Soft Tissue Tumors of the Abdominal Wall: Analysis of Disease Patterns and Treatment

Hypothesis  Abdominal wall tumors, though clinically similar, have varying degrees of biological behavior.

Design  Retrospective review of prospective databases.

Setting  Memorial Sloan-Kettering Cancer Center.

Patients  Eighty-five patients with abdominal wall soft tissue tumors.

Main Outcome Measures  Primary endpoints included time to first local recurrence, distant metastases, and disease-related mortality. Survival analysis was performed by Kaplan-Meier method, and comparisons were made by log-rank analysis.

Results  Thirty-nine desmoids, 32 soft tissue sarcomas (STS), and 14 dermatofibrosarcoma protuberans (DFSP) underwent surgery directed at achieving margin-negative resection. Unlike DFSP, most STS (77%) and desmoids(87%) were deep lesions requiring full-thickness abdominal wall resection and mesh reconstruction. Median follow-up time was 53 months, 101 months, and 31 months, with 5-year local recurrence–free survival rates of 97%, 100%, and 75%, for desmoids, DFSP, and STS, respectively. Desmoid tumors resected with positive microscopic margins had higher local failure rates (68% [positive margin] vs 100% [negative margin] 5-yr local recurrence–free survival, P<.05). For STS, high grade, deep location, and size at or above 5 cm were adverse prognostic factors for disease-specific and distant recurrence–free survival (P<.05); patients experiencing local recurrence was associated with decreased 5-year relapse-free survival rates (87% [primary] vs 50% [local recurrence], P<.05). Characteristically, no DFSP or desmoid developed distant metastases. Soft tissue sarcomas had significantly lower relapse-free survival rates than DFSP or desmoids (P<.05).

Conclusion  Abdominal wall tumors demonstrate a broad spectrum of biological behavior. Desmoids and DFSP are a local problem. High grade, size at or above 5 cm, and deep location predict distant failure and tumor-related mortality for patients with STS. Complete surgical resection is the recommended treatment approach to achieve local control. Stratification by prognostic factors will facilitate selection of patients with STS for adjuvant systemic therapies.

SOFT TISSUE tumors of the abdominal wall, though clinically similar, have many distinct histologic subtypes. For optimal treatment planning and outcome, consideration of the type of primary tumor is crucial. Some tumor variants are known for their aggressive biologic behavior. Included among these are soft tissue sarcomas (STS), desmoid tumors, and dermatofibrosarcoma protuberans (DFSP).

The natural history of abdominal wall soft tissue tumors is poorly understood, in large part because of their rarity. Consequently, only a few small retrospective series representing these lesions exist. These rare neoplasms can be challenging clinical problems in terms of resection and reconstruction. The treatment of choice for most abdominal wall soft tissue tumors is surgical resection with an adequate margin of uninvolved tissue.

While the most common soft tissue neoplasm of the abdominal wall is the desmoid tumor, a similar incidence of STS is seen in this site, and these should be considered in the differential diagnosis. Soft tissue sarcomas are mesenchymal neoplasms comprising 1% of adult malignant growths. They have a high incidence of local recurrence (25%) and have a propensity for distant metastases, accounting for a substantial mortality rate of 50%.¹ Less than 5% of sarcomas appear as primary abdominal wall tumors. Prognostic factors for local and distant recurrence and disease-specific survival are not well defined for sarcomas originating in this anatomic site. Truncal sarcomas are commonly included with extremity sarcomas in prognostic analyses based on perceived similar factors for outcome; however, specific data to support this assumption are lacking.

Desmoid tumors and DFSP are rare, slow-growing neoplasms recognized for their progressive, locally infiltrative nature. Dermatofibrosarcoma protuberans is a low-grade STS. Desmoid tumors, while they seem to be histologically benign, behave clinically like low-grade fibrosarcomas. Despite their proclivity for local recurrence following resection, desmoid tumors do not, and DFSP rarely do, metastasize to distant sites. Only the fibrosarcomatous variant of DFSP demonstrates a more aggressive behavior with potential for distant metastasis. The association of margin-positive resection with subsequent local recurrence in patients with abdominal wall desmoid tumors is controversial. Several studies have addressed this issue with varying conclusions.^2-6

The purpose of the current study was to define the clinical behavior and prognostic indicators of outcome for abdominal wall soft tissue tumors in a well-characterized, relatively large cohort of patients treated and followed up prospectively at a single institution. In patients with abdominal wall STS, we identified those at high risk for disease recurrence and death. For those patients with DFSP and desmoid tumors, factors affecting local control were evaluated. In this article, management strategies for abdominal wall tumors are discussed, and treatment recommendations are provided.

