If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
NTRK gene fusions are targetable oncogenic drivers independent of tumor type.
Selective TRK inhibitor larotrectinib is active in tumors with NTRK gene fusions.
A diverse array of methods can be used to detect NTRK gene fusions.
The most common testing approach was RNA-based next-generation sequencing.
54 different NTRK fusion partners identified in 225 patients with TRK fusion cancer.
NTRK gene fusions are targetable oncogenic drivers independent of tumor type. Prevalence varies from highly recurrent in certain rare tumors to <1% in common cancers. The selective TRK inhibitor larotrectinib was shown to be highly active in adult and pediatric patients with tumors harboring NTRK gene fusions.
We examined the techniques used by local sites to detect tumor NTRK gene fusions in patients enrolled in clinical trials of larotrectinib. We also report the characteristics of the detected fusions in different tumor types.
The analysis included 225 patients with 19 different tumor types. Testing methods used were next-generation sequencing (NGS) in 196 of 225 tumors (87%); this was RNA-based in 96 (43%); DNA-based in 53 (24%); DNA/RNA-based in 46 (20%) and unknown in 1 (<1%); FISH in 14 (6%) and PCR-based in 12 (5%). NanoString, Sanger sequencing and chromosome microarray were each utilized once (<1%). Fifty-four different fusion partners were identified, 39 (72%) of which were unique occurrences.
The most common local testing approach was RNA-based NGS. Many different NTRK gene fusions were identified with most occurring at low frequency. This supports the need for validated and appropriate testing methodologies that work agnostic of fusion partners.
The neurotrophic receptor tyrosine kinase genes, NTRK1, NTRK2 and NTRK3, encode tyrosine kinases (TRKA, TRKB and TRKC, respectively) involved in the development, maintenance, and function of the vertebrate nervous system [
]. Somatic chromosomal rearrangements resulting in fusions between the 3′ region of one of the NTRK genes and the 5′ region of one of a series of partner genes (NTRK gene fusions) have been identified at varying prevalence in a wide range of human tumors. The product of such a fusion may result in constitutive ligand-independent activation of the TRK kinase and associated downstream signaling and may drive cancer development [
Recent clinical studies of the highly selective TRK inhibitor larotrectinib and the TRK/ROS1/ALK multikinase inhibitor entrectinib have confirmed that TRK fusion proteins are targetable oncogenic drivers in an extensive range of different solid tumors [
]. These studies led to the regulatory approval of larotrectinib and entrectinib by the US Food and Drug Administration (FDA) and other global regulatory agencies for the tumor-agnostic treatment of patients with solid tumors harboring an NTRK gene fusion [
The initial FDA approval of larotrectinib was based on a pooled analysis of the first 55 consecutively enrolled adult and pediatric patients with a range of primary non-central nervous system (CNS) solid tumors harboring an NTRK gene fusion who were included in one of three phase I or II studies [
]. The objective response rate by investigator assessment in this analysis was 80%. An updated pooled efficacy analysis of an expanded population of 159 patients enrolled in these studies subsequently reported an investigator assessed response rate of 79% and a median duration of response of 35.2 months, with high response rates in both adult (73%) and pediatric (92%) patient subgroups [
]. Responses were seen across a broad range of tumor types and although these pooled analyses did not include enrolled patients with primary CNS tumors, promising activity in such tumors has also separately been reported for larotrectinib [
]. A more diverse spectrum of NTRK gene fusions has also been identified in a wide range of common cancers, including non-small cell lung cancer (NSCLC), colorectal cancer, non-secretory metastatic breast cancer and melanoma, generally at overall frequencies of <1% of total cases [
The high level of activity of the selective TRK inhibitor larotrectinib in patients with tumors harboring NTRK gene fusions creates a clinical imperative to identify all patients with such tumors so that they can receive the most appropriate and effective targeted treatment. However, the highly variable prevalence of these chromosomal rearrangements in different tumor types suggests that a single diagnostic strategy is unlikely to be optimal across all settings. A further complication is that the NTRK2 and NTRK3 genes contain large repetitive element rich intronic regions within which fusion breakpoints may be located. Consequently, the detection of all NTRK gene fusions may be challenging with existing DNA-based next-generation sequencing (NGS) analysis approaches [
]. Seeking to address these issues, several expert author groups, including one recruited by the European Society for Medical Oncology, have proposed testing algorithms to facilitate the rational identification of patients with tumors harboring NTRK gene fusions [
In order to provide some context to the discussion of how to most effectively identify such patients, we now report applied detection methods and the nature of the NTRK gene fusions identified across tumor types in patients included in the phase I and II larotrectinib studies.
