Experimental Brain Research

, Volume 184, Issue 4, pp 469–478

Locomotor response to levodopa in fluctuating Parkinson’s disease

Authors

    • Department of NeurologyMount Sinai School of Medicine
  • Hamish G. MacDougall
    • Department of NeurologyMount Sinai School of Medicine
    • School of PsychologyUniversity of Sydney
  • Jean-Michel Gracies
    • Department of NeurologyMount Sinai School of Medicine
  • William G. Ondo
    • Department of NeurologyBaylor College of Medicine
Research Article

DOI: 10.1007/s00221-007-1113-y

Cite this article as:
Moore, S.T., MacDougall, H.G., Gracies, J. et al. Exp Brain Res (2008) 184: 469. doi:10.1007/s00221-007-1113-y

Abstract

The aim of this study was to quantify the dynamic response of locomotion to the first oral levodopa administration of the day in patients with fluctuating Parkinson’s disease (PD). Stride length, walking speed, cadence and gait variability were measured with an ambulatory gait monitor in 13 PD patients (8 males) with a clinical history of motor fluctuations. The Unified Parkinson’s Disease Rating Scale (UPDRS) gait score (part 29) was also determined by a movement disorders specialist from video recordings. Subjects arrived in the morning in an ‘off’ state (no PD medication) and walked for a maximum length of 100 m. They then took their usual morning dose of oral levodopa and repeated the walking task at 13 min intervals (on average) over a 90 min period. Changes in stride length over time were fit with a Hill (Emax) function. Latency (time until stride length increased 15% of the difference between baseline and maximum response) and the Hill coefficient (shape of the ‘off–on’ transition) were determined from the fitted curve. Latency varied from 4.7 to 53.3 min post-administration [23.31 min (SD 14.9)], and was inversely correlated with age at onset of PD (R = −0.83; P = 0.0004). The Hill coefficient (H) ranged from a smooth hyperbolic curve (0.9) to an abrupt ‘off–on’ transition (16.9), with a mean of 8.1 (SD 4.9). H correlated with disease duration (R = 0.67; P = 0.01) and latency (R = 0.67; P = 0.01), and increased with Hoehn & Yahr stage in the ‘off’ state (P = 0.02) from 5.7 (SD 3.5) (H&Y III) to 11.9 (SD 4.7) (H&Y IV). Walking speed correlated with changes in mean stride length, whereas cadence and gait variability did not. UPDRS gait score also reflected improving gait in the majority of subjects (8), providing clinical confirmation of the objective measures of the locomotor response to levodopa. Increasing abruptness (H) of the ‘off–on’ transition with disease duration is consistent with results from finger-tapping studies, and may reflect reduced buffering capacity of pre-synaptic nigrostriatal dopaminergic neurons. Ambulatory monitoring of gait objectively measures the dynamic locomotor response to levodopa, and this information could be used to improve daily management of motor fluctuations.

Keywords

Stride lengthWalking speedCadenceGait variabilityYoung onset

Introduction

Parkinson’s disease (PD) results from a progressive loss of dopaminergic and other sub-cortical neurons (Braak et al. 2004). One of the cardinal features of PD is locomotor dysfunction; shortened stride length, increased variability of stride, reduced walking speed (Hausdorff et al. 1998, 2003a; Schaafsma et al. 2003), and freezing (Bloem et al. 2004). Levodopa, the metabolic precursor to dopamine, is commonly used to manage the motor symptoms of PD by replacing endogenous dopamine at the striatum. Although initially effective, almost 60% of patients have developed motor fluctuations after 3 years of treatment (McColl et al. 2002). These include the ‘off–on’ phenomenon (abrupt and unpredictable responses to individual doses of levodopa, Nutt 1987) and ‘wearing off’ (declining dose duration). In particular, fluctuators have been noted to have a delayed and inconsistent response to the first levodopa dose of the day (Chana et al. 2004). Levodopa-related fluctuations in locomotor function can significantly limit mobility and complicate management of PD (Ondo 2003).

