Gut microbial clusters in children show different cardiometabolic responses to wholegrains – a post hoc analysis of a randomized wholegrain trial in children

Abstract

The beneficial effects of wholegrain intake on cardiometabolic risk may be mediated by the gut microbiota. In adults, cardiometabolic responses appear to be influenced partly by the structure of the gut microbiome ecosystem. However, this has not been investigated in children. We aimed to identify gut microbial clusters among children and explore their role in the cardiometabolic response to high wholegrain intake. The present study utilized data from a previous randomized cross-over study involving 51 Danish 8–13 year-old children with high BMI, provided with wholegrain oats and rye (“WG”) and refined grain (“RG”) for 8 weeks in random order. Anthropometry, body composition, cardiometabolic markers in blood, and blood and faecal short-chain fatty acids were assessed at 0, 8 and 16 weeks. Faecal microbiota composition was assessed by 16S rRNA gene amplicon sequencing and microbial clusters were identified by partitioning around medoid clustering. We used linear mixed models to investigate the modifying effects of the baseline microbial cluster on the cardiometabolic effects of wholegrain. We identified two clusters dominated by Faecalibacterium (Cluster 1, n = 25), and Clostridium (Cluster 2, n = 26), respectively. Cluster modified the effects of WG on BMI z-score, fat mass index, and waist circumference (all P_interaction ≤ 0.002), which were reduced among children in Cluster 1 and increased in those in Cluster 2. Cluster also modified the effect on HDL cholesterol (P_interaction = 0.03) and tended to modify the effect on faecal butyrate. In conclusion, two distinct gut microbial clusters were identified in schoolchildren, which appeared to affect how effectively a high wholegrain intake reduced fat mass.

---

Introduction

Wholegrain intake has consistently been inversely associated with type 2 diabetes (T2D),¹ cardiovascular disease (CVD),² and weight gain^3,4 in observational studies. Randomized trials, particularly investigating wholegrain oat and rye intake, show beneficial effects on cardiometabolic risk markers in adults,^5–12 and emerging research suggests similar effects among children.^13–17 The cardiometabolic benefits of wholegrain intake have been suggested to be mediated by fermentation of cereal fibres by the gut microbiota, leading to production of short-chain fatty acids (SCFA),^{10,12,17–20} which have demonstrated beneficial metabolic effects in humans and animal models.^21–24 However, the overall evidence of the effects of wholegrain is inconclusive due to large between-study heterogeneity.²⁵ While variations in the results of trials may partly be attributed to the administration of various wholegrain types, individualized responses to wholegrain intake may also play a role, as explorative studies have shown that the effects of wholegrain consumption on cardiometabolic risk markers in adults can be modified by baseline gut microbial composition.^10,26–28

The establishment of stable gut microbial community structures and clusters, also referred to as “enterotypes”, has traditionally been recognized to occur within the first few years of life, influenced by factors such as birth mode, breastfeeding, and number of siblings.²⁹ Emerging research has challenged this understanding, revealing continued gut microbial development throughout mid-childhood, partly influenced by habitual diet, in particular dietary fibre intake,^30–33 but gut microbial clusters have yet to be thoroughly investigated across diverse young populations during the transition into adulthood. Three robust cross-national enterotypes, characterized by the variation in the abundances of Bacteroides, Prevotella, and Ruminococcus, are typically identified in adults.^34,35 Enterotypes with high abundance of Prevotella and Bacteroides have been linked to long-term dietary patterns in adults³⁶ and the ratio between these two taxa has demonstrated stability even following an intervention diet rich in dietary fibre.³⁷ Noticeably, previous studies have shown that diets high in wholegrain or dietary fibre can have beneficial effects on weight loss,^26,27 glucose metabolism²⁸ and blood lipids¹⁰ among adults with a high Prevotella abundance. However, the potential modifying effect of pre-treatment gut microbiota on cardiometabolic risk markers has never been explored in dietary intervention trials among children.

The present exploratory study was based on the randomized cross-over trial, KORN, among Danish 8–13 year-olds with a high body mass index (BMI).¹⁷ The study showed that high intake of wholegrain oats and rye (“WG”), compared to refined grains (“RG”), reduced serum low-density lipoprotein cholesterol (LDL-C), modulated the relative abundance of specific microbial taxa, and increased SCFA in both plasma and faecal samples. In the present study, we aimed to identify gut microbial clusters present before the intervention and investigate whether these clusters modify the cardiometabolic response to wholegrain intake in schoolchildren.

Experimental

Study design, participants and randomization

The present study used data from a 2 × 8 week randomized crossover study among children that was conducted at the Department of Nutrition, Exercise and Sports (NEXS), University of Copenhagen, Denmark during September 2020 to May 2021.¹⁷ The study protocol was approved by the Committees on Biomedical Research Ethics for the Capital Region of Denmark (no. H-19010737) and registered with clinicaltrials.gov (NCT04430465, Study Record | Beta ClinicalTrials.gov). Written consent was obtained from all custody holders of the children prior to enrolment.

Eligible children living in the Capital Region of Denmark were identified through the national civil registry and invited by digital invitation to custody holders. In order to be included in the trial, the children had to be 8–13 years old, have a reported BMI >1 SD above the median according to sex and age-standardized Danish growth curves,¹⁷ eat grain products daily, speak Danish, and have at least one parent fluent in Danish. Exclusion applied for children with allergies to the study products, the use of dietary fibre or probiotic supplements, energy restricted diets, serious illness, medication that could interfere with the intervention, participation in other trials or cohabitation with another participating child.¹⁷

Children were randomized to a treatment sequence in blocks of six at the end of the baseline examination visit. Study participants and staff were not blinded during the intervention, but investigators were blinded during laboratory analysis. The study was conducted during the COVID-19 pandemic, which involved lockdown of schools and sports activities from December 9, 2020, to February 8, 2021, so children were home-schooled during this period.

We included 55 children in the trial; 52 of them completed the intervention and the 51 children who had available baseline microbiota data were included in the present exploratory study.

