Freezing of Gait and its Associations in the Early and Advanced Clinical Motor Stages of Parkinson’s Disease: A Cross-Sectional Study

Abstract

Background:

Freezing of gait is a common disabling symptom of Parkinson’s disease (PD) with limited treatment options. The pathophysiological mechanisms of freezing behaviour are still contentious.

Objective:

To investigate the prevalence of freezing of gait and its associations with increasing disease severity to gain a better understanding of the underlying pathophysiology.

Methods:

This exploratory study included 389 idiopathic PD patients, divided into four groups; early and advanced PD with freezing of gait, and early and advanced PD without freezing of gait. Motor, cognitive and affective symptoms, REM sleep behaviour disorder and autonomic function were assessed.

Results:

Regardless of disease stage, patients with freezing of gait had more severe motor symptoms and a predominant non-tremor phenotype. In the early stages, freezers had a selective impairment in executive function and had more marked REM sleep behaviour disorder. Autonomic disturbances were not associated with freezing of gait across early or advanced disease stages.

Conclusion:

These findings support the notion that impairments across the frontostriatal pathways are intricately linked to the pathophysiology underlying freezing of gait across all stages of PD. Features of REM sleep behaviour disorder suggest a contribution to freezing from brainstem pathology but this does not extend to more general autonomic dysfunction.

INTRODUCTION

Freezing of gait (FOG) is a common symptom in Parkinson’s disease (PD) that is defined as a sudden involuntary motor arrest during walking [1]. With limited treatment options [2], FOG has a substantial impact on independence, quality of life and nursing home placement [3, 4]. Previous work has reported that FOG is present in around 80% of patients in the more advanced stages of PD whilst only 10% of patients experience this symptom in the early stages [5, 6].

A number of phenotypic features have been associated with the presence of FOG. In relation to motor features, FOG has been commonly associated with a non-tremor dominant subgroup rather than tremulous PD [7, 8]. Patients with this phenotype typically experience a more rapid disease progression and have more severe motor impairments, such as increased muscle rigidity with a concomitant deterioration of gait and balance [7, 9]. In addition, non-tremor dominant patients typically manifest a selective pattern of executive dysfunction [7] and a higher risk of dementia [10].

One recent study has confirmed that these distinct phenotypic differences are present between patients with and without FOG in the early clinical stages of PD [11]. Specifically, patients in this study had disease duration of less than five years and were Hoehn and Yahr stage 2 or less. Those patients with FOG had a non-tremor motor phenotype with selective executive impairments but did not differ from non-freezers on measures of tremor and more general cognition. Moreover, in this sample of early clinical stage patients, no differences were found on measures of autonomic function, REM sleep behaviour disorder (RBD) or mood between freezers and non-freezers [11]. This finding is in distinction from earlier reports with less stringently selected patients who had more varied disease stages and durations [12–15].

The finding that FOG was not related to autonomic dysfunction, RBD or mood disorder in patients with early disease [11] suggests that the underlying pathophysiology of freezing at this stage may be more closely related to disturbances in frontostriatal rather than brainstem circuitry. However, it must be acknowledged that good evidence exists implicating the mesencephalic locomotor region and pedunculopontine nuclei (PPN) of the brainstem in gait and freezing [16–18]. Early involvement of the frontostriatal pathways in FOG, however, contradicts the commonly accepted Braak hypothesis of disease spread in PD emphasises early brainstem pathology [19]. Although Braak’s model elegantly integrates motor and non-motor symptoms, it is unable to currently account for the appearance of non-motor functions that precede the motor symptoms [20], suggesting that the pathological degradation of regions outside the brainstem might be responsible for the complex, non-motor symptoms of PD. Little is known about whether patients who experience freezing in the more advanced stages of PD have the same clinical phenotype that has been characterized for those in the early clinical stages of disease. Furthermore, it is not clear whether non-freezers in the more advanced stages of PD demonstrate a distinct clinical phenotype that might also inform the mechanisms underlying FOG. The aim of this exploratory cross-sectional study was to determine how key phenotypic features, including motor symptoms, cognition, mood, RBD and dysautonomia might be related to the freezing phenomenon FOG in the advanced stages (H&Y ≥ 2.5) and if this demonstrates a similar pattern as in the early stages (H&Y 1-2). We hypothesized that patients with FOG show impairments on variables related to frontostriatal dysfunctions. Conversely, we expected no differences on variables related to brainstem pathology in both the early and advanced stages.

