Abstract
Background
Although NTRK fusions are actionable targets for a wide array of solid tumors, treatment-relevant biomarkers are heterogeneous and remain incompletely characterized for real-world NTRK fusion-positive populations. This retrospective cohort study provides a detailed molecular characterization of functional NTRK fusions and identifies biomarker associations in a real-world pan-cancer population.
Methods
We retrospectively analyzed a cohort of 345 patients with functional NTRK fusions identified by targeted next-generation sequencing (NGS). Microsatellite instability (MSI), tumor mutational burden (TMB), homologous recombination deficiency (HRD), and Epstein-Barr virus (EBV) status were assessed on an available-case basis.
Results
Among NTRK fusions, NTRK1 was the most common (67.2%), followed by NTRK3 (20.0%) and NTRK2 (12.8%); meanwhile, TPM3 and ETV6 were the most common gene fusion partners, with ETV6 being enriched in NTRK3 fusions. MSI-high (MSI-H) status occurred in 17.0% of MSI-tested cases and was strongly lineage-dependent, concentrated in colorectal/intestinal tumors (67.1%) but rare in thoracic/lung tumors (0.84%). MSI-H tumors exhibited a markedly elevated TMB compared with microsatellite stable (MSS) tumors, whereas TMB did not differ across NTRK subtypes. A high HRD score was observed in 14.2% of HRD-tested cases, and EBV positivity was rare (0.6%). The co-alteration landscape revealed both lineage-associated events and MSI-linked signatures (e.g., RNF43/ACVR2A enrichment in MSI-H tumors).
Conclusions
Functional NTRK fusion-positive tumors comprise biologically and clinically distinct subsets defined by tumor type and MSI/TMB context. Concurrent reporting of MSI/TMB (and HRD/EBV when available) together with NTRK fusion status may facilitate integrated clinical interpretation, support precision treatment selection, and refine trial stratification in clinical practice.
Keywords: NTRK fusion, pan-cancer, gene fusion partner, microsatellite instability (MSI), tumor mutational burden (TMB)
Highlight box.
Key findings
• In the analysis of a real-world pan-cancer cohort (n=345), NTRK1 fusions were the most common NTRK fusions (67.2%), with TPM3 (NTRK1) and ETV6 (NTRK3) being the predominant partners.
• Microsatellite instability (MSI)-high (MSI-H) status was highly lineage-specific: common in colorectal/intestinal (67.1%) but rare in thoracic tumors (0.84%). MSI-H correlated with high tumor mutation burden (TMB).
• High homologous recombination deficiency (HRD) scores were found in 14.2% of tested cases, while Epstein-Barr virus (EBV) positivity was rare (0.6%). Co-alterations linked to lineage and MSI status were identified.
What is known and what is new?
• NTRK fusions are actionable oncogenic drivers across tumor types, but the landscape of their real-world biomarkers has not been extensively characterized.
• This study developed a systematic, real-world molecular profile of NTRK fusion by quantifying the prevalence of MSI-H, TMB, HRD, and EBV alongside NTRK fusion subtypes and partners, revealing strong lineage-specificity for MSI-H.
What is the implication, and what should change now?
• Tumor lineage and concurrent biomarkers (MSI/TMB) define distinct biological subsets within NTRK fusion-positive cancers, which may be relevant to clinical management.
• MSI/TMB (and HRD/EBV if available) should be integrated into clinical reporting alongside NTRK fusion status to guide optimized treatment selection and refine trial stratification.
Introduction
Neurotrophic tyrosine receptor kinase (NTRK1/2/3) gene fusions are uncommon in solid tumors, but constitute a clinically important, tumor-agnostic therapeutic target due to their role as dominant oncogenic drivers. In population-scale datasets, NTRK fusions are generally detected at <0.5% in common cancer types (1). The clinical value of identifying NTRK fusions is supported by the marked and durable efficacy of TRK inhibitors across different tumor types (2). In key trials on larotrectinib and entrectinib, patients with advanced or metastatic NTRK fusion-positive tumors achieved high objective response rates and sustained clinical benefit, independent of tumor histology (3-6). These results have made timely and accurate NTRK fusion testing an essential component of contemporary precision oncology practice.
However, the routine detection of NTRK fusions in clinical practice remains technically challenging. This is largely attributed to the heterogeneity of fusion partners, variability in genomic breakpoints, and frequently suboptimal RNA quality in real-world specimens. Screening methods such as pan-TRK immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH), as well as molecular approaches such as reverse transcription polymerase chain reaction (RT-PCR) and next-generation sequencing (NGS), each possess distinct strengths and limitations. Notably, interassay inconsistencies have been reported across these platforms, further complicating diagnostic standardization (7). Of particular note, meta-analytic evidence suggests that pan-TRK IHC is useful for screening but is not a reliable substitute for molecular confirmation (8), highlighting the importance of definitive molecular assays for accurate NTRK fusion detection (9). In other research, RNA-based NGS has been validated as the gold standard method for detecting NTRK gene fusions (10).
