NTRK Gene Fusions in Solid Tumors and TRK Inhibitors: A Systematic Review of Case Reports and Case Series

The approval of larotrectinib and entrectinib for cancer patients harboring an NTRK gene fusion has represented a milestone in the era of “histology-agnostic” drugs. Among the clinical trials that led to the approval of these two drugs, most of the enrolled patients were affected by soft tissue sarcomas, lung, and salivary gland cancer. However, as next-generation sequencing assays are increasingly available in the clinical setting, health care professionals may be able to detect NTRK gene fusions in patients affected by tumor types under or not represented in the clinical trials. To this aim, we systematically reviewed MEDLINE from its inception to 31 August 2022 for case reports and case series on patients with NTRK gene fusion-positive tumors treated with TRK inhibitors. A virtual cohort of 43 patients was created, excluding those enrolled in the above-mentioned clinical trials. Although our results align with those existing in the literature, various cases of central nervous system tumors were registered in our cohort, confirming the benefit of these agents in this subgroup of patients. Large, multi-institutional registries are needed to provide more information about the efficacy of TRK inhibitors in cancer patients affected by tumor types under or not represented in the clinical trials.


Rationale
In the last decade, the advent of precision medicine has revolutionized the treatment landscape of several solid tumors [1]. The rapidly expanding knowledge in genomics, proteomics, and transcriptomics has led to the discovery of new molecular alterations and markers of immune phenotypes shared by multiple tumor types regardless of their site of origin [2]. In this scenario, the availability of pharmacological agents specifically and selectively targeting those alterations and markers has led to the approval of the so-called "histology-agnostic" drugs [3]. This new therapeutic approach has determined a paradigm shift in cancer treatment, paving the way for a new class of biomarker-driven anticancer agents that transcend tumor histologies [4]. Since 2017, the Food and Drug Administration (FDA) has approved six anticancer agents with a "histology-agnostic" indication: two immune checkpoint inhibitors (for both cancers with high tumor mutational burden or mismatch-repair deficiency/microsatellite instability) [5][6][7] and four targeted therapies (for tumors harboring a BRAF V600E mutation or a neurotrophic tyrosine receptor kinase (NTRK) gene fusion) [8][9][10].
The NTRK gene family is composed of three members (NTRK1, NTRK2, and NTRK3) that encode for three tropomyosin receptor kinases (TRK) (TrkA, TrkB, and TrkC, respectively), physiologically involved in the development of the central and peripheral nervous system [11]. The occurrence of NTRK gene rearrangements determines a constitutive activation of TRK receptors, potentially leading to cancer cell transformation [11]. In this framework, several studies have recently demonstrated that these alterations are oncogenic drivers of several tumors [12]. They can be detected with high frequencies (up to 90%) in some rare cancers (secretory breast carcinoma, mammary analogue secretory carcinoma, congenital infantile fibrosarcoma) and with lower frequencies (less than 1%) in a range of common adult patients' cancers (non-small cell lung, colorectal, thyroid, and salivary gland cancers) [13].
Two oral targeted agents are currently available for unresectable locally advanced or metastatic cancer patients harboring an NTRK gene fusion: larotrectinib [9] and entrectinib [10]. The former was approved in 2018, based on the results of a prespecified combined analysis of three clinical trials published by Drilon et al. [14]. The latter was approved in 2019, after the publication of a pooled analysis of three clinical trials by Doebele et al. [15]. Although the above clinical trials enrolled patients with NTRK gene fusion-positive tumors regardless of their site of origin, most were affected by soft tissue sarcomas, lung, and salivary gland cancer [14,15]. In contrast, some tumor types were under or not represented, such as gastrointestinal (except for colorectal cancer), genitourinary, and gynecological malignancies [14,15]. However, as genomic medicine is constantly evolving, next-generation sequencing (NGS) assays for comprehensive genomic profiling have become increasingly available in clinical and research settings [16]. Therefore, health care professionals may have the chance to detect an NTRK gene fusion even in patients with tumor histologies under or not represented in clinical trials [14,15], thus facing the opportunity to provide patients with TRK inhibitors despite a lack of data.

