Gastroenterology 2022;162:1911–1932

A Systematic Review and Meta-analysis of Dietary Interventions

Modulating Gut Microbiota and Cardiometabolic
Diseases—Striving for New Standards in Microbiome Studies
Ilias Attaye,1,* Moritz V. Warmbrunn,1,* Aureline N. A. F. Boot,1 Suze C. van der Wolk,1
Barbara A. Hutten,2 Joost G. Daams,3 Hilde Herrema,1 and Max Nieuwdorp1

CLINICAL AT
1
Department of Internal and Vascular Medicine, Amsterdam University Medical Centers, location AMC, Amsterdam, the
Netherlands; 2Department of Epidemiology and Data Science, Amsterdam Cardiovascular Sciences, Amsterdam University
Medical Centers, location AMC, Amsterdam, the Netherlands; and 3Medical Library, Amsterdam University Medical Centers,
location AMC, Amsterdam, the Netherlands

This article has an accompanying continuing medical education activity, also eligible for MOC credit, on page e16. Learning
Objective: Upon completion of this CME activity, successful learners will be able to discuss dietary interventions that modulate gut
microbiota and cardiometabolic diseases.

See Covering the Cover synopsis on page 1787;

See commentary on page 1822. C ardiometabolic diseases (CMDs), such as metabolic
syndrome (MetSyn), type 2 diabetes (T2D), nonal-
coholic fatty liver disease (NAFLD), and atherosclerosis, are
part of a spectrum of diseases with shared causes and fea-
BACKGROUND & AIMS: Cardiometabolic diseases (CMDs) have tures.1 CMDs can be initiated by a Western lifestyle char-
shared properties and causes. Insulin resistance is a risk factor acterized by insufficient physical activity and high caloric
and characteristic of CMDs and has been suggested to be intake.2 This high caloric intake results in obesity and low-
modulated by plasma metabolites derived from gut microbiota grade inflammation, which often precedes MetSyn, T2D,
(GM). Because diet is among the most important modulators of NAFLD, and cardiovascular disease (CVD).3–5 Several factors
GM, we performed a systematic review of the literature to contributing to the progression of obesity to CMDs are
assess whether CMDs can be modulated via dietary in- known; for example, accumulation of ectopic fat, such as
terventions targeting the GM. METHODS: A systematic review
visceral adipose tissue with increased lipolysis, results in an
of the literature for clinical studies was performed on Ovid
increased fatty acid influx in muscle cells. This impairs
MEDLINE and Ovid Embase. Studies were assessed for risk of
mitochondrial function and ultimately insulin signaling,
bias and patterns of intervention effects. A meta-analysis with
contributing to insulin resistance.6 In addition to this,
random effects models was used to evaluate the effect of di-
etary interventions on clinical outcomes. RESULTS: Our search excessive amounts of white adipose tissue induce an in-
yielded 4444 unique articles, from which 15 randomized flammatory response in the body, marked by low-grade
controlled trials and 6 nonrandomized clinical trials were inflammation. This process is associated with the progres-
included. The overall risk of bias was high in all studies. In sion to T2D, NAFLD, and CVD.7,8 The development of CMDs
general, most dietary interventions changed the GM composi- thus is multifactorial, but shared risk factors and charac-
tion, but no consistent effect could be found. Results of the teristics of CMDs suggest pathophysiological similarities and
meta-analyses showed that only diastolic blood pressure is therefore potentially also overlap in treatment options.
decreased across interventions compared with controls (mean However, what factors drive the progression of insulin
difference:
3.63 mm Hg; 95% confidence interval,
7.09 resistance toward T2D and CVD remain to be elucidated.
to
0.17; I2 ¼ 0%, P ¼ .04) and that a high-fiber diet was Because a Western diet is a major risk factor for CMDs,
associated with reduced triglyceride levels (mean dietary aspects are important for cardiometabolic health.
difference:
0.69 mmol/L; 95% confidence interval,
1.36
to
0.02; I2 ¼ 59%, P ¼ .04). Other CMD parameters were not *Authors share co-first authorship.
affected. CONCLUSIONS: Dietary interventions modulate GM
composition, blood pressure, and circulating triglycerides. Abbreviations used in this paper: BCAA, branched-chain amino acids; CI,
confidence interval; CHD, coronary heart disease; CMD, cardiometabolic
However, current studies have a high methodological hetero- disease; CVD, cardiovascular disease; GM, gut microbiota; HDL-c, high-
geneity and risk of bias. Well-designed and controlled studies density lipoprotein cholesterol; MetSyn, metabolic syndrome; NAFLD,
nonalcoholic fatty liver disease; RoB, Cochrane Risk of Bias tool; rRNA,
are thus necessary to better understand the complex interac- ribosomal RNA; RoB, Cochrane Risk of Bias; SCFA, short-chain fatty
tion between diet, microbiome, and CMDs. PROSPERO: acids; T2D, type 2 diabetes.
CRD42020188405 Most current article
© 2022 The Author(s). Published by Elsevier Inc. on behalf of the AGA
Keywords: Cardiometabolic Diseases; Dietary Intervention; Gut Institute. This is an open access article under the CC BY-NC-ND license
Microbiome; Metabolic Syndrome; Type 2 Diabetes. (http://creativecommons.org/licenses/by-nc-nd/4.0/).
0016-5085
https://doi.org/10.1053/j.gastro.2022.02.011
 1912 Attaye et al Gastroenterology Vol. 162, No. 7

WHAT YOU NEED TO KNOW Methods

BACKGROUND AND CONTEXT Inclusion Criteria and Search Strategy
Studies were eligible to be included in this systematic review
Cardiometabolic diseases have similar characteristics as
if they were performed as a randomized or nonrandomized
well as causes and are associated to gut microbiota.
However, the role of dietary microbiome interventions to clinical trial and measured a component of CMD (ie, MetSyn,
treat cardiometabolic diseases is still unclear. T2D, NAFLD, atherosclerosis). Moreover, a dietary intervention,
defined as a change in dietary macronutrient composition or
NEW FINDINGS dietary pattern, had to be performed in the clinical trial as well as
Dietary microbiome studies can improve cardiometabolic a form of microbiome analysis, such as 16S ribosomal (r)RNA
diseases, but no consistent effect could be observed in sequencing or shotgun metagenomic sequencing. Studies were
the gut microbiota. This may be due to the high risk of excluded if the intervention contained prebiotics, probiotics, or
CLINICAL AT

bias by selection of results in the available studies. synbiotics, if the participants were aged <18 years, and if the
LIMITATIONS participants had other conditions that could influence CMD
(Supplementary Document 1). Studies were excluded if they
Included studies were heterogeneous in medication use,
were not published in English, Dutch, or German.
disease progression, antibiotic use, and dietary
interventions, which limits the generalizability of the The systematic literature search was performed on July 6,
presented findings. The duration of interventions also 2020, by an experienced librarian using a predefined protocol
differed between studies, which may confound gut according to the Preferred Reporting Items for Systematic Re-
microbiota findings. views and Meta-Analyses (PRISMA) guidelines26
(Supplementary Document 1). The search strategy can be
IMPACT
summarized as follows: ([gut microbiota] and [dietary in-
This study shows that dietary microbiome interventions terventions] and [cardiometabolic diseases] or [relevant tri-
can be effective to improve cardiometabolic disease als]). Animal studies were excluded. This strategy was adapted
parameters. However, the observed high risk of for Ovid MEDLINE and Ovid Embase. See Appendix
methodologic bias calls for new standards in the (Supplementary Document 2) for full search details. The pro-
microbiome field.
gram Rayyan QCRI27 was used as a screening tool. This study
was registered in the International Prospective Register of
For example, a high salt intake promotes inflammation and Systematic Reviews (PROSPERO) on July 5, 2020, under code:
hypertension, and a high caloric intake (eg, high in sugars, CRD42020188405. Patients were not involved in drafting of the
such as fructose, or saturated fat, or both) is related to in- study design. All authors had access to the study data and
sulin resistance and CVD.9–13 However, diet can also reviewed and approved the final manuscript.
improve CMD, as the Mediterranean diet can improve car-
diovascular health as well as prevent and treat diabetes.14,15 Risk of Bias Assessment
In addition to this, studies suggest that a high intake of fi- The risk of bias was assessed by 2 independent authors
bers, which are degraded by gut microbiota (GM) to short- using the Cochrane Risk of Bias (RoB) tool for randomized
chain fatty acids (SCFA), is associated with insulin studies28 and the Newcastle-Ottawa Scale for nonrandomized
sensitivity.16,17 studies.29 For nonrandomized studies, a Newcastle-Ottawa
The GM has also been shown to influence host meta- Scale score of 7 indicates a good methodological quality.30
bolism. Feces transplantation from lean donors to in- Disagreements were resolved via discussion.
dividuals with MetSyn improved insulin sensitivity and
increased the abundance of bacteria that produce the SCFA Data Extraction
butyrate.18 In fact, microbiota-derived metabolites have From the 4444 articles, 21 met the inclusion and exclusion
been shown to be associated with properties of CMDs.19 criteria (Figure 1).31 The following data were extracted from all
For example, microbially produced imidazole propionate articles by 2 independent authors: study name, country of
impairs insulin signaling in mice, and production of this study, study design, type of CMD, the dietary intervention,
metabolite was linked to dietary histidine intake.20 In study duration, method of dietary compliance measurement,
addition, the microbiota-derived metabolite trimethylamine study population, baseline characteristics, assessment time
N-oxide from dietary choline and phosphatidylcholine is points, microbiota sequencing technique, main microbiota
associated with CVD and NAFLD in T2D.21–23 This suggests outcome, and main clinical outcome (Table 1). Disagreements
that diet indirectly modulates clinical parameters of CMDs were resolved via discussion. An extended version of Table 1 is
and the risk of developing CMDs through the GM. How- added as Supplementary Document 3.
ever, dietary interventions that modulate the GM to
improve CMDs show conflicting results thus far.24,25 To get Meta-analysis
an overview on the current level of knowledge concerning Unique randomized studies with an intervention and a con-
dietary interventions that modulate host metabolism trol arm that reported absolute data were pooled to compare the
through the GM, we performed a systematic review and effect of dietary interventions. We calculated the mean difference
meta-analysis on the current literature to evaluate the ef- (95% confidence interval [CI]) between the mean change of the
fect of dietary interventions on the GM and CMD-related outcome measures for fasting glucose, fasting insulin, homeo-
outcomes. stasis model assessment-insulin resistance, body weight, body
June 2022 New Standards for Dietary GM Studies in CMD 1913

