Background
Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder that is caused by the expression of mutant huntingtin secondary to a polyglutamine (CAG) expansion in exon 1 of the huntingtin gene [
1]. The disease is characterised by the dysfunction, and then loss, of specific neuronal populations especially in the striatum as well as in the cerebellar cortex, thalamus, cerebral cortex and hippocampus [
2,
3] in association with early posterior white matter changes [
3‐
5]. Typically patients present in mid life with an array of motor signs including chorea and bradykinesia as well as psychiatric and cognitive impairments [
6]. Many studies have sought to identify the earliest changes in HD and subtle impairments in motor and cognitive function before predicted disease onset have been reported [
7,
8]. Others have sought to more objectively track disease progression once the disease has become manifest and this includes a range of motor, cognitive and imaging approaches [
4,
9]. Such objective markers are increasingly needed as we move towards a time when disease modifying therapies for HD are coming to trial [
10].
Currently the gold standard for looking at motor impairments in HD is the Unified Huntington’s Disease Rating Scale (UHDRS), which was primarily designed for manifest HD [
11]. The UHDRS however, is susceptible to subjective error and inter-rater variability and has limited sensitivity in early HD patients [
12]. We, and others, have therefore sought to find simple motor measures which may be accurately used for tracking disease. In the TRACK-HD baseline analysis, voluntary self paced finger tapping was shown to be a sensitive task in premanifest patients and is associated with disease burden in HD patients [
4]. Speed finger tapping tasks used in cross-sectional studies has also been shown to be a sensitive early marker of change in premanifest and manifest HD, with the deficits being more pronounced in later stages of HD [
13]. These late change deficits also correlate with atrophy on MRI and clinical scores in HD patients thereby linking structure to function [
13]. In another cross-sectional study HD patients had larger variability in hand and finger tapping rates compared with controls and this correlated with cognitive impairments in these patients [
14]. In a pre-diagnostic study changes in button tapping showed a significant change in rate of decline as the subjects neared disease onset [
13]. In the Predict-HD study, premanifest HD patients also showed variability in speed and self paced tapping although this needs to be validated with longitudinal follow up [
15]. Collectively, the data from the TRACK-HD study showed that tapping speed was a robust measure in premanifest and HD patients at 12, 24, and 36 months [
5,
9,
16]. Premanifest patients also showed a decrease in the number of taps performed in the Predict-HD study at 2 years follow up [
17]. Our group has previously shown in a 10 year follow up study that the rate of decline in hand tapping correlated with UHDRS motor scores [
18].
In addition to hand and finger tapping, other motor measures have been found to be sensitive markers of disease progression including grip force and tongue protrusion tasks. Similar to tapping, performance on these tasks also deteriorates over time in both premanifest patients close to disease onset and in manifest patients in comparison to controls [
5,
9,
17]. The oculomotor system has also been studied in HD and correlates with other motor features [
19], even in longitudinal follow up [
20]. In particular, changes in saccades such as increased error rates, saccade latency and increased variability of saccade latency has been shown in premanifest and HD patients, with increasing abnormalities in advanced HD patients [
19]. Over a three year period premanifest and HD patients displayed significantly increased saccade latencies compared to controls, which could be used as a predictor of time to disease onset in premanifest patients [
20].
Finally, attempts have been made to look at more complex motor tasks such as the use of the peg board test, which measures the time taken to insert 25 pegs from a rack into a series of appropriate holes [
21]. It has been shown that patients are impaired in this task, although there does not appear to be any significant difference in patients in the premanifest versus manifest stages of the illness [
21,
22]. When testing complex tasks, execution speed of rapid alternating motion sequences are slower in HD patients compared to controls [
23].
In the analysis reported in this paper, we have used a simple hand tapping task, which we have previously shown to be useful in a small cohort of patients followed over time [
18]. We now report on the utility of this test in a much larger numbers of patients followed over a longer period of time (up to 14 years), to show that this single measure gives a robust annual decline irrespective of disease stage and therefore could be used as one of an array of assessment tools in trials in manifest HD.
