Desmoid tumors (also known as aggressive fibromatoses) were first described as arising from the anterior abdominal wall6. In 1923, Nichols reported on extra-abdominal desmoid tumors that had developed at other sites, and it is now recognized that desmoid tumors can arise from any mesenchymal tissue.
These tumors are extremely difficult to treat. They are relatively rare and, hence, they are seen infrequently by most physicians. Because they are locally aggressive, some physicians have considered them to be low-grade sarcomas10. Such an assumption may lead to overtreatment. In contrast, some tumors may be undertreated, and this may lead to problems with local control.
In a previous article12, one of us (D. J. P.) and colleagues reviewed the cases of 194 patients who had been managed for an extra-abdominal desmoid tumor at a tertiary-care referral center between 1908 and 1980. In that study, 132 patients (68 per cent) had a local recurrence at an average of 1.4 years after the first treatment. Girls and women were more likely to have a local recurrence, as were patients who were more than thirty years old. An increased risk was associated with certain anatomical locations, especially the foot and calf, and with an intralesional or a marginal resection12.
The present study was undertaken to review our more recent experience with extra-abdominal desmoid tumors.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Read in part at the Annual Meeting of The American Orthopaedic Association, Sun Valley, Idaho, June 6, 1994.
‡Department of Orthopedics (D. J. P.), Division of Anatomic Pathology (A. G. N.), and Division of Radiation Oncology (I. A. P.), Mayo Clinic, 200 First Street S.W., Rochester, Minnesota 55905. Please address requests for reprints to Dr. Pritchard.
We retrospectively reviewed the records of 103 patients who had had an extra-abdominal desmoid tumor in the limb, limb girdle, or trunk and had initially been managed between 1981 and 1989. During that period, twenty-five patients were seen for consultation only and were not managed at our institution, forty-five patients had Gardner syndrome and an intra-abdominal tumor, and thirteen patients had a desmoid tumor of the abdominal wall and were managed by general surgeons; none of these patients were included in the study. Of the 103 patients, fifty-two had been managed elsewhere and had a recurrent tumor when they were first seen at our institution. One patient who had a non-resectable lesion was lost to follow-up after four months and was not included in the series. The remaining fifty patients had an untreated desmoid tumor and were followed for at least two years (average, forty-eight months). Follow-up information was obtained for all patients by direct examination or by letter.
Specimens from each patient were available for histological examination and were reviewed by one of us (A. G. N.), a pathologist. Macroscopically, each tumor was seen to be a mass with a white hard-cut surface. The greatest dimension of the smallest tumor was 1.8 centimeters and that of the largest one was thirty centimeters. The mean diameter of the tumors was 8.9 centimeters (median, eight centimeters; mode, thirteen centimeters).
Each tumor consisted of a proliferation of spindle-shaped fibroblasts with plump nuclei and mildly eosinophilic cytoplasm (Fig. 1). The nuclei were clear; the chromatin was distributed along the nuclear membrane and the nucleoli were inconspicuous. The tumor cells were arranged in parallel rows and were separated from one another by a variable but usually large amount of collagen. However, twenty-one of the tumors showed an unusually high degree of cellularity; in twenty of them, the collagenous extracellular matrix was scanty. Most of the tumors lacked mitotic activity; an occasional mitotic figure was identified in only ten. No tumor displayed necrosis.
The family history and the patient's history, especially regarding trauma, were recorded for all patients. The results of routine blood studies were noted as well, and plain radiographs were available for all patients. The findings of magnetic resonance imaging were documented because we were interested in whether the use of that modality for diagnosis improves the results of operative treatment. We also reviewed the operative records and classified the resection as intralesional, marginal, or wide, according to the recommendations of the Musculoskeletal Tumor Society2. The records on radiation therapy also were reviewed. Any adjuvant systemic therapy was also noted. Systemic and radiation treatment in addition to the operation had been prescribed at the discretion of the treating physician; there had been no set standard of treatment.
Of the fifty patients who were initially managed at our institution, nineteen were male and thirty-one were female (male:female ratio, 1:1.6). The average age at the time of the diagnosis was forty years (range, nine to eighty-three years). Eight patients (16 per cent) had had an episode of trauma to the site of involvement before a mass or pain had developed. In addition, four patients had had a previous operation, with subsequent development of tumor in the scar at the operative site. Forty-nine patients felt a mass, twenty-four had pain, and four had decreased mobility of an adjacent joint. Four patients had neurological symptoms due to nerve compression. The average duration of symptoms was six months (range, one month to ten years). One patient was known to have had a mass for ten years before the diagnosis. Thirty-one patients had had a biopsy before they were referred to us, but they had not had any additional treatment.
