The association between gut microbiota composition and cardiometabolic parameters in healthy adults

Abstract

Background

The human gut microbiota comprises approximately 100 trillion microbial cells and produces a wide range of metabolites. Its composition is shaped by factors such as geography, dietary habits, and genetic background. Dysbiosis—an imbalance in this microbial ecosystem—has been associated with the development of metabolic diseases. This study aimed to characterize the gut microbiota composition in adults without diagnosed chronic diseases and assess its potential associations with cardiometabolic parameters.

Methods

Volunteers over the age of 18 residing in Izmir province who met the inclusion criteria were enrolled in the study. Fecal and blood samples were collected from all participants. Bacterial DNA from fecal samples was extracted, and the full-length 16 S rRNA was amplified. Full-length 16 S rRNA PCR amplicons were sequenced using Oxford Nanopore Technologies. Taxonomic classification, Firmicutes/Bacteroidetes ratio, and Shannon index were determined using Massbiome Fecal Microbiome Analysis (Massive Bioinformatics, Türkiye). Dominant bacterial genera were also analyzed using the Epi2Me database. Blood samples were analyzed for metabolic health markers. Associations between bacterial taxa and blood-based metabolic parameters were examined statistically.

Results

A total of 82 participants were included in the study. According to BMI, 34 (41.4%) of the participants were classified as having a healthy weight, 23 (28.1%) as overweight, and 25 (30.5%) as obese. Segatella (Prevotella) copri was identified as the dominant species in most participants (51%). Other dominant species included Ruminococcus torques (13%), Faecalibacterium prausnitzii (13%), Faecalitalea cylindroides (5%), Bacteroides uniformis (4%), and Phocaeicola vulgatus (3%). S. copri was identified as the dominant bacterial species in 45 participants (54.8%) according to the Epi2Me database. The Shannon index and Firmicutes/Bacteroidetes ratio varied among the three BMI groups. Both Lachnospira and Ruminococcus genera had a significant negative correlation with BMI, indicating that higher levels of BMI are associated with lower abundances of these genera. In the regression analysis, Lachnospira was found to be less abundant at higher HbA1c levels. HOMA-IR was found to be a significant predictor for the relative abundance of Vescimonas. Ruthenibacterium is found to be more abundant at higher HDL levels.

Conclusion

The dominant bacterial taxa differed from those reported in other populations, suggesting region-specific microbial profiles. Notably, specific gut microbial genera were associated with metabolic health indicators, including BMI, HbA1c, HOMA-IR, and HDL level. These findings underscore the potential role of gut microbiota in metabolic regulation and support the need for further region-specific research. This study is among the few in Türkiye focusing on healthy adults and contributes to the understanding of gut microbiota in relation to cardiometabolic parameters.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04261-4.

Introduction

The gut microbiota is a complex community of microorganisms that begins to establish at birth and develops according to geography, eating habits, and genetic background [1]. This community harbors 100 trillion microbial cells and produces metabolites, including short-chain fatty acids, vitamins, anti-inflammatory antioxidants, neurotoxins, carcinogens, and immunotoxins. These metabolites influence metabolic processes, mental health, energy balance, and the immune system by regulating the expression of relevant genes [2, 3]. A decrease in beneficial microorganisms, an increase in pathogenic bacteria, and a loss of microbial diversity constitute dysbiosis, which has been linked to the development of metabolic diseases [2].

The gut microbiota is estimated to contain 500–1000 bacterial species, most of which are anaerobic [4]. Approximately 90% of these bacteria belong to the Firmicutes and Bacteroidetes phyla [2, 4]. The healthy human gut microbiota is primarily composed of families such as Bacteroidaceae, Clostridiaceae, Prevotellaceae, Eubacteriaceae, Ruminococcaceae, Bifidobacteriaceae, Lactobacillaceae, Enterobacteriaceae, Saccharomycetaceae, and Methanobacteriaceae [5].

