Effects of bilateral pallidal or subthalamic stimulation on gait in advanced Parkinson's disease

Effects of Bilateral Pallidal or Subthalamic Stimulation on Gaitin Advanced Parkinson’s Disease

N. Allert, MD,1* J. Volkmann, MD,1 S. Dotse, MD,1 H. Hefter, MD,1 V. Sturm, MD,2

and H.-J. Freund, MD1

1Department of Neurology, Heinrich-Heine University Düsseldorf, Germany2Department of Stereotaxy and Functional Neurosurgery, University of Cologne, Germany

Abstract: Bilateral high-frequency stimulation of the internalglobus pallidus (GPi) and the subthalamic nucleus (STN) bothalleviate akinesia, rigidity, and tremor in idiopathic Parkinson’sdisease. To test the specific effect of these procedures on gait,we used quantitative gait analysis in addition to relevantsubscores of the Unified Parkinson’s Disease Rating Scale in agroup of 10 patients with advanced Parkinson’s disease treatedby GPi stimulation and eight patients treated by STN stimula-tion. Patients were assessed before and 3 months after surgery.Thirty age-matched healthy subjects served as controls. Thenon-random selection allowed a descriptive but no direct sta-

tistical comparison of the respective procedure. Gait analysisshowed significant stimulation-induced improvements of spa-tiotemporal gait and step parameters in both patient groups.Moreover, the effects on step length and cadence suggested adifferential effect of both basal ganglia targets. Hence, the in-crease in gait velocity in the STN group was almost exclusivelydue to a significant increase in step length, while in the GPigroup statistically non-significant increases in both step lengthand cadence contributed. © 2001 Movement Disorder Society.

Key words: gait; Parkinson’s disease; deep brain stimula-tion; globus pallidus; subthalamic nucleus

In recent years there has been a resurgence of interestin surgical procedures to improve the cardinal motorsymptoms of Parkinson’s disease (i.e., tremor, akinesia,rigidity, and postural instability). High-frequency stimu-lation of the internal globus pallidus (GPi) and the sub-thalamic nucleus (STN) have attracted particular atten-tion, as stimulation via implanted electrodes has the ad-vantage of less permanent side effects, the possibility ofperforming safe bilateral procedures, reversibility, andadaptability during the course of the disease. So far, bothbasal ganglia targets have been shown to be effective inalleviating the Parkinsonian motor symptoms. From theexperience that stimulating the smaller volume of theSTN requires lower energies and that chronic STNstimulation allows a significant reduction in anti-parkinsonian medication, most centers currently preferSTN stimulation over GPi. Still, no randomized clinical

comparative study allows a direct comparison of the ef-fects of both targets on differential motor symptoms.

Gait disorders among other motor deficits are a hall-mark of idiopathic Parkinson’s disease (PD) and are par-ticularly disabling due to progressive deterioration ofmobility, increased risk of falls and loss of patients’ in-dependence. The pathophysiology of parkinsonian gaitdisorders is still poorly understood but impaired controlof brainstem locomotor areas by different descendingconnections from different regions of the basal gangliaappears to be of pivotal importance.1,2 Most of the clini-cal outcome studies of high-frequency stimulation of GPior STN have evaluated the effect on gait as part of clini-cal assessment scores most notably the Unified PD Rat-ing Scale (UPDRS). Bilateral stimulation of STN inthese studies resulted in significant improvements ofclinical gait rating scores in the medicationoff state inthe order of 30%.3,4 High-frequency stimulation of GPi,in contrast, resulted in less consistent results, suggestingdifferential effects depending on the exact site ofstimulation.5–7 Our series of nine patients treated by bi-lateral stimulation of GPi showed significant improve-ments of gait with reduction of a mean clinical

*Correspondence to: Dr. Niels Allert, Department of Neurology,Heinrich-Heine-University, Moorenstrasse 5, D-40225 Düsseldorf,Germany.

Received 6 July 2000; Revised 23 February 2001; Accepted 26February 2001

Published online 10 October 2001.

Movement DisordersVol. 16, No. 6, 2001, pp. 1076–1085© 2001 Movement Disorder SocietyPublished by Wiley-Liss, Inc.DOI 10.1002/mds.1222

1076

gait and posture disability subscore by 57% in the medi-cationoff state.8

The clinical evaluation of parkinsonian gait disorders,however, is difficult since locomotion is impaired bydifferent types of motor deficits representing differentpathophysiological mechanisms. Gait hypokinesia (slow-ness) is the most frequent feature of gait disorder whichusually occurs even in early stages of the disease. It ischaracterized by a reduced gait velocity; small, shufflingsteps with decreased stride length; reduced arm swing;and a bent-forward posture of the trunk. Freezing phe-nomena (start hesitation and blockades) and postural in-stability with pro- and retropulsion represent more com-plex dysfunctions often observed in the later stages of thedisease. Moreover, different compensatory strategiesmay be employed by the patients. For this reason, quan-titative gait analysis has been used to better characterizespecific changes in PD,9–13 as well as to evaluate phar-macological14 and nonpharmacological therapeutical in-terventions quantitatively.15,16

