This is a study of a database collected in previous studies and approved by St Vincent’s Hospital Melbourne Human Research and Ethics Committee. The criterion for selecting 228 PwP from this database was that contemporaneous scores from the UPDRS III and PKG data were available: 90% of whom also had the Parkinson’s Disease Questionnaire (PDQ39) and the other three UPDRS scales. The clinical characteristics of the participants are shown in Table 1. Included are 84 PwP who participated in two previously reported studies (15, 16) where oral therapy was used to treat bradykinesia or where Deep Brain Stimulators or changes in oral therapy were used to reduce dyskinesia. Demographics and selection are described in the results sections and in Tables 1 and 5. Also included were 157 subjects aged over 60, recruited from bowls and golf clubs, University of the 3rd Age, and Probis who had no previous concern of neurodegenerative disorders or gait disorders requiring use of walking aids. These Controls subjects wore the PKG for 6 days, but no clinical scales are available. All studies were carried out in accordance with the guidelines issued by the National Health and Medical Research Council of Australia for Ethical Conduct in Human Research (2007, and updated May 2015) and in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Clinical characteristics of participating subjects and inclusion criteria into the analyses are discussed below.
Table 1 Demographics and clinical scores of participants
|
PwP§
|
Controls§
|
Number
|
228
|
157
|
Gender
|
35% F: 65% M
|
61% F: 39% M
|
Age
|
71 (9.7)
|
69.6 (8.7)
|
Disease Duration
|
6.0 (6.0)
|
n/a
|
Unified Parkinsons Disease Rating Scale
|
|
|
Part I
|
10.0 (8.0)
|
n/a
|
Part II
|
11.0 (9.0)
|
n/a
|
Part III (“ON”)
|
36.0 (18.0)
|
n/a
|
Part IV
|
4.5 (6.0)
|
n/a
|
Total
|
62.5 (29.3)
|
n/a
|
Levodopa Equivalent Daily Dose (LEDD)
|
675 (500)
|
n/a
|
Parkinsons Disease Questionnaire
|
28 (29)
|
n/a
|
median Dyskinesia Score (mDKS)
|
2.1 (3.7)
|
2.6 (2.8)
|
Adjusted median Dyskinesia Score
|
1.3 (2.5)
|
1.5 (2.2)
|
median Bradykinesia Score (mBKS)
|
25 (9.6)
|
22.0 (2.8)
|
Active median Bradykinesia score
|
23.9 (7.8)
|
21.4 (3.2)
|
§ Values are mean and standard deviation (in parenthesis) of each variable from PwP and Control subjects.
The PKG System
The PKG system consists of a wrist-worn data logger, a series of algorithms that produce data points for bradykinesia (23) and dyskinesia (23) every two minutes and a series of graphs and scores that synthesise this data into a clinically useful format known as the PKG. The PKG plots the two-minute bradykinesia and dyskinesia scores against the time of day and shows when medications are due. The numerical output is summarised in the Glossary that follows. The reader is referred to the company’s website for further details (http://www.globalkineticscorporation.com.au/) and to other publications (15, 23-28).
Glossary of PKG terms
The PKG system assumes that there is a continuum in the distribution of the kinematics from Controls to PwP and that with treatment, the kinematics of some PwP can be normalised. Consequently, Controls as well as PwP have bradykinesia and dyskinesia scores, and in PwP, these scores can enter the control range if treatment is optimal.
Bradykinesia Score: This is the bradykinesia score for each 2 minute epoch produced over all days that the PKG was worn (23).
Epoch: Each 2-minute period of recording is called an epoch and analysed as previously described (23). The following scores are estimated from the part of the day between 09:00-18:00, excluding epochs where the logger senses that it was not being worn, or the Bradykinesia Score≥80 (usually sleep (25, 27)). As well, inactivity is removed for some scores. Inactivity is when so few movements are made over 2 minutes that clinical assessment of bradykinesia would not be possible and identified when the 30 minutes centre weighted moving median bradykinesia score is greater than 40. Available epochs are epochs remaining after those related to inactivity, sleep or occurring when the logger is not worn have been removed.
Median Bradykinesia Score (mBKS): The mBKS is the 50th percentile of all bradykinesia scores from epochs between 09:00 and 18:00, when the logger was worn and bradykinesia score≤80, over all available days that the PKG was worn (usually 6 days)(23).
Active median Bradykinesia Score. This is the 50th percentile of the bradykinesia score for all from epochs over days that the PKG was worn (usually 6 days). As well, epochs with inactivity are removed: inactivity being when so few movements are made over 2 minutes that clinical assessment of bradykinesia would not be possible and identified when the 30 minutes centre weighted moving median bradykinesia score is greater than 40.
Percent time bradykinesia (PTB). This is explained below after reader is introduced to Severity Levels (below).
Dyskinesia Score: This is the dyskinesia score for each 2 minute epoch produced over all days that the PKG was worn (23).