In recent years, research on the human microbiome has expanded rapidly, revealing associations between gut microbiota composition and various diseases. In addition to numerous known factors, geographical and population-specific variables also affect gut microbiota composition. Therefore, characterizing the gut microbiota of healthy individuals in specific regions is essential for advancing local microbiome–disease research.

Short-read sequencing technologies are commonly used to analyze gut microbiota composition, typically offering taxonomic resolution at the family or genus level. In this study, full-length 16 S rRNA sequencing using the Oxford Nanopore platform was employed, enabling species-level identification and providing a more detailed profile of gut microbiota composition.

The objective of this study was to characterize the gut microbiota in adult individuals without diagnosed chronic diseases and to evaluate its potential associations with cardiometabolic parameters.

Methods

Participants

Volunteers over the age of 18 residing in İzmir province who met the inclusion criteria were recruited for this study. Body mass index (BMI) was calculated using weight and height squared (kg/m²). Participants were classified as healthy weight (≥ 18.5 to < 25 kg/m²), overweight (≥ 25 to < 29.9 kg/m²), or obese (≥ 30 kg/m²) according to body mass index (BMI) guidelines from the World Health Organization. Individuals meeting the exclusion criteria listed in Table 1 were not included in the study.

Table 1.

Exclusion criteria

Exclusion Criteria

▪ Use of antibiotics, pump inhibitors drugs, probiotics, or prebiotics in the last 3 months ▪ Clinical diagnosis of chronic lung, liver, kidney, and cardiovascular diseases ▪ Patients with cancer or a history of cancer ▪ Diagnosed with neurodegenerative, gastrointestinal, and psychiatric diseases ▪ Individuals with alcohol and substance addiction ▪ Patients with Human Immunodeficiency Virus (HIV), Hepatitis B virus (HBV), or Hepatitis C Virus (HCV) ▪ Changed eating habits in the last month ▪ Patients with chronic diarrhea or chronic constipation ▪ Underwent surgery in the last month ▪ Women who are menstruating, pregnant, or breastfeeding

Data on covariates, including age, sex, disease status, medication use, and smoking status, were collected through a questionnaire completed by each participant. Participants’ consumption frequency of various food items was assessed through a food frequency questionnaire, which included dairy products (such as milk, yogurt, cheese), meat and animal products (such as red meat, poultry, egg), plant-based foods (such as leafy green vegetables, legumes, fruits), grain-based foods (such as bread, rice, pasta), and ready-to-eat foods. Each item was scored on a 6-point scale, which was later converted to a 0–5 scale, where 0 indicated “never” and 5 indicated “daily” consumption. For each food category, an average score was calculated based on the relevant food items to derive composite dietary intake scores. Kruskal-Wallis tests were conducted to assess whether consumption frequency differed across BMI groups for each food category. Additionally, cluster analysis was performed in SPSS version 29.0 using the composite category scores to identify distinct dietary patterns among participants. A three-cluster solution based on four major food categories (dairy, animal-based, plant-based, and grain-based foods) was selected for further analysis due to its interpretability.

Sample collection

Fecal and blood samples were collected from all participants. Fecal samples were collected using the Stool Sample Collection and Stabilization Kit (Canvax, ES), which is specifically designed for gut microbiota analysis and allows stabilized samples to be stored at room temperature for several months (https://www.canvaxbiotech.com/ruo-grade-shop/sample-collection/stool-samples/stool-sample-collection-stabilization-kit/). The collected fecal samples were transported to the Bioprocess and Microbiota Laboratory at Dokuz Eylul University BioIzmir Application and Research Center. Upon arrival, samples were immediately processed for DNA extraction.

The study was approved by the Dokuz Eylul University Noninterventional Research Ethics Committee (ethics numbers 22.03.2023, 2023/09–33), and written informed consent was obtained from all participants.

