Isometric force matching in the TRACK-HD study - a novel quantitative assessment for clinical applications?

Abstract

Background

Huntington's disease is a rare, progressive, neurodegenerative disease. Capturing symptomatic progression and treatment effects reliably in clinical therapeutic trials has shown to be a challenging task, facing the problem of small cohorts and a variable phenotype. Hence, robust and sensitive outcome measures are needed, to assess efficacy and safety of novel therapies in clinical studies with small cohorts.

Objective

Objectives of this study were to assess feasibility, discriminative potential, and correlation to clinical and imaging endpoints of the Q-Motor isometric force matching task with visual feedback. Furthermore, a statistical comparison with the Q-Motor grasping and lifting task should be assessed.

Methods

220 huntingtin gene expansion carriers (HGEC) and 110 non-huntingtin gene expansion carrier (Non-HGEC) participants of the observational TRACK-HD study completed the Q-Motor force matching assessment, along with a standard battery of clinical tests (UHDRS) and MRI assessments. During the Q-Motor force matching assessments, patients were reproducing a target force in precision grip with help of visual feedback. Q-Motor utilizes a highly sensitive force transducer to record force feedback. HGEC participants were categorized into four groups by a CAG- and age-based survival score.

Results

Q-Motor force matching allowed for very good discrimination between HGEC and Non-HGEC participant groups (p < 0.001 for all but least affected HGEC group) and between different HGEC groups (all p < 0.05). Strong correlations with UHDRS scores (TMS = −0.708, TFC = −0.638), CAG-Age product scores (Survival Score = 0.626) and imaging outcomes (caudate volume = −0.609, striatum volume = −0.624) were observed. A statistical difference between correlation strength of Q-Motor force matching and Q-Motor grasping and lifting tasks could not be observed.

Conclusions

Cross-sectional analysis of Q-Motor force matching showed promising results, outperforming clinical rating scales in sensitivity. Further efforts are required to assess longitudinal robustness of the task, and to further explore its potential of capturing cognitive effects by increasing cognitive load during the task.

Keywords

Huntington's disease motor disorders technology assessment haptic devices visual feedback hand grip strength chorea outcome assessment feasibility studies

Introduction

Huntington's disease (HD) is a progressive, neurodegenerative, autosomal dominant CAG-repeat disease. The disease pattern involves disruption and atrophy of white matter particularly in the vicinity of the basal ganglia and cortical regions.¹ The basal ganglia and the basal ganglia-cortical loops are suspected to play a central role in many processes that are known to be impaired in persons with HD (PwHD). They include reactive control and timing control, as well as sequential hand movements and movement planning, likely caused by reduction in somatosensory input to cortex, and decreased facilitation of cortical neurons.^1–3

Currently, there is no effective therapy to slow down the progression of HD. However, there are some promising therapeutic approaches reducing the supposedly toxic mutant Huntingtin protein (mHTT) or targeting other neuroprotective effects.^4,5 As recent studies have shown, there is a strong need for quantitative, objective measures, to reliably and sensitively assess the efficacy and safety of such approaches. This is concluded from repeated observations showing that clinical rating scales suffer from high inter and intra rater variability and appear to be vulnerable to patient and rater induced placebo effects, which mask potential treatment effects.^6–8

One quantitative measure frequently used in HD clinical trials is the quantitative motor assessment battery Q-Motor.⁹ The Grasping and Lifting task of the Q-Motor battery, revealed increased grip force variability that correlated with HD severity and progression. The measure allows for differentiation between gene-expansion carriers and Non-HGEC controls up to a decade before clinical disease onset in the clinically pre-symptomatic/premanifest stages.¹⁰ The isometric force matching Q-Motor task investigated in this study was designed to specifically assess grip force control with visual feedback and under well-defined conditions, i.e., with the intention to reduce the impact of chorea and motor impersistence regularly observed in the Grasping and Lifting task investigated previously.^11,12 Isometric force matching has already been shown to be a sensitive tool to measure motor performance in HD in a small pilot study, as well as in other movement disorders or diseases that involve motor impersistence.^13–15

The data presented here was generated in the TRACK-HD study, a multicenter observational study, that was conducted in four clinical sites in London (UK), Paris (France), Leiden (Netherlands) and Vancouver (Canada) between 2008 and 2011. The study was designed as adoptive study with a baseline visit in year 1, and 3 years of annual follow-up visits. The Q-Motor isometric force matching tasks were introduced in the course of the study and conducted in 220 individuals that were huntingtin gene expansion carriers (HGEC) and 110 control participants that were non-huntingtin gene expansion carriers (Non-HGEC). The objective of this work was to investigate the feasibility, sensitivity and applicability of the Q-Motor force matching task in clinical studies in Huntington's disease and to explore its possible usefulness as quantitative outcome measures.

