early vascular risk factors for later dementia

 a recent article assessed long-term vascular risk factors associated with future incident dementia, quantifying the early presence of hypertension, diabetes, and smoking to the subsequent development of dementia both before and after age 80, taking into account apolipoprotein ε4, and calculating the population attributable fractions (PAFs) for these risk factors individually and in combination: see dementia early risk for later dementia JamaNeuro2025 in dropbox or doi:10.1001/jamaneurol.2025.1495

Details:

-- 7731 participants in the Atherosclerosis Risk and Communities (ARIC) study with 33 years of follow-up from 1987 to 2020, in the current ARIC-NCS study (ARIC-Neurocognitive Study)

    -- this study included participants from Jackson, Mississippi; Forsyth County, North Carolina; Minneapolis suburbs, Minnesota; and Washington County, Maryland

    -- these sites reflected significant diversity in terms of socioeconomic factors, including race

-- overall demographics by vascular risk factors status and age at measurement:

    -- age 45 to 54: 7731 participants, mean age 49; 58% female; 29% Black/72% white; 18% less than high school education/42% high school/39% more; apolipoprotein ε4 carriers having at least one allele 31%; hypertension 42%; diabetes 9%; current smoking 28%; alcohol use never 24%/ former 17.6%/current 59%; BMI 27; total cholesterol 210 and HDL 52 mg/dL; follow-up time 27 years, incident dementia by age 80: 801 participants/10% of sample

    -- age 55 to 64: 12274 participants; mean age 59; 55% female; 24% Black/77% white; 23% less than high school education/41% high school/36% more; apolipoprotein ε4 carriers at least one allele 30%, hypertension 55%; diabetes 15%; current smoking 21%; alcohol use never 23%/former 24%/current 53%; BMI 28; total cholesterol 212 mg/dL and HDL 51 mg/dL; follow-up time 18 years; incident dementia by age 80: 995 participants/8% of sample

    -- age 65 to 74: 6787 participants; mean age 70; 56% female; 20% Black/80% white; 19% less than high school education/42% high school/39% more; apolipoprotein ε4 carriers at least one allele 31%, hypertension 71%; diabetes 24%; current smoking 11%; alcohol use never 22%/former30%/current 48%; BMI 28; total cholesterol 195 mg/dL and HDL 51 mg/dL; follow-up time 8 years, incident dementia by age 80: 422 participants/6% of sample

-- apolipoprotein ε4 genotype is the plasma variation in the apolipoprotein E (APOE) that is associated with the later risk of Alzheimer's disease (AD), with increasing association depending on the number of alleles present (one allele confers 2- to 3-fold risk of AD; 2 copies, as in those homozygous for APOE, confers an 8- to 12-fold increased risk vs noncarriers). About 60% of individuals who are homozygous for APOE ε4 seem to develop Alzheimer's, and about 40% of those with AD are not APOE ε4 carriers

-- hypertension was defined as a systolic pressure of at least 130 mmHg, diastolic blood pressure of at least 80 mmHg, or the use of blood pressure medications

-- diabetes was defined as a fasting glucose of at least 126 mg/dL, nonfasting glucose of at least 200 mg/dL, self-reported physician's diagnosis, of the use of diabetes medications

-- current smoking was self-reported

-- Main outcome measures: incident dementia; population attributable fractions (PAFs) were estimated before age 80, and separately after age 80; they also assessed dementia risk from having at least one vascular risk factor by different ages at risk factor measurement

-- Follow up was over 33 years

 Results:

-- The overall results of the study are best reflected in the following graphs from the study (easier to understand than my rattling off way-too-many numbers...)

-- the first graph is risk factor prevalence at the prespecified age groups, with hazard ratios and population-attributable fractions of the vascular risk factors associated with the development of incident dementia by age 80

-- the second graph is the same but for those developing dementia after age 80

--Overall interpretation of those developing dementia by age 80

    -- graph A (top left): the point prevalence of having a vascular risk factor, with the bars representing the prevalence of at least one risk factor as well as the specific risk factors measured, per each age group at entry

    -- graph B (middle left): the hazard ratio for each risk factor leading to dementia by age 80, adjusted for age, sex, race, education, at least one allele of APOE4, alcohol consumption, BMI, total cholesterol, HDL cholesterol, and leisure-time physical activity (the latter was available only for those 45-54yo) [note that there is a huge difference in risk of developing dementia if someone has 2 vs 1 alleles of APOE4, though this study put these 2 conditions together...]