CLINICAL AT
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses PRISMA flowchart31 describes the pipeline of
the selection process of included studies.
mass index, low-density lipoprotein-cholesterol, high-density Overall, all clinical trials had a high risk of bias
lipoprotein-cholesterol (HDL-c), triglycerides, and systolic (Figure 2). The main explanation for this high risk of bias
blood and diastolic blood pressure in the dietary intervention was that all studies scored a “high risk” in domain 5: “Se-
group and in the control group if absolute values were available. lection of the Reported Outcomes.” Studies scored high in
Then, the pooled mean difference was calculated by using the domain 5 because none of the included studies registered a
weighted sum of these differences, with the weight being the detailed prespecified analysis plan for their microbiome
reciprocal of the combined variance for each study. Heteroge- analyses (RoB 5.1), which makes it likely that results have
neity of the included studies was estimated using the I2 statistic. been selected on the basis of multiple eligible analyses of
All analyses were performed using a random-effects model. A Z the data (RoB 5.3). For detailed descriptions of other do-
test was performed to test the overall effect. The analyses were
mains see Supplementary Document 4.
performed using Review Manager 5.4.0 (The Cochrane Collabo-
Nonrandomized studies scored poorly on selection
ration, http://www.cochrane.org).
criteria of the study participants, controlling for con-
founders, and having an adequate follow-up period
Results (Table 2). No nonrandomized study scored above the
A search of MEDLINE and Embase databases found 4444 threshold for “good” methodological quality of 7 points.
articles after removal of duplicates. During the screening of
abstract and title, 4357 articles were excluded (Figure 1), Metabolic Syndrome
adapted from Moher et al.31 The most important reasons for We identified 8 studies that fulfilled all criteria. Of these
exclusion were wrong publication type (eg, narrative re- 8 studies, 6 were randomized controlled trials,32–37 1 was a
views), no clinical trial, and no dietary intervention or nonrandomized controlled trial,38 and 1 was a crossover
change in macronutrients. Then, 87 full-text articles were randomized controlled trial.39 The study populations were
screened for eligibility, and 66 publications were excluded, relatively small: the largest included 51 participants,36 and
leaving 21 human studies that met all inclusion criteria. the smallest study included only 9 participants.32 The di-
etary interventions varied greatly, but mainly focused on
Quality of Included Studies carbohydrates and fiber intake.33,36,39 Also, 2 of the 8 diets
The methodological quality and risk of bias of the 21 were energy-reduced diets.32,35 The duration of the studies
included studies was assessed according to the Cochrane varied from 4 weeks39 to 2 years.33
RoB 2 for Randomized trials28 and the Newcastle-Ottawa Only 1 study investigated GM by using shotgun
Scale for nonrandomized studies.29 sequencing.32 All other studies focused on 16S rRNA
 CLINICAL AT

Table 1.Characteristics of the Included Studies

1914 Attaye et al
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Balfegó et al RCT T2D Standard T2D diet 6 months; Week 0, n ¼ 19, female: [1] BMI baseline: 30.5 ± No statistical differences in [1] Changes in fasting
(2016)43 enriched with 100 g month 6 57.9% [1]; n ¼ 4.36 kg/m2; 6 months: the abundance of the glucose, HbA1c,
of sardines 5 days 16, female: 50% 30.0 ± 4.36 kg/m2; bacterial groups fasting insulin, and
a week instead of [2] Age: 60.0 ± 7.4 years. analyzed between the HOMA-IR were
protein foods intake [2] BMI baseline: 28.8 groups. Phylum similar and not
(sardine group: SG), ± 3.2 kg/m2; 6 Firmicutes were statistically different
[1] T2D standard diet months: 28.5 ± 3.2 decreased in both between 2 diet
without sardines [2] kg/m2; Age: 61.2 ± groups after 6 months, groups. Only [2]
9.6 years; and the Firmicutes/ decreased
Comorbidities:
; Bacteroidetes ratio significantly HbA1c
Medication use: none decreased in [1]. values compared
Escherichia coli with baseline. Both
concentrations were dietary interventions
higher at the end of the decreased fasting
study in both [1] and [2]. insulin and HOMA-
Proportions of IR [1]: insulin,
Bacteroides-Prevotella HOMA-IR, [2]:
increased in the SG, insulin, HOMA-IR.
while there was a Adiponectin
decreasing trend in the increased
Firmicutes/ significantly in [1]
Bacteroidetes ratio in group from at
the control group baseline to at 6
months.
Inflammatory
markers: no
differences between
intervention groups
(TNF-a increased
significantly, and
IL10 did not show
any significant

Gastroenterology Vol. 162, No. 7

change)
Table 1. Continued

June 2022
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Frost et al Non-RCT T2D [1] 3-phase (P) weight 15-week diet with 3 N ¼ 12, female: [1] Age: 57.2 years, no SD Collinsella was the only [1] Led to weight loss
(2019)44 loss program. P-1: 6 phases; week 0, 67% reported; mean BMI genera that had an and improved
weeks’ food week 6, and of all subjects at altered (lower) glycemic index in all
substitution by low- week 15 baseline: 39.6 kg/m2; abundance after P-3 subjects
calorie formula diet Medication use: þ (15 weeks)
providing 800 kcal/d.
P-2: 4 weeks’
reintroduction of
normal diet with 1200
kcal/d. P-3: 5 weeks
with 1200–1500 kcal/
d intake to stabilize
weight. Furthermore,
subjects received a
structured physical
training course
Huang et al Non-RCT T2D Okinawan-based Nordic 12-week N ¼ 30, female: BMI: 29.9 ± 4.1 kg/m2; [1] Resulted in no changes [1] Weight loss and
(2018)46 diet [1]. High intervention, 16- 57% Age: 57.5 ± 8.2 years; in Enterobacteriaceae lower BMI and waist
proportion of week follow-up T2D duration: 10.4 ± abundance, microbial circumference
vegetables and (28 weeks in 7.6 years; Medication diversity, or SCFAs.
legumes, a moderate total); Week 0, use: þ Metformin resulted in
intake of fish products 12, 28 higher counts of
and alcohol, a low Enterobacteriaceae
consumption of dairy compared with the
and meat products group without

New Standards for Dietary GM Studies in CMD 1915

etc. No control metformin treatment.
groups Metformin did not affect
microbiota diversity
Karusheva Crossover RCT T2D [1] BCAA-rich diet 4 weeks; Time N ¼ 12, female: Age: 54 ± 4 years; BMI: Lower abundance of Clamp hepatic and
et al followed by BCAA- point 0, week 33.3% 30.8 ± 2.8 kg/m2; Firmicutes after BCAA whole-body insulin
(2019)42 reduced diet and [2] 4, week 8 HbA1c: 49 ± 10 mmol/ reduction [2] and higher sensitivity did not
BCAA-reduced diet mol; Medication Bacteroidetes after change; oral
followed by BCAA- use: þ both interventions [1, 2] glucose sensitivity
rich diet index was higher
after BCAA-reduced
diet, whereas meal
insulin secretion
was lower

CLINICAL AT
 CLINICAL AT

Table 1. Continued

1916 Attaye et al
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome
Ohlsson et al Non-RCT T2D Okinawan-based Nordic 12 weeks diet, N ¼ 30, female: [1] Age: 57.5 ± 8.2 years; [1] Decreased serum [1] Improved glucose
(2019)45 diet, with lower followed by 16 57% at 12 BMI: 29.9 (± 4.1); isovaleric acid after 12 and lipid
carbohydrate but weeks follow-up. weeks and N ¼ Medication use: NR weeks and butyrate metabolism and
higher fiber, fat, and Total duration 28 23 at 28 weeks after 29 weeks. The diet increased satiety,
protein energy weeks; Baseline, did not affect microbial while decreasing
content [1] week 12 and diversity or sweet cravings
week 28 Enterobacteriaceae
levels
Medina-Vera RCT T2D 15-day reduced-energy 3 Months; 5 study n ¼ 28 [1], female: [1] Age: 50.4 ± 8.7 years; [1] Increased alpha [1] Reduced AUC for
et al (2019)40 diet (
500 kcal/ visits. OGTT and 68% and n ¼ 25 BMI 30.7 ± 4 kg/m2; diversity and increased glucose,
d compared with fecal sample [2] female: Medication use: þ, all levels of F prausnitzii triglycerides, and
usual diet). After 15 collection at 54.2% on metformin. [2] Age: and A muciniphila while total cholesterol and
days, subjects baseline and 49.8 ± 10.6 years; reducing levels of P reduced levels of
continued their after 3 months BMI 31.5 ± 6 kg/m2; copri. [1] Also CRP, HbA1c, and
energy-deficit diet Medication use: þ, stimulated Bif longum FFAs. [1] and [2]
with a dietary baseline difference and increased Bac reduced LPS levels
portfolio, constituting between group [1] fragilis
nopal, chia seeds, soy and [2] with regards to
protein and inulin [1], antidiabetic drugs
or placebo, consisting (more in [2])
of calcium caseinate
and maltodextrin [2]
Zhao et al RCT T2D Acarbose þ WTP diet 84 days; Day 0, 28, n ¼ 27 [1], n ¼ 16 Age: 58.4 ± 32.2 years [1] Butyric acid concentrations [1] Increased
(2018)41 (high-fiber diet 56, 84 [2]. Overall and 59.7 ± 24 years increased significantly postprandial GLP-1
composed of whole female: 58% [2]; BMI not known; after diet [1]. Acetic acid levels. HbA1c
grains, traditional Medication: did not increase. [1] reduction in both
Chinese medicinal everybody on Increased fecal acetate, groups after 28
foods, and prebiotics) acarbose butyrate weeks; However, at
[1] & acarbose þ concentrations. The the end of the
usual diet T2D acetate-producing Bif intervention, there
guidelines for T2D [2] pseudocatenulatum was a greater
was one of the most reduction of HbA1c