Between July 1, 1982, and August 7, 1999, we identified 86 patients with primary and recurrent abdominal wall soft tissue tumors, including desmoid tumors, STS, and DFSP, from a prospective database at Memorial Sloan-Kettering Cancer Center (MSKCC), New York, NY.

Primary endpoints included time to first local recurrence, distant metastases, and disease-related mortality rate. The following clinicopathologic factors were correlated with the study end points: (1) patient factors (patient age, sex, and symptoms); (2) tumor factors (status at initial visit [primary vs local recurrence] and tumor depth); (3) and pathologic factors (tumor size, histologic subtype, histopathologic grade, and status of microscopic surgical margins [positive, negative, or close]). All patients were treated with appropriate surgical resection. Some received adjuvant radiation and/or chemotherapy (generally doxorubicin and ifosfamide) based on prognostic factors predicting an increased risk of local or distant disease failure at the discretion of the Multidisciplinary Soft Tissue Sarcoma group at MSKCC or as part of clinical trials. Because adjuvant therapy was nonuniform and not randomized, these treatment-related factors were not studied.

A primary tumor was defined as a localized lesion previously untreated or biopsied (incisional or inadequate excisional biopsy) prior to definitive surgical therapy. The abdominal wall was defined as the region bordered superiorly by the costal margins, inferiorly by the inguinal ligaments, and laterally by the posterior axillary lines. Tumors originating in the abdominal wall and extending full thickness into the peritoneal cavity to involve intraperitoneal structures were included in this analysis. Tumor size was defined as the maximum diameter of the tumor at radiographic imaging or pathologic analysis. Tumors were classified as small (<5 cm), intermediate (5-10 cm), or large (>10 cm). Tumor grade was classified as low or high based on the degree of cellularity, differentiation and vascularity, nuclear pleomorphism, number of mitoses per high-power field, and amount of stromal necrosis.⁷ Tumor depth was characterized as superficial or deep relative to the deep fascia. Macroscopic margins were assessed at the time of surgery, and microscopic margins were assessed histopathologically. Complete resection was defined as the absence of gross residual disease following surgical excision of the tumor. A microscopically negative margin was defined as a tumor being farther than 1 mm (more than one half of an ×10 microscopic field) from the inked margin. A microscopically positive margin was defined as a tumor being present at the inked margin. A close microscopic margin was defined as a tumor present within less than 1 mm (less than one half of an ×10 microscopic field) from the inked margin. Local recurrence was defined as a documented tumor being within or contiguous to the previously excised field 3 or more months following primary therapy.

Slides were reviewed to confirm the histologic diagnosis. A histologic grade was assigned to nondesmoid tumors according to established criteria.⁷ Classical DFSP was diagnosed when histologic examination demonstrated a tumor arising in the dermis that consisted of irregular, interwoven fibrocellular fascicles composed of uniform spindle-shaped cells having a storiform appearance.^8-11 The fibrosarcomatous variant of DFSP was diagnosed when there was histologic evidence of high-grade fibrosarcomatous change in 5% or more of the lesion.¹² The fibrosarcomatous areas had a fascicular, herringbone growth pattern and displayed increased mitotic activity and cellularity.¹³

Desmoid tumor was diagnosed when microscopic examination demonstrated an infiltrative, poorly circumscribed lesion composed of elongated fibroblasts and myofibroblasts characterized by elongated, tapered cytoplasm; elongated, vesicular, typical-appearing nuclei; and multiple small nucleoli. There was no cellular pleomorphism or mitotic activity. The cells were linearly arranged, forming broad bundles or fascicles in a unidirectional spreading pattern in the absence of necrosis.¹⁴ The neoplastic cells were separated and encompassed by abundant collagen.²

Surgical treatment decisions were based on review of pathologic material obtained from biopsy or prior resections and size and location of the tumor. The goal of surgical treatment was excision of the tumor with negative gross and microscopic margins of resection. During surgery, dissection of the tumor was performed through uninvolved tissue at a distance from the grossly palpable lesion, including biopsy incisions and needle biopsy tracts where appropriate. We routinely attempted to achieve a gross margin greater than 2 cm. Full-thickness resection of the tumor-containing abdominal wall with a grossly negative margin was performed when the lesion closely approximated or invaded the peritoneum. Exploratory laparotomy was performed to evaluate the intraperitoneal extent of disease if there was clinical or radiologic evidence to suggest transperitoneal disease extension. Intraperitoneal structures involving the tumor were resected en bloc when possible in an effort to obtain complete gross tumor resection. Tumors adherent to or invading rib, pubic, or iliac bone were resected en bloc with a segment of these osseus structures as indicated.

