Dietary Acculturation Is Associated With Altered Gut Microbiome, Circulating Metabolites, and Cardiovascular Disease Risk in US Hispanics and Latinos: Results From HCHS/SOL

Abstract

Background: Dietary acculturation, or adoption of dominant culture diet by migrant groups, influences human health. We aimed to examine dietary acculturation and its relationships with cardiovascular disease (CVD), gut microbiota, and blood metabolites among US Hispanic and Latino adults.

Methods: In the HCHS/SOL (Hispanic Community Health Study/Study of Latinos), US exposure was defined by years in the United States (50 states and Washington, DC) and US nativity. A dietary acculturation pattern was derived from 14 172 participants with two 24-hour dietary recalls at baseline (2008-2011) using least absolute shrinkage and selection operator regression, with food groups as predictors of US exposure. We evaluated associations of dietary acculturation with incident CVD across ≈7 years of follow-up (n=211/14 172 cases/total) and gut microbiota (n=2349; visit 2, 2014 to 2017). Serum metabolites associated with both dietary acculturation-related gut microbiota (n=694) and incident CVD (n=108/5256 cases/total) were used as proxy measures to assess the association of diet-related gut microbiome with incident CVD.

Results: We identified an empirical US-oriented dietary acculturation score that increased with US exposure. Higher dietary acculturation score was associated with higher risk of incident CVD (hazard ratio per SD, 1.33 [95% CI, 1.13-1.57]), adjusted for sociodemographic, lifestyle, and clinical factors. Sixty-nine microbial species (17 enriched from diverse species, 52 depleted mainly from fiber-utilizing Clostridia and Prevotella species) were associated with dietary acculturation, driven by lower intakes of whole grains, beans, and fruits and higher intakes of refined grains. Twenty-five metabolites, involved predominantly in fatty acid and glycerophospholipid metabolism (eg, branched-chain 14:0 dicarboxylic acid** and glycerophosphoethanolamine), were associated with both diet acculturation-related gut microbiota and incident CVD. Proxy association analysis based on these metabolites suggested a positive relationship between diet acculturation-related microbiome and risk of CVD (r=0.70, P<0.001).

Conclusions: Among US Hispanic and Latino adults, greater dietary acculturation was associated with elevated CVD risk, possibly through alterations in gut microbiota and related metabolites. Diet and microbiota-targeted interventions may offer opportunities to mitigate CVD burdens of dietary acculturation.

Keywords: Hispanic or Latino; acculturation; cardiovascular diseases; gut microbiome; metabolome.

Figure 1.. Dietary acculturation pattern in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) (n=14,172).

A. Overview of analytical strategy. We identified a dietary acculturation pattern that reflects U.S. exposure in the HCHS/SOL (1.1, n=14,172) and examined the prospective association of the derived dietary pattern with incident CVD (1.2, n=14,172). Then, we examined the association between dietary acculturation pattern and gut microbial features (2, n=2349). Finally, we identified serum metabolites that were associated with both diet-related microbiota and incident CVD (3.1 and 3.2, n=5256 and n=694), and then assessed the association between diet-related species and incident CVD using the identified metabolites as proxy measures (3.3, n=5256 and n=694). Numbers adjacent to arrows represent the sample size in each analysis. B. Least Absolute Shrinkage and Selection Operator regression (LASSO) coefficients for 25 selected food groups in the dietary acculturation score. C-D. Distribution of dietary acculturation score by years living in the U.S. and U.S. nativity in the HCHS/SOL and the National Health and Nutrition Examination Survey (NHANES). The box plots show median and 1.5 times interquartile range for each group. Letters indicate significance groups: groups labelled with the same letters are not significantly different (pairwise Wilcoxon rank sum tests, p<0.05).

Figure 2.. Association of dietary acculturation pattern with incident CVD, accounting for the relationship with other healthy dietary patterns (n=14,172).

