Ding J, et al. - The predictive models that possibly incorporated a set of CT image features for preoperatively differentiating the high grade (Fuhrman III–IV) from low grade (Fuhrman I–II) clear cell renal cell carcinoma (ccRCC) were comparatively analyzed by the experts. The enrollment consisted of patients with ccRCC treated with a partial or radical nephrectomy. During this study, the Texture-score was estimated through a linear combination of the 4 selected texture features. It was discovered that the Texture-score based models could enable the preoperative discrimination of the high from low grade ccRCC.

Methods

• The enrollment comprised of 114 subjects with ccRCC treated with a partial or radical nephrectomy in the training cohort.
• Researchers analyzed 6 non-texture features, including Pseudocapsule, Round mass, maximal tumor diameter (Diameter_max), intratumoral artery (Artery_tumor), enhancement value of the tumor (TEV) and relative TEV (rTEV), for each tumor.
• The texture features were extracted from the CT images of the section with the largest area of renal mass at both corticomedullary and nephrographic phases.
• In order to screen the most valuable texture features to calculate a texture score (Texture-score) for each patient, least absolute shrinkage and selection operator (LASSO) was found to be useful.
• Using logistic regression model in the training cohort, the high from low grade ccRCC at nephrectomy was discriminated.
• The predictors possibly consisted of all non-texture features in Model 1, all non-texture features and Texture-score in Model 2, and Texture-score in Model 3.
• The performance of the predictive models were investigated and compared in an independent validation cohort composed of 92 cases with ccRCC.

Results

• Findings reported good inter-rater agreement for each non-texture feature and Texture-score (the concordance correlation coefficient or Kappa coefficient > 0.70).
• Using a linear combination of the 4 selected texture features, the Texture-score was measured.
• The 3 models demonstrated good discrimination of the high from low grade ccRCC in the training cohort and the area under receiver operating characteristic curve (AUC) was 0.826 in Mode 1, 0.878 in Model 2 and 0.843 in Model 3.
• A prominent different AUC was discovered between Model 1 and Model 2.
• In the validation cohort, application of the predictive models yielded a discrimination (AUC > 0.670).
• The Texture-score based models with or without the non-texture features (Model 2 and 3) presented with a better discrimination of the high from low grade ccRCC (P < 0.05).