Older Adults with Amnestic Mild Cognitive Impairment Exhibit Exacerbated Gait Slowing under Dual-Task Challenges

Abstract

Background

With age, performance of motor tasks becomes more reliant on cognitive resources to compensate for the structural and functional declines in the motor control regions in the brain. We hypothesized that participants with amnestic mild cognitive impairment (aMCI) are more prone to motor dysfunctions than cognitively normal older adults under dual-task conditions where competitive demands challenge cognitive functions while performing a motor task simultaneously.

Methods

Sixteen aMCI participants (females=9, age=64±5yrs, clinical dementia rating score=0.5) and 10 age- and education-matched cognitively normal adults (females=5, age=62±6yrs) participated. Using a 10-meter-walk test (10MW), gait velocity was recorded at baseline and under 4 different dual-task (DT) conditions designed to challenge working memory, executive function, and episodic memory. Specifically, DT1: verbal fluency; DT2: 5-digit backward span; DT3: serial-7 subtraction; and DT4: 3-item delayed recall. Physical function was measured by Timed Up-and-Go (TUG), simple reaction time (RT) to a free-falling yardstick, and functional reach (FR).

Results

No difference was found in physical functions, aerobic fitness, and exercise cardiopulmonary responses between aMCI participants and controls. However, aMCI participants showed more pronounced gait slowing from baseline when compared to the controls (p<0.05; p=0.001; p<0.001; p<0.001, respectively).

Conclusions

Our finding supports the theory of shared resource of motor and cognitive control. Participants with aMCI manifested more gait slowing than cognitively-normal older adults under DT conditions, with the largest differences during tests of working and episodic memory. The outcome of dual-task assessment shows promise as a potential marker for detection of aMCI and early Alzheimer disease.

1. INTRODUCTION

Both physical [1] and many cognitive functions [2] decline with aging, and mounting evidence suggests that older adults require more cognitive resources to perform even simple motor tasks [3-5]. More specifically, with age, performance of motor tasks becomes more reliant on the cognitive resources to compensate for the structural and functional declines of the motor control network including the primary and secondary motor cortex, prefrontal lobe, cerebellum and basal ganglia pathways [6]. Recent neuroimaging studies have demonstrated increased recruitment of the prefrontal cortex activity measured by blood-oxygen-level-dependent (BOLD) signal in older adults when performing simple hand and foot movements, suggesting an increased demand on cognitive function when performing motor tasks [4, 7, 8]. Consequently, motor slowing may occur when the supply of cognitive resource is limited and/or unable to meet the competitive demands of required tasks [9].

Conventionally, gait was considered as an automated motor function with minimal involvement of higher-level cognitive input. However, associations between executive function and motor function performance have been observed in non-demented older adults [10], and a link between executive function and gait disturbances has been identified [11]. Executive function (EF) refers to higher level cognitive functions that process information collected from multiple cortical systems to regulate behavior [12]. These functions are crucial for performing goal-oriented behaviors throughout daily activities. With age, executive function appears to deteriorate as the frontal lobes are more susceptible to age-related changes [13]. Along these lines, it has been shown that reductions in gait velocity are an important risk factor for cognitive decline and may be a harbinger of dementia in older adults [14, 15]. Furthermore, reduction in gait velocity, increases in gait variability, balance impairment, and falls are observed more frequently in older adults with cognitive dysfunction, potentially implicating the role oflimited shared resources between physical and cognitive function [16].

The term amnestic mild cognitive impairment (aMCI) has been used to describe a transitional phase between normal aging and Alzheimer's disease (AD),and patients with aMCI are at a greater risk of developing AD [17]. Based on the above factors, we hypothesized that cognitive impairment in aMCI may be associated with more pronounced motor (gait) slowing during dual-task conditions [18] in which cognitive function are challenged while walking. Previous studies have shown that decline in gait velocity under dual-task conditions is a valuable tool to predict falls in the elderly [3], but little is known about the effects of cognitive impairment in relation to gait slowing. This study explored the possibility whether changes in gait velocity under dual-task conditions also can be used as a potential marker to distinguish aMCI patients from normal control subjects.

