OUP user menu

Cholinergic dysfunction contributes to gait disturbance in early Parkinson’s disease

Lynn Rochester, Alison J. Yarnall, Mark R. Baker, Rachel V. David, Susan Lord, Brook Galna, David J. Burn
DOI: http://dx.doi.org/10.1093/brain/aws207 2779-2788 First published online: 8 September 2012

Summary

Gait disturbance is an early feature in Parkinson’s disease. Its pathophysiology is poorly understood; however, cholinergic dysfunction may be a non-dopaminergic contributor to gait. Short-latency afferent inhibition is a surrogate measure of cholinergic activity, allowing the contribution of cholinergic dysfunction to gait to be evaluated. We hypothesized that short-latency afferent inhibition would be an independent predictor of gait dysfunction in early Parkinson’s disease. Twenty-two participants with Parkinson’s disease and 22 age-matched control subjects took part in the study. Gait was measured objectively using an instrumented walkway (GAITRite), and subjects were asked to walk at their preferred speed for 2 min around a 25-m circuit. Spatiotemporal characteristics (speed, stride length, stride time and step width) and gait dynamics (variability described as the within subject standard deviation of: speed, stride time, stride length and step width) were determined. Short-latency afferent inhibition was measured by conditioning motor evoked potentials, elicited by transcranial magnetic stimulation of the motor cortex, with electrical stimuli delivered to the contralateral median nerve at intervals ranging from N20 (predetermined) to N20 + 4 ms. Short-latency afferent inhibition was determined as the percentage difference between test and conditioned response for all intervals and was described as the group mean. Participants were optimally medicated at the time of testing. Participants with Parkinson’s disease had significantly reduced gait speed (P = 0.002), stride length (P = 0.008) and stride time standard deviation (P = 0.001). Short-latency afferent inhibition was also significantly reduced in participants with Parkinson’s disease (P = 0.004). In participants with Parkinson’s disease, but not control subjects, significant associations were found between gait speed, short-latency afferent inhibition, age and postural instability and gait disorder score (Movement Disorders Society Unified Parkinson’s Disease Rating Scale) and attention, whereas global cognition and depression were marginally significant. No other gait variables were associated with short-latency afferent inhibition. A multiple hierarchical regression model explored the contribution of short-latency afferent inhibition to gait speed, controlling for age, posture and gait symptoms (Postural Instability and Gait Disorder score—Movement Disorders Society Unified Parkinson’s Disease Rating Scale), attention and depression. Regression analysis in participants with Parkinson’s disease showed that reduced short-latency afferent inhibition was an independent predictor of slower gait speed, explaining 37% of variability. The final model explained 72% of variability in gait speed with only short-latency afferent inhibition and attention emerging as independent determinants. The results suggest that cholinergic dysfunction may be an important and early contributor to gait dysfunction in Parkinson’s disease. The findings also point to the contribution of non-motor mechanisms to gait dysfunction. Our study provides new insights into underlying mechanisms of non-dopaminergic gait dysfunction, and may help to direct future therapeutic approaches.

  • Parkinson’s disease
  • gait
  • short-latency afferent inhibition
  • cholinergic dysfunction
  • attention

Introduction

Gait dysfunction presents early in Parkinson’s disease (Baltadjieva et al., 2006), is a significant cause of disability and has limited management options available in the later stages of the disease because of its refractory nature to dopaminergic therapy and surgery (Sethi, 2008). The pathophysiological correlates of gait disturbance in Parkinson’s disease are poorly understood. However, evidence is accumulating that cholinergic dysfunction is a non-dopaminergic contributor to gait, even in the early stages (Bohnen and Albin, 2011; Yarnall et al., 2011), although the influence and role of cholinergic dysfunction remains unclear.

Two major cholinergic projection systems in the brain could conceivably contribute to gait disturbance: the subcortical system originating in the pedunculopontine nucleus in the brainstem and the cortical system ascending from the nucleus basalis of Meynert in the basal forebrain. Cholinergic neurons in the pedunculopontine nucleus are believed to have a powerful influence on motor control of gait and posture (Jenkinson et al., 2009; Karachi et al., 2010). Thalamic acetylcholinesterase activity, derived mainly from the terminals of neurons in the pedunculopontine nucleus, reflects cholinergic activity (Bohnen and Albin, 2011) and is reduced in early Parkinson’s disease (Shimada et al., 2009; Gilman et al., 2010; Bohnen and Albin, 2011). It is also more severely reduced in Parkinson’s disease fallers compared with non-fallers (Bohnen et al., 2009a), further reflecting a relationship with gait and postural dyscontrol. Both cortical and subcortical systems play a role in attentional control and executive function (Yarnall et al., 2011) that contribute to gait disturbance in Parkinson’s disease (Yogev-Seligmann et al., 2005; Rochester et al., 2008; Lord et al., 2010a, b, 2011). These studies, however, provide only indirect evidence for the role of the cholinergic system in gait disturbance, and there is as yet no direct evidence for a cholinergic contribution to gait.

