Precipitating Mechanisms of Falls in Preclinical Alzheimer’s Disease

Participants

Participants in studies at the Knight Alzheimer’s Disease Research Center (Knight ADRC) at Washington University in St. Louis were invited to participate in this study, a 4-year longitudinal observational study of fall risk, functional mobility, and AD. Individuals were invited to participate if they were over the age of 65, cognitively normal as indicated by a Clinical Dementia Rating ^® (CDR) [14] score of 0 at the time of assessment, and underwent amyloid PET and MRI scans within 2 years of the CDR and cognitive assessments. Additional study details have been described elsewhere [15]. This study was approved by the Institutional Review Board at Washington University in St. Louis (201807135). All participants provided written informed consent. All data used in the present study are from the first year of the ongoing parent study (see Fig. 1 for timeline of study procedures).

OPEN IN VIEWER

Fig. 1

Timeline of study procedures. ADRC, Alzheimer’s Disease Research Center.

CDR status

The CDR uses a 5-point scale to characterize 6 domains of cognitive and functional performance that are applicable to AD and other dementias [14]. The domains include memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. CDR scores are determined through semi-structured interviews with a licensed clinician, the participant, and a reliable informant such as a family member or friend. A CDR score of 0 indicates cognitive normality, 0.5 = very mild dementia, 1 = mild dementia, 2 = moderate dementia, and 3 = severe dementia. All participants in this study were CDR 0 at the time of evaluation.

APOE status

DNA samples were obtained and genotyped according to methods published previously [16]. An Illumina 610 or Omniexpress chip was used for genotyping. For the purposes of these analyses, APOE status was converted from a genotype to a binary variable. If participants had at least one copy of the Apolipoprotein (APOE) ɛ4 allele, they were deemed “APOE positive” and “APOE negative”otherwise.

In-home visit

As described in Bollinger et al. [15], an occupational therapist blinded to participants’ AD biomarker levels completed an in-home visit lasting approximately 120–180 min. The occupational therapist conducted assessments related to functional mobility, the peripheral nervous system, and fall covariates (Table 1). The visit was typically completed in 1 session but was completed over 2 sessions for 4 participants due to participant fatigue or request. Participants received their results from the home visit and fall risk assessments based on established fall risk cutoff scores [17].

Fall measures

A 12-month calendar-journal was given to each participant to record whether or not a fall occurred. Falls were reported monthly via an automated call or e-mail [16]. Participants were compensated with $5 on a reloadable gift card after they completed their fall report each month. A fall was defined as any unintentional movement to the floor, ground, or an object below knee level [18]. If a fall was reported, an interviewer called the participant to verify that the fall met this operational definition. The measures of interest included days to first reported fall, total number of falls reported over the year (i.e., fall propensity), and highest severity of falls reported over the year [1]. These graded fall outcomes were chosen because they capture and retain important information on subtleties of falls. Fall severity was measured using a published algorithm [1]: 0, no falls; 1, one fall without serious injury that did not require medical attention; 2, a fall with minor injury that received medical attention but not a hospital admittance or more than one fall; and 3, a fall with major injury that prompted a hospital admission. Additional details about monthly fall monitoring have been published elsewhere [15]. Fall monitoring for the present study began at enrollment and continued for the first 12 months of the parent study.

Assessment of cognition

Participants were given a series of computerized and paper-and-pencil tasks to measure different facets of cognition. A cognitive composite similar to the Preclinical Alzheimer’s Cognitive Composite (PACC) [19, 20] was created by averaging the standardized scores from the following tests: Free and Cued Selective Reminding Test Free Recall, Animal Verbal Fluency Naming Total Score, Craft Story 21 Delayed Recall, and the total correct score from the Number Symbol Test (a computerized task developed and validated at the Knight ADRC that assesses similar constructs to the Wechsler Adult Intelligence Scale Digit Symbol Substitution task). Specifically, each participant’s raw score on each task was standardized to the mean and standard deviation of the cohort’s baseline completion of that particular test. The z-scores were then averaged together such that higher scores indicated better performance [21].

Clinically informed composites

We used our team members’ expertise in occupational therapy, neurology, and psychology to group the potential fall mechanisms collected into clinically meaningful categories. We refer to these groupings as composites, with each capturing different facets of functioning associated with falls (see Table 1) [3, 10]. The composites were: sensation, fall risk factors/comorbidities, cognition, activities of daily living [22], strength, fall precautionary behaviors, and functional mobility. Each composite was created by averaging the standardized scores from the measures listed in Table 1. Scores with the superscript “r” were reverse scored (standardized scores multiplied by –1) so that higher scores indicated better performance across all composites. Descriptions of measures used to create each composite and rationales are provided in Table 1.

Neuroimaging AD biomarkers

Amyloid PET imaging was performed with [¹¹C] Pittsburgh Compound B (PiB) or [¹⁸F]-Florbetapir (AV45) and was acquired on a Biograph mMR scanner (Siemens Medical Solutions, Malvern, PA). PiB PET scans included data 30–60 min post-injection, and AV45 PET scans included data 50–70 min post-injection. All data were converted to standardized uptake value ratios with the cerebellar cortex used as a reference region before then being converted to the Centiloid scale [23, 24]. PET data were processed with an in-house pipeline using regions of interest (ROIs) derived from FreeSurfer (https://github.com/ysu001/PUP) [25]. This approach corrects for the spillover signal from adjacent ROIs and non-brain tissue on the basis of the scanner point spread function and the relative distance between regions. This partial volume correction approach accounts for spillover not only from different areas in the brain, but also from non-brain regions into the brain. Amyloid deposition was quantified with the average across the left and right lateral orbitofrontal, medial orbitofrontal, rostral middle frontal, superior frontal, superior temporal, middle temporal, and precuneus regions.