Abdominal wall reconstruction was dictated by the extent of resection and remaining residual defect. Primary tension-free closure of the abdominal wall was performed when feasible; however, in most cases reconstruction was performed using prosthetic mesh. The posterior deep fascial layer was reconstructed using absorbable mesh followed by nonabsorbable mesh reconstruction of the anterior fascial layer. The mesh was sutured in a tension-free manner to the edges of the fascial defect. Wound closure was performed following placement of a subcutaneous drainage catheter attached to a closed suction collection device. Myocutaneous rotation or free flaps was rarely indicated for wound closure (n = 4).

At the discretion of the Multidisciplinary Soft Tissue Sarcoma group or as part of clinical trials, patients were treated with adjuvant radiation therapy based on prognostic factors predicting an increased risk of local recurrence. Techniques used to deliver radiation therapy included external beam radiation therapy (n = 8) and brachytherapy with iridium-192 seed implantation at 5 days following surgery (n = 1). Total adjuvant radiation dose ranged from 4350 to 6500 cGy. One patient with desmoid tumor and no patients with DFSP received radiation therapy.

Summary statistics were obtained using established methods. Associations between categoric variables were evaluated using the Fisher exact test. Time to local recurrence and distant recurrence, recurrence-free survival, and disease-related mortality were all calculated from the date of surgery. Deaths resulting from disease were treated as an endpoint for disease-specific survival. All other deaths were regarded as censored events. Relapse-free survival was calculated from the time of surgery to any local, regional nodal, or distant disease recurrence. The rate of recurrence or death was estimated using the method of Kaplan and Meier.¹⁵ Univariate influence of prognostic factors on study endpoints was analyzed using the log rank test. No multivariate analysis was carried out owing to the moderate size of this cohort. Patients with DFSP and desmoid tumors rarely experience metastasis or die of disease^2,13,16,17; as a result, patients with these diagnoses were excluded from analyses of distant metastases and disease-specific survivals. Statistical analysis was performed using JMP software (JMP; SAS Institute Inc, Cary, NC). A P value of less than .05 was considered statistically significant.

During the study period, 40 patients visited MSKCC for treatment of abdominal wall desmoid tumors. One patient was excluded from analysis on the basis of short follow-up and incomplete clinicopathologic information. Of the remaining patients, 35 (90%) had localized abdominal wall primary, and 4 (10%) had locally recurrent tumors previously treated elsewhere (Table 1). Unlike STS, abdominal wall desmoid tumors demonstrated a predilection for the younger female patient (34 patients [87%] were women). All but 1 female developed the desmoid tumor during her childbearing years. The female-male ratio was 7:1. Median age at presentation for this cohort of patients was 34 years (range, 23-79 years). The majority (90%) was diagnosed between the ages of 23 years and 40 years. Of 33 patients with documented clinical presentation, 26 (79%) (24 primary, 2 local recurrence) had a painless abdominal wall mass, and 7 (21%) (5 primary, 2 local recurrence) had a painful abdominal wall mass. Much like the STS, desmoid tumors were hard, poorly circumscribed lesions demonstrating a locally infiltrative growth pattern. Two patients in this study had familial adenomatous polyposis. None of these patients developed clinical evidence of regional lymphatic or distant metastatic disease.

The median follow-up time was 53 months (range, 4-212 months). Three patients (8%) developed locally recurrent disease. One patient developed local recurrence after primary surgical therapy, underwent reresection with microscopic margin involvement, and remained alive 14 years later with stable subclinical residual disease. Another patient had an abdominal wall desmoid tumor 6 years following total abdominal colectomy for familial adenomatous polyposis; he developed 2 subsequent local recurrences that were resected. Eight years later, he underwent abdominoperineal resection for rectal cancer and died 4 years later of metastatic disease, clinically free of local abdominal wall desmoid tumor. The remaining patient with local disease recurrence visited MSKCC with a second local recurrence. He developed a third recurrence after resection and remains disease free 12 years after the last resection. All 3 patients who developed local recurrence had involved microscopic margins. The 5-year local recurrence–free survival (LRFS) for patients with abdominal wall desmoid tumors was 97%.

