Preoperative prediction of macrotrabecular-massive hepatocellular carcinoma based on B-Mode US and CEUS

Objectives: To develop a preoperative prediction model to identify macrotrabecular-massive hepatocellular carcinoma (MTM-HCC) and evaluate the model's diagnostic performance in differentiating MTM-HCC from HCC. Methods: We conducted a mono-center retrospective study in a grade A tertiary hospital in China. Consecutive patients with suspected HCC from February 2019 to December 2020 were eligible for inclusion. All consenting patients underwent CEUS examination and were histologically diagnosed. Based on the clinical and US features between the two groups, we developed a binary logistic regression model and a nomogram for predicting MTM-HCC. Results: A total of 161 patients (median age, 57 years; interquartile range, 48-64 years; 129 men) were included in the analysis. Twenty-seven of the HCCs (16.8%) were of the MTM subtype. Binary logistic regression analysis indicated that PVP hypoenhancement (OR = 15.497; 95% CI: 1.369, 175.451; p = 0.027), AFP > 454.6 ng/mL (OR = 8.658; 95% CI: 3.030, 24.741; p < 0.001), ALB <= 29.9 g/L (OR = 3.937; 95% CI: 1.017, 15.234; p = 0.047), halo sign (OR = 3.868; 95% CI: 1.314, 11.391; p = 0.016), and intratumoral artery (OR = 2.928; 95% CI: 1.039, 8.255; p = 0.042) were predictors for MTM subtype. Combining any two criteria showed a high sensitivity (100.0%); combining all five criteria showed a high specificity (99.2%); and the AUC value of the logistic regression model was 0.88 (95% CI: 0.81, 0.92). Conclusions: BMUS and CEUS could be used for identifying patients suspected of having MTM-HCC. Combining clinical information, BMUS, and CEUS features could achieve a noninvasive diagnosis of MTM-HCC.

Automated summary

This study developed a preoperative prediction model to identify MTM-HCC and evaluated its diagnostic performance in differentiating MTM-HCC from HCC. BMUS and CEUS can be used for identifying patients suspected of having MTM-HCC, and a noninvasive diagnosis of MTM-HCC can be achieved by combining clinical information, BMUS, and CEUS features.