Median Dyskinesia Score (mDKS): This is the 50th percentile of all dyskinesia scores from between 09:00 and 18:00, when the logger was worn and bradykinesia score≤80, from all days that the PKG was worn (23).
Adjusted median Dyskinesia Score. This is the 50th percentile of the available dyskinesia score for all days that the PKG was worn excluding those in which either walking or tremor were detected. Walking refers to maintained perambulation detected using a supervised gradient boosting decision tree model to identify walking levels with sufficient energy to influence the dyskinesia signal. This model used features obtained from conventional gait detection and from the pattern of harmonics of the acceleration signal during the epoch under examination. Using a previously described tremor detector (26), epochs with tremor were also identified. Epochs with “walking” identified by the algorithm were removed as it is possible for dyskinesia and walking to occur in the same epoch and thus, these epochs are uninterpretable. If tremor was detected in the epoch and its dyskinesia score≥10, the epoch’s dyskinesia score was set to zero. The assumption was that under these circumstances the elevated dyskinesia score may have been mostly due to tremor, rather than dyskinesia. Epochs where bradykinesia score>80 or the logger was not worn were removed.
Percent time Dyskinesia (PTD). Percent time in dyskinesia (PTD) was estimated as the percent time of available epochs whose dyskinesia score≥10 and in which neither walking nor tremor were detected (as for Adjusted median Dyskinesia Score). The 75th percentile of the remaining epochs in the control population was 10, hence dyskinesia score≥10 was chosen as the threshold for PTD. PTD was estimated on available epochs between 09:00-18:00 on the 6 days the PKG was worn, and expressed as the percentage of all available epochs in that period. As the number of available epochs will vary from person to person, time in bradykinesia is expressed as a percentage, allowing comparison between subjects.
The first dose time. The PKG logger is programmed to vibrate at specific times to remind subjects to take their medications. Subjects can acknowledge when they consume the medications by “swiping” the smart screen on the watch. The first dose time is the 5 epochs (10 minutes), centred on the first reminder after 05:00.
Time of peak levodopa effect: This is the time when levodopa was estimated to have had its peak effect. It is calculated as the peak of the smoothed bradykinesia score time series from 46 minutes to 90 minutes after first dose time, using data from all recorded days (24).
Definition of Severity Levels of Bradykinesia
Severity Levels were fully described in a previous study describing how PKG data could be used to predict the absolute change in the UPDRS III produced by a levodopa challenge test (24). Here we briefly reiterate aspects of this model that are relevant to Severity Levels and a levodopa response (LR).
The UPDRS III scale was divided into six Severity Levels (Table 2 (29)) and the UPDRS III scores measured prior to and after the levodopa challenge were sorted into these six Levels. An algorithm assigned each PKG epoch to a Severity Level corresponding to the range of UPDRS III scores (Table 2). In the current study, this algorithm is used to measure the proportion of time a PwP is in bradykinesia and to measure Levodopa Response (LR). Note that the terms “Bradykinesia Severity Level” and “Severity Level” are used interchangeably, even though tremor and other information as well as the PKG’s bradykinesia scores were used in developing the score.
Table 2 PKG Severity Levels compared to UPDRS III
Target range
|
In target
|
|
Above target
|
Bradykinesia Severity Level
|
0
|
1
|
2
|
3
|
4
|
5
|
|
UPDRS III Interval
|
0-10
|
10-22.5
|
22.5-35
|
35-47.5
|
47.5-60
|
≥ 60
|
|
This table shows the range of MDS-UPDRS III corresponding to each bradykinesia severity level. Epochs in Severity levels 0, 1 and 2 were in target, whereas those in level 3 or above were above target.
Targets
Bradykinesia is considered “in target” and “controlled” when “Severity Level<2.5” and “above target” and “uncontrolled” when the “Severity Level≥2.5” and a “significant” LR is when the Severity Level changes by ≥1.15. This is based on a previous study (24) that used data from clinical levodopa challenge tests to show that changes in Severity Level≥1.15 were clinically significant and indicated a response to levodopa. A Severity Level of 2.5 (i.e., the midpoint between levels 2 and 3 in Table 2) approximates an mBKS of 26 and a UPDRS III of 35 and an LR of 1.15 is equivalent to ~14-point reduction in the UPDRS. The performance of the model is affected if the sample of available epochs becomes too small. This results in increased variability, which is particularly relevant at the time of the first dose, because some subjects continue to sleep or are inactive at that time. Excess Variability is present when the standard deviation of Severity Levels is greater than one unit of Severity Level at the times of the first dose and Effect Time. Cases with excess variability (34%) were removed when estimating LR and other dependent estimates. This is relevant for the early morning period (time of first dose) because sleep and or inactivity was more common and because there was only a maximum of 30 epochs available in early morning periods over 6 days (see Early Morning bradykinesia score, below).
Definition of parameters that depend on estimates of Severity Levels
In Target or Controlled. This relates to when the Severity Level was below the target range (Severity Level <2.5 or bradykinesia score<26). The term “on” has been reserved for the subjective symptom of levodopa action observed by the PwP.