Study products

Children were provided with study products containing wholegrain from oats and rye, “WG”, and refined grain, “RG”, for 2 eight week periods, in random order. The products included breakfast cereals, breads, pasta, etc., and were provided ad libitum to substitute for the grain products habitually eaten by the children during a whole day as previously described.¹⁷ Children consumed 108 ± 38 and 3 ± 2 g d⁻¹ wholegrain in the WG and RG periods, respectively, which was verified by a profound between-period difference in plasma alkylresorcinols (P < 0.001).¹⁷

Questionnaires

Questionnaires about the household's highest education, children's pubertal stage, physical activity and screen time were completed online within one week prior to the first visit as previously described.¹⁷

Dietary intake

Dietary intake was recorded by the parents with help from the children during the last week prior to each examination visit, using a 4-day weighted dietary record on three consecutive weekdays and one weekend day as previously described.¹⁷ Intake of study products was included in the 4-day dietary records, but the reported daily intake was based on daily self-recording of the consumed study product (in decilitres or grams) throughout the trial duration.

Anthropometry, body composition and blood pressure

At 0, 8 and 16 weeks, we measured the children's height, weight, waist circumference, blood pressure, and body composition by whole body DXA, as detailed.¹⁷ Age- and sex-adjusted z-scores for BMI were calculated based on the WHO's growth standards using the WHO 2007 R macro package, WHO2007_R.zip.³⁸ The fat mass index (FMI) was calculated as kg of total body fat divided by the squared height in m.

Blood and faecal samples

At each time point, we collected fasting blood samples which were processed and analysed according to specific protocols as previously described.¹⁷ We quantified serum LDL-C, high-density lipoprotein cholesterol (HDL-C), total cholesterol (TC) and triglycerides (TG), plasma insulin and glucose, serum C-reactive protein (CRP) and interleukin 6 (IL-6) and plasma SCFA, including acetate, propionate and butyrate.

Faecal samples were collected and analysed for concentrations of acetate, propionate and butyrate using liquid chromatography with tandem mass spectrometry (LC-MS/MS) as previously described.¹⁷ The gut microbiota composition was determined by 16S rRNA gene amplicon sequencing (Illumina NextSeq). Details of DNA extraction, library preparation, sequencing and construction of the amplicon sequencing variant (ASV) table have been provided previously.³⁹ Phylogenetic annotation of the gut microbiome including agglomeration to the genus level was performed using previously described procedures.³⁹ The microbiota composition was classified using Jensen–Shannon distance and partitioning around medoid (PAM) clustering.³⁵ The Calinski–Harabasz (CH) index was applied to determine the optimal number of clusters and the silhouette index was calculated using R.⁴⁰ We tested for significant differences between clusters using PERMANOVA (999 permutations) with the vegan R package (v2.6-2).

Statistical analyses

Descriptive data are presented as mean ± SD or median (25^th–75^th percentiles) for normally distributed and skewed variables, respectively. Background characteristics, dietary intake and cardiometabolic markers were compared between Cluster 1 and Cluster 2, using unpaired Student's t-test or Wilcoxon rank-sum test, for normally distributed and skewed continuous variables, respectively, and Pearson's chi-squared test for categorical data. The relative abundances of the top 15 genera were compared between clusters, using a non-parametric test with p values adjusted using Holm–Bonferroni false discovery rate correction.

To investigate the effects of WG compared to RG on cardiometabolic risk markers and SCFAs and the involvement of microbial clusters, we used linear mixed models with a treatment × cluster interaction. In the full model, fixed effects included treatment (WG/RG), cluster (1/2), the treatment × cluster interaction term (4 levels), baseline value and home-schooling due to COVID-19 (% of all days), while the subject was modelled as a random effect to adjust for within-subject correlation. Treatment effects were analysed as mean differences (95% CI) between WG and RG within each microbial cluster and potential treatment modification by cluster was analysed using an F-test, comparing the full model with a similar model without the interaction term. Models were verified based on residual and normal probability plots. Insulin, TG, CRP and IL-6 were log-transformed prior to analyses, and estimates were back-transformed, as previously described.⁴¹

All statistical analyses were performed in R version 4.2.1, using particularly the lmerTest package version 3.1-3.⁴² Statistical significance was established at P < 0.05.

Results and discussion

The 51 included children were on average 11 years of age and had a BMI z-score of 1.5 ± 0.6 (mean ± SD); 47% were females. The children were separated into two distinct gut microbial clusters by Jensen–Shannon distance and PAM clustering with 25 children in Cluster 1 and 26 children in Cluster 2 (Fig. 1A). The average silhouette index of the clusters was 0.138, indicating that the 2 clusters separate, but with the clustering of some subjects being associated with some uncertainty as also evident by visual inspection. However, importantly, the allocation to the two clusters remained robust after the start of the intervention (chi-squared test for comparison of cluster allocation at the baseline and first visit, P = 0.114). The children in the two clusters did not differ in any of the assessed variables at the baseline, although LDL cholesterol tended to be higher in Cluster 2 and dietary fibre intake tended to be higher in Cluster 1 (Table 1). Cluster 1 tended to have higher diversity than Cluster 2, as determined by the Shannon diversity index (P = 0.09), but clusters did not differ in species richness as determined by the Chao1 index (Fig. 1B). Following adjustment for multiple testing, the children in Cluster 1 showed a higher relative abundance of Faecalibacterium, whereas those in Cluster 2 had higher relative abundance of Clostridium cluster XI and Clostridium sensu stricto (Fig. 1C).

Fig. 1 (A) At the baseline, the children from the KORN study (n = 51) separated into two microbial clusters (Cluster 1 and Cluster 2) using the partitioning around medoids algorithm. The between cluster P-value is based on comparison using PERMANOVA (999 permutations). (B) α-Diversity within the two microbial clusters defined by the Shannon index and Chao1 index. Data are given as median (25th–75th percentiles) and P-values are based on the Wilcoxon rank sum test comparisons between the clusters. (C) Distribution of the top 15 generas’ relative abundances in the two clusters. The red dots represent median values plotted on a log10 scaled x-axis. Adjusted p-values for between cluster comparisons of relative abundance were determined by the non-parametric test with Holm–Bonferroni false discovery rate correction.