Gastroenterology Vol. 162, No. 7

significantly promoted in the diet group [1].
SCFA producers Adequate glycemic
control (HbA1c
<7%) was
significantly higher
in [1] (89% vs 50%
in [2])
Table 1. Continued

June 2022
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Haro et al RCT Obese Mediterranean diet [1] & 1 year; Week 0 and N ¼ 20, female: 0% [1] Age: 62.2 ± 14.31 [1] Increased the [1] & [2] significantly
(2016)52 population low-fat, high-complex 1 year years; BMI: 32.8 ± abundance of increased insulin
with CHD carbohydrate [2] 2.24 kg/m2; Roseburia genus. [2] sensitivity index
Medication use: þ. [2] Decreased the based on an OGTT
Age: 61.4 ± 11.63 abundance of the
years; BMI: 31.6 ± Roseburia genus and
3.58 kg/m2; increased F. prausnitzii.
Medication use: þ Also, fecal metabolome
changes are correlated
to the changes in the
microbiota induced in
[1] and [2]
Chen et al RCT NAFLD with 220 g/d yogurt [1] & 220 24 weeks; Week n ¼ 48, female: Age: 33-66 years; BMI: [1] Decreased the [1] Decreased fat mass,
(2019)49 MetSyn and g/d milk [2] 0 and Week 24 100% [1] and [1] 31.81 ± 2.76 kg/m2 abundance of HOMA-IR; fasting
no T2D n ¼ 44, female: & [2] 32.18 ± 3.18 kg/ Firmicutes phylum, insulin; 2-h insulin
100% [2]; n ¼ 20 m2; Medication use:
Clostridiales order and and 2-h AUC for
[1] & n ¼ 20 [2] Blautia and insulin compared
for MRI analysis Eubacterium with [2]
and fecal ventriosum genera. [1]
microbiota Also decreased
analysis Erysipelotrichaceae
family and
Ruminococcus genus
and increased
Phascolarctobacterium

New Standards for Dietary GM Studies in CMD 1917

genus with a decrease
in Pseudobutyrivibrio
and Dialister genera

CLINICAL AT
 CLINICAL AT

Table 1. Continued

1918 Attaye et al
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Jian et al RCT NAFLD (12 of 38) Overfeeding diet of 1000 3 weeks; Week 0, 3, N ¼ 38 overweight Age: 48 ± 2 years; BMI: At baseline the Firmicutes- Increased liver fat in the
(2021)47 kcal/d on top of 12 and obese 31 ± 1 kg/2; to-Bacteroidetes ratio SAT [I1] group
habitual diet rich in subjects, of Comorbidities: was significantly higher compared with
saturated fat (SAT, which 12 had overweight/ obesity; in the subjects with UNSAT and CARB.
59E% fat) [I1], or NAFLD (>5.56% Medication:
NAFLD and positively Also, insulin
unsaturated fat liver fat, female: associated with liver fat. resistance, activity
(UNSAT, 60E% fat) 50%). [I1] n ¼ SAT [I1] increased of liver enzymes,
[I2], or simple sugars 14; n ¼ 12 [I2]; Proteobacteria, while conc. plasma HDL
(CARB, 24E% fat) [I3] n ¼ 12 [I3] UNSAT [I2] increased and LDL and LBP-
butyrate producers. to-CD14 ratio
CARB [I3] increased increased
Lactococcus and significantly in the
Escherichia coli. SAT group liver
Especially Bilophila was enzymes, fasting
associated to glucose, free fatty
overfeeding induced acids, insulin, lipids
liver fat
de Faria Ghetti RCT NAFLD Diet intervention (47.4% 12 weeks; 0 and 12 n ¼ 20, female: 40% [1] Age: 48.3 ± 10.3 Small-intestinal bacterial Diet group: Reduction
et al (2019)48 carbohydrates, weeks [1]. n ¼ 20, years; Hypertension: overgrowth frequency in body weight,
27.7% lipids, 24.9% female: 55% [2] 35%; T2D: 10%. [2] was 30% in the diet BMI, waist
proteins, 30.3 g Age: 50.6 ± 10.3 group and 45% in the circumference,
fibers) þ nutritional years; Hypertension: control group. An percentage of fat,
orientation [1] and 50%; T2D: 30% increase in density of serum aspartate
only nutritional total microorganisms in aminotransferase,
orientation [2] diet group was alanine
detected, while in the aminotransferase,
control group reduced Î3-glutamyltrans-
Bacteroidetes and ferase, glycemia,
Verrucomicrobiales HOMA-IR, total
were observed cholesterol, and
triacylglycerols. No
differences in

Gastroenterology Vol. 162, No. 7

control group
Table 1. Continued

June 2022
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Pataky et al Prospective NAFLD Hypocaloric hyperproteic 3 weeks; Week n¼ 15 NAFLD (NAS Overall age: median 50 [1] Changed lower-level [1] Decreased BMI by
(2016)50 experimental diet (HHD) intake 0 and 3 3 (2–5); n ¼ 15, (range, 47–55) years; taxi belonging to 3.6%, liver fat by
single-arm 1059 kcal/d, 125.5 female: 26.7% BMI: 34.6 (32.4, 36.7) Clostridiales order: 65%, and CRP by
study protein g/d, 42.3 lipids kg/m2; Comorbidities: decreased Lachnospira, 19%. [1] Decreased
g/d, 38.0 overweight/obesity; increased Blautia, GGT, glucose,
carbohydrates g/d, Medication:
Butyricicoccus and triglycerides, total
20.1 sugar g/d, 3.1 changes in 10 OTUs. cholesterol, and
fructose g/d, 4.6 HDD also led to adipose tissue index
polyols g/d, 1.4 starch changes in
g/d, 24.8 fibers g/d [1] Bacteroidales order,
with decreases in 10
OTU’s of Bacteroides.
No changes in gut
microbiota function.
Haro et al RCT MetSyn with Mediterranean diet: 35% 2 years; Week 0 and N ¼ 239 (n ¼ 138 MetSyn Age: 60.16 ± [I1] Increases the No differences found
(2016)51 CHD fat (22% 2 years MetSyn, female: 8.69 years; Glucose: abundance of between the
monounsaturated, 20% and n ¼ 126.47 ± 61.09 mg/ P. distasonis, B. participants
6% polyunsaturated, 101 without dL. Non-MetSyn Age: thetaiotaomicron, F. assigned to [I1] or
and 7% saturated) [I1] MetSyn as 61.41 ± 94.45 years; prausnitzii, B. [I2] diet or MetSyn
or low-fat high- control group, Glucose: 102.27 ± adolescentis, and or non-MetSyn
carbohydrate diet: female: 14% 32.46 mg/dL; B. longum. There was a
28% fat (12% Medication:
negative correlation
monounsaturated, between the
8% polyunsaturated, abundance of above
and 8% saturated) [I2] species and waist

New Standards for Dietary GM Studies in CMD 1919

circumference and for
some also with
triglyceride plasma
levels.

CLINICAL AT
 CLINICAL AT

Table 1. Continued

1920 Attaye et al
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Santos-Marcos RCT MetSyn with Male vs female. 3 years; Week 0 and N ¼ 246, patients MetSyn women: Age Increased abundance of Positive relation with
et al (2019)64 CHD Mediterranean diet: 3 years with and without 62.95 ± 8.89 years. Desulfovibrio, Acidaminococcus
35% fat (22% MetSyn in men BMI MetSyn women: Roseburia, and genus and waist
monounsaturated, and women: 1. 32.18 ± 4.09 kg/m2. Holdemania genera in circumference in
6% polyunsaturated, MetSyn-Women MetSyn men: Age MetSyn-Men compared men with MetSyn,
and 7% saturated) [I1] (n ¼ 79), 2. 61.63 ± 8.80 years. with MetSyn-Women and for
or low-fat high- MetSyn-Men BMI 31.83 ± 4.53 kg/ after the low-fat [I1] Desulfovibrio genus
carbohydrate diet: (n ¼ 79) 3. m2. Groups were all diet. No differences in a negative
28% fat (12% NonMetSyn- matched by main the abundance of these relationship with
monounsaturated, Women (n ¼ 44), metabolic variables. bacterial genera HDL-c in men with
8% polyunsaturated, and 4. Comorbidities: CHD between sexes were MetSyn. Positive
and 8% saturated) [I2] NonMetSyn- with an event 6 observed after [I1] diet, relationship for
Men (n ¼ 44), months before but there were higher genus for the S24-4
overall female: enrolling with levels on an unknown family, plasma
50% conventional genus from the triglyceride levels,
treatment. Rikenellaceae family in insulin levels and
Medication:
MetSyn-Men compared Acidaminococcus
with MetSyn-Women genus a genus
after [I1] diet, whereas of the
no differences were Ruminococcaceae
observed after [I2] diet family in women
with MetSyn
Lappi et al RCT MetSyn Low-fiber (4%) wheat 12 weeks; Week N ¼ 51 (n¼ 24 Age: 60 ± 6 years; In both groups no changes No differences in
(2013)36 bread [1] & high- 0 and Week 12 [1] & n¼ 27 [2]), BMI: 31 ± 4 kg/m2; in microbiota clustering. glucose levels and
fiber (7-15%) rye female: 51% Medication: 19% decrease of inflammation
bread [2] antihypertensives in Bacteroidetes in parallel markers (CRP).
53% to a 53% decrease in
the alkylrecorsinol
concentration and
increase of bacteria
related to B vulgatus, B
plebeius, and Prevotella

Gastroenterology Vol. 162, No. 7

tannerae decreased,
whereas that of bacteria
related to Collinsella
and members of the
Clostridium clusters IV
(17%) and XI increased
[1]. No changes in
abundance [2].
Table 1. Continued

