Identification of potential diagnostic biomarkers for Parkinson's disease

Abstract

The identification of biomarkers for early diagnosis of Parkinson's disease (PD) prior to the onset of symptoms may improve the effectiveness of therapy. To identify potential biomarkers, we downloaded microarray datasets of PD from the Gene Expression Omnibus database. Differentially expressed genes (DEGs) between PD and normal control (NC) groups were obtained, and the feature selection procedure and classification model were used to identify optimal diagnostic gene biomarkers for PD. A total of 1229 genes (640 up-regulated and 589 down-regulated) were obtained for PD, and nine DEGs (PTGDS, GPX3, SLC25A20, CACNA1D, LRRN3, POLR1D, ARHGAP26, TNFSF14 and VPS11) were selected as optimal PD biomarkers with great diagnostic value. These nine DEGs were significantly enriched in regulation of circadian sleep/wake cycle, sleep and gonadotropin-releasing hormone signaling pathway. Finally, we examined the expression of GPX3, SLC25A20, LRRN3 and POLR1D in blood samples of patients with PD by qRT-PCR. GPX3, LRRN3 and POLR1D exhibited the same expression pattern as in our analysis. In conclusion, this study identified nine DEGs that may serve as potential biomarkers of PD.

Keywords: Parkinson's disease; biomarker; differentially expressed genes; integrated analysis.

Figure 1

Identification of optimal gene biomarkers for PD. (A) The importance value of each DEG ranked according to the mean decrease in accuracy by using a random forest analysis. (B) The variance rate of classification performance when increasing numbers of the predictive DEGs. (C–K) Box‐plots displaying the expression levels of PTGDS (C), GPX3 (D), SLC25A20 (E), CACNA1D (F), LRRN3 (G), POLR1D (H), ARHGAP26 (I), TNFSF14 (J), and VPS11 (K); the y‐axis represents gene expression levels.

Figure 2

ROC results. (A–C) The ROC results of SVM (A), random forest (B) and decision tree (C) models based on the nine DEGs between PD and NCs. (D–F) The ROC results of SVM (D), random forest (E) and decision tree (F) models of the confirmation by dataset GSE72267 for these three models. The value before the parenthese represents the cut‐off. The values in the parenthese represent the specificity and sensitivity, respectively. The value below represents the AUC.

Figure 3

PPI network. The red and green ellipses represent proteins encoded by up‐ and down‐regulated differentially expressed mRNAs between PD and normal controls. Blue ellipses represent other proteins. Ellipses with a black border are the nine DEGs which were selected as the optimal diagnostic gene biomarkers for PD.

Figure 4

The qRT‐PCR results of DEGs between PD and NCs. (A) GPX3, (B) SLC25A20, (C) LRRN3, and (D) POLR1D. The P‐value > 0.05 was assessed by one‐way ANOVA. The error bars represent SD. n = 3.