Expression of FRS2 in atypical lipomatous tumor/well-differentiated liposarcoma and dedifferentiated liposarcoma: an immunohistochemical analysis of 182 cases with genetic data

Background The fibroblast growth factor receptor substrate 2 (FRS2) gene is located close to MDM2 and CDK4 within the 12q13-15 chromosomal region. FRS2 gene was recently found to be consistently amplified in atypical lipomatous tumor (ALT)/well-differentiated liposarcoma (WDL) and dedifferentiated liposarcoma (DDL), suggesting the detection of FRS2 amplification could be a diagnostic tool for ALT/WDL/DDLs. However, the expression of FRS2 protein and diagnostic value of FRS2 immunohistochemistry (IHC) has not been evaluated in a large cohort of ALT/WDL/DDLs. Methods A SNOMED search of hospital surgical pathology files from January 2007 to July 2020 identified 182 ALT/WDL/DDLs with available materials. FRS2 fluorescence in situ hybridization (FISH) and IHC were performed on 182 ALT/WDL/DDLs and 64 control samples. The expression of FRS2 was also compared with that of classic immunomarkers (MDM2 and CDK4) of this tumor entity. Results This study included 91 ALT/WDLs and 91 DDLs. The FISH results showed 172 of 182 (94.5%) cases were FRS2-amplified, and 10 cases were FRS2-nonamplified. Immunostaining results showed 171 (94.0%) ALT/WDL/DDLs were positive for FRS2 and 11 cases (6.0%) were FRS2-immunonegative. In 172 FRS2-amplified cases, 166 (96.5%) were FRS2-immunopositive, and 6 (3.5%) were negative. Among 10 FRS2-nonamplified ALT/WDL/DDL cases, 5 cases were FRS2-immunonegative, and 5 tumors displayed 1+ staining for this marker. In 64 control cases, none of them exhibited FRS2 amplification. Forty-seven (73.5%) control cases were negative for FRS2 immunostaining, while 17 cases (26.5%) were FRS2-immunopositive. Fifteen of these false positive samples (15/17, 88.2%) showed 1+ positivity and only 2 cases (2/17, 11.8%) displayed 2+ positivity. In ALT/WDL/DDLs, the sensitivity of FRS2 immunostaining was slightly lower than MDM2 (FRS2 vs. MDM2: 94.0% vs 100.0%) and CDK4 (FRS2 vs. CDK4: 94.0% vs 97.0%). However, the specificity of FRS2 (73.5%) was slightly higher than that of MDM2 (67.8%) and CDK4 (64.4%). Conclusion This study indicated that FRS2 IHC had relatively good consistency with FRS2 FISH, suggesting that FRS2 immunostaining could be utilized as an additional screening tool for the diagnosis of ALT/WDL/DDL. It must be emphasized that MDM2/CDK4/FRS2 especially MDM2 FISH remains the gold standard and the most recommended method to diagnose this entity.

Conclusion: This study indicated that FRS2 IHC had relatively good consistency with FRS2 FISH, suggesting that FRS2 immunostaining could be utilized as an additional screening tool for the diagnosis of ALT/WDL/DDL. It must be emphasized that MDM2/CDK4/FRS2 especially MDM2 FISH remains the gold standard and the most recommended method to diagnose this entity.
The fibroblast growth factor receptor substrate 2 (FRS2) gene is also located in the 12q13-15 region close to the MDM2 and CDK4 genes. In 2011, Wang et al. reported the consistent amplification of the FRS2 gene in ALT/WDL/DDLs (57 of 57, 100%) [14]. Later, Zhang et al. verified FRS2 amplification in 15 DDLs, with an amplification frequency of 93% (14/15) [15]. Our recent work also identified that the FRS2 gene is highly amplified in 146 ALT/WDL/DDL cases (136/146, 93.2%) [16]. These results indicated that FRS2 have diagnostic utility for this tumor group. Because the lack of commercial FRS2 FISH probes and the evaluation of FISH requires specific equipment and experienced cytogeneticists. Furthermore, IHC is a more economical alternative and could be carried out in most hospitals in clinical practice. Hence, it is necessary to examine the utility of FRS2 IHC for the diagnosis of ALT/WDL/DDL. Moreover, FRS2 acts as a key adaptor protein in the fibroblast growth factor receptor (FGFR) pathway, subsequently activating downstream signaling pathways [17]. Hence, the detection of FRS2 protein expression in ALT/WDL/ DDL may also provide clues for their therapy in the future. However, the expression of FRS2 in ALT/WDL/ DDL and the utility of FRS2 IHC for the differential diagnosis of this disease spectrum have not been widely investigated. FRS2 IHC was only performed by Zhang et al. in 11 DDLs, among which 9 (82%) cases were positive for FRS2 [15].
Here, we performed FRS2 immunostaining in 182 ALT/WDL/DDL and 64 control cases, and the results were compared with FRS2 FISH data of the 182 cases to evaluate the sensitivity, specificity and diagnostic value of FRS2 IHC for this entity. Moreover, the MDM2 and CDK4 IHC results of most cases were reviewed to compare FRS2 and these classic immunomarkers for ALT/ WDL/DDL.