We specifically aimed to investigate the hypotheses whether isometric force matching with visual feedback (1) allows for discrimination between (Non-HGEC) and (HGEC) groups; (2) correlates with core HD clinical endpoints and/or imaging measures and (3) delivers more sensitive data about fine motor force control of participants than the Q-Motor Grasping and Lifting task.

Materials and methods

The TRACK-HD study

The study was approved by the local ethics committees; all enrolled participants gave written informed consent. Q-Motor tests were conducted by trained and certified Q-Motor raters. TRACK-HD was designed as an adoptive biomarker study; novel tests were introduced at annual visits following steering committee recommendations. During the visits 3 and 4 of TRACK-HD, the Q-Motor Force Matching task was introduced as part of the assessment battery driven by exploratory data, published later.¹³ Therefore in this work, only TRACK-HD year 3 data was used for the cross-sectional analyses of Force Matching data. Basic demographics of the study population at year 3 are listed in Table 1.

Table 1.

Basic demographic data of the study population at baseline (year 3 of the TRACK-HD study), grouped by the traditional staging model of HD (cont = controls (non-HGEC), preHD = premanifest HD, HD = symptomatic HD).

	Cont	preHD	HD
n	110	70	150
age [median (IQR)]	47.6 (16.1)	41.4 (12.4)	49 (16.9)
education [median (IQR)]	4 (2)	4 (1)	4 (2)
gender [% female]	0.56	0.54	0.55

Participant grouping

The TRACK-HD study enrolled 366 participants, with 332 participants attending the 24 moths visit (year 3) of the study. Two hundred and twenty individuals carrying the mutant huntingtin gene (mHTT) and 110 control participants without the gene mutation (Non-HGEC) completed the Q-Motor Force Matching assessment in year 3. HGEC groups were stratified by the Unified Huntington's disease Rating Scale (UHDRS) diagnostic confidence score into premanifest and manifest groups of matched population sizes.¹⁰ The premanifest group was further divided into two subgroups preHD A and preHD B, by the median of the individuals’ predicted years to HD diagnosis, with preHD B being closer to a predicted clinical onset of HD than preHD A. The manifest group was divided into the subgroups HD stage 1 and HD stage 2, based on the UHDRS Total Functional Capacity score. HD stage 1 was less impaired than HD stage 2.

First assessment of isometric Force Matching started in year 3 of the TRACK-HD study. Considering this visit as baseline for our analyses, we found that the stratification at baseline no longer matched the original criteria, as the disease had progressed. Re-grouping the participants using the same stratification on the other hand resulted in rather unbalanced subgroup population sizes. Thus, we divided participants in subgroups solely based on the survival model, presented in Langbehn et al.; a parametric model that predicts the probability of surviving until at least the given age without developing neurological symptoms.¹⁶ We applied the model to all HGEC participants, regardless whether they were clinically symptomatic or not. Since the model takes age and CAG repeat length as its sole input parameters, it makes the grouping strategy independent from clinical ratings and thus from subjective bias.

In lack of common cutoffs for this grouping strategy, and based on the originally used sub-groups pre-HD 1 and 2 and manifest HD stage 1 and 2, we chose cutoffs at the quartiles of the survival score to subdivide HGEC participants into four groups from $S (age, CAG)$ = [0–0.25) with the highest risk, over [0.25–0.5) and [0.5–0.75) to [0.75–1), the lowest risk of having developed motor symptoms at the current age.

These grouping criteria are un-biased from the physicians’ assessments of diagnostic confidence and the UHDRS sub-scores, and independent from the population sample that is here analyzed. Furthermore, they result in a more homogeneous distribution across the sub-groups in our sample than the original grouping criteria used at baseline of TRACK-HD. Analogue to Table 1, Table 2 shows the demographic distribution per group as stratified by survival score.

Table 2.

Basic demographic data of the study population at baseline (year 3 of the TRACK-HD study), distributed in four HGEC groups (HGEC1, HGEC2, HGEC3, HGEC4) by survival score, plus non-HGEC control group.

