Primary Leiomyosarcoma of Extragnathic Bone: Clinicopathologic Features and Reevaluation of Prognosis

Leiomyosarcoma (LMS) is a malignant neoplasm with histologic, immunophenotypic, and ultrastructural evidence of smooth muscle differentiation. Excluding cutaneous LMS, most cases involve the female genital tract or, less commonly, the retroperitoneum, abdominal and pelvic cavities, or deep soft tissues of the extremities. Rarely, LMS may present as a primary tumor of bone, where it is hypothesized to arise from intraosseous blood vessels, pluripotent mesenchymal stem cells, or intermediate cellular forms such as myofibroblasts.1–4 Since the initial report by Evans and Sanerkin5 in 1965, 131 cases of primary LMS of extragnathic bone have been reported in the English literature. Because it is so uncommon, however, it is likely that primary LMS of bone is often misdiagnosed as undifferentiated pleomorphic sarcoma (previously known as malignant fibrous histiocytoma), fibrosarcoma, fibroblastic osteosarcoma, rhabdomyosarcoma, myofibroblastic sarcoma, or other unspecified sarcoma of bone. In contrast, some reported cases of LMS of bone appear to be better classified as one of these other unusual sarcomas of bone. Based on these prior reports, LMS of bone has been considered to portend a dismal prognosis.6–12 We present 3 cases and review the clinical, radiographic, and pathologic characteristics of this tumor. In addition, the prognosis of primary LMS of extragnathic bone is reevaluated, considering only those reported cases in which the published histomorphologic, immunophenotypic, and/or ultrastructural data are conclusive of smooth muscle differentiation.

A computerized search of the surgical pathology files at Vanderbilt University Medical Center (Nashville, Tennessee) from 1999 to 2008 yielded 3 cases of primary LMS of bone, after metastases were excluded. These cases were reviewed and clinical data were recorded from the patient's electronic medical record. A literature search for primary LMS of extragnathic bone was conducted using MEDLINE. Key words used for this search were “leiomyosarcoma” and “bone.” The search was limited to reports written in the English language. Cases arising in the craniofacial bones and those determined to be metastases from other primary sites were excluded. Each case report or series was reviewed for pertinent clinical, radiographic, and pathologic data. Twenty-seven of the 131 reported cases of primary LMS of extragnathic bone were excluded because of inadequate illustrational evidence of smooth muscle differentiation defined by standard histomorphologic, immunophenotypic, or ultrastructural criteria.13,14 

For purposes of comparison, all tumors were restaged according to the 6th edition of the American Joint Committee on Cancer TNM staging system for bone sarcomas.15 Histologic grades were recorded using a binary grading system (low or high grade) by conversion from the reported 3- or 4-tiered system grade (intermediate grade sarcomas in the 3-tiered system were considered high grade). For cases in which histologic grade was not explicitly reported, photomicrographs and accompanying histologic descriptions were reviewed and assessed for French Federation of Cancer Centers Sarcoma Group grading criteria (degree of differentiation, mitotic activity, and tumor cell necrosis).16 Inadequate data were published for accurate grading of 22 of the 104 cases reviewed in this series. The prognostic value of clinical and pathologic parameters, including age, gender, anatomic location, associated pathologic fracture, histologic grade, American Joint Committee on Cancer stage, and local recurrence was evaluated by standard univariate methods using InStat v3.06 and Prism v5.01 (GraphPad Software, Inc, La Jolla, California).

A 34-year-old woman presented with a 1-year history of right knee pain. Radiographs showed a 6 × 4-cm osteolytic lesion with indistinct borders in the metaphysis of the distal femur (Figure 1, A). Magnetic resonance imaging revealed a 7-cm lesion that was hypointense on T1- and hyperintense on T2-weighted imaging; soft tissue extension was also observed. An incisional biopsy revealed atypical spindle cells suggestive of high-grade sarcoma. Immunohistochemical stains demonstrated strong and diffuse reactivity for muscle-specific actin (HHF-35), α-smooth muscle actin (1A4), and desmin. The diagnosis of LMS was made. Subsequent staging studies revealed no evidence of pulmonary metastasis. Six cycles of mesna, ifosfamide, Adriamycin, and dacarbazine were administered before wide surgical excision was performed. Eighteen months postoperatively, there is no evidence of disease recurrence.

