SFTs are rare neoplasms that usually involve the pleura or other serosal surfaces. Its origin has been controversial. Previous experimental reports showed that submesothelial cells had the ability to differentiate into mesothelial cells [34, 35]. Despite they are frequently regarded as pleural tumors, a variety of extrathoracic locations has been reported like the meninges and the soft tissues of the neck [8–10], pelvis [11, 16, 17] and retroperitoneum [14, 15], liver [12, 13], and the soft tissue [18–33]. Soft tissue SFTs still represent a rare entity with 46 cases described in the English literature, mainly from a histopathological point of view. They principally affect adults between the fourth and seventh decades of life [16, 17]. Clinically, extrathoracic SFTs cause symptoms that relate to tumor’s size and location. Systemic symptoms such as hypoglycemia, arthralgia, osteoarthritis, and clubbing have also been reported. These symptoms usually subside upon tumor’s resection . In our cases, apart from focal swelling and mild to marked displacement of neighboring anatomic structures (e.g. muscles, neurovascular bundles), no other regional or systemic symptoms occurred and laboratory analysis ranged within normal limits.
Radiological findings concerned twenty seven well-defined masses with round or ovoid shapes displacing adjacent anatomic structures. On unenhanced CT images, pleural SFTs appeared isodense relative to muscle and some contained hypodense areas, while on enhanced CT images they usually exhibited mild to marked heterogeneous contrast uptake [35, 36]. According to the Rosado-de-Christenson et al. study , isodensity on pre contrast CT scans correlates with hypercellular areas and capillary networks while markedly enhanced regions represent hypervascular areas, intermediate enhanced areas correlate with hypocellularity, and patchy hypodensities correspond to necrotic, myxoid or cystic changes. Of our eighteen lesions imaged with CT scans, two were isodense, five appeared relatively hypodense compared to neighboring muscles and eleven were mixed of isodense and hypodense areas. On enhanced images thirteen lesions were markedly and heterogeneously enhanced including patchy unenhanced areas (Figures 3B, 3C, 4C) while only four exhibited a more homogeneous enhancement pattern (Figures 1B, 2C). In reference to the dynamic characteristics, Fuksbrumer et al.  and Moser et al.  stressed the importance of marked and delayed enhancement and washout in characterizing liver SFTs. In our study, one case (Figure 1B) that was assessed with dynamic CT protocols, presented with mild arterial enhancement and more intense and prolonged enhancement on late venous and delayed phases. These features related histologically to cellular areas (prolonged enhancement) admixed with collagenous stroma (delayed enhancement).
According to the published MR features of SFTs, isointensity to muscle is most commonly displayed on T1-weighted sequences along with an increase of signal intensity on post gadolinium images and variable appearance on T2-weighted sequences. It is considered that the variety of signal intensity on T2-weighted images depends upon the different nature of the tumor components, namely, the amount of collagen and cellularity (fibroblasts), and on the presence of degeneration [9, 34]. This series included seven masses of equal signal intensity on T1-weighted images relative to muscle (Figure 5A) and five lesions of heterogeneous-mixed intensity (Figure 3B, 3C). In most MR studies on mature fibrous tissue, the observed hypointensity on both T1 and T2-weighted sequences is related to the presence of hypocellularity and abundant collagenous stroma whereas fibrous tissue of relative increased cellularity and vascularity increases the signal intensity on T2-weighted sequences (Figure 5B) [34, 37, 38]. The aforementioned was noticed in all nine lesions that showed mixed to high signal intensities on T2-weighted images. Gadolinium enhancement of SFTs is generally a result of tumor vascularization that may vary from hypovascularity to hypervascularity. Of the ten cases that received gadolinium, six revealed intense heterogeneous enhancement and the remaining four cases were more homogeneously enhanced (Figure 5C). Similarly to the Tateishi et al. study , the enhancing portions corresponded histologically to areas of abundant cellularity combined with microvessels. Patchy heterogeneous areas of variable size were depicted in totally ten MRI exams (Figures 3F, 3G). Four of these cases demonstrated areas of low to intermediate and markedly high signal intensity on T2WSE and the remaining six showed areas of mild to no uptake on post contrast images and were correlated to hypocellular collagenous stroma (n=10) with co-existing necrotic (n=3) or degenerative (n=2) changes. Hypocellular collagenous stroma was also correlated with the patchy unenhanced areas of the remaining nine (n=9) contrast CT cases with co-existing necrotic changes in only one case (n=1). On histologic analysis there were additionally two other cases with necrotic changes; one of which was imaged solely with MRI and appeared homogeneous both on pre and post gadolinium images and the other was imaged only with U/S and contained a few scattered hypoechoic areas. Moreover, thin hypointense on T2WSE, linear or curvilinear structures were observed in eight out of fourteen MRI cases. These structures remained unenhanced and were attributed to smooth hypocellular fibrous septations. Calcifications are suggestive of necrosis but they are only sporadically reported in extrathoracic SFTs . Likewise in this study, it was depicted in only six cases (22,2%).
