Characterising the covariance pattern between lifestyle factors and structural brain measures: a multivariable replication study of two independent ageing cohorts

Abstract

Modifiable lifestyle factors have been shown to promote healthy brain ageing. However, studies have typically focused on a single factor at a time. Given that lifestyle factors do not occur in isolation, multivariable analyses provide a more realistic model of the lifestyle-brain relationship. Here, canonical correlation analyses (CCA) examined the relationship between nine lifestyle factors and seven MRI-derived indices of brain structure. The resulting covariance pattern was further explored with Bayesian regressions. CCA analyses were first conducted on a Danish cohort of older adults (n = 251) and then replicated in a British cohort (n = 668). In both cohorts, the latent factors of lifestyle and brain structure were positively correlated (UK: r =.37, p < 0.001; Denmark: r =.27, p < 0.001). In the cross-validation study, the correlation between lifestyle-brain latent factors was r =.10, p = 0.008. However, the pattern of associations differed between datasets. These findings suggest that baseline characterisation and tailoring towards the study sample may be beneficial for achieving targeted lifestyle interventions.