	Controls	HGEC1	HGEC2	HGEC3	HGEC4
n	110	68	69	35	48
age [median (IQR)]	47.6 (16.1)	40.1 (9.8)	44.2 (11.2)	53.7 (15.3)	56.6 (12.1)
education [median (IQR)]	4 (2)	4 (2)	4 (2)	4 (2)	4 (2.2)
gender [% female]	0.56	0.56	0.59	0.54	0.48

In the analyses we included Q-Motor data correlations with the survival score and with the CAP (CAG-Age product) score, since CAP scores are frequently used tools in HD clinical research e.g., for patient stratification. The CAP score as used for our analyses is computed as $CAP = age \cdot (CAG - L) / K$ , with $L = 35.5$ and $K = 1$ (also referred to as Disease Burden Score). However, a major difference between CAP scores and the Langbehn survival model is the non-linearity between symptom progression and CAG-Age relation, which is captured by the Langbehn survival model but is not reflected in the generally simpler CAP scores.

Q-Motor force matching

The quantitative motor (Q-Motor) battery was designed to supplement the UHDRS Total Motor Score (UHDRS-TMS), the clinical gold standard to assess motor performance and progression in HD, with objective data.⁹ Q-Motor captures and tracks movement characteristics that are known to be affected in HD, including involuntary movements (chorea), bradykinesia, increased motor variability and impairments in motor persistence (i.e., motor impersistency). Movement characteristics of tapping and isometric force tasks are detected using a pre-calibrated and temperature-controlled six-axis force transducer (ATI Force/Torque SI-20-1, ATI Industrial Automation, Apex, NC) measures with a sampling frequency of ∼350 Hz at a precision of 0.01 N in z-direction.¹⁷

Q-Motor Force Matching was an isometric force persistence task using a device specifically designed to assess motor impersistency (Figure 1). The participant was seated comfortably in front of the device. The force transducer was mounted to the device, which was firmly attached on the table in front of a monitor, that displayed a target force (orange line) and the current force applied on the force transducer (blue line).

Figure 1.

A: Task set-up. The participant is seated in front of a screen; the grasping device is mounted fixed on the table. B: The device with attached force transducer is grasped in precision grip, using the thumb and index finger. C: The monitor displays the target force (orange line) and currently applied force (blue line).

When the examiner started the task, an acoustic signal was audible. The participant was instructed to grasp the force transducer in the precision grip with thumb and index finger of the assessed hand, and to match the applied force (blue line on the display) with the target force (orange line on the display) as good as possible. After 20 s another acoustic signal marked the end of the trial.

The task has been performed with two target forces, low (2 N) and high (10 N), and was recorded first with the right and then with the left hand. Each task comprised six consecutive trials including an initial practice trial that was not evaluated. All data was recorded using the software WinSC (Department of Physiology, University of Umeå, Sweden).

Definitions of Q-motor measures applied

Static phase

Participants grasp the device after they hear the start sound and first forces are recorded until the trial ends signaled by an audio cue played 20 s after the start. The “static phase” is defined as the 15 s interval from second 5–20 aiming to assess performance of participants after being able to adjust applied forces to the target force and reaching a fairly steady-state or stationary force level.

Force

The ATI-Mini40 force transducer applied is a six-axis force sensor measuring forces in three dimensions (x, y, z) as well as corresponding shearing forces. The z-force channel denotes the force applied perpendicular to the grasping surface of the sensor and is the force channel used and analysed in this study. It is referred to as “force” hereafter.

Force deviation

“Force deviation” is defined as the difference between measured force and displayed target force level.

Force variability

“Force variability” is defined as the coefficient of variation of the static phase force, calculated as standard deviation of the force divided by the mean force level measured during the static phase (see above).

Force deviation variability

Analogue to force variability, the “force deviation variability” is defined as coefficient of variation of the force deviation.

Controls

Group of non gene-expanded (Non-HGEC) participants included in the study.

HGEC1

Group of gene-expanded participants with a survival score higher than 0.75 ((0.75, 1]) that has a low risk of showing symptoms of HD.

HGEC2

Group of gene-expanded participants with a survival score between 0.5 and 0.75 ((0.5, 0.75]) that has an increased risk of showing symptoms of HD.

HGEC3

Group of gene-expanded participants with a survival score between 0.25 and 0.5 ((0.25, 0.5]) that has a high risk of showing symptoms of HD.

HGEC4

Group of gene-expanded participants with a survival score lower or equal 0.25 ((0, 0.25]) that has the highest risk of showing symptoms of HD.

Statistics

Statistical analyses were conducted with R version 4.1.2, using the tidyverse framework.^18,19 If not explicitly noted otherwise, R-core packages have been used for computations. We considered a type I error of $α = 0.05$ acceptable; statistical outcomes with p-values < 0.05 were considered statistically significant.

