MDCT-based scoring system for differentiating angiomyolipoma with minimal fat from renal cell carcinoma

Abstract

Background

Subtype-related various computed tomography (CT) features of renal cell carcinoma (RCC) are a confusing factor in differentiating angiomyolipoma with minimal fat (AMLmf) from RCC. To overcome RCC heterogeneity, a scoring system, which integrates multiple discrimitive parameters can be helpful for differentiating AMLmf from RCC.

Purpose

To develop a MDCT-based scoring system for differentiating AMLmf from RCC.

Material and Methods

In 407 patients with pathologically confirmed 48 AMLmfs and 359 RCCs (247 clear cell RCCs, 67 papillary RCCs, and 45 chromophobe RCCs), MDCT features (ratio of long-to-short diameter, enhancement characteristics, tumor attenuation on unenhanced scan, tumor margin, calcification), age, and sex were compared between AMLmf and RCCs. Based on logistic regression, a scoring system for diagnosing AMLmf over RCC was built, and its diagnostic accuracy was evaluated.

Results

Scores suggesting AMLmf, i.e. the logit function as used in logistic regression analysis, were calculated as follows: $\text{Score}=\frac{e^{6.16.\text{A}-0.003.\text{B}+1.20.\text{C}+0.97.\text{D}+2.13.\text{E}-0.05.\text{F}}}{1+e^{6.16.\text{A}-0.003.\text{B}+1.20.\text{C}+0.97.\text{D}+2.13.\text{E}-0.05.\text{F}}}$ , where A = ratio of long-to-short diameter, B = enhancement amount in early excretory phase, C = homogeneous enhancement, D = tumor attenuation on unenhanced scan, E = sex, and F = age. Area under receiver-operating characteristics curve of scoring system was 0.919. With a score of 0.204 or higher, the scoring system yielded greatest accuracy (90%, 368/407) for diagnosing AMLmf over RCC, which was greater than that of any single MDCT or clinical parameter (53–85%) (P < 0.05). With a score of 0.317 or higher, sensitivity and specificity were 68% (32/48) and 95% (340/359).

Conclusion

MDCT-based scoring system can improve diagnostic performance of MDCT in differentiating AMLmf from RCC and help patients with AMLmf to avoid unnecessary surgery with high specificity.