    -- graph C (bottom left): the population-attributable fractions (PAFs) for vascular risk factors for incident dementia by age 80, with confidence intervals (many with overlapping confidence intervals).

    -- graph A (top right): PAF stratified by presence or absence of any APOE4 ε4 allele, finding that the attributable risk for the vascular risk factors is only statistically significant for the older age group at baseline (overlapping confidence intervals for the others)

    -- graph B (middle right): PAF stratified by race, finding trend to being higher in those self-identified as being Black, but no statistically significant difference for each age category (overlapping confidence intervals)

    -- graph C (bottom right): PAF by sex, also with trends to more dementia in women than men, but no statistically significant difference for each age category (overlapping confidence intervals)

-- Overall interpretation of those developing dementia after age 80, where the number of people developing dementia was far greater than in the younger group:

    -- graph on left: the relative risk of developing dementia increased with each risk factor with increasing age, though they were not statistically significant given the overlapping confidence intervals for all of them

    -- the PAFs for all of the risk factors had overlapping confidence intervals (not statistically significant); smoking was more evident in the younger cohort who later developed dementia, but the numbers of smokers did decrease dramatically with age and was present in only 79 people aged 65-74

therefore, 

-- lots of people have vascular risk factors associated with later dementia by age 80 (a majority at all ages, though tends to increase with age)

-- the calculated risk for dementia by age 80 is typically in the 50% to 100% increase for each of the risk factors, though this relative risk decreases with entry age (presumably because those with risk factors in the youngest group have them much longer than those who are older who develop those risk factors later in life), though the absolute risk remains higher with advancing age

-- individuals with 0 alleles for the APOE ε4 genotype had a PAF reflecting a much higher likelihood of developing dementia by age 80

-- the relationship between the vascular risk factors and developing dementia after age 80 is much less evident (though the incidence of dementia after age 80 is more common, there were no specific statistically significant associations between the risk factors and incident dementia)

-- the most common risk factor cluster was hypertension with diabetes, increasing in prevalence from 4.6% at age 45-54, to 17.9% at age 65-74

--sensitivity analyses:

    -- no real difference if one excludes people with baseline cardiovascular disease

    -- no real difference by statistically manipulating the populations to mathematically account for the expected confounding associated with the different age categories (ie, they created a "pseudo-population" of people with balanced confounders through Inverse Probability Weighting)

 Commentary:

-- this study chose hypertension, diabetes, and smoking as the significant vascular risk factors, but not elevated cholesterol, obesity, or alcohol consumption, since a former ARIC-NCS study had suggested that the first three measured in midlife and early life were stronger predictors of subsequent cognitive decline 

-- the presumed mechanism for dementia per these three vascular risk factors is arteriosclerotic cerebral small vessel disease associated with cognitive decline; this current study was independent of specific Alzheimer neuropathology (ie it assessed dementia, not by subtypes)

-- the selection of these four communities was to have a diverse sampling of the US population. There were notable differences in particular with the racial composition from area to area, and the largest number of Black individuals included were from a single site (Jackson, Mississippi), so it is not possible to differentiate racial from the very different social conditions in this site from the other three). Of course, the current understanding of racial differences in general does not reflect genetic differences (especially since actual genetic testing suggests large admixtures of different genetic backgrounds in many individuals), but more profoundly reflects differences in other major socioeconomic factors such as stress, types of jobs and exposures, intensity and diversity of medical conditions associated with these, access to healthcare, financial disparities, housing quality and crowding, differential access to healthy foods and gyms etc, etc

-- the overall findings of the current study:

    -- 22% to 44% of dementia cases by age 80 could be attributed to poor vascular health through age 75

         -- this increase with age is largely related to several issues: the combination of multiple risk factors increases with age. for example, aging is associated with more obesity, and that is associated with more hypertension and diabetes. also there is typically less leisure-time physical activity

