Association of dietary sodium intake with late‐life cognitive decline and postmortem neuropathology in community‐dwelling older adults

Abstract

INTRODUCTION

Studies link sodium intake to cognition, but most were cross‐sectional, and associations with neuropathology remain unclear.

METHODS

1520 community‐dwelling adults (mean age 80.7 years) had annual evaluations (mean follow‐up 7.6 years), including 19 neuropsychological tests and a food frequency questionnaires (FFQ). Postmortem neuropathologies were documented in 717 deceased participants. Associations of dietary sodium intake estimated from the FFQ with cognitive decline and neuropathology were examined using adjusted regression models.

RESULTS

Compared to the middle sodium intake quintile (median 2264 mg/day), the lowest quintile (1764 mg/day), but not the highest (2800 mg/day), was associated with faster decline in global cognition, particularly episodic memory and semantic memory. Among the deceased, the lowest quintile was associated with higher immunohistochemistry‐based tau tangle density, but not with amyloid burden or infarcts.

DISCUSSION

Low dietary sodium intake is associated with faster cognitive decline and more neuropathology, specifically tangles. Dietary sodium likely plays a role in brain health.

Highlights

• The relation of dietary sodium to brain health is not well understood.

• In this community‐based study, FFQs were used to estimate dietary sodium intake.

• Longitudinal neurocognitive and postmortem neuropathology evaluations were conducted.

• Among 1520 older adults without known dementia followed annually for 7.6 years, low sodium intake was associated with faster decline in global cognition.

• Among a subset who died and came to autopsy, low sodium intake was associated with higher tangle density, but not with amyloid burden or infarcts.

1. BACKGROUND

Dementia is one of the most prominent and debilitating chronic conditions of aging, with no known cure, making its prevention a top public health priority. ¹ , ² One important strategy that has been proposed to prevent dementia involves dietary interventions. ³ , ⁴ In particular, some studies have found that greater adherence to a healthy diet is associated with slower late‐life cognitive decline and a lower burden of postmortem Alzheimer's disease (AD) neuropathology, a major contributor to dementia. ⁵ , ⁶ , ⁷ Although various healthy diets share common features, they may differ in key components, such as sodium intake. Some diets emphasize limiting sodium to 2300 mg/day or less, ⁸ while others do not impose such restrictions. ⁷ , ⁹ This raises the question of whether sodium intake is associated with cognitive health.

High sodium intake is well established as a risk factor for hypertension, ¹⁰ which in turn is a known contributor to dementia. ¹¹ Given this link, several studies have examined the relationship between sodium levels and cognitive function, but the findings remain mixed. ¹² Some research suggests that lower sodium intake is associated with better cognition, ¹³ , ¹⁴ , ¹⁵ , ¹⁶ while other studies found no significant relationship. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ One study even reported that lower sodium intake was linked to poorer cognitive performance. ²¹ These inconsistencies may stem from differences in the demographic and clinical characteristics of study populations, as well as differences in study design and the relatively small sample sizes in most prior studies. Furthermore, little is known about brain mechanisms underlying the potential link between sodium intake and cognition. A 4‐week dietary intervention study found that, in cognitively normal middle‐aged adults, a Mediterranean‐like low sodium (approximately 1300 mg/day for 2000 calories) diet improved cerebrospinal fluid biomarkers for AD and cerebral perfusion, whereas a Western‐like high‐sodium (approximately 3200 mg/day for 2000 calories) diet had the opposite effects. ²² However, whether these findings extend to older adults or have lasting effects remains unclear. Moreover, we are not aware of any previous study that examined the association of sodium intake with postmortem neuropathologic changes.

The goal of this study was to investigate the association between baseline daily dietary sodium intake, as assessed by validated food frequency questionnaire (FFQ) data, and longitudinally measured cognitive function (global cognition and five individual cognitive domains) in 1520 participants of the community‐based, clinical‐pathologic Rush Memory and Aging Project (MAP). Furthermore, in a subset of 717 participants who were deceased and autopsied, we sought to examine the association of mean sodium intake from baseline through follow‐up with postmortem neurodegenerative neuropathologies (including amyloid, tau tangles, transactive response DNA binding protein 43 kDa [TDP‐43], and Lewy bodies) and cerebrovascular neuropathologies (brain infarcts, atherosclerosis, arteriolosclerosis, and cerebral amyloid angiopathy, i.e., cerebral amyloid angiopathy [CAA]). We hypothesized that dietary sodium intake was associated with cognitive decline and neuropathology, without specifying the direction of these associations due to mixed findings on cognition and scarce prior data on neuropathology.

2. METHODS

2.1. Study population

All participants of this study were from the MAP, a longitudinal, community‐based, clinical‐pathologic cohort study of cognitive aging. ²³ MAP received Institutional Review Board approval from Rush University Medical Center and began enrolling older adults without known dementia from retirement communities across the Chicago metropolitan area since 1997. Participants signed an informed consent to undergo annual clinical evaluations including dietary assessment (started in 2004) and neuropsychological testing, and a subset signed an anatomical gift act to donate their brain at the time of death. ²³

For this study, we included only participants who had at least one complete dietary assessment, with the first complete dietary assessment serving as the analytical baseline. We also required the included participants to have no dementia diagnosis at the analytical baseline and to have neuropsychological testing data available at both the analytical baseline and at least one follow‐up visit. Of a total of 2425 MAP participants enrolled in the parent study, at the time of analysis, 1520 older adults met the inclusion criteria. The age at the analytical baseline ranged from 57.9 to 98.7 years, with a mean of 80.7 years. Of these, 855 died, and 744 underwent an autopsy (87% autopsy rate). Of those who were autopsied, 717 had neuropathologic data available.

