Tumor and Perirenal Adipose Tissue Radiomic Models for Pathological T-Stage Prediction and Biological Exploration in...

Prediction of pathological T stage can assist in guiding treatment approaches for clear cell renal cell carcinoma (ccRCC). The objectives of this research were to assess the value of radiomic models based on tumor and perirenal adipose tissue (PRAT) for preoperative prediction of pathological T-stage of ccRCC and to conduct radiogenomic correlation analysis.

Patients recruited were randomly stratified into the training set and the internal test set in a 7:3 proportion. A total of 50 patients from The Cancer Imaging Archive (TCIA) were included as an external test set. Radiomics features were derived from the volume of interest within tumors and PRATs to generate the tumor model, the PRAT model, and the combined model. The T-stage was independently assessed by two experienced radiologists. Subsequently, analysis of biological functions was performed utilizing RNA-seq data obtained from The Cancer Genome Atlas (TCGA) database.

A total of 251 patients were divided into the training set (n = 139), internal test set (n = 62), and external test set (n = 50). The AUC values of the tumor, PRAT, and combined models were 0.84, 0.82, and 0.88 in the external test set, respectively. The combined model achieved the highest AUC value among all models and was notably superior to radiologists in both internal and external test sets. Signaling pathways associated with tumor progression and neural signal transduction were found to be activated in the TCGA set.

The CT-based combined radiomics model demonstrated superior classification performance in predicting the pathological T stage.