Many laboratory and clinical studies have demonstrated the ability of levodopa to increase stride length and walking speed, typically evaluating gait over short distances (<10 m) prior to administration (‘off’) and at the peak dose effect (‘on’) of the levodopa medication cycle (see Morris et al. 2001). However, the temporal dynamics of the locomotor response to levodopa (i.e., how gait transitions from the ‘off’ to ‘on’ state), necessary for objective assessment of fluctuations, has not been successfully elucidated. The only published attempt (MacKay-Lyons 1998) periodically assessed gait on a fixed 7-m walkway over the levodopa cycle but found no consistent changes in stride length. This may be due to the contrived nature of the laboratory walking task, which can temporarily enhance performance in PD patients (Yekutiel 1993), and the limited number of strides available for analysis. The Unified Parkinson’s Disease Rating Scale (UPDRS) (Fahn et al. 1987), arguably the most widely utilized measure in research studies (Mitchell et al. 2000), can differentiate between ‘off’ and ‘on’ states, but is of limited value for assessing motor fluctuations (Shannon 2004).

The fine-motor dynamic response to a standardized dose of levodopa (100 mg) has been assessed using finger-tapping at 15-min intervals post-administration (Contin et al. 1996, 2001). The drug effect–time curve (taps/min vs. time-since-administration) followed a hyperbolic profile in early stages of the disease, with a sharper sigmoidal ‘off–on’ transition in advanced PD. However, it is unclear whether finger-tapping accurately reflects the locomotor response to levodopa; no correlation was found between the ‘off–on’ variation in tapping and gait (Vokaer et al. 2003), suggesting dissociation in the effects of levodopa on upper limb movement and locomotor function.

The goals of the current study were twofold: (i) to determine whether a new ambulatory system for monitoring of gait, which has proven sensitive to stride length changes in response to levodopa (Moore et al. 2007), could be used to objectively assess locomotor fluctuations in response to the first levodopa dose of the day; and (ii) whether the abruptness of the dynamic locomotor response from ‘off’ to ‘on’, modeled using a Hill function, was related to disease duration, as observed in previous finger-tapping studies (Contin et al. 1996, 2001).

Methods

Thirteen participants (8 males) diagnosed with idiopathic PD (UK PD Society Brain Bank diagnostic criteria) were recruited by a movement disorders specialist at the Parkinson’s Disease Center and Movement Disorders Clinic (PDCMDC) at Baylor College of Medicine (Table 1). All patients had a clinical history of motor fluctuations (delayed ‘on’, sharp ‘off–on’ transitions, ‘wearing off’ and/or freezing), and no known non-dopaminergic lesions or cognitive impairment (based on clinical assessment). Age ranged from 45 to 79 years [62.3 (SD 9.8)], height from 155 to 193 cm [173.8 (11.6)], age-at-onset of PD from 24 to 69 years [45.7 (SD 13.1)], time-since-onset from 6 to 28 years [17.2 (SD 7.5)], and Hoehn and Yahr (H&Y) stage III–IV (‘off’ state). All participants were taking an oral levodopa/carbidopa combination, with the morning levodopa dose ranging from 100 to 450 mg [165 mg (SD 103)] and total daily dose from 400 to 2,350 mg [1,074 mg (SD 654)]. The study was approved by the Institutional Review Boards at the Mount Sinai School of Medicine and Baylor College of Medicine and Affiliated Hospitals, and was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki. Participants gave informed consent prior to their inclusion in the study.
Table 1

Patient characteristics

Subject ID

Age (years)

Onset (years)

Time since Dx (years)

LD (morning) mg

LD (total) mg

H&Y ‘off’

H&Y ‘on’

