Development and Validation of a Vascular Endothelial Growth Factor A-associated Prognostic Model for Unresectable Hepatocellular Carcinoma

PURPOSE: High serum vascular endothelial growth factor (VEGF) levels have been identified as an independent risk factor for hepatocellular carcinoma (HCC). We aimed to construct a VEGF-included prognostic model to accurately perform individualized predictions of survival probability for patients with unresectable HCC. PATIENTS AND METHODS: From October 2018 to March 2021, 182 consecutive newly diagnosed patients with unresectable HCC were retrospectively enrolled. Baseline serum VEGF-A and other characteristics were collected for all patients. Univariate Cox regression analysis and LASSO regression model were applied to develop the prognostic model, enhanced bootstrap method with 100 replicates was performed to validate its discrimination and calibration. We compared the final model with China Liver Cancer (CNLC) stage, American Joint Committee on Cancer (AJCC) stage, Barcelona Clinic Liver Cancer (BCLC) stage, and the model without the “VEGF”. Finally, the established model was stratified by age. RESULTS: The VEGF-associated prognostic model we established has high accuracy with an overall C-index of 0.7892 after correction for optimistic estimates. The area under the curve (AUC) of the time-dependent receiver operating characteristic (ROC) curves at 6-month, 1-year, and 2-year after correction were 0.843, 0.860, 0.833, respectively, and the calibration of the model was 0.1153, 0.1514, and 0.1711, respectively. The final model showed significant improvement in predicting OS when compared to the other models according to Harrell’s C-index, The AUC of the time-dependent ROC, area under the decision curve analysis (AUDC), integrated discrimination improvement (IDI), and continuous net reclassification index (NRI). CONCLUSION: The VEGF-associated prognostic model may help to predict the survival probabilities of HCC patients with favorable performance and discrimination. However, further validation is required since we only verified this model using internal but not external data.