An Oncogenic NTRK Fusion in a Patient with Soft-Tissue Sarcoma with Response to the Tropomyosin-Related Kinase Inhibitor LOXO-101

Tropomyosin-related kinase (TRK) is a receptor tyrosine kinase family of neurotrophin receptors that are found in multiple tissues types. Three members of the TRK proto-oncogene family have been described: TRKA, TRKB, and TRKC, coded by the NTRK1, NTRK2, and NTRK3 genes, respectively. The TRK receptor family is involved in neuronal development, including the growth and function of neuronal synapses, memory development and maintenance, and the protection of neurons after ischemia or other types of injury (1).

TRK was originally identified from a colorectal cancer as an oncogene fusion containing 5′ sequences from the tropomyosin-3 (TPM3) gene and the kinase domain encoded by the 3′ region of the neurotrophic tyrosine kinase, receptor, type 1 gene (NTRK1; refs. 2, 3). TRK gene fusions follow the well-established paradigm of other oncogenic fusions, such as those involving ALK and ROS1 that have been shown to drive the growth of tumors and can be successfully inhibited in the clinic by targeted drugs (4, 5). Oncogenic TRK fusions induce cancer cell proliferation and engage critical cancer-related downstream signaling pathways, such as MAPK and AKT, as recently reviewed by Vaishnavi and colleagues (6). The discovery of numerous oncogenic rearrangements involving not only NTRK1 but also the related TRK family members NTRK2 and NTRK3, and the demonstration that these oncogenes can be targeted by kinase inhibitors, has renewed interest in this oncogene family (6, 7). Although there are numerous different 5′ gene fusion partners identified, all share an in-frame, intact TRK kinase domain. This high variability observed in the 5′ gene partner is similar to ROS1, where at least 25 different 5′ partners have been identified; however, all studied variants seem to respond to targeted therapy (4, 8).

LOXO-101 (the hydrogen sulfate salt of ARRY-470, shown in Supplementary Fig. S1) is a small molecule that was designed to block the ATP binding site of the TRK family of receptors, with 2 to 20 nmol/L cellular potency against the TRKA, TRKB, and TRKC kinases (7). LOXO-101 was evaluated for off-target kinase enzyme inhibition against a panel of 226 non-TRK kinases at a compound concentration of 1,000 nmol/L and ATP concentrations near the K_m for each enzyme. In the panel, LOXO-101 demonstrated greater than 50% inhibition for only one non-TRK kinase (TNK2 IC₅₀, 576 nmol/L; ref. 7). There were neither relevant hERG inhibition nor prolonged QT findings in any preclinical species tested.

A multicenter phase I dose-escalation study in patients with advanced solid tumors (ClinicalTrials.gov no. NCT02122913) was initiated in 2014 to evaluate the safety and pharmacokinetics of LOXO-101. Patients are not required to have TRK fusions or other TRK alterations. Patients are dosed once or twice daily for 28 days of continuous dosing in escalating cohorts. Patients remain on treatment until evidence of disease progression or intolerable toxicity. Preliminary pharmacokinetic and safety data from this ongoing study indicate that free plasma levels of LOXO-101 are at biologically relevant concentrations to inhibit TRK oncogenes.

This report describes the first and only patient enrolled to date on this dose-finding study with a demonstrated TRK fusion and the rapid clinical tumor regression seen with the treatment of a selective TRK inhibitor, LOXO-101.