Adjunct PD meds

3

66

40

26

100

565

IV

III

Pramipexole

6

65

42

23

100

800

III

II

Tolcapone

7

45

24

21

100

600

IV

II

Tolcapone

9

65

37

28

100

1,250

III

III

10

65

54

11

250

1,900

III

II

Pramipexole

11

72

50

22

450

2,350

IV

III

Pramipexole

12

68

52

16

150

1,750

IV

III

Pramipexole

13

66

60

6

150

600

III

II

Entacapone, Pramipexole

14

51

42

9

250

1,250

III

II

Entacapone, Ropinirole

15

60

35

25

200

1,600

III

II

Ropinirole, Deprenyl, Amantadine

16

79

69

10

100

400

III

II

Pramipexole

17

47

30

17

100

500

IV

III

Pramipexole

18

69

60

9

100

400

III

II

Ropinirole

Patients arrived 8:00 a.m. at the PDCMDC without having taken their usual morning PD medications (time since previous dopaminergic medication administration was at least 12 h), in an ‘off’ state. Participants walked without assistance at a self-determined pace within an internal corridor (Fig. 1a). Maximum distance traversed was 100 m, less if patients indicated that they were no longer able (or desired) to walk at any point. The visual environment was typical of a clinic, with windowless beige unadorned walls and a featureless grey/brown carpet. The most compelling visual cue was likely the doorway through which patients entered and exited the corridor from a conference room; a number of closed doorways were also distributed along the hallway (Fig. 1a). Immediately following the initial walking trial patients took their usual morning dose of PD medications (Table 1), then periodically repeated the walking task over a 90 min epoch post-administration. Although an attempt was made to regulate the interval between locomotor activity it varied considerably from 5 to 30 min [13.2 (SD 4.7)], primarily due to difficulty walking in the ‘off’ state (in particular for subject’s 7 and 17—see Fig. 3). Distance walked was dependent on patient ability, ranging from 1 to 90 m [51 m (SD 27)] when ‘off’ and from 30 to 94 m [60 m (SD 21)] in the ‘on’ state. Data collection was terminated at 90 min or earlier if the ‘on’ state had been reached over at least two periods of locomotor activity.
https://static-content.springer.com/image/art%3A10.1007%2Fs00221-007-1113-y/MediaObjects/221_2007_1113_Fig1_HTML.gif
Fig. 1

a Floor plan of the area of the Parkinson’s Disease Center and Movement Disorders Clinic (PDCMDC) at Baylor College of Medicine used in the current study. Participants were instrumented in a conference room, and periodically walked along the adjacent hallway. The red (subject 7) and blue (subject 13) traces indicate the approximate path taken by two subjects prior to levodopa administration (determined from video recordings). The grey shaded area indicates the region where walking speed, gait variability and step frequency were calculated in the steady state over a period of 12 s. b Stride data from S7 in the ‘off’ state corresponding to the red path in a. Stride length was small and variable, and the subject experienced a freezing episode while turning (point 4). c Stride data from S13 in the ‘off’ state, corresponding to the blue path in a

‘Off’ and ‘on’ states were determined from subject feedback and a brief clinical evaluation by a neurologist familiar with the patients; all subjects were clinically ‘off’ prior to levodopa administration and had reached an ‘on’ state by the end of testing. All walking trials were recorded on a digital video camera and assessed after testing by a movement disorder specialist without prior knowledge of the stride length data or the relative timing of the video records with respect to levodopa administration. The UPDRS part 29 (gait) (Fahn et al. 1987; Shannon 2004) was determined using an 9 point scale at intervals of 0.5 (range from 0 to 4; 0—normal gait; 1—walks slowly, may shuffle with short steps, no festination or propulsion; 2—walks with difficulty, but requires little or no assistance, may have festination, short steps or propulsion; 3—severe disturbance of gait, requiring assistance; 4—cannot walk at all, even with assistance). Episodes of freezing of gait were also identified from the video records.

During the walking task the length of every stride of the left leg was measured using a novel ambulatory device developed by the authors (Moore et al. 2007). An Inertial Measurement Unit, 9 V battery and Bluetooth serial transmitter were mounted around the left shank (just above the ankle) using an elasticized strap and Velcro. The stride monitor was unobtrusive (weighing less than 130 g) and did not interfere with locomotion; a critical point as distractions significantly increase gait variability in PD (Hausdorff et al. 2003b). Vertical linear acceleration and pitch angular velocity of the left leg were transmitted wirelessly at a rate of 100 Hz to a Pocket PC carried by an investigator, who typically walked 5 m behind the participant. After each walking trial, leg movement data were processed using custom analysis software to calculate the length of every stride taken (Fig. 1b). Accuracy of stride length measurement was 5 cm, although the sensitivity of the device to small changes in stride was considerably higher (∼1 cm) (Moore et al. 2007).

