The effects of dual-task cognitive interference on gait and turning in Huntington’s disease

Huntington’s disease (HD) is characterized by motor, cognitive, and psychiatric dysfunction. HD progression causes loss of automaticity, such that previously automatic tasks require greater attentional resources. Dual-task (DT) paradigms and fast-paced gait may stress the locomotor system, revealing deficits not seen under single-task (ST). However, the impact of gait “stress tests” on HD individuals needs further investigation. Therefore, the aims of this study were to investigate whether: 1) fast-paced and dual-task walking uncover deficits in gait and turning not seen under single-task, 2) cognitive and gait outcomes relate to fall incidence, and 3) gait deficits measured with wearable inertial sensors correlate with motor symptom severity in HD as measured by the Unified Huntington’s disease Rating Scale-total motor score (UHDRS-TMS). Seventeen HD (55 ± 9.7 years) and 17 age-matched controls (56.5 ± 9.3 years) underwent quantitative gait testing via a 25m, two-minute walk test with APDMTM inertial sensors. Gait was assessed under a 1) ST, self-selected pace, 2) fast-as-possible (FAP) pace, and 3) verbal fluency DT. The UHDRS-TMS and a cognitive test battery were administered, and a retrospective fall history was obtained. During ST, DT, and FAP conditions, HD participants demonstrated slower gait, shorter stride length, and greater lateral step and stride length variability compared to controls (p<0.00001 to 0.034). Significant dual-task costs (DTC) were observed for turns; HD participants took more time (p = 0.013) and steps (p = 0.028) to complete a turn under DT compared to controls. Higher UHDRS-TMS correlated with greater stride length variability, less double-support, and more swing-phase time under all conditions. Decreased processing speed was associated with increased gait variability under ST and FAP conditions. Unexpectedly, participant’s self-reported falls did not correlate with any gait or turn parameters. HD participants demonstrated significantly greater DTC for turning, which is less automatic than straight walking, requiring coordination of body segments, anticipatory control, and cortical regulation. Turn complexity likely makes it more susceptible to cognitive interference in HD.

Introduction recruited from the community. All participants were co-recruited for a previously published study assessing the effects of dual-tasking and cognition on balance in HD [14]. Inclusion criteria were 1) a clinical diagnosis of HD by a movement disorders/HD expert (JGG) [19], 2) >21 years of age, 3) ability to ambulate for two minutes without an assistive device, and 4) the ability to follow protocol-specific directions with confirmation from family member and/or caregiver. Exclusion criteria included a diagnosis of Juvenile HD, as well as those who have had lower limb or back orthopedic surgery in the past 12 months, or any other disorders negatively affecting gait. Controls were recruited based on the same criteria, with the additional exclusion of cognitive impairment. Participants were classified as having a choreatic, hypokinetic-rigid, or mixed phenotype, as previously described [20]. Informed consent was obtained from all participants in accordance with the RUMC Institutional Regulatory Board.

Gait assessments
Quantitative gait analysis under self-selected single task (ST), fast-as-possible (FAP) and dualtask (DT) conditions was performed using the well validated, reliable inertial sensor system with gait metrics generated by Mobility Lab TM software (APDM, Oregon, USA) [21]. Six Opal TM wearable sensors were placed on the wrists, dorsum of feet, sternum (2 cm below the sternal notch), and lumbar trunk (at the L5, the approximate center of mass). Participants performed three, 2-minute walk tests [22,23] on a 25-meter walkway under a 1) self-selected (SS) pace, 2) fast-as-possible (FAP) pace, and 3) cognitive-motor DT condition (DT) at their normal pace, with rests between trials as needed. Participants were instructed to walk at their "normal", comfortable walking speed for the self-selected and DT trials, and to walk as fast as they safely could, without running, for the FAP trial. During the DT trial, participants were asked to perform an animal naming verbal fluency task, with the instruction that no animal could be repeated. Participants were carefully monitored during all trials for safety. A baseline 2-minute animal naming task was also performed while seated, and this was randomized to be conducted either before or after the gait tests for each participant. This randomization was done to assess whether participants prioritized cognition or gait while dual-tasking. The participants were not told to prioritize the cognitive or motor task during the DT gait condition.
The main outcome variable selected for analysis were 1) cadence (steps/min), 2) stride length (m), 3) gait speed (m/s), 4) swing (% gait cycle), 5) double support (% gait cycle), 6) turn duration (s), 7) number of steps to complete a turn, 8) lateral step variability (m), which groups 3 consecutive steps and derives the extent of perpendicular deviation of the middle foot placement from the first and third step, 9) stride length coefficient of variation (CoV), another common measure of gait variability, and 10) step duration. The extent of DT interference, or the dual-task cost (DTC) in gait and turn performance was defined as DTC (%) = ((DT-ST)/ ST) � 100, as previously described [24].
Higher UHDRS-TMS values reflect the presence of more severe motor symptoms. The Berg Balance Scale (BBS) [31] and the Activities-Specific Balance Confidence Scale (ABC) [32] were used to assess participants' awareness of their postural stability and collect performance-based balance information. Higher scores on all cognitive tests and functional balance scales are indicative of better performance. Additionally, participants were asked to recall how many falls they had in the past 12 months.

