Journal Article

IDDF2023-ABS-0080 Utilizing tumor microenvironment microbial profiles and host gene expressions for reliable survival subtyping in liver hepatocellular carcinoma

Zhang, Haohong, Ning, Kang

2023

Overview

BackgroundLiver hepatocellular carcinoma (LIHC) is a challenging and deadly cancer with poor prognosis and treatment options. Despite recent advances in genomics and immunotherapies, a deeper understanding of the molecular mechanisms underlying LIHC survival is crucial to identify novel therapeutic targets. One promising area of research is the tumor microbiome, a complex community of microbes found in tumors and surrounding tissue. However, the intricate relationships between microbial profiles and host gene expressions that drive the development of LIHC and influence patient survival remain unclear.MethodsTo address this challenge, we developed ASD-LIHC (autoencoder-based subtypes detector for LIHC), a semi-supervised deep learning framework that extracts survival-related features from tumor microbiome and transcriptome data to differentiate LIHC patients into survival subtypes based on their overall survival time (IDDF2023-ABS-0080 Figure 1a). We tested our framework on a cohort of 360 LIHC patients from The Cancer Genome Atlas (TCGA) database.ResultsUsing ASD-LIHC, we identified two statistically distinct survival subtypes in these LIHC patients. Our framework provided improved risk-stratification (log-rank test, P = 8.12E-6) compared to traditional PCA methods (log-rank test, P = 0.87), accurately predicted survival subtypes, and identified important biomarkers for classifying survival subtypes, which are likely not sensitive about clinical stages (IDDF2023-ABS-0080 Figure 1b). Furthermore, our analysis revealed that the high-risk group had more cancer-related pathways compared to the low-risk group, and we identified potential pathways of interaction between microbes and genes that may play a role in LIHC survival (IDDF2023-ABS-0080 Figure 1c,d). For instance, Arcobacter,Methylocella, and Isoptericola may regulate host survival through interactions with host genes enriched in critical signaling pathways in cancer, particularly the HIF-1 signaling pathway, indicating these species as potential therapy targets to improve LIHC patient prognosis.ConclusionsOverall, our study sheds light on the complex interplay between microbes and genes in LIHC survival and has important implications for its monitoring, management, prevention, and treatment. Our findings may guide the development of specific treatments and future drug design, ultimately improving outcomes for patients with this devastating disease.Abstract IDDF2023-ABS-0080 Figure 1

• Publisher: BMJ Publishing Group Ltd and British Society of Gastroenterology,BMJ Publishing Group LTD
• Subject: Cancer therapies / Clinical Hepatology / Deep learning / Drug development / Genes / Genomics / Hepatocellular carcinoma / Immunotherapy / Liver / Liver cancer / Medical prognosis / Microbiomes / Molecular modelling / Patients / Prognosis / Risk groups / Signal transduction / Survival / Therapeutic targets / Transcriptomes / Tumor microenvironment / Tumors