Related Experiment Video
Updated: Apr 15, 2026
Author Spotlight: Finding New Therapeutic Targets for Malignant Peripheral Nerve Sheath Tumor Through Genome-Scale shRNA Screens
Published on: August 25, 2023
Leveraging Network-Based Transcriptome Analysis from Mouse Tumor Models and Explainable Artificial Intelligence to
Haruna Imamura1, Sufeng Chiang1, Megumi Kuronishi2
1The Systems Biology Institute, Saisei Ikedayama Bldg. 5-10-25 Higashi Gotanda, Shinagawa 141-0022, Tokyo, Japan.
Background/Objectives:
Understanding the mechanisms of drug response plays an essential role in predicting effects prior to drug administration and advancing personalized medicine by optimizing treatment strategies. This study aimed to identify gene combinations that can predict the antitumor activity of lenvatinib, which is a multi-targeted tyrosine kinase inhibitor.
Methods:
Cancer- and drug-response-related gene sets were identified by mapping gene expression profiles of previously reported syngeneic mouse tumor models onto a protein-protein interaction network and extracting subnetworks comprising nodes where high expression levels were clustered. The scores for these network modules were calculated using the expression data of mouse tumor models prior to drug administration. These scores were used to train a machine learning (ML) model of drug response to lenvatinib by narrowing down the parameter space using hepatocellular carcinoma patient-derived xenograft (HCC PDX) models acquired in this study.
Results:
Using this integrative framework, we identified several network modules including those involved in the nerve growth factor signaling pathway, Wnt signaling pathway, and interleukin signaling pathways, that were consistently prioritized as informative features across PDX models and human patient data from The Cancer Genome Atlas.
Conclusions:
These network modules exhibit biological functions that are linked to the known targets of lenvatinib in the cancer cells or the tumor microenvironment, highlighting their potential relevance as determinants of drug response.
Insights
This study identifies key gene network modules, including those in nerve growth factor and Wnt signaling pathways, that predict lenvatinib antitumor activity for personalized cancer medicine.
Area of Science:
- Oncology
- Systems Biology
- Pharmacogenomics
Background:
- Understanding drug response mechanisms is crucial for personalized medicine.
- Lenvatinib is a multi-targeted tyrosine kinase inhibitor used in cancer treatment.
Purpose of the Study:
- Identify gene combinations predicting lenvatinib's antitumor activity.
- Advance personalized medicine by optimizing cancer treatment strategies.
Main Methods:
- Integrated gene expression profiles from mouse models and HCC PDX models.
- Utilized protein-protein interaction networks to identify relevant gene modules.
- Employed machine learning to train a drug response prediction model.
Main Results:
- Identified network modules in nerve growth factor, Wnt, and interleukin signaling pathways.
- These modules were prioritized across patient-derived xenograft and TCGA data.
- Modules showed consistent relevance as predictive features.
Conclusions:
- Identified network modules are biologically linked to lenvatinib targets.
- These modules may serve as determinants of lenvatinib response.
- Findings support the potential for predicting lenvatinib efficacy.