To identify the set of independent Force Matching outcome variables with the highest sensitivity for HD progression, we applied a random forest model using the party package R function party::ctree, for each test condition separately (low target force, left hand; low target force, right hand; high target force, left hand; high target force right hand).^20,21 Variable importance was assessed under consideration of correlations among the predictors.

For group comparisons, the Wilcox Rank Sum test has been used to determine group differences. Visualization of group comparisons was done using notched box plots, with boxes that denote the inter quartile range (IQR) of the data sample, and whiskers that reach to the highest and lowest value within a $1.5 \cdot IQR$ area. Notches expand over $1.58 \cdot IQR / \sqrt{n}$ as a good approximation for the 95% confidence interval to the median.

Correlation to clinical measures and brain imaging results were computed using Spearman's $ρ$ , which is robust against outliers and non-normal distributed data. Clinical assessments for correlation analyses included the Symbol Digit Modality Test (SDMT), Total Functional Capacity (TFC) and Total Motor Score (TMS). TRACK-HD visits included annual MR brain imaging of participants. Volumetric data from the MRI scans were available for these analyses. We used log-transformed results of striatum, caudate and putamen volumes, normalized to the intracranial brain volume (ICV) for correlation analyses.

To determine and separate the influence of age, sex, education and study site, we computed a linear model using the formula

l o g (force variability) \sim group + age + sex + education + study_site

with log-transformed force variability as the dependent variable. Independent variables included age defined as integer age in years at visit date, education as rating of the six point International Standard Classification of Education (ISCED-97) scale, group as nominal factor with the levels “Non-HGEC”, “HGEC1”, “HGEC2”, “HGEC3” and “HGEC4”, sex as binary factor (0 = male, 1 = female) and study_site as nominal factor with the levels “leiden”, “london”, “paris” and “vancouver”.

Comparisons of correlation coefficients were done using the WRS package.²² We followed Wilcox et al., using a percentile bootstrap approach with B = 500 bootstrap samples, to investigate whether differences in correlation coefficients between Q-Motor Grasping and Lifting task and Q-Motor Force Matching task can be considered statistically significant.²³

Results

Sample demographics

Year 3 of the TRACK-HD study was completed by 110 Non-HGEC participants, 101 pre-symptomatic and 119 symptomatic HGEC participants (see Table 1), as enrolled for year 1 of the study. For three subjects of the symptomatic group, no valid data was available for the low target force assessment, while 4 data points of symptomatic HGECs were missing for the high target force matching data.

While demographic characteristics were generally well balanced, the group ages differed between groups; the pre-symptomatic HGEC group was significantly younger than Non-HGEC controls (p < 0.001) and symptomatic HGECs (p < 0.001).

Group comparisons

Figure 2 shows boxplots of two sample parameters of the matching task (coefficient of variation of the applied force (force variability) and absolute deviation of residual force (force deviation variability)) plotted against the classical staging model groups (top) and the groups that resulted from the survival scores (bottom). While the performance distribution of participants from less to more affected participants followed a sigmoid shape in the staging groups, it was almost linear when looking at survival score groups.

Figure 2.

Comparison of force matching variables performance in different grouping strategies. Top: Force matching variability and force deviation variability of the left hand grouped by classical staging model. Bottom: Force matching variability and force deviation variability of the left hand grouped into four equidistant survival score subgroups. The sample sizes of subgroups using the survival score strategy are more balanced and the relationship with the Q-Motor variables was more linear, compared to the classical staging model.

With the exception of the least affected HGEC group (HGEC1), all HGEC subgroups differed significantly from the Non-HGEC control group (p < 0.001). In the survival score grouping model, also the four HGEC groups were differentiated from each other (HGEC1 ∼ HGEC2 p < 0.001; HGEC2 ∼ HGEC3 p < 0.05; HGEC3 ∼ HGEC4 p < 0.001), using Wilcox Rank Sum tests.

In the following analysis, group comparisons of force matching performance are reported for the survival score groups, only.

Figure 3 shows the un-adjusted p-values with 95% confidence intervals from comparisons of HGEC groups against Non-HGEC group for all variables and test paradigms, using Wilcox Rank Sum tests. In all categories, the force coefficient of variation showed the best performance in terms of differentiation from controls.

Figure 3.