Of the 4 patients who had locally recurrent abdominal wall desmoid tumors following resection elsewhere, 3 remained disease-free after re-resection. One patient with familial adenomatous polyposis had a first local abdominal wall recurrence with synchronous unresectable mesenteric disease. This patient was free of local abdominal wall disease 33 months after resection, and there has been no mesenteric disease progression. This patient is being treated with adjuvant tamoxifen therapy. One patient visited MSKCC with first recurrence, another with second local recurrence; both were re-resected and remained disease-free 20 and 21 months later, respectively. Median time to development of first local recurrence was 19 months (range, 8-35 months). At last follow-up, 37 patients (95%) were alive with no clinical evidence of disease, and 1 is alive with disease or has died of other causes.

During the study period, 32 patients visited MSKCC for the treatment of their abdominal STS. The majority (81%) had localized primary abdominal wall sarcomas. Two patients (6%) had primary disease and synchronous metastases, 3 patients (9%) had locally recurrent disease, and 1 patient (3%) had distant metastases (Table 1). There were 12 women (38%) and 20 men (62%) ranging in age from 23 to 80 years (median, 49 years) at their initial visit. Of the 26 patients with only primary STS, the majority (77%) had a painless abdominal wall mass, typically firm and poorly circumscribed, with underlying skeletal muscle and fascial fixation.

The median follow-up duration was 31 months (range, 7-180 months). Thirteen patients (41%) developed recurrent disease. Five patients (16%) have developed local recurrences. Median time to development of first local recurrence was 10 months (range, 9-20 months). One patient has had 4 surgically treated local recurrences and remained alive with disease 77 months after initial treatment and 6 months after most recent surgical treatment. A second patient developed locally recurrent disease 10 months after primary resection, underwent reexcision and remained free of disease 81 months later. Another patient had a locally recurrent STS reexcised 9 months after primary therapy and died postoperatively of respiratory failure. The remaining 2 patients developed distant disease recurrence 6 months and 9 months, respectively, following resection of locally recurrent tumors, and both died of disease. Both of these patients had high-grade, deep, primary STS greater than 5 cm in size. Five-year LRFS for patients with abdominal wall STS was 75%. The 5-year relapse-free survivals for patients who presented with primary and locally recurrent disease were 87% and 50%, respectively.

Metastases occurred in 8 additional patients (25%) after treatment of localized primary STS, all of whom have died of disease. Median time to distant disease failure was 12 months (range, 9-75 months). The 3 patients (9%) who had distant metastases at their initial visits subsequently died of the disease. All patients who died of the disease had high-grade, deep, primary tumors, 85% of which were 5 cm or larger in size. At last follow-up, 44% of patients were disease free, 38% had died of disease, and 9% are living with the disease or have died of other causes.

Of the 3 patients with distant disease at initial visit, 2 had liver metastases in the setting of sarcomatosis, and 1 had multiple bone metastases. Ten patients developed metachronous metastases as described above, of which 5 had distant disease spread to the lung, and 1 had additional mediastinal, soft tissue, and bone metastases. Three of the remaining 5 patients had sarcomatosis, 1 with concomitant liver metastases. Two patients developed distant soft tissue metastases, 1 with synchronous ileofemoral nodal and bone metastases. Thus, 5 (38%) of 13 patients had widespread intraperitoneal disease as the type of distant disease failure, and another 38% demonstrated metachronous lung metastases.

Fourteen patients (8 women, 6 men) ranged in age from 26 to 73 years (median, 37 years) at their initial visit (Table 1). In 86% of patients, the tumor developed between the ages of 24 and 50 years. The tumor's seemingly benign appearance and indolent nature caused patients to wait extended periods of time prior to seeking medical attention. All patients in this series noted a painless abdominal wall skin nodule that prompted them to seek medical attention after a time interval that ranged from 1 to 35 years (median, 5 years). The tumor was typically a discrete, firm, dense, subcutaneous nodule adherent to the skin but unattached to underlying fascia or musculature. Twelve patients (86%) visited MSKCC for surgical treatment following diagnostic biopsy or incomplete excision elsewhere. Two patients had their first local recurrence after prior local excision elsewhere. Reexcision in the patients with margin involvement after excision elsewhere revealed residual disease in all cases. None of these patients developed clinical evidence of regional lymphatic or distant metastatic disease.

The median follow-up duration was 101 months (range, 12-187 months). During this time, 1 (7%) of these patients developed locally recurrent disease. This patient had a first local recurrence, underwent re-resection, experienced recurrence on 4 separate occasions, and was treated each time with local reexcision. The patient was free of disease 8 years following the last resection. At last follow-up, all patients were alive with no clinical evidence of disease.

Median tumor size was 6 cm (range, 0.3-12.1 cm). Virtually all patients (97%) had tumors that involved or were deep to the posterior abdominal wall fascia (Table 2). No patient had a macroscopically positive margin or incomplete resection. Eleven of 39 patients (28%) had microscopic margin involvement at pathologic evaluation.

