Diagnostic Performance of Multidetector Row Computed Tomography, Superparamagnetic Iron Oxide-Enhanced Magnetic Resonance Imaging, and Dual-Contrast Magnetic Resonance Imaging in Predicting the Appropriateness of a Transplant Recipient Based on Milan Criteria

Abstract

To retrospectively evaluate the diagnostic performance of multidetector row CT (MDCT), superparamagnetic iron oxide (SPIO)-enhanced MRI (S-MRI), and dual-contrast MRI (DC-MRI) in predicting the appropriateness of recipients with hepatocellular carcinoma (HCC) for liver transplantation (LT), based on Milan criteria. This retrospective study received Institutional Review Board approval. Requirement for patient informed consent was waived. During a 3-year period, 80 patients who underwent LT were enrolled in this study. However, 2 patients in whom >10 HCCs were present were excluded from the analysis of detection performance of imaging modalities for HCC. MDCT and DC-MRI examinations with the sequential use of SPIO and gadolinium were performed in all patients. Interval readings for MDCT, S-MRI, and DC-MRI were performed. Two radiologists independently recorded confidence levels using a 4- and 5-point scale for the presence of HCC and for the appropriateness regarding LT, respectively. Image interpretation was compared with histopathological results on a lesion-by-lesion basis. Diagnostic performance of the 3 imaging techniques was compared using jackknife alternative free-response receiver operating characteristic and ROC analyses. Eighty-two HCCs were detected in 38 of 78 patients. Twenty-seven HCCs were larger than 2 cm in diameter and 55 HCCs were smaller than 2 cm in diameter. Among 80 patients included for the assessment of eligibility for LT, 69 recipients were categorized as appropriate and the remaining 11 patients were found to be inappropriate for LT based on Milan criteria. In terms of detecting HCCs, the reader-averaged figure of merit was highest for DC-MRI (0.764), followed by S-MRI (0.702) and MDCT (0.672). The use of DC-MRI was significantly better than the use of the other 2 modalities specifically for HCCs smaller than 2 cm in diameter (P < 0.001) although not for those larger than or equal to 2 cm (P = 0.125-1). The AZ value for predicting the appropriateness for LT was highest with the use of S-MRI (0.841), followed by the use of DC-MRI (0.830) and the use of MDCT (0.790). However, significant differences were not seen for the predictions determined by both radiologists (P = 0.384-1). This result might be because of the small number of patients who had a critical number of HCCs (ie, 2 approximately 4 HCCs). DC-MRI showed significantly better diagnostic performance in transplantation candidates for the detection of HCCs, particularly small HCCs, than both MDCT and S-MRI. However, for assessing the appropriateness of a transplantation recipient based on Milan criteria, MDCT, S-MRI, and DC-MRI showed comparable diagnostic accuracy without a statistical difference.