Clinicopathological features of 70 desmoid-type fibromatoses confirmed by β-catenin immunohistochemical staining and CTNNB1 mutation analysis

Desmoid-type fibromatosis (DF) is a locally aggressive neoplasm characterized by mutations in the CTNNB1 gene, which encodes the β-catenin protein. We reviewed 85 cases of DF and performed Sanger sequencing for detecting mutations in CTNNB1 and immunostaining for detecting β-catenin localization. We included 70 DF samples, of which 56 cases demonstrated nuclear β-catenin localization and 43 cases harboured CTNNB1 mutations. CTNNB1-mutant DF samples consistently displayed nuclear β-catenin expression and were derived from larger-sized tumours compared to samples with wild-type CTNNB1. When we further classified DF cases into 2 subgroups based on the type of specimen, excised specimens with nuclear β-catenin expression frequently displayed CTNNB1 mutation and no statistical correlation between nuclear β-catenin expression and CTNNB1 mutation was observed in biopsies. When we classified CTNNB1 mutation cases into 2 subgroups (DF with T41A or T41I, and DF with S45F or S45P), T41A or T41I mutations were observed more frequently in males than in females. Additionally, DF tumours harbouring S45F or S45P mutations were located more frequently in the abdominal wall than tumours with T41A or T41I mutations. In conclusion, CTNNB1 mutation correlates with nuclear β-catenin expression in larger or excised DF tumours, and DF harbouring CTNNB1 mutations manifest variable clinical presentations.


Introduction
Desmoid-type fibromatosis (DF; also called desmoid tumour, deep fibromatosis, or aggressive fibromatosis) is a borderline soft tissue tumour that displays variable aggressiveness and a high proclivity for recurrence [1,2]. Because of the high recurrence rate and preventable morbidity associated with repeated surgeries, the treatment options for DF vary from surgery to watchful waiting with careful follow-up [2][3][4]. DF tumours are classified based on their location as tumours in the abdominal wall, intra-abdominal cavity, or extra-abdominal sites. It is well known that up to 80% of sporadic lesions, regardless of tumour location, harbour mutations in the CTNNB1 gene, which encodes the β-catenin protein [5]. The most common CTNNB1 mutations in DF are observed in exon 3 and include T41A, S45F, and S45P [1,6]. A novel pathogenic alteration in DF, the T41I mutation, was recently reported [7]. When mutations occur, β-catenin degradation is impaired in the cytoplasm, leading to its accumulation in the nucleus [6,[8][9][10]. Nuclear beta-catenin staining was seen in a very limited subset of sof tissue tumor types, including desmoid-type fibromatosis, solitary fibrous tumor, endometrial stromal sarcoma and synovial sarcoma [11]. β-catenin is expressed in approximately 70−75% of DF and is used to differentiate DF from postoperative scar formation caused by other soft tissue tumours such as fibroblastoma [12]. Therefore, nuclear staining using an antibody against β-catenin is used as a marker for diagnosing DF [6,13].
Smooth-muscle actin (SMA) is usually expressed in DF, but is an unspecific marker and is not expressed in all cases of DF [12,14,15]. In diagnostic practice, it is hard to interpret subcellular β-catenin expression in the biopsied specimens or in small-sized tissues because tumour cells stain variably for β-catenin. Furthermore, genetic studies for identifying CTNNB1 mutations cannot be performed for all specimens used for the diagnosis of DF in pathology laboratories. Further investigation is required to determine whether immunohistochemical staining for β-catenin in different specimen types and in varying tumour sizes is indicative of mutations in the CTNNB1 gene.
In this study, we reviewed 85 cases of pathologically diagnosed DF and performed β-catenin immunohistochemical staining and CTNNB1 gene sequencing on tumour samples. Furthermore, we compared the β-catenin expression with CTNNB1 mutation status and analysed the clinicopathological features according to the molecular subtypes of DF and the types of specimens.

