Preoperative and postoperative risk prediction models based on perirenal fat and immune infiltration for small renal tumors: A multicenter cohort study.

Abstract

e16514 Background: The obesity paradox in clear cell renal cell carcinoma (ccRCC) has drawn attention, with studies suggesting perirenal fat may serve as an immune cell reservoir, influencing the tumor microenvironment and prognosis. However, its specificity and clinical relevance compared to BMI remain uncertain, and its prognostic utility in patient management is underexplored. This study evaluates the prognostic significance of perirenal fat thickness (PRFT) and its potential to improve prediction models, particularly for small renal tumors (≤4 cm). Methods: PRFT was measured and categorized in a multicenter cohort of 620 ccRCC patients, and its association with survival was analyzed. Immune infiltration was assessed using single-cell RNA sequencing and multiplex immunofluorescence. Correlations between PRFT and immune cell infiltration were explored. Prognostic models were developed for small renal tumors (≤4 cm). Results: Tumor-side perirenal fat clustering demonstrated superior prognostic value over BMI. Reduced tumor-side perirenal fat correlated with increased infiltration of exhausted CD8+ T cells (CD8E). Prognostic models integrating PRFT and immune infiltration showed strong predictive performance for progression-free survival, aiding preoperative risk assessment and postoperative management. Limitations include reliance on postoperative data and lack of external validation. Conclusions: Perirenal fat is a promising prognostic marker, particularly in small renal tumors (≤4 cm). By integrating PRFT and immune infiltration into preoperative and postoperative models, we have developed a personalized and systematic approach for treatment decision-making in early-stage renal tumors, providing a practical framework for improved management and tailored clinical strategies. Performance comparison of prognostic models: 10-fold cross-validation and time-dependent AUC analysis. Model/AUC Fold1 Fold2 Fold3 Fold4 Fold5 Fold6 Fold7 Fold8 Fold9 Fold10 Mean Full model 3-years 4-years 5-years 6-years Preoperative assessment FATe-Surg(Our model) 0.62 0.71 0.82 0.69 0.62 0.82 0.86 0.85 0.64 0.78 0.74 0.74 0.80 0.81 0.78 0.81 Age 0.50 0.59 0.84 0.66 0.57 0.85 0.79 0.85 0.58 0.68 0.69 0.69 0.79 0.79 0.73 0.77 Size 0.59 0.56 0.41 0.23 0.28 0.48 0.54 0.51 0.25 0.62 0.45 0.51 0.64 0.55 0.50 0.47 Age+Size 0.50 0.62 0.83 0.60 0.54 0.84 0.79 0.84 0.54 0.68 0.68 0.69 0.78 0.78 0.72 0.77 Postoperative management FATe-Follow(Our model) 0.63 0.72 0.83 0.73 0.65 0.79 0.83 0.83 0.65 0.74 0.74 0.75 0.84 0.85 0.82 0.84 Age 0.50 0.59 0.84 0.66 0.57 0.85 0.79 0.85 0.58 0.68 0.69 0.69 0.79 0.79 0.73 0.77 SSIGN Suspicion is high collinearity, unable to perform 10-fold cross-validation. 0.53 0.58 0.59 0.57 0.55 Age+SSIGN 0.63 0.80 0.80 0.73 0.77