Lack of Prognostic Value of CTNNB₁ Mutation Profile in Desmoid-Type Fibromatosis

The incidence of desmoid-type fibromatosis (DF) is approximately five individuals per million people yearly. The peak incidence age at diagnosis is approximately 40 years (1), with two thirds of the patients being female, and the median tumor size is 6 cm (1). DF is a monomorphic proliferation of monoclonal fibroblast-like cells that form an invasive, infiltrative, and retractile soft-tissue tumor mass without metastatic spreading.

Notably, the clinical course of DF is unpredictable; some DFs are spontaneously stable (40%–50%), some spontaneously regress (20%), and some increase in size, requiring treatment (30%–40%). In recent years, considering the uncertainty in tumor course, the risk of local relapse after surgical resection, and the potential harms of surgery, there has been a shift in practice guidelines from surgery to active surveillance. Similarly, a stepwise approach to treatment, primarily systemic treatment, has been adopted in a selected subset of cases with significant tumor progression or functional impairment (2).

The accumulation of β-catenin is a molecular driver of DF. Approximately 10% to 15% of DF cases are associated with germline adenomatous polyposis coli (APC) pathogenic variants and familial adenomatous polyposis (FAP). Conversely, 85% to 90% of DFs are sporadic and are associated with somatic β-catenin gene (CTNNB₁) exon 3 pathogenic variants. The three major CTNNB₁ mutations were found in two hotspots in exon 3: codon 41 (p.T41A) and codon 45 (p.S45F and p.S45P; refs. 3, 4). Apart from CTNNB₁ and APC, pathogenic variants and other molecular alterations remain rare and nonrecurrent. Because of other recurrent molecular alterations, different CTNNB₁ pathogenic variants are logically regarded as factors explaining the natural history of DF.

Several studies and one meta-analysis have explored CTNNB₁ mutation as a prognostic factor (5–9), with p.S45F being associated with poor outcomes and a high rate of local recurrence after surgery. Nevertheless, the results of these studies remain conflicting (5–9). Furthermore, all these studies are retrospective and focused on DF resection (5–9), although surgery is no longer regarded as the first-line management of DF. Therefore, in 2016, we launched a prospective nationwide cohort study to improve the understanding of the outcomes of DF (ALTITUDES study). We report herein the first analysis of this cohort study focusing on the prognostic value of CTNNB₁ pathogenic variants.

ALTITUDES is a prospective nationwide clinical-biological cohort of incident DF diagnosed between January 2016 and December 2020 and confirmed through a central pathologic review. Following recommendations from the French National Cancer Institute, suspected cases of DF required a second pathology opinion by the labeled French Sarcoma Group, which included molecular confirmatory tests (next-generation sequencing or Sanger sequencing for identifying CTNNB₁ mutations). Analysis of the CTNNB₁ mutational status was performed at diagnosis. In compliance with the French Regulation, this study was approved by the National Ethics Committee [approval by the ethics committee was on December 3, 2015 (CPP Nord-Ouest I), and by the French Drug Agency on November 20, 2015 (ANSM)]. Written informed consent was obtained from each patient. This study was registered as NCT02867033.

The inclusion criteria for study participants were: (i) incident case of sporadic or FAP-associated DF diagnosed after January 1, 2016, in France (metropolitan regions; overseas departments were excluded); (ii) diagnosis confirmed by pathology review in the French Sarcoma Group; (iii) affiliation to the National Health Insurance; and (iv) signed informed consent (both parents’ signature in patients aged <18 years). The exclusion criteria were: (i) administrative or legal measures of liberty privation and (ii) inability or unwillingness to provide consent.

The ALTITUDES study did not include specific therapeutic interventions and no precise guidelines for choosing one of the possible first-line approaches but entailed a prospective collection of clinical data and biobanking. The management was decided by both the attending physician and patient. Since the ALTITUDES study is not a randomized trial, the efficacy of surgical resection, active surveillance, and systemic treatment, were not directly compared, as there are apparent indication biases. This was a hypothesis-generating study and did not have a formal sample size calculation.

We selected patients with unifocal DF treated by surgery, active surveillance, or systemic treatment. Patients with multifocal tumors and those who had undergone radiotherapy or cryotherapy as first-line therapies were excluded.

The primary endpoint was event-free survival (EFS) from the date of diagnosis (9). Regarding the heterogeneity of first-line management, the events considered were: (i) local relapse after complete (R0/R1) resection, (ii) disease progression according to the local investigator after R2 resection or during active surveillance or systemic treatment, or (iii) deaths. EFS was censored at the date of the last contact when no event was reported in the follow-up, and at the date of second-line treatment if a treatment was started because of worsening symptoms but without documented progression or relapse (only 14 cases, 2.2%).