2. METHODS

2.1. Subjects

The Institutional Review Board of the University of Texas Southwestern Medical Center (UTSW) and Texas Health Presbyterian Hospital Dallasapproved this study. Informed consent was obtained from all study participants. Two groups were included – participants with aMCI and non-MCI controls. Subjects with aMCI were part of a larger study in people with aMCI (Clinical Trial ID: NCT01146717). The primary recruitment mechanisms included: 1) emails and newsletters of the Texas Health Prebyterian Hospital Dallas, part of the Texas Health Resources hospital network in the Dallas-Ft.Worth metropolitan region; and 2) advertisement through local media and health fairs. All enrollment and testing were performed at our facilities, making this a single-site study. The diagnosis of aMCI was based on the Petersen criteria [19] as modified by the Alzheimer's Disease Neuroimaging Initiative (ADNI) (http://adni-info.org),i.e., a Clinical Dementia Rating (CDR) score of 0.5 (memory box score ≥0.5), objective evidence of memory impairment as measured by the Wechsler Memory Scale-Revised Logical Memory subtest [20], a subjective memory complaint, an absence of other cognitive impairments, and preserved activities of daily living. The Mini-Mental State Exam (MMSE) [21] was performed to assess global cognitive function. This included clinical evaluations performed at the UT Southwestern Medical Center Alzheimer's Disease Center by board-certified neurologists and neuropsychologists to exclude other conditions that may cause memory problems and to exclude subjects with non-amnestic MCI. All subjects were free of major medical problems based on a detailed medical history and physical exams including 12-lead electrocardiogram (ECG) and echocardiogram. Subjects were excluded if they were smokers or used recreational drugs. They were also excluded if they had 1) clinical evidence of cerebrovascular diseases (e.g. history of stroke, transient ischemic attack or the presence of cortical infarct on MRI), 2) met criteria for dementia or non-amnestic MCI, 3) DSM-IV-TR Axis I psychopathology, 4) other neurologic disorder such as traumatic brain injury, 5) diabetes mellitus (fasting plasma glucose>126 mg/dL or Hemoglobin A1C>7.0%), 6) uncontrolled hypertension (systolic pressure>160 and/or diastolic pressure>110 mmHg), or 7) other cardiovascular diseases.In addition, to ensure that subjects were similar in terms of activity levels and rule out conditions which may affect gait velocity, participants were excluded if they had 8) physical disability which prevented them from walking independently without an assistive device, 9) vigorous physical activity > 90 minutes/week, or 10) were taking anticholinergic, antipsychotic or other drugs that may influence cognitive or motor function. All participants were carefully screened and otherwise healthy older adults with a relatively sedentary lifestyle according to the American College of Sports Medicine and American Heart Association (ACSM/AHA) guidelines for physical activity recommendations for older adults [22].

2.2. Design

This was a cross-sectional study.

2.3. Dual-Task (DT) Conditions

All subjects were asked to refrain from strenuous exercise, caffeine or alcohol at least 12 hours prior to testing. Using a 10-meter-walk test (10MW) [23], gait velocity was recorded at baseline. Subjects were instructed to walk at a comfortable pace; practices were given until subjects were able to consistently walk at their comfortable speed [24]. They were then asked to repeat the 10MW under 4 different dual-task (DT) conditions designed to challenge working memory and executive function as they walked. The order of DT conditions was randomized, and tasks were selected to reflect tasks commonly used in neuropsychological or cognitive screening evaluations [25]. Specifically, DT-1 required subjects to generate as many words as they could that begin with letter “C” during the 10MW (letter “C” verbal fluency). Cognitive performance was assessed by calculating words generated per second. In DT-2, subjects were asked to remember a 5-number digit string and recite it backward during the 10MW (5-digit backward span). Subject was awarded 1 point for each digit in the right order (max score = 5). In DT-3, subjects were asked to perform a serial 7 subtraction task (i.e., subtracting 100-7= 93; 93-7= 86, etc.) continuously during the 10MW.Each correct subtraction was awarded 1 point. DT-4 required subjects to recall 3 words (door, nickel, bus) that they were asked to memorize 15 minutes prior to the test and to rearrange the words in alphabetical order and say the words immediately after they were instructed to begin the 10MW. Subjects were awarded 1 point for each word correctly recited, and an additional point for each word in the right order (max score = 6).

2.4. Physical Function Assessment

All subjects completed a set of physical function tests including timed-up-and-go, reaction time, and functional reach to assess motor control and balance performance. The reliability and reproducibility of these tests have been validated extensively in older adults as well as in clinical populations [26-28]. The timed-up-and-go test is a simple test used to assess a person's mobility and requires both static and dynamic balance. It requires subjects to rise from a chair without using their arms, walk three meters, turn around, walk back to the chair, and sit down [26]. The simple reaction time test (commonly used in sports medicine and physical therapy settings) required subjects to stop a free-falling yardstick against wall before it could reach the ground when released at shoulder height at random times. Distance dropped was measured in centimeters [27]. Functional reach was tested by placing a yardstick on the wall, parallel to the floor, at the height of the acromion of the subject's dominant arm. The subjects were asked to stand with the feet a comfortable distance apart, and forward flex the dominant arm to approximately 90 degrees. The subjects were then asked to reach forward as far as possible without taking a step or touching the wall. The distance between the start and end point was measured [28].