Cholinergic activity in the brain can be estimated with short-latency afferent inhibition (SAI), a technique that non-invasively assesses an inhibitory circuit in the sensorimotor cortex. The circuit is believed to depend mainly on cholinergic activity in cortical and subcortical cholinergic systems (Di Lazzaro et al., 2002; Oliviero et al., 2005; Chen et al., 2008). SAI thus offers a means of non-invasively determining in vivo putative cholinergic loss. Previous studies investigating SAI in Alzheimer’s disease, where the neurodegenerative process is characterized by attrition of cholinergic interneurons, showed that SAI was reduced (Di Lazzaro et al., 2002, 2007) and that administration of an acetylcholinesterase inhibitor not only improved cognition but normalized SAI (Di Lazzaro et al., 2002). Studies in patients with Parkinson’s disease and visual hallucinations, believed to reflect functional involvement of cortical cholinergic circuits, have also shown reduced SAI compared with those without hallucinations (Manganelli et al., 2009).

Earlier studies have mainly focussed on the role of cholinergic function in cognition. To date, SAI has not been used to probe the role of cholinergic systems in gait in Parkinson’s disease. Because of the increasing evidence of cholinergic dysfunction in patients with Parkinson’s disease with gait disorders, we hypothesized that SAI would be an independent predictor of impaired gait in the early stages of Parkinson’s disease, reflecting a contribution of cholinergic dysfunction.

Materials and methods

Subjects and recruitment

Twenty-two patients with early Parkinson’s disease (defined by duration of motor symptoms <36 months) and 22 age-matched control subjects were recruited. Subjects who fulfilled inclusion/exclusion criteria were recruited consecutively (to avoid ascertainment bias) from a regional movement disorders clinic. Inclusion criteria were as follows: diagnosis of idiopathic Parkinson’s disease made by a movement disorder specialist according to Queen Square Brain Bank criteria (Hughes et al., 1992); ability to mobilize independently and absence of cerebrovascular disease, dementia and clinically significant depressive episode. Patients/controls were also excluded if there were contraindications to magnetic stimulation (e.g. metallic heart valve, cranial aneurysm clips or previous seizures). Moreover, because SAI must be performed at rest (i.e. no background EMG activation), patients with significant resting tremor [tremor score of >2 on Movement Disorder Society Unified Parkinson’s Disease Rating Scale (MDS-UPDRS)] were excluded (Goetz et al., 2008). Patients treated with cholinesterase inhibitors, amantadine, benzodiazepines, neuroleptic or anti-depressants were also excluded. All subjects scored ≥24/30 on the Mini-Mental State Examination (Folstein et al., 1975). The study had ethical approval from Newcastle and North Tyneside Research Ethics Committee, and all participants gave informed written consent to the study procedures. Subjects attended Newcastle Biomedicine’s Clinical Ageing Research Unit, where testing took place.

The following descriptive and clinical data were collected to describe and compare the participants with Parkinson’s disease and control subjects: age; sex; handedness; 15-item Geriatric Depression Scale (Yesavage et al., 1983); MDS-UPDRS part III (Parkinson’s disease only) (Goetz et al., 2008); Hoehn and Yahr stage (Parkinson’s disease only; Hoehn and Yahr, 1967); the National Adult Reading Test (Crawford et al., 1990); subtests of the Cambridge Neuropsychological Test Automated Battery (CANTAB; Cambridge Cognition Ltd; Robbins et al., 1998): One-touch Stockings of Cambridge, Pattern Recognition Memory and Spatial Recognition Memory; Cognitive Drug Research Battery (Bracket); Power of Attention score (Wesnes et al., 2002) and Montreal Cognitive Assessment (Dalrymple-Alford et al., 2010). These tests were completed by all participants. All participants with Parkinson’s disease were taking dopaminergic medications, and l-DOPA equivalent doses were calculated for each patient (Tomlinson et al., 2010).

Quantitative gait evaluation

A 7-m-long instrumented walkway (GAITRite, CIR Systems Inc.) captured individual footfall data, using embedded pressure sensors. This is a valid and reliable method of assessing the spatiotemporal parameters of gait in healthy older adults and patients with Parkinson’s disease (Bilney et al., 2003). Gait was measured during a 2 min continuous walk around a 25-m circuit, and the mat was positioned along one side of the circuit allowing data to be sampled intermittently as each participant walked repeatedly over the mat. Continuous gait was evaluated to capture steady-state locomotion and avoid acceleration and deceleration. Participants were asked to walk at their preferred speed, and data were captured 1 h after medication intake. Approximately five passes over the mat were made, from which spatiotemporal gait characteristics [gait speed (m/s), stride length (m), stride time (s) and step width (cm)] and gait dynamics (variability of gait speed, stride length, stride time and step width described as the standard deviation) were determined. Gait dynamics were determined as the within-person standard deviation of pooled strides/steps where appropriate, and at least 20 strides (40 steps) were included for each participant. Data were collected at 240 Hz, saved onto a computer and were analysed using proprietary software. This test was completed by all participants. The data were analysed by a researcher who was blinded to outcomes from the neurophysiological evaluation and data analysis. Gait evaluation was carried out on a separate day from neurophysiological evaluation, no more than 2 weeks apart.