Even when an NTRK fusion is confirmed, a positive result alone is often insufficient to address subsequent clinical decisions. In clinical practice, treatment selection (11,12) frequently depends on the broader biomarker context, most notably microsatellite instability (MSI) (13) and tumor mutational burden (TMB) (14,15), and, when available, homologous recombination deficiency (HRD) (16,17) and Epstein-Barr virus (EBV) (18). The variability of these biomarkers across tumor lineages significantly influences therapeutic strategies, treatment sequencing, and clinical trial eligibility. However, the joint distribution and co-alteration patterns of these biomarkers in real-world NTRK fusion-positive populations has not been well described.
Defining these co-alterations may further clarify the biological subsets within NTRK fusion-positive tumors and improve their clinical interpretation. Specifically, MSI-associated signatures and lineage-associated molecular events may identify clinically distinct subgroups that are not discernible based on NTRK subtype alone. Accordingly, we established a real-world pan-solid tumor cohort of 345 patients with functional NTRK fusions identified through targeted NGS. For this cohort, we characterized NTRK fusion subtypes and partner genes, summarized co-alterations, and examined the associations with MSI, TMB, HRD, and EBV on an available-case basis. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0365/rc).
Methods
Study design and participants
This retrospective cohort study analyzed a real-world population of patients with solid tumors harboring functional NTRK fusions. Eligible patients were retrospectively identified in China between 2018 and 2025. The inclusion criteria were as follows: (I) histologically confirmed solid malignancy; (II) detection of a functional NTRK fusion via a validated NGS assay; and (III) availability of formalin-fixed paraffin-embedded (FFPE) tumor tissue or blood samples. Cases with insufficient molecular data or ambiguous fusion annotation were excluded. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Review Committee of Shanghai Fourth People’s Hospital (approval No. SYLL2023031). As this was a retrospective analysis of de-identified data obtained from routine clinical practice, the requirement for informed consent was waived by the ethics committee.
Data sources and variables
Demographic, clinical, and pathological data were extracted from electronic medical records via a standardized case report form. Molecular data included NTRK fusion subtype, genomic breakpoints, fusion partner genes, co-occurring alterations in oncogenic pathways, and status of the following biomarkers: MSI, TMB, HRD, and EBV infection. Assay types [DNA/RNA-targeted NGS or circulating tumor DNA (ctDNA) NGS], sequencing platforms, and reporting thresholds were documented for each case (19).
Bias and quality control
To minimize selection bias, all NGS assays were performed in College of American Pathologists (CAP)/Clinical Laboratory Improvement Amendments (CLIA)-accredited and ISO-15189-certified laboratories under validated workflows. Variant annotation was completed according to international guidelines. For biomarker assessment, MSI was determined via NGS-based profiling, TMB-high (TMB-H) status was defined as ≥10 mutations/Mb, HRD positivity was determined according to genomic scarring scores or pathogenic alterations in homologous recombination repair genes, and EBV status was determined via a validated NGS-based algorithm on four viral genes.
NGS for tissue samples
After confirming a tumor cell content exceeding 20% by hematoxylin and eosin staining of FFPE sections, the samples were subjected to concurrent DNA and RNA extraction. Library preparation and sequencing were performed on a NovaSeq 6000 platform (Illumina, San Diego, CA, USA) at a mean depth of 1,000×. Paired-end raw reads were aligned to the human reference genome (hg19) using the Burrows-Wheeler Aligner (version 0.7.12). Polymerase chain reaction (PCR) duplicates were subsequently removed, and alignment metrics were generated with Picard (version 1.130) and SAMtools (version 1.1.19). Notably, RNA-based sequencing offers a higher sensitivity for detecting gene fusions compared to DNA-based approaches, as it circumvents the confounding effects of intronic regions. A subset of forty tissue specimens was concurrently assessed using both DNA and RNA sequencing with the 3DMed Onco Core Tissue Kit (3D Medicines Inc., Shanghai, China).
NGS for ctDNA
Within 2 hours of collection, blood samples were centrifuged in Streck tubes at 1,600 ×g for 10 minutes at 4 ℃. Approximately 5 mL of the resulting plasma supernatant was transferred to a fresh tube and subjected to a second centrifugation at 16,000 ×g for 10 minutes at 4 ℃ to eliminate residual cells and debris. Cell-free DNA (cfDNA) was then extracted from the clarified supernatant using the QIAamp Circulating Nucleic Acid Kit (Qiagen, Hilden, Germany) in accordance with the manufacturer’s instructions. Sequencing libraries were constructed from cfDNA using the KAPA Hyper Prep Kit (KAPA Biosystems, Roche, Basel, Switzerland) following the recommended protocol, and each library was labeled with unique molecular identifiers (UMIs). After PCR amplification and purification for target enrichment, library concentration and fragment size distribution were assessed using a Qubit 3.0 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) and a LabChip GX Touch HT Analyzer (PerkinElmer, Shelton, CT, USA), respectively. The enriched libraries were subsequently sequenced on a NovaSeq 6000 platform (Illumina) with 100-bp paired-end reads.
