Certain Risk Factors for Patients with Desmoid Tumors Warrant Reconsideration of Local Therapy Strategies

Abstract

OBJECTIVE

To evaluate treatment outcomes in patients with desmoid tumors (DT) receiving local therapy with surgery alone, radiation therapy (RT) alone, or combined modality RT and surgery (CMT).

METHODS

We performed a cross-sectional cohort study of 412 patients with non-mesenteric DTs who received local therapy at our institution between 1965 and 2018.

RESULTS

Median follow-up time was 95 months (range, 1–509). Local recurrence occurred in 127 patients (31%) at a median time of 21 months (range, 3–295). The 5-year LC was 67%. Patient or tumor factors that were significantly associated with poorer 5-year LC on multivariable analysis included: age ≤30 years (57% vs. 75% > 30; HR 1.73, P=0.004), extremity location (57% vs. 71% non-extremity; HR 1.77, P=0.004), and large tumors (>10cm: 59%; HR 2.17, P=0.004; 5–10cm: 65%; HR 1.71, P=0.02; vs ≤5cm 76%). Subset analyses of these high risk patients revealed no local therapy strategy as superior for young patients ≤30 years (surgery P=0.33, HR 1.42; RT P=0.38, HR 1.36) or for large tumors >10cm (surgery P=0.46, HR 1.55; RT P=0.91, HR 0.91). However, for patients with extremity tumors, surgery alone was significantly associated with inferior LC (surgery P<0.001, HR 5.15; RT P=0.38, HR 1.51).

CONCLUSION

Local therapy provides durable tumor control in the majority of patients with desmoid tumors. However, young patients, extremity location, or large tumors are at increased risk of recurrence. When active treatment is indicated, systemic therapy should perhaps be considered a first-line option in these high risk subsets. Prospective multi-institutional studies evaluating this strategy are warranted.

Precis:

Patients with desmoid tumors who are young or have tumors that are large or in the extremities have an increased risk of relapse following local therapy. Therefore, when treatment is indicated in these higher risk subsets, systemic therapies should likely be considered the first-line therapy.

INTRODUCTION

Desmoid tumors (DTs), also referred to as aggressive fibromatoses, often exhibit an unpredictable clinical course ranging from indolent to destructive. Initial management of asymptomatic patients should consist of a “watch and wait” strategy as there are a subset of patients whose DTs may spontaneously regress or stabilize thereby negating the need for active treatment; of those who are initially observed, half will require treatment eventually. ^1–6 Conversely, some DTs demonstrate a more aggressive, infiltrative, or destructive growth pattern that necessitates local or systemic interventions.^7–10 Ultimately, treatment is indicated for progressing or symptomatic DTs, and multidisciplinary input is critical in managing these complex patients to tailor appropriate therapies on an individual basis.¹¹

Once treatment has become necessary, upfront therapy has commonly consisted of local strategies with either wide local excision or RT. In patients where functional outcomes, tumor size, or tumor location have precluded resection, RT has provided a favorable treatment option. Tumor control outcomes, both for patients undergoing surgery or RT, typically range between 70–80%.^{7–9,12–14} Alternatively, systemic therapies have commonly been reserved for the salvage setting or for patients where anatomy precludes surgery or RT given the more modest tumor control outcomes associated with non-cytotoxic therapies and the toxicities associated with chemotherapy.¹⁵ However, newer systemic agents have activity and toxicity profiles that may warrant revisiting the “standard” DT management algorithm which has prioritized local therapies.^4,16

In addition to the realization that not all DTs require active intervention, there is increasing awareness that not all patients who receive local therapy have favorable outcomes. Here we explore a large cohort of patients with desmoid tumors who received local therapy consisting of either surgery alone, RT alone, or a combination of the two. We aim to identify risk factors for local recurrence in order to provide clinically relevant treatment recommendations to assist in day-to-day management of these complex patients.

