The promise of TRK inhibitors in pediatric cancers with NTRK fusions

Introduction

Neurotrophic tyrosine receptor kinases (NTRK) fusions in pediatric cancers are uncommon overall, though highly enriched in specific rare cancers, such as infantile fibrosarcoma (IFS), cellular congenital mesoblastic nephroma (CMN), secretory breast carcinoma, and mammary analog secretory carcinoma of the salivary gland (MASC). The frequency of NTRK fusions in these diagnoses can surpass 90%, with most harboring an ETV6-NTRK3 fusion^1–4.

Approximately 85% of IFS tumors are driven by an ETV6-NTRK3 fusion, with reports of NTRK1 and other NTRK3 fusions in ETV6-NTRK3-negative cases^2,5–12. Similar to IFS, the majority of CMN cases are characterized by an ETV6-NTRK3 fusion, with rare cases involving other NTRK fusions^8,12–18. Both secretory breast carcinoma and salivary gland MASC are nearly universally defined by ETV6-NTRK3 fusions. A number of other pediatric cancers harbor NTRK fusions at a lower frequency, including subsets of glioma (up to 40% of non-brainstem high-grade gliomas and 4% of diffuse midline gliomas)¹⁹, papillary thyroid carcinoma (PTC) (26%)^20–23, and melanocytic tumors (21% of atypical Spitz subtype and 21% of spitzoid melanoma)²⁴, among others.

With the observation that NTRK fusions are enriched in select histologic subtypes, coupled with the success of first-generation tropomyosin receptor kinase (TRK) inhibitors (as further described below), researchers broadened their analysis to evaluate NTRK alterations in large cohorts of pediatric cancers in an effort to better define the spectrum of these aberrations. In 2018, Okamura, et al.²⁵ published a study of 9,966 adult and 3,501 pediatric patient samples, with data received from The Cancer Genome Atlas (TCGA) and the St. Jude PeCan database, respectively. Results showed a frequency of NTRK fusions of 0.31% in adult and 0.34% in pediatric cases. Within the pediatric population, NTRK fusions were seen in melanoma (11.1%), gliomas (5.3% of high-grade cases and 2.5% of low-grade cases), and hematologic malignancies (0.14%), with all 3 NTRK genes observed as fusion partners.

Similarly, Zhao, et al.²⁶ published a single institution study in 2021 that analyzed 1,327 tumors from 1,217 pediatric patients, and identified NTRK fusions in 2.22% of all samples. The frequency of cancers harboring NTRK fusions was highest in PTC (13%), followed by central nervous system (CNS) tumors (1.9%, gliomas and mixed neuronal glial tumors), non-CNS/non-PTC solid tumors (1.8%, including IFS and salivary gland MASC), and hematologic malignancies (0.4%). Assessment of NTRK subtypes showed a predominance of NTRK3 fusions (most commonly ETV6-NTRK3) across all histology subtypes, a proclivity of NTRK1 fusions in PTC, and a selectivity of NTRK2 fusions in CNS tumors. Of interest, 5 novel fusions were identified, 4 of which involved NTRK2 in CNS tumors and the remaining engaged NTRK1 in a PTC case. Each of these studies are limited to sequencing data from a single institution and thus potential selection or referral bias. Still, taken together, these recent findings suggest that NTRK fusions are more common in pediatric than adult cancers, that these aberrations are implicated in a broader range of tumor types than previously considered, and that a more extensive list of fusion partners is being observed, including novel fusions across all 3 NTRK genes.

Historic outcomes in histologic subtypes commonly harboring NTRK fusions are widely variable, though those with favorable outcomes often required invasive surgeries for a cure. While patients with IFS have excellent outcomes, as exhibited by a 5-year overall survival (OS) rate averaging 90%, this can come at the cost of significant morbidity due to amputations or disfigurement required for full surgical resection^26–27. Similarly, outcomes associated with CMN are excellent with an OS reaching 96% across a multi-study review, though nearly all require nephrectomy for cure^8,13,29. Both secretory breast carcinoma and salivary gland MASC have a good prognosis when disease is localized, but require surgical intervention and sometimes radiation^4,30–35. Treatment approaches and outcomes across gliomas, PTC and melanocytic tumors fall along a broad range, though high-grade CNS tumors portend a particularly poor prognosis^19,36–37.

In parallel to research focused on understanding the landscape of NTRK fusions in pediatric cancers, a number of early phase clinical trials were conducted using potent TRK inhibitors in a targeted fashion. Results from these trials showed excellent outcomes in patients with NTRK fusions, and case scenarios within these studies have suggested reduced morbidity. Thus, although NTRK fusions are still considered rare events, the benefit in using TRK inhibitors in this selective population was considered substantial. This review highlights the success of these recent clinical trials, and describes ongoing and promising research in this area. We will focus on type I first- and second-generation TRK inhibitors with specific activity; other multi-kinase inhibitors exist that have varying degrees of activity against NTRK fusions, but have not been clinically developed for this indication and will not be discussed.

NTRK as an oncogene and therapeutic target

A detailed review of NTRK fusion pathology, molecular characteristics, and mutational profile is provided by Hsiao, et al., Zhong, et al., and Morrissette, et al., respectively in this series. A brief overview will be provided here for the purpose of better understanding TRK inhibitor mechanisms of action and acquired resistance.

