Association between Gut Microbiota and Inflammation: Mediation Analysis Using Waist Circumference

Abstract

Chronic low-grade inflammation (CLI) is implicated in the development of multiple metabolic diseases. The gut microbiota (GM) activates different signaling pathways and induces phenotypic changes, offering an exciting opportunity to treat CLI. We evaluated the mediation of waist circumference on the association of GM with serum cytokines. In this cross-sectional study of 331 children, we measured 5 gut bacterial species, namely, Lactobacillus (L.) casei, L. paracasei, L. reuteri, Staphylococcus (S.) aureus, and Akkermansia (A.) muciniphila, as well as anthropometry, serum cytokines, and other covariates. We evaluated adjusted regression models, path analysis, and structural equation modeling to obtain path coefficients (PCs) for direct, indirect (waist circumference-mediated), and total effects. We found that L. paracasei was directly associated with lower interleukin-10 (IL-10) levels (PC = −173.5 pg/mL). We also observed indirect associations between S. aureus with lower adiponectin levels (PC = −0.1 µg/mL and −0.09 µg/mL). Finally, A. muciniphila was indirectly associated with higher adiponectin levels (PC = 0.1 µg/mL). Our findings suggest the importance of considering the GM composition and waist circumference when evaluating inflammatory-related factors, providing a basis for future research to identify potential strategies to intervene in inflammatory processes and prevent metabolic diseases in childhood.

Introduction

The gut microbiota (GM) is a dynamic ecosystem of microorganisms, primarily bacteria, residing in the human gastrointestinal tract (Jaswal et al., 2023). Its colonization and composition vary depending on diet, genetics, lifestyle, infections, and antibiotic use. These bacteria establish a symbiotic relationship with humans, playing a crucial role in health, especially in childhood, influencing various physiological processes, such as immune response, metabolism, and inflammation (Turroni et al., 2020; Yao et al., 2021). The symbiotic interaction depends on the balance and composition of bacterial communities. When this balance is disrupted, a phenomenon known as intestinal dysbiosis emerges, which has been linked to chronic inflammatory conditions, suggesting a potential causal relationship. Several studies support a significant association between the specific composition of the GM and various inflammation markers (Ling et al., 2022; Meng et al., 2022; Sun et al., 2020). The GM, as well as its metabolites and products, interact with the immune system through various pathways, influencing the response of T cells, B cells, dendritic cells, and macrophages (Yang and Cong, 2021). In addition to gut dysbiosis, waist circumference has emerged as a marker of central obesity that contributes significantly to metabolic dysfunction and low-grade inflammation (CLI) in children (Steene-Johannessen et al., 2010). It is proposed that central obesity could act as a mediator in the association between GM composition and inflammation. As adipocytes expand through hyperplasia or hypertrophy in response to excessive food consumption, these hypertrophic adipocytes secrete proinflammatory cytokines and reduce the expression of anti-inflammatory cytokines (Hildebrandt et al., 2023; Zhang et al., 2023). Together, GM and central obesity contribute to the alteration of the inflammatory response, causing immunological intolerance and affecting physiological processes in various tissues and organs, increasing the risk of developing chronic diseases (Furman et al., 2019). However, further research is needed to establish the exact mechanisms of mediation and causal relationships. Therefore, this study aimed to assess the mediation of waist circumference in the association between Akkermansia (A.) muciniphila, Lactobacillus (L.) casei, L. paracasei, L. reuteri, and Staphylococcus (S.) aureus with interleukin (IL)-6, IL-10, tumoral necrosis factor-alpha (TNF-α), and adiponectin levels in children.

Material and Methods

Design and study population

The data originated from a previous cross-sectional study conducted between 2012 and 2014 and that had received approval from the Ethics, Research, and Biosafety Commissions of the National Institute of Public Health, Mexico (Estrada-Velasco et al., 2014). All children and their parents provided written assent and consent. Unrelated children residing in the 4 zones of Mexico City (north, south, east, and west) were selected by nonprobabilistic sampling. Children diagnosed with infectious diseases or gastrointestinal disorders and who had taken antibiotics 2 months before the interview were excluded. Information from 331 individuals of age 6–12 with complete data on GM, anthropometry, and cytokine levels was analyzed.

Statistical power calculation

We used Alexander M. Schoemann’s methodology to calculate statistical power using the Monte Carlo method through the R programming language (Schoemann et al., 2017). We performed 5,000 simulations with a confidence level of 95%. Based on the results obtained, it was established that a minimum sample size of 230 individuals is needed to estimate indirect effects. In our case, the sample consists of 331 individuals, implying a low probability of not detecting a genuine association when it is present.

Gut microbiota

We collected a 200 mg of stool sample for DNA extraction using the QIAamp^® Fast DNA Stool Mini Kit (Qiagen, Germany). DNA concentration and purity were assessed using a Thermo Scientific^TM NanoDrop^TM Lite spectrophotometer (Madison, USA). We used quantitative polymerase chain reaction (qPCR) on the StepOnePlus^TM Real-Time PCR System (Woodlands, Singapore) to amplify a variable region of the ribosomal RNA of the 16S gene of A. muciniphila, L. casei, L. paracasei, L. reuteri, and S. aureus using specific and universal primers (Supplementary Table S1) (Bacchetti De Gregoris et al., 2011; Collado et al., 2007; Haarman and Knol, 2006). Each qPCR was conducted in duplicate with 5 μL of Maxima SYBR Green/ROX qPCR Master Mix (2X) Thermo Scientific^TM (Carlsbad, USA), 1 μL of the sense oligonucleotide, 1 μL of the antisense oligonucleotide, 2 μL of Fermentas^® nucleic acid-free water (Carlsbad, U.S.A.), and 1 μL of the DNA sample (5 ng/μL for S. aureus and 10 ng/μL for the other species), resulting in a final volume of 10 μL. We estimated the relative abundance (RA) using the comparative 2^−ΔCt method, where ΔCt represents the difference between the specific primer’s Ct and the universal primer’s Ct (Schmittgen and Livak, 2008).

