The role of NTRK fusion and overexpression of the TRK receptor has great clinical relevance in soft tissue sarcomas as targeted therapy option is available for those patients [1–4]. Extensive investigation has been performed to identify novel entities or tumor subtypes with the involvement of the NTRK genes.

The neurotrophic tyrosine kinase receptor (NTRK) gene family consists of three genes: NTRK1, NTRK2 and NTRK3 [5]. These genes are localized on different chromosomes, respectively: 1q23.1, 9q21.33, and 15q25.3 [5]. These genes encode three tyrosine kinase receptors (TRKA, TRKB and TRKC), that play a role in the regulation of apoptosis, survival, differentiation and proliferation in several cell types through well-known downstream signaling pathways, including PLCγ/PKC, MAPK/ERK and PI3K/AKT [5]. The tyrosine kinase receptors consist of extracellular, ligand-binding domain, transmembrane domain and intracellular, tyrosine kinase domain [5, 6]. The most important gene alterations affecting the NTRK gene family are fusions [6, 7]. NTRK fusions are characteristic in some rare tumors, like infantile fibrosarcoma [8], and congenital mesoblastic nephroma [9], but can occur in common malignancies as well, like lung cancer [10], colorectal cancer [11], melanoma [12], though less frequently. However, the fusions are the most investigated and best known, other gene alterations can occur in the NTRK genes. Recently, oncogenic point mutations within the extracellular, transmembrane and tyrosine kinase domains as well were identified in haematological malignancies [13, 14]. Furthermore, a deletion in the extracellular domain of TRKA showed oncogene capability in acute myeloid leukemia [15]. Also, a TRKA splice variant promoted tumorigenic cell behavior [16]. Although, NTRK1 amplification occurs in 8% of breast cancers [17], and several non-fusion NTRK alterations have been described in 14% of adult and pediatric tumors as well, their role in oncogenesis is yet to be explored [18, 19]. The NTRK gene family is at the center of attention nowadays, due to the highly effective, selective TRK inhibitors [20]. The US Food and Drug Administration approved the usage of specific TRK inhibitors, larotrectinib [1] and entrectinib [2], in case of any malignancy with NTRK fusion, regardless of the exact entity. Due to acquired resistance mutations within the tyrosine kinase domain, second-generation TRK inhibitors, repotrectinib [3] and selitrectinib [4], were developed.

Pan-TRK immunohistochemistry, which indicates the presence of TRKA and/or TRKB and/or TRKC protein(s), is a valuable tool to identify tumors without NTRK gene rearrangements, as it has a high negative predictive value [21]. Pan-TRK immunopositivity can be detected in some soft tissue sarcomas showing myogenic phenotype or myogenic differentiation, without NTRK fusion [22, 23].

In our previous work we investigated 131 dedifferentiated liposarcoma (DDLPS) cases and found 75 pan-TRK immunohistochemically positive cases [24]. Among the pan-TRK positive cases we found 6 cases with recurrent NTRK1 c.1810C>T (p.H604Y) and c.1838G>T (p.G613V) tyrosine kinase domain mutation pair and interestingly 3 of this 6 cases showed NTRK1 amplification. This surprisingly high frequency prompted us to extend our investigation and to estimate the real amplification frequency in this 75 cases. One additional pan-TRK positive case was included, so the total number of investigated cases were 76.

Our hypothesis was to investigate the relation between pan-TRK immunopositivity and possible genetic alterations by using NTRK1/2/3 locus specific break apart probe set in order to estimate the frequency of genomic rearrangement and/or regional or overall copy number alterations related to any of these three regions.

We analyzed 76 individual cases of dedifferentiated liposarcoma with myogenic heterologous differentiation and pan-TRK immunopositivity from our institutional and consultation archives (diagnosed between 2014 and 2024) in this study. The diagnosis was based on clinical features, characteristic morphology and immunohistochemical and/or molecular examination of MDM2 and/or CDK4 overexpression or amplification. Previously executed immunohistochemistry with desmin and alpha smooth muscle actin antibody were used to show myogenic heterologous differentiation, and pan-TRK antibody was used to detect NTRK1-2-3 overexpression. This study was approved by the Semmelweis University Regional and Institutional Committee of Science and Research Ethics (TUKEB 155/2012).

Immunohistochemistry was performed on deparaffinized, rehydrated whole slide formalin-fixed, paraffin-embedded (FFPE) block sections with the Bond-Max automated staining system (Leica Biosystems, Wetzlar, Germany), after applying antibody-specific epitope retrieval techniques, using the following antibody: pan-TRK (ready-to-use, EPR17341, Ventana). The intensity and extent of immunoreactivity were evaluated by two soft tissue pathologists (ZS and KD). Tumor samples were considered positive with pan-TRK antibody if >1% of tumor cells exhibited immunostaining at any intensity above the background. Staining intensities of tumor cells were graded as no staining (0), very weak (1), weak (2), moderate (3), strong (4) or very strong (5) and numbers of stained tumor cells as the extent of staining were graded as 0% (0), 1%–19% (1), 20%–39% (2) 40%–59% (3), 60%–79% (4), 80%–100% (5).

