Soft tissue and bone tumors are a very heterogenous group of neoplasms which makes prognosis difficult to assess. Moreover, given their rarity of occurrence, standardization of diagnostic criteria and precise treatment protocols is challenging. The majority of localized sarcomas are treated with excision mostly followed by adjuvant radiotherapy to control local recurrence [1, 2]. Few sarcoma types are chemo-sensitive, such as rhabdomyosarcoma, osteosarcoma and Ewing sarcoma [2, 3]. In the metastatic setting, the 5-year overall survival rates do not exceed 20%, highlighting that effective treatment of advanced disease remains a challenge. There is clear need for new therapeutic options in advanced sarcomas [4].

Some of the molecular alterations that are found in tumors are druggable and this has been a significant turning point in cancer treatment. One of the most important representatives are tyrosine kinase inhibitors (TKI) that are used for the treatment of diverse tumor types, such as imatinib for gastrointestinal stromal tumors (GIST) [5]. Unfortunately, today most of the oncogenic driver alterations remain undruggable.

Very recently, it has been proven that tumors with fusions of one of the three neurotrophic tyrosine receptor kinase genes (NTRK1, NTRK2 and NTRK3) can be treated with tyrosine receptor kinase (TRK) -inhibitors. The first generation TRK-inhibitors, larotrectinib and entrectinib [6], show excellent response rates and durable effects in tumors that harbor fusion of one of the NTRK genes, regardless of tumor type or site of origin. The adverse events are well tolerated. Still, a major obstacle for those inhibitors is the development of resistance. Therefore, new generation TRK-inhibitors are developed, such as LOXO-195 and repotrectinib [7, 8].

Despite the promising results of TRK-inhibition, NTRK fusions are rare genetic events. Among sarcomas, infantile fibrosarcomas show fusions of the NTRK3 gene in more than 90% of the cases [9]. The rates of NTRK fusions in the remaining sarcoma types are unfortunately very low. Moreover, no clear morphologic criteria are established for the recognition of the NTRK-fused sarcomas. A NTRK fusion has to be demonstrated by means of molecular analysis. Immunohistochemistry can be performed as an enrichment strategy to select tumors for subsequent molecular analysis In a previous publication, we suggested an algorithm for determining which sarcomas are more likely to carry this fusion [10].

In this study, we investigated the expression of the pan-TRK antibody in different types of soft tissue and bone tumors by immunohistochemistry and we correlated the expression of the positive cases with the presence of NTRK fusions. Furthermore, we summarized the clinicopathological characteristics, in an attempt to give more information about the identification of these tumors, in conjunction with the cases that are described in the literature.

TRK is a member of the tyrosine kinase family, predominantly known for its role in neuronal cell differentiation. There are three receptors (TRK A, B, and C) encoded by the three NTRK genes, NTRK 1, 2, and 3, respectively [9]. TRK is a transmembrane receptor, that upon its binding with a ligand, undergoes dimerization. TRK dimerization leads to activation of three signaling pathways: the PI3K/AKT, the MAPK and the PLCγ pathway, all of which impact cell proliferation, cell growth, and cell survival [14] (Figure 3). Fusion of one of the three NTRK genes leads to ligand independent constitutive activation of the TRK signaling pathway that can induce uncontrolled cell growth and proliferation, neo-angiogenesis and cell migration [15].

NTRK fusions are highly actionable driver alterations that are found across many different tumor types. TRK inhibitors are very active against tumors that harbor one of the NTRK fusions. These drugs have a high response rate and durable responses [14, 15]. Hence, detection of NTRK-fused tumors can have important therapeutic consequences for the patient.

With regard to sarcomas, Infantile Fibrosarcoma (IF) shows the ETV6-NTRK3 fusion in more than 90% of the cases [9]. No other specific sarcoma types are corelated with the presence of fusions of one of the NTRK genes. Moreover, sarcomas with an NTRK fusion show no other actionable alterations within the same tumor [16]. Thus, sarcomas with a known driver oncogenic alteration, as for instance alveolar rhabdomyosarcoma or synovial sarcoma, are less likely to harbor fusions of one of the three NTRK genes. This is in line with our results, where the two NTRK-fused tumors where not of a histologic type that carries a specific genetic alteration. Consequently, testing for NTRK fusions should be prioritized in sarcomas with high NTRK fusion frequency, while testing in sarcomas with a canonical oncogenic alteration might not have any diagnostic or therapeutic value [17].

Currently, correct identification of NTRK-fused sarcomas is only possible by means of molecular techniques. However, recently there have been attempts to describe the morphological features of tumors carrying this molecular alteration. To date, various morphological features have been described in NTRK-fused sarcomas, the most consistent being spindle cell morphology [10]. Growth patterns attributed to these tumors are summarized in our previous review[10] and include lipofibromatosis-like [18, 19], hemangiopericytoma-like [20], fibrosarcoma-like [21–23], neoplasms resembling inflammatory myofibroblastic tumors [24, 25] and dermatofibrosarcoma protuberans-like [26]. Moreover, spindle cell tumors with myxoid features [27] as well as characteristic blood vessels with perivascular thick collagen deposition [28] are also reported.