RESEARCH IN CONTEXT

1. Systematic review: Using PubMed, we reviewed the literature which showed an association of dietary sodium intake with late‐life cognitive decline. Although no human study has examined associations of sodium intake with postmortem neuropathologies, one study has identified association of sodium intake with cerebrospinal fluid markers of AD pathology in middle‐aged adults.

2. Interpretation: Our study found that low baseline sodium intake was associated with faster late‐life cognitive decline, suggesting adequate sodium intake may be important for maintaining brain health. Among the deceased participants, low sodium intake was associated with a higher density of neurofibrillary tangles, but not with Aβ accumulation. These findings suggest that low sodium intake might contribute to hyperphosphorylated tau accumulation and cognitive impairment through unidentified mechanisms.

3. Future directions: Future studies should further explore how age and other demographic, lifestyle, and clinical factors may affect the association of sodium with cognitive decline, as well as examine underlying mechanisms linking sodium with neuropathology.

2.2. Neuropsychological testing

At each study visit, participants completed 21 neuropsychological tests. Nineteen of these tests were grouped to form composite measures of global cognition and five cognitive domains: episodic memory (seven tests), working memory (four tests), semantic memory (four tests), perceptual speed (two tests), and visuospatial abilities (two tests). ²⁴ Raw scores of individual tests were converted to z‐scores, using the baseline mean and standard deviation from the entire cohort. These z‐scores of individual tests were then averaged to calculate composite scores for global cognition or individual cognitive domains. Detailed information about individual tests has been published elsewhere. ²⁵ , ²⁶

2.3. Dietary assessment

Dietary sodium intakes were estimated using a modified Harvard semi‐quantitative FFQ that was validated for use in older Chicago community residents. ²⁷ The FFQ determines the typical frequency of intake of more than 137 food items and dietary supplements over the past 12 months. The reported frequency of food consumption was multiplied by the nutrient content of natural portion sizes (e.g., one apple) or sex‐specific mean portion sizes from the oldest men and women in national surveys. The nutrient content of each food was based on United States Department of Agriculture (USDA) food composition tables and supplementary nutrient databases (https://fdc.nal.usda.gov/). Daily nutrient intakes for each participant were computed by summing over the frequency‐based nutrient content of all the food items. All nutrients, including sodium intake, were calorie‐adjusted using the residual regression method within sex. ²⁸

Sodium intake was computed as milligrams per day and assessed in quintiles. We designated the middle quintile as the reference group, considering that both low and high sodium intakes could adversely affect cognition. Additionally, the dietary guidelines for Americans 2020–2025 (available at DietaryGuidelines.gov) recommend limiting sodium intake to no more than 2300 mg/day, which is very similar to the median intake of our middle quintile (2264 mg/day). According to What We Eat in America (WWEIA), the National Health and Nutrition Examination Survey (NHANES) August 2021–August 2023 (available at ars.usda.gov/nea/bhnrc/fsrg), adults aged 71 years and older consumed an average of 3139 mg/day for men and 2472 mg/day for women.

The Mediterranean dietary approaches to stop hypertension (DASH) intervention for neurodegenerative delay (MIND) diet score assesses adherence to a dietary pattern designed to support brain health. This diet is modeled after the Mediterranean and DASH diets, with modifications informed by research on nutrition and dementia. ⁹ To calculate the MIND diet score, ²⁹ food items from the FFQ are classified into 10 brain‐healthy and five unhealthy food groups. Each group is scored from 0 to 1 (including increments of 0.5) based on the frequency of consumption of the associated food items. The total MIND diet score ranges from 0 to 15, with higher scores reflecting greater adherence to the brain‐healthy MIND diet.

2.4. Clinical evaluations

At enrollment (study baseline) and at each annual visit, participants underwent a comprehensive clinical evaluation. Mild cognitive impairment (MCI) and dementia were identified, as previously reported. ²³ , ³⁰ Self‐reported demographics (including date of birth, sex, and years of education) were collected at enrollment. In addition, the following data collected at the time of the first dietary assessment (analytical baseline) were included in this study: body mass index (BMI); mean systolic and diastolic blood pressures (calculated as the average of two seated and one standing measurements); hours of weekly physical activity (total hours spent in five categories: walking for exercise, gardening or yard work, calisthenics or general exercise, bicycling, and swimming or water exercises); hours of cognitive activity; history of hypertension; use of antihypertensive or diuretic medications; and history of diabetes, stroke, heart attack, and smoking. Apolipoprotein E (APOE) ε4 allele data were also available.

2.5. Brain autopsy and neuropathologic assessment

Brain autopsies were performed following a standardized procedure as described in detail elsewhere. ³⁰ Briefly, the brain was removed, and the cerebral hemispheres were cut into 1‐cm coronal slabs. Slabs from one hemisphere were frozen to −80°C. Slabs not designated for freezing were fixed in 4% paraformaldehyde for 3 to 21 days before the macroscopic examination for infarcts and the dissection of diagnostic blocks. The blocks were paraffin‐embedded, cut into 6‐µm sections, and mounted on microscope slides. Neuropathologic diagnoses were made by board‐certified neuropathologists blinded to clinical data.