Tumor characteristics for abdominal wall soft tissue tumors are shown in Table 2. The majority of tumors were of high histologic grade (78%), and deep (78%) to or involving the posterior investing fascia. Malignant fibrous histiocytomas (22%), fibrosarcomas (16%), and synovial sarcomas (16%) were the most common histologic variants (Figure 1). Median tumor size was 6.5 cm (range, 2.5-31.0 cm). All patients with primary or locally recurrent tumors had complete resection of grossly evident disease. Microscopic surgical resection margins were negative in 25 patients (78%), of whom 7 had no microscopic residual disease evident after re-resection. Microscopic margins were positive or unknown in 7% and 14% of cases, respectively.

There were 12 classic and 2 fibrosarcomatous variants of DFSP involving the abdominal wall in this series (Table 2). Median tumor size was 2.8 cm (range, 0.3-8.0 cm). Eleven lesions (79%) were larger than 5 cm. The majority (79%) of these tumors were located superficial to the investing fascia. Wide (≥2 cm) and narrow (<2 cm) gross margin resections were performed in 12 patients (86%) and 2 patients (14%), respectively, of whom 1 from each group had a positive microscopic margin involvement by tumor. Microscopic margin assessment showed 11 negative margins (79%), 1 close margin (7%), and 2 positive margins (14%).

Of the 39 patients with abdominal wall desmoid tumors, 3 (8%) had wide local excision involving partial abdominal wall thickness, and 25 (64%) underwent wide full-thickness abdominal wall resection (Table 3). Surgical records were not available for 2 patients. The remaining 9 patients had en bloc resection of omentum, major vein, bowel, bone, and/or other viscera. Four abdominal wall wounds (10%) were closed primarily, and 33 (85%) were reconstructed with prosthetic mesh. One patient received adjuvant radiotherapy for an extensive locally recurrent lower abdominal desmoid; he remained free of clinical evidence of disease 20 months into follow-up.

Of the 32 patients with abdominal wall STS, 1 (3%) had an unresectable local recurrence, and 3 (9%) had distant metastases. The remaining 28 patients underwent resection for primary or locally recurrent disease, including 3 (9%) who underwent wide local excision involving partial abdominal wall thickness resection, and 17 (53%) who underwent wide full-thickness abdominal wall resection (Table 3). Three had en bloc resection of either bowel and/or bone, or other viscera, and 1 patient had in continuity resection of an external iliac vein involved with tumor. Three patients had en bloc inguinal lymphadenectomy, 1 with orchiectomy as part of the resection. The method of reconstruction of the abdominal wall defect consisted of primary closure in 12 patients (38%), mesh reconstruction in 15 patients (47%), and autogenous tensor fascia lata or sartorius muscle flap coverage in 1 patient.

Nine patients (28%) were treated with adjuvant radiotherapy. Techniques used to deliver radiation therapy included external beam radiation therapy in 8 patients (25%) and brachytherapy with iridium-192 seed implantation 5 days postoperatively in 1 patient (3%). Doses ranged from 4350 to 6500 cGy. Seven of the patients received adjuvant radiation following surgical resection of primary tumors, 1 following resection of local recurrence and isolated spine metastasis. Most patients received radiation therapy for large (>5 cm), high-grade, deeply infiltrative tumors following microscopic margin negative resections. Of the 7 patients treated with primary adjuvant radiotherapy, 3 are living with no clinically evident disease.

Ten patients (31%) were treated with chemotherapy consisting of various doxorubicin-based and ifosfamide-based regimens. All had high-grade, deep tumors larger than 5 cm. Eight patients treated for metastatic disease have died of the disease after a median time of 8 months (range, 2-48 months). One patient was treated with adjuvant chemotherapy following resection of a local recurrence, and remained alive with the disease 48 months later. Another patient received adjuvant therapy after resection of a primary STS and was living disease-free 87 months into follow-up.

Of the 14 patients treated for DFSP of the abdominal wall, 7 were treated with wide partial-thickness and wide full-thickness abdominal wall resection (Table 3). Twelve (86%) had primary wound closure of their abdominal wound, and 2 (14%) required myocutaneous flap reconstruction. No patient with DFSP received radiation therapy. All patients were alive and disease free at last follow-up.