Objective
To this aim, we evaluated the benefit of TRK inhibitors in a virtual cohort of patients affected by NTRK gene fusion-positive solid tumors derived from case reports (CR) and case series (CS) available in the literature. Although CR and CS are characterized by a high risk of biases due to their nature, they have consistently advanced medical knowledge on rare conditions, providing helpful information for clinical practice. To our knowledge, CR and CS on cancer patients harboring an NTRK gene fusion and treated with TRK inhibitors have never been systematically reviewed.

Protocol and Registration
The literature search was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines for systematic reviews [17] (Supplementary Material S1). The protocol was designed a priori by all the authors and was registered on the Open Science Framework website (https://osf.io/ngz6s/ [accessed on 1 October 2022]).
We included CR and CS on pediatric and adult patients affected by locally advanced or metastatic NTRK gene fusion-positive solid tumors treated with larotrectinib or entrectinib. Only publications in English or European languages were considered. Furthermore, CS were deemed eligible only if single-case descriptions were provided. Letters to the editor or other types of publications reporting CR or CS were also considered if they satisfied all the previous criteria. On the contrary, CR and CS on NTRK gene fusion-positive solid tumor patients treated with TRK inhibitors not yet approved by the FDA were excluded. In addition, eligible publications reporting on patients enrolled in the clinical trials that led to the approval of larotrectinib and entrectinib were excluded, as well [14,15]. Indeed, this systematic review aimed to create a virtual cohort of patients different from the populations enrolled in the clinical trials. Finally, patients who received TRK inhibitors through an early access program or were enrolled in observational or non-interventional clinical trials (e.g., post-marketing safety surveillance studies) were considered eligible.

Search
The electronic PubMed database was searched from inception to 31 August 2022 to identify all relevant publications. No research filters were used. The search strategy was decided on after a discussion among the authors. It was composed of four different syntaxes that were consecutively evaluated: (i) larotrectinib AND (case report OR case series); (ii) entrectinib AND (case report OR case series); (iii) larotrectinib AND NTRK; (iv) entrectinib AND NTRK. The results were uploaded to a reference management software (Zotero), and duplicates were later removed.
Two groups of three authors (G.M.I., M.C., S.G. and S.R., S.G., B.C.) independently scrutinized the available results following a two-stage study selection process. Firstly, all titles and abstracts were screened for potential relevance. Secondly, full texts of potentially appropriate results were retrieved and further assessed for eligibility. An agreement of the three authors of each group was required for exclusion at both stages. At the end of the selection process, the two groups discussed the list of publications to include in the review. In parallel, all the references of eligible studies were also hand-searched for other potential publications. In case of disagreements on the study selection, a consultation with two additional authors (M.R. and F.T.) was required. Finally, the two groups of authors achieved a complete consensus on the included papers and reported the results in the PRISMA flow diagram [17].

Data Charting Process
Three authors (G.M.I., S.R., F.T.) created a data charting template using Microsoft Excel software. As in the study selection process, the two groups of authors charted the data independently, discussing the results in an interactive process. Once a CS was analyzed, the single-case data were extracted individually. Disagreements in the charting process required consultation with two additional authors (M.R. and F.T.) and were resolved by consensus.
The variables extracted were: first author, journal of publication, year of publication, type of evidence (CR or CS), age and sex of the patient, primary tumor type, histological subtype, site of metastases (before being treated with NTRK inhibitors), NTRK gene fusion diagnostic modality, type of NTRK gene fusion, previous treatments, type of TRK inhibitor received, best radiological response (according to Response Evaluation Criteria in Solid Tumors (RECIST)), and outcomes.
Variables charted from eligible publications were described using numbers and proportions for categorical variables while mean, standard deviation, median, and interquartile range were used for continuous variables. The descriptive analyses were performed using R Studio (version 1.4.1106) software. Quantitative analyses were performed considering the overall population composed of both adult and pediatric patients, as reported in the clinical trials that led to the approval of larotrectinib and entrectinib. No inferential or predictive statistics analyses were performed.