Additional gait parameters were derived in the steady state from stride measures over 12 s as subjects walked along a straight section of hallway (Fig. 1a, shaded area), with a buffer of at least three strides prior to and following this epoch. Walking speed was calculated from the distance traversed over 12 s, gait variability from the SD of stride length (expressed as a percentage of mean stride length over the epoch), and step frequency from a power spectral analysis of vertical shank acceleration over 12 s (MacDougall and Moore 2005). One subject (S12) could not traverse the shaded area of the hallway (Fig. 1a) in the ‘off’ state and 16 min after levodopa administration (see Fig. 2c); walking speed, gait variability and step frequency for these trials were estimated from the available data.
https://static-content.springer.com/image/art%3A10.1007%2Fs00221-007-1113-y/MediaObjects/221_2007_1113_Fig2_HTML.gif
Fig. 2

Representative locomotor response (60 s of stride length data from the left leg) to levodopa administration (at LD + 0 min) in two patients. a Subject 13, with disease duration of 6 years, H&Y III in the ‘off’ state. The shaded area represents mean stride length for the first walking trial (LD + 0). b Stride data from S13, fitted with a Hill function, exhibited a smooth hyperbolic transition (H = 0.9) from ‘off’ to ‘on’. The shaded area represents latency (7.3 min). c A subject with advanced PD (S12, 16 years since onset, H&Y IV ‘off’). The shaded area represents mean stride length for the first walking trial (LD + 0). d Stride data from S13, fitted with a Hill function, displayed a sigmoidal response with a sharp ‘off–on’ transition (H = 4.4). The shaded area represents latency (17.9 min)

A drug effect–time curve for the locomotor response to levodopa was determined by fitting all stride length values over the 90-min period for each participant [total number of strides per subject ranged from 228 to 682; mean 416 (SD 114)] with a sigmoid Hill (Emax) function (Holford and Sheiner 1981), which has previously been used to model the dynamics of the finger-tapping response to levodopa (Contin et al. 1996, 2001). Goodness of fit of models to experimental data was assessed by the correlation coefficient between observed and computed values (Contin et al. 2001). The latency of the locomotor response to levodopa was calculated from the fitted Hill curve for each subject as the time from oral administration until stride length increased 15% of the difference between baseline and maximum response (Contin et al. 2001). The Hill coefficient, H, a dimensionless value which describes the shape of the fitted stride–time curve, was also determined (Holford and Sheiner 1981); H has previously been used to characterize the finger-tapping transition from ‘off’ to ‘on’ (Contin et al. 1996, 2001). A regression analysis was performed to determine the relationship of latency and the Hill coefficient to patient age, age-at-onset of PD, time-since-onset of PD, and the morning and total daily levodopa dose. Correlations were considered significant for P < 0.05.

Results

Data from two subjects illustrate the stride monitoring technique and the range of dynamic locomotor responses to levodopa (Fig. 2). A participant with a relatively recent diagnosis of PD (S13; 6 years post-onset, H&Y III when ‘off’) exhibited a hyperbolic response (Fig. 2a, b) with stride length increasing smoothly (H = 0.9) from a baseline of 0.73 m prior to levodopa (LD) administration (LD + 0 min) to 0.99 m 69 min post-administration. In contrast, a patient at a more advanced stage of PD (S12; 16 years post-onset, H&Y IV ‘off’) was essentially restricted to using a wheelchair in the ‘off’ state, managing a total of less than 5 m with small shuffling strides (<0.3 m) immediately prior to and 16 min after levodopa administration (Fig. 2c). This subject exhibited a sigmoidal response to levodopa, with a sharp transition (H = 4.4) from ‘off’ to ‘on’ with stride length stabilizing at 0.68 m by LD + 40 min (Fig. 2d).

Stride data from all 13 participants were well fit by a Hill (Emax) function (Fig. 3), with a strong correlation between observed and computed values (P < 0.001). All subjects exhibited a significant increase (P = 0.003) in stride length in response to levodopa, from an average of 0.49 m (SD 0.22) in the ‘off’ state prior to administration (first data point) to 0.83 m (SD 0.28) when ‘on’ (final data point). Walking speed also increased in response to levodopa [P = 0.0006; ‘off’: 0.56 m/s (SD 0.23); ‘on’: 0.87 m/s (SD 0.15)]; but step frequency [P = 0.9; ‘off’: 2.0 Hz (SD 0.7); ‘on’: 2.0 Hz (SD 0.43)] and gait variability [P = 0.5; ‘off’: 0.7% (SD 0.8); ‘on’: 1.1% (SD 1.4)] did not vary significantly. Walking speed closely followed changes in stride in response to levodopa, with a strong correlation between speed (Fig. 3, blue squares) and mean stride length (Fig. 3, red circles) [R = 0.89 (SD 0.06); P = 0.026 (SD 0.025); mean of individual correlations from 13 subjects]. There was no evidence of a correlation between mean stride length and step frequency [R = 0.64 (SD 0.29); P = 0.25 (SD 0.28)] or gait variability [R = 0.47 (SD 0.15); P = 0.39 (SD 0.19)] (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00221-007-1113-y/MediaObjects/221_2007_1113_Fig3_HTML.gif
Fig. 3