Wound-related complications occurred in 8 patients (9%). These included 2 wound seromas that resolved with percutaneous needle aspiration, 1 wound abscess requiring operative debridement and mesh removal, 1 flap necrosis requiring operative debridement and tensor fascia lata flap coverage, and 4 cases of cellulitis that resolved with antibiotic therapy. Patient follow-up and assessment of functional outcome were obtained from clinical record review, tumor registry information, private physician records, patient correspondence, and telephone interview. The patient with the wound abscess underwent delayed ventral hernia repair, and uneventful recovery ensued. One patient developed a functional disturbance following surgery. This patient developed edema in a lower extremity following radical resection of a recurrent lower abdominal sarcoma en bloc with superficial and deep groin lymphadenectomy. His functional performance decrement was estimated as less than 25%. He required a compression stocking to mollify his symptoms, but did not require narcotic analgesics for occasional dependency-related leg discomfort. The remaining patients had no functional limitations; they were able to conduct their activities of daily living unencumbered and sustain prolonged activity without performance decrement.

Univariate comparisons were performed for each endpoint listed in Table 4, and 5-year relapse-free, disease-specific, and distant disease–free survivals are listed for prognostic factors associated with statistically significant categorical differences. In the univariate analysis for abdominal wall STS, high grade, deep tumor location, and a size of 5 cm or larger emerged as adverse prognostic factors for 5-year relapse-free, disease-specific, and distant disease failure survivals (Figure 2). Patients experiencing local recurrence had significantly lower 5-year relapse-free survivals than those with primary STS (87% [primary] vs 50% [local recurrence]). Patients with soft tissue sarcomas had a significantly lower overall 5-year relapse-free survival than those with DFSP and desmoids (Figure 3). Univariate log-rank analysis of adverse prognostic factors for LRFS rate is precluded by inadequate statistical power by virtue of the relatively few local recurrences in this cohort of patients. Small sample size in the DFSP subpopulation also limits the statistical analysis of prognostic factors for LRFS.

Of the 27 patients with desmoid tumors resected with negative or close (22 and 5 patients, respectively) microscopic margins, none have developed local recurrence. Three of the 11 patients with abdominal wall desmoids undergoing positive microscopic margin resection developed local recurrence. Involved microscopic margins were associated with a higher rate of local failure (P = .01). Five-year LRFS for positive and negative microscopic margins following resection of desmoid tumors were 68% and 100%, respectively.

We have previously shown that histologic grade is a more significant predictor of outcome for extremity STS than histologic tumor subtype.¹⁸ To define further the relative influence of grade as opposed to histology on disease relapse and tumor-related mortality, we conducted a 3-tiered univariate analysis. In the first part of the analysis, all low-grade tumors (STS, desmoids, and DFSP) were compared with all high-grade STS. Tumors of high histologic grade had significantly shorter 5-year relapse-free, LRFS, distant disease–free, and disease-specific survivals than low-grade tumors (P<.001). The second part of the analysis excluded high-grade STS, comparing low-grade STS, desmoids, and DFSP according to the same outcome measures. There were no differences in 5-year relapse-free, LRFS, distant disease–free, and disease-specific rates among the 3 groups. The final analysis compared low-grade STS with low-grade non-STS (desmoids + DFSP) tumors. Similarly, there was no significant difference in outcome between the comparison groups. The stratified analyses suggest that tumor grade has a more significant influence on outcome than does tumor histology.

In this study, we sought to define the clinical behavior and prognostic indicators of outcome for abdominal wall soft tissue tumors. We evaluated desmoid tumors, DFSP, and other types of STS. The most common tumor in this patient cohort was a desmoid. A similar proportion of the lesions were STS. Thus, not all soft tissue tumors of the abdominal wall are desmoids, and sarcomas must be considered in the diagnosis and management of neoplasms of this site. The clinicopathologic factors of high-grade, deep tumor location, and large tumor size (≥5 cm) were associated with sarcoma-related mortality, and they correlated with subsequent distant metastases. It is not surprising to have similar factors predicting distant metastases and disease-specific mortality. The finding of presentation with locally recurrent disease as an adverse prognostic factor for relapse-free survival is consistent with previous observations.¹⁸ The increased risk of relapse observed with local recurrence is likely related to the risk of subsequent local failure since presentation status did not significantly influence distant failure and tumor-related mortality rates. The difference in prognostic factors for local and distant failure is in agreement with larger comprehensive analyses of outcome predictors for extremity sarcomas.^18,19 Though the limited number of microscopically positive margins following resection in this study limits formal statistical analysis, 1 study evaluating disease-specific and disease-free survival for patients with extremity sarcomas surviving for more than 5 years identified positive microscopic margin as a significant predictor of post–5-year metastases and sarcoma-related mortality.²⁰ Long-term follow-up of these patients is indicated. It seems appropriate to consider patients with large, deep abdominal wall sarcomas with high histologic grade and positive microscopic margins following resection eligible for investigational adjuvant systemic therapy.

