Pelvic osteosarcomas are difficult to resect. The authors reviewed their institution's experience with patients who had such tumors to characterize the patients' clinical findings and to assess the impact of surgical resection on outcome.
Pelvic osteosarcomas are difficult to resect. The authors reviewed their institution's experience with patients who had such tumors to characterize the patients' clinical findings and to assess the impact of surgical resection on outcome.
A review was conducted of the records from patients with pelvic osteosarcoma who were treated at the authors' institution between January, 1970 and March, 2004.
Among 442 patients with osteosarcoma, 19 patients (4%) had high-grade tumors arising in the pelvic bones, including the ilium in 15 patients, the pubis in 2 patients, and the sacrum in 2 patients. The median patient age at diagnosis was 16.8 years. Four tumors were secondary to radiation therapy. Five patients had metastases in the lung (n = 4 patients) or bone (n = 1 patient) at diagnosis. Ten tumors were chondroblastic. The median greatest tumor dimension for the 13 tumors with known size was 10 cm. Ten patients had unresectable pelvic tumors, and 9 patients underwent hemipelvectomy (2 internal and 7 external); complete resection with negative margins was achieved in 5 patients. Four patients survived, including one patient who survived with disease. Of the three patients who survived disease-free, one patient underwent complete resection, one patient underwent incomplete resection (nonviable tumor at the soft tissue margin) with a good response to chemotherapy, and one patient with a sacral tumor underwent radiotherapy only for local control. Of the 9 patients who underwent resection, 7 experienced disease recurrence (n = 5 patients) or progression (n = 2 patients) at distant sites and died. All patients with metastatic disease at diagnosis died.
Pelvic osteosarcomas often were large and unresectable. A high propensity for metastasis contributed to the poor outcome of patients with pelvic osteosarcoma. New therapeutic approaches are needed. Cancer 2005. © 2005 American Cancer Society.
Osteosarcoma of the pelvis accounts for ≈ 5–11% of all osteosarcomas.1–5 Surgery is the primary treatment for patients with osteosarcoma; the removal of all macroscopic and microscopic tumor with a cuff of normal tissue is required for local control. Osteosarcomas arising in the pelvic bones are difficult to resect because of their location, local extension, and large size, and affected patients reportedly have a poor outcome.2, 4, 6–8 In a study of 1702 patients with high-grade osteosarcoma of the extremity and trunk,8 the 5-year overall survival probability for all patients was 65.3%, whereas the same survival probability among 77 patients with pelvic tumors was only 28.9%. Other studies have shown a similar wide discrepancy in survival probability based on tumor location and a very poor prognosis for patients with pelvic tumors.4, 9
Very limited information is available about the clinical behavior and management of osteosarcoma of the pelvis in children because of its low relative incidence; all reports describe pediatric and adult patients together. To define the natural history of these tumors better and to establish management recommendations for children and young adults, we reviewed our institution's experience with osteosarcoma of the pelvis over the past 34 years. In this report, we characterize the clinical findings and treatment outcome of children and young adults with these tumors in relation to the extent of surgical resection. Our review differs from previous reports, because our patients were younger and had no underlying bone disorders, such as Paget disease.3, 4, 6
Through a search of our solid tumor data base, we identified all patients who were diagnosed with osteosarcoma of the pelvic bones and treated at St. Jude Children's Research Hospital between January, 1970 and March, 2004. Information regarding the clinical characteristics, treatment, and outcome of these patients was collected by a review of their medical records. Methods used to evaluate the primary tumor in all patients included plain radiography and bone scintigraphy; and, in the more recent era, computerized tomography (CT) scans and magnetic resonance imaging (MRI) also were used. To exclude metastasis, negative results obtained by plain radiography of the chest and bone scintigraphy were required; for patients who were treated after 1987, negative thoracic CT scan results also were required.