Case selection
This study was approved by the West China Hospital Institutional Review Board. A SNOMED search of the hospital surgical pathology files from January 2007 to July 2020 identified 182 ALT/WDL/DDLs with available materials for further study. Histological sections as well as previous MDM2 and CDK4 IHC slides of the tumors were reviewed independently by two pathologists with soft tissue tumor pathology expertise (H.Z. and H.C.) and 2 general surgical pathologists (W.J. and T.L.) independently. Classification was performed according to the World Health Organization criteria, and grading was carried out following the 'modified' Federation Nationale des Centres de Lutte Contre le Cancer (FNCLCC) grading system [2,18]. Clinicopathological information was collected from clinical records and pathology reports. The control cases were obtained independently from clinical cases, including conventional lipoma (n = 25), spindle cell lipoma (n = 7), atypical spindle cell lipomatous tumor (n = 3), myxoid liposarcoma (n = 2), pleomorphic liposarcoma (n = 8), myxofibrosarcoma (n = 5), leiomyosarcoma (n = 5), schwannoma (n = 1), undifferentiated pleomorphic sarcoma (n = 3), osteosarcoma (n = 1), low-grade myofibroblastic sarcoma (n = 1), and normal fat (n = 3).

Fluorescence in situ hybridization (FISH)
FRS2 FISH was performed on 182 ALT/WDL/DDLs, among which 146 cases had archival FRS2 FISH records that were published previously [16], and additional FRS2 FISH analyses were performed on the remaining 36 cases. MDM2 FISH was performed on all FRS2 nonamplified ALT/WDL/DDL cases. FISH assays were carried out according to an established laboratory protocol [8,19]. A bacterial artificial chromosome (BAC) clone for FRS2 (RP11-956E11) was purchased from the Children's Hospital Oakland Research Institute (CHORI, Oakland, CA, USA). Preparation and validation of the FRS2 probe was performed according to a previously published study [14]. A Vysis LSI MDM2 dual-color probe (Abbott Molecular, Des Plaines, IL, USA) was used for the MDM2 FISH assay. Tumors were scored by two investigators counting 100 nuclei in a blinded fashion. The hemorrhage and necrotic areas were excluded as much as possible when making the evaluation. The amplification of FRS2 or MDM2 was defined as FRS2/CEP12 or MDM2/CEP12 ≥ 2.0, while a ratio < 2.0 was considered nonamplified and a ratio < 2.0 with more than two signals was considered polysomic for CEP12.

Statistical analysis
The sensitivity and specificity of FRS2 IHC were evaluated in the ALT/WDL and DDL subgroups to assess the ability of the immunomarker to correctly classify ALT/ WDL/DDL from their histologic mimics. Student's t-test was used to compare continuous variables, and McNemar and kappa tests were utilized to evaluate the agreement between FRS2 FISH and FRS2 IHC. The chisquare test was used to compare the positive expression rates of FRS2 with the other immunohistochemical markers by SPSS version 20.0 (IBM Corp, Armonk, NY, USA). P-values < 0.05 were considered significant.

Diagnostic values of FRS2 immunostaining for ALT/WDL/ DDL
The diagnostic sensitivity and specificity of FRS2 IHC in distinguishing ALT/WDL/DDL from other histological mimics were 94.0 and 73.5%, respectively (Table 3). With the standard of 2+ staining, the specificity reached 96.9%, while the sensitivity decreased to 66.0%. The sensitivity and specificity of FRS2 IHC for distinguishing ALT/WDL from normal fat tissue and other benign adipose tumors (lipoma, spindle cell lipoma) were 91.2 and 80.0%, and those of 2+ FRS2 were 54.9 and 100.0%, respectively. In the differentiation of DDL from other