Materials and methods
Patients with tumors harboring an NTRK gene fusion included in this analysis were pooled from three clinical trials of larotrectinib. These were a phase I dose-escalation study in adults with solid tumors (NCT02122913) [
] and a phase II basket study in adults and children aged 12 years and older with solid tumors harboring an NTRK gene fusion (NCT02576431). The data cutoff for the current analysis, which included only those patients for whom an independent radiology review committee assessment was available, was July 20, 2020.
In each of these studies, pathologic diagnoses and molecular analyses were sourced locally by investigators, with no central confirmation prior to enrollment. The molecular pathology reports were provided and applied testing methods were reviewed and recorded centrally. Where a tumor had been analyzed locally by multiple methods, such as FISH followed by NGS for confirmation, and where this information was included in the local report, the technique most definitively demonstrating the NTRK gene fusion was recorded as the testing approach. The assessment for NGS reflected whether an RNA- or DNA-based method was used for the identification of the NTRK gene fusion. In some instances, reports indicated that both RNA- and DNA-based NGS had been used to analyze a sample. In such cases, NTRK gene fusion identification might have been based on one or both of these methods, with such information not always described in the molecular pathology reports. The category RNA/DNA was therefore applied if the NTRK fusion identification was either based on both input materials, or if it was not known on which input material the final fusion calling was based.
Infantile fibrosarcomas, congenital mesoblastic nephromas, and secretory breast cancers with a documented ETV6 rearrangement (or NTRK3 rearrangement) were considered to have an inferred ETV6-NTRK3 gene fusion/inferred NTRK3 gene fusion based on the known incidence of the alteration in these tumor types and are included into the groups of ETV6-NTRK3 or NTRK3 gene fusions as applicable within Fig. 1, Fig. 2, Fig. 3.
]. Genome-scale plots and chord diagrams were generated using a modification of the circlize package. The chromosomal localization of fusion partners was identified using HUGO Gene Nomenclature Committee at the European Bioinformatics Institute database (https://www.genenames.org/). The band coordinates were determined using the Bulk Sequence - Cytogenetic Conversion Service (https://www.ncbi.nlm.nih.gov/genome/tools/cyto_convert/); Bethesda (MD): National Library of Medicine (US), National Center for Biotechnology Information.
The analysis included 225 patients, with 19 different solid tumor types (including one case of cancer of unknown primary). Of the included patients, 129 (57%) of 225 were adults and 96 (43%) were children under 18 years of age. Patients were predominantly white (162 [72%] of 225), 115 (51%) were male and 110 (49%) were female. As documented by institutional assessment, the most common tumor types were soft tissue sarcomas, reported in 92 (41%) of 225 patients (including infantile fibrosarcomas in 40 [18%] patients), primary CNS tumors (33 [15%]) and thyroid tumors (28 [12%]) (Table 1).
The testing methods that were used to identify NTRK gene fusions varied to some extent by tumor type (Table 1), with sequencing approaches used exclusively for calling primary CNS tumors and gene-specific non-sequencing methods used predominantly for infantile fibrosarcomas. The most commonly used method overall for NTRK gene fusion confirmation in 196 (87%) of 225 tumors was NGS; this NGS fusion detection was RNA-based in 96 (43%) of 225 tumors, DNA-based in 53 (24%), DNA/RNA-based in 46 (20%), and was unknown in 1 (<1%) based on the provided report. In tumor types in which NTRK fusions have been shown to be highly recurrent molecular changes, alternative screening methods were commonly used without further sequencing-based confirmation, including FISH in 14 (6%) cases, polymerase chain reaction (PCR)-based approaches in 12 (5%; predominantly reverse transcriptase [RT]-PCR), and NanoString analysis in 1 case (<1%). In addition to these approaches, Sanger sequencing and a chromosome microarray were used to detect NTRK gene fusions in single cases (<1%). The sole use of non-sequencing methods such as FISH or PCR to detect fusions was most notable for infantile fibrosarcomas (22 [55%] of 40 cases) and secretory breast cancers (2 [50%] of 4 cases).
