A 16-miRNA Prognostic Model to Predict Overall Survival in Neuroblastoma

Abstract

Neuroblastoma is the most malignant childhood tumor. The outcome of neuroblastoma is hard to predict due to the limitation of prognostic markers. In our study, we constructed a 16-miRNA prognostic model to predict the overall survival of neuroblastoma patients for early diagnosis. A total of 205 DE miRNAs were screened using RNA sequencing data from GSE121513. Lasso Cox regression analysis generated a 16-miRNA signature consisting of hsa-let-7c, hsa-miR-135a, hsa-miR-137, hsa-miR-146a, hsa-miR-149, hsa-miR-15a, hsa-miR-195, hsa-miR-197, hsa-miR-200c, hsa-miR-204, hsa-miR-302a, hsa-miR-331, hsa-miR-345, hsa-miR-383, hsa-miR-93, and hsa-miR-9star. The concordance index of multivariate Cox regression analysis was 0.9, and the area under the curve (AUC) values of 3-year and 5-year survival were 0.92 and 0.943, respectively. The mechanism was further investigated using the TCGA and GSE90689 datasets. Two miRNA–gene interaction networks were constructed among DEGs from two datasets. Functional analysis revealed that immune-related processes were involved in the initiation and metastasis of neuroblastoma. CIBERSORT and survival analysis suggested that lower CD8 T-cell proportion and higher SPTA1 expressions were related to a better prognosis. Our study demonstrated that the miRNA signature may be useful in prognosis prediction and management improvement.