For assessing AD neuropathologic change (ADNC), a modified Bielschowsky silver stain was used to visualize neuritic plaques and neurofibrillary tangles. The counts of these two neuropathologic markers within an area of 1 mm² from sections of midfrontal, superior temporal, inferior parietal, and entorhinal cortices were summarized as one continuous, standardized measure, global ADNC score. In addition, quantitative measures of amyloid burden and neurofibrillary tangle density were obtained using immunohistochemistry with antibodies to amyloid beta (Aβ) and phosphorylated tau. ³¹ Given the non‐normal distribution of these measures, the raw data were square‐root‐transformed for further analyses.

Levels of ADNC in each deceased and brain autopsied participant were also categorized semi‐quantitatively (low, intermediate, and high) using the 2012 National institute on Aging‐Alzheimer's Association guidelines for the neuropathologic assessment of AD. ³²

The presence of Stage 2 or 3 limbic‐predominant age‐related TDP‐43 encephalopathy neuropathologic change (LATE‐NC) was evaluated by identifying TDP‐43 cytoplasmic inclusions in neurons and glia using immunohistochemistry in eight brain regions, employing an anti‐phospho‐TDP‐43 (pS409/410) monoclonal antibody (TAR5P‐1D3, Ascenion; 1:100). ³³ , ³⁴

The presence of Lewy bodies was assessed by immunohistochemistry with an anti‐alpha‐synuclein monoclonal antibody (LB509, Zymed; 1:50), using the avidin‐biotin method with alkaline phosphatase as the color developer. The following seven regions were examined for Lewy bodies: substantia nigra, entorhinal cortex, anterior cingulate cortex, midfrontal cortex, superior or middle temporal cortex, inferior parietal cortex, and amygdala. ³⁵

Cerebrovascular disease and gross infarcts were identified on macroscopic examination and classified by number, location (cortical, subcortical), and volume (in square millimeters). Each gross infarct was then dissected and confirmed on microscopic examination, and classified by age (acute, subacute, and chronic). Microinfarcts were identified under the microscope in blocks of nine brain regions that were stained with H&E, and their location and age were also recorded. ³⁶ Only chronic gross infarcts and chronic microinfarcts were considered in this study.

Cerebral vessel pathology was also examined systematically. ³⁷ Atherosclerosis severity was graded using a semi‐quantitative scale from 0 (no atherosclerosis) to 6 (severe atherosclerosis) following visual inspection of vessels in the circle of Willis. Similarly, arteriolosclerosis severity was graded using a scale from 0 (no arteriolosclerosis) to 7 (complete small vessel occlusion) during histologic examination of the anterior basal ganglia. Last, CAA severity was graded in several neocortical brain regions, based on the degree of immunohistochemical labeling with anti‐Aβ. ³⁸ For all cerebral vessel pathologies, final scoring was summarized on an ordinal four‐point scale – none (0), mild (1), moderate (2), or severe (3).

2.6. Statistical approach

Initial analyses included a graphical examination of the variables and their correlations. Dietary sodium intake was categorized into quintiles, which allowed for a balanced comparison across different levels of sodium intake, helping to identify potential dose‐response relationships with cognitive decline while reducing the influence of extreme values. The associations of baseline calorie‐adjusted dietary sodium intake among participants without dementia, with longitudinally assessed cognitive function, were performed using a series of linear mixed‐effects models. All models were adjusted for age, sex, education, MIND diet score, and the interaction of each of these variables with time. Further analyses added terms for other potential confounders/mediators to the basic model, including BMI, systolic and diastolic blood pressures, hours of weekly physical activity, cognitive activity, total calorie intake, history of hypertension, antihypertensive/diuretic use, history of diabetes, history of stroke, history of heart attack, history of smoking, serum sodium level, and the presence of the APOE ε4 allele.

In a subset of deceased and autopsied participants who had neuropathologic data available, we examined the associations of mean dietary sodium intake (over all the years with FFQ data) with postmortem neuropathologic measures. Linear regression was used for quantitative ADNC variables: silver stain‐based global ADNC score and immunohistochemistry‐based amyloid burden and tau tangle density.

Logistic regression was used to examine the relation of sodium intake with the presence of neurodegenerative pathology (intermediate to high levels of ADNC, Stage 2 and 3 LATE‐NC, and Lewy bodies), cerebrovascular pathology, including the presence (vs. absence) of infarcts (any chronic infarcts, gross chronic infarcts, chronic microinfarcts, chronic cortical infarcts, and chronic subcortical infarcts), and the severity (two‐levels: none or mild vs. moderate or severe) of cerebral blood vessel pathology (atherosclerosis, arteriolosclerosis, and CAA).

All models were adjusted for age, sex, education, and MIND diet score, with or without additional covariates as mentioned earlier. Analyses were conducted using SAS/STAT software, version 9.4 of the SAS system for Linux (SAS Institute). A two‐tailed hypothesis was assumed.