METHODS

We reviewed the medical records of 412 consecutive patients diagnosed with non-mesenteric DTs who received definitive local therapy consisting of either surgery, RT or combined modality therapy (CMT) with surgery and RT at The University of Texas MD Anderson Cancer Center (MDACC) between 1965 and 2018. The patients were identified from institutional tumor registries and their medical records were retrospectively reviewed. Inclusion criteria for this study included having non-mesenteric DTs pathologically confirmed by our soft tissue pathology group and receipt of definitive local therapy at MDACC. This series expands upon a recently published study evaluating outcomes for patients with desmoid tumors who received RT as part of their management;⁸ patients who underwent surgery alone for their DTs were identified and included in this analysis to provide a comparison cohort for additional insights on management of patients with DT.

The management of DTs has evolved over the decades. Current practice paradigms recommend a “watch and wait” strategy for patients with newly diagnosed DTs. Once patients develop clinical progression or symptoms, local therapy has been recommended. Decisions regarding type of local therapy were made by the multidisciplinary team based on multiple patient- and tumor characteristics. Systemic therapies have typically been used as a salvage treatment after recurrence following local therapy or when local therapy could not be applied without undue morbidity. However, recently our multidisciplinary team is more commonly using systemic therapy in the upfront setting.

Follow-up and Statistical Analysis

At diagnosis or presentation to our center, patients are commonly followed every 3 months to establish the tempo of disease progression or regression. In patients requiring local therapy, follow up typically consists of an evaluation with diagnostic imaging every 3–6 months for the first two years after treatment with decreasing frequency the further out from therapy depending on disease status and symptom burden.

Disease status was determined by diagnostic radiology commonly using MRI. Local control was defined as no evidence of recurrence in the setting of resection or disease stability on serial imaging in cases where there was tumor present at the time of RT. Relapse was either confirmed with tissue biopsy or assessed with serial imaging to document progression of the suspected recurrence.

Treatment-related complications were identified retrospectively, categorized, and scored in severity as mild (requiring no intervention), moderate (requiring medical management), or severe (requiring surgical intervention or hospitalization). Surgical toxicities commonly were in the postoperative period, though late toxicities were also identified. Surgical pain was excluded as many patients presented with pain syndromes that made it difficult to differentiate retrospectively whether the pain was a result of the desmoid tumor versus surgical management.

Descriptive statistics were used to evaluate baseline characteristics. Fisher’s exact test and chi-squared analysis were employed to analyze categorical data. The Kaplan-Meier method was used to estimate actuarial rates of overall survival (OS) and local control (LC) with survival times calculated from completion of local therapy to the occurrence of the outcome. Log-rank tests were applied to assess significance of differences between actuarial curves. Multivariate analyses were conducted using the Cox proportional hazards model. Significant (P ≤ 0.05) estimated hazard ratios (HR) and 95% confidence intervals (CI) are reported. IBM SPSS Statistics 22 was used for data analysis.

RESULTS

Patient Characteristics and Treatment

Patient demographics and tumor characteristics are listed in Table 1. The median patient age was 35 years (range, 1–85 years) with 40% (n=162) of the cohort ≤30 years of age, and there was nearly a 2:1 female to male ratio. The most common tumor location was in the trunk, which comprised 61% of cases (n=253) (anterior abdominal wall n=64, chest wall n=60, back/paraspinal musculature n=39, axilla/shoulder n=38, buttock n=18, pelvis n=17). Extremity location was the next most common location (n=106, 26%) (upper extremity n=56 - proximal arm n=35, forearm n=15; lower extremity n=50 - thigh n=18, leg n=17), and nearly all the head and neck tumors (n=53, 13%) were in the neck (n=50). The median tumor size was 6 cm (range, 1–24) for the entire cohort, and DTs in the lower extremity and trunk were disproportionally larger than tumors in the upper extremity or neck (P=0.004).

Table 1.

