Nichol Miller was in a wheelchair and on oxygen when her doctors went back to a genomic test they had already run. Tumors from a soft tissue sarcoma that started in her leg filled 90% of her lungs. The report, reread, showed something nobody had acted on: an NTRK gene fusion. In March 2015 she became the first patient with that fusion to enter a trial of a drug that would later be sold as Vitrakvi, and within 3 days she could sit up and speak in full sentences. A month later the oxygen was gone (Weintraub, CURE, 2019).
Her story has 2 halves, and the second one is the point of this guide. The drug worked because her tumor carried a specific genetic glitch, and the drug was only tried because someone looked for that glitch. More than a decade on, 3 FDA-approved pills exist for NTRK fusion cancer, and the biggest obstacle between a patient and those pills is still the test. In the largest study of US patients whose tumors were known to carry the fusion, only 81 of 216, or 37.5%, ever received a TRK inhibitor (Sledge et al., Nature Communications, 2025). Those are the people who were tested. Nobody can count the ones who were not.
NTRK Is Said En-Track, and the Other Names You Will See
The name is a stumbling block before anything else is. NTRK is usually said "en-track", and the protein family it codes for, TRK, is said "track", which is why Bayer named its drug Vitrakvi and its support line TRAK Assist. People type it the way they hear it, as entrack, en track or n-trk, and search engines do not always connect those spellings to the gene. The letters stand for neurotrophic tyrosine receptor kinase. There are 3 of these genes, NTRK1, NTRK2 and NTRK3, and they make 3 proteins, TRKA, TRKB and TRKC, which in healthy adults do most of their work in the developing nervous system and then go quiet.
A fusion is what happens when a piece of one of those genes breaks off and joins another gene. The joined-up gene makes a hybrid protein with the TRK engine permanently switched on, and that engine tells the cell to keep dividing. The fusion is the whole cause of the cancer in many of these tumors, which is why blocking the engine with a pill works so well. You will see the same thing called an NTRK gene fusion, a TRK fusion, TRK fusion cancer, NTRK fusion-positive solid tumor, or by the specific partner, with ETV6-NTRK3 the most common pairing. They all mean the same test result, and they all mean the same 3 drugs apply.
What Happens to NTRK Fusion Cancer Patients Who Are Found but Not Treated
Caris Life Sciences runs one of the largest tumor-sequencing operations in the country, and in 2025 its scientists drew on a database of 295,316 sequenced tumors and linked the NTRK fusion-positive patients to insurance claims to see what happened after a tissue-agnostic approval. Tissue-agnostic means the FDA approved the drug for any solid tumor carrying the marker, regardless of organ, and NTRK fusions were the second marker to earn that kind of approval. The result was blunt. Between 2019 and 2022, 216 patients in the dataset had an NTRK fusion and 81 were treated with a TRK inhibitor. A second electronic-record database, ConcertAI, gave almost the same answer, 17 of 45. The authors titled that section of the paper "poor clinical uptake of drugs for rare tissue-agnostic indications" (Sledge et al., 2025).
Sledge et al., Nature Communications, March 2025. Untreated patients were more often MSI-high or had a high tumor mutation burden, which suggests some were given immunotherapy instead, but the paper could not explain most of the gap.
The Caris number is the end of a pipe with several leaks upstream of it. A survey of 28 community oncologists who had each treated patients with TRK fusion cancer found that 68% tested for NTRK before starting first-line treatment, and fewer than half, 46%, said they include a TRK inhibitor in the plan once a fusion is confirmed (Klink et al., Targeted Oncology, 2022). At 6 US academic cancer centers, among 55 patients eventually found to carry the fusion, the median wait from cancer diagnosis to the NTRK test was 85 days, and only 35% were tested before their first treatment, 16% in the years before the drugs were approved and 56% after (Willis et al., JMCP, 2024). In community lung cancer practices, NTRK testing in metastatic non-squamous lung cancer rose from 34.6% to 55.7% after the drugs arrived, which still leaves 4 in 10 untested (Sireci et al., Clinical Lung Cancer, 2023).