June 2022
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Haro et al RCT MetSyn with Mediterranean diet [1] & 2 years; Week 0 and n ¼ 33, female: 0%, MetSyn-OB: Age: 59.03 MetSyn-OB group had a Both diets reduced
(2017)33 CHD low-fat, high-complex 2 years (MetSyn-OB), ± 10.51 years; BMI: marked gut microbial triglycerides levels
carbohydrate diet [2] n ¼ 32, female: 32.42 ± 4.37 kg/m2; dysbiosis compared to in the MetSyn-OB
in MetSyn obese 0%, (Non- Medication useþ. Non-MetSyn-OB and group, whereas no
(MetSyn-OB), non- MetSyn-OB) and Non-MetSyn-OB: Non-MetSyn-Non-OB. effects were seen in
MetSyn Obese (non- n ¼ 41, female: Age: 63.72 ± 9.62 This dysbiosis was the Non-MetSyn-
MetSyn-OB) and 0% (Non- years; BMI: 32.88 ± reversed by both [1] OB and Non-
Non-MetSyn non- MetSyn-Non- 3.17 kg.m2; and [2]. No microbiota MetSyn-Non-OB
obese (Non-MetSyn- OB) Medication use: þ. changes were observed groups
Non-OB) subjects Non-MetSyn-Non- in the Non-MetSyn-OB
OB: Age: 67.73 ± 8.90 and Non-MetSyn-Non-
years; BMI: 27.05 ± OB groups. The
1.73 kg/m2; Bacteroides and
Medication use: þ Prevotella genera
increased, reducing the
Bacteroides/Firmicutes
ratio
Louis et al Cohort from RCT MetSyn and Optifast 52 Nestlé 1-year intervention, n ¼ 9 in persistent Age and initial BMI PF [1] Resulted in weight Subjects with MetSyn
(2016)32 NAFLD Program. 52-week followed by 1 success (PS) and NS group: NR reduction during the had higher
lifestyle intervention year of follow- group) and n ¼ 7 Overall study group first 3 months in PS and Firmicutes/
with 12 weeks very- ups; Week 0, 3, in no persistent BMI: 43.1 (95% CI NS. The weight loss Bacteroides ratio at
low caloric diet (800 6, 12, 18 and 24 success (NS) 39.5-46.8) kg/m2; was accompanied with baseline compared
kcal/d intake), months group. 9/16 Medication use: NR improved HOMA-IR with obese non-
followed by MetSyn and 13/ and lower fatty liver MetSyn subjects. [1]

New Standards for Dietary GM Studies in CMD 1921

reintroduction phase 16; NAFLD, sex index Led to lower
of normal food intake not shown Roseburia levels in
during months 3-6 the weight loss
and stable weight phase. [1] Led to an
maintenance phase increase in
during month 7-12 [1] Akkermansia
throughout the
intervention.
Furthermore,
Akkermansia is the
strongest indicator
between PS and
NS. NAFLD subjects
had lower
Subdoligranulum
levels

CLINICAL AT
 CLINICAL AT

Table 1. Continued

1922 Attaye et al
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome

Bellikci-Koyu RCT MetSyn Kefir (180 mL/d) diet [1] & 12 Weeks; Week 0, n ¼ 12, female: 83% [1] Age: 52 (47.50-60.50) [1] Significantly increased [1] Improved HOMA-IR,
et al (2019)34 control group week 4, 8 and 12 [1] and n ¼ 10, years; BMI: 30.67 Actinobacteria, which insulin, and blood
received unfermented female: 60% [2] (26.94-34.66) kg/m2; was also seen by [2]. pressure. But was
milk (180 mL/d) [2] Medication use: NR, No changes in alpha or not significant when
but no antidiabetic or beta diversity after [1] & compared with
lipid-lowering drugs. [2] control group [2]. [2]
[2] Age: 53 (45-60) Did not significantly
years; BMI: 32.38 affect insulin or
(29.18-34.59) kg/m2; HOMA-IR and only
Medication use: NR lowered systolic
but no antidiabetic or blood pressure
lipid-lowering drugs. significantly
Data reported as
median (IQR).
Clark et al Non-RCT MetSyn Diet of 50% fruit and 9 weeks; Week N ¼ 17, female: Age: 22.2 ± 3.4 years; [1] Deceased No significant changes
(2018)38 vegetable intake [1] 0 and week 9 64.7% BMI: 37.95 ± 5.04 kg/ Erysipelotrichaceae at in MetSyn
m2; Medication use: the family levels and parameters after the
NR increased dietary intervention.
Caulobacteraceae The diet did induce
an increase in LDL
cholesterol and
sodium

Gastroenterology Vol. 162, No. 7

Table 1. Continued

June 2022
Study duration
and assessment Baseline Main microbiota Main clinical
Study Study design CMD Dietary intervention points Study population characteristics outcome outcome
Guevara-Cruz RCT MetSyn Lifestyle intervention with Lifestyle n ¼ 11 NW, n ¼ 18 [1] NR, Age: NR; BMI: [1] Increased Prevotella, [1] NW group:
et al (2019)35 an energy reduced intervention for 2 MetSyn and n ¼ 22.4 ± 0.83 kg/m2; Faecalibaterium, Decreased fasting
(
2092 J/d deficit) weeks, 11 OCIII-MetSyn Medication use:
. Ruminococcus, and glucose; in MetSyn:
and low-saturated fat afterwards [1] and n ¼ 10 MetSyn: Sex: NR; Roseburia in NW Decreased systolic
diet (LSFD) for 2 randomization to NW, n ¼ 17 Age; NR; BMI: 34.3 ± compared with [2]. [1] and diastolic blood
weeks, followed by a [1] lifestyle MetSyn and n ¼ 3.78 kg/m2; Increased Bacteroides pressure and in
2-month period with a intervention or 9 OCIII-MetSyn Medication use:
. and Faecalibacterium, OCIII-MetSyn:
LSFD and functional [2] placebo for 2 [2]. Sex: NR OCIII-MetSyn: Sex: but decreased Decreased AUC of
food diet, consisting months in NR; Age: NR; BMI: Prevotella and insulin compared
of a mixture of normal weight 43.0 ±3.35 kg/m2; Roseburia in MetSyn with [2]
dehydrated nopal, (NW), MetSyn Medication use:
. [2] compared with [2]. [1]
chia seeds, oats, and class III NW: Sex: NR; Age: Increased Bacteroides,
soybean protein, obesity- MetSyn NR; BMI: 23.3 ±1.71 Faecalibacterium, and
inulin, sweetener, and (OCIII-Met-S); kg/m2; Medication Oscillospira, but
flavoring Week 0, 2 15 use:
. MetSyn: Sex: decreased Prevotella in
days, 45 days NR; Age: NR; BMI: OCIII-MetSyn group
and 75 days 32.3 ± 3.13 kg/m2; compared with [2]. [1]
Medication use:
. Decreased Prevotella/
OCIII-MetSyn: Sex: Bacteroides ratio in
NR; Age: NR; BMI: MetSyn and OCIII-
44.0 ±3.42 kg/m2; MetSyn group
Medication use:
compared with [2]. [1]
Increased Akkermansia
muciniphila in NW
group, Bacteroides

New Standards for Dietary GM Studies in CMD 1923

ovatus and F prausnitzii
in MetSyn group and A
muciniphila and F
prausnitzii in
OCIIIþMetSyn group
Hald et al RCT, crossover MetSyn Isocaloric healthy 4 weeks, with 4 n ¼ 9 [1] and n ¼ 10 Age: 60 (35-75) years; [1] Decreased bacterial No changes in body
(2016)39 design carbohydrate diet, weeks washout; [2], sex: NR BMI: 30.6 (25.9-41) species richness and weight, body fat, or
enriched with week 0 and kg/m2; Medication increased inter-and waist circumference
arabinoxylan and week 4 use: þ. Data reported intraindividual variation regardless of diet
resistant starch [1] as median (IQR) of the intestinal
and low-fiber microbiota
Western-style diet [2]

AUC, area under the curve; BMI, body mass index; CD14, cluster of differentiation14; CRP, C-reactive protein; FFAs, free fatty acids; GGT, g-glutamyltransferase; GLP-1, glucagon-
like peptide 1; HbA1c, hemoglobin A1c; HOMA-IR, homeostatic model assessment for insulin resistance; IL, interleukin; IQR, interquartile range; LBP, Lipopolysaccharide-binding
protein; LDL, low-density lipoprotein; LPS, lipopolysaccharide; MRI, magnetic resonance imaging; NAS, NAFLD Activity Score; NR, not reported; OGTT, oral glucose tolerance
test; OTU, operational taxonomic unit; RCT, randomized clinical trial; SD, standard deviation; TNF-a, tumor necrosis factor-a.

CLINICAL AT
 1924 Attaye et al Gastroenterology Vol. 162, No. 7
CLINICAL AT

Figure 2. Quality assessment of included studies. Risk of bias assessment according to Cochrane Checklist for Randomized
studies.26 Low, low risk of bias; Some, some arguments for risk of bias; High, high risk of bias.