Out of Target or Uncontrolled. This relates to when the Severity Level was above the target range (Severity Level ≥2.5 or bradykinesia score≥26). The term “off” has been reserved for the subjective symptom of loss of levodopa action observed by the PwP.
Percent time bradykinesia (PTB). PTB was estimated as the percent time in Severity Levels 3, 4 and 5. PTB was the number of epochs (excluding inactivity, sleep and the logger not worn as per Active median bradykinesia score) in Severity Levels 3, 4 and 5 between 09:00-18:00 on the 6 days the PKG was worn, expressed as the percentage of all available epochs in that period. As the number of available epochs will vary from person to person, time in bradykinesia is expressed as a percentage, allowing comparison between subjects.
Early Morning Bradykinesia Score: The Bradykinesia Severity Level estimated at the time of the first dose. This score was not estimated if there was no dose reminder or the first dose reminder was earlier than 05:00 or if more than 50% of the epochs were unavailable (24). Early Morning Bradykinesia was considered to be present when the Severity Level was≥2.5, in keeping with the definition of PTB.
The levodopa response (LR). The LR was estimated by calculating the magnitude of improvement in bradykinesia Severity Level at time of peak levodopa effect compared with first dose time (Fig. 1a). An improvement in Severity Level of 1.15 predicted an improvement of 14 UPDRS III points which also approximated a 30% improvement (24). We will use the terminology of “significant and non-significant” LR to refer to LR that are greater or less than 1.15 severity points, respectively.
Wearing-Off. If the LR declines by ≥1 Severity Levels within 2 hours of time of peak levodopa effect, it was defined as Wearing-Off (Fig. 1a). The rationale was to detect a decline in levodopa action occurring at ~3 hours from dose consumption.
Assessment of fluctuations
The LR is used to distinguish between Fluctuators (those whose LR is significant) and Non-Fluctuators (those whose LR is not significant) (Fig. 1b and Fig. 1c). Each were then divided as to whether their activity was in target (controlled) or out of target (uncontrolled). The abbreviations are provided here because they are also used in several of the Figures.
Non-Fluctuators-Controlled (NFC). These are PwP with a non-significant LR and whose early morning Bradykinesia level was already in the Controlled range (in target) prior to the first dose. Presumably, these would include PwP with early non fluctuating PD.
Non-Fluctuator-Uncontrolled (NFU). These are PwP with a non-significant LR not significant but whose early morning Bradykinesia level are in the Uncontrolled range (out of target). PwP with high bradykinesia and little or no LR are presumably undertreated or unresponsive to levodopa and indeed may have another akinetic-rigid syndrome.
Fluctuators: PwP whose response (LR) to the first morning levodopa dose was significant. However, this LR may not be large enough to cause bradykinesia scores to fall below target. In these circumstances, despite fluctuations, there is no reduction in the time spent in the target range (although bradykinesia scores are improved). As well, the LR might persist up to and beyond the next dose (no “wearing-off”) or there may be a rise in bradykinesia scores (“wearing-off”) prior to the next dose. Consider for example a PwP whose first morning dose significantly changes objectively measured scores (e.g., UPDRS III by 30 points) from being above target to below target 30 minutes later. The next dose is due in 5 hours, but after 3 hours in target, bradykinesia rises above target and remains high until the next dose 90 minutes later ( Fig. 1a). For the five hours between the first and second dose this PwP is above target 120 mins (30+90). The medication regimen is then changed to be 3 hourly: it still takes 30 mins following the first dose until target is reached but the subject now stays within target until the next dose (and possibly for the rest of the day). Both are fluctuators (recognised by the significant LR, Fig. 1b) but will be categorised separately. Similarly, subjects whose LR is significant but not sufficient to reach target are also recognised as a separate case. To accommodate these concepts four categories of fluctuators were recognised (Fig. 1b).
Fluctuator whose response enters the Controlled range and Persists in that range (FCP). Early morning “off” with significant LR that results in bradykinesia levels that enter the target range and does not “wear-off”.
Fluctuator whose response enters the Controlled range, but the response Wears OFF (FCWO). Early morning “off” with significant LR that results in bradykinesia levels that enter the target range but the LR “wear-off”.
Fluctuator whose best response remains in the Uncontrolled range and Persists in that range (FUP). Early morning “off” with significant LR but the bradykinesia levels remain above target but does not “wear-off”.
Fluctuator whose best response is in Uncontrolled range, but the response Wears OFF (FUWO). Early morning “off” with significant LR but the bradykinesia levels remain above target and “wear-off”.
Clinical Scales
Clinical Scales were performed within a month or less of wearing the PKG. The UPDRS was the Movement Disorder Society version (MDS-UPDRS) and was performed during the day while participants were taking their usual medication. All MDS-UPDRS III scoring was done by St Vincent’s Neurology Department staff who had received the MDS-UPDRS training.