Patient and Tumor Characteristics

Variable	All Patients Value or No. (%)	Surgery alone Value or No. (%)	RT alone Value or No. (%)	Surgery + RT Value or No. (%)	P-value
Number of patients	412	218 (53)	109 (26)	85 (21)
Follow-up time, months
Median	95	90	78	163	<0.001
Range	1–509	1–386	3–447	1–509
Age, years
Median	35	36	36	28	0.02
Range	1–85	1–77	8–85	12–73
Age, grouped
≤ 30 years	162 (39)	79 (36)	37 (34)	46 (54)	0.007
> 30 years	250 (61)	139 (64)	72 (66)	39 (46)
Sex
Male	140 (34)	64 (42)	42 (39)	34 (40)	0.11
Female	272 (66)	154 (58)	67 (61)	51 (60)
Tumor Site
Head and Neck	53 (13)	16 (7)	26 (24)	11 (13)	<0.001
Upper extremity	56 (14)	26 (12)	14 (13)	16 (19)
Trunk	253 (61)	157 (72)	53 (49)	43 (51)
Lower extremity	50 (12)	19 (9)	16 (15)	15 (18)
Maximum Tumor Dimension, cm
Median	6	5	7.2	7.0
Mean	7.1	5.8	9.0	7.6	<0.001
Range	1–24	1–22	1–24	2–23
Tumor size
≤ 5 cm	152 (37)	104 (48)	20 (18)	28 (33)	<0.001
5–10 cm	168 (41)	73 (34)	54 (50)	41 (48)
>10 cm	74 (18)	24 (11)	35 (32)	15 (18)
Unknown	18 (4)	17 (8)	0 (0)	1 (1)
Number of prior surgeries
0	268 (65)	168 (77)	62 (57)	38 (45)	<0.001
1	96 (23)	38 (17)	29 (27)	29 (34)
2	27 (6)	6 (3)	10 (9)	11 (13)
3+	21 (5)	6 (3)	8 (7)	7 (8)

Abbreviations: RT, radiation therapy; Gy, Gray

Of the 412 patients in this series, 218 (53%) received surgery alone at MDACC as definitive local management, whereas 109 (26%) received RT alone and 85 (21%) were treated with combined modality surgery and RT. The majority of patients (n=268, 65%) presented with de novo disease having not received prior surgery while 96 (23%) patients had one prior outside attempt at definitive surgical resection and 48 (11%) patients had 2 or more. Patients who were dispositioned to receive surgery alone had smaller DTs (≤5cm 48% vs 18% RT vs 33% CMT, P<0.001) and were more commonly truncal (72% vs 49% RT vs 51% CMT, P<0.001). Patients who received CMT were younger (≤30 yrs 54% vs 36% surgery vs 34% RT, P=0.007) and had tumors located more commonly in the extremities (37% vs. 28% RT vs. 21% surgery).

The median dose for patients treated with RT alone was 56 Gy (range, 56–75 Gy), whereas it was 50.4 Gy (range 50–66 Gy) for those who received CMT.

Survival and Local Control

The 5-year and 10-year OS rates were 98% and 95%, respectively, and the median follow-up time among all patients from the completion of local therapy was 95 months (interquartile range [IQR], 47–170 months). Patients who received CMT had a longer median follow up (163 months vs. 78 months RT alone vs. 90 months surgery alone, P<0.001), which is likely attributable to a practice strategy that has moved away from CMT in the past decade.

Local relapse occurred in 127 patients (31%) at a median time of 21 months (IQR, 12–38 months). The 5-year and 10-year LC rates were 67% and 65%, respectively. Among the entire cohort, there were several patient- or tumor-related factors on univariate analysis that were significantly associated with poorer 5-year LC including: younger age (≤30 yrs 57% vs >30 yrs 75%, P=0.001), extremity location (57% vs non-extremities 71%, P=0.002), and larger tumors (>10cm 59% vs 5–10cm 65% vs ≤5cm 76%, P=0.02) (Table 2, Figure 1 A-C). Additionally, when not adjusting for any potentially confounding factors or patient selection variables, single modality treatment had inferior 5-year LC compared to CMT (77% vs. 64% surgery vs. 65% RT, P=0.04).

