Blood Biomarkers Associated with Cognitive Decline in Early Stage and Drug-Naive Parkinson’s Disease Patients

Early diagnosis of Parkinson’s disease (PD) continues to be a major challenge in the field. The lack of a robust biomarker to detect early stage PD patients has considerably slowed the progress toward the development of potential therapeutic agents. We have previously evaluated several RNA biomarkers in whole blood from participants enrolled in two independent clinical studies. In these studies, PD patients were medicated, thus, expression of these biomarkers in de novo patients remains unknown. To this end, we tested ten RNA biomarkers in blood samples from 99 untreated PD patients and 101 HC nested in the cross-sectional Parkinson’s Progression Markers Initiative by quantitative real-time PCR. One biomarker out of ten, COPZ1 trended toward significance (nominal p = 0.009) when adjusting for age, sex, and educational level. Further, COPZ1, EFTUD2 and PTBP1 mRNAs correlated with clinical features in PD patients including the Hoehn and Yahr scale, Movement Disorder Society revision of Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) and Montreal Cognitive Assessment (MoCA) score. Levels of EFTUD2 and PTBP1 were significantly higher in cognitively normal PD patients (PD-CN) compared to cognitively impaired PD patients (PD-MCI). Interestingly, blood glucose levels were significantly higher in PD and PD-MCI patients (≥ 100 mg/dL, pre-diabetes) compared to HC. Collectively, we report the association of three RNA biomarkers, COPZ1, EFTUD2 and PTBP1 with clinical features including cognitive decline in early drug-naïve PD patients. Further, our results show that drug-naïve PD and PD-MCI patients have glucose levels characteristic of pre-diabetes patients, suggesting that impaired glucose metabolism is an early event in PD. Evaluation of these potential biomarkers in a larger longitudinal study is warranted.


Introduction
Parkinson's disease (PD) is a debilitating neurodegenerative disease characterized by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta. Although PD has been largely recognized as a classical movement disorder, non-motor symptoms including

Study Participants
PPMI is an ongoing 5-year observational, longitudinal, and multi-center study of drug-naïve patients with early stage PD (n = 423) and HC (n = 196) recruited from 32 clinical sites in Australia, United States of America (USA) and Europe [18]. A power analysis based on each biomarker from our previous studies [12][13][14][15], indicated that a minimum fold change of 1 between PD and controls could be detected with a 95% power using 100 samples per group. Thus, we selected a total of 200 participants, 99 PD patients and 101 age and sex matched HC nested in the PPMI study PD patients that had a dopamine transporter deficit assessed by DaTscan imaging. PD patients with a Hoehn and Yahr stage I or II were chosen for this study so that we could relate the results to our earlier studies on medicated patients at these stages of the disease. Healthy individuals were cognitively normal, free of neurological disorder, and with no detectable dopamine transporter deficit. Among the 99 PD patients, we selected 35 patients with cognitive impairment as defined by a Montreal Cognitive Assessment (MoCA) score of less than 26. Demographic and clinical characteristics about the study participants are shown in Table 1. The Institutional Review Boards of Rosalind Franklin University of Medicine and Science and all sites participating in the PPMI study approved the study. Written informed consent was obtained from all participants before inclusion in the study. All participants were evaluated for

Statistical Analysis
Statistical analysis was performed using STATISTICA 12 (StatSoft, OK, USA) and GraphPad Prism version 5 (GraphPad Software, Inc., CA, USA). A Student-t-test (unpaired, two tailed) was used to assess the differences between two groups and a chi-square test was used to analyze categorical data. Pearson correlation was performed for all correlations. A Pearson correlation analysis of the RIN values and expression of each biomarker showed no significant associations (p = 0.9). The relative abundance of each biomarker was independently assessed using a general linear regression model adjusting for age, sex, and educational level. A Bonferroni adjustment was used to correct for multiple comparisons and a p-value of 0.005 or less was considered significant.
In addition, we evaluated the expression of each RNA biomarker independently in PD patients with cognitive decline compared to cognitively normal PD patients. For this analysis, we used a general linear regression model adjusting for age, gender, education, Hoehn and Yahr, and MDS-UPDRS. A receiver operating characteristic curve (ROC) analysis was performed to determine the diagnostic accuracy of each biomarker. A nominal p-value of 0.05 or less was considered significant

