MATERIALS AND METHODS
The cohort consisted of patients diagnosed with chondrosarcoma, Ewing sarcoma, and osteosarcoma of the short bones from 1973 to 2013. Patient data were obtained using the case-listing session procedure in the SEER program of the National Cancer Institute. The SEER database holds only anonymized data, which are freely available. This study was performed in accordance with standard guidelines and was approved by the Ethics Committee of the Second Affiliated Hospital, School of Medicine, Zhejiang University.
All primary tumors that originated in the short bones were identified using the lesion number C40.1 (short bones of upper limb and associated joints) and C40.3 (short bones of lower limb and associated joints). The International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) was then used to identify patients with chondrosarcoma (ICD-O-3 codes 9180, 9220, 9221, 9231, 9240, 9242, and 9243), Ewing sarcoma (ICD-O-3 code 9260), and osteosarcoma (ICD-O-3 codes 9180–9187 and 9192–9195). All patients were diagnosed by histologic confirmation either from biopsy or surgical pathology. Patients who were diagnosed from only clinical presentation, radiography, autopsy, or death records were excluded. Data extracted from the SEER database included patient demographics, stage at diagnosis, tumor grade, tumor type, tumor sequence, cause of death, year of diagnosis, surgical treatment, and survival time.
Statistical analyses were performed using the SPSS statistical package version 21.0 and Microsoft Excel 2016. OS was defined as the time from diagnosis to death from any cause, and CSS was defined as the time from diagnosis to death specific to the cancer-related diagnosis. Descriptive epidemiologic data were calculated for all variables. OS and CSS curves were calculated using the Kaplan–Meier method. Observations were censored if the patient was alive at the time of the last follow-up. The effects of demographic variables, tumor characteristics, and treatment variables were compared using log-rank tests for categorical values. The Cox proportional hazards regression model was used to determine the independent prognostic factor for OS and CSS using univariate and multivariate analyses. The hazard ratios (HRs) with corresponding 95% CIs were used to show the effect of factors on OS and CSS. Differences were deemed statistically significant if p <0.05.
In total, 457 patients were eligible for the study, including 245 patients with chondrosarcoma, 115 patients with osteosarcoma, and 97 patients with Ewing sarcoma of hands or feet. Chondrosarcoma was the most common type in hands, feet, or both. Demographically, 53.4% and 46.6% of patients were male and female, respectively. Most patients (84.2%) were white. The mean age at diagnosis was 41.9 years, and more than half of the patients (62.4%) were born in or after the year 2000. Histologically, 42.2% of cases were low grade, 22.5% were high grade, and 35.2% were an unknown tumor grade. Information regarding the extent of disease was available in most cases (92.3%), and the majority presented as locally invasive disease (46.2%). Forty-eight patients (10.5%) had developed metastasis at presentation. Most patients were diagnosed with first malignant primary tumors (93.7%). After diagnosis, 88% of patients underwent surgical treatment. A total of 176 patients (38.5%) died; among these, 95 died of the tumor itself (cancer-specific death). The OS rates of the entire cohort at 5 and 10 years were 75.7% and 66.1%, respectively. The CSS rates at 5 and 10 years were 78.7% and 73.7%, respectively. The 5- and 10-year OS rates for patients with malignant osseous foot tumors were 72.2% and 63.7%, and the respective 5- and 10-year CSS rates were 74.1% and 68.0%. For patients with malignant osseous hand tumors, the OS rates were 81.2% and 70.9% and the CSS rates were 86.3% and 83.4%, respectively (Table 1).
Univariate survival analysis revealed that older age (age >40 years) was significantly associated with a worse OS (HR=1.618, p=0.002; Table 2; Figure 1A) but not with a worse CSS (HR=0.797,p=0.287; Table 3). Similarly, gender was associated with significant differences in OS (male vs. female HR=0.733, p=0.042; Table 2) but not in CSS (male vs. female HR=1.136, p=0.544; Table 3). A more recent year of diagnosis was associated with improved CSS (1990s vs. 1970s, HR=0.287,p=0.002; ≥2000s vs. 1970s, HR=0.332, p=0.002; Table 3; Figure 2A) but not with OS (p>0.05;Table 2). For both OS and CSS, race showed no significant effect on survival (p>0.05; Tables 2 and3). Compared with patients with bone sarcomas of the hands, patients with sarcomas of the feet had worse CSS (HR=1.912, p=0.010; Table 3) but not OS (p>0.05; Table 2). There was a significant difference in both OS and CSS based on the extent of disease at presentation (OS: regional vs. localized, HR=1.504, p=0.024; and distant vs. localized, HR=5.722, p=0.000; CSS: regional vs. localized, HR=2.100, p=0.010; and distant vs. localized, HR=12.902, p=0.000; Tables 2 and 3; Figures 1B and 2B). Univariate analysis indicated that tumor grade was an important prognostic factor for both OS and CSS (both p=0.000; Tables 2 and 3; Figures 1C and 2C), with high tumor grade portending a worse prognosis. Compared with chondrosarcoma, the prognosis of patients with Ewing sarcoma was much worse in terms of OS (Ewing sarcoma vs. chondrosarcoma, HR=1.758, p=0.002;Table 2) and CSS (Ewing sarcoma vs. chondrosarcoma, HR=4.550, p=0.000; Table 3). In terms of treatment, patients who did not receive surgical treatment had worse OS and CSS than patients who underwent surgery (HR=2.258, p=0.000 and HR=3.614, p=0.000, respectively; Tables 2 and 3; Figures 1D and 2D). Additionally, there were significant differences in OS (second or more vs. first, HR=1.879, p=0.020; Table 2; Figure 1E) but not CSS (p=0.214; Table 3) based on tumor sequence.
(To view a larger version of Table 2, click here.)
(To view a larger version of Figure 1, click here.)
(To view a larger version of Table 3, click here.)
(To view a larger version of Figure 2, click here.)
On multivariate analysis for all patients (Table 2), age at diagnosis (HR=2.418; 95% CI, 1.635–3.574; p=0.000), stage (regional vs. localized, HR=1.480, 95% CI, 1.034–2.119, p=0.032; distant vs. localized, HR=3.839, 95% CI, 2.375–6.205, p=0.000), tumor grade (HR=1.643; 95% CI, 1.025–2.636; p=0.039), surgery (HR=1.792; 95% CI, 1.139–2.818; p=0.012), and tumor sequence (HR=2.284; 95% CI, 1.304–4.000; p=0.004) were independent predictors of OS. The results of multivariate analysis of the parameters that influence CSS are presented in Table 3. Decade of diagnosis (1980s vs. 1970s, HR=0.382, 95% CI, 0.168–0.866, p=0.021; 1990s vs. 1970s, HR=0.171, 95% CI, 0.069–0.424, p=0.000; ≥2000s vs. 1970s, HR=0.269, 95% CI, 0.128–0.563,p=0.000), stage (distant vs. localized, HR=7.040; 95% CI, 3.708–13.364; p=0.000), tumor grade (HR=3.385; 95% CI, 1.523–7.523; p=0.003), and surgery (HR=1.890; 95% CI, 1.133–3.151;p=0.015) were independent prognostic factors for CSS. Moreover, multivariate analysis showed no statistically significant difference among primary tumor location and tumor type.