Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jan 1.
Published in final edited form as: J Acad Nutr Diet. 2021 Jul 3;122(1):78–98. doi: 10.1016/j.jand.2021.05.023

The gut microbiome is associated with circulating dietary biomarkers of fruit and vegetable intake in a multiethnic cohort

Cara L Frankenfeld 1,2, Meredith AJ Hullar 3, Gertraud Maskarinec 4, Kristine R Monroe 5, John A Shepherd 6, Adrian A Franke 7, Timothy W Randolph 8, Lynne R Wilkens 9, Carol J Boushey 10, Loïc Le Marchand 11, Unhee Lim 12, Johanna W Lampe 13
PMCID: PMC9019929  NIHMSID: NIHMS1797130  PMID: 34226163

Abstract

Background:

Results from observational studies suggest high diet quality favorably influences the human gut microbiome. Fruit and vegetable consumption is often a key contributor to high diet quality.

Objective:

To evaluate measures of gut bacterial diversity and abundance in relation to serum biomarkers of fruit and vegetable intake.

Design:

Secondary analysis of cross-sectional data.

Participants/setting:

Men and women from Los Angeles, CA and Hawai’i who participated in the Multiethnic Cohort - Adiposity Phenotype Study from 2013–2016 (n=1709).

Main outcome measures:

Gut microbiome diversity and composition in relation to dietary biomarkers.

Statistical analyses:

Carotenoid (beta-carotene, alpha-carotene, cryptoxanthins, lutein, lycopene, and zeaxanthin), tocopherol (alpha, beta+gamma, and delta), and retinol concentrations were assessed in serum. Alpha and beta diversity and composition of the gut microbiome were classified based on 16S rRNA gene sequencing of bacterial DNA from self-collected fecal samples. Global differences in microbial community profiles in relation dietary biomarkers were evaluated using multivariable permutational analysis of variance (PERMANOVA). Associations of alpha diversity (Shannon index), beta diversity (weighted and unweighted UniFrac) with center log-ratio-transformed phyla and genera abundances were evaluated using linear regression, adjusted for covariates.

Results:

Increasing total carotenoid, beta-carotene, alpha-carotene, cryptoxanthin, and lycopene concentrations were associated with higher gut bacterial diversity (Shannon Index) (p<0.001). Total tocopherol, alpha-tocopherol, and delta-tocopherol concentrations contributed significantly to more than 1% of the microbiome variation in gut bacterial community: total tocopherol: 1.74%; alpha-tocopherol: 1.70%; and delta-tocopherol: 1.16% (p<0.001). Higher total carotenoid was associated with greater abundance of some genera relevant for microbial macronutrient metabolism (p<0.001).

Conclusions:

Objective biomarkers of fruit and vegetable intake, particularly carotenoids, were favorably associated with gut bacterial composition and diversity in this multiethnic population. These observations provide supportive evidence that fruit and vegetable intake is related to gut bacterial composition; more work is needed to elucidate how this influences host health.

Keywords: microbiome, food, nutrient, biomarkers, ethnicity

Introduction

Despite the complex relationship between the gut microbiota and host physiology, evidence supports that the gut microbiota influences the overall health of the human host.1–9 Microbial species in the gut utilize nutrients consumed by the host as sources of fuel for growth and proliferation, and dietary composition may influence which microbial species can thrive in the gut environment. Some short-term, human intervention studies of fiber, polyphenols, omega-3 fatty acids, or specific foods have resulted in beneficial changes to the gut microbiome profile or diversity.10–15 These observations support the role of diet in influencing gut microbiome profiles and inform about biological mechanisms. However, these results may not be generalizable to the general population for a variety of reasons, such as smaller sample size, limited range of dietary intake, and homogeneity of personal characteristics, notably health status, specific to a given study. Under usual dietary intake conditions, individuals consume a variety of foods, and interventions are not generally designed to evaluate whole diet relationships with the microbiome. Observational studies in humans provide a needed complement to short-term intervention studies to contribute to an overall picture of dietary influences on the gut microbiome.

Maskarinec et al (2019) showed previously that overall dietary quality assessed using self-reported intake was associated with the gut microbiome and that fruit and vegetable intake was associated with gut microbiome diversity.16 Observational studies of diet rely on subjective and objective measures of intake, and these measures can provide complementary information to provide an overall profile of dietary intake and exposure. Self-reported assessments, such as food frequency questionnaires (FFQs), provide an estimate of usual intake for participants, which informs how long-term diet may contribute to health outcomes. FFQs also allow evaluation of health outcomes in relation to foods or food groups, although they and other self-reported methods may be prone to reporting or recall bias.17,18 Objective measures, such as biomarkers of intake, are not subject to such biases, and provide complementary information about dose and allow for the evaluation of health outcomes in relation to nutrients across a variety of foods. Biomarkers of dietary intake provide a way to reflect intake from a variety of food sources without bias due to self-report. For example, serum carotenoid concentration demonstrate relatively high correlations with fruit and vegetable intakes or overall diet quality across different populations.19–24 To consider diet measured by objective biomarkers, this analysis evaluated associations between serum biomarkers of food intake and gut microbiome diversity and microbiota abundance. To capitalize on unique features of this study population, exploratory analyses were also conducted to evaluate whether associations of dietary biomarkers with microbiome differed across sex or ethnicity.

Methods

The Multiethnic Cohort Adiposity Phenotype Study

The Multiethnic (MEC) cohort and details about data collection methods are published elsewhere.25 Briefly, the MEC is an on-going, prospective study of diet, lifestyle, and genetic risk factors for cancer and other chronic diseases conducted in Hawaii, and Los Angeles, California. Participants in this study were part of a body fat imaging study, the Adiposity Phenotype Study (APS), that recruited participants between 2013 and 2016.26 Individuals were ineligible to participate in APS for any of the following reasons: current body mass index (BMI) was below 18.5 kg/m2 or above 40 kg/m2; antibiotic use in prior three months; current or former smoking within the past two years; presence of soft or metal implants (other than knee or hip replacement); having diabetes requiring insulin; or, other serious health conditions. Individuals with weight change of more than 9 kg in the past year or treatments or procedures that could potentially influence outcomes of interest in APS were re-evaluated for eligibility after six months. Stratified recruitment was conducted to ensure representation across sex, race/ethnicity and six BMI categories. Institutional Review Boards at the University of Hawaii and University of Southern California approved the protocol and written informed consent was provided by all participants. In order to capture diet broadly, both objective (biomarkers) and subjective (self-reported diet assessed via MEC quantitative food frequency questionnaire (FFQ)) methods were used to evaluate dietary intake. The MEC FFQ was developed specifically for this cohort and participants are asked to reported on usual intake in the past year.25,27

Serum Carotenoids and Tocopherols

Lipid-soluble nutrients (including 15 carotenoids and three tocopherols) were analyzed by an isocratic heart-cut 2-dimensional HPLC assay with photo-diode array detection, as detailed elsewhere.28 Assays were validated by participation in quality assurance programs for fat-soluble vitamins organized by the National Institute for Standards and Technology. Coefficients of variation for duplicate analyses for this analytic method range from 2% for β+γ-tocopherol to 11% for trans-zeaxanthin.28 Lutein was the sum of cis1, cis2, cis-anhydro, trans, and trans-anhydro lutein concentrations. Cryptoxanthin was the sum of cis-beta, trans-alpha, and trans-beta cryptoxanthin concentrations. Beta-carotene was the sum of cis beta-carotene and trans beta-carotene concentration.

