https://orcid.org/0000-0002-7511-6388
Abstract
An accurate diagnosis is crucial to determine the treatment modality for desmoid-type fibromatosis, although the histopathological diagnosis is occasionally difficult to make. Many desmoid-type fibromatosis have been reported to have hotspot mutation of β-catenin gene (CTNNB1). In the present study, we performed a systematic review to verify the usefulness of CTNNB1 mutation analysis in the diagnosis of desmoid-type fibromatosis.
A literature search from January 1990 to August 2017 was conducted. Three reviewers independently assessed and screened the literature for eligibility and determined the final articles to be evaluated. Data regarding the sensitivity, specificity, accuracy and usefulness of CTNNB1 mutation analysis in the diagnosis of desmoid-type fibromatosis were recorded. We rated each report according to the Grading of Recommendations Development and Evaluation approach.
The search yielded 90 studies, seven of which were included after the first and second screenings. The positive rate of CTNNB1 mutation in desmoid-type fibromatosis was 86.8%, but the cohort of six of the seven reports was already diagnosed histopathologically as desmoid-type fibromatosis. Therefore, the usefulness of CTNNB1 mutation analysis in a cohort that is difficult to diagnose histopathologically is not clear in this review. Nevertheless, CTNNB1 mutation showed very high specificity in desmoid-type fibromatosis, indicating the usefulness of CTNNB1 mutation analysis in its diagnosis in combination with histological examination.
Because the lack of data precludes any useful comparison with histological diagnosis, the evidence level is low. However, considering its specificity, CTNNB1 mutation analysis may be useful in cases in which the histopathological diagnosis is difficult.
Introduction
Desmoid-type fibromatosis (DF), also known as aggressive fibromatosis, is characterized by benign and locally infiltrative (myo-) fibroblastic neoplasms (1). It accounts for 0.03–0.1% of all solid tumors and 3.6% of fibrous tissue neoplasms (2,3). Due to not only its scarcity, but also enigmatic behavior, a standard treatment strategy for DF has not yet been established.
Extra-abdominal DF is locally aggressive and at high risk of local recurrence after planned surgery even with a wide surgical margin (range 14.1–68%) (4–7), but lacks metastatic potential (1). Furthermore, some DF cases show stabilization or spontaneous regression without treatment (8). Considering these inconsistent behaviors, the therapeutic approach to extra-abdominal DF has been shifting from surgery with a wide surgical margin to conservative therapy (9,10).
To properly formulate a treatment strategy for DF, which is ‘initial wait and see policy’, DF must be correctly diagnosed. However, the histopathological diagnosis is notably difficult in some cases (11). Intra-nuclear staining of β-catenin has traditionally been used for the diagnosis. However, the specificity is poor (12,13), and DF cases with no β-catenin staining in the nucleus are also occasionally observed (14). On the other hand, several studies have reported that ~85–90% of DF harbor hotspot mutations within β-catenin gene (CTNNB1). Most of the mutations are codons 41 and 45 of exon 3, with T41A being the most common mutation type, followed by S45F (15,16). These suggest that mutation analysis for CTNNB1 might be useful in their diagnosis. This systematic review (SR) aims to examine the usefulness of CTNNB1 mutation analysis in the diagnosis of DF. Based on its results, we would like to propose a recommendation statement for clinical guidelines for DF.
Materials and methods
Conflicts of interest and selection of SR members
No commercial companies were involved in this consensus meeting. Members of the Systematic Review Committee were elected by the Chair (Y.N.) and invited to join the Clinical Practice Guidelines Development Committee.
PICO development and importance of outcomes
This SR for the PICO question ‘Is mutation analysis of β-catenin useful for the diagnosis of DF?’ was composed of (P) Participants—extra-abdominal DF, other soft tissue tumors and lesions that resemble DF in histopathology; (I) Intervention—CTNNB1 mutation analysis; (C) Comparison—histopathological examination; and (O) Outcomes—sensitivity and accuracy of DF diagnosis. Initially, (P) was defined as extra-abdominal DF because this SR was conducted to formulate medical guidelines for extra-abdominal DF. However, the papers extracted in the review process included intra-abdominal desmoids.