Demographic characteristics of study participants
There were no significant differences in mean age and sex distribution between PD patients and HC. However, PD patients had a small but significant difference in the number of years they attended school compared to HC ( Table 1). As expected, comparisons of clinical features showed significant differences in Hoehn and Yahr stage, Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS), MoCA score, University of Pennsylvania Smell Identification Test (UPSIT), Geriatric Depression Scale (GDS) and Scale for Outcomes in Parkinson's disease for Autonomic Symptoms (SCOPA) scores between PD patients and controls (Table 1). We also compared the blood glucose levels in PD patients and controls. Mean blood glucose level was within the normal range in HC (<100 mg/dL) but was significantly higher than normal in the PD group ( 100 mg/dL) ( Table 1).

Evaluation of RNA biomarkers in blood of PPMI study participants
A total of ten RNA biomarkers were tested in the PPMI cohort by quantitative PCR assays. No significant difference was observed in the expression of 9 of the markers, including EFTUD2, PTBP1, C5ORF4, APP, SOD2, HNF4A, WLS, ZNF160 and MACF1, between PD patients and controls in the univariate analysis (S1 Table). Interestingly however, a small but significant difference in the expression of COPZ1 was identified in the blood of PD patients compared to HC. Specifically, the relative abundance of COPZ1 (p = 0.005) was significantly up-regulated in PD patients compared to HC, although overlap in expression levels between the two groups was observed (Fig 1, S1 Table). In addition, the analysis was performed using a general linear regression model adjusting for confounding variables including age, sex and educational level. Relative abundance of COPZ1 trended toward significance in PD patients compared to HC after adjusting for confounding variables (ß = 0.18, nominal p = 0.009).
Pearson correlation analysis was performed to determine whether the relative abundance of each RNA biomarker was independent of age, gender and educational level and clinical outcome variables. The relative abundance of COPZ1 was independent of age, gender and education (S2 Table). Interestingly, the relative abundance of three biomarkers, COPZ1, EFTUD2 and PTBP1 correlated with measures of disease severity in PD patients. For instance, COPZ1 correlated positively with the Hoehn and Yahr scale, MDS-UPDRS total, and MDS-UPDRS parts II and III. Relative abundance of EFTUD2 correlated negatively with age, Hoehn and Yahr scale, MDS-UPDRS total, MDS-UPDRS parts I, II and III and correlated positively with MoCA (S2 Table). Relative abundance of PTBP1 correlated negatively with MDS-UPDRS parts I and positively with MoCA. In addition, COPZ1 and EFTUD2 correlated negatively with the UPSIT and GDS scores, respectively.
EFTUD2 and PTBP1 are associated with cognitive decline in PD Since cognitive decline is common in PD, we evaluated the clinical differences between cognitively normal PD patients (PD-CN) and PD patients with mild cognitive impairment (PD-MCI). Cognitive impairment was defined using the recommended MoCA cutoff greater than 26 as shown previously [20,21]. HC were excluded from this analysis. There were no significant differences in age, educational level, disease duration, and disease severity scales including the Hoehn and Yahr and MDS-UPDRS between PD-CN and PD-MCI (Table 2). However, there were a significant higher number of male participants in the PD-MCI group (Table 2). Interestingly, mean blood glucose levels are significantly higher than normal in PD-MCI (mean (SD) 95% CI: 105.8 (15.27) [100.6-111.0]) compared to PD-CN (Table 2). Relative abundance of COPZ1 mRNA in the PPMI study. Relative abundance of COPZ1 in PD patients (black circles) compared to HC (white circles). Relative abundance of COPZ1 was calculated using GAPDH mRNA as a reference gene. Error bars represent the 95% confidence interval. A p-value of less than 0.005 was regarded as significant based on a Student t-test (two-tailed). Pearson correlation analysis showed significant and positive correlations among COPZ1 EFTUD2 and PTBP1. For example, COPZ1 correlated with EFTUD2 (r = 0.14, p = 0.04) and PTBP1 (r = 0.32, p = 0.0001). The highest correlation value was observed between EFTUD2 and PTBP1 (r = 0.61, p = 0.001).

