Multiregional radiomics nomogram for postoperative recurrence risk stratification in localized clear cell renal cell carcinoma

Abstract

Abstract Background The selection of individualized treatment options based on the risk of recurrence is crucial in the adjuvant treatment of clear cell renal cell carcinoma(ccRCC). Multiregional radiomics might noninvasively obtain potential information about the intratumoral and peritumoral heterogeneity of ccRCC and reveal the prognostic information behind the images. This study aimed to develop a CT-based multiregional radiomics nomogram to improve the stratification of postoperative recurrence risk in patients with localized ccRCC. Methods A total of 395 patients with pathologically diagnosed ccRCC were included in the training (n = 281) and internal validation set (n = 114). Multiregion radiomics features from both the intratumoral and peritumoral areas were extracted. The selection of radiomics features and clinicopathological factors was performed using the least absolute shrinkage and selection operator (LASSO) Cox regression. A final model (FM) for the radiomics nomogram was developed, which incorporated the selected clinicopathological and radiomics features predictors based on multivariate Cox proportional hazard regression. The performance of the model was assessed using receiver operator characteristic (ROC) analysis. Results The radiomics nomogram demonstrated excellent prediction performance in both the training and validation sets. The discriminatory ability of the radiomics nomogram was superior to that of the clinical model (C-index 0.926 vs .0.898, P < 0.05). Decision curve analysis revealed that the nomogram had more net benefit than the clinical model. Conclusions The use of a radiomic nomogram with multiregion features improved the stratification of postoperative recurrence risk in patients with localized ccRCC and can be considered a valuable tool in clinical decision-making.