Waist circumference

Waist circumference was measured using a flexible tape measure, using the lower edge of the last palpable rib and the upper edge of the iliac crest as equidistant reference points. Training professionals performed measurements following standardized techniques adapted to the pediatric population (Harrington et al., 2013).

Serum cytokines

Blood samples were collected through antecubital venipuncture after a 12-h fasting period. These samples were subsequently centrifuged to isolate the serum. The serum levels of IL-6, IL-10, TNF-α, and adiponectin were quantified using enzyme-linked immunoadsorption assay, following the instructions provided by the manufacturer (PeproTech, Rocky Hill, USA). The absorbance readings were conducted using a Labsystems Multiskan MS^® (Vantaa, Finland).

Data collection

An 18-question questionnaire was used to collect data on sex, age, and family history of overweight-obesity and type 2 diabetes (T2D). This questionnaire included socioeconomic, sociodemographic, and personal pathological-hereditary history information.

Physical activity

Physical activity data were collected using the Mexican Students’ Activity and Inactivity Questionnaire (CAINM). This questionnaire measures the daily hours of moderate or vigorous activity during the past month (Hernández et al., 2000).

Macronutrients intake

We collected dietary information using a food frequency questionnaire (FFQ). This semiquantitative questionnaire comprises questions on the frequency of intake in the past year. The children, accompanied by their parents, indicated the frequency of daily, weekly, or monthly consumption of each food item in the questionnaire. We calculated the average daily intake of carbohydrates, lipids, and proteins based on the reported frequency of each food. This average intake was multiplied by the equivalent grams constituting a standard portion based on the Mexican National Health and Nutrition Survey (ENSANUT, 2012) (Romero-Martínez et al., 2013).

Statistical analysis

For the exploratory data analysis, we categorized the children into 2 groups based on the median waist circumference within our population. We considered the proximity to established cutoff points by other researchers: a normal waist circumference (<63.6 cm) and a high waist circumference (≥63.6 cm) (Fredriksen et al., 2018; Yamanaka et al., 2021). We tested our data’s normality using the Shapiro–Wilk test. We estimated medians, 25th percentile, 75th percentile, and percentages for the variables of interest. The Mann–Whitney U statistic (for continuous variables) and chi-square (for categorical variables) were used to explore group differences. For the inferential analysis, we obtained the tertiles of RA of the 5 bacterial species (Supplementary Table S2), taking as reference the tertile of lowest risk for each species as described in the literature. We performed linear and logistic regression models to evaluate the association between GM, waist circumference, and cytokine levels. We used path analysis to assess waist circumference mediation to calculate path coefficients (PCs) for each bacterial species’ direct, indirect, and total effects on cytokine levels. Subsequently, latent variables (microbiota profiles) were generated to evaluate the joint effect of bacterial species that were significantly and marginally associated with cytokine levels in path analysis. Finally, structural equation modeling was used to examine waist circumference mediation, estimating the PC for direct, indirect, and total effects of microbiota profiles on adiponectin levels (Harris and Gleason, 2022). All models were adjusted for potential confounders identified in directed acyclic graphs (Supplementary Fig. S1). We found no significant variability in the estimator when we evaluated physical activity and family history of T2D as potential confounding variables in the statistical models. The minimal sufficient adjustment set includes age, sex, family history of overweight-obesity, and daily intake of carbohydrates, lipids, and proteins. Statistical significance was determined at P < 0.05. All analyses were executed using Stata^® version 17 software.

Results

We analyzed the general characteristics of 331 children, with an average age of 9 years, categorized according to waist circumference size. We estimated a 51% prevalence of high waist circumference. In the group of children with high waist circumference, we observed a higher median older age (10 years vs. 8 years), a higher prevalence of family history of overweight-obesity (56% vs. 45%), a higher RA of S. aureus (6 × 10⁻⁵ vs. 3 × 10⁻⁵), and lower levels of adiponectin (5 µg/mL vs. 6 µg/mL) compared with children with normal waist circumference. No differences were detected between the 2 groups in terms of physical activity distribution, sex, family history of T2D, daily macronutrient intake, and RA of L. paracasei, L. casei, L. reuteri, and A. muciniphila, as well as in IL-6, IL-10, and TNF-α levels (Table 1).

Table 1.

