Propensity score analysis

Because grouping was based only on perioperative AVT rather than randomization, propensity score matching (PSM) was used to reduce the imbalance between the two groups. All possible clinicopathological covariates, such as tumor number, tumor size, and vascular invasion, which might have affected the target outcomes were included when conducting the PSM. Propensity scores were evaluated using a logistic regression model in this study. Patients were matched at a 1:1 ratio using the caliper matching method within 0.2 of the standard deviation from the propensity score logit, basing on previous literatures [22–24]. And finally, we obtained the score-matched pairs for subsequent analyses.