We and others have previously demonstrated that cancer cells harboring oncogenic TRK can be inhibited in vitro and in vivo by drugs that target the TRK kinase family (7, 9). To demonstrate activity of LOXO-101 in preclinical models harboring different variants of TRK oncogenes, we performed proliferation assays in three cell line models harboring TRK gene fusions: CUTO-3.29 is derived from a patient with lung adenocarcinoma harboring the MPRIP–NTRK1 gene fusion; the KM12 cell line is a colorectal cancer cell line harboring the TPM3–NTRK1 fusion (7); and the MO-91 cell line is derived from an acute myeloid leukemia patient harboring the ETV6–NTRK3 fusion (10). Measurement of proliferation following treatment with LOXO-101 demonstrated a dose-dependent inhibition of cell proliferation in all three cell lines (Fig. 1A–C). The IC₅₀ was less than 100 nmol/L for CUTO-3.29 (Fig. 1A) and less than 10 nmol/L for KM12 (Fig. 1B) and MO-91 (Fig. 1C), consistent with the known potency of this drug for the TRK kinase family. Consistent with the inhibition of cellular proliferation, we also observed inhibition of phosphorylation of the MPRIP–TRKA oncoprotein and ERK1/2 in CUTO-3.29 cells (Fig. 1D), inhibition of TPM3–TRKA, pAKT, and pERK1/2 in KM12 cells (Fig. 1E), and inhibition of the TEL–TRKC oncoprotein (encoded by ETV6–NTRK3), pAKT, and pERK1/2 in MO-91 cells using low doses of LOXO-101 (Fig. 1F). pAKT was not inhibited in the CUTO-3.29 cells by LOXO-101, suggesting that AKT signaling is not TRK-dependent in this tumor.

We next determined whether LOXO-101 could inhibit tumor growth in vivo. Athymic nude mice injected with KM12 cells were treated with LOXO-101 orally daily for 2 weeks (Fig. 1G). Dose-dependent tumor inhibition was observed, demonstrating the ability of this selective compound to inhibit tumor growth in vivo. Together, these and previously published results indicate that TRK fusions are constitutively activated, regulate critical downstream signaling pathways, such as MAPK and AKT, and are inhibited by a highly specific TRK inhibitor.

In September 2014, a 41-year-old woman presented with a firm mass in her left groin. Initial imaging confirmed a 10-cm mass within the musculature of the anterior thigh; open biopsy revealed an undifferentiated sarcoma. Initial staging scans demonstrated multiple bilateral 4- to 13-mm pulmonary nodules consistent with metastatic disease.

The patient's diagnostic, open tumor biopsy was tested using the FoundationOneHeme panel (Foundation Medicine). This multitarget comprehensive genomic profiling (CGP) assay using DNA and RNA sequencing of hundreds of cancer-related genes demonstrated the presence of a gene fusion encoding exons 1 to 2 of the lamin A/C gene (LMNA) and exons 11 to 17 of the NTRK1 gene, resulting in the LMNA–NTRK1 fusion gene (Fig. 2A). CGP also showed the loss of the tumor suppressor CDKN2A/B (not shown), but no other known oncogenic mutations.

Subsequently, a break-apart FISH assay performed on the patient's tumor sample exhibited a predominantly single 3′ NTRK1 (red fluorescence signal) pattern in 64% of tumor nuclei, consistent with a genomic alteration involving the NTRK1 gene locus, most likely secondary to a genomic deletion between the two genes given the location and orientation of both LMNA and NTRK1 on the large arm of chromosome 1 (Fig. 2B). mRNA expression of the LMNA–NTRK1 fusion transcript from the chromosomal deletion was confirmed by RT-PCR and sequencing (Fig. 2C).

In order to assess both protein expression and functional activity of the fusion oncoprotein, we applied a proximity ligation assay (PLA) to the patient's tumor sample. PLAs are unique because they can detect functional signaling complexes between a kinase and one of its adaptors in situ (11). In this assay, we measured TRKA complexed with its preferred adaptor, SHC1, which binds to Y496 in the TRKA kinase domain (Supplementary Fig. S2; ref. 12). We have validated this assay in both human cell lines and formalin-fixed patient-derived tumor xenograft (PDX) tumor samples (Supplementary Fig. S3). RNAi knockdown of NTRK1 disrupts TRKA–SHC1 complexes in the CUTO-3 cell line harboring the MPRIP–NTRK1 fusion gene (Supplementary Fig. S3A–S3C), as does inhibition with the pan-TRK selective inhibitor LOXO-101 (Supplementary Fig. S3D and S3E). The TRK PLA detects functional signaling complexes in a formalin-fixed, paraffin-embedded (FFPE) tumor sample from a PDX, CULC001, harboring the MPRIP–NTRK1 gene fusion but not the PDX CULC002, which does not harbor a known oncogenic driver mutation (Supplementary Fig. S3F and S3G). The TRK–SHC PLA can also detect non-oncogenic signaling complexes as shown by a positive signal in a region of peripheral nerve tissue of the CULC001 PDX, where the TRK family of receptors have high expression and activity mediated by the neurotrophins (Supplementary Fig. S3H and S3I; ref. 13). Application of this assay to the patient's tumor sample demonstrated robust signaling associated with tumor nuclei, but only a weak signal in the blood vessel (human endothelial cells express TRKA; ref. 14) consistent with oncogenic signaling by the LMNA–TRKA oncoprotein (Fig. 2D and E). The TRK–SHC1 PLA demonstrated a negative result on a tumor sample from an ALK⁺ non–small cell lung cancer patient, whereas the ALK–GRB2 PLA was positive (Supplementary Fig. S4A and S4B), further demonstrating the ability of this assay to detect oncogenic signaling in human tumor samples.