sequencing, with differences in the variable region that was did not report the method used.42 The dietary interventions
being sequenced. The diets significantly impacted GM mainly focused on SCFA and SCFA-producing strains,
composition and alpha diversity in all studies, except 1 whereas other metabolites were not studied, with the
study, which compared the effect of milk against kefir con- exception of 1 study.42 The dietary interventions had posi-
sumption in participants with MetSyn.34 tive effects on glycemic indices in 4 studies40,41,44,45 and no
The diets had little to no effects on clinical outcomes, significant effect in 2 studies, despite changes in GM
with only 3 of 8 diets reporting improved glucose homeo- composition and function.42,43
stasis.34,35,37 Other components of the MetSyn, such as Karusheva et al42 was the only research group that per-
dyslipidemia, poor blood pressure regulation, and abdom- formed a crossover study in this group. The authors studied a
inal fat, were altered in a few studies,32,33,35,37 while others branched-chain amino acid (BCAA)-enriched diet against a
reported no significant changes. Of particular importance BCAA-poor diet for 4 weeks in 12 individuals with T2D. This
was the study of Hald et al,39 which was the only study with study found that both diets introduced similar changes in the
a crossover design. Their study evaluated a diet enriched in GM composition, which was not accompanied by improved
healthy carbohydrates and starch during 4 weeks compared insulin resistance, as measured via the gold standard
with a low-fiber Western-style diet. The authors concluded hyperinsulinemic clamp methodology. However, a reduced
that a healthy carbohydrates and starch diet led to increased BCAA diet did improve the oral glucose sensitivity index.
production of fecal SCFA and an increased intraindividual
and interindividual variation of the GM composition. How-
Nonalcoholic Fatty Liver Disease
ever, the authors could not relate these findings to improved
We found 5 studies that included individuals with
clinical outcomes.
NAFLD. Most studies (4 of 5) were randomized controlled
clinical trials,32,47–49 and 1 was a single-arm prospective
Type 2 Diabetes intervention trial.50 The study population varied between
Seven of the included studies focused on T2D. Most of 13 and 44 individuals with NAFLD. The duration also varied
these studies (4 of 7) consisted of randomized controlled widely between studies, ranging from only 3 weeks to 1
trials,40–43 with 1 study having a crossover design.42 The year of intervention with a 1-year follow -up period. Most
study population varied between 1242 and 4341 individuals studies focused on altered caloric and fat intake: a hypo-
with T2D. The studies lasted 4 weeks42 to 6 months.43 The caloric diet was investigated in 2 studies and a hypercaloric
diets mainly focused on fiber and protein intake, with 2 diet in 1 study, with different macronutrient additions.
studies focusing on reduced caloric intake in general.40,44 The 2 studies assessing hypocaloric diets32,50 used 16S
The included studies focused on different methods to rRNA and shotgun sequencing to assess the taxonomic di-
sequence the GM. Two studies used 16S rRNA versity. The first study found decreased fecal levels of Lach-
sequencing,40,44 1 used shotgun sequencing,41 1 used ter- nospira and increased Blautia and Butyricicoccus.50 The
minal restriction fragment length polymorphism,45 and 2 second study found lower fecal levels of Subdoligranulum and
used quantitative polymerase chain reaction.44,46 One study higher abundance of Lactococcus and Paraprevotella, with no
 June 2022 New Standards for Dietary GM Studies in CMD 1925

Adequate Total score

changes in metabolic pathways, except an increase of the

of outcome Follow-up follow-up of 9 points

Score
naphthalene degradation pathway in individuals with

3/9
4/9
4/9
3/9
5/9
2/9
NAFLD.32
Another randomized controlled trial assessed the effect
of overfeeding by adding 1000 kcal/d on top of the
habitual diet using different macronutrients (saturated
fats, unsaturated fats, or sugars).47 This study found that
0
0
0
1
1
0
Outcome (max 3 stars)

NAFLD patients at baseline already have higher Blautia

and Lachnospiracaecae and a reduction in the genera
Bacteroides, Alistepes, and Clostridium, as well as lower
Bacteroidetes at the phylum level. Overall, the GM stayed
0
0
1
1
1
0

CLINICAL AT
relatively stable in all groups during overfeeding, regard-
less of the macronutrient. However, some specific changes
Assessment

were found, such as increases of Proteobacteria after

saturated fat overfeeding, whereas unsaturated fat
0
1
1
0
1
1
increased butyrate producers. For example, sugar over-
feeding increased Lactococcus, Escherichia coli, and Bilo-
phila. No differences were found in alpha diversity.47 For a
Comparability (max 2 stars)

full overview of results, see Table 1.

for confounders
Study controls

Coronary Heart Disease

Table 2.Risk of Bias Assessment According to the Newcastle-Ottawa Scale for Nonrandomized Studies29

Three randomized controlled trial subanalyses from the

0
0
0
0
0
1

same cohort evaluated the effect of the Mediterranean diet

on GM composition and clinical parameters in healthy,
overweight, and MetSyn individuals with coronary heart
disease.33,51,52 In all studies, participants had their last
cardiac event >6 months before the start of the study. Study
population varied between 20 and 239 individuals. The
present at start of
interest was not
that outcome of

studies used different techniques to evaluate microbiota

Demonstration

composition, focusing on different 16S rRNA regions and

the study

quantitative polymerase chain reaction. Changes in micro-

1
1
1
1
1
1

biota composition also varied between studies. In 2 studies,

the Mediterranean diet increased Faecalibacterium praus-
nitzii33,51 and Parabacteroides distasonis, whereas the low-
fat high-fiber diet in another study increased F prausnitzii
Ascertainment of

and decreased P distasonis. However, the Mediterranean

diet also increased P distasonis.52 Overall, a Mediterranean
Selection (max 4 stars)

exposure

and a low-fat high-fiber diet both improved insulin sensi-

1
1
1
0
1
0

tivity52 and reduced plasma triglyceride levels.33 In addi-

tion, a negative association between waist circumference
and fecal bacteria such as Bacteroides thetaiotaomicron, P
distasonis, F prausnitzii, Bifidobacterium longum, Rumino-
Selection

coccus flavefaciens subgroup, and Bifidobacterium ado-

lescentis was reported.51
0
0
1
0
0
0

Meta-analysis: Effect of Dietary Intervention on

Representativeness

Clinical Outcomes
of sample

Because CMDs have similar properties, we aimed to

evaluate whether a shared effect of dietary interventions on
1
1
0
0
0
0

clinical outcomes could be observed. Therefore, we pooled

all randomized studies with an intervention group as well as
a control group from unique study populations and per-
formed meta-analyses on relevant cardiometabolic out-
45

Pataky (2016)50

comes. Six studies34,35,41,43,48,49 with 8 intervention groups

Huang (2018)46
Ohlsson (2019)
32
Clark (2018)38

44
Louis (2016)
Frost (2019)

were eligible for the meta-analysis. Of these 6 studies, 2

included individuals with MetSyn, 2 included individuals
Study

with T2D, 1 included individuals with NAFLD and MetSyn,

and 1 study included MetSyn only. The studied outcomes
 1926 Attaye et al Gastroenterology Vol. 162, No. 7
CLINICAL AT

Figure 3. Forest plots of dietary interventions on clinical outcomes across CMDs. (A) Results of fasting blood glucose of
studies eligible for the meta-analysis. (B) Results of diastolic blood pressure and (C) triglycerides based on a meta-analysis
with random effects model. IV, inverse variance. SD, standard deviation. The solid squares indicate the mean difference
and are proportional to the weights used in the meta-analysis. The horizontal lines represent the 95% CI. The diamond in-
dicates the weighted mean difference, and the lateral tips of the diamond indicate the associated CIs.

were body weight, body mass index, fasting glucose, fasting evaluated outcomes. For a detailed overview of results see
insulin, homeostasis model assessment-insulin resistance, Supplementary Document 5.
body weight, body mass index, HDL-c, low-density Because CMDs comprises a spectrum of diseases, we also
lipoprotein-cholesterol, triglycerides, and systolic and dia- performed meta-analyses stratified to prespecified CMD
stolic blood pressure. When the 6 studies were pooled conditions. When only studies with participants with Met-
together, only in diastolic blood pressure a mean decrease Syn were assessed,34,35,49 none of the outcomes were
of
3.63 mm Hg (95% CI,
7.09 to
0.17 mm Hg; I2 ¼ significantly different between the intervention and control
0%, P ¼ .04) was observed compared with the control groups. However, dietary interventions in 2 NAFLD
group after the dietary interventions (Figure 3 and studies48,49 appeared to follow a trend toward higher HDL-c
Supplementary Document 5), as was assessed in 3 studies compared with the control group (mean difference, 0.10;
with 6 intervention groups. There was also a trend toward 95% CI, 0.00–0.20; I2 ¼ 0%, P ¼ .06). Studies that included
lower triglyceride levels (mean difference,
0.26 mmol/L; individuals with T2D41,43 did not show any differences be-
95% CI,
0.58 to 0.06 mmol/L; I2 ¼ 59%, P ¼ .11) and tween the groups.
lower fasting plasma glucose levels (mean We also aimed to assess what the effect of specific
difference,
0.18 mmol/L; 95% CI,
0.39 to 0.04 mmol/L; dietary interventions is on CMDs. In the 2 studies41,48 that
I2 ¼ 0%, P ¼ .11) (Figure 3), after dietary interventions. used a high-fiber dietary intervention, the mean triglyc-
Included studies for the meta-analyses did not always eride decrease was
0.69 mmol/L (95% CI,
1.36
report all outcomes of interest; therefore, the number of to
0.02 mmol/L; I2 ¼ 59%, P ¼ .04) compared with the
studies used for the meta-analyses differed between control group (Figure 4). Insufficient studies were
June 2022 New Standards for Dietary GM Studies in CMD 1927

CLINICAL AT
Figure 4. Forest plots of meta-analyses per diet and CMD separately. (A) HDL-c levels after a dietary intervention in studies
with NAFLD populations. (B) Triglyceride levels after dietary intervention in studies with a high-fiber or low-fiber diet. IV, inverse
variance. SD, standard deviation. The solid squares indicate the mean difference and are proportional to the weights used in
the meta-analysis. The horizontal lines represent the 95% CI. The diamond indicates the weighted mean difference, and the
lateral tips of the diamond indicate the associated CIs.