Genomic DNA extraction and PCR amplification

DNA from the fecal samples was extracted via the Quick-DNA Fecal/Soil Microbe Kits (Zymo Research, USA) according to the manufacturer’s protocol. The total DNA concentration of each sample was measured with a Qubit 4 fluorometer (Thermo Fisher Scientific, USA). The template DNA was stored at −20 °C for PCR amplification. The full-length 16 S rRNA region of the bacteria was amplified and sequenced. Each PCR reaction was prepared using 10 ng of normalized template DNA and 10 pmol of 16 S rRNA primers, in a total reaction volume of 10 µl. The primer sequences and PCR cycling conditions are shown in Table 2 [6]. The PCR products were run on a 1% agarose gel (Cleaver Scientific, UK) at 120 volts and 400 amps for 30 min. The DNA bands were imaged and analysed via an imaging system (Azure Biosystem, USA).

Table 2.

Primer sequences and cycling conditions for full-length 16 S rRNA

Primer Sequences	Cycling Condition
Forward: 27 F AGAGTTTGATCMTGGCTCAG Reverse:1492R GGTTACCTTGTTACGACTT	94 °C, 30 s Initial Denaturation 94 °C, 20 s Denaturation } 35 Cycles 58 °C, 1 min Annealing } 35 Cycles 72 °C, 1 min Extension } 35 Cycles 72 °C, 10 min Final Extension

Library Preparation and 16 S rRNA gene sequencing

Library preparation was performed via the Native Barcoding Kit 96 V14 (SQK-NBD114.96, Oxford Nanopore Technologies) according to the kit protocol (https://store.nanoporetech.com/native-barcoding-kit-96-v14.html). Briefly, the library preparation protocol was performed in three main steps: (i) End-Prep of PCR Products: 200 fmol of the PCR products were processed for end-repair using NEB Next Ultra II End Repair/dA-Tailing Module (NEB, E7546), (ii) Barcode Ligation: unique barcodes were ligated to a 1/20 portion of the end-repaired product using the NEB Blunt/TA Ligase Master Mix (NEB, M0367). All the barcoded samples were pooled and purified using 0.4x AMPure XP Beads (Mobiomx Magnetic beads, MBD01-02-03-04). (iii) Adaptor Ligation: The pooled and purified samples were ligated with the Native Adapter using T4 DNA Ligase (Mobiomx, RL01), and the excess adaptor was removed with 0.4x AMPure XP Beads. The resulting library was quantified by a Qubit 4 fluorometer (Thermo Fisher Scientific). The genome library containing 20 ng of DNA was loaded into a MinION MK1B device (Oxford Nanopore Technologies) connected to MinKNOW software on Linux. Sequencing was stopped after 24–72 h once the number of reads for each sample reached 10 kb.

Bioinformatics analysis

FAST5 files obtained from the Nanopore sequencing process were converted into FASTQ format via the guppy basecaller (guppy, Oxford Nanopore Technologies, UK). Adaptors and low-quality FASTQ reads were trimmed via the Trimmomatic tool. After clearing the sequences, reads 1250–1750 bp long were filtered with Trimmomatic, and the remaining reads were excluded from the analysis. A minimum Phred score of 20 was used to filter out low-quality reads in FASTQ files. The quality of the filtered reads was checked via the FASTQC tool. Sequencing depth per sample was assessed during the quality control steps, and a minimum threshold of 10,000 reads was applied to exclude low-depth samples from downstream analyses. Rarefaction was performed using QIIME2’s core-metrics-phylogenetic pipeline, with a sampling depth set at 10,000 sequences per sample to normalize sequencing effort across all samples.

The trimmed reads were aligned to the Massive Bioinformatics database produced using 16 S rRNA genes of bacterial whole genomes submitted to NCBI using the NCBI BLAST tool. Using the taxonomic details of the 16 S rRNA genes to which the reads were aligned and the number of reads aligned to the genes, a biome file was produced. Taxonomic classification, the Firmicutes/Bacteroidetes (F/B) ratio, and the Shannon index were determined via microbiome analysis (Massive Bioinformatics, Türkiye). MassBiome Fecal Microbiome Analysis version V2 (01-Mar-2024) was used for this study. Dominant bacterial species and genera were also analysed via the Epi2Me database (16 S Classification - v5.0.11).

