Emerging Targeted Therapy for Tumors with NTRK Fusion Proteins

Tropomyosin-related kinase A, B, and C (TRKA, TRKB, and TRKC) are receptor tyrosine kinases encoded by the genes neurotrophic tyrosine receptor kinase 1, 2, and 3 (NTRK1, NTRK2, and NTRK3), respectively. TRKs are membrane-spanning receptors composed of extracellular ligand-binding, transmembrane, and intracellular ATP-binding domains (1). The extracellular domains of TrkA, TrkB, and TrkC exhibit high structural similarity, composed of three leucine-rich motifs flanked by two cysteine clusters and two immunoglobulin-like I set domains (2). The immunoglobulin-like regions are believed to encompass the ligand-binding sites. TRKs serve as signal receptors for neurotrophins, their cognate ligands. Nerve growth factor, brain-derived growth factor, and neurotrophin 3/4 are neurotrophic factors that activate TrkA, TrkB, and TrkC, respectively. TRKs play a pivotal role in the physiology, development, and function of the peripheral and central nervous systems (3, 4). Ligand–receptor interaction results in receptor dimerization and subsequent phosphorylation of the kinase domain. Activated kinases promote cell proliferation, differentiation, and survival by triggering downstream intracellular signal transduction pathways (refs. 5–9; Fig. 1).

NTRK rearrangements are the most common alterations in NTRK-mutated tumors (10). Our discussion herein focuses on the role of NTRK fusions in cancer and ongoing clinical trials involving TRK inhibitors.

NTRK oncogenic fusions arise from exact intrachromosomal or interchromosomal rearrangements that juxtapose the kinase domain-containing 3′ region of NTRK with the 5′ region of NTRK's gene partner. Chimeric fusion proteins promote tumorigenesis via constitutive ligand-free activation of intracellular biological pathways and signal transduction cascades that control cell-cycle progression, proliferation, apoptosis, and survival (11). Preclinical data demonstrated that chimeric oncogenic fusions may lead to partial or complete deletion of the immunoglobulin-like domain of TRK, which has an inhibitory influence on downstream signaling pathways in the absence of activating ligands (2). Several NTRK fusion partners have been identified so far and shown to contribute to the development of various cancer types (Table 1).

NTRK oncogenic fusions are infrequent but recurrent events observed in various types of congenital and acquired cancers (Table 2). The exact frequency of NTRK fusions in solid tumors remains unclear. The variations in frequencies among different studies and tumors subtypes may be biased by screened study cohorts and NTRK fusion detection techniques. In an analysis of over 11,000 patients conducted by Caris Diagnostics, TRK fusion proteins were detected by immunohistochemistry (IHC) in only 26 patients (0.23%; Gatalica and colleagues abstract TARG-17-A047). Most common fusions detected were TPM3 (Tropomyosin 3)-NTRK1, and ETV6 (ETS Variant 6)-NTRK3 (6 cases each). Furthermore, Stransky and colleagues (12) reported various NTRK fusions in 9 of 20 screened cancer samples. The estimated prevalence varies among histologic subtypes and fusion partners. The annual incidence of NTRK fusion–driven tumors is estimated to be 1,500 to 5,000 cases in the United States (13). Specific findings for NTRK fusions by tumor type are described below.

NTRK fusions are rare in lung cancer. Using next-generation sequencing (NGS) and FISH, Vaishnavi and colleagues (14) detected two novel NTRK1 gene fusion partners: Myosin Phosphatase Rho Interacting Protein (MPRIP) and CD74. The estimated frequency of NTRK fusion in this study was 3.3%, with 3 of 91 patients having NTRK1 fusions (14). The screened cohort, however, did not exhibit any other chromosomal alterations except for NRTK gene rearrangements. Furthermore, in a phase I study, Farago and colleagues (15) performed an anchored multiplex polymerase chain reaction (AMP) test to screen 1,378 cases of non–small cell lung cancer (NSCLC) for NTRK1, NTRK2, NTRK3, ALK (Anaplastic Lymphoma Kinase), and ROS1 (ROS Proto-Oncogene1) fusions. Utilizing FISH, NTRK1 gene fusions were detected in 2 patients: one with a TPM3-NTRK1 rearrangement, whereas the second patient, who had stage VI lung adenocarcinoma, harbored a novel NTRK1 gene partner, SQSTM1(Sequestosome 1). NTRK fusions were estimated to occur at a rate of 0.1% (95% confidence interval, 0.01%–0.5%; ref. 15). Because FISH was used to confirm the presence of these genetic alterations in both studies (14, 15), the discrepancy in the reported frequencies between these two studies may be attributed to their difference in sample size and screened patient populations.