Table 2.

Univariate Analysis of Factors Potentially Affecting Actuarial Rates of Local Control at 5 years for All Patients

Variable	LC 5-yr control, %	LC 10-yr control, %	P Value
Treatment Years
2005 or before	66	64	0.75
After 2005	71	69
Age, grouped
≤ 30 years	57	56	0.001
> 30 years	75	71
Sex
Male	67	66	0.89
Female	67	65
Tumor Site
Head and Neck	71	71	0.02
Upper extremity	60	54
Trunk	72	71
Lower extremity	52	47
Tumor Site, grouped
Extremities	57	51	0.002
Non-extremities	71	71
Tumor size
≤ 5 cm	76	75	0.02
5–10 cm	65	62
>10 cm	59	55
Prior Treatment
Yes	63	61	0.19
No	70	67
Local Treatment
Surgery alone	64	62	0.04
Radiation alone	65	62

Abbreviations: LC, local control; RT, radiation therapy; Gy, Gray

FIGURE 1.

Local recurrences for patient with desmoid fibromatosis stratified by patient age (A) tumor location (B) and tumor size (C). D-F represent subset analyses of local control strategies for these high risk subsets including: young patients (D), extremity tumors (E), and large tumors >10cm (F).

A multivariable analysis was constructed to adjust for these factors found to be significant in univariate analyses and all remained significantly associated with inferior LC: young age ≤30 years (HR 1.73, P=0.004, 95% CI 1.19–2.51), extremity location (HR 1.77, P=0.004, 95% CI 1.20–2.60), large tumor size (>10cm: HR 2.17, P=0.004, 95% CI 1.28–3.67; 5–10cm: HR 1.71, P=0.02, 95% CI 1.10–2.66), and treatment with surgery alone compared to CMT (HR 2.35, P=0.001, 95% CI 1.40–3.97); RT alone was not significantly associated with poorer LC compared to CMT (HR 1.72, P=0.06, 95% CI 0.98–3.01) (Table 3).

Table 3.

Multivariate Analyses of Factors Affecting Actuarial Rates of Local Control for the Entire Cohort

Variable	Hazard Ratio (HR)	Confidence Interval (95%)	P-value	Comparison Variable
ENTIRE COHORT
Age, grouped
≤ 30 years	1.73	1.19–2.51	0.004	> 30 years
Tumor Site, grouped
Extremity	1.77	1.20–2.60	0.004	Non-extremity
Tumor size
5–10cm	1.71	1.10–2.66	0.02	<5cm
>10cm	2.17	1.28–3.67	0.004
Treatment Approach
Surgery alone	2.35	1.40–3.97	0.001	CMT
RT alone	1.72	0.98–3.01	0.06

Abbreviations: HR, hazard ratio; RT, radiation therapy; CMT, combined modality therapy

Stratified Local Control Analysis

Several subset analyses were performed in order to further evaluate optimal local therapy strategies for characteristics associated with poorer local control. Separate multivariable analyses were performed for young patients, for patients with large tumors, and for patients with tumors of the extremity.

For patients ≤30 years old (n=162), when adjusting for tumor size, location, sex, and treatment strategy, the only factor associated with inferior LC was large tumor size (5–10 cm: HR 2.02, P=0.04, 95% CI 1.04–3.92; >10cm: HR 3.08, P=0.002, 95% CI 1.53–6.22); neither of the single treatment modality approaches were inferior compared to CMT (RT alone, P=0.38, surgery alone, P=0.33) (Table 4, Figure 1D).

Table 4.

