Article Text

Download PDFPDF

Comparative analysis of gait and speech in Parkinson's disease: hypokinetic or dysrhythmic disorders?
  1. Stéphanie Cantiniaux1,2,
  2. Marianne Vaugoyeau2,
  3. Danièle Robert3,
  4. Christine Horrelou-Pitek3,
  5. Julien Mancini4,5,
  6. Tatiana Witjas1,
  7. Jean-Philippe Azulay1,2
  1. 1Department of Neurological Sciences, Movement Disorders Unit, hôpital de la Timone, Marseille cedex, France
  2. 2CNRS—Université Provence, Site Saint Charles, Marseille, France
  3. 3Department of Otorhinolaryngology, Hôpital de la Timone, Marseille, France
  4. 4Aix Marseille Université, LERTIM, Marseille, France
  5. 5Department of Public Health and Medical Information t, hôpital de la Timone, Marseille cedex, France
  1. Correspondence to Jean-Philippe Azulay, Service de Neurologie et Pathologie du Mouvement. Hôpital de la Timone, 264, rue Saint-Pierre, 13385 Marseille cedex 05, France; jean-philippe.azulay{at}ap-hm.fr

Abstract

Gait and speech are automatic motor activities which are frequently impaired in Parkinson's disease. Obvious clinical similarities exist between these disorders but were never investigated. We propose to determine whether there exist any common features in Parkinson's disease between spatiotemporal gait disorders and temporal speech disorders.

Gait and speech were analysed on 11 Parkinsonian patients (PP) undergoing deep-brain stimulation of the subthalamic nucleus (STN-DBS) and 11 control subjects under three conditions of velocity (natural, slow and speed). The patients were tested with and without l-dopa and stimulator ON or OFF. Locomotor parameters were recorded using an optoelectronic system. Speech parameters were recorded with a headphone while subjects were reading a short paragraph.

The results confirmed that PP walk and read more slowly than controls. Patient's difficulties in modulating walking and speech velocities seem to be due mainly to an inability to internally control the step length and the interpause-speech duration (ISD).

STN-DBS and levodopa increased patients' walking velocity by increasing the step length. STN-DBS and levodopa had no effect on speech velocity but restored the patients' ability to modulate the ISD. The walking cadence and speech index of rythmicity tended to be lower in patients and were not significantly improved by STN-DBS or levodopa. Speech and walking velocity as well as ISD and step length were correlated in both groups. Negative correlations between speech index of and walking cadence were observed in both groups.

Similar fundamental hypokinetic impairment and probably a similar rhythmic factor similarly affected the patients' speech and gait. These results suggest a similar physiopathological process in both walking and speaking dysfunction.

  • Parkinson's disease
  • locomotion
  • speech
  • deep brain stimulation
  • motor control
  • movement disorders
View Full Text

Statistics from Altmetric.com

Introduction

Basal ganglia dysfunction leads to the lack of automatic execution of learnt motor sequences21 and to the development of Parkinsonian motor symptoms,19 including changes in gait and speech.

Gait disorders in Parkinson's disease (PD)16 constitute one of the main factors affecting patients' independence and quality of life.18 PP walk more slowly, with shorter step length and a compensatory increased walking cadence.5 24 25 Levodopa improves spatiotemporal parameters such as walking velocity and step length, but not temporal parameters such as step time, swing time and walking cadence.3 5 26 Studies have shown that Subthalamic-Nucleus Deep Brain Stimulation (STN-DBS) improves walking velocity by increasing the step length without affecting the walking cadence,7 20 while a modulation of cadence has been reported by others.30

Approximately 70% of Parkinsonian patients (PP) experience speech difficulties attributed to bradykinetic and hypokinetic articulatory movements, orofacial hypomimy, rigidity and rest tremor.15

PPs' speech timing deficits seem to have some points in common with their spatiotemporal gait deficits. Speech velocity often decreases in PP6 22 just like walking velocity. The interpause-speech duration (ISD) is usually shortened in PP,9 13 in line with the step length.