Data sources and searchers
A literature search was conducted by a professional librarian belongs to the Japan Medical Library Association (N.Y. in acknowledgment), using the database of MEDLINE (Pubmed), the Japan Medical Abstracts Society (Ichushi in Japanese), the Cochran Library, the Guideline in National Quality Measures Clearinghouse and National Institute for Health and Care Excellence. This search was conducted in September 2017 using the search terms: ‘fibromatosis, aggressive’ and ‘beta catenin’ or ‘CTNNB1 protein, human’: publication date from 1 January 1990 to 31 August 2017; English; Japanese. Three reviewers independently reviewed all of the studies selected by the librarian and assessed them for eligibility based on the title and abstract as the first screening. The second screening was performed after careful reading of the entire article. Excluded studies were those that did not report CTNNB1 mutation status, series with only a small number of cases and individual case reports. We excluded articles that studied only DF with Gardner syndrome or intra-abdominal locations.
Rating the quality of evidence
The three reviewers independently rated the quality of the evidence according to Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach (17). To assess bias risk, the Cochran risk of bias tool was used (18). After the evaluation by the three reviewers of the extracted individual studies, an ‘evaluation sheet’ was generated for the selected important outcome; the sensitivity, specificity, accuracy and usefulness of CTNNB1 mutation analysis in the diagnosis of DF, taking into account, the limitations of the study design, the risk of integration bias, the factors of increase and the indirectness to PICO. The number of event patients and at risk patients for evaluation of sensitivity and specificity was also recorded and evaluated. Based on the evaluation sheets, the reviewers created the ‘body of evidence’ for the clinical question, taking into account the limitations of study design, risk of bias, inconsistencies, indirectness, imprecision and other factors according to the methods suggested by Minds Tokyo GRADE center as A (strong), B (moderate), C (weak) and D (very weak). A ‘qualitative systematic review’ and a ‘summary of systematic review report’ were created by the reviewers and provided by them to the clinical guideline committee. A recommendation was determined by consideration of the expert committee of the quality of the evidence under GRADE approach, cost of the CTNNB1 mutation analysis, insurance coverage in Japan and patient preference. The Clinical Guidelines Committee voted on recommendations and strengths in this clinical question. Seventy-five percent consent was required for determination. If the initial vote did not reach 75% consent, voting members re-discussed the issue before voting again. An explanation of the recommendation was prepared by the members of the Guidelines Committee who were responsible for this clinical question for the readers of the clinical guideline of DF.
Results
The literature review identified 108 potential articles: 107 from Pubmed and one from Ichushi. After the evaluation of a title and abstract as the first screening, 90 articles were excluded. Subsequently, we performed full-text evaluation as the second screening for eligibility. Finally, seven articles were deemed relevant and included in the final analysis (Fig. 1) (13,15,16,19–22). The details of these seven studies are listed in Table 1, namely study design, number of participants, number of CTNNB1 mutation-positive cases and important outcomes determined by the guideline committee (sensitivity and accuracy of CTNNB1 mutation analysis in the diagnosis of DF). Two studies were reported from the same institute (16,20); however, duplicated cases (recurrent cases) were excluded from the newer literature (20) to ensure the accuracy of the analysis (Table 1).
PRISMA flow diagram of study screening and selection.
Summary of studies
| Reference | Study design | Participants | Detected number of CTNNB1 mutation | Outcome | ||
|---|---|---|---|---|---|---|
| DF | Other | Sensitivity (%) | Accuracy (%) | |||
| Amary et al. (15) | Case series retrospective | DF 76 other 57 | 66/76 | 0/57 | 86.8 | 92.5 |
| Lazar et al. (16) | Case series retrospective | DF 138 | 117/138 | N/A | 84.8 | N/A |
| Salas et al. (19) | Case series retrospective | DF 182 | 150/182 | N/A | 82.4 | N/A |
| Colombo et al. (20) | Case series retrospective | DF 37 other 10 | 30/37 | 0/10 | 81.1 | 85.1 |
| Le Guellec et al. (13) | Case series retrospective | DF 254 other 175 | 223/254 | 0/175 | 87.8 | 92.7 |
| Aitken et al. (21) | Case series retrospective | DF 144 other 15 | 133/144 | 0/15 | 92.4 | 93.1 |
| Crago et al. (22) | Case series retrospective | DF 117 | 104/117 | N/A | 88.9 | N/A |
| Reference | Study design | Participants | Detected number of CTNNB1 mutation | Outcome | ||
|---|---|---|---|---|---|---|
| DF | Other | Sensitivity (%) | Accuracy (%) | |||
| Amary et al. (15) | Case series retrospective | DF 76 other 57 | 66/76 | 0/57 | 86.8 | 92.5 |
| Lazar et al. (16) | Case series retrospective | DF 138 | 117/138 | N/A | 84.8 | N/A |
| Salas et al. (19) | Case series retrospective | DF 182 | 150/182 | N/A | 82.4 | N/A |
| Colombo et al. (20) | Case series retrospective | DF 37 other 10 | 30/37 | 0/10 | 81.1 | 85.1 |
| Le Guellec et al. (13) | Case series retrospective | DF 254 other 175 | 223/254 | 0/175 | 87.8 | 92.7 |
| Aitken et al. (21) | Case series retrospective | DF 144 other 15 | 133/144 | 0/15 | 92.4 | 93.1 |
| Crago et al. (22) | Case series retrospective | DF 117 | 104/117 | N/A | 88.9 | N/A |
DF, desmoid-type fibromatosis; N/A, not applicable.