available to perform a meta-analysis of the separate ef- interventions and study methods such as study duration
fects of specific macronutrient changes such as a high- and microbiome experimental and computational tech-
protein vs low-protein diet, caloric restriction, or the niques. In addition to this, we found a high risk of bias in all
Mediterranean diet. performed clinical trials, as assessed via the Newcastle-
Ottawa Scale for nonrandomized studies29 and the
Cochrane Checklist for Randomized studies.28 Notably, all
Discussion studies scored a “high risk of bias” in the domain “Selection
This study is the first systematic review to date sum- of the Reported Results.” In addition to this, several studies
marizing clinical trials that investigated the effects of dietary used different GM sequencing techniques known to signifi-
modulations on GM composition and function in individuals cantly influence microbiota outcomes.54,55
with CMDs. This study identified clinical trials that investi- Previous systematic reviews already focused on evalu-
gated MetSyn, T2D, NAFLD, and CVD as forms of CMD. ating the effect of dietary interventions on a single CMD. A
However, it is important to note that CMD represents a systematic review56 on the effect of dietary intervention in
broad spectrum of disorders and that clinical microbiome MetSyn animal and human studies included 31 publications,
trials are still lacking for some diseases (eg, atheroscle- among them several reviews. While this is an interesting
rosis).53 Dietary interventions have been shown to modu- approach and provides the opportunity to take advantage of
late all clinical parameters of relevance to CMD, with GM the plethora of available information about the microbiome,
composition and function playing a crucial role in this ef- it also provides challenges on how to interpret their find-
fect.25 Our meta-analysis revealed that overall, dietary in- ings, because this information gathering from varying
terventions lowered diastolic blood pressure and that a sources is difficult to interpret. Willson and Situ56 found
high-fiber diet reduces diastolic but not systolic blood high SCFA levels were related to improved glycemic control
pressure. We also observed a trend in individuals with in participants with MetSyn. In the study of Hald et al,39 a
NAFLD that favored higher HDL-c levels in the intervention randomized controlled trial included in our study, no
diet. However, the exact interplay between dietary in- improvement of glycemic control was observed after a high-
terventions and GM composition and function in CMDs re- fiber diet, which indeed resulted in increased levels of fecal
mains to be elucidated because we could not identify SCFA. However, our meta-analysis revealed that a high-fiber
consistent microbiota changes related to clinical outcomes. diet reduces blood triglyceride levels, which is also in line
This may be due to the high risk of selection bias and het- with previous findings.57
erogeneity in study design as was observed in our study. A different study from Houghton et al58 focused not only
The main qualitative finding of this study was that on lifestyle interventions, such as dietary interventions and
macronutrient intake change by dietary interventions af- probiotic administration, but also on exercise interventions
fects GM profiles, but no consistent effect of diet on GM in T2D studies. This study included 8 studies, and changes
composition or function could be identified when looking at in clinical parameters and alpha diversity were observed. In
all groups together or separately. The main quantitative line with our study, no consistent changes were observed in
finding was that blood pressure and lipid spectra are the abundance of single bacteria in feces between inter-
affected by dietary interventions, but the effects do not vention and control groups. However, this study included
relate to universal GM changes or CMD-specific GM changes. different interventions, as also 3 interventions were evalu-
This might be explained by the heterogeneity of dietary ated with probiotic or synbiotic interventions, and
 1928 Attaye et al Gastroenterology Vol. 162, No. 7

macronutrient intake was not changed in 2 studies because showed changes in phyla ratios with Firmicutes. Because
fibers or the starch-converting enzyme transglucosidase the directions of the changes in phyla ratio or relative
were supplemented only.58 abundance were not consistent between groups, we cannot
Some interventions have already been evaluated in a conclude that dietary intervention effects are directly re-
systematic review across different CMDs. Grosso et al59 flected in phyla ratios. This is in line with findings of Magne
evaluated the role of the Mediterranean diet in CVD risk et al,65 which suggested in their review that Firmicutes/
factors, such as lipid profiles, inflammatory markers, and Bacteroidetes associations to obesity may depend on
hypertension, and also on MetSyn and T2D. The authors methodological differences between studies, sample pro-
concluded that their qualitative findings support the notion cesses, or correction for potential confounders. The variety
that the Mediterranean diet seems to be beneficial for in sex distribution and duration of interventions make it
several clinical markers such as blood pressure, insulin also more difficult to compare the results of these studies.
CLINICAL AT

resistance, and markers of inflammation; however, they also However, even though it might be tempting to describe the
concluded that more homogeneity of study methods and relative abundance of single bacterial strains, the GM
design is necessary to be able to get conclusive insights on should be considered as a biotope, and more emphasis on
the effect of this diet.59 In our review, we encountered guilds or trophic networks of intestinal symbiotic bacteria
studies that reported beneficial effects of the Mediterranean is probably necessary to better understand the role of GM
diet on insulin sensitivity and triglycerides; unfortunately, in health and disease.66
these studies were not eligible for the meta-analysis. Although ample evidence exists that dietary in-
We found that the GM composition in most evaluated terventions can significantly alter CMD (risk),67–69 causality
studies was changed by the dietary intervention. However, is still not shown with regards to the role of GM herein. All
affected GM composition was often not related to clinical randomized controlled trials included for qualitative anal-
improvements. Furthermore, observed clinical improve- ysis in our work scored a high risk of bias due to no pre-
ments could not be linked to consistent changes in GM published analysis plan of microbiota data, even though this
across different CMDs and separately with our qualitative was the primary outcome of several studies, thus increasing
approach. the risk of selection bias. This does not necessarily mean
In the overall group, we saw that diastolic blood pres- that selection bias influenced study results; however, it
sure was reduced based on 4 studies34,35,41,49 reporting 6 should lead to caution when interpreting the results of our
dietary interventions. This is in line with findings from a qualitative and quantitative results.
recent meta-analysis focusing on dietary interventions Another factor that introduces heterogeneity is that 11
against hypertension, which found that low-sodium diets studies excluded participants if they previously used anti-
lead to systolic and diastolic blood pressure reductions, biotics, whereas 10 studies did not report this
whereas the Mediterranean diet only reduces diastolic (Supplementary Document 3). In addition to this, we found
blood pressure.60 The 4 studies,34,35,41,49 however, did not that the time of prior antibiotics use differed between
use the Mediterranean diet as an intervention, although studies from no time definition and from 1 month to 6
important characteristics of the Mediterranean diet were months. Because antibiotics use profoundly affects the GM
part of most interventions, such as a high fiber intake and composition and it takes at least 4 weeks until most of the
fermented dairy products such as yoghurt.34,48,49,61 GM are restored,70 having different limits for antibiotics use
SCFA are microbial metabolites produced from dietary may confound microbiome-related findings. A broader
fibers.62 The GM is also associated with blood pressure; in perspective on selection bias in the microbiome field can be
fact, SCFA-producing bacteria such as Roseburia spp. were found in the accompanying commentary on this paper.71
associated with lower blood pressure.63 One study in our Studies assessing the effect of dietary interventions on
qualitative analysis found that the Mediterranean diet also clinical parameters through the GM involving CMDs other
increased the relative abundance of Roseburia and that a than MetSyn, NAFLD, and T2D, such as chronic kidney dis-
low-fat high-carbohydrate diet decreased the abundance,52 ease or CVD, have been studied poorly, and systematic re-
as did a lifestyle intervention with reduced energy views that provide an overview of the quality and number of
intake.35 Interestingly, the study of Santos-Marcos et al64 clinical studies are lacking. This systematic review is the
found that there are sex differences in response to a low- first to provide an overview and judgment of quality of all
fat diet: men had higher Roseburia levels than women af- clinical studies that investigated the role of dietary in-
ter a 3-year intervention period. In addition to this, a 12 terventions and GM composition and function with regard
week very low-caloric diet among individuals with MetSyn to a broad spectrum of CMDs.
and NAFLD decreased Roseburia relative abundance.32 The We found that MetSyn was the most studied CMD among
previously mentioned studies included individuals with all clinical trials. This is not surprising, because MetSyn is a
various combinations of MetSyn only,35 MetSyn and CHD,64 frequently occurring cluster of risk factors that contribute
MetSyn and NAFLD,32 and obesity and CHD.52 This illus- significantly to cardiovascular morbidity and mortality.72
trates that somewhat similar effects can be observed be- Other CMDs that were identified in this study were scarcely
tween different CMD spectra and calls for studies that allow investigated, and no clear conclusion can be drawn because
better comparison between studies. these studies varied in dietary intervention, microbial
At the phylum level, 2 studies42,49 showed a decrease in outcome, and populations investigated. Studies that investi-
relative Firmicutes abundance, and four32,33,43,47 studies gated MetSyn are more comparable and focused mainly on
June 2022 New Standards for Dietary GM Studies in CMD 1929

modulating carbohydrate and fiber intake.33,36,39 These Lastly, control interventions differed between the
macronutrients were predominantly investigated because studies, as some control intervention comprised milk
previous studies have shown that dietary interventions and intake,34,49 usual care with only dietary recommendations,41
weight loss can improve MetSyn, with a specific emphasis on or a placebo group that received additional caseinate,
the hyperglycemic component.73,74 This is of note, because maltodextrin, and sweeteners,35 making it more difficult to
insulin resistance plays a crucial role in CMDs.75 However, generalize our findings.
what type of dietary intervention is best suited in preventing
or even reversing MetSyn is currently unclear.76
With regards to the GM, some clinical trials included in this Conclusion
study focused on altered SCFA levels or producers. Other Dietary interventions modulating CMDs affect the GM
more recently identified microbial metabolites20,77 are yet to composition. Dietary interventions changing the GM also
influence clinical parameters, such as diastolic blood pres-