Recordings

Surface EMGs were recorded from the first dorsal interosseous of the symptomatic arm (patients) or the dominant hand (controls) with adhesive Ag–AgCl gel electrodes (Biosense Medical Ltd). The active electrode was placed over the muscle belly and the reference electrode over the proximal metacarpophalangeal joint. Somatosensory evoked potentials were recorded through adhesive electrodes (Neuroline 720, Ambu) applied to the scalp (contralateral to median nerve stimulation) after appropriate skin preparation, using a bipolar montage with the non-inverting electrode 2 cm anterior to C3/C4 (according to the 10–20 system), the inverting electrode 2 cm posterior to C3/C4 (depending on the side of stimulation) and the reference electrode placed on the forehead. Signals were amplified (EMG gain: 1000–2000; EEG gain: 50 000) and bandpass filtered (EMG: 30 Hz–2 kHz; EEG: 3 Hz–2 kHz), using a Digitimer D360 system before being digitized at 5 kHz by a Power1401 interface (Cambridge Electronic Design Ltd) connected to a computer running Spike2 software (Cambridge Electronic Design Ltd).

Median nerve stimulation

Somatosensory evoked potentials were obtained by stimulating the median nerve in the more symptomatic arm in patients (and on the dominant side in controls). Stimuli (single pulses; pulse width: 200 µs) were delivered to the median nerve at the wrist using a constant current stimulator (Digitimer DS7AH) through adhesive electrodes (cathode proximal; Biosense Medical Ltd). The intensity of the stimulus was adjusted to just above motor threshold, as determined by a visible twitch in the abductor pollicis brevis muscle.

Magnetic stimulation

Transcranial magnetic stimulation of the motor cortex was performed using a high power Magstim 200 (Magstim Co.) and circular transcranial magnetic stimulation coil (130-mm-diameter). The vertex (Cz) was measured using standard clinical procedures and was marked with indelible ink. The circular coil was then placed with the vertex marker positioned in the centre of the coil and the handle of the coil (held by the experimenter) posterior. An anticlockwise coil-current was used to stimulate the left hemisphere (right hand) and a clockwise coil-current for the right hemisphere (left hand). The direction of the coil current used (and, therefore, the hemisphere stimulated) varied between patients (and controls). In patients with Parkinson’s disease, transcranial magnetic stimulation was delivered to the hemisphere opposite the most affected side, whereas in control subjects the dominant hemisphere was targeted. Motor evoked potentials were recorded from the contralateral first dorsal interosseus muscle. Resting motor threshold was determined as the percentage of maximum stimulator output that elicited a liminal motor evoked potential (∼50 µV in 5 of 10 trials) at rest. Motor evoked potentials were digitized (see earlier) and stored for later analysis, when they were analysed blind to the diagnostic category.

Short latency afferent inhibition

SAI was carried out using a modified version of the protocol described by Tokimura et al. (2000). Motor evoked potentials were conditioned by median nerve stimulation. Conditioning stimuli delivered to the median nerve preceded cortical transcranial magnetic stimulation by varying interstimulus intervals. These were determined relative to the latency of the N20 component of the average somatosensory evoked potential to median nerve stimulation (averages of 2000 raw sweeps). SAI was randomly tested at five different interstimulus intervals, with 10 trials at each interstimulus interval (from N20 in 1-ms increments until N20 + 4 ms), with 20 unconditioned (test) stimuli also delivered randomly (Fig. 1). The peak-to-peak amplitude of the conditioned motor evoked potentials at each interstimulus interval were averaged and expressed as a percentage of the averaged unconditioned motor evoked potential (baseline). To reduce variability, the conditioned responses were combined across all interstimulus intervals and were expressed as the percentage of the unconditioned motor evoked potentials to provide a grand mean of SAI, as described by others (Di Lazzaro et al., 2000; Nardone et al., 2005). Participants were given visual feedback to maximize complete relaxation, and any EMGs contaminated by interference were excluded. SAI requires participants to be tested at rest with no background EMG activity in the target muscle, and as such, we had to exclude participants with Parkinson’s disease with excessive tremor. This may have biased our sample to those with a greater risk of developing a more postural instability and gait disturbance motor phenotype. Subjects were tested while taking their usual optimized dopaminergic medication to reduce discomfort.