The description of the study population was based on classical statistical methods: number and percentages for categorical data and median and extreme values for continuous data. EFS was tabulated using the Kaplan–Meier method and calculated from the date of diagnosis to the occurrence of an event or death. Prognostic factors were analyzed using univariate and multivariate Cox models. We evaluated the effect of CTNNB₁ mutation (p.S45F vs. all other cases) on EFS according to the type of first-line approach (surgery, active surveillance, or systemic treatment) by including an interaction term in the multivariate model. A similar method was used to evaluate the heterogeneity of CTNNB₁ mutations on EFS according to patient age, tumor site, and tumor size. Estimates are reported with their 95% confidence intervals (CI), and analyses were performed at a two-sided 5% alpha level. Statistical analyses were conducted using STATA statistical software version 15·0 (StataCorp, LLC, College Station, Texas).

The data generated in this study are available upon request from the corresponding author, subject to approval by the study management committee and appropriate data transfer agreements. Genetic data have been submitted to ClinVar platform (Submission N°SUB11163120).

Overall, 628 patients were enrolled in the ALTITUDES study, and 516 (82.2%) met the eligibility criteria (Supplementary Fig. S1). Table 1 summarizes patient characteristics. Most cases were regarded as sporadic, without a personal or family history of polyposis (433 informative cases; 83.9%). In contrast, personal or family history of polyposis was observed in 42 cases (8.1%). In 33 cases (6.4%), either personal history or family history was missing. In eight cases (1.6%), data were missing for personal and family histories.

The five most common primary tumor locations were as follows: abdominal wall (168; 32.7%), chest wall and paraspinal trunk (78; 15.1%), mesentery (42; 8.2%), breast (27; 5.3%), and shoulder girdles (25; 4.9%). Overall, limb DF was observed in 68 patients (13.2%).

The median tumor size at diagnosis was 50 mm (range, 5–530 mm). The first-line strategy was as follows: surgery in 120 cases (23.3%), active surveillance in 352 cases (68.2%), and systemic treatments (hormonal therapies, chemotherapy, or molecular targeted therapies) in 44 cases (8.5%). Most surgical procedures as the first treatment were performed outside labeled centers, and the quality of the surgery could be assessed in 72 out of 120 cases (60.0%). R0/R1 and R2 were achieved in 62 and 10 cases, respectively. The nature of the systemic treatments is provided in Supplementary Table S1. Among the 352 patients managed by active surveillance, at least 41 received nonsteroidal anti-inflammatory drugs (11.6%).

CTNNB₁ mutations or deletions were found in 435 (84.3%) cases (Table 2). The frequency of the classical mutations was 50.6% for p.T41A (261 cases), 13.0% for p.S45F (67 cases), and 16.3% for p.S45P (84 cases). Notably, DF harboring the p.S45F mutation was significantly associated with male sex, extra-abdominal wall locations, pain (measured by Visual Analogic Scale), and large tumor size (Table 1 and Fig. 1). In contrast, the distribution of CTNNB₁ alterations appeared similar between the three first-line strategies.

The median follow-up period was 24.7 months (range, 0.4–59.7). During the follow-up period, 129 patients experienced an event (3 deaths: 1 caused by rapidly progressing intra-abdominal DF, 1 due to another cancer, and 1 due to stroke). Overall, the EFS probability at 2 and 3 years was 70.3% (95% CI, 65.5%–74.7%) and 66.2% (95% CI, 60.5%–71.2%), respectively (Supplementary Fig. S2). In univariate analysis, the CTNNB₁ mutation profile was not significantly associated with EFS (P = 0.65; Fig. 2). In comparison with no mutation or other mutations, the HR associated with p.S45F mutation was 1.06 (95% CI, 0.65–1.73; P = 0.81). As expected, in univariate analysis, EFS differed according to first-line strategy and tumor size (Table 3). In multivariate analysis including both tumor size and first-line strategy, when comparing patients with p.S45F mutation to all other cases, the adjusted HR was 0.91, (95% CI, 0.55–1.49), with P = 0.71. Supplementary Figure S3 shows that we did not observe any significant heterogeneity in this estimate across tumor site subgroups or first-line strategy subgroups. On the other hand, we observed borderline heterogeneity across age and tumor size subgroups. Finally, with sample size limitations, Supplementary Figures S4 and S5 show the absence of prognostic value of p.S45F in term of EFS, in patients managed by active surveillance in one hand (Supplementary Fig. S4) and those treated by surgical resection (Supplementary Fig. S5) in the other hand.