Table 1 Baseline characteristics by microbial cluster (mean ± SD)

	Cluster 1 (n = 25)	Cluster 2 (n = 26)	P-Value^a
n = 24 for all blood measurements in Cluster 1. MVPA, moderate to vigorous physical activity.a P-Values for comparison of Clusters 1 and 2 using Student's t-test, Wilcoxon rank-sum test and Pearson's chi-squared test as appropriate.b Median (25^th–75^th percentiles).c Household's longest education.
Female sex (n, %)	13, 52	11, 42	0.680
Age (years)	11.1 ± 1.8	11.1 ± 1.9	0.969
Tanner stage (>1) (n, %)	15, 60	15, 58	1.000
Body weight (kg)	50.1 ± 11.3	49.8 ± 11.8	0.931
BMI (z-score)	1.4 ± 0.6	1.6 ± 0.6	0.408
Fat mass index (kg m^−2)	8.0 ± 2.2	8.4 ± 1.8	0.574
Waist circumference (cm)	78.2 ± 8.8	77.7 ± 9.0	0.852
Diastolic blood pressure (mmHg)	65 ± 3	66 ± 4	0.249
Systolic blood pressure (mmHg)	103 ± 5	104 ± 6	0.304
LDL cholesterol (mmol L^−1)	2.1 ± 0.5	2.5 ± 0.7	0.060
HDL cholesterol (mmol L^−1)	1.4 ± 0.2	1.3 ± 0.2	0.500
Total cholesterol (mmol L^−1)	3.8 ± 0.6	4.1 ± 0.8	0.169
Total: HDL cholesterol	2.9 ± 0.6	3.2 ± 0.7	0.130
Triacylglycerol (mmol L^−1)^b	0.58 (0.48–0.93)	0.65 (0.51–0.85)	0.547
Insulin (pmol L^−1)^b	59 (49–82)	73 (50–87)	0.322
Glucose (mmol L^−1)	5.1 ± 0.4	5.1 ± 0.4	0.428
HOMA-IR^b	2.2 (1.7–3.2)	2.8 (2.0–3.4)	0.320
C-reactive protein (mg L^−1)^b	0.35 (0.21–0.67)	0.44 (0.21–0.82)	0.534
IL-6 (pmol L^−1)^b	0.94 (0.59–1.17)	0.91 (0.61–1.54)	0.698
Plasma acetate (μmol L^−1)	88 ± 60	95 ± 76	0.690
Plasma propionate (μmol L^−1)	1.0 ± 0.5	1.1 ± 0.5	0.499
Plasma butyrate (μmol L^−1)	0.8 ± 0.6	0.9 ± 0.8	0.679
Fecal acetate (μmol g^−1)	161 ± 84	151 ± 59	0.654
Fecal propionate (μmol g^−1)	50 ± 22	57 ± 38	0.457
Fecal butyrate (μmol g^−1)	65 ± 36	54 ± 25	0.190
MVPA (min d^−1)^b	60 (41–146)	66 (36–136)	0.828
Screen time (min d^−1)^b	60 (30–97)	56 (27–133)	0.748
Education ≥master's degree (n, %)^c	11, 44	12, 46	1.000
Total energy intake (MJ per day)	7.9 ± 1.8	7.8 ± 1.9	0.886
Dietary fat intake (E%)	33 ± 6	31 ± 4	0.219
Dietary protein intake (E%)	16 ± 2	15 ± 2	0.390
Dietary carbohydrate intake (E%)	51 ± 6	54 ± 4	0.127
Dietary fiber intake (g d^−1)	20.6 ± 7.0	17.0 ± 7.3	0.080

---

The effect of WG versus RG on BMI z-scores, FMI, waist circumference (P_interaction ≤ 0.002) and HDL cholesterol (P_interaction = 0.030) differed between Cluster 1 and Cluster 2 (Table 2). All measures of obesity decreased among the children in Cluster 1 following WG intervention, while the children in Cluster 2 exhibited increases in the adiposity measures and increasing HDL cholesterol (Table 2 and Fig. 2). A significant increase in faecal butyrate with WG compared to RG was only observed for children in Cluster 1 and this was supported by a tendency of effect modification by cluster (P_interaction = 0.09) (Table 2). In concurrence, children in Cluster 1 showed increased plasma SCFAs (acetate, propionate and butyrate) following the WG intervention, whereas this was not observed in Cluster 2 (Table 2).

Fig. 2 Difference in adiposity measures and HDL cholesterol with wholegrain versus refined grain intervention stratified for Cluster 1 and Cluster 2. Differences are shown as median (25th–75th percentiles) wholegrain response relative to the response to refined grain with a red no-difference reference line.

Table 2 Estimated treatment effects (mean difference, 95% CI) in body composition, cardiometabolic risk markers and SCFAs stratified by microbial cluster and test for interaction between the treatment and microbial cluster

