Association between the dietary index for gut microbiota and Alzheimer's disease: A cross‐sectional study from the National Health and Nutrition Examination Survey (2004 to 2018)

Abstract

INTRODUCTION

Emerging evidence implicates gut microbiota (GM), shaped by diet, in Alzheimer's disease (AD) pathogenesis. However, the association between the dietary index for GM (DI‐GM) and AD remains unclear.

METHODS

This cross‐sectional study analyzed data from 28,830 adults aged ≥20 years in the 2004–2018 National Health and Nutrition Examination Survey (NHANES). The DI‐GM score, derived from dietary recalls, comprised beneficial to GM score (BGMS) and unfavorable to GM score (UGMS) components. AD was identified via self‐report, medications, or death certificates. Multivariable weighted logistic regression, restricted cubic spline, and subgroup analyses were performed.

RESULTS

Contrary to expectations, higher DI‐GM score and UGMS were associated with increased AD prevalence (DI‐GM: odds ratio [OR] = 1.24, 95% CI: 1.02 to 1.52, p = 0.033; UGMS: OR = 1.36, 95% CI: 1.10 to 1.69, p = 0.005).

DISCUSSION

The DI‐GM was positively associated with AD prevalence, suggesting that imbalanced plant‐based diets low in protein or key nutrients may elevate AD risk despite presumed microbiota benefits.

Highlights

• Higher DI‐GM and UGMS were significantly associated with greater AD prevalence in US adults.

• Restricted cubic spline analyses showed linear and non‐linear associations of DI‐GM and UGMS with AD, respectively.

• Results challenge prior assumptions that higher DI‐GM scores are uniformly linked to health benefits.

• Imbalanced plant‐based diets low in protein or key nutrients may adversely affect cognitive aging despite presumed microbiota benefits.

1. INTRODUCTION

Alzheimer's disease (AD), a neurodegenerative disorder characterized by progressive neuronal damage and irreversible cognitive decline, accounts for 60% to 80% of the 55 million dementia cases worldwide and is projected to triple in prevalence by 2050 as a result of aging populations, profoundly impacting public health through increased mortality, morbidity, healthcare costs, and burden on family caregivers, thereby contributing substantially to societal burdens. ¹ , ² AD pathology, marked by amyloid beta (Aβ) plaques and tau protein tangles, develops decades before symptoms appear, and the absence of a cure or reliable treatments underscores the urgency of early prevention targeting modifiable risk factors. ³ Growing evidence identifies the gut microbiome as a key factor in AD onset and progression, particularly through the microbiota–gut–brain axis, which regulates neuroinflammation and amyloid metabolism. ⁴ , ⁵ , ⁶ Schneider et al. ⁷ proposed the diet–microbiota–gut–brain axis as a novel frontier for brain health diagnostics and therapeutics across the lifespan. Additionally, diet shapes gut microbiota (GM) composition and function, emphasizing the therapeutic potential of diet–microbiome–disease interactions. ⁸ As a modifiable lifestyle factor, diet may play a key role in brain health through various biological mechanisms, with evidence supporting the potential of personalized dietary interventions to prevent cognitive decline and dementia. ⁹

The dietary index for GM (DI‐GM), developed from a comprehensive literature review and validated through its association with biomarkers of GM diversity, is an innovative dietary quality index that reflects GM diversity, with higher scores indicating a healthier GM. ¹⁰ Therefore, the DI‐GM provides a valuable criterion for studying dietary influences on GM and health, enabling more precise dietary recommendations. Moreover, Zhang et al. ¹¹ found that the DI‐GM was negatively correlated with depression prevalence, total Patient Health Questionnaire‐9 score, and specific depression symptoms. Given the rising interest in leveraging diet to modulate GM and reduce AD risk, we first examined the association between DI‐GM and AD, utilizing adult data from the National Health and Nutrition Examination Survey (NHANES).

2. METHODS

2.1. Study population

NHANES is a continuous cross‐sectional study conducted by the National Center for Health Statistics (NCHS) using a complex, stratified, multistage probability sampling method to assess the health and nutritional status of the non‐institutionalized US national population. NHANES data are publicly accessible at https://wwwn.cdc.gov/nchs/nhanes. NHANES protocols were approved by the Institutional Review Board of the NCHS, and written informed consent was obtained from all participants. Our study involved a total of 48,735 participants aged ≥20 years from 2004 to 2018. The exclusion criteria included missing data on AD information (n = 8980) and missing data on covariates (n = 10,925). A total of 28,830 participants were included in the final analysis, as shown in Figure 1.