This study provides a prospective assessment of the distribution of CTNNB₁ alterations in newly diagnosed DFs. We found a p.S45F mutation in 13.0% of cases. p.S45F was associated with factors classically associated with poor outcome (male sex, extra-abdominal wall location, pain, and large size). Nevertheless, in univariate and multivariate analyses (including both tumor size and first-line strategy), we did not find that p.S45F was associated with EFS.

The CTNNB₁ mutation profile is controversial in existing literature, and current evidence must be interpreted when considering treatment options. Most of the available retrospective studies assessed the prognostic value of CTNNB₁ pathogenic variants after R0/R1 resection. Timbergen and colleagues recently published an international meta-analysis of seven studies in this setting. The study population consisted of 329 resected DF (56% of R0 and 38% of R1 resection), with a median size of 55 mm, occurring mostly in female patients (75%) with a median age of 38 years. The distribution of CTNNB₁ mutations was as follows: p.T41A (47%), p.S45F (20%), p.S45P (7%), and “wild-type” (26%). During a median follow-up period of 49 months, 25% of DF cases relapsed. In the univariate analysis, the poor prognostic factors were age, tumor size, location in the extremities or head and neck, and p.S45F mutation. Multivariate analysis revealed three prognostic factors for local relapse: tumor size [HR = 1.53, (1.03–2.28), P = 0.034], extremity location [HR = 4.15, (2.14–8.05), P < 0.001], and absence of CTNNB₁ mutation [HR = 0.44, (0.21–0.92), P = 0.029; ref. 4]. Timbergen and colleagues stressed the relationship between p.S45F and tumor size (4).

In this study, we also found an association between tumor size and p.S45F mutation; tumors harboring p.S45F tended to be larger (median tumor size: 61 mm vs. 50 mm). Multivariate analysis revealed that tumor size, rather than CTNNB₁ mutation profile, was the main factor associated with EFS.

Subsequently, Nishida and colleagues reported a series of 88 resected DFs managed in Japanese centers. In their multivariate analysis, the poor prognostic factors for local relapse were extremity location [HR = 5.59, (2.55–12.20), P < 0.001], recurrent DF [HR = 3.1, (1.05–9.18), P = 0.41], p.S45F mutation (HR = 2.92, (1.12–7.60), P = 0.28], and R1 versus R0 [HR = 2.39, (1.04–5.46), P = 0.039; ref. 10). Today, the recommended strategy for DF is active surveillance rather than surgery. To the best of our knowledge, no previous study has reported the prognostic value of CTNNB₁ mutations in patients managed by active surveillance. Moreover, data on patients with progressive DF treated with systemic treatments remain sparse.

Kasper and colleagues reported that in 34 patients with DF treated with imatinib, tumor control was higher in DF harboring the p.S45F mutation (11). Notably, in this study, p.S45F was particularly frequent (13/34; 38%; ref. 11). In a retrospective study of 90 patients with DF treated with oral vinorelbine, Mir and colleagues reported that p.S45F or p.S45P pathogenic variants were associated with a longer time to treatment failure than p.T41A or wild-type DF [HR, 2.78 (1.23–6.27); P = 0.04; ref. 12]. Conversely, Nishida and colleagues reported that in 38 patients in a phase II trial with methotrexate-vinblastine, the p.S45F mutation was not associated with longer progression-free survival [HR = 0.44, (0.28–8.47), P = 0.614; ref. 13). Two of these three studies suggested that the CTNNB₁ mutation in codon 45 could predict a good response to systemic treatment. Overall, these findings indicate that p.S45F could be a poor prognostic factor for relapse after surgery, but could also be a good predictive factor for response to systemic treatment. However, these findings must be interpreted with caution because of these studies’ small sample size and retrospective nature. In this prospective study, we found that CTNNB₁ mutational profile does not influence the EFS of newly diagnosed DF, regardless of the first-line strategy.

This study had some limitations. The first-line treatments were heterogeneous; there were various strategies according to the center of initial treatment (sarcoma versus non-sarcoma centers), and there was also an obvious indication bias. Consequently, ALTITUDES data should not be used to directly compare the efficacy of surgery and active surveillance. Furthermore, regarding the heterogeneity of first-line strategy, EFS definition included different items: relapse after complete surgery, disease progression after R2 resection or during active surveillance, and systemic treatments. These events did not have the same clinical values. Disease progression during active surveillance did not imply systematic starting of treatment; some of these progressions were transient with subsequent regression. In contrast, the occurrence of disease progression during systemic treatment is a more severe clinical event. Nevertheless, interaction (Supplementary Fig. S3) and subgroup analyses (Supplementary Figures S4 and S5) showed that p.S45F was not associated with EFS, regardless of the first-line strategy. Longer follow-up is required for study the correlation between mutational status and occurrence of tumor regression, especially is case of active surveillance. Lastly, APC mutations are not available in suspected FAP-associated DF; systematic sequencing of APC is ongoing and will analyze later.

