Neuropathologic burden and the degree of frailty in relation to global cognition and dementia

Abstract

Objective

To test the hypothesis that degree of frailty and neuropathologic burden independently contribute to global cognition and odds of dementia.

Methods

This was a secondary analysis of a prospective cohort study of older adults living in Illinois. Participants underwent an annual neuropsychological and clinical evaluation. We included 625 participants (mean age 89.7 ± 6.1 years; 67.5% female) who died and underwent autopsy. We quantified neuropathology using an index measure of 10 neuropathologic features: β-amyloid deposition, hippocampal sclerosis, Lewy bodies, tangle density, TDP-43, cerebral amyloid angiopathy, arteriolosclerosis, atherosclerosis, and gross and chronic cerebral infarcts. Clinical consensus determined dementia status, which we coded as no cognitive impairment, mild cognitive impairment, or dementia. A battery of 19 tests spanning multiple domains quantified global cognition. We operationalized frailty using a 41-item frailty index. We employed regression analyses to model relationships between neuropathology, frailty, and dementia.

Results

Both frailty and a neuropathology index were independently associated with global cognition and dementia status. These results held after controlling for traditional pathologic measures in a sample of participants with Alzheimer clinical syndrome. Frailty improved the fit of the model for dementia status (χ²[2] 72.64; p < 0.0001) and explained an additional 11%–12% of the variance in the outcomes.

Conclusion

Dementia is a multiply determined condition, to which both general health, as captured by frailty, and neuropathology significantly contribute. This integrative view of dementia and health has implications for prevention and therapy; specifically, future research should evaluate frailty as a means of dementia risk reduction.

As populations age, the burden of dementia increases, lending force to the need for a contemporary understanding of dementing illnesses. Dementia is a multiply determined clinical condition of several causes including Alzheimer disease (AD) pathology, as well as other pathologies, i.e., mixed pathology.^1–5 Recently, we demonstrated that frailty moderated the relationship between the pathologic features of AD and the clinical syndrome of Alzheimer (referred to here as Alzheimer dementia); specifically, higher frailty was associated with increased odds of dementia among people with low levels of plaques and tangles.⁶ This work is in line with several other studies that suggest that frailty is associated with the burden of neuropathology,⁷ with the accumulation of AD-related biomarkers,⁸ with cognition,^9,10 and dementia status.¹¹ This emerging conceptualization of late-life AD as a multiply-determined age-related condition leads to questions of how diverse types of neuropathology contribute to dementia, and how frailty may influence this relationship across the cognitive spectrum.

Prior work has shown that a composite metric of physical frailty based on the categorical measure captured in the frailty phenotype and its individual components are associated with the level and rate of cognitive decline and predicts incident mild cognitive impairment (MCI) and Alzheimer dementia.^10,12–17 Moreover, physical frailty together with other related motor constructs (i.e., bradykinesia, rigidity, parkinsonian gait and tremor) were more strongly associated with incident Alzheimer dementia.¹⁸ Physical frailty and its components including grip strength, gait speed, body mass index, and physical activity are associated with incident Alzheimer dementia and related disorders.^4,9,19–21 Modeling the trajectories of both cognition and physical frailty simultaneously showed that the rate of progression of physical frailty and cognitive decline in the same individuals were strongly correlated, and were correlated with AD.⁹ Given the strong relationship between frailty and cognitive decline, it is possible that they share a common pathologic basis.

We extend this work by using broader definitions of both frailty and of neuropathology to better understand the mechanisms by which dementia arises. This approach, which considers age-related deficit accumulation both in general health and in brain pathology, responds to shortcomings of previous investigations that have not been able to explicitly address the relationship between aging and dementia. In particular, we can consider the extent to which general age-related deficit accumulation is related to the accumulation of neuropathology. Specifically, our objective was to determine the relative contributions of neuropathology and the degree of frailty, both measured using a deficit accumulation approach, to global cognition and odds of dementia. We hypothesized that frailty would independently contribute to both outcomes.

Methods

Standard protocol approvals, registrations, and patient consents

The Rush Memory and Aging Project (MAP) was approved by an institutional review board of Rush University Medical Center, Chicago, IL. Written informed consent was obtained from all participants (or guardians of participants) enrolled in the study (consent for research). Participants also signed a repository consent that allowed their data to be repurposed for other studies.