Unadjusted estimates and 95% confidence intervals from Wilcox Rank Sum tests, comparing the control group to the individual survival score subgroups for each test paradigm (high and low forces, right and left hand). The subgroup with the lowest risk to show symptoms soon (HGEC1) is the only one that did not significantly differ from controls in every test.

A detailed table of the un-adjusted estimates, confidence intervals and p-values is presented in the Supplementary Material (see Table 6).

Variables of importance and correlations

While the force coefficient of variation qualitatively showed the best performance of extracted matching task features, we decided to verify this impression with a statistical approach, using a random forest model under consideration of correlations among the predictors.

The importance distribution was similar in all test cases, Table 3 shows the mean results, ordered from highest to lowest.

Table 3.

Mean variable importance, conditionally determined using a random forest model over each test paradigm.

Variable	Mean importance
PLCV	0.00835
PLSD	0.00503
SRMD	0.00152
SRAD	0.00093
PLMN	0.00024
SRSD	0.00022
SRMN	−0.00013

For each test paradigm, the force coefficient of variation (force matching variability) turned out to be the most influential and important variable. In line with this finding, the force coefficient of variation also yielded the overall highest correlation with clinical scores and imaging results (see Appendix Table 7).

Correlations and linear regressions

All correlation coefficients between Q-Motor variables and clinical scores or imaging outcome were performed using Spearman's $ρ$ , which is suited for non-parametric and parametric data and is known to yield results that are robust against outliers.²³

Highest correlation was observed between force variability (low target force, left hand) and the UHDRS-TMS ( $ρ = 0.708$ ) while there was generally good correlation with other clinical and demographic scores, either (UHDRS-TFC: $ρ = - 0.638$ , Survival Score: $ρ = - 0.626$ ).

A detailed correlation table of all variables of interest can be found in the Supplementary Material (see Figure 6 and Table 7).

In Figure 4, correlations between the best performing force matching coefficient of variation left hand (force variability) and various outcome measures of the Track-HD study were plotted, with 95% confidence interval ellipses in ordinal scale data and linear regression lines in interval scale data for individual groups.

Figure 4.

Top: scatter plot with 95% confidence interval ellipses, showing force matching coefficient of variation of the left hand (force variability) versus UHDRS total functional capacity (TFC) and Total Motor Score (TMS). Bottom: Scatter plots with linear regression lines and standard error of Spearman correlations between force variability versus Single Digit Modality Test correct score (SDMT) and striatum volume change normalized to intra cranial volume (Striatum / ICV).

Strongest correlations were observed with the clinical TMS, ranging from $ρ = 0.45$ to $ρ = 0.68$ ( $p < 0.001$ ) in the HGEC groups, with no significant correlation in the control group. While the group of Non-HGEC controls did not cover a large range in clinical scales TMS and TFC, and thus yielded no significant correlations with Q-Motor variables, a correlation between Q-Motor Force Matching and normalized striatum volume was weak but significant ( $ρ = - 0.26, p = 0.009$ ).

Correlations for the CAP and Survival Score (Figure 5) were not computed per group, since both variables, as well as the grouping strategy, were based on age and CAG as factors. Both showed strong correlations (CAP: $ρ = 0.62, p < 0.001$ ; Survival Score: $ρ = - 0.63, p < 0.001$ ) with the force matching coefficient of variation.

Figure 5.

Linear regression with standard error of spearman correlations between force matching coefficient of variation of the left hand (force variability) versus the disease burden score (CAP) and survival score.

Linear model

Except for the least affected gene-expanded group HGEC1, all groups of HGECs showed significantly more variance in the Force Matching task ( $p < 0.001$ ) compared to Non-HGEC controls. Next to a weak site-effect, we also observe significant effects in sex ( $p < 0.05$ , males tend to show less variability than females) and education ( $p < 0.001$ , less variability was observed in participants with more years of education). The complete coefficients table can be found in Table 4.

Table 4.

Coefficient table of the linear model. With respect to further factors like age, sex, education and study site, the least affected HGEC group HGEC1 does not show significant performance differences compared to controls, while the difference in all other groups reach highly significant levels.

Variable	Estimate	Std. Error	t-value	p-value
bin: HGEC1	0.09490	0.08537	1.11165	0.26713
bin: HGEC2	0.48223	0.08223	5.86420	0.00000***
bin: HGEC3	0.88467	0.10449	8.46632	0.00000***
bin: HGEC4	1.43700	0.09506	15.11613	0.00000***
age	−0.00315	0.00327	−0.96320	0.33618
sex	−0.13787	0.05937	−2.32229	0.02085*
education	−0.10143	0.02409	−4.21023	0.00003***
study site: london	0.11129	0.08422	1.32139	0.18733
study site: paris	0.19618	0.08485	2.31219	0.02141*
study site: vancouver	0.16201	0.08273	1.95817	0.05109

Comparison to Q-motor lifting task performance

When comparing Q-Motor Force Matching results with grasping force variability of the Q-Motor Lifting task, we found higher correlation coefficients for Force Matching variability for all computed correlations (Table 5).