The sonographic appearance of extremity SFTs is not pathognomonic and in this series, generally comprised of a heterogeneous hypoechoic, well defined mass, with or without calcifications deeply located in the soft tissues. Similarly, X-rays findings were not specific and mainly include that of a radio-dense soft tissue mass with ill defined margins and occasionally gross calcifications. In one lesion, DSA revealed marked vascularity of the lesion originating from the superficial and deep femoral vessels.
Histologically SFTs are well circumscribed and consist of a varying number of spindle cells randomly arranged in a collagenous background of variable vascularity. Some cases present with a hemangiopericytoma-like pattern of irregular branching vessels and occasionally storiform or herringbone patterns (Figure 4E) are seen . Immunohistochemistry is essential in differentiating these tumors from other spindle-cells neoplasms . Positivity for CD34, CD99 and bcl-2 is an indicator of SFTs [3, 16]. Pathological criteria for aggressiveness include tumor size >5 cm, high cellularity, nuclear pleomorphism, high mitotic rate (more than 4 mitosis in 10 HPF) and necrosis . In the present study 55,5% of cases (15/27) contained aggressive features; one tumor (max diameter: 5,2 cm) locally recurred at 66 mo postoperatively, one case (max diameter: 15 cm) presented with concomitant lung metastases and two others (max diameters: 7 cm and 8,6 cm) metastasized postoperatively (to bones, lungs and soft tissues) at 46 mo and 85 mo respectively. The first lesion that locally recurred at 66 mo follow-up was characterized by a homogeneous enhancing pattern on CT and MRI images and showed no evidence of aggressiveness while the lesion with concomitant metastases showed aggressive features on histology and hypodense-unenhanced areas on post contrast CT. The remaining two cases that metastasized postoperatively were characterized as aggressive on histological grounds while one of them was homogeneously enhanced and the other was heterogeneously enhanced on CT scans. As a result, we can not predict the biological behavior of soft tissue SFTs based solely either on histopathologic or imaging evaluation.
The behavior of extrathoracic SFTs is unpredictable. Some histologically aggressive tumors behave in a benign fashion while some morphologically histologically benign lesions behave aggressively [16, 32]. Ten to 13% of cases recur and/or metastasize [22, 24]. They may also recur or metastasize after complete surgical resection even in the absence of atypical histological features. Therefore, complete surgical resection and long-term follow-up is recommended for patients with extrathoracic SFTs . Moreover, Cranshaw et al.  showed that extrathoracic SFTs have a higher rate of malignant behavior than that classically described and recommended that in the presence of atypical features extrathoracic SFTs should be managed and followed up in the same manner as other high-grade soft tissue tumors, which is not what we found in our cases.
In conclusion, imaging features of extremity SFTs lack specificity and correlate with the variable histopathological appearances. However, soft tissue SFTs often (16/27, 59,2%) are isodense and isointense masses mixed with hypodense and hypointense areas on unenhanced CT and MR T1W. Depending upon tumor’s cellularity, vascularity, collagen distribution and/or degeneration, heterogeneous enhancement and areas of mixed to high signal intensity on MRT2W are observed. Although rare, a high index of suspicion of a soft tissue SFT should be raised when a deep heterogeneous-mixed soft tissue mass with marked enhancement is encountered.