A scoring model based on MRI features for predicting early recurrence after surgical resection of hepatocellular carcinoma

TL;DR

This study creates a scoring model using MRI features to predict early recurrence of liver cancer after surgery, helping doctors plan personalized treatments.

Contribution

A novel MRI-based scoring model is developed to predict early recurrence of hepatocellular carcinoma after surgical resection.

Findings

Tumor number, margin, peritumoral enhancement, and macrovascular invasion were identified as independent predictors of early recurrence.

The scoring model achieved ROC values of 0.873 and 0.847 in training and validation cohorts, with high sensitivity and specificity.

The model's predictive ability increases with higher score groups, enabling risk stratification for early recurrence.

Abstract

Based on MRI features, a scoring model was constructed to predict early recurrence after surgical resection of hepatocellular carcinoma (HCC).

A total of 310 patients from two centers with HCC (212 in the training cohort, 98 in the validation cohort) were collected from January 2017 to October 2023, all patients underwent preoperative MRI-enhanced examinations and were pathologically diagnosed after resection and were divided into early recurrence group and non-early recurrence group based on follow-up results. Clinical, laboratory, and MRI features of patients were collected and subjected to statistical analysis. Univariate analysis and multivariable analysis were used to identify independent predictive factors. The independent predictive factors for early recurrence of liver cancer were weighted using regression coefficient-based scores and construct a score model integrating preoperative variables. Subsequently, receiver operating characteristic (ROC) curves and calibration curves were created to evaluate the performance of the scoring model. The overall score distribution was divided into four groups to show the probability of distinguishing early recurrence.

After multifactor analysis, tumor number, tumor margin, peritumoral enhancement, and macrovascular invasion were identified as independent predictors of early recurrence in preoperative variables. Among them, the tumor margin predictor was assigned 3 points, while the remaining predictors were each assigned 2 points. With a cutoff value of 3.5 points, the ROC value of the score model were 0.873 and 0.847, with sensitivities of 83.9% and 81.3%, and specificities of 77.8% and 73.8%. According to the scores, the predictive ability of early recurrence increased across the four groups.

The established scoring model effectively predicts early recurrence after surgical resection of HCC. The simplicity of the scoring model facilitates clinical application, aiding in the development of personalized treatment plans before surgery.

Figures