Locomotor response to levodopa in all 13 subjects. Stride length (SL) data were fit with a Hill (Emax) function (blue trace). Latency to drug effect (green shaded region) and the Hill coefficient (shape of the ‘off–on’ transition) were determined from the Hill curve. Red circles mean stride length (m); blue squares walking speed (m/s); F indicates one or more episodes of freezing occurred. Walking speed reflected changes in stride length, with a strong correlation between speed and mean SL

https://static-content.springer.com/image/art%3A10.1007%2Fs00221-007-1113-y/MediaObjects/221_2007_1113_Fig4_HTML.gif
Fig. 4

Step frequency (magenta circles), gait variability (orange squares) and UPDRS gait score (green diamonds) over the test epoch in all 13 subjects. Note that the scale of the UPDRS gait score (in green to the left of S9’s plot) has been inverted such that an improvement in gait (i.e., a reduction in gait score) exhibits an upward trend to match the increase in stride length. The Hill fit to stride length (blue trace, determined from the stride data in Fig. 3) is also shown. Step frequency and gait variability were unrelated to changes in mean stride length, whereas UPDRS gait score correlated with mean SL in 8 of the 13 subjects (indicated by asterisk next to subject ID)

The latency to increased stride length varied from 4.7 to 53.3 min post-administration [23.3 min (SD 14.9)] (Fig. 3, green shaded regions). Subjects experiencing episodes of freezing (either when initiating gait or turning—see Figs. 1b, 3) had significantly longer latency [P = 0.02; 30.4 min (SD 13.5); N = 8] than non-freezers [11.8 min (SD 8.7); N = 5]. Latency to drug effect was strongly related to age-at-onset of PD (P = 0.0004); the younger the onset the longer the latency (Fig. 5a). Not surprisingly, latency also correlated with age (although an order of magnitude less significant at P = 0.002), as patients with young onset tended to be the younger subjects, and vice versa (Table 2). Adjusting for age and age-at-onset as independent predictors with a multivariable regression analysis revealed that a younger age-at-onset reached the threshold of significance as a sole predictor of latency (P = 0.05); age alone did not (P = 0.46).
https://static-content.springer.com/image/art%3A10.1007%2Fs00221-007-1113-y/MediaObjects/221_2007_1113_Fig5_HTML.gif
Fig. 5

a Latency to increased stride length was inversely correlated with age at onset of PD. b The Hill coefficient was positively correlated with disease duration, c latency

Table 2

Regression analyses for the Hill coefficient and latency to drug effect

 

Hill coefficient

Latency

R

P

R

P

Age

−0.20

0.51

−0.76

0.002b

Age at onset of PD

−0.53

0.06

−0.83

0.0004b

Time since Dx

0.67

0.01a

0.46

0.11

LD (morning dose)

0.31

0.30

0.12

0.70

LD (total daily dose)

0.09

0.80

0.19

0.51

Latency

0.67

0.01a

X

X

The ‘X’ means ‘not applicable’ (i.e., latency cannot be correlated with itself)

LD levodopa, SL stride length

aHill coefficient was correlated with disease duration and latency

bLatency was correlated with age and age-at-onset of PD (although a multivariable analysis accounting for age revealed that age-at-onset alone was a predictor of latency)

The dynamic locomotor response to levodopa ranged from a smooth hyperbolic curve (H = 0.9) to an abrupt ‘off–on’ transition (H = 16.9), with a mean of 8.1 (SD 4.9). The Hill coefficient (Table 2) was positively correlated with time-since-onset of PD (P = 0.01) and latency (P = 0.01) (Fig. 5b, c). Moreover, H was significantly larger (ANOVA; P = 0.02) in patients at H&Y stage IV in the ‘off’ state [11.9 (SD 4.7)] than those staged at H&Y III [5.7 (SD 3.5)]. The Hill coefficient and latency showed no evidence of a relationship to the levodopa dose (morning or total). Evaluation of the patient’s motor state using UPDRS part 29 (gait) provided some clinical confirmation of the locomotor transitioning from ‘off’ to ‘on’ (Fig. 4, green diamonds). UPDRS ‘gait’ score correlated with mean stride length (Fig. 4, asterisk next to subject ID) in 8 of the 13 subjects [R = −0.89 (SD 0.09), P = 0.02 (SD 0.02); mean of 8 individual correlations].