We performed pre- and postoperative gait analysis inaddition to clinical rating in patients with advanced PDselected for surgical stimulating procedures in the GPi orSTN in order to better evaluate the particular effect ofthese treatments on gait. Differences in preoperative pa-tient characteristics only allowed a descriptive but nodirect statistical analysis between GPi and STN stimula-tion. Part of the gait analysis data in patients treated bybilateral GPi stimulation have been published previ-ously.8

METHODS

Subjects

Eighteen patients with advanced bilateral PD and 30age-matched healthy control subjects with no history ofneurological or orthopedic disease were included in thisprospective, non-randomized study. Ten PD patientswere treated by bilateral stimulation of the GPi and eightPD patients were treated by bilateral stimulation of the

STN. All patients suffered from motor fluctuations and/or L-dopa-induced dyskinesias despite optimized drugtreatment.

Between 1996 and 1997, we predominantly treated PDpatients selected for surgery by GPi stimulation.8 Withevidence of a probably better efficacy to relieve parkin-sonian symptoms by bilateral stimulation of the STN,6

the majority of our PD patients eligible for surgery un-derwent the latter procedure after 1997. The decision foreither procedure (GPi or STN stimulation) did not reflectpreselective clinical differences, nor did it follow astrictly random approach. Moreover, since the aim wasto evaluate clinical efficacy of the respective procedurerather than to compare both statistically, gait analysis inthe STN group was performed in a series of eight con-secutive patients whose baseline gait parameters werenot matched with those of the GPi group. The groupcharacteristics are summarized in Table 1. A totalequivalent L-dopa dose was calculated for each patientby using the following conversion factors: 100 mg ofstandard release L-dopa was equivalent to 125 mg ofsustained release L-dopa, 75 mg standard release L-dopaplus catechol-O-methyltransferase (COMT) inhibitors, 1mg pergolide, 4 mg ropinirole, 0.7 mg pramipexole, 1.5mg cabergoline, or 20 mga-dihydroergocryptine. Thestudy protocol was approved by the local ethics commit-tee and all patients gave informed consent.

Procedure and Patient Evaluation

Stereotactic surgery was performed as described pre-viously.8 All PD patients selected for surgery underwenta careful evaluation of motor and non-motor functionsbefore and after surgery using a modified core assess-ment program for intracerebral transplantation (CAPIT)protocol which included the UPDRS, the Hoehn & Yahrscale, and the Schwab and England Activities of DailyLiving. Motor symptoms and disability were assessed bythe same unblinded neurologist (N.A. or J.V.) at baseline(within 1 to 2 weeks before surgery) in the practically

TABLE 1. Demographic patient characteristics

Target structure Globus pallidus Subthalamic nucleus

No. patients 10 8PD duration 11.3 ± 2.6 11.8 ± 5.2Age at surgery (yr) 55.2 ± 9.8 57.4 ± 3.2L-dopa equivalent preop 760.2 ± 326.9 753.4 ± 326.0L-dopa equivalent postop 673.3 ± 320.6 355.0 ± 206.3Hoehn & Yahr preop, medicationoff 3.7 ± 0.5 4.5 ± 0.6Hoehn & Yahr preop, medicationon 3.1 ± 0.6 2.4 ± 0.9Schwab & England preop, medicationoff 53.5 ± 13.0 36.3 ± 18.0Schwab & England preop, medicationon 76.0 ± 14.3 77.5 ± 13.9

PD, Parkinson’s disease.

EFFECTS OF GPi OR STN STIMULATION ON GAIT 1077

Movement Disorders, Vol. 16, No. 6, 2001

defined medicationoff andon state, and 3 months aftersurgery under the following conditions: (1) after over-night drug withdrawal (>12 hours) with stimulationswitchedoff for at least 1 hour, (2) 30 minutes afterturning on stimulation without taking medication, (3) inthe “beston,” 0.5 to 2 hours after intake of either theregular morning dose of antiparkinsonian drugs (in thebeginning of our study) or a 1.5-fold higher “challengingdose” (later in the course of our study) with stimulationswitchedoff for at least 1 hour, and (4) in the “beston”after turningon stimulation for at least 30 minutes. Anincreased “challenging dose” to test the effect of anti-parkinsonian drugs was chosen, since a regular morningdose proved not to induce the best medicationon re-sponse. Unfortunately, this methodological change con-founds the possibility of direct comparisons between GPiand STN patients, since the former were exposed to aregular morning dose while all STN patients were testedwith a “challenging dose.”

