We investigated the prospective associations between midlife obesity and subsequent trajectories of brain atrophy in the BLSA. We report no associations between either central or global midlife obesity with subsequent rates of global and regional brain volume changes against the background of generalized age-related atrophy in non-demented older individuals. Although sparse, relationships between rates of brain atrophy and obesity begin to emerge when data up to the point of dementia diagnosis are included for those individuals who become impaired during the course of the study. The associations seem to be predominantly present in relation to global obesity and largely driven by frontal and temporal GM changes ().
When examining the associations with regional volume changes across the range of BMI and WC values as opposed to employing a cut-off point to define obesity, a slightly different pattern of results emerges. Higher BMI is associated with steeper decline in cingulate volume. Additional associations emerge in relation to larger WC, namely steeper rates of atrophy in whole brain and WM matter volume, especially once the data up to the point of dementia diagnosis are included for those individuals who become impaired during follow-up.
The pattern of result obtained across the range of BMI and WC values differed from the categorical analysis employing a cut-off to define obesity most likely due to greater power afforded by the continuous analysis. Majority of the findings, however, do not survive corrections for multiple comparisons. Overall, our results suggest little in terms of relationships between midlife obesity and rates of brain volume changes in non-demented older adults but that obesity may act as a modifier of regional brain atrophy in those who are succumbing to age-related impairment. To fully explore this possibility, a larger number of impaired individuals followed over longer intervals will allow for a more direct evaluation of the effect that impending cognitive impairment has on the relationship between brain volumes and obesity, currently not possible in the BLSA which is limited to those who become impaired during the course of the study.
Current literature, albeit still limited and predominantly cross-sectional in nature, identifies several regions of smaller volumes in association with obesity, namely the whole brain volume [Debette et al., 2010
; Enzinger et al., 2005
; Raji et al., 2010
; Ward et al., 2005
], GM [Pannacciulli et al., 2006
; Taki et al., 2008
], prefrontal regions [Enzinger et al., 2005
; Raji et al., 2010
; Walther et al., 2010
] and the temporal lobe [Jagust, 2007
; Walther et al., 2010
] including the hippocampus [Jagust et al., 2005
; Raji et al., 2010
], in various age groups of non-demented individuals. Given that we see no relationship between obesity and the rates of regional brain atrophy in the non-demented, both when employing an obesity cut-off and when looking across the range of values, our findings suggest that the existing cross-sectional studies may have overestimated the relationships between brain volumes and obesity, given that 1.) cross-sectional studies, especially those focusing on middle-aged adults or younger, may not be effective at screening out individuals who will eventually become impaired, and 2.) even the inclusion of data prior to diagnosis for those who became impaired may change the results substantially.
The study is not without limitations. Our sample is not population-based; the majority of the sample is highly educated and Caucasian. This, however, should not undermine many unique advantages of the study, such as the large number of follow-ups, relatively short periods between assessments, state-of-the art image processing methods, and the extensive screening and characterization of this cohort. The relative homogeneity of the sample can also be viewed as an advantage given that the majority of our sample has good access to medical care and has remained relatively healthy over the follow-up interval. Potentially, the biggest limitation is that obesity measurements at age 50 were not available for all NI participants and had to be estimated based on both their available data at later ages and the larger BLSA sample (N = 3005) which was demographically similar to the neuroimaging subsample. Although the number of globally obese individuals in this cohort was low, identical pattern of results was obtained in the non-demented regardless of the type of obesity (global vs. central) or analyses performed (continuous vs. using a cut-off to define obesity).
To the best of our knowledge, this is the first study exploring the relationship between midlife obesity and subsequent longitudinal brain volume changes in a number of regions of interest which have not been previously investigated in relation to obesity. Our longitudinal design addresses some of the limitations in the existent literature imposed by their cross-sectional nature or limited follow-ups. Together, our results suggest that that the relationships between global and central obesity and age-related brain atrophy may be small to non-existent in the non-demented older individuals. Furthermore, global and central obesity may have differential contributions to brain volume changes, especially in samples containing individuals who later become cognitively impaired.
More importantly, our finding suggests that the results and hence, the potential discrepancies in findings between different studies may be in part due to the fundamental difference in analyses and participant sampling. The fact that relationships with regional brain volumes seem to emerge when data are included, although only prior to dementia diagnosis, for those who eventually became impaired, suggest that the relationships between obesity and structural brain integrity should be further investigated separately with respect to normal and pathological aging. Indeed, a study by Ho and colleagues (2010)
suggests higher BMI is associated with lower brain volumes in frontal, temporal, parietal, and occipital lobes in those with AD and Mild Cognitive Impairment. Longitudinal studies will remain critical in determining the clinical relevance of observed brain changes in relation to obesity in both normal aging and cognitive impairment. Further studies are needed to elucidate the mechanisms by which midlife obesity may affect brain atrophy and subsequent risk for dementia. Such studies can better inform the development of animal models and set the stage for more effective population and epidemiological studies. Moreover, molecular mechanisms mediating the relationship between obesity parameters, brain structure, and dementia have yet to be elucidated [see Whitmer, 2007
for a review].