METHODS

Subjects

Data was prospectively collected from 389 patients between 2009 and 2013 at the Parkinson’s Disease Research Clinic at the Brain and Mind Research Institute, University of Sydney. Idiopathic PD was diagnosed based on the UKPDSBB criteria [21] and confirmed by a trained Neurologist (SJGL). Exclusion criteria included the presence of other neurological diseases or other conditions that are linked to gait impairment or dementia as rated by the Movement Disorder Society (MDS) Task Force criteria [22]. Freezers were firstly identified as those who were observed to freeze during MDS-UPDRS assessments (question 3.11). Additionally, patients were included if they a positive score on FOG-Questionnaire item 3 (“Do you feel that your feet get glued to the floor while walking, making a turn or when trying to initiate walking (freezing)?”) [23]. This item is the most accessible tool and has previously been shown to be a reliable screening instrument [24] and was included to avoid false negative classification that is common when relying on clinically observed ON-state FOG only. Patients were divided into four different groups, on the basis of the presence of FOG and the severity of PD according to the modified Hoehn and Yahr (H&Y) motor clinical staging [25] distinguished by postural instability. Patients in H&Y stages 1 and 2 were classified as early PD patients with FOG (E-FOG) and early PD patients without FOG (E-NF). Patients in H&Y stages 2.5–5 with FOG were classified as advanced patients with (A-FOG) and without FOG (A-NF). All patients were assessed on their regular medication and their dopamine dose equivalence (LEDD; mg/day) [26] was recorded. Written informed consent for this study was obtained from each patient, and the University ofSydney Human Research and Ethics Committee approved the study.

Data collection

Age and time since diagnosis were recorded for each patient. The Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) [27] was obtained as well as the motor sub-score, excluding question 3.11 (“Freezing of gait”). Using a previously described method [7], the tremor score was derived from the average of questions 2.10 and 3.15–3.18 of the MDS-UPDRS and a global score for non-tremor features was obtained by averaging questions 2.1, 2.9, 2.11, 2.12, 3.1–3.10, 3.12–3.14.

All patients were assessed on the Mini Mental State Examination (MMSE) [28] to measure general cognitive function for descriptive purposes. More detailed neuropsychological testing included psychomotor speed (TMT_A) and attentional set-shifting (TMT_B) [29]. To control for psychomotor speed in part B of the TMT, the time to complete part A was subtracted from the time to complete part B (TMT_B-A) [30]. The Digit Span (DS) Forward and Backward sub-tests (total score) were used to assess attention and working memory and Logical Memory (LM) Retention score (Wechsler Memory Scale –III) assessed verbal memory [31]. Verbal (letters F, A, S) and semantic fluency (Animals) was also assessed [32].

The Hospital Anxiety and Depression Scale (HADS) was obtained to assess affective function [33]. HADS depression and anxiety scores were compared between the groups, as well as the percentage of patients who scored above the cut-off score of 8, which indicates clinically significant symptomatology [33].

The RBD Screening Questionnaire (RBDSQ) [34] was used to investigate symptoms of RBD. The autonomic function score was derived from averaging questions 1.10, 1.11, 1.12, 1.13 and 2.2 of the MDS-UPDRS.

Statistical analyses

All data analyses were performed using IBM SPSS Version 22. Due to the differences in sample size and inequalities of variances, Kruskal Wallis tests were used to determine differences between the four groups. To investigate whether there are any phenotypical differences between the groups, E-FOG was compared to E-NF, and A-FOG was compared to A-NF. In addition, to investigate how phenotypical changes progress, E-FOG was compared to A-FOG and E-NF was compared to A-NF. Significant results were followed up with pairwise comparisons with a Mann Whitney test to determine which groups were significantly different from each other. In view of the exploratory nature of the study, p-values were reported before corrections for multiple comparisons were made due to the increased risk of false negatives [35]. However, Dunn’s Bonferroni approach was also implemented and significant p-values that survived this correction are marked with a ⁺ in Table 2. Furthermore, Jonckheere tests were used to calculate effect sizes. Gender differences, depression and anxiety were analysed using the Pearson’s Chi-Square test. Z-scores for TMT [36] and verbal fluency [37] were calculated and LM scores [38] were adjusted for age. Z-scores based on normative data were not computed for DS Forward and Backward tests, as normalized data was not available for these scores individually. Furthermore, the raw scores of the TMT were used to measure TMT_B-A. In keeping with previous studies, the cut off score for TMT_A was set at 180 s and at 300 s for TMT_B [39]. All analyses used an alpha of 0.05 and were two-tailed.