Multivariate Analyses of Factors Potentially Affecting Actuarial Rates of Local Control for the Entire Cohort and Two Subsets of Patients Receiving RT Alone and CMT

Variable	Hazard Ratio (HR)	Confidence Interval (95%)	P-value	Comparison Variable
YOUNG AGE ≤30 SUBSET
Tumor size 5–10cm	2.02	1.04–2.92	0.04	<5cm
>10cm	3.08	1.53–6.22	0.002
Treatment Approach
Surgery alone	1.42	0.70–2.86	0.33	CMT
RT alone	1.36	0.68–2.72	0.38
Tumor Site, grouped
Extremity	1.32	0.79–2.21	0.29	Non-extremity
LARGE TUMORS >10CM
Age, grouped
≤ 30 years	2.84	1.32–6.12	0.008	> 30 years
Tumor site, grouped
Extremity	2.13	1.04–4.37	0.04	Non-extremity
Treatment Approach
Surgery alone	1.55	0.49–4.94	0.46	CMT
RT alone	0.91	0.36–2.52	0.91
EXTREMITY TUMORS
Tumor size
5–10cm	3.22	1.47–7.05	0.004	<5cm
>10cm	5.49	2.26–13.32	<0.001
Treatment Approach
Surgery alone	5.15	2.24–11.85	<0.001	CMT
RT alone	1.51	0.60–3.79	0.38
Age, grouped
< 30 years	1.42	0.76–2.64	0.27	> 30 years

Abbreviations: HR, hazard ratio; RT, radiation therapy; CMT, combined modality therapy

In patients with large tumors >10cm (n=74), when adjusting for age, location, and treatment strategy, both young age ≤ 30 years (HR 2.84, P=0.008, 95% CI 1.32–6.12) and extremity location (HR 2.13, P=0.04, 95% CI 1.04–4.37) emerge as significantly associated with inferior LC; again, neither single treatment modality approach was inferior compared to CMT (RT alone, P=0.91, surgery alone, P=0.46) (Table 4, Figure 1F).

For extremity tumors (n=106), when adjusting for age, size, and treatment strategy, inferior LC was observed in large tumors (5–10cm: HR 3.22, P=0.004, 95% CI 1.47–7.05; >10cm: HR 5.49, P<0.001, 95% CI 2.26–13.32) and in patients who received surgery alone (HR 5.15, P<0.001, 95% CI 2.24–11.85) (Table 4, Figure 1E).

The treatment cohorts were then analyzed using separate multivariable analyses to identify the highest risk factors associated with local relapse. In patients receiving RT, the only factor associated with poorer LC was young age ≤30 year for both the RT alone subset (n=109) (HR 2.56, P=0.004, 95% CI 1.34–4.88) or CMT subset (n=85) (HR 4.23, P=0.01, 95% CI 1.41–12.66). For patients that received surgery alone, factors associated with inferior LC included: extremity location (HR 2.40, P=0.01, 95% CI 1.42–4.05), large tumor size (5–10cm: HR 2.24, P=0.005, 95% CI 1.28–3.93; >10cm: HR 3.17, P=0.005, 95% CI 2.41–7.08), and prior surgical resection at the tumor site (HR 4.13, P<0.001, 95% CI 2.41–7.07).

Local control for the de novo presentation subset

To better interpret the results in the modern context of the recommended initial “wait and see” strategy, a separate analysis of patients presenting de novo having not received prior treatment was performed. Similar outcomes were observed in this subset as the entire cohort.

Seventy five patients of 268 (28%) developed local relapse resulting in a 5-year LC rate of 70%. Factors that were significantly associated with poorer 5-year LC on univariate analysis included: younger age (≤ 30 yrs 51% vs >30 yrs 80%, P<0.001), extremity location (52% vs non-extremities 74%, P<0.001), and larger tumors (>10cm 61% vs 5–10cm 68% vs ≤5cm 80%, P=0.04). No other factors were significantly associated with LC including treatment era (P=0.92), sex (P=0.54), or treatment modality (P=0.20). On multivariable analysis, only younger age (HR 2.83, P<0.001, 95% CI 1.72–4.68) and extremity location (HR 1.89, P=0.02, 95% CI 1.12–3.17) remained significantly associated with poorer LC.

