NTRK gene fusions as novel targets of cancer therapy across multiple tumour types

Introduction

The treatment of solid tumours is dramatically changing in recent years thanks to the enhancement of molecular diagnostic technologies leading to identification of an increasing number of specific actionable oncogenic abnormalities such as gene activating point mutations, in-frame insertions/deletions and amplification or rearrangements. The concept of precision medicine consists in the accomplishment of therapy individualised to each tumour by exploiting these alterations as predictive biomarkers as well as targets of therapy. Neurotrophic tropomyosin receptor kinase (NTRK) gene rearrangements have recently emerged as targets for cancer therapy, because novel compounds have been developed that are selective inhibitors of the constitutively active fusion proteins that arise from these molecular alterations. Developments in this field are being aided by next generation sequencing methods as tools for unbiased gene fusion discovery. In this article, we review the role of NTRK gene fusions across several tumour histologies, and the promises and challenges of targeting such genetic alterations for cancer therapy.

Tropomyosin receptor kinase (Trk) family of receptors

The Trk receptor family comprises three transmembrane proteins referred to as Trk A, B and C (TrkA, TrkB and TrkC) receptors, and are encoded by the NTRK1, NTRK2 and NTRK3 genes, respectively. These receptor tyrosine kinases (TK) are expressed in human neuronal tissue, and play an essential role in both the physiology of development and function of the nervous system through activation by neurotrophins (NTs).1 The latter are specific ligands known as nerve growth factor (NGF) for TrkA, brain-derived growth factor (BDGF), and NT-4/5 for TrkB and NT3 for TrkC, respectively.2 All three Trk receptors are structured with an extracellular domain for ligand binding, a transmembrane region and an intracellular domain with a kinase domain. The binding of the ligand to the receptor triggers the oligomerisation of the receptors and phosphorylation of specific tyrosine residues in the intracytoplasmic kinase domain. This event results into the activation of signal transduction pathways leading to proliferation, differentiation and survival in normal and neoplastic neuronal cells1 (figure 1). The binding of TrkA receptor by NGF causes the activation of the Ras/Mitogen activated protein kinase (MAPK) pathway, which leads to increased proliferation and cellular growth through extracellular signal-regulated kinase (ERK) signalling. Other pathways such as phospholipase C-γ (PLCγ) and PI3K are also activated. TrkC coupling with NT3 causes preferential activation of the PI3/AKT pathway preventing apoptosis and increasing cell survival, whereas TrkB transduces the BDNF signal via Ras-ERK, PI3K and PLCγ pathway, resulting in neuronal differentiation and survival.1 The Trk receptor kinases play a key role in central and peripheral nervous system development as well as in cell survival. The proper regulation of Trk receptor levels and their activation is critically important in cell functioning, and the upregulation of Trk receptors has been reported in several central nervous system-related disorders (eg, TrkB in epilepsy, neuropathic pain, or depression).3

The NTRK1 gene is located on chromosome 1q21-q22,4 and its mutations disrupting the function of the TrkA protein are found in patients affected by congenital insensitivity to pain with anhidrosis (CIPA) syndrome.5 In 1999 Indo et al cloned the full-length NTRK1 human gene encoding a 790-residue or 796-residue protein (TrkA receptor) with an intracellular domain containing a juxtamembrane region, a TK domain and a short C terminal tail.6

The NTRK2 gene is mapped on chromosome 9q22.17 and contains 24 exons,8 coding for a protein of 822 amino acid residues (TrkB receptor). The full-length TrkB receptor contains an N-terminal signal sequence, followed by a cysteine-rich domain, a leucine-rich domain, a second cysteine-rich domain, 2 immunoglobulin (Ig)-like domains that make up the BDNF-binding region, a transmembrane domain, a Src homology 2 domain containing (SHC)-binding motif, a TK domain near the C terminus and a C-terminal PLCγ-docking site.

The NTRK3 gene is located on chromosome 15q25,9 and its transcription product known as TrkC was isolated and characterised by Lamballe et al10 in 1991. TrkC receptor is a glycoprotein of 145 kD preferentially expressed in the human hippocampus, cerebral cortex and in the granular cell layer of the cerebellum.

Molecular alterations of NTRK genes in various malignancies across histologies

Gene fusions of NTRK genes represent the main molecular alterations with known oncogenic and transforming potential.11 Less common oncogenic mechanisms that have been described are in-frame deletion of NTRK1 in acute myeloid leukaemia12 and a TrkA alternative splicing in neuroblastoma.13 In all reported Trk oncogenic gene fusions, the 3’ region of the NTRK gene is joined with a 5’ sequence of a fusion partner gene by an intrachromosomal or interchromosomal rearrangement, and the oncogenic chimaera is typically a constitutively activated or overexpressed kinase (figure 2). Table 1 shows the NTRK fusions reported to date and their associated cancer types.