Taxonomic profiling was performed at multiple levels (phylum to species), with a focus on genus and species-level data. Statistical analyses were then conducted to examine associations between microbial taxa and cardiometabolic health markers.

Biochemical analysis of blood metabolic health parameters

Blood samples were taken following an overnight fast. All blood parameters were analysed using fresh blood samples immediately after collection at the Central Laboratory of Dokuz Eylül University Hospital. Fasting plasma glucose, total cholesterol, High-density lipoprotein (HDL), Low-density lipoprotein (LDL), triglyceride, and C-reactive protein (CRP) levels were analysed via photometric methods via an AU5800 autoanalyzer (Beckman Coulter, USA). Hemoglobin A1C (HbA1c) was measured via a TOSOH G8 HPLC analyser (Tosoh Corporation, Japan). For homocysteine analysis, blood samples were collected in EDTA plasma tubes, stored on ice, and promptly separated to maintain stability. Homocysteine levels were determined using the solid-phase competitive chemiluminescent immunoassay method via the DPC IMMULITE 2500 analyser (Siemens Healthcare Diagnostics, USA). Fasting insulin levels were also measured by immunoassay tests via the SIEMENS ADVIA Centaur CP analyser (Siemens Healthcare Diagnostics, USA). The homeostatic model assessment for insulin resistance (HOMA-IR) score was calculated from fasting glucose and insulin levels via the following formula: HOMA-IR = (insulin × glucose (mg/dL))/405.

Metabolic health status of participants was assessed using the National Cholesterol Education Program’s Adult Treatment Panel III (NCEP ATP III) criteria. According to these guidelines, an individual is classified as having metabolic syndrome risk if they meet at least three of the following five risk factors: abdominal obesity: waist circumference greater than 102 cm for men or 88 cm for women, elevated triglycerides: fasting triglyceride levels of 150 mg/dL or higher, reduced HDL cholesterol: fasting HDL cholesterol levels of less than 40 mg/dL for men or less than 50 mg/dL for women, elevated blood pressure: systolic blood pressure greater than 130 mmHg or diastolic blood pressure greater than 85 mmHg, or the use of antihypertensive medication, elevated fasting glucose: fasting blood glucose levels of 100 mg/dL or higher, or the use of medication for elevated blood glucose [7]. In the present analysis, BMI > 30 was used in place of waist circumference to indicate obesity, while hypertension was identified based on clinical diagnosis rather than blood pressure value. Due to logistical constraints and the inability to re-contact participants, obtaining multiple blood pressure measurements per individual was not feasible. Moreover, a single blood pressure measurement is not considered reliable for hypertension classification [8]. Therefore, hypertension status was determined based on self-reported physician diagnosis or antihypertensive medication use.

Statistical analysis

Basic statistical analyses were performed via SPSS (IBM) version 29.0 for Windows. The Kruskal‒Wallis test was used to test differences between BMI groups. p values less than 0.05 were considered statistically significant. In order to measure the effect size of the Kruskal-Wallis test, eta-squared (η²) was calculated using the formula;

where H is the Kruskal-Wallis test statistic, k is number of groups and n is the total number of subjects across all groups. The magnitude of the effect was interpreted based on standard benchmarks, where small, medium, and large effect sizes correspond to values of 0.01, 0.06, and 0.14, respectively [9].

Linear regression analyses were conducted to examine the relationships between the relative abundance of microbial taxa and independent variables, including BMI and blood metabolic health marker levels. Each variable was individually tested in a single linear regression model. To control for multiple testing in the initial single regression analyses, p-values were adjusted using the Benjamini-Hochberg False Discovery Rate (FDR) correction. Variables with a FDR-adjusted p -value less than 0.1 were included in the final model. All models were adjusted for age, sex, and smoking status. Model optimization was further refined using backward stepwise selection based on the Akaike Information Criterion (AIC) to balance model complexity and goodness of fit. The residuals of each model were tested for normality using the Shapiro-Wilk test, and residual plots were inspected for heteroscedasticity. Data that were not normally distributed were transformed using the natural logarithm (ln), while relative abundance data containing zero values were transformed using the ln (1 + X) formula. For two variables where log transformation did not normalize residual distributions, square root transformation and rank-based inverse normal transformation were applied. Partial eta squared values and observed power were calculated for each model to assess effect size and statistical power.