A 46-year-old man presented with acute left knee pain and swelling. Radiographs showed a 6-cm osteolytic lesion involving the medial tibial plateau with associated soft tissue swelling and pathologic fracture. Incisional biopsy revealed fascicles of spindle cells with mild to moderate nuclear atypia. Immunohistochemical stains were positive for muscle-specific actin, α-smooth muscle actin, and desmin, consistent with LMS. The patient underwent 2 cycles of neoadjuvant Adriamycin and cisplatin. After results of staging studies were negative for distant disease, wide resection of the tumor was performed and followed by reconstruction with a tibial osteoarticular hemiallograft. Within 18 months, the allograft failed and he was converted to a total knee replacement. He has no signs of local recurrence or metastasis 8 years after surgery.

A 42-year-old woman complained of right ankle pain for 1 year. Radiographs demonstrated a solitary osteolytic lesion occupying the distal 6.5 cm of the fibula. Magnetic resonance imaging confirmed the presence of soft tissue extension involving the tibiofibular syndesmosis and peroneal tendons (Figure 1, B). Incisional biopsy revealed a low-grade LMS. Immunohistochemical studies confirmed the diagnosis. Staging studies revealed no distant disease. Radical resection and ankle arthrodesis was performed. The presence of coagulative tumor necrosis, significant cytologic atypia, and conspicuous mitotic activity observed on sampling the resection specimen prompted alteration of the diagnosis to high-grade LMS. Currently, the patient is undergoing adjuvant chemotherapy.

Primary LMS of bone is extremely rare, with only 107 well-documented cases reported, including the current series (Table 1). The incidence of this tumor is highest between the fourth and seventh decades (median age, 47 years; range, 9–87 years). Males and females are affected equally. Most patients (77%) cited pain as their primary complaint. Swelling was also a common presenting symptom. Sixteen of 80 evaluable cases (20%) presented with pathologic fracture. Similar to the more common primary sarcomas of bone, LMS most often involves the appendicular skeleton (Table 2). Potential risk factors were identified in approximately 10% of cases (Table 3).

Plain radiographs characteristically show a solitary osteolytic lesion with a permeative appearance, indistinct margins, irregular cortical erosion, and minimal to no sclerosis or periosteal reaction. In the long bones, most cases are metaphyseal, with occasional extension into the diaphysis or epiphysis. Although intralesional calcifications are not considered typical radiographic features of LMS of bone or soft tissue, a recent study demonstrated that nearly 20% of these tumors may contain areas of calcification, mimicking more commonly mineralized neoplasms such as osteosarcoma.17 Computed tomography scans typically show an intramedullary lesion with associated cortical destruction and extension into the surrounding soft tissues. These findings can also be demonstrated by magnetic resonance imaging, in which the lesion is usually hypointense or isointense relative to skeletal muscle on T1-weighted imaging and hyperintense on T2-weighted imaging. The radiographic characteristics of histologically low-grade LMS of bone may be slightly different. The lesional margins are usually more distinct and may often be delimited by a surrounding rim of sclerotic bone, thereby mimicking benign tumors such as nonossifying fibroma or giant cell tumor of bone.1,18–20 

Grossly, LMS of bone often appears well circumscribed and is usually firm and white to gray, with areas of tumor necrosis or hemorrhage in some cases (Figure 2). The histologic features of LMS of bone are identical to LMS arising in more common anatomic locations.13,14 The classic pattern is that of fascicles of spindle-shaped cells, intersecting at wide, often perpendicular angles (Figure 3, A). Tumor cells have blunt-ended, cigar-shaped nuclei and abundant, eosinophilic, often fibrillary cytoplasm. Occasionally, multinucleated giant cells may be seen.1,9,12,18,21–25 Despite the gross circumscription of these tumors, extensive infiltration of adjacent trabecular bone is seen on microscopic examination (Figure 3, B). Importantly, there is no associated osteoid or chondroid matrix deposition by tumor cells. Most primary LMS of bone are of high histologic grade; only 23 of 85 (27%) previously reported cases were low grade.