Study design & participants

We analyzed data from MAP, described in depth elsewhere.²² Briefly, MAP is a clinical–pathologic cohort study that since 1997 has enrolled over 2,100 older adults from residential facilities, senior and subsidized housing, church groups, and social service agencies in Northeastern Illinois. This cohort has been followed for over 20 years with annual clinical evaluations. Participants were eligible for enrollment if they were able and willing to sign both the informed consent and an Anatomical Gift Act for donation of their brain, spinal cord, nerve, muscle, and other biospecimens at death.

Measures

Neuropathologic index

Trained neuropathologists completed a neuropathologic assessment at autopsy, including the following 10 neuropathologic features.

1. Overall level of β-amyloid (Aβ) was quantified as percent area of cortex occupied by Aβ as stained using immunohistochemistry over 8 regions: hippocampus, entorhinal cortex, midfrontal cortex, inferior temporal, angular gyrus, calcarine cortex, anterior cingulate cortex, and superior frontal cortex.⁶ As there are no appropriate clinical cut points for this continuous measure, Aβ was recoded between 0 and 1 based on quartiles (0–0.67; 0.68–4.16; 4.17–8.17; 8.18–22.94).

2. Typical hippocampal sclerosis was evaluated as not present/possible or present but atypical (0) and definitely present (1) according to severe neuronal loss and gliosis in CA1 or subiculum.²³

3. Lewy body pathology was determined by distribution of α-synuclein (as stained for α-synuclein immunostain) in midfrontal, midtemporal, inferior parietal, anterior cingulate, entorhinal and hippocampal cortices, basal ganglia, and midbrain. Severity was coded as not present (0), nigral-predominant (0.33), limbic-type (0.67), and neocortical type (1).²⁴

4. Cortical tangle density was quantified as the mean density (per mm²) of 8 regions: hippocampus, entorhinal cortex, midfrontal cortex, inferior temporal, angular gyrus, calcarine cortex, anterior cingulate cortex, and superior frontal cortex, as stained by immunochemistry. This measure of paired helical filaments tau tangles was recoded for analysis into quartiles (0–1.69; 1.70–4.00; 4.01–8.78; 8.79–78.52).⁶

5. TDP-43 pathology was staged using data from 6 regions (hippocampus, CA1 and dentate, amygdala, midfrontal, midtemporal, and entorhinal cortices), using immunohistochemistry. It was coded as no pathology (0), stage 1, localized to amygdala only (0.33), stage 2, extension to hippocampus/entorhinal cortex (0.67), and stage 3, extension to neocortical (1)²³; this coding was also used for the limbic-predominant age-related TDP-43 encephalopathy neuropathologic changes (LATE-NC) sensitivity analyses.

6. Arteriolosclerosis was evaluated using a semi-quantitative grading system of no occlusion (0), mild occlusion (0.33), moderate occlusion (0.67), or severe occlusion (1). It is based on histologic changes including intimal deterioration, smooth muscle degeneration, and fibrohyalinotic thickening in the anterior basal ganglia.²⁵

7. Cerebral amyloid angiopathy was quantified as a summary score of amyloid deposition in meningeal and parenchymal vessels using immunostains in the midfrontal, midtemporal, angular, and calcarine cortices.²⁶ A 4-level staging technique was applied where 0 = no deposition, 0.33 = mild deposition, 0.67 = moderate deposition, and 1 = severe deposition.

8. Cerebral large vessel atherosclerosis was evaluated using semi-quantitative 4-level grading applied to vertebral, basilar, posterior cerebral, middle cerebral, and anterior cerebral arteries and their proximal branches. Coding of 0 indicates no significant atherosclerosis; 1 indicates mild atherosclerosis, with small amounts in several arteries but no significant occlusion; 2 indicates moderate atherosclerosis in up to half of visualized major arteries, with <50% occlusion of any single vessel; and 3 indicates severe atherosclerosis, present in more than half of all visualized vessels or more than 75% occlusion of 1 or more vessels.²⁷

9. Acute gross cerebral infarcts, as determined by neuropathologic evaluation and confirmed by histopathologic dissection, was classified as no gross infarctions (0), or 1 or more gross infarctions, regardless of age and location (1).²⁸

10. Chronic gross cerebral infarcts, as determined by neuropathologic evaluation and confirmed by histopathologic dissection, was classified as no gross chronic infarctions (0), or 1 or more gross chronic infarctions, regardless of age and location (1).²⁸

Traditional AD staging measures

Braak staging,²⁹ a measure of the number and distribution of tangle pathology, was scored between 1 and 6 (where higher scores indicate more pathology). Consortium to Establish a Registry for Alzheimer's Disease (CERAD) score³⁰ is another measure of traditional AD pathology based on Braak score and neuritic plaque number and distribution. Both measures were semiquantitatively scored by a technician and reviewed by a neuropathologist.