Results

Participant characteristics and metabolic profiles

A total of 82 participants (66 females, 16 males; mean age 43.64 ± 9.7; age range 19–64) were included in the current study. According to BMI, 34 (41.4%) of the participants were classified as having a healthy weight, 23 (28.1%) as overweight, and 25 (30.5%) as obese.

In the questionnaires, five participants reported a diagnosis of type II diabetes, and three of them indicated they were using antidiabetics. Two of these participants were classified as overweight, while three were in the obese group. HbA1c and fasting glucose level data were consistent with the reported disease status. No other participants had an HbA1c level above 6.5% or fasting glucose levels exceeding 126 mg/dL. Three participants reported having high cholesterol, with two of them using cholesterol-lowering medications. All three participants with high cholesterol were in the overweight group. Additionally, five participants reported hypertension diagnosis; four were classified as obese, and one was in the healthy weight group. Among them, three participants reported receiving treatment for hypertension.

When classified according to the NCEP ATP III metabolic syndrome criteria, one overweight and nine obese participants met the diagnostic criteria. Although the number of participants with metabolic syndrome was limited, a preliminary comparison was conducted. This group showed lower alpha diversity and reduced relative abundances of Bacteroides, Lachnospira, and Phocaeicola, whereas S. copri and Prevotella appeared more abundant. Due to the small sample size and lack of statistical testing, these observations should be interpreted with caution.

Dietary habits

The food consumption frequency questionnaire revealed that all participants had similar eating habits consistent with a Mediterranean diet. None of the participants followed restrictive diets such as vegan, vegetarian, or gluten-free diets. All participants reported consuming foods from each food group daily. The consumption of ready-to-eat foods was reported to be infrequent, with 46.6% of participants consuming such foods less than once a month. Overall, the participants appeared to adhere to balanced diets. Additionally, 40% of participants reported consuming yogurt daily, and 82.7% consumed yogurt more than twice a week. Ayran (a yogurt-based drink) consumption was also reported to be frequent. Food consumption frequencies are similar across all weight groups.

Kruskal-Wallis tests revealed no statistically significant differences in the consumption frequencies of dairy, animal-based, plant-based, or grain-based food categories across BMI groups (p > 0.05 for all comparisons). Cluster analysis using the four-category scores resulted in three distinct dietary patterns. Cluster 1 consisted of participants with moderate intake across all food categories (“Moderate Consumers”), Cluster 2 represented individuals with relatively higher intake of all food types, especially plant-based and grain-based foods (“High Intake/Plant-Leaning Consumers”), and Cluster 3 was characterized by low grain consumption with moderate intake of other categories (“Low-Grain Consumers”). A chi-square test was conducted to examine the relationship between cluster membership and BMI categories, but no significant association was found. Following the identification of dietary clusters, Kruskal-Wallis tests were performed to compare gut microbiota composition at the genus level, alpha diversity indices, Firmicutes/Bacteroidetes (F/B) ratio, and various cardiometabolic health parameters across the three clusters. None of these comparisons reached statistical significance, as all p-values were greater than 0.05.

Metabolic parameters across BMI groups

The relationship between BMI and blood cardiometabolic parameters is investigated as these are known to change with obesity and may serve as contributing factors to the observed variations in gut microbiota composition across different BMI groups. The general characteristics and blood metabolic health marker levels of the participants according to BMI groups (healthy weight, overweight, and obese) are shown in Table 3. Kruskal-Wallis test was conducted to compare blood marker levels across the healthy weight, overweight, and obese groups. Significant differences were observed for fasting glucose (H = 7.53, p = 0.023, η2 = 0.07), fasting insulin (H = 27.73, p < 0.001, η2 = 0.34), HOMA-IR (H = 28.3, p < 0.001 η2 = 0.35), HbA1c (H = 12.7, p = 0.002, η2 = 0.14), triglyceride (H = 19.18, p < 0.001, η2 = 0.22), CRP (H = 17.63, p < 0.001, η2 = 0.2). Post-hoc pairwise comparisons revealed that the obese group had significantly higher levels of these markers compared to the healthy weight group (p < 0.05). The number of smokers was significantly higher in the overweight and obese groups compared to the healthy weight group, as determined by a Chi-square test (p = 0.007). The data were adjusted according to smoking status in further analyses.

