Phenotype of postural instability/gait difficulty in Parkinson disease: relevance to cognitive impairment and mechanism relating pathological proteins and neurotransmitters

Parkinson disease (PD) is identified as tremor-dominant (TD) and postural instability and gait difficulty (PIGD) phenotypes. The relationships between motor phenotypes and cognitive impairment and the underlying mechanisms relating pathological proteins and neurotransmitters in cerebrospinal fluid (CSF) are unknown. We evaluated the motor symptoms and cognitive function by scales, and detected the levels of pathological proteins and neurotransmitters in CSF. TD group and PIGD group had significantly higher levels of total tau, tau phosphorylated at the position of threonine 181(P-tau181t), threonine 231, serine 396, serine 199 and lower β amyloid (Aβ)1–42 level in CSF than those in control group; PIGD group had significantly higher P-tau181t level and lower Aβ1–42 level than those in TD group. In PD group, PIGD severity was negatively correlated with MoCA score and Aβ1–42 level in CSF, and positively correlated with Hoehn-Yahr stage and P-tau181t level in CSF. In PIGD group, PIGD severity was negatively correlated with homovanillic acid (HVA) level in CSF, and HVA level was positively correlated with Aβ1–42 level in CSF. PIGD was significantly correlated with cognitive impairment, which underlying mechanism might be involved in Aβ1–42 aggregation in brain and relevant neurochemical disturbance featured by the depletion of HVA in CSF.


Methods
Subjects. Patients with PD. In total, 520 PD patients were recruited from the Department of Geriatrics and Neurology, Beijing Tiantan Hospital, Capital Medical University. Demographic information, including age, age of onset, sex, education, disease duration, predominantly affected side, disease severity, Montreal Cognitive Assessment (MoCA)-Beijing version 25 , Mini-Mental State Examination (MMSE) and levodopa equivalent daily dose (LED) were recorded. Patients were diagnosed with PD according to the criteria of Parkinson's UK Brain Bank 26 .
Control subjects. In total, 28 age-matched controls from the Department of Geriatrics and Neurology, Beijing Tiantan Hospital were selected. The inclusion criteria were as follows: (1) no neurological symptoms and signs; (2) no intracranial diseases; (3) no essential tremor, PD, secondary parkinsonism, or Parkinson-plus syndrome; (4) no dysarthria or mental illness that affected expression; (5) no alcohol or drug abuse. All these controls had no family history of PD or tremor.
The controls were also patients, but their diseases, such as peripheral neuropathy and headache caused by high intracranial pressure, etc, were not related to and had no effect on the results of this investigation.

Assessments of disease severity and motor symptoms of PD patients. The severity of PD was assessed by Hoehn and Yahr (H-Y) stage.
Each PD patient was evaluated using the items in the Unified Parkinson's Disease Rating Scale (UPDRS) III, in which items 20 and 21 were for tremor, item 22 was for rigidity, items 23-26 and 31 were for bradykinesia, and items 27-30 were for PIGD. The score for each motor symptom was calculated by summing up the scores of corresponding items in UPDRS III.
Motor phenotypes were determined as either TD phenotype (n = 309) or PIGD (n = 211) phenotype following the classification algorithm proposed by Jankovic et al. 27 . Following the original classification methods, the ratio of the mean UPDRS tremor scores (8 items) to the mean UPDRS PIGD scores (5 items) was used to define TD phenotype (ratio ≥ 1.5), PIGD phenotype (ratio ≤ 1), or indeterminate phenotype (ratios > 1.0 and <1.5). Finally, total 520 patients with PD were eligible for this study with 60 cases excluded due to the indeterminate motor phenotype.
Assessment of cognitive function for PD patients. Cognitive function of PD patients was evaluated by the MoCA-Beijing version 25 and MMSE. The MoCA-Beijing required education adjustment, i.e., one point was added to the total score for those with education < 12 years 28 . The MoCA-Beijing had 7 cognitive domains, including visuospatial/executive function, attention, recall, orientation, abstract, language and naming. Patients with MoCA-Beijing score > 26 points were defined as cognitively normal. We compared the score of each item and the percentage of completely correct in each item of MoCA-Beijing between PIGD group and TD group. Beijing Tiantan Hospital review board (KY2013-003-03) has approved this study. All participants completed the written informed consents. This study met the guidelines of Capital Medical University, which abode by the Helsinki Declaration on ethical principles for medical research involving human subjects.
Collections of CSF and serum samples. Before fasting, CSF (3 mL in a polypropylene tube via lumbar puncture) was collected from patients with PD whose conditions allowed us to withdraw the antiparkinsonian drugs for 12 to14 hours. Approximately 0.5 ml volume of CSF was aliquotted into separate Nunc cryotubes and kept frozen at − 80 °C until used in assays. Each aliquot dedicated for each measure to avoid freeze-thawing and potential degradation of protein.
Measurements of the levels of pathological proteins in CSF. The levels of pathological proteins, including Aβ 1-42 , T-tau, P-tau181t, P-tau396s and tau phosphorylated at the following positions: threonine 231 (P-tau231t) and serine 199 (P-tau199s), in CSF from PD patients and control participants were determined by using an enzyme-linked immunosorbent assay.

