Development and validation of a pretreatment prognostic index to predict death and lung metastases in extremity osteosarcoma

BACKGROUND: To develop a prognostic index to predict the 5-year overall survival (OS) and 5-year lung metastasis-free survival (LMFS) of patients with extremity osteosarcoma at the time of diagnosis. METHODS: We retrospectively evaluated 454 patients with extremity osteosarcoma at our center from 2005 to 2013. The cohort was randomly divided into training and validation sets. The association of potential risk factors with OS and LMFS was assessed by Cox proportional hazards analysis in the training set, and a prognostic index was created according to scores that were proportional to a regression coefficient for each factor. This prognostic index was assessed in the validation set. RESULTS: For the 5-year OS, 5 independent prognostic factors were identified: tumor size, Enneking stage, pretreatment platelet, alkaline phosphatase(ALP), and neutrophils. The multivariate Cox model identified tumor size, pretreatment platelets, ALP, and neutrophils as associated with the 5-year LMFS. A prognostic index for death and lung metastases was calculated. Three risk groups were defined for each survival point: low, intermediate, and high risk for the 5-year OS; low, intermediate, and high risk for the 5-year LMFS. The C statistic for the 5-year OS was 0.723 in the training set and 0.710 in the validation set. The C statistic for the 5-year LMFS was 0.661 and 0.693 respectively. CONCLUSION: This prognostic index is based on routine tests and characteristics of extremity osteosarcoma patients and is a useful predictor of OS and lung metastases. This index could be applied to clinical practice and trials for individualized risk-adapted therapies.