The biologic role of NTRK genes (NTRK1, NTRK2, and NTRK3) and the resultant protein product (TRKA, TRKB, and TRKC, respectively) is in nervous system development and function via signal transduction of neurotrophins. The TRK proteins are transmembrane kinases with an extracellular ligand-binding domain, a single transmembrane domain, and an intracellular catalytic domain. The solvent front is a region in the catalytic domain where kinase inhibitors typically make contact³⁸. Within the catalytic domain, an activation loop labeled with a conserved aspartic acid-phenylalanine-glycine, or Asp-Phe-Gly (DFG), motif is positioned to regulate enzyme activity - behind a gatekeeper phenylalanine in the active site (thereby preventing adenosine triphosphate, or ATP, binding) when inactive, and rotated out of the active site in response to ligand binding in the extracellular domain (thereby exposing the ATP binding site)³⁹. Following ligand binding, the intracellular domain undergoes consecutive phosphorylation using phosphates derived from ATP. Downstream of this signal transduction is activation of several pathways known to play an oncogenic role, including MAPK/ERK, PI3K/AKT and PLCγ/PKC^40–41. Various aberrations deriving oncogenic potential can occur in NTRK genes, the majority of which are fusions. Here, the 3′-region of the NTRK gene is fused with a 5′ region of a fusion partner, resulting in a novel protein product that is constitutively active⁴².

TRK inhibitors are categorized into 4 subtypes according to their binding location and subsequent interactions with TRK^43–44. Type I inhibitors bind the target kinase in an active confirmation (DFG-in) to exert competitive inhibition at the ATP-binding site, and are the main subtype being investigated clinically. Type II inhibitors bind the target kinase in an inactive confirmation (DFG-out) and can be more selective than Type I inhibitors. Type III inhibitors bind to the kinase domain, but outside of the ATP-binding site, leading to potential selectivity at the level of TRK isoforms. Finally, type IV inhibitors bind outside of the kinase domain and function to interrupt ligand-receptor or protein-protein interactions⁴⁵. As type I inhibitors are, to our knowledge, the only drugs to reach pediatric clinical trials for the treatment of NTRK fusion cancers, we will focus on these agents.

Type I TRK inhibitors can be further divided into generations. First-generation TRK inhibitors nearly universally elicit substantial efficacy initially. However, the clinical benefit is often limited due to acquisition of kinase domain mutations that derive steric interference with the inhibitor. To address this, second-generation TRK inhibitors were designed to limit the compound surface area in the active site⁴⁵. Both first- and second-generation TRK inhibitors will be further discussed below in the context of clinical trials broadly and the pediatric experience more specifically.

First-generation TRK inhibitors

Interest in TRK inhibitors emerged from preclinical studies where larotrectinib and entrectinib showed potent inhibition against TRKA, TRKB, and TRKC, with IC₅₀ values of 5–11 nM^46–47 and 1–5 nM⁴⁸, respectively. While larotrectinib exhibits selective inhibition of TRK^46,49, entrectinib serves as a multi-kinase inhibitor with additional activity against ROS1 and ALK^50–51.

Early phase clinical trials of both agents in pediatric and combined pediatric-adult trials showed promising results in tumors with NTRK fusions in a histology-agnostic manner. Importantly, the quick and durable responses were seen in many patients with locally advanced or metastatic disease, and for whom previous lines of therapy had failed. Additionally, activity was seen in patients with primary CNS tumors and CNS metastases, providing benefit in yet another challenging clinical space.

With these favorable results, the US Food and Drug Administration (FDA) granted breakthrough designation for larotrectinib in 2016 and entrectinib in 2017 for the treatment of NTRK fusion-positive tumors, and accelerated approval for the same indication for larotrectinib (any age) in 2018 and entrectinib (age ≥12 years) in 2019 (Figure 1). The clinical trials leading to this designation are further discussed below.

Figure 1.

Timeline of early phase clinical trials, FDA breakthrough designation and accelerated approval for larotrectinib and entrectinib.

Larotrectinib

The first clinical trial testing larotrectinib in a pediatric patient population was the SCOUT study: A Study to Test the Safety and Efficacy of the Drug Larotrectinib for the Treatment of Tumors With NTRK-fusion in Children [NCT02637687]. The phase I component is complete and results were published by Laetsch and DuBois, et al.³ in 2018 (Table 1), while the phase II portion is ongoing at the time of this publication (Table 2).

Table 1.

Pediatric-specific results from key clinical trials of first-generation TRK inhibitors.