Statistical analyses
Clinical characteristics were compared between the HD and healthy control group using twotailed Student t-tests for parametric and normally distributed measures, or the Mann-Whitney U test for variables that did not have normal distributions. Differences in gait variables and animal naming cognitive assessment under ST and DT conditions, and the DTC for each of the primary outcome variables between HD participants and healthy controls were examined with the same statistical tests. For those gait variables observed to be significantly impaired in HD under all conditions compared to controls, a repeated-measures analysis of variance (ANOVA) (parametric) or a Friedman test (non-parametric) with Bonferroni post-hoc tests were performed to assess if those impairments were further exacerbated under FAP and DT conditions.
Correlations between gait and turn measures and cognitive test scores, UHDRS-TMS, and retrospective falls were examined in the HD group using Spearman's rho. The statistical significance for these comparisons was set at p = 0.05 given the exploratory nature of this work. However, due to the large number of variables tested and correlations performed, the correlations that remained significant after Bonferroni corrections using an adjusted p-value of < 0.001 were also indicated with a 'b' superscript. An exploratory linear regression analysis was then performed incorporating disease duration and UHDRS-TMS as covariates to further investigate potential cognitive and gait relationships.

Participant characteristics
Seventeen individuals with HD and 17 age-matched healthy controls were enrolled in a larger dual-task study examining the effects of cognition on balance and gait in Huntington's disease; participant demographic and clinical characteristics are listed in Table 1. UHDRS-TMS scores ranged from 7-39. Five of the seventeen HD participants were not on any medications at the time of testing. The most common medications reported were an NMDA antagonist (n = 4), benzodiazepine (n = 4), cholinesterase inhibitor (n = 4), antipsychotic (n = 2), and selective serotonin reuptake inhibitors (SSRI) (n = 2). No participants were taking vesicular monoamine transporter 2 (VMAT2) inhibitors. Eight HD participants were characterized as having a mixed phenotype, 4 as choreatic, and 4 as hypokinetic-rigid, one participant did not have a UHDRS-TMS recorded. Ten out of the 17 HD individuals reported having >1 fall in the past twelve months. HD participants showed significant deficits in global cognition (MoCA, p = 0.0009), response inhibition (Stroop, p = 0.007), attention/processing speed (SDMT, p<0.0001), verbal fluency (animal naming ST and DT, p<0.0001), visuospatial processing (JLO, p = 0.0083) and working memory (Digit Span, p = 0.0087) compared to controls. HD participants' delayed recall (CERAD Word List) performance was not significantly different from controls. There were no significant DTC for the animal naming cognitive task during the two-minute walk (DT) compared to the ST while seated in HD subjects compared to controls. HD participants reported lower balance confidence on the ABC (p = 0.0001), worse scores on the BBS (p<0.0001), and a greater number of falls within the past twelve months (p = 0.0007) compared to controls (Table 1).