		Cluster 1		Cluster 2		Treatment × cluster
		WG–RG^a	P-Value^a	WG–RG^a	P-Value^a	P-Value^b
Bold indicates p-values of 0.05 or below. N = 51 for all analysis. RG, refined grain; WG, wholegrain.a Estimates and p-values from mixed model with treatment × cluster interaction, baseline value and homeschooling as fixed effects and subject as random effect.b P-Values from F-test comparing model with treatment × cluster interaction, baseline value and homeschooling as fixed effects and subject as random effect with a similar model without an interaction term.c Median (25^th–75^th percentiles). Model estimates and confidence intervals were back-transformed to their original scale.
Anthropometry	BMI (z-score)	−0.06 (−0.11, −0.01)	0.022	0.06 (0.01, 0.11)	0.04	0.002
	Fat mass index (kg m^−2)	−0.30 (−0.48, −0.12)	0.002	0.24 (0.07, 0.42)	0.01	<0.001
	Waist circumference (cm)	−0.89 (−1.73, −0.06)	0.041	1.21 (0.39, 2.03)	0.006	<0.001
Blood pressure	Diastolic blood pressure (mmHg)	−0.25 (−1.52, 1.01)	0.696	−0.53 (−1.77, 0.71)	0.410	0.758
	Systolic blood pressure (mmHg)	−0.37 (−2.11, 1.36)	0.675	−1.321 (−3.02, 0.38)	0.134	0.437
Blood lipids	LDL cholesterol (mmol L^−1)	−0.09 (−0.23, 0.06)	0.240	−0.19 (−0.33, −0.04)	0.013	0.330
	HDL cholesterol (mmol L^−1)	−0.01 (−0.08, 0.05)	0.711	0.09 (0.03, 0.15)	0.009	0.030
	Total cholesterol (mmol L^−1)	−0.18 (−0.37, 0.02)	0.084	−0.09 (−0.28, 0.11)	0.397	0.504
	Total:HDL cholesterol	−0.13 (−0.28, 0.02)	0.104	−0.28 (−0.43, −0.13)	<0.001	0.158
	Triacylglycerol (mmol L^−1)^c	−0.06 (−0.21, 0.08)	0.377	−0.15 (−0.28, −0.01)	0.045	0.411
Glucose metabolism	Insulin (pmol L^−1)^c	2.32 (−11.65, 16.28)	0.747	−4.09 (−17.84, 9.66)	0.563	0.514
	Glucose (mmol L^−1)	−0.03 (−0.16, 0.10)	0.643	−0.08 (−0.21, 0.05)	0.213	0.567
Inflammatory markers	C-reactive protein (mg L^−1)^c	−0.14 (−0.33, 0.05)	0.154	−0.13 (−0.32, 0.06)	0.178	0.964
	Interleukin-6 (pmol L^−1)^c	0.08 (−0.16, 0.32)	0.526	−0.11 (−0.35, 0.12)	0.352	0.257
Short-chain fatty acids	Plasma acetate (μmol L^−1)	41 (11, 70)	0.008	20 (−10, 49)	0.196	0.297
	Plasma propionate (μmol L^−1)	0.26 (0.06, 0.46)	0.015	0.04 (−0.16, 0.24)	0.698	0.134
	Plasma butyrate (μmol L^−1)	0.42 (0.14, 0.70)	0.004	0.26 (−0.02, 0.54)	0.067	0.417
	Fecal acetate (μmol g^−1)	13 (−26, 52)	0.524	−2.1 (−41, 37)	0.916	0.588
	Fecal propionate (μmol g^−1)	0 (−11, 11)	0.995	−7 (−18, 3)	0.197	0.350
	Fecal butyrate (μmol g^−1)	27 (12, 42)	<0.001	9 (−5, 24)	0.220	0.090

---

Among Danish school children aged 8–13 with high BMI, we identified two gut microbial clusters that remained stable after the start of the intervention. The clustering appeared to be driven by taxa indicative of a shift between two groups of SCFA-producing bacteria. Cluster 1 was characterized by higher relative abundance of the butyrate-producing Faecalibacterium whereas Cluster 2 tended to be less diverse and was dominated by higher relative abundance of Clostridium cluster XI and Clostridium sensu stricto. Intriguingly, the identified clusters modified the effect of wholegrain on BMI, FMI, and waist circumference as only children in Cluster 1 had reduced adiposity measures in response to WG. Furthermore, cluster modified the effect of WG on HDL cholesterol, with raised concentrations following WG among children in Cluster 2 only. Together, these results suggest that differences in microbiome structures could affect metabolic responses to high wholegrain intake in children.

A previous study of 281 Dutch 6–10-year-olds found that the children could be divided into three clusters or enterotypes dominated by Bifidobacterium, Bacteroides and Prevotella, respectively. The Bifidobacterium enterotype was characterized by lower α-diversity compared to other clusters.³⁰ In the present study, which focused on slightly older children with a high habitual intake of wholegrains, we identified only two distinct clusters. The lower alpha-diversity and increased abundance of Clostridium sensu stricto in Cluster 2 could reflect an immature and less adult-like gut microbiota, since this taxon comprises a pronounced butyrate-producing member of the infant gut microbiota.⁴³ Cluster 1 showed higher levels of Faecalibacterium, which is a prominent butyrate-producing taxon that increases with age⁴³ and pre-dominates the adult gut microbiota.^44,45 While no age difference was observed between clusters, children in Cluster 1 showed a tendency of higher dietary fibre intake. Dietary fibre has been shown to be a key factor in the transition towards a more mature adult-like microbiota,⁴⁶ which rather than reaching a stable state around 3 years of age as previously believed, may continue to develop into childhood and adolescence.^29,47

The reduction in body fatness in response to WG among children in Cluster 1 is in line with data from adults, where gut microbial enterotypes favouring the dietary fibre-associated taxa, e.g. Prevotella, have been shown to predict wholegrain induced weight loss.^26,27 Similarly, more mature gut microbiota compositions with a high dietary fibre-degrading potential have been inversely associated with markers of insulin resistance in both Dutch³⁰ and Mexican children.⁴⁸ Although in the original study, butyrate was linked to LDL cholesterol, the reduction in body fatness in Cluster 1 in the present analysis may be due to the higher butyrate-producing capacity of Faecalibacterium relative to the capacity of Clostridium sensu stricto.⁴³ Although we found no significant difference in faecal butyrate between clusters at the baseline, there was a trend of effect modification on faecal butyrate in the same direction as the obesity measures. Furthermore, significant changes in both plasma butyrate and faecal butyrate were observed in Cluster 1, suggesting that these children particularly responded to the intervention. Butyrate has been linked to induced thermogenesis, fat oxidation,⁴⁹ and release of appetite-regulating hormones in response to wholegrains.⁵⁰ Moreover, findings from a 6-month clinical trial in children with obesity suggest that oral butyrate supplementation may lower BMI and have beneficial effects on glucose metabolism and inflammation.⁵¹ However, this warrants further investigation in mechanistic studies or studies with better power for stratified analyses. Combined with previous findings, our results suggest that some beneficial responses to wholegrains may be more pronounced in children with a more mature gut microbiota composition, likely due a greater capacity to degrade dietary fibres. Notably, we observed an increased relative abundance of Faecalibacterium in response to WG in the original study,¹⁷ highlighting the potential of high wholegrain consumption to promote gut microbiome maturation towards a more metabolically advantageous composition during mid-childhood.