Trial and results source	Number, age of patients	Histologic diagnoses	Molecular changes	Treatment-related AE*	Efficacy results (ORR)
SCOUT, phase I results (larotrectinib) Laetsch and DuBois, et al.3	24 (22 evaluated for response) 4 months to 18 years	IFS (n=8), other STS (n=7), PTC (n=2), DIPG (n=2), other (n=5)	NTRK fusion-positive: 17  NTRK1 x9  NTRK2 x1  NTRK3 x7 NTRK fusion-negative: 7	All dosing levels: 17% At RP2D: 22% Grade 4 or 5: 0%	Fusion-positive: 93% Fusion-negative: 0%
Pooled analysis from SCOUT and NAVIGATE, interim results (larotrectinib) Hong, et al.53	52 (51 evaluable for blinded independent response assessment) 1 month to 18 years (24 patients <1 year)	IFS (n=28), other STS (n=19), PTC (n=2), CMN (n=1), melanoma (n=1)	NTRK fusion subtypes not separated from adult data	Grade 3 or 4: 17%	Pediatric cohort: 92%
Pooled analysis from SCOUT and NAVIGATE, expanded dataset (larotrectinib) van Tilburg, et al.54	78 (investigator assessment only) 1 month to 18 years	IFS (n=44), other STS (n=29), PTC (n=2), CMN (n=2), melanoma (n=1)	NTRK fusion-positive: 78  NTRK1 x31  NTRK2 x3  NTRK3 x44	Grade 3 or 4: 22% Serious AE: 5%	All patients: 88% IFS: 100%
STARTRK-NG, interim results (entrectinib) Robinson, et al.58	39 (all evaluated for response) 5 months to 20 years	NBL (n=15), primary CNS tumor (n=14), non-NBL solid tumor (n=10)	Fusion-positive: 22  NTRK1 x3  NTRK2 x4  NTRK3 x7  ROS1 x5  ALK x3 ALK mutation: 1 No aberration: 16	DLTs from phase I included elevated creatinine, dysgeusia, fatigue and pulmonary edema	Fusion-positive: 77% Fusion-negative: 6%

AE = adverse event, ORR = objective response rate, IFS = infantile fibrosarcoma, STS = soft tissue sarcoma, PTC = papillary thyroid carcinoma, DIPG = diffuse intrinsic pontine glioma, CMN = congenital mesoblastic nephroma, CNS = central nervous system, NBL = neuroblastoma, RP2D = recommended phase II dose, DLT = dose limiting toxicity.

^*grade >3 or higher

Table 2.

Ongoing pediatric-specific clinical trials of first-generation TRK inhibitors.

Trial and phase	Age of patients	Histologic diagnoses	Molecular changes	Primary outcome	Study status
SCOUT, phase II (larotrectinib) [NCT02637687]	Up to 21 years	IFS (cohort 1), advanced Non-CNS solid tumor, r/r (cohort 2) Primary CNS tumor, r/r (cohort 3)	NTRK fusion ETV6 rearrangement (for IFS, CMN, and secretory breast carcinoma)	ORR based on RECIST v1.1, RANO, or INRC	Recruiting (as of 11/23/21)
NCI-COG Pediatric MATCH, phase II (larotrectinib) [NCT03213704]	1 to 21 years	r/r non-CNS solid tumor, primary CNS tumor, LCH, NHL (single cohort)	NTRK fusion	ORR based on RECIST v1.1 and RANO	Recruiting (as of 11/24/21)
Phase II, upfront/neoadjuvant (larotrectinib) [NCT03834961]	Up to 30 years	IFS (cohort A), untreated Non-CNS solid tumor (excluding IFS) and primary CNS tumor (excluding HGG), untreated (cohort B) Acute leukemia, r/r (cohort C)	NTRK fusion ETV6 rearrangement (for IFS and CMN)	ORR for IFS prior to local control, based on RECSIT v1.1 Outcomes for non-IFS histologies are part of secondary outcomes	Recruiting (as of 10/19/21)
STARTRK-NG, phase I/II (entrectinib) [NCT02650401]	Up to 18 years	Non-CNS solid tumor and primary CNS tumor (phases I and II)	NTRK, ROS1, and ALK fusions and other molecular aberrations (phase I) NTRK1 and ROS1 fusions (phase II)	MTD/RP2D (phase I) ORR based on RECIST v1.1 or RANO (phase II)	Phase I closed for enrollment Phase II recruiting (as of 11/11/21)

IFS = infantile fibrosarcoma, CMN = congenital mesoblastic nephroma, CNS = central nervous system, r/r = relapsed/refractory, ORR = objective response rate, RECIST v1.1 = Response Evaluation Criteria in Solid Tumours (RECIST) 1.1, RANO = Response Assessment in Neuro Oncology, INRC = International Neuroblastoma Response Criteria, NCI = National Cancer Institute, COG = Children’s Oncology Group, LCH = Langerhans Cell Histiocytosis, NHL = non-Hodgkin lymphoma, HGG = high grade glioma, MTD = maximum tolerated dose, RP2D = recommended phase II dose.

The phase I portion of the SCOUT study was open to patients with relapsed or refractory solid or primary CNS tumors, regardless of underlying genomic aberrations, for the dose escalation cohort. This was followed by a dose expansion cohort that narrowed enrollment to those with a documented NTRK fusion (including NTRK fusions implied by documentation of an ETV6 rearrangement in the case of IFS, CMN or secretory breast carcinoma). A total of 24 patients ages 4 months to 18 years enrolled, including 17 patients with NTRK fusions and 7 without. Histologic subtypes of those with NTRK fusions included IFS (n=8), other soft tissue sarcomas (n=7), and PTC (n=2). Of these patients, 11 had locally advanced disease and all but 2 had received prior therapy for their IFS. NTRK fusions involved NTRK1 (n=9), NTRK2 (n=1), and NTRK3 (n=7). The primary endpoint for the dose escalation portion was safety of larotrectinib, including dose-limiting toxicity (DLT), of which only 1 occurred (grade 3 transaminitis). The side effect profile for this study was overall favorable, with primarily grade 1 and 2 treatment-related adverse effects (TRAE), most frequently mild transaminitis, hematologic abnormalities and nausea/vomiting. Within the study population, 4 (17%) experienced a grade 3 TRAE, and no grade 4 or 5 events were reported. Neurologic toxicities were rare, with only grade 1 fatigue in 3 (13%) subjects. A maximum tolerated dose (MTD) was not reached, thus the highest dosing level (100 mg/m² twice daily) was used as the recommended phase II dose (RP2D).