FIGURE 1.

Study flow chart. AD, Alzheimer's disease; BMI, body mass index; DM, diabetes mellitus; HEI‐2015, Health Eating Index‐2015; PIR, poverty income ratio.

2.2. Ascertainment of AD

AD cases were identified based on self‐reported physician diagnosis, use of AD‐related medications (Rivastigmine, Galantamine, Donepezil, or Memantine), or a recorded AD diagnosis on death certificates in the National Death Index. ¹²

RESEARCH IN CONTEXT

1. Systematic review: Emerging evidence underscores the diet–microbiota–gut–brain axis as vital to brain health. The DI‐GM, constructed from beneficial and unfavorable components, quantifies diet quality relative to GM, with higher scores indicating a healthier microbial profile. Despite known health associations, the relevance of DI‐GM to AD remains unclear.

2. Interpretation: Using NHANES data (2004 to 2018), this study is the first to examine DI‐GM in relation to AD prevalence in US adults. Unexpectedly, higher DI‐GM and UGMSs were associated with greater AD prevalence. These findings challenge presumed benefits of higher DI‐GM scores, suggesting that imbalanced plant‐based diets lacking sufficient protein or essential nutrients may elevate AD risk despite microbial advantages.

3. Future directions: The counterintuitive findings warrant reassessing microbiota‐targeted dietary guidance for AD. Longitudinal and mechanistic studies are needed to clarify causal pathways and inform personalized nutritional strategies.

2.3. Assessment of DI‐GM

The DI‐GM consists of 14 foods or nutrients, with beneficial components including fermented dairy, chickpeas, soybeans, whole grains, fiber, cranberries, avocados, broccoli, coffee, and green tea (not recorded in NHANES for specific tea types) and unfavorable components including red meat, processed meat, refined grains, and a high‐fat diet (≥40% energy from fat). ¹⁰ The DI‐GM score was calculated utilizing 24‐h dietary recall data derived from NHANES 2004 to 2018. For beneficial components, a score of 1 was assigned when consumption ≥ sex‐specific median, otherwise a score of 0, with the scores summed to yield beneficial to GM score (BGMS, range: 0 to 9); for unfavorable components, a score of 0 was assigned when consumption ≥ sex‐specific median or 40% (for high‐fat diet), otherwise a score of 1, resulting in unfavorable to GM score (UGMS, range: from 0 to 4). The DI‐GM score (range: 0 to 13) was obtained by summing scores for each component and grouped according to 0 to 3, 4, 5, and ≥6. ¹¹

2.4. Covariates

Several potential confounding variables were included based on published research findings and clinical relevance, including sociodemographic characteristics (age, sex, race/ethnicity, education level, marital status, and family poverty‐to‐income ratio [PIR]), lifestyle factors (body mass index [BMI], physical activity, smoking status, alcohol consumption, and the Healthy Eating Index‐2015 [HEI‐2015]), and comorbidities (hypertension, diabetes mellitus [DM], heart disease, and stroke). ¹² , ¹³ , ¹⁴ , ¹⁵ Definitions and classification criteria are detailed in Table S1.

2.5. Statistical analysis

All analyses adhered to NHANES analytic guidelines, incorporating Mobile Examination Center (MEC) sampling weights to account for the complex, multistage probability sampling design and to ensure nationally representative estimates. Continuous variables were presented as weighted means (standard errors [SEs]) and categorical variables as weighted counts and percentages (%). Group comparisons were conducted using the Wilcoxon rank‐sum test for continuous variables and the Rao–Scott chi‐squared test for categorical variables, accounting for the survey design.

The association between DI‐GM and AD was assessed using multivariable weighted logistic regression models, with odds ratios (ORs) and 95% confidence intervals (CIs) reported. Model 1 was unadjusted. Model 2 was adjusted for sociodemographic variables: age, sex, race/ethnicity, education level, marital status, and PIR. Model 3 was further adjusted for lifestyle factors: BMI, physical activity, smoking status, alcohol consumption, and HEI‐2015. Model 4 was additionally adjusted for comorbidities: hypertension, DM, heart disease, and stroke.

Restricted cubic spline (RCS) analyses with three knots were employed to explore potential non‐linear dose–response relationships between DI‐GM and AD, dividing the range of DI‐GM at the 10th, 50th, and 90th quantiles to fit curves. Threshold analysis was conducted to identify cut‐off values, followed by segmented logistic regression to assess the association at different levels of DI‐GM. To examine the robustness of the association, subgroup analyses were performed based on age, sex, race/ethnicity, marital status, education level, PIR, BMI, smoking status, alcohol consumption, hypertension, DM, heart disease, and stroke.