Salvage

For the 127 patients that developed local recurrence, the median follow up time after relapse was 81 months (range, 1–429). Eighty-nine (70%) patients had successful treatment of their local relapse (47% surgery, 19% RT, 34% systemic therapy) as defined as either no-evidence-of-disease, stable tumor, or regressing tumor, whereas the other 28 patients (30%) continued to have progressing disease or efficacy of salvage therapy was not yet determined.

Treatment-related Toxicities

Seventy-four patients (18%) had an identified treatment-related toxicity at a median time post-treatment of 35 months (IQR, 10–120) resulting in a 5-year complication free survival of 88%. The majority of the treatment toxicities were either mild (n=18, 24%), or moderate in severity (n=22, 30%), whereas 34 patients (46% of those with treatment related toxicity) did have more severe complications that required surgical intervention or hospitalization. There was no significant difference in the proportion of complications (P=0.32) by treatment, but patients receiving surgery alone did have a higher proportion of complications that required surgical intervention or hospitalization (P=0.008), commonly related to infection (n=14, 31%) or wound healing issues (n=18, 41%). When evaluating the association between treatment era and RT-related late toxicity given changes in treatment techniques, there was a higher risk in the earlier era when stratifying at 1990 prior to 3D techniques becoming more prevalent (27% ≤1990 vs 15% >1990; P=0.02), but stratifying at 2000 (P=0.51) and 2005 (P=0.80) when intensity modulated RT usage was increasing were both non-significant. Finally, there were 3 patients (1% of those receiving RT) that developed a radiation-associated secondary malignancy at a median time of 15.1 years (range, 10.8–17) of which all had DTs of the trunk.¹⁷

DISCUSSION

Desmoid tumors are clinically complex, and require expert multidisciplinary input for management of the disease and its symptoms. Importantly, not all patients require therapeutic intervention; treatment should be reserved for DTs that are progressing or symptomatic. For those in whom treatment is indicated, local therapy – consisting of either surgical resection or RT – remains a feasible option, though systemic treatments are increasingly utilized. In the current study, we evaluated one of the largest cohorts in the literature of patients with non-mesenteric DTs to investigate outcomes following local therapy and risk factors associated with poorer outcomes. While the majority of patients achieve durable local control, we observed several subsets of patients that do not have uniformly favorable outcomes. Factors that predicted for higher relapse included younger age, extremity locations, and larger tumors. For patients with these higher risk features, perhaps reconsideration of front-line therapy is warranted. Systemic therapy should be considered a first-line treatment in these high risk patients, and local therapy should be reserved for salvage.

While the paradigm for DT management is shifting, local therapy remains an effective treatment option for patients when active surveillance is no longer indicated. Among the entire cohort, our study observed a 5- and 10-year LC rate of 67% and 65%, respectively, which is consistent with previously published studies with rates ranging between 65% to about 80%.^7–10,18 Factors that emerged as independent predictors of relapse regardless of local therapy included young patients ≤30 years (HR 1.73), extremity tumors (HR 1.77), and larger tumors (5–10cm, HR 1.71; >10cm, HR 2.17). Previous studies focusing only on surgical or RT cohorts have identified similar factors.^7–9 For example, a large surgical series by Crago and colleagues observed a 5-year local-recurrence free survival (LRFS) of 69% among the entire cohort with poorer outcomes in young patients (<26 years, 5-y LRFS 52%, HR 4.27), extremity or head and neck locations (5-y LRFS 60%, HR 5.02), and large tumors >10cm (5-y LFFS 57%, HR 1.94).

Given these higher risk subsets, we independently assessed each group to evaluate if one local control strategy was superior over another. For young patients or patients with large tumors, we observed suboptimal local control outcomes, and no particular local intervention was associated with significantly superior tumor control. Therefore, in these two high risk subsets in particular, we believe systemic therapy should be considered a first-line treatment option. Local therapy may still be warranted but should be considered a second-line or salvage option.

In contrast, however, for patients with extremity DTs, surgical resection alone resulted in inferior local control. This finding was also observed by Crago and colleagues. They reported that for patients with extremity tumors, adjuvant RT was associated with an absolute reduction in local recurrence of 15% (LRFS 71% with RT vs. 56% without).⁹ Therefore, when treatment is indicated for tumors in this location, local therapy should likely include RT; alternatively, systemic therapy could also be considered a first-line strategy.