Methods
Patients
237 patients were recruited from the regional NHS HD clinic at the Cambridge Centre for Brain Repair between 1998 and 2012, and hand tapping data was collected as part of their routine clinical assessment. Patients were seen typically at intervals of approximately 6 or 12 months, although some were seen more frequently for reasons of clinical care and management. All patients included in the analysis were seen for at least one year and for at least three visits and all were taking their normal medication for their HD. There were no inclusion or exclusion criteria outside of this, and thus and the patients are a representative sample of patients with HD. This analysis was classified as a service evaluation and registered with the Patient Safety Unit at Addenbrooke’s Hospital (Project Register Number: 3026). Since hand tapping data was not initially collected for research purposes and the analysis used fully anonymised data, ethics review was not required and informed consent was not obtained. All patients had a positive genetic test for HD or had features of the disease and came from a known HD family and were diagnosed as having manifest disease when they had a diagnostic confidence level score of >3. The calculated long allele CAG length for the patients in our cohort ranged from 37 to 62 repeats, this was calculated by PCR analysis as previously described [
24]. Many of the patients had their genetic test before 2001 when the exact repeat number was not routinely recorded. In those where a CAG length was known (n = 151), their demographic information is summarised in Table
1. The patients overall total functional capacity (TFC) [
25,
26] was assessed using the UHDRS which ranges from 0–13, with 13 equating to normal independent life and 0 to a total dependence on others. The motor UHDRS score was calculated at the patients first visit and ranged from 0–86. Higher scores indicate a more severe motor impairment [
11].
Table 1
Demographic information at first visit
Age
| | | | |
Male | 49.1 (13.7) | 14.0 – 75.6 | 47.3 (13.5) | 14.1 – 75.6 |
Female | 50.0 (12.8) | 18.5 – 77.9 | 49.3 (12.9) | 18.8 – 77.9 |
Follow up (years)
| | | | |
Male | 4.6 (2.6) | 1.0 – 12.9 | 4.7 (2.3) | 1.3 – 10.6 |
Female | 4.9 (2.5) | 1.0 – 14.1 | 4.5 (2.3) | 1.0 – 14.1 |
Number of visits
| | | | |
Male | 7.1 (4.9) | 3 – 25 | 6.8 (4.2) | 3 – 22 |
Female | 6.6 (4.2) | 3 – 26 | 6.6 (3.5) | 3 – 26 |
TFC
| | | | |
Male | 9.5 (3.4) | 2 – 13 | 10.4 (2.9) | 3 – 13 |
Female | 9.1 (3.4) | 3 – 13 | 9.5 (3.3) | 3 – 13 |
UHDRS motor score
| | | | |
Male | 23.3 (17.1) | 0 – 86 | 20.2 (13.6) | 0 – 51 |
Female | 25.8 (18.4) | 0 – 70 | 23.3 (17.5) | 0 – 62 |
Disease duration (years)
| | | | |
Male | 5.3 (3.3) | 0.1 – 17.0 | 4.4 (2.4) | 0.1 – 10.5 |
Female | 5.4 (3.2) | 0.2 – 15.8 | 4.7 (2.4) | 0.4 – 14.2 |
CAG length
| | | | |
Male | 44.6 (5.1) | 39 – 62 |
Female | 43.4 (3.9) | 37 – 61 |
Hand tapping device
The original hand-tapping device was designed as a simple objective measure of motor function for use in the routine assessment of patients in clinic. It consists of two buttons 6 cm in diameter, mounted with their centres 30 cm apart. The subject is asked to alternately tap one button after the other as rapidly as possible using their right hand for 30 seconds, and then again with their left hand. The total number of taps for each hand is recorded manually and then summed to give a total number of taps for both hands. This device is different to that used previously by us, which was a more sophisticated device that automatically downloaded data onto a computer and calculated inter tap intervals [
18]. Hence, the device used in the present study only allowed us to collect data on the total number of taps.