- Tested at all, in community lung cancer practiceNTRK testing in metastatic non-squamous lung cancer after the TRK inhibitors were approved. It was 35% before. (Sireci et al., Clin Lung Cancer 2023, 500 patients)
- Tested with RNA sequencing, which catches the fusions DNA missesIn the same clinics, 45% of lung cancer patients had DNA-based sequencing and only 30% had RNA-based sequencing. (ASCO 2024 abstract 8589, community clinics in 2023)
- Tested before the first treatment startsAmong 55 patients eventually found NTRK-positive, 35% had the result before first-line therapy; median wait was 85 days. (Willis et al., JMCP 2024, 6 academic centers)
- Given a TRK inhibitor once the fusion is confirmedThe Caris claims analysis above: 81 of 216 confirmed patients were treated. (Sledge et al., Nat Commun 2025, 216 patients)
The honest summary is that at every point where a decision has to be made, between a third and two thirds of patients go the wrong way, and a patient has to clear every one of them.
Why a DNA-Only Panel Misses NTRK3 Fusions and What RNA Sequencing Adds
Most comprehensive genomic profiling reads DNA. That works for a point mutation, a single misspelled letter, and it works for many fusions. It fails for a specific reason with NTRK3. The parts of the gene where the breaks happen, the stretches between the coding segments called introns, are enormous. In NTRK3 the 3 introns where most breakpoints fall span 193,000 letters of DNA and are full of repetitive sequence, so panel designers leave them out because they cannot be read reliably (Solomon et al., 2020). A fusion whose break sits in an unread intron is invisible to the panel.
RNA sequencing sidesteps the problem. RNA is the copy of the gene the cell makes after the introns are cut out, so the fusion appears as a clean join between 2 genes no matter where in the intron the break was. When Memorial Sloan Kettering compared its DNA panel against its RNA panel across 33,997 patients, DNA caught 81.1% of the fusions RNA found. The pattern of misses is what matters for a patient. Every one of the 29 ETV6-NTRK3 fusions was caught, because that partner is common enough that the panel was built to see it. Of 10 NTRK3 fusions with any other partner, DNA caught 1.
Solomon et al., Modern Pathology, 2020. Across all 3 genes, DNA sequencing found 81.1% of the fusions that RNA sequencing found; the misses were concentrated in this second row plus 5 NTRK1 and NTRK2 fusions.
There is a cheaper screening test, an antibody stain on the tumor slide called pan-TRK immunohistochemistry, which looks for the TRK protein itself. In the same MSK study it found 87.9% of fusions, with the same weak spot, NTRK3 fusions at 79.4%. In thyroid cancer specifically, the stain missed 4 of 26 fusion-positive tumors and every miss was an ETV6-NTRK3 (Int J Mol Sci, 2022). The European Society for Medical Oncology's recommendation reflects all of this: in tumor types where the fusion is rare, go straight to sequencing, preferably RNA, or screen with the stain and sequence the positives; in tumor types where the fusion is nearly universal, a confirmatory test such as FISH or RNA is enough (Marchiò et al., Annals of Oncology, 2019).
How Rare NTRK Fusions Are in Lung, Thyroid, Salivary Gland and Other Cancers
Rare is the wrong single word for this fusion, because its frequency swings across 3 orders of magnitude depending on the tumor. Across 295,676 sequenced patients at Foundation Medicine the overall rate was 0.30%, about 1 in 330, and in children it was 1.34% (Westphalen et al., npj Precision Oncology, 2021). In common adult cancers the rate stays below 0.5%; a systematic review of 160 studies concluded that "a thousand patients with common cancer types would need testing to identify two or three patients with NTRK fusions" (O'Haire et al., Scientific Reports, 2023). In a few rare tumors the fusion is the defining feature, present in 7 of 10 infantile fibrosarcomas and 9 of 10 secretory carcinomas of the breast.
