Identification of an Autophagy-Related Risk Signature Correlates With Immunophenotype and Predicts Immune Checkpoint Blockade Efficacy of Neuroblastoma.

Abstract

Neuroblastoma is one of the malignant solid tumors with the highest mortality in childhood. Targeted immunotherapy still cannot achieve satisfactory results due to heterogeneity and tolerance. Exploring markers related to prognosis and evaluating the immune microenvironment remain the major obstacles. Herein, we constructed an autophagy-related gene (ATG) risk model by multivariate Cox regression and least absolute shrinkage and selection operator regression, and identified four prognostic ATGs (BIRC5, GRID2, HK2, and RNASEL) in the training cohort, then verified the signature in the internal and external validation cohorts. BIRC5 and HK2 showed higher expression in MYCN amplified cell lines and tumor tissues consistently, whereas GRID2 and RNASEL showed the opposite trends. The correlation between the signature and clinicopathological parameters was further analyzed and showing consistency. A prognostic nomogram using risk score, International Neuroblastoma Staging System stage, age, and MYCN status was built subsequently, and the area under curves, net reclassification improvement, and integrated discrimination improvement showed more satisfactory prognostic predicting performance. The ATG prognostic signature itself can significantly divide patients with neuroblastoma into high- and low-risk groups; differentially expressed genes between the two groups were enriched in autophagy-related behaviors and immune cell reactions in gene set enrichment analysis (false discovery rate q -value < 0.05). Furthermore, we evaluated the relationship of the signature risk score with immune cell infiltration and the cancer-immunity cycle. The low-risk group was characterized by more abundant expression of chemokines and higher immune checkpoints (PDL1, PD1, CTLA-4, and IDO1). The risk score was significantly correlated with the proportions of CD8+ T cells, CD4+ memory resting T cells, follicular helper T cells, memory B cells, plasma cells, and M2 macrophages in tumor tissues. In conclusion, we developed and validated an autophagy-related signature that can accurately predict the prognosis, which might be meaningful to understand the immune microenvironment and guide immune checkpoint blockade.