Fecal Microbiome Analysis

Participants self-collected fecal samples in RNAlater and stored them in their home freezers before mailing or bringing to the clinic, as described previously.29 Briefly, stool samples were stored in RNAlater at −80 °C at study centers and shipped in bulk on dry ice to Fred Hutchinson Cancer Research Center (Fred Hutch), thawed and homogenized, and genomic DNA was extracted as described previously.29 To optimize bacterial genomic DNA extraction, bead beating was applied at 45s (2x) each with samples placed on ice in between.29 Quality control samples, duplicate participant samples, and processing blanks were used to assess variation in library preparation and sequencing batches. Paired-end sequencing of the V1-V3 region of the 16S rRNA gene was performed at Molecular Diagnostics, LLP (Shallowater, TX) on the MiSeq to obtain 2×300 bp paired-end reads (Illumina, San Diego, CA). At Fred Hutch, sequences were processed using QIIME 1.830 and Silva 138.131 to classify sequences to bacterial taxonomy and included parameters in QIIME to exclude low abundant sequences, singletons, and chimeras, and final filtering at the genera level, genera which appeared in <10% of the subjects were removed.29,30 Counts of taxonomic placement (phyla, genera and operational taxonomic units (OTUs)), alpha diversity estimates (Shannon diversity index), and beta diversity matrices (unweighted and weighted UniFrac) were calculated and exported for statistical analysis. Stool samples were collected over a 3-year period of time and were processed and sent for sequencing in 31 batches. ComBat-seq was applied to account for potential batch effects.32 This method aims to remove unwanted (non-biological) variation that is manifested, possibly, in the different batches. It also aims to preserve biological variability and, so, in addition to batch number, an algorithm was provided with each participant’s sex and ethnicity. Additionally, the algorithm used a set of quality-control samples that were run within each batch. The resulting adjusted counts were used in all subsequent analyses.

Other covariates

Other characteristics of individuals included in statistical models were collected from self-report: age (in years), sex (male or female), race/ethnicity (White, African American, Native Hawaiian, Japanese American, or Latino), antibiotic use in the past year (no, yes, unknown), moderate or vigorous physical activity (<0.5, 0.5 to <1, 1 to <1.5, 1.5+ hours/day), smoking status (never, former), and log-transformed energy intake (in kcal, calculated from responses to FFQ). Total body fat measured using dual energy X-ray absorptiometry (DEXA) (as described elsewhere)26, month of clinic visit (January-March, April-June, July-Sept, Oct-Dec) and laboratory batch (A, B, C, D) were also included as covariates in the analysis.

Study sample and statistical analysis

From 1,861 APS participants, further exclusions were made for unavailable data on: microbiome (n=74), serum biomarkers (n=8), dietary information (n=34), total body fat from DEXA (n=21), antibiotic usage in the past 12 months (n=27), or physical activity (n=2). Given some overlap in individuals for missing characteristics, the final sample size for analysis was 1,709. Due to updated microbiome database (see above) and analyses, this sample size differs from prior publications.16 Descriptive statistics were applied to describe the study sample. The associations of serum biomarker quartiles with alpha diversity (Shannon Index) were evaluated using median regression, adjusted for covariates. Permutational multivariate analysis of variance (PERMANOVA) was used to evaluate the relationship between overall microbiome profile based on unweighted and weighted UniFrac distances (R vegan package, ADONIS command) and serum biomarker quartiles, adjusted for age, sex (not in sex-stratified models), ethnicity (not in ethnicity-stratified models), season, physical activity, antibiotic use in the past 12 months, smoking status, fat mass, dietary energy intake, alcohol intake, diabetes, and laboratory batch. PERMANOVA is a statistical analysis that partitions the multivariate variation across a dissimilarity measure, specifically here, the Unifrac weighted and unweighted matrices, according to an ANOVA design with p-values obtained using distribution-free permutation techniques.33,34 This is commonly used in evaluation of the microbiome because of the skewed nature of microbiota abundance across individuals. The test allows for global comparison of total gut bacterial community profiles across characteristics of individuals, such as biomarker concentration quartiles, while adjusting for potential confounding variables. Unweighted and weighted UniFrac distances on OTUs (operational taxonomic unit defined as clusters of sequences that are 97% similar) were chosen in order to incorporate phylogenetic information into the distance measures.35 Unweighted UniFrac provides information about the presence/absence of particular bacteria and weighted UniFrac describes abundance. Higher value of R-squared from PERMANOVA (percent variation explained) was used to identify groups (serum beta-carotene and serum total tocopherol) to evaluate differences in phyla and genera across the categories using median regression. Regression models use the centered log ratio (CLR) values of phylum and genus abundance, adjusted for covariates. The CLR transform essentially uses the logarithm of all taxon counts and centers each sample around its mean and thus accounts for potential differences in the total sequence count in each sample.36 The R compositions packages was used to perform the CLR for phyla and genera counts following the adjustment for batch using ComBat. The resulting values were used as the dependent variables in subsequent statistical analyses. Bonferroni-adjusted p-values were applied for the three biomarkers with the 10 phyla and 152 genera (p<0.000103). Exploratory analyses of the global microbiome test (PERMANOVA) and associations of serum biomarker quartiles with alpha diversity for subgroups defined by sex and ethnicity were conducted through restriction to subgroups (males, females, white, Latino, African American, Japanese American, Native Hawaiian). R statistical software was used for microbiome composition analyses as described above,37 and descriptive analyses and linear regressions were conducted using Stata SE.38

Results

Participants

Participants ranged in age from 60 to 77 years, with a mean age of 69.2 years (Table 1). As expected from the recruitment strategy, proportions of participants across sex, race/ethnicity, BMI, and time of year of stool sample collection were similar. Approximately 16% of the participants had a history of diabetes, more than half reported an hour or more of moderate or vigorous physical activity per day, and 21% had taken antibiotics in the previous year. Mean age was similar across sex and race/ethnicity, but other personal characteristics exhibited some differences, such as a lower prevalence of diabetes history among whites, lower amounts of moderate or vigorous physical activity in African Americans, and higher percent body fat in African Americans and Latinos.

Table 1.

