Objective:To explore the value of combining gut microbiota and clinical features for preoperative microvascular invasion (MVI) prediction in hepatocellular carcinoma (HCC).Methods:Clinical data and fecal samples were collected from 71 HCC patients who underwent curative resection at Ningbo Second Hospital between Jan 2023 and Aug 2024. Among them, 41 patients were assigned to the training set and 30 to the validation set. Gut microbiota composition was analyzed using 16S rRNA sequencing. Redundancy analysis (RDA) was used to evaluate the influence of clinical features on the microbiota. Differences in alpha and beta diversity between the MVI-negative and MVI-positive groups were assessed. Differential genera were identified using the Wilcoxon test and LEfSe analysis. A random forest model and Logistic regression were employed to screen key differential genera, followed by ROC analysis. Genera with high ROC values were further validated in the validation set.Results:RDA indicated that MVI was a key factor influencing gut microbiota composition. The random forest model (AUC=0.925), combined with Logistic regression analysis, identified four genera: Acidovorax ( OR=0.618), Tissierella ( OR=1.293), Chitinophaga ( OR=4.596), and Virgisporangium ( OR=0.960), as well as two clinical features: tumor diameter ( OR=0.668) and liver cirrhosis ( OR=14.011), as independent risk factors. ROC analysis showed that in the training set, the combination of Chitinophaga (AUC=0.71) and tumor diameter (AUC=0.75) had the best diagnostic performance (AUC=0.87). In the validation set, the combination of Virgisporangium (AUC=0.80) and tumor diameter (AUC=0.79) yielded the highest diagnostic performance (AUC=0.87). Conclusions:A genomics-based model combining gut microbiota and clinical features shows promising predictive value for noninvasive preoperative assessment of MVI status in HCC patients.

Objective:To explore the value of combining gut microbiota and clinical features for preoperative microvascular invasion (MVI) prediction in hepatocellular carcinoma (HCC).Methods:Clinical data and fecal samples were collected from 71 HCC patients who underwent curative resection at Ningbo Second Hospital between Jan 2023 and Aug 2024. Among them, 41 patients were assigned to the training set and 30 to the validation set. Gut microbiota composition was analyzed using 16S rRNA sequencing. Redundancy analysis (RDA) was used to evaluate the influence of clinical features on the microbiota. Differences in alpha and beta diversity between the MVI-negative and MVI-positive groups were assessed. Differential genera were identified using the Wilcoxon test and LEfSe analysis. A random forest model and Logistic regression were employed to screen key differential genera, followed by ROC analysis. Genera with high ROC values were further validated in the validation set.Results:RDA indicated that MVI was a key factor influencing gut microbiota composition. The random forest model (AUC=0.925), combined with Logistic regression analysis, identified four genera: Acidovorax ( OR=0.618), Tissierella ( OR=1.293), Chitinophaga ( OR=4.596), and Virgisporangium ( OR=0.960), as well as two clinical features: tumor diameter ( OR=0.668) and liver cirrhosis ( OR=14.011), as independent risk factors. ROC analysis showed that in the training set, the combination of Chitinophaga (AUC=0.71) and tumor diameter (AUC=0.75) had the best diagnostic performance (AUC=0.87). In the validation set, the combination of Virgisporangium (AUC=0.80) and tumor diameter (AUC=0.79) yielded the highest diagnostic performance (AUC=0.87). Conclusions:A genomics-based model combining gut microbiota and clinical features shows promising predictive value for noninvasive preoperative assessment of MVI status in HCC patients.