NTRK gene fusions are also uncommon in colorectal cancer, and their estimated frequency likewise varies across different studies (0.5%–2.0%; refs. 16, 17). Several studies have demonstrated TPM3-NTRK1 fusions in colorectal cancer patients (16–18). Creancier and colleagues (17) performed IHC and quantitative reverse transcriptase-PCR tests to detect NTRK rearrangements in 408 cases of patients belonging to all clinical stages of colorectal cancer (I, II, III, and IV). Two cases (0.5%) were NTRK fusion positive. A TPR (Translocated Promoter Region)-NTRK1 oncogenic fusion was identified in a 53-year-old female with stage II, poorly differentiated adenocarcinoma. This patient also harbored wild-type (wt) KRAS (Kirsten RAt Sarcoma viral oncogene homolog), NRAS (Neuroblastoma RAS), and BRAF (B Rapidly Accelerated Fibrosarcoma), but is MSI-positive/microsatellite instability high (MSI-H) with loss of MLH1 (MutL Homolog 1)/PMS2 (Postmeiotic Segregation Increased 1 Homolog 2) and no MLH1 promoter methylation. In addition, TPM3-NTRK1 fusion was detected in a 66-year-old male with moderately differentiated adenocarcinoma of the left colon, also having wt KRAS, NRAS, and BRAF, and is MSI positive. A novel gene fusion with oncogenic potential, LMNA (Lamin A/C)-NTRK1, was detected by FISH in a patient with liver and adrenal gland metastases of colorectal cancer (19). In another study (20), a 61-year-old colorectal cancer patient with high MSI-H, wt RAS, BRAF, and EGFR harbored a novel SCYL3 (SCY1 Like pseudokinase protein 3)-NTRK1 rearrangement. Moreover, a retrospective study found NTRK fusions occurring in 2.5% of 2,044 heavily pretreated patients with metastatic colorectal cancer (21). This is in contrast with a study by Pietrantonio and colleagues (22), wherein they identified NTRK gene rearrangements in 13 of 346 (4%) metastatic colorectal cancer patients. Ten of the 13 patients [(76.9%), P < 00.1] with NTRK-rearranged tumors also had MSI-H status. Genetic alterations were detected using a targeted NGS technique utilizing Foundation One, MSK-IMPACT, and Minerva panel (22). Although NTRK gene fusions were screened and detected using highly specific techniques, the sample size was small and may have overestimated the exact prevalence of NTRK fusions in patients with MSI-H metastatic colorectal cancer.

In general, the estimated prevalence rate of NTRK fusions in patients with papillary thyroid carcinoma (PTC) does not exceed 12% (10) and varies among study populations according to geographical distribution and methods of detections (23). NTRK fusion oncogenes were also detected in 7 of 27 (26%) PTC patients in a pediatric population. Patients having NTRK-rearranged PTC presented with extensive disease and had worse prognosis than those with BRAF mutations (24). Although ETV6-NTRK3 is a rare somatic gene fusion in sporadic thyroid cancers, it was found to be more common in radiation-related tumors (25).

The rate of occurrence of NTRK fusions in brain tumors varies based on age group: 40% in pediatric versus 3% in adult-type tumors (12, 26, 27). NTRK2 fusions are most frequently detected in glioblastoma multiforme (GBM; Gatalica and colleagues abstract TARG-17-A047), whereas NTRK1 rearrangements were detected in only 3 of 115 (3%) patients with GBM so far (27).

