New classification of dynamic computed tomography images predictive of malignant characteristics of hepatocellular carcinoma.

Aim:   The aim of this study was to elucidate whether the histopathological characteristics of hepatocellular carcinoma (HCC) can be predicted from baseline dynamic computed tomography (CT) images.

Methods:   This retrospective study included 86 consecutive patients with HCC who underwent surgical resection between January 2000 and September 2008. The arterial- and portal-phase dynamic CT images obtained preoperatively were classified into four enhancement patterns: Type-1 and Type-2 are homogeneous enhancement patterns without or with increased arterial blood flow, respectively; Type-3, heterogeneous enhancement pattern with septum-like structure; and Type-4, heterogeneous enhancement pattern with irregular ring-like structures. We also evaluated the predictive factors for poorly-differentiated HCC, specific macroscopic type of HCC (simple nodular type with extranodular growth [SNEG] and confluent multinodular [CMN]) by univariate and multivariate analyses.

Results:   The percentages of poorly-differentiated HCC according to the enhancement pattern were three of 51 nodules (6%) of Type-1 and -2, three of 24 (13%) of Type-3, and eight of 11 (73%) of Type-4. The percentages of SNEG/CMN according to the enhancement pattern were 12 of 51 nodules (24%) of Type-1 and -2, 13 of 24 (54%) of Type-3, and five of 11 (45%) of Type-4. Multivariate analysis identified Type-4 pattern as a significant and independent predictor of poorly-differentiated HCC (P < 0.001) while Type-3 pattern was a significant predictor of SNEG/CMN (P = 0.017).

Conclusion:   Heterogeneity of dynamic CT images correlates with malignant characteristics of HCC and can be potentially used to predict the malignant potential of HCC before treatment.