Risk of Bias Assessment
The risk of bias assessment of included publications was performed using the methodological tool proposed by Murad et al. [18]. The two groups of authors responsible for the "selection process" assigned a binary response (yes = 1 and no = 0) to every question required by the tool. Subsequently, an aggregated score was formulated for each included CR or CS. In case of disagreements, a consultation with two additional authors (M.R. and F.T.) was required.

Study Characteristics
The results of the literature search and the study selection process are displayed through a PRISMA diagram in Figure 1

Study Characteristics
The results of the literature search and the study selection process are displayed through a PRISMA diagram in Figure 1. A total of 38 publications were included in this systematic review: 32 CR and 6 CS. They were published between 2018 and 2022 ( Figure 2). Particularly, 21 CR and 2 CS reported about adult patients, while 11 CR and 4 CS were on pediatric patients. All the publications were written in English. Among the ineligible publications, 28 CR/CS were excluded because they reported data about patients enrolled in the clinical trials that led to the approval of larotrectinib and entrectinib. Furthermore, 1 CR was excluded because the patient was affected by a hematologic tumor. The results of the included single sources of evidence are described in Table 1. A total of 38 publications were included in this systematic review: 32 CR and 6 CS. They were published between 2018 and 2022 ( Figure 2). Particularly, 21 CR and 2 CS reported about adult patients, while 11 CR and 4 CS were on pediatric patients. All the publications were written in English. Among the ineligible publications, 28 CR/CS were excluded because they reported data about patients enrolled in the clinical trials that led to the approval of larotrectinib and entrectinib. Furthermore, 1 CR was excluded because the patient was affected by a hematologic tumor. The results of the included single sources of evidence are described in Table 1.

Synthesis of Results
Data extracted from the included CR and CS allowed us to create a virtual cohort of 43 patients affected by NTRK gene fusion-positive tumors treated with TRK inhibitors, including 25 adult and 18 pediatric patients ( Table 2).
In terms of treatment, larotrectinib and entrectinib were administered in 81.4% and 16.3% of patients, respectively. Furthermore, in one case (2.3%), the patient was treated with both drugs. TRK inhibitors were used as first-and second-line therapy in 34.9% of cases. In addition, in 16.3% of cases, TRK inhibitors were used as third line and, in 6.9% of cases, they were used in the subsequent lines. The best radiological response to TRK inhibitors was partial response (74.5%), while a complete response was achieved in 20.9% of cases. In contrast, only 2.3% of cases had stable or progressive disease, respectively. Finally, at the time of publication, 72.1% patients were alive with disease, 11.6% died due to progressive disease, and no information was reported for 16.3% of cases.

Quality Assessment
All the included publications were evaluated with a tool proposed by Murad et al. [18]. The aggregated scores assigned to every CR and CS included in the study are available in Supplementary Material S2.

