Identification of serum microRNA signatures associated with autism spectrum disorder as promising candidate biomarkers

Abstract

Background MicroRNAs (miRNAs) are short non-coding RNA molecules with a well-recognized role in gene expression mostly at the post-transcriptional level. Recently, dysregulation of miRNAs and miRNA-mRNA interactions has been associated with CNS diseases, including numerous psychiatric disorders. Dynamic changes in the expression profiles of circulating miRNA are nowadays regarded as promising non-invasive biomarkers that may facilitate the accurate and timely diagnosis of complex conditions. Methods In this study, we investigated the gene expression patterns of four miRNAs, which were previously reported to be dysregulated in pooled serum samples taken from Autism Spectrum Disorder (ASD) patients and typically developing children. The performance of a diagnostic model for ASD based on these four miRNAs was assessed by a receiver operating characteristic (ROC) curve analysis, which evaluates the diagnostic accuracy of the investigated miRNA biomarkers for ASD. Finally, to examine the potential modulation of CNS-related biological pathways, we carried out target identification and pathway analyses of the selected miRNAs. Results Significant differential expression for all the four studied miRNAs: miR-500a-5p, miR-197-5p, miR-424-5p, and miR-664a-3p, was consistently measured in the samples from ASD patients. The ROC curve analysis demonstrated high sensitivity and specificity for miR-500a-5p, miR-197-5p, and miR-424-5p. With all miRNA expression data integrated into an additive ROC curve, the combination of miR-500a-5p and miR-197-5p provided the most powerful diagnostic model. On the other hand, the mRNA target mining showed that miR-424-5p and miR-500-5p regulate pools of target mRNA molecules which are enriched in a number of biological pathways associated with the development and differentiation of the nervous system. Conclusions The steady expression patterns of miR-500a-5p, miR-197-5p, miR-424-5p, and miR-664a-3p in ASD children suggest that these miRNAs can be considered good candidates for non-invasive molecular biomarkers in the study of ASD patients. The highest diagnostic potential is manifested by miR-500a-5p and miR-197-5p, whose combined ROC curve demonstrates very strong predictive accuracy.