Obesity in Midlife and Late-Life Dementia
Obesity in Midlife and Late-Life Dementia
In a large population-based sample of men and women, we found that being overweight or obese in midlife was not significantly associated with lower total brain volume, focal brain vascular damage, or dementia. These results were robust to advanced alternative model specifications in our sensitivity analyses.
Strengths of the study include the large, well-described cohort; the availability of mid- and late-life data, including standardized height and weight measures; clinical measures and information on potential confounders measured both in midlife and late life; MRI measures of percentage of TBV and percentage of WMLV obtained using a previously validated automated segmentation algorithm; standardized criteria for diagnosing BIs and CMBs; dementia diagnosis established through clinical consensus; mortality data on the total original cohort; and formal correction for potential selective attrition.
Several issues related to the interpretation of our findings should be noted. Central rather than overall adiposity might be a more valid indicator of the resultant metabolic changes, but those measures were not taken at the midlife examination. However, BMI is widely used in adults to reliably capture both total and central adiposity. Further, we investigated the percentage of TBV as measure of global brain atrophy by inferring it from cross-sectional data. It is possible that a high BMI could have specific localized effects on smaller regional brain volumes, for instance hippocampal volume. Hypothesis-generating follow-up studies are needed to assess this issue. We took several steps to investigate whether bias from selective participant loss might have influenced our results. Participation rates were higher and mortality rates were lower in normal-weight participants than in those who were overweight or obese in midlife. Further, compared with participants who were excluded, those in the analytic sample had a slightly better cardiovascular risk profile. However, to affect the study conclusions, survival bias would have to be in the direction of missing data on obese participants with a lower risk of dementia and brain damage, and this is likely minimal or unlikely.
Within the analytic sample, overweight and obese participants had significantly worse mid- and late-life cardiovascular risk factor profiles for mortality and brain damage than normal-weight participants (including higher blood pressure and higher rates of diabetes). When those variables were included in the models, their effect was very modest. Moreover, our results did not change when we applied inverse probability weighting to our models to formally account for the potential influence of selective attrition on the association of midlife overweight and obesity with dementia and brain damage. Further, results were unchanged for dementia when participants with missing MRI data were included in the sample. However, these additional analyses cannot completely account for the selected attrition. Across the categories of BMI and brain outcomes, there is a small but relatively consistent tendency for obese participants to have less risk for cerebral vascular pathology than overweight or normal-weight participants. Persons with a high BMI and low vascular disease burden might have a different risk for cardiovascular disease–related outcomes or other outcomes also associated with cerebral lesions.
Prospective evidence of the association of midlife BMI with brain atrophy and brain vascular damage (the structural changes underlying dementia) in late life is patchy. Our mid- to late-life prospective analyses of brain structural measures assessed via MRI (including brain atrophy and both focal (BI and CMB) and diffuse (WMLV) vascular damage), dementia, and mortality in a single population-based study have not been reported elsewhere. In addition, no prior studies on the association of overweight and obesity in midlife with dementia have formally accounted for the potential influence of selective attrition.
In a younger cohort, associations of midlife BMI with structural brain measures assessed via MRI were not significant after controlling for covariates comparable to the ones that we considered in the present study. In a smaller sample of relatively healthy older adults, neither central nor global midlife obesity was associated with brain atrophy, and measures of vascular brain damage were not available. Recently, in a meta-analysis, Anstey et al. suggested that there was a positive association between BMI and dementia risk. However, those results were largely driven by 2 large studies that relied upon available medical records to ascertain dementia. This might have introduced bias. Higher BMI is plausibly associated with greater morbidity and higher rates of hospitalization; therefore, dementia detection among participants who were leaner might have been lower because of less frequent health care use. Moreover, there might be issues related to residual confounding and measurement bias in studies in which investigators did not adjust for education level or when BMI calculations are based on self-reported height and weight. Further, not included in the meta-analysis were 3 studies that found no significant association between overweight and obesity measured in midlife and dementia in late life after adjustment for sociodemographic and health characteristics. Our results on the lack of a significant association of midlife overweight and obesity with dementia are consistent with those of studies that accounted for similar potential confounders, and the prevalence of midlife overweight and obesity in the present study did not differ from those observed in samples from previous studies.
It is biologically possible that excess adipose tissue plays a role in the long-term development of cerebral diseases. Mechanisms might include insulin resistance, glucose intolerance, and diabetes; adipocyte-produced hormones and proinflammatory cytokines (i.e., tumor necrosis factor-α and interleukin-6); and vascular risk factors and diseases. Moreover, the association of the risk variant of the fat mass and obesity-associated protein (FTO) gene with reduced brain volume but not with diffuse cerebrovascular damage (which is consistent with our findings on percentage of WMLV) provides indirect evidence of a possible unfavorable effect of high adiposity on the brain in late life. Nevertheless, other studies suggest that there are mechanisms that support a more complex association. For example, there might be neuropathological changes in the brain that build up several years before the clinical onset of dementia and might be associated with weight loss decades before old age. In recent studies, a reduced risk of dementia has been reported with increasing levels of leptin, a cytokine that is produced in the adipose tissue and known to enhance cognitive function and to exert neuroprotective effects. In this context, our results are consistent with a recent hypothesis, which posits that the association of obesity with dementia and brain damage could be largely confounded by genotype and life circumstances that may influence lifelong adiposity levels and vascular risk profile, as well as late-life cognitive ability and neuropathology.
The identification of modifiable factors on the pathway to dementia is key for prevention. Our results suggest that a clinical marker of high total adiposity in midlife might not be associated with a higher risk of brain pathology and dementia at a later age. Therefore, weight management interventions might not be effective for prevention of dementia. Nevertheless, the maintenance of a healthy body weight should be encouraged because excess body weight is associated with poorer health.
Discussion
In a large population-based sample of men and women, we found that being overweight or obese in midlife was not significantly associated with lower total brain volume, focal brain vascular damage, or dementia. These results were robust to advanced alternative model specifications in our sensitivity analyses.
Strengths of the study include the large, well-described cohort; the availability of mid- and late-life data, including standardized height and weight measures; clinical measures and information on potential confounders measured both in midlife and late life; MRI measures of percentage of TBV and percentage of WMLV obtained using a previously validated automated segmentation algorithm; standardized criteria for diagnosing BIs and CMBs; dementia diagnosis established through clinical consensus; mortality data on the total original cohort; and formal correction for potential selective attrition.
Several issues related to the interpretation of our findings should be noted. Central rather than overall adiposity might be a more valid indicator of the resultant metabolic changes, but those measures were not taken at the midlife examination. However, BMI is widely used in adults to reliably capture both total and central adiposity. Further, we investigated the percentage of TBV as measure of global brain atrophy by inferring it from cross-sectional data. It is possible that a high BMI could have specific localized effects on smaller regional brain volumes, for instance hippocampal volume. Hypothesis-generating follow-up studies are needed to assess this issue. We took several steps to investigate whether bias from selective participant loss might have influenced our results. Participation rates were higher and mortality rates were lower in normal-weight participants than in those who were overweight or obese in midlife. Further, compared with participants who were excluded, those in the analytic sample had a slightly better cardiovascular risk profile. However, to affect the study conclusions, survival bias would have to be in the direction of missing data on obese participants with a lower risk of dementia and brain damage, and this is likely minimal or unlikely.
Within the analytic sample, overweight and obese participants had significantly worse mid- and late-life cardiovascular risk factor profiles for mortality and brain damage than normal-weight participants (including higher blood pressure and higher rates of diabetes). When those variables were included in the models, their effect was very modest. Moreover, our results did not change when we applied inverse probability weighting to our models to formally account for the potential influence of selective attrition on the association of midlife overweight and obesity with dementia and brain damage. Further, results were unchanged for dementia when participants with missing MRI data were included in the sample. However, these additional analyses cannot completely account for the selected attrition. Across the categories of BMI and brain outcomes, there is a small but relatively consistent tendency for obese participants to have less risk for cerebral vascular pathology than overweight or normal-weight participants. Persons with a high BMI and low vascular disease burden might have a different risk for cardiovascular disease–related outcomes or other outcomes also associated with cerebral lesions.
Prospective evidence of the association of midlife BMI with brain atrophy and brain vascular damage (the structural changes underlying dementia) in late life is patchy. Our mid- to late-life prospective analyses of brain structural measures assessed via MRI (including brain atrophy and both focal (BI and CMB) and diffuse (WMLV) vascular damage), dementia, and mortality in a single population-based study have not been reported elsewhere. In addition, no prior studies on the association of overweight and obesity in midlife with dementia have formally accounted for the potential influence of selective attrition.
In a younger cohort, associations of midlife BMI with structural brain measures assessed via MRI were not significant after controlling for covariates comparable to the ones that we considered in the present study. In a smaller sample of relatively healthy older adults, neither central nor global midlife obesity was associated with brain atrophy, and measures of vascular brain damage were not available. Recently, in a meta-analysis, Anstey et al. suggested that there was a positive association between BMI and dementia risk. However, those results were largely driven by 2 large studies that relied upon available medical records to ascertain dementia. This might have introduced bias. Higher BMI is plausibly associated with greater morbidity and higher rates of hospitalization; therefore, dementia detection among participants who were leaner might have been lower because of less frequent health care use. Moreover, there might be issues related to residual confounding and measurement bias in studies in which investigators did not adjust for education level or when BMI calculations are based on self-reported height and weight. Further, not included in the meta-analysis were 3 studies that found no significant association between overweight and obesity measured in midlife and dementia in late life after adjustment for sociodemographic and health characteristics. Our results on the lack of a significant association of midlife overweight and obesity with dementia are consistent with those of studies that accounted for similar potential confounders, and the prevalence of midlife overweight and obesity in the present study did not differ from those observed in samples from previous studies.
It is biologically possible that excess adipose tissue plays a role in the long-term development of cerebral diseases. Mechanisms might include insulin resistance, glucose intolerance, and diabetes; adipocyte-produced hormones and proinflammatory cytokines (i.e., tumor necrosis factor-α and interleukin-6); and vascular risk factors and diseases. Moreover, the association of the risk variant of the fat mass and obesity-associated protein (FTO) gene with reduced brain volume but not with diffuse cerebrovascular damage (which is consistent with our findings on percentage of WMLV) provides indirect evidence of a possible unfavorable effect of high adiposity on the brain in late life. Nevertheless, other studies suggest that there are mechanisms that support a more complex association. For example, there might be neuropathological changes in the brain that build up several years before the clinical onset of dementia and might be associated with weight loss decades before old age. In recent studies, a reduced risk of dementia has been reported with increasing levels of leptin, a cytokine that is produced in the adipose tissue and known to enhance cognitive function and to exert neuroprotective effects. In this context, our results are consistent with a recent hypothesis, which posits that the association of obesity with dementia and brain damage could be largely confounded by genotype and life circumstances that may influence lifelong adiposity levels and vascular risk profile, as well as late-life cognitive ability and neuropathology.
The identification of modifiable factors on the pathway to dementia is key for prevention. Our results suggest that a clinical marker of high total adiposity in midlife might not be associated with a higher risk of brain pathology and dementia at a later age. Therefore, weight management interventions might not be effective for prevention of dementia. Nevertheless, the maintenance of a healthy body weight should be encouraged because excess body weight is associated with poorer health.