Characteristics of 1709 participants included in microbiome analyses from the Multiethnic Cohort Study who participated in the Adiposity Phenotype Study, stratified by sex and ethnicity

All Female Male African American Hawaiian Japanese Latino White
Age, years, mean ± SD 69.2 ± 2.7 69.1 ± 2.7 69.2 ± 2.8 69.8 ± 2.7 68.6 ± 3.4 68.9 ± 2.5 69.7 ± 2.7 69.0 ± 2.3
Sex, n (%)
 Female 844 (49.4) 865 (100) 0 (0.0) 106 (39.4) 135 (48.2) 229 (53.3) 171 (50.1) 203 (52.2)
 Male 865 (50.6) 0 (0.0) 844 (100) 163 (60.6) 145 (51.8) 201 (46.7) 170 (49.9) 186 (47.8)
Race or ethnicity, n (%)
 Black 269 (15.7) 163 (18.8) 106 (12.6) 269 (100) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)
 Hawaiian 280 (16.4) 145 (16.8) 135 (16.0) 0 (0.0) 280 (100) 0 (0.0) 0 (0.0) 0 (0.0)
 Japanese 430 (25.2) 201 (23.2) 229 (27.1) 0 (0.0) 0 (0.0) 430 (100) 0 (0.0) 0 (0.0)
 Latino 341 (20.0) 170 (19.7) 171 (20.3) 0 (0.0) 0 (0.0) 0 (0.0) 341 (100) 0 (0.0)
 White 389 (22.8) 186 (21.5) 203 (24.1) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 389 (100)
History of diabetes, n (%)
 No 1441 (84.3) 738 (85.3) 703 (83.3) 221 (82.2) 235 (83.9) 362 (84.2) 255 (74.8) 368 (94.6)
 Yes 268 (15.7) 127 (14.7) 141 (16.7) 48 (17.8) 45 (16.1) 68 (15.8) 86 (25.2) 21 (5.4)
Moderate or vigorous activity, n (%)
 <0.5 hr/day 339 (19.8) 192 (22.2) 147 (17.4) 81 (30.1) 44 (15.7) 83 (19.3) 87 (25.5) 44 (11.3)
 0.5 to <1 hr/day 377 (22.1) 206 (23.8) 171 (20.3) 65 (24.2) 64 (22.9) 87 (20.2) 86 (25.2) 75 (19.3)
 1 to <1.5 hr/day 357 (20.9) 178 (20.6) 179 (21.1) 53 (19.7) 51 (18.2) 95 (22.1) 66 (19.4) 92 (23.7)
 1.5+ hr/day 636 (37.2) 289 (33.4) 347 (41.1) 70 (26.0) 121 (43.2) 165 (38.4) 102 (29.9) 178 (45.8)
Antibiotics in past year, n (%)
 No 1338 (78.3) 648 (74.9) 690 (81.8) 212 (78.8) 222 (79.3) 353 (82.1) 259 (76.0) 292 (75.1)
 Yes 371 (21.7) 217 (25.1) 154 (18.2) 57 (21.2) 58 (20.7) 77 (17.9) 82 (24.0) 97 (24.9)
Smoking status, n (%)
 Never 1047 (61.3) 608 (70.3) 439 (52.0) 156 (58.0) 166 (59.3) 252 (58.6) 229 (67.2) 244 (62.7)
 Former 662 (38.7) 257 (29.7) 405 (48.0) 113 (42.0) 114 (40.7) 178 (41.4) 112 (32.8) 145 (37.3)
Stool sample collection, season, n (%)
 January-March 439 (25.7) 221 (25.5) 218 (25.8) 75 (27.9) 75 (26.8) 108 (25.1) 92 (27.0) 89 (22.9)
 April-June 414 (24.2) 208 (24.0) 206 (24.4) 57 (21.2) 69 (24.6) 125 (29.1) 69 (20.2) 94 (24.2)
 July-September 439 (25.7) 218 (25.2) 221 (26.2) 71 (26.4) 55 (19.6) 125 (29.1) 79 (23.2) 109 (28.0)
 October-December 417 (24.4) 218 (25.2) 199 (23.6) 66 (24.5) 81 (28.9) 72 (16.7) 101 (29.6) 97 (24.9)
Mean total body fat by dual-energy X-ray absorptiometry (kg), mean ± SD 25.3 ± 8.7 27.6 ± 8.9 22.9 ± 7.7 31.7 ± 9.7 24.7 ± 7.8 20.5 ± 6.0 28.9 ± 7.9 23.4 ± 7.7
Mean total energy intake (kcal)a, mean ± SD 1870 ± 910 1740 ± 900 2000 ± 910 1890 ± 1780 1950 ± 1010 1730 ± 700 1940 ± 920 1890 ± 830
Open in a new tab
a

Energy intake was assessed by quantitative food frequency questionnaire that assesses usual intake over the past year.25,27

Dietary Biomarkers

Serum biomarkers did not appear to be collinear when quartiles were compared; the strongest agreement between two biomarkers was seen for beta-carotene and total carotenoids quartiles (60.6% percent agreement, kappa=0.47, Table 2).

Table 2.

Percent agreement and kappa values for the serum biomarker quartiles in in 1709 individuals in the Multiethnic Cohort Study.

Statistic Serum Nutrient Quartiles Total carotenoids Beta-carotene Alpha-carotene Cryptoxanthin Lutein Lycopene Retinol Alpha-tocopherol Beta+gamma-tocopherol Delta-tocopherol Total tocopherol Zeaxanthin
Percent Agreement a Total carotenoids 100.0
Beta-carotene 60.6 100.0
Alpha-carotene 48.1 51.7 100.0
Cryptoxanthin 52.5 44.9 40.1 100.0
Lutein 59.0 41.3 40.0 41.5 100.0
Lycopene 44.9 36.4 33.4 34.1 37.0 100.0
Retinol 31.1 27.0 25.4 26.4 33.5 27.7 100.0
Alpha-tocopherol 39.0 37.8 33.2 36.2 35.7 34.2 29.7 100.0
Beta+gamma-tocopherol 24.6 22.6 23.8 23.2 25.2 28.3 25.1 22.5 100.0
Delta-tocopherol 36.0 33.1 31.0 32.7 32.7 32.2 26.6 33.9 37.7 100.0
Total tocopherol 38.0 36.0 33.8 35.0 36.0 35.0 30.3 78.1 29.7 38.9 100.0
Zeaxanthin 44.9 36.4 32.1 37.6 50.8 32.7 31.5 34.0 27.2 33.6 34.2 100.0
Kappa b Total carotenoids 1.00
Beta-carotene 0.47 1.00
Alpha-carotene 0.31 0.36 1.00
Cryptoxanthin 0.37 0.27 0.20 1.00
Lutein 0.45 0.22 0.20 0.22 1.00
Lycopene 0.27 0.15 0.11 0.12 0.16 1.00
Retinol 0.08 0.03 0.01 0.02 0.11 0.04 1.00
Alpha-tocopherol 0.19 0.17 0.11 0.15 0.14 0.12 0.06 1.00
Beta+gamma-tocopherol 0.00 −0.03 −0.02 −0.02 0.00 0.04 0.00 −0.03 1.00
Delta-tocopherol 0.15 0.11 0.08 0.10 0.10 0.10 0.02 0.12 0.17 1.00
Total tocopherol 0.17 0.15 0.12 0.13 0.15 0.13 0.07 0.71 0.06 0.19 1.00
Zeaxanthin 0.27 0.15 0.09 0.17 0.34 0.10 0.09 0.12 0.03 0.11 0.12 1.00
Open in a new tab
a