Immunohistochemistry is often required for accurate diagnosis, particularly for poorly differentiated anaplastic tumors. Muscle-specific actin was positive in 56 of 60 reported cases (93%), and α-smooth muscle actin was positive in 59 of 62 cases (95%). Desmin staining can be helpful, but only about half of the reported cases (42 of 74) were positive for it (Figure 4, A). It should be noted that these markers are not specific for smooth muscle and are positive in myofibroblastic sarcoma and pleomorphic rhabdomyosarcoma. Focal or weak staining may also be seen in undifferentiated pleomorphic sarcoma and osteosarcoma.26,27 H-caldesmon and smooth muscle myosin may be used to demonstrate smooth muscle differentiation and exclude myofibroblastic sarcoma but have not been extensively studied in this diagnostic setting.8,28–30 In equivocal cases, ultrastructural examination can confirm the presence of smooth muscle differentiation, particularly when myofibroblastic sarcoma is a diagnostic consideration.28,29,31 

Cytokeratin expression was described in 10 of 29 cases (34%) of LMS of bone. However, 8 of these 10 keratin-positive cases are from 2 studies published more than 10 years ago in which the immunohistochemical protocols are not described in detail.23,24 Similarly, 4 of 5 reported S100-positive cases were presented in a single report.12 Only one of an additional 20 cases stained for S100 protein was reportedly positive.10 

The cytopathologic features of LMS are dependent on the histologic grade of the neoplasm. Well-differentiated tumors often yield hypocellular smears containing fascicles of spindle cells with relatively abundant dense fibrillar to granular cytoplasm, elongated, blunt-ended nuclei, and minimal nuclear atypia.32 Single intact cells are usually scant. In contrast, aspirates of high-grade LMS are typically cellular and composed of many single or small, disorganized groups of variably sized spindle cells that show pronounced nuclear pleomorphism, hyperchromasia, and prominent nucleoli (Figure 4, B). Sanerkin2 initially described the presence of numerous binucleate and multinucleated giant cells in fine-needle aspirates. Evidence of smooth muscle differentiation in high-grade LMS may be extremely difficult to discern in cytologic specimens without the assistance of ancillary studies.

Before definitive therapy is initiated, the possibility of metastasis from another site of origin, most commonly the gynecologic tract or retroperitoneum, must be excluded. Radiographic features that would suggest the possibility of metastasis include multiple, small tumors or involvement of unusual anatomic sites, such as the axial skeleton or proximal femur.24,33–35 Although LMS arising in soft tissue could theoretically involve adjacent bone, concern for this possibility should be extremely low, as even high-grade soft tissue sarcomas tend to follow the path of least resistance and are more likely to slowly erode cortical bone rather than aggressively invade the underlying medullary cavity.36 

Once the diagnosis of primary LMS of bone is confirmed, the treatment of choice is wide surgical resection. Approximately 20% of reported cases were treated by amputation; however, most of these cases predated limb salvage techniques. Both chemotherapy and external beam radiation therapy have been used to augment surgical treatment, although the efficacy of chemotherapy has not been confirmed. Some authors have reported that LMS of bone is relatively radioresistant.1,11 For example, in their review of 33 cases (26 of which were extragnathic), Antonescu et al1 were not able to demonstrate a therapeutic benefit of adjuvant therapy in the treatment of primary LMS of bone. Efficacy of adjuvant therapy was difficult to evaluate in this series because of the lack of uniformity in treatment regimens and the limited number of cases available for comparison. Therefore, recommendations regarding adjuvant therapy for primary LMS of bone are difficult to substantiate.