In addition to the above clinical rating, quantitativegait analysis was performed in the medicationoff statebefore surgery and 3 months after surgery with and with-out stimulation (see below).

Data Analysis

For the clinical evaluation of gait, we used two sub-scores of the UPDRS, one relating to the patient’s self-assessment of gait-related motor ability and one relatedto the examiner’s assessment. The patient’s self-assessment is provided by part II of the UPDRS (activi-ties of daily living [ADL] items 5 to 17), of which items13 (falling), 14 (freezing), and 15 (walking) relate morespecifically to gait. This subscore is subsequently termed“ADL gait score.” The examiner’s assessment is pro-vided by part III (items 18 to 31) with items 26 (legagility for both legs), 27 (arise from chair), 28 (posture),29 (postural stability), and 30 (gait) being related to gait.This subscore is subsequently termed “gait motor score.”Each item of the UPDRS is scored from 0 to 4 with 0reflecting no impairment and 4 reflecting severe impair-ment. The resultant compound ADL gait and gait motorscore ranged between 0 to 12 and 0 to 24, respectively.

In addition to the above clinical rating, quantitativegait analysis was conducted in patients and controls us-ing a commercially available computer-assisted gait ana-lysing system (Ultraflex system; Infotronic, Tubbergen,The Netherlands). The system consists of a pair of stan-dardized shoes with pressure-sensitive soles connected toa portable data-recording module fixed to the patient’sback. Each sole contains eight pressure sensors evenlydistributed in five rows between heel and toes with one

sensor at the heel and toes, respectively, and pairs ofsensors between. Data were recorded at a frequency of100 Hz (data stored on an IBM-compatible PC). For dataanalysis, Ultraflex Portable Datalogger System version2.05 was used. Since the aim of this study was to evalu-ate the effect of surgery on medicationoff-related par-kinsonian gait, we restricted this quantitative gait analy-sis to the medicationoff state. Hence, patients underwentthe analysis in the medicationoff state at baseline and 3months after surgery, once with stimulationoff and oncewith stimulationon. Patients and control subjects wereasked to walk at a comfortable speed over a straightdistance of 40 m on a hospital corridor of approximately50 m length and 2.5 m width. The recording time wasrestricted to a maximum of 40 seconds.

Analysis of the gait data by means of the applicationsoftware provided the mean values of cycle duration,cadence, symmetry (steptime L/steptime R), single sup-port, double support, stance, and step time with the re-spective standard deviations. As an expression of thedynamic foot–floor contact taking place during gait, theprogression of the resultant normalized force vector(center of force, COF) of either the eight sensors of eachfoot separately (single COF) or the 16 sensors of bothfeet together (total COF) was displayed in an arbitrarilychosen cartesian coordinate system reflecting a planartwo-dimensional (2D) projection of both foot soles dur-ing walking. The resultant line graphs were called“single COF” (separate force vectors for each foot) or“total COF” (single bipedal force vector). In addition tothis spatial distribution of the normalized force vector,we further analyzed the force production at differentstages of the stance period. Heel contact in the beginningof the stance period (“heel on”) and foot take-off with theforefoot at the end of the stance period (“push-off”) bothresult in a characteristic double peak force pattern duringnormal gait (Fig. 3A). The corresponding line graphswere called “stance forces.” Means and standard devia-tions were provided for the forces of both peaks and ofthe minimum between (reflecting midstance). The entiredata set could be visualized on a force graph, allowingthe exclusion of freezing episodes from the analysis(freezing resulted in an interruption of the alternatingstance–swing pattern of both feet and the interval foranalysis could be set manually). Furthermore, the veloc-ity was determined by the time to walk 40 meters or thedistance walked in 40 seconds, respectively. If no freez-ing occurred during the recording, this velocity reflectedgait velocity. In case of freezing, the determined velocityof locomotion was corrected for the freezing periods todetermine gait velocity. The mean step length was calcu-

N. ALLERT ET AL.1078


lated from gait velocity and cadence. For pairwise compari-sons of the UPDRS gait scores at baseline and at 3 months(stimulation on or off), we used the Wilcoxon signed ranktest. For multiple comparisons, a Bonferroni adjustment ofthe level of significance (P < 0.05 orP < 0.01) was used toavoid a type I error. Gait parameters were analyzedwithin groups using analyses of variance (ANOVAs) forrepeated measurements and Scheffe´’s exact test for pair-wise comparisons between measurements at baseline andafter surgery with stimulation switched on or off.