Sarcomas represent a wide spectrum of uncommon tumors. Yet, in phase I and phase II studies which included 17 different NTRK fusion–positive tumor types detected by FISH or NGS, 21 of 55 (38%) patients were diagnosed with sarcoma including 3 patients with gastrointestinal stromal tumor (GIST; ref. 13). Drilon and colleagues (13) showed that sarcoma including soft tissue, infantile fibrosarcoma, and GIST comprises the largest cohort of cancer patients to harbor NTRK fusions in their study. Moreover, two other studies identified 1 patient each with ETV6-NTRK3 fusion GIST (28, 29). Interestingly, both patients exhibited WT KIT/PDGFR/BRAF disease. Overall, the estimated prevalence rate of NTRK fusions in sarcomas ranges from 1% in adult-type sarcomas to 92% in patients with congenital fibrosarcoma (12, 13, 30). Furthermore, Doebele and colleagues (31, 32) detected a novel LMNA-NTRK1 fusion using the Foundation One Heme assay in a 41-year-old woman with metastatic soft-tissue sarcoma to the lungs.

Secretory breast cancer and mammary analog secretory carcinoma (MASC) of the salivary gland are rare tumors with distinct clinical and pathologic features. However, they harbor the same underlying pathognomonic genetic alteration, ETV6-NTRK3, resulting from the chromosomal rearrangement t(12;15)(p12;q26.1) (33–35). ETV6-NTRK3 fusion is detected in 92% and 100% of secretory breast cancer and MASC cases, respectively (34, 36).

NGS provides a precise method to detect NTRK gene fusions (37). In addition to high sensitivity and specificity, it detects gene partners that might have clinical implications in future studies. Although NGS has changed the landscape of detecting chromosomal rearrangements driving tumors, several challenges remain for NTRK fusion testing (38). For example, the most popular commercially available DNA NGS panels, such as Foundation One, may not detect certain NTRK fusions. The addition of RNAseq to NGS testing has shown high sensitivity and specificity rates, 93% and 100% respectively, in detecting clinically actionable gene fusions (39). Data showed that RNAseq had led to unbiased results as well. In addition, RNAseq requires no prior knowledge of fusion partners or intronic/exonic break points. RNAseq use is now beyond research goals and has been incorporated into clinical practice (40).

Although FISH is considered the gold standard in detecting gene fusions, it can only detect a single target at a time. For instance, commonly used break apart FISH probe scan detect gene fusion but not the fusion partner. In addition, designing multiple probes for detecting NTRK fusions partners is cost ineffective and time consuming, making it not amenable for high-throughput screening (38).

Hechtman and colleagues (41) showed that pan-TRK fusion IHC test had sensitivity and specificity rates of 95.2% and 100%, respectively. Authors concluded that the pan-TRK fusion IHC test is a time- and tissue-efficient method for detecting NTRK fusions. However, researchers at MD Anderson Cancer Center were not able to replicate these findings (unpublished data).

A two-step diagnostic method incorporating rapid IHC screening that uses a cocktail of antibodies including anti–pan-Trk antibodies, followed by anchored multiplex PCR (AMP; ref. 38), showed that IHC screening had a 100% negative predictive value for excluding samples devoid of gene rearrangements (38).

The estimated prevalence rates of chromosome rearrangements range from 17% to 20% in cancer (12, 42). Over the past few years, various TKIs targeting the TRK family members have been developed and tested in clinical trials. The most promising thus far are summarized below, with a complete list provided in Table 3.

MGCD-516 is a novel small-molecule multikinase inhibitor that targets MET, AXL, MER, as well as members of the VEGFR, platelet-derived growth factor receptor (PDGFR), discoidin domain receptor tyrosine kinase 2 (DDR2), and TRK families (43). A phase Ia/II trial (NCT02219711) enrolling patients with advanced solid tumors is also ongoing, with MGCD-516 administrated at escalating doses within a 21- or 28-day cycle.