Definition of NGS-based MSI
MSI status was determined via the small-panel NGS on the MSI (SPANOM) algorithm. Briefly, 2,539 candidate microsatellite loci were screened and evaluated for sequencing coverage with whole-exome sequencing data [49 MSI-H and 96 microsatellite stable (MSS) samples]. For each sample, the percentage of loci classified as unstable was calculated; a proportion >0.4 was defined as MSI-high (MSI-H); otherwise, it was classified as MSS.
TMB calculation
TMB was calculated as the total number of somatic single-nucleotide variants (SNVs) and insertions or deletions (InDels) per megabase of the sequenced coding region, with driver mutations excluded. The count encompassed all variant types—including missense, silent, stop-gain, stop-loss, in-frame, and frameshift mutations—detected within the coding regions of the targeted genes.
HRD assessment
HRD was assessed with a customized 3DMed HRD assay (3D Medicines Inc.) based on targeted sequencing. The 3DMed-HRD algorithm characterizes genome-wide “genomic scar” signals using >10,000 single-nucleotide polymorphisms distributed across the genome and generates an HRD score as the sum of loss of heterozygosity, telomeric allelic imbalance, and large-scale state transition, with adjustment for tumor ploidy and purity. HRD positivity was defined as an HRD score ≥42 and/or the presence of a deleterious BRCA1/2 mutation.
EBV assessment
EBV status was inferred from targeted sequencing with an NGS-based EBV detection algorithm. Four EBV genes (EBNA-1/2/3 and BZLF1) were used for scoring. An EBV score was computed as the median of the normalized depths (NorDepths) of these four EBV genes, where NorDepth was the EBV-gene coverage normalized by the overall panel sequencing depth. The optimal cutoff for EBV positivity was determined through receiver operating characteristic analysis as an EBV score ≥0.05695; the EBV score correlated with EBV load as measured by quantitative PCR (BamHI W fragment), and the analytical limit of detection was reported as 5% at a sequencing depth of 100× in EBV-positive cell line dilution experiments.
Statistical analysis
All statistical analyses were conducted with R version 4.3.0. Categorical variables are expressed as counts and percentages and were compared with χ2 tests or Fisher exact tests as appropriate. Continuous variables are expressed as the median and interquartile range (IQR). Between-group comparisons were compared via Mann-Whitney U tests (two groups) or Kruskal-Wallis tests (≥3 groups). All analyses were performed on an available-case basis. All tests were two-sided, with P<0.05 being considered statistically significant.
Results
Patient cohort and baseline characteristics
A total of 345 patients from China with functional NTRK fusions identified between 2018 and 2025 were included in this pan-solid tumor cohort (Figure 1). The median age was 63 years (IQR, 54–71 years; range, 4–91 years), and 178 (51.6%) were female. Tumor types were heterogeneous, with the most common being thoracic/lung tumors (148/345, 42.9%) and colorectal/intestinal tumors (80/345, 23.2%), followed by gastric cancer (13/345, 3.8%), prostate cancer (13/345, 3.8%), sarcoma/mesenchymal tumors (13/345, 3.8%), breast cancer (11/345, 3.2%), thyroid cancer (11/345, 3.2%), and other tumor types (46/345, 13.3%). Among specimen types, FFPE slides accounted for 246 (71.30%) and cfDNA (plasma) for 99 (28.70%) (Table 1).
Figure 1.
Genomic landscape of the 345 NTRK fusion tumors that were analyzed for MSI, TMB, HRD, and EBV via NGS. EBV, Epstein-Barr virus; HRD, homologous recombination deficiency; MSI, microsatellite instability; MSI-H, microsatellite instability-high; MSS, microsatellite stable; NA, not available; NGS, next-generation sequencing; NTRK, neurotrophic tyrosine receptor kinase; TMB, tumor mutational burden.
Table 1. Baseline clinicopathologic characteristics of the NTRK fusion-positive cohort (n=345).
| Characteristic | Values |
|---|---|
| Age (years) | |
| Median [interquartile range] | 63 [54–71] |
| Range | 4–91 |
| Sex, n (%) | |
| Female | 178 (51.6) |
| Male | 167 (48.4) |
| Tumor type (major categories), n (%) | |
| Thoracic/lung | 148 (42.9) |
| Colorectal/intestinal | 80 (23.2) |
| Gastric | 13 (3.8) |
| Prostate | 13 (3.8) |
| Sarcoma/mesenchymal | 13 (3.8) |
| Breast | 11 (3.2) |
| Thyroid | 11 (3.2) |
| Other | 46 (13.3) |
| Specimen type (index report), n (%) | |
| FFPE slides | 246 (71.3) |
| cfDNA (plasma) | 99 (28.7) |
cfDNA, cell-free DNA; FFPE, formalin-fixed paraffin-embedded; NTRK, neurotrophic tyrosine receptor kinase.