Summary of studies
| Reference | Study design | Participants | Detected number of CTNNB1 mutation | Outcome | ||
|---|---|---|---|---|---|---|
| DF | Other | Sensitivity (%) | Accuracy (%) | |||
| Amary et al. (15) | Case series retrospective | DF 76 other 57 | 66/76 | 0/57 | 86.8 | 92.5 |
| Lazar et al. (16) | Case series retrospective | DF 138 | 117/138 | N/A | 84.8 | N/A |
| Salas et al. (19) | Case series retrospective | DF 182 | 150/182 | N/A | 82.4 | N/A |
| Colombo et al. (20) | Case series retrospective | DF 37 other 10 | 30/37 | 0/10 | 81.1 | 85.1 |
| Le Guellec et al. (13) | Case series retrospective | DF 254 other 175 | 223/254 | 0/175 | 87.8 | 92.7 |
| Aitken et al. (21) | Case series retrospective | DF 144 other 15 | 133/144 | 0/15 | 92.4 | 93.1 |
| Crago et al. (22) | Case series retrospective | DF 117 | 104/117 | N/A | 88.9 | N/A |
| Reference | Study design | Participants | Detected number of CTNNB1 mutation | Outcome | ||
|---|---|---|---|---|---|---|
| DF | Other | Sensitivity (%) | Accuracy (%) | |||
| Amary et al. (15) | Case series retrospective | DF 76 other 57 | 66/76 | 0/57 | 86.8 | 92.5 |
| Lazar et al. (16) | Case series retrospective | DF 138 | 117/138 | N/A | 84.8 | N/A |
| Salas et al. (19) | Case series retrospective | DF 182 | 150/182 | N/A | 82.4 | N/A |
| Colombo et al. (20) | Case series retrospective | DF 37 other 10 | 30/37 | 0/10 | 81.1 | 85.1 |
| Le Guellec et al. (13) | Case series retrospective | DF 254 other 175 | 223/254 | 0/175 | 87.8 | 92.7 |
| Aitken et al. (21) | Case series retrospective | DF 144 other 15 | 133/144 | 0/15 | 92.4 | 93.1 |
| Crago et al. (22) | Case series retrospective | DF 117 | 104/117 | N/A | 88.9 | N/A |
DF, desmoid-type fibromatosis; N/A, not applicable.
All seven articles selected were retrospective case series studies. Their sample size ranged from 47 to 429 including DF and DF mimickers. The number of DF cases ranged from 37 to 254. Mimickers of DF included mesenchymal lesions such as scar, fasciitis/myositis, other fibroblastic lesions and sarcomatoid carcinoma. In six of the seven reports, intra-abdominal DF was included in the cohort, and in the other one report, the tumor location was not noted. For the CTNNB1 mutation analysis, five reports used Sanger sequencing, one report used next-generation sequencing (NGS) (21), and another used both (22).
The sensitivity of the CTNNB1 mutation ranged 81.1–92.4%, and the accuracy rate was 85.1–93.1%. However, considering that six of the seven reports focused on a collection of cases already diagnosed histopathologically as DF, we could not verify the usefulness of CTNNB1 mutation analysis in comparison with histopathological examination in this review. On the other hand, four studies analyzed the CTNNB1 mutation for mimickers of DF (13,15,20,21). A salient finding was that no cases positive for CTNNB1 mutation were found in any of the mimickers (0/257), suggesting the specificity of CTNNB1 mutation analysis to be 100%,
The ‘body of evidence’ for this clinical question for the selected important outcomes was created and integrated by the three reviewers (Table 2). We assessed the strength of the ‘body of evidence’ as pertaining to outcome. The strength level was B for both the sensitivity and accuracy of CTNNB1 mutation in the diagnosis of DF. Because all seven studies were case-series studies, the strength of the evidence started with C for two outcomes. With regard to sensitivity, all seven studies demonstrated very high rates of 81.1–92.4%, the effects of the intervention (CTNNB1 mutation analysis) are considered to be significant. Therefore, the strength of the evidence for sensitivity was raised from C to B. Similarly, with regard to accuracy, since all four studies describing the accuracy indicated high rates of 85.1–93.1%, the effects of the intervention (CTNNB1 mutation analysis) are considered to be significant. As well, the strength of the evidence for accuracy was raised from C to B.
