The frail brain

Abstract

Background and aims: Frailty in older adults has been associated with reduced brain health and greater risk of cognitive decline. However, structural and functional brain signatures of frailty remain understudied, with very little research focus on the neuroscience of frailty. The studies published to date use different frailty operationalisations and metrics of neuroimaging biomarkers across different cohorts with lack of agreement between results. My aims were: to explore cross-sectional associations between a frailty index (FI) and grey matter volume on magnetic resonance imaging (MRI); to identify the frailty features that had a core role in the frailty-cortex volume relationship; to identify brain regions and connectivity patterns associated with frailty; and to explore the association of frailty with cognitive and functional performance.

Methods: This was a cross-sectional observational study from a population-based study (The Irish Longitudinal Study on Ageing: TILDA). Participants aged ? 50 years who underwent the wave 3 MRI substudy were included. Cortex, basal ganglia, and each of the Desikan-Killiany regional volumes were measured. White matter fibre integrity was quantified using diffusion tensor imaging (DTI). Age- and sex-adjusted correlations were performed with a 32-item self-reported FI that included deficits commonly tested for in clinical and research settings. A graph theory analysis of the network composed by each FI item and cortex volume was performed. Connectome-based predictive modelling (CPM) was applied to the FI using resting state functional MRI data from the same cohort. A Spearman partial correlation model was constructed adjusting by age, sex and level of education to explore the correlation between the FI and measures of global cognition and physical performance.

Results: In 523 participants of the structural MRI cohort (mean age 69 years, 49% men), lower cortex and thalamic volumes were independently associated with higher FI. Sensory and functional difficulties, diabetes, polypharmacy, knee pain, and poorer self-reported health were the main FI associations with cortex volume. In the network analysis, cortex volume had a modest influence within the frailty network. Regionally, higher FI was significantly associated with lower volumes in both orbitofrontal and temporal cortices. DTI analyses revealed inverse associations between the FI and the integrity of some association bundles, mostly involving frontal lobes with relative preservation of the left parietal cortex. For the functional MRI cohort, 347 participants were included (mean age 68 years, 49% men). From CPM, 204 edges were obtained that positively correlated with the FI and composed the ?frailty network? characterised by connectivity of the visual network (right); and 188 edges that negatively correlated with the FI and formed the ?robustness network? characterised by connectivity in the basal ganglia. Both networks? highest degree node was the caudate but with different patterns: from caudate to visual network in the frailty network; and to default mode network (DMN) in the robustness network. The FI was correlated with physical performance metrics but not with metrics of global cognition, reinforcing the matching between the FI and the brain connectivity pattern found (main predicted connectivity in basal ganglia); and with the structural brain patterns found (high order integration areas and association white matter tracts mostly involving frontal lobes).

Conclusions: 1. Frailty defined as self-reported accumulation of health deficits had recognisable signatures on brain structure (lower volume in high order integration areas) and loss of integrity in association white matter tracts (mostly involving frontal lobes with relative preservation of the left parietal cortex).

2. The features that composed the FI had low weight vis-?-vis total cortex volume when their interactions were analysed as a network. Functional and sensory impairments were the FI features that were most related with cortex volume.

3. A ?frailty network? was identified that was predominantly characterised by connectivity of the visual network. On the other side of the continuum, a ?robustness network? was predominantly characterised by connectivity of the basal ganglia. Moreover, both networks? highest degree node was located in the caudate, but with different connectivity pattern: from caudate to visual network in the frailty network; and to DMN-related areas in the robustness network.

4. The FI in this sample of TILDA participants captured a state of impairment of daily physical performance but not significant cognitive impairment, matching the structural and connectivity patterns found.