Sarcomas of the abdominal wall are commonly included with extremity sarcomas in prognostic analyses, and they are thought to have similar factors for outcome; however, specific data to support this assumption are limited by the low incidence of abdominal wall sarcomas. Pisters et al¹⁸ have evaluated prognostic factors for 1041 patients with localized sarcomas of the extremities. They identified presentation with locally recurrent disease and positive histologic margins as independent predictors of local recurrence. Large, high-grade, deep, locally recurrent tumors were at significant risk for distant metastases and tumor-related mortality. In addition, microscopic margin involvement by tumor predicted disease-specific mortality. A definitive evaluation on the relationship between site (abdominal wall vs extremity) and subsequent outcome adjusted for clinical and pathologic factors cannot be achieved with adequate statistical power because of the relatively small number of patients with abdominal wall tumors. However, a comparison of 5-year disease-specific and recurrence-free survival rates for these sites controlling for biological factors suggests that sarcomas of the abdominal wall and extremities have similar outcome predictors and natural histories. Thus, it seems reasonable to include abdominal wall sarcomas with extremity sarcomas in the design of clinical trials.

Surgery remains the primary mode of therapy for STS of the abdominal wall. The principal objective of surgery is complete removal of the tumor with dissection at a distance from the lesion through uninvolved tissue. A 2- to 3-cm margin of normal tissue is recommended in light of the propensity for clinically inapparent local tumor spread. In addition, full-thickness abdominal wall resection is indicated in most cases to obtain adequate local control. Partial-thickness abdominal wall resection is occasionally appropriate for the small (<5 cm), low-grade sarcomas located superficial to the abdominal fascia. In general, for lesions closely approximating the peritoneum, full-thickness resection is indicated. When the question arises as to the adequacy of tumor clearance in proximity to bony structures, it is advisable to resect adjacent segments of rib, pubis, or iliac crest en bloc. Reconstruction of the abdominal wall defect can readily be achieved with prosthetic mesh achieving satisfactory functional outcome in most cases.

The limited number of patients who received radiotherapy in this study does not permit analysis of the impact of radiation on local and distant tumor control. A randomized prospective trial conducted at our institution compared the effect of adjuvant brachytherapy with that of no further therapy following complete resection of STS involving the extremity or superficial trunk. The study revealed that brachytherapy significantly improved local disease control after complete resection of high-grade sarcomas.²¹ However, this improvement in local control did not translate into reduction of distant metastases or improvement in disease-specific survivals. There were relatively few patients with truncal sarcomas (n = 23) in that study, and the local control data for extremity lesions could not be extrapolated to truncal tumors on this basis. Our policy is to incorporate adjuvant external beam radiotherapy as part of the treatment of large (≥5 cm) low-grade sarcomas, and adjuvant brachytherapy is applied to tumors of high histologic grade. Admittedly, the evidence that tumors of low histologic grade derive the same local control benefit from adjuvant external beam radiation that high-grade tumors do from brachytherapy is lacking.

Dermatofibrosarcoma protuberans is a rare, indolent, low-grade sarcoma recognized for its progressive, locally infiltrative nature. A painless, long-standing, slow-growing subcutaneous nodule represents the most common presentation of this disease. The seemingly benign appearance and behavior of the tumor often cause patients to wait extended periods prior to seeking medical attention. Successful treatment of this disease centers on wide local excision to assure adequate surgical margins. Since the majority of abdominal wall DFSP are small (<5 cm) and superficial, they can often be treated with wide local excision and primary closure without the need to resect full abdominal wall thickness. When doubt arises as to the adequacy of the gross resection margin, particularly at time of reexcision or treatment of recurrent tumors, extending the scope of resection to include full-thickness excision is advisable.

The most significant factor that predicts local control of DFSP is extent of resection. Local control of DFSP is related to gross surgical margin. The difficulty in consistently achieving negative microscopic margin resection of DFSP is attributable to the histologic characteristics of this tumor, with peripheral clinically inapparent fingerlike projections of tumor cells that resemble normal collagen, inconspicuously blending into the tumor's periphery.^8-11 The high local recurrence rates following surgical treatment are likely attributable to the failure to excise these tumor projections that make definitive histologic margin assessment challenging.^22,23

Minimum gross resection margin to assure local control is undefined, but most authorities would agree that a margin of normal tissue between 2 cm and 3 cm from the gross tumor boundary including skin, subcutaneous fat and underlying fascia must be achieved to ensure adequate surgical extirpation.¹¹ In the current series, 1 of 12 patients who underwent wide gross margin resection (≥2 cm) and 1 of 2 undergoing a less than 2 cm gross margin resection had microscopic margin involvement by tumor. In a recent series of 33 patients with DFSP of the head and neck, we found that a gross resection margin of or greater than 2 cm was a significant predictor of negative histologic margin.¹⁶ No local recurrences developed after wide margin resection (≥2 cm), whereas the local recurrence rate following resection with margins smaller than 2 cm was 19%.