Figure 1

Short-latency afferent inhibition. Conditioned and unconditioned motor evoked potentials recorded from: (A) control subject (aged 66 years); (B) patient with Parkinson’s disease (aged 83 years). Conditioned motor evoked potentials are plotted in black and baseline (unconditioned) motor evoked potentials are plotted in grey for comparison. Average motor evoked potentials (n = 20) have been plotted. Interstimulus intervals are indicated above each motor evoked potential and are expressed in terms of N20 latency (determined individually for each patient/control subject). (C) Time course of inhibition of SAI for participants with Parkinson’s disease (n = 22) and control subjects (n = 22) at all interstimulus intervals. Mean and standard deviation are shown.

Although paired-pulse transcranial magnetic stimulation paradigms were all initially described using figure-of-eight coils (outside diameter was 90 mm for each circle) placed over the motor ‘hotspot’ of the muscle being studied, equally reliable results can be obtained using a circular coil (130 mm outside diameter) placed over the vertex (Trompetto et al., 1999; Zoghi et al., 2003; Badawy et al., 2011). Moreover, the circular coil has a number of advantages over the figure-of-eight coil: (i) motor evoked potential amplitude is exquisitely sensitive to translational and vertical and horizontal rotational changes in coil orientation, which is a major problem with figure-of-eight coils where the flat junction of the coils is placed over the convexity of the skull. With the circular coil, the hollow is placed over the vertex, and even with a hand held coil, there is little movement-related motor evoked potential variability; (ii) the vertex is easily marked using anatomical landmarks allowing the centre of the circular coil to be rapidly, accurately and reproducibly repositioned during a study; and (iii) it is not necessary to map the ‘hotspot’ with a circular coil, unlike the figure-of-eight coil, reducing overall experimental time, which is critical to ensure data are collected while the patient remains optimally medicated.

Statistical analysis

All data met the assumptions for normality apart from stride time standard deviation, which was log transformed. Univariate and bivariate analysis were used to describe the data. Independent t-tests were used to compare participants with Parkinson’s disease and controls. Pearson product correlation coefficients were calculated to explore associations between gait variables and dependent variables. Regression analysis was used to explore the primary hypothesis that SAI was an independent predictor of gait. Gait disturbance in Parkinson’s disease is multifactorial, with contributions from age, motor disease severity, depression, executive function and attention (Yogev-Seligmann et al., 2005; Rochester et al., 2008; Lord et al., 2010a, b, 2011). Therefore, we hypothesized a priori that age, motor disease severity and cognition would make a substantial contribution and we controlled for them, so as to discern the independent contribution of cholinergic dysfunction to gait pathology in early Parkinson’s disease. Gait was entered in the first level, and variables we wished to control for were entered in the second level. Colinearity diagnostics (eigenvalues and condition indices) were inspected to test for multi-colinearity, and the Durbin–Watson statistic was used to identify autocorrelation (values <1 and >3 were problematic). All data met assumptions for regression analysis, and we aimed to recruit a minimum of five participants per variable. A significance value of P < 0.05 was set. IBM SPSS version 19 was used to analyse the results.

Results

Demographic and clinical characteristics

Demographic and clinical characteristics are shown in Table 1. Control participants were well matched for age and National Adult Reading Test scores. Depression scores were low in both groups. Participants with Parkinson’s disease had a mean score of 29.14 (9.14) on the MDS-UPDRS III and a median Hoehn and Yahr score of 2, indicating a range of mild to moderate disease severity, despite the early course of the disease [mean (SD) disease duration was 19.83 (8.60)]. Nineteen patients were taking l-DOPA, nine patients were taking dopamine agonists, 13 patients were taking monoamine oxidase inhibitors and three patients were taking a catechol-O-methyltransferase inhibitor. There were relatively more male participants in the Parkinson’s disease group compared with control subjects; however, we feel that this is unlikely to influence the findings, and there is no evidence in the literature to suggest that it would. Cognitive function was more impaired for all neuropsychological tests in participants with Parkinson’s disease compared with control subjects.

View this table:
Table 1

Demographic and descriptive data for all outcomes for participants with Parkinson’s disease and control subjects

OutcomePatients with Parkinson’s disease (n = 22)Control subjects (n = 22)P value
Age (years)70.18 (9.67)67.43 (8.43)0.320
Male/female16/69/13
MDS-UPDRS III (ON)29.14 (9.54)
PIGD score0.80 (0.51)
Hoehn and Yahr stageI (1); II (11); III (10)
l-DOPA equivalent scores304.86 (130.191)
Disease duration (months)19.83 (8.60)
National Adult Reading Test112.59 (10.76)116.50 (8.43)0.187
Montreal Cognitive Assessment (0–30)24.50 (4.02)27.64 (1.97)0.002*
CANTAB OTS (problems solved)14.55 (5.19)17.50 (1.32)0.023*
CANTAB PRM (mean correct latency)2164.26 (467.62)2386.89 (655.99)0.236
CANTAB SRM (% correct)75.71 (13.54)86.56 (5.69)0.645
CDR PoA (ms)1401.52 (227.59)1258.56 (126.94)0.015*
GDS2.05 (2.44)1 (1.27)0.082
SAI (%)23.11 (28.38)45.85 (21.17)0.004*
  • PIGD score = selected items from the MSD-UPDRS; CANTAB OTS = One-touch stocking of Cambridge test; CANTAB PRM = Pattern Recognition Memory test; CANTAB SRM = Spatial Recognition Memory test; CDR PoA = Power of Attention score; GDS = Geriatric Depression Scale.