PP are subject to dysprosody and show less variability in the fundamental frequency and intensity and more speech velocity and pausing abnormalities than control subjects (CS).6 14 17

Dysfluent behaviour has also been described in PP,4 12 consisting mainly of repetitive speech phenomena involving the repetition of syllables, words or phrases. Skodda and Schlegel28 reported that during a standardised reading task, the speech velocity of PP accelerates more strongly while speaking, and the total number of pauses decreases significantly, which indicates that their speech rhythm and timing organisation processes are impaired. Physicians often have the feeling that paroxystic speech velocity accelerations develop in line with gait festinations and dysrhythmic pauses, and that repetitive speech phenomena such as stuttering, iterations and palilalia are somewhat comparable with freezing of gait. Few studies have focused on the effects of levodopa on speech-timing parameters. Levodopa does not improve speech velocity32 or stuttering.1 4 Speech dysfluencies may be due to an increase or decrease in the dopamine levels present in the brain. Two studies have shown that selective left-side STN-DBS may have negative effects on prosody,27 31 but the quantitative effects of STN-DBS on specific temporal speech parameters have never been assessed.

Our aim was to compare gait and speech patterns of PP and CS, and to analyse the patients' responses to levodopa and STN-DBS. Another aim was to determine whether there exist any correlations between the spatiotemporal gait and temporal speech parameters with a view to establishing whether, in line with what occurs with gait, hypokinesia may be mainly responsible for these patients' speech timing disorders and whether accelerated timing strategies may be used as compensatory mechanisms.

Material and methods

Subjects

Eleven PP, seven males and four females (mean age 64.5) and 11 age-matched CS, seven males and four females (mean age 66.3) with no neurological history took part in these experiments after giving their informed consent. The project was approved by the local ethical committee. All the PP had undergone bilateral STN-DBS and were stable when the study was conducted. PP and controls with any other disorder possibly affecting their walking or speaking abilities were excluded from the study.

PPs' level of functional disability was determined on the Unified Parkinson's Disease Rating Scale motor and axial scores in four treatment conditions (OFF and ON states of STN-DBS and levodopa). The patients' characteristics are given in table 1.

Table 1

Medical characteristics of the 11 Parkinsonian patients: sex, age and disease duration, Unified Parkinson's Disease Rating Scale (UPDRS) part III and axial scores, effect of l-dopa and STN as a percentage of the OFF/OFF score, in four states of treatment performed at the time of testing

Test procedure

Gait and speech recordings were conducted in the four treatment conditions, always in the same order. Gait recordings were run always just before speech recordings.

  1. PP were tested in the OFF STN-DBS state at least 12 h after undergoing STN-DBS treatment and in the OFF levodopa state at least 12 h after their last dose of dopaminergic treatment (DT) (OFF/OFF).

  2. The PP were tested in the ON STN-DBS state at least 1 h after undergoing STN-DBS treatment and in the ON levodopa state at the peak effect of a suprathreshold dose of levodopa (ON/ON).

  3. The PP were tested in the ON STN-DBS state at least 12 h after undergoing STN-DBS treatment and in the OFF levodopa state at least 12 h after taking their last DT dose (ON/OFF).

  4. The PP were tested in the OFF STN-DBS state at least 3 h after undergoing STN-DBS treatment and in the ON levodopa state at the peak effect of a suprathreshold dose of levodopa (OFF/ON).

The gait and speech recordings were conducted just once on the CS.

Gait recordings

Tasks

Subjects were instructed to walk on level ground at their natural speed (NS), slowest as compared with the first task (SS) and as fast as possible without running (FS).

Experimental recordings

Kinematic analysis was performed with the ELITE TV image processing system.8 Twenty-two reflective markers 1 cm in diameter were taped onto the subjects' skin at specific symmetrical anatomical points. The recordings were taken with six TV cameras at a sampling rate of 100 Hz.

Gait parameters

Based on the kinematic recording, temporal and spatial gait parameters were measured during each trial: walking velocity (m/s), step length (mm) and walking cadence (step/min).

Speech recordings

Tasks

Subjects had to read a short printed text at their natural speed, slowest as compared with the natural condition and as fast as possible.