RESULTS

Clinical Sssessment of Gait

Assessment of gait by the patient

The group data of the ADL gait scores are presented inFigure 1 (upper part). At baseline in the medicationoff

state, the GPi group revealed slightly less severe dis-ability subscores than the STN group (5.3 ± 2.8 vs. 6.5± 2.7). L-dopa induced similar reductions of ADL gaitby 47.2% (GPi) and 53.8% (STN), respectively. At 3months after surgery, high-frequency stimulation in themedicationoff state resulted in a significant reduction ofthe baselineoff-score by 58.5% in the GPi group. Therespective reduction in the STN group by 32.7% wassmaller and statistically not significant. Furthermore, inthe GPi group, stimulation in the medicationon staterevealed a significant improvement of the ADL gaitscore compared to the baseline medicationon state. Inthe STN group, in contrast, stimulation in the medica-tion on state was not significantly different from thebaseline medicationon state.

Assessment of gait by an examinerThe group data of the gait motor scores are presented

in Figure 1B. At baseline in the medicationoff state, the

FIG. 1. Improvement of gait related disability subscores of part II (activities of daily living) and part III (motor score) of the Unified Parkinson’sDisease Rating Scale (UPDRS) by internal globus pallidus (GPi) and subthalamic nucleus (STN) stimulation. The 3 months follow-up evaluation iscompared with the baseline results before surgery (*P < 0.05; **P <0.01). The patients’ self-assessment score (activities of daily living [ADL]-gait)revealed significant improvements both with and without medication in the GPi but not in the STN group. The motor examination, in contrast, revealedsignificant improvements of the medicationoff scores in both patient groups while the medicationon score was only reduced in the GPi group. Thecombined medical and stimulation therapy resulted in similar reductions of the baseline disability subscore in both patient groups by 70%.



GPi patients revealed a less severe gait disability motorscore. The reduction of the gait motor score induced byL-dopa was more pronounced in the STN group (64.2%vs. 36.0% in the GPi group). However, the apparentlysmaller L-dopa response in the GPi group has to beweighed against the possibility of a submaximal effectdue to a smaller testing dose in the GPi patients. Aftersurgery, stimulation significantly improved the medica-tion off gait motor score by 44.9% in the GPi group andby 61.6% in the STN group in comparison to baselinevalues. Stimulation in the medicationon state led to sig-nificant improvement of the gait motor score in the GPigroup, but again this may be due to a submaximal effectof the baseline L-dopa testing dose. In the STN group, incontrast, stimulation had no effect on the medicationongait motor scores. Compared with the baseline medica-tion off state, the additive effect of stimulation and medi-cation challenge was similar in both patient groups withreductions of the gait motor scores by 66.9% (GPi group)and 66.4% (STN group), respectively.

Quantitative Assessment of Gait

Gait velocity, cadence, and step length

The effect of surgery and high-frequency stimulationon gait velocity, cadence, step length, and freezing ispresented in Table 2. At baseline in the medicationoffstate, two STN patients were wheelchair-bound and un-able to stand or walk unassisted. Hence, these patientswere excluded from the statistical analysis for the sig-nificance of parameter changes. Compared with age-matched controls, the reduction of gait velocity by 68.3%in the STN group was more prominent than the reductionby 38.3% in the GPi group. This reduction in gait veloc-ity was primarily due to a decrease in step length by69.7% (STN) and 39.4% (GPi), respectively. Cadence

did not reveal significant differences between Parkinsonpatients and controls. After surgery, no significantchange of any of these gait parameters was found ineither group when stimulation was switched off. Withstimulation on, in contrast, gait velocity increased in bothpatient groups by 184.2% (STN) and 50.0% (GPi) group,respectively. In the STN group, this increase in gait ve-locity was due to a significant increase in step length by170.0%, while cadence was unchanged. In the GPigroup, the increase in gait velocity resulted from con-comitant increases in both step length by 40.0% andcadence by 10.0%, which were both statistically not sig-nificant.

In the medicationoff state, freezing episodes duringgait analysis were recorded at baseline in four out ofeight STN patients, and in two out of 10 GPi patients.After surgery, freezing persisted in the stimulation offstate in all four STN- and in one GPi patient. With stimu-lation on, no freezing was recorded in the STN patientsbut still persisted in one GPi patient.

Temporal stride parameters

Steptime, single and double support time were evalu-ated for each foot separately and later pooled. The com-bined data are presented in Table 3. At baseline, steptimein both patient groups was more or less in the same rangeas in controls and showed no significant changes aftersurgery. The variability index of steptime as a measure ofthe variation of steptime from step to step was signifi-cantly increased in both patient groups with higher val-ues for the STN group. After surgery, stimulation sig-nificantly reduced this variability index in the STN pa-tient group and led to a temporally more regular gaitpattern. A slight improvement found in the GPi group did

TABLE 2. Comparison of gait velocity, step length, and cadence in Parkinson patients before and 3 months after bilateralGPi/STN stimulation compared with age-matched controls

Control(n 4 30)

GPi stimulation group(n 4 10)

STN stimulation group(n 4 8)

Preop Postop/stim off Postop/stim on Preop Postop/stim off Postop/stim on

Gait velocity (m/sec) 1.20 ± 0.16 0.74 ± 0.29 1.00 ± 0.32 1.11 ± 0.33 0.38 ± 0.40 0.49 ± 0.40 1.08 ± 0.34NS * NS *