Table 5.

Comparison of correlation coefficients. Spearman's $ρ$ has been computed to determine correlations between Q-Motor Matching and Lifting task results and clinical variables as well as imaging outcomes. Correlation coefficients were compared, using a percentile bootstrap approach.

	$ρ_{L i f t}$	$ρ_{M a t c h}$	$Δ ρ$	p-value	Confidence interval
TMS	0.66913	0.70826	−0.03913	0.224	−0.105–0.027
TFC	−0.58357	−0.63838	0.05480	0.104	−0.0118–0.113
CAP	0.57450	0.58636	−0.01186	0.756	−0.101–0.0778
striatum_log	−0.54125	−0.62413	0.08288	0.048	0.00127–0.165
putamen_log	−0.53168	−0.61945	0.08777	0.032	0.00847–0.174
caudate_log	−0.54030	−0.60891	0.06862	0.096	−0.0168–0.155
brain/ICV_log	−0.45904	−0.49833	0.03929	0.348	−0.0479–0.132
caudate/ICV_log	−0.58949	−0.59847	0.00898	0.872	−0.0734–0.0929
putamen/ICV_log	−0.59488	−0.59640	0.00152	0.976	−0.0684–0.082
striatum/ICV_log	−0.60299	−0.61024	0.00726	0.904	−0.0671–0.0858

However, comparison of correlation coefficients using the bootstrapping approach as suggested by Wilcox et al. showed no statistically significant differences between Force Matching and Lifting correlation coefficients.²³

Discussion

We found that assessing the Force Matching task in the multicenter TRACK-HD study was feasible with a very high level of data availability and -quality.

Grasping and holding an object is a task we perform in everyday life, e.g., when eating and drinking, brushing our teeth or holding a pen or smartphone. Impairment of grip force control has a direct impact on these daily activities and as such on quality of life. The measures defined and reported are therefore aimed at assessing functionally relevant tasks.

One limitation of this study were unbalanced groups when looking at classical HD categories (pre-manifest, stage 1, stage 2, etc), induced by the late implementation of the Force Matching task into TRACK-HD in year 3. Furthermore, being an observational study, neither participants nor physicians and raters were blinded for the participants’ genetic status, which may have introduced a bias in the clinical rating scales the HD categories depend on.

We addressed these limitations, using an alternative approach to HD categories, using the Langbehn Survival Score which is in nature similar to the CAG-Age product (CAP) scores and does not rely on qualitative rating scales.¹⁶ Another possibility of categorizing participants could have been the HD Integrated Staging System (HD-ISS), a framework that has been developed to become the standard tool e.g., for cohort stratification in HD clinical research.²⁴ However, in lack of some variables required for correct computation of HD-ISS stages, a reliable comparison with our data was not possible. Particularly the volumetric ratios of putamen and caudate to intracranial brain volume have to be computed using a specified imaging processing pipeline, to receive robust landmarks for HD-ISS stage 0 and 1. The Independence Scale which serves as landmark for stage 2 and 3 was not included in the TRACK-HD protocol at all.

Force Matching allowed for very good discrimination between HGEC groups and Non-HGEC group, particularly for HGEC groups with a Survival Score of $\leq 0.75$ .

In contrast to earlier findings in a smaller cohort (see Medzech et al.), the correlations were evident not only in 2 N target force matching but also with the 10 N target force.¹³ Still, the lower target force differentiated more sensitively between Non-HGEC and HGEC groups, particularly in early stages of the disease, while high forces tended to be harder to match for more affected participants.

Grasping force variability showed strong and highly significant correlations with clinical rating scales such as the UHDRS-TMS and TFS, with CAG-Age product scores like CAP and the Survival Score and also with imaging variables, particularly with caudate and striatum volumes.

The Force Matching assessment was designed to assess aspects of grasping force control and -variability in a more defined environment, compared to the already available Q-Motor Grasping and Lifting task, which does record grasping force variability but does not include a visual feedback or the premise to match a target force.

