Evaluating the severity of microvascular invasion in hepatocellular carcinoma, by probing the combination of enhancement modes and growth patterns through magnetic resonance imaging

TL;DR

This study uses MRI to identify imaging traits linked to microvascular invasion severity in liver cancer, helping doctors make better treatment decisions.

Contribution

The study introduces a novel method combining MRI enhancement modes and tumor growth patterns to predict microvascular invasion severity in hepatocellular carcinoma.

Findings

Four MRI features were significantly associated with microvascular invasion (MVI) severity.

Nomograms achieved high accuracy in predicting MVI and its M2 grade with AUCs of 0.885 and 0.805, respectively.

Enhancement modes and growth patterns were confirmed as independent risk factors for MVI severity.

Abstract

Microvascular invasion (MVI), particularly its severity, correlates with prognosis in hepatocellular carcinoma (HCC), however, it remains uncertain which imaging traits are associated with MVI grades. Predicting MVI status precisely pre-surgery assists clinicians in making optimal treatment decisions.

213 HCC patients with surgically confirmed were assigned into three groups based on the severity of MVI (M0, M1, and M2). Clinical and imaging features were compared between each group. Univariate and multivariate analyses were used to identify the significant variables associated with MVI severity. Subsequently, nomograms were constructed to estimate MVI and its M2 grade by crucial factors. Nomograms were assessed for accuracy, clinical value, and efficacy using the area under the curve (AUC), calibration curve, and decision curve analysis (DCA).

Four factors associated with MVI (P < 0.05) were related, including non-solitary growth types, no/mini enhanced mode, peritumoral enhancement on arterial phase, and peritumoral hypointensity on hepatobiliary phase. Only the ratio of the maximum and minimum tumor diameter (Max/Min-R), confluent multinodule growth type, and non-washin/washout enhanced modes of those MVI-positive patients showed a strong correlation with M2 grade. The areas under the receiver operating characteristic (ROC) curves were 0.885 (95% confidence intervals [CI]: 0.833–0.937) in identifying MVI and 0.805 (95% CI: 0.703–0.908) in predicting its M2 grade, respectively. The nomograms demonstrated a high goodness-of-fit and clinical benefits in DCA and calibration curve.

Enhancement modes and tumor growth patterns of preoperative MRI were independent risk factors of MVI severity, which were valuable for facilitating individualized decision-making.

Figures