Frailty index

The frailty index is a health state measure that reflects the extent of illness and vulnerability to adverse outcomes, including death.³¹ The frailty index = (number of health deficits present)/(number of health deficits measured). For example, a person with 5 of 30 potential deficits measured has a frailty index score of 5/30 = 0.17. We used a 41-item frailty index from health variables obtained at each clinical evaluation, as detailed elsewhere.⁶ Variables included in the frailty index employed here can be found in table 1. Candidate variables included symptoms, signs, comorbidities, and function. We screened variables for inclusion based on standard procedures³²; those strongly related to the outcome (i.e., cognitive variables) were excluded. Sensitivity analyses in this sample have previously shown that excluding functional items such as activities of daily living or items weakly associated with the outcome do not change the direction or significance of results.⁶ Higher frailty index values indicate poorer health. For ease of interpretation, we multiplied frailty index values by 10 so that the odds ratio (OR) would represent the proportion change for an increased increment of 0.1 of the frailty index (an additional 3–4 deficits).

Table 1.

Items included in frailty index

Global cognition

A battery of 17 cognitive tests spanning multiple domains including episodic, semantic, and working memory, perceptual speed, and visuospatial ability quantified global cognition.³³ Raw scores were converted to z scores calculated on the basis of original baseline cohort data.³⁴ Negative scores indicate global cognition below average of the entire cohort from the baseline evaluation, while positive scores indicate above average global cognition. Scores were taken at last study evaluation before death (0.9 ± 1.2 years antemortem).

Dementia status

Clinical consensus at each annual assessment determined dementia status. A 3-step process was employed: (1) computer scoring of cognitive testing³³; (2) clinical judgment by a neuropsychologist blinded to participant demographics; (3) diagnostic classification by a clinician (neurologist, geriatrician, geriatric nurse practitioner, or neuropsychologist)³⁴ based on the National Institute of Neurologic and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association. We categorized this into no cognitive impairment, MCI, or dementia (including possible and probable Alzheimer and other dementias). For the purpose of some analyses (receiver operating characteristic [ROC] curves), we created dummy variables for binary outcomes (no cognitive impairment vs MCI, and MCI vs dementia).

Confounders

We evaluated age, sex, education, and time from last assessment to death as confounders. Age was calculated from birth date to date of cognitive testing and used as years; self-reported education at the baseline evaluation was measured in years; sex was a self-report binary variable determined at baseline. Time interval from last assessment to death was measured in years.

Statistical analysis

We employed descriptive analysis techniques to describe characteristics of the sample and used analysis of variance and χ² tests to compare means. As there were few missing data, we employed complete case analysis. We used ordinal regression (logit function) to evaluate the relationship between the neuropathologic index, frailty index, and dementia status and linear regression to evaluate the relationship of the neuropathologic index and the frailty index with respect to global cognition. In sensitivity analyses with binary outcomes, we used logistic regression models. For ordinal regression models, we transformed β coefficients and confidence intervals (CIs) into ORs for ease of interpretation. We also evaluated interaction terms between neuropathology and frailty. We adjusted all analyses for age, sex, education, and time interval between last assessment and death. Change in deviance (χ² of −2 log likelihood values) evaluated improvement in model fit. We also reported Nagelkerke pseudo R² as a measure of model fit. We calculated absolute risk reduction and numbers needed to treat for median and 75th percentile frailty index values in relation to dementia status (MCI was excluded). We calculated area under the ROC curve (AUC) to evaluate the sensitivity and specificity of the independent variables (neuropathologic index and frailty index) in discriminating no cognitive impairment from MCI and MCI from dementia. SPSS version 25.0 was employed for all analyses. R was used to plot figures.

Data availability

Data from MAP are available by request at radc.rush.edu.

Results

Demographics

At the time we froze the data set in January 2017, there were 645 autopsied participants with completed neuropathologic assessment. We examined 625 participants with sufficient neuropathologic, cognitive, and clinical data to create a frailty index (see figure 1 for inclusion flow chart). Variables of interest were not missing any data. Participants were 89.7 ± 6.1 years of age at time of death, and mostly female (see table 2 for full demographic information). Most participants were classified as having AD according to the CERAD pathologic staging measure, though less than half had a diagnosis of dementia (table 2). The frailty index and neuropathologic index had near normal distributions that became more right skewed with increasing cognitive impairment. Prevalence of dementia increased as a function of both neuropathology and frailty (figure 2).