Characteristics of NTRK fusions in the pan-solid tumor cohort
Among the 345 cases, NTRK1 fusions were the most common (232/345, 67.2%), followed by NTRK3 (69/345, 20.0%) and NTRK2 (44/345, 12.8%) (Figure 2A). The most frequent fusion partners in the overall cohort were TPM3 (21.2%), ETV6 (15.1%), PHF20 (5.8%), LMNA (2.9%), and TPR (2.6%) (Figure 2B). Fusion partner distribution varied by NTRK subtype: ETV6 was predominantly associated with NTRK3 fusions (50/69), whereas TPM3 was most commonly observed with NTRK1 fusions (73/232) (Figure 2C). In addition, 36 of 345 (10.43%) patients harbored at least one other actionable functional fusion driver alongside the NTRK fusion, most frequently ALK (n=8), EGFR (n=3), FGFR3 (n=3), and RET (n=3) (table available at https://cdn.amegroups.cn/static/public/tcr-2026-1-0365-1.xlsx).
Figure 2.
NTRK fusion landscape. (A) Proportions of NTRK1–3. (B) Most common fusion partners. (C) Partner distribution by NTRK subtype. NTRK, neurotrophic tyrosine receptor kinase.
Co-alteration spectrum in NTRK fusion-positive tumors
Across the cohort, the most prevalent co-altered genes (SNV/InDel and/or copy number variation) were TP53 (136/345, 39.4%), EGFR (75/345, 21.7%), PIK3CA (36/345, 10.4%), RNF43 (31/345, 9.0%), KRAS (26/345, 7.5%), RB1 (25/345, 7.2%), CDKN2A (22/345, 6.4%), ARID1A (20/345, 5.8%), ACVR2A (20/345, 5.8%), and BRCA2 (18/345, 5.2%) (Figure 3 and table available at https://cdn.amegroups.cn/static/public/tcr-2026-1-0365-1.xlsx).
Figure 3.
Co-alteration spectrum in NTRK fusion-positive tumors. (A) Most commonly co-altered genes. (B) Co-alterations stratified by MSI status (MSI-H vs. MSS). MSI, microsatellite instability; MSI-H, microsatellite instability-high; MSS, microsatellite stable; NTRK, neurotrophic tyrosine receptor kinase.
Importantly, co-alterations characteristic of MSI-H tumors (e.g., RNF43, ACVR2A, TGFBR2, and ARID1A) were enriched in the MSI-H subset, whereas canonical oncogenic/tumor suppressor events (e.g., TP53, EGFR, PIK3CA, RB1, KRAS, and CDKN2A) were more prominent in MSS tumors (Figure 3B and table available at https://cdn.amegroups.cn/static/public/tcr-2026-1-0365-1.xlsx).
Distribution of molecular biomarkers
Biomarker data availability varied across the 345 patients included in the cohort. MSI status was available for 306 of 345 patients. MSI-H was identified in 52/306 (17.0%), corresponding to 15.1% of the entire cohort (Table 2). The prevalence of MSI-H varied substantially by tumor type: among colorectal/intestinal tumors, MSI-H occurred in 51/76 (67.1%) of MSI-tested cases, whereas it was rare in thoracic/lung tumors (1/119, 0.84%) (Figure 4A). Within this MSI-H colorectal/intestinal subgroup (n=51), we identified somatic alterations in MLH1 (7/51, 13.7%), MSH2 (7/51, 13.7%), MSH6 (17/51, 33.3%), and PMS2 (7/51, 13.7%), consistent with MMR pathway disruption detectable at the DNA level; MSH3 alterations (16/51, 31.4%) were also frequent, consistent with an MSI-associated frameshift-prone background. TMB values were available for 179 of 345 patients, with a median TMB of 6.24 muts/Mb (IQR, 3.24–13.16 muts/Mb). Under commonly used cutoffs, 55 of 179 (30.7%) patients had TMB ≥10 muts/Mb, and 35 of 179 (19.6%) had TMB ≥20 muts/Mb (Table 2). HRD scores were available for 155 of 345 patients, with a median HRD score of 12 (IQR, 4–28). With a prespecified threshold of HRD ≥42 used to define HRD-high status, 22 of 155 (14.2%) patients were classified as HRD-high (Table 2). EBV testing was available for 165 of 345 patients, among whom only 1 (1/165, 0.6%) was EBV-positive (Table 2), precluding meaningful inferential analyses.
