Imaging Features of Gadoxetic Acid–enhanced and Diffusion-weighted MR Imaging for Identifying Cytokeratin 19-positive Hepatocellular Carcinoma: A Retrospective Observational Study

Purpose To determine the preoperative magnetic resonance (MR) imaging findings potentially most useful for predicting cytokeratin 19 (CK19)–positive hepatocellular carcinoma (HCC) and to evaluate the prognosis after curative resection in patients with a single HCC lesion positive for CK19 compared with patients with HCC who are negative for CK19. Materials and Methods The institutional review board approved this study and waived the requirement for informed consent. Two hundred four patients with CK19-negative HCC and 38 with CK19-positive HCC who underwent curative resection after gadoxetic acid–enhanced and diffusion-weighted MR imaging were retrospectively evaluated in a single institution. Two radiologists evaluated preoperative findings at MR imaging. Significant findings for differentiating the two groups were identified at univariate and multivariate analyses. By using receiver operating characteristic analysis, the optimal cut-off values for quantitative variables were determined. Recurrence-free survival rates after surgery were also compared between groups. Results At multivariate analysis, irregular tumor margin (P = .024), arterial rim enhancement (P < .001), lower tumor-to-liver signal intensity (SI) ratio on hepatobiliary phase (HBP) images (≤0.522; P = .01), and lower tumor-to-liver apparent diffusion coefficient (ADC) ratio (≤0.820; P < .001) were independent significant factors to predict CK19-positive HCC. When three of these four criteria were combined, 63.2% (24 of 38; 95% confidence interval: 46.0%, 78.2%) of CK19-positive HCCs were identified with a specificity of 90.7% (185 of 204; 95% confidence interval: 46.0%, 78.2%). When all four criteria were satisfied, specificity was 99.5% (203 of 204; 95% confidence interval: 97.3%, 100%). Recurrence-free survival rates were significantly lower in patients with CK19-positive HCCs compared with those with CK19-negative HCCs after curative resection (63.9% vs 90.0% at 1 year, 63.9% vs 79.9% at 2 years, and 54.8% vs 70.2% at 3 years, P = .001 by log-rank test). Conclusion At gadoxetic acid–enhanced and diffusion-weighted MR imaging, irregular margin, arterial phase rim enhancement, lower tumor-to-liver ADC ratio, and lower tumor-to-liver SI ratio at HBP imaging may be helpful to predict CK19-positive HCC with early recurrence (<2 years) after curative resection. © RSNA, 2017 Online supplemental material is available for this article.