All statistical analyses were conducted using R software (version 4.3.3). A two‐sided p value < 0.05 was considered statistically significant.

3. RESULTS

3.1. Participant characteristics

Table 1 presents the characteristics of a sample representing 159.36 million US adults with a mean age of 47.71 years (SE: 0.23), of whom 0.52 million were identified as AD. Participants with AD were typically older, predominantly non‐Hispanic White, and married, had notably lower educational attainment and income levels and BMI, engaged less in physical activity, were less likely to smoke or consume alcohol, and displayed a higher prevalence of hypertension, DM, heart disease, and stroke. Additionally, they had elevated DI‐GM scores, particularly in the DI‐GM ≥6 group, with significantly higher UGMS.

TABLE 1.

Characteristics of study participants.

Characteristics	Total	AD	Without AD	p value
Weighted population, n (in millions)	159.36	0.52	158.83
Age, mean (SE), year	47.71 (0.23)	74.72 (1.04)	47.62 (0.23)	<0.001
Age, n (in millions), %				<0.001
< 60	117.80 (73.92)	0.04 (6.99)	117.76 (74.14)
60 to 69	22.34 (14.02)	0.06 (11.59)	22.28 (14.03)
70 to 79	13.10 (8.22)	0.20 (37.54)	12.90 (8.12)
≥80	6.12 (3.84)	0.23 (43.87)	5.89 (3.71)
Sex, n (in millions), %				0.089
Female	82.31 (51.65)	0.32 (60.44)	81.99 (51.62)
Male	77.05 (48.35)	0.21 (39.56)	76.84 (48.38)
Race/ethnicity, n (in millions), %				<0.001
Non‐Hispanic Black	16.55 (10.39)	0.04 (6.89)	16.52 (10.40)
Non‐Hispanic White	113.54 (71.25)	0.44 (84.29)	113.10 (71.21)
Mexican American	11.82 (7.41)	0.02 (3.04)	11.80 (7.43)
Other Hispanic	7.47 (4.69)	0.02 (4.52)	7.45 (4.69)
Other	9.97 (6.26)	0.01 (1.26)	9.97 (6.27)
Education level, n (in millions), %				<0.001
Less than high school	22.79 (14.30)	0.15 (28.87)	22.64 (14.25)
High school or equivalent	37.37 (23.45)	0.12 (23.44)	37.25 (23.45)
Above high school	99.20 (62.25)	0.25 (47.69)	98.95 (62.30)
Marital status, n (in millions), %				<0.001
Married	91.33 (57.31)	0.33 (64.19)	90.99 (57.29)
Living with partner	12.04 (7.55)	0.01 (1.19)	12.03 (7.58)
Never married	26.95 (16.91)	0.01 (1.20)	26.94 (16.96)
Other	29.04 (18.22)	0.17 (33.42)	28.87 (18.17)
PIR, mean (SE)	3.08 (0.03)	2.61 (0.15)	3.08 (0.03)	0.003
PIR, n (in millions), %				<0.001
≤1.30	31.17 (19.56)	0.10 (18.37)	31.08 (19.57)
1.31 to 3.50	57.31 (35.96)	0.29 (56.39)	57.01 (35.90)
> 3.50	70.87 (44.47)	0.13 (25.24)	70.74 (44.54)
BMI, mean (SE), kg/m2	29.04 (0.08)	27.51 (0.52)	29.05 (0.08)	0.005
Physical activity, mean (SE), minutes/week	748.03 (15.08)	210.80 (87.83)	749.80 (15.14)	<0.001
Smoking status, n (in millions), %				<0.001
Never	86.88 (54.52)	0.28 (53.09)	86.61 (54.53)
Former	40.80 (25.60)	0.22 (42.53)	40.57 (25.54)
now	31.68 (19.88)	0.02 (4.39)	31.65 (19.93)
Alcohol consumption, n (in millions), %				<0.001
Never	16.84 (10.57)	0.12 (22.35)	16.73 (10.53)
Former	22.60 (14.18)	0.17 (32.36)	22.43 (14.12)
Mild	59.42 (37.29)	0.17 (31.81)	59.25 (37.31)
Moderate	28.03 (17.59)	0.07 (12.53)	27.97 (17.61)
Heavy	32.46 (20.37)	0.00 (0.94)	32.45 (20.43)
HEI‐2015, mean (SE)	53.55 (0.19)	53.09 (1.37)	53.55 (0.19)	0.741
Hypertension, n (in millions), %	60.71 (38.10)	0.37 (70.92)	60.34 (37.99)	<0.001
DM, n (in millions), %	21.51 (13.50)	0.18 (33.69)	21.34 (13.43)	<0.001
Heart disease, n (in millions), %	5.69 (3.57)	0.07 (12.60)	5.63 (3.54)	<0.001
Stroke, n (in millions), %	4.61 (2.89)	0.13 (24.22)	4.48 (2.82)	<0.001
DI‐GM score, mean (SE)	4.61 (0.02)	5.02 (0.18)	4.61 (0.02)	0.028
DI‐GM, n (in millions), %				0.042
0 to 3	38.05 (23.87)	0.08 (14.89)	37.97 (23.90)
4	39.45 (24.76)	0.13 (25.11)	39.32 (24.76)
5	37.84 (23.75)	0.11 (20.23)	37.73 (23.76)
≥6	44.02 (27.62)	0.21 (39.77)	43.81 (27.58)
BGMS, mean (SE)	2.34 (0.02)	2.37 (0.15)	2.34 (0.02)	0.826
UGMS, mean (SE)	2.27 (0.01)	2.65 (0.09)	2.27 (0.01)	<0.001