General Characteristics of Children Categorized by Waist Circumference Size

Characteristics	Normal waist circumference n = 162 (49%)	High waist circumference n = 169 (51%)	p values
Characteristics	Age (years)	8 (7–9)	p values	10 (8–11)	<0.001
Physical activity (METs/h/week)	291 (150–550)	342 (162–631)	0.174
Sex
Boy (%)	52	48	0.502
Girl (%)	49	51	0.502
Family history of overweight-obesity
Yes (%)	45	56	0.011
No (%)	56	47	0.011
Family history of type 2 diabetes
Yes (%)	40	60	0.054
No (%)	52	48	0.054
Macronutrient intake
Carbohydrates (g/day)	271 (207–350)	275 (213–345)	0.896
Lipids (g/day)	75 (56–94)	75 (60–93)	0.677
Proteins (g/day)	69 (53–85)	69 (54–86)	0.711
Exposure variables
Staphylococcus (S.) aureus (RA)	3.0 × 10⁻⁵ (7.38 × 10⁻⁶–2.16 × 10⁻⁴)	6.0 × 10⁻⁵ (1.25 × 10⁻⁵–2.77 × 10⁻⁴)	0.022
Lactobacillus (L.) paracasei (RA)	2.9 × 10⁻⁴ (2.36 × 10⁻⁵–2.71 × 10⁻³)	3.4 × 10⁻⁴ (2.26 × 10⁻⁵–4.52 × 10⁻³)	0.569
L. casei (RA)	3.7 × 10⁻⁴ (4.83 × 10⁻⁵–4.84 × 10⁻³)	5.5 × 10⁻⁴ (5.96 × 10⁻⁵–8.43 × 10⁻³)	0.247
L. reuteri (RA)	1.1 × 10⁻⁴ (1.80 × 10⁻⁵–5.73 × 10⁻⁴)	1.4 × 10⁻⁴ (2.16 × 10⁻⁵–1.05 × 10⁻³)	0.268
Akkermansia muciniphila (RA)	2.5 × 10⁻³ (2.92 × 10⁻⁵–5.01 × 10⁻²)	4.6 × 10⁻³ (6.32 × 10⁻⁵–8.43 × 10⁻²)	0.119
Outcome variables
IL-6 (pg/mL)	22 (7–73)	34 (10–104)	0.228
IL-10 (pg/mL)	58 (0.7–252)	54 (0.4–247)	0.882
TNF-α (pg/mL)	1,106 (748–1,533)	1,043 (803–1,451)	0.668
Adiponectin (µg/mL)	6 (4.9–6.5)	5 (4.2–5.7)	<0.001

Values are shown in median (25th percentile–75th percentile) or percentages. Differences between groups were calculated using the Mann–Whitney U test for continuous variables or the χ² test for categorical variables. RA, relative abundance; IL-6, interleukin 6; IL-10, interleukin 10; TNF-α, tumor necrosis factor alpha. Statistically significant differences p < 0.05 are marked in bold.

We performed logistic regression modeling to evaluate the association between tertiles of RA of the 5 bacterial species tested and waist circumference. We found that children with medium and high RA of S. aureus had higher odds of having a high waist circumference compared with children who had low RA [odds ration (OR), 2.3 (95% CI: 1.4, 3.6) and OR, 1.7 (95% CI: 1.1, 2.7)]. Children with low and medium RA of A. muciniphila had lower odds of having a high waist circumference compared with children who had high RA [OR, 0.6 (95% CI: 0.4, 0.9) and OR, 0.5 (95% CI: 0.3, 0.8)]. We did not find significant associations with the rest of the species (Supplementary Table S3).

We applied linear regression models to evaluate the association between bacterial RA and waist circumference with cytokine levels. We observed that children with low RA of L. paracasei had a reduction in IL-10 levels of 174.9 pg/mL, on average, compared with children with high RA [β = −174.9 (95% CI: −316.3, −33.5)]. Likewise, those with medium RA of A. muciniphila had an average increase in adiponectin levels of 0.3 µg/mL compared with children with high RA [β = 0.3 (95% CI: 0.03, 0.6)]. Finally, we noted that children with a high waist circumference had reduced adiponectin levels by 0.7 µg/mL, on average, compared with children with normal waist circumference [β = −0.7 (95% CI: −1.0, −0.5)]. We did not find significant associations with the rest of the species (Supplementary Table S4).

Table 2 shows the path coefficients of the direct, indirect (waist circumference-mediated), and total effects of bacterial RA on cytokine levels obtained by path analysis (Supplementary Fig. S2). We observed that children with low RA of L. paracasei had a reduction in IL-10 levels of 173.5 pg/mL, on average, compared with children with high RA [PC = −173.5 (95% CI: −312.4, −34.5)]. We found that waist circumference is a complete mediator in the association between S. aureus and A. muciniphila with adiponectin levels. In addition, we found that children with medium and high RA of S. aureus had a reduction in adiponectin levels of 0.1 µg/mL and 0.2 µg/mL than children with low RA, respectively [PC = −0.1 (95% CI: −0.2, −0.05) and PC= −0.09 (95% CI: −0.2, −0.01)]. Children with medium RA of A. muciniphila had an increase in adiponectin levels of 0.1 µg/mL compared with children with high RA [PC = 0.1 (95% CI: 0.03, 0.2)]. We did not find significant associations with the rest of the species.

Table 2.

