Prognostic value of nutritional and inflammatory biomarkers in patients with non-small cell lung cancer and bone metastasis: A retrospective study.

To evaluate the prognostic significance of nutritional-inflammatory biomarkers in non-small cell lung cancer (NSCLC) patients with bone metastases. The present study constructed prognostic models using machine learning methods and assessed their performance, aiming to develop a clinically practical nomogram. A retrospective analysis of 233 patients with NSCLC and confirmed bone metastasis (BM) was conducted. The present study analyzed clinical and laboratory data, including 10 nutritional-inflammatory indicators. The present study used univariate and multivariate Cox regression, Least Absolute Shrinkage and Selection Operator (LASSO), Random Forest and extreme gradient boosting to select variables and construct Cox models. Performance was assessed via C-index, time-dependent area under the curve, Brier score, calibration curve and Akaike information criterion (AIC). A nomogram was developed based on the best-performing model. Multivariate Cox regression identified history of primary tumor surgery [hazard ratio (HR)=0.35; P<0.001], hemoglobin (HR=0.99; P=0.02), prognostic nutritional index (HR=0.98; P=0.002), CYFRA21-1 (HR=1.01; P<0.001), neuron-specific enolase (NSE; HR=1.03; P<0.001) and total cholesterol (HR=1.05; P=0.006) as independent prognostic factors. Individual nutritional-inflammatory biomarkers demonstrated limited discrimination (C-index range: 0.48-0.58). By contrast, integrated models incorporating these markers markedly improved predictive performance. The LASSO_yes model achieved the highest C-index (0.74; 95% CI: 0.70-0.77), with a 24-month area under the curve of 0.79 and the lowest Akaike information criterion (AIC; 1684.6). Based on the best-performing model, a prognostic nomogram was constructed to estimate individualized survival probabilities. Nutritional-inflammatory biomarkers provide incremental prognostic value when incorporated into integrated models. The LASSO-based nomogram may provide a potentially practical tool for individualized survival prediction in patients with NSCLC with BM, although external validation is still required before broader clinical application.