3. RESULTS

3.1. Sample characteristics

The characteristics of the 1520 participants included in this study are presented in Table 1, organized into sodium intake quintile groups based on estimated calorie‐adjusted daily intake levels at the first dietary assessment available. The analytical sample was primarily White (93%) and female (75%), with an average age of 80.7 years (± 7.2) at the time of first dietary assessment (analytical baseline). The mean follow‐up duration was 7.6 years. Characteristics of the participants were similar among the five sodium intake groups (quintiles), except that the highest quintile group had a lower proportion of females and a higher prevalence of diabetes diagnoses compared to the other quintile groups (Table 1).

TABLE 1.

Participant characteristics by daily dietary sodium intake quintiles.

Quintile group	Q1	Q2	Q3	Q4	Q5
Sample size	304	304	304	304	304
Sodium intake, mg/day, median (interquartile range)	1764 (1609 to 1861)	2064 (2006 to 2116)	2264 (2209 to 2320)	2490 (2424 to 2560)	2800 (2709 to 2997)
Demographics
Age, years	80.0 ± 7.4	80.5 ± 7.3	80.0 ± 7.4	81.2 ± 6.8	81.6 ± 7.1
Female sex, % *	88.2	81.6	75.3	70.1	60.2
Education, years	14.9 ± 3.1	15.1 ± 2.8	15.6 ± 3.1	15.7 ± 3.3	15.7 ± 3.0
Dietary assessment
Total energy, kcal/d	1660 ± 757	1758 ± 556	1775 ± 474	1779 ± 448	1682 ± 428
MIND diet score	7.1 ± 1.7	7.6 ± 1.7	7.7 ± 1.7	7.6 ± 1.8	7.7 ± 1.8
Lifestyle factors
Physical activity, hours/week	3.3 ± 3.7	3.4 ± 3.3	3.7 ± 4.2	4.0 ± 4.0	3.5 ± 3.4
Cognitive activity	3.2 ± 0.7	3.2 ± 0.6	3.3 ± 0.6	3.2 ± 0.6	3.2 ± 0.6
Physical exam
Body mass index, kg/m2	27.5 ± 5.5	27.4 ± 5.1	27.4 ± 5.1	27.6 ± 4.7	27.8 ± 4.9
Systolic blood pressure, mmHg	134 ± 20	133 ± 19	132 ± 18	134 ± 17	132 ± 19
Diastolic blood pressure, mmHg	75 ± 12	75 ± 11	75 ± 12	75 ± 12	75 ± 11
Medical conditions
Hypertension, %	58.2	55.9	52.6	57.9	55.6
Antihypertensive use, %	70.7	70.7	64.5	69.4	74.3
Diabetes, % *	12.8	10.9	10.5	14.1	19.4
Stroke, %	10.6	11.6	10.0	8.0	9.7
Laboratory testing
APOE ε4, %	22.9	25.9	23.6	19.7	17.5

Note: Data are presented as mean ± SD unless otherwise specified. Q1‐Q5 refers to first‐fifth quintiles.

^* p < 0.05 in chi‐squared test. n = 1520.

3.2. Associations of dietary sodium intake with cognitive decline

We first examined the associations of dietary sodium intake with global cognitive function using linear mixed models adjusted for age, sex, education, the MIND diet score, and the interaction of each variable with time (Table 2). A total of 11,502 observations of cognition data from 1520 participants were included. A detailed frequency distribution of the number of cognitive assessments is provided in the Supporting Information (Table S1 and Figure S1). As noted in the “Methods” section, we used the middle quintile of sodium intake as the reference because it approximates the recommended limit of 2300 mg/day and accounts for the potential non‐linear relationship between sodium intake and cognition. We did not find any associations between higher sodium intake quintiles (Q4 and Q5) and either the level of global cognition or the rate of its decline. In contrast, the two lowest quintiles (Q1 and Q2) of sodium intake were both associated with lower baseline levels of global cognition. In addition, the lowest quintile (Q1) of sodium intake was also associated with a faster rate of global cognitive decline. These associations remained significant after additionally adjusting for total energy intake, physical activity, cognitive activity, BMI, systolic blood pressure, diastolic blood pressure, antihypertensive use, diabetes, and stroke (data not shown). However, Q1 was no longer associated with global cognitive decline after additionally adjusting for APOE ε4 (estimate = −0.015, SE = 0.010, p = 0.087). To examine whether the association differed by APOE ε4 status, we conducted stratified analyses. APOE ε4 genotype data were available for 1194 participants (78.6% of the full sample). Among participants without any APOE ε4 allele (n = 929; 7856 observations), sodium intake in the lowest quintile was not associated with global cognitive decline (estimate = −0.012, SE = 0.010, p = 0.236). Similarly, among participants with one or two APOE ε4 alleles (n = 265; 2185 observations), no significant association was observed (estimate = −0.028, SE = 0.023, p = 0.238).

TABLE 2.

Associations of dietary sodium intake with cognitive decline.