Figure 1. Participant inclusion flow chart.

FI = frailty index; MAP = Rush Memory and Aging Project.

Table 2.

Descriptive characteristics

Figure 2. Dementia status by level of neuropathologic burden and frailty.

Outcome: dementia status

Ordinal regression models demonstrated that both the neuropathologic index and frailty index were independently associated with dementia status (table 3), after adjusting for age, sex, education, and time from last assessment to death. The relationship between neuropathology and dementia was similar across levels of frailty (figure 3). When the frailty index was added to the model with the neuropathologic index, each remained statistically significant and model fit significantly improved (deviance change χ²[2] = 183.92, p < 0.0001). Pseudo R² increased from 0.28 to 0.40, suggesting that the frailty index increased the explained variance by 12%. The frailty by neuropathology interaction was nonsignificant. Absolute risk reduction (ARR) was calculated for dementia status: if a frailty index cutoff of 0.41 (median) was used, ARR = 0.40, and number needed to treat is 2.50; if a frailty index cutoff of 0.55 (75th percentile) was used, ARR = 0.38, and number needed to treat is 2.60. This suggests that for each 2.5–2.6 people in whom severe frailty is avoided, one case of dementia will be averted. The first ROC used the outcome no cognitive impairment vs MCI (n = 384), where both the neuropathologic index and frailty index significantly classified the outcome: AUC 0.64 (95% CI 0.58–0.69) and AUC 0.58 (95% CI 0.52–0.63), respectively. Similarly, when the outcome MCI vs dementia (n = 414) was used, both the neuropathologic index and frailty index again were significant: AUC 0.70 (95% CI 0.65–0.75) and AUC 0.68 (95% CI 0.63–0.73), respectively. Results did not differ when stratified by sex.

Table 3.

Ordinal regression demonstrating relationship between the neuropathologic index, frailty index, and dementia status

Figure 3. The relationship between neuropathology and dementia across levels of frailty.

(A) Probability of mild cognitive impairment (MCI) (vs no cognitive impairment) as a function of neuropathologic burden, stratified by frailty level. (B) Probability of dementia (vs MCI) as a function of neuropathologic burden, stratified by frailty level. (C) Probability of dementia (vs no cognitive impairment) as a function of neuropathologic burden, stratified by frailty level.

Outcome: global cognition

Linear regression models also demonstrated that higher neuropathologic index and frailty index scores were independently associated with poorer cognition (table 4). When the frailty index was added to the model with the neuropathologic index, each remained statistically significant, and model fit improved: pseudo R² increased from 0.28 to 0.39. The interaction between frailty and neuropathology was significantly associated with cognition (unstandardized β coefficient = −0.06, 95% CI −0.07 to −0.05, p < 0.0001). Results did not differ when stratified by sex.

Table 4.

Linear regression demonstrating relationship between the neuropathologic index, frailty index, and global cognition

Sensitivity analyses

We conducted a sensitivity analysis using logistic regression with an outcome of no cognitive impairment vs dementia that fit with a clinical syndrome of AD (here referred to as Alzheimer dementia) (n = 448) to test performance of the neuropathologic index and frailty index after accounting for Braak and CERAD pathologic AD staging measures. We found that the neuropathologic index and frailty index remained significant after controlling for CERAD score and Braak stage. When the frailty index was added to the model, fit improved (deviance change χ²[2] = 56.19, p < 0.0001) and pseudo R² increased from 0.46 to 0.56 (10%). ROC results also held in a sensitivity analysis where we restricted the outcome to no cognitive impairment vs Alzheimer dementia. Further, when the MCI group was first collapsed into the nondementia group, and subsequently into the dementia group, neither analysis demonstrated significant interactions between frailty and neuropathology (OR 0.97, 95% CI 0.91–1.05, p = 0.46; and OR 0.98, 95% CI 0.92–1.05, p = 0.52, respectively).

Ordinal regressions were used to evaluate the relationship between frailty and LATE-NC with all-cause dementia status. Frailty and LATE-NC were independent predictors of dementia status (unstandardized β coefficients = 0.27 [95% CI 0.21–0.33, p < 0.001] and 0.28 [95% CI 0.19–0.37, p < 0.001], respectively), but did not interact. Frailty was significantly higher in those with dementia but did not differ according to LATE-NC staging. Results did not differ when stratified by sex.

