Association of the frailty index with functional brain connectivity in the Irish Longitudinal Study on Ageing (TILDA)

Abstract Background Frailty, expressed as accumulation of health deficits in older adults, could contribute to the risk of cognitive decline. We cross‐sectionally studied the brain connectivity signatures of frailty in The Irish Longitudinal Study on Ageing (TILDA). Method We included participants from Wave 3 of TILDA who underwent resting‐state functional MRI. Connectivity maps were based on Shen’s parcellation and connectome‐based predictive modeling was used to find associations between connectivity and frailty as measured by Frailty Index (FI) (Table 1). On each iteration of a 10‐fold cross‐validation the training folds were utilized to perform partial Spearman correlations (adjusted for age, sex, and mean framewise displacement) between each brain connection and FI scores. Connections with p<0.001 correlations were selected. Network strength was computed by summing the correlations across all connections selected and a linear model built using the network strength to predict FI score. After 10 folds are complete the performance was measured with linear R values between predicted and true FI scores. This process was repeated 1000 times (with random cross‐validation splits) and mean performance metrics obtained. A permutation test was employed in which the above procedure was repeated 1000 times, with FI scores shuffled each time. Significance of selected connections was calculated as the number of permutations scores that were equal to or greater than the proper mean score, divided by 1000. Result 347 participants were included (mean age 67.7 (SD 7.3) and 51.6% men). We obtained a connectivity network that positively correlated with frailty (“frailty network”) (R = 0.169, p = 0.006) and another that negatively correlated with frailty (“robustness network”) (R = ‐0.193, p = 0.006); and are depicted in Figure 1. The “frailty network” was more widespread and showed higher connectivity from right occipital to both prefrontal cortices, subcortical and left temporal cortex, whereas the “robustness network” exhibited higher functional connectivity in subcortical to left parietal and bilateral temporal cortices. Conclusion Functional brain connectivity could be influenced by frailty. Further research should study the relationship of brain connectivity and different tools for frailty and establish how different patterns of deficit accumulation affect brain health across lifespan.