Step length (cm) 66 ± 7 40 ± 14 50 ± 17 56 ± 18 20 ± 22 26 ± 20 54 ± 16NS NS NS *

Cadence (steps/min) 108 ± 10 110 ± 14 122 ± 9 121 ± 11 114 ± 18 110 ± 12 119 ± 6NS NS NS NS

Freezing (no. patients) 2 1 1 4 4 0

All patients data were recorded in the medicationoff state.Statistical analysis: preop compared with postop/stim off and postop/stim on, respectively.*P < 0.05.NS, not significant; GPi, internal globus pallidus; STN, subthalamic nucleus.



not reach significance. In PD patients, the step timeswere often found to be significantly different betweenboth feet, reflecting a characteristic asymmetry of thedisorder. This asymmetry (expressed as the deviationfrom 1 of the ratio of left and right step time) was morepronounced in the STN group (0.11 ± 0.04) than in theGPi group (0.05 ± 0.03). While after surgery no changesof this asymmetry factor were found in the GPi group(0.05 ± 0.09 at stimulation off and 0.07 ± 0.11 at stimu-lation on), high-frequency stimulation in the STN groupresulted in a decrease (from 0.11 ± 0.10 to 0.03 ± 0.03).Another typical feature of the hypokinetic gait is achange in the relation of single and double support times,with decreases in single and increases in double supporttimes. At baseline, theses changes were more prominentin the STN group (decrease of the single/double supportratio to 1.57 ± 0.92) than in the GPi group (2.38 ± 0.75).In controls, this ratio was found to be 3.19 ± 0.44. Aftersurgery, stimulation resulted in increases of this ratio by18% in the GPi group and 81% in the STN group, com-pared with the baseline data. The improvement in theSTN group was due to a 42% decrease in double supporttime and a 24% increase in single support time. Thecorresponding changes in the GPi group (a 15% decreasein double support time and a 5% decrease in single sup-port time) were smaller and statistically not significant.

Foot–floor contactThe foot–floor contact during gait was analyzed by the

progression of force vectors from the hindfoot area to theforefoot area (single COF). Figure 2 presents the groupdata of patients and controls with each diagram depictingthe superimposed means of all subjects in the respectivepatient or control group. Compared with controls, thesingle COF pattern in the medicationoff state was sig-nificantly altered in both patient groups, and again moreprominent in the STN group than in the GPi group. Ingeneral, the changes reflected a shift of the COF to theforefoot area with a reduced anterior–posterior span.Hence, floor contact of the foot occurred at more anteriorareas than in controls (arbitrary units in a cartesian co-ordinate system): −451 ± 418 in GPi; −242 ± 286 in STNcompared with −737 ± 44 in controls (left and right footdata pooled). In contrast, the area of foot take-off showedonly minor differences between Parkinson patients andcontrols. After surgery, stimulation resulted in signifi-cant improvements of this anterior shift at foot landing:in the GPi-group to −671 ± 198 and in the STN group to−682 ±154. Stimulation thereby resulted in an increase ofthe length of the single COF progression by 18% in theGPi group (not significant,P < 0.16) and by 58% in theSTN group (P < 0.0001).

Apart from the spatial distribution of force vectors

TABLE 3. Comparison of temporal step parameters in Parkinson patients before and 3 months after bilateral GPi/STNstimulation compared with age-matched controls

Control(n 4 30)

GPi stimulation group(n 4 10)

STN stimulation group(n 4 8)

Preop Postop/stim off Postop/stim on Preop Postop/stim off Postop/stim on

Steptime (msec) 555 ± 51 556 ± 73 494 ± 35 498 ± 41 535 ± 74 552 ± 57 505 ± 23NS NS NS NS

Var. ind. 3.62 ± 1.05 7.59 ± 4.95 5.23 ± 3.62 5.52 ± 4.29 11.26 ± 7.70 9.48 ± 4.15 4.07 ± 1.19NS NS NS *

Asymmetry index(steptime L/R-1) 0.02 ± 0.01 0.05 ± 0.03 0.05 ± 0.09 0.07 ± 0.11 0.11 ± 0.04 0.11 ± 0.10 0.03 ± 0.03

NS NS NS NSSi. supp. (msec) 421 ± 34 373 ± 69 352 ± 57 354 ± 67 297 ± 82 302 ± 68 367 ± 26

NS NS NS *Var. ind. 3.47 ± 1.07 11.44 ± 11.57 6.92 ± 5.52 6.20 ± 6.56 14.57 ± 6.89 15.34 ± 7.40 5.22 ± 2.24

NS NS NS *Dbl. supp. (msec) 135 ± 24 170 ± 46 142 ± 34 145 ± 66 238 ± 111 250 ± 106 138 ± 37