As one of the secondary endpoints of the SCOUT study, antitumor activity was measured in terms of objective response rate (ORR). This early efficacy analysis showed a stark difference in response between those with NTRK fusions and those without. Specifically, the ORR across the 15 patients with evaluable NTRK fusion-positive tumors was 93% (4 with complete response, CR; 10 with partial response, PR; 1 with progressive disease, PD) vs 0% in the 7 patients with NTRK fusion-negative tumors (all PD). Interestingly, at this data cut-off, the single patient with on-therapy PD in the NTRK fusion-positive group had an initial partial response to larotrectinib with >90% tumor regression, then subsequently progressed after 8 months. Further genomic analysis on this patient’s tumor revealed a new solvent front resistance mutation (NTRK3 G623R). Of those patients who achieved a PR, 4 were then able to undergo full surgical resection with likely reduced morbidity.

To further explore the benefits of larotrectinib in the neoadjuvant setting, DuBois, et al.⁵² published a post hoc analysis of the phase I larotrectinib study in 2018, with a detailed analysis of a subset of patients who underwent surgical resection while on study. Included were 5 patients with locally advanced NTRK fusion-positive sarcomas (3 with IFS and 2 with other soft tissue sarcomas), most of whom had progressed through prior lines of cytotoxic chemotherapy (n=4). All 5 patients achieved a PR while on larotrectinib, and then proceeded with surgical resection. Histologic evaluation of these tumors included complete (n=2) or near-complete (n=1) pathologic response with negative surgical margins, and viable tumor (n=2) with positive surgical margins. These latter 2 patients remained on larotrectinib post-operatively. Importantly, no surgical complications were reported as a result of prior larotrectinib therapy.

The notable efficacy of larotrectinib in the treatment of pediatric and adult patients with NTRK fusion-positive tumors became quickly apparent as several early phase clinical trials opened. With this, two pooled analyses were conducted and included patients from three clinical trials testing larotrectinib in patients with NTRK fusion-positive tumors: a phase I study in adults [NCT02122913]; the aforementioned SCOUT phase I/II study in children [NCT02637687]; and a phase II study of adolescents and adults, known as NAVIGATE [NCT02576431]. The first combined study, published by Drilon and Laetsch, et al.⁴⁷ in 2018, was performed to better define efficacy and to support rapid approval of larotrectinib. Data from 55 patients, ages 4 months to 76 years, were included for analysis. This patient group represented 17 tumor types with a predominance of NTRK1 and NTRK3 fusions and 14 unique partners. The primary endpoint of ORR was 75–80% (variation due to independent vs investigator review; 7 with CR, 34 with PR, 7 with SD, 5 with PD, and 2 unevaluable, assigned by central assessment). Secondary endpoints included duration of response (DOR) and progression-free survival (PFS), which was not reached at the time of publication. However, at 1-year, 71% of patients with CR/PR had ongoing response and 55% of all patients were progression-free. In total, 6 patients lacked a primary response to larotrectinib (i.e. PD as best response). Evaluation of 4 of these patients showed 1 with an acquired solvent front mutation in the setting of prior TRK inhibitor use and 3 with negative TRK immunohistochemistry (IHC) testing on the tumor (indicating either the absence of an NTRK fusion due to false positive initial testing or a lack of protein expression despite the molecular finding). The data reported in this study supported FDA approval of larotrectinib for treatment of NTRK fusion-positive tumors in children and adults, including use in the neoadjuvant setting, regardless of tumor type.

A few years later, in a follow-up pooled analysis, Hong, et al.⁵³ published results from expanded cohorts of the 3 original trials, including the initial 55 patients and an additional 104 subjects (Table 1). Ages ranged from 1 month to 84 years, with 33% under the age of 18. A similar proportion of NTRK1 and NTRK3 fusions were represented, and 29 unique partners identified. Here too, the primary endpoint of ORR was 79% (16% CR, 63% PR), consistent with the original analysis. Interestingly, when viewing ORR across age ranges, pediatric patients fared better than adults, with an ORR of 92% vs 73%, respectively. These results and the realization that pediatric patients exhibited higher response rates than adult subjects were shown again with larger series of patients and more mature data. With these developed data sets, van Tilburg, et al.⁵⁴ reported an ORR of 88% across a cohort of 78 children with non-CNS NTRK fusion-positive tumors in 2021 (Table 1) and Drilon, et al.⁵⁵ showed an ORR of 71% across a cohort of 116 adults with tumors harboring NTRK fusions in 2020. Whether this difference reflects age-related or histology-specific differences is unclear. Regardless, these large scale analyses serve to confirm efficacy of larotrectinib in a genomic-focused, histology- and age-agnostic setting, with additional support for safety.

In a more focused assessment of larotrectinib, Perreault, et al.⁵⁶ presented results from a pooled analysis of larotrectinib use in adults and children with NTRK fusion-positive primary CNS tumors. This work included 9 patients from the NAVIGATE trial and 24 from the SCOUT trial, with 79% from the pediatric age group. Of importance in the treatment of primary CNS tumors and CNS metastases, larotrectinib was detectible in the cerebrospinal fluid (CSF) of 2 patients who underwent sampling in the SCOUT trial. The majority of subjects in this combined study had high-grade gliomas (58%) and were previously treated with systemic therapy (82%). The primary endpoint of ORR was 30%, though 73% achieved disease control surpassing 24 weeks (defined as ORR and SD combined).