Direct, Indirect, and Total Effects of Gut Microbiota on Cytokine Levels

Relative abundance	IL-6			IL-10			TNF-α			Adiponectin
Relative abundance	PC	95% CI	p values	PC	95% CI	p values	PC	95% CI	p values	PC	95% CI	p values
Direct effect
Staphylococcus (S.) aureus ^a
Medium tertile	11.5	−28.8, 51.9	0.574	42.2	−95.3, 179.8	0.548	5.8	−148.9, 160.6	0.941	0.1	−0.1, 0.4	0.298
High tertile	−15.7	−56.9, 25.5	0.455	32.7	−108.0, 173.5	0.649	36.6	−131.8, 205.1	0.670	0.08	−0.2, 0.3	0.533
Lactobacillus (L.) reuteri ^a
Medium tertile	5.3	−35.8, 46.4	0.800	18.9	−121.1, 158.9	0.792	49.5	−112.6, 211.5	0.594	−0.02	−0.3, 0.2	0.858
High tertile	−12.4	−52.4, 27.5	0.542	−22.7	−158.9, 113.5	0.744	32.6	−125.2, 190.5	0.685	−0.1	−0.4, 0.1	0.311
L. paracasei ^b
Low tertile	−10.1	−51.2, 31.0	0.631	−173.5	−312.4, −34.5	0.014	−147.2	−310.2, 15.7	0.077	−0.008	−0.3, 0.2	0.950
Medium tertile	11.3	−29.9, 52.5	0.593	−100.3	−239.8, 39.2	0.159	−149.0	−313.1, 15.0	0.075	0.07	−0.2, 0.3	0.573
L. casei ^b
Low tertile	4.9	−36.4, 46.2	0.816	−52.7	−193.1, 87.8	0.462	−73.2	−234.3, 87.8	0.373	−0.06	−0.3, 0.2	0.654
Medium tertile	−0.4	−42.8, 41.9	0.984	−51.9	−196.1, 92.2	0.480	−146.6	−313.2, 19.9	0.084	0.1	−0.1, 0.4	0.386
Akkermansia (A.) muciniphila ^b
Low tertile	29.1	−12.7, 70.9	0.173	46.2	−96.9, 189.3	0.527	−65.1	−231.9, 101.7	0.444	0.1	−0.1, 0.4	0.329
Medium tertile	−10.7	−50.7, 29.3	0.600	−19.9	−156.8, 116.9	0.775	−16.6	−172.6, 139.3	0.834	0.2	−0.1, 0.4	0.144
Indirect effect (waist circumference-mediated)
S. aureus ^a
Medium tertile	4.8	−2.8, 12.4	0.217	7.5	−16.9, 32.0	0.546	−1.1	−33.3, 31.1	0.947	−0.1	−0.2, −0.05	0.002
High tertile	2.6	−2.3, 7.4	0.296	4.0	−9.7, 17.8	0.564	−0.5	−14.6, 13.6	0.947	−0.09	−0.2, −0.01	0.028
L. reuteri ^a
Medium tertile	0.8	−2.6, 4.2	0.647	1.3	−5.3, 7.9	0.688	−0.08	−2.7, 2.6	0.952	−0.03	−0.1, 0.04	0.419
High tertile	−0.2	−3.3., 2.9	0.895	−0.4	−5.8, 5.1	0.897	0.04	−1.5, 1.6	0.954	−0.03	−0.1, 0.04	0.389
L. paracasei ^b
Low tertile	−0.9	−4.5, 2.6	0.601	−1.4	−7.8, 5.1	0.671	1.9	−12.1, 16.0	0.786	0.03	−0.04, 0.1	0.375
Medium tertile	−2.1	−6.5, 2.2	0.341	−3.1	−13.9, 7.6	0.564	2.5	−15.5, 20.6	0.784	0.02	−0.05, 0.1	0.611
L. casei ^b
Low tertile	−2.4	−6.9, 2.21	0.309	−3.6	−15.7, 8.6	0.563	3.0	−14.8, 20.9	0.739	0.04	−0.03, 0.1	0.264
Medium tertile	−4.0	−10.4, 2.4	0.220	−6	−25.3, 13.3	0.542	4.8	−23.0, 32.6	0.736	0.07	−0.002, 0.1	0.058
A. muciniphila ^b
Low tertile	−1.9	−6.0, 2.2	0.374	−3.2	−13.9, 7.4	0.551	0.3	−7.1, 7.6	0.943	0.07	−0.0005, 0.1	0.052
Medium tertile	−3.5	−9.4, 2.33	0.237	−6.1	−24.2, 11.9	0.507	0.5	−14.4, 15.5	0.943	0.1	0.03, 0.2	0.008
Total effect
S. aureus ^a
Medium tertile	16.4	−23.4, 56.2	0.420	49.7	−85.7, 185.2	0.472	4.7	−146.6, 156.1	0.951	0.001	−0.3, 0.3	0.994
High tertile	−13.2	−54.3, 28.0	0.532	36.8	−103.4, 177.0	0.607	36.2	−131.7, 204.1	0.673	−0.007	−0.3, 0.3	0.957
L. reuteri ^a
Medium tertile	6.1	−35.1, 47.3	0.772	20.2	−119.8, 160.3	0.777	49.4	−112.6, 211.4	0.550	−0.05	−0.3, 0.2	0.691
High tertile	−12.6	−52.7, 27.4	0.537	−23.1	−159.4, 113.2	0.740	32.7	−125.2, 190.5	0.685	−0.2	−0.4, 0.1	0.224
L. paracasei ^b
Low tertile	−11.0	−52.2, 30.2	0.600	−174.9	−313.8, −35.9	0.014	−145.3	−307.7, 17.1	0.079	0.03	−0.2, 0.3	0.850
Medium tertile	9.1	−32.3, 50.4	0.664	−103.4	−242.7, 35.8	0.145	−146.5	−309.7, 16.6	0.078	0.1	−0.2, 0.3	0.494
L. casei ^b
Low tertile	2.5	−38.8, 43.8	0.905	−56.2	−196.4, 83.9	0.431	−70.2	−230.3, 89.9	0.390	−0.02	−0.3, 0.2	0.905
Medium tertile	−4.4	−46.6, 37.7	0.836	−57.9	−201.0, 85.1	0.427	−141.8	−306.1, 22.4	0.091	0.2	−0.1, 0.5	0.167
A. muciniphila ^b
Low tertile	27.2	−14.6, 69.1	0.203	42.9	−99.9, 185.9	0.555	−64.8	−231.5, 101.8	0.446	0.2	−0.1, 0.5	0.135
Medium tertile	−14.2	−54.1, 25.5	0.483	−26.1	−161.9, 109.8	0.707	−16.1	−171.4, 139.2	0.839	0.3	0.03, 0.6	0.027