NS NS NS *Var. ind. 8.88 ± 2.42 10.38 ± 4.54 11.78 ± 5.17 11.21 ± 5.85 16.43 ± 10.47 15.95 ± 7.30 9.85 ± 2.63

NS NS NS NSSi./dbl. support ratio 3.19 ± 0.44 2.38 ± 0.75 2.64 ± 0.82 2.80 ± 0.88 1.57 ± 0.92 1.57 ± 1.01 2.84 ± 0.74

NS NS NS *

All patient data were recorded in the medicationoff state.Statistical analysis: preop compared with postop/stim off and postop/stim on, respectively.*P < 0.05.GPi, internal globus pallidus; STN, subthalamic nucleus; NS, not significant.Var. ind., variability index; si. supp., single support; dbl. suppl., double support.



during the stance period, the dynamic foot–floor contactwas further assessed with regard to the temporal evolu-tion of the force production (stance forces). Normal con-trols showed the characteristic double-peak pattern asdepicted in Figure 3A. PD patients preoperativelyshowed a reduction of the relative differences betweenpeak and midstance forces and, hence, a flattended ap-pearance of the double-peak force production at stance.In fact, in most of the more disabled STN patients, thepattern was reduced to a single peak, reflecting thepathological roll-off of the foot (right foot of the patientpresented in Fig. 3B). Since the push-off with the fore-foot is particularly important for the forward propulsionof the body, we analyzed the peak force at the end ofstance in relation to the midstance force. In control per-sons, the push-off resulted in an increase of the mid-stance force by 18 ± 12%. At baseline, this parameterwas reduced to 10 ± 10% in the GPi group and 2 ± 3 inthe STN group. After surgery, stimulation significantlyincreased the ratio in STN patients to 13 ± 15%. Anincrease to 18 ± 22% in the GPi group was statisticallynot significant (P < 0.253).

The total COF gave further information about the bi-pedal progression of force vectors. The alternating trans-fer of load from the forefoot to the contralateral hindfootarea resulted in a characteristic total COF pattern withthe shape of a butterfly. Figure 4 presents the group dataof patients and controls with each diagram depicting thesuperimposed means of all subjects in the respective pa-tient or control group. Similar to the single COF graphs,PD patients in the medicationoff state demonstrated sig-nificant alterations. Although these alterations were notvery uniform, in general there was a tendency to a re-duced anterior–posterior span of the total COF, reflectinga diminished forward propulsion of the COF at the singlesupport phase of the gait cycle and a forward shift to theforefoot area probably reflecting the bent forward pos-ture. Moreover, in patients prior to surgery, the equilib-rium point of force transition between both feet was of-ten shifted laterally from the midline, reflecting impairedbipedal coordination. In the STN group, the changes

FIG. 3. A: The dynamic foot–floor contact during walking results inpeak forces in the beginning and in the end of the stance period whenthe body weight is first supported by the hindfoot area and then trans-ferred to the forefoot. Weight distribution on the whole foot at mid-stance results in the reduced force between both peaks.B: Example ofrecordings from a Parkinson patient treated by bilateral STN stimula-tion. At baseline, the normal double-peak force pattern of the foot–floorcontact is significantly reduced on both sides, more prominent on themore affected right side. Stimulation results in a significantly improvedand more symmetrical pattern. Correspondingly, gait velocity in thispatient increased from 0.8 to 1.4 m/sec and step length from 0.44 to0.76 meter.

FIG. 2. The progression of the center of force (COF) on each footduring walking (single COF). The mean single COF of each patient andcontrol is superimposed for comparison of the stimulation effect. Allpatient data were recorded in the medicationoff state.



were more prominent than in the GPi group. In bothpatient groups (again more pronounced in the STNgroup) high-frequency stimulation led to a restoration ofthe symmetrical, “butterfly-like” total COF found in nor-mal controls with an increase of the anterior–posteriorspan by 63% in GPi and 156% in STN and a reduction/normalization of the anterior shift of the total COF.

DISCUSSIONThe results of this study clearly demonstrate signifi-

cant improvements of temporal and spatial gait param-eters of Parkinson patients in the medicationoff stateboth by stimulating STN and GPi. The statistical analysissuggests more prominent improvements when stimulat-ing STN; however, this has to be weighed against dif-ferences in patient characteristics. At baseline, gait inpatients treated by STN stimulation was more severelyaffected in the medicationoff state and revealed higherdisability scores in the patients’ ADL-gait scores, theexaminer’s gait motor score and the gait analysis param-eter. Furthermore, the improvement induced by L-dopa

appeared to be more prominent in the STN group than inthe GPi group, although a change in the L-dopa test inboth patient groups may have contributed to this differ-ence. However, although the relative improvement ofclinical scores and gait parameters induced by stimula-tion was more significant in the STN group, the absolutevalues of most gait parameters in the medicationoff statewith stimulationon was more or less comparable in bothpatient groups. In fact, these values were comparablewith values of normal controls suggesting a ceiling effectof possible improvements.

