1 Introduction

Desmoid tumors (DT), also known as aggressive fibromatoses, are rare monoclonal fibroblastic tumors that can affect any part of the body, with an estimated yearly incidence of 2–4 new cases per million people [1]. Management of DTs is often a multi-faceted discussion between the multi-disciplinary medical teams and the patient. Because DTs typically affect young adults with an age peak of 30–40 years [2, 3], any intervention must consider long-term or late effects that may manifest decades following therapy. Although DTs do not metastasize, they exhibit a variable clinical trajectory ranging from spontaneous regression to aggressive local recurrence and/or progression, which can be debilitating and even life-threatening when involving vital structures. Therefore, careful patient selection for intervention is essential.

Treatment options for DTs include active surveillance (AS), surgery, radiation therapy (RT), non-steroidal anti-inflammatory drugs (NSAIDs), endocrine therapy, tyrosine kinase inhibitors, gamma secretase inhibitors, chemotherapy, and investigational approaches such as high intensity focused ultrasound, isolated limb perfusion, and cryoablation [4-8]. In recent decades, the management of DTs has evolved with a shift towards a more conservative approach. For asymptomatic patients, regardless of tumor site or size, guidelines from the international Desmoid Tumor Working Group (DTWG) in 2020 and 2024 [3, 9] advise a period of initial AS under a specialist team. Beyond this, the selection of the most appropriate therapy depends on factors including tumor symptoms and location, whether the tumor is associated with familial adenomatous polyposis (FAP) and patient preferences. Where surgical morbidity is limited and adequate surgical margins can be achieved, surgical resection as upfront therapy can be considered. Otherwise, RT offers excellent long-term control, but this must be carefully balanced with the risk of late toxicities, such as fibrosis, functional impact, and secondary malignancies [10-13]. Modern medical therapies have proven efficacy for DTs, though as a group they comprise a heterogenous range of agents. Recent consensus guidelines have deprioritized endocrine therapy and NSAIDs due to lack of efficacy, but highlighted active agents including tyrosine kinase inhibitors, gamma-secretase inhibitors, and chemotherapy [3, 9]. Selection of these agents will be influenced by drug accessibility, tumor progression, symptoms, and patient tolerance and acceptance. FAP-associated DTs tend to be more aggressive with multi-focality and have higher morbidity and mortality compared to sporadic DTs, and hence this subpopulation of patients may benefit from more aggressive medical management [14].

Due to the rarity of DTs, there is a paucity of robust data comparing efficacy or informing optimal sequencing of treatment modalities. Clinicians often must extrapolate beyond the parameters of published reports to tailor treatment for the patient under their care, balancing achieving local control with treatment risks to achieve optimal long-term health outcomes. Here, we report our experience in managing patients with DTs, where the vast majority of our patients with DTs were discussed in a multi-disciplinary meeting. We analyzed trends in patterns of care and tumor control, and imaging response outcomes following various treatment strategies over a 12-year period at both patient and treatment episode levels. These real-world results from a large cohort add to the limited body of data for a rare disease with heterogeneous treatment strategies.

2 Materials and Methods

3 Results

3.2 Treatment Trends

Trends in utilization of treatment types were assessed by dividing the 2010–2021 period into three 4-year periods. For 1L interventions only, adoption of AS was stable across this time at around 32% (Figure 1A, Table S2). However, use of 1L-NSAIDs/tamoxifen increased dramatically over this period, from 6.9% (2010–2013) to 47.3% (2018–2021). The inverse was observed for use of 1L-RT, declining from 34.5% (2010–2013) to 3.6% (2018–2021). Interestingly, there was a dip in the proportion of intra-abdominal tumors between 2014 and 2017 (7.8%, compared to 17.2% over 2010–2013 and 21.8% over 2018–2021), which may account for the drop in 1L-surgery in the same period. Otherwise, there were no significant shifts in tumor site over time to confound this observation.

For all lines of intervention, a dominant trend of increasing NSAID/tamoxifen use was also observed, from 10.3% (2010–2013) to becoming the mainstay treatment strategy at 54.5% (2018–2021) (Figure 1B, Table S3). RT had the sharpest decrease in utilization from 33.3% (2010–2013) to 8.2% (2018–2021). Surgery use had a fall in proportion after the 2010–2013 period (28.2%) but remained stable thereafter (10.1%–13.4%).

3.3 Treatment Switch

Our TTNI definition (refer Section 2) serves as a surrogate for tumor control, as patients were not requiring a change in therapy. Surgery and RT conferred the highest TTNI with 77.1% and 81.6% of patients estimated to not require progression-related treatment switch at 2 years, and 72.3% and 66.1% at 5 years, respectively (Figure 2). For AS and NSAID/tamoxifen, these rates were 61.9% and 53.7% at 2 years, and 52.0% and 41.2% at 5 years, respectively (Figure 2). For chemotherapy/targeted therapy/trial drug, 58.3% were estimated to not require a change in treatment at 2 years, and all patients in this subgroup had switched by 5 years (Figure 2). When restricting to the 1L setting only, the TTNI curves were not substantially different (Figure S1).