Across all tumor types, NTRK3 gene fusions were the most commonly identified (definitively or inferred) being reported in 108 (48%) of 225 tumors, followed by NTRK1 gene fusions in 87 (39%) and NTRK2 gene fusions in 30 (13%). Figs. 1A–C and 2 summarize the NTRK gene fusion distribution by tumor type. NTRK3 gene fusions were most prevalent in salivary gland tumors, regardless of histology (100%; 22 of 22 cases), gastrointestinal stromal tumors (100%, 4 of 4), congenital mesoblastic nephromas, cellular subtype (100%; 2 of 2) and infantile fibrosarcomas (85%; 34 of 40). NTRK1 gene fusions were most prevalent in colon cancers (88%; 7 of 8 cases), NSCLCs (86%; 12 of 14) and other soft tissue sarcomas (67%; 32 of 48). NTRK2 gene fusions were generally uncommon across tumor types, with the exception of primary CNS tumors, in which this was the most frequently involved NTRK gene (73%; 24 of 33 cases, including in 5 of 7 adult and 19 of 26 pediatric patients).
Figs. 1A–C and 3 summarize the NTRK fusion partner distribution by tumor type. There were 54 different fusion partners identified across the series, with 39 (72%) of 54 documented in only a single tumor. While the majority of fusion partners (51 [94%] of 54) in this series tended to be specific to one particular NTRK gene, both BCR-NTRK2 and BCR-NTRK3 gene fusions were each identified in primary CNS tumors (BCR-NTRK2, n = 3; BCR-NTRK3, n = 1), SQSTM1-NTRK1 and SQSTM1-NTRK3 gene fusions in lung cancers (SQSTM1-NTRK1, n=1; SQSTM1-NTRK3, n = 2) and one infantile fibrosarcoma (SQSTM1-NTRK1), and SPECC1L-NTRK2 and SPECC1L-NTRK3 gene fusions in primary CNS tumors (SPECC1L-NTRK2, n=2) and one soft tissue sarcoma (SPECC1L-NTRK3). Twenty-four different fusion partners were associated with NTRK1, 22 with NTRK2, and 11 with NTRK3. Of these 57 chromosomal rearrangements, 27 (47%) were intrachromosomal events, including 17 (71%) of 24 associated with NTRK1, 6 (27%) of 22 associated with NTRK2 and 4 (36%) of 11 associated with NTRK3(Fig. 1D–F).
The most commonly identified (definitively or inferred) gene fusion was between ETV6 and NTRK3. This rearrangement was documented in 93 (41%) of 225 tumors. In particular, it was reported for all of the salivary gland tumors, breast cancers, gastrointestinal stromal tumors and congenital mesoblastic nephromas, cellular subtype harboring NTRK3 gene fusions, and it was present in 32 (94%) of the 34 infantile fibrosarcomas harboring NTRK3 gene fusions. In most cases, this rearrangement was definitively indicated by the molecular test used. However, in 11 tumors, including nine cases of infantile fibrosarcoma and two cases of secretory breast cancer, considering the high documented prevalence of the ETV6-NTRK3 fusion in these tumor types, the presence of this fusion was inferred from positive gene rearrangement results of ETV6, NTRK3, or ETV6 and NTRK3 break-apart FISH analysis (Table 1). Of these 11 patients, 10 (91%) had a complete or partial response to larotrectinib, indicating that the use of FISH in these high-prevalence settings was sufficient to identify an appropriate target population. The remaining patient with secretory breast cancer had a best response of stable disease. A subsequent NGS analysis of circulating tumor DNA from this patient confirmed the presence of the ETV6-NTRK3 fusion. Initially inferring the presence of an NTRK gene fusion in these settings by FISH was therefore a clinically effective approach in relation to identifying patients who might benefit from larotrectinib treatment.
Two other recurrent gene fusions were identified in several tumor types, with both detected commonly in the group of other soft tissue sarcomas; these were TPM3-NTRK1 in 37 (16%) of 225 cases overall (including 5 [62%] of 8 colon cancers) and LMNA-NTRK1 in 18 (8%) cases overall. The diversity of gene fusion events within a particular tumor type was most notable in primary CNS tumors, with 24 different fusion partners identified in the 33 tumors included in this category; 17 (71%) of these 24 gene fusion events were documented only in a single tumor.
