Brain-predicted age difference scores (Brain-PADs) leverage machine learning to estimate the degree to which an individual’s brain magnetic resonance images (MRIs) appear younger or older than expected given their chronological age (Cole, 2020, Cole and Franke, 2017). Higher biological brain aging relative to chronological age (i.e., a brain that looks “older” than expected) predicts poor outcomes in older adults (Cole and Franke, 2017, Franke and Gaser, 2019). For example, higher Brain-PADs are associated with psychiatric and major depressive disorders (Blake et al., 2023, Han et al., 2021). Brain age measures are also associated with mild cognitive impairment (MCI) diagnoses and increased likelihood of progression to Alzheimer's Disease (AD) and related dementias (Gaser et al., 2013, Liem et al., 2017, Lowe et al., 2016). Individuals with AD, for example, have predicted brain ages about 10 years older than their chronological age (Franke et al., 2010).
With brain reserve defined as “an individual’s total neural resources or their neurobiological capital at a given point in time” (Arenaza-Urquijo and Vemuri, 2020, Kremen et al., 2022), brain-PAD serves as a good index of this construct. Brain-PADs integrate across multiple modalities (e.g., capturing grey and white matter) and are trained to weigh regions more heavily when they are sensitive to age differences derived from comparisons between cross-sectional assessments, perhaps providing advantages compared to other MRI measures such as whole brain volume. Importantly, we have demonstrated that young adult general cognitive ability and favorable lifestyle characteristics, including not smoking, low alcohol consumption, healthy diet, and social engagement, are associated with less brain aging in midlife (Franz et al., 2021, Whitsel et al., 2022). However, few studies have examined how brain reserve, operationalized with Brain-PADs, relates to cognitive trajectories in midlife, prior to the onset of MCI or AD.
Executive functions (EFs) are particularly important to examine in relation to brain reserve and brain aging. They represent a set of processes that control and regulate behavior (Friedman and Miyake, 2017), often including measures of response inhibition, task-set shifting, and/or working memory updating. These abilities are highly correlated, so a construct of ‘common executive function’ ability is used to quantify them and their relation to health and aging outcomes (Friedman et al., 2008, Gustavson et al., 2019, Gustavson et al., 2023b). EFs are not only some of the first cognitive abilities to decline in normal aging (Bakkour et al., 2013, Buckner, 2004), but are also highly relevant to pathological aging. For example, EF deficits predict progression from MCI to dementia and may do so better than traditional memory-based MCI classifications (Junquera et al., 2020).
Some studies have demonstrated that having a brain age comparatively younger than chronological age is associated with better concurrent EF ability, including performance on the trail making test (Cole, 2020) and visual (but not auditory) EF tasks (Huang et al., 2022). To date, however, there appear to be no studies examining whether Brain-PADs predict longitudinal changes in EF. Quantifying whether Brain-PADs relate to EF changes across midlife and early old age would further demonstrate their utility in capturing brain reserve (and related risk factors) before cognitive changes manifest.
When investigating these associations, it will be important to consider cognitive reserve and AD genetic risk factors such as APOE genotype alongside brain reserve, including their potential interactions. Cognitive reserve, which we define as an “individual’s total or overall cognitive resources” at a given point in time (Kremen et al., 2022), is an important factor which may help identify individuals who do not exhibit the cognitive or functional deficits expected based on their brain pathology (Barulli and Stern, 2013, Kremen et al., 2022). Our focus in the present report is on young adult cognitive reserve, which is not confounded by later aging effects. On one hand, cognitive reserve may moderate the association between brain reserve and executive function, suggesting nonadditive effects of brain and cognitive reserve in prediction of cognitive changes. Alternatively, the absence of moderation would be consistent with cognitive and brain reserve conveying independent, additive protective effects. The few existing studies on this topic are consistent with the latter model in which cognitive reserve and (concurrent) neuroimaging measures independently relate to executive function performance, but do not interact (Gustavson et al., 2023a, Krch et al., 2019). However, it will be important to test this possibility using a longitudinal design where both cognitive and brain reserve are measured prior to EF changes. Similarly, brain reserve may predict EFs more strongly in individuals at higher AD genetic risk (e.g., APOE e4 genotype). Indeed, APOE dosage scores appeared to drive the association between AD polygenic scores and EF changes across midlife in a prior study using this sample (Gustavson et al., 2023b). It will also therefore be important to examine whether AD genetic risk and brain reserve convey independent, additive effects, or if one set of risk factors magnifies the impact of the other.
Finally, it will be important to understand how brain reserve relates to cognitive changes across midlife and old age. Specifically, both Brain-PADs and EFs in midlife are highly heritable and stable over midlife (Gillespie et al., 2022, Gustavson et al., 2018). For example, using the BARACUS approach (Liem et al., 2017) to calculating brain age, our group found that 74% of the variance in Brain-PAD between ages 51 and 72 years was captured by a common set of genetic influences (Gillespie et al., 2022). Thus, it is possible that any association of EF and Brain-PAD would reflect shared genetic associations, consistent with Brain-PADs indexing risk factors (including genetic risk) that precede cognitive changes. Twin analyses, which parse variance in independent variables into genetic, shared environmental, and nonshared environmental influences are useful in disentangling these associations.
We examined whether brain reserve in midlife would be associated with concurrent EF abilities and change in EF abilities over the next 12 years using data from the Vietnam Era Twin Study of Aging (VETSA) project. MRI measures were available at mean age 56 and comprehensive EF assessments were available in this sample at three waves of assessment (mean age 56, 62, and 68). Using latent growth curve models, our first goal was to evaluate whether Brain-PADs (at baseline) were associated with baseline and change in EF abilities across this time period. Second, we evaluated whether young adult cognitive reserve (i.e., general cognitive ability assessed 35 years before VETSA began) and APOE genotype moderated observed associations between brain reserve and EFs. Finally, we used the classical twin design to estimate genetic and environmental correlations between brain reserve and EF.
留言 (0)