This analysis demonstrates the link between bone metastasis distribution and prognosis in renal cell carcinoma, suggesting more accurate risk assessment.
Key Points
To evaluate the prognostic significance of bone metastasis distribution and develop a predictive model for renal cell carcinoma.
Stratified 122 patients by MSKCC/Motzer risk score at first bone metastasis diagnosis.
Classified patients into locoregional, stochastic, and extensive groups based on lesion distribution.
Conducted univariate analysis, logistic regression, and Kaplan-Meier survival analysis for associations.
Developed a random survival forest model for survival prediction, validated through train-validation splits.
Locoregional and spinal metastasis were identified as predictors of higher MSKCC/Motzer risk stratification.
Pelvic metastasis was linked to a shorter median overall survival (32 vs 49 months; p<0.05).
The RSF model showed high predictive performance with AUCs of 0.90 at 1 year and 0.87 at 3 years, outperforming Cox regression (median AUC 0.89 vs 0.59).