2.5. Measurement of Aerobic Fitness

Subjects underwent maximal exercise testing on a separate visit. Maximal oxygen uptake (VO_2max) was assessed using a modified Astrand-Saltin protocol involving incremental exercise on a treadmill [29]. Maximal oxygen uptake (VO_2max) was defined as the highest oxygen uptake (VO₂) measured from at least a 40-second Douglas bag during the last stage of testing. The criteria to confirm that VO_2max was achieved included 1) an increase in VO₂ < 150 ml/min, despite increasing work rate of 2% grade (plateau); 2) a respiratory exchange (RER) ratio > 1.1; 3) HR within 5 beats/min of age-predicted maximal values (220–age); and 4) blood lactate > 8.0mmol/l. In all cases, at least three of these criteria were achieved, confirming the identification of VO_2max per the ACSM guidelines [30]. During exercise testing, arterial pressure (Tango+, Suntech), electrocardiogram (Mortara, Mortara Instrument), cardiac output (C2H2 rebreathing method) and ventilation were measuredcontinuously to monitor cardiopulmonary responses.Of note, our previous studies have demonstrated that by using these methods, VO_2max can be measured reliably in sedentary elderly subjects [31, 32].

2.6 Statistical Analysis

Statistical analysis was performed using Sigma Plot 11.0 (SSTI, San Jose, CA). Student's t-tests were conducted to detect group differences in physical and aerobic fitness under baseline conditions. Two-way repeated measures analysis of variance (AVOVA) was used for between and within group comparisons of gait velocity and cognitive performance under dual-task conditions using Tukey test for post-hoc pair wise comparisons. A statistically significant level is considered as P < 0.05.Student's t-tests were also used to detect the group differences in percentage changes in gait velocity from the baseline under dual-task conditions. Since these measurements (% changes in gait velocity under dual-task conditions) were the primary outcome of this study, Bonferroni corrections were conducted for multiple comparisons (P<0.0125) to strengthen the conclusions of this study.

3. RESULTS

Subject characteristics are presented in Table 1. No differences were found in age, gender distribution, or education between groups. Groups were also similar in mean MMSE score, and as expected, aMCI subjects scored lower on Logical Memory than controls. The groups did not differ in terms of physical functions or aerobic fitness (Table 2). Using twoway repeated measures ANOVA, we found significant interactions between group and dual-task conditions (p<0.001, F=13.49). Interestingly, although no difference was found in baseline gait velocity (m/s) between groups, aMCI participants exhibited slower gait velocity (m/s) under all DT conditions (Table 3). In addition, aMCI participants also showed lower cognitive performance in 3 of 4 dual-task conditions as shown in Table 4 (i.e. DT2 = digits in the right order, max score = 5, DT3 = number of times of correct subtractions, DT4 = sum of words recalled correctly + ordered correctly, max score = 6).

Table 1.

Subject characteristics.

	aMCI (n = 16)	Control (n = 10)
Male / Female	7 / 9	5 / 5
Age (years)	64.4 ± 5.3	61.8 ± 6.5
Education (years)	16.2 ± 2.7	15.9 ± 2.5
MMSE	28.8 ± 1.5	28.9 ± 1.0
Height (cm)	169.1 ± 5.7	168.4 ± 6.9
Weight (kg)	75.3 ± 9.2	70.4 ± 10.5
BMI (kg/m2)	26.4 ± 3.5	25.0 ± 4.6
Logical Memory (immediate)	11.8 ± 2.6*	15.6 ± 4.8
Logical Memory (delayed)	9.8 ± 2.2*	13.6 ± 4.5

Values are means ± SD.

^*aMCI subjects significantly lower than controls, P<0.05. MMSE = Mini Mental State Exam 3MI = body mass index

Table 2.

Physical fitness.

	aMCI	Control
Timed Up-and-Go (sec)	9.31 ± 1.34	9.50 ± 1.06
Reaction Time (cm dropped)	37.9 ± 15.6	37.8 ± 9.2
Functional Reach (cm displaced)	29.7 ± 6.5	28.4 ± 9.2
VO2max (ml/kg/min)	23.4 ± 5.9	24.3 ± 4.5

Values are mean ± SD. Timed Up-and-Go (sec) = lower value indicates faster gait speed Reaction Time (cm dropped) = lower value indicates faster reaction time Functional Reach (cm displaced) = higher value indicates better functional balance VO_2max (ml/kg/min) = higher value indicates better aerobic fitness

Table 3.