Measurements of the levels of neurotransmitters in CSF.
The levels of DA and its two metabolites of HVA, DOPAC, Ach, serotonin (5-HT) and norepinephrine (NE) in CSF from PD patients were tested by high performance liquid chromatography (HPLC). Henomenex 150*2 mm, 150*3 mm chromatographic columns and LC-MS-MS 6410 chromatographic instrument were from Agilent Company (USA), and standard sample was from Sigma Company (USA).

Data analyses.
Statistical analyses were performed with SPSS Statistics 20.0 (IBM Corporation, New York, USA).
Demographic information, motor symptoms and MoCA score were compared between TD and PIGD groups. The levels of Aβ 1-42 , T-tau, P-tau181t, P-tau231t, P-tau396s and P-tau199s in CSF were compared among control, TD and PIGD groups.
Continuous variables, if they were normally distributed, were presented as means ± standard deviations and compared by ANOVA test. Bonferroni correction was performed in further comparisons between two groups. P value was significant when it was < 0.017. Continuous variables, if they were not normally distributed, were presented as median (quartile) and compared by nonparametric test. P value was significant when it was < 0.017 in further comparisons between two groups. Discrete variables were compared by Chi square test.
Spearman correlation analyses were made between PIGD score and the levels of Aβ 1-42 , T-tau, P-tau181t, P-tau231t, P-tau396s, and P-tau199s in CSF, between PIGD score and the levels of neurotransmitters in CSF, and between the levels of Aβ 1-42 and P-tau181t and HVA level in CSF in PD group.
Multiple linear regression models were established, in which PIGD score in PD group was set as dependent variable, whereas age, age of onset, sex, education, predominantly affected side, MoCA score, disease duration, H-Y stage, Aβ 1-42 and P-tau181t were set as independent variables. P value was significant when it was < 0.05.

Results
Demographic and clinical variables of control, TD and PIGD groups. Among the 520 PD patients,  Table 1. PIGD group displayed significantly higher PIGD score (p < 0.001) and lower tremor score (p < 0.001) than the TD group, as well as higher scores of rigidity (p = 0.04) and bradykinesia (p < 0.001). PIGD group had significantly more patients at H-Y stage of 4-5 than TD group.
The percentage of completely correct in the item of recall in MoCA in PIGD group was significantly lower than that in TD group (p = 0.04). No statistical differences were seen in age, age of onset, gender, education, predominantly affected side and LED. There were no significant differences in total scores of MoCA, MMSE and the score of each item in MoCA, as well as the percentage of completely correct in other items between PIGD group and TD group ( Table 1).
The levels of pathological proteins in CSF from control, TD and PIGD groups. In TD and PIGD groups, Aβ 1-42 level in CSF was prominently decreased and the levels of T-tau, P-tau181t, P-tau231t, P-tau396s and P-tau199s in CSF were all significantly elevated when compared with control group (Table 2).
In PIGD group, Aβ 1-42 level in CSF was significantly reduced and P-tau181t level in CSF was evidently increased when compared with TD group (Table 2). P-tau199s level in CSF was elevated when compared with TD group ( Table 2).
The levels of neurotransmitters in CSF from control, TD and PIGD groups. In TD and PIGD groups, the levels of Ach, HVA and 5-HT in CSF were prominently lower than those in control subjects. Ach level in CSF in PIGD group was significantly lower than that in TD group (Table 2).