    -- those with no allele for APOE ε4 had a larger role for the vascular risk factors in subsequent dementia at age 80, confirming the extensive literature on this genetic role in incident dementia

    -- there are some countervailing risk factor changes that could affect the intensity of vascular risk factors. For example, blood pressure in general tends to decrease in the elderly (https://gmodestmedblogs.blogspot.com/2018/01/decreasing-blood-pressure-in-elderly.html ). Also, for several reasons, vascular risk factors may be treated more aggressively in older people (younger ones may not seek the same level of medical care, clinicians may be more reticent to follow younger patients closely to monitor and treat mild elevations of risk factors, though these risk factors over time likely lead to higher cumulative levels of vascular disease and damage....)

-- Another recent study affirmed and extended the relationship between diet and dementia, focusing on the Mediterranean Diet (MedDiet): see dementia Med diet counters genetics NatMed2025 in dropbox, or doi.org/10.1038/s41591-025-03891-5 )

    -- as background, the prior PREDIMED study found that the Mediterranean diet was associated with decreased later cognitive decline and Alzheimer's, as well as improved anatomical brain structures on imaging: https://gmodestmedblogs.blogspot.com/2015/11/mediterranean-diet-and-brain-volume.html

    -- this study assessed the interplay of APOE4 ε4 and Mediterranean diet in 2 long-term groups of people, along with Mendelian randomization analysis

    -- assessment: 4215 women in the prospective Nurses' Health Study (NHS) with 34 years of followup, with replication in 1490 men in the Health Professionals Follow-up Study (HPFS); and a 2-sample Mendelian randomization, specifically targeting metabolomics (which, as opposed to genetics, assesses potential exogenous factors that might serve as targets for intervention, in this case diet)

        -- the huge advantage of these groups of people is the large numbers assessed and the long-term followup with assiduous detail to changes in diet by assessment every 2 years; the drawback is that the sample was all health professionals, almost all white.

    -- to validate the dementia outcome, they assessed plasma phosphorylated tau-217, an established biomarker for early Alzheimer's

    -- findings in this rather complex study:

        -- overall, individuals with greater adherence for the MedDiet had significantly lower risk of dementia and better cognitive function

        -- the most pronounced protective association of the MedDiet was in people who were APOE4 homozygotes vs either heterozygotes or noncarriers (ie, the higher the genetic proclivity to dementia, the higher the benefit of MedDiet), with similar patterns in both of the NHS and HPFS cohorts

            -- this correlation was most pronounced in those who adhered best to the MedDiet

        -- individual components of MedDiet most strongly associated with the metabolomic patterns consistent with cognitive health included nuts, fruit, and monounsaturated fats (ie, drinking moderate amounts of wine as in the full Mediterranean diet was not identified as decreasing dementia risk)

        -- 57 metabolites were identified that varied with dementia risk in APOE4 homozygotes; eg, there was an inverse association with glycerides which was evident only in APOE4 homozygotes

         -- Mendelian randomization analysis identified 19 putative causal relationships between metabolites and cognitive outcomes, including protective effects of 4-guanidinobutanoate, carotenoids, and N6-carbamoylthreonyladenosine.

            -- interestingly, 4-guanidinobutanoate is a gamma-aminobutyric acid (GABA)-related metabolite involved in inhibitory neurotransmission and may counteract neural excitotoxicity, known to contribute to dementia pathogenesis

            -- of likely clinical significance to clinicians, gabapentin does not directly affect GABA receptors: https://pmc.ncbi.nlm.nih.gov/articles/PMC3499716/; gabapentin modulates the release of GABA by binding to a protein called the α2δ subunit of voltage-dependent calcium channels. This binding prevents calcium from entering the neuron, which thereby reduces the release of excitatory neurotransmitters, including glutamate

            -- and there are studies finding that gabapentinoids are associated with increased dementia risk: https://pmc.ncbi.nlm.nih.gov/articles/PMC10266423/

                -- yet another of many reasons to avoid prescribing gabapentinoids: https://gmodestmedblogs.blogspot.com/2025/04/pain-and-gabapentinoids-no-clear.html