As both tasks have been assessed in the TRACK-HD study, we could compare the performance in correlations using a percentile bootstrap approach. All correlation coefficients were increased in the Force Matching task, however, significant differences could only be observed in correlations with the non-normalized striatum and putamen volumes. Still, in summary the results suggest an increase in sensitivity when using the Force Matching task to assess grasping force.

Our findings suggest, that sensitivity for a plain force matching task may suffer from a floor effect in patient populations with little to no motor symptoms. While the task may be a useful marker of disease progression in trials of symptomatic therapies, pre-symptomatic trials are gaining popularity in pharmaceutical HD research. However, the visual feedback design of the presented task yields the potential of adding additional motor and cognitive challenges to the assessments e.g., through variable target patterns or in dual-task designs. It could potentially be used to answer further questions, regarding fine motor control, force endurance and fatigue, and shifting sensitivity towards less motor symptomatic populations.

Further analyses of longitudinal performance will be necessary to assess the question whether force matching can still be a sensitive marker for subtle performance changes also in pre-manifest HGECs. In combination with the robust and sensitive hardware and the high degree of standardization of the task, careful tuning of task paradigms may enable detection of smallest performance differences.

Quantitative digital assessments like the one presented here potentially can increase effect sizes while reducing population sizes in clinical trials, resulting in more accurate and cost effective trials and less burden for participants, compared to categorical, qualitative scales. We have previously observed that compared to categorical clinical rating scales, quantitative Q-Motor data was free of patient- and rater bias, yielding placebo-free results.^6–8 Several Q-Motor measures were more sensitive than clinical scales in detecting signals early and in treatment studies.^{6–8,10,25,26} The data observed in this study suggest that the Q-Motor Force Matching task investigated here is a candidate for further exploration as part of the Q-Motor assessment battery in future studies.

Supplemental Material

sj-docx-1-hun-10.1177_18796397251386987 - Supplemental material for Isometric force matching in the TRACK-HD study - a novel quantitative assessment for clinical applications?

Supplemental material, sj-docx-1-hun-10.1177_18796397251386987 for Isometric force matching in the TRACK-HD study - a novel quantitative assessment for clinical applications? by Robin Schubert, Pascal Barallon, Benjamin Habbel, Michael Deppe and Ralf Reilmann in Journal of Huntington's Disease

Footnotes

Acknowledgments

We acknowledge the contribution of the TRACK-HD study team and sites in Leiden, London, Paris & Vancouver. TRACK-HD including the development of Q-Motor tasks at GHI was funded by the CHDI-Foundation. Q-Motor tasks can be made available for the research community in collaborations and sponsored projects. The Isometric Force Matching task is protected by patent DE102012018124A1 (QuantiMedis GmbH. Isometrische finger-greifkraft-ziel-messvorrichtung. DE102012018124A1, 2012. (accessed 21 February 2025)).

ORCID iDs

Robin Schubert

Pascal Barallon

Benjamin Habbel

Michael Deppe

Ralf Reilmann

Ethical approval and informed consent statements

The study was approved by the local ethics committees. All enrolled participants gave written informed consent.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The George-Huntington-Institute received funding for Q-Motor assessments and data analyses in the TRACK-HD study.

Declaration of conflicting interests

Robin Schubert, Benjamin Habbel and Pascal Barallon are employees of the George-Huntington-Institute and QuantiMedis, both developing and distributing the Q-Motor system for clinical studies and research.

Ralf Reilmann is founder and CEO of the George-Huntington-Institute and QuantiMedis.

Data availability

The TRACK-HD data set can be obtained from CHDI, by submitting a request on .

Not-included Q-Motor data can be made available on request.

Artificial intelligence policy

No artificial intelligence has been used for generation or enhancement of content of any kind in any part of this manuscript.

Supplemental material

Supplemental material for this article is available online.

References

Rao

Gordon

Marder

. Coordination of fingertip forces during precision grip in premanifest huntingtons disease. Mov Disord 2011 Mar; 26: 862–869.

Muratori

Reilmann

Gordon

. Coordination of fingertip forces during precision grasping in multiple system atrophy. Neuropsychologia 2003 Jan; 41: 1498–1508.

Fellows

Schwarz

Schaffrath

, et al. Disturbances of precision grip in huntingtons disease. Neurosci Lett 1997 Apr; 226: 103–106.

Kieburtz

Reilmann

Olanow

. Huntingtons disease: current and future therapeutic prospects. Mov Disord 2018 May; 33: 1033–1041.

Wild

Tabrizi

. Therapies targeting DNA and RNA in huntingtons disease. Lancet Neurol 2017 Oct; 16: 837–847.