Table 2. Molecular biomarker distributions in the NTRK fusion-positive cohort.
| Biomarker | Values |
|---|---|
| MSI status (available n=306) | |
| MSI-H | 52/306 (17.0) |
| Non-MSI-H | 254/306 (83.0) |
| TMB (available n=179) | |
| TMB (muts/Mb) | 6.24 [3.24–13.16] |
| TMB ≥10 muts/Mb | 55/179 (30.7) |
| TMB ≥20 muts/Mb | 35/179 (19.6) |
| HRD (available n=155) | |
| HRD score | 12 [4–28] |
| HRD-high (HRD score ≥42) | 22/155 (14.2) |
| HRD-low (HRD score <42) | 133/155 (85.8) |
| EBV status (available n=165) | |
| EBV-positive | 1/165 (0.6) |
| EBV-negative | 164/165 (99.4) |
Data are presented as n/N (%) or median [interquartile range]. EBV, Epstein-Barr virus; HRD, homologous recombination deficiency; MSI, microsatellite instability; MSI-H, microsatellite instability-high; NTRK, neurotrophic tyrosine receptor kinase; TMB, tumor mutational burden.
Figure 4.
Associations of MSI/TMB/HRD with tumor type and NTRK subtype. (A) MSI-H prevalence across tumor-type categories (tested subset). (B) MSI-H by NTRK subtype. (C) TMB by NTRK subtype. (D) TMB in MSI-H and MSS. (E) HRD by NTRK subtype. HRD, homologous recombination deficiency; MSI, microsatellite instability; MSI-H, microsatellite instability-high; MSS, microsatellite stable; NTRK, neurotrophic tyrosine receptor kinase; TMB, tumor mutational burden.
Associations between NTRK subtypes and MSI, TMB, HRD
Among the MSI-tested cases (n=306), MSI-H frequency varied by NTRK subtype, occurring in 20.3% (41/202) of NTRK1, 0% (0/39) of NTRK2, and 16.9% (11/65) of the NTRK3 subtypes. This indicated a significant association between NTRK subtype and MSI status (P=0.008) (Figure 4B). In contrast, TMB did not significantly differ across NTRK subtypes among patients with available TMB values (Kruskal-Wallis P=0.70) (Figure 4C). Consistent with known biology, MSI-H tumors exhibited a substantially higher TMB compared with MSS tumors (median 70.95 vs. 5.59 muts/Mb; P=1.17e−15) (Figure 4D). HRD scores suggested a borderline difference across NTRK subtypes (Kruskal-Wallis P=0.050), with NTRK3 tumors displaying a numerically lower median HRD score compared with NTRK1 and NTRK2 tumors (Figure 4E).
Discussion
A real-world pan-solid tumor cohort of 345 functional NTRK fusion-positive cancers was analyzed to characterize NTRK fusions within the clinically relevant biomarker context, with MSI, TMB, HRD, and EBV being considered in conjunction with the co-alteration landscape. The key finding of this study is that clinically meaningful heterogeneity among NTRK fusion-positive tumors is correlated with tumor lineage and MSI/TMB status and not solely NTRK subtype.
Our study confirmed the heterogeneous distribution of NTRK fusions and the predominance of lung and colorectal cancers (CRCs). This aligns with a previous literature review that reported NTRK fusions are present across multiple cancers, most commonly non-small cell lung cancer, breast cancer, soft tissue sarcomas, and colorectal carcinoma (20). It further found that NTRK fusions are not isolated driver events but are deeply intertwined with the molecular background of the tumor lineage. We observed in our study that, at the fusion level, NTRK1 was the predominant subtype, and there was a nonrandom fusion partner distribution, with TPM3 and ETV6 being among the most common partners (ETV6 being enriched for NTRK3), which aligns with previous pan-cancer analyses. For instance, Cocco et al. reported that ETV6-NTRK3 fusion is pathognomonic for certain histologies, such as secretory breast carcinoma and cellular/mixed congenital mesoblastic nephroma, with a >90% prevalence in selected cohorts (21). Similarly, a large-scale study of 11,502 samples reported that ETV6-NTRK3 and TPM3-NTRK1 were predominant fusions, corroborating our findings on partner distribution. However, their study noted that pan-Trk IHC reliably detected NTRK1/2 but not all NTRK3 fusions, suggesting assay type significantly impacts fusion identification (22).
Regarding co-alterations, we observed that MSI-H was present in a minority of MSI-tested cases, but its distribution was strongly lineage-dependent—concentrated in colorectal/intestinal tumors and rare in thoracic/lung tumors. Consistent with this biology, MSI-H tumors showed markedly elevated TMB compared with MSS tumors, whereas TMB did not meaningfully differ across NTRK subtypes. One cohort study across 17 cancer types also found that TRK fusion frequently co-occurred with MSI-H in CRC, but it was associated with reduced concurrent oncogenic drivers (P<0.001) and lower TMB (P<0.001) in other cancers (23). These results indicate that the clinical interpretation of the NTRK fusion must be integrated with its concomitant tumor microenvironment and co-biomarkers. This offers a critical rationale for designing clinical trials based on precise molecular stratification rather than on fusion status alone. In the context of MSI-H, elevated TMB reflects a high neoantigen burden that is a key driver of immune checkpoint inhibitor (ICI) response, providing a mechanistic basis for exploring combinations of TRK inhibition and immunotherapy.
