Perioperative tislelizumab plus lenvatinib treatment for resectable hepatocellular carcinoma at high risk of recurrence: single-arm phase II trial.

Postoperative recurrence significantly impacts long-term survival in patients with resectable hepatocellular carcinoma (HCC). In this single-arm, phase II trial (NCT04834986), 27 patients with resectable HCC at high recurrence risk received perioperative tislelizumab (anti-PD-1) plus lenvatinib (multi-kinase inhibitor). Co-primary endpoints were safety and feasibility. Secondary endpoints included objective response rate (ORR), 1-year and 2-year disease-free survival (DFS), median DFS, and pathologic complete response (pCR). Post-hoc exploratory machine learning analyses identified potential predictive factors. The regimen was feasible and generally well tolerated; grade 3 treatment-related adverse events occurred in 7.4%. ORR was 55.6%. Among 20 patients undergoing surgery, The Median DFS was 18.8 months (20.1-month follow-up), with 1- and 2-year DFS of 74.0% and 49.8%, and 20% achieved pCR. No unexpected safety signals were observed. The post-hoc machine learning model based on Tabular Prior Data Fitted Network, comprising six pretreatment hematological measures, demonstrated classification of candidate selection in the training set (area under the curve [AUC], accuracy [ACC], and average precision [AP], all 1.0) and the test set (AUC = 0.917, ACC = 0.778, AP = 0.966). In conclusion, tislelizumab-lenvatinib is a potential perioperative regimen for resectable HCC at high recurrence risk, showing preliminary efficacy and safety.