Experimental recordings

Subjects were seated in a quiet room. Temporal acoustic recordings obtained via a headphone were analysed using the Prosody software program, which is part of the Software Environment for Speech Analysis and Evaluation (SESANE).

Speech parameters

Based on the acoustic data, the following parameters were measured at each trial:

  • Speech velocity (syllables/s);

  • ISD in seconds: this corresponds to the time interval between two pauses and was determined by tallying the duration occurring between two pauses;

  • Speech Index of Rhythmicity (SPIR): this is an index to evaluate speech rhythmicity in the same way as the walking cadence; as the alternation between pauses and speech gives rhythm to speech just as the alternation between steps and double foot support phases gives rhythm to gait, SPIR corresponds to the number of speech interpauses per minute.

Statistical procedure

Because of the small size and the non-normal distribution of our samples, all the tests applied were non-parametric tests. For each subject and condition, all variables have been averaged over the three trials. In the OFF/OFF situation, two PP were unable to walk. Nevertheless, we have used the data of these patients in the other conditions of treatment. The statistics given are medians and interquartiles.

Differences between CS and PP were tested with a Mann–Whitney U test. The task effect in each group and the treatment effect in the PP group were analysed with a Friedman ANOVA. In order to isolate statistical differences when the hypothesis of differences was verified, we have realised a multiple comparison procedures (post-hoc tests) using a Student–Newman–Keuls test. The correlations between gait and speech parameters were tested with a Spearman test.

Results

Gait analysis

Effects of PD

In order to assess the effects of PD on gait parameters, the gait characteristics of PP in the OFF/OFF state of treatment were compared with those of the CS.

Whatever the experimental situations, gait velocity of PP was slower than that of CS (T(1.18)=61, p<0.05; T(1.18)=62, p<0.05 and T(1.18)=58, p<0.05 in conditions NS, SS and FS, respectively); the mean step length was shorter (T(1.18)=57, p<0.01; T(1.18)=59, p<0.01 and T(1.18)=59, p<0.01 in conditions NS, SS and FS, respectively); whatever the experimental task, the cadence was similar in both groups.

Effects of treatment

The gait characteristics of PP in each mode of treatment are given in figure 1.

Figure 1

Task effect on gait parameters. From top to the bottom: medians and quartiles of walking velocity, step length and walking cadence values recorded in Parkinsonian patients in the three experimental conditions (from left to right) natural speed (NS), slow speed and fast speed (SS and FS) in the four states of treatment, from left to right: without treatment (OFF/OFF), without STN-DBS but with levodopa (OFF/ON), with STN-DBS but without levodopa (ON/OFF) and with both STN-DBS and levodopa (ON/ON).

Walking velocity

The statistical analysis revealed a global effect of the treatment on walking velocity in all experimental condition (χ2=9.41, p<0.01; χ2=17.93, p<0.001 and χ2=15, p<0.01 for NS, SS and FS, respectively). Walking velocity increased similarly with levodopa, DBS and a combination of both compared with the complete OFF state.

Step length

In the NS condition, we showed a significant global effect of treatment on step length (χ2=15.27, p=0.002). Statistical analysis showed a significant increase in the step length in the ON/ON state of treatment in comparison with the other states. The Student–Newman–Keuls post hoc analysis revealed a significant decrease in the step length in the OFF/OFF state of treatment as compared with the ON/OFF and OFF/ON states of treatment. No significant differences were observed between OFF/ON and ON/OFF states, indicating that levodopa and DBS similarly improved step length and that the combination of both therapies had an additional effect.

In SS and FS, we showed a significant global effect of treatment on step length (χ2=15.8, p=0.001 and χ2=11.8, p<0.01, for SS and FS respectively). The step length was significantly shorter in the OFF/OFF state than in the other. No significant differences were observed between the states OFF/ON, ON/OFF or ON/ON.

Walking cadence

In the NS and SS conditions, the walking cadence did not differ significantly between all states of treatment. In the SS condition, a significant overall effect of treatment was observed (χ2=8.33, p<0.05). The multiple comparisons revealed that the walking cadence was significantly lower in the OFF/OFF state of treatment than in the other.

