Scope of FNAC in the diagnosis of soft tissue tumors-A study from a tertiary cancer referral center in India

Background Fine needle aspiration cytology (FNAC) forms one of the first diagnostic tools in the evaluation of tumors. Its role in diagnosing soft tissue tumors (STT) has been fairly documented, as well as debated. Present study was aimed at evaluating its scope in diagnosing 127 cases of soft tissue tumors. Methods Conventional Pap and MGG staining was available in all the cases. Immunocytochemistry (ICC) was performed in 15 cases. Histopathological details were available in 115 cases. Results 50% cases were referred for a primary diagnosis, while 26.8% & 22.8% cases were evaluated for recurrent and metastatic lesions, respectively. Extremities were the commonest sites. On FNAC, 101 cases (79.5%) were labeled as malignant, whereas 10 cases (7.9%) were labeled as benign. The remaining 16 cases (11%) were not categorized and were labeled as 'unsure/not specified'. Histopathological confirmation in 115 cases, gave a diagnostic accuracy of 98%, with a positive predictive value of 98% in malignant cases and a negative predictive value of 100% in benign cases. Two cases were false positive. Among the various cytological categories, 60 cases (47.2%) were of spindle cell type, followed by 32 (25.2%) of round cell type and 14 cases (11%) of lipomatous type. Other 12 cases (9.4%) were of pleomorphic type; 7 (5.5%) cases of epithelioid type and remaining 2 cases were of myxoid type. All the round cell, pleomorphic and myxoid type of tumors were sarcomas, whereas 73.3% cases of spindle cell type were labeled as 'malignant'. Exact cytological sub typing was offered in 58 cases, with rhabdomyosarcoma (RMS) as the most frequently sub typed tumor. The two false positive malignant cases were of fibromatosis and a pigmented schwannoma, on biopsy. Out of 28 metastatic lesions, lymph nodes were the commonest site for metastasis, with epithelioid tumors that formed highest percentage of metastatic cases. Conclusion FNAC is fairly specific and sensitive in STT diagnoses for primary, recurrent and metastatic lesions. The cytological types, especially round cell and pleomorphic sarcomas, can be quickly identified. Clinicopathological correlation with ICC as an adjunct, are valuable in exact sub typing.


Background
Soft tissue tumors (STT) are rare neoplasms. At our hospital, 2.5% of the cancer cases are constituted by soft tissue sarcomas. Most of these cases are referrals, including a larger proportion of malignant cases, invariably with a high grade and stage. [1] These tumors pose a significant diagnostic challenge as a result of their morphologic overlap and biological heterogeneity [2]. Among its various diagnostic aids, lately, FNAC has been gaining importance as a result of its cheap cost, easy performance, safety, along with fair specificity and specificity, especially in terms of sorting out malignant cases. Apart from its use in recurrent and metastatic cases, FNAC has been identified as a useful diagnostic technique in the initial diagnosis of STTs [3][4][5][6][7][8].
The present study was aimed at evaluating 127 cases of FNAC smears from soft tissue tumors, diagnosed over a period of 5 years. The other objectives included identification of cyto-histological concordance in terms of malignancy; stratification of the tumors into various cytomorphological categories and to note the extent of specific sub-typing in individual cases.

Methods
127 aspirates (unaided and guided) from STTs were evaluated over a period of 5 years (2001-2006). Only cases with relevant clinical data were included. All cases included conventional Papanicolaou (Pap) and May Grunwald Giemsa (MGG) stained smears. Inadequate smears were the ones with 'no' or scanty cells. Non-neoplastic lesions, on biopsy, were excluded. Histological details, in 115 cases, were accessed from the clinical charts and the Hospital Diagnostic Information System (DIS).
On examination, all cases were placed into 6 cytomorphological categories, namely spindle cell, round cell, pleomorphic, myxoid, epithelioid/polygonal cell and lipomatous type [9]. In cases of mixed components, the specific subtype was assigned based on the predominant morphological pattern. Immunocytochemistry was performed in 15 cases for an exact sub typing, on smears and cell blocks. Exact sub typing was offered in the clinical context, with ICC as an adjunct.