The presence of the LMNA–NTRK1 fusion detected by a FoundationOneHeme assay and then validated by FISH and RT-PCR combined with the evidence of TRKA protein expression and functional activity of the TRK pathway in the patient's tumor sample suggested that the patient has a TRK-driven cancer suitable for treatment with a TRK-specific inhibitor.

After the initial diagnosis in September 2014, an aggressive treatment plan was agreed upon, and she enrolled on a phase II trial of sorafenib with chemotherapy, preoperative radiation, and limb-sparing surgery (ClinicalTrials.gov no. NCT02050919). After 2 weeks of sorafenib 400 mg daily, she received epirubicin 30 mg/m² daily and ifosfamide 2,500 mg/m² daily with mesna for 3 consecutive days, with continuation of daily sorafenib. The tumor became progressively more painful during these 5 weeks of systemic therapy. During simulation for preoperative radiation, extension of the tumor was noted cranially within the psoas muscle, precluding the safe administration of effective radiation doses due to predicted bowel toxicity. She therefore came off protocol and proceeded to surgical resection. Resection of the primary tumor achieved negative margins, and review of the pathologic specimen confirmed 90% tumor necrosis. A restaging chest CT (shown in Fig. 3A) obtained 9 weeks after initial scans showed worsening metastatic disease, with the largest nodule now measuring 18 mm. The patient's postoperative course was complicated by a polymicrobial wound infection, requiring repeated wound debridement and prolonged antibiotic therapy. Repeat chest CT was obtained before resumption of chemotherapy and demonstrated dramatic progression over the prior 9 weeks, with multiple pulmonary nodules greater than 3 cm, the largest nearly 7 cm, and a large left pleural effusion. In February 2015, after placement of a tunneled pleural drain and initiation of supplemental home oxygen, the patient received one dose of doxorubicin 75 mg/m² while awaiting enrollment on the LOXO-101 clinical trial.

Based on multiple lines of genetic and functional biomarker data suggesting the presence of a TRK driver oncogene, the patient was referred for consideration of enrollment into the phase I trial of LOXO-101 (ClinicalTrials.gov no. NCT02122913) in February 2015. In March 2015, the patient was found eligible for the trial and provided written informed consent. The baseline CT scan showed continued tumor progression with multiple large pulmonary metastases in both lungs, although the pleural effusion had resolved following placement of the pleural drain (Fig. 3B). On clinical presentation, the patient had significant exertional dyspnea and required 5 L of supplemental oxygen to maintain an oxygen saturation of 90%. Baseline laboratory values were notable for an elevated CA125 tumor marker level (Supplementary Fig. S5). The patient was assigned to receive LOXO-101 100 mg twice daily. During cycle 1, the patient was seen weekly for pharmacokinetic and safety data collection. No drug-related adverse events were noted, and the patient experienced weekly improvement in her exertional dyspnea during this 4-week period. The CA125 levels normalized over cycle 1. A CT was performed prior to the start of cycle 2 day 1, which demonstrated a marked improvement in multiple pulmonary metastases and was deemed a partial response by RECIST 1.1 (Fig. 3C). Additional CT scans on cycle 5 day 1 (after 4 months of LOXO-101 dosing) demonstrated almost complete tumor disappearance of the largest tumors (Fig. 3D). Clinically, the patient had significantly improved exertional dyspnea and was no longer requiring supplemental oxygen, with an oxygen saturation of 97% on room air. After 4 months of dosing, the patient did not have any adverse events that were attributed to LOXO-101.