Task effect: The gait characteristics in each experimental task are given on Figure 2

Walking velocity

A significant effect of task was observed in CS (χ2=22, p<0.001) and in treated PP (χ2=22, p<0.001; χ2=20, p<0.001 and χ2=18.2, p<0.001 in the OFF/ON, ON/OFF and ON/ON states, respectively): The walking velocity increased significantly between SS and FS. Without treatment, a significant effect of task was also observed in PP (χ2=12.67, p<0.001), but interestingly, no significant difference was found between NS and FS, indicating that PP were unable to increase significantly their velocity.

Figure 2

Treatment effect on gait parameters. Medians and quartiles of walking velocity, step length and walking cadence values recorded in the natural speed (in white), in the slow speed (in grey) and in the fast speed (in black) conditions, from left to right: without treatment (OFF/OFF), without STN-DBS but with levodopa (OFF/ON), with STN-DBS but without levodopa (ON/OFF) and with both STN-DBS and levodopa (ON/ON) and in control subjects (CS).

Step length

A significant effect of task on step length was observed in CS (χ2=22, p<0.001) and in treated PP, whatever the treatment the step length increased significantly between the SS and the FS condition (χ2=22, p<0.001; χ2=16.8, p<0.001 and χ2=14.6, p<0.001 in the OFF/ON, ON/OFF and ON/ON states, respectively). In untreated PP, a significant effect of task was also observed (χ2=11.56, p=0.001), but no significant difference was found to exist between NS and FS conditions.

Walking cadence

A significant effect of task was observed in CS (χ2=16.55, p<0.001) and in treated PP (χ2=16.22, p<0.001; χ2=22, p<0.001; χ2=18.2, p<0.001 and χ2=20, p<0.001 for OFF/OFF, OFF/ON, ON/OFF and ON/ON states of treatment, respectively). The walking cadence increased significantly between SS, NS and FS.

Speech analysis

Effects of PD

During NS and SS tasks, speech velocity did not differ significantly between the two groups. In the FS task, the speech velocity was significantly lower in PP (T=93.5, p<0.05).

In the NS task, ISDs' values were similar in both groups. In the SS task, the ISD's value of PP was significantly longer than that of CS (T=160.5, p<0.05), whereas in the FS task, the ISD of PP was significantly shorter than that of CS (T=91, p<0.05).

In NS and FS tasks, the SPIR did not differ significantly between the two groups. In the SS task, the SPIR values of PP were significantly lower than those of CS (T=93, p<0.05).

Effects of treatment

In order to assess the effects of treatment, speech performances of PP were compared depending on the treatment undergone. The speech performances of PP are given in figure 3.

Figure 3

Task effects on speech parameters. Medians and quartiles of speech velocity, ISD and speech index of rythmicity (SPIR) recorded in Parkinsonian patients in the three experimental conditions (natural speed, slow speed and fast speed) in the four states of treatment, from left to right: without treatment (OFF/OFF), without STN-DBS but with levodopa (OFF/ON), with STN-DBS but without levodopa (ON/OFF) and with both STN-DBS and levodopa (ON/ON).

Speech velocity

Whatever the experimental condition, no significant differences were observed between the four states of treatment.

ISD

Concerning the ISD, the Friedman ANOVA showed significant differences between the four states of treatment during NS (χ2=9.275, p<0.05) and FS condition (χ2=1.175, p<0.01) but not during the SS condition. ISD increased significantly with levodopa, DBS and combination of both therapies as compared with the complete OFF state.

SPIR

The SPIR did not differ significantly between any of the states of treatment during NS and SS conditions. During the FS condition, the statistical analysis showed the existence of a significant overall effect of treatment on SPIR (χ2=8.46, p<0.05), which were significantly higher in the OFF/OFF state of treatment than in the other states.