Discussion
The diagnosis of ALT/WDL/DDL based only on histological features could sometimes be challenging. Owing to the consistent amplification of MDM2 and CDK4 in ALT/ WDL/DDL, FISH and IHC for the identification of MDM2 and CDK4 amplification or overexpression have been widely used for the differential diagnosis of this entity [7][8][9][10][11][12][13]. Recently, the amplification of the FRS2 gene in ALT/WDL/DDL was identified by several studies, including our research group [14][15][16]. The use of FRS2 IHC was assessed by Zhang et al. in only 11 DDLs [15]. Here, we performed FRS2 IHC in 182 ALT/WDL/DDL and 64 histologic mimics with corresponding genetic data.
In 64 FRS2-nonamplified control cases, 47 (73.5%) cases were negative for FRS2 immunostaining and 17 were FRS2-immunopositive, including 2 cases with 2+ positivity and 15 cases with 1+ positivity. It should be mentioned that the FISH analysis of the pleomorphic liposarcoma with 2+ positivity showed polysomic for CEP12, which may lead to FRS2-immunopositivity in this case. Moreover, among the 15 cases with 1+ FRS2 expression, 2 tumors also harbored a polysomic pattern (1 pleomorphic liposarcoma and 1 undifferentiated pleomorphic sarcoma). Other changes such as necrosis or hemorrhage were found in 5 cases (2 leiomyosarcomas, 1 pleomorphic liposarcoma, 1 lipoma and 1 spindle cell lipoma). Therefore, the evaluation should be careful in the cases exhibiting 1+ positivity with hemorrhage and/ or necrosis.
This series showed that the diagnostic sensitivity and specificity of FRS2 IHC were 94.0 and 73.5%, respectively, in distinguishing ALT/WDL/DDL from other histological mimics. The results suggested that FRS2 could be used as a screening tool for ALT/WDL/DDL. In addition, we analyzed the utility of FRS2 IHC for differentiating ALT/WDL and DDL from their own morphological simulators. When we used 1+ or 2+ as the threshold value, the sensitivity and specificity of FRS2 to distinguish ALT/WDL from fat tissue and benign adipocytic tumors were 91.2 and 80.0%, 54.9 and 100%, respectively. In differentiating DDL from other spindle cell tumors, the sensitivity and specificity of FRS2 were 96.7 and 65.5%, respectively. Interestingly, when using 2+ staining as the cutoff value in the differentiation of DDL with its histologic mimic, the sensitivity and specificity of FRS2 IHC changed to 76.9 and 93.1%, respectively. Hence, when one case had 2+ FRS2 immunostaining positivity, ALT/WDL/DDL should be highly suspected.
The total positive rate of FRS2 immunostaining in our cohort was generally similar to that of existing diagnostic markers (MDM2 and CDK4). In our series, the sensitivity of FRS2 immunostaining seemed to be slightly lower than that of MDM2 (FRS2 vs. MDM2: 94.0% vs 100.0%) and CDK4 (FRS2 vs. CDK4: 94.0% vs 97.0%), whereas the sensitivities of MDM2 and CDK4 in previous studies  The main therapeutic strategy for ALT/WDL/DDL is surgical resection; in the meantime, targeted therapy has been developed and is worthy of further exploration, especially for unresectable tumors. In recent years, MDM2 antagonists and CDK4 inhibitors have been used in clinical and preclinical studies, and some of them obtained favorable outcomes [29][30][31][32]. In our study, the analysis of FRS2 gene amplification and expression can help in the diagnosis of ALT/WDL/DDL. Moreover, as the key adaptor of the FGFR pathway, the overexpression of FRS2 in ALT/WDL/DDL indicated that it could play an important role in the targeted therapy of this disease. Zhang et al. identified activation of FGFR/FRS2 pathways in liposarcoma, and the use of FGFR inhibitors could significantly inhibit the growth of liposarcoma cells. Other groups also reported that FGFR inhibitors could inhibit cell proliferation in FRS2-amplified DDL cell lines, further demonstrating the possible therapeutic use of the FGFR/FRS2 pathway in ALT/WDL/DDL [15,33,34]. In addition, recent studies showed that pharmacologically targeting FRS2 inhibited FGF/FGFR-mediated oncogenic signaling and tumor progression in prostate cancer and gastric cancer cell lines [35]. Further investigations are warranted to further evaluate the therapeutic prospects of the FGFR/FRS2 pathway in ALT/WDL/ DDL.
In summary, our study analyzed the expression of FRS2 protein in 182 ALT/WDL/DDL cases and compared the result with genetic data. The FRS2 immunostaining had relatively good consistency with FRS2 FISH. Moreover, FRS2 immunostaining had a slightly lower sensitivity but a slightly higher specificity than that of classic IHC markers (MDM2 and CDK4). It also should be noted that FRS2 IHC performed in a similar fashion to MDM2 and CDK4, which are also imperfect in their way with a subset of cases showing false immunopositivity in MDM2/CDK4/FRS2-nonamplified samples. This new marker could be utilized as an additional screening tool for the diagnosis of ALT/WDL/DDLs in laboratories without the possibility of performing in situ hybridization assays. Importantly, careful histological inspection, immunohistochemical and molecular tests can help to arrive at correct diagnosis for challenging cases. It must be emphasized that MDM2/CDK4/FRS2 especially MDM2 FISH remains the gold standard and the most recommended method to diagnose this entity. In the future, the amplification and overexpression of the FRS2 gene in this disease spectrum might provide new clues for the targeted therapy of ALT/WDL/DDL.