TRK fusions have been identified as a rare subset of a number of diverse tumor histologies (15, 16). The genomic structures of these TRK fusions and the inferred protein structures have suggested that these genes are oncogenic drivers, and previously reported preclinical work in addition to that shown here supports this hypothesis (6, 7, 9). The first patient with a documented NTRK fusion in this clinical trial demonstrated substantial tumor regression when treated with a selective inhibitor of TRK, providing clinical validation of this molecular target.

TRK fusions are relatively unknown in soft-tissue sarcoma (STS), and this clinical response also indicates the potential for actionable molecular targets for this disease. There is a large unmet clinical need for better treatments in advanced STS, a heterogeneous group of tumors of mesenchymal origin encompassing more than 50 histologic subtypes. More than 12,000 Americans will be diagnosed with STS this year, and 4,700 people will die from the disease (17). Inoperable metastatic STS is treated palliatively, with a median overall survival of approximately 1 year and a 5-year survival rate of less than 20% (18). The standard of care for most advanced STS subtypes remains single-agent doxorubicin, with an expected response rate of approximately 20% (19–23). Although a significant proportion of sarcomas have known chromosomal translocations that are valuable for diagnostic purposes, these alterations have generally not informed treatment decisions and have often proven to be difficult therapeutic targets. The majority of sarcoma-associated translocations involve genes encoding transcription factors, such as the pathognomonic EWS–FLI1 fusion of Ewing sarcoma. Fusions involving kinases or growth factors have been identified in several sarcoma subtypes characterized by low metastatic potential, including dermatofibrosarcoma protuberans (DFSP). DFSP is driven by a COL1A1–PDGFB fusion in fibroblasts, resulting in unregulated production of mature PDGFB. Treatment with imatinib results in dramatic tumor shrinkage (24). Similarly, inflammatory myofibroblastic tumors (IMT) harbor fusions involving ALK or ROS1, and treatment with crizotinib results in rapid clinical response (25–27). The only previously known defining TRK fusion in sarcoma is the ETV6–NTRK3 fusion in infantile fibrosarcoma, a locally aggressive but rarely metastasizing mesenchymal tumor exclusively affecting children under the age of one.

The LMNA–NTRK1 gene fusion has been previously reported in Spitzoid nevi and is constitutively activated when expressed in cells resulting in activation of ERK1/2, AKT, and PLCγ, demonstrating its oncogenicity (9). LMNA has also been identified as a gene fusion partner of ALK (26). Foundation Medicine has previously tested 1,272 STS samples with the FoundationOneHeme CGP Test, resulting in the detection of 8 NTRK1 or NTRK3 fusions, including the patient described in this case report (Table 1). Thus, Foundation Medicine's detection rate for NTRK fusions in STS is 0.63% [95% confidence interval (CI), 0.32%–1.24%], which we believe represents a lower estimate of the true prevalence of these genetic lesions due to the complexities of sarcoma disease ontology classification and the likelihood of false negatives. Notably, 6 of the 8 sarcoma patients with NTRK fusions are under the age of 25 (Fisher exact test, P value = 4 × 10⁻⁴), and 4 of the 8 are under the age of 5 (Fisher exact test, P value = 2 × 10⁻⁵), indicating an increased detection rate of NTRK fusions among pediatric patients (4.1%; 95% CI, 1.8%–9.3%) and particularly those under the age of 5 (14.3%; 95% CI, 5.7%–31.5%). Also of interest, one of the gene fusions detected combines the majority of the NTRK3 gene (exons 1–17) to the 3′ end of the HOMER2 gene (exons 2–9), which contains a dimerization domain (coiled-coil domain), and therefore represents a 3′ gene fusion event that has been described for multiple other RTK-encoding genes, such as EGFR, AXL, and FGFR3 (17, 28, 29). Because this patient also had a CDKN2A/B deletion, we decided to examine whether co-occurrence of these two lesions was common in sarcomas. We found that 4 of the 8 sarcoma patients with an NTRK fusion also harbored a CDKN2A/B deletion (Table 1); 14% of all sarcomas tested by Foundation Medicine have been found to harbor CDKN2A/B deletions (30).