Immunocytochemistry (ICC) technique on smears
Smears were fixed in methanol, followed by 3 changes in 100% absolute alcohol for 5 min each. This was followed by washing under running tap water for 5 min. Next, the smears were subjected to endogenous peroxidase for 30 min (H 2 O 2 + methanol). This was followed by washing under running tap water, antigen retrieval (low heat with microwave or enzyme) and 3 changes of wash with Tris Buffer solution (TBS) ~5 min. Subsequently, blocking serum was applied, followed by incubation with primary antibody ~1 hour and TBS washing. Further, secondary antibody was applied for 30 min, followed by washing with TBS and application of avidin-biotin complex (ABC). Afterwards, TBS washing was carried out with subsequent staining with Diamino Benzidine (DAB) reagent. Harris' hematoxylin was used as a counter stain. Finally, the smears were dehydrated with absolute alcohol, cleared with xylene and mounted with DPX. The cases were accompanied with appropriate controls. A limited panel of markers was carried out on the smears, including vimentin, desmin, S-100, cytokeratin (CK), MIC-2, myogenin, neuron specific enolase (NSE), synaptophysin, chromogranin and myeloperoxidase (MPO). Immunohistochemistry (IHC) was performed by immunoperoxidase procedure. On biopsy a wide panel of markers was carried out for exact subtyping. In addition, smooth muscle actin (SMA), Myo D-1, BCL-2, epithelial membrane antigen (EMA), Leukocyte common antigen (LCA), ALK-1, CD30, NSE, synaptophysin and chromogranin (Dako, Produkionsveg, Glostrup, Denmark) constituted as other markers.

Statistical analysis
Statistical analysis was carried out using SPSS (version 14) software. Descriptive analysis was performed using frequency and percentages. In terms of malignancy, cytology was compared with histopathological findings by calculating specificity, sensitivity, and concordance along with positive and negative predictive values (PV). The significance of various cytological categories in terms of benign, malignant and 'unsure'/not specified' was analyzed using chi-square test. P value < 0.05 was considered significant.

Results
Out of 127 cases, maximum cases (27,21.3%) were noted in the age group 21-30 years. Lower extremities were the common sites, with thigh as the most common site of occurrence in 20 cases (15.7% Among the various cytomorphological categories, 14 cases (11%) were of lipomatous tumors, 12 cases (9.4%) were of pleomorphic type, 32 cases (25.2%) were of round cell type, 60 (47.2%) of spindle cell type, 7 cases (5.5%) of epithelioid type and the remaining 2 cases were of myxoid type. All the round cell, epithelioid, pleomorphic and myxoid tumors were sarcomas. Out of 60 cases of spindle cell subtype, 44 (73.3%) were malignant and 16 cases (26.6%) were not categorized, either as benign or malignant.
Out of 16 'unsure'/'non classified' cases; on biopsy, 8 cases turned out to be malignant, namely 3 spindle cell sarcomas, 2 MPNSTs, 1 synovial sarcoma, 1 gastrointestinal stromal tumor (GIST) and 1 pleomorphic sarcoma (NOS). The remaining 8 cases included 2 of intermediate malignancy, namely, 1 case, each of a dermatofibrosarcoma protuberans (DFSP) and fibromatosis. Another 1 case was of proliferative fasciits. Another 3 cases included 2 of schwannoma and 1 case of a neurofibroma. Remaining 2 cases could not be categorized due to lack of biopsy results With the help of clinical details and ICC, exact sub typing was possible in 58 cases. Out of 14 lipomatous tumors, commonest were lipomas (10 cases), including 1 case of a pleomorphic lipoma; 2 cases were of myxoid liposarcomas and remaining 2 cases were of pleomorphic liposarcomas. Out of 60 cases of spindle cell category, specific diagnoses in 14 cases included leiomyosarcoma (3 cases), MPNST (2 cases), synovial sarcoma (3 cases), melanoma of soft parts (1 case). One case was offered a differential diagnosis of MPNST vs. melanoma of soft parts, another of a MPNST vs. synovial sarcoma and remaining 1 case was of sarcoma vs. a sarcomatoid carcinoma (poorly differentiated). In the round cell category, exact sub typing was more possible i.e. in 20 cases that included RMS (10 cases), PNET/Ewing's (9 cases) along with 1 case of a neuroblastoma. In the pleomorphic group, 8 cases were labeled as pleomorphic cell sarcoma, 2 as pleomorphic rhabdomyosarcomas and 1 case as an extraskeletal oste-osarcoma. One case was offered a differential of a pleomorphic sarcoma vs. poorly differentiated carcinoma, however, lacked biopsy results.