Table 3.

General characteristics of the participants according to BMI group. Data are represented as the median (minimum value–maximum value). The p-value for the difference between BMI groups was calculated via the kruskal‒wallis test. For smokers, the p-value was calculated via the chi-square test

	Healthy Weight	Overweight	Obese	p-value
Sample Size(Male/Female)	34 (M = 9, F = 25)	23 (M = 5, F = 18)	25 (M = 2/F = 23)
Age (years)	40.62 (19–59)	47.04 (30–64)	44.67 (21–62)	0.054
BMI (kg/m2)	22.4 (19.1–24.9)	27.5 (25-29.7)	35.2 (30.1–51.1)
Fasting Glucose (mg/dL)	91 (69–104)	86 (70–173)	94 (84–172)	0.023
Fasting Insulin (mU/L)	6.9 (3.2–24.3)	10.7 (3.39–19.8)	15.7 (6.85–52.61)	< 0.001
HOMA IR	1.6 (0.72–5.1)	2.5 (0.72–4.79)	3.8 (1.52–10.9)	< 0.001
Glycosylated haemoglobin (HbA1c) (%)	5.3 (5–5.7)	5.4 (4.7–8)	5.6 (4.7–7.3)	0.002
Homocysteine (Umol/L)	11 (5.76–32)	12 (6.36–18.6)	10.1 (7.6–21.4)	0.735
Total Cholesterol (mg/dL)	211 (119–352)	205 (166–284)	198 (135–297)	0.681
Triglyceride (mg/dL)	70.5 (38–141)	94 (39–419)	106 (58–273)	< 0.001
HDL (mg/dL)	63.5 (35–106)	50 (38–80)	49 (29–72)	0.19
LDL (mg/dL)	140.4 (69.2-231.4)	133 (91.6–400)	120.6 (74.2-185.4)	0.275
CRP (mg/dL)	1.5 (0.2–9.5)	2.9 (0.4–82.7)	3.1 (0.5–27.9)	< 0.001
Smokers (%)	15.1	30.4	54.1	0.007

Dominant bacterial species

The dominant bacterial species identified among the participants are shown in Fig. 1. Segatella (Prevotella) copri (S. copri) was identified as the dominant species in most participants (51%). Other dominant species included Ruminococcus torques (13%), Faecalibacterium prausnitzii (13%), Faecalitalea cylindroides (5%), Bacteroides uniformis (4%), and Phocaeicola vulgatus (3%). Segatella (Prevotella) copri was identified as the dominant bacterial species in 45 participants (54.8%) via the Epi2Me database.

Fig. 1.

Dominant bacterial species

Distribution of bacteria BMI-Based participant groups

The relative abundances of the 15 most commonly identified species and genera among the participants are presented as percentages in Fig. 2. S. copri, R. torques, F. prausnitzii, Anaerostipes hadrus, P. vulgatus, Bacteroides, Blautia, Coprococcus, Lachnospira, and Roseburia were detected in the gut microbiota of all the participants. Escherichia coli, Enterococcus faecium, and Clostridium were identified in some participants. The 15 most frequently identified bacterial genera and their cumulative reads via the Epi2Me database are shown in Fig. 3.

Fig. 2.

Distribution of bacteria BMI-based participant groups. Bar plots depicting the relative abundances (%) of gut microbiota in BMI-based participant groups were presented at the (a) genus and (a) species levels

Fig. 3.