Relationships between PIGD score and the levels of neurotransmitters in CSF from PIGD group.
In total PD patients, there were no correlations between PIGD score and the levels of neurotransmitters in CSF. However, in PIGD group, PIGD score was prominently and negatively correlated with HVA level in CSF (r = − 0.575, p = 0.016) ( Table 4).

Relationships between the levels of Aβ 1-42 and neurotransmitters in CSF from PIGD group.
In PIGD group, we furtherly conducted correlation between the levels of Aβ 1-42 , P-tau181t and HVA level in CSF. Correlation analyses showed that Aβ 1-42 level was significantly and positively correlated with HVA level in CSF (r = 0.588, p = 0.035) (Supplemental Table 2).

Discussion
In this study, PIGD group displayed significantly higher scores of PIGD, rigidity and bradykinesia than those in TD group (Table 1). Previous longitudinal studies reported that PIGD score worsened at similar rates as bradykinesia and rigidity, whereas tremor score was stable over time 29,30 . Thus, the development of symptoms and signs of PIGD might be one of the pivotal element of PD progression. This study indicated that PIGD group had more patients at H-Y stage 4-5 than those in TD group. Studies showed difference 31 or no difference 32 in the disease severity scored by the H-Y stage between PIGD group and TD group. As disease progressed, PD patients might gradually developed cognitive impairment, thus, disease severity revealed by more advanced H-Y stage might be related to cognitive deterioration.   In this study, the percentage of completely correct in the item of recall in MoCA in PIGD group was significantly lower than that in TD group (Supplemental Table 3), indicating that more PD patients with PIGD phenotype suffered from cognitive impairment. In the correlation analysis, PIGD score was significantly and negatively correlated with MoCA score (Table 3), which was consistent with the hypothesis that cognitive impairment might be associated with the PIGD subtype. Postural control is referred to the control of the body's position with the aim of maintenance of balance and orientation 33,34 . A previous study showed that more cognition-related  Table 2. The levels of pathological proteins and neurotransmitters in CSF from control, TD and PIGD groups. Kruskal-Wallis test was used to compare T-tau level in CSF among control, TD and PIGD groups; ANOVA was used to compare the levels of Aβ 1-42 , P-tau231t, P-tau181t, P-tau199s, P-tau396s, Ach, DA, HVA, DOPAC, 5-HT and NE in CSF among control, TD and PIGD groups; p < 0.017 was defined as statistically significant. P1: Control group vs. TD group; Kruskal-Wallis test was used to compare T-tau level in CSF between control and TD groups; p < 0.017 was defined as statistically significant. Two-tailed t-test was used to compare the levels of Aβ 1-42 , P-tau231t, P-tau181t, P-tau199s, P-tau396s, Ach, DA, HVA, DOPAC, 5-HT and NE in CSF between control and TD groups, p < 0.017 was defined as statistically significant. P2: Control group vs. PIGD group; Kruskal-Wallis test was used to compare T-tau level in CSF between control and PIGD groups; p < 0.017 was defined as statistically significant. Two-tailed t-test was used to compare the levels of Aβ 1-42 , P-tau231t, P-tau181t, P-tau199s, P-tau396s, Ach, DA, HVA, DOPAC, 5-HT and NE in CSF between control and PIGD groups; p < 0.0017 was defined as statistically significant. P3: TD group vs. PIGD group; Kruskal-Wallis test was used to compare T-tau level in CSF between TD and PIGD groups; p < 0.017 was defined as statistically significant. Two-tailed t-test was used to compare CSF levels of Aβ 1-42 , P-tau231t, P-tau181t, P-tau199s, P-tau396s, Ach, DA, HVA, DOPAC, 5-HT and NE between TD and PIGD groups; p < 0.017 was defined as statistically significant. Aβ 1-42 = β amyloid (Aβ )  ; T-tau = total tau; P-tau181t = tau phosphorylated at threonine 181; P-tau231t = tau phosphorylated at threonine231; P-tau396s = tau phosphorylated at serine 396; P-tau 199s = tau phosphorylated at serine 199; Ach = acetylcholine; DA = dopamine; DOPAC = dihydroxyphenylacetate; HVA = homovanillic acid; 5-HT = serotonin; NE = norepinephrine. *P < 0.017, **P < 0.01.  Table 3. Associations of PIGD severity with the levels of Aβ 1-42 and P-tau181t in CSF in univariate and multivariate linear regression models with adjustment for potential confounders. MoCA = Montreal Cognitive Assessment; Aβ 1-42 = β amyloid (Aβ ) 1-42 ; P-tau181t = tau phosphorylated at threonine 181. *P < 0.05, **P < 0.01.