Overall, the sequencing of treatment modalities after 1L intervention was highly variable (Figure 3). At the time of study analysis, approximately two-thirds of patients undergoing AS, surgery, and RT had not required subsequent therapy (60%, 58% and 67%, respectively). One patient had planned pre-operative RT followed by surgery for an upper limb tumor. No patients received adjuvant RT following surgery. For non-AS interventions, TTNI was not different whether in first or subsequent-line settings (p = 0.39).

Common reasons for stopping medical therapy were clinical futility as defined by clinician-assessed disease or symptom progression (60.7%) and toxicity (25.0%) (Table S4). An issue with compliance was seen particularly for NSAIDs, accounting for 11–13% of cases. No RT-related malignancies were observed in this study.

3.4 Imaging Response

Imaging responses at 6, 12, and 24 months for every treatment episode, except for surgery (not relevant) and focused ultrasound (small numbers), are depicted in Figure 4. Notably, there was an increasing proportion of spontaneous partial and even complete responses over 24 months for the AS cohort, despite not having received therapeutic intervention (5% CR and 12% PR at 24 months, respectively). This should be borne in mind when interpreting responses in the other cohorts. Medical therapy primarily stabilized disease growth, while RT has a moderate chance of offering PR, but not CR (33%, 42% and 35% PR at 6, 12, and 24 months).

4 Discussion

This study reports on the largest cohort of patients and treatment episodes for DTs in Australia over a 12-year contemporary period. Our results confirm that the conventional modalities of surgery and RT offer the advantage of tumor debulking or shrinkage (not just stabilization) on top of excellent long-term control, though their use has been offset by a worldwide shift towards more conservative management with AS and NSAID/tamoxifen [3, 9]. The drop in the use of surgery particularly is reflected in global trends in patterns of practice [15, 16], though not universally [17]. Importantly, our data corroborates that AS is safe in appropriately selected patients [18-20], with 52% not requiring therapeutic intervention at 5 years. This is similar to other institution-based studies, such as that by Rastrelli et al., which reported 44% of patients managed with AS not requiring change in management [21]. Consistent with other reports, 17% of DTs in our cohort underwent spontaneous regression at 2 years (5% CR) without any therapeutic intervention, and patients whose tumors remained controlled while on AS at 2–3 years were unlikely to require subsequent active intervention [19, 22, 23]. Nonetheless, tumors in the AS subgroup in our cohort were smaller than those in active intervention cohorts, and notably, 37% of these patients reported pain at baseline (Table 1). This practice is in contrast to the 2020 DTWG guidelines (published towards the end of this study period) which recommended AS based on an asymptomatic presentation, regardless of tumor size or site [3]. It will be interesting to observe if a shift in accordance with these guidelines will result in a difference in TTNI for the AS strategy.

The use of NSAID/tamoxifen increased significantly over the study period, while AS as 1L management remained steady. As an explanation for this trend, we postulate that a proportion of these patients may have been commenced on NSAIDs for pain relief, while still being effectively managed as AS. This is supported by the observation that 60% of patients who received 1L NSAID (±tamoxifen) reported pain as a symptom. The clinical burden of DTs is often high, with pain reported in up to 63% of patients who participated in a French Desmoid Advocacy group survey [24]. Another study by Penel et al. demonstrated that the presence of pain at baseline, compared to patients who were pain-free, was associated with lower quality of life, higher functional impairment, and poorer event free survival [25]. Because these medical treatments are considered to have low toxicity, they could also have been commenced to manage patient (or even clinician) anxiety. In fact, a recent study considered NSAIDs and tamoxifen as ineffective for DT and categorized patients on these agents as under AS [18]. Furthermore, in that cohort (baseline symptoms not reported), the majority (69%) of patients were on NSAIDs, while only 25% were strictly not on any treatment. It is worth noting, however, that in our study the 1L-AS and 1L-NSAID/tamoxifen subgroups were not identical in regard to tumor size, site, and presenting symptoms (Table S1), which may explain the different TTNI curves between the two strategies (Figure 2). Aligned with international consensus, our unit has now moved away from using NSAID/tamoxifen as upfront therapy to AS in most patients and an earlier adoption of modern targeted therapies when treatment is required.

In this context, the group of chemotherapy, targeted therapy, and trial drugs has increased in use over time, reflecting the increased availability of effective medical therapy options. The recent FDA approval of nirogacestat in progressing DT based on the Phase 3 De-Fi trial was an exciting new addition in the armamentarium of systemic therapies beyond cytotoxic chemotherapy and tyrosine kinase inhibitors such as sorafenib and pazopanib [6]. However, many questions remain regarding these therapies, including the optimal duration and sequencing of therapies. The possible long-term toxicities of this class of drugs are also a concern in a disease that often afflicts a younger population. Superficially, it may appear surprising that the TTNI for these agents was only modest and not better than that of NSAID/tamoxifen, but tumors in this cohort were significantly larger at the time of treatment commencement, with many of the clinical trials using these agents requiring progressive disease at study entry. Furthermore, the limited cohort size hampers the confidence of this observation. This was due to difficult access to targeted therapy and trial drugs during the study period, with patients often requiring to self-fund targeted therapies outside trials. Off-label use of TKIs on compassionate access programs and a clinical trial involving a gamma secretase inhibitor (AL102) became an option at our institution only towards the end of this cohort period (Table S2); thus, this data does not reflect the more readily adopted use of these agents. While comparison with existing reports is highly challenging due to the wide range of agents used and small study numbers, our results fit within most studies of TKIs and chemotherapy agents, which report 12 month progression-free survival rates generally between 60% and 80%, and overall response rates anywhere between 5% and 60% (being higher with cytotoxic chemotherapy) [3, 4, 19].