  • Data are displayed as mean (SD) unless otherwise stated. Groups were compared using independent t-tests unless otherwise stated.

  • *P ≥ 0.05 is considered significant.

Gait characteristics

Gait characteristics are shown in Table 2. Gait characteristics were selectively affected in the participants with Parkinson’s disease, with only gait velocity, stride length and stride time variability (SD) being significantly impaired in subjects with Parkinson’s disease compared with controls.

View this table:
Table 2

Descriptive gait characteristics for participants with Parkinson’s disease and control subjects

OutcomeParticipants with Parkinson’s disease (n = 22)Control subjects (n = 22)P value
Spatio-temporal gait characteristics
    Gait speed (m/s)1.12 (0.24)1.31 (0.15)0.002*
    Stride length (m)1.23 (0.23)1.39 (0.14)0.008*
    Stride time (s)1.12 (0.11)1.07 (0.09)0.117
    Step width (cm)9.09 (3.16)8.52 (2.47)0.505
Gait dynamics
    Gait speed SD (m/s)0.046 (0.017)0.042 (0.014)0.379
    Stride length SD (m)0.037 (0.013)0.030 (0.011)0.06
    Stride time SD (s)0.032 (0.013)0.019 (0.007)0.001*
    Step width SD (cm)1.93 (0.62)2.23 (0.55)0.09
  • Data are shown as mean (±SD). Groups were compared using independent t-tests unless otherwise stated.

  • *P ≥ 0.05 is considered significant.

Short latency afferent inhibition

Recordings were made from the dominant hand in control subjects (19 right handed and three left handed) and the symptomatic arm in participants with Parkinson’s disease (11 right and 11 left-side disease dominance). Figures 1A and B show examples of conditioned and unconditioned motor evoked potentials obtained from a control subject and a patient with Parkinson’s disease during SAI testing. Figure 1C shows the difference in SAI at each interstimulus interval in patients with Parkinson’s disease and control subjects, with a clear separation between the groups at each interstimulus interval. All control subjects showed inhibition of motor evoked potentials at each interstimulus interval. Participants with Parkinson’s disease also showed inhibition but not at all interstimulus intervals (Fig. 1A). The grand mean (pooled data across each interstimulus interval) was determined for each participant, and individual values are shown in Fig. 2. While all control participants showed inhibition, participants with Parkinson’s disease showed significantly reduced SAI (Table 1 and Fig. 2), although there was a wide range of responses.

Figure 2

Scatterplot of percentage inhibition of SAI in participants with Parkinson’s disease (n = 22) and control subjects (n = 22) with mean and standard deviation displayed alongside for reference. PD = Parkinson’s disease.

Relationship between gait variables and demographic, clinical, neuropsychological and short-latency afferent inhibition outcomes

To inform our selection of gait variable and refine an explanatory model of gait, we explored the association between those gait variables that were significantly different from control subjects to identify independent gait characteristics that represented discreet elements of gait control, allowing inferences to be made about the role of cholinergic dysfunction to the neural control of gait. We also explored the relationship between all gait variables and SAI. In participants with Parkinson’s disease, only gait speed (r = 0.606; P = 0.003) and stride length (r = 0.568; P = 0.006) were significantly correlated with SAI. As gait speed and stride length were also highly correlated (r = 0.915; P ≤ 0.001), we selected gait speed as the dependent variable for the gait model because of its higher correlation coefficient with SAI. In control participants, SAI was not associated with gait variables, so further explanatory analysis was confined to participants with Parkinson’s disease (Table 3).

View this table:
Table 3

Bivariate analysis of explanatory variables for gait speed in participants with Parkinson’s disease and control subjects

Participants with Parkinson’s disease (n = 22)Control subjects (n = 22)
SAI (%)
    r0.6060.040
    P0.003*0.860
Age
    r−00.610−00.309
    P0.003*0.162
Montreal Cognitive Assessment
    r0.4100.227
    P0.0580.309
PIGD score (MDS-UPDRS III)
    r−00.453
    P0.034*
l-DOPA equivalent score
    r−0.038
    P0.868
Disease duration (months)
    r0.175
    P0.437
National Adult Reading Test
    r0.1230.119
    P0.5860.598
GDS
    r−0.410−0.179
    P0.0580.425
CANTAB OTS (problems solved)
    r0.219−0.030
    P0.3540.913
CANTAB PRM (mean correct latency)
    r−0.263−0.478
    P0.2500.061
CANTAB SRM (% correct)
    r−00.1890.165
    P0.4130.540
CDR PoA (ms)
    r−00.5170.033
    P0.014*0.886
  • *Significant values. P-value of 0.05 was considered significant. PIGD score = selected items from the MDS-UPDRS; CANTAB OTS = One-touch stocking of Cambridge test; CANTAB PRM = pattern recognition memory test; CANTAB SRM = spatial recognition memory test; CDR PoA = power of attention score; GDS = Geriatric Depression Scale.