Tumor size was based on the maximum tumor dimension obtained from findings on diagnostic imaging studies or from surgical pathology specimens, when available. Surgical resection was complete when the tumor was resected with bone and soft tissue margins that were free of tumor, as shown by histologic analysis, and was incomplete if the tumor was unresectable or was resected with bone or soft tissue margins that were positive for tumor, also as shown by histologic analysis. External hemipelvectomy was defined as surgical resection of pelvic tumors with amputation of the distal limb, and internal hemipelvectomy was defined as surgical resection of pelvic tumors with preservation of the distal limb.10 Assessment of histologic response of the tumor to preoperative chemotherapy used the four-grade system described by Huvos and colleagues.11, 12 This retrospective review was approved by the Institutional Review Board.
Duration of survival was defined as the interval between diagnosis and either the last follow-up visit or death due to any cause. Duration of event-free survival (EFS) was defined as the interval between diagnosis and either the last follow-up visit or the first event (recurrence, disease progression, second malignancy, or death due to any cause). Distributions of the probability of survival and EFS were estimated by using the method of Kaplan and Meier,13 and standard errors were calculated by the method of Peto et al.14 Differences in outcome distributions were examined by using the exact log-rank test.
Among 442 patients with osteosarcoma who were treated between January, 1970 and March, 2004, 19 patients (4%) had high-grade tumors arising in the pelvic bones. Only 1 patient had a low-grade osteosarcoma and was excluded from this review; this patient underwent complete resection (external hemipelvectomy) only and was alive 17.9 years after diagnosis. Table 1 summarizes the clinical characteristics of the 19 patients who had high-grade osteosarcoma. The median age at diagnosis was 16.8 years (range, 10.8–28.7 years). Ten patients were female. Fifteen patients were Caucasian, 3 patients were African American, and 1 patient was Hispanic. Fifteen patients had tumors arising in the ilium, 2 patients had tumors arising in the pubis, and 2 patients had tumors arising in the sacrum. At the time of diagnosis, 5 patients had metastases in the lung (n = 4 patients) or bone (n = 1 patients). Four tumors arose in the field of previous irradiation, including a third malignancy, which developed 5.7 years after treatment of rectal carcinoma and 13 years after treatment of osteosarcoma of the tibia. The most common histologic type was chondroblastic osteosarcoma (n = 10 patients) followed by osteosarcoma not further specified (n = 6 patients). Data about tumor size were available for 13 patients; the median greatest tumor dimension was 10.0 cm (range, 5.0–18.0 cm).
|Patient no. (yr of diagnosis)||Age in yrs||Race/gender||Primary Tumor site (size in cm)a||Histologic type of osteosarcoma||Surgery (bone/soft tissue margins)||Initial chemotherapy (XRT dose)||Initial recurrence/progression site(s)||Outcome (survival duration in yrs)b|
|Localized at diagnosis|
|1 (1978)||19.9||B/M||Ilium (NA)||Not further specified||None (tumor unresectable)||M, C, D||Local||DOD (0.6)|
|2 (1979)||15.6||W/F||Ilium (NA)||Chondroblastic||None (tumor unresectable)||M||Locoregional||DOD (1.4)|
|3 (1979)||17.7||W/M||Pubis (NA)||Chondroblastic||External hemipelvectomy (negative/negative)||M, C, D||Lung||Died (2.7)c|
|4 (1981)||12.4||B/F||Ilium (10.0)||Chondroblastic||None (tumor unresectable)||D, P||Local||DOD (1.4)|
|5 (1987)||12.3||W/F||Ilium (18.0)||Chondroblastic||External hemipelvectomy (negative/nonviable tumor at margin)||I, M, D, P||None||NED (16.5)|
|6 (1990)||11.8||W/M||Ilium (12.0)||Chondroblastic||None (tumor unresectable)||I||Local||DOD (1.2)|
|7 (1991)d||22.7||W/F||Ilium (7.0)||Not further specified||Internal hemipelvectomy (negative/negative)||I, M, D, P||Lung||DOD (3.0)|
|8 (1991)||14||H/F||Ilium (18.0)||Chondroblastic||External hemipelvectomy (positive/negative)||Carbo, I, D, P, XRT (50.4 Gy)||Lumbar spine||Died of toxicity (0.7)e|
|9 (1991)||16.9||W/M||Ilium (NA)||Not further specified||External hemipelvectomy (negative/negative)||Carbo, I, D, M||None||NED (11.2)|