The anatomical distribution of the tumors was wide (Fig. 2). The region of the shoulder, including the axilla, was the most common site. There were few tumors in distal locations. Four patients had two separate lesions in the same general location, and one had two separate lesions at sites remote from each other (both thighs). Forty-eight of the tumors were deep-seated, and two were considered to be superficial.
In the preoperative assessment, no imaging modality could accurately delineate the boundaries of the tumor. In the early years of the study, computerized tomography generally was performed, but the images usually were not adequate to determine the precise extent of the lesion and its relationship to surrounding muscle. In the later years of the study, magnetic resonance imaging was done (twenty patients). We believe that magnetic resonance imaging provides more accurate definition of the margins of the tumor and is therefore more helpful for the planning of the resection and radiation therapy.
Initial Treatment
Two patients who had a desmoid tumor in the popliteal space and one who had a tumor in the axilla were managed with biopsy and observation only. These lesions were thought to be non-resectable. One of the patients who had a popliteal tumor was followed for eleven years, and the lesion had caused no symptoms and had not increased in size. The other popliteal tumor was still stable after forty-nine months of observation. The patient who had the axillary lesion was observed for more than four years (until 1989), but no information was available after that time.
Three patients had radiation therapy only. One of them had a large non-resectable tumor of the axilla that was treated with 5047 centigray; the tumor remained stable for two years and then began to grow. An intralesional debulking procedure was performed. The patient was followed for an additional two years, and the disease was stable but caused symptoms. Another patient had an axillary lesion that involved the brachial plexus; a cervical laminectomy and an operation on the carpal tunnel had been done before the desmoid tumor was found. This patient was managed with 6120 centigray of radiation. The patient was followed for fifty-six months, at which time the disease was stable and caused few symptoms. Another patient who had an axillary lesion initially was observed for one year, during which there was evidence of tumoral progression; 6480 centigray of radiation then was administered. The patient was followed for four years, during which time she received hormonal treatment with medroxyprogesterone (Depo-Provera) and tamoxifen, but the tumor gradually increased in size. At the time of the latest follow-up, the patient had edema and occasional episodes of paresthesia. Another patient, who was followed for four months but was not included in this study, had radiation therapy (6600 centigray) only for a non-resectable lesion in the forearm. At the time of the latest follow-up evaluation, the size of the lesion had continued to decrease.
Of the remaining forty-four patients, thirty-four were managed with an operation only and ten, with an operation and radiation therapy (Table I). Thirty (68 per cent) of the forty-four patients were free of disease at five years. The tumor did not recur in twenty-two (65 per cent) of the patients who had had operative management only: eleven of the thirteen who had had a wide resection, ten of the nineteen who had had a marginal resection, and one of the two who had had an intralesional resection (Table I). No patient had had an amputation.
Of the ten patients who were managed with an operation and radiation therapy, nine received radiation therapy postoperatively and one, preoperatively and postoperatively. The resection was wide in two patients, marginal in four, and intralesional in four (Table I). Eight of these ten patients were free of local recurrence at five years; one patient who had had a marginal resection and one who had had an intralesional procedure had a local recurrence.
Local Recurrence
Over-all, a local recurrence developed in fourteen (32 per cent) of the forty-four patients, at an average of eighteen months (range, five to twenty-eight months). There were two local recurrences in one extremity: one proximal and one distal to the original site. No patient died of the disease.
The recurrent lesion was treated with an operation or radiation therapy, or both, in thirteen patients (Table II). In one other patient, it was treated at another institution, and the type and result of treatment were unknown. Three of the six patients who had only an operation for the local recurrence had a second local recurrence. Of the seven patients who had radiation therapy alone or in conjunction with an operation, only one had a second local recurrence; thus, nine patients were free of local recurrence at the latest follow-up examination. None of the thirteen patients had an amputation.
Desmoid tumors are non-metastasizing, locally aggressive soft-tissue tumors11. They frequently invade normal structures and are associated with a high rate of morbidity. Histologically, they are classified as benign, but their locally invasive and destructive behavior is similar to that of low-grade fibrosarcomas.
Most desmoid tumors arise from extra-abdominal sites, including the extremities, the chest wall, and the head and neck. Intra-abdominal (mesenteric) and abdominal sites are associated with familial polyposis coli, especially Gardner syndrome. Also, it is not uncommon to find desmoid tumors in the abdominal wall post partum. Although the lesion can affect individuals of all ages, patients usually are between twenty and forty years old. In most series, the majority of patients have been women.