Sources in the list below. The salivary secretory carcinoma figure is a rounded midpoint of the pooled range of 83% to 90%. Pediatric papillary thyroid is a single 27-patient series from the northeast United States, so its exact rate is uncertain, but every series agrees it is far above the adult rate.
The ladder explains the testing policy. Nobody would screen every lung cancer with a dedicated NTRK test for a 1 in 400 hit, which is why guidelines fold NTRK into the broad panel that lung cancer patients should be getting anyway. At the top of the ladder the logic flips: a pathologist who sees a secretory carcinoma expects the fusion and confirms it. The middle rungs, thyroid, sarcoma and salivary gland at 1 to 3%, are where patients get lost, because the fusion is common enough to matter and rare enough that nobody thinks of it.
NTRK Fusion Thyroid Cancer, Where Testing Is Recommended When Radioactive Iodine Stops Working
Ashton Leeds was a preschooler in Claresholm, Alberta, when his mother noticed a lump on his neck. The diagnosis was stage 4 papillary thyroid cancer that had spread to his lungs, and radioactive iodine, the standard second step after surgery, did not clear it. A researcher asked to add his tumor to a rare-tumor database and found an NTRK fusion. He started larotrectinib at Seattle Children's in July 2017, his family traveling from Alberta every month, and within 2 months most of the tumors were gone (ABC7, 2018; CURE, 2019).
His case is typical in a way that matters. NTRK fusions are far more common in childhood thyroid cancer than in adult disease, 26% in one series of 27 children with papillary thyroid cancer and no radiation history (Prasad et al., Cancer, 2016), against about 1.25% in adults. They are also enriched in thyroid cancers that follow radiation exposure. The tumors tend to be papillary, often with a follicular or solid growth pattern, and they are the ones that fail radioactive iodine, which is precisely when the National Comprehensive Cancer Network says to test. Its patient guideline puts it plainly: testing for NTRK fusions, along with RET, ALK and BRAF, "is recommended for recurrent, advanced, or metastatic cancers that can't be treated with RAI therapy".
The drugs work well here. In the pooled larotrectinib trials, 28 evaluable thyroid patients had a 71% response rate, and the split by subtype is the useful number: 86% in differentiated (papillary and follicular) thyroid cancer against 29% in anaplastic thyroid cancer, the aggressive form (Waguespack et al., European Journal of Endocrinology, 2022). The current Vitrakvi label reports 63% across 30 thyroid patients. For entrectinib, 7 of 13 thyroid patients responded, 53.8% (Demetri et al., Clinical Cancer Research, 2022). Susan Spinosa, who founded the patient group NTRKers, was diagnosed with thyroid cancer in 2004, went through neck surgeries and several courses of radiation, and found her fusion in 2018 only after she asked for genomic testing herself (NTRKers).
The next idea being tested is to use the pill to make radioactive iodine work again. The LANTERN trial (NCT05783323), run by Children's Hospital of Philadelphia with UCSF, St. Jude, MD Anderson and Seattle Children's, gives 6 months of larotrectinib to patients with NTRK fusion differentiated thyroid cancer and lung spread, then radioactive iodine, to see whether the drug restores the tumor's ability to soak up the iodine. The primary endpoint is complete clearance of the lung metastases, which happens in fewer than a quarter of patients on iodine alone.
Secretory Carcinoma of the Salivary Gland, the Tumor Defined by ETV6-NTRK3
Until 2010 this tumor did not have a name. Pathologists filed it as acinic cell carcinoma or as an unclassified adenocarcinoma of the parotid gland. That year a Czech group led by Alena Skálová described 16 cases that looked like secretory carcinoma of the breast and carried the same ETV6-NTRK3 fusion, and called it mammary analogue secretory carcinoma, or MASC (Skálová et al., American Journal of Surgical Pathology, 2010). The World Health Organization shortened the name to secretory carcinoma in its 2017 classification of head and neck tumors. The fusion is present in 83% to 90% of cases, which makes this one of the few cancers where the diagnosis and the drug target are the same thing.