Percent agreement is calculate as the number of observations whose quartile classification agrees for the two values being compared divided by the total number of observations

b

kappa is calculated as (observed agreement − expected agreement)/(1-expected agreement)

Microbiome Diversity

The mean Shannon index in the study sample was 6.56 (SD=0.72, range: 2.81–8.23). In the combined population, higher total carotenoid and specific carotenoid (beta-carotene, alpha-carotene, cryptoxanthin, and lycopene) concentrations were correlated with higher Shannon Index values (Figure 1), whereas lower total tocopherol, alpha-tocopherol, and retinol concentrations were correlated with higher Shannon Index values. Similar direction or null relationships were observed in males vs. females and across ethnic groups (Figure 2).

Figure 1.

Figure 1.

Open in a new tab

Difference in mean Shannon index between the highest category and lowest category of serum biomarkers in 1709 individuals in the Multiethnic Cohort Study Adiposity Phenotype Study. Beta-coefficient and standard error for each quartile value is presented with the lowest quartile as the reference group and the p-for-trend for each association across the quartiles from median regression is presented. Quartile values are in ng/ml.a

Figure 2.

Figure 2.

Figure 2.

Figure 2.

Figure 2.

Open in a new tab

Difference in mean Shannon index between the highest category and lowest category of serum biomarkers in 1709 individuals in the Multiethnic Cohort Study, stratified by sex (A) and ethnicity (B). Beta-coefficient and standard error for each quartile value is presented with the lowest quartile as the reference group and the p-for-trend for each association across the quartiles from median regression is presented. Quartile values are in ng/ml.a

Microbiome Profile

Total tocopherol, alpha-tocopherol, and delta-tocopherol concentrations contributed significantly to more than 1% of the microbiome variation in PERMANOVA analyses using the unweighted UniFrac microbiome beta-diversity matrix (p<0.001, Table 3). There were no notable differences or contributions when stratified by sex. By race and ethnicity, tocopherol and retinol concentrations contributed more strongly to microbiome profiles in African American and Native Hawaiian participants to a greater degree than other racial or ethnic groups. Total carotenoids and many of the specific carotenoid concentrations contributed to more than 1% of the variation in microbiome profiles for African Americans, Native Hawaiian, Latino, and White participants with unweighted UniFrac beta-diversity matrix, and additionally for these groups plus Japanese American participants with weighted UniFrac beta-diversity matrix. However, many of the observations were not statistically significant in the stratified analyses (p>0.001).

Table 3.

Percentage of microbiome variation explained in unweighted and weighted UniFrac analyses for self-reported and biomarker measures of dietary intake in 1709 individuals in the Multiethnic Cohort Study, stratified by sex and ethnicity. Percentage variation >1% indicated in bold.

All Female Male African American Hawaiian Japanese Latino White
% Var. p-value % Var. p-value % Var. p-value % Var. p-value % Var. p-value % Var. p-value % Var. p-value % Var. p-value
Unweighted Unifrac
Carotenoids
Total carotenoids 0.30 0.061 0.54 <0.001 0.59 <0.001 1.13 0.199 1.59 <0.001 0.94 0.003 1.17 0.003 1.08 0.005
Beta-carotene 0.25 0.099 0.61 <0.001 0.56 <0.001 1.33 0.040 1.30 0.025 0.85 0.011 1.31 <0.001 1.22 <0.001
Alpha-carotene 0.19 0.244 0.55 <0.001 0.55 <0.001 1.44 0.012 1.16 0.114 0.98 0.002 1.54 <0.001 1.12 0.002
Cryptoxanthin 0.35 0.025 0.54 <0.001 0.58 <0.001 1.18 0.121 1.15 0.131 0.92 0.007 1.17 0.003 1.30 <0.001
Lutein 0.15 0.363 0.38 0.107 0.45 0.023 1.06 0.415 1.12 0.199 0.73 0.169 0.90 0.192 1.53 <0.001
Lycopene 0.45 0.009 0.52 <0.001 0.42 0.044 1.20 0.115 1.27 0.043 0.66 0.425 1.02 0.044 0.86 0.094
Zeaxanthin 0.15 0.370 0.38 0.106 0.36 0.232 1.13 0.234 0.92 0.715 0.67 0.422 0.88 0.254 0.87 0.075
Retinol
Retinol 0.18 0.280 0.43 0.020 0.42 0.048 1.06 0.360 1.20 0.070 0.71 0.220 0.78 0.693 0.70 0.566
Tocopherols
Total tocopherol 1.74 <0.001 0.72 <0.001 0.71 <0.001 1.35 0.016 1.39 0.011 0.77 0.088 0.77 0.730 0.89 0.068
Alpha-tocopherol 1.70 <0.001 0.67 <0.001 0.74 <0.001 1.24 0.068 1.35 0.016 0.85 0.024 0.85 0.406 1.00 0.014
Beta+gamma-tocopherol 0.19 0.230 0.40 0.061 0.39 0.104 1.06 0.380 1.22 0.061 0.79 0.065 0.90 0.217 1.02 0.009
Delta-tocopherol 1.16 <0.001 0.72 <0.001 0.41 0.061 1.09 0.328 1.54 0.003 0.62 0.643 1.18 0.007 0.79 0.217
Weighted Unifrac
Carotenoids
Total carotenoids 0.31 0.095 0.76 0.005 0.69 0.019 1.25 0.265 1.43 0.135 1.32 0.016 0.95 0.318 1.32 0.031
Beta-carotene 0.43 0.027 0.93 <0.001 0.82 0.007 1.87 0.024 1.43 0.118 1.22 0.037 0.71 0.616 1.69 0.005
Alpha-carotene 0.30 0.114 0.83 0.002 0.58 0.060 1.03 0.523 1.45 0.103 1.04 0.073 1.39 0.072 1.53 0.007
Cryptoxanthin 0.31 0.086 0.78 0.008 0.70 0.013 1.53 0.108 1.12 0.355 2.04 <0.001 1.28 0.100 1.48 0.013
Lutein 0.29 0.112 0.60 0.035 0.55 0.077 1.06 0.493 1.12 0.356 0.85 0.199 0.95 0.328 1.49 0.009
Lycopene 0.30 0.133 0.73 <0.001 0.42 0.218 1.37 0.192 1.49 0.117 0.86 0.199 0.80 0.514 0.96 0.150
Zeaxanthin 0.37 0.050 0.58 0.041 0.43 0.208 0.92 0.659 0.67 0.932 0.85 0.225 0.64 0.758 0.84 0.303
Retinol
Retinol 0.16 0.473 0.49 0.090 0.34 0.430 0.88 0.717 0.82 0.745 0.83 0.216 0.65 0.734 0.73 0.460
Tocopherols
Total tocopherol 0.16 0.496 0.46 0.134 0.49 0.119 0.86 0.768 1.61 0.060 0.65 0.478 1.06 0.215 0.89 0.238
Alpha-tocopherol 0.16 0.464 0.48 0.098 0.43 0.219 0.72 0.912 1.54 0.089 0.56 0.649 1.03 0.253 0.82 0.331
Beta+gamma-tocopherol 0.33 0.074 0.97 <0.001 0.33 0.505 1.09 0.435 1.45 0.136 0.85 0.204 1.04 0.245 1.45 0.012
Delta-tocopherol 0.27 0.149 0.62 0.013 0.36 0.421 0.79 0.849 1.07 0.400 1.97 <0.001 0.81 0.495 1.04 0.128
Open in a new tab