Variable
regions, such as caudate, fusiform gyrus and inferior parietal lobes 35 , were affected in PIGD patients. One study reported that cognitive performance was a predictor of postural control in individuals with PD 36 . PD patients could increase attention to counteract the balance disturbance in late stage 37 , therefore, PD patients exhibited more impaired postural control in the cognitive-motor dual-task condition in later stage because of distraction by cognitive task. Thus, cognitive-motor dual-task training could improve static postural control in PD patients 38,39 . These data implied that cognitive pathways might be involved in the PIGD phenotype. We then investigated the potential mechanisms underlying the correlation between PIGD phenotype and cognitive impairment. Decreased Aβ 1-42 level in CSF indicated Aβ aggregation and deposition in brain 9 . Substantial evidence suggested that lower Aβ 1-42 level in CSF was related to cognitive impairment 40 . Previous study reported that Aβ 1-42 level in CSF prominently decreased at later stage in PD patients with dementia 13 . In this study, our data demonstrated that PIGD severity was correlated with lower Aβ 1-42 level in CSF, independent of age, education level, H-Y stage and cognition (Table 3). Thus, we speculated that PIGD might share common pathological mechanisms with cognitive impairment in PD patients featured by the increased Aβ 1-42 aggregation in brain.
Under normal conditions, tau contributes to the integrity of the cytoskeleton. Excessive phosphorylation of tau in PD patients results in impaired cell integrity, loss of its physiological function and even cell death. Tau in CSF, thought to be a marker of neuronal death, was reported to be increased in PD patients with dementia 41 . Furthermore, increased tau level in CSF was found in PD patients with disease duration of less than 2 years, indicating that tau pathology might occur in the initial stage of the disease 42 . In this study, the results indicated that the levels of T-tau and all P-tau variables in PIGD group and TD group were significantly higher than those in control group. In total PD patients, PIGD severity was correlated with the increased P-tau181t level in CSF beyond age, education, H-Y stage and cognition ( Table 3), implying that tau pathology might be a potential underlying mechanism of PIGD. Further comparison revealed that P-tau181t level in CSF in PIGD group was significantly higher than that in TD group (Table 2), indicating that PIGD patients might have severer or faster neurodegeneration than TD patients. P-tau181t was reported to be related to cognitive impairment in Alzheimer's disease (AD) patients 43 , implying that P-tau181t might serve as a potential pathological marker linking PIGD and cognitive impairment in PD patients.
Up to now, there were just a few neurochemical studies, in which the levels of different neurotransmitters in CSF were examined in PD patients 44 . In this study, the data showed that HVA level in CSF in PIGD group and TD group were significantly lower than that in control group, and further comparison showed prominently declined HVA level in CSF in PIGD group compared with TD group (Table 3). We for the first time found that PIGD severity in PIGD group was significantly and negatively correlated with HVA level in CSF (Table 4), and a previous study reported that PIGD severity might be related to the degeneration of frontal dopaminergic pathways 45 , indicating that dopaminergic dysfunction was involved in PIGD. HVA was the main end-product of dopamine metabolism. Studies reported that the measurements of dopamine metabolites, rather than the parent compound, were the best reflection of dopamine's moment-by-moment turnover in nerve terminals 30,46 .