Figure 2

The chimeric Trk protein, composed by the TK domain with ATPase activity and a TM loop along with a fusion partner. Oligomerisation of the chimeric protein is the main proposed mechanisms for increased tumour cell survival and proliferation via the known pathways of Trk receptors. (A) NTRK1 gene with 17 exon sequences and respective TrkA protein regions.5 (B) Mechanism of entrectinib (E) action and the known mechanisms of acquired resistance (point mutation G595R and G567C) to E.50 Cys, cysteine clusters; Ig1 and Ig2, first and second immunoglobulin-like motifs, respectively; LRM, leucine-rich motifs; SP, signal peptide; TK, tyrosine kinase; TM, transmembrane.

Colorectal adenocarcinoma: It has been believed for more than three decades since the seminal work of Fearon and Vogelstein14 on molecular carcinogenesis of colorectal carcinoma (CRC), that rearrangement of oncogenes are of such low prevalence, as compared with DNA sequencing alteration or amplifications, that their role is almost negligible in the genesis of this tumour type. The first published report of a NTRK rearrangement in CRC dates back to 1986,15 when a TPM3-NTRK1 translocation was detected in a tumour biopsy, and thereafter very little has been reported about these gene defects in CRC. However, the NTRK gene fusions and their oncogenic potential in this tumour as oncogenes may have been underestimated, mostly because of the absence, until recently, of targeted therapies exploiting these gene abnormalities. These circumstances resemble what previously occurred with ALK (Anaplastic Lymphoma Kinase) gene fusions in non-small cell lung cancer (NSCLC), in which despite evidence of ALK fusions since 2007,16 the scientific interest was raised only after synthesis of compounds with ALK-specific inhibitory activity.17 In particular, in 2014, Ardini et al reported the characterisation of the TPM3-NTRK1 gene rearrangement as a recurring, although rare, event in CRC, and also discovered entrectinib (NMS-P626; RXDX-101) as a novel, highly potent and selective pan-Trk inhibitor. Entrectinib suppressed TPM3-TRKA phosphorylation and downstream signalling in KM12 cells and showed remarkable antitumour activity in mice bearing KM12 tumours. Also, in 2015, Créancier et al18 reported the 0.5% prevalence of NTRK fusions in 408 CRC clinical samples, including a TPM3-NTRK1 (TRK-T2 fusion). Recently, in the molecular screening within the phase I first-in-human study of entrectinib (EudraCT number: 2012-000148-88), an abnormal expression of the TrkA protein was identified in tumour and liver metastases of a patient with CRC refractory to standard therapy, and molecular characterisation unveiled a novel LMNA-NTRK1 rearrangement within chromosome 1, with oncogenic potential. The patient was treated with entrectinib, achieving objective partial response with decrease of metastatic lesions in the liver and adrenal gland.19 Further, as a part of the molecular screening of the Memorial Sloan Kettering IMPACT (MSK-IMPACT) programme,20 Braghiroli et al21 reported a 4% (2 of the 49 cases) of incidence of NTRK unidentified fusions in appendiceal adenocarcinoma.

Lung adenocarcinoma: Gene rearrangements have already emerged as therapeutic targets in NSCLC,22 since Food and Drug Administration (FDA) and European Medicines Evaluation Agency (EMEA) approval of crizotinib for patients with NSCLC harbouring EML4-ALK translocations. In 2013, Vaishnavi et al23 described two different gene fusions involving the NTRK1 gene that lead to constitutive TrkA TK domain activation. The first was characterised by a rearrangement of the 5’ portion of the myosin phosphatase Rho-interacting protein (MPRIP) gene fused to the 3’ portion of NTRK1; the resultant protein (RIP-TrkA) encoded by this fusion showed in cultured cells autophosphorylation of the fusion protein at critical tyrosine residues in cultured cells, implying its constitutive activation. The second gene fusion was characterised by a rearrangement between the CD74 and NTRK1 gene. In the same study, the authors also reported a TPM53–NTRK1 fusion (similar to that already described in CRC). A total of 3.3% patients in this study (3/91) harboured NTRK rearrangements potentially susceptible to TrkA inhibitors. In 2014, Stransky et al24 identified a novel TRIM24-NTRK2 gene fusion in lung adenocarcinoma through an unbiased computational pipeline designed for the identification of gene fusions in the data set from The Cancer Genome Atlas (TCGA).