Reilmann

Rouzade-Dominguez

M-L

Saft

, et al. A randomized, placebo-controlled trial of AFQ056 for the treatment of chorea in huntington’s disease. Mov Disord 2015 Mar [cited 2018 May 17]; 30: 427–431.

Reilmann

McGarry

Grachev

, et al. Safety and efficacy of pridopidine in patients with huntingtons disease (PRIDE-HD): a phase 2, randomised, placebo-controlled, multicentre, dose-ranging study. Lancet Neurol 2019 Feb; 18: 165–176.

Reilmann

Anderson

Feigin

, et al. Safety and efficacy of laquinimod for huntington’s disease (LEGATO-HD): a multicentre, randomised, double-blind, placebo-controlled, phase 2 study. Lancet Neurol 2024 Mar; 23: 243–255.

Reilmann

Schubert

. Chapter 18 - motor outcome measures in Huntington disease clinical trials. In: Feigin

Anderson

(eds) Handbook of clinical neurologyHuntington Disease144. Amsterdam, Netherlands: Elsevier, 2017, pp.209–225.

10.

Tabrizi

Langbehn

Leavitt

, et al. Biological and clinical manifestations of huntington’s disease in the longitudinal TRACK-HD study: cross-sectional analysis of baseline data. Lancet Neurol 2009 Sep [cited 2020 Apr 9]; 8: 791–801.

11.

Gordon

Quinn

Reilmann

, et al. Coordination of prehensile forces during precision grip in huntingtons disease. Exp Neurol 2000 May; 163: 136–148.

12.

Reilmann

Bohlen

Kirsten

, et al. Assessment of involuntary choreatic movements in huntingtons disease-toward objective and quantitative measures. Mov Disord 2011 Jun; 26: 2267–2273.

13.

Medzech

Sass

Bohlen

, et al. Impaired isometric force matching in upper and lower limbs revealed by quantitative motor assessments in huntington’s disease. J Huntington’s Dis 2019 Oct; 8: 483–492.

14.

Rohrbach

Buettner

Pollex

, et al. Quantitative examination of isometric tongue protrusion forces in children with oro-facial dysfunctions or myofunctional disorders. J Oral Rehabil 2018 Jan; 45: 228–234.

15.

Klehmet

Beutner

Hoffmann

, et al. Quantitative grip force assessment of muscular weakness in chronic inflammatory demyelinating polyneuropathy. BMC Neurol. 2019 Dec [cited 2020 Apr 21]; 19: 118.

16.

Langbehn

Brinkman

Falush

, et al. A new model for prediction of the age of onset and penetrance for huntingtons disease based on CAG length. Clin Genet 2004 Mar; 65: 267–277.

17.

Bechtel

Scahill

Rosas

, et al. Tapping linked to function and structure in premanifest and symptomatic huntington disease. Neurology 2010 Dec; 75: 2150–2160.

18.

R Core Team . R: A language and environment for statistical computing [Internet]. Vienna, Austria: R Foundation for Statistical Computing, 2021. https://www.R-project.org/ (accessed 21 February 2025).

19.

Wickham

Averick

Bryan

, et al. Welcome to the tidyverse. J Open Source Software 2019; 4: 1686.

20.

Hothorn

Buehlmann

Dudoit

, et al. Survival ensembles. Biostatistics 2006; 7: 355–373.

21.

Strobl

Boulesteix

A-L

Kneib

, et al. Conditional variable importance for random forests. BMC Bioinformatics 2008; 9: 307.

22.

Wilcox

Schönbrodt

. The WRS package for robust statistics in R (version 0.38.1). 2021. https://github.com/nicebread/WRS (accessed 21 February 2025).

23.

Wilcox

. Comparing dependent robust correlations. Br J Math Statistic Psychol 2016 Oct; 69: 215–224.

24.

Tabrizi

Schobel

Gantman

, et al. A biological classification of Huntington's disease: the Integrated Staging System. Lancet Neurol 2022; 21: 632–644.

25.

Tabrizi

Reilmann

Roos

RAC

, et al. Potential endpoints for clinical trials in premanifest and early Huntington’s disease in the TRACK-HD study: analysis of 24 month observational data. Lancet Neurol 2012 Jan; 11: 42–53.

26.

Tabrizi

Scahill

Owen

, et al. Predictors of phenotypic progression and disease onset in premanifest and early-stage huntington’s disease in the TRACK-HD study: analysis of 36-month observational data. Lancet Neurol 2013 Jul; 12: 637–649.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.57 MB