Out of 7 cases of tumors with epithelioid morphology, all were malignant. The cytological diagnoses included 1 case of an epithelioid sarcoma and 1 case of an epithelioid MPNST. Remaining 5 cases were labeled as sarcomas with an epithelioid morphology. On histopathology, 1 of these turned out as a malignant granular cell tumor; 1 epithelioid leiomyosarcoma; 1 Alk1 positive anaplastic large cell lymphoma (ALCL) and another 1 as a pleomorphic sarcoma (NOS). One case turned out to be a myxoid liposarcoma with round/polygonal cells, the latter that were seen in the smears. In the myxoid category, both the tumors were designated as myxofibrosarcomas. (Figures 1,2,3,4). While one of these was in the thigh of a 60 years old male, the other case that turned out as a pleomorphic sarcoma, myxoinflammatory type on biopsy, was identified in the leg of a 52 years old male. (Table No.

Discussion
Soft tissue tumors (STTs) have been diagnosed with the 'time-honored' histopathology that is recognized as the 'gold standard' for their evaluation. However, in the current era, where 'needle is preceding the scalpel' and the biopsy material is getting limited, it would be prudent to discuss the role and scope of FNAC in diagnosing STTs. The role of FNAC has mostly been noted in soughting recurrent and metastatic STT cases [6][7][8]. Nevertheless, there are studies indicating its role in primary diagnosis of these tumors [3][4][5]. The present study was carried out to highlight various aspects of cytological diagnosis of STTs, in a series of 127 cases, wherein maximum cases (50%) were submitted exclusively for a primary diagnosis. An increased number of overall malignant (84.25%) vs. benign cases (12.5%) in our series were comparable to a study by Miralles et al [10] and contrasting to another other study [11]. The reason was the predominance of referral cases, including a greater proportion of sarcomas that our center receives. Moreover, non-neoplastic lesions were excluded in this study. In terms of diagnosis, complex heterogeneity of STTs is known to be a limiting factor in their exact categorization. With the advent of ancillary techniques like IHC, flow cytometry, cytogenetics and molecular techniques, the objectivity of diagnosing a STT has been considerably enhanced. In a study, Kilpatrick et al [4] have presented the extent to which cytology can be utilized in effective diagnosis of STTs. Subsequently, there have been studies highlighting application of the ancillary techniques in exact sub typing of these tumors [7,8,12,13]. Apart from providing material for these studies, FNAC is easy, safe and cost effective. While the fear of tumor seedling is hardly noted, its effectivity is best exemplified with more access to mass lesions in form of multiple aspirates [4]. In our series, none of the cases had complications or tumor seedlings.

A. Malignant peripheral nerve sheath tumor
In terms of diagnostic efficacy, 100% sensitivity and 87% specificity in our cases was comparable to results of Nagira et al [11], wherein the respective values were 92% and 97% respectively. In another series, Wakely et al [14] reported 100% sensitivity and 97% specificity in STT diagnosis with FNAC. Layfield et al [3] achieved 95% sensitivity and specificity while dealing with these lesions. In our study, 2 cases (1.57%) were false positive (FP) and none was false negative (FN). While a study on 517 STT aspirates by Akerman et al [15] revealed a 2.9% false positive rate, the subsequent studies by Wakely et al [14] and by Kilpatrick et al [4] yielded a single case of false negativity and nil false positivity. This was in contrast to a study by Nagira et al [11], who identified higher figures for false positivity and false negativity. Our results were comparable to the documented range of < 1%-5% (FP) and 2-15% (FN) [3,4,11,16]. Nil FN in our series was due to the fact that 16 cases were diagnosed as 'unsure'/not specified. Out of these 8 turned out to be malignant, 1 of intermediate malignancy and 5 benign, on biopsy. An overall concordance of 98% is comparable to results of Shah et al [17].