Patients and samples
We retrospectively recruited 85 patients who were diagnosed with DF from 2005 to 2013 at Severance Hospital, Yonsei University College of Medicine. Medical records were reviewed for clinical features and radiological and pathological findings. To select the most representative formalin-fixed paraffin-embedded (FFPE) tissues for immunohistochemical staining and molecular studies, samples were mounted on slides, stained with haematoxylin and eosin, and reviewed by two pathologists (JH Jung & SK Kim). The algorithm for the selection of eligible cases of DF in this study is shown in Fig 1. We excluded the uninterpretable cases which did not show internal positive control staining of β-catenin by immunohistochemistry. All the uninterpretable cases had wild type CTNNB1 genes in Sanger sequencing. Finally, 70 specimens of DF were included. Preoperative radiological imaging was available for 64 patients.
All methods and experimental protocols using human tissues were carried out in accordance with relevant guidelines and regulations approved by the Institutional Review Board of Severance Hospital, Yonsei University Health System, and the patients' consent form were waived (IRB no. 4-2016-0687).

Immunohistochemistry
FFPE blocks were cut into 4-μm-thick sections and processed using heat-induced epitope retrieval. Immunohistochemical staining was performed in an automated immunostainer (Ventana Discovery1 XT, Ventana Medical System, Inc., Oro Valley, AZ, USA). Antibodies against β-catenin (1:400, Cell Marque, #224ㅡ15) and SMA (1:1000, DAKO, #M0851) were utilized. Two pathologists (JY Ahn & JH Jung) blinded to the pathologic information independently evaluated the immunohistochemical staining. β-catenin staining was classified as negative or positive according to the subcellular staining pattern. Cytoplasmic, membranous, or Golgi staining pattern was regarded as negative expression and nuclear staining as positive expression when observed in over 1% of tumour cells. We regarded positive expression for SMA when it displayed cytoplasmic expression in more than 1% of the tumour area. We defined negative or positive staining regardless of staining intensity for both beta-catenin and SMA.

Statistical analysis
All statistical analyses were performed using SPSS version 18.0 (IBM, Chicago, IL, USA) and GraphPad Prism version 7.04 (GraphPad software, San Diego, CA, USA). The relationships between groups were compared using the χ2 test, Fisher's exact test, or unpaired t-test. Twosided P values < 0.05 were considered statistically significant.

Clinical features of DFs
We reviewed 85 cases of pathologically diagnosed DF and performed immunohistochemical staining for ß-catenin and CTNNB1 gene mutation tests on all the samples (Fig 1). Nuclear ß- catenin expression was identified in 56 out of 85 DF samples, whereas CTNNB1 gene mutation was observed in 43 out of 85 DF samples. Among the 85 cases of DF, we excluded 15 cases which did not show internal positive control staining of β-catenin and these concurrently displayed negative ß-catenin staining and a wild-type CTNNB1 gene. Thus, we excluded these ßcatenin-negative and CTNNB1-wild type cases for further analyses.
Next, we analysed clinicopathological features of 70 patients with DF (Table 1). All patients were between 2 months and 84 years old (mean age: 36.01 ± 20.89 years) with 29 males and 41 females. Nine cases manifested in the abdominal wall (12.86%), 9 in the abdominal cavity (12.86%), and 52 in extra-abdominal sites (74.29%). Tumour sizes were measured during preoperative imaging and ranged from 0.9 cm to 17.0 cm in size (mean size: 6.733 ± 4.424 cm). Sixty tumours were excised and 10 cases were biopsied. Forty-three cases out of 60 patients from whom tumours were excised underwent follow-up and 19 cases of recurrence were observed (44.19%). We performed immunohistochemical staining for ß-catenin and SMA and observed nuclear ß-catenin expression in 56 cases (80.0%; Fig 2) and SMA positivity in 61 cases (87.14%), which were found to be consistent with the literature [5]. There was no statistical difference between ß-catenin and SMA expressions in DFs (S1 Table).

CTNNB1 mutations in desmoid-type fibromatosis
Using Sanger sequencing, we detected CTNNB1 mutations in 43 out of 70 patients (61.43%; Fig 2C). We divided DF patients into 2 groups according to ß-catenin expression and CTNNB1 mutation status -those with ß-catenin expression or CTNNB1 mutations and those without ß-catenin expression and CTNNB1 mutations -and compared clinical features between the groups (S2 Table). There was no significant statistical difference in sex, age at diagnosis, tumor site, tumor size, procedures, recurrence and SMA expression between the groups. Next, we divided DF patients into 2 groups -those with CTNNB1 mutations and those without CTNNB1 mutations -and compared clinical features between the groups ( Table 2).
The results demonstrated that DF cases harbouring CTNNB1 mutations were derived from larger-sized tumours (7.655 ± 0.642 cm) compared with DF cases without CTNNB1 mutation (4.973 ± 0.952 cm; p = 0.020). Immunostaining analysis found that ß-catenin nuclear expression was more frequent in DF cases harbouring CTNNB1 mutation (39/43, 90.70%) than in DF cases without CTNNB1 mutation (17/27, 62.96%; p = 0.012). There were no significant differences between the two groups with respect to gender, age at diagnosis, tumour site, specimen type or recurrence.