|10 (1995)||18.3||W/F||Sacrum (10.0)||Osteoblastic||None (tumor unresectable)||Carbo, I||Local||DOD (0.7)|
|11 (1997)d||16.8||W/M||Ilium (12.8)||Not further specified||Internal hemipelvectomy, positive/negative||D, P||Lung||DOD (1.3)|
|12 (1997)||10.8||W/F||Sacrum (5.0)||Telangiectatic||None (tumor unresectable)||Carbo, I, D, M, XRT (98.2 Gy)||None||NED (7.0)|
|13 (1998)d||28.7||W/M||Ilium (9.0)||Not further specified||None (tumor unresectable)||Carbo, I, E||Opposite pubis||AWD (4.8)|
|14 (2000)||19.8||W/M||Ilium (16.5)||Chondroblastic||External hemipelvectomy (negative/negative)||CPT-11, I, D, P, M, XRT (51 Gy)||Lung||DOD (2.6)|
|Metastatic at diagnosis|
|15 (1976)||13.6||W/F||Ilium (NA)||Chondroblastic||External hemipelvectomy (negative/negative)||V, M, D||Lung||DOD (1.9)|
|16 (1983)||17.5||W/M||Ilium (NA)||Chondroblastic||External hemipelvectomy (negative/positive)||M, B, C, A, P, D, E||Lung||DOD (1.9)|
|17 (1988)d||14.2||W/F||Ilium (9.6)||Not further specified||None (tumor unresectable)||I||Local, lung||DOD (0.3)|
|18 (1993)||22.9||B/M||Pubis (7.0)||Telangiectatic||None (tumor unresectable)||Carbo, I||Local||DOD (6.1)|
|19 (1996)||14.1||W/F||Ilium (12.0)||Chondroblastic||None (tumor unresectable)||Carbo, I, P, D, XRT (59.4 Gy)||Local, lung||DOD (1.1)|
The initial treatment and outcomes of the 19 patients are shown in Table 1. Ten patients had unresectable tumors, including 7 of 14 patients with localized disease and 3 of 5 patients with metastasis (P = 1.0). The primary tumors were considered unresectable by the surgeon because of involvement of the sacrum (n = 7 patients), the sacroiliac joint (n = 6 patients), or both (n = 9 patients) or because of the presence of an extensive soft tissue mass that displaced the bladder and rectum (n = 1 patient). The remaining nine patients underwent hemipelvectomy (two internal and seven external), and negative bone margins were achieved in seven patients (one internal and six external). One of 7 patients with negative bone margins (Patient 16) had positive soft tissue margins, and another patient (Patient 5) had nonviable tumor cells present at the soft tissue margin; both patients had undergone external hemipelvectomy.
Four patients received radiotherapy. Two of 10 patients with unresectable tumors (Patients 12 and 19) underwent radiotherapy for local control of the tumor. Patient 12 had a telangiectatic osteosarcoma of the sacrum that was treated with radiotherapy with a cone-down to a total dose of 98.2 grays (Gy), and the patient remained alive 7 years after diagnosis with a stable, residual abnormality indicated by MRI. Despite initial response to chemotherapy (carboplatin and ifosfamide), Patient 19 experienced disease progression at the primary site and lung after radiotherapy and died 1.1 years after diagnosis. Two other patients (Patients 8 and 14) received preoperative radiotherapy (approximate dose, 50 Gy) before they underwent hemipelvectomy. Negative bone margins were achieved only in Patient 14, but this patient eventually died of recurrent disease in the lung. Two patients (Patients 14 and 19) received concomitant intravenous cisplatin therapy and radiotherapy.
Chemotherapy varied with the treatment era and the protocols that were active at the time of diagnosis. Sixteen of 19 patients received chemotherapy with at least 2 agents that are active against osteosarcoma (doxorubicin, cisplatin, carboplatin, ifosfamide, or methotrexate); 3 of those 16 patients survived without evidence of disease, and 1 patient remained alive with disease. All 3 patients who initially received either methotrexate alone or ifosfamide alone (Patients 2, 6, and 17) experienced disease progression and died. Preoperative chemotherapy was used to treat six of nine patients who underwent resection of their primary tumor. Negative bone and soft tissue resection margins were achieved in four of those six patients, and negative bone margins with nonviable tumor cells at the soft tissue margin were achieved in one patient. Histologic response (> 90% tumor necrosis) at the time of hemipelvectomy was observed in 2 of those 6 patients (Patients 5 and 14); chemotherapy included at least 3 active drugs against osteosarcoma in all 6 patients.