The 15 most frequently identified bacterial genera and cumulative reads from the Epi2Me database

Alpha diversity and F/B ratio across BMI groups and their associations with metabolic parameters

The boxplots that illustrate Shannon index that indicates alpha diversity of bacterial communities and F/B ratio across three BMI groups: healthy weight, overweight and obese are shown in Fig. 4. The median alpha diversity is 5.26 (interquartile range (IQR): 0.73) for the healthy weight group, 4.95 (IQR: 0.95) for the overweight group, and 4.3 (IQR: 1.13) for the obese group. The median F/B ratios are 2.27 (IQR: 2.24) for the healthy weight group, 1.69 (IQR: 0.93) for the overweight group, and 1.06 (IQR: 1.75) for the obese group. The Kruskal-Wallis test indicated significant differences in the F/B ratio (H = 11.13, p = 0.004, η2 = 0.12), and alpha diversity (H = 10.71, p = 0.005, η2 = 0.11) across three BMI groups. Pairwise comparisons revealed that both the F/B ratio and alpha diversity were significantly greater in the healthy weight group compared to the obese group (p = 0.004, p = 0.003, respectively).

Fig. 4.

Boxplot of Shannon index and F/B ratio across BMI groups. Statistical analyses were performed via the Kruskal‒Wallis test for three groups. F/B ratio and alpha diversity were significantly different among the three BMI groups (p = 0.004 and p = 0.005, respectively). Pairwise comparisons revealed that the F/B ratios and alpha diversity were significantly greater in the healthy weight group than in the obese group (p = 0.004 and p = 0.003, respectively)

A regression model was conducted to examine the associations between alpha diversity, BMI, and metabolic parameters. After selection of significant predictor parameters and model fitting, the final regression model included BMI, fasting glucose, age, and smoking status, where age and smoking were included as covariates for data correction. The overall model was significant, F(4, 75) = 3.45, p = 0.012, R² = 0.155. Among the predictors, fasting glucose was the only variable with a statistically significant effect. Results from the regression analysis suggest that alpha diversity decreases as fasting glucose increases, with a significant negative relationship between the two (β = −0.242, p = 0.034). This indicates that higher levels of fasting glucose are associated with reduced alpha diversity in the gut microbiota. The final regression model of F/B ratio included BMI, fasting glucose, HOMA-IR, age, smoking status (F(5, 72) = 3.63, p = 0.006, R² = 0.201). BMI and fasting glucose have statistically significant effect; for BMI β = −287, p = 0.046; for fasting glucose β = −256, p = 0.038. The model indicates that higher BMI values and higher fasting glucose levels are associated with lower F/B ratio. Sex was initially included in both models but was excluded in the final analysis as it did not significantly contribute to the explanation of alpha diversity or F/B ratio, and its removal resulted in a more parsimonious models without a substantial reduction in adjusted R². Partial eta squared values and observed power of the model are presented in Table 4.

Table 4.

Results of the linear regression models. The beta coefficients, p-values, Eta squared (η²), and observed power for the significant predictors in the analysis are reported

Alpha Diversity Model R squared = 0.155	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	0.012	0.155	0.835
Intercept	-	< 0.001	0.632	1.000
BMI	−0.192	0.107	0.034	0.363
Fasting Glucose	−0.242	0.034	0.058	0.567
Age	0.047	0.665	0.003	0.071
Smoking Status	−0.128	0.255	0.017	0.205

F/B Ratio Model R squared = 0.201	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	0.006	0.201	0.906
Intercept	-	< 0.001	0.198	0.986
BMI	−0.287	0.046	0.054	0.516
Fasting Glucose	−0.256	0.038	0.059	0.552
HOMA-IR	0.066	0.654	0.003	0.073
Age	−0.008	0.94	0	0.051
Smoking Status	−0.107	0.355	0.012	0.151

Ruminococcus Model R squared = 0.245	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	< 0.001	0.245	0.985
Intercept	-	< 0.001	0.438	1
BMI	−0.424	< 0.001	0.165	0.971
Age	−0.164	0.113	0.032	0.354
Sex	−0.188	0.073	0.041	0.433
Smoking Status	0.013	0.903	0	0.052