Study quality of evidence according to GRADE guidelines
| Outcomes | No. of studies | Study design | Risk of bias | Inconsistency | Imprecision | Indirectness | Others | Level of evidencea | Significance of outcomeb |
|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | 7 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Accuracy | 4 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Outcomes | No. of studies | Study design | Risk of bias | Inconsistency | Imprecision | Indirectness | Others | Level of evidencea | Significance of outcomeb |
|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | 7 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Accuracy | 4 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
aA, strong; B, moderate; C, low; D, very low.
bSignificance of outcome was determined by guideline committee when setting up PICO by a 1–10 point evaluation (10: most important).
Study quality of evidence according to GRADE guidelines
| Outcomes | No. of studies | Study design | Risk of bias | Inconsistency | Imprecision | Indirectness | Others | Level of evidencea | Significance of outcomeb |
|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | 7 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Accuracy | 4 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Outcomes | No. of studies | Study design | Risk of bias | Inconsistency | Imprecision | Indirectness | Others | Level of evidencea | Significance of outcomeb |
|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | 7 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
| Accuracy | 4 | Case series | Not serious | Not serious | Not serious | Not serious | Not serious | B | 8 |
aA, strong; B, moderate; C, low; D, very low.
bSignificance of outcome was determined by guideline committee when setting up PICO by a 1–10 point evaluation (10: most important).
‘Body of evidence’, ‘qualitative systematic review’ and a ‘summary of systematic review report’ were prepared by the reviewers and were submitted to the guideline committee. Based on these reports, considering that the CTNNB1 analysis is not covered by insurance and not yet available at many facilities in Japan, the guideline committee is responsible for this clinical question proposed that ‘we weakly recommend CTNNB1 mutation analysis in the diagnosis of DF’. This recommendation obtained a 100% consensus on the vote by the entire guideline group.
Discussion
DF is an ‘enigmatic’ soft tissue tumor that is locally aggressive, but has no metastatic potential and may spontaneously regress (1). Because of its characteristic behavior, it is necessary to conduct multidisciplinary treatment according to its particular algorithm. The treatment strategy for DF is described separately from that of other soft tissue tumors in the National Comprehensive Cancer Network (NCCN) and European Society for Medical Oncology (ESMO) guidelines (23,24). In some cases, it is difficult to make a sufficiently reliable histopathological diagnosis to be able to differentiate it from other soft tissue tumors, fibroblastic lesions and scars (11). This study is the first review to examine the usefulness of CTNNB1 mutation analysis in the diagnosis of DF.
Six of the seven reports were a collection of cases already diagnosed histopathologically as DF, and so we could not verify the usefulness of CTNNB1 mutation analysis in cases with difficulty in histopathological diagnosis in this review. Furthermore, CTNNB1 mutation analysis can be performed only in limited facilities in Japan and is not a genetic testing with insurance coverage. We cannot recommend performing the analysis for all cases of suspected DF. First, morphological assessment with hematoxylin and eosin stain should be performed, and CTNNB1 analysis had better be performed when diagnosis is difficult. However, the specificity of the CTNNB1 mutation in DF was very high, and interestingly, no CTNNB1 mutation was detected in any of the non-DF diseases (mimickers) in this review. Colombo et al. reported 47 cases in which the histopathological diagnosis was difficult to make, 37 of which could be diagnosed as DF by CTNNB1 mutation analysis (20). The results of this study suggested that CTNNB1 mutation analysis can be a useful aid in the diagnosis of DF, particularly in cases difficult to diagnose histopathologically. Among seven articles listed in this review, Lazar et al. described that nuclear β-catenin staining was observed in 98% of specimens, and CTNNB1 mutations were observed in 85% of DF (16). Whereas Amary et al. reported that all DF cases and 72% of the mimics tested showed nuclear positivity for β-catenin, indicating immunohistochemistry is a sensitive but not a specific test for DF (15). A recent study reported the low positivity of nuclear staining of β-catenin in DF. It demonstrated that, among 40 cases of DF, only 21 (52%) had nuclear β-catenin staining, and among 32 tumors with known CTNNB1 mutation, 17 (53%) had nuclear β-catenin staining (22). A more recent study, which was not listed in this review, reported that 84% of DF cases showed nuclear staining of β-catenin, and among 17 cases in which nuclear immunostaining was absent, CTNNB1 mutation was observed in five cases (14). Together, in cases with difficulty to diagnose histopathologically, particularly cases with equivocal immunohistochemical results of β-catenin expression, CTNNB1 mutation analysis will be a useful tool for DF diagnosis. The mutations of CTNNB1 have been also reported in some types of carcinomas (hepatocellular carcinoma, uterine endometrium carcinoma and colorectal carcinoma) and sarcomas (malignant fibrous histiocytoma, de-differentiated liposarcoma and synovial sarcoma) (25–30). The location of the CTNNB1 mutation in these reported cases of sarcomas is mostly outside the hotspots of DF, so that sensitivity and accuracy of CTNNB1 mutation analysis for DF might be still high. In addition, these malignant lesions can be diagnosed easily by their clinical and histopathological features without CTNNB1 mutation analysis.