Microbiome Composition

Differences in phyla and genera were explored across the quartiles of total carotenoid, total tocopherol, and retinol concentrations. None of the serum biomarkers were significantly associated with any of ten phyla (p>0.001; data not shown). Serum retinol was not associated with any genera. Higher serum total carotenoid concentrations were associated with higher abundance of 8 genera: Clostridium sensu stricto 1, Coprococcus 2, Eubacterium eligens group, Lachnospira, Lachnospiraceae NC2004 group, Lachnospiraceae ND3007 group, Ruminococcaceae, and Ruminococcus 1; and, was associated with lower abundance of Ruminoclostridium 5 (Figure 3). Higher total tocopherol concentrations were associated with higher abundance of Eryspilepatoclostridium, and lower abundances of Eubacterium coprostanoligens group and Ruminococcaceae UCG-002.

Figure 3.

Figure 3.

Open in a new tab

Significant associations of quartiles of serum total tocopherol and serum beta-carotene with specific bacterial genera in 1709 individuals in the Multiethnic Cohort Study Adiposity Phenotype Study. Beta-coefficient and standard error for the center log-transformed regression coefficients for each quartile value of dietary biomarker is presented with the lowest quartile as the reference group and the p-for-trend for each association across the quartiles from median regression is presented with bacterial name in the axis.

Discussion

In this multiethnic study, serum biomarkers of food intake were associated with specific bacteria, a higher percent of variation explained within racial or ethnic groups, and several associations with bacterial diversity. In all participants, serum tocopherol concentrations explained a higher percentage of variation in the overall bacterial community structure (beta diversity), but there were some potentially important associations with serum total and some specific carotenoids when stratified by race/ethnicity.

Higher gut bacterial diversity is generally believed to be beneficial to the host, and low diversity may reflect a gut bacterial imbalance often associated with poor health/disease (dysbiosis). Low-diversity dysbiosis has been linked to a number of diseases, including liver disease, inflammatory bowel disease, and Clostridium difficile infection.39 In this analysis, serum biomarkers that are generally associated with fruit and vegetable intake,21,40 such as higher concentrations of carotenoids, cryptoxanthin, lutein, and lycopene, were associated with higher alpha diversity (Shannon Index). Higher serum total carotenoid concentration was also associated with higher abundance of several bacteria that are involved in carbohydrate or other macronutrient fermentation, such as those in the family Lachnospiraceae and Ruminococcaceae.41–43 There is a limited body of work that has evaluated carotenoids and the gut microbiome. In a study of cystic fibrosis patients, higher beta-carotene intake was associated with overall microbiome profiles, and associated with higher abundance the genus Clostridium, the order Gemellales, and the phylum Firmicutes; and, associated with lower abundance of Bacteroides.44 A proposed mechanism for the relationship between carotenoids and the gut microbiome is that carotenoids can increase the IgA production, and maturation of gut-associated immune tissues45, although the specific mechanistic actions for IgA on microbiome composition are not clear.46 Another possible explanation is that carotenoids may serve as a marker of dietary fiber or other nutrients in vegetables.47 In a study of overweight or obese pregnant women in Finland, an association was observed between higher Index of Dietary Quality (IDQ) and higher alpha diversity (Shannon and Chao indices), phylogenetic diversity, and counts of OTUs.48 The authors further evaluated particular foods with gut microbial diversity, and observed that higher intakes of whole grains and vegetables were significantly associated with higher diversity. Several studies have observed that adherence to a Mediterranean diet, which is a plant-based diet emphasizing fruits, vegetables, whole grains, legumes, and nuts, is associated favorably with the presence of genera in the Ruminococcaceae and Lachnospiraceae families that metabolize these dietary components of a plant-enriched diet.49–52

Associations of higher abundance were observed for higher serum total tocopherol concentrations with bacteria that may be harmful, including associations with Erysipelatoclostridium, which are bacteria that can cause host infections.53 Intervention studies with tocopherol in various animal models have observed gut microbiome changes.54–56 Gamma-tocopherol, which contributes to total tocopherol concentrations, provides an indicator of inflammatory status,57 and scavenges reactive nitrogen oxide species that can decay the integrity of gut cell membranes. It has also been hypothesized that scavenging of reactive oxidant species may enhance overall anaerobic bacterial metabolism potential required for dietary fiber digestion.55–57 However, more research is needed to specifically evaluate tocopherol and the microbiome in humans. Altogether, these results suggest that micronutrients may be important contributors to the overall microbiome or strong indicators of overall diet quality. However, this is a cross-sectional analysis and temporality cannot be established; for example, as hypothesized for reactive oxidant species influences on dietary fiber digestion,55–57 the microbiome may influence serum nutrient concentrations through their digestive action on dietary fiber, and further studies using intervention, longitudinal, or cohort designs are needed to evaluate temporality.

Although there were some limitations in evaluation due to sample size when stratified by race and ethnicity, there were some suggestions of associations that may be worth evaluating in larger studies. Ethnicity attenuated the associations between diet and microbiome profiles, suggesting that there may be some important overlap between diet and ethnicity. In stratified analyses, correlations of biomarkers with bacterial diversity were similar across ethnic groups. The percentage variation explained was sometimes greater when stratified by race/ethnicity, and the biomarkers that explained more variation differed across race/ethnicity. These observations align with other work that has observed that race or ethnicity may be strong contributors to the variation in microbiome across individuals, even when living in the same geographic area.58 There are some hypotheses as to why this could occur that warrant further evaluation with larger datasets and different populations. First, within racial or ethnic groups, it is possible that the foods that make up the food group or contribute to the biomarker are more homogeneous, which could provide a stronger association if there are some foods that are more strongly associated with the gut microbiome. These observations support the need for future work in this and other populations to evaluate specific dietary differences across ethnic groups and microbiome profiles. Second, it is possible that these values are different from one another and the sample size here did not allow for us to formally test for interactions between diet and serum biomarkers and race/ethnicity.

