Multi-omics clustering combined with multiple machine learning to identify epigenetic features in low-grade glioma patients.

BACKGROUND

Epigenetic heterogeneity has been demonstrated in a wide range of cancers including leukemia, colorectal cancer, and glioma; in the case of low-grade glioma (LGG), its association with tumor malignant progression and immune microenvironmental regulation has been less reported. This study aimed to further explore the application of epigenetics in LGG.

METHODS

Epigenetic subtypes were identified based on consensus clustering of multi-omics data, and prognostic models were constructed based on 101 machine learning combinations. Biological differences among different risk groups were explored by functional enrichment analysis, and finally Jumonji Domain-Containing 8 (JMJD8) was identified as a potential molecular marker in LGGs by pan-cancer survival analysis.

RESULTS

We revealed three epigenetic molecular subtypes in LGG patients, and the chromosomal regulator-associated risk score, defined as an independent prognostic factor for LGGs, was effective in predicting the response to treatment in LGG patients. JMJD8, a key model gene, was also strongly associated with the prognosis of LGG patients.

CONCLUSIONS

This study reveals the molecular pattern and prognostic model of epigenetic inheritance in LGG, and the key gene JMJD8 is a potential prognostic biomarker of LGG, which is a guide for the clinical application of epigenetics.