Continuous variables are presented as weighted means (SEs) and categorical variables as weighted counts (percentages [%]). The DI‐GM score, comprising BGMS and UGMS, was categorized as 0–3, 4, 5, and ≥6. Smoking status was categorized as never (< 100 lifetime cigarettes), former (≥100 but quit), and now (≥100 and currently smoke). Alcohol consumption was categorized by average daily intake over the past 12 months as never (< 12 lifetime drinks), former (≥12 lifetime drinks but none in the past year), mild (≤1 drink/day for females, ≤2 for males), moderate (≤2 for females, ≤3 for males), and heavy (≥3 for females, ≥4 for males). The HEI‐2015 consisted of 13 components (range: 0–100).

Abbreviations: AD, Alzheimer's disease; BGMS, beneficial to gut microbiota score; BMI, body mass index; DI‐GM, dietary index for gut microbiota; DM, diabetes mellitus; HEI‐2015, Health Eating Index‐2015; PIR, poverty income ratio; SE, standard error; UGMS, unfavorable to gut microbiota score.

3.2. Association between DI‐GM and AD

As shown in Table 2, higher DI‐GM scores were associated with an increased AD prevalence in the crude model, with an OR of 1.19 (95% CI: 1.02 to 1.38). After adjusting for sociodemographic factors, the association was attenuated and no longer significant (OR: 1.05, 95% CI: 0.89 to 1.24). However, after further adjustments for lifestyle factors and comorbidities, the association became significant, with ORs of 1.23 (95% CI: 1.01 to 1.50) and 1.24 (95% CI: 1.02 to 1.52), respectively. Thus, for each one‐point increase in the DI‐GM score, AD prevalence increased by 24%. DI‐GM scores were also categorized into four groups (0 to 3 as the reference group, 4, 5, and ≥ 6). In the crude model, the DI‐GM ≥ 6 group had a significantly higher AD prevalence (OR: 2.32, 95% CI: 1.14 to 4.69). This association was attenuated after adjusting for sociodemographic factors (OR: 1.28, 95% CI: 0.62 to 2.62) but became significant again in fully adjusted models that incorporated lifestyle and comorbidities (ORs: 2.25, 95% CI: 1.04 to 4.85, and 2.33, 95% CI: 1.09 to 5.01, respectively). Further analysis examined BGMS and UGMS. BGMS showed no significant association with AD prevalence across all models. In contrast, UGMS was consistently associated with a higher AD prevalence. In the crude model, the OR was 1.46 (95% CI: 1.21 to 1.77), and this association remained significant after adjusting for all covariates (OR: 1.36, 95% CI: 1.10 to 1.69). This implies that for each one‐point increase in UGMS, AD prevalence increases by 36%.

TABLE 2.

Association between DI‐GM and AD among NHANES 2004–2018 participants.