        -- all statistically significant interactions found were specific to those individuals who were homozygotes for APOE4; the 49 significant metabolites found in regards to dementia risk were associated with these individuals homozygous for APOE4. this plasma metabolomic profile associated with dementia risk was evident decades before disease onset (ie, this decreases possibility of reverse causation)

        -- overall, they found that their baseline model associated with dementia risk included age, family history of dementia, education level, smoking status, history of depression or regular antidepressant use, and the MedDiet index

            -- adding APOE ε4 in their polygenic risk score for Alzheimers-related dementias improved the predictive model

            -- and further adding metabolites predictive of dementia risk further improved the model

        -- this study brings up the rather paradoxical finding that those at the highest genetic risk for dementia (homozygous for apolipoprotein ε4) had the most benefit from the Mediterranean diet. of interest, this finding follows a previous genetic study that found that those at the highest genetic predisposition to weight gain actually had the most beneficial effect when they adhered to a healthier diet (see http://gmodestmedblogs.blogspot.com/2018/01/dietary-effect-strong-when-high-genetic.html). no clear reason for this. but, to me as a non-geneticist, perhaps there is evolutionary pressure for humans with APOE4 ε4 to developing other genetic mutations that confer some dementia protection in the setting of a healthy diet. of course, the only way that there can be such evolutionary pressure on the genome is if that pressure occurred in fertile couples. And the risk of dementia in the setting of homozygous apolipoprotein ε4 is associated with early-onset AD, which can occur in women around age 45 and in men around age 50 (https://alz-journals.onlinelibrary.wiley.com/doi/10.1002/alz.12955), both of which could still be associated with having live offspring

            -- also, of note, the presence of high Lp(a) levels was found in the LOAD study (late-onset AD) to be a risk factor for developing AD. for more on Lp(a), a very strong risk factor for general cardiovascular disease, see https://gmodestmedblogs.blogspot.com/2025/06/high-lpa-increases-risk-of-recurrent.html

-- a prior genetic study utilizing the large UK Biobank of 196,383 individuals with 8 years of followup found that a favorable lifestyle (diet, smoking, physical activity, less alcohol) also found that lifestyle decreased the risk of dementia even in those with high-risk genetic proclivity to dementia: https://gmodestmedblogs.blogspot.com/2019/07/dementia-genetics-and-lifestyle-both.html 

 Limitations of the main study:

-- perhaps the biggest limitations are:

    -- we do not know about changes over time in these risk factors: the results could be very different if these baseline numbers changed over the next few years than if they were persistent

    -- we do not know about therapies (what medications were taken for the different vascular risk factors, how consistently they were taken, or how the levels of incident dementia correlated with the adequacy of risk factor control)

    -- we do not know about changes in lifestyles that clearly could affect the persistence of these risk factors: what was the quality of the food eaten and how did that change over time, what was the intensity of exercise and how did that change over time, what were the levels of stress and general social supports over time, what were the levels of environmental and occupational risk factors, etc.

-- there were clearly substantial differences in these risk factors by the defined age cohorts above. Mathematical modeling helped control for these differences, but that is less robust than having actual individuals with similar potential confounders

-- though leisure-time physical activity is very likely related to the development of dementia (https://gmodestmedblogs.blogspot.com/2022/11/dementia-mortality-prediction-model.html), it was only recorded for the youngest age bracket in the study, limiting our understanding of the role of physical activity in those over 54yo

-- several of the data elicited were from self-report, potentially limiting their accuracy

-- reverse causation is always a concern (did dementia lead to differences in diet, exercise, etc)?? but this is much less of an issue with the very long followup in this study

so,

-- many of the vascular risk factors begin early in life and are associated with increasing vascular disease over time, and they accrue as increasing risk factor combinations with age

-- which very likely means that early identification of these risk factors and aggressive early treatment of them through nonpharmacologic and pharmacologic means will undercut the cumulative adverse vascular effects of these risk factors and lead to lower levels of dementia over time

-- these articles do suggest that though genetic proclivity to dementia as mediated by apolipoprotein ε4 status is clear in the medical literature, it is not determinant; there may well be a perhaps complex interaction between identified genetic factors and social conditions

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them