CLINICAL AT
be addressed in clinical trial settings. However, observational
studies did find that MetSyn is associated with a reduced GM sure, in overall CMD, and a high-fiber diet decreased tri-
diversity and altered levels of gut-derived metabolites such as glyceride levels in participants with NAFLD and T2D.
SCFA and trimethylamine N-oxide.78–80 Interestingly, our However, included studies had a high risk of bias, and a
meta-analysis did not show specific changes for participants broad variety in study design and methods was observed.
with MetSyn. However, SCFAs have also been shown to affect Therefore, future studies with a standardized approach to
lipid metabolism, for example, by inhibiting lipolysis in adi- evaluate GM are necessary to better understand the com-
pocytes.81 In fact, rectal administration of acetate and pro- plex relation between diet, GM, and CMDs.
pionate resulted in a decrease of free fatty acids and a
concomitant increase in triglycerides.82 Supplementary Material
This is in contrast to beneficial effects observed from Note: To access the supplementary material accompanying
recent literature, as SCFAs alleviated inflammation and this article, visit the online version of Gastroenterology at
steatosis in mice with nonalcoholic steatohepatitis.83 This is www.gastrojournal.org, and at https://doi.org/10.1053/
in line with our findings, where a high-fiber diet in partici- j.gastro.2022.02.011.
pants with NAFLD48 and T2D41 reduced triglyceride levels
compared with the control group. However, it should also be
noted that the presence of SCFA producers does not
necessarily mean that this influences SCFA levels per se, References
because the SCFA production depends on local metabolism
1. Miranda JJ, Barrientos-Gutiérrez T, Corvalan C, et al.
and may vary between individuals. More clinical studies
Understanding the rise of cardiometabolic diseases in
focusing on the mechanistic role of SCFA in lipid metabolism
low-and middle-income countries. Nat Med 2019;
and GM of patients with CMDs are necessary to better un-
25:1667–1679.
derstand this complex interplay.
2. Kopp W. How Western diet and lifestyle drive the
pandemic of obesity and civilization diseases. Diabetes
Metab Syndr Obes Targets Ther 2019;12:2221–2236.
Limitations 3. Rooij SR de, Nijpels G, Nilsson PM, et al. Low-grade
This systematic review has several limitations. Most chronic inflammation in the Relationship between Insulin
included studies could only be assessed in a qualitative Sensitivity and Cardiovascular Disease (RISC) popula-
manner, and the meta-analysis only assessed clinical out- tion: associations with insulin resistance and car-
comes due to the heterogeneity of used methods and out- diometabolic risk profile. Diabetes Care 2009;
comes. In addition to this, we only included studies with a 32:1295–1301.
control and intervention group for the meta-analysis, which 4. Guarner V, Rubio-Ruiz ME. Low-grade systemic inflam-
decreases the power of our results. Moreover, this review mation connects aging, metabolic syndrome and car-
could only identify a relatively small number of clinical diovascular disease. Interdiscip Top Gerontol 2015;
studies, which did not cover the full range of CMDs. This 40:99–106.
finding indicates that the relationship between diet, GM 5. Pereira SS, Alvarez-Leite JI. Low-grade inflammation,
composition and function, and CMDs has been poorly studied obesity, and diabetes. Curr Obes Rep 2014;3:422–431.
and skewed to certain CMDs, namely MetSyn and T2D. 6. Yang Q, Vijayakumar A, Kahn BB. Metabolites as regu-
Owing to the large variation in methodology between the lators of insulin sensitivity and metabolism. Nat Rev Mol
studies, it was not possible to draw definitive conclusions Cell Biol 2018;19:654–672.
on how diet modulates the GM composition and function 7. Bensinger SJ, Tontonoz P. Integration of metabolism
and affects CMDs. and inflammation by lipid-activated nuclear receptors.
Additionally, most participants in the included studies Nature 2008;454:470–477.
had comorbidities that might not reflect the general popu- 8. Ferrante AW. Obesity-induced inflammation: a metabolic
lation and could thus limit the generalizability of our find- dialogue in the language of inflammation. J Intern Med
ings. Furthermore, most of the studies performed did not 2007;262:408–414.
report on significant confounders, such as medication use 9. Kjeldsen SE. Hypertension and cardiovascular risk:
and lifestyle, that could affect (microbial) outcomes. general aspects. Pharmacol Res 2018;129:95–99.
 1930 Attaye et al Gastroenterology Vol. 162, No. 7

10. Wenstedt EF, Verberk SG, Kroon J, et al. Salt increases 27. Ouzzani M, Hammady H, Fedorowicz Z, et al. Rayyan—a
monocyte CCR2 expression and inflammatory re- web and mobile app for systematic reviews. Syst Rev
sponses in humans. JCI Insight 2019;4:e130508. 2016;5:210.
11. Chen Z, Watanabe RM, Stram DO, et al. High calorie 28. Higgins JPT, Altman DG, Gotzsche PC, et al. The
intake is associated with worsening insulin resistance Cochrane Collaboration’s tool for assessing risk of bias
and b-cell function in Hispanic women after gestational in randomised trials. BMJ 2011;343:d5928–d5928.
diabetes mellitus. Diabetes Care 2014;37:3294–3300. 29. The Ottawa Hospital Research Institute. Newcastle–
12. Cosentino F, Grant PJ, Aboyans V, et al. 2019 ESC Ottawa Quality Assessment Scale Cohort Studies.
Guidelines on diabetes, pre-diabetes, and cardiovascular Available at: http://www.ohri.ca/programs/clinical_
diseases developed in collaboration with the EASD. Eur epidemiology/nosgen.pdf.31. Accessed June 1, 2021.
Heart J 2020;41:255–323. 30. Munckhof ICL van den, Kurilshikov A, Ter Horst R, et al.
13. American Diabetes Association. 8. Obesity management
CLINICAL AT

Role of gut microbiota in chronic low-grade inflammation

for the treatment of type 2 diabetes: Standards of Medical as potential driver for atherosclerotic cardiovascular
Care in Diabetes—2020. Diabetes Care 2020;43:S89–S97. disease: a systematic review of human studies. Obes
14. Georgoulis M, Kontogianni MD, Yiannakouris N. Medi- Rev 2018;19:1719–1734.
terranean diet and diabetes: prevention and treatment. 31. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting
Nutrients 2014;6:1406–1423. items for systematic reviews and meta-analyses: the
15. Martínez-González MA, Gea A, Ruiz-Canela M. The PRISMA statement. PLoS Med 2009;6:e1000097.
Mediterranean diet and cardiovascular health. Circ Res 32. Louis S, Tappu R-M, Damms-Machado A, et al. Char-
2019;124:779–798. acterization of the gut microbial community of obese
16. Canfora EE, Jocken JW, Blaak EE. Short-chain fatty patients following a weight-loss intervention using whole
acids in control of body weight and insulin sensitivity. metagenome shotgun sequencing. PLoS One 2016;11:
Nat Rev Endocrinol 2015;11:577–591. e0149564.
17. Pingitore A, Chambers ES, Hill T, et al. The diet-derived 33. Haro C, García-Carpintero S, Rangel-Zúñiga OA, et al.
short chain fatty acid propionate improves beta-cell Consumption of two healthy dietary patterns restored
function in humans and stimulates insulin secretion microbiota dysbiosis in obese patients with metabolic
from human islets in vitro. Diabetes Obes Metab 2017; dysfunction. Mol Nutr Food Res 2017;61:1700300.
19:257–265. 34. Bellikci-Koyu E, Sarer-Yurekli BP, Akyon Y, et al. Effects
18. Vrieze A, Van Nood E, Holleman F, et al. Transfer of in- of regular kefir consumption on gut microbiota in pa-
testinal microbiota from lean donors increases insulin tients with metabolic syndrome: a parallel-group, ran-
sensitivity in individuals with metabolic syndrome. domized, controlled study. Nutrients 2019;11:2089.
Gastroenterology 2012;143:913–916.e7. 35. Guevara-Cruz M, Flores-López AG, Aguilar-López M,
19. Herrema H, Niess JH. Intestinal microbial metabolites in et al. Improvement of lipoprotein profile and metabolic
human metabolism and type 2 diabetes. Diabetologia endotoxemia by a lifestyle intervention that modifies the
2020;63:2533–2547. gut microbiota in subjects with metabolic syndrome.
20. Koh A, Molinaro A, Ståhlman M, et al. Microbially pro- J Am Heart Assoc 2019;8:e012401.
duced imidazole propionate impairs insulin signaling 36. Lappi J, Salojärvi J, Kolehmainen M, et al. Intake of
through mTORC1. Cell 2018;175:947–961.e17. whole-grain and fiber-rich rye bread versus refined wheat
21. Wang Z, Tang WHW, Buffa JA, et al. Prognostic value of bread does not differentiate intestinal microbiota
choline and betaine depends on intestinal microbiota- composition in Finnish adults with metabolic syndrome.
generated metabolite trimethylamine-N-oxide. Eur J Nutr 2013;143:648–655.
Heart J 2014;35:904–910. 37. Chen Z, Franco OH, Lamballais S, et al. Associations of
22. León-Mimila P, Villamil-Ramírez H, Li XS, et al. Trime- specific dietary protein with longitudinal insulin resis-
thylamine N-oxide levels are associated with NASH in tance, prediabetes and type 2 diabetes: the Rotterdam
obese subjects with type 2 diabetes. Diabetes Metab Study. Clin Nutr 2020;39:242–249.
2021;47:1011830. 38. Clark RL, Famodu OA, Holásková I, et al. Educational
23. Tang WHW, Wang Z, Levison BS, et al. Intestinal mi- intervention improves fruit and vegetable intake in young
crobial metabolism of phosphatidylcholine and cardio- adults with metabolic syndrome components. Nutr Res
vascular risk. N Engl J Med 2013;368:1575–1584. 2019;62:89–100.
24. Hills R, Pontefract B, Mishcon H, et al. Gut microbiome: 39. Hald S, Schioldan AG, Moore ME, et al. Effects of ara-
profound implications for diet and disease. Nutrients binoxylan and resistant starch on intestinal microbiota
2019;11:1613. and short-chain fatty acids in subjects with metabolic
25. Attaye I, Pinto-Sietsma S-J, Herrema H, et al. A crucial syndrome: a randomised crossover study. PLoS One
role for diet in the relationship between gut microbiota 2016;11:e0159223.
and cardiometabolic disease. Annu Rev Med 2020; 40. Medina-Vera I, Sanchez-Tapia M, Noriega-López L, et al.
71:149–161. A dietary intervention with functional foods reduces
26. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting metabolic endotoxaemia and attenuates biochemical
items for systematic reviews and meta-analyses: the abnormalities by modifying faecal microbiota in people
PRISMA statement. BMJ 2009;339:b2535–b2535. with type 2 diabetes. Diabetes Metab 2019;45:122–131.
June 2022 New Standards for Dietary GM Studies in CMD 1931

41. Zhao L, Zhang F, Ding X, et al. Gut bacteria selectively analytical “microscope” affect the biological interpreta-
promoted by dietary fibers alleviate type 2 diabetes. tion? Microorganisms 2019;7:393.
Science 2018;359:1151–1156. 56. Willson K, Situ C. Systematic review on effects of diet on
42. Karusheva Y, Koessler T, Strassburger K, et al. Short- gut microbiota in relation to metabolic syndromes. J Clin
term dietary reduction of branched-chain amino acids Nutr Metab 2017;1:2.
reduces meal-induced insulin secretion and modifies 57. Hannon BA, Thompson SV, Edwards CG, et al. Dietary
microbiome composition in type 2 diabetes: a random- fiber is independently related to blood triglycerides
ized controlled crossover trial. Am J Clin Nutr 2019; among adults with overweight and obesity. Curr Dev
110:1098–1107. Nutr 2019;3:nzy094.
43. Balfegó M, Canivell S, Hanzu FA, et al. Effects of sardine- 58. Houghton D, Hardy T, Stewart C, et al. Systematic review
enriched diet on metabolic control, inflammation and gut assessing the effectiveness of dietary intervention on gut
microbiota in drug-naïve patients with type 2 diabetes: a microbiota in adults with type 2 diabetes. Diabetologia