Previous studies have used gait analysis in parkinso-nian patients and healthy elderly controls in an attempt tocharacterize parkinsonian features, thereby gaining in-sight into the pathophysiology of gait and its relationshipto other cardinal motor impairments such as rigidity andbradykinesia.9,11–14 The observation of a relative in-crease in cadence and a relative decrease in step lengthwith regard to the achieved gait velocity has led to thehypothesis that an impairment in the regulation of stridelength is the fundamental deficit in PD, while the in-crease in cadence is a compensatory mechanism.12,17,18

In accordance with this view, the reduction in gait ve-locity observed in our study was primarily due to a re-duction in step length by 39% in the GPi group and 70%in the STN group compared with controls. Cadence, andconsequently steptime, were not significantly differentfrom controls. High-frequency stimulation significantlyimproved step length by 170% in STN patients, suggest-ing a direct influence on the underlying pathophysiologi-cal deficit while the improvement in GPi patients wassmaller (increase by 40%) and statistically not signifi-cant, although again the baseline characteristics of thetwo groups were very different and the final step lengthvalues were essentially identical. This difference maysimply be due to a ceiling effect in the Gpi group. How-ever, the relative contribution of cadence to the increasein gait velocity appeared to be more important in the GPigroup, possibly reflecting a differential effect with amore prominent role of compensatory gait strategies.

The effect of L-dopa on spatiotemporal gait param-eters was investigated by Blin and colleagues.14 In thisstudy, stride length, interpreted as a parameter of energy,also significantly improved and thus was found to beL-dopa-sensitive. Temporal parameters such as strideand swing duration and stride duration variability, incontrast, did not change significantly and were thereforeconsidered to be L-dopa-resistant. Interestingly, wefound the effect of bilateral stimulation of STN not onlyresulted in a significant increase in step length but also ina significant increase in single swing time and a signifi-cant reduction of the variability of step time. However,

FIG. 4. The progression of the COF between both feet during walking(total COF). The mean total COF of each patient and control is super-imposed for comparison of the stimulation effect. All patient data wererecorded in the medicationoff state.



whether the improvement of “L-dopa-resistant” step pa-rameters reflects a non-dopaminergic mechanism in-volved in stimulating STN or whether it is due to themore advanced stage of the disease of those patientsincluded in our study remains speculative. Gait analysisafter intake of L-dopa was not included in this studybecause all patients suffered from severe L-dopa-induceddyskinesias which would not allow the separatation ofparkinsonian changes from dyskinetic changes in gaitparameters.

In contrast to the gait analysis data, the patients’ self-assessment also showed significant improvements of gaitin both patient groups, but surprisingly suggested a betterefficacy in the GPi group with a reduction of the ADLgait subscore by 59% as opposed to 33% in the STNgroup. The reason for this differential patients’ self-estimation of gait is unclear. One hypothesis could bethat the higher disability subscore of ADL gait in theSTN group might result from a different gait patterninduced by stimulating STN as opposed to GPi. STNstimulation resulted in increased single support and re-duced double support times as well as important in-creases in step length. With regard to the patient’s sta-bility this “strategy” appears to be more at risk than thesuggested “strategy” of GPi patients with proportionallylonger bipedal floor contact and, instead, smaller butconcomitant increases in cadence/decreases in steptimeto increase overall gait velocity. The danger of thesedifferent gait strategies might become obvious in morerealistic daily life situations than the gait analysis task(i.e., when walking has to entail changes in direction, thepassage of obstacles and the possible need for immediatechanges in gait velocity). Whether differences in affec-tive and emotional processes influence the self-estimation of motor symptoms awaits further investiga-tion. In our experience, a considerable number of STNpatients after surgery show a syndrome of abulia mani-festating with fatigue and loss of initiative. Similar af-fective changes induced by STN stimulation have beenreported by Krack and colleagues6 and such changesmight contribute to the discrepancy between the patients’own awareness of gait related motor ability and the quan-titative gait analysis.

The gait analysis used in our study primarily concen-trates on the evaluation of the hypokinetic type of par-kinsonian gait disturbances and, consequently, the ob-served improvements induced by stimulation mainly re-flect influences on gait hypokinesia. Posture was alsofound to improve significantly in both patient groupswhen assessed clinically by part III of the UPDRS or asdeduced from the gait analysis which showed an im-provement of the anterior shift of the COF. However,

freezing and postural instability with increased risk offalls are particularly important disturbances which arenot sufficiently evaluated by the present data. Both as-pects of gait disorders are difficult to assess. Freezingoften is time- and location-dependent and can occur as amedicationoff- as well ason symptom.19 The patientsincluded in this study primarily suffered from differentdegrees ofoff-freezing which improved after L-dopa in-take. In both patient groups, freezing appeared to beimproved by stimulation, although the effect could not bereliably quantified. Similarly, postural stability as as-sessed by the pull-test within the UPDRS was found toimprove significantly in both patient groups but the func-tional relevance of this test to assess postural stability indaily living circumstances remains uncertain and war-rants further studies.