As a next step, we determined the core set of variables for the explanatory model of gait. We explored the association between gait speed, SAI, demographic, clinical and neuropsychological measures using bivariate correlations (Table 3). Figure 3 shows the associations between gait speed with SAI for participants with Parkinson’s disease and control subjects. In participants with Parkinson’s disease, there was a significant relationship between gait speed and SAI, with slower gait speed being associated with reduced SAI (less inhibition). Gait speed in participants with Parkinson’s disease was also significantly associated with age and attention (power of attention score) and the mean postural instability and gait disturbance score (determined from five items of the Items 2.12, 2.13, 3.10, 3.11, 3.12: MDS-UPDRS III). Global cognitive impairment (Montreal Cognitive Assessment) and depression (Geriatric Depression Scale) were of borderline significance.

Figure 3

Correlation of gait speed and SAI in Parkinson’s disease (n = 22) and control participants (n = 22) showing a significant association for Parkinson’s disease only.

Explanatory characteristics of gait speed in Parkinson’s disease

Variables that were significantly associated or of borderline significance with gait speed were included in a hierarchical multiple regression model to determine the explanatory characteristics of gait with SAI entered into the first level. In the second level, we controlled for symptoms related to the severity of postural instability and gait disorder, using the postural instability and gait disturbance mean score (from the MDS-UPDRS III). We also included age, Power of Attention score and depression (Geriatric Depression Scale). Because Power of Attention score and Montreal Cognitive Assessment were highly correlated with each other, we selected Power of Attention score because of its greater correlation coefficient.

Reduced SAI (less inhibition) was an independent determinant of reduced gait speed, explaining 37% of variability. SAI remained a significant determinant of gait speed after controlling for age, cognition and motor disease severity relating to postural control and gait. In the final model, reduced SAI and impaired attention (Power of Attention score) were independent determinants, explaining 72% of variability in gait speed. Postural and gait symptoms, age and depression were not significant. Model characteristics are shown in Table 4.

View this table:
Table 4

Regression model statistics and coefficients of variables for walking speed in patients with Parkinson’s disease (n = 22)

IndependentβP-value
Model 1
    SAI0.6060.003*
R2 = 0.37; P-value of change = 0.003
Model 2
    SAI0.3810.024*
    Age−0.2810.126
    PIGD score−0.1260.479
    PoA−0.3590.028*
    GDS−0.1730.319
R2 = 0.72; P-value of change = 0.009
  • Standardized regression coefficients (β) and P values are reported. R2 and P-value of change are reported for the model.

  • *Significant explanatory variables and a P value of ≤0.05 is considered significant. PIGD score = selected items from the MDS-UPDRS; PoA = Power of Attention score; GDS = Geriatric Depression Scale.

Discussion

This is the first study to our knowledge to apply SAI as a putative marker of cholinergic dysfunction to evaluate the role of cholinergic disturbance in gait dysfunction in Parkinson’s disease. SAI was an independent predictor of gait speed in Parkinson’s disease in contrast to control subjects, thus confirming our hypothesis. Our results demonstrate two important features: firstly, that cholinergic dysfunction makes a significant and independent contribution to gait dysfunction in patients with Parkinson’s disease, and secondly, that these findings are present relatively early in the disease course.

Alterations in gait with advanced disease have been widely reported for Parkinson’s disease, whereas early changes have received less attention. In de novo Parkinson’s disease gait performance remains relatively high, although a subtly altered gait pattern is already present (Baltadjieva et al., 2006). Similar previously described features of gait disturbance were also apparent in our patient group with early disease, even though gait remained within the reported range for age-matched subjects (Oh-Park et al., 2010). It is interesting that despite participants with Parkinson’s disease being optimally medicated, differences persisted. Therefore, dopaminergic medication does not fully ameliorate gait dysfunction, further supporting a role for non-dopaminergic contribution to gait.

Fundamental to the design of this study were convergent lines of evidence to support SAI as a putative marker of cholinergic dysfunction. For example, SAI is abolished with administration of scopolamine, a non-specific muscarinic receptor antagonist (Di Lazzaro et al., 2000). SAI is decreased in forms of dementia with a known cholinergic basis (Freitas et al., 2011; Nardone et al., 2011) and can be normalized by administration of an acetylcholinesterase inhibitor (Di Lazzaro et al., 2002, 2004b, 2007). Evidence of reduced SAI in patients with Parkinson’s disease, however, is complicated by the influence of dopaminergic medications, the relatively small sample size of studies and participant heterogeneity. Studies have shown that SAI may be normal in non-medicated Parkinson’s disease (Sailer et al., 2003), decreased on the more affected side in subjects with Parkinson’s disease on medication (Sailer et al., 2003) and enhanced in drug-free Parkinson’s disease (Di Lazzaro et al., 2004a; Nardone et al., 2005). A more recent study found that SAI was reduced in subjects with Parkinson’s disease with visual hallucinations (related to degeneration of cholinergic neurons in the pedunculopontine nucleus), a finding taken to indicate functional involvement of cholinergic circuits (Manganelli et al., 2009). We found a reduction in SAI in our group of participants with Parkinson’s disease with early disease compared with controls. These findings may be accounted for by the inclusion of a larger homogenous cohort tested in an optimally medicated state.