Currently, the most mature data on larotrectinib use in the pediatric population was presented by van Tilburg, et al⁵⁴ in 2021 and included data on 78 pediatric patients with non-CNS NTRK fusion-positive tumors (75 from the SCOUT trial and 3 from the NAVIGATE trial). Importantly, this data set was advanced enough to capture survival outcomes at 3 years. The primary objective of ORR was consistent with earlier results of larotrectinib at 88% overall (evaluable patients, n=76) and 100% within a subset of patients with IFS. Secondary objectives included DOR, PFS and OS at 3-years, which were reported as 63%, 60%, and 97%, respectively. These results demonstrated that larotrectinib can achieve a rapid and durable response when used in a population selected for its primary target.

At the time of this publication, 3 pediatric early phase clinical trials with larotrectinib are ongoing, including the phase II portion of the SCOUT trial discussed above (Table 2). The two other trials are open through the Children’s Oncology Group (COG), including: a phase II trial of larotrectinib in relapsed or refractory solid tumors harboring a NTRK fusion, as part of the larger National Cancer Institute (NCI) Pediatric Match trial [NCT03213704]; and a phase II trial of larotrectinib in previously untreated solid tumors harboring a NTRK fusion or relapsed acute leukemia [NCT03834961]. The latter study will provide valuable data on larotrectinib in the upfront and/or neoadjuvant setting in a prospective fashion.

Entrectinib

The first clinical trial evaluating entrectinib in the pediatric population is ongoing, but interim analyses have repeatedly shown encouraging results and have contributed to drug approval. Similar to the early larotrectinib studies, the STARTRK-NG trial, A Study Of Entrectinib in Children and Adolescents With Locally Advanced Or Metastatic Solid Or Primary CNS Tumors and/or Who Have No Satisfactory Treatment Options [NCT02650401], includes a dose escalation cohort that is open to ‘all comers’ (patients with relapsed or refractory solid or primary CNS tumors, regardless of underlying genomic alterations), followed by a dose expansion cohort that narrows the enrollment to target disease and genomic alterations. Specifically, the expansion cohort was opened to patients with solid and primary CNS tumors with NTRK, ROS1, or ALK fusions, and neuroblastoma (regardless of molecular changes) (Table 2).

In 2019, Robinson, et al.⁵⁷ published an abstract with interim results based on the first 29 patients enrolled, with ages ranging from 5 months to 20 years. The primary endpoints included MTD/RP2D and ORR, the former of which was determined to be 500 mg/m² daily. Of note, all responses were seen at doses >400 mg/m² daily. DLTs included elevated creatinine, dysgeusia, fatigue and pulmonary edema. Of the 28 evaluable patients, 12 had objective responses (2 with CR, 9 with PR). Similar to the early larotrectinib studies, the presence of a targeted fusion was a dichotomizing feature. The ORR in the fusion-positive group (n=11) was 100% (5 with high-grade CNS tumors, 6 with extracranial solid tumors) and only one of the fusion-negative patients responded. This fusion-negative patient’s tumor had an ALK mutation (F1174L) and achieved a CR. Interestingly, the ALK F1174L mutation is known to be crizotinib-resistant and can arise de novo.

In a subsequent abstract in 2020, Robinson, et al.⁵⁸ reported on a second interim analysis, which included 39 patients from the same study, dosed on the same schedule, and falling within the same age range (Table 1). The results again demonstrated effectiveness of entrectinib in target fusion-positive cases, with an ORR in the fusion-positive group of 77% (7 with CR, 10 with PR, 3 with SD, 2 with PD) vs 6% in the fusion-negative group (1 with CR, 3 with SD, 10 with PD, 3 unevaluable). Common side effects included weight gain, elevated creatinine, cytopenias, transaminitis, and bone fractures, of which more than half were determined to be treatment related. Importantly, this series included 14 children with high-grade fusion-positive CNS tumors (11 with NTRK fusions), with an ORR of 64% (4 with CR, 5 with PR, 3 with SD, 2 with PD), thus highlighting the potential of TRK fusion inhibition in this challenging tumor type.

Acquired resistance to first-generation TRK inhibitors

Amid success of first-generation TRK inhibitors in NTRK fusion-positive tumors, resistance patterns emerged in a manner that paralleled ALK and ROS1 fusion-positive cancers in the setting of ALK or ROS1-directed TKI^59–61. After studying sequential cases where patients developed PD after initial response to first-generation TRK inhibitors (as further described below), 3 types of ‘on target’ resistance mechanisms were described. These acquired resistance models resulted from amino acid substitutions in 3 regions of the TRK kinase domain - the solvent front, the gatekeeper residue, and the xDFG motif of the activation loop. These molecular changes created steric interference, or altered the ATP-binding affinity or kinase domain conformation^62–63 (Figure 2). To date, resistance mutations have been mostly documented in NTRK1 and NTRK3, which is thought to be due to the low frequency of NTRK2-based fusions and not necessarily an inherent difference in risk of developing resistance^47,64.

Figure 2.

Development of acquired resistance to first-generation type I TRK inhibitors and use of 2^nd-generation type I TRK inhibitors to overcome resistance.