Of the 49 cases for which tumor stage could be assigned, 7 patients (all with high-grade LMS) presented with metastatic disease and 1 patient with low-grade LMS had discontinuous tumors within a rib (stage III). Similar to other primary sarcomas of bone, the most common site of metastasis is the lung; skeletal metastases (particularly to the axial skeleton) were also common (Table 4). Univariate Kaplan-Meier analysis of the reported cases demonstrated that tumors of advanced American Joint Committee on Cancer TNM stage and high histologic grade are associated with adverse outcomes (Table 5). Patients with stage I and IIA tumors show prolonged overall and disease-free survival compared with those with stage IIB, III, or IV tumors (Figure 5, A). A post hoc comparison limited to patients with nonmetastatic, high-grade LMS confirmed that patients with stage IIA tumors (≤8 cm in greatest dimension) have longer overall survival compared with patients with stage IIB tumors (>8 cm in greatest dimension; log-rank test, P = .007).

High histologic grade also correlated with decreased overall and disease-free survival (Table 5; Figure 5, B). Among cases with a reported follow-up duration of at least 2 years, 24 of 47 patients (51%) with high-grade tumors died a median of 19 months after definitive surgery (range, 1–72 months). An additional 11 patients (23%) were alive with disease at last follow-up. Distant metastasis occurred in 27 patients (57%) with high-grade LMS (median interval to metastasis, 12 months; range, 3–84 months). During comparable follow-up intervals, 4 deaths (29%) occurred in 14 patients with low-grade LMS (median survival interval, 65 months; range, 18–115 months). Two additional patients (14%) with low-grade LMS were alive with metastatic disease 30 and 72 months after surgery. Six patients (43%) with low-grade LMS developed metastatic disease (median metastasis-free interval, 36 months; range, 12–78 months).

Local recurrence was relatively uncommon, occurring in 15% of patients with high-grade LMS and 14% of patients with low-grade LMS, usually within 3 years of initial surgical resection. Locally recurrent disease had no discernible impact on overall survival (log-rank test, P = .73). Although tumors arising in the axial skeleton correlated with decreased overall survival, this analysis was confounded by the tendency for axial tumors to present in advanced American Joint Committee on Cancer stage (Fisher exact test, P = .007). Patients with postradiation LMS had significantly shorter median overall survival (15 months) than those with tumors not arising in previously irradiated fields (115 months; log-rank test, P = .01). Due to the small number of cases available for comparison, it is unclear whether LMS of bone arising in the setting of previous chemotherapy, Paget disease of bone, or adjacent to orthopedic implants portends a worse prognosis, as demonstrated for other primary bone sarcomas.37–40 There were no statistically significant differences in either overall or disease-free survival based on age, sex, or associated pathologic fracture.

Presentation of LMS as a primary bone tumor is extraordinarily uncommon. Despite publication of one relatively large series, the clinical, radiographic, and histopathologic features of these tumors and their associated prognostic factors are not well defined.1 The radiologic differential diagnosis is often broad and includes fibroblastic osteosarcoma, dedifferentiated chondrosarcoma, or undifferentiated pleomorphic sarcoma. Depending on the dominant morphologic features, the pathologic differential diagnosis can also be quite diverse. Immunohistochemical or ultrastructural analysis is required to confirm the diagnosis in many cases.

Primary LMS of bone has traditionally been considered to have a dismal prognosis.6–12 However, this conclusion was based on a limited number of case reports and small series, some of which were poorly documented. In this review, the 5-year overall survival rate of LMS of bone (59%) is similar to those of conventional intramedullary osteosarcoma and Ewing sarcoma (each approximately 50%) and chondrosarcoma (68%).41 

We thank Jean McClure for editorial assistance.