Abstract
The purpose of this study was to directly (head-to-head) compare the per-lesion diagnostic performance of contrast-enhanced computed tomography (CT) (also referred to as CT hereafter) and gadoxetic acid disodium (Gd-EOB-DTPA)-enhanced magnetic resonance (MR) imaging (also referred to as MRI hereafter) for the detection of hepatocellular carcinoma (HCC). Studies reporting direct per-lesion comparison data of contrast-enhanced multidetector CT and Gd-EOB-DTPA-enhanced MR imaging that were published between January 2000 and January 2015 were analyzed. The data of each study were extracted. Systematic review, paired meta-analysis, and subgroup analysis were performed. Twelve studies including 627 patients and 793 HCC lesions were analyzed. The sensitivity estimates of MRI and CT were, respectively, 0.86 (95% CI 0.76-0.93) and 0.70 (95% CI 0.58-0.80), with significant difference (P < 0.05). The sensitivity estimates were both 0.94 (95% CI 0.92-0.96) (Chi-square 4.84, degrees of freedom = 1, P > 0.05). In all subgroups, Gd-EOB-DTPA-enhanced MR imaging was more sensitive than multidetector CT for the detection of HCC, and specificity estimates of both tests maintained at a similarly high level in all conditions: sensitivity estimates of both tests were reduced in studies where patients were diagnosed with HCC solely by liver explant or in those where HCC lesions were small (≤2 cm, especially when ≤1 cm). But in all situations, sensitivities of MRI were higher than those of CT with or without significance. Gd-EOB-DTPA-enhanced MR imaging showed better per-lesion diagnostic performance than multidetector CT for the diagnosis of HCC in patients with cirrhosis and in small hepatic lesions.