The response rates are the best in the field. Among 24 salivary gland cancer patients treated with larotrectinib, every one of whom carried ETV6-NTRK3, 92% responded (Le et al., The Oncologist, 2024); the Vitrakvi label puts it at 84% in 25 patients. Entrectinib's trials included 24 patients with secretory carcinoma and 20 responded, 83.3% (Demetri et al., 2022). The practical lesson for a patient with a salivary gland tumor is that the question is not whether to test but whether the pathologist recognized the tumor in the first place, since a slide filed as acinic cell carcinoma may never be sent for the fusion test. Adults with any salivary gland cancer carry the fusion at 2.43%, 8 times the rate in solid tumors overall, so it belongs on the panel for the whole group, not only the secretory subtype.
NTRK Fusion Lung Cancer, 1 in 400 Patients and Easy to Miss
Lung cancer produces more adult NTRK fusion patients than any other tumor type, 136 in the Foundation Medicine series, for the simple reason that there is so much lung cancer. The rate is low, 0.23% in a screen of 4,872 non-small cell lung cancers across 47 institutions, and the fusion shows up in smokers and non-smokers, in every age group and in both main histologies (Farago et al., JCO Precision Oncology, 2018). It clusters in one place: tumors with no other driver mutation. When Memorial Sloan Kettering took 254 lung adenocarcinomas that a DNA panel had called driver-negative and ran RNA sequencing, 36 of them, 14%, turned out to carry an alteration the DNA panel had missed, 27 of them fusions, and 8 of the 10 patients who then got a matched drug benefited from it (Benayed et al., Clinical Cancer Research, 2019).
That is why NCCN requires NTRK1, NTRK2 and NTRK3 in any broad molecular panel run on advanced non-small cell lung cancer, alongside EGFR, ALK, ROS1, KRAS, BRAF, MET, RET and HER2, and why the testing numbers above are so troubling: in community clinics in 2023, 45% of lung cancer patients got DNA sequencing and 30% got RNA sequencing (ASCO 2024). For the patients who are found, the drugs perform as they do everywhere else. Larotrectinib produced responses in 73% of 15 evaluable lung cancer patients, with a median duration of response of 33.9 months and median survival of 40.7 months (Drilon et al., JCO Precision Oncology, 2022); the label says 70% in 30. Entrectinib, which was designed to cross into the brain, gave 14 responses in 22 lung patients, 63.6%, and the Rozlytrek label records brain metastases shrinking in 3 of 4 adults who had them.
Vitrakvi, Rozlytrek and Augtyro Side by Side
The 3 approved drugs are all pills taken at home, and the choice between them comes down to age, brain involvement and whether a first drug has already stopped working. Vitrakvi (larotrectinib) is the only one approved solely for NTRK fusion cancer, the only one with full rather than accelerated approval, and the only one with an oral solution and no lower age limit. Rozlytrek (entrectinib) is approved for patients older than 1 month, also treats ROS1-positive lung cancer, and has the strongest brain penetration data. Augtyro (repotrectinib), for patients 12 and older, is the drug for the day the first one stops working: 50% of patients whose tumors had become resistant to a prior TRK inhibitor responded, including 60% of those with the specific solvent-front mutations that block the older drugs. The full label-by-label comparison, including side effects and the 4 drugs in trials, is on the NTRK fusion cancer treatments compared page.
Vitrakvi prescribing information, May 2026. The tumor types at the top are the ones where the fusion is the whole story; brain tumors respond least; Rozlytrek and Augtyro were both designed to cross into the brain.
The pipeline is thinner than the approvals suggest. As of October 2026 only 2 experimental NTRK drugs are recruiting at US sites, TL118 in a Phase 2 trial at MD Anderson, Texas Oncology in Dallas and Adventist Health in Glendale, and ANS03 in a first-in-human Phase 1 trial at Memorial Sloan Kettering. Two next-generation drugs, zurletrectinib and eratrectinib, were approved in China in December 2025 and June 2026 with response rates of 89% and 69% in single-arm trials, and neither has a US filing. Taletrectinib's NTRK program and 2 smaller candidates have been terminated, and selitrectinib has had no new trials since its Phase 1 study completed in January 2023. The approved drugs are themselves in trials for new settings, including the CARE study of Augtyro in children under 12 and GLOBOTRK, a St. Jude trial of Rozlytrek as the first treatment for infants with NTRK-fused brain tumors.