Testing approaches for the identification of tumor-specific NTRK gene fusions fall broadly into two categories: single analyte methods, investigating one gene or gene fusion at a time, such as FISH and RT-PCR; and comprehensive multiplex approaches, which enable the investigation of potentially very large numbers of genes in parallel, such as DNA- and RNA-based NGS. While single analyte methods may provide relatively rapid initial screening solutions in settings where specific well-characterized NTRK gene fusions occur at high frequency, NGS approaches are likely to be more appropriate testing methods in histologies where NTRK gene fusions are relatively rare events and where a wide range of NTRK fusion partner genes and alternative targetable driver mutations may be expected. The pre-analytical considerations, turnaround times, and advantages and disadvantages of the different clinical laboratory techniques that may be used to identify tumors harboring NTRK gene fusions have previously been described in detail [
Our analysis shows that the testing methods used to detect NTRK gene fusions prior to enrollment in larotrectinib clinical trials varied to some extent depending on tumor type and perhaps according to local practice. In particular, in the rare tumor types in which certain gene fusions such as ETV6-NTRK3 are highly recurrent genetic events, approximately half of the tests used by local sites to identify fusions were solely based on gene-specific FISH or PCR approaches without additional confirmation by sequencing. The use of gene-specific testing approaches was most common in patients with infantile fibrosarcoma. However, of note, when NGS was used to assess the tumor genetic landscape in infantile fibrosarcomas, 20% (8/40) of tumors were found to have NTRK gene fusions other than ETV6-NTRK3. This highlights that even in tumor types in which specific NTRK gene fusions are highly recurrent events [
], a testing method that is agnostic of the partner gene is recommended to identify all patients who may benefit from TRK inhibitor treatment, in particular as larotrectinib activity has been seen regardless of NTRK gene fusion characteristics [
]. However, given that the turnaround time for NGS analysis may necessitate a delay of several weeks before treatment can commence, and that access to such testing may be restricted in some settings, the optimum diagnostic approach in rare indications with a high prevalence of ETV6-NTRK3 gene fusions might be to initially screen tumors with ETV6 and/or NTRK3 FISH (break-apart or dual-fusion probes) or RT-PCR [
]. If such tumors are negative for rearrangement using these gene-specific methods, NGS reflex testing, ideally RNA-based, should be considered.
The most common method local investigators used to identify tumor NTRK gene fusions was RNA-based NGS, either alone or in combination with DNA-based NGS. One advantage of using an RNA-based NGS approach is that commercial kits are available which cover all fusions without any prior knowledge of the partner gene or breakpoints. In addition, only expressed gene fusions will be detected, and the process allows for in frame versus out of frame confirmation for all fusions [
]. Also importantly, the intron architecture of the involved genes is not a factor in RNA-based analyses. This is particularly relevant in relation to detecting NTRK2 and NTRK3 fusions, as these genes have several unusually large, repetitive element rich introns within which fusion breakpoints may be located [
]. These characteristics present particular challenges for comprehensive coverage of NTRK2 and NTRK3 in DNA-based NGS assays. However, as tumors are increasingly and routinely tested for mounting numbers of predictive biomarkers, and in order to derive the maximum amount of information from limited available tissue samples, it is likely that a broad-based approach of analyzing both DNA and RNA by NGS in order to detect all classes of somatic mutational changes, as occurred for a sizable cohort of patients in this study, will be optimal. If cost or reimbursement implications preclude the routine use of such approaches, it may be that triaging tumors for such analyses based on validated immunohistochemistry assays or genomic triaging on the basis of the absence of other driver mutations may be effective options [
]. Of particular note in this context, it has been shown that the positive predictive value of pan-TRK immunohistochemistry may be 100% in some tumor types, including carcinomas of the colon, lung, thyroid, pancreas and biliary tract, which highlights the potential use of this approach as a screening method to prioritize tumors in certain indications for comprehensive testing by NGS [
The use of partner-agnostic NGS approaches to identify patients with somatic NTRK gene fusions has highlighted the variability of these rearrangements within and across tumor types. Overall, 54 different partners were associated with the 225 gene fusion events, with 39 (72%) of those partners documented in only a single tumor. This variability was particularly apparent for primary CNS tumors, with 24 different fusion partners identified in the 33 tumors included in this category, 17 of which were documented only in a single tumor. Whereas gene fusions involving NTRK2 were relatively uncommon in other tumor types, being reported in only 6 of 192 (3%) non-primary CNS tumors, 24 (73%) of 33 primary CNS tumors harbored such fusions. A similar pattern of distribution in primary CNS tumors has previously been reported in a large-scale survey of NTRK gene fusions in human cancer [
In conclusion, NTRK gene fusions occur with a wide range of possible partners with the majority of particular fusions occurring at a low frequency across multiple tumor types. This pattern of occurrence supports the need for validated and appropriate routine biomarker testing methods that provide complete coverage for all three NTRK genes regardless of partners and breakpoints. It is likely that different diagnostic approaches will be optimal in different clinical settings.