A. Spearman correlation between dietary acculturation score and healthy dietary patterns. All p-values<0.001. B. Hazard ratios (95% CIs) of cardiovascular disease (CVD) according to dietary acculturation score, analyzed in continuous (per standard deviation [SD] increment, stratified by sex) and in quartiles (the lowest quartile [Q1] as the reference) format. Survey multivariable Cox proportional hazards models were applied, adjusting for age, sex, field center, Hispanic/Latino background, educational attainment, income, insurance coverage, cigarette use, alcohol use, prevalent diabetes, physical activity, anti-hypertensive medication use, lipid-lowering medication use, family history of CVD, BMI, and total energy intake. The first panel shows results of the fully-adjusted model with covariates mentioned above, while other panels show results of the fully-adjusted model with additional and separate adjustment for healthy dietary patterns: Dietary Approaches to Stop Hypertension [DASH], Healthy Eating Index [HEI-2015], alternate Mediterranean Diet [aMED], and healthful Plant based Diet Index [hPDI].

Figure 3.. Gut microbiome species and functional pathways associated with dietary acculturation pattern (n=2349).

A. The phylogenetic tree included 69 microbial species from 8 taxonomic classes that were associated with the dietary acculturation pattern. We used Analysis of Composition of Microbiomes (ANCOM) models to identify the dietary acculturation-related species (detection level ≥0.7, false discovery rate [FDR]-adjusted p-value (q-value)<0.1), adjusting for age, sex, field center, Hispanic/Latino background, educational attainment, income, insurance coverage, cigarette use, alcohol use, prevalent diabetes, physical activity, anti-hypertensive medication use, lipid-lowering medication use, BMI, total energy intake, antibiotics usage, probiotics usage, and Bristol stool score. Node colors for species reflects taxonomic class. Heatmap shows the effect sizes from multivariable linear regression with dietary acculturation score (continuous variable, per 1-SD increase) as predictor and centered log-ratio (CLR)-transformed abundance of each microbial species as outcome. The heights of the outmost grey bars are proportional to the mean relative abundance of each species. B. The volcano plot shows the association between dietary acculturation pattern and microbial species. Beta coefficients and p-values were from multivariable linear regression models described in A. Species were colored if detection level>=0.7, but species were only labeled if detection level>=0.9 to preserve legibility. C. Partial Spearman correlation between dietary acculturation-related microbiome score and individual food groups, adjusting for age, sex, and field center. Scores were derived by Z-score standardizing the CLR-transformed abundance of species associated with dietary acculturation pattern, followed by summing (if positive) or subtracting (if negative) species according to the association direction with dietary acculturation. Food intakes (servings/day) were adjusted for total energy intake using the residual method and then standardized by Z-score. Dot color indicates direction of the correlation between food group and dietary acculturation score. D. The volcano plot shows the association between dietary acculturation pattern and microbial functional pathways. Beta coefficients and p-values were from multivariable linear regression models. Pathways were colored if detection level ≥0.7.

Figure 4.. Dietary acculturation-related gut microbiome, metabolites, and CVD.

A. Number of serum metabolites (baseline) associated with incident CVD in multivariable Cox proportional hazards models at p-value<0.05 (n=5256), adjusting for age, sex, field center, Hispanic/Latino background, educational attainment, income, insurance coverage, cigarette use, alcohol use, physical activity, prevalent diabetes, anti-hypertensive medication use, lipid-lowering medication use, BMI, total energy intake, family history of CVD, estimated glomerular filtration rate (eGFR), and U.S. nativity. B. Twenty-five metabolites that were associated with both dietary acculturation-related microbiome score (metabolites at visit 2) and incident CVD (metabolites at baseline). Heatmap shows the partial Spearman correlation between dietary acculturation-related microbiome score and metabolites (false discovery rate [FDR]-adjusted p-value<0.1 (q-value), n=694), adjusting for age, sex, field center, and eGFR. Forest plot of the multivariable Cox proportional hazards models shows the hazard ratios and 95% confidence intervals of incident CVD per standard deviation (SD) increase of serum metabolites (p<0.05, n=5256), adjusting for covariates in A. C. The proxy association between dietary acculturation-related microbiome score and CVD, using 25 metabolites as the proxy measures. We correlated the effect sizes for associations between metabolites and CVD (i.e., natural-log-transformed HRs standardized by Z-score, adjusting for same covariates as in [A], n=5256) with the effect sizes for associations between metabolic signatures and gut microbiota (i.e., beta coefficients standardized by Z-score, adjusting for same covariates as in [B], n=694), to derive a Spearman correlation coefficient (r) between these two sets of effect estimates, indicating the prospective relationship of the diet acculturation-related gut microbiome with CVD.