Novel Score Predicts Liver Cancer Risk with 93% Accuracy
Researchers have developed a new machine-learning model that predicts the risk of liver cancer, specifically hepatocellular carcinoma (HCC), with 93% accuracy. The study, led by Xian-Yang Qin at the RIKEN Center for Integrative Medical Sciences (IMS) in Japan, identifies a “MYCN niche” – a cluster of genes associated with increased MYCN expression – as a key indicator of tumor development and recurrence. Published in Proceedings of the National Academy of Sciences, the findings offer a potential strategy for early detection and improved clinical outcomes for this deadly cancer.
The Challenge of Liver Cancer
Liver cancer is the third leading cause of cancer death worldwide, responsible for over 800,000 deaths annually. The American Cancer Society reports that the high mortality rate is due to late detection and a recurrence rate between 70% and 80%. Early identification of at-risk individuals is crucial for effective intervention.
MYCN: A Key Driver of Tumorigenesis
The research focuses on the MYCN protein, a known contributor to liver cancer development, particularly in damaged livers. Researchers aimed to understand how MYCN contributes to cancer, hypothesizing that its overexpression could serve as a biomarker. To investigate this, the team used a hydrodynamic tail vein injection-based transposon system to induce MYCN overexpression in mice.
The study found that combining MYCN overexpression with always-active AKT led to tumor development in 72% of mice within 50 days. These tumors exhibited characteristics consistent with human hepatocellular carcinoma. Overexpressing either gene alone did not result in tumor formation, highlighting the synergistic effect of MYCN and AKT.
Identifying the MYCN Niche with Spatial Transcriptomics
To understand the early microenvironmental cues that trigger liver tumorigenesis, the researchers employed spatial transcriptomics. This technique maps gene expression within a tissue, revealing where specific genes are active. Analyzing gene expression over time in a mouse model of metabolic dysfunction-associated liver cancer, they identified a cluster of 167 genes differentially expressed in tumor-free liver sections with increased MYCN levels. This cluster was termed the “MYCN niche.”
A Machine-Learning Model for Risk Prediction
Based on the spatial transcriptomics data, the researchers developed a machine-learning model that assesses gene-expression patterns and generates a “MYCN niche score” indicating the likelihood of a tumor developing. The model demonstrated 93% accuracy in identifying MYCN niches.
MYCN Niche Score Predicts Recurrence and Poor Outcomes in Humans
Applying the MYCN niche score to human hepatocellular carcinoma datasets revealed a strong correlation between higher scores and increased risk of tumor recurrence and poorer clinical outcomes. Notably, the score was more predictive when derived from non-tumor tissue than from tumor tissue, suggesting its potential for identifying precancerous microenvironments.
Future Directions
“We have developed a clinically actionable strategy to identify high-risk patients by profiling gene expression in non-tumor liver tissue,” said Xian-Yang Qin, of the RIKEN Center for Integrative Medical Sciences. “By integrating spatial transcriptomics with machine learning, we have established a MYCN niche score that predicts recurrence risk and detects precancerous microenvironments predisposed to de novo liver tumorigenesis.”
Future research will focus on elucidating the biological mechanisms underlying the machine-learning-derived spatial feature scores and understanding how cancer-permissive environments are established and maintained.
About Xian-Yang Qin
Xian-Yang Qin is a Senior Research Scientist at the RIKEN Center for Integrative Medical Sciences Laboratory for Cellular Function Conversion Technology in Japan. His research focuses on the MYCN oncogene and its role in hepatocellular carcinoma. RIKEN is a leading research institution in Japan.