Microscopic Fat on Chemical-Shift MRI in Solid Renal Tumours: Clear Cell Carcinoma, Fat-Poor Angiomyolipoma and the Added Value of T2-Weighted Heterogeneity and Necrosis

Background/Objectives: To assess whether microscopic fat on chemical-shift MRI, T2-weighted heterogeneity and necrosis separate clear cell renal cell carcinoma (ccRCC) and fat-poor angiomyolipoma (AML) from other solid renal tumours and from each other. Methods: Seventy pathologically proven solid renal tumours without macroscopic fat (38 ccRCCs, 17 papillary RCCs, five chromophobe RCCs, six AMLs, and four oncocytomas) consecutively imaged at 1.5 T between 2017 and 2021 were reviewed retrospectively by one radiologist blinded to histology. Visual signal loss, the chemical-shift index (CSI) and a tumour-to-kidney signal-intensity drop ratio were recorded; signal loss > 20% defined microscopic fat. T2-weighted heterogeneity and necrosis/cystic degeneration (T2-weighted and contrast-enhanced T1-weighted images) were recorded. Groups were compared with non-parametric and Fisher’s exact tests. Results: Median CSI was 43.8% in AML, 14.3% in ccRCC and ≤11.7% in the other subtypes (p < 0.001). A CSI > 20% was observed in 20 of 44 ccRCCs or AMLs and in 1 of 26 other tumours (p < 0.001; sensitivity 45.5%, specificity 96.2%); the tumour-to-kidney ratio and visual assessment performed similarly. AML had higher CSI than ccRCC (p = 0.033), but threshold positivity did not differ significantly (5/6 vs. 15/38; p = 0.077). T2-weighted heterogeneity (84% vs. 17%; p = 0.002) and necrosis/cystic degeneration (82% vs. 0%; p < 0.001) were more frequent in ccRCC than AML. An exploratory AML-positive criterion combining CSI > 20% with T2 homogeneity increased specificity from 60.5% to 92.1%, while sensitivity decreased from 83.3% to 66.7%. A negative CSI was more frequent in papillary RCC (59% vs. 19%, p = 0.004). Conclusions: Opposed-phase signal loss > 20% is a specific but insensitive marker of ccRCC or AML. Morphology provided exploratory complementary information, but the six-AML subgroup limits precision.