LOXO-101 (ARRY-470) has previously been demonstrated to have no effect on cell lines that lack a TRK oncogene (7). In this report, three different cell line models, each with different tumor histologies and TRK fusions, demonstrated dose-dependent inhibition by LOXO-101 in vitro. Although there were subtle differences in signaling, such as AKT dependence on TRK signaling, it remains unclear whether these differences are caused by the cell context (histologic origin or genetic context) or the gene fusion itself. For example, the persistence of AKT signaling in the CUTO-3.29 cells may account for the slightly higher IC₅₀ of this cell line compared with KM12 and MO-91 cells where pAKT was reduced with LOXO-101. Identification of additional patients and cell line models may help answer this question in the future. Finally, murine xenograft experiments with a colorectal cancer cell line harboring an NTRK1 fusion demonstrated dose-dependent inhibition of tumor growth by LOXO-101.

The tumor regression and clinical response seen in this LMNA–NTRK1 fusion patient provide clinical validation of a new molecular target in oncology, and establish that at least this NTRK1 fusion is a molecular driver of this patient's clinical disease. Additional follow-up of this patient to understand the durability of the response and the potential future mechanisms of resistance will be important to investigate how TRK inhibition has affected this tumor. Additional patients with this and other TRK fusions will need to be treated with this or other TRK inhibitors to understand the clinical utility of TRK inhibition in this population and other potential TRK fusion populations. Preclinical data using diverse cancer cell lines originating from different tumor types suggest that the oncogene driver may be the dominant factor in determining response to targeted therapy, rather than the histologic subtype. Molecular evaluation of all tumor types is increasingly essential to properly identify investigational and approved drugs to better treat molecularly defined cancers.

NCT02122913 is an ongoing multicenter phase I dose-escalation study evaluating the safety and pharmacokinetics of LOXO-101, a selective pan-TRK, in unselected patients with metastatic or advanced solid tumors without standard therapy options. The study is approved by Institutional Review Boards (IRB) at all institutions that enroll patients, and eligible patients provide written informed consent to participate. The study is sponsored by Loxo Oncology and is conducted in accordance with the Declaration of Helsinki and Good Clinical Practices. LOXO-101 is a 3-urea-substituted pyrazolo[1,5a]-pyrimidine (7) provided in 100 mg capsules by the clinical trial sponsor (Loxo Oncology). CAS registry numbers for LOXO-101 include 1223403-58-4 and 1223405-08-0. Enrolled patients receive escalating doses of LOXO-101 according to a modified 3+3 design, and receive LOXO-101 daily or twice daily until intolerable toxicity, disease progression, or withdrawal of consent. In patients with measurable disease, efficacy is assessed per RECIST 1.1 criteria.

DNA and RNA were extracted and adaptor-ligated sequencing libraries were captured by solution hybridization using custom bait-sets targeting 405 cancer-related genes and 31 frequently rearranged genes by DNA-seq, and 265 frequently rearranged genes by RNA-seq (FoundationOneHeme; Foundation Medicine). All captured libraries were sequenced to high depth (Illumina HiSeq) in a Clinical Laboratory Improvements Amendment (CLIA)–certified College of American Pathologists (CAP)–accredited laboratory (Foundation Medicine), averaging >500× for DNA and >6M unique pairs for RNA. Sequence data from gDNA and cDNA were mapped to the reference human genome (hg19) and analyzed through a computational analysis pipeline to call genomic alterations present in the sample, including substitutions, short insertions and deletions, rearrangements, and copy-number variants.

NTRK1 break-apart FISH was performed on 4-μm slides from FFPE tumor samples as previously described using the Vysis LSI NTRK1 (Cen) SpectrumGreen and Vysis LSI NTRK1 (Tel) SpectrumRed (Abbott Molecular; # 08N43-030 and 08N43-020, respectively; ref. 7).

RT-PCR was performed as previously described using the forward primer to LMNA (LMNA F1, 5′gagggcgagctgcatgat3′; ref. 9) and the reverse primer to NTRK1 (NTRK1 R1, 5′cggcgcttgatgtggtgaac3′). DNA sequencing of the RT-PCR product was performed using Sanger DNA Sequencing at the Pathology Core at the University of Colorado.