Summary of Evidence
In recent years, the advances in molecular diagnosis have led to a significant change in how cancer patients are treated, shifting from a "one size fits all" therapeutic paradigm toward a "precision medicine" approach with the development of new agents targeting specific genomic abnormalities . This revolution has been witnessed by the progressive discovery of an increasing number of actionable molecular alterations, gaining the chance of improving cancer patients' survival with biomarker-driven drugs [1]. Nowadays, several "agnostic therapies" have been approved for patients harboring specific genomic alterations, based on the possibility of administering targeted therapies across different tumor histologies and regardless of the tumor site of origin [59].
In recent years, NTRK gene fusion has represented one of the most groundbreaking discoveries among the biomarkers targeted by agnostic therapies. The family of NTRK genes (NTRK1, NTRK2, NTRK3) encodes for TrkA, TrkB, and TrkC receptors, respectively [11]. They are composed of an intracellular domain, a transmembrane region, and an extracellular domain for ligand binding [60]. Beyond their physiological involvement in the nervous system's development, the constitutive activation of Trk receptors mediated by the occurrence of NTRK gene fusions leads to the uncontrolled growth of cancer cells [11]. Particularly, the interaction between TRK receptors and their ligands triggers the activation of signal transduction pathways implicated in tumorigenesis, including Ras/Mitogen activated protein kinase (MAPK), the phosphatidylinositol-3-kinase (PI3K)/Akt, and the mammalian target of rapamycin (mTOR) pathways (Figure 3) [61].
prospectively sequenced patients. Seventy-six cases (0.28%) with confirmed NTRK fusions were identified, mainly represented by salivary gland cancer, soft tissue sarcomas, and thyroid cancers [63]. In 2020, Forsythe et al. reported the results of a systematic review and meta-analysis aiming at describing the NTRK gene fusion incidence among available studies published from 1987 to 2020. The authors showed that rare tumors, such as secretory breast cancer, infantile fibrosarcoma, secretory salivary gland cancer, papillary thyroid carcinoma (pediatric), and congenital mesoblastic nephroma, were characterized by the highest NTRK gene fusion frequencies (from 10 to 92.8%) [64]. In 2021, Westphalen et al. reported the results of a retrospective study which aimed to evaluate the genomic landscape and prevalence of NTRK gene fusions in a large real-world database of comprehensive genomic profiling data (FoundationCORE). Among more than 295,000 analyzed cancer cases, salivary gland cancers (2.43%), soft tissue sarcomas (1.27%), and thyroid cancers (1.25%) were the most common tumor types harboring NTRK gene fusions [65].  In 2013, Vaishnavi et al. described, for the first time, an NTRK1 gene rearrangement in a cohort of patients affected by non-small cell lung cancer [62]. Since then, several studies aiming to describe the genomic landscape and the prevalence of NTRK gene fusions in solid tumors have been published. In 2019, Rosen et al. reported the analysis of genomic and clinical data about NTRK gene-positive tumors identified among more than 26,000 prospectively sequenced patients. Seventy-six cases (0.28%) with confirmed NTRK fusions were identified, mainly represented by salivary gland cancer, soft tissue sarcomas, and thyroid cancers [63]. In 2020, Forsythe et al. reported the results of a systematic review and meta-analysis aiming at describing the NTRK gene fusion incidence among available studies published from 1987 to 2020. The authors showed that rare tumors, such as secretory breast cancer, infantile fibrosarcoma, secretory salivary gland cancer, papillary thyroid carcinoma (pediatric), and congenital mesoblastic nephroma, were characterized by the highest NTRK gene fusion frequencies (from 10 to 92.8%) [64]. In 2021, Westphalen et al. reported the results of a retrospective study which aimed to evaluate the genomic landscape and prevalence of NTRK gene fusions in a large real-world database of comprehensive genomic profiling data (FoundationCORE). Among more than 295,000 analyzed cancer cases, salivary gland cancers (2.43%), soft tissue sarcomas (1.27%), and thyroid cancers (1.25%) were the most common tumor types harboring NTRK gene fusions [65].
From a diagnostic point of view, several assays have been developed to accurately identify patients harboring an NTRK gene fusion, including IHC, FISH, RT-PCR, and DNA-or RNA-based NGS. However, the advantages and disadvantages of each diagnostic modality must be weighed when evaluating the tissue specimen required for NTRK gene fusions analysis [66]. To this aim, the European Society of Medical Oncology Translational Research and Precision Medicine Working Group published recommendations for a rational approach for detecting NTRK gene fusions based on the prevalence of these alterations among different tumor histotypes [67]. In tumors with highly frequent NTRK fusions, the best methodologies to use as confirmatory tests are FISH, RT-PCR, or targeted RNA NGS assays. Differently, in tumors where NTRK fusions are recognized not to be highly prevalent, NGS targeted panels (DNA-or RNA-based) are the recommended tests of choice. In case of an NTRK gene fusion detection by the use of NGS-based assays, a further confirmatory test with IHC is recommended [67]. Alternatively, if a sequencing platform is not available, IHC may be used as a screening tool, followed by an NGS targeted panel, in case of a positive result [67].
Nowadays, two targeted agents are available for patients harboring NTRK gene fusions: larotrectinib and entrectinib. The former was granted accelerated approval by the FDA in 2018 [9] after Drilon et al. published a prespecified combined analysis of three clinical trials evaluating the activity of larotrectinib in patients with locally advanced or metastatic NTRK fusion-positive solid tumors. The studies involved in the analysis were a phase 1 trial on adults (LOXO-TRK-14001), a phase 1-2 trial on children (SCOUT), and a phase 2 "basket" trial involving adolescents and adults (NAVIGATE). Among the 55 patients enrolled, the overall response rate was 75% (95% CI, 61-85%) after a median follow-up of 9.4 months [14]. After two years, Hong et al. reported the results of an expanded pooled efficacy analysis on 159 patients enrolled across the same 3 clinical trials. Although only 153 patients were evaluable for response, the objective response rate was 79% (95% CI, 72-85%) after a median follow-up of 11.1 months. In total, 24 (16%) patients achieved a complete response, and the median progression-free survival was 28. The studies involved in the analysis were two phase I trials (ALKA-372-001 and STARTRK-1) and one phase II trial (STARTRK-2). The objective response rate was 57% (95% CI, 43.2-70.8%) among the enrolled 54 patients, after a median follow up of 12.9 months. The complete response rate was 7% and the median duration of response was 10 months (95% CI, 7.1-NE) [15].