In this study, 81.1–92.4% of DF had CTNNB1 mutation, and the remaining cases were CTNNB1 wild type. In five of the seven reports, Sanger sequencing was used for CTNNB1 mutation analysis, and the detection rate of the mutation in DF was 81.1–87.8%. On the other hand, Aiken et al. reported a relatively high detection rate using NGS (21). Crago et al. performed whole exome sequence on eight DF cases with CTNNB1 mutation negative by Sanger sequencing. They found CTNNB1 mutations in three of the eight cases by NGS analysis (22). The results indicated that even if the CTNNB1 mutation is negative by Sanger method, there are actually not fewer cases with positive CTNNB1 mutation. They also detected mutations of APC in two of the eight cases (22), and suggested that wild type with Sanger method includes cases of Gardner’s syndrome with attenuated familial adenomatous polyposis and/or cases with sporadic mutation of APC. Recently, large intragenic deletions of CTNNB1, and mutations of AKT1 and/or BRAF with or without CTNNB1 hotspot mutations were also reported in the DF cases (31,32). Conversely, we should keep in mind that cases with CTNNB1 wild type may have included non-DF lesions because occasionally it is difficult to differentiate DF from its mimickers. More accurate analysis of CTNNB1 mutation and diagnosis of DF with CTNNB1 wild type by Sanger method will be important topics for future research.
There were several limitations in this study. First, as mentioned above, the majority of DF in this review were collections of the cases that had already been histopathologically diagnosed as DF. Therefore, we could not compare the usefulness of CTNNB1 mutation analysis compared with that of histopathological examination. Second, most reports included intra-abdominal DF. Intra-abdominal DF may contain cases with germ-line APC mutation (Gardner syndrome) and sporadic CTNNB1 or APC mutated cases. The significance of CTNNB1 mutation analysis might be more important for extra-abdominal cases. Third, all studies were retrospective case series, with low levels of evidence. However, randomized controlled trial study would not be practical for comparison of CTNNB1 mutation analysis and histopathological evaluation. Accumulated data from results of case series would be sufficient to make a recommendation of this test. Fourth, the histopathological diagnostic criteria for DF might vary at individual facilities. To compare the usefulness of histopathological examination and CTNNB1 mutation analysis for the diagnosis of DF, a non-randomized multicenter case-recorded study including non-DF lesions would be needed to resolve future research questions.
Conclusion
We could not verify the usefulness of CTNNB1 mutation analysis in comparison with histopathological examination for diagnosis of DF in this review. However, considering high specificity and accuracy of CTNNB1 mutation analysis, the analysis could be a useful diagnostic tool for DF that is difficult to diagnose histopathologically. For these reasons, we make a weak recommendation for the role of CTNNB1 mutation analysis in the diagnosis of DF.
Acknowledgements
We thank Japan Medical Library Association and Mr. Naohiko Yamaguchi for their support to literature search. We also thank Ms. Eri Ishihara and Yoko Kawai for their secretarial assistance for this study.
Funding
This work was supported in part by the Ministry of Health, Labor and Welfare of Japan, and the Ministry of Education, Culture, Sports, Science and Technology of Japan [Grant-in-Aid 17H01585 for Scientific Research (A)], and the National Cancer Center Research and Development Fund (29-A-3) and with support from the Japanese Orthopedic Association.
Conflict of interest statement
None declared.
Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