Gait velocity (GV) under dual-task conditions (DT).

	aMCI	Control	P-value
Baseline GV (m/s)	1.06 ± 0.11	1.14 ± 0.19	0.262
DT-1 (verbal fluency) GV (m/s)	0.92 ± 0.10	1.06 ± 0.20	0.058
DT-2 (5-digit backward span) GV (m/s)	0.91 ± 0.13	1.08 ± 0.20	0.017
DT-3 (serial-7 subtraction) GV (m/s)	0.88 ± 0.13	1.08 ± 0.22	0.005
DT-4 (delayed recall & sorting) GV (m/s)	0.85 ± 0.12	1.06 ± 0.19	0.002

Values are mean ± SD. Two-way repeated measures ANOVA with Tukey test for pairwise comparisons. Higher values indicate faster gait velocity. P values are for group comparisons under the same dual task conditions. P<0.05 was used to indicate a statistically significant comparison.

Table 4.

Cognitive performance under dual-task conditions (DT).

	aMCI	Control	P-value
DT-1 (verbal fluency)	0.30 ± 0.11	0.41 ± 0.15	0.855
DT-2 (5-digit backward span)	2.19 ± 2.10	4.10 ± 1.20	0.003
DT-3 (serial-7 subtraction)	1.81 ± 1.97	3.80 ± 1.32	0.002
DT-4 (delayed recall & sorting)	3.69 ± 2.21	5.80 ± 0.42	0.001

Values are mean ± SD. Two-way repeated measures ANOVA with Tukey test for pairwise comparisons. Higher values indicate better cognitive performance. P values are for group comparisons under the same dual task conditions. P<0.05 was used to indicate a statistically significant comparison. DT1 = correct words generated per second DT2 = digits in the right order, max score = 5 DT3 = number of times of correct subtractions DT4 = sum of words recalled correctly + ordered correctly, max score = 6

More importantly, after normalizing to baseline velocity (expressed in %), aMCI participants showed pronounced gait slowing compared to controls in 3 of 4 dual-task conditions as shown in Fig. (1) (DT1: –12.9±6.1 vs. –8.0±3.8%, p<0.05; DT2: –14.8±5.4 vs. –5.2±4.9 %, p<0.001; DT3: -17.0±7.6 vs. –5.5±6.0 %, p=0.001; DT4: –20.3±6.4 vs. – 5.9±4.7 %, p<0.001). In addition, Logical Memory Delayed Recall score showed a significant correlation with normalized gait velocity decline in DT-2 (r=0.57, p<0.005), DT-3 (r=0.53, p<0.01), and DT-4 (r=0.62, p<0.005). Specifically, greater decline in gait velocity is associated with lower performance on Logical Memory Delayed Recall.

Fig. (1).

Changes in gait velocity relative to baseline in aMCI participants and controls.aMCI group exhibited more gait slowing than controls under dual-task (DT) conditions (DT1: p<0.05; DT2: p<0.001; DT3: p=0.001; DT4: p<0.001). P<0.0125 (Bonferonni corrected) was used to indicate a statistically significant comparison. Error bars = standard error of the mean.

4. DISCUSSION

Our data revealed that aMCI participants showed more gait slowing than normal older adults under dual-task conditions. The largest group differences in gait slowing were observed during cognitive challenges of concentration/working memory and episodic memory. In addition, aMCI participants showed lower cognitive performance scores on tasks of 5-digit backward span, serial-7 subtraction, and 3-item recall/ sort. Moreover, we found that the Logical Memory Delayed Recall performance was negatively associated with gait velocity declines on tasks requiring working or episodic memory. Notably, there was no difference in baseline gait velocity, timed up-and-go, reaction time, functional reach performance, and aerobic fitness between aMCI participants and normal controls, suggesting that the accentuated gait slowing in aMCI participants under dual-task conditions is likely to be mediated by a central mechanism. Taken together, these findings support the theory of shared resource of motor and cognitive control in older adults.

With age, cognitive declines are manifested mainly in the domains of executive function and working memory (fluid cognitive ability) which may be related to the deteriorations in attention, mental control, processing speed, and visual-spatial processing [33]. Older adults also exhibit declines in physical functions such as motor control, balance, and gait velocity when compare to young adults [34]. These age-related declines in cognitive and physical function are likely to be inter-correlated and may negatively affect the individual's ability to perform daily activities and jeopardize the independence later in life [35]. Recent studies have shown that gait velocity in patients with MCI and AD was reduced when compared with cognitively normal controls under simple walking conditions and during dual-tasks of verbal fluency and serial-seven subtractions [24, 36]. However, comorbidities and physical fitness level were not controlled, which may have confounded the observed changes in gait velocity in these studies [24, 36].

