Construction and validation of an immune infiltration-related risk model for predicting prognosis and immunotherapy response in low grade glioma.

BACKGROUND

Low grade glioma (LGG) is considered a heterogeneous tumor with highly variable survival and limited efficacy of immunotherapy. To identify high-risk subsets and apply immunotherapy effectively in LGG, the status and function of immune infiltration in the glioma microenvironment must be explored.

METHODS

Four independent glioma cohorts comprising 1,853 patients were enrolled for bioinformatics analysis. We used ConsensusClusterPlus to cluster patients into four different immune subtypes based on immune infiltration. The immune-infiltration signature (IIS) was constructed by LASSO regression analysis. Somatic mutation and copy number variation (CNV) analyses were performed to explore genomic and transcriptomic traits in the high- and low- risk groups. The correlation between response to programmed cell death 1 (PD-1) blockade and the IIS risk score was confirmed in an in vivo glioma model.

RESULTS

Patients were clustered into four different immune subtypes based on immune infiltration, and the high immune infiltration subtype was associated with worse survival in LGG. The high immune infiltration subtype had stronger inflammatory response, immune response and immune cell chemotaxis. The IIS, consisting of EMP3, IQGAP2, METTL7B, SLC1A6 and TNFRSF11B, could predict LGG malignant progression, which was validated with internal clinical samples. M2 macrophage infiltration positively correlated with the IIS risk score. The high-risk group had significantly more somatic mutations and CNVs. The IIS risk score was related to immunomodulatory molecules and could predict immunotherapy clinical benefit. In vivo, immunotherapy-sensitive glioma model exhibited higher IIS risk score and more infiltration of immune cells, especially M2 macrophages. The IIS risk score was decreased in an immunotherapy-sensitive glioma model after anti-PD1 immunotherapy.

CONCLUSION

Different immune subtypes of LGG had unique immune cell infiltration characteristics, and the high immune infiltration subtype was associated with immunosuppressive signaling pathways. A novel IIS prognostic model based on immune infiltration status was constructed for immunophenotypic classification, risk stratification, prognostication and immunotherapy response prediction in LGG.