Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma: A Chinese Retrospective Multicenter Study Based on Global Meta-Analysis.
Microvascular invasion (MVI) has been identified as a risk factor for the prognosis of patients with hepatocellular carcinoma (HCC). However, it can only be diagnosed pathologically, and thus no widely applicable preoperative MVI risk prediction model has been established. The aim of this study was to develop a preoperative predictive model for MVI.
The model derivation set was derived from a global meta-analysis involving 39,253 patients. Model validation was performed at the First Affiliated Hospital of Nanjing Medical University (Nanjing set), which included 538 participants, and the First Affiliated Hospital of Sun Yat-sen University (Guangzhou set), which included 111 participants. The model's discriminative ability, assessed by the concordance index (C-statistic), and calibration curves for MVI probability were evaluated.
Through meta-analysis, we identified 40 MVI risk factors and utilized the three most widely used preoperative indicators (alpha-fetoprotein [AFP], tumor size, and tumor margin) to establish an MVI risk prediction model. The C-statistics of the scoring model were 0.805 (95% CI 0.765-0.844) in the Nanjing set and 0.808 (95% CI 0.729-0.887) in the Guangzhou set. Calibration of the established preoperative MVI risk prediction model was adequate (all χ2 values < 20).
The preoperative MVI risk prediction model developed via global meta-analysis exhibits good discriminative ability in different Chinese validation cohorts. Preoperative assessment of MVI status can guide the selection of hepatectomy type, surgical margin width, and neoadjuvant therapy administration, thereby ultimately improving patient prognosis.
The model derivation set was derived from a global meta-analysis involving 39,253 patients. Model validation was performed at the First Affiliated Hospital of Nanjing Medical University (Nanjing set), which included 538 participants, and the First Affiliated Hospital of Sun Yat-sen University (Guangzhou set), which included 111 participants. The model's discriminative ability, assessed by the concordance index (C-statistic), and calibration curves for MVI probability were evaluated.
Through meta-analysis, we identified 40 MVI risk factors and utilized the three most widely used preoperative indicators (alpha-fetoprotein [AFP], tumor size, and tumor margin) to establish an MVI risk prediction model. The C-statistics of the scoring model were 0.805 (95% CI 0.765-0.844) in the Nanjing set and 0.808 (95% CI 0.729-0.887) in the Guangzhou set. Calibration of the established preoperative MVI risk prediction model was adequate (all χ2 values < 20).
The preoperative MVI risk prediction model developed via global meta-analysis exhibits good discriminative ability in different Chinese validation cohorts. Preoperative assessment of MVI status can guide the selection of hepatectomy type, surgical margin width, and neoadjuvant therapy administration, thereby ultimately improving patient prognosis.
Authors
Xu Xu, Zhang Zhang, Yang Yang, Deng Deng, Li Li, Wang Wang, Song Song, Xu Xu, Wu Wu, Wu Wu, Xu Xu
View on Pubmed