The first case of acquired resistance was published by Russo, et al.⁶² in 2016, where an adult patient with NTRK1 fusion-positive colorectal cancer progressed after 4 months of response to entrectinib. Sequential circulating tumor DNA (ctDNA) specimens revealed two new mutations in the kinase domain around the time of disease progression (NTRK1 G595R in the solvent front and NTRK1 G667C in the xDFG motif). This was followed shortly by a case report from Drilon, et al.¹ where an adult patient with MASC progressed similarly after a period of response to entrectinib. Tumor sequencing in this patient showed a new solvent front mutation (NTRK3 G623R, a paralogue to NTRK1 G595R).

Similar mechanisms of resistance were seen with larotrectinib, suggesting that the risk lies at the level of the drug class. In their 2018 publication, Drilon and Laetsch, et al.⁴⁷ described resistance patterns seen in 10 pediatric and adult patients in the pooled analysis, all of whom had an objective response or SD for 6 months prior to developing PD. Sequencing was performed on either tumor specimen or plasma (ctDNA) for 9 of these patients and revealed all three types of mutations across the cohort (7 solvent front, 2 gatekeeper, and 2 xDFG). Interestingly, 3 patients developed more than 1 resistance mutation, as also seen in the 2016 entrectinib-related case report.

‘Off target’ mechanisms have also been described, with a common theme of alternate mechanisms of pathway activation, most commonly affecting the MAPK pathway. In 2019, Cocco and Schram, et al.⁶⁵ published a series of adult patients who underwent extensive genomic sequencing after a work-up for ‘on target’ mutations returned negative. Results showed a number of inciting genomic aberrations serving as ‘bypass’ mechanisms, such as a BRAF V600E mutation, MET amplification, and KRAS, MAP2K1 and ERBB2 mutations. Included in this analysis was the patient noted above with colorectal cancer and acquired resistance mutations to entrectinib, who then later progressed on a second-generation TRK inhibitor, selitrectinib, despite initial response. Further investigation revealed several KRAS mutations, which were predicted to be MAPK pathway activating.

With these differing mechanisms of resistance, proposed treatment strategies are tailored accordingly. In the case of an ‘on target’ mutation, one might consider treatment with an alternate TRK inhibitor that provides activity against the ‘on target’ resistance mutation, such as transitioning from a first- to second-generation inhibitor in a sequential fashion. While not formally studied in a clinical trial, there are case reports of adult patients responding to type II multi-kinase inhibitors, such as cabozantinib⁶⁶, following detection of acquired resistance to a first-generation type I inhibitor, suggesting a possible strategy for overcoming resistance that may merit future study. For the ‘off target’ mutations, experience has shown that a combinatorial approach, such as a TRK inhibitor paired with a MET inhibitor, may allow for disease control yet again. Central to this concept is the role of repeat molecular profiling when acquired resistance is suspected.

Pediatric experience

Fewer cases of acquired resistance to first-generation TRK inhibitors have been reported in children than in adults. It is unclear if this represents a reduced proclivity to development, a longer time to acquisition, or simply fewer pediatric patients being treated with TRK inhibitors at large. One of the few reported cases is of a 2 year old female with recurrent IFS of the right neck and skull base, which harbored an ETV6-NTRK3 fusion. Prior to starting larotrectinib, the patient progressed on multiple lines of therapy and after several surgical resections. Once on larotrectinib, tumor regression of more than 90% was realized. Unfortunately, PD was detected 8 months later, and a new NTRK3 G623R mutation was identified on tumor biopsy by whole exome sequencing (WES).⁶⁷

Second-generation TRK inhibitors

Second-generation TRK inhibitors were designed as fused macrocyclics to address ‘on target’ resistance. Their small size and low molecular weight allow navigation around steric impedance in the kinase domain and the ability, once again, to reach the ATP-binding pocket. These agents generally exhibit more potent inhibition than their first-generation counterparts, and this activity is replicated across all 3 types of acquired resistance^63,68–69. For example, in preclinical studies, selitrectinib exhibited potent inhibition against NTRK1 G595R, NTRK1 G667C, NTRK3 G623R, and NTRK3 G696A (IC₅₀ of 2, 9.8, 2.5 and 2.3 nM, respectively), and repotrectinib against NTRK1 G595R, NTRK2 G639R, NTRK3 G623E, and NTRK3 G623R (IC₅₀ of 0.4, 0.6, 1.4, and 0.2 nM, respectively)³⁹ (Figure 2).

Selitrectinib

Selitrectinib is currently under evaluation through A Phase I/II Study of the TRK Inhibitor Selitrectinib in Adult and Pediatric Subjects With Previously Treated NTRK Fusion Cancers [NCT03215511] (Table 3), which is open to patients as young as 1 month of age, and through an expanded access/compassionate use protocol [NCT03206931].

Table 3.

Ongoing pediatric-specific clinical trials of second-generation TRK inhibitors.

