An epigenetic biomarker for adult high-functioning autism spectrum disorder

Abstract

Increasing evidence suggests that epigenetic mechanisms play a role in the etiology of autism spectrum disorder (ASD). To date, several studies have attempted to identify epigenetic biomarkers for ASD. However, reliable markers remain to be established and most of these studies have focused on pediatric patients with ASD. In this study, we sought to find an epigenetic DNA methylation biomarker from peripheral blood for adult patients with high-functioning ASD. DNA methylation profiles were analyzed using the Illumina 450 K methylation array. To identify robust candidate markers, we employed two types of machine-learning algorithms for marker selection. We identified a potential marker (cg20793532) for which is the AUC value was 0.79. Notably, cg20793532 was annotated to the PPP2R2C gene, which was hypermethylated and down-regulated in blood from ASD patients compared to that in the controls. Although requiring careful interpretation, this pilot study seems to provide a potential blood biomarker for identifying individuals with high-functioning ASD.

Figure 1

The classification performance based on methylation profiling was performed with two machine-learning algorithms. (A) Variable importance ranking with the Mean Decrease Gini index calculated by the random forest (RF) algorithm. The 15 highest importance values are shown. (B) The Greedy Algorithm was applied to 2000 selections of random 80% subsets of samples. The eight most frequently chosen markers are shown with their selection rates.

Figure 2

Pyrosequencing validation of cg20793532. (A) Comparison of methylation levels of cg20793532 CpG1 between controls and ASD patients in the discovery set. Boxplots represent pyrosequencing-based methylation levels. The line within the box represents the median. Student’s t-test showed significant differences (P = 9.08E-5). (B) The receiver operating characteristic (ROC) curve analysis for evaluating the diagnostic performance of the methylation levels of cg20793532 CpG1. AUC with 95% confidence intervals (95% CI) was calculated (AUC = 0.79, 95% CI: 0.67–0.89). (C) Comparison of methylation levels of cg20793532 CpG2 between controls and ASD patients in the discovery set. Student’s t-test showed significant differences (P = 1.82E-4). (D) The ROC curve analysis of the methylation levels of cg20793532 CpG2. AUC with 95% confidence intervals (95% CI) was calculated (AUC = 0.78, 95% CI: 0.67–0.88).

Figure 3

Differential gene expression of PPP2R2C. PPP2R2C gene was significantly down-regulated in blood from ASD patients (n = 29) compared to that in the controls (n = 29). Microarray expression data was obtained from Kimura et al. Student’s t-test showed significant differences (P = 3.98E-4).