Discussion
Mutations in the CTNNB1 gene, which encodes the β-catenin protein, is highly prevalent in DF and is considered to contribute to tumour recurrence [16]. Using Sanger sequencing and immunostaining, we detected a positive correlation between CTNNB1 gene mutation and nuclear β-catenin expression in the DF specimens. It is known that up to 80% of sporadic DF harbour mutations in the CTNNB1 gene [5]. However, we detected CTNNB1 mutations in 43 out of 70 patients (61.43%). We assume that the lower positive rate of DF with CTNNB1 mutation is due to the limitation of biopsied tissues. We included 10 biopsied FFPE tissues in this study, and we cannot exclude the possibility that the biopsied samples might not contain the enough amount of tumor cells for sequencing. We believe that the positive rate of CTNNB1 mutation could increase if the research was carried out with more biopsied tissues.
We identified a patient harbouring both T41A and S45F mutations in his tumour. The patient was a 48-year-old man with a 4.5-cm sized chest wall mass encasing the rib. Additionally, we detected a T41I mutation in a DF patient, who was a 3-month-old boy with a 5 × 4.8-cm sized scalp mass. A recent study reported that three CTNNB1 T41I variants were preferentially detected in paediatric DF patients (n = 3/22; 13.6%), with only a single instance of DF in an adult (n = 1/170; 0.6%) [7]. Among the mutations, the S45F mutation is implicated as a factor that increases the rate of tumour recurrence after excision [2,[16][17][18]. In this study, we reviewed 43 out of 70 DF patients with follow-up data from our institution. However, no significant differences in recurrence were observed between the DF groups with or without CTNNB1 gene mutations. When we investigated recurrences among DF patients with CTNNB1 gene mutations, there was no statistical difference between the two groups -DF with T41A or T41I mutation, and DF with S45F or S45P.
We speculated that stabilization and subsequent nuclear localization of β-catenin could be representative of CTNNB1 mutation status. Indeed, immunostaining results demonstrated that nuclear β-catenin expression was observed more frequently in DF cases with CTNNB1 mutation than in DF cases without CTNNB1 mutation. We presented four cases with CTNNB1 mutation which had beta-catenin expression with cytoplasmic/membranous/Golgi staining pattern. The relationship between CTNNB1 gene mutation, β-catenin nuclear localization, and Wnt pathway activation is complex. According to the literatures, nearly half of CTNNB1 mutant cases of endometrial carcinoma had only 5-10% of tumor cells with β-catenin nuclear localization, and the authors suggested that immunohistochemistry could be an initial screen, with CTNNB1 sequencing employed when nuclear localization of β-catenin is absent [9]. Audard et al. report that 37% of tumors with β-catenin mutations they examined were without cytosolic/nuclear staining of β-catenin in hepatocellular carcinoma [19]. Furthermore, CTNNB1 mutation was observed more frequently in the larger-sized tumours. Therefore, if the excised tumour is small in size, we need to know that β-catenin immunostaining is indicative of the presence of CTNNB1 mutation, particularly in the case of recurrent tumours. In this study, we divided DF cases into 2 groups according to the specimen types: biopsied tissues and excised tissues. Notably, the correlation between nuclear localization of βcatenin and CTNNB1 mutation was not identified in biopsied DF tissues. Therefore, we suggest that the CTNNB1 mutation can be identified in a biopsied specimen even though β-catenin expression is found in the cytoplasm or the membrane and vice versa.
In conclusion, we found that nuclear localization of β-catenin correlated with the CTNNB1 gene mutation status in DF, and DF demonstrated different clinicopathological features based on the mutations they harbour. We newly identified a patient harbouring both T41A and S45F mutations in his tumour. We demonstrated that CTNNB1 mutation correlates with nuclear βcatenin expression in larger or excised DF tumours, therefore caution is advised in the interpretation of the immunohistochemical and mutation data when derived from biopsied samples or small tumours.