Abstract
BACKGROUND: Microvascular invasion (MVI), as an important stage of tumor cell invasion of blood vessels, is a key step for hepatocellular carcinoma (HCC) to metastasize into the liver and remote organs. Transcatheter arterial chemoembolization (TACE) is currently the most widely used adjuvant treatment after HCC surgery, which is used to treat tumor micrometastasis, and can inactivate residual tumor foci and reduce the recurrence rate. However, there is still controversy concerning its efficacy. The objective of our clinical study is to use propensity score matching and random forest models to determine whether HCC patients with MVI can benefit from postoperative adjuvant TACE therapy. Additionally, based on adjuvant TACE treatment, we aim to develop individualized random forest prediction models for the 5-year disease-free survival (DFS) rate and overall survival (OS) rate in HCC patients with MVI. METHODS: Three hundred and twelve HCC patients with MVI and without MVI were divided into TACE and non-TACE (N-TACE) groups by propensity matching scoring method. Kaplan-Meier method was used to compare DFS and OS. Cox regression analysis evaluated the effect of adjuvant TACE on DFS and OS. The most influential characteristic variables of DFS and OS were used to construct the random forest prediction models. RESULTS: After matching, the 1-, 3-, and 5-year DFS and OS of HCC patients with MVI in the TACE group were significantly higher than those in the N-TACE group (P<0.05). There was no significant difference in DFS and OS of HCC patients without MVI. Adjuvant TACE was an independent predictor of DFS and OS in HCC patients with MVI. Moreover, the random forest prediction models showed the key role of postoperative adjuvant TACE in DFS and OS of HCC patients with MVI. CONCLUSIONS: Postoperative adjuvant TACE improves the survival of HCC patients with MVI. The random forest prediction model has a good value for predicting the survival of HCC patients with MVI.