There are some notable strengths of this work to highlight. One, the multiethnic composition of the cohort allowed for the exploration of differences by ethnicity, including some ethnicities that are less represented in the current gut microbiome literature. Two, biomarkers of dietary intake provide an evaluation of the relationship between diet and gut microbiome that is independent of self-report. Three, the sample size for the main analyses was large for observational studies of the microbiome. There are also some limitations of this analysis that should be considered. One, the analysis was cross-sectional, and it is not possible to evaluate the temporal relationship between dietary intake and gut microbiome profiles in analysis. However, relationships between a single time point of microbiome data with diet quality at two time points of dietary intake collection in this cohort were similar16, suggesting that dietary intake, and correspondingly related biomarker concentrations, may be reasonably consistent over time and the direction of association is that diet influences the microbiome composition. Additional studies with longitudinal designs with multiple time points for diet and microbiome assessment would help to better elucidate temporal relationships between the gut microbiome and diet. Despite this limitation, results of this work provide informative directions for future dietary analyses. Two, dietary biomarkers and gut microbiome profiles were measured only once, and both diet and the gut microbiome exhibit intraindividual variation.59 However, lipid-soluble micronutrients reflect intake over a longer time period than water-soluble nutrients, and other work indicates that a single microbiome sample provides a reasonable reflection of microbiome community.29 The bias expected from this intraindividual variation is a decreased ability to detect associations, but internal validity should remain high. Three, QIIME 1.8 is an older analysis program for microbiome work and the use of OTUs has been criticized for clustering sequences at potentially arbitrarily defined threshold.60,61 While this is major limitation and the results may have been a bit stronger with output from Qiime2, these within-study comparisons here are likely not biased toward false associations by use of the older version of QIIME. Four, the older population of the study may limit generalizability to younger populations, and more work is generally needed to understand generational and age differences in the gut microbiome. Five, smaller sample sizes in the race/ethnicity strata limited power to detect associations. Due to the small sample sizes in some ethnic groups, exploratory analyses were not conducted for microbiome composition stratified by sex and ethnicity. More work in larger and diverse study populations is warranted to better understand how inherent characteristics (such as race, sex or age) interact with diet to influence the gut microbiome. While these limitations support careful interpretation and context to the results, there are still notable observations that inform understanding of the human microbiome.

Conclusion

Overall, biomarkers of food intake, particularly serum concentrations of carotenoids and tocopherols, were associated with gut microbiome profiles. Objective biomarkers offer some advantages in that they provide an aggregation of nutrients across foods and are not subject to reporting errors. Notably, these results suggest that total carotenoids, or the foods that they represent, may be beneficial for the microbiome, and this may reflect adherence to a diet higher in fruits and vegetables. There was some suggestion that there may be ethnic differences, but this needs further evaluation. Overall, these observations from a cross-sectional study using an objective biomarker of dietary intake provide a foundation for further dietary intervention studies or prospective cohort studies to elucidate the underlying dietary, environmental, social, or inherent mechanisms contributing to understanding of the diet-microbiome relationship across populations.

Research Snapshot.

Research Question:

How are dietary biomarkers associated with gut bacterial diversity and abundance?

Key Findings:

In this cross-sectional study of 1709 members of a multiethnic cohort, higher serum tocopherol and specific serum carotenoid concentrations were associated with higher bacterial diversity and explained some variation in the bacterial abundance. Higher serum retinol concentrations were significantly associated with lower diversity and did not contribute to overall variation in bacterial abundance.

Acknowledgments

Funding/financial disclosures:

This work was supported by US National Cancer Institute (NCI) grants P01CA168530, U01CA164973, P30 CA071789, P30 CA015704, UL1TR000130.

Footnotes

Conflict of interest disclosures:

None of the authors report a conflict of interest related to the study

Contributor Information

Cara L. Frankenfeld, George Mason University, 4400 University Drive MS 5B7, Fairfax, VA, 22030; Associate Professor and Program Director, Master of Public Health Program; University of Puget Sound, 1500 N. Warner St, Tacoma, WA 98416.

Meredith A.J. Hullar, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109.

Gertraud Maskarinec, University of Hawaii Cancer Center.

Kristine R. Monroe, Keck School of Medicine, University of Southern California.

John A. Shepherd, Population Sciences in the Pacific Program, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Adrian A. Franke, Cancer Biology Program, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Timothy W. Randolph, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109.

Lynne R. Wilkens, Population Sciences in the Pacific Program, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Carol J. Boushey, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Loïc Le Marchand, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Unhee Lim, Population Sciences in the Pacific Program, University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, HI 96813.

Johanna W. Lampe, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109.