Discussion

The results of this study reveal for the first time the temporal dynamics of the stride length response to oral levodopa in patients with advanced PD. Consistent with previous studies of the dynamic response of finger-tapping to levodopa (Contin et al. 1996, 2001), the locomotor transition from ‘off’ to ‘on’, represented here by the Hill coefficient, was positively correlated with disease duration and increased with H&Y stage in the ‘off’ state. The magnitude of the locomotor Hill coefficient (expressed as median and range to allow comparison) was similar to that previously observed in finger-tapping studies (Contin et al. 2001) in H&Y stage III (locomotor median 5.2, range 0.9–11.3; finger-tapping median 7, range 2–29) and IV patients (locomotor 12, 4.4–16.9; finger-tapping 18, 4–49). There was also some evidence that a simple clinical measure of gait (UPDRS part 29) reflected the locomotor transition, correlating with changes in mean stride length in the majority of subjects. Recent PET studies have shown that patients with ‘wearing off’ fluctuations exhibit a threefold increase in (estimated) synaptic dopamine levels 1 h after levodopa administration relative to stable-responders (de la Fuente-Fernandez et al. 2001; Sossi et al. 2004). Normally part of an exogenous dopamine load is taken up via the dopamine transporter receptor into the pre-synaptic dopaminergic neurons and released physiologically, thus ‘buffering’ post-synaptic receptor dopamine stimulation. The increase in dopamine receptor stimulation in fluctuators is postulated to result from reduced buffering capacity as the dopamine-producing cells die off. This may underlie the increasing abruptness of the ‘off’ to ‘on’ transition with disease duration observed in both the current locomotor and previous finger-tapping studies (Contin et al. 1996, 2001).

Mean stride length in ‘off’ (0.49 m) and ‘on’ (0.83 m) states were consistent with previously reported values from laboratory studies of 0.4–0.9 m (‘off’) (Siegel and Metman 2000) and 0.8–1.0 m (‘on’) (Morris et al. 1996), and significantly less than healthy older adults (1.2–1.5 m) (Kerrigan et al. 1998; Ostrosky et al. 1994). Walking speed mirrored the changes in mean stride length in response to levodopa, and in the ‘on’ state (0.87 m/s) was consistent with previously reported values for PD patients of 0.67–1.0 m/s (Morris et al. 1996; O’Sullivan et al. 1998), but again less than age-matched healthy adults (1.25–1.5 m/s) (Kerrigan et al. 1998; Ostrosky et al. 1994). Mean step frequency (cadence) was within the healthy adult range at 2 Hz (MacDougall and Moore 2005) and, as previously reported by O’Sullivan et al. (1998), did not change in response to levodopa administration. However, high-frequency (3.2 Hz) ‘shuffling’ (Schaafsma et al. 2003) was observed in one subject (S17) in the ‘off’ state (Fig. 4). There were no consistent changes in gait variability in response to levodopa. These results support previous findings that stride length and walking speed are ‘dopa-sensitive’ (Blin et al. 1991), whereas temporal characteristics of gait (cadence) are ‘dopa-resistant’.

There was evidence that the latency to drug effect, i.e., increased stride length, was inversely related to age-at-onset of PD, although the result was diminished by the corresponding relationship with age. The causal mechanism of delayed ‘on’ is unknown. Motor fluctuations are more common and commence sooner after initiation of levodopa therapy in younger onset PD (<40 years) (Quinn et al. 1987; Schrag et al. 1998; de la Fuente-Fernandez et al. 2000), and a recent study has suggested a greater imbalance between dopamine turnover and dopamine synthesis, storage and release in young-onset PD that may induce larger swings in synaptic dopamine levels (Sossi et al. 2006). Increased latency to drug effect of the first levodopa dose also appears to be related to reduced levodopa plasma levels (Chana et al. 2004) and delayed gastric emptying (Chana et al. 2004; Muller et al. 2006). A potential weakness of the current study was that food intake was not monitored, although it appears likely that any disparity in protein content between subjects [which may delay absorption and reduce cerebral uptake (Nutt et al. 1984)] would increase variability in the relationship between latency and age-at-onset, and potentially weaken the observed correlation. The Hill coefficient was also correlated with latency; the longer the delay until stride length increase the sharper the transition from ‘off’ to ‘on’, consistent with the observation that patients with motor fluctuations tend to have a delayed response to the first oral levodopa dose of the day (Chana et al. 2004).