	Model 1		Model 2		Model 3		Model 4
Characteristics	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value
DI‐GM score	1.19 (1.02, 1.38)	0.025	1.05 (0.89, 1.24)	0.580	1.23 (1.01, 1.50)	0.041	1.24 (1.02, 1.52)	0.033
DI‐GM group
0 to 3	Reference		Reference		Reference		Reference
4	1.63 (0.80, 3.33)	0.180	1.24 (0.60, 2.55)	0.556	1.42 (0.69, 2.91)	0.335	1.45 (0.71, 2.98)	0.304
5	1.37 (0.67, 2.77)	0.382	0.88 (0.42, 1.84)	0.734	1.17 (0.56, 2.44)	0.677	1.20 (0.57, 2.50)	0.632
≥6	2.32 (1.14, 4.69)	0.020	1.28 (0.62, 2.62)	0.498	2.25 (1.04, 4.85)	0.039	2.33 (1.09, 5.01)	0.030
BGMS	1.02 (0.84, 1.25)	0.826	0.94 (0.75, 1.17)	0.555	1.06 (0.84, 1.34)	0.605	1.07 (0.85, 1.36)	0.589
UGMS	1.46 (1.21, 1.77)	<0.001	1.22 (0.99, 1.50)	0.057	1.36 (1.10, 1.69)	0.005	1.36 (1.10, 1.69)	0.005

Model 1 was unadjusted. Model 2 was adjusted for age, sex, race/ethnicity, education level, marital status, and PIR. Model 3 was further adjusted for BMI, physical activity, smoking status, alcohol consumption, and HEI‐2015. Model 4 was additionally adjusted for hypertension, DM, heart disease, and stroke. The DI‐GM score, comprising BGMS and UGMS, was categorized as 0 to 3, 4, 5, and ≥6.

Abbreviations: AD, Alzheimer's disease; BGMS, beneficial to gut microbiota score; BMI, body mass index; CI, confidence interval; DI‐GM, dietary index for gut microbiota; DM, diabetes mellitus; HEI‐2015, Health Eating Index‐2015; NHANES, National Health and Nutrition Examination Survey; OR, odds ratio; PIR, poverty income ratio; UGMS, unfavorable to gut microbiota score.

Figure 2 illustrates the relationship between the DI‐GM score and AD prevalence. Panel A depicts a linear relationship between DI‐GM score and AD prevalence (p for non‐linearity = 0.081). Panel B shows minimal association between BGMS and AD prevalence, as the ORs remain close to 1 across all score ranges (p for non‐linearity = 0.595). Panel C reveals a strong positive association between UGMS and AD prevalence, with a steep rise in ORs beyond a score of 3, indicating significant non‐linearity (p for non‐linearity = 0.019).

FIGURE 2.

Association between DI‐GM and AD prevalence among NHANES 2004‐2018 participants by RCS. The model was adjusted for age, sex, race/ethnicity, education level, marital status, PIR, BMI, physical activity, smoking status, alcohol consumption, HEI‐2015, hypertension, DM, heart disease, and stroke. The DI‐GM score comprises BGMS and UGMS. Red dots represent the reference values. (A) Linear association between DI‐GM score and AD prevalence (p for non‐linearity = 0.081). (B) Non‐significant linear trend between BGMS and AD prevalence (p for non‐linearity = 0.595). (C) Panel illustrates a strong positive association between UGMS and AD prevalence, with ORs rising sharply beyond a score of 3, indicating significant non‐linearity (p for non‐linearity = 0.019). AD, Alzheimer's disease; BGMS, beneficial to gut microbiota score; BMI, body mass index; CI, confidence interval; DI‐GM, dietary index for gut microbiota; DM, diabetes mellitus; HEI‐2015, Health Eating Index‐2015; NHANES, National Health and Nutrition Examination Survey; OR, odds ratio; PIR, poverty‐to‐income ratio; RCS, restricted cubic spline; UGMS, unfavorable to gut microbiota score.

Figure 3 shows that higher DI‐GM scores were significantly associated with increased AD prevalence in adults aged <60 and 70 to 79 years, non‐Hispanic Whites, individuals with at least high school education, married participants, those with BMI < 23.0 kg/m², mild or moderate alcohol consumers, and those without DM or stroke, with a marginal association observed in those without heart disease.

FIGURE 3.