A basic cytological approach towards making a STT diagnosis begins with the familiarity with normal structures, along with myxoid or metachromatic stromal fragments and a variety of dyscohesive cells like spindly, round, pleomorphic, polygonal that are indicators of a STT, on aspirates [9,18]. These features are integrated with clinico radiological findings. The value of radiological findings has been fairly discussed [11,19]. Infact, improved localization of the ST lesion with radiological techniques has revolutionized the amount of aspirate one can achieve, especially in deep-seated lesions. An adequate aspirate is analyzed and non-neoplastic lesions are sorted out. This is followed by placing neoplastic lesions in benign, borderline/intermediate and malignant groups. This approach was followed in the present study, with the exclusion of secondary soft tissue deposits of carcinomas. The sarcomas were not graded similar to a study by Nagira et al [11], even though attempts at cytologic grading of sarcomas have been made [20]. In their study, Mathur et al [20] identified a low concordance of cytological grading in cases of grade 1 vs. high-grade sarcomas. Our series focused upon the value of cytomorphological sub typing into the six categories, as noted earlier [9,11]. While all 32 cases of round cell category; 7 of epithelioid type; 12 of pleomorphic type and 2 of myxoid type were placed into malignant groups, 44 of the total 60 cases (73.3%) of spindle cell type could be labeled as sarcomas. Of the remaining 16 cases, 8 turned out to be malignant; 2 of borderline malignancy and 4 as benign. Remaining 2 cases could not be ascertained, in view of lack of biopsy results. The relationship between cytological categorization with benignancy or malignancy of the subtypes was statistically significant. Among the lipomatous group, more cases (10/14) were benign than malignant (4/14). (p < 0.05). While the results for round cell, pleomorphic and lipomatous tumors were comparable to the findings of Nagira et al [11]; we observed more cases of spindle and all of polygonal and myxoid types, as malignant. The reasons include referral of mostly malignant cases to our centre.
Tumors with epithelioid/polygonal shapes, exactly catego-rized on biopsy  In terms of exact sub typing, which was offered in 58 cases, all lipomatous and 70% pleomorphic tumors were suitably categorized. One case of a pleomorphic lipoma showed presence of mature fibroadipose fragments with 'floret-like' giant cells, displaying nuclear molding. Presence of this tumor in a superficial location i.e. neck, in an elderly male, was a helpful clue in ruling out a liposarcoma [21]. 70% of pleomorphic sarcomas were catego-rized in congruence with histopathology. Four out of 10 cases could not be categorized, even on histology with IHC. The reason is the referral of poorly differentiated sarcomas to our center. In their study, Berardo et al [22] concluded lack of any cytological features that could differentiate between malignant fibrous histiocytomas with other pleomorphic sarcomas.    [23]. One case of soft tissue neuroblastomas showed neuropils and false rosettes. This was objectively confirmed with NSE positivity and MIC2 negativity. Neuroblastomas were lesser in number in our study, as these tumors are well diagnosed with imaging findings, biochemical tests and biopsy, at our Center. FNAC was carried out in few cases.
Among spindle cell tumors, specific diagnoses were assigned in few cases viz. leiomyosarcoma, MPNST, synovial sarcoma in 3 cases and melanoma in 1 case. While the former 2 were substantiated with clinical details, including location, synovial sarcoma was diagnosed based on specific cytomorphological features. Diagnosis of synovial sarcoma in 3 cases included 2 metastatic cases and one recurrent case. Cytomorphology of synovial sarcoma on cytology is complex. In unusual situations, ICC and t (x: 18) translocations (SYT-SSX) can be useful in forming this exact subtype [13,24]. One case of melanoma of soft parts was ascertained with prominently nucleolated spindly cells and few polygonal cells that exhibited diffuse S-100 positivity. This was confirmed with additional HMB45 positivity, on IHC [25]. On biopsy, level of exact subyping was significantly increased in spindle cell tumors. Another case of an endometrial stromal sarcoma was presumed to be a leiomyosarcoma, on FNAC. Retrospectively, presence of hypercellular smears with ovoid cells and prominent basement membrane-like material reminiscent of proliferating vessels were identified as useful features. Recently, a similar case has been reported [26]. Out of 16 cases that were not classified, 4 turned out benign on biopsy. The remaining 8 cases included 2 of intermediate malignancy, namely, 1 case, each of a dermatofibrosarcoma protuberans (DFSP) and fibromatosis. Another 1 case was of proliferative fasciits, on biopsy. In case of adequate material and typical clinical settings, diagnosis of a DFSP can be suggested [27]. The other daunting tumor is fibromatosis that is recognized as a pitfall in STT diagnosis [3,5,13]. This was identified as 1 false positive case in our study, along with another case of MPNST that turned out to be a pigmented schwannoma, on histopathology.