Structural MRI data were acquired on a Siemens Biograph mMR or Trio 3T scanner. T1-weighted images were acquired with a magnetization-prepared rapid acquisition gradient echo sequence acquired in the sagittal orientation with a repetition time of 2300 ms, an echo time of 2.95 ms, a flip angle of 9°, 176 slices, an in-plane resolution of 240×256, and a slice thickness of 1.2 mm. Images underwent volumetric segmentation with FreeSurfer 5.3 (http://freesurfer.net) to identify ROIs for further analysis. Hippocampal volumes were adjusted for head size with a regression approach and summed across hemispheres [26, 27]. Cortical thickness values were obtained for each hemisphere for a limited number of ROIs reflecting brain atrophy patterns in AD [28]. Cortical thickness was calculated as the shortest distance between the cortical gray/white boundary to the gray/cerebrospinal fluid boundary [29]. Neuroimaging AD biomarkers measured within 2 years of enrollment in the present study, which is also when the monthly fall monitoring began, were used for the analysis (see Fig. 1).

Statistical analysis

We used Pearson correlations to examine associations among the fall measures, clinically informed composites, and neuroimaging AD biomarkers (as continuous variables). We also conducted multiple regression models to test for the contribution of each clinically informed composite above and beyond that of the covariates (i.e., age, gender, education, APOE4 status, polypharmacy, sensation, and comorbidities of alcoholism, depression, and pain; see Table 1). Variables with significant correlations with one another were not included in the same models, as they would be subject to potential multicollinearity issues. This applied to the clinically informed and cognitive composites (Fig. 2). Therefore, non-nested models were used, and model fit was evaluated based on the R² and Akaike information criterion (AIC). For all models, reference groups are indicated in the output tables (Tables 3 and 4). For all results reported, statistical significance was set at p <  0.05 two-tailed, unless otherwise noted. Cohen’s d [30], a measure of effect size, was reported for significant t-tests. Adjusted degrees of freedom were reported such that unequal variances were assumed for comparisons, and the Welsh approximation was applied. All analyses were conducted using the R statistical computing environment [31].

OPEN IN VIEWER

Fig. 2

Relationships Among Age, Composites, Neuroimaging Biomarkers, and Falls. Significant correlations at *p <  0.05, **p <  0.01, and ***p <  0.001. ADL, Activities of Daily Living.

Sample characteristics

A total of 182 participants were included in the study. Demographic information and sample characteristics are displayed in Table 2. A total of 227 falls were reported over the 12-month monitoring period. Falls per person ranged from 0–16; the median number of falls per person was 1.

Relationships among composites, neuroimaging biomarkers, and falls

Cortical thickness, a measure of neurodegeneration, was associated with functional mobility (r = 0.2, p <  0.05). Although the cognitive composite was associated with AD biomarkers and clinical composites, we found that the activities of daily living (ADL), strength, fall precaution, and functional mobility composites all significantly correlated with the fall outcome measures, particularly fall propensity (rs≤–0.4, ps <  0.01; see Fig. 2). The cognitive composite, in contrast, was not associated with any of the fall outcome measures (rs = 0.03–0.09). Participants with higher ADL function, strength, fall precaution, and functional mobility reported fewer total falls and less severe falls over the 12-month monitoring period. The same relationships appeared for fewer days to first fall; however, they were weaker and did not reach significance.

Predicting fall propensity and severity

ADL and functional mobility composites significantly predicted the total number of falls reported by participants over the 12-month monitoring period (fall propensity; see Table 3). Furthermore, examination of R² and AIC values suggested that functional mobility was the domain that best predicted fall propensity compared to the other domains in the analysis.

Next, we performed a similar analysis to identify the domain that best predicted fall severity. Examination of R² and AIC values again indicated that functional mobility was the domain that best predicted fall severity compared to the other predictors in the model (See Table 4).

Limitations

The findings of this study should be considered in light of its limitations. First, we only analyzed a single time point for the clinical, cognitive, and biomarker measures in this initial examination. These relationships will be examined throughout the larger longitudinal observational study to better understand how these mechanisms impact fall risk and severity over time. Second, the data failed to produce significant relationships among AD biomarkers and fall measures as would be expected based on Keleman et al. [37]. As there are many ways to measure cognition and other clinical constructs of interest, it is possible that using more sensitive measures and other analytical procedures may capture preclinical AD changes and yield a different pattern of results. Finally, participants consisted of older adults who were comprehensively phenotyped and often engaged in imaging and fluid biomarker studies, and therefore are not representative of the general population. Although a primary goal of the Knight ADRC is to diversify enrollment to include underrepresented persons in aging and dementia research, the participants in this study were primarily White and highly educated (see Table 2).

Conclusions

This study extends upon previous work showing a relationship between preclinical AD biomarkers and falls in CN older adults. Functional mobility best predicted fall propensity and severity beyond a host of covariates, where other clinical measures such as strength and cognition did not. Cognition was more closely associated with AD neuropathology (amyloid and neurodegeneration) compared to other clinical measures, and functional mobility was associated with neurodegeneration but to a lesser extent. These findings suggest that, although subtle changes in cognition may be more closely associated with AD neuropathology, functional mobility indicators better predict falls in CN older adults. Together, these findings suggest that functional mobility plays an important role in fall risk among older adults at risk for AD and may be an ideal mechanism for future fall prevention strategies.