Values show path coefficients (PC) and their confidence intervals (CI). Path analysis adjusted for age, sex, family history of overweight-obesity, and carbohydrate, lipid, and protein intake. IL-6, interleukin 6; IL-10, interleukin 10; TNF-α, tumor necrosis factor alpha. Statistically significant differences p < 0.05 are marked in bold. Reference group: ^aLow tertile, ^bHigh tertile.

Finally, we explored whether waist circumference mediates the joint effect of microbiota profiles on adiponectin levels. Table 3 presents the path coefficients obtained by structural equation modeling (Supplementary Fig. S3). The results reveal higher adiponectin levels when the RA of L. casei and A. muciniphila predominate over the RA of S. aureus [PC = 0.4 µg/mL (95% CI: 0.1, 0.6) and PC = 0.7 µg/mL (95% CI: 0.04, 1.4)]. Conversely, lower adiponectin levels are observed when the RA of S. aureus is predominant over the RA of L. casei and A. muciniphila [PC = −0.08 (95% CI: −0.1, −0.01)]. Our analysis also showed that waist circumference is a complete mediator when the RA of L. casei and A. muciniphila predominate. When the RA of S. aureus predominates, waist circumference acts as a partial mediator. We did not find significant associations with the rest of the GM profiles.

Table 3.

Direct, Indirect, and Total Effects of Gut Microbiota Profiles on Adiponectin Levels

Gut microbiota	Adiponectin
Gut microbiota	PC	95% CI	p values
Direct Effect
Profile 1	−0.03	−0.7, 0.6	0.938
Profile 2	0.1	−1.5, 1.8	0.854
Profile 3	−0.1	−0.4, 0.2	0.513
Profile 4	−0.1	−0.6, 0.4	0.711
Profile 5	−0.3	−0.5, −0.07	0.009
Profile 6	0.03	−0.2, 0.2	0.794
Indirect effect (waist circumference-mediated)
Profile 1	0.4	0.1, 0.6	0.005
Profile 2	0.7	0.04, 1.4	0.039
Profile 3	0.1	−0.1, 0.3	0.270
Profile 4	−0.1	−0.3, 0.04	0.132
Profile 5	−0.08	−0.1, −0.01	0.018
Profile 6	−0.06	−0.13, 0.005	0.069
Total effect
Profile 1	0.3	−0.4, 1.0	0.345
Profile 2	0.8	−0.9, 2.7	0.340
Profile 3	0.02	-0.2, 0.2	0.857
Profile 4	−0.2	−0.8, 0.3	0.412
Profile 5	−0.4	−0.6, −0.1	0.001
Profile 6	−0.03	−0.3, 0.2	0.783

Values show path coefficients (PC) and their confidence intervals (CI). Structural equation modeling adjusted for age, sex, family history of overweight-obesity, and carbohydrate, lipid, and protein intake. Statistically significant differences p < 0.05 are marked in bold.

Profile 1: Low Staphylococcus (S.) aureus, High Lactobacillus (L) casei, and High Akkermansia (A.) muciniphila.

Profile 2: Low S. aureus, Medium L. casei, and Medium A. muciniphila.

Profile 3: Low S. aureus, Low L. casei, and Low A. muciniphila.

Profile 4: High S. aureus, High L. casei, and High A. muciniphila.

Profile 5: High S. aureus, Medium L. casei, and Medium A. muciniphila.

Profile 6: High S. aureus, Low L. casei, and Low A. muciniphila.

Discussion

Our results provide evidence of the effect of waist circumference on the association between GM and cytokine levels. Before our mediation analysis, we explored the association between some bacterial species and waist circumference. We found that S. aureus was associated with increased odds of having a high waist circumference. Several studies have reported higher levels of S. aureus in overweight or obese children than in normal-weight children (Kalliomäki et al., 2008). Even from the first days after birth, it has been observed that male infants with the presence of Staphylococcus spp. in feces had higher birth weights compared with infants in whom this bacterium was not detected. This finding identifies possible trajectories toward obesity in the early stages of life (Collado et al., 2010; White et al., 2013).

Similar results have been reported regarding our findings on the association between A. muciniphila and lower odds of having a high waist circumference. A study examining GM and health in individuals from the United States and the United Kingdom observed that those with a higher presence of A. muciniphila had up to 22% lower risk of being obese than those of normal weight (Zhou et al., 2020). In addition, when A. muciniphila supplementation was provided to overweight or obese individuals over 3 months, an effect on decreasing total body weight, fat mass, and hip circumference was observed compared with those who did not receive the treatment (Depommier et al., 2021).

