Related Experiment Video
Updated: May 28, 2026
Detection of a Circulating MicroRNA Custom Panel in Patients with Metastatic Colorectal Cancer
Published on: March 14, 2019
DeepMetabio-mCRC Screener: A Multi-Omics Deep Learning Framework for Early Risk Prediction and Biomarker Discovery in
Hongyu Zhang1, Ke Wang2, Runqiu Guo1
1Zhejiang Province Key Laboratory of Anti-Cancer Drug Research, College of Pharmaceutical Science, Innovation Institute for Artificial Intelligence in Medicine of Zhejiang University, Zhejiang University, Hangzhou 310058, China.
A new DeepMetabio-mCRC Screener tool accurately predicts colorectal liver metastasis (CRLM) risk using multi-omics data. It identifies aminocarboxymuconate-semialdehyde decarboxylase (ACMSD) as a key biomarker for diagnosis and treatment decisions in colorectal cancer (CRC).
Area of Science:
- Oncology
- Metabolomics
- Bioinformatics
- Machine Learning
Background:
- Colorectal liver metastasis (CRLM) is the leading cause of mortality in colorectal cancer (CRC) patients.
- Current predictive tools and biomarkers for CRLM are insufficient, necessitating advanced diagnostic approaches.
- Metabolic reprogramming plays a crucial role in CRC progression and metastasis.
Purpose of the Study:
- To develop an integrated multi-omics framework, DeepMetabio-mCRC Screener, for early prediction of CRLM.
- To identify novel metabolic biomarkers associated with CRLM development and progression.
- To validate the clinical utility of the developed screener and identified biomarkers in CRC patients.
Main Methods:
- Trained a convolutional neural network using 620 metabolism-related genes from 1,077 CRC transcriptomic profiles.
- Integrated transcriptomic data with serum metabolomics to identify core metabolic features linked to metastasis.
- Validated the DeepMetabio-mCRC Screener model performance against established machine learning models.
- Clinically validated the identified biomarker, aminocarboxymuconate-semialdehyde decarboxylase (ACMSD), in CRC patient cohorts.
Main Results:
- The DeepMetabio-mCRC Screener achieved high predictive accuracy (AUC 0.92-0.97), outperforming existing models.
- Identified 22 core metabolic features, including retinol and tryptophan metabolism, associated with CRLM.
- Elevated ACMSD levels in CRLM patients correlated with advanced stage, recurrence risk, and immune-inflamed TME.
- ACMSD knockdown suppressed CRC cell migration and reduced pro-inflammatory and immune-responsive pathways in vitro.
Conclusions:
- DeepMetabio-mCRC Screener is a robust tool for early risk prediction of CRLM.
- ACMSD is a promising multifunctional biomarker for CRLM diagnosis, prognosis, and therapeutic guidance.
- Targeting ACMSD may offer a novel therapeutic strategy for managing CRLM by modulating cell migration and immune response.
