PURPOSE

To identify a predictive biomarker of sorafenib for hepatocellular carcinoma personalized therapy.

EXPERIMENTAL DESIGN

The patients treated with or without sorafenib after hepatocellular carcinoma recurrence from multicenters were matched with propensity score matching analysis. The expression levels of Fms-like tyrosine kinase 3 (FLT3) in hepatocellular carcinoma specimens of the matched patients ( = 276) were analyzed by IHC. The optimal cut-off point of FLT3 levels for overall survival (OS) was defined via Cutoff Finder. Subgroup analysis of OS was employed to investigate the association between FLT3 levels and sorafenib benefit. The predictive value was assessed via Cox regression models with an interaction term. Hepatocellular carcinoma and paratumoral normal tissues were used to investigate the expression and copy-number variation of FLT3. Patient-derived xenograft (PDX) models were used to confirm the association between FLT3 levels and sorafenib response.

RESULTS

Patients with FLT3-high hepatocellular carcinoma exhibited a superior OS upon sorafenib treatment. High FLT3 levels were predictive of sorafenib benefit in terms of OS ( = 0.00006). Copy-number losses and decreased expression of FLT3 in hepatocellular carcinoma were detected in about 64% of patients. Moreover, the PDXs derived from tumors with high FLT3 levels also displayed a better response to sorafenib.

CONCLUSIONS

Sorafenib may be able to delay tumor progression in patients with FLT3-high hepatocellular carcinoma. This potential biomarker needs to be further validated in independent cohorts prior to helping stratify patients for precision therapy in advanced hepatocellular carcinoma.