An important finding that we want to highlight is the association between L. paracasei and lower levels of IL-10. Lactobacillus paracasei has been used as a probiotic because of its anti-inflammatory capacity and involvement in reducing proinflammatory cytokine expression (Choi et al., 2020). Comparing our results with those of clinical trials is complicated by the nature of our cross-sectional research design. Furthermore, it is crucial to note that these findings have been reported mainly in adult studies. Regarding the pediatric population, studies have not identified differences in IL-10 levels between children who received L. paracasei supplements and those who received a placebo (Lin et al., 2014). These differences may be due to the biological variability inherent in IL-10 levels in healthy children, which plays a crucial role in assessing the anti-inflammatory effect of this bacterium.

Also, we found that the RA of S. aureus and A. muciniphila was associated with adiponectin levels across waist circumference. Waist circumference plays a crucial mediating role that could affect metabolism and the risk of metabolic diseases.

Reduced abundance of A. muciniphila affects barrier function in the intestine, causing the mucosal layer to become thinner and facilitates intestinal permeability (Raineri et al., 2022; Rodrigues et al., 2022). In this situation, S. aureus colonizes the intestine, interacts with the mucosal layer, and contributes to lipopolysaccharide (LPS) moving into the bloodstream owing to changes in intestinal permeability. Once in the blood, LPS binds to Toll-like receptor 4 (TLR4), initiating a signaling cascade that triggers the body’s immune response (Harberts et al., 2020). In addition to changes in the bacterial composition in the gut, in individuals with high waist circumference, enlarged adipocytes favor an oxygen-poor environment, causing endogenous hypoxemia. This hypoxia leads to fibrosis in white adipose tissue, blebbing in brown adipose tissue, and the recruitment of macrophages in the adipose tissue, promoting dysregulated cytokine expression (Wang et al., 2022).

When analyzing the joint effects of S. aureus, A. muciniphila, and L. casei, we observed a trend, although not all profiles showed statistically significant differences. Profiles predominant in S. aureus were associated with lower adiponectin levels. In contrast, profiles abundant in A. muciniphila and L. casei were associated with higher adiponectin levels.

Within the limitations of our study, we should mention that we used waist circumference as a proxy for central obesity. The prevalence of central obesity in children varies widely owing to factors such as the type of population studied, standardized measurement techniques, and proper data analysis (Monzani et al., 2016). These factors make it challenging to create universal cutoff points to identify children with true central obesity. However, waist circumference is a valid and commonly used tool in population-based studies to assess central obesity in children (Irenewati et al., 2020). Training personnel conducted waist circumference measurements using standardized techniques for more precise results. In addition, we established cutoff points based on the median of our population.

The FFQ is one of the most widely used techniques in population research because of its ability to measure long-term dietary patterns in simple, cost-effective, and efficient manner. We acknowledge the inability of participants to recall their intake because the information was collected by means of self-report. However, we addressed this problem by adjusting the statistical models by daily macronutrient intake (“all-component model”) to provide a more accurate estimate (Shim et al., 2014; Tomova et al., 2022). In addition, to ensure uniformity, the FFQ was administered in the same way to all participants in the presence of their parents.

We know the limitations inherent in our study design when evaluating mediation; because we measured both exposures and events simultaneously, it was difficult to establish a causal relationship. In recent years, there has been a focus on asking more robust causal questions, even in observational studies (Hamaker et al., 2020; Hernán, 2018). Although we are working with a cross-sectional design, proper data analysis allows us to explore causal hypotheses. We have applied methods such as path analysis and structural equation modeling, which allow us to understand the possible causal relationships between the variables incorporated in our statistical models. We are aware that our results should be interpreted with caution. However, we consider the probability of reverse causality low, as several prospective studies have shown that changes in microbiota composition can alter adiposity phenotypes and, in turn, affect cytokine expression.

Through mediation analysis, we identified the possible pathway by which the RA of bacterial species is associated with adiponectin, considering the effect of waist circumference. Understanding this mechanism makes us consider that GM and central adiposity are crucial in assessing factors associated with developing CLI in children. These findings provide a solid basis for future research and emphasize the need to explore interventions to modulate GM and waist circumference. These strategies could help prevent inflammation-associated metabolic diseases in childhood.

Conclusions

We found that L. paracasei was directly associated with IL-10 levels. Staphylococcus aureus and A. muciniphila were indirectly associated with adiponectin levels through waist circumference. Our findings expand the understanding of the relationship between GM, waist circumference, and cytokine levels in CLI. Identifying these pathways provides us with valuable opportunities for developing future intervention strategies to improve metabolic health and decrease the risk of other inflammation-linked diseases in the pediatric population.

Footnotes

Authors’ Contributions

J.C.A.-G.: Conceptualization, Data curation, Methodology, Formal analysis, Writing—Original Draft, and Visualization. M.B.-R.: Investigation, Data curation, Writing—Review and Editing, and Visualization. C.E.D.-B.: Formal analysis, Writing—Review and Editing, and Supervision. V.H.B.-M.: Validation, Investigation, and Writing—Review and Editing. M.C.: Resources, Writing—Review and Editing, and Funding acquisition. A.L.-M.: Conceptualization, Methodology, Validation, Investigation, Writing—Review and Editing, and Project administration. A.I.B.-G.: Conceptualization, Methodology, Resources, Writing—Review and Editing, Project administration, and Funding acquisition.

