Subcentimeter hepatocellular carcinoma in treatment-naïve patients: noninvasive diagnostic criteria and tumor staging on gadoxetic acid-enhanced MRI.

Abstract

It is controversial to adopt non-invasive diagnostic criteria of hepatocellular carcinoma (HCC) in subcentimeter lesions. This study was aimed to define the optimal noninvasive diagnostic criteria of subcentimeter HCC and to evaluate the effect on tumor staging. We included 110 treatment-naïve patients at risk of HCC and eligible for curative treatment who had subcentimeter lesions (n = 136) on gadoxetic acid-enhanced magnetic resonance imaging (MRI) performed between January 2013 and December 2013. Modified diagnostic criteria for subcentimeter HCC were developed using logistic regression analysis. Accuracies of MR staging with and without using the modified criteria were compared by generalized estimating equation test using pathologic staging as reference standards. Subgroup analysis was performed for patients with co-existing HCC ≥ 1 cm (co-HCC). The modified criteria (presence of co-HCC, arterial phase hyperenhancement, and hypointensity on transitional phase [TP]) showed 61.5% (95% CI, 41.6-78.2) of sensitivity and 98.2% (95% CI, 93.0-99.5) of specificity. Including subcentimeter HCCs improved the accuracy of MR staging from 84.5 to 94.5% (p = 0.001). Fifty percent of subcentimeter lesions found in patients with co-HCCs were HCC, whereas 5.9% of them without co-HCCs were HCC (p = 0.001). In the subgroup with co-HCCs, the accuracy of MR staging with subcentimeter HCCs was improved from 69.0% to 92.8% (p = 0.001). Including subcentimeter HCCs based on the modified diagnostic criteria (co-existing HCC ≥ 1 cm, arterial phase hyperenhancement, and hypointensity on TP) improved MR staging accuracy. • Fifty percent of non-benign appearing subcentimeter lesions found in patients with co-HCCs were HCC, whereas 5.9% of them without co-HCCs were HCC (p = 0.001). • Including subcentimeter HCCs improved the accuracy of MR staging from 84.5 to 94.5% (p = 0.001).