References

  • 1.Tilg H, Moschen AR. Microbiota and diabetes: an evolving relationship. Gut. 2014;63(9):1513–1521. doi: 10.1136/gutjnl-2014-306928 [DOI] [PubMed] [Google Scholar]
  • 2.Qin J, Li Y, Cai Z, et al. A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature. 2012;490(7418):55–60. doi: 10.1038/nature11450 [DOI] [PubMed] [Google Scholar]
  • 3.Le Chatelier E, Nielsen T, Qin J, et al. Richness of human gut microbiome correlates with metabolic markers. Nature. 2013;500(7464):541–546. doi: 10.1038/nature12506 [DOI] [PubMed] [Google Scholar]
  • 4.Fu J, Bonder MJ, Cenit MC, et al. The Gut Microbiome Contributes to a Substantial Proportion of the Variation in Blood Lipids. Circ Res. 2015;117(9):817–824. doi: 10.1161/CIRCRESAHA.115.306807 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Turnbaugh PJ, Hamady M, Yatsunenko T, et al. A core gut microbiome in obese and lean twins. Nature. 2009;457(7228):480–484. doi: 10.1038/nature07540 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Ley RE. Obesity and the human microbiome. Curr Opin Gastroenterol. 2010;26(1):5–11. doi: 10.1097/MOG.0b013e328333d751 [DOI] [PubMed] [Google Scholar]
  • 7.Flier JS, Mekalanos JJ. Gut check: testing a role for the intestinal microbiome in human obesity. Sci Transl Med. 2009;1(6):6ps7. doi: 10.1126/scitranslmed.3000483 [DOI] [PubMed] [Google Scholar]
  • 8.Hullar MA, Lampe JW. The gut microbiome and obesity. Nestle Nutr Work Ser. 2012;73:67–79. doi: 10.1159/000341288 [DOI] [PubMed] [Google Scholar]
  • 9.Kelly TN, Bazzano LA, Ajami NJ, et al. Gut Microbiome Associates With Lifetime Cardiovascular Disease Risk Profile Among Bogalusa Heart Study Participants. Circ Res. 2016;119(8):956–964. doi: 10.1161/CIRCRESAHA.116.309219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Costantini L, Molinari R, Farinon B, Merendino N. Impact of Omega-3 Fatty Acids on the Gut Microbiota. Int J Mol Sci. 2017;18(12). doi: 10.3390/ijms18122645 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Haro C, Villatoro M, Vaquero L, et al. The Dietary Intervention of Transgenic Low-Gliadin Wheat Bread in Patients with Non-Celiac Gluten Sensitivity (NCGS) Showed No Differences with Gluten Free Diet (GFD) but Provides Better Gut Microbiota Profile. Nutrients. 2018;10(12). doi: 10.3390/nu10121964 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Holscher HD, Caporaso JG, Hooda S, Brulc JM, Fahey GCJ, Swanson KS. Fiber supplementation influences phylogenetic structure and functional capacity of the human intestinal microbiome: follow-up of a randomized controlled trial. Am J Clin Nutr. 2015;101(1):55–64. doi: 10.3945/ajcn.114.092064 [DOI] [PubMed] [Google Scholar]
  • 13.Sawicki CM, Livingston KA, Obin M, Roberts SB, Chung M, McKeown NM. Dietary Fiber and the Human Gut Microbiota: Application of Evidence Mapping Methodology. Nutrients. 2017;9(2). doi: 10.3390/nu9020125 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Wilson R, Willis J, Gearry RB, et al. SunGold Kiwifruit Supplementation of Individuals with Prediabetes Alters Gut Microbiota and Improves Vitamin C Status, Anthropometric and Clinical Markers. Nutrients. 2018;10(7). doi: 10.3390/nu10070895 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Duenas M, Munoz-Gonzalez I, Cueva C, et al. A survey of modulation of gut microbiota by dietary polyphenols. Biomed Res Int. 2015;2015:850902. doi: 10.1155/2015/850902 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Maskarinec G, Hullar MAJ, Monroe KR, et al. Fecal Microbial Diversity and Structure Are Associated with Diet Quality in the Multiethnic Cohort Adiposity Phenotype Study. J Nutr. June 2019. doi: 10.1093/jn/nxz065 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Prentice RL. Measurement error and results from analytic epidemiology: dietary fat and breast cancer. J Natl Cancer Inst. 1996;88(23):1738–1747. doi: 10.1093/jnci/88.23.1738 [DOI] [PubMed] [Google Scholar]
  • 18.Naska A, Lagiou A, Lagiou P. Dietary assessment methods in epidemiological research: current state of the art and future prospects. F1000Research. 2017;6:926. doi: 10.12688/f1000research.10703.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Frankenfeld CL, Lampe JW, Shannon J, et al. Fruit and vegetable intakes in relation to plasma nutrient concentrations in women in Shanghai, China. Public Heal Nutr. 2012;15(1):167–175. doi: 10.1017/s1368980011001029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Lampe JW, Huang Y, Neuhouser ML, et al. Dietary biomarker evaluation in a controlled feeding study in women from the Women’s Health Initiative cohort. Am J Clin Nutr. 2017;105(2):466–475. doi: 10.3945/ajcn.116.144840 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Aumueller N, Boushey CJ, Franke AA, et al. Diet quality measured by four a priori-defined diet quality indices is associated with lipid-soluble micronutrients in the Multiethnic Cohort Study (MEC). Eur J Clin Nutr. 2019;73(5):703–713. doi: 10.1038/s41430-018-0272-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Townsend MS, Shilts MK, Styne DM, et al. Vegetable behavioral tool demonstrates validity with MyPlate vegetable cups and carotenoid and inflammatory biomarkers. Appetite. 2016;107:628–638. doi: 10.1016/j.appet.2016.09.002 [DOI] [PubMed] [Google Scholar]
  • 23.Shiraishi M, Haruna M, Matsuzaki M, Murayama R, Sasaki S. Validity of a diet history questionnaire estimating β-carotene, vitamin C and α-tocopherol intakes in Japanese pregnant women. Int J Food Sci Nutr. 2013;64(6):694–699. doi:. 10.3109/09637486.2013775225 [DOI] [PubMed] [Google Scholar]
  • 24.Marchand L Le, Hankin JH, Carter FS, et al. A Pilot Study on the Use of Plasma Carotenoids and Ascorbic Acid as Markers of Compliance to a High Fruit and Vegetable Dietary Intervention. Cancer Epidemiol Biomarkers Prev. 1994;3(3):245–251. [PubMed] [Google Scholar]
  • 25.Kolonel LN, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol. 2000;151(4):346–357. doi: 10.1093/oxfordjournals.aje.a010213 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Lim U, Monroe KR, Buchthal S, et al. Propensity for Intra-abdominal and Hepatic Adiposity Varies Among Ethnic Groups. Gastroenterology. 2019;156(4):966–975.e10. doi: 10.1053/j.gastro.2018.11.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Stram DO, Hankin JH, Wilkens LR, et al. Calibration of the dietary questionnaire for a multiethnic cohort in Hawaii and Los Angeles. Am J Epidemiol. 2000;151(4):358–370. doi: 10.1093/oxfordjournals.aje.a010214 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Franke AA, Morrison CM, Custer LJ, Li X, Lai JF. Simultaneous analysis of circulating 25-hydroxy-vitamin D3, 25-hydroxy-vitamin D2, retinol, tocopherols, carotenoids, and oxidized and reduced coenzyme Q10 by high performance liquid chromatography with photo diode-array detection using C18 and C30 column. J Chromatogr A. 2013;1301:1–9. doi: 10.1016/j.chroma.2013.05.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Fu BC, Randolph TW, Lim U, et al. Temporal Variability and Stability of the Fecal Microbiome: The Multiethnic Cohort Study. Cancer Epidemiol Biomarkers Prev. 2019;28(1):154–162. doi: 10.1158/1055-9965.EPI-18-0348 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 2010;7(5):335–336. doi: 10.1038/nmeth.f.303 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.deNBI. Silva: High Quality Ribosomal Databases. https://www.arb-silva.de/.