Data Availability

Data are contained within the article or in Supplementary Material.

Ethics Committee Approval

This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics (CI: 1129, N°1294 [27 August 2012]), Research (N°1129 [11 September 2012]), and Biosafety (CB:1120-CI:1129 [11 September 2012]) Commissions of the National Institute of Public Health (INSP), Mexico.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The study was funded by the National Council of the Humanities, Sciences, and Technologies (CONAHCYT; Grant Numbers: SSA/IMSS/ISSSTE-CONACYT 2015–262133 and FSSEP02-CB-2018; Application A1-S-33221).

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Table S4

References

Bacchetti De Gregoris

, Aldred

, Clare

, et al. Improvement of phylum- and class-specific primers for real-time PCR quantification of bacterial taxa. J Microbiol Methods, 2011; 86(3):351–356.

Choi

, Moon

, Shin

, et al. Lactobacillus paracasei-derived extracellular vesicles attenuate the intestinal inflammatory response by augmenting the endoplasmic reticulum stress pathway. Exp Mol Med, 2020; 52(3):423–437; doi: 10.1038/s12276-019-0359-3

Collado

, Derrien

, Isolauri

, et al. Intestinal integrity and Akkermansia muciniphila, a mucin-degrading member of the intestinal microbiota present in infants, adults, and the elderly. Appl Environ Microbiol, 2007; 73(23):7767–7770; doi: 10.1128/AEM.01477-07

Collado

, Isolauri

, Laitinen

, et al. Effect of mother’s weight on infant’s microbiota acquisition, composition, and activity during early infancy: A prospective follow-up study initiated in early pregnancy. Am J Clin Nutr, 2010; 92(5):1023–1030; doi: 10.3945/ajcn.2010.29877

Depommier

, Everard

, Druart

, et al. Serum metabolite profiling yields insights into health promoting effect of A. muciniphila in human volunteers with a metabolic syndrome. Gut Microbes, 2021; 13(1):1994270; doi: 10.1080/19490976.2021.1994270

Estrada-Velasco

, Cruz

, Garcia-Mena

, et al. La obesidad infantil como consecuencia de la interacción entre firmicutes y el consumo de alimentos con alto contenido energético [Childhood obesity is associated to the interaction between firmicutes and high energy food consumption]. Nutr Hosp, 2014; 31(3):1074–1081; doi: 10.3305/nh.2015.31.3.8302

Fredriksen

, Skar

, Mamen

. Waist circumference in 6-12-year-old children: The Health Oriented Pedagogical Project (HOPP). Scand J Public Health, 2018; 46(21_suppl):12–20; doi: 10.1177/1403494818767790

Furman

, Campisi

, Verdin

, et al. Chronic inflammation in the etiology of disease across the life span. Nat Med, 2019; 25(12):1822–1832; doi: 10.1038/s41591-019-0675-0

Haarman

, Knol

. Quantitative real-time PCR analysis of fecal Lactobacillus species in infants receiving a prebiotic infant formula. Appl Environ Microbiol, 2006; 72(4):2359–2365; doi: 10.1128/AEM.72.4.2359-2365.2006

10.

Hamaker

, Mulder

, van IJzendoorn

. Description, prediction and causation: Methodological challenges of studying child and adolescent development. Dev Cogn Neurosci, 2020; 46:100867; doi: 10.1016/j.dcn.2020.100867

11.

Harberts

, Liang

, Yoon

, et al. Toll-like receptor 4-independent effects of lipopolysaccharide identified using longitudinal serum proteomics. J Proteome Res, 2020; 19(3):1258–1266; doi: 10.1021/acs.jproteome.9b00765

12.

Harrington

, Staiano

, Broyles

, et al. Waist circumference measurement site does not affect relationships with visceral adiposity and cardiometabolic risk factors in children. Pediatr Obes, 2013; 8(3):199–206; doi: 10.1111/j.2047-6310.2012.00106.x

13.

Harris

, Gleason

. Application of path analysis and structural equation modeling in nutrition and dietetics. J Acad Nutr Diet, 2022; 122(11):2023–2035; doi: 10.1016/j.jand.2022.07.007

14.

Hernán

. The C-word: Scientific euphemisms do not improve causal inference from observational data. Am J Public Health, 2018; 108(5):616–619; doi: 10.2105/AJPH.2018.304337

15.

Hernández

, Gortmaker

, Laird

, et al. Validez y reproducibilidad de un cuestionario de actividad e inactividad física para escolares de la ciudad de México [Validity and reproducibility of a questionnaire on physical activity and non-activity for school children in Mexico City]. Salud Pública Méx, 2000; 42(4):315–323; doi: 10.1590/S0036-36342000000400006

16.

Hildebrandt

, Ibrahim

, Peltzer

. Cell death and inflammation during obesity: “Know my methods, WAT(son)”. Cell Death Differ, 2023; 30(2):279–292; doi: 10.1038/s41418-022-01062-4

17.

Irenewati

, Arkhaesi

, Wistiani

. Association of BMI measurements to waist circumference and waist-to-height ratio in overweight and obese children. PI, 2020; 60(3):131–136; doi: 10.14238/pi60.3.2020.131-6

18.

Jaswal

, Todd

, Behnsen

. Neglected gut microbiome: Interactions of the non-bacterial gut microbiota with enteric pathogens. Gut Microbes, 2023; 15(1):2226916; doi: 10.1080/19490976.2023.2226916

19.

