchard D et al. Lancet Oncol. 2016;17(5):642-50. 

[10] Tiacci E  et al.N Engl J Med. 2015;373(18):1733-47.

[11] Subbiah V et al. J Clin Oncol. 2018;36(1):7-13.

[12] Diamond EL et al. JAMA Oncol. 2018;4(3):384-388.

[13] Vaishnavi,A, Le,A, and Doebele, RC et al. Cancer Discov. 2015; 5(1): 25–34.

[14] Hong, DS et al. Ann Oncol. 2019 Feb; 30(2): 325–331.

[15] Drilon, A et al. N Engl J Med. 2018 Feb 22; 378(8): 731–739.

[16] Hong, DS et al. Lancet Oncol. 2020 Apr;21(4):531-540

[17] GD Demetri, et al. ESMO 2018. abstract#LBA17

[18] Peter, s et al. N Engl J Med. 2017;377(9):829-838.

[19] Farago AF et al. J Thorac Oncol. 2015;10(12):1670-4

[20] Ziegler, DS et al.Br J Cancer. 2018; 119(6): 693–696

[21] Rosen E, Schram A, Young R, et al. JCO Precis Oncol 2019

[22] Drilon A, DuBois S, Farago A, et al. J Clin Oncol 2019; 37(15): 2006.

[23] Robinson, GW et al. J Clin Oncol 37, 2019 (suppl; abstr 10009)

[24] Hechtman JF et al. Am J Surg Pathol. 2017;41(11):1547-1551.

[25] Hong, DS et al AACR 2020.!/9045/presentation/10962.

[26] Lovely, CM, Shaw, AT et al. Clin Cancer Res. 2014;20(9):2249-56.

[27] Estrada-Bernal, A et al. AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr C65.

[28] Estrada-Bernal, A et al.AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-118.

[29]Doebele, RC et al. ESMO 2019

[30] Drolon, AE et al. Journal of Clinical Oncology 2018 36:15_suppl, 2513-2513

[31] Hyman, D et al. AACR; Cancer Res 2019;79(13 Suppl): Abstract CT127.