TSR-011 is an oral dual ALK (IC₅₀, 0.7 nmol) and pan-TRK (IC₅₀, <3 nmol) inhibitor that had been tested in a phase I/IIa clinical trial (NCT02048488) to determine its safety, tolerability, RP2D, and antitumor activity in patients with advanced tumors refractory to previous treatment with ALK inhibitors (44). TSR-011 was administered orally in dose escalation (30–480 mg 2 or 3 times a day) to 23 patients. This trial demonstrated that TSR-011 was safe and well tolerated at a fractionated dose of 60 mg daily. Dose-limiting toxicities (DLT) were prolonged QTc and dysesthesia. Three of 5 patients with ALK-rearranged NSCLC achieved partial response. TSR-011 efficacy is being investigated in ALK- and NTRK-rearranged tumors.

Entrectinib is a novel, highly potent oral ATP-competitive, pan-TRK, ROS1, and ALK TKI with low to sub-nanomolar antienzymatic efficacy (IC₅₀, 0.1–1.7 nmol/L; ref. 45). Two large multicenter 3+3 phase I clinical trials, ALKA-372-001 and STARTRK-1, were designed and conducted to determine the safety, efficacy, and antitumor activity of entrectinib in patients with advanced or metastatic solid tumors harboring NTRK1/2/3, ALK, and ROS1 rearrangements (46). A total of 119 patients (54 in ALKA-372-001, 65 in STARTRK-1) received treatment using different doses and schedules. Of them, only 60 patients possessed the aforementioned gene rearrangements. The majority of the patients (82%, 98/119) received three or more prior lines of treatment. Entrectinib was given to patients in the ALKA-372-001 on three different schedules, whereas entrectinib was administered daily for 28 days to patients in the STARTRK-1 trial. No DLTs were observed in the ALKA-372-001 trial, whereas grade 3 fatigue and grade 3 cognitive disturbances were observed in the STARTRK-01 trial with a daily entrectinib dose of 800 mg. The most common adverse events (AE) of any grade were fatigue (46%, 55/119), dysgeusia (42%, 50/119), paresthesia (29%, 34/119), nausea (28%, 33/119), and myalgia (23%, 27/119). A daily entrectinib dose of 600 mg was determined to be the maximum tolerated dose as well as the RP2D (46).

Twenty-five patients were enrolled in the phase II portions of the trials, four of whom had NTRK fusions. The median progression-free survival duration in patients with NTRK-rearranged tumors was not reached as of the data cutoff date (95% confidence interval, 3.5 months–not reached; ref. 46). The objective response rate was 100% (95% confidence interval, 44%–100%) in patients with NTRK-rearranged tumors, which include NSCLC (SQSMT1-NTRK1), metastatic colorectal cancer (LMNA-NTRK1), MASC (ETV6-NTRK3; refs. 15, 19, 46, 47), and glioneuronal tumor BCAN (Brevican)-NTRK1 (48). The majority of the responses occurred within the first two cycles of treatment. The authors concluded that entrectinib is a safe, well-tolerated pan-TRK/ROS1/ALK inhibitor, with patients having NTRK fusion–rearranged malignancies exhibiting the most clinically promising responses (46).

Larotrectinib is a highly selective, potent, ATP-competitive, and small-molecule pan-TRK inhibitor with an IC₅₀ in the low nanomolar range (31). The safety and efficacy of larotrectinib in treatment of locally advanced or metastatic solid tumors were investigated in a series of multicenter phase I and II clinical trials. A total of 55 patients with 17 different types of NTRK fusion–driven solid tumor [median age, 45 years (range, 0.3–76.0 years); Eastern Cooperative Oncology Group score ≤3] were enrolled in three trials: 8 in an adult phase I trial, 12 in the SCOUT pediatric phase I/II trial, and 35 in the NAVIGATE phase II basket trial (13). Larotrectinib was administered at 100 mg twice daily. Fifty-five percent of the patients were treatment-naïve or had received one prior line of treatment, whereas 31% had received at least three lines. As of the data cutoff date (July 17, 2017), the objective, partial, and complete response rates according to investigator assessment were 80% (95% confidence interval, 67%–90%), 64%, and 16%, respectively. Nine percent of the patients had stable disease, and 71% of responses were ongoing at 1-year follow-up. The median duration of response and progression-free survival had not been reached after a median follow-up durations of 8.3 months and 9.9 months, respectively. The median time to first response was 1.8 months. The one-year progression-free survival was 55% (13).