CUTO-3 cell lines were initiated from the malignant pleural effusion of a patient with stage IV lung adenocarcinoma harboring the MPRIP–NTRK1 gene fusion as previously described (7, 31); IRB-approved informed consent to derive immortal cell lines was obtained from the patient. CUTO-3.29 was derived from the same malignant pleural effusion by single-cell cloning. KM12 and MO-91 have been previously described (7, 10). All cell lines in this study were authenticated by DNA fingerprinting by short tandem repeat analysis using the AmpFLSTR Identifiler PCR Amplification Kit (Cat. # 4322288 from Invitrogen Life Technologies/Applied Biosystems) in July 2015 by the Barbara Davis Center Molecular Biology Service Center at the University of Colorado.

IRB-approved informed consent to generate patient-derived murine xenografts was obtained from the relevant patients. Animal care and procedures were approved by the Institutional Animal Care and Use Committee Office of the University of Colorado Anschutz Medical Campus. Pleural fluid (CULC001) from a lung adenocarcinoma patient harboring an MRPIP–NTRK1 gene fusion was centrifuged, and the resulting cell pellet was suspended in 5 mL ACK buffer (Lonza) for 2 minutes allowing for the complete lysis of red blood cells. Lysis was halted by the addition of 20 mL PBS and centrifuging the samples. The pellet was washed twice with PBS prior to being suspended in DMEM-supplemented media as above. One hundred microliters of cells (1 × 10⁶ per flank) suspended in a 1:1 mix of DMEM and Matrigel (BD) were injected subcutaneously into the flanks of 5 nude mice. Tumor tissue from an oncogene-negative lung adenocarcinoma patient (CULC002) was cut into 3 × 3 × 3 mm pieces that were transferred to DMEM supplemented with 10% FBS, 200 units/mL penicillin, and 200 μg/mL streptomycin. Tumor pieces were dipped in Matrigel (Corning) and inserted into incisions on each flank of 5 nude mice. Propagation and maintenance of resulting xenografts were previously described (32).

Cells were seeded onto glass coverslips (in a 48-well plate) or chamber slides at 25k to 75k cells/well. Cells were treated with the indicated doses and times then fixed for 15 minutes by shaking at room temperature in 4% paraformaldehyde. Cells were rinsed twice in PBS, and then the Duolink in situ PLA Kit from SigmaAldrich in mouse/rabbit (Red) was used according to the manufacturer's protocol (cat. # DUO92101). Antibody concentrations were optimized using immunofluorescence prior to PLA experiments. FFPE tissue PLAs from mice or patients were prepared as described in histology. In addition, samples were treated with 300 mmol/L glycine for 15 minutes prior to the blocking step; otherwise, the assay was performed according to the manufacturer's protocol. Cells were mounted using Prolong gold anti-fade reagent (with DAPI) and cured overnight prior to imaging. Images were taken either on a Nikon standard inverted fluorescent microscope at 40×, or on the 3I Marianas spinning disc confocal in the University of Colorado Anschutz Medical Campus Advanced Light Microscopy Core at 40× or 100×. The following antibodies were used: TRK (C17F1; 1:100) and ALK (D5F3; 1:200–1:400) from Cell Signaling Technology, SHC1 (1:50) from Novus, and GRB2 (610111; 1:200–1:400) from BD.

All proliferation assays were performed in media supplemented with 5% FBS as previously described, using Cell Titer 96 MTS (Promega). Cells were seeded 500 to 2,000 cells per well and treated for 72 hours at the drug concentrations described on each graph. Each assay was performed in triplicate in at least 3 independent biologic replicates. Data were plotted, and IC₅₀ values were calculated using GraphPad software.

Athymic nude mice were obtained from Harlan Laboratories and maintained under aseptic conditions. The care and treatment of experimental animals were in accordance with institutional guidelines. KM12 cells (5 × 10⁵) were injected subcutaneously into the dorsal flank area of the mice. Tumor volume was monitored by direct measurement with calipers and calculated by the formula: length × (width²)/2. Following the establishment of tumor and when the tumor size was between 150 and 200 mm², mice were randomly selected to receive diluent, 60 mg/kg/dose, or 200 mg/kg/dose of LOXO-101. LOXO-101 was administered by oral gavage once daily for 14 days. After the last dose, tissue and blood were collected at 3, 6, and 24 hours after treatment.