Moreover, in this study, CTNNB₁ sequencing was not centralized but was prospectively done at diagnosis, and the CTNNB₁ mutational profile is consistent with literature data (3, 4, 6, 7, 10). The median follow-up period was relatively short (27 months). Nevertheless, we know that most of the events usually occur within 2 years of diagnosis (2, 4). Therefore, beyond EFS, other outcomes should be assessed in further studies, such as health-related quality of life or pain. The accumulation of β-catenin is a driver of DF. In this study, we found that CTNNB₁ mutation profile explains the natural course of DF. The prognostic value of other biomarkers should also be explored. For example, the overexpression of WNT target genes (such as AXIN2, DKK1, and CCND1) (13–15), Mutations in genes other than CTNNB₁ (such as AKT1 or KIT; refs. 16, 17), immune infiltrates (18), macrophages (19), tertiary lymphoid structures (20), microvessel density (21, 22), or circulating DNA of CTNNB₁ (23). This will require large studies with biobanking such as ALTITUDES, the study from Fondazione IRCCS Istituto Nazionale dei Tumori (NCT02547831), and the study from the Netherlands (NTR4714).

In conclusion, in this large prospective study, the CTNNB₁ mutational profile was associated with some unfavorable prognosis factors (size, locations, pain) but not associated with EFS; new studies are warranted to identify biomarkers predicting the outcome of DF.

S. Piperno-Neumann reports personal fees from Immunocore and Atlanthera outside the submitted work. P. Boudou-Rouquette reports other support from PharmaMar and personal fees from Ipsen outside the submitted work. J.-E. Kurtz reports nonfinancial support from AstraZeneca and personal fees from GlaxoSmithKline and Clovis outside the submitted work. A. Italiano reports grants and personal fees from Bayer and Roche; grants from Pfizer, Bristol-Myers Squibb, Merck Sharp & Dohme, and AstraZeneca; grants and nonfinancial support from Merck; and nonfinancial support from Transgene and Birdie Pharmaceuticals outside the submitted work. A. Le Cesne reports personal fees from PharmaMar, Bayer, and Deciphera outside the submitted work. D. Orbach reports other support from private fund during the conduct of the study. J. Thery reports grants from Institut Curie and Ligue Nationale Contre le Cancer during the conduct of the study. O. Mir reports personal fees from Amgen, AstraZeneca, Bayer, Blueprint Medicines, Bristol-Myers Squibb, Eli Lilly, Ipsen, Lundbeck, Merck Sharpe & Dohme, Pfizer, Roche, Servier, and Vifor Pharma outside the submitted work. No disclosures were reported by the other authors.

N. Penel: Conceptualization, data curation, supervision, methodology, writing–original draft, writing–review and editing. S. Bonvalot: Data curation, writing–review and editing. A.-M. Bimbai: Software, formal analysis. A. Meurgey: Data curation, writing–review and editing. F. Le Loarer: Data curation, writing–review and editing. S. Salas: Data curation, writing–review and editing. S. Piperno-Neumann: Data curation, writing–review and editing. C. Chevreau: Data curation, writing–review and editing. P. Boudou-Rouquette: Data curation, writing–review and editing. P. Dubray-Longeras: Data curation, investigation, writing–review and editing. J.-E. Kurtz: Data curation, writing–review and editing. C. Guillemet: Data curation, writing–review and editing. E. Bompas: Data curation, writing–review and editing. A. Italiano: Data curation, writing–review and editing. A. Le Cesne: Data curation, writing–review and editing. D. Orbach: Data curation, writing–review and editing. J. Thery: Investigation. M.-C. Le Deley: Formal analysis, methodology, writing–review and editing. J.-Y. Blay: Data curation, writing–review and editing. O. Mir: Writing–review and editing.

This work was supported by the Ligue Nationale Contre le Cancer, InterSarc, and a personal grant from a donor (S. Wisna). This work was also funded by a grant from nonprofit institutions that support academic research: Ligue Nationale Contre le Cancer, InterSarc, and Institut Curie Fondation (personal grant from a donor, S. Wisna). The study design was set up by the study coordinator (N. Penel), the Sponsor (Center Oscar Lambret), and their statisticians. The authors would like to thank the patients and their families for their participation in this study, and the staff members involved in the study management: Emilie Decoupigny and Marie Vanseymortier. We would also like to thank the patient advocacy group “SOS Desmoïde” and the French National Cancer Institute (INCa) for funding the labeled networks for the management of sarcomas (including Inter-Sarc).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.