Subgroup analyses of association between DI‐GM and AD prevalence among NHANES 2004–2018 participants. This forest plot presents ORs and 95% CIs for the association between DI‐GM and AD prevalence across stratified subgroups. Significant associations were observed in adults aged <60 and 70 to 79 years, non‐Hispanic Whites, individuals with at least high school education, married participants, those with BMI < 23.0 kg/m², mild or moderate alcohol consumers, and those without DM or stroke with borderline significance in those without heart disease. AD, Alzheimer's disease; BMI, body mass index; CI, confidence interval; DI‐GM, dietary index for gut microbiota; DM, diabetes mellitus; NHANES, National Health and Nutrition Examination Survey; OR, odds ratio; PIR, poverty‐to‐income ratio.

4. DISCUSSION

This study is the first to demonstrate that higher DI‐GM score, particularly in the DI‐GM ≥ 6 group, and UGMS were significantly associated with increased AD prevalence, while BGMS showed no significant association. RCS analysis revealed a non‐linear relationship with AD prevalence, notably increasing for UGMS exceeding 3. Importantly, each one‐point increase in UGMS corresponded to a 36% increase in AD prevalence. These findings underscore the critical role of dietary factors in AD prevalence, suggesting that targeted dietary interventions may help modify disease progression.

Numerous studies have established that the microbiota–gut–brain axis influences AD by regulating glial functions and targeting barriers like the intestinal and blood–brain barriers, meninges, and peripheral immune system through immunological, neurological, and metabolic pathways involving metabolites, neurotransmitters, and gut hormones. ⁴ , ⁵ , ⁶ Diets shape GM composition and function, profoundly influencing brain health, thereby substantiating the diet–microbiota–gut–brain axis proposed by Schneider et al. ⁷ The DI‐GM quantifies diets promoting a healthy GM and is positively correlated with indirect biomarkers of microbial diversity, such as enterodiol and enterolactone. ¹⁰ However, it is noteworthy that our findings deviated from both expectations and those reported by Zhang et al., ¹¹ suggesting that higher DI‐GM scores, although generally associated with healthier GM, may not consistently confer benefits. Specifically, it is important to clarify that higher UGMS reflects minimal intake of refined grains, red and processed meat, and high‐fat diets, indicating a dietary pattern heavily weighted toward plant‐based consumption. This nutritional imbalance may compromise the intake of high‐quality protein, essential fatty acids, and key micronutrients, which are vital to cognitive and physical function in aging and may contribute to the elevated AD prevalence observed. Moreover, this interpretation aligns with findings by Ngabirano et al., ¹⁶ who reported higher dementia risk among individuals with very low meat intake, and by Jigeer et al., ¹⁷ who observed that low‐quality vegetarian diets were correlated with poorer aging outcomes compared to omnivorous diets. Wang et al. ¹⁸ noted that the cognitive effects of plant‐based diets remain inconclusive and largely depend on nutrient adequacy and overall diet quality, despite their potential benefits to GM. Taken together, these findings underscore the need to balance microbial benefits and nutritional adequacy when evaluating the cognitive implications of plant‐based diets.

Mounting evidence links dietary proteins and their constituent amino acids to dementia prevention by preserving neuronal structure, modulating inflammation, and supporting muscle maintenance. ¹⁹ Tynkkynen et al. ²⁰ found that elevated circulating levels of branched‐chain amino acids, such as isoleucine, leucine, and valine, were associated with a reduced risk of dementia and AD. Experimental findings in tauopathy models revealed that protein malnutrition downregulated synaptic components and modestly accelerated brain atrophy, while supplementation with essential amino acids mitigated neuroinflammation and maintained brain homeostasis without altering tau burden. ²¹ In parallel, a cross‐sectional study in cognitively normal older adults showed that lower dietary protein intake correlated with greater cerebral Aβ burden, implicating a protective role for dietary protein in early AD pathology. ²² Moreover, a high‐tryptophan diet attenuated cognitive impairment and Aβ deposition by modulating GM and suppressing neuroinflammation via aryl hydrocarbon receptor activation and nuclear factor kappa B (NF‐κB) inhibition. ²³ Complementing these findings, Ming et al. ²⁴ reported that low‐protein diets trigger intestinal inflammation through necroptosis, mediated by nuclear fragile X mental retardation‐interacting protein 1‐DNA damage response signaling, further implicating inadequate protein intake in gut–brain axis disruption relevant to neurodegeneration. Consistent with this, Keum et al. ²⁵ found that higher protein intake was related to better episodic memory in older adults without dementia, particularly among apolipoprotein E ε4 (APOE ε4) carriers, a major genetic risk factor for AD. In addition, adequate protein intake helps prevent frailty, ²⁶ recognized as an upstream determinant of dementia pathogenesis. ²⁷ Overall, these findings emphasize that both the quantity and source of dietary protein warrant consideration in strategies for cognitive aging.