	Estimate (SE, p value)
Predictor	Global cognition	Episodic memory	Working memory	Semantic memory	Perceptual speed	Visuospatial ability
Sodium intake
Q1	−0.112 (0.039, 0.004)	−0.137 (0.051, 0.008)	−0.116 (0.056, 0.036)	−0.084 (0.045, 0.065)	−0.069 (0.055, 0.211)	−0.173 (0.053, 0.001)
Q2	−0.081 (0.039, 0.037)	−0.086 (0.051, 0.091)	−0.093 (0.055, 0.093)	−0.029 (0.045, 0.524)	−0.056 (0.054, 0.301)	−0.095 (0.053, 0.071)
Q3	Ref	Ref	Ref	Ref	Ref	Ref
Q4	−0.025 (0.039, 0.512)	−0.022 (0.051, 0.668)	−0.083 (0.055, 0.130)	0.004 (0.045, 0.931)	0.002 (0.055, 0.965)	−0.003 (0.053, 0.954)
Q5	−0.019 (0.039, 0.622)	0.013 (0.052, 0.799)	−0.106 (0.056, 0.057)	−0.003 (0.046, 0.944)	−0.011 (0.055, 0.840)	0.021 (0.054, 0.693)
Interactions of sodium intake with time
Q1*Time	−0.022 (0.009, 0.012)	−0.027 (0.011, 0.011)	−0.011 (0.008, 0.157)	−0.025 (0.010, 0.010)	−0.015 (0.008, 0.073)	−0.008 (0.008, 0.336)
Q2*Time	−0.014 (0.009, 0.114)	−0.018 (0.011, 0.092)	−0.006 (0.008, 0.452)	−0.026 (0.010, 0.007)	−0.018 (0.008, 0.032)	−0.008 (0.008, 0.319)
Q3*Time	Ref	Ref	Ref	Ref	Ref	Ref
Q4*Time	−0.006 (0.009, 0.486)	−0.009 (0.011, 0.399)	0.004 (0.008, 0.665)	−0.012 (0.010, 0.222)	−0.001 (0.008, 0.917)	0.001 (0.008, 0.909)
Q5*Time	−0.006 (0.009, 0.493)	−0.007 (0.011, 0.509)	−0.001 (0.009, 0.902)	0.000 (0.010, 0.996)	−0.005 (0.009, 0.561)	−0.008 (0.008, 0.339)

Note: Six linear mixed models were adjusted for age, sex, education, MIND diet score, and their interactions with time. Data in bold denote statistical significance. Q1–Q5 represent quintile groups of estimated daily dietary sodium intake from food sources, adjusted for total energy intake. A total of 11,502 observations of cognition data from 1520 participants were included.

Because the relation of sodium intake to cognition may differ by age, we conducted additional analyses stratified by the median age (81.5 years): Among the older age group (age ≥ 81.5 years; n = 760, 5020 observations), the lowest quintile of sodium intake was associated with faster decline in global cognition (estimate = −0.035, SE = 0.014, p = 0.013). Among the younger age group (age < 81.5 years; n = 760, 6482 observations), the lowest quintile of sodium intake was not associated with decline in global cognition (estimate = −0.010, SE = 0.011, p = 0.361). No other associations between sodium intake and global cognitive decline were observed in either age group.

Given that the highest sodium intake quintile group had a lower proportion of females compared to the other groups, we conducted additional stratified analyses by sex. Among females (n = 1141; 8787 observations), the lowest quintile of sodium intake was associated with a faster rate of global cognitive decline (estimate = −0.023, SE = 0.010, p = 0.026). This association was not statistically significant in males (n = 379; 2715 observations; estimate = −0.036, SE = 0.020, p = 0.078). Similar to the full sample analysis, we did not observe significant associations between high sodium intake and cognitive decline in either sex.

Given that sodium intake was estimated via food frequency questionnaire and may be influenced by early cognitive changes, we conducted a sensitivity analysis excluding participants who developed MCI or dementia within 2 years of baseline. This analysis included 8059 observations from 977 participants, representing 64% of the full sample (n = 1520). We found no association between the lowest quintile of sodium intake and decline in global cognition (estimate = −0.010, SE = 0.009, p = 0.268) or any other associations between sodium intake and global cognitive decline.

Next, we conducted secondary analyses using similarly adjusted linear mixed models to examine the associations of sodium intake and cognitive function within each of the five separate domains: episodic memory, working memory, semantic memory, perceptual speed, and visuospatial abilities (Table 2). We found that the lowest quintile of sodium (Q1) intake was associated with lower baseline level of episodic memory, working memory, and visuospatial ability. Also, Q1 was associated with a faster rate of cognitive decline in episodic memory and semantic memory. Further, Q2 was associated with a faster rate of decline in semantic memory and perceptual speed. There was no other association between sodium intake and the level or rate of decline in individual cognitive domains.

3.3. Associations of dietary sodium intake with neurodegenerative pathologies

In a subset of 717 deceased and autopsied participants, we examined the associations of mean dietary sodium intake with neurodegenerative diseases. We first examined associations of sodium intake with ADNC burden (assessed with silver stain‐based global ADNC score and immunohistochemistry‐based amyloid burden and tau tangle density) using separate linear regression models adjusted for age at death, sex, education, and MIND diet score (Table 3). Although none of the quintile groups of sodium intake were associated with global ADNC score or amyloid burden, the lowest quintile (Q1) of sodium intake was associated with a higher tau tangle density. This association remained significant after additionally adjusting for total energy intake, physical activity, cognitive activity, BMI, systolic blood pressure, diastolic blood pressure, antihypertensive use, diabetes, stroke, and APOE ε4. We did not find any association between high‐sodium‐intake quintiles (Q4 and Q5) and tau tangle density.

TABLE 3.

Associations of dietary sodium intake with quantitative ADNC burden.

