Radio-genomics analysis of contrast-enhanced MRI for predicting therapeutic response to lenvatinib-based systemic therapy in hepatocellular carcinoma.

e16325 Background: Systemic therapies, including tyrosine kinase inhibitors (TKIs) and immune checkpoint inhibitors (ICIs), are vital for treating unresectable hepatocellular carcinoma (HCC). Currently, lenvatinib has been recommended as a first-line treatment option for patients with unresectable HCC. However, patients always showed inconsistent responses due to heterogeneity of HCC. Thus, to predict the patients who can benefit most from lenvatinib-based therapies is urgently needed. Radiomics, capable of extracting high-throughput quantitative imaging features, has emerged as a promising approach for predicting drug efficacy. This study aims to investigate the role of Radio-genomics analysis in predicting the therapeutic response of HCC patients to lenvatinib-based systemic therapies. Methods: A total of 158 patients with HCC who underwent contrast-enhanced magnetic resonance imaging (CE-MRI) prior to lenvatinib-based systemic therapy between September 2018 and December 2020 were retrospectively recruited. These patients were randomly allocated into training cohort (n = 126) and validation cohort (n = 32). Objective response was defined as either complete response (CR) or partial response (PR) based on the Response Evaluation Criteria in Solid Tumours (RECIST) version 1.1. 63 patients with available RNA sequencing data were included to uncover the underlying biological mechanisms. Results: A total of 14 radiomic features were pinpointed for the radiomics model. The T2 sequence radiomics model demonstrated the most promising predictive performance. In the training cohort, it achieved an area under the curve (AUC) of 0.903 (95% confidence interval (CI): 0.893 - 0.922), and in the validation cohort, the AUC was 0.885 (95% CI: 0.863 - 0.909). Furthermore, among HCC patients, those predicted as non-responders had significantly poorer overall survival (mOS = 23.9 VS 24.7, P = 0.041) and progression-free survival (mPFS = 6.2 VS 11.2, P < 0.001). During the treatment period, 3 (2%), 63 (40%) patients experienced CR and PR, respectively. In the biological validation cohort, 63 patients were classified into the Response group (n = 18) and the non-Response group (n = 45). Through gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, Response group showed upregulation of pathways related to lymphocyte differentiation and T cell activation. Pathways associated with cell metabolism were found to be enriched in the non- Response group. Conclusions: We have developed a novel radiomics model which exhibits capabilities in predicting therapeutic efficacy and prognosis for patients with HCC receiving lenvatinib-based systemic therapy. Moreover, we have found tumour cell metabolism and immune cell differentiation are related to treatment response.