Immunoblotting was performed as previously described (7). Briefly, cells were lysed in RIPA buffer with Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific) and diluted in loading buffer (LI-COR Biosciences). Membranes were scanned and analyzed using the Odyssey Imaging System and software (LI-COR). The following antibodies were used from Cell Signaling: pTRK Y490 (rabbit polyclonal, #9141; 1:1,000), pERK1/2 XP T202/Y204 (#9101; 1:2,000), ERK1/2 (1:2,000), pAKT S473 (rabbit mAb, #4060; 1:2,000), and AKT mouse clone D3A7 (#2920; 1:2,000). TRK (C-14) rabbit polyclonal antibody (1:1,000) was purchased from Santa Cruz Biotechnology. GAPDH (MAB374; 1:5,000) and pTYR (4G10; 1:2,000) were from Millipore.

CIs for the detection rate of NTRK fusions in samples from patients with sarcoma were calculated using the 1-sample proportions test. The disease histology classification was based on the Foundation Medicine disease ontology as of April 2015. The enrichment of NTRK fusions in younger patient groups was tested using the Fisher exact test. All statistical testing was performed in R v 3.1.3. Comparison of treatment arms for the in vivo mouse xenograft studies was performed using repeated measures ANOVA with Bonferroni correction using GraphPad software.

R.C. Doebele reports receiving commercial research grants from Abbott Molecular, Loxo Oncology, and Mirati Therapeutics; is a consultant/advisory board member for Loxo Oncology; and has received licensing fees from Abbott Molecular (licensed patent), Blueprint Medicines (licensed cell line), and Chugai (licensed cell line). A.T. Le has received licensing fees from Abbott Molecular (licensed patent). M. Varella-Garcia reports receiving a commercial research grant from Abbott Molecular and has ownership interest in a patent for use of FISH probes for molecular diagnosis. D.L. Aisner has received speakers bureau honoraria from AstraZeneca and Clovis. Y. Li has ownership interest in Foundation Medicine. P.J. Stephens has ownership interest in Foundation Medicine. D. Morosini has ownership interest in Foundation Medicine. J.A. Low has ownership interest in Loxo Oncology. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R.C. Doebele, L.E. Davis, J.A. Low

Development of methodology: R.C. Doebele, A. Vaishnavi, A. Estrada-Bernal, A. Jimeno, M. Varella-Garcia, B.B. Tuch, N. Nanda, J.A. Low

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.C. Doebele, A.T. Le, A. Estrada-Bernal, S. Keysar, A. Jimeno, M. Varella-Garcia, D.L. Aisner, D. Morosini, M. Fernandes, J.A. Low

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.C. Doebele, A. Vaishnavi, A.T. Le, A. Estrada-Bernal, M. Varella-Garcia, Y. Li, P.J. Stephens, B.B. Tuch, N. Nanda, J.A. Low

Writing, review, and/or revision of the manuscript: R.C. Doebele, L.E. Davis, A. Vaishnavi, A.T. Le, A. Estrada-Bernal, S. Keysar, A. Jimeno, M. Varella-Garcia, D.L. Aisner, Y. Li, P.J. Stephens, D. Morosini, B.B. Tuch, M. Fernandes, N. Nanda, J.A. Low

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.E. Davis, A.T. Le, B.B. Tuch, M. Fernandes, N. Nanda, J.A. Low

Study supervision: R.C. Doebele, M. Fernandes, J.A. Low

The authors thank the patient and her family for her participation in this clinical trial, and prior patients at the University of Colorado for graciously donating the tumor tissue that was used to generate cell lines and PDXs. They also thank Nikki Ayodeji and the remainder of the clinical trial staff at the University of Colorado, and Carol Hill at Loxo Oncology.

This study was funded by the State of Colorado and University of Colorado Technology Transfer Office Bioscience Discovery Evaluation Grant Program, the University of Colorado Lung Cancer SPORE (funded by NCI grant P50CA058187), a Loxo Oncology research grant, and a V Foundation Scholar Award (to R.C. Doebele). The study was also supported by NIH/NCI CCSG P30CA046934 (Molecular Pathology Shared Resource; to M. Varella-Garcia and D.L. Aisner).

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.