Previous study reported that cholinergic dysfunction was related to gait disturbance in PD patients 47 . However, there was no study investigating the relationships between Ach level in CSF and motor phenotypes in PD. The present study was therefore the first to show that Ach level in CSF in PIGD group was remarkably decreased compared to that in TD group and control group ( Table 2). This study demonstrated no association between Ach level in CSF and PIGD severity in PIGD group. Most of PIGD patients in our study were from out-patients clinic, and their PIGD symptoms were relatively mild. Therefore, the decreased Ach level in CSF might not play a key role in the neurobiological mechanisms of PIGD at that time.
Previous study demonstrated that aggravation of PIGD in advanced PD patients was related partially to impaired serotonergic transmission 48 . Reduced serotonergic neurotransmission impaired learning and memory function, whereas an increased serotonergic neurotransmission was associated with an improved cognitive and behavioral performance, not only in rodents, but also in AD patients 45 . A recent functional Magnetic Resonance Imaging (MRI) study demonstrated that the dysfunction of prefrontal-parietal network might be associated with the prominent gait impairments of PIGD subtype 35 . 5-HT projections originating from the raphe nuclei converged at several key target structures involved in memory, ascended within the medial forebrain bundle and were projected to prefrontal cortex 45 . Based on the above studies, it implied that the 5-HT projections abnormity might have an impact on PIGD. This study showed that 5-HT level in CSF in PIGD group and TD group were significantly lower than that in control group; furthermore, in PIGD group, 5-HT level in CSF in PIGD group was decreased when compared with TD group (Table 2). Yet, there was no correlation between PIGD severity and 5-HT level in CSF, which might be explained that patients in our study were at relatively early stage and the role of 5-HT dysfunction played on cognition in PIGD patients might not be significant.
This study also explored the relationship between the levels of neurotransmitters and Aβ 1-42 , T-tau and P-tau in CSF in PIGD group. Data showed that Aβ 1-42 level in CSF was prominently and positively correlated with HVA level in CSF (Supplemental Table 2), indicating that Aβ 1-42 level might accelerate dopamine degradation, causing the decreased HVA level in CSF and finally leading to PIGD.
In this investigation, P-tau 181t level in CSF was not correlated with each neurotransmitter level in CSF, which might be explained that most of PD patients were not demented (75.86% PD patients had MMSE > 24 scores ) PIGD severity (points) R P value HVA in CSF (ng/mL) − 0.575 0.016* and a mount of them (64.81%) were at the stage of no more than H-Y stage 2. We might, therefore, speculate that Aβ 1-42 played a major role on cognitive impairment for PD patients at early stage, consistent with a recent study reporting that tau probably revealed the degree of neurodegeneration, whereas Aβ 1-42 was a more specific marker for cognitive impairment. In summary, the findings of this study revealed a relationship between PIGD and cognitive impairment in PD patients. The results here provided a neuropathological mechanism underlying the correlation between PIGD and cognitive impairment. The decreased HVA level in CSF was closely associated with PIGD severity in PD patients. Aβ 1-42 deposition might contribute to HVA dysfunction in brain related to both PIGD and cognitive impairment in PD patients.
There are following limitations in this study. Relatively few CSF samples were analyzed due to the difficulties of obtaining CSF from PD patients and controls, which might weaken the statistical power of the analyses, such as of the correlations between PIGD severity with the levels of 5-HT and Ach in CSF. Cognitive function evaluated by MoCA was relatively rough and a formal neuropsychological test battery containing detailed cognitive domains would be used in the future study in order to provide more information.