Several recent studies have demonstrated notable improvements in the efficacy of newer targeted agents over historical systemic standards for DT management. One of the recent randomized studies by Gounder and colleagues reported a 2-year PFS of 81% in patients receiving sorafenib compared to 36% in those receiving placebo, in which the latter control group also demonstrated the potential advantage of observation versus therapeutic intervention.⁴ Similar outcomes were observed in patients treated with pazopanib; Toulmonde and colleagues demonstrated significant clinical activity of pazopanib having 84% of patients free of progression at 6 months compared to 45% in those receiving standard cytotoxic chemotherapy with methotrexate/vinblastine.¹⁶ Despite these practice changing findings, questions remain about systemic therapy. First, how long should patients remain on these therapies? Indefinitely? If so, how is that tolerated? Second, what is the duration of the responses? These are early data that warrant longer follow up, particularly given that the local therapy literature has reported equally favorable outcomes at 2 years. Third, how does prolonged systemic therapy use influence quality of life metrics compared to a local therapy strategy?

The treatment paradigm for DTs has certainly shifted in the last decade. In keeping with the recent consensus recommendations from the Desmoid Tumor Working Group, we recommend a period of active surveillance for all patients with stable DTs or asymptomatic lesions.¹¹ When treatment is indicated, our study suggests local therapy should be considered in patient subsets that are at low risk of recurrence: older patients, non-extremity tumors, and smaller tumors; whereas, in patients with high risk of relapse, systemic therapy should be a first-line option. Our findings support and expand upon the recently proposed treatment algorithm by the Desmoid Tumor Working Group that stratifies patients by tumor location.¹¹ We recommend that patient age and tumor size also be used in the risk assessment for determining management strategies. Importantly however, regardless of recurrence risks, local therapy should not be withheld in clinical scenarios where tumor progression would become debilitating or prevent a successful local control procedure with further intervention delay. Specifically, this includes patients with increasing functional disability from their tumor or poorly controlled pain, or patients for whom critical normal structures are at risk of invasion and perturbation from progressive desmoid fibromatosis.

Importantly, in addition to tumor control priorities, toxicities associated with the various treatments need to be considered. In particular, this study observed a 1% risk of secondary malignancies, which is higher than expected. Previous studies have demonstrated increased secondary malignancy risk in young patients and in non-extremity locations, which likely explains this observation given that patients with DTs are disproportionately young and have truncal involvement compared to the general population who receive RT.¹⁹ This risk should be factored in when considering RT use in young patients.

Our study was comprised of a large, homogenous cohort of patients with non-mesenteric DTs who received consistent local therapy, but despite multiple strengths, there are limitations that are important to consider when interpreting the results. First, there is selection bias in the local therapies recommended for patients; attempts at adjusting for this in multivariable models was performed but the retrospective nature of the study does not allow for removal of all biases. Secondly, the patients were treated over a long time period with evolving standards, and there could be influence on changes in diagnostic imaging and treatment that were not corrected for despite an attempt to account for era in the analyses. Furthermore, attempts were made to identify treatment decisions using modern paradigms but extracting this data retrospectively was unreliable. Despite these known limitations, we were better able to adjust for potential biases compared to most previous studies given the large, detailed dataset.

In conclusion, DTs are clinically unpredictable and can cause significant morbidity. Local therapy consisting of surgery, RT, or surgery and RT provides durable local control in a majority of patients. However, young patients, large tumors, or extremity tumors are at increased risk of relapse. Systemic therapy should be considered a first-line treatment in these high-risk subsets, but prospective evaluation of this strategy is warranted. For patients with extremity tumors, surgery alone as local therapy may have inferior tumor control compared to strategies that include RT. Initial referral to large expert centers is recommended for multidisciplinary input and management of desmoid tumors.