Detailed Abstract
[E-poster - Liver (Liver Disease/Surgery)]
[EP 004] Comprehensive Multi-omics Analysis for Resectable Hepatocellular Carcinoma Uncovers Biomarkers to Predict Microvascular Invasion.
Incheon KANG 1, Woo Young KWON 1, Sung Hwan LEE 1
1 Department of Surgery, Bundang CHA, REPUBLIC OF KOREA
Background : Microvascular invasion (MVI) is a key factor in predicting cancer recurrence in resectable hepatocellular carcinoma (HCC). A precision strategy using cancer-specific multi-omics features to predict MVI at HCC diagnosis is needed.
Methods : We analyzed gene expression in resected human HCC (Discovery cohort, n=240) to identify a transcriptomic signature predicting MVI. Using the cancer dependency map (DepMap) project data, we conducted comprehensive analyses, including multi-omics characterization, drug sensitivity screening, and in-silico prediction methods. The MVI-predictive transcriptomic signature was validated in independent cohorts (TCGA-LIHC; n=373, KOREA; n=188, TOKYO; n=183, MODENA; n=78, ZHONGSHAN; n=159).
Results : A 1028-gene MVI signature was identified, showing significant prediction accuracy in the validation cohort (AUC=0.865, p<0.01). Multi-omics analysis highlighted aggressive tumor biology linked to the MVI signature and identified specific biomarkers.
Conclusions : Multi-omics profiling in resectable HCC identifies biomarkers predicting MVI, aiding in distinguishing tumors with aggressive biology. This approach could lead to a precision strategy for identifying HCC patients who would benefit most from surgical resection, informed by ongoing clinical trials based on this study.
Methods : We analyzed gene expression in resected human HCC (Discovery cohort, n=240) to identify a transcriptomic signature predicting MVI. Using the cancer dependency map (DepMap) project data, we conducted comprehensive analyses, including multi-omics characterization, drug sensitivity screening, and in-silico prediction methods. The MVI-predictive transcriptomic signature was validated in independent cohorts (TCGA-LIHC; n=373, KOREA; n=188, TOKYO; n=183, MODENA; n=78, ZHONGSHAN; n=159).
Results : A 1028-gene MVI signature was identified, showing significant prediction accuracy in the validation cohort (AUC=0.865, p<0.01). Multi-omics analysis highlighted aggressive tumor biology linked to the MVI signature and identified specific biomarkers.
Conclusions : Multi-omics profiling in resectable HCC identifies biomarkers predicting MVI, aiding in distinguishing tumors with aggressive biology. This approach could lead to a precision strategy for identifying HCC patients who would benefit most from surgical resection, informed by ongoing clinical trials based on this study.
SESSION
E-poster
E-Session 03/21 ALL DAY