Papillary thyroid carcinoma (PTC): A few years after the first published paper of a NTRK rearrangement in colorectal cancer, Bongarzone et al25 in 1989 described an oncogenic version of NTRK1 in PTC. Further works showed that oncogenic NTRK1 rearrangements in PTC are the consequence of the fusion of the TK domain of NTRK1 oncogene with 5-terminal sequences of at least three different genes. TRK-T1 and TRK-T2 are two different hybrid forms derived from chromosome inversion and different portions of TPR (Translocated Promoter Region) gene on chromosome 1q25 activate them. Another NTRK1 oncogene obtained by fusion with the TFG gene (TRK fused gene) on chromosome 3 is TRK–T3. All these oncogenic forms of NTRK1 gene encode cytoplasmic hybrid proteins that are constitutively phosphorylated at tyrosine residues.26 Somatic rearrangements of the NTRK1 gene in PTC usually do not exceed 12%, but range quite widely across different populations (from 15% to 50% in the Italian population27 to <10% in French,28 Japanese29 and Chinese30). A rare chromosomal rearrangement in sporadic thyroid cancer, but more frequent in radiation-related tumours, is ETV6-NTRK3, which results from an interchromosomal translocation t(12;15)(p13;q25) that juxtaposes exons 1–4 of ETV6 to exons 12–18 of NTRK3. The breakpoint differs from those reported in congenital fibrosarcomas and secretory breast cancers.31 Recently, via unbiased genomic approaches, novel gene rearrangements have been described in PTC, such as the PPL-NTRK1 (from periplakin, PPL)32 and the RBPMS-NTRK3 (from the RNA-binding protein with multiple splicing, RBPMS) gene fusion.24

Human secretory breast carcinoma: This is a rare but distinct subtype of infiltrating ductal carcinoma that was originally described in children and adolescents, but is now known to occur with equal incidence in adults. In 2002, Tognon et al33 reported the ETV6-NTRK3 gene fusion t(12;15)(p12;q26.1) as a pathognomonic genetic feature of this rare carcinoma.

Glioblastoma: Gene fusions occur in approximately 30–50% of patient with glioblastoma (GBM) samples and the Trk family could play a very important role. In 2013, Frattini et al34 analysed 185 GBM samples and found two in-frame fusions involving NTRK1 (BCAN-NTRK1 and NFASC-NTRK1). In the same year, Shah et al35 confirmed such genomic events as recurrent in a cohort of 24 GBM samples. In their work based on wide genomic screening of formalin-fixed paraffin-embedded (FFPE) specimens, Zheng et al32 reported one in-frame fusion involving exon 21 of ARHGEF2 (encoding Rho/Rac guanine nucleotide exchange factor 2) and exon 10 of NTRK1, and two in-frame fusions involving exon 5 of CHTOP (encoding the chromatin target of PRMT1) and exon 10 of NTRK1, of the 115 brain tumour FFPE analyses. In 2014, Wu et al36 applied a whole genome, whole exome and/or transcriptome sequencing to 127 samples of paediatric high-grade glioma (HGG), identifying recurrent fusions involving the neurotrophin receptor genes NTRK1, 2, or 3 in 40% of non-brainstem HGG.

Miscellaneous tumours: The recent efforts for identifying targetable genomic alterations through the use of next generation sequencing (NGS) are leading to the identification of novel and recurrent gene fusions in other various types of cancer. Ross et al37 reported a NGS screening of 28 FFPE samples of intrahepatic cholangiocarcinoma, in which a novel gene fusion RABGAP1L-NTRK1 was identified from a liver biopsy of a 62-year-old woman. Further, a recurrent gene fusion involving the ETV6 and the NTRK3 gene (ETV6-NTRK3) has been described in congenital fibrosarcoma.38

Trk inhibitors

The growing number of identified gene fusions involving the NTRK gene enhanced the interest of the scientific community in the development of drugs with inhibitory capacity for the TK domain of Trk. Bertrand et al39 published the high-resolution crystal structures of TrkA and TrkB in their apo-forms, as well as in complex with three nanomolar inhibitors. At least 40 residues of the kinase domain in its Asp-Phe-Gly (DFG)-in conformation, which potentially interact with a ligand in the ATP-binding site, are highly conserved between Trk proteins. Only 2 of the 40 residues are different between TrkA and TrkB, whereas the TrkB and TrkC ATP-binding sites are identical,39 suggesting an easier design of pan-, rather than selective, inhibitors of the three distinct isoforms of Trk receptors. Even though this may be an issue in developing selective Trk agonists for the treatment of neurological diseases,3 ,40 it may be indeed advantageous for cancer treatment where inhibition of all three Trk members’ gene fusions translates into wider antitumour activity.41

There is a limited number of reported Trk receptor tyrosine-kinase inhibitors in the literature and only a few of these have been tested in clinical trials (table 1).

