Atypical renal angiomyolipoma versus renal cell carcinoma; a diagnostic dilemma. Useful Computed Tomography findings to discriminate these tumors

Objective: To compare various computed tomographic features of atypical angiomyolipomawith those of size-matched renal cell carcinoma. Materials and methods: A retrospective study of sixty-eight patients (17 with atypical angiomyolipomas and 51 withRenal cell carcinoma) who had undergone nephrectomy due to a presumptive diagnosis of renal cell carcinoma < 45mm in diameter determined by preoperative Triphasic Computed Tomography (CT) (non-contrast corticomedullary and early excretory phases) was performed. Two radiologists who were unaware of the final diagnosis retrospectivelyexamined the general characteristics of the tumor, their attenuation in the non-contrast phase and enhancement characteristics. A statistical analysis was performed with R software applying the Logit model to differentiate atypical angiomyolipoma from renal cell carcinoma based on the tomographic findings of a renal mass. The final model included tumor contour, attenuation features in the non-contrast scans and the enhancement pattern over time. Results: The most valuable CT findings to differentiate anatypical angiomyolipoma were the hyperdense attenuation of the tumor in the non-contrast phase and its prolonged enhancement pattern with a likelihood of 10.49 (p=0.0381) and 36.71 (p = 0.0009), respectively. Based on the value of each finding included in the model, we calculated likelihood ratios, sensitivity (0.2941) and specificity (0.9804) A ROC curve was developed to determine the optimal cut off point (0.9694) to discriminate atypical angiomyolipoma that confirmed their presence. Conclusion: Triphasic helical CT may be useful in differentiating atypical angiomyolipoma fromrenal cell carcinoma in which hyperdense attenuation of the tumor in the non-contrast scans and prolonged enhancement pattern proved to be the most valuable CT features. © 2016 Sociedad Argentina de Radiologia. Published by Elsevier Spain, S.L.U. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)