In contrast, children in Cluster 2 showed improved HDL cholesterol in response to WG. Noticeably, children in Cluster 2 had higher LDL cholesterol at the baseline compared to children in Cluster 1, which might pre-dispose this group to a greater potential for improvements in the lipid profile in response to the WG treatment, as previously observed in wholegrain trials.⁹

The study had a very low drop-out rate and high compliance based on self-reported intake and plasma biomarkers of WG intake. The cross-over design is a strength since it can efficiently be used to compare relatively small treatment effects by eliminating any between-individual variance due to differences in genetics and background diets. Another strength of our study is the use of gold standard outcome measurements such as DXA scans, fasting blood samples and robust faecal microbiome analysis. Children up to 4–6 years of age have previously been shown to exhibit greater microbial plasticity than adults, indicating that enterotypes may not remain stable during a dietary intervention.⁵² Nevertheless, the observed microbial clusters remained consistent after initiating the grain intervention. This apparent stability in the present study may be because most participants were aged 8–13 years, a period beyond the early developmental window typically associated with enterotype establishment. Alternatively, the relatively short duration of the intervention was insufficient to induce a shift in enterotype composition. However, even a 6-month dietary intervention with more fruits, vegetables and wholegrains was insufficient to alter enterotypes in adults.³⁷ The interaction between treatment and baseline gut microbial clusters may introduce residual confounding in the analyses due to other host traits associated with these clusters, such as unmeasured clinical baseline values. As the trial was not primarily designed for microbial cluster analysis, the reduced statistical power resulting from data stratification and potential interaction with baseline microbiota composition may partly explain the attenuation or loss of some wholegrain intervention effects, observed in the original study,¹⁷ in this stratified analysis. Finally, there is a risk of false positive findings, which was partly mitigated through conservative correction for multiple testing in the microbiota data analyses. Moreover, the consistent direction of adiposity outcomes supports the validity of the current findings.

Conclusions

In conclusion, this exploratory study identified two distinct gut microbial clusters among Danish children aged 8–13 years with high BMI. These clusters appeared to modify the effects of wholegrain intervention, favouring children with relatively high levels of Faecalibacterium for reducing fat mass, and those enriched with taxa within the Clostridiales order for increasing HDL-C. While the findings need to be validated to establish causality, the observations are intriguing and provide a rationale for investigating the possible pivotal role of gut microbiome maturation during childhood in diet-induced cardiometabolic responses.

Author contributions

M. B. M. M. helped plan and conducted the study, analysed data, wrote the paper and had primary responsibility for the final content. C. T. D. and R. L. designed the study. D. S. N. and Y. Z. carried out gut microbiota characterization and bioinformatics. C. T. D., D. S. N., H. M. R., L. C., and L. L. contributed with writing the manuscript and the interpretation of results. All authors read and approved the final manuscript.

Conflicts of interest

None of the authors reported a conflict of interest related to the study.

Abbreviations

AR	Alkylresorcinols
B	Baseline
FMI	Fat mass index
RG	Refined grain
WG	Wholegrain

Data availability

Data for the article is collected from human participants and cannot be made available due to legal and ethical confidentiality requirements.

Acknowledgements

We would like to thank all the children and their families who participated in the study.

The work was supported by the Independent Research Fund Denmark|Health and disease, the Swedish Research Council and the Chinese Scholarship Council. Study products were provided by Lantmännen, Semper, Il Fornaio, REMA 1000, Barilla, Mejnerts Mølle and Skærtoft Mølle. The funding sources had no involvement in the study design, data collection, analysis or interpretation or publication.