Our findings also have direct clinical implications for biomarker testing strategies. A study on American Association for Cancer Research Genomics, Evidence, Neoplasia, Information, Exchange (AACR GENIE) data (24) reported that NTRK-positive and RET-positive CRCs uniquely harbored high TMB (66.6 vs. 35 mut/Mb), surpassing both other RTK-positive CRCs and non-CRC tumors, while all NTRK-positive CRCs with known MSI status exhibited deficient DNA mismatch repair (dMMR). The study proposed TMB-H in CRCs as a practical screen to enrich for clinically actionable NTRK/RET fusions (24). Another pan-cancer study of 67,883 Chinese patients found that novel NTRK fusions (prevalence 0.18%) and NTRK-positive tumors in CRC showed elevated overall TMB, a complex interplay between NTRK fusions and TMB (25). Our data support this notion, as the TMB-H values in our cohort were confined to MSI-H, lineage-defined subsets. These patterns provide a practical reference for molecular interpretation and can inform assay performance evaluation in routine testing. These findings suggest that MSI-H/TMB-H can serve as an enrichment strategy for identifying NTRK fusions, and that the underlying dMMR may be an early event in tumorigenesis, creating a permissive genomic context for the emergence of driver fusions.
We further found that within NTRK fusion-positive tumors, TMB-H is principally a function of MSI-H status and tumor lineage, not the NTRK gene. It should be noted that the immunotherapy-relevant biomarker profile (MSI-H/TMB-H) in NTRK fusion-positive tumors is not predictable from the fusion but reflects underlying lineage-associated hypermutated states. Therefore, although NTRK fusions and MSI-H/TMB-H can co-occur, their interplay may modulate therapeutic outcomes, and thus integrated biomarker assessment is necessary. The clinical consequence of this biomarker context was identified in a previous study of 187 patients with metastatic CRC. The study reported that NTRK fusions were detected in 5.3% of dMMR/MSI cases (particularly in MLH1-methylated, RAS/BRAF wild-type tumors), indicating that NTRK fusion may help identify a molecularly defined subset within MSI-H CRC with distinct immunotherapy outcomes (26). The lineage-associated distinction in the molecular context in our study may hold certain clinical implications. Specifically, in NTRK fusion-positive CRC, the co-occurrence of MSI-H and TMB-H may support the exploration of combination strategies involving NTRK inhibitors and ICIs, whereas in NTRK fusion-positive lung cancer, the currently available evidence appears to support NTRK inhibitor monotherapy. This differential therapeutic strategy is grounded in tumor biology: MSI-H drives immunogenicity through defective DNA repair and mutation accumulation, whereas NTRK fusions primarily drive proliferation via constitutive kinase activation; their coexistence may create distinct therapeutic windows for combined or sequential treatment approaches.
The NGS-based co-alteration frequency analysis in our study further supports this interpretation, revealing both lineage-associated molecular events and MSI-linked genomic signatures, including enrichment of RNF43 and ACVR2A alterations in MSI-H tumors. The observation is supported by multiple studies, which have consistently identified RNF43 and ACVR2A as being frequently mutated in MSI-H tumors across various cancer lineages, largely due to their coding microsatellite sites that are vulnerable to frameshift mutations caused by defective mismatch repair (27). Clinically, RNF43 or ACVR2A mutations may further refine therapeutic strategies in NTRK fusion-positive tumors, supporting broader molecular profiling to guide treatment sequencing beyond TRK inhibition alone. In intestinal tumors, this pattern is consistent with NTRK fusions occurring during MSI-driven tumorigenesis, while in MSI-rare lineages, co-alterations may reflect other oncogenic programs. Recognizing these distinct biological contexts provides a framework for discussing a positive NTRK fusion result with patients, particularly as it relates to treatment options and therapeutic sequencing.
HRD-high was observed in a subset of HRD-tested cases, indicating that NTRK fusions can coexist with broader DNA repair deficiencies, while EBV positivity was rare. Another study that analyzed four gastric carcinoma subtypes (EBV-associated, dMMR, conventional, and hepatoid/enteroblastic differentiated) reported the complete absence of NTRK alterations among all 51 EBV-associated cases, further highlighting their molecular distinctiveness (28). Their study emphasized the unique biology of EBV-associated cancers, which typically exhibit distinct immune microenvironments characterized by T-cell infiltration and programmed death-ligand 1 (PD-L1) expression. Consistent with this, another study found that EBV+ pMMR gastric cancers respond better to ICI and may benefit further from combined programmed cell death protein 1 (PD-1)/cytotoxic T-lymphocyte-associated protein 4 (CTLA)-4 inhibition (18). The direct association between NTRK alterations and HRD remains unestablished, although previous work has confirmed that HRD exhibits significant variability in its prognostic impact across different cancer types (29). The co-occurrence of NTRK fusions with HRD-high may define tumors with heightened genomic instability and potentially distinct therapeutic vulnerabilities, such as sensitivity to poly ADP-ribose polymerase (PARP) inhibitors or platinum-based chemotherapy. Although HRD and EBV status were uncommon at the cohort level, when detected, they may further clarify the clinically relevant context and support reporting practices that integrate key biomarkers with the fusion result.