Task effects: The speech characteristics in each experimental task are given on Figure 4

Speech velocity

A significant effect of task on speech velocity was observed in CS (χ2=16.545, p<0.01) and in treated PP (χ2=14.727, p<0.01; χ2=20.182, p<0.001 and χ2=14.364, p<0.01 in the OFF/ON, ON/OFF and ON/ON states, respectively): the speech velocity increased significantly between the SS and the FS condition. A significant effect of task was also observed in the OFF/OFF PP (χ2=17.636, p<0.01): speech velocity increased significantly between NS and FS condition, but not between SS and NS conditions.

Figure 4

Treatment effects on speech parameters. Medians and quartiles of speech velocity, ISD and speech index of rythmicity (SPIR) recorded in the natural speed (in white), in the slow speed (in grey) and in the fast speed (in black) conditions, from left to right: without treatment (OFF/OFF), without STN-DBS but with levodopa (OFF/ON), with STN-DBS but without levodopa (ON/OFF) and with both STN-DBS and levodopa (ON/ON) and in control subjects (CS).

ISD

A significant effect of task on ISD was found in CS (χ2=13.273, p<0.01) and in treated PP (χ2=10.093, p<0.01; χ2=8.727, p<0.05 and χ2=13.273, p<0.01 in the OFF/ON, ON/OFF and ON/ON states, respectively): the ISD increased significantly between the SS and FS condition. It is worth noting that the increase in the ISD between NS and FS was significant only in the ON/ON state of treatment. In PP in the OFF/OFF state of treatment, no significant differences were detected.

SPIR

A significant effect of task on SPIR was observed in CS (χ2=8.909, p<0.001): the SPIR values decreased significantly in this group between SS and NS and between NS and FS. No significant effects of task on SPIR were detected in PP.

Correlations between gait and speech parameters

In order to measure the strength of association between gait and speech parameters, we have used the Spearman coefficient of correlation. We have correlated gait velocity with speech velocity, walking cadence with SPIR and step length with ISD. We have brought together all the velocity of task executions.

Statistical correlations between walking velocity and speech velocity

Significant correlations were found to exist between walking and speech velocities in CS (r=0.84, p<0.001), in PP in the OFF/OFF (r=0.717, p<0.001), in the ON/OFF (r=0.505, p<0.01) and in the OFF/ON states of treatment (r=0.4325, p<0.05) but not in the ON/ON sate.

Correlations between step length and ISD

A significant correlation was found to exist between step length and ISD in CS (r=0.609, p<0.001), in PP in the ON/OFF (r=0.486, p<0.01) and ON/ON (r=0.421, p<0.05) state of treatment. No statistical correlations were observed in the OFF/OFF and OFF/ON states of treatment.

Correlations between walking cadence and SPIR

A significant negative correlation was found between walking cadence and SPIR in CS (r=−0.702, p<0.001) in PP in the OFF/OFF (r=−0.418, p<0.05) and in the OFF/ON state of treatment (r=−0.438, p<0.05), but not in PP in the ON/OFF or ON/ON states of treatment.

Discussion

Our results demonstrate large similarities between the gait's pattern and speech's involvement in PD and in their responses to STN-DBS and levodopa treatment.

Gait and speech impairments in PD

The spatiotemporal gait disorders observed in PD are in line with those previously described: without treatment, walking velocity and step length of PP are decreased, while walking cadence is not significantly different.3 25 29 Neither walking velocity nor step length increased when the untreated patients were asked to walk as fast as possible, suggesting that they were unable to regulate their step length on the basis of internal control mechanisms. In contrast, the patients' internal walking cadence regulation process seems to be preserved, since their walking cadence increased significantly between SS and NS and between NS and FS. This fundamental hypokinetic deficit has previously been described by Morris et al.25

The results of the speech analysis brought to light the existence of speech timing disorders showing some analogies with the gait impairments. Without any treatment, PP tended to speak more slowly than the CS, and their ISD often tended to decrease in comparison with the control values, in parallel with the decreasing step length. The SPIR showed some similarities with the walking cadence, since it was equal to the control values at NS and FS. Few data are available to compare our results, Canter6 and Metter and Hanson22 observed changing speech velocities in PP. The ISD has often been reported to be shorter in PP than in CS during both reading and spontaneous speech.9 13 The speech velocity also changes with the coexistence of bradylalia and tachylalia.11 The present results show in addition that the speech velocity of PP without treatment did not differ between SS and NS, and that their ISD remained unchanged under all the experimental conditions. In line with the changes observed in their walking patterns, PP therefore have some difficulty in modulating their speech velocities because they are no longer able to control the ISD internally. However, the internal speech rhythm regulation processes seem to be impaired also, since the SPIR remained unchanged at NS and FS condition.