CLINICAL AT
pilot randomized trial. Lipids Health Dis 2016;15:78. 2018;61:1700–1711.
44. Frost F, Storck LJ, Kacprowski T, et al. A structured weight 59. Grosso G, Mistretta A, Frigiola A, et al. Mediterranean
loss program increases gut microbiota phylogenetic di- diet and cardiovascular risk factors: a systematic review.
versity and reduces levels of Collinsella in obese type 2 Crit Rev Food Sci Nutr 2014;54:593–610.
diabetics: a pilot study. PLoS One 2019;14:e0219489. 60. Gay HC, Rao SG, Vaccarino V, et al. Effects of different
45. Ohlsson B. An Okinawan-based Nordic diet improves dietary interventions on blood pressure: systematic re-
glucose and lipid metabolism in health and type 2 dia- view and meta-analysis of randomized controlled trials.
betes, in alignment with changes in the endocrine profile, Hypertens Dallas Tex 1979 2016;67:733–739.
whereas zonulin levels are elevated. Exp Ther Med 2019; 61. Zhao L, Zhang Q, Ma W, et al. A combination of quer-
17:2883–2893. cetin and resveratrol reduces obesity in high-fat diet-fed
46. Huang F, Nilholm C, Roth B, et al. Anthropometric and rats by modulation of gut microbiota. Food Funct 2017;
metabolic improvements in human type 2 diabetes after 8:4644–4656.
introduction of an Okinawan-based Nordic diet are not 62. Cummings JH, Pomare EW, Branch WJ, et al. Short
associated with changes in microbial diversity or SCFA chain fatty acids in human large intestine, portal, hepatic
concentrations. Int J Food Sci Nutr 2018;69:729–740. and venous blood. Gut 1987;28:1221–1227.
47. Jian C, Luukkonen P, Sädevirta S, et al. Impact of short- 63. Verhaar BJH, Collard D, Prodan A, et al. Associations
term overfeeding of saturated or unsaturated fat or between gut microbiota, faecal short-chain fatty acids,
sugars on the gut microbiota in relation to liver fat in and blood pressure across ethnic groups: the HELIUS
obese and overweight adults. Clin Nutr Edinb Scotl 2021; study. Eur Heart J 2020;41:4259–4267.
40:207–216. 64. Santos-Marcos JA, Haro C, Vega-Rojas A, et al. Sex
48. de Faria Ghetti F, De Oliveira DG, De Oliveira JM, et al. differences in the gut microbiota as potential de-
Effects of dietary intervention on gut microbiota and terminants of gender predisposition to disease. Mol Nutr
metabolic-nutritional profile of outpatients with non- Food Res 2019;63:e1800870.
alcoholic steatohepatitis: a randomized clinical trial. 65. Magne F, Gotteland M, Gauthier L, et al. The Firmicutes/
J Gastrointest Liver Dis 2019;28:279–287. Bacteroidetes ratio: a relevant marker of gut dysbiosis in
49. Chen Y, Feng R, Yang X, et al. Yogurt improves insulin obese patients? Nutrients 2020;12:1474.
resistance and liver fat in obese women with nonalcoholic 66. Wu G, Zhao N, Zhang C, et al. Guild-based analysis for
fatty liver disease and metabolic syndrome: a randomized understanding gut microbiome in human health and
controlled trial. Am J Clin Nutr 2019;109:1611–1619. diseases. Genome Med 2021;13:22.
50. Pataky Z, Genton L, Spahr L, et al. Impact of hypocaloric 67. Esposito K, Maiorino MI, Bellastella G, et al. A journey
hyperproteic diet on gut microbiota in overweight or into a Mediterranean diet and type 2 diabetes: a sys-
obese patients with nonalcoholic fatty liver disease: a tematic review with meta-analyses. BMJ Open 2015;5:
pilot study. Dig Dis Sci 2016;61:2721–2731. e008222.
51. Haro C, Garcia-Carpintero S, Alcala-Diaz JF, et al. The 68. Siervo M, Lara J, Chowdhury S, et al. Effects of the Di-
gut microbial community in metabolic syndrome patients etary Approach to Stop Hypertension (DASH) diet on
is modified by diet. J Nutr Biochem 2016;27:27–31. cardiovascular risk factors: a systematic review and
52. Haro C, Montes-Borrego M, Rangel-Zúñiga OA, et al. meta-analysis. Br J Nutr 2015;113:1–15.
Two healthy diets modulate gut microbial community 69. Brandhorst Sebastian, Longo Valter D. Dietary re-
improving insulin sensitivity in a human obese popula- strictions and nutrition in the prevention and treatment
tion. J Clin Endocrinol Metab 2016;101:233–242. of cardiovascular disease. Circ Res 2019;124:
53. Frostegård J. Immunity, atherosclerosis and cardiovas- 952–965.
cular disease. BMC Med 2013;11:117. 70. Harris VC, Haak BW, Handley SA, et al. Effect of
54. Rintala A, Pietilä S, Munukka E, et al. Gut microbiota antibiotic-mediated microbiome modulation on rotavirus
analysis results are highly dependent on the 16S rRNA vaccine immunogenicity: a human, randomized-control
gene target region, whereas the impact of DNA extrac- proof-of-concept trial. Cell Host Microbe 2018;24:
tion is minor. J Biomol Tech 2017;28:19–30. 197–207.e4.
55. Siegwald L, Caboche S, Even G, et al. The impact of 71. Warmbrunn M, Attaye I, Herrema H, Nieuwdorp M.
bioinformatics pipelines on microbiota studies: does the Protocol standardization of microbiome
 1932 Attaye et al Gastroenterology Vol. 162, No. 7

studies—daunting but necessary. Gastroenterology 82. Wolever TM, Brighenti F, Royall D, et al. Effect of rectal
2022;162:1822–1824. infusion of short chain fatty acids in human subjects. Am
72. Expert Panel on Detection, Evaluation, and Treatment of J Gastroenterol 1989;84:1027–1033.
High Blood Cholesterol in Adults. Executive summary of 83. Deng M, Qu F, Chen L, et al. SCFAs alleviated steatosis
the third report of the National Cholesterol Education and inflammation in mice with NASH induced by MCD.
Program (NCEP) Expert Panel on Detection, Evaluation, J Endocrinol 2020;245:425–437.
and Treatment of High Blood Cholesterol in Adults (Adult
Treatment Panel III). JAMA 2001;285:2486–2497.
73. Case CC, Jones PH, Nelson K, et al. Impact of weight loss
Received June 9, 2021. Accepted February 1, 2022.
on the metabolic syndrome. Obes Metab 2002;4:407–414.
74. Samson SL, Garber AJ. Metabolic syndrome. Endocrinol Correspondence
Address correspondence to: Max Nieuwdorp, MD, PhD, Amsterdam UMC,
Metab Clin North Am 2014;43:1–23.
CLINICAL AT

Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands. e-mail:

75. Ormazabal V, Nair S, Elfeky O, et al. Association between [email protected].
insulin resistance and the development of cardiovascular Acknowledgments
disease. Cardiovasc Diabetol 2018;17:122. The authors thank E.N. Hamulyak for critical input on the study design.
76. de la Iglesia R, Loria-Kohen V, Zulet M, et al. Dietary
CRediT Authorship Contributions
strategies implicated in the prevention and treatment of Ilias Attaye, MD (Conceptualization: Lead; Formal analysis: Equal;
metabolic syndrome. Int J Mol Sci 2016;17:1877. Methodology: Equal; Software: Lead; Supervision: Lead; Visualization: Lead;
Writing – original draft: Lead; Writing – review & editing: Lead). Moritz
77. Nemet I, Saha PP, Gupta N, et al. A cardiovascular Valerian Warmbrunn, MD (Conceptualization: Lead; Data curation: Equal;
disease-linked gut microbial metabolite acts via adren- Formal analysis: Lead; Investigation: Equal; Methodology: Equal;
ergic receptors. Cell 2020;180:862–877.e22. Visualization: Lead; Writing – original draft: Lead; Writing – review & editing:
Lead). Aureline N.A.F. Boot, BSc (Data curation: Supporting; Investigation:
78. Barrea L, Annunziata G, Muscogiuri G, et al. Trimethyl- Supporting). Suze C. van der Wolk, BSc (Formal analysis: Supporting;
amine-N-oxide (TMAO) as novel potential biomarker of Investigation: Supporting). Barbara A. Hutten, PhD (Data curation: Equal;
Formal analysis: Supporting). Joost. G. Daams, MA (Investigation: Equal).
early predictors of metabolic syndrome. Nutrients 2018; Hilde Herrema, PhD (Conceptualization: Supporting; Writing – review &
10:1971. editing: Equal). Max Nieuwdorp, MD, PhD (Conceptualization: Equal; Funding
acquisition: Lead; Methodology: Equal; Writing – review & editing: Equal).
79. DiNicolantonio JJ, McCarty M, OKeefe J. Association of
moderately elevated trimethylamine N-oxide with car- Conflicts of interest
diovascular risk: is TMAO serving as a marker for hepatic The authors disclose no conflicts.
insulin resistance. Open Heart 2019;6:e000890.
80. Festi D, Schiumerini R, Eusebi LH, et al. Gut microbiota Funding
and metabolic syndrome. World J Gastroenterol 2014; Moritz V. Warmbrunn is supported by a Netherlands CardioVascular Research
Committee (CVON) IN-CONTROL-II grant (2018-27). Ilias Attaye is supported
20:16079. by a Joint Programming Initiative a Healthy Diet for a Healthy Life (JPI HDHL)
81. Morrison DJ, Preston T. Formation of short chain fatty MICRODIET grant (5290510105). Max Nieuwdorp is supported by a ZonMw
Vici grant 2020[09150182010020] and a Le Ducq consortium grant
acids by the gut microbiota and their impact on human 17CVD01. Hilde Herrema is supported by a Senior Fellowship of the Dutch
metabolism. Gut Microbes 2016;7:189–200. Diabetes Research Foundation (2019.82.004).