Four of 19 patients were alive (3 with no evidence of disease and 1 with disease) a median of 9.1 years after diagnosis (range, 4.8–16.5 years). At the time of the current analysis, the length of follow-up for the survivors was 16.5 years, 11.2 years, 7.0 years, and 4.8 years. All survivors had been seen or contacted within 15 months of the time of the analysis. Thirteen patients died of disease, 1 patient died of treatment-related toxicity with evidence of disease, and 1 patient died of postoperative complications while the disease was in remission. Sixteen patients had progressive or recurrent disease a median of 4.0 months after diagnosis (range, 1.1–16.4 months). All but 1 of these patients died: Patient 13 remained alive with disease. Five-year estimates of survival and EFS for the 19 patients were 26.3% ± 10.1% and 15.8% ± 7.2%, respectively (Fig. 1). Ten-year estimates of survival and EFS were 19.7% ± 10.2% and 15.8% ± 8.4%, respectively.
Of the nine patients who underwent resection of the primary tumor, none experienced local recurrence as the initial event. Two patients (Patients 5 and 9) who had localized disease at diagnosis and underwent hemipelvectomy with negative bone margins (Patient 5 had nonviable tumor at the soft tissue margin) did not experience any events and survived disease-free 16.5 years and 11.2 years after diagnosis. The remaining 7 patients, 5 of whom had positive bone or soft tissue margins (n = 3 patients) or metastatic disease (n = 2 patients) at diagnosis experienced disease recurrence or progression at distant sites (lung, n = 6 patients; spine, n = 1 patient) as the initial event a median of 9.6 months (range, 3.3–16.4 months) after diagnosis. All 6 patients who had pulmonary metastases underwent thoracotomy for resection of pulmonary lesions (median number of thoracotomies, 2; range, 1–6 thoracotomies), but each patient subsequently developed recurrent unresectable metastatic disease, and 3 patients (2 patients with positive margins positive and 1 patient with negative margins) also developed subsequent local recurrences. All 7 patients with distant recurrent disease or disease progression died a median of 1.9 years (range, 0.7–3.0 years) after diagnosis.
Of the 10 patients with unresectable tumors, 8 patients had local (n = 7 patients) or locoregional (n = 1 patient) disease progression as the initial event a median of 2 months (range, 1.1–8.7 months) after diagnosis. Two of those patients (with metastatic disease at diagnosis) also had progressive disease in the lung. All 8 patients who had local or locoregional disease progression died, including 7 patients who died between 0.3 years and 1.4 years after diagnosis and 1 patient with telangiectatic osteosarcoma (Patient 18) who died 6.1 years after diagnosis. Two patients who had unresectable primary tumors did not have local disease progression as an initial event. One of those 2 patients was treated with high-dose radiotherapy for a telangiectatic tumor of the sacrum (Patient 12) and has survived for 7 years after diagnosis. The initial event for the other patient (Patient 13) was metastasis to the opposite pubic bone; this patient was alive with disease 4.8 years after diagnosis. The median survival for the 10 patients who had unresectable tumors was estimated at 1.3 years (95% confidence interval, 0.7–6.1 years).
We analyzed patient outcome with respect to age, histologic features of the tumor (chondroblastic vs. other), tumor size, disease stage (localized vs. metastatic) at diagnosis, and extent of surgical resection (Table 2). Analysis of tumor size as a predictor of outcome was performed by using data for the 13 patients who had known tumor size. No statistically significant differences in survival or EFS distributions were found in relation to the analyzed factors. Of the 3 patients who survived disease-free, 1 patient (Patient 9) underwent complete resection, and another (Patient 5) underwent incomplete resection (nonviable tumor at the soft tissue margin) with a good response to chemotherapy. All five patients who had metastatic disease at the time of diagnosis experienced disease progression and died.