RESULTS

In total 148 patients were classified as non-freezers (38% ) whereas 241 patients (62% ) were classified as freezers. The percentages of freezers and non-freezers across all H&Y stages are presented in Fig. 1. Sixty-nine patients were identified as E-FOG, the E-NF group consisted of 113 patients. The A-FOG consisted of 172 patients and the A-NF of 35 patients.

Clinical characteristics

The medians and 25th and 75th percentiles of demographic data are presented in Table 1. Table 2 demonstrates the statistical differences between the four groups. As expected, patients in the advanced stages were older than patients in the early stages. No differences in age or gender were observed between the FOG and NF patients. Both the freezer groups had a significantly longer time since diagnosis than their respective non-freezer groups. E-FOG patients had significantly higher LEDD scores compared to E-NF. Interestingly, LEDD scores did not differ between the early and advanged stages. Only three E-FOG patients (5% ) had a LEDD score of zero and four of the A-FOG patients (2% ) was untreated or refused to take medication. Seven patients in the E-NF group (7% ) and five patients in the A-NF group (15% ) had a LEDD score of zero. In the A-NF group, all unmedicated patients had discontinued medications due to side-effect ofconfusion.

Motor function

As demonstrated in Tables 1 and 2, both the A-FOG and A-NF groups displayed higher total MDS-UPDRS, motor MDS-UPDRS-III and non-tremor scores comparing to the E-FOG and E-NF group respectively. A significant increase in tremor score was observed in A-NF compared to E-NF group. Compared to E-NF, E-FOG had significantly higher scores on the total MDS-UPDRS, and both FOG groups performed more poorly on the non-tremor measures compared to the NF groups. Despite higher scores for the NF groups, the tremor score was not significantly different between FOG and NF patients after controlling for multiple comparisons.

Cognition

Compared to E-FOG patients, the A-FOG group had significantly worse scores on the MMSE and LM Retention. Compared to E-NF patients, A-NF patients scored significantly worse on the MMSE and TMT_B-A. No other differences were observed between these groups. E-FOG patients scored more poorly thanE-NF patients on the MMSE, TMT_A, TMT_B,TMT_B-A, DS-Backward, phonemic and semantic verbal fluency, however, after controlling for the multivariate approach only the differences on TMT_A, phonemic and semantic verbal fluency performance remained significant between the groups. In the advanced stages, compared to the A-NF group, the A-FOG group had significantly poorer scores on DS- Backward, but this difference did not survive after controlling for multiple comparisons. However, due to the small number of patients in the A-NF group, these results should be interpreted with caution.

Mood

Depression scores increased in both FOG and NF as the disease progressed. In contrast, anxiety was not related to disease stage (see Tables 1 and 2). Freezers in both the early and advanced stages had higher anxiety and depression scores compared to non-freezers, however the difference did not reach significance when using a multivariate approach. Clinically significant anxiety was found in 25.0% in the E-FOG group and in 31.9% of the patients in the A-FOG group, compared to 9.9% and 17.1% in the E-NF and A-NF groups. In 6.3% of the E-NF patients and 22.9% of the A-NF patients, clinically significant depression symptoms were reported, whilst the prevalence of depression in the E-FOG and A-FOG groups were 17.9% and 39.2% respectively.

Autonomic function and REM sleep behaviour disorder

Compared to early disease stage, both the A-FOG as well as the A-NF group had significantly higher autonomic scores, but no differences were found between freezers and non-freezers in both early and advanced disease. RBD was independent of disease stage in both groups. Both FOG groups had higher RBDSQ scores than than the NF patients, reaching significance in the early clinical disease stages.