Statistical analysis
The main outcome for all analyses was the total number of taps, which was the sum of the number of taps made with the left and right hand. Even though the data are counts (non-negative integers), the data distribution was approximated well by a normal model (the values were far away from zero), which was therefore used instead of a Poisson or negative binomial model. The first analysis used all 237 patients and estimated the change in number of taps over time with a mixed-effects model. Fixed factors were time (since first visit), age at first visit, sex, and time by sex interaction. The random factors were patient and patient by time interaction (varying intercepts and varying slopes). The serial dependence of observations within patients was modeled using an exponential correlation structure. The second analysis used the 151 patients for which information on CAG length was available. The same mixed-effects model as the first analysis was used with the addition of CAG length as a continuous variable and the removal of the non-significant time by sex interaction.
Finally, the relationship between tapping and TFC was examined with a mixed-effects model using the 227 patients for which TFC data was available. The model included a fixed effect of age, sex, and TFC, and random effects for patient and patient by TFC interaction. Analysis was conducted with R (3.0.0).
Discussion
This analysis used a simple motor hand tapping task to follow disease progression in HD patients, at all stages of the disease, to ascertain whether tapping changes reliably over time. We found that in this patient cohort the rate of voluntary hand tapping declined at a steady rate (linearly) of 5.1 taps per year on average (Figure
1C) and this rate of decline was similar between sexes. It extends our previous study in this area by looking at larger numbers of patients over longer time periods, on treatment, and thus has the advantage that it recruited all patients attending clinic and so represents real life practice.
Several studies have now demonstrated that simple, rather than complex, motor tasks are the most sensitive markers of disease onset and progression in HD, and include tasks such as hand and finger tapping [
9,
27,
28], peg insertion task, grooved pegboard task [
29], tongue force [
5], and decision making reaction time [
30]. The current study reinforces this point that a simple motor biomarker is useful for tracking disease course and that the impairment in voluntary movement captured in these patients using this approach is related to functional disability. We did not assess whether hand tapping would be a suitable surrogate biomarker for TFC or any other physiological endpoint such as striatal volume, nor whether tapping has any prognostic or predictive value. However, we have previously shown that changes in hand tapping over time correlates with UHDRS motor scores [
18] and thus the number of taps could be used as a biomarker to assess the efficacy of compounds or other therapeutic interventions as it is noninvasive, inexpensive, and related to standard motor scores. In addition, tapping has an important advantage over "wet" biomarkers as it is not affected by sample quality, differences between labs or clinics in terms of sample preparation and handling, or differences between batches within a lab. It is therefore less likely to be affected by extraneous variables that can introduce bias and variability, making the analysis and interpretation of results more complex [
31].
Although this study has clear advantages in terms of size and representativeness of the sample, there are a number of limitations. These include the lack of a control group and a paucity of other clinical or imaging measures with which to correlate our hand tapping data. As this is a retrospective study based on patients attending the clinic for their routine appointment, we did not include testing of controls as part of the study design. This would be useful to include in future studies to better understand the age dependent effects on the rate of hand tapping. The smaller number of patients with known CAG repeat lengths also reflects the fact that the study is retrospective and that many patients were diagnosed at a time when these were not routinely recorded as part of normal clinical practice. Finally, it should be realised that this task can also be affected by mood and fatigue, subject to a practice effect, and can be impaired in patients with orthopaedic or rheumatological problems [
12]. We did not look specifically at any of these factors but over long time scales any practice effect would likely be negligible. In addition, the device does not rely on pressure sensitive taps or large movements, and therefore joint problems are unlikely to play a major part in determining the hand tap score.
Competing interests
The authors have no competing interests to disclose.
Authors' contributions
RAB planned the analysis and collected the data. SEL performed the statistical analysis and interpretation. LMC helped collect and analyse the patient data. All authors contributed to writing the manuscript and read and approved the final version.