	Estimate (SE, p‐value)
Predictor	Global ADNC Score	Amyloid Burden	Tau Tangle Density
Sodium intake
Q1	0.145 (0.087, 0.095)	0.100 (0.090, 0.265)	0.279 (0.109, 0.010)
Q2	0.054 (0.085, 0.529)	0.125 (0.088, 0.157)	0.180 (0.106, 0.091)
Q3	Ref	Ref	Ref
Q4	0.022 (0.086, 0.802)	0.035 (0.088, 0.696)	0.029 (0.107, 0.785)
Q5	−0.133 (0.091, 0.147)	−0.063 (0.094, 0.500)	−0.165 (0.113, 0.145)

Note: Three linear regression models were adjusted for age, sex, education, and MIND diet score. Q1–Q5 represent quintile groups of estimated daily dietary sodium intake from food sources, adjusted for total energy intake. n = 717.

Abbreviation: ADNC, Alzheimer's disease neuropathologic change.

Next, we examined the associations of sodium intake with the presence of ADNC, TDP‐43, and Lewy bodies using logistic regressions (Table 4). We did not find any significant associations.

TABLE 4.

Association of dietary sodium intake with presence of ADNC (intermediate to high levels), LATE‐NC (Stage 2 or 3), and Lewy body pathology.

	OR, Estimate (SE, p‐value)
Predictor	ADNC	LATE‐NC	Lewy bodies
Sodium intake
Q1	1.589, 0.463 (0.256, 0.070)	1.463, 0.380 (0.251, 0.130)	0.769, −0.263 (0.275, 0.340)
Q2	1.158, 0.147 (0.244, 0.548)	1.415, 0.347 (0.245, 0.157)	0.784, −0.244 (0.269, 0.363)
Q3	Ref	Ref	Ref
Q4	1.160, 0.149 (0.246, 0.546)	1.003, 0.003 (0.252, 0.991)	0.832, −0.185 (0.265, 0.486)
Q5	0.964, −0.037 (0.256, 0.886)	1.092, 0.088 (0.266, 0.742)	0.992, −0.008 (0.272, 0.976)

Note: Three logistic regression models were adjusted for age, sex, education, and MIND diet score. Q1–Q5 represent quintile groups of estimated daily dietary sodium intake from food sources, adjusted for total energy intake. n = 717.

Abbreviations: ADNC, Alzheimer's disease neuropathologic change; LATE‐NC, limbic‐predominant age‐related TDP‐43 encephalopathy neuropathologic change.

3.4. Association of dietary sodium intake with cerebrovascular pathologies

We studied the associations of sodium intake with the presence of chronic brain infarcts using separate logistic regression models (Table 5). We did not find any association of sodium intake with the presence of any chronic brain infarcts, including with the presence of gross infarcts, microinfarcts, cortical infarcts, or subcortical infarcts.

TABLE 5.

Association of dietary sodium intake with presence of brain infarcts.

	OR, estimate (SE, p value)
		Size of infarct		Location of infarct
Predictor	Any infarct	Gross infarcts	Micro‐infarcts	Cortical infarcts	Subcortical infarcts
Sodium intake
Q1	0.996, −0.004 (0.241, 0.988)	1.011, 0.011 (0.246, 0.963)	0.950, −0.051 (0.253, 0.839)	0.757, −0.279 (0.254, 0.272)	0.802, −0.220 (0.252, 0.382)
Q2	1.122, 0.115 (0.237, 0.626)	0.938, −0.064 (0.242,0.791)	1.202, 0.184 (0.243, 0.450)	0.986, −0.014 (0.243, 0.953)	0.887, −0.120 (0.244, 0.624)
Q3	Ref	Ref	Ref	Ref	Ref
Q4	1.101, 0.097 (0.239, 0.687)	0.989, −0.011 (0.244, 0.963)	1.139, 0.130 (0.247, 0.598)	0.983, −0.018 (0.245, 0.943)	0.911, −0.093 (0.249, 0.709)
Q5	1.051, 0.049 (0.251, 0.845)	1.088, 0.084 (0.256, 0.743)	1.164, 0.152 (0.259, 0.559)	1.120, 0.114 (0.255, 0.656)	0.991, −0.009 (0.260, 0.972)

Note: Three logistic regression models were adjusted for age, sex, education, and MIND diet score.Q1–Q5 represent quintile groups of estimated daily dietary sodium intake from food sources, adjusted for total energy intake. n = 717. Analyses evaluated for outcomes of chronic infarcts only.

Last, we examined the associations of sodium intake with the presence of moderate to severe cerebral vessel pathologies (Table 6). We found that both first and fourth quintiles of sodium intake (Q1 and Q4) were associated with higher odds of presence of moderate to severe arteriolosclerosis. These associations remained significant after additionally adjusting for other covariates (data not shown). We did not find an association of second and fifth quintile of sodium intake (Q2 and Q5) with moderate to severe arteriolosclerosis. Additionally, we did not find any association of sodium intake with the presence of moderate to severe atherosclerosis or CAA.

TABLE 6.

Association of dietary sodium intake with presence of moderate to severe cerebral vessel pathologies.