Declaration of Competing Interest
E.R.R. reports institutional reimbursement from Bayer for time on advisory boards and as part of clinical trials. J.H. is a full-time employee of Neogenomics, and reports research funding and advisory board fees from Bayer and honoraria from WebMD. M.R., J.S., J.W., K.S. and H.N. are employees of Bayer and R.N. works for an organization that carries out contract research for Bayer. C.M.L. reports that her spouse is employed by Bayer. D.S.H. reports institutional research/grant funding from AbbVie, Adaptimmune, Aldi-Norte, Amgen, Astra-Zeneca, Bayer, BMS, Daiichi-Sankyo, Deciphera, Eisai, Erasca, Fate Therapeutics, Genentech, Genmab, Infinity, Kite, Kyowa, Lilly, LOXO, Merck, Medimmune, Mirati, Mologen, Navier, NCI-CTEP, Novartis, Numab, Pfizer, Pyramid Bio, SeaGen, Takeda, Turning Point Therapeutics, Verstatem, and VM Oncology; travel, accommodation, expenses from: Bayer, Genmab, AACR, ASCO, SITC and Telperian; consulting, speaker or advisory roles with: Adaptimmune, Alpha Insights, Acuta, Alkermes, Amgen, Aumbiosciences, Atheneum, Axiom, Barclays, Baxter, Bayer, Boxer Capital, BridgeBio, CDR-life AG, COR2ed, COG, Ecor1, Genentech, Gilead, GLG, Group H, Guidepoint, HCW Precision, Immunogen, Infinity, Janssen, Liberium, Medscape, Numab, Oncologia Brasil, Pfizer, Pharma Intelligence, POET Congress, Prime Oncology, Seattle Genetics, ST Cube, Takeda, Tavistock, Trieza Therapeutics, Turning Point, WebMD, and Ziopharm; and other ownership interests in relation to: OncoResponse (founder) and Telperian Inc (advisor). A.D. reports honoraria/advisory board roles with: Ignyta/Genentech/Roche, Loxo/Bayer/Lilly, Takeda/Ariad/Millenium, TP Therapeutics, AstraZeneca, Pfizer, Blueprint Medicines, Helsinn, Beigene, BergenBio, Hengrui Therapeutics, Exelixis, Tyra Biosciences, Verastem, MORE Health, Abbvie, 14ner/Elevation Oncology, Remedica Ltd., ArcherDX, Monopteros, Novartis, EMD Serono, Melendi, Liberum, Repare RX, Nuvalent, Merus, AXIS, Chugai Pharm, and EPG Health; associated research paid to institution from Pfizer, Exelixis, GlaxoSmithKlein, Teva, Taiho, and PharmaMar; royalties from Wolters Kluwer; other from Merck, Puma, Merus, and Boehringer Ingelheim; and CME honoraria from Medscape, OncLive, PeerVoice, Physicians Education Resources, Targeted Oncology, Research to Practice, Axis, Peerview Institute, Paradigm Medical Communications, WebMD, MJH Life Sciences, Med Learning, Imedex, Answers in CME, and Clinical Care Options. T.W.L. reports consulting roles with Bayer, Cellectis, Novartis, Deciphera, Jumo Health, Y-mAbs Therapeutics and research support from: Bayer, Pfizer and Novartis. S.R.-C. declares that he has no competing interests.
We thank the patients and their families, many of whom travelled long distances to participate in the clinical studies. This study was funded by Bayer. Medical writing support was provided by Jim Heighway of Cancer Communications and Consultancy (Knutsford, UK), and was funded by Bayer.
Neurotrophins: roles in neuronal development and function.