Overall, the population included in the above-mentioned clinical trials was composed of patients affected by different tumor histologies harboring an NTRK gene fusion. The most frequent tumor histotypes were soft tissue sarcomas, lung, salivary gland, and thyroid cancer. In contrast, gastrointestinal (except for colorectal cancer), genito-urinary, and gynecological malignancies were under or not-represented [14,15]. However, since the rapid evolution of precision medicine has determined a reduction in the cost of molecular profiling, NGS assays have become increasingly available in clinical practice. Accordingly, health care professionals are more likely to face the detection of NTRK gene fusions in patients affected by tumor histologies under or not represented in the clinical trial, with the consequent lack of data on the efficacy of TRK inhibitors. Therefore, we systematically reviewed the available literature for CS and CR on NTRK gene fusion-positive tumors treated with TRK inhibitors.
The publication of CR and CS has often played an essential role in advancing medical knowledge on rare conditions . In this context, the creation of a virtual cohort of patients from CR and CS on NTRK gene fusion-positive solid tumors treated with TRK inhibitors has a double value. On the one hand, it provides health care professionals with a single source of evidence that is easy to access and summarizes data derived from multiple clinical experiences. On the other hand, the choice to exclude both CR and CS on patients enrolled in the clinical trials lends this virtual cohort of patients the opportunity to be compared with populations enrolled in the clinical trials [14,15]. Nevertheless, since the intrinsic bias of both CR and CS is related to their nature and the inclination to publish more reports of positive rather than negative responses to treatments, it is essential to be cautious in the extrapolation of emerging data to clinical practice [18].
The data obtained from this patients' cohort appear in line with those available in the current literature. Indeed, patients were affected by the most frequent tumor types as those reported in the clinical trials that led to the approval of larotrectinib and entrectinib [14,15], such as soft tissue sarcoma, thyroid, and salivary gland tumors (Table 3). In our opinion, a reasonable explanation for these similar results relies on the different frequencies of NTRK gene fusions among different tumor histotypes, as reported in the above-mentioned multiple retrospective studies [63][64][65]. Although the NTRK gene fusions are rare genomic abnormalities, they can be detected with high frequencies in some rare cancers (secretory breast carcinoma, mammary analogue secretory carcinoma, congenital infantile fibrosarcoma) and with lower frequencies in a range of common adult patients' cancers [13].
In our cohort, we found a higher prevalence of patients affected by primary CNS tumors (27.9%) compared to those reported in the abovementioned clinical trials' combined analyses. Nowadays, these patients lack effective therapies [71] and, thus, our results confirm the indication of the current guidelines that encourage performing molecular testing for NTRK gene fusion in CNS tumors [72]. However, it is important to cautiously interpret the extremely positive outcomes reported by the patients included in our cohort due to the "publication bias" related to CR and CS. In this direction, a recently published post-hoc analysis of two clinical trials that led to the approval of larotrectinib showed an objective response rate of 30% (95% CI, 16-49) with a 24-week disease control rate of 73% (95% CI: 54-87) for primary CNS tumors [73]. Nevertheless, considering the reported benefit of administering TRK inhibitors in this subgroup of patients and the data available in the literature, the detection of an NTRK gene fusion in CNS tumors may represent a "game-changer" in treating those malignancies.
Finally, our results confirm how TRK inhibitors represent a significant therapeutic strategy for metastatic cancer patients harboring an NTRK gene fusion, and, thus, all the cancer patients harboring this molecular alteration should be evaluated for specific inhibitors [74]. Although these drugs are generally administered as a single-agent treatment in the metastatic setting, we found a cervical cancer patient treated with the combination of chemotherapy plus larotrectinib [30] and two pediatric patients treated with "adjuvant" maintenance larotrectinib after definitive surgical resection of a kidney sarcoma and anaplastic astrocytoma [38]. These reports are interesting because they focus on new potential strategies of TRK inhibitors' administration in terms of clinical setting (adjuvant vs. metastatic disease) and combinations (single agent vs. combined treatment). In our opinion, considering the importance of maximizing the benefit of these drugs, we believe that these strategies should be further assessed in the near future. In parallel, it is essential to provide guidelines to help healthcare professionals determine the best time to administer these anticancer agents. Recently, a Belgian expert consensus for the tumor-agnostic treatment of NTRK gene fusion-driven solid tumors with larotrectinib has been published [75]. The authors distinguished three categories of patients affected by NTRK gene fusion-driven solid tumors: (i) those affected by advanced solid tumors with non-satisfactory standard-of-care (SoC) therapies, (ii) those with advanced solid tumors with satisfactory SoC therapies, and (iii) those affected by locally advanced tumors [75]. For the first group, it was suggested to consider the use of larotrectinib as a first-line treatment, considering the high unmet medical need for these patients. Concerning the second group, the authors suggested larotrectinib as a consideration for second or later treatment line (after failure of SoC). For the last group, larotrectinib should be considered as a neoadjuvant therapy [75].

