Abstract
BACKGROUND: Recently, nomograms have been used as models for risk prediction in malignant tumor because they can predict the outcome of interest for a certain individual based on many variables. This study aimed to establish an effective prognostic nomogram for osteosarcoma based on the clinicopathological factors and microRNA-203. RESULTS: The results showed that miR-203 expression was significantly lower in osteosarcoma tissues compared with the corresponding adjacent tissues (P < 0.001). Patients with low miR-203 expression had poor overall survival (OS) in osteosarcoma. The histological type, tumor size, AJCC stage and miR-203 expression were integrated in the nomogram. The nomogram showed significantly better prediction of OS than for patients with non-metastatic osteosarcoma. The ROC curve also showed higher specificity and sensitivity for predicting 3- and 5-year osteosarcoma patients' survival compared with AJCC stage. The decision curve analysis also indicated more potential of clinical application of the nomogram compared with AJCC staging system. Moreover, our findings were supported by the validation cohort. MATERIALS AND METHODS: We retrospectively investigated 301 patients with non-metastatic osteosarcoma. Data from primary cohort (n = 198) were used to develop multivariate nomograms. This nomogram was internally validated for discrimination and calibration with bootstrap samples and was externally validated with an independent patient cohort (n = 103). CONCLUSIONS: Our proposed nomogram showed more accurate prognostic prediction for patients with non-metastatic osteosarcoma.