The local control of extra-abdominal desmoid tumors can be difficult. In a previous report12, which included patients who were managed between 1908 and 1980, one of us (D. J. P.) and colleagues noted that 132 of 194 patients who had had an extra-abdominal desmoid tumor had a local recurrence, an average of 1.4 years after the first treatment. Over-all, 65 per cent had no evidence of disease after as many as ten courses of treatment. One hundred and twenty (90 per cent) of the 134 patients who had had an intralesional or a marginal resection had a local recurrence or residual disease, whereas twenty-five (48 per cent) of the fifty-two patients who had had a wide resection had a local recurrence or residual disease. It also was noted that, with each additional operative treatment, the rate of so-called cure was only about 20 per cent.
Recent studies have suggested that the margin of resection on histological examination accurately reflects the true extent of resection and the subsequent risk of local recurrence1,4,7,8,10,13. Unfortunately, the exact risk of local recurrence in any one patient is difficult to determine because investigators have studied small numbers of patients, grouped patients who had primary disease with those who had recurrent disease, included different definitions of margins, grouped together intra-abdominal and extra-abdominal sites, or included patients who had adjuvant therapy, such as irradiation, a course of tamoxifen, or chemotherapy. A review of studies (including the present one) in which the margins of resection were indicated after gross total resection of a mesenteric or abdominal desmoid tumor demonstrated that the risk of local recurrence is lower when the operative margins are negative for tumor cells than when they are close (less than two millimeters from positive tissue) or histologically positive (Table III)1,4,7,8,10,13. Sherman et al. reported a local recurrence in one of nine patients who had negative or uncertain margins after operative management and adjuvant radiation therapy. McCollough et al. reported good local control of eleven of fourteen lesions that were treated with adjuvant radiation therapy for positive, questionable, or reactive margins. These results were for both primary and recurrent disease. Posner et al. conducted multivariate analysis of 138 patients and found that the margin of resection on histological examination was the most important factor influencing local control. Furthermore, positive operative margins were more likely to be found after more limited resections. In contrast to these findings, Miralbell et al. reported local control in seventeen (81 per cent) of twenty-one patients in whom the margins were histologically positive or close after resection of a primary desmoid tumor.
The combined results of multiple studies (including the current report) show that radiation therapy, with or without subtotal resection, resulted in local control in sixty-five (78 per cent) of eighty-three patients who had gross disease (Table IV)3-5,8,10,13. Radiation therapy alone is preferred for patients who have an inoperable tumor or gross residual disease after operative debulking and for those in whom operative management would lead to major morbidity or loss of function.
It has been well established that radiation therapy in conjunction with more conservative operative treatment can help in the local control of sarcomas in the extremities. Therefore, patients who have a desmoid tumor would be expected to have a similar benefit15,19. Radiation therapy has not been used consistently in most series, which makes it difficult to evaluate the most appropriate application. Several reports have suggested that the addition of radiation therapy improves local control after resection for gross residual disease3. However, the value of radiation for patients who have close or histologically positive margins is less clear. McCollough et al. reported excellent local control in patients who had radiation therapy for histologically evident residual disease after resection for primary or recurrent disease. These results are similar to those of Miralbell et al. after operative management only.
The effect of chemotherapy on desmoid tumors has not been tested adequately because of the indolent and non-metastasizing nature of the disease. Reports of the results of chemotherapy with agents such as vincristine, dactinomycin, cyclophosphamide, methotrexate, and doxorubicin have been primarily anecdotal17. It has been unclear, in these reports, how many other patients did not have improvement after chemotherapy. The results of hormonal therapy (primarily with tamoxifen, an anti-estrogen) for desmoid tumors have also been reported anecdotally14,18. This therapy is based on the observation that estrogen may be a growth factor for fibroblasts and fibroblastic tumors. A substantial number of desmoid tumors occur in women during their child-bearing years or in association with pregnancy, which also supports the hormonal relationship. Although the reports are encouraging, without information about the true rate of response this treatment also is investigational.
From the previous study12, we learned that observation is sufficient for the treatment of some desmoid tumors. Patients who had had numerous, extensive procedures and did not want additional treatment chose to have periodic monitoring. In sixty of sixty-eight patients who made this decision, the size of the tumor remained stable and caused few symptoms during an average of 6.3 years of follow-up. In six of the patients, the tumor became smaller during the period of observation; it progressed, necessitating additional treatment, in only two patients.
The present study provides evidence that the ability to control local disease has improved. Better imaging techniques may have contributed to better preoperative planning; however, magnetic resonance imaging was used only in the latter half of the study period and the technique of magnetic resonance imaging and the interpretation of scans has undoubtedly improved (Figs. 3-A and 3-B). Hence, no definitive statement can be made about the usefulness of this modality5. The discovery of synchronously appearing multicentric extra-abdominal desmoid tumors with magnetic resonance imaging has been reported16.
In summary, desmoid tumors are locally aggressive and difficult to treat. Adequate resection offers the best chance for a so-called cure. Adjuvant radiation therapy is probably beneficial, but the precise indication for its use is not well defined.