References

1. D. Aune, T. Norat, P. Romundstad and L. J. Vatten, Whole grain and refined grain consumption and the risk of type 2 diabetes: A systematic review and dose–response meta-analysis of cohort studies, Eur. J. Epidemiol., 2013, 28, 845–858.
2. D. Aune, N. Keum, E. Giovannucci, L. T. Fadnes, P. Boffetta, D. C. Greenwood, S. Tonstad, L. J. Vatten, E. Riboli and T. Norat, Whole grain consumption and risk of cardiovascular disease, cancer, and all cause and cause specific mortality: Systematic review and dose–response meta-analysis of prospective studies, Br. Med. J., 2016, 353, i2716.
3. O. Sadeghi, M. Sadeghian, S. Rahmani, V. Maleki, B. Larijani and A. Esmaillzadeh, Whole-grain consumption does not affect obesity measures: An updated systematic review and meta-analysis of randomized clinical trials, Adv. Nutr., 2020, 11, 280–292.
4. K. Pol, R. Christensen, E. M. Bartels, A. Raben, I. Tetens and M. Kristensen, Whole grain and body weight changes in apparently healthy adults: A systematic review and meta-analysis of randomized controlled studies, Am. J. Clin. Nutr., 2013, 98, 872–884.
5. P. L. Hollænder, A. B. Ross and M. Kristensen, Whole-grain and blood lipid changes in apparently healthy adults: A systematic review and meta-analysis of randomized controlled studies, Am. J. Clin. Nutr., 2015, 102, 556–572.
6. Z. Li, H. Yan, L. Chen, Y. Wang, J. Liang, X. Feng, S. Hui and K. Wang, Effects of whole grain intake on glycemic control: A meta-analysis of randomized controlled trials, J. Diabetes Invest., 2022, 13, 1814–1824.
7. S. Marventano, C. Vetrani, M. Vitale, J. Godos, G. Riccardi and G. Grosso, Whole grain intake and glycaemic control in healthy subjects: A systematic review and meta-analysis of randomized controlled trials, Nutrients, 2017, 9, 19.
8. A. Reynolds, J. Mann, J. Cummings, N. Winter, E. Mete and L. Te Morenga, Carbohydrate quality and human health: A series of systematic reviews and meta-analyses, Lancet, 2019, 393, 434–445.
9. A. Whitehead, E. J. Beck, S. Tosh and T. M. Wolever, Cholesterol-lowering effects of oat β-glucan: A meta-analysis of randomized controlled trials, Am. J. Clin. Nutr., 2014, 100, 1413–1421.
10. A. K. Eriksen, C. Brunius, M. Mazidi, P. M. Hellström, U. Risérus, K. N. Iversen, R. Fristedt, L. Sun, Y. Huang, N. P. Nørskov, K. E. B. Knudsen, C. Kyrø, A. Olsen, A. Tjønneland, J. Dicksved and R. Landberg, Effects of whole-grain wheat, rye, and lignan supplementation on cardiometabolic risk factors in men with metabolic syndrome: A randomized crossover trial, Am. J. Clin. Nutr., 2020, 111, 864–876.
11. J. Suhr, S. Vuholm, K. N. Iversen, R. Landberg and M. Kristensen, Wholegrain rye, but not wholegrain wheat, lowers body weight and fat mass compared with refined wheat: A 6-week randomized study, Eur. J. Clin. Nutr., 2017, 71, 959–967.
12. S. Vuholm, D. S. Nielsen, K. N. Iversen, J. Suhr, P. Westermann, L. Krych, J. R. Andersen and M. Kristensen, Whole-grain rye and wheat affect some markers of gut health without altering the fecal microbiota in healthy overweight adults: A 6-week randomized trial, J. Nutr., 2017, 147, 2067–2075.
13. P. Hajihashemi, L. Azadbakht, M. Hashemipor, R. Kelishadi and A. Esmaillzadeh, Whole-grain intake favorably affects markers of systemic inflammation in obese children: A randomized controlled crossover clinical trial, Mol. Nutr. Food Res., 2014, 58, 1301–1308.
14. P. Hajihashemi, L. Azadbakht, M. Hashemipour, R. Kelishadi, P. Saneei and A. Esmaillzadeh, Whole grain intake favorably affects blood glucose and serum triacylglycerols in overweight and obese children: A randomized controlled crossover clinical trial, Nutrition, 2021, 87–88, 111200.
15. P. Hajihashemi, L. Azadbakht, M. Hashemipour, R. Kelishadi, P. Saneei and A. Esmaillzadeh, The effects of whole grain intake on anthropometric measures in overweight and obese children: A crossover randomised clinical trial, Br. J. Nutr., 2021, 126, 1459–1465.
16. H. C. Koo, B. K. Poh and A. T. Ruzita, Intervention on whole grain with healthy balanced diet to manage childhood obesity (GReat-Child™trial): Study protocol for a quasi-experimental trial, SpringerPlus, 2016, 5, 840.
17. M. T. B. Madsen, R. Landberg, D. S. Nielsen, Y. Zhang, O. M. R. Anneberg, L. Lauritzen and C. T. Damsgaard, Effects of wholegrain compared to refined grain intake on cardiometabolic risk markers, gut microbiota and gastrointestinal symptoms in children: A randomized crossover trial, Am. J. Clin. Nutr., 2024, 119, 18–28.
18. S. M. Gråsten, K. S. Juntunen, K. S. Poutanen, H. K. Gylling, T. A. Miettinen and H. M. Mykkänen, Rye bread improves bowel function and decreases the concentrations of some compounds that are putative colon cancer risk markers in middle-aged women and men, J. Nutr., 2000, 130, 2215–2221.
19. G. H. McIntosh, M. Noakes, P. J. Royle and P. R. Foster, Whole-grain rye and wheat foods and markers of bowel health in overweight middle-aged men, Am. J. Clin. Nutr., 2003, 77, 967–974.
20. K. Iversen, J. Dicksved, C. Zoki, R. Fristedt, E. Pelve, M. Langton and R. Landberg, The effects of high fiber rye, compared to refined wheat, on gut microbiota composition, plasma short chain fatty acids, and implications for weight loss and metabolic risk factors (the RyeWeight Study), Nutrients, 2022, 14, 1669.