As total protein intake gains recognition for its cognitive benefits, ²⁸ the implications of specific sources, particularly red and processed meat, have drawn increasing scientific scrutiny. High intakes of plant‐based proteins may benefit older adults but carry risks of malnutrition and fiber intolerance. ²⁹ By contrast, red meat is the richest source of bioavailable heme‐iron critical for muscle and cardiovascular health and supplies all essential amino acids along with neuroprotective micronutrients such as vitamin B12. ¹⁸ , ³⁰ In older adults, moderate consumption of red meat may better preserve muscle mass than other dietary protein sources and confer additional benefits, though overconsumption poses metabolic disorders. ³¹ A meta‐analysis by Guasch‐Ferré et al. ³² demonstrated that cardiometabolic effects of red meat varied by comparator, showing less favorable lipid profiles when replaced with plant proteins, but comparable or improved outcomes relative to fish or refined carbohydrates. Complementing this, a UK Biobank study linked both genetic and lifestyle risks for coronary artery disease to vascular dementia (VaD), whereas AD was moderately associated with genetic risk alone. ³³ These findings imply that red meat‐related cardiometabolic pathways may be more relevant to VaD than AD, consistent with our observation of stronger DI‐GM–AD associations in those without cardiometabolic disease. Similarly, Xu et al. ³⁴ reported that meat intake was associated with improved memory and a fourfold lower dementia risk in seniors (≥68 years). In the UK Biobank, Zhang et al. ³⁵ found that each 25 g/day increase in processed meat intake elevated risks of all‐cause dementia and AD, whereas a 50 g/day increment in unprocessed red meat showed a modest protective effect. Additionally, cognitive benefits were observed with higher dietary frequencies of thin pork. ³⁰ Therefore, this evolving view suggests that insufficient intake of high‐quality animal protein may contribute to cognitive vulnerability in aging, supporting moderate inclusion of healthy animal‐based foods in age‐specific dietary strategies. ³⁶

Dietary fat, as a modifiable and increasingly recognized determinant of cognitive function, may partly explain the associations observed in this study. Fatty acids, classified by saturation (saturated [SFA], monounsaturated [MUFA], polyunsaturated [PUFA]), and configuration (cis [CFA], trans isomers [TFA]), exert distinct biological effects. A 2025 umbrella review synthesizing 167 meta‐analyses and Global Burden of Disease data challenged the conventional belief that lowering total fat intake enhances health. ³⁷ It found no links between fat reduction and type 2 DM, stroke, cardiovascular mortality, or cancer mortality. However, it identified increased risks of AD and cognitive decline with higher SFA and TFA intake, while MUFA and PUFA, particularly omega‐3 (n‐3) and omega‐6 (n‐6), showed protective effects. Notably, the review indicated that each 0.1 g/day increment in docosahexaenoic acid (DHA) intake corresponded to a 14% lower risk of dementia and a 37% lower risk of AD. Interestingly, lard, compared with corn and canola oils, was found to increase GM α‐diversity in mice, ³⁸ suggesting that not all dietary fats exert uniformly adverse effects, with respect to the microbiota–gut–brain axis. These neuroprotective effects may stem from the multifaceted actions of n‐3 PUFAs, which serve as essential structural components of neuronal membranes. ¹⁹ , ³⁹ Structurally, DHA and eicosapentaenoic acid integrate into glycerophospholipids, modulating membrane fluidity, protein dynamics, ion permeability, and cholesterol homeostasis while preserving blood–brain barrier integrity, thereby mitigating age‐ and AD‐related dysfunction. Mechanistically, they attenuate Aβ pathology by shifting amyloidogenic amyloid precursor protein processing toward non‐amyloidogenic pathways, enhancing degradation via insulin‐degrading enzyme and neprilysin and promoting clearance through autophagy, immune uptake, and glymphatic drainage. Concurrently, they suppress neuroinflammation through G‐protein coupled receptor 120 (GPR120) and peroxisome proliferator‐activated receptor‐mediated inhibition of Toll‐like receptor 4/NF‐κB signaling, lipoxygenase/cyclooxygenase‐derived anti‐inflammatory metabolites, and displacement of pro‐inflammatory arachidonic acid in phospholipids. They further sustain brain energetics by upregulating mitochondrial respiration genes, stabilizing calcium buffering, and boosting ATP production. In addition, Barmaki et al. ⁴⁰ demonstrated in an in vitro AD model that DHA exerted dual neuroprotective actions through inhibition of peptidyl arginine deiminase 4 (PAD4) and stimulation of autophagy. Collectively, these mechanisms highlight late‐life essential fatty acid deficiency as a modifiable risk factor in AD pathogenesis.