Discussion
Validation of candidate biomarkers for PD in several independent clinical studies is an important step toward their effective translation to the clinic. One crucial step in this process is the evaluation of biomarkers in drug-naïve patients to ensure that commonly prescribed medications for PD do not have an effect in their expression. Here we tested ten blood RNA biomarkers that were evaluated previously in two independent clinical studies, in blood of untreated PD patients and controls enrolled in the PPMI study.
Three candidate blood biomarkers, COPZ1, EFTUD2 and PTBP1, correlated with clinical features in early stage drug-naïve PD patients suggesting they may be useful for patient stratification according to disease severity measures. Specifically, COPZ1 significantly correlated with Hoehn and Yahr scale, MDS-UPDRS total, MDS-UPDRS parts II and III and UPSIT score. Similarly, EFTUD2 showed significant correlations with Hoehn and Yahr, MDS-UPDRS total, MDS-UPDRS parts I, II, and III, MoCA and GDS scores. In addition, PTBP1 significantly correlated with MDS-UPDRS part I, and MoCA score. These findings suggest that COPZ1, EFTUD2 and PTBP1 may be useful biomarkers to classify PD patients with regards to symptoms. Nonetheless, these results need to be taken with caution since the biomarker correlations with clinical features were relatively weak. Thus, evaluation of these potential biomarkers in a larger longitudinal study will be required to assess their prognostic utility.
One biomarker, COPZ1 trended toward significance (nominal p = 0.009) after adjusting for age, sex and educational level. It is important to note, that we observed overlap in the relative abundance of COPZ1 between PD and controls and low diagnostic capacity. In addition, we previously reported that the relative abundance of COPZ1 was up-regulated in blood of medicated PD patients enrolled in two independent clinical studies [11,12]. It should be noted that the magnitude of the change in COPZ1 expression was greater in our earlier reports using medicated PD patients compared to unmedicated patients in this report (5-fold vs 1.5-fold). Together these results suggest that medication has an impact on the expression of COPZ1.
Cognitive decline is a common non-motor aspect of PD that frequently progresses to dementia. In fact, up to 80% of PD patients with mild cognitive impairment will eventually develop dementia [22]. Thus, biomarkers that predict changes in cognitive abilities are essential for the development of therapeutic strategies for PD [23]. Several potential biomarkers of cognitive decline have been identified. For example, CSF levels of amyloid beta peptides (Aß), but EFTUD2 and PTBP1 mRNAs as biomarkers for cognitive decline in PD. A. Relative abundance of EFTUD2 in PD patients with normal cognition (circles) compared to PD patients with mild cognitive impairment (triangles). B. Relative abundance of PTBP1 in PD patients with normal cognition (circles) compared to PD patients with mild cognitive impairment (triangles). Error bars represent the 95% confidence interval. A p-value of less than 0.05 was regarded as significant based on a Student t-test (two-tailed). PD-CN is cognitively normal PD patients, PD-MCI is PD patients with mild cognitive impairment.
doi:10.1371/journal.pone.0142582.g002 not total tau, were associated with cognitive impairment in untreated PD patients enrolled in the Norwegian ParkWest study [24]. Contrary to these findings, Aß peptides, total tau and phosphorylated tau (p-tau) did not correlate with cognitive measures in untreated patients in a larger cohort of patients enrolled in Deprenyl and Tocopherol Antioxidative Therapy of Parkinsonism (DATATOP) study. However, higher p-tau and p-tau/Aß42 levels in the CSF predicted cognitive decline in PD patients after levodopa treatment [25]. In addition, serum levels of the insulin growth factor 1 (IGF-1) and epidermal growth factor (EGF) are associated with cognitive decline in early stage and drug-naïve patients with PD [26][27][28][29][30]. Approximately 10% of PD patients met the criteria for cognitive impairment in the PPMI study [21]. In this study, relative abundance of EFTUD2 and PTBP1 was significantly up-regulated in PD-CN compared to PD-MCI independently of age, gender, disease duration, and disease severity measures. To the best of our knowledge, these are the first blood RNA biomarkers that have been associated with cognitive decline in PD. Future studies in larger and well-characterized longitudinal studies will be important to assess their predictive performance in PD patients.