Overall, our findings support a combined reporting and interpretation approach: NTRK fusion status should ideally be reported concurrently with MSI and TMB (along with HRD/EBV when available) to facilitate clinically actionable and context-informed decision-making. This approach may assist clinicians in balancing TRK inhibition against immunotherapy options, guiding treatment sequencing, and refining trial stratification in routine practice. Such integrated reporting captures the full spectrum of tumor drivers and immune microenvironment status, enabling more precise prediction of differential responses to targeted therapy and immunotherapy.
There were certain limitations to this study inherent to its retrospective real-world design. First, heterogeneity in biomarker availability and the limited sample size for rare tumor subtypes may reduce the generalizability of the findings, and lineage composition and testing indication biases could have acted as residual confounders. Second, the lack of treatment exposure and outcome data precluded an assessment of how the concomitant biomarker landscape modifies therapeutic responses to TRK inhibitors or immunotherapy. The inability to adjust for these important clinical variables through multivariate modeling means that the observed molecular associations could have been influenced by unmeasured clinical confounders. Third, as our cohort comprised exclusively NTRK fusion-positive cases, we lacked a fusion-negative control group for comparative analyses, preventing identification of co-mutations significantly enriched or absent in NTRK fusion-positive tumors. Future studies incorporating matched fusion-negative cohorts are needed to address this question.
Conclusions
Functional NTRK fusion-positive tumors comprise biologically and clinically distinct subsets defined by tumor type and MSI/TMB status rather than by NTRK fusion subtype alone. We therefore recommend routine co-reporting of MSI/TMB with NTRK fusions, and HRD/EBV when clinically indicated, to inform treatment decisions, particularly distinguishing patients suitable for TRK inhibitor monotherapy from those with MSI-H who may also benefit from immunotherapy. This integrated approach supports precision treatment selection and refines trial stratification.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Review Committee of Shanghai Fourth People’s Hospital (approval No. SYLL2023031). As this was a retrospective analysis of de-identified data obtained from routine clinical practice, the requirement for informed consent was waived by the ethics committee.
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0365/rc
Funding: None.
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0365/coif). Q.Z., R.S., X.W. and M.H. are from 3D Medicines Inc., Shanghai, China. The other authors have no conflicts of interest to declare.
(English Language Editor: J. Gray)
Data Sharing Statement
Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0365/dss
References
- 1.O'Haire S, Franchini F, Kang YJ, et al. Systematic review of NTRK 1/2/3 fusion prevalence pan-cancer and across solid tumours. Sci Rep 2023;13:4116 . 10.1038/s41598-023-31055-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Drilon A. TRK inhibitors in TRK fusion-positive cancers. Ann Oncol 2019;30:viii23-30. 10.1093/annonc/mdz282 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Drilon A, Laetsch TW, Kummar S, et al. Efficacy of Larotrectinib in TRK Fusion-Positive Cancers in Adults and Children. N Engl J Med 2018;378:731-9. 10.1056/NEJMoa1714448 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Doebele RC, Drilon A, Paz-Ares L, et al. Entrectinib in patients with advanced or metastatic NTRK fusion-positive solid tumours: integrated analysis of three phase 1-2 trials. Lancet Oncol 2020;21:271-82. 10.1016/S1470-2045(19)30691-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Laetsch TW, DuBois SG, Mascarenhas L, et al. Larotrectinib for paediatric solid tumours harbouring NTRK gene fusions: phase 1 results from a multicentre, open-label, phase 1/2 study. Lancet Oncol 2018;19:705-14. 10.1016/S1470-2045(18)30119-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Desai AV, Bagchi A, Armstrong AE, et al. Efficacy and safety of entrectinib in children with extracranial solid or central nervous system (CNS) tumours harbouring NTRK or ROS1 fusions. Eur J Cancer 2025;220:115308. [DOI] [PubMed] [Google Scholar]
- 7.Marchiò C, Scaltriti M, Ladanyi M, et al. ESMO recommendations on the standard methods to detect NTRK fusions in daily practice and clinical research. Ann Oncol 2019;30:1417-27. 10.1093/annonc/mdz204 [DOI] [PubMed] [Google Scholar]
- 8.Hechtman JF, Benayed R, Hyman DM, et al. Pan-Trk Immunohistochemistry Is an Efficient and Reliable Screen for the Detection of NTRK Fusions. Am J Surg Pathol 2017;41:1547-51. 10.1097/PAS.0000000000000911 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hondelink LM, Schrader AMR, Asri Aghmuni G, et al. The sensitivity of pan-TRK immunohistochemistry in solid tumours: A meta-analysis. Eur J Cancer 2022;173:229-37. 10.1016/j.ejca.2022.06.030 [DOI] [PubMed] [Google Scholar]