Effect of levodopa and STN-DBS on gait and speech deficits

As regards the patients' responses to treatment, their walking velocities were greatly improved by both STN-DBS and levodopa, probably due to an increase in the step length, whereas the walking cadence remained unchanged in NS and FS. In all tasks, no significant difference between STN-DBS and levodopa was observed, but a cumulative effect was observed in the NS condition. Two previous studies show that STN-DBS, like levodopa, improves walking velocity by increasing step length without changing walking cadence.7 20 These findings suggest that STN-DBS may involve the same action mechanisms as levodopa via the dopaminergic basal ganglia loop, while the cadence, which is a rhythmic parameter, is probably regulated by a non-dopaminergic structure.

Levodopa and STN-DBS had no effect on the patients' speech velocity but increased the ISD. Combined treatment restored the patients' ability to modulate their ISD. The ISD therefore seems to be the main clinical manifestation of speech akinesia, since this parameter is regulated by the dopaminergic basal ganglia loop. The effects of treatment on the SPIR were often not significant. This finding probably means that in the case of both speech and gait, the rhythm is regulated by a non-dopaminergic structure.

Gait and speech deficits in PD: a common feature?

This is the first time that correlations have been made between gait and speech parameters in PD. In CS and PP undergoing no treatment or under STN-DBS treatment alone, a significant correlation was found between walking velocity and speech velocity. In other words, CS and PP who walk faster also speak faster. This outcome seems to be modified by levodopa.

A significant positive correlation between step length and ISD was also found. This confirms that step length and ISD decrease are clinical manifestations of both gait and speech akinesia, that they are both responsive to levodopa and STN-DBS, and that just as akinesia is the main gait deficit, speech akinesia may be the main speech deficit in PD. The negative correlation between walking cadence and SPIR in CS underlines the different strategies used to modulated gait and speech velocity: to walk faster, CS increase the number of steps they make, whereas to speak faster, they decrease the number of speech interpauses and pauses. This result suggests that the SPIR is a good index to analyse speech rhythmicity, just as the walking cadence is a good index to analyse gait rhythmicity. In PD, a similar phenomenon was observed only under levodopa, suggesting that walking cadence and SPIR are regulated by different non-dopaminergic structures and that rhythmic perturbations do not always affect speech and gait concurrently, but that they may occur earlier in speech, where a twofold deficit occurs due to the patients' inability to internally regulate either the number of pauses or the ISD.

Conclusions

As recently suggested by Moreau et al,23 who established that oral festinations were correlated with gait festinations in PP, the present results underline several common features between gait and speech troubles in PD. The decrease in PPs' walking velocity and speech velocity result from a decrease in step length and ISD: these akinetic parameters, which are strongly correlated, are regulated by the dopaminergic basal ganglia loop. Walking cadence and SPIR, which are the rhythmic gait and speech parameters, do not seem to be regulated by a non-dopaminergic structure.

Experimental studies on animals have shown that the PPN of the MLR2 is involved in the regulation of the locomotor rhythm.10 On the basis of our results, we propose an equivalent model for the regulation of the spatial and rhythmic parameters of gait and the temporal and rhythmic parameters of speech (figure 5). Further studies are required to confirm such a hypothesis and determine what structures are involved.

Figure 5

Model for the regulation of the spatial and rhythmic parameters of gait (left) and the temporal and rhythmic parameters of speech (right).

Acknowledgments

We are grateful to all the subjects, who participated in this study, and to J Anderson-Blanc, for revising the English manuscript.

References

View Abstract

Footnotes

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the Marseille 2.

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.