Future Perspectives
We believe that the creation of prospective, international patients' registries would represent an essential tool to acquire "real-word" data on the efficacy of TRK inhibitors among NTRK gene fusion positive tumors, including histotypes under or not represented in previous clinical trials. Currently, two ongoing studies aim to collect data prospectively in these patients: the REALTRK registry [76] and the TRacKING registry [77]. The former aims to analyze the treatment reality and outcomes of NTRK gene fusion-positive patients treated with TRK inhibitors until at least 36 months after their inclusion in the study [76]. The latter aims to study the real-life management of patients with rare actionable fusions, including those harboring an NTRK gene fusion [77].

Limitations
The present systematic review has some limitations. Firstly, a limited number of CR and CS lacked relevant clinical data and were at high risk of several biases. Secondly, we did not include CS reporting on aggregate patients' data instead of individual data. Unfortunately, these limitations are common in studies that evaluate CR and CS [78,79], but they were considered before designing the protocol. In addition, the search strategy was designed to be extensive, and both the data extraction and selection were performed with a "two-stage process" to minimize bias.

Conclusions
The results of our systematic review confirmed the efficacy of TRK inhibitors in cancer patients harboring an NTRK gene fusion. Although the patients of our virtual cohort were mainly affected by the most frequently diagnosed tumor histologies in patients enrolled in the available clinical trials on TRK inhibitors, we reported a higher prevalence of CNS tumors, confirming the benefit of these agents even in this subgroup of patients. Large, multi-institutional registries are needed to provide more information about the efficacy of TRK inhibitors in cancer patients affected by under or not represented histologies included in the clinical trial that led to the approval of entrectinib and larotrectinib.