Routine Laboratory Biomarkers for Survival Prediction After Limb-Salvage Surgery in Osteosarcoma: Development of a...

Background: Despite advances in treatment strategies, osteosarcoma patients undergoing limb-salvage surgery (LSS) still face a considerable risk of recurrence and unsatisfactory long-term survival. Therefore, reliable pretreatment baseline prognostic assessment tools are needed to improve postoperative risk stratification and guide individualized treatment. Methods: This single-center retrospective prediction model development study analyzed 118 patients with pathologically confirmed osteosarcoma treated at Shengjing Hospital, China Medical University, between January 2019 and January 2025. After applying predefined inclusion and exclusion criteria, 81 patients were included in the final analysis. All predictors were defined as before-treatment baseline measurements at the diagnosis stage: alkaline phosphatase (ALP) and lactate dehydrogenase (LDH) were serum biochemical markers, whereas fibrinogen (FIB) and D-dimer were plasma coagulation-related markers. Ki-67 index, age, sex, tumor size, and lung metastasis status were also assessed at this before-treatment baseline. A diagnosis-stage baseline prognostic model for overall survival (OS) among patients undergoing LSS was subsequently constructed. Internal validation of model performance was performed using bootstrap resampling only, with no external validation performed. Results: Kaplan-Meier analysis demonstrated 1-, 3-, and 5-year survival rates of 87.3%, 74.3%, and 64.1%, respectively. Survival curve analysis revealed that patients with elevated levels of all four laboratory biomarkers were associated with shorter OS. Multivariate Cox regression analysis revealed that baseline ALP remained independently associated with OS, while FIB showed borderline significance (p = 0.050). The final OS prognostic model incorporated ALP, LDH, FIB, and D-dimer. The apparent Harrell's concordance index (C-index) of the model was 0.838 for the full dataset, with an optimism-corrected C-index of 0.780 after Bootstrap internal validation. The area under the receiver operating characteristic curve (AUC) of the model for predicting 1-, 3-, and 5-year OS in the full dataset was 0.831, 0.864, and 0.871, respectively; the optimism-corrected AUC values after Bootstrap validation were 0.771, 0.810, and 0.814. Conclusions: In this single-center retrospective cohort, higher baseline levels of ALP, LDH, FIB, and D-dimer were associated with adverse clinicopathological characteristics and poorer survival outcomes in patients with osteosarcoma. The LASSO-Cox-based prognostic model showed promising discrimination and calibration after internal Bootstrap validation. However, given the limited sample size and absence of external validation, these findings should be considered preliminary, and prospective multicenter external validation is required before the model can be considered for routine clinical implementation.