21. M. Müller, M. A. Gonzalez Hernandez, G. Goossens, D. Reijnders, J. Holst, J. Jocken, H. Eijk, E. Canfora and E. Blaak, Circulating but not faecal short-chain fatty acids are related to insulin sensitivity, lipolysis and GLP-1 concentrations in humans, Sci. Rep., 2019, 9, 1–9.
22. J. Jocken, M. A. Gonzalez Hernandez, N. Hoebers, C. van der Beek, Y. Essers, E. Blaak and E. Canfora, Short-chain fatty acids differentially affect intracellular lipolysis in a human white adipocyte model, Front. Endocrinol., 2018, 8, 372.
23. M. C. P. Cleophas, J. M. Ratter, S. Bekkering, J. Quintin, K. Schraa, E. S. Stroes, M. G. Netea and L. A. B. Joosten, Effects of oral butyrate supplementation on inflammatory potential of circulating peripheral blood mononuclear cells in healthy and obese males, Sci. Rep., 2019, 9, 775.
24. G. den Besten, A. Bleeker, A. Gerding, K. van Eunen, R. Havinga, T. H. van Dijk, M. H. Oosterveer, J. W. Jonker, A. K. Groen, D. J. Reijngoud and B. M. Bakker, Short-chain fatty acids protect against high-fat diet-induced obesity via a PPARγ-dependent switch from lipogenesis to fat oxidation, Diabetes, 2015, 64, 2398–2408.
25. S. A. Kelly, L. Hartley, E. Loveman, J. L. Colquitt, H. M. Jones, L. Al-Khudairy, C. Clar, R. Germano, H. R. Lunn, G. Frost and K. Rees, Whole grain cereals for the primary or secondary prevention of cardiovascular disease, Cochrane Database Syst. Rev., 2017, 8, CD005051.
26. L. Christensen, S. Vuholm, H. M. Roager, D. S. Nielsen, L. Krych, M. Kristensen, A. Astrup and M. F. Hjorth, Prevotella abundance predicts weight loss success in healthy, overweight adults consuming a whole-grain diet ad libitum: A post hoc analysis of a 6-wk randomized controlled trial, J. Nutr., 2019, 149, 2174–2181.
27. M. F. Hjorth, H. M. Roager, T. M. Larsen, S. K. Poulsen, T. R. Licht, M. I. Bahl, Y. Zohar and A. Astrup, Pre-treatment microbial Prevotella-to-Bacteroides ratio, determines body fat loss success during a 6-month randomized controlled diet intervention, Int. J. Obes., 2018, 42, 580–583.
28. P. Kovatcheva-Datchary, A. Nilsson, R. Akrami, Y. S. Lee, F. De Vadder, T. Arora, A. Hallen, E. Martens, I. Björck and F. Bäckhed, Dietary fiber-induced improvement in glucose metabolism is associated with increased abundance of Prevotella, Cell Metab., 2015, 22, 971–982.
29. A. Bergström, T. H. Skov, M. I. Bahl, H. M. Roager, L. B. Christensen, K. T. Ejlerskov, C. Mølgaard, K. F. Michaelsen and T. R. Licht, Establishment of intestinal microbiota during early life: a longitudinal, explorative study of a large cohort of Danish infants, Appl. Environ. Microbiol., 2014, 80, 2889–2900.
30. H. Zhong, J. Penders, Z. Shi, H. Ren, K. Cai, C. Fang, Q. Ding, C. Thijs, E. E. Blaak, C. D. A. Stehouwer, X. Xu, H. Yang, J. Wang, J. Wang, D. M. A. E. Jonkers, A. A. M. Masclee, S. Brix, J. Li, I. C. W. Arts and K. Kristiansen, Impact of early events and lifestyle on the gut microbiota and metabolic phenotypes in young school-age children, Microbiome, 2019, 7, 2.
31. Y. Ou, C. Belzer, H. Smidt and C. de Weerth, Development of the gut microbiota in healthy children in the first ten years of life: Associations with internalizing and externalizing behavior, Gut Microbes, 2022, 14, 2038853.
32. E. B. Hollister, K. Riehle, R. A. Luna, E. M. Weidler, M. Rubio-Gonzales, T. A. Mistretta, S. Raza, H. V. Doddapaneni, G. A. Metcalf, D. M. Muzny, R. A. Gibbs, J. F. Petrosino, R. J. Shulman and J. Versalovic, Structure and function of the healthy pre-adolescent pediatric gut microbiome, Microbiome, 2015, 3, 36.
33. R. Agans, L. Rigsbee, H. Kenche, S. Michail, H. J. Khamis and O. Paliy, Distal gut microbiota of adolescent children is different from that of adults, FEMS Microbiol. Ecol., 2011, 77, 404–412.
34. P. I. Costea, F. Hildebrand, M. Arumugam, F. Bäckhed, M. J. Blaser, F. D. Bushman, W. M. de Vos, S. D. Ehrlich, C. M. Fraser, M. Hattori, C. Huttenhower, I. B. Jeffery, D. Knights, J. D. Lewis, R. E. Ley, H. Ochman, P. W. O'Toole, C. Quince, D. A. Relman, F. Shanahan, S. Sunagawa, J. Wang, G. M. Weinstock, G. D. Wu, G. Zeller, L. Zhao, J. Raes, R. Knight and P. Bork, Enterotypes in the landscape of gut microbial community composition, Nat. Microbiol., 2018, 3, 8–16.
35. M. Arumugam, J. Raes, E. Pelletier, D. Le Paslier, T. Yamada, D. R. Mende, G. R. Fernandes, J. Tap, T. Bruls, J.-M. Batto, M. Bertalan, N. Borruel, F. Casellas, L. Fernandez, L. Gautier, T. Hansen, M. Hattori, T. Hayashi, M. Kleerebezem, K. Kurokawa, M. Leclerc, F. Levenez, C. Manichanh, H. B. Nielsen, T. Nielsen, N. Pons, J. Poulain, J. Qin, T. Sicheritz-Ponten, S. Tims, D. Torrents, E. Ugarte, E. G. Zoetendal, J. Wang, F. Guarner, O. Pedersen, W. M. de Vos, S. Brunak, J. Doré, M. Antolín, F. Artiguenave, H. M. Blottiere, M. Almeida, C. Brechot, C. Cara, C. Chervaux, A. Cultrone, C. Delorme, G. Denariaz, R. Dervyn, K. U. Foerstner, C. Friss, M. van de Guchte, E. Guedon, F. Haimet, W. Huber, J. van Hylckama-Vlieg, A. Jamet, C. Juste, G. Kaci, J. Knol, K. Kristiansen, O. Lakhdari, S. Layec, K. Le Roux, E. Maguin, A. Mérieux, R. Melo Minardi, C. M'Rini, J. Muller, R. Oozeer, J. Parkhill, P. Renault, M. Rescigno, N. Sanchez, S. Sunagawa, A. Torrejon, K. Turner, G. Vandemeulebrouck, E. Varela, Y. Winogradsky, G. Zeller, J. Weissenbach, S. D. Ehrlich, P. Bork and H. I. T. C. Meta, Enterotypes of the human gut microbiome, Nature, 2011, 473, 174–180.