Eight patients (15%) needed dose reductions, with tumor regression maintained in all of them (one complete response, five partial responses, and one stable disease). Majority of AEs (93%) were grade 1or 2. Grade 3 or 4 AEs were anemia (11%), fatigue (5%), increased alanine transaminase or aspartate transaminase level (7%), nausea (2%), and dizziness (2%). Grade 3 treatment-related AEs were noted in less than 5% of the patients. Overall, larotrectinib is a safe, well-tolerated pan-TRK inhibitor in adults and children, and may be a new standard of care for NTRK-rearranged tumors.

Larotrectinib and entrectinib are the most clinically effective TKIs that target TRK fusion proteins. Although entrectinib also targets ROS1 and ALK fusion proteins, larotrectinib is, by far, the only highly selective pan-TRK inhibitor in clinical trials. Both drugs are safe and well tolerated, have ability to cross the blood–brain barrier, and control brain metastatic disease (13, 46). Although entrectinib's antitumor activity was tested in two phase I trials, responses were limited to 25 patients and only 4 had NTRK-rearranged tumors with partial response. Larotrectinib, on the other hand, showed robust outcomes in a series of phase I and II trials, which enrolled 55 patients with 17 unique NTRK fusion–positive solid tumors, who achieved overall response rate (ORR) and complete response rates of 80% and 16%, respectively (13). The FDA awarded Orphan Drug Designation to Entrectinib in 2015. Likewise, in May 2018, the FDA granted Priority Review for larotrectinib for the treatment of adult and pediatric NTRK-rearranged tumors.

Resistance to larotrectinib is driven by three different categories of mutations: (1) Solvent front mutations (NTRK1 p.G595R, NTRK3 p.G623R); (2) Gatekeeper mutations (NTRK1 p.F589L); and (3) xDFG (NTRK1 p.G667S, NTRK3 p.G696A; ref. 13). Solvent front and xDFG mutations involve the nucleotide-binding and activating loop of the kinase domain, respectively, and sterically change the larotrectinib-binding site that decreases larotrectinib's inhibitory properties and potency (13). Two patients with colorectal cancer who experienced resistance to larotrectinib treatment were found to have the NTRK p.G595R mutation (Table 4; ref. 13)

NTRK1 p.G595R and NTRK1 p.G667C are point mutations in the ATP-binding pocket of TrkA chimeric fusion proteins. These mutations were described in a colon cancer patient with LMNA-NTRK1 rearrangement who had developed resistance to entrectinib. Whereas higher, clinically achievable doses of entrectinib can overcome NTRK1 p.G667C–mediated resistance in cancer cells, no other TRK inhibitors available in clinical trials (e.g., larotrectinib, TSR-011) have demonstrated activity against NTRK1 p.G595R.

NTRK3 p.G623R is a point mutation that mediates the resistance of ETV6-NTRK3–rearranged tumors to treatment with either entrectinib or larotrectinib (13, 49, 50).

LOXO-195 is a novel and highly selective second-generation pan-TRK inhibitor developed to overcome NTRK1 p.G595R–mediated resistance to TRK inhibitors. NCT03215511 is a multicenter, open-label phase I/II clinical trial designed to evaluate the safety and efficacy of LOXO-195 in patients with NTRK-rearranged solid tumors.

Patients with NTRK-rearranged tumors have achieved robust and durable responses to treatment with TRK inhibitors in clinical trials. Hence, targeting NTRK fusion proteins is an effective strategy to improve outcomes in patients with NTRK-rearranged malignancies, and incorporating molecular and mutational analysis results into cancer treatment planning is crucial.

D.S. Hong is a consultant/advisory board member for Bayer. No potential conflicts of interest were disclosed by the other author.

The authors would like to thank Kathrina Marcelo-Lewis, PhD, of the Department of Investigational Cancer Therapeutics at The University of Texas MD Anderson Cancer Center, for assisting in the editing of this article.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.