DISCUSSION

This large-scale cross-sectional study investigated the association between different characteristics of PD and FOG relative to disease severity. Here, we provide an in-depth comparison of the clinical phenotype of patients with and without FOG, at early versus advanced clinical disease stages. In doing so, a number of critical novel findings have emerged, which inform possible mechanistic explanations of the emergence of FOG across disease stages.

Regardless of disease stage, patients with FOG performed worse on their motor assessment particularly in relation to non-tremor symptoms. However, freezers and non-freezers demonstrated similar tremor scores in both clinical stages, but an increase in tremor score between the clinical stages was only found in non-freezers. Although tremor and FOG share some characteristics such as an increased prevalence during mental stress and an episodic character, there are significant distinct mechanisms underlying each symptom [7]. Specifically, the cerebello-thalamo-cortical circuit is likely to be involved in the incidence of tremor [40] rather than arising from frontostriatal pathways.

Previous studies have demonstrated a selective cognitive impairment in freezers that typically affects executive functions such as verbal fluency [41], conflict resolution [42, 43] and set-shifting [44]. Indeed, the current study demonstrated that FOG patients scored worse on measures for phonemic and semantic verbal fluency in the early stages. Additionally, though non-significant after controlling for multiple comparisons, freezers performed more poorly on set shifting in the early stages of the disease and working memory in both the early and advanced stages. In contrast, memory retention performance, which is known to reflect temporal lobe function, did not differ between groups across disease stages, thus reinforcing the possible selectivity of frontostriatal impairment in FOG. Although current results do not show a clear division between FOG and NF patients, a difference between these patient groups is plausible and in accordance with recent work using functional neuroimaging techniques. Namely, it has been shown that there is a paroxysmal functional decoupling of activity between neural networks including the cognitive control network and the basal ganglia in patients with FOG [45]. Such impaired information processing in the frontostriatal system might be influenced by a faulty selective gating mechanism across the basal ganglia, which would normally be responsible for the rapid updating of information [46]. However, it is important to note that not all freezers present with executive dysfunction, pointing out the likelihood of different causes of FOG [47]. Indeed, FOG is not due to executive dysfunction per se, but rather to impairments in conflict processing independent of the processing system of which the conflict occurred. As such, motor, cognitive, affective or perceptual conflict may underlie freezing behaviour, given that dysfunctional activity within each of these circuits is processed via a final common neural pathway, i.e. the subthalamic nucleus [48].

The current study demonstrated that in contrast with depression, anxiety is not related to disease stage. Freezers showed a trend towards significantly higher average anxiety scores. This is in accordance with an elegant study recently published, showing differences on depression and anxiety scores between FOG and non-FOG patient. However, this difference disappeared when using a multivariate approach [9]. The exact etiology of mood disturbances remains unknown [49]. The interplay of emotion and FOG might have an underlying neurobiological basis in shared frontostriatal limbic circuitry, an interpretation consistent with a SPECT study that found that bilateral ventromedial prefrontal cortex showed decreased perfusion in freezers compared to non-freezers [50]. However, previous work has also highlighted involvement of the locus coeruleus [51, 52]. Therefore, as the current results are inconclusive, the association between freezing and anxiety and depression needs further exploration, as well as its influence on other variables associated with FOG, such as executive functioning, methods of assessments and cut-off scores, and their underlying pathophysiology. A relationship between these phenomena has potentially significant implications for future treatment strategies, such as targeted cognitive-behavioural therapy that would be designed to manage anxiety [53], a non-pharmacological approach that would be advantageous in a patient population often requiring polypharmacy [2].

Higher RBD scores found in this study are in contrast with our previous study that used more stringent inclusion criteria, pointing out a role of disease progression. However, only in the early stages the RBD scores remained significantly different between the FOG and NF groups when using a multivariate approach. RBD suggest brainstem pathology [54], yet associations with executive dysfunction and mild cognitive impairment have been frequently reported [55, 56]. Finally, autonomic functions did not differ between FOG and NF regardless of disease stage. These findings suggest that the pathological processes within the brainstem underlying autonomic functions are not intrinsically associated with freezing behaviour [57].