Discussion

In this analysis of 625 older adults, an index measure of frailty and of neuropathology were each significantly associated with dementia status and a measure of global cognitive functioning derived from neuropsychological tests. Sensitivity analyses demonstrated that even among clinical cases of Alzheimer dementia, both the neuropathologic index and frailty index independently contributed to odds of dementia over and above traditional pathologic measures. Given the interest in the newly defined LATE-NC,³⁵ we also tested the relationship between frailty, LATE-NC, and dementia status, and found results consistent with our original analyses, suggesting that regardless of how we define neuropathology, frailty continues to play an important role in dementia risk.

Taken together, these results suggest that dementia (including Alzheimer dementia) is a multiply determined condition. Importantly, frailty as measured using the deficit accumulation approach improved the fit of the model for both dementia status and global cognition and explained 11%–12% more of the variance in the outcomes (dementia status and cognition, respectively). This suggests that overall health, as quantified by the frailty index, independently affects cognitive processes and that frailty may be a useful target for prevention and treatment of dementia. Given the relatively large sample size, few recruitment criteria, and robustness of the findings to sensitivity analyses, we expect these results to be generalizable to other populations of older adults at risk of developing cognitive impairment.

The relationship between neuropathology and cognition has long been contentious in AD research. Many reports, particularly from community-based samples, emphasize the discrepancy.^1,2,24,36,37 For example, many people who were diagnosed clinically with Alzheimer dementia show relatively low burden of plaques and tangles on autopsy. Conversely, other individuals who did not appear to have dementia during life have high levels of neuropathology on autopsy. The evidence of discordance suggests that Alzheimer dementia may not be a homogeneous disease state that is marked by specific and discrete pathology, but rather may be a multiply determined condition that arises from an imbalance between increasing pathologic mechanisms and decreasing resilience that comes with age as health deficits accumulate. In this analysis, frailty is associated with cognition and dementia independent of neuropathology, consistent with other reports of AD risk factors,^38,39 and evidence relating frailty to cognitive decline and dementia.^10,17,40 Interestingly, the relationship between neuropathology and cognition changed over levels of frailty, but the relationship between neuropathology and dementia diagnosis did not change over levels of frailty. This finding may be a reflection of the differential contributions of frailty to cognitive change vs more global impairment (i.e., reduced functional capacity) characteristic of dementia. Taken together, these results suggest that frailty may be a measure of the latent resilience capacity continuum in the face of nonmodifiable pathology. This is likely not specific to the frailty index measurement.^1,2 It also is a way to respond to the critique of AD being studied as though all that mattered about age was exposure duration, whereas “the problems of old age come as a package.”⁴¹

These analyses extend prior work demonstrating that frailty moderates the relationship between AD pathology and dementia⁶ by identifying a second mechanism through which frailty independently may provide cognitive reserve. Further, the present work confirms other reports that many neuropathologic lesions are associated with dementia status and global cognition, even among people with clinical Alzheimer dementia.¹ We build on this by demonstrating that the number of neuropathologic deficits, rather than their particular nature, is crucial in determining who develops dementia, and that this mechanism is in addition to frailty. This is consistent with our previous work on the deficit accumulation model of frailty.^42,43 In essence, frailty helps us quantify individual differences in resilience, or the ability to stave off dementia, despite the accumulation of neuropathology.

Our results are consistent with other reports linking frailty and dementia. Frailty, operationalized both as an index and a phenotype, has been associated with cognitive decline and incident Alzheimer dementia.^6,10 Further, both physical and psychiatric conditions work in tandem to influence dementia risk.⁴⁴ Previous work has also linked frailty to the neuropathologic features of AD^20,45 as well as other dementia-related neuropathologic features.¹ In acknowledging the complexity of dementia, and its context in aging,⁴⁶ we have added to the literature by demonstrating that frailty influences odds of cognitive impairment and dementia, even in the face of neuropathology, which suggests frailty interventions may be useful at any stage of neuropathologic accumulation. Further, it supports the suggestion that discoveries in aging research may also offer candidate therapeutics for age-related illnesses,⁴⁷ of which late-life cognitive decline may be a leading example.⁵