Kalliomäki

, Collado

, Salminen

, et al. Early differences in fecal microbiota composition in children may predict overweight. Am J Clin Nutr, 2008; 87(3):534–538; doi: 10.1093/ajcn/87.3.534

20.

Lin

, Fu

, Lin

, et al. Evaluation of the effect of Lactobacillus paracasei (HF.A00232) in children (6–13 years old) with perennial allergic rhinitis: A 12-week, double-blind, randomized, placebo-controlled study. Pediatr Neonatol, 2014; 55(3):181–188; doi: 10.1016/j.pedneo.2013.10.001

21.

Ling

, Cheng

, Chen

, et al. Changes in fecal microbiota composition and the cytokine expression profile in school-aged children with depression: A case-control study. Front Immunol, 2022; 13:964910; doi: 10.3389/fimmu.2022.964910

22.

Meng

, Feng

, Hao

, et al. Changes in gut microbiota composition with age and correlations with gut inflammation in rats. PLoS One, 2022; 17(3):e0265430; doi: 10.1371/journal.pone.0265430

23.

Monzani

, Rapa

, Prodam

, et al. High Discrepancy in abdominal obesity prevalence according to different waist circumference cut-offs and measurement methods in children: Need for age-risk-weighted standardized cut-offs? PLoS One, 2016; 11(1):e0146579; doi: 10.1371/journal.pone.0146579

24.

Raineri

EJM

, Altulea

, van Dijl

. Staphylococcal trafficking and infection-from ‘nose to gut’ and back. FEMS Microbiol Rev, 2022; 46(1):fuab041; doi: 10.1093/femsre/fuab041

25.

Rodrigues

, Elias-Oliveira

, Pereira

ÍS

, et al. Akkermansia muciniphila and Gut Immune System: A good friendship that attenuates inflammatory bowel disease, obesity, and diabetes. Front Immunol, 2022; 13:934695; doi: 10.3389/fimmu.2022.934695

26.

Romero-Martínez

, Shamah-Levy

, Franco-Núñez

. Encuesta Nacional de Salud y Nutrición 2012 [National Health and Nutrition Survey]. Salud Publica Mex, 2013; 55(Supl.2):332–S340.

27.

Schmittgen

, Livak

. Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc, 2008; 3(6):1101–1108; doi: 10.1038/nprot.2008.73

28.

Schoemann

, Boulton

, Short

. Determining power and sample size for simple and complex mediation models. Soc Psychol Personal Sci, 2017; 8(4):379–386; doi: 10.1177/1948550617715068

29.

Shim

, Oh

, Kim

. Dietary assessment methods in epidemiologic studies. Epidemiol Health, 2014; 36:e2014009; doi: 10.4178/epih/e2014009

30.

Steene-Johannessen

, Kolle

, Reseland

, et al. Waist circumference is related to low-grade inflammation in youth. Int J Pediatr Obes, 2010; 5(4):313–319; doi: 10.3109/17477160903497035

31.

Sun

, Luo

, Liang

, et al. Bifidobacterium alters the gut microbiota and modulates the functional metabolism of T regulatory cells in the context of immune checkpoint blockade. Proc Natl Acad Sci U S A, 2020; 117(44):27509–27515; doi: 10.1073/pnas.1921223117

32.

Tomova

, Arnold

, Gilthorpe

, et al. Adjustment for energy intake in nutritional research: A causal inference perspective. Am J Clin Nutr, 2022; 115(1):189–198; doi: 10.1093/ajcn/nqab266

33.

Turroni

, Milani

, Duranti

, et al. The infant gut microbiome as a microbial organ influencing host well-being. Ital J Pediatr, 2020; 46(1):16; doi: 10.1186/s13052-020-0781-0

34.

Wang

, Sun

, Wu

, et al. Hypoxia as a double-edged sword to combat obesity and comorbidities. Cells, 2022; 11(23):3735; doi: 10.3390/cells11233735

35.

White

, Bjørnholt

, Baird

, et al. Novel developmental analyses identify longitudinal patterns of early gut microbiota that affect infant growth. PLoS Comput Biol, 2013; 9(5):e1003042; doi: 10.1371/journal.pcbi.1003042

36.

Yamanaka

, Davis

, Wilkens

, et al. Determination of child waist circumference cut points for metabolic risk based on acanthosis nigricans, the Childreńs Healthy Living Program. Prev Chronic Dis, 2021; 18:E64; doi: 10.5888/pcd18.210021

37.

Yang

, Cong

. Gut microbiota-derived metabolites in the regulation of host immune responses and immune-related inflammatory diseases. Cell Mol Immunol, 2021; 18(4):866–877; doi: 10.1038/s41423-021-00661-4

38.

Yao

, Cai

, Ye

, et al. The role of microbiota in infant health: From early life to adulthood. Front Immunol, 2021; 12:708472; doi: 10.3389/fimmu.2021.708472

39.

Zhang

, Yang

, Limei

, et al. The mediating effect of central obesity on the association between dietary quality, dietary inflammation level and low-grade inflammation-related serum inflammatory markers in adults. Int J Environ Res Public Health, 2023; 20(5):3781; doi: 10.3390/ijerph20053781

40.

Zhou

, Zhang

, Wang

, et al. Gut bacteria Akkermansia is associated with reduced risk of obesity: Evidence from the American gut project. Nutr Metab, 2020; 17:90; doi: 10.1186/s12986-020-00516-1