Trial and phase	Age of patients	Histologic diagnoses	Molecular changes	Primary outcome	Study status
Phase I/II, adult and pediatric (selitrectinib) [NCT03215511]	Separate cohort for <12 years and ≥12 years for both phase I and II	Non-CNS solid tumor and primary CNS tumor, r/r including prior TRKi	NTRK fusion ETV6 rearrangement (for IFS and CMN)	MTD/RP2D (phase I) ORR based on RECIST v1.1 and RANO (phase II)	Active, but not recruiting (as of 11/11/21)
TRIDENT-1, phase I/II, adult and pediatric (repotrectinib) [NCT03093116]	12 years and older (phase II; adults only in phase I)	Non-CNS solid tumor and primary CNS tumor, no prior TRKi (cohort 5, phase II), +TRKi (cohort 6, phase II) Cohorts 1 –4 are for NSCLC and assumed not pediatric-focused	NTRK and ROS1 fusions (phase II)	ORR based on RECIST v1.1 (phase II)	Recruiting (as of 11/17/21)
CARE, phase I/II, pediatric and young adult (repotrectinib) [NCT04094610]	Up to 12 years (phase I), up to 25 years (phase II)	Non-CNS solid tumor, primary CNS tumor, and ALCL, r/r (phase I) Non-CNS solid tumor and primary CNS tumor no prior TRKi (cohort 1, phase II), +TRKi (cohort 2, phase II) ALCL, other (cohort 3, phase II)	NTRK fusion (cohorts 1 and 2, phase II) NTRK, ROS1, and ALK fusions and other molecular aberrations (phase I and cohort 3, phase II)	MTD/RP2D (phase I) ORR based on RECIST v1.1 or RANO (phase II)	Recruiting (as of 11/4/21)

CNS = central nervous system, r/r = relapsed/refractory, TRKi = TRK inhibitor, IFS = infantile fibrosarcoma, CMN = congenital mesoblastic nephroma, MTD = maximum tolerated dose, RP2D = recommended phase II dose, ORR = objective response rate, RECIST v1.1 = Response Evaluation Criteria in Solid Tumours (RECIST) 1.1, NSCLC = non-small cell lung cancer, ALCL = anaplastic large cell lymphoma.

In 2017, Drilon, et al.⁶³ published a pilot/first-in-human series of selitrectinib, which included 2 patients - the adult with progressive colorectal cancer and the 2 year old with recurrent IFS, both discussed above. The investigation included an intra-patient PK-guided dose escalation, which both patients tolerated well, allowing for goal target coverage. The adult responded quickly to selitrectinib, with a durability extending past the point of publication. The child had an initial PR to selitrectinib, but unfortunately progressed 3 months later and passed away.

In 2019, Hyman, et al.⁷⁰ presented an abstract with pooled data from the phase I trial (n=20) and the expanded access protocol (n=11), including adult (n=24) and children (n=7), all of whom had NTRK fusion-positive tumors and at least one prior exposure to a TRK inhibitor. The median duration on a prior TRK inhibitor was 9.5 months, during which the majority developed an ‘on target’ mutation (n=20). Of the remaining patients, only 3 were found to have an ‘off target’ mutation (all others were unknown or inconclusive). ORR for those with an ‘on target’ mutation was 45% (9 with CR/PR, 6 with SD, 3 with PD, 2 unevaluable), while those with bypass mechanisms did not respond (2 with PD, 1 unevaluable). Selitrectinib was found to be tolerable overall, with DLTs seen only in adults, and included ataxia, dizziness, and vomiting.

Repotrectinib

Repotrectinib is currently being studied in the pediatric population through 2 early phase trials: the TRIDENT-1 study, A Study of Repotrectinib in Patients With Advanced Solid Tumors Harboring ALK, ROS1, or NTRK1–3 Rearrangements [NCT03093116], and A Study of Repotrectinib in Pediatric and Young Adult Subjects Harboring ALK, ROS1, OR NTRK1–3 Alterations [NCT04094610] (Table 3). These trials differ from those evaluating selitrectinib in a few key ways: first, the target fusion list includes ALK and ROS1, in addition to NTRK, in a similar way to entrectinib; second, the phase II portions test repotrectinib in both TRK inhibitor-naïve and TRK inhibitor-refractory patients through separate expansion cohorts; and third, the latter study allows enrollment for ALK, ROS1, and NTRK alterations outside of fusions (i.e. mutations, amplifications) in a separate cohort.

Interim analyses of the TRIDENT-1 study has shown the most data and promise in patients with ROS1 fusion-positive non-small-cell lung cancer (NSCLC), though results on NTRK fusion-positive solid tumors are still encouraging. In 2021, Cho, et al.⁶⁰ published results from the phase II portion of the study, which included 39 adolescent and adult patients, 6 of whom were categorized as ‘NTRK fusion-positive, TRK inhibitor-refractory solid tumors.’ The primary endpoint of ORR was 50%, vs 40% and 67% in ROS1 fusion-positive, TKI inhibitor-refractory NSCLC with 2 and 1 prior TKI, respectively, and vs 86% in ROS1 fusion-positive, TKI inhibitor-naïve NSCLC. Overall, repotrectinib was considered to be well tolerated, with 90% of patients in the phase II assessment tolerating planned dose escalation from 160 mg once daily to twice daily. Similarly for selitrectinib, dizziness was a common adverse event seen with repotrectinib, potentially reflecting a class effect.

Acquired resistance to second-generation TKI inhibitors

The mechanisms of acquired resistance mutations to second-generation TKI inhibitors, such as selitrectinib and repotrectinib, are less well understood. However, recent case reports have highlighted patients who progressed on multiple lines of TRK inhibitors, with compelling evidence for development under selective pressure. For example, Oliver, et al.⁶⁷ reported on a 10 month old female with progressive IFS of the anterior mediastinum, shortly after her passing. Her course prior to initiation of TRK inhibitors included traditional cytotoxic chemotherapy and a surgical resection. Once started on larotrectinib, she had a temporary response, followed by progression in the setting of a new solvent front mutation (NTRK3 G623R). She then began treatment with selitrectinib, and again had an initial response for a few months, followed by disease progression. Sequencing studies at that time showed a new gatekeeper mutation (NTRK3 F617L) within the same clone. From there, she underwent repeat surgical resection followed by continued selitrectinib in combination with conventional chemotherapy. Unfortunately, she passed away 2 months later.