Gait: Single task, fast-as-possible, and DT conditions
Independent sample t-tests indicated there was not a significant difference between the left and right foot for all gait parameters; therefore, the average was calculated and used for all future analyses. Four spatiotemporal gait parameters were consistently observed to be significantly different between the HD and control groups under all three conditions (SS, FAP, DT); these were gait speed, stride length, lateral step variability, and stride length variability (CoV) (Fig 1). HD participants exhibited significantly slower gait speed (p = 0.034, 0.0005, 0.004), shorter stride length (p = 0.0004, 0.029, 0.005), and greater lateral step variability (p < 0.00001, 0.003, < 0.00001) and stride length variability (p = 0.00001, 0.001, 0.00001) compared to controls (Table 2). Additionally, HD individuals took longer to complete a turn compared to controls under the FAP condition (p = 0.045). Compared to SS pace trials, HD individuals were able to significantly increase their gait speed under FAP conditions (p < 0.00001); no difference was observed during DT. Stride length was observed to increase during FAP conditions compared to SS (p < 0.00001); no difference was seen under DT. Lateral step variability increased during DT conditions compared to SS (p = 0.0026); no difference was seen under the FAP condition. After correcting for multiple comparisons, stride length variability was not significantly different between conditions. No significant differences were found for turning across conditions in HD. Although controls increased their gait speed (p < 0.00001) and stride length (p = 0.00002) during FAP trials compared to SS trials, no significant differences were  Dual-task gait in Huntington's disease observed for lateral step and stride length variability between conditions and no differences were observed between DT and SS conditions. The only significant DTC in HD participants were found for turn variables such that the HD group exhibited significantly greater turn durations (p = 0.013) (Fig 2A) and more steps to complete a turn (p = 0.029) (Fig 2B), than controls.

Correlations between gait, cognition, UHDRS-TMS, and falls
Correlations with cognitive tests were limited to our cognitive domains of interest: executive function (Digit Span, Stroop and animal naming), information processing speed (SDMT), and visuospatial processing (JLO). Poorer performance on the SDMT and animal naming was significantly associated with increased gait variability in individuals with HD. Lower SDMT scores were associated with increased lateral step variability under SS pace (r = -0.623, p = 0.008) (Fig 3A), as well as increased stride length CoV under SS (r = -0.547, p = 0.023) ( Fig  3B) and FAP (r = -0.725, p = 0.001) (Fig 3C) conditions. Furthermore, poorer performance on animal naming was associated with greater stride length CoV during the FAP trials (r = 0.706, p = 0.002). When the significant cognitive variables were entered into the exploratory regression model controlling for disease duration and UHDRS-TMS, the significant associations between cognition and gait were no longer observed.
UHDRS  (Table 3). No gait or cognitive variables were found to be correlated with the number of falls self-reported in the past 12 months. All p-

Discussion
Our study found that cognitive interference is significant when individuals with HD are turning during the gait cycle. These results highlight the complexity of turning and how modifying gait patterns to complete a turn requires both motor and cognitive resources [33]. Turning is believed to rely on frontal lobe functioning significantly more than straight walking, so that the negative effects of cognitive impairment on the postural adjustments of turning are more pronounced [33]. Additionally, turning necessitates more inter-limb and spinal segment coordination and is greatly impacted by cognitive functioning, such as processing speed [34,35], which is notably impaired in HD [10]. This knowledge is in alignment with our finding of significant cognitive interference during turns in HD. These results expand upon our previous findings where individuals with HD exhibited significant cognitive interference under environmental conditions that markedly challenge postural stability [14]. In addition, a previous study in HD indicated that multitasking was the most reported cause of falls [36], supporting the theory that cognitive deficits, in combination with loss of automaticity, can result in decreased postural stability in HD, especially during the postural adjustments required for turning.
The fact that there were no elevated DTC for spatiotemporal aspects of gait or cognition during straight walking suggests that the HD group as a whole subscribed to the "posture- Dual-task gait in Huntington's disease second" strategy of dual-tasking observed in PD, where both elements of the DT are treated with equal attention [37][38][39]. Employing this strategy becomes an issue though, due to limited cognitive resources and impaired postural control, resulting in neither task being adequately accomplished [37].
Interestingly, the same four spatiotemporal gait parameters were consistently found to be impaired in HD compared to healthy controls under all testing conditions: gait speed, stride length, lateral step variability, and stride length variability. The gait domains of pace and variability are commonly reported as abnormal in previous gait studies in other movement disorders [8,[40][41][42]. However, a number of gait variables are included in these domains and a set of sensitive variables for the HD population has not been validated. Lateral step variability is not a commonly reported outcome measure, although our study suggests it should be investigated in the future, as it was a notable abnormal feature of HD participants. Gait variability is reported to be increased in HD [36,43,44] and is thought to be a result of a disruption of the basal ganglia's cueing mechanism to the supplementary motor cortex [18]. Disrupted cueing can then cause inefficient generation of movement timing and greater movement variability, increasing fall risk [18]. These findings lay the groundwork for future, more targeted gait studies in HD, suggesting that gait speed, stride length, stride length variability, and lateral step variability could all be sensitive outcome measures for future clinical trials.
We found that DT gait in HD, but not healthy controls, increases lateral step variability. Increased gait variability has been shown to be associated with increased gait instability and Table 3. Relationship between UHDRS-TMS and gait variables. Step Duration r = 0.203 p = 0.451