Our results should be interpreted with caution. Temporality could not be ascertained, and thus causality cannot be determined; neuropathologic features were quantified at postmortem autopsy, while cognitive and frailty measures were taken at last study evaluation before death (an average of 11 months antemortem). This raises the possibility that neuropathologic burden could have changed significantly between the time of last assessment and autopsy, though there is consensus that most neuropathologic features accumulate slowly.⁴⁸ It is also possible that the association reflects diverse phenotypes affected by the same pathology. Future work should address the longitudinal relationship between frailty, brain-based features of dementia, and clinical expression of the disease. Further, our study sample was not representative. All participants were recruited from retirement communities in a fixed geographic location. It is likely that this sample differs on levels of frailty, social vulnerability, and prevalence of dementia vs the general population, therefore population-based samples with analytics that can control for known risks such as social vulnerability and cardiovascular risk would be ideal to confirm the results presented here. Cause of death should also be controlled for in future investigations. ARR and number needed to treat are typically used for interventional trials and not observational studies, but we aimed to use these figures to show the potential effect of frailty as a risk factor.

A neuropathologic index of 10 features was associated with global cognition and dementia status. Then neuropathologic index performed significantly better at discriminating Alzheimer dementia status than traditional AD pathologic staging measures. Adding frailty to a model with the neuropathologic index significantly improved the fit of the model in discriminating the odds of dementia. Future research should focus on elucidating mechanisms linking frailty and dementia, evaluate frailty as a means of dementia risk reduction, and employ longitudinal analysis to understand how dementia develops and progresses.

Glossary

Aβ

β-amyloid

AD

Alzheimer disease

ARR

absolute risk reduction

AUC

area under the receiver operating characteristic curve

CERAD

Consortium to Establish a Registry for Alzheimer's Disease

CI

confidence interval

LATE-NC

limbic-predominant age-related TDP-43 encephalopathy neuropathologic changes

MAP

Rush Memory and Aging Project

MCI

mild cognitive impairment

OR

odds ratio

ROC

receiver operating characteristic

Appendix. Authors

Study funding

L.M.K. Wallace is supported by a doctoral fellowship from the Canadian Institutes of Health Research (CIHR). M.K. Andrew's work on frailty and dementia is part of a Canadian Consortium on Neurodegeneration in Aging (CCNA) investigation into how multimorbidity modifies the risk of dementia and the patterns of disease expression (Team 14). The CCNA receives funding from the CIHR (CNA-137794) and partner organizations. Kenneth Rockwood's work on frailty and cognition is supported by CIHR PJT-156114 and by the Dalhousie Medical Research Foundation Kathryn Allen Weldon Chair of Alzheimer Disease Research. The Rush Memory and Aging Project is supported by NIH grant R01AG17917.

Disclosure

L.M.K. Wallace is supported by a Doctoral Research Award from the Canadian Institutes of Health Research (CIHR). O. Theou and S. Darvesh report no disclosures relevant to the manuscript. D. Bennett reports grants from the NIH. A.S. Buchman reports no disclosures relevant to the manuscript. M. Andrew reports grants from GSK, Pfizer, Sanofi, and the Canadian Frailty Network unrelated to the current work. S. Kirkland reports no disclosures relevant to the manuscript. J.D. Fisk receives research grant support from the CIHR, the Multiple Sclerosis Society of Canada, Crohn's and Colitis Canada, and royalty and distribution fees from MAPI Research Trust. K. Rockwood reports personal fees from Lundbeck for attending an advisory board meeting in 2017. K. Rockwood is President and Chief Science Officer of DGI Clinical, which in the last 5 years has contracts with pharma and device manufacturers (Baxter, Baxalta, Shire, Hollister, Nutricia, Roche, Otsuka) on individualized outcome measurement. Otherwise any personal fees are for invited guest lectures and academic symposia, received directly from event organizers, chiefly for presentations on frailty. He is Associate Director of the Canadian Consortium on Neurodegeneration in Aging, which is funded by the CIHR, and with additional funding from the Alzheimer Society of Canada and several other charities, as well as, in its first phase (2013–2018), from Pfizer Canada and Sanofi Canada. He receives career support from the Dalhousie Medical Research Foundation as the Kathryn Allen Weldon Professor of Alzheimer Research and research support from the CIHR, the Nova Scotia Health Research Foundation, the Capital Health Research Fund, and the Fountain Family Innovation Fund of the Nova Scotia Health Authority Foundation. Go to Neurology.org/N for full disclosures.