Gamma-aminobutyric acidergic (GABAergic) interneurons also have a role in the inhibitory circuit underlying SAI, as shown in experiments in which the partial GABAA-agonists diazepam and lorazepam were administered to healthy control subjects (Di Lazzaro et al., 2005). In these experiments, diazepam increased SAI, whereas lorazepam decreased SAI. This observation is presumably explained by differences in the composition of alpha subunits present in the postsynaptic and presynaptic (cholinergic terminals) GABAA receptors and, therefore, variable receptor affinities. Thus, while GABAergic interneurons are a component of the SAI circuit, because GABA both reduces and increases SAI, the net effect of GABA released from GABAergic interneurons on the measurement of SAI is likely to be minimal (assuming that degenerative processes affecting GABAergic interneurons are neither interneuron subtype nor region specific).

To be confident of the underlying pharmacological basis for changes identified via SAI, other non-invasive electrophysiological measures can be used to assess the relative muscarinic and GABAergic contributions. Short-interval intracortical inhibition is one such example, which is increased by diazepam and lorazepam and unaffected by scopolamine (Di Lazzaro et al., 2000, 2005). However, there is evidence that short-interval intracortical inhibition can be enhanced by nicotine (Orth et al., 2005); therefore, loss of cholinergic or GABAergic interneurons would reduce short-interval intracortical inhibition and SAI, confounding interpretation.

While degeneration of cholinergic interneurons in the pedunculopontine nucleus is a recognized histological finding in Parkinson’s disease, particularly in patients with an associated gait disorder (Karachi et al., 2010), loss of cortical GABAergic interneurons is not a feature of Parkinson’s disease (Halliday et al., 2005). Therefore, it is reasonable to assume that a reduction in SAI is related to loss of cortical or brainstem cholinergic interneurons. Consequently, we would argue that SAI is a potentially useful proxy measure of cholinergic dysfunction and points to impaired cholinergic function in early Parkinson’s disease compared with controls.

We described a range of gait characteristics; however, only gait speed and step length were significantly associated with SAI. Gait speed was selected as our dependent gait variable to explore the role of cholinergic dysfunction in determining gait alteration. In relation to our primary question, SAI was an independent predictor of walking speed, explaining 37% of variability. Furthermore, it remained independent after controlling for motor deficit associated with posture and gait dysfunction (Postural Instability and Gait Disturbance score), age, attention (Power of Attention score) and depression (Geriatric Depression Scale). The model explained 72% of variance and only reduced inhibition (less SAI) and attention (decreased Power of Attention score), significantly explained slower gait. The focus of work exploring the cholinergic system has largely been related to its direct role in cognition, and the contribution of this system to gait has remained poorly understood (Bohnen et al., 2009a). Therefore, these results provide evidence for an important and early contribution of the cholinergic system to gait dysfunction and to the important influence of non-motor function on gait even in early Parkinson’s disease.

We found that cognition, and particularly attention (using Power of Attention score as a composite measure), was an independent explanatory characteristic of gait speed in early Parkinson’s disease, a finding that concurs with previous reports in more advanced Parkinson’s disease (Yogev-Seligmann et al., 2005; Rochester et al., 2008; Lord et al., 2010a, b, 2011). SAI was not associated with any of the other explanatory variables in the model in keeping with the assumption of independence. The independence of SAI and attention as determinants of gait speed in early Parkinson’s disease is interesting. Reduced attention is a feature of cholinergic dysfunction (Bohnen et al., 2006); however, its independence from SAI suggests that it is underpinned by a different neurochemical substrate in early Parkinson’s disease and may reflect a dopaminergically mediated contribution (Cools et al., 2001).

Age was not an independent contributor, which was suprising. Gait speed reduces with age, and age has previously been shown to contribute to gait speed in Parkinson’s disease (Rochester et al., 2008). Despite this, given the known association of age with speed, we would recommend the need to control for age as a contributory factor in studies exploring pathophysiology of gait. Depression was also not an independent determinant in contrast to earlier work that has shown that this contributes to gait in Parkinson’s disease (Rochester et al., 2008). This may reflect the relative importance of depression compared with cholinergic function in gait control.