While 2 myxoid tumors were termed as myxofibrosarcomas, one of these turned out to be a pleomorphic sarcoma, myxoinflammatory type. Lack of exact categorization of sarcomas with epithelioid morphology (5 cases) was due to limited cellularity and tumor heterogeneity. One case labeled as sarcoma with epithelioid morphology turned out to be a myxoid liposarcoma, with a round/polygonal cell component. The remaining 2 cases in this category were unusual. While one case turned out to be a malignant granular cell tumor, the other was an Alk-1 positive ALCL. Granular cell tumors have been recognized as a pitfall in FNAC of breast lesions [28]. The diagnosis was ascertained with S100 positivity, on biopsy [29]. The case of an Alk1 + ALCL was noted in a young adolescent male, with multifocal lesions Smears showed dyscohesive polygonal cells with multilobated nuclei and moderate amount of finely vacuolated cytoplasm [30]. Positivity for LCA, CD30, EMA and Alk-1 confirmed this diagnosis, on biopsy. The morphological overlapping was While Costa et al [31] achieved 20.9% rates in terms of exact sub typing of sarcomas, we, like Nagira et al [11], observed an overall higher rate. In their review, Kilpatrick et al [4] have presented a range of 21-74% of exact sub typing of sarcomas, based on various studies. Exact sub typing can be increased with application of ICC [13]. In the present study, routine ICC was attempted in limited cases; mostly of the round cell type, as exact sub typing in these cases has a definite bearing on management. Maximum cases that lacked exact sub typing were of the spindle cell type. Thirteen cases were diagnosed as spindle cell sarcomas, even on histology. The reasons in many of these included referral cases with lack of paraffin blocks and negativity of a specific lineage, even with IHC, in the remaining few. Exact sub typing of spindle cell tumors is a limiting factor, even in other studies [9]. Prognostication of soft tissue sarcomas has been recommended with histological grading and staging [32]. Recent studies [33,34] on cytomorphology of leiomyosarcomas and neurogenic tumors constitute an 'add-on' to refining the existing cytomorphology of these subtypes.
Apart from its primary diagnosis; the role of FNAC is recognized in metastatic STT evaluation [7,8]. Evaluation of 'foreign cells' in metastatic locations, in cases with known clinical context, is less challenging than in situations initial metastatic presentations. In such cases, clinical details are imperative before ascertaining a diagnosis. In our series, out of 28 cases with metastatic lesions, maximum metastatic deposits were reported in lymph nodes (55.1%), 'more-so' of the groin, followed by liver and lungs. While lymph nodes have been observed as the commonest metastatic location in earlier studies [7,8], in our study, groin nodes, were found to be more commonly involved. This might be as a result of more STS occurring in extremities. While maximum metastatic cases were of spindle cell type (considering their higher absolute number), the percentage was highest with epithelioid tumors. This is because of tendency of sarcomas, with epithelioid morphology, to have increased chances for metastasis, like carcinomas. An earlier series observed RMS as the commonest metastasizing tumor [8]. In terms of exact subtypes, RMS was the commonly metastasizing tumor in our series.
To sum up, FNAC has a definite role in the diagnosis of STTs in primary, recurrent and metastatic lesions, for a timely management. Cytological categorization is effective, especially for sorting out round cell and pleomorphic tumors. Spindle and epithelioid cell tumors are challenging, especially in terms of exact sub typing. Exact sub tying can be enhanced with applications of ICC. Recognition of 'pitfall' lesions like aggressive fibromatosis is vital. Further scope of STT evaluation on cytology can be increased with more studies dealing with application of ancillary techniques on aspirate samples.