What NTRK Testing Costs and Why Medicare Denies 1 in 4 Claims
Since March 16, 2018, Medicare has covered next-generation sequencing nationally for anyone with recurrent, metastatic or advanced stage III or IV cancer who has not had the same test before and intends to pursue treatment, provided the test is an FDA-approved companion diagnostic with an approved use in that cancer (CMS National Coverage Determination 90.2). FoundationOne CDx has been an approved companion diagnostic for Vitrakvi since October 23, 2020 and for Rozlytrek since June 9, 2022, and Medicare paid $3,500 for it when it first set a rate in 2018. Tests that are not FDA-approved, and RNA sequencing in particular, are left to regional Medicare contractors to decide, which is one reason the RNA step in the chart above leaks so badly.
Coverage on paper is not payment in practice. A study of 29,919 Medicare claims for cancer sequencing between 2016 and 2021 found that 23.3% were denied, and the denial rate rose over time, from 16.8% before the national policy to 27.4% after its 2020 amendment. Claims from independent laboratories, which is where most sequencing happens, had 2.76 times the odds of denial compared with claims from hospitals (Kang et al., JAMA Network Open, 2025). In a survey of 200 US oncologists, surgeons and pathologists published in 2024, the most common reimbursement barriers to sequencing were prior authorization (72%), unfamiliarity with new billing codes (68%) and paperwork (67.5%). Foundation Medicine states that 90% of its Medicare patients and 77% of commercially insured patients owe nothing, so a denial is usually a fight between the lab and the insurer rather than a bill, but it can delay a result by weeks, and 85 days was already the median wait.
One email when NTRK Fusion Cancer trials change or an FDA decision lands. No newsletter, no spam.
The Question to Ask Your Oncologist About NTRK Testing
Susan Spinosa's turning point, as she told Bayer's Vitrakvi patient-story site, was a doctor who looked at her chart and said her thyroid cancer was not acting the way it should, so they should do genomic testing. On her own group's site she describes how, after years of surgery and radiation that did not hold, she asked for the test herself. Nichol Miller's fusion was sitting in a report that had already been filed. Ashton Leeds's was found because a researcher wanted his tumor for a database. None of the 3 was found by routine.
The question is worth asking at 3 moments: at diagnosis of any advanced solid tumor, when a first treatment stops working, and whenever a pathology report uses the words secretory carcinoma, infantile fibrosarcoma, or papillary thyroid cancer in a child. Ask whether comprehensive genomic profiling was ordered, whether it included RNA sequencing or a pan-TRK stain, and whether the report lists NTRK1, NTRK2 and NTRK3 as covered genes. If the tumor was tested years ago on a smaller panel, ask for it to be rerun. If there is no tissue left, the blood-based FoundationOne Liquid CDx is approved for Rozlytrek, with the caveat that a negative blood result does not rule the fusion out. The NTRKers group runs a free trial and treatment center finder with phone and chat support, and the TargetCancer Foundation's TRACK study offers remote genomic testing with a virtual tumor board to people with rare cancers, including unexplained ones.
Questions People Ask About NTRK Fusion Cancer
How do you pronounce NTRK?
NTRK is usually said "en-track", and TRK, the protein family it codes for, is said "track". The pronunciation is informal rather than official, but it is the one used in clinics, and Bayer built both the drug name Vitrakvi and its support program TRAK Assist around it. The letters stand for neurotrophic tyrosine receptor kinase.
What is an NTRK gene fusion?