As highlighted earlier, the role of cholinergic dysfunction in gait may be mediated through influences on motor control by the pedunculopontine nucleus or through cognitive control either by the nucleus basalis of Meynert or the pedunculopontine nucleus and its role in attention (Inglis et al., 2001). It was interesting that only gait speed was related to SAI. Others have suggested that the temporal dynamics of gait (variability in stride time) may be more sensitive to the pathology of Parkinson’s disease, being potentially mediated by non-dopaminergic pathology (Baltadjieva et al., 2006). Our findings do not support this contention. Postural dyscontrol may also contribute to gait disturbance, and it is thought to be a predominant feature of cholinergic dysfunction. Although we found the posture and gait score was associated with gait, it was not retained as an independent predictor in the model. These findings imply that gait dysfunction may be predominantly mediated through cognitive disturbance highlighting the important contribution of cognition to early gait dysfunction, a feature that has also been reported in later stage Parkinson’s disease (Rochester et al., 2008; Lord et al., 2010a, b, 2011). Of relevance to this are recent findings that alpha oscillations in the pedunculopontine nucleus correlate with gait performance in more advanced Parkinson’s disease, and alpha activity is proposed to contribute to attentional control allowing for more efficient motor performance (Thevathasan et al., 2012). In addition, gait speed has been shown to be sensitive to cognitive function and predictive of cognitive decline in initially non-impaired older adults (Waite et al., 2005; Verghese et al., 2007). Therefore, it is plausible that the relationship with SAI may be explained by a more profound non-motor (i.e. cognitive) contribution to gait rather than a motor contribution. The precise nature of cholinergic dysfunction, however, cannot be identified in this study, and this question remains an interesting area for further study.

As stated earlier, Parkinson’s disease is widely accepted to be a multi-system disease with variable pathology and rate of progression. Cholinergic denervation (shown through acetylcholinesterase inhibitor activity in imaging studies) in Parkinson’s disease is equally variable in its rate and progression (Bohnen and Albin, 2009b; Shimada et al., 2009). The distribution of SAI in patients with Parkinson’s disease in our study also demonstrated considerable heterogeneity, despite significant group differences between participants with Parkinson’s disease and control subjects. SAI may similarly reflect differences in the rate or degree of cholinergic denervation and explain the variability of gait dysfunction in Parkinson’s disease.

There is a complex functional relationship between the cholinergic system, gait and cognition in Parkinson’s disease. Early evidence of cholinergic disturbance in mediating gait dysfunction may also be informative in identifying those subjects at greater risk of more rapid cognitive decline. In support of this, the postural instability and gait disorder motor phenotype is predictive of cognitive decline and dementia, suggesting that gait and postural control are underpinned by concomitant degeneration in cortical and subcortical cholinergic systems (Alves et al., 2006; Burn et al., 2006; Bohnen and Albin, 2011). Additionally, falls, a consequence of poor gait, are predicted by attentional deficits (Allcock et al., 2009). Whether a combination of reduced SAI and gait speed is a risk factor for falls and cognitive decline is a question we are addressing in a longitudinal study. Early intervention with cholinesterase inhibitors may delay onset of falls and cognitive decline in Parkinson’s disease, and with emerging evidence to support the use of these agents to reduce fall frequency (Chung et al., 2010). It would also be of interest to evaluate the effect of a single dose of cholinesterase inhibitors on gait to probe more specifically the cholinergic influence at this early stage of the disease.

There are some limitations to this study that should be acknowledged. Participants with Parkinson’s disease were tested ON medication, thus offsetting the influence of dopaminergically mediated dysfunction, and this feature may blunt the potential role of postural instability and gait disturbance. Other studies indicate that motor disease severity makes an important contribution to gait in more advanced Parkinson’s disease (Rochester et al., 2008). Our results suggest that SAI may be a more sensitive and specific marker of underlying pathology than measures of motor disease severity.

Our finding that SAI was a strong determinant of gait speed supports a role for cholinergic dysfunction in gait control. Cholinergic dysfunction may play an important role in gait in early Parkinson’s disease potentially representing a therapeutic target. Further studies are required to evaluate whether gait and SAI may be useful biomarkers for falls risk and cognitive decline.

Funding

L.R. is supported by the UK National Institute for Health Research Biomedical Research Centre for Ageing and Age-Related Disease award to the Newcastle upon Tyne Hospitals NHS Foundation Trust. A.J.Y. is supported by funding from the Michael J. Fox Foundation; the Lockhart Parkinson’s Fund and Parkinson’s UK. M.R.B. is supported through the National Institute for Health Research; Medical research Council; Wellcome Trust and Academy of Medical Sciences. D.J.B. is supported by the UK National Institute for Health Research Biomedical Research Unit in Lewy Body Dementia award to the Newcastle upon Tyne Hospitals NHS Foundation Trust.

Abbreviations
CANTAB
Cambridge Automated Neuropsychological Testing Battery
GABA
gamma-amino butyric acid
MDS-UPDRS
Movement Disorders Society Unified Parkinson’s Disease Rating Scale
SAI
short-latency afferent inhibition

References

View Abstract