Acknowledgments:This study was supported by the DeutscheForschungsgemeinschaft (SFB 194 A 17).

REFERENCES

1. Dietz V. Neurophysiology of gait disorders: present and futureapplications. [Review]. Electroencephalogr Clin Neurophysiol1997;103:333–355.

2. Nutt JG, Marsden CD, Thompson PD. Human walking and higher-level gait disorders, particularly in the elderly [see comments].[Review]. Neurology 1993;43:268–279. Comment in: Neurology1994;44:779–780.

3. Limousin P, Pollak P, Benazzouz A, Hoffmann D, Le Bas J-F,Broussolle E, Perret JE, Benabid A-L. Effect on parkinsonian signsand symptoms of bilateral subthalamic nucleus stimulation. Lancet1995;345:91–95.

4. Limousin P, Krack P, Pollak P, Benazzouz A, Ardouin C, Hoff-mann D, Benabid AL. Electrical stimulation of the subthalamicnucleus in advanced Parkinson’s Disease. N Engl J Med 1998;339:1105–1111.

5. Bejjani BP, Damier P, Amulf I, Papadopoulos S, Bonnet AM,Vidailhet M, Agid Y, Pidoux B, Cornu P, Dormont D, Marsault C.Deep brain stimulation in Parkinson’s disease: opposite effects ofstimulation in the pallidum. Mov Disord 1998;13:969–970.

6. Krack P, Pollak P, Limousin P, Hoffmann D, Xie J, Benazzouz A,Benabid AL. Subthalamic nucleus or internal pallidal stimulationin young onset parkinson’s disease. Brain 1998;121:451–457.

7. Tronnier VM, Fogel W, Kronenbuerger M, Krause M, SteinvorthS. Is the medial globus pallidus a site for stimulation or lesioningin the treatment of Parkinson’s disease? Stereotact Funct Neuro-surg 1997;69:62–68.

8. Volkmann J, Sturm V, Weiss P, Kappler J, Voges J, KoulousakisA, Lehrke R, Hefter H, Freund H-J. Bilateral high-frequencystimulation of the internal globus pallidus in advanced Parkinson’sdisease. Ann Neurol 1998;44:953–961.

9. Blin O, Ferrandez AM, Serratrice G. Quantitative analysis of gaitin Parkinson patients: increased variability of stride length. J Neu-rol Sci 1990;98:91–97.

10. Ferrandez AM, Blin O. A comparison between the effect of inten-tional modulations and the action of L-dopa on gait in Parkinson’sdisease. Behav Brain Res 1991;45:177–183.

11. Knutsson E. An analysis of parkinsonian gait. Brain 1972;95:475–486.

12. Morris ME, Iansek R, Matyas TA, Summers JJ. The pathogenesisof gait hypokinesia in Parkinson’s disease. Brain 1994;117:1169–1181.



13. Murray MP, Sepic SB, Gardner GM, Downs WJ. Walking patternsof men with parkinsonism. Am J Phys Med 1978;57:278–294.

14. Blin O, Ferrandez AM, Pailhous J, Serratrice G. Dopa-sensitiveand Dopa-resistant gait parameters in Parkinson’s disease. J NeurolSci 1991;10:351–354.

15. Nieuwboer A, De Weerdt W, Dom R, Nuttin B, Peeraer L, PattynA. Walking ability after implantation of a pallidal stimulator:analysis of plantar force distribution in patients with parkinson’sdisease. Parkinsonism Related Disord 1998;4:189–199.

16. Thaut MH, McIntosch GC, Tice RR, Miller RA, Rathbun J, Brault

JM. Rhythmic auditory stimulation in gait training for Parkinson’sdisease patients. Mov Disord 1996;11:193–200.

17. Ebersbach G, Sojer M, Valldeoriola F, Wissel J, Müller J, TolosaE, Poewe W. Comparative analysis of gait in Parkinson’s disease,cerebellar ataxia and subcortical arteriosclerotic encephalopathy.Brain 1999;122:1349–1355.

18. Morris M, Iansek R, Matyas T, Summers J. Abnormalities in stridelength-cadence relation in parkinsonian gait. Mov Disord 1998;13:61–69.

19. Giladi N, Kao R, Fahn S. Freezing phenomenon in patients withparkinsonian syndromes. Mov Disord 1997;12:302–305.



Effects of bilateral pallidal or subthalamic stimulation on gait in advanced Parkinson's disease

Documents