New radiologic classification of renal angiomyolipomas.

PURPOSE: To introduce a new radiologic classification of renal angiomyolipoma (AML).

MATERIALS AND METHODS: Between 1995 and 2014, CT or MR images in 98 patients with histologically proven 98 AMLs were reviewed independently by a radiologist and a resident. The lesions were classified as (a) 53 fat-rich AML (≤-10HU), (b) 22 fat-poor AML (>-10HU) with tumor-to-spleen ratio (TSR) <0.71 or signal intensity index (SII) >16.5%, and (c) 23 fat-invisible AML (>-10HU) with TSR ≥0.71 and SII ≤16.5%. Inter-reader agreement was assessed with a weighted kappa value. Fat-poor and fat-invisible AMLs were compared in terms of attenuation value, TSR, and SII using unpaired t-test.

RESULTS: The weighted kappa value was 0.956 (95% confidence interval, 92.0-99.1%). When a region of interest (ROI) was placed within the most hypodense area on unenhanced CT or within the most signal-dropped area on chemical shift image, the mean attenuation values, TSRs, and SIIs of fat-poor versus fat-invisible AMLs were 19.5±8.1 HU versus 38.1±9.9 HU, 0.59±0.19 versus 0.96±0.01, and 43.7±16.9% versus -5.4±21.1%, respectively (p<0.0001). When a ROI was placed within the other area on CT or chemical shift images, 90.1% (48/53) of fat-rich AMLs were mis-classified as fat-poor or fat-invisible AML and 50% (11/22) of fat-poor AMLs as fat-invisible AML.

CONCLUSION: The new radiologic classification of renal AML is feasible for clinical practice. ROI location is important in differentiating the types of AMLs.