Braak and colleagues have identified distinct stages in the progression of PD that spreads rostrally from several nuclei in the brainstem with little inter-individual variability [19]. Despite this pattern of progression, disease heterogeneity is well recognised in PD. Specifically, in relation to FOG, the results of our study do not indicate that early or advanced brainstem pathology uniquely contributes to FOG. Indeed, a recent positron emission tomography study demonstrated that the effects of cholinergic degeneration associated with FOG were driven by neocortical degeneration but not by PPN-thalamic degeneration, which was more associated with postural reflex impairments [58]. This finding would be aligned with our observations and suggests that pathology within the PPN by itself is unlikely to be the underlying cause of FOG in PD.

Whilst the partial amelioration of FOG bydopaminergic agents presumably operating through frontostriatal networks is well documented [59, 60], there is evidence for a dopamine resistant form of freezing [9, 61]. Furthermore, cholinergic degeneration in PD has been associated with a non-tremor dominant phenotype [62] along with executive and attentional dysfunction [63], depression and apathy [64]. These observations highlight the complex interplay required between complementary networks and neurotransmitter systems that when compromised lead to FOG via dysfunction across cortical and subcortical regions. Furthermore, the paroxysmal nature of FOG suggests that FOG does not result from structural lesions, but rather from bouts of decreased neuronal activity. Indeed, FOG is best conceptualized as a functional disorder with multiple possible pathological underpinnings across a range of different areas in the brain.

Limitations and further directions

Whilst the current study raises several important clinical issues in relation to FOG in PD, it does suffer with a number of limitations. Group sizes were not equal and the data was only attained at one time point. Therefore, it was not possible to determine when FOG had initially presented in the different groups. As such, more accurate temporal relationships between neuropathology and phenotypic features could not be explored. All patients were assessed on their optimal ON state and no differences were observed for time since last dose. However, it is hard to elicit ON-state FOG in the clinical setting; Levodopa helps in reducing the frequency and duration of ‘off’-related FOG [65], and thus many patients report significant FOG at home [49, 66]. Therefore, we have included a subjective questionnaire to identify freezers. Future studies will be necessary to assess the effect of dopaminergic medication, including potential wearing-off phenomena along with other neurotransmitter disturbances in PD patients with FOG. Furthermore, the inability to match subgroups on all demographic features may also have influenced our findings, but it should be acknowledged that it might not be possible to control for the effects of FOG, such as the use of higher dopaminergic therapy in an attempt to reduce freezing episodes. Both FOG groups had a significantly longer time since diagnosis. Therefore, future longitudinal studies observing the evolution of FOG within patients may offer important further insights to those offered here. In addition, the executive functioning scores should be interpreted with caution. Only 15–17 patients in the A-NF group performed these tests and not all differences remained significant after controlling for multiple comparisons. Finally, whilst we acknowledge that using solely behavioural tests in this study limits our speculation regarding the pathophysiology underlying FOG, our results are aligned with previous functional imaging data [67, 68].

CONCLUSIONS

In conclusion, this study suggests that FOG is associated with non-tremor motor severity, selective executive dysfunction, mood disturbance and RBD but not more general autonomic dysfunction or tremor. In patients with more severe tremor, the frontostriatal pattern of neuropathology might be less affected in these patients compared to patients in the non-tremor phenotype. The current findings do not support the notion that distinctly different pathophysiological processes may underlie the freezing phenomenon across advancing disease stages, however the differences between the groups become less strict with progressing pathology. Future prospective clinicopathological studies may help better understand these associations and could provide insights into novel therapies.

FINANCIAL DISCLOSURES

JM Hall is supported by a University of Western Sydney Postgraduate Research Award. JM Shine is supported by a NHMRC CJ Martin Fellow Award. C O’Callaghan is supported by a NHMRC Neil Hamilton Fairley Award. CC Walton is supported by an Australian Postgraduate Award at the University of Sydney. M Gilat is supported by an International Postgraduate Research Scholarship at the University of Sydney. SL Naismith is supported by a National Health and Medical Research Council Career Development Award No. 1008117. SJG Lewis is supported by a National Health and Medical Research Council Practitioner Fellowship No. 1003007.

CONFLICT OF INTEREST

The authors report no conflicts of interest.