Our study found no significant association between BGMS and AD, suggesting that beneficial components within the DI‐GM may not necessarily reduce AD prevalence. While Prokopidis et al. ⁴¹ reported a positive association between fiber intake and cognitive function, particularly in the Digit Symbol Substitution Test, with benefits plateauing at 34 g/day in adults aged 60 years and older, the overall effect remains limited. Soybean and its products, though beneficial for GM, do not consistently improve dementia or cognitive function. For instance, higher tofu consumption may impair memory and increase dementia risk in the elderly. ³⁴ , ⁴² The potential mechanisms include formaldehyde in tofu, which may cause oxidative damage to the hippocampus and frontal cortex, areas critical for cognitive function ⁴³ ; high tofu intake exacerbating preclinical hypothyroidism, a condition associated with cognitive decline ⁴⁴ ; genistein acting as an estrogen antagonist at high levels, potentially accelerating neuronal death when exposure follows damage, as proposed by the healthy‐cell bias theory ⁴⁵ ; and the negative impact of genistein on cognitive function in the elderly. ⁴⁶ Moreover, Borkent et al. ⁴⁷ reported that replacing animal‐based protein sources with plant‐based foods in older adults reduced both protein quantity and quality, with the risk of inadequate protein intake still a concern. Analogously, evidence on coffee intake and dementia risk is conflicting. A recent meta‐analysis indicated a protective effect for 1 to 3 cups daily, ⁴⁸ while Lefèvre‐Arbogast et al. ⁴⁹ found that coffee reduced dementia risk only in “slower caffeine metabolizers” with the CYP1A2 polymorphisms, showing protection at ≥4 cups/day, but a J‐shaped trend in “faster metabolizers” with reduced risk up to 3 cups/day and increased risk beyond.

This study has several limitations. First, while the cross‐sectional design precludes causal inference between DI‐GM and AD, reverse causation remains possible, as dietary patterns may be shaped either by neurobehavioral changes in preclinical AD or by greater reliance on accessible or processed foods among frail or cognitively impaired individuals. Second, bias may stem from residual confounding due to measurement error, as well as recall inaccuracies in self‐reported DI‐GM and related variables, particularly among cognitively impaired participants. Nevertheless, the large NHANES sample and standardized 24‐h dietary recall protocol likely limit such bias to non‐differential misclassification with respect to AD status, which generally attenuates associations rather than inflating them, suggesting that the observed effects may underestimate the true magnitude. Third, while grounded in GM evidence, the DI‐GM remains unvalidated for cognitive outcomes and may inadequately reflect overall diet quality, as it was developed from a limited number of studies per food or food group, excluding nutrient‐rich items insufficiently examined for microbiota relevance. Further refinements may be guided by emerging evidence. Lastly, although AD develops over decades, the DI‐GM was based on dietary intake at the time of data collection. Nevertheless, adult dietary patterns are generally stable unless modified intentionally for weight loss or medical reasons, supporting their validity as a proxy for long‐term intake in population studies.

In conclusion, this study demonstrated significant positive associations between DI‐GM, DI‐GM ≥6 group, and AD prevalence, with UGMS showing a strong positive and non‐linear relationship with AD prevalence. These findings challenge prior evidence suggesting benefits of higher DI‐GM scores, underscoring the pivotal role of unfavorable components within the DI‐GM in AD onset and progression and emphasizing the importance of maintaining nutritional balance in late life. Longitudinal and interventional studies incorporating dietary assessments and biomarker trajectories are warranted to elucidate causal pathways and inform precision nutrition strategies for AD prevention and management.

CONFLICT OF INTEREST STATEMENT

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

CONSENT STATEMENT

NHANES is carried out by the Centers for Disease Control and Prevention in collaboration with the National Center for Health Statistics (NCHS). The NHANES study protocol underwent review and approval by the NCHS Research Ethics Review Committee. Written informed consent was obtained from all participants.

Supporting information

Supporting Information

Supporting Information

Liu J, Huang S. Association between the dietary index for gut microbiota and Alzheimer's disease: A cross‐sectional study from the National Health and Nutrition Examination Survey (2004 to 2018). Alzheimer's Dement. 2025;17:e70170. 10.1002/dad2.70170