Desmoid tumors are histologically benign neoplasms characterized by slow, locally invasive growth. They have no metastatic potential. Surgical resection remains the treatment of choice for abdominal wall desmoids. Because of their notoriously aggressive behavior, they have a strong proclivity for local recurrence following resection, with reported rates ranging from 24% to 77%.^2,4,24,25

The association of microscopic margin involvement following resection with local failure is controversial. Several studies have addressed this issue with variant conclusions.^2-6 One large series⁴ of a heterogeneous population of patients with desmoid tumors (based on site and mode of presentation) found that presentation with locally recurrent disease and microscopically positive margins following surgical resection were independent prognostic factors for local recurrence. A recent study² of extremity and trunk desmoid tumors included 21 patients with abdominal wall lesions. Positive resection margins in that series were not predictive of local recurrence of desmoid tumors involving the trunk or abdominal wall.

The natural history of desmoid tumors remains enigmatic. This is underscored by series that report no tumor progression and even regression when these neoplasms are managed with observation only.^24,26 Rock et al²⁴ reported clinical stability in 60, and regression in 6 of 68 patients with desmoid tumors observed during a 6-year period. In our population of abdominal wall desmoid tumors, we have found microscopically positive margins following resection to be associated with the development of subsequent local recurrence.

Our findings support the overall favorable natural history of abdominal wall desmoid tumors and suggest the importance of negative margin resections in achieving local control of these tumors. Surgical principles that apply to the treatment of abdominal wall desmoid tumors are similar to those for STS. Wide margin resection should involve biopsy incisions and needle biopsy tracts where applicable. Full-thickness resection of the tumor-containing abdominal wall with grossly negative margin is recommended when the lesion closely approximates or involves the peritoneum. Full-thickness abdominal wall resection is generally required to completely resect desmoid tumors, for they characteristically involve the deep investing fascia. Exploration of the abdominal cavity is indicated to determine the extent of intraperitoneal disease when clinical or radiologic findings suggest transperitoneal disease extension. Intraperitoneal organs or adjacent bony structures involved by tumor ought to be resected en bloc when possible, in an effort to obtain complete tumor resection. Abdominal wall reconstruction most often requires and is readily achieved through the use of prosthetic mesh to achieve a tension-free closure.

Pulmonary metastases are a common site for distant disease failure of extremity sarcoma.²⁷ In the current study, 38% of patients developed lung metastases, and 38% had sarcomatosis. High-grade, large (≥5 cm), extremity, and superficial trunk sarcomas are at high risk for distant metastases.²⁷ Since the lung is a significant target site of distant disease spread, it is a major focus for screening studies for disease recurrence in the follow-up of patients. The notable incidence of peritoneal spread of disease for STS of the abdominal wall merits consideration of staging radiographic imaging of the abdomen at the time of initial visit.

In this study, the clinical behavior and management of soft tissue tumors of the abdominal wall were evaluated. Not all abdominal wall neoplasms are desmoids, and sarcomas must be considered in the differential diagnosis. Abdominal wall desmoids, STS, and DFSP demonstrate a broad spectrum of biologic behavior. Initial biopsy is usually mandatory to facilitate treatment planning. Desmoid tumors remain exclusively a local problem. Dermatofibrosarcoma protuberans of the abdominal wall is virtually always a local issue, with rare distant metastases associated with its fibrosarcomatous variant. Prognostic factors associated with adverse survival outcome are defined for abdominal wall sarcomas, and they are identical to factors predicting outcomes for other sites. The treatment approach for all 3 diseases remains to be aggressive, complete surgical resection to achieve local control. Repair of abdominal wall defects, with excellent functional results, can be sufficiently achieved with prosthetic mesh reconstruction. Stratification on the basis of prognostic factors for distant disease failure and tumor-related mortality will facilitate selection of patients with STS for adjuvant systemic therapies.

This study is supported by US Public Health Service grant CA47179.

Corresponding author and reprints: Murray F. Brennan, MD, Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021 (e-mail: brennanm@mskcc.org).