Conclusion

Results from early phase clinical trials of first-generation TRK inhibitors (larotrectinib, entrectinib) showed remarkable potential in adults and children with tumors harboring varying NTRK fusions. The brisk and durable responses observed were quite specific in that patients with fusion-positive tumors had an ORR reaching 100% in some series, while those with fusion-negative tumors rarely responded. With these promising results, the FDA granted expedited approval for larotrectinib and entrectinib in 2018 and 2019, respectively, for the treatment of NTRK fusion-positive tumors, irrespective of histology.

Certain rare subsets of pediatric cancers are enriched with NTRK fusions, including IFS, cellular CMN, secretory breast carcinoma, and salivary gland MASC, with frequencies surpassing 90%. With the knowledge of a targeted, effective, and tolerable therapy, efforts have been made to further characterize the landscape of NTRK fusions in pediatric cancers. Two recent molecular studies of large, unselected pediatric cancer cohorts estimated this frequency at 0.34% and 2.22%, a discrepancy likely explained by differences in the prevalence of NTRK fusion-enriched subtypes in the studied cohorts. These studies also demonstrated that the breadth of tumor types harboring NTRK fusions as well as the number of unique fusion partners is more expansive than previously considered.

The clinical benefit of treatment with TRK inhibitors has also been broad. For example, outcomes in patients with IFS in the pre-TRK inhibitor era were favorable, with an OS averaging 90%. However, because of the locally aggressive nature of these tumors, morbid surgical procedures were often required to obtain a cure. With the advent of TRK inhibitors and the favorable responses seen, the need for such debilitating approaches can be reduced or even eliminated in some cases. In other tumor subtypes, such as high-grade gliomas, treatment options are limited and outcomes remain persistently poor. The response rates seen with TRK inhibitors in this patient population gives hope that targeted and effective therapy will bring positive steps towards improving outcomes and realizing a cure.

Amidst this success, acquired resistance to first-generation TRK inhibitors have developed in a pattern that mirrors ALK and ROS1 fusion-positive tumors after therapy with targeted inhibitors. Two types of resistance mechanisms have been identified: ‘on target’ mutations that occur in the TRK kinase domain, and ‘off target’ mutations that exert their effect through alternate pathways. This has led to the emergence of a treatment paradigm that is adaptive and driven by molecular changes, such that the former mechanism can be theoretically salvaged with a next-generation TRK inhibitor, while the later benefits from a combinatorial approach with addition of a targeted agent that covers the ‘bypass’ mutation.

Given the rapidity with which first-generation TRK inhibitors received FDA approval, ongoing and upcoming studies will continue to provide data on critical questions that remain. While the efficacy and clinical benefit of NTRK inhibitors is clearly evident, there are several components of management that have yet to be fully elucidated. For example, the appropriate length of therapy (overall and in the context of expected surgical resection) is unclear, as well as if serial ctDNA studies can aid in this assessment. The optimal use of NTRK inhibition in patients with primary CNS tumors, especially high-grade tumors for whom prognosis is inherently poor, can only be conjectured at this point. Studies evaluating whether and how to best use these agents in combination with chemotherapy and/or radiation for this patient population are needed. Further, the optimal sequencing of TRK inhibitors for all patients with tumors harboring NTRK fusions is unknown, especially in the setting of acquired resistance at the level of both first- and second-generation agents, and the risk of development under the forces of selective pressure.

With the conclusion of early phase clinical trials testing the anti-neoplastic properties of TRK inhibitors occurring less than 5 years ago, enough time has not yet elapsed to fully evaluate and understand long-term toxicities. While first-generation TRK inhibitors exhibit a favorable overall safety profile in the short-term, especially when compared with other TKIs^64,72–74, the adverse effects associated with extended use has yet to be explored. Since TRK signaling is intricately involved in nervous system development and maintenance⁷⁵, the long-term influences of these agents are essential to characterize, especially in the pediatric population. Reassuringly, through preclinical studies, the implications of TRK inhibition on a developing nervous system is thought to be less consequential in the post-embryonic period⁵⁹. However, an increased frequency of some on-target adverse effects (those related to TRK signaling in nervous system development and maintenance) are seen in the pediatric population when compared to the adult population, such as weight gain^57,72, and in second-generation TRK inhibitors with more potent inhibition (such as dizziness, ataxia, and gait disturbances)⁷⁰. With these correlations, full evaluation of toxicity profiles is imperative to assure safety in long-term use, which is a clinical scenario surfacing with increased frequency with these promising targeted agents.

As discussed, the frequency and breadth in pediatric cancers harboring NTRK fusions continues to expand as more large-scale sequencing studies are conducted. Currently, the extent of evaluation for NTRK fusions is quite variable. While these are rare events, their identification can lead to durable responses to well tolerated targeted therapy. Thus, screening for NTRK fusions in the setting of a new diagnosis of tumors with a high prevalence of NTRK fusions and all high-risk, relapsed and treatment-refractory tumors should be strongly considered.