A fusion is a genetic accident in which part of an NTRK gene (NTRK1, NTRK2 or NTRK3) breaks off and joins a different gene. The combined gene makes a TRK protein that is permanently switched on and drives the cell to keep dividing. The fusion is found in about 0.3% of solid tumors overall, under 0.5% of common cancers, 1 to 3% of thyroid, sarcoma and salivary gland cancers, and 70 to 90% of infantile fibrosarcoma and secretory carcinoma.
Who should be tested for NTRK fusions?
Anyone with an advanced or metastatic solid tumor who is about to start treatment or whose treatment has stopped working, because the FDA-approved TRK inhibitors work in any tumor type that carries the fusion. Guidelines specifically require NTRK on broad panels for advanced lung cancer and recommend it for thyroid cancer that no longer responds to radioactive iodine. Children with papillary thyroid cancer and anyone with a tumor called secretory carcinoma or infantile fibrosarcoma should be tested early.
Can a genomic test miss an NTRK fusion?
Yes. DNA-only sequencing panels found 81.1% of the fusions that RNA sequencing found in a 33,997-patient Memorial Sloan Kettering study, and most of the misses, 9 of 14, involved NTRK3 with an uncommon partner gene, where DNA panels caught 1 of 10. The reason is that the breakpoints sit in stretches of DNA too large and repetitive for panels to read. If a DNA panel was negative and the tumor has no other driver mutation, ask about RNA-based testing.
What is the difference between pan-TRK immunohistochemistry and NTRK sequencing?
Pan-TRK immunohistochemistry is a stain applied to the tumor slide that shows whether TRK protein is present; it is fast and cheap and found 87.9% of fusions in the MSK study, but it misses some NTRK3 fusions and gives false positives in tumors with nerve or muscle features. Sequencing reads the gene itself and identifies the partner. ESMO recommends sequencing, ideally RNA-based, in tumor types where the fusion is rare, or a stain followed by sequencing of positives.
Does Medicare cover NTRK fusion testing?
Medicare covers FDA-approved next-generation sequencing tests nationally for recurrent, metastatic or advanced stage III or IV cancer under National Coverage Determination 90.2, which has applied since March 2018. FoundationOne CDx is an approved companion diagnostic for Vitrakvi and Rozlytrek. Coverage of tests that are not FDA-approved, including most RNA sequencing, is decided by regional contractors, and a 2025 JAMA Network Open study found 23.3% of Medicare sequencing claims were denied between 2016 and 2021.
What are the response rates for NTRK fusion thyroid, salivary and lung cancer?
For larotrectinib (Vitrakvi): 71% in thyroid cancer overall, 86% in differentiated and 29% in anaplastic thyroid cancer; 92% in salivary gland cancer; 73% in lung cancer. For entrectinib (Rozlytrek): 53.8% in thyroid, 83.3% in secretory carcinoma of the salivary gland, 63.6% in lung cancer. These come from single-arm trials of 13 to 28 patients per tumor type and are not head-to-head comparisons.
Which NTRK fusion cancer drug is best?
No trial has compared them. Vitrakvi and Rozlytrek are both first-line options with response rates near 60% in adults. Vitrakvi has full approval, an oral solution and no age limit; Rozlytrek is approved for patients older than 1 month and has stronger data in brain metastases. Augtyro is approved for patients 12 and older and is the drug used after a first TRK inhibitor stops working, with a 50% response rate in that setting.
Are there NTRK clinical trials in the United States?
As of October 2026, 2 investigational NTRK drugs are recruiting at US sites: TL118, a Phase 2 pan-TRK inhibitor at MD Anderson, Texas Oncology in Dallas and Adventist Health in Glendale (NCT06010342), and ANS03, a Phase 1 ROS1 and TRK inhibitor at Memorial Sloan Kettering (NCT06716138). The approved drugs are also in trials for new settings, including Augtyro in children under 12 (CARE, NCT04094610), Rozlytrek as first treatment for infants with NTRK-fused brain tumors at St. Jude (GLOBOTRK, NCT06528691), and larotrectinib before radioactive iodine in thyroid cancer (LANTERN, NCT05783323).