Prediction of Microvascular Invasion in Recurrent Hepatocellular Carcinoma Prior to the Second Surgery Using Enhanced MRI and Variations in Clinical Features

Abstract

Background: The prediction of microvascular invasion (MVI) in recurrent hepatocellular carcinoma (HCC) holds significant clinical importance, as it has the potential to alter treatment strategies. However, current research in this field remains relatively limited, particularly studies utilizing two preoperative clinical variations to predict MVI in recurrent HCC. Objectives: To improve the efficacy of MVI diagnosis in recurrent HCC, this study aims to incorporate the variations of clinical features into preoperative characteristics of contrast-enhanced magnetic resonance imaging (CE-MRI). Patients and Methods: This single-center, historical cohort study enrolled 72 patients who underwent primary HCC surgery at our hospital and later underwent repeat surgery for recurrent HCC. First, the correlation between the imaging and clinical characteristics of primary HCC and MVI in recurrent HCC were analyzed. Subsequently, to predict MVI in recurrent HCC, two models were developed: Model 1 incorporated second preoperative CE-MRI and clinical characteristics of the recurrent tumor, while model 2 added the clinical variations on the basis of model 1. Univariate and multivariate logistic regression analyses were used to identify independent predictors of MVI in recurrent HCC. The performance of the two models was compared using the DeLong test. Finally, the time to recurrence between patients with MVI and those without MVI in the recurrent tumor was compared. Results: The CE-MRI and clinical features of the first tumor were not statistically related to recurrent tumoral MVI. The factors influencing MVI in recurrent HCC include the variations of eosinophil counts and the variations of size, arterial peritumoral enhancement, non-smooth margins, and multifocality. Model 1 had an area under the curve (AUC) of 0.769 (95% Confidence Interval [CI]: 0.652 - 0.881, P < 0.001), while model 2 had an AUC of 0.839 (95% CI: 0.739 - 0.940, P < 0.001), so model 2 performed better (z = -2.170, P = 0.030). Additionally, tumors with MVI had a significantly shorter median recurrence time of 850.0 days (95% CI: 670.5 - 1029.5) compared to tumors without MVI, which had a median recurrence time of 1270.0 days (95% CI: 1061.8 - 1478.2) (log-rank P = 0.010). Conclusion: The CE-MRI of the primary tumor, along with clinical characteristics, was not correlated with MVI in recurrent tumors. By incorporating variations in clinical characteristics, the prediction of MVI in recurrent tumors can be improved, especially the variation in eosinophil counts, which was independently associated with MVI in recurrent tumors. Notably, patients with MVI in secondary tumors experienced shorter recurrence intervals. These findings improve surveillance for HCC recurrence, boost confidence in preoperative MVI prediction, and hold significant clinical value for postoperative follow-up in HCC patients without MVI.