36. G. D. Wu, J. Chen, C. Hoffmann, K. Bittinger, Y. Y. Chen, S. A. Keilbaugh, M. Bewtra, D. Knights, W. A. Walters, R. Knight, R. Sinha, E. Gilroy, K. Gupta, R. Baldassano, L. Nessel, H. Li, F. D. Bushman and J. D. Lewis, Linking long-term dietary patterns with gut microbial enterotypes, Science, 2011, 334, 105–108.
37. H. M. Roager, T. R. Licht, S. K. Poulsen, T. M. Larsen and M. I. Bahl, Microbial enterotypes, inferred by the prevotella-to-bacteroides ratio, remained stable during a 6-month randomized controlled diet intervention with the New Nordic Diet, Appl. Environ. Microbiol., 2014, 80, 1142–1149.
38. M. de Onis, A. W. Onyango, E. Borghi, A. Siyam, C. Nishida and J. Siekmann, Development of a WHO growth reference for school-aged children and adolescents, Bull. W. H. O., 2007, 85, 660–667.
39. J. L. Castro-Mejía, B. Khakimov, Ł. Krych, J. Bülow, R. L. Bechshøft, G. Højfeldt, K. H. Mertz, E. S. Garne, S. R. Schacht, H. F. Ahmad, W. Kot, L. H. Hansen, F. J. A. Perez-Cueto, M. V. Lind, A. J. Lassen, I. Tetens, T. Jensen, S. Reitelseder, A. P. Jespersen, L. Holm, S. B. Engelsen and D. S. Nielsen, Physical fitness in community-dwelling older adults is linked to dietary intake, gut microbiota, and metabolomic signatures, Aging Cell, 2020, 19, e13105.
40. O. Koren, D. Knights, A. Gonzalez, L. Waldron, N. Segata, R. Knight, C. Huttenhower and R. E. Ley, A guide to enterotypes across the human body: Meta-analysis of microbial community structures in human microbiome datasets, PLoS Comput. Biol., 2013, 9, e1002863.
41. R. P. Laursen, S. M. Dalskov, C. T. Damsgaard and C. Ritz, Back-transformation of treatment differences – an approximate method, Eur. J. Clin. Nutr., 2014, 68, 277–280.
42. A. Kuznetsova, P. B. Brockhoff and R. H. B. Christensen, lmerTest Package: Tests in linear mixed effects models, J. Stat. Softw., 2017, 82, 1–26.
43. O. Appert, A. R. Garcia, R. Frei, C. Roduit, F. Constancias, V. Neuzil-Bunesova, R. Ferstl, J. Zhang, C. Akdis, R. Lauener, C. Lacroix and C. Schwab, Initial butyrate producers during infant gut microbiota development are endospore formers, Environ. Microbiol., 2020, 22, 3909–3921.
44. M. Vital, A. Karch and D. H. Pieper, Colonic butyrate-producing communities in humans: An overview using omics data, mSystems, 2017, 2, e00130-17.
45. A. Rivière, M. Selak, D. Lantin, F. Leroy and L. De Vuyst, Bifidobacteria and butyrate-producing colon bacteria: Importance and strategies for their stimulation in the human gut, Front. Microbiol., 2016, 7, 979.
46. M. F. Laursen, L. B. Andersen, K. F. Michaelsen, C. Mølgaard, E. Trolle, M. I. Bahl and T. R. Licht, Infant gut microbiota development is driven by transition to family foods independent of maternal obesity, mSphere, 2016, 1, e00069-15.
47. C. J. Stewart, N. J. Ajami, J. L. O'Brien, D. S. Hutchinson, D. P. Smith, M. C. Wong, M. C. Ross, R. E. Lloyd, H. Doddapaneni, G. A. Metcalf, D. Muzny, R. A. Gibbs, T. Vatanen, C. Huttenhower, R. J. Xavier, M. Rewers, W. Hagopian, J. Toppari, A.-G. Ziegler, J.-X. She, B. Akolkar, A. Lernmark, H. Hyoty, K. Vehik, J. P. Krischer and J. F. Petrosino, Temporal development of the gut microbiome in early childhood from the TEDDY study, Nature, 2018, 562, 583–588.
48. J. N. Martinez-Medina, R. Flores-Lopez, B. E. López-Contreras, H. Villamil-Ramirez, D. Guzman-Muñoz, L. R. Macias-Kauffer, P. León-Mimila, O. Granados-Portillo, B. E. Del-Rio-Navarro, F. J. Gómez-Perez, C. A. Aguilar-Salinas, N. Torres, A. R. Tovar, S. Canizales-Quinteros and S. Moran-Ramos, Effect of gut microbial enterotypes on the association between habitual dietary fiber intake and insulin resistance markers in Mexican children and adults, Nutrients, 2021, 13, 3892.
49. Z. Li, C. X. Yi, S. Katiraei, S. Kooijman, E. Zhou, C. K. Chung, Y. Gao, J. K. van den Heuvel, O. C. Meijer, J. F. P. Berbée, M. Heijink, M. Giera, K. Willems van Dijk, A. K. Groen, P. C. N. Rensen and Y. Wang, Butyrate reduces appetite and activates brown adipose tissue via the gut-brain neural circuit, Gut, 2018, 67, 1269–1279.
50. P. Amiri, S. A. Hosseini, S. Ghaffari, H. Tutunchi, S. Ghaffari, E. Mosharkesh, S. Asghari and N. Roshanravan, Role of butyrate, a gut microbiota derived metabolite, in cardiovascular diseases: A comprehensive narrative review, Front. Pharmacol., 2022, 12, 837509.
51. S. Coppola, R. Nocerino, L. Paparo, G. Bedogni, A. Calignano, C. Di Scala, A. F. de Giovanni di Santa Severina, F. De Filippis, D. Ercolini and R. Berni Canani, Therapeutic effects of butyrate on pediatric obesity: A randomized clinical trial, JAMA Network Open, 2022, 5, e2244912.
52. K. V. Ruggles, J. Wang, A. Volkova, M. Contreras, O. Noya-Alarcon, O. Lander, H. Caballero and M. G. Dominguez-Bello, Changes in the gut microbiota of urban subjects during an immersion in the traditional diet and lifestyle of a rainforest village, mSphere, 2018, 3, e00193-18.

---