- 10.Repetto M, Chiara Garassino M, Loong HH, et al. NTRK gene fusion testing and management in lung cancer. Cancer Treat Rev 2024;127:102733 . 10.1016/j.ctrv.2024.102733 [DOI] [PubMed] [Google Scholar]
- 11.U.S. Food and Drug Administration. FDA grants accelerated approval to pembrolizumab for first tissue/site agnostic indication. 2017. Available online: https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-pembrolizumab-first-tissuesite-agnostic-indication
- 12.U.S. Food and Drug Administration. FDA approves pembrolizumab for adults and children with TMB-H solid tumors. 2020. Available online: https://www.fda.gov/drugs/drug-approvals-and-databases/fda-approves-pembrolizumab-adults-and-children-tmb-h-solid-tumors
- 13.Westphalen CB, Krebs MG, Le Tourneau C, et al. Genomic context of NTRK1/2/3 fusion-positive tumours from a large real-world population. NPJ Precis Oncol 2021;5:69 . 10.1038/s41698-021-00206-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome Med 2017;9:34 . 10.1186/s13073-017-0424-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Xiao J, Li W, Huang Y, et al. A next-generation sequencing-based strategy combining microsatellite instability and tumor mutation burden for comprehensive molecular diagnosis of advanced colorectal cancer. BMC Cancer 2021;21:282 . 10.1186/s12885-021-07942-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Knijnenburg TA, Wang L, Zimmermann MT, et al. Genomic and Molecular Landscape of DNA Damage Repair Deficiency across The Cancer Genome Atlas. Cell Rep 2018;23:239-254.e6. 10.1016/j.celrep.2018.03.076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chen Y, Wang X, Du F, et al. Association between homologous recombination deficiency and outcomes with platinum and platinum-free chemotherapy in patients with triple-negative breast cancer. Cancer Biol Med 2023;20:155-68. 10.20892/j.issn.2095-3941.2022.0525 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bai Y, Xie T, Wang Z, et al. Efficacy and predictive biomarkers of immunotherapy in Epstein-Barr virus-associated gastric cancer. J Immunother Cancer 2022;10:e004080 . 10.1136/jitc-2021-004080 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhao Y, Du R, Chen M, et al. The fusion characteristics of RET fusion in pan-cancer among the Chinese population: A comprehensive genomic analysis. Transl Oncol 2025;55:102384 . 10.1016/j.tranon.2025.102384 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Marchetti A, Ferro B, Pasciuto MP, et al. NTRK gene fusions in solid tumors: agnostic relevance, prevalence and diagnostic strategies. Pathologica 2022;114:199-216. 10.32074/1591-951X-787 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Cocco E, Scaltriti M, Drilon A. NTRK fusion-positive cancers and TRK inhibitor therapy. Nat Rev Clin Oncol 2018;15:731-47. 10.1038/s41571-018-0113-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Gatalica Z, Xiu J, Swensen J, et al. Molecular characterization of cancers with NTRK gene fusions. Mod Pathol 2019;32:147-53. 10.1038/s41379-018-0118-3 [DOI] [PubMed] [Google Scholar]
- 23.Rosen EY, Goldman DA, Hechtman JF, et al. TRK Fusions Are Enriched in Cancers with Uncommon Histologies and the Absence of Canonical Driver Mutations. Clin Cancer Res 2020;26:1624-32. 10.1158/1078-0432.CCR-19-3165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Arter ZL, Lee ATM, Nagasaka M, et al. Tumor Mutation Burden Survey of AACR GENIE Database Revealed NTRK (NTRK+) and RET (RET+) Fusions Positive Colorectal Carcinoma (CRC) as Distinct Subsets. Cancer Med 2025;14:e70665 . 10.1002/cam4.70665 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Qi C, Zhou T, Bai Y, et al. China special issue on gastrointestinal tumors-NTRK fusion in a large real-world population and clinical utility of circulating tumor DNA genotyping to guide TRK inhibitor treatment. Int J Cancer 2023;153:1916-27. 10.1002/ijc.34522 [DOI] [PubMed] [Google Scholar]
- 26.Svrcek M, Cayre A, Samaille T, et al. High prevalence of NTRK fusions in sporadic dMMR/MSI mCRC RAS/RAF wild-type: an opportunity for a post-immune checkpoint inhibitors progression rescue strategy. ESMO Gastrointest Oncol 2024;5:100084 . 10.1016/j.esmogo.2024.100084 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Yang X, Lian B, Zhang N, et al. Genomic characterization and immunotherapy for microsatellite instability-high in cholangiocarcinoma. BMC Med 2024;22:42 . 10.1186/s12916-024-03257-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Deshpande V, Bal M. Enteroblastic gastric cancer subtype holds therapeutic clues. J Clin Pathol 2024;77:605-7. 10.1136/jcp-2023-209346 [DOI] [PubMed] [Google Scholar]
- 29.Dong L, Li L, Zhu L, et al. Multiomics analysis of homologous recombination deficiency across cancer types. Biomol Biomed 2024;25:71-81. 10.17305/bb.2024.10448 [DOI] [PMC free article] [PubMed] [Google Scholar]