	OR, Estimate (SE, p‐value)
Predictor	Atherosclerosis	Arteriolosclerosis	CAA
Sodium intake
Q1	1.152, 0.141 (0.266, 0.595)	1.895, 0.639 (0.271, 0.018)	1.125, 0.117 (0.247, 0.635)
Q2	0.797, −0.226 (0.269, 0.399)	1.431, 0.358 (0.268, 0.182)	1.389, 0.329 (0.240, 0.171)
Q3	Ref	Ref	Ref
Q4	0.939, −0.063 (0.267, 0.813)	1.948, 0.667 (0.270, 0.014)	0.798, −0.226 (0.249, 0.365)
Q5	0.750, −0.288 (0.293, 0.325)	1.446, 0.369 (0.292, 0.207)	0.980, −0.021 (0.258, 0.936)

Note: Three logistic regression models were adjusted for age, sex, education, and MIND diet score. Q1–Q5 represent quintile groups of estimated daily dietary sodium intake from food sources, adjusted for total energy intake. Data in bold denotes statistical significance. n = 717.

Abbreviation: CAA, cerebral amyloid angiopathy.

4. DISCUSSION

In this study of dietary sodium intake and late‐life cognition in 1520 community‐dwelling older adults, we found that, compared to the middle quintile of sodium intake (∼2300 mg/day), the lowest quintile was associated with a faster decline in global cognition. In analyses with separate cognitive domains, the lowest quintile of sodium intake was associated with declines in episodic memory and semantic memory, while the second quintile was linked to declines in semantic memory and perceptual speed. In a subset of 717 deceased participants, the lowest quintile of sodium intake was associated with a higher burden of tau tangle density. But there was no association with ADNC, amyloid, TDP‐43, or Lewy bodies. For cerebrovascular outcomes, both the lowest and fourth highest quintiles of sodium intake were associated with increased odds of moderate to severe brain arteriolosclerosis. No associations were observed between sodium intake and other cerebrovascular pathologies including infarcts or other vessel pathologies. Taken together, these findings suggest that compared to the recommended sodium intake of 2300 mg/ day, lower dietary sodium intake of 1800 mg/day is associated with faster late‐life cognitive decline and higher tau tangle density, whereas both low and high sodium intake are associated with arteriolosclerosis.

Most previous studies on the association of dietary sodium intake with global cognition were cross‐sectional and yielded mixed results. ¹² Some found that higher sodium intake was associated with cognitive impairment, ¹⁵ , ¹⁶ while others found no significant relationship. ¹⁷ , ¹⁸ , ¹⁹ One cross‐sectional study of 925 older adults (mean age 74.5 years, 60% female) found that lower dietary sodium was associated with poor cognition, and higher dietary sodium was associated with better cognition. ²¹ Only a few studies examined the longitudinal association between dietary sodium intake and global cognitive decline in community‐dwelling older adults. One study of 1262 older adults (mean baseline age 74.2 years, 51% female) found sodium intake measured by FFQ was independently associated with greater global cognitive decline over a 3‐year period in sedentary but not physically active older adults. ³⁹ Another study estimated sodium intake using FFQ in 1194 participants (mean baseline age 74.2 years, 54% female) and found no association with cognitive decline over 6.9 years. ²⁰ A recent study estimating sodium intake through urinary sodium excretion in 2041 Chinese older adults (mean baseline age 68.5 years, 52% female) found that higher sodium intake was linked to faster cognitive decline over an average follow‐up of 11.4 years. ⁴⁰ However, given the relatively young mean baseline age, it may not fully reflect the association between sodium intake and cognitive decline in more advanced old age. These mixed findings across study designs may stem from the reliance of most studies on a single screening test of global cognition, such as the Mini‐Mental State Examination (MMSE), which may lack sensitivity to detect subtle changes. ⁴¹ In our study, using a comprehensive battery of neuropsychological tests and statistically robust linear mixed models, we found that low sodium intake (Q1 and Q2) was associated with a lower baseline level of global cognition, in line with findings from the cross‐sectional study of 925 older adults (mean age 74.5 years, 60% female). ²¹ Further, we showed that the lowest quintile of sodium intake (Q1) was associated with a faster subsequent decline in global cognition, an arguably more clinically relevant outcome, even after accounting for lifestyle and clinical factors. This may be driven by domain‐specific effects, with possible cumulative contributions from modest, non‐significant declines in others. Adjustment for APOE ε4 attenuated the association, suggesting genetic confounding or reduced power from a smaller sample. Stratified analyses by APOE ε4 showed no association in either group, likely reflecting limited power. The results of stratified analyses by age and by sex suggest that the observed association between low sodium intake and faster global cognitive decline may be driven by participants aged 81.5 years and older and by females. The sensitivity analysis excluding participants who developed MCI or dementia within 2 years showed no association, suggesting early cognitive changes may drive the main analysis findings, though limited power could also play a role. We did not find an association of high sodium intake with the level or decline of global cognition. This can be because the highest quintile of sodium intake was not particularly high. It is also possible that participants experiencing early stages of cognitive decline might have altered eating habits, leading to lower sodium intake. ⁴²

Most studies on sodium intake and cognitive domains were cross‐sectional with a small sample size and reported null findings. ¹² One exception, the previously mentioned cross‐sectional study of 925 older adults (mean age 74.5 years, 60% female), found that lower sodium intake was associated with poorer performance on the Trail Making Test Part B, a measure of executive function, but only among individuals aged 80 and older. ²¹ Consistent with these findings, in our sample of 1520 older adults with a mean age of 81 years, we observed that the lowest quintile of sodium intake was associated with lower levels of working memory, a component of executive function, and visuospatial abilities. Additionally, we found that lower sodium was also associated with a lower level of episodic memory and a faster decline in episodic memory, the earliest affected domain in both normal aging and AD. ⁴³ We further found that low sodium intake was associated with faster decline in semantic memory and perceptual speed. Associations remained after adjusting for APOE ε4. Although our findings suggest that low sodium intake may adversely affect multiple cognitive domains in older adults, results from other populations have differed. For example, in a randomized controlled trial of 160 sedentary men and women (mean age 65.4 years, 46% female) with cognitive impairment but no dementia and risks factors for cardiovascular diseases, researchers found that reduced sodium intake through the DASH diet was associated with better executive function after 6 months. ⁴⁴ These contrasting findings underscore the need for future research to explore how age, cardiovascular risk, and other lifestyle and clinical factors (e.g., depression, and others) may modify the relationship between sodium intake and cognitive trajectories in older adults.