The biomarkers examined in this study are related to biological processes associated with PD. For example, COPZ1 encodes a subunit of the cytoplasmic coatomer complex, which is involved in autophagy and protein trafficking. EFTUD2, elongation factor Tu GTP binding domain containing 2, which encodes the splicing factor U5-116 kD, is involved in RNA splicing via the spliceosome. In this regard, the involvement of aberrant RNA splicing in blood of PD has been noted in numerous studies [7,11,12,15], reviewed in [31]. In this regard, a recent study identified a network of dysregulated splicing factors in blood of PD patients supporting the idea that splicing may be both inefficient and dysregulated in PD [15].
The polyprimidine tract binding protein1 (PTBP1) belongs to the family of heterogeneous nuclear ribonucleoproteins (hnRNPs) that are involved in pre-mRNA processing, metabolism and transport. PTBP1 plays a pivotal role in the mRNA stabilization and translation of insulin in the pancreas and it has been associated with diabetes [32]. Thus, altered expression of PTBP1 may be an important indicator of impaired glucose and insulin signaling in de novo PD patients.
In this context, mounting evidence indicates that impaired glucose metabolism, insulin signaling and diabetes are associated with PD [13,33,34]. Treatment with levodopa exacerbates glucose intolerance in PD patients [35,36], but abnormal blood glucose levels in untreated PD patients have not yet been reported. According to the American Diabetes Association (http:// www.diabetes.org/diabetes-basics/diagnosis/), fasting blood glucose levels 100 mg/dL are characteristic of subjects with pre-diabetes. Accordingly, blood glucose levels were significantly higher than normal in PD (101.6 mg/dL) and PD-MCI (105.8 mg/dL) patients. Thus, we report for the first time that drug-naïve PD and PD-MCI patients have glucose levels characteristic of pre-diabetes patients, suggesting that impaired glucose metabolism is an early event in PD. These results are representative of a subset of participants enrolled in PPMI study and therefore warrant further investigation.
Collectively, we report the association of three RNA biomarkers with clinical features including cognitive decline in early drug-naïve PD patients. Nevertheless, the diagnostic performance of these biomarkers is low and below the ideal characteristics for a diagnostic biomarker. There are several important factors that need to be considered when interpreting the results from this study and its discrepancies from our previous studies. First, our biomarkers were tested previously in very homogenous clinical studies in the USA [11,12,14,15], where most of the PD patients were medicated. PPMI is an international multicenter study of drugnaïve PD patients and HC recruited from 32 different clinical sites including Europe, Australia and USA. Thus, other factors related to the diverse ethnic backgrounds, diet, and epigenetic changes, for example, may explain the discrepancies in biomarker expression observed in this study. In addition, even with the rigorous standard procedures for sample collection, processing and storage followed by the PPMI study, there is a potential for pre-analytical variability introduced by the different clinical sites [37].
Unbiased approaches are needed to identify highly sensitive and specific diagnostic biomarkers for the identification of early stage and pre-symptomatic PD patients [38]. We foresee that emerging high-throughput technologies combined with network analyses using early stage untreated and/or pre-symptomatic PD patients will be instrumental in the discovery of new potential biomarkers [16]. Evaluation of these biomarkers in larger longitudinal studies is needed to test further their prognostic and diagnostic utility.
Supporting Information S1 Table. Relative expression levels of blood biomarkers in HC and patients with PD. a Assessed by Student t-test (two-tailed). CI is the 95% confidence interval. A p-value of 0.005 or less was considered significant. (DOC)