There was no evidence of a correlation between the locomotor response (Hill coefficient and latency) and morning or total levodopa dose. This is consistent with the observation that the motor response to levodopa is essentially ‘all or nothing’ (Nutt 1987); that is, increasing a clinically effective dose does not significantly improve the response. Moreover, the dynamics of the finger-tapping response to levodopa were also independent of dose (Contin et al. 1996, 2001). Patients with moderate to advanced PD are commonly prescribed various adjunct treatments in an attempt to minimize motor fluctuations in response to levodopa. In this study the most common adjuncts were dopamine agonists (pramipexole, ropinirole) that directly stimulate post-synaptic dopamine receptors (10 patients), and COMT inhibitors (tolcapone, entacapone) to reduce peripheral metabolism of levodopa (4 patients). These adjunct medications augment the effects of oral levodopa, generally assumed clinically to increase the effective levodopa dose by around 10%. We believe these adjuncts were unlikely to have influenced the lack of correlation observed in the current study between the levodopa dose and the Hill coefficient and latency [dopamine agonists and MAO-B inhibitors did not affect the dynamic finger-tapping response to levodopa (Contin et al. 1996)]. However, the combination of levodopa and adjunct medications would need to be considered if the locomotor response was used to assess the effectiveness of pharmacological management of PD symptoms in individual patients.

Stress, attention and distractions can have both a positive and negative impact on Parkinsonian gait, and it is well established that visual and audio cues can improve locomotor performance (see Morris et al. 2001 for review). A striking example was the improvement in stride length when PD patients walked on a treadmill with transverse lines appearing periodically across the belt, which was not observed when the lines were parallel to belt motion (Hanakawa et al. 1999). SPECT imaging from this study demonstrated increased activity in the right lateral premotor cortex following walking with the transverse lines, suggesting that PD patients used non-affected visuomotor networks to compensate for impaired basal ganglia function. A possible disadvantage of ambulatory monitoring of PD patients is the potential for such attentional cues to influence gait unbeknownst to the clinician; however, this may be offset by the ecological validity of assessing movement in a natural setting (Dunn et al. 1994; Teasdale and Stelmach 1988; Czaja and Sharit 2003). Moreover, laboratory gait studies have not been able to discern the dynamics of the locomotor response to levodopa (MacKay-Lyons 1998), and arguably the most controlled walking environment, the treadmill, provides a ‘bottom-up’ drive to regulate stride (Hirasaki et al. 1999) that may confound the study of ‘top-down’ impairment of gait control in PD. In the current study the visual cues experienced by subjects were essentially similar in steady state gait, with little in the way of contrast along the walls and floor of the corridor (particularly in the transverse direction). However, the visual environment may well have influenced freezing, which occurred in several subjects (Fig. 3) when passing through doorways or turning.

Motor fluctuations, by definition, are variable and likely influenced by many factors. The relatively small number of subjects in the current study limited statistical analysis of the effects of levodopa dose, disease duration and age at onset. In particular, the novel finding of a relationship between young onset PD and latency was confounded by age. Larger studies accounting for patient age and disease severity are required to better understand this complex phenomenon. Nevertheless, the results of this study demonstrate that long-term monitoring of stride length is sensitive to the dynamic gait response to levodopa and provides a continuous, objective assessment of locomotor fluctuations that could be performed at the clinic or in the patient’s home. This information may prove a useful adjunct to the subjective patient feedback typically used in the clinical management of PD, as well as reflect changes in the dopaminergic system as the disease progresses.

Acknowledgments

Dr. Steven Moore was supported in part by NASA grant NNJ04HF51G.

Copyright information

© Springer-Verlag 2007