|Variable||No. of patients||No. of deaths||Survival||Event-free survival|
|Five-yr estimate (± 1 SE) (%)||P value||Five-yr estimate (± 1 SE) (%)||P value|
|Age at diagnosis|
|< 16 yrs||9||7||22.2 (11.3)||22.2 (11.3)|
|≥ 16 yrs||10||8||30.0 (14.5)||0.56||10.0 (6.7)||0.97|
|Chondroblastic||10||9||10.0 (6.7)||10.0 (6.7)|
|Other histologies||9||6||44.4 (16.6)||0.32||22.2 (11.3)||1.0|
|< 10 cm||5||3||60.0 (21.9)||20.0 (12.6)|
|≥ 10 cm||8||7||12.5 (8.3)||0.16||12.5 (8.3)||0.72|
|Stage at diagnosis|
|Localized||14||10||28.6 (12.1)||21.4 (9.5)|
|Metastatic||5||5||20.0 (12.6)||0.36||0 (0)||0.49|
|Extent of resection|
|Negative margins||5||4||20.0 (12.6)||20.0 (12.6)|
|Positive margins or unresectable||14||11||28.6 (12.1)||0.34||14.3 (7.6)||0.096|
In this report, we present our institution's experience with 19 children and young adults who were treated for pelvic osteosarcoma over a period of 34 years. Our study was limited by its retrospective nature, the heterogeneity of treatment over the years, and the small number of patients. However, considering the rarity of pelvic osteosarcoma, this review provides valuable information.
The incidence of pelvic osteosarcoma in our study (4%) was similar to that reported previously (4.2–5.6%).3, 6, 15, 16 Most tumors arose from the ilium, and 53% were chondroblastic. This proportion of chondroblastic histology is higher than that seen for osteosarcoma in extremities, in which the majority of tumors are osteoblastic. In other series, 36–58% of pelvic osteosarcomas were chondroblastic.3, 7, 15
In our series, the 5-year survival estimates were similar to those stated in reports of adult and pediatric patients (18–34%),3, 6, 7, 16 although our patients were younger. The anatomic location of these tumors accounts in part for the poor prognosis, because complete resection is much more difficult to achieve in the pelvis than in the extremity. Ten of 19 tumors in our series were unresectable, although it is possible that, with advances in diagnostic imaging and surgical techniques, some of these tumors currently may be considered resectable. Complete resection was achieved in only five of our patients. In a retrospective review of 38 adult and pediatric patients, Bielack et al. showed that the failure of local control of axial tumors was 60%, but the failure of local control of extremity tumors was only 5.7%.15 Ozaki et al.16 described 67 adult and pediatric patients who had pelvic osteosarcomas that did not arise from the sacrum. Adverse prognostic variables included the presence of metastasis at diagnosis as well as the lack of tumor resection or intralesional tumor resection. In 3 different series of adult and pediatric patients (36–40 patients each),3, 6, 7 surgical margin was consistently a significant prognostic factor. Our study did not show a significant difference in outcome based on extent of resection. This may have been related to the small number of our patients, the use of suboptimal chemotherapy, or an inherently aggressive nature of osteosarcoma arising in the pelvis.
Sacral involvement has long been considered a poor prognostic factor in patients with pelvic osteosarcoma, mainly because it makes successful surgical resection more difficult.17 Although the ilium was the most commonly involved site in our series and in other reports,3, 7 extension into the sacrum or sacroiliac joint is very common at the time of presentation. With advances in orthopedic procedures, however, challenges regarding surgery that involves the sacrum possibly may be overcome.18, 19 It is noteworthy that one of our patients with a sacral tumor survived after treatment with radiotherapy only for local control.
In the current study, tumor size, represented by the greatest tumor dimension, had no prognostic significance. This result may have been related to the small number of our patients. In addition, analysis of tumor volume, which probably is a better representative of tumor size, was not possible due to unavailability of data. Large tumor size was a poor prognostic factor in other studies of patients with osteosarcoma of the pelvis3, 7 or extremity.8, 20
In our series, 26% of the tumors were metastatic at diagnosis. This proportion is comparable to that reported by Bielack et al., who noted that osteosarcoma of the trunk was more likely to be metastatic at diagnosis than tumors of the extremity (34% vs. 12%; P < 0.001).8 Wurtz et al.21 evaluated the effect of delays in treatment of a variety of pelvic sarcomas (with 16 of 70 identified as osteosarcomas) on survival. Despite a high incidence of a prolonged delay to diagnosis, there was no significant association between the duration of symptoms and any tumor grade, stage, or survival; this finding suggests that some of these tumors inherently are more aggressive and, thus, may be likely to metastasize by the time of diagnosis.
