Preoperative grading scale to predict survival in patients undergoing resection of malignant primary osseous spinal neoplasms

Background context: Large population-based studies of malignant primary osseous spinal neoplasms are lacking and are necessary to have sufficient statistical power to determine if various patient-related factors are in fact significant indicators of prognosis. Purpose: Using a 30-year US national cancer registry (Surveillance, Epidemiology, and End Results [SEER]), we introduce a preoperative grading scale that is associated with survival in patients undergoing surgical resection for malignant primary osseous spinal neoplasms. Study design: Large-scale retrospective study. Patient sample: SEER registry. Outcome measure: Survival. Methods: The SEER registry (1973-2003) was queried to identify adult patients undergoing surgical resection of histologically confirmed primary spinal chordoma, chondrosarcoma, or osteosarcoma via International Classification of Disease for Oncology, Third Edition coding. Variables independently associated with survival were determined via Cox proportional hazards regression analysis for all tumor types. A grading scale comprising these independent survival predictors was then developed and applied to each histology-specific tumor cohort. Results: Three hundred forty-two patients who underwent surgical resection of a malignant primary osseous spinal neoplasm (114 chordoma, 156 chondrosarcoma, and 72 osteosarcoma) were identified. Overall median survival after surgical resection was histology specific (osteosarcoma: 22 months; chordoma: 100 months; and chondrosarcoma: 160 months). Increasing age (years) and increasing tumor invasion (confined to periosteum; invasion through periosteum into adjacent tissues; and distal site metastasis) were the only variables independently associated with decreased survival (p<.05) for all tumor types. For spinal chordoma, sacrum/pelvic location (p<.05) and earlier year of surgery (p<.005) were also independently associated with decreased survival. Using variables of patient age, extent of local tumor invasion, and metastasis status in a five-point grading scale, increasing score (1-5) closely correlated (p<.001) with decreased survival for chordoma, chondrosarcoma, and osteosarcoma. Conclusions: In our analysis of a US population-based cancer registry (SEER), a grading scale consisting of age, metastasis status, and extent of local tumor invasion was associated with overall survival after surgical resection of chordoma, chondrosarcoma, and osteosarcoma of the spine. Although this analysis could not take into account specific chemotherapy regimens and variations in surgical technique, this grading scale may offer valuable prognostic data based on variables available to the surgeon and patient before surgery and may help guide level of aggressiveness in subsequent treatment strategies.