Selecting the most appropriate endpoint for assessment is challenging for DTs. Overall response rates are a helpful endpoint for medical therapies but not so much for surgery or RT. Progression-free survival based on radiological evaluation is often used but is complicated by a reliance on regular imaging as well as the natural history of DTs, whereby tumors can have alternating intervals of progression, stability, and even regression. Furthermore, many DTs have complex radiological appearances that are not easily assessable by RECIST, hence creating a level of subjectivity [26]. Assessment of change in MRI T2 signal intensity is used in our center to detect treatment response, with T2 hyperintensity representing viable DT activity, but this evaluation is not yet standardized. Quality of life measures are highly important, but they are poorly documented in a consistent manner outside of clinical trials and can be difficult to analyze. In this study, we used a composite TTNI endpoint, for which an event is only triggered if the change in therapy is due to progressive disease or symptom burden. Otherwise, the therapy switch is considered a censored observation. Thus, a TTNI event in this study can be interpreted as progression due to the futility of the current treatment. This endpoint is objective, easily discerned from medical records, and meaningful for clinicians and patients.

Limitations of this study include caveats related to its retrospective and single-institution nature. As alluded to above, given the progress in treatment options for DTs since our study period ended [9], our study population does not fully reflect the most contemporaneous practice in specialist centers, including the use of newer systemic therapies and locally ablative techniques such as cryoablation. Additionally, patients with DTs who were successfully managed with AS in the community were not fully captured in our database. For all patients, including those referred to PMCC after failure of AS in the community, the ACCORD database only recorded AS as an intervention if explicitly stated as such by the clinician. To address this, we automatically coded all patients who received their first active intervention ≥ 6 months from diagnosis as having received AS upfront, while also acknowledging this may not have captured all patients who were managed with initial AS intent. Another limitation is missing data on imaging responses due to a proportion of patients, especially those on AS, undergoing follow-up imaging outside PMCC with no documentation in our system. The use of TTNI as an endpoint is therefore more reliable. Finally, TTNI outcomes in our study are not meant to be compared across treatment strategies as they are heavily influenced by patient selection. Ultimately, this study aims to provide a contemporary snapshot of the management of DTs in our institution to inform clinician and patient discussions using real-world data, rather than to necessarily provide recommendations for clinical practice.

5 Conclusion

In this large real-world cohort of DT patients and treatment episodes, we highlight the heterogeneity and changing trends in the management of DTs. AS, as recommended in DTWG guidelines, is a valid primary treatment for asymptomatic patients and is associated with a high rate of tumor control. RT and surgery remain effective treatments that offer excellent long-term control of DTs in patients not suitable for AS. Importantly, the landscape of medical therapies, including new entries such as gamma secretase inhibitors, continues to evolve and has the potential to significantly shape the DT management paradigm. Similarly, the emergence of newer local ablative therapies, such as cryoablation, holds promise for treating DTs. Trials are ongoing to establish the optimal timing, sequencing, and its integration with medical therapies.

Author Contributions

Joseph Sia: formal analysis (lead), investigation (equal), methodology (equal), supervision (equal), visualization (lead), writing – original draft (lead), writing – review and editing (lead). Stephanie Tan: data curation (lead), writing – original draft (equal), writing – review and editing (equal). Narayan Mohanakrishnan: data curation (equal). Kelvin Yu: data curation (equal), methodology (equal), supervision (equal). Samuel Y. Ngan: writing – review and editing (equal). Sarat Chander: writing – review and editing (equal). Jayesh Desai: writing – review and editing (equal). Jeremy Lewin: writing – review and editing (equal). Anne Hamilton: writing – review and editing (equal). Stephen J. Luen: writing – review and editing (equal). David E. Gyorki: writing – review and editing (equal). Hayden Snow: writing – review and editing (equal). Claudia DiBella: writing – review and editing (equal). Sarah O'Reilly-Harbidge: writing – review and editing (equal). Lisa M. Orme: writing – review and editing (equal). Julie Chu: conceptualization (equal), methodology (equal), supervision (equal), writing – review and editing (equal). Susie Bae: conceptualization (equal), methodology (equal), supervision (equal), writing – review and editing (equal).

Ethics Statement

Approval by the Peter MacCallum Cancer Centre (PMCC)'s Human Research Ethics Committee was obtained for this study, including a waiver of the requirement for consent (approval number QA/73959/PMCC).

Conflicts of Interest

The authors declare no conflicts of interest.