Prediction of microvascular invasion of hepatocellular carcinomas with gadoxetic acid-enhanced MR imaging: Impact of intra-tumoral fat detected on chemical-shift images.

Purpose To investigate the impact of intra-tumoral fat detected by chemical-shift MR imaging in predicting the MVI of HCC. Materials and methods Gadoxetic acid-enhanced MR imaging of 365 surgically proven HCCs from 365 patients (306 men, 59 women; mean age, 55.6 years) were evaluated. HCCs were classified into two groups, fat-containing and non-fat-containing, based on the presence of fat on chemical-shift images. Fat-containing HCCs were subdivided into diffuse or focal fatty change groups. Logistic regression analyses were used to identify clinical and MR findings associated with MVI. Results Based on MR imaging, 66 tumors were classified as fat-containing HCCs and 299 as non-fat-containing HCCs. Among the 66 fat-containing HCCs, 38 (57.6%) showed diffuse fatty changes and 28 (42.4%) showed focal fatty changes. MVI was present in 18 (27.3%) fat-containing HCCs and in 117 (39.1%) non-fat-containing HCCs ( P = 0.07). Univariate analysis revealed that serum alpha-fetoprotein (AFP) and tumor size were significantly associated with MVI ( P < 0.001). A multiple logistic regression analysis showed that log AFP (odds ratio 1.178, P = 0.0016), tumor size (odds ratio 1.809, P < 0.001), and intra-tumoral fat (odds ratio 0.515, P = 0.0387) were independent variables associated with MVI. Conclusion Intra-tumoral fat detected with MR imaging may suggest lower risk for MVI of HCC and, therefore, a possibly more favorable prognosis, but the clinical value of this finding is uncertain. [ABSTRACT FROM AUTHOR]