Entrectinib, formerly RXDX-101 and NMS-E628, is an orally bioavailable inhibitor of the TK TrkA, TrkB and TrkC, as well as of C-ros oncogene 1 (ROS1) and anaplastic lymphoma kinase (ALK). Entrectinib can cross the blood–brain barrier, and could thus potentially be effective in the treatment of brain metastases and GBM by activating gene fusions of NTRK, ROS1 or ALK. Currently, two phase I trials are ongoing with this drug;42–44 the RP2D of this drug is known and initial reports of substantial antitumour activity in cancers harbouring NTRK1 gene fusions have been published.19 ,45 During screening for accrual in the European phase I trial of entrectinib (EudraCT Number: 2014-001326-15), the investigators identified a novel LMNA–NTRK1 gene fusion in a patient with metastatic CRC who was enrolled in the trial. The patient achieved a meaningful response after only 1 month of treatment, suggesting for the first time that the NTRK gene may play a role as actionable oncogenic driver in this tumour type as well.19 In February 2015, entrectinib was granted FDA Orphan Drug Designation for the treatment of TrkA, TrkB and TrkC positive non-small cell lung and colorectal tumours.41

LOXO-101 is a pan-Trk inhibitor with highly selective activity against the Trk kinase family. To date, in the phase I study, the maximum tolerated dose (MTD) has not yet been reached and pharmacokinetics show good systemic exposure of LOXO-101 after oral dosing, reaching approximately 98% inhibition of TrkA/B/C at peak concentrations with once daily dosing of 50 or 100 mg.46 The clinical activity of this inhibitor was recently reported47 in a case of a 41-year-old woman enrolled in the phase I trial who showed remarkable tumour response after treatment with LOXO-101.47 The report by Doebele et al47 also identified a LMNA–NTRK1 gene fusion, previously unreported in sarcoma.

Altiratinib (DCC-2701)48 and sitravatinib (MGCD516)49 are multikinase inhibitors with reported in vitro activity against TrkA and TrkB, and were both recently tested in phase I clinical trials. Other multikinase inhibitors with claimed anti-Trk activity include TSR-011, PLX7486, DS-6051b, F17752 and cabozantinib (XL184), all in development in phase I/II trials (table 2).

Acquired resistance to Trk inhibitors

As with other targeted therapies, the onset of acquired (secondary) resistance may limit the efficacy of Trk inhibitors. Given their recent introduction into the therapeutic armamentarium of cancer treatment, very limited data are available regarding mechanisms underlying resistance. Recently, a study by Russo et al unveiled gene alterations associated with entrectinib-acquired resistance. In particular, it has been reported that entrectinib induced a remarkable clinical response in mCRC carrying a LMNA-NTRK1 translocation, and tumour resistance developed after 4 months of response to discontinuous treatment (the cycle on which the patient was treated consisted of 4 days on, 3 days off, for 3 weeks, followed by a 1 week break in dosing), when the patient experienced progression of disease.19 To characterise the molecular basis of acquired resistance to entrectinib, circulating tumour DNA (ctDNA) was collected longitudinally (liquid biopsies) during treatment in this individual case, and a tumour biopsy was obtained before entrectinib treatment and transplanted into immunodeficient mice in order to obtain a patient-derived xenograft (PDX), which was challenged with the same entrectinib regimen until resistance.50 The study of ctDNA extracted from plasma samples collected before treatment initiation and at clinical relapse revealed two novel NTRK1 genetic alterations, p.G595R and p.G667C, in the kinase domain of the protein, which were not detected in the ctDNA pretreatment but emerged in the circulation as early as 4 weeks from beginning of treatment. This finding was strengthened by the observation of the emergence of one of the same molecular alterations, the NTRK1 p.G695R, in the engrafted tumour biopsy expanded to multiple cohorts of mice treated with dosage and schedule of entrectinib matching those of the patient. The establishment of independent preclinical models with acquired resistance to entrectinib finally reinforced the conclusion that p.G595R or p.G667C NTRK1 mutations are mechanisms of resistance to entrectinib. The authors also concluded that there is probably a dose-dependent effect affecting the emergence of each of the two mutations: NTRK1 p.G667C emerged with exposure of low concentration of the inhibitor (absent with higher dose), and it is weaker than p.G595R in conferring resistance.50 In the reported case,19 indeed, the Trk pan-inhibitor entrectinib was administered with an intermittent dosing regimen that may have promoted or anticipated the development of resistance. Nevertheless, it is still unknown whether continuous or intermittent dosing will affect the emergence and/or the type of acquired mutations.

Conclusions and future perspectives

Gene fusions have been recognised as drivers in tumours for more than three decades, providing useful insights into carcinogenic processes. New deep-sequencing technologies are now illuminating a landscape of previously undetected gene fusions in a multitude of cancers.51 In such an expanding scenario, NTRK gene fusions are emerging as novel targets across multiple tumour types, due to the growing availability of new drugs with anti-Trk activity.

Clearly, a major challenge in the development of these inhibitors is the low incidence in each single tumour histology that is counterbalanced by a wide distribution of Trk alterations across multiple histologies, thus increasing the number of patients who can be reached and treated with matched therapeutics in basket trials at the front edge of present clinical cancer research.20 ,52