Preoperative prediction of microvascular invasion classification in hepatocellular carcinoma based on clinical features and MRI parameters

Microvascular invasion (MVI) in hepatocellular carcinoma (HCC) is a critical pathological factor and the degree of MVI influences treatment decisions and patient prognosis. The present study aimed to predict the MVI classification based on preoperative MRI features and clinical parameters. The present retrospective cohort study included 150 patients (training cohort, n=108; validation cohort, n=42) with pathologically confirmed HCC. Clinical and imaging characteristics data were collected from Shengli Oilfield Central Hospital (Dongying, China). Univariate and multivariate logistic regression analyses were conducted to assess the association of clinical variables and MRI parameters with MVI (grade M1 and M2) and the M2 classification. Nomograms were developed based on the predictive factors of MVI and the M2 classification. The discrimination capability, calibration and clinical usefulness of the nomograms were evaluated. Multivariate analysis revealed an association between the Lens culinaris agglutinin-reactive fraction of alpha-fetoprotein, protein induced by vitamin K absence-II and tumor margin and MVI-positive status, while peritumoral enhancement and tumor size were demonstrated to be marginal predictors, but were also included in the nomogram. However, among MVI-positive patients, only peritumoral hypointensity and tumor size were demonstrated to be risk factors for the M2 classification. The nomograms, incorporating these variables, exhibited a strong ability to discriminate between MVI-positive and MVI-negative patients with HCC in both the training and validation cohort [area under the curve (AUC), 0.877 and 0.914, respectively] and good performance in predicting the M2 classification in the training and validation cohorts (AUC, 0.720 and 0.782, respectively). Nomograms incorporating clinical parameters and preoperative MRI features demonstrated promising potential as straightforward and effective tools for predicting MVI and the M2 classification in patients with HCC. Such predictive tools could aid in the judicious selection of optimal clinical treatments.