FISH was performed on interphase nuclei from 3 µm sections of FFPE blocks using the ZytoLight SPEC Dual Color Break Apart Probes of the NTRK1/2/3 genes (ZytoVision, Bremerhaven, Germany). The slides were analyzed by assessing a minimum of 100 cells each in a tumor region of interest. Rearrangement of the NTRK genes was defined as a minimum of 20% tumor cells with split signals (classical pattern) of the corresponding NTRK probes. Polysomy or copy number gain was defined by an average copy number of all NTRK (1/2/3) signals ≥3. Amplification was defined if a NTRK gene showed at least twice in number of signals compared to the average of the other two NTRK gene type. In other words, the different NTRKs (locating on different chromosomes) served as internal controls to assess whether there is a polysomy or either of them is amplified. Because we used break-apart NTRK probes, we could distinguish “whole” amplification (the whole gene was affected) or “partial” amplification (either the 3′ end/red signal or the 5′ end/green signal was amplified). FISH signals were scored by one pathologist (ZS) and one biologist (GP).

Targeted DNA and RNA libraries were prepared according to the TruSight Tumor 170 reference guide (Illumina, San Diego, CA, United States). The TruSight Tumor 170 kit included the analysis of 151 genes for insertion, deletion, single nucleotide variant, copy number variation and 55 genes for fusions and splice variants and 59 genes for amplification. Genomic DNA and total RNA were isolated from formalin-fixed, paraffin-embedded specimens using the QIAamp^® DNA FFPE Tissue Kit (Qiagen) and the High Pure FFPET RNA Isolation Kit (Roche Diagnostics, Mannheim, Germany). Estimations of tumor cell percentages of the samples were performed by histopathological examinations prior to the isolation processes. DNA and RNA concentrations were measured using the Qubit™ dsDNA HS Assay and Qubit™ RNA HS Assay Kits (Thermo Fisher Scientific) on a Qubit™ 4 Fluorometer. 120 ng DNA in 52 μL volume was sheared (200 cycles, peak power: 75 W, duty factor: 10, treatment time: 510 s, at 7°C setpoint) using a Covaris M220 Focused-ultrasonicator (Covaris, Woburn, United States). 80 ng of DNase digested RNA was used for library preparation of each sample The size of double-stranded DNA fragments and RNA molecules was confirmed after shearing using a Tapestation 4,200 (Agilent, Cheshire, UK). Library preparation workflow of Illumina TruSight Tumor 170 assay was performed according to the manufacturer’s protocol. The final libraries were paired-end sequenced at a 2 × 101 bp read length, using Illumina NextSeq 1000/2000 P3 (200 cycles) reagent kits on a NextSeq 2000 platform. Bioinformatic analysis was performed using TruSight Tumor 170 v2.0.2 Local App (Illumina, San Diego, CA, United States). For further clinical interpretation, we used Genoox’s Franklin web-based analysis tool, which applies variant filtering and further annotations: pathogenicity scores, population-frequency, protein structure predictions, relevant clinical guidelines and therapeutic options for variants.

Publicly available data sets from TCGA and cBioPortal were analysed for the presence of structural variants involving NTRK1/2/3 gene loci. From three studies analyzing soft tissue sarcomas, well-differentiated and dedifferentiated liposarcoma cases were further analyzed to interrogate the frequencies of NTRK1/2/3 genes and MDM2 and CDK4 genes. From the 2551 cases presented in these three studies 265 unique well-differentiated or dedifferentiated liposarcoma samples were included for the analysis using cBioPortal [25–29].

NTRK gene family statuses by FISH (normal, polyploidy and amplification) were investigated based on pan-TRK immunohistochemistry (extent and intensity values). In order to compare three samples, one can use ANOVA test; however, in this case, there are two assumptions to be justified: 1. Normality of residuals (the errors used for the estimation of the error terms are normally distributed), 2. Homogeneity of variance (the level of variance for a particular variable is constant across the sample). For testing normality, we used one-sample Kolmogorov-Smirnov normal test; and Levene Statistic was used to test homogeneity of variances. Finally, to compare normal, polysomic and amplification groups, parametric one-way ANOVA and non-parametric Kruskal-Wallis tests were run.

The statistical analysis showed that in all groups (normal, polysomy and amplification) the assumption of normality of the residues is not met, neither in terms of extent nor in terms of intensity. The homogeneity of variance was found to be fulfilled in terms of the extent. Since simulation studies using a variety of non-normal distributions have shown that the false positive rate is not affected very much by the violation of the normality assumption [30, 31], we used one-way ANOVA to investigate extent. It resulted in p = 0.002, which means that the groups are significantly different, moreover post hoc test showed the following p values: p = 0.075 for normal and polysomy groups, p = 0.684 for polysomy and amplification groups, and p = 0.02 for normal and amplification groups. The conclusion is that normal and amplification groups are significantly different in term of extent. In term of intensity, the homogeneity of variance was found to not be fulfilled, thus instead of ANOVA, non-parametric Kruskal-Wallis test was used. It resulted in p = 0.116 meaning that we should retain the null hypothesis, namely that there is no significant difference between the three groups in term of intensity, however, the tendency for increased intensity from normal to amplification group can be established (Table 2).