Self-Selected UHDRS-TMS Fast as Possible UHDRS-TMS
Step Duration r = -0.053 p = 0.845 Step Dual-task gait in Huntington's disease falls in other populations [45]. Although we did not find gait variability parameters to be associated with falls in the present study, perhaps due the low sample size, further prospective studies with larger subject numbers might find that increased gait variability may potentially serve as a marker or predictor of future fallers and thus have clinical utility.
A reduction in attention and information processing speed, as assessed by the SDMT, were found to be associated with greater gait variability in HD. Prior studies have indicated that lower processing speed is associated with impaired stability and increased falls in MS [46], impaired turning in PD [35], slower gait speed in an aging population [47], and worse gait and balance as assessed by the Tinetti Mobility Test in HD [9]. However, our observed association between cognition and gait was no longer significant after controlling for disease duration and UHDRS-TMS in our exploratory regression analysis. We posit that this was due to our low sample size. Future studies with larger subject numbers, as well as obtaining a total UHDRS score in order to have a more accurate measure of disease severity, will hopefully allow us to determine the impact of cognition and disease severity on gait deficits in HD.
Similar to the findings of our balance study in HD [14], we did not find the number of retrospective falls to correlate with any cognitive and gait parameters in HD individuals. We believe the lack of retrospective fall correlations can be attributed to the small sample size of this study. Additionally, retrospective fall reporting relies on a participant's memory recall and self-awareness and individuals with HD often exhibit a lack of disease insight, making a selfreported fall history prone to under-reporting. Therefore, prospective recording, caregiver corroboration, or a fall detection monitoring device would provide a more accurate fall report. Subsequent studies will include a larger cohort with more accurate means of reporting falls or prospective fall reporting to strengthen this analysis.
Previous studies in HD, PD, and cerebellar ataxia reported that participants spent more time in double support and stance phase and less time in swing phase [48][49][50][51][52][53] as a possible compensatory measure to maintain postural stability [52,53]. However, we did not find this to be the case in our HD cohort; swing and double support time were not significantly different from controls under any gait condition. We attribute these findings to the variability of motor symptoms in our HD group.
Increased motor symptom severity was associated with greater gait variability suggesting those with greater motor impairment have a more unstable gait. Higher UHDRS-TMS, indicating greater motor symptom severity, were significantly correlated with less time in double support and more time in swing phase. These results are difficult to explain but suggest further investigation into the relationship between gait and motor severity. Future studies utilizing wearable inertial sensors might aid in better characterization of choreatic gait and the poorly defined "stutter step" gait pattern exhibited by some individuals with HD [54]. The relationship we observed between gait and UHDRS-TMS is difficult to interpret due to the variability of gait impairment within our sample. Therefore, more studies with larger subject numbers and stratification based on motor severity are needed to create a thorough spatiotemporal and kinematic profile of HD gait and how gait relates to HD motor severity.
The strengths of this study include: 1) the use of a sensitive inertial sensor system to assess gait in HD under challenging conditions reflective of everyday scenarios, 2) significant contributions to the growing body of work characterizing the impact of cognition and cognitive dual-tasking on turns during ambulation in HD and 3) the use of an extensive neuropsychological test battery to assess multiple cognitive domains and their interaction with gait deficits. This study is not without limitations. As previously mentioned, future studies will have larger subject numbers and include participants with varying severities of motor and cognitive symptoms to potentially stratify HD participants and examine phenotypic differences in dual-task capabilities and costs. Future studies would also benefit from complementary neuroimaging or neurophysiological data to understand the neural mechanisms underlying gait control in HD and how basal ganglia cortical connectivity, volumetric changes, and/or activation patterns relate to turn deficits and dual-tasking.

Conclusion
In conclusion, individuals with HD exhibit detrimental effects of cognitive interference while turning, highlighting the complexity of turning and the dynamic motor and cognitive coordination necessary to safely complete a turn. Additionally, impaired attention and processing speed was associated with more gait variability and is a domain that should be investigated further as an indicator of fall risk in HD. More thorough studies need to be done to quantitatively characterize the choreatic gait pattern and distinguish it from other movement disorders, improving clinical gait assessment in patients with HD.