Previous evidence linking sodium intake to neurodegenerative neuropathologies is limited and mostly based on non‐human data. A 2019 study reported that dietary salt induced hyperphosphorylation of tau followed by cognitive dysfunction in mice. ⁴⁵ Similarly, another study found that elevated salt intake in mice induced tau hyperphosphorylation and synaptic loss during aging. ⁴⁶ A recent study showed both increased hippocampal amyloid and phosphorylated tau with cognitive impairment in mice fed a high‐salt diet. ⁴⁷ However, the relevance of these findings to humans remains uncertain, as the high‐salt diets used greatly exceed human salt consumption. ⁴⁵ , ⁴⁸ The only relevant human study published found that a Mediterranean‐like low‐sodium diet improved cerebrospinal fluid biomarkers for AD and cerebral perfusion, whereas a Western‐like high‐sodium diet had the opposite effects in cognitively normal middle‐aged adults. ²² In our study, we examined the association of sodium intake with multiple neuropathologies and including across both neurodegenerative and cerebrovascular pathologies. We found that individuals in the lowest quintile of sodium intake had a higher density of hyperphosphorylated tau. However, no association was observed between sodium intake and amyloid burden, the presence of intermediate to high levels of ADNC, and other neurodegenerative neuropathologies (e.g., LATE‐NC, Lewy bodies). These findings suggest that while dietary sodium intake in older adults might not be directly linked to classical ADNC, low sodium intake might contribute to hyperphosphorylated tau accumulation and cognitive impairment through alternative mechanisms, such as oxidative stress. ⁴⁹

We also examined the association of sodium intake with multiple cerebrovascular neuropathologies. We found that both high and low sodium intake were associated with increased odds of moderate to severe arteriolosclerosis. This suggests a potential U‐shaped relationship. Both extremes may harm vascular brain health. ⁵⁰ High sodium intake may exacerbate hypertension and endothelial dysfunction, ⁵¹ while chronic low sodium intake could activate compensatory mechanisms such as renin‐angiotensin system and sympathetic nervous system leading to vascular remodeling. ⁵² , ⁵³ We found no association between the highest quintile (Q5) of sodium intake and increased odds of moderate or severe arteriolosclerosis. While this may reflect the presence of protective factors such as genetic predispositions or medication use, it is also possible that selection bias contributed to this finding, as individuals with poorer health may have been less likely to participate.

This study has several limitations that should be noted when interpreting the findings. First, we considered only sodium from foods reported on the questionnaire, excluding salt added during cooking or at the table. Second, we estimated sodium intake using a self‐reported FFQ, which may be subject to recall bias and inaccuracies related to cognitive impairment, although the participants were without baseline dementia and the FFQ validated. ²⁷ Third, the lowest and the highest sodium intake quintiles were not substantially different from the middle quintile, which may have limited statistical power to detect associations. Fourth, selection bias is possible given the advanced baseline age, as individuals in poorer health or with higher sodium intake may have died earlier and contributed less data, leading to underestimation of associations. Finally, because the participants were predominantly White and highly educated older adults, our findings may not be generalizable to more diverse or younger populations. Despite these limitations, this study has important strengths. First, we used an extensive battery of neuropsychological tests to assess global cognition and five specific cognitive domains, providing a nuanced and robust evaluation of cognitive function. Second, the annual follow‐up assessments conducted over an average of 7.6 years generated high‐density longitudinal data, allowing us to rigorously model the trajectory of late‐life cognitive decline. Third, the systematic and comprehensive neuropathologic evaluation enhanced our ability to examine the association between sodium intake and common dementia‐related pathologies.

In conclusion, our findings reveal a complex relationship between sodium intake and brain health in late life, suggesting that maintaining a balanced sodium intake may be important for preserving cognitive function and limiting neuropathologic burden.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare. Author disclosures are available in the supporting information.

CONSENT STATEMENT

The Rush MAP was approved by the Institutional Review Board of Rush University Medical Center. All participants signed an informed consent before enrollment, and a subset of participants signed an anatomical gift act to donate their brain tissues for research at the time of death.

Supporting information

Supporting Information

Supporting Information

Tong H, Capuano AW, Agarwal P, et al. Association of dietary sodium intake with late‐life cognitive decline and postmortem neuropathology in community‐dwelling older adults. Alzheimer's Dement. 2025;21:e70706. 10.1002/alz.70706

DATA AVAILABILITY STATEMENT

All data supporting the conclusions of this article can be requested for research purposes via Rush Alzheimer's Disease Center Research Resource Sharing Hub at radc.rush.edu.