In the current study, high-dose radiotherapy was an effective local-control measure in 1 patient with unresectable tumor of the sacrum (Patient 12); and, preoperatively, this treatment may have facilitated complete tumor resection in another patient (Patient 14). Estrada-Aguilar et al.22 described good initial local control in 5 patients with pelvic osteosarcoma who were treated with radiotherapy and concurrent intraarterial cisplatin therapy, but only 2 patients remained alive (56 months and 77 months after diagnosis) at the time of the report. Cisplatin potentiates the cytotoxic effects of radiotherapy by inhibiting repair pathways activated by sublethal doses of radiation and was used concurrently with radiotherapy in two of our patients (in one of whom achieved negative margins). Although radiotherapy did not seem to improve survival or to enhance local control of osteosarcoma of the trunk, the treatment positively influenced the prognosis of patients with unresectable osteosarcoma of the pelvis.15, 16
The poor prognosis for patients with pelvic osteosarcoma has been blamed in part on the relatively high incidence of secondary osteosarcoma in the pelvic region.6, 23 In the current study, the proportion of patients who had osteosarcoma secondary to radiation therapy was 21%, emphasizing the higher percentage of secondary tumors in the pelvic area. In other studies, the proportion of secondary tumors in the pelvis has ranged from 12.5% to 15.8%, whereas the proportion in the extremities has been 1.9%3, 7, 8; most of those tumors were secondary to radiotherapy. In the series reported by Grimer et al.,6 33% of pelvic osteosarcomas were secondary: Most were due to Paget disease, and insignificant numbers were secondary to radiation therapy. In that series group, patients with secondary disease were more likely to have had metastasis at diagnosis. Only one of our four patients with secondary osteosarcoma had metastatic disease at diagnosis, and one patient remained alive (with disease). A study found that, provided local tumor control is achieved, patients with osteosarcoma that occurs as a second malignant disease (17 of 30 tumors were radiation-related) who are treated with combined-modality therapy may have a prognosis that approaches that of patients with otherwise comparable, primary osteosarcoma.24
A few studies have evaluated new therapies for unresectable tumors. Carrie et al.25 reported the use of photon and fast-neutron radiotherapy after initial chemotherapy in four patients with localized unresectable pelvic osteosarcoma. There was no change in the radiologic extent of the tumors, but the performance status of the patients improved, and their pain decreased. At the time of the report by Carrie et al., the length of follow-up was 22–40 months, and all 4 patients were alive with no evidence of local progression or metastatic disease. Anderson et al.26 reported improvement in pain among 30 patients with osteosarcoma or bone metastases who were treated with escalating doses of samarium-153 ethylene diamine tetramethylene phosphonate (153Sm-EDTMP), a bone-seeking radiopharmaceutical. Bruland et al.27 administered 153Sm-EDTMP to a male patient age 35 years whose recurrent osteosarcoma was localized to the spine. The patient had subjective pain relief and clinical improvement. Further studies are needed to evaluate the efficacy of these and other novel therapies in patients with unresectable osteosarcoma.
Osteosarcoma of the pelvis is an aggressive tumor that tends to present with bulky disease, is a major challenge to the surgical oncologist, and is difficult to control with available types of chemotherapy and radiation therapy. The overall survival probability is quite poor for patients with osteosarcoma of the pelvis. Internal hemipelvectomy is the preferred surgical treatment, but this type of surgery often is not feasible. The poor prognosis for patients with pelvic osteosarcoma and the presence or absence of metastasis always should be considered in the decision-making about treatment with more aggressive surgery that severely would impair function and cosmesis. Because of the rarity of this disease, large cooperative trials of novel therapeutic approaches will be necessary to improve outcome.
The authors thank Alvida M. Cain for data management and Julia Cay Jones for editing the article.