  • 32.Zhang Y, Parmigiani G, Johnson WE. ComBat-seq: batch effect adjustment for RNA-seq count data. NAR genomics Bioinforma. 2020;2(3):lqaa078. doi: 10.1093/nargab/lqaa078 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Anderson MJ. A new method for non-parametric multivariate analysis of variance. Aust Ecol. 2001;26:32–46. [Google Scholar]
  • 34.McArdle BH, Anderson MJ. Fitting multivariate models to community data: a comment on distance based redundacy analysis. Ecology. 2001;82:290–297. [Google Scholar]
  • 35.Lozupone C, Hamady M, Knight R. UniFrac - An online tool for comparing microbial community diversity in a phylogenetic context. BMC Bioinformatics. 2006;7:1–14. doi: 10.1186/1471-2105-7-371 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Aitchison J The Statistical Analysis of Compositional Data. London: Chapman & Hall; 1986. [Google Scholar]
  • 37.R Core Team. R: A language and environment for statistical computing. 2020. https://www.r-project.org/.
  • 38.StataCorp. Stata Statistical Software: Release 15. 2017.
  • 39.Kriss M, Hazleton KZ, Nusbacher NM, Martin CG, Lozupone CA. Low diversity gut microbiota dysbiosis: drivers, functional implications and recovery. Curr Opin Microbiol. 2018;44:34–40. doi: 10.1016/j.mib.2018.07.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Guillermo C, Boushey CJ, Franke AA, et al. Diet Quality and Biomarker Profiles Related to Chronic Disease Prevention: The Multiethnic Cohort Study. J Am Coll Nutr. July 2019:1–8. doi: 10.1080/07315724.2019.1635921 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Vacca M, Celano G, Calabrese FM, Portincasa P, Gobbetti M, De Angelis M. The Controversial Role of Human Gut Lachnospiraceae. Microorganisms. 2020;8(4). doi: 10.3390/microorganisms8040573 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Flint HJ, Duncan SH, Scott KP, Louis P. Interactions and competition within the microbial community of the human colon: links between diet and health. Env Microbiol. 2007;9(5):1101–1111. doi: 10.1111/j.1462-2920.2007.01281.x [DOI] [PubMed] [Google Scholar]
  • 43.Flint HJ, Scott KP, Duncan SH, Louis P, Forano E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012;3(4):289–306. doi: 10.4161/gmic.19897 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Li L, Krause L, Somerset S. Associations between micronutrient intakes and gut microbiota in a group of adults with cystic fibrosis. Clin Nutr. 2017;36(4):1097–1104. doi: 10.1016/j.clnu.2016.06.029 [DOI] [PubMed] [Google Scholar]
  • 45.Lyu Y, Wu L, Wang F, Shen X, Lin D. Carotenoid supplementation and retinoic acid in immunoglobulin A regulation of the gut microbiota dysbiosis. Exp Biol Med (Maywood). 2018;243(7):613–620. doi: 10.1177/1535370218763760 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Catanzaro JR, Strauss JD, Bielecka A, et al. IgA-deficient humans exhibit gut microbiota dysbiosis despite secretion of compensatory IgM. Sci Rep. 2019;9(1):1–10. doi: 10.1038/s41598-019-49923-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Wallström P, Wirfält E, Lahmann PH, Gullberg B, Janzon L, Berglund G. Serum concentrations of beta-carotene and alpha-tocopherol are associated with diet, smoking, and general and central adiposity. Am J Clin Nutr. 2001;73(4):777–785. doi: 10.1093/ajcn/73.4.777 [DOI] [PubMed] [Google Scholar]
  • 48.Laitinen K, Mokkala K. Overall Dietary Quality Relates to Gut Microbiota Diversity and Abundance. Int J Mol Sci. 2019;20(8). doi: 10.3390/ijms20081835 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Haro C, Garcia-Carpintero S, Rangel-Zuniga OA, et al. Consumption of Two Healthy Dietary Patterns Restored Microbiota Dysbiosis in Obese Patients with Metabolic Dysfunction. Mol Nutr Food Res. 2017;61(12). doi: 10.1002/mnfr.201700300 [DOI] [PubMed] [Google Scholar]
  • 50.Gutierrez-Diaz I, Fernandez-Navarro T, Salazar N, et al. Adherence to a Mediterranean Diet Influences the Fecal Metabolic Profile of Microbial-Derived Phenolics in a Spanish Cohort of Middle-Age and Older People. J Agric Food Chem. 2017;65(3):586–595. doi: 10.1021/acs.jafc.6b04408 [DOI] [PubMed] [Google Scholar]
  • 51.Mitsou EK, Kakali A, Antonopoulou S, et al. Adherence to the Mediterranean diet is associated with the gut microbiota pattern and gastrointestinal characteristics in an adult population. Br J Nutr. 2017;117(12):1645–1655. doi: 10.1017/S0007114517001593 [DOI] [PubMed] [Google Scholar]
  • 52.De Filippis F, Pellegrini N, Vannini L, et al. High-level adherence to a Mediterranean diet beneficially impacts the gut microbiota and associated metabolome. Gut. 2016;65(11):1812–1821. doi: 10.1136/gutjnl-2015-309957 [DOI] [PubMed] [Google Scholar]
  • 53.Mancabelli L, Milani C, Lugli GA, et al. Identification of universal gut microbial biomarkers of common human intestinal diseases by meta-analysis. FEMS Microbiol Ecol. 2017;93(12). doi: 10.1093/femsec/fix153 [DOI] [PubMed] [Google Scholar]
  • 54.Choi Y, Lee S, Kim S, et al. Vitamin E (alpha-tocopherol) consumption influences gut microbiota composition. Int J Food Sci Nutr. July 2019:1–5. doi: 10.1080/09637486.2019.1639637 [DOI] [PubMed] [Google Scholar]
  • 55.Surai PF, Kochish II, Romanov MN, Griffin DK. Nutritional modulation of the antioxidant capacities in poultry: the case of vitamin E. Poult Sci. 2019;98(9):4030–4041. doi: 10.3382/ps/pez072 [DOI] [PubMed] [Google Scholar]
  • 56.Belanche A, Kingston-Smith AH, Newbold CJ. An Integrated Multi-Omics Approach Reveals the Effects of Supplementing Grass or Grass Hay with Vitamin E on the Rumen Microbiome and Its Function. Front Microbiol. 2016;7:905. doi: 10.3389/fmicb.2016.00905 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Thompson MD, Cooney RV. The Potential Physiological Role of γ-Tocopherol in Human Health: A Qualitative Review. Nutr Cancer. 2020;72(5):808–825. doi: 10.1080/01635581.2019.1653472 [DOI] [PubMed] [Google Scholar]
  • 58.Deschasaux M, Bouter KE, Prodan A, et al. Depicting the composition of gut microbiota in a population with varied ethnic origins but shared geography. Nat Med. 2018;24(10):1526–1531. doi: 10.1038/s41591-018-0160-1 [DOI] [PubMed] [Google Scholar]
  • 59.Thursby E, Juge N. Introduction to the human gut microbiota. Biochem J. 2017;474(11):1823–1836. doi: 10.1042/BCJ20160510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Edgar RC. Updating the 97% identity threshold for 16S ribosomal RNA OTUs. Bioinformatics. 2018;34(14):2371–2375. doi: 10.1093/bioinformatics/bty113 [DOI] [PubMed] [Google Scholar]
  • 61.Prodan A, Tremaroli V, Brolin H, Zwinderman AH, Nieuwdorp M, Levin E. Comparing bioinformatic pipelines for microbial 16S rRNA amplicon sequencing. PLoS One. 2020;15(1):1–19. doi: 10.1371/journal.pone.0227434 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES