Invited Commentary: Body Mass Index and Risk of Dementia—Potential Explanations for Life-Course Differences in Risk Estimates and Future Research Directions

Abstract

The relationship between body mass index (BMI) and health outcomes of older adults, including dementia, remains controversial. Many studies find inverse associations between BMI and dementia among older adults, while in other studies high BMI in midlife is associated with increased dementia risk. In this issue, Li et al. (Am J Epidemiol. 2021;190(12):2503–2510) examine BMI from mid- to late life and risk of dementia using the extensive follow-up of the Framingham Offspring Study. They found changing trends in the association between BMI and dementia from a positive association for midlife (ages 40–49) to an inverse trend in late life. Their work demonstrates the importance of studying dementia risk factors across the life course. Midlife obesity might be an important modifiable risk factor for dementia. However, because incipient dementia can lead to weight loss, reverse causation remains a key source of bias that could explain an inverse trend between BMI and dementia in older ages. The extent of other biases, including unmeasured confounding, inaccuracy of BMI as a measure for adiposity, or selective survival, are also unclear. Triangulating evidence on body composition and dementia risk could lead to better targets for dementia intervention, but future work will need to evaluate specific pathways.

Abbreviations

AD

Alzheimer disease

BMI

body mass index

 

Editor ’s note: The opinions expressed in this article are those of the author and do not necessarily reflect the views of the American Journal of Epidemiology.

Dementia is a clinical syndrome of severe cognitive impairment that affects daily activities of living, and it is one of the leading causes of death (1). There are no effective treatments to slow disease progression in most cases of dementia, including Alzheimer disease (AD), the most common cause of dementia (1). There is substantial interest in identifying modifiable risk factors for dementia and in developing effective strategies to prevent or delay dementia onset. Potentially modifiable factors of interest include education, social isolation, hearing loss, cardiovascular disease, diabetes, smoking, physical activity, and obesity (2).

Obesity is hypothesized to increase dementia risk through cell-signaling proteins released by adipose (fat) tissue and the negative effects of obesity on cardiometabolic function and disease (3). However, observational findings on body mass index (BMI) are inconsistent across the life course. Generally, obesity or high BMI in midlife is associated with higher risk of dementia (4–8). But in late-life studies, there is often an inverse association between high BMI and dementia (5, 7–9). Not all studies have found these same associations (6). Few studies have examined BMI across the mid-to-late life span in the same subjects.

Using extensive longitudinal data from the Framingham Heart Study Offspring Cohort, Li et al. (10) found that higher BMI in ages 40–49 was associated with increased risk of dementia, but that for ages 50–59, there was no association. By ages ≥70 a trend towards an inverse association between BMI and dementia risk reached borderline significance. Accounting for competing risk of death attenuated estimates but did not change inferences. These results support prior findings that obesity in midlife but not late life is associated with increased risk of dementia; however, their use of smaller age categories suggested that a switch towards inverse association might begin even in ages typically considered midlife. Their study highlights the importance of a life-course perspective in dementia research but also raises potential challenges in understanding the effects of mid- versus late-life risk factors for dementia. Here, I will expand upon the explanations raised by Li et al. for the associations of BMI and dementia in mid- and late life as well as discuss future research directions.

CLARIFYING EVIDENCE ON MIDLIFE RISK FACTORS COULD INFORM DEMENTIA INTERVENTIONS AND TIMING

A challenge in health research is the difficulty of studying and identifying risk factors across the life course. This is especially true for dementia, which is typically not diagnosed until ages 75–85 years but is preceded by years of neurodegeneration and subtle cognitive and behavioral changes (11). Most studies on dementia, however, occur among older adults who might be already experiencing neurodegeneration and cognitive decline. Cohort studies that originally studied cardiovascular disease, such as the Framingham Heart Study and Offspring cohort leveraged by Li et al., offer substantial promise for studying life-course risk factors by continuing data collection and examining cognitive outcomes into late life (10). Studies that can assess which early or midlife risk factors predict dementia could help identify causes of dementia and could inform at what ages to target for preventive interventions.

The Lancet Commission on dementia prevention estimated that 1% of dementia is attributable to obesity (BMI ≥30) in midlife and up to 6% is attributable to health factors associated with obesity, such as diabetes, hypertension, and low physical activity (2). Together this is equivalent to the estimated dementia burden attributable to the strongest genetic variant (7% for the apolipoprotein E ε4 allele) (2). This suggests that interventions targeting healthy lifestyles in midlife, including weight loss, could significantly reduce dementia burden. However, these estimates of burden reduction should be treated with caution because they are based on observational studies that are susceptible to biases. The long preclinical period of dementia, in particular, remains a key source of bias that can mask whether, when, and by what magnitude high BMI increases risk for dementia.

ACCOUNTING FOR REVERSE CAUSATION OF PRECLINICAL DISEASE IS A CHALLENGE IN OBSERVATIONAL DEMENTIA STUDIES

Li et al. (10) found that the positive association between BMI and dementia flips to a null or negative association as early as ages 50–59 years. A null or inverse association with dementia risk in later life could be due to a number of reasons, similar to the “obesity paradox” in other health settings (12). While higher BMI in late life might have some physiological benefits (13, 14), many potential biases or combinations of factors could also help explain an “obesity paradox” (12). Such a result could be due to survivor bias; competing risk of death; reverse causation or confounding by weight loss in preclinical disease; confounding by health status, physical fitness, or obesity duration; or inaccuracy of BMI as a measure of adiposity. In the context of dementia, the impact of preclinical dementia-related pathophysiology on research findings in later life is especially important to consider and difficult to rule out.

Weight loss is common in those with dementia and predicts subsequent AD or dementia diagnosis (5, 15–17), AD neuropathologic burden at autopsy (18), and AD neuroimaging biomarkers even among cognitively normal older adults (19, 20). An association between BMI and dementia could conflate weight loss due to preclinical dementia disease with any negative health effects of BMI level. For example, in one study, high BMI was associated with brain atrophy even in late life, once individual trends in weight loss were incorporated into models (21). By examining longitudinal trajectories of BMI, another study found that individuals who developed dementia in late life were more likely to have higher BMI in midlife than those without dementia, but they experienced weight loss up to 12 years before dementia diagnosis (7).

THE EARLIEST AGES OF DEMENTIA PATHOLOGY–RELATED WEIGHT LOSS MIGHT BEGIN DECADES PRIOR TO DIAGNOSIS

There are conflicting estimates of when lower BMI or weight loss becomes associated with dementia; most suggest 8–12 years before dementia diagnosis, but estimates range up to 20 years (5, 7, 8). A study of individuals with genetic mutations for early-onset AD found that BMI trajectories began to diverge 18 years before expected symptom onset compared with BMI trajectories of family members without the mutations (22). These individuals have rare autosomal dominant mutations for AD with dementia onset expected before age 65. Thus, any changes in BMI are likely attributable to AD pathophysiology rather than effects of other chronic diseases or aging. If these findings can be extended to late-life dementia, then identifying the ages at which null or inverse association between high BMI and dementia begins might clarify timing of the earliest clinical manifestations of dementia pathologies. This would be useful both for identifying potential intervention windows for slowing dementia onset and for identifying when the biasing effects of reverse causation could occur in observational studies on life-course risk factors for dementia.

Li et al. (10) found that high BMI was not associated with increased risk of dementia as early as ages 50–59, and although they did not find significant inverse associations, there is a consistent trend in later age groups. These suggest that this trend might begin over 20 years before dementia diagnosis (1). This flip in the association is earlier than other studies (4, 6–8) but might explain some null or inverse findings of high BMI and dementia when using a broader age ranges to define midlife BMI (9, 23). Although this hypothesis is attractive, future work is needed to establish when the earliest signs of AD and related dementias begin. Additionally, other potential biases, such as earlier mortality in unhealthy individuals with obesity, might also drive some of the inverse association between late-life BMI and risk of dementia.

REMAINING QUESTIONS FOR FUTURE INVESTIGATION

Many questions remain regarding the role of BMI and body composition on dementia risk. BMI is a convenient measure but might not capture unhealthy excess weight or adiposity (amount of fat tissue) accurately (3). Future studies are needed to clarify what specific aspects of obesity, duration, and trajectories of weight gain/loss are most strongly associated with increased risk of dementia. Some individuals with overweight- or obese-category BMI could be metabolically “healthy” (14); one question is whether middle-aged adults with high BMI, but healthy behaviors or metabolically healthy obesity, are also at higher risk for dementia. Higher BMI is also correlated with higher levels of lean mass whereas lower BMI and weight loss are associated with frailty and other factors that can increase dementia risk (13, 14). Thus, high BMI might be less accurate as a representation of unhealthy obesity in older ages, in particular, due to selective survival of those with “less risk obesity” (13). These factors could help account for some of the inverse association seen with dementia in older ages, although the effects of weight loss due to incipient dementia is probably the main driver of this trend. It will be important to tease apart these additional potential explanations.

There might important subgroup differences as well. Obesity rates vary by sex/gender and geographic region, and they are highest among those with low socioeconomic status and, in the United States, among non-Hispanic Black and Hispanic persons (24); however, prior research has been conducted primarily in White and well-educated populations. Although some prior work has examined sex/gender differences (23) and associations in African Americans (25), additional studies on life-course BMI and dementia in diverse populations are needed. Finally, there might also be differences in associations by dementia subtypes, which warrant further investigation. Some research suggests that midlife obesity could be a risk for vascular or non-AD dementia (23, 26) and that weight loss is associated only with nonvascular dementias including AD (23).

FUTURE RESEARCH DIRECTIONS FOR EVIDENCE TRIANGULATION

There are challenges to answering these questions with traditional observational studies. It is generally not possible to establish temporal ordering of BMI or other measures of adiposity relative to established markers of dementia pathologies or to control for all potential confounders. Li et al. (10) were able to examine BMI across an impressive range of follow-up; however, similar to other, prior studies their data cannot easily evaluate specific causal mechanisms. Studies that attempt to test specific causal pathways or have different underlying assumptions (27), such as Mendelian randomization (MR) or mediation analyses, might further help triangulate evidence around BMI, obesity, and dementia. MR studies take advantage of genes as instruments or analytical tools that are not as susceptible to unmeasured confounding or reverse causation (28). To date, MR studies have not found strong evidence that genes for high BMI cause dementia (29, 30), questioning the role of midlife BMI on dementia risk. However, some genes increase BMI but result in metabolically favorable adiposity that is associated with reduced risk of cardiovascular disease and diabetes (31). Combining MR and other study approaches that can study adiposity subgroups and metabolic pathways might yield further insights into cases of dementia.

CONCLUSIONS

Li et al. (10) highlight the potential importance of midlife risk factors for dementia and the need for further understanding of risk factors across the life course. Future studies are needed to confirm: 1) at what stages in the adult life course (if any) does obesity increases risk for dementia? And 2) at what age does weight loss due to dementia pathologies begin? The most promising future research directions and approaches might be those that can isolate specific mechanisms and hypotheses regarding BMI, body composition, and dementia.