On a molecular basis, NTRK fusions are the driving oncogenic mechanisms, with fusions of the NTRK3 gene being the most frequent, followed by fusions of the NTRK1 gene, while only a few show NTRK2 fusions [10].

In our case, both tumors with NTRK fusion presented with a spindle cell morphology. None of the tumors showed a characteristic growth pattern like those previously mentioned. It is suggested that the majority of the NTRK-rearranged mesenchymal neoplasms display a combination of morphological patterns, which could be a helpful clue in tumor recognition [29]. Indeed, one of the tumors that we present also showed hyalinization of the blood vessels while the other displayed thickened collagen bundles between the tumor cells.

In many cases in literature, the tumors expressed CD34 and/or S100-protein on immunohistochemistry [18–23, 26, 28, 30]. None of the NTRK-fused cases in our series were positive for these immunohistochemical markers. However, this could be a useful diagnostic tool in cases of a spindle cell tumor without a specific growth pattern, in order to pinpoint those neoplasms that need further investigation for the presence of the fusion.

Pan-TRK immunohistochemistry is a reliable screening method for the detection of NTRK fusions with sensitivity and specificity exceeding 95% [12, 31]. Careful interpretation of the immunohistochemical staining is though recommended, since its specificity is low in sarcomas with neural or myogenic differentiation, as wild-type TRK protein is physiologically expressed in neural and smooth muscle tissue [32].

In our samples, two out of the seven immunohistochemical positive cases (nearly 28,6%) correlated with the presence of a NTRK fusion. This is in line with a very recently published study of Nozzoli et.al., where two of the eight immunohistochemical positive cases that had available RNA material for NGS, proven to harbor an NTRK fusion [33]. The ratio of sarcomas with a fusion in relation to the total population of the 70 cases examined in our series is approximately 2,8%. False-negative cases, with negative pan-TRK immunohistochemistry while the tumor was proven to harbor a NTRK fusion, are rarely described [29, 31, 34]. The staining pattern of the antibody also varies in intensity and localization. Different subcellular staining patterns have been documented such as cytoplasmic, cell membranous, nuclear with or without peri-nuclear accentuation, all of which are considered as positive [22, 31, 35, 36]. All these staining patterns suggest the presence of the fusion, and molecular testing is needed for confirmation. Subsequently, there has been an attempt to demonstrate a correlation between the staining pattern and the NTRK gene fusion. Hence, it has been shown that fusions of the NTRK1 gene mostly correlate with a diffuse cytoplasmic staining, while nuclear and rather weak pan-TRK staining is frequently mentioned with NTRK3 gene fusions [29]. Furthermore, nuclear positivity is correlated with ETV6-NTRK3 fusion in different tumor types [31, 37]. This is consistent with our findings, where one of the tumors showed pan-TRK weak nuclear positivity, and this tumor harbored an ETV6-NTRK3 rearrangement.

The prognosis of NTRK1- and NTRK3- fused sarcomas in correlation with histomorphology, has also been a subject of investigation. Sarcomas with NTRK1 gene fusions can present a low- or a high-grade histomorphology. While morphologically high-grade NTRK1-fused tumors display an aggressive course in the majority of the cases, morphologically low-grade NTRK1-fused sarcomas can have a favorable clinical course [10, 38]. On the other hand, sarcomas with fusions of the NTRK3 gene and especially those with ETV6 fusion partner, are mainly aggressive neoplasms, even those with intermediate cytological atypia [10, 39]. Low-grade cytomorphology is usually not a feature of these fusions. This is in contrast with our findings. We describe a tumor with ETV6-NTRK3 fusion with a rather indolent course. The tumor was present for almost six to seven years prior to the diagnosis and showed no signs of recurrence or metastatic spread almost a year after the excision. Morphologically, the tumor showed a cellular spindle cell proliferation without pronounced cytological atypia; mitotic activity was apparent but rather limited, while no necrosis was seen. An interesting feature was the hyalinization and presence of multinucleated cells in the blood vessel wall.

In addition, we describe a TFG-NTRK3 fused sarcoma with an aggressive course. From a histological point of view, the tumor was a high-grade neoplasm with marked cytological atypia, increased mitotic activity as well as areas of necrosis. No CD34 or S100-protein expression was observed. The patient also developed disseminated disease. These findings are again in contrast with the features of the limited TFG-NTRK3 fused mesenchymal tumors documented in the literature [18, 39]. TFG-NTRK3 sarcomas are amongst the rare NTRK3-fused sarcoma cases with a rather favorable histological and clinical picture. Opposite to our case, the tumors are known to display mostly an intermediate cytological grade or a lipofibromatosis-like morphology. Immunohistochemical positivity of the neoplastic cells with CD34 and S100-protein were documented, in contrast to our case. The prognosis of these tumors is reported favorable with no evidence of recurrence or metastasis in the (rather limited) follow up period.

Among these two tumors, the one with the best prognosis was a superficial lesion on the acral skin, while the one with the worst clinical course was located in the deep soft tissues. It therefore seems that in addition to the presence of an NTRK fusion, tumor location might also has a prognostic value.