This study extends previous observations by showing that in otherwise healthy aMCI participants and control subjects, there were no group differences in gait velocity under simple walking conditions and other physical function measures including the timed-up-and-go, simple reaction time, and functional reach. However, reductions in gait velocity under dual-task conditions were much more pronounced in aMCI participants than normal controls, and these differences were greatest during working and episodic memory challenges. The underlying mechanisms of reduced gait velocity during dual tasks are not clear. Based on the theory of shared resources of motor and cognitive control discussed above [6], it is possible that when the available cognitive resources (e.g. memory and executive function) were reduced in aMCI participants, the motor slowing during walking were unmasked when the cognitive supply could not meet the demand for performing physical function.

Interestingly, accumulating evidence indicates that perturbation of default mode network (DMN), activities in certain brain regions during resting-state, may yield clinical significance in patients with aMCI [37]. It is well established that older adults with higher cognitive reserve showed lower deactivations within the DMN and lower task-related activity, which reflect adequate neural efficiency [37]. In contrast, higher deactivations in DMN suggesting increases in DMN compensatory activity have been reported in patients with aMCI and AD [38]. It is our speculation that these DMN disruptions may reflect reduced cognitive resources associated with the motor slowing we observed in aMCI participants.

In addition, neuroimaging data suggesting that brain white matter integrity is likely to play an important role in age-related changes in cognitive and physical function has emerged [39]. For example, white matter hyperintensity volume has been associated with executive dysfunction, reductions in gait velocity and motor control in older adults [40]. Furthermore, MRI studies using diffusion tensor imaging have shown that gait disturbance such as gait slowing and gait variability was associated with the loss of white matter integrity in the regions such as the corpus callosum, cingulum bundle, and the posterior limb of the internal capsule that are known to be linked to motor cortex [39, 41]. In addition, patients with MCI and AD have shown more white matter deterioration in some of these regions when compared with controls [42, 43]. These findings indicate that further studies are needed to understand the role of DMN and brain structural integrity in gait slowing in aMCI patients.

The findings of this study may have potential clinical significance although we acknowledge that further study of sensitivity and specificity of this test is required for detection of aMCI and early AD. First, reduction in gait velocity has been used to predict falls in older adults [44]. An early study using dual-task paradigm has suggested that older adults who could not “walk and talk” subsequently fell, while those subjects who could walk and talk were much less prone to future falls [45]. Of note, thirty percent of people over 65 years of age living in the community fall each year [46] and fall-related injuries among older adults are associated with substantial economic costs [47]. Secondly, MCI has been identified as an independent predictor of falls in community older adults [48] and incidence of falls in people with cognitive impairments estimated to be twice that of cognitively intact older adults [49]. The findings of this study may be useful in identifying older adults without major medical risks or major cognitive impairment that may be at increased risk for falls. Secondly, gait slowing has been recognized as an important risk factor for cognitive decline and dementia in older adults [14]. We also demonstrated that currently existing brief neuropsychological tasks may be administered under dual-task conditions in the clinic setting, with simple instruments (i.e. stop watch and measuring tape), to potentially increase the sensitivity of detecting early MCI in primary care settings.

4.1. Study Limitations

We acknowledge the limitation of a small sample size of this study and note that the results should be interpreted with caution. However, the more pronounced gait slowing was observed consistently in all aMCI participants compared to controls, suggesting the robust nature of this test. In addition, given the limitations of cross-sectional study design, differences in the gait velocity under dual-task conditions could not be attributed solely to cognitive impairment of aMCI participants. Furthermore, we implemented a stringent screening protocol to control for potential confounding factors such as co-morbidity and aerobic fitness level which may influence gait velocity in older adults. In addition, sex, age, educational and global cognitive levels between groups were similar. Future investigation should aim to validate the sensitivity and specificity of gait slowing under dual-task conditions in subjects with aMCI and to determine its role in Alzheimer's disease progression/conversion via longitudinal design with a larger cohort.

CONCLUSION

We found that gait velocity under simple walking conditions in otherwise healthy aMCI participants was preserved when compared with normal controls. However, participants with aMCI showed more accentuated gait slowing under dual-task conditions when concentration, executive and memory functions were challenged. These findings support the theory of shared resources of motor and cognitive control in older adults and suggest that the measurement of gait velocity under dual-task conditions can potentially be administered in addition to the current screening tools for detection of aMCI and early Alzheimer disease.