Lachnospira Model R squared = 0.235	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	0.001	0.235	0.965
Intercept	-	< 0.001	0.169	0.971
BMI	−0.255	0.031	0.06	0.582
HbA1c	−0.344	0.006	0.098	0.805
Age	0.17877371	0.122	0.032	0.339
Sex	−0.02155547	0.841	0.001	0.055
Smoking Status	−0.10677146	0.328	0.013	0.163

Vescimonas Model R squared = 0.145	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	0.043	0.145	0.738
Intercept	-	0.001	0.144	0.931
HOMA-IR	−0.311	0.009	0.092	0.761
CRP	−0.195	0.081	0.042	0.416
Age	0.015	0.891	0	0.052
Sex	−0.065	0.562	0.005	0.089
Smoking Status	−0.052	0.655	0.003	0.073

Ruthenibacterium Model R Squared = 0.103	Coefficient β	p value	Partial Eta Squared	Observed Power
Corrected Model	-	0.038	0.103	0.677
Intercept	-	0.075	0.041	0.43
HDL	0.311	0.008	0.089	0.772
Age	−0.066	0.553	0.005	0.091
Sex	−0.047	0.674	0.002	0.07

Associations between bacterial taxa, BMI, and metabolic health markers

The relative abundances of the genera Ruminococcus (H = 19.07, p < 0.001, η2 = 0.22), Coprococcus (H = 10.52, p = 0.005, η2 = 0.11), Lachnospira (H = 9.58, p = 0.008, η2 = 0.10), and Vescimonas (H = 9.10, p = 0.011, η2 = 0.09) were significantly lower in the obese group than in the healthy weight group. In contrast, the relative abundances of S. copri (H = 6.82, p = 0.03, η2 = 0.06) and Lactobacillaceae (H = 6.83, p = 0.03, η2 = 0.06) were significantly greater in the obese group than in the healthy weight group.

Multiple linear regression models were constructed to analyse the associations between the relative abundances of bacteria, metabolic parameters, and BMI, with data adjusted for age, sex, and smoking status. Only the relative abundances of Ruminococcus and Lachnospira were significantly associated with BMI. BMI was found to be significant predictors in the Ruminococcus model after adjusting for age, sex, and smoking status (F(4, 77) = 6.261, p < 0.001, R² = 0.245). The model indicates that Ruminococcus is less abundant at higher BMI values (β= −0.424, p < 0.001).

BMI and HbA1c were found to be significant predictors in Lachnospira model after adjusting for age, sex and smoking status F(5, 75) = 4.60, p = 0.001, R² = 0.235). Lachnospira was found to be less abundant at higher BMI (β= −0.255, p = 0.03) and higher HbA1c levels (β= −0.344, p = 0.006). In the regression model for Vescimonas, HOMA-IR and CRP were included, adjusted for age, sex, and smoking status (F(5,72) = 2.43, p = 0.04, R² = 0.145). HOMA-IR was found to be a significant predictor for relative abundance of Vescimonas (β=−0.311, p = 0.009). Relative abundance of Ruthenibacterium is found to be associated with HDL when adjusted for age and sex (F=(3,77) = 2.94, p = 0.03, R² = 103). Ruthenibacterium is found to be more abundant at higher HDL levels (β = 0.311, p = 0.008). No other significant associations were found between other microbial taxa and metabolic health parameters. Eta squared values and observed power for each model are reported in Table 4.

Discussion

A healthy gut microbiota, characterized by high microbial diversity, is crucial for processes such as polysaccharide biodegradation, short-chain fatty acid production, and the synthesis of vitamins and essential amino acids. Dysbiosis, or imbalance in the gut microbiota, can lead to altered metabolite production and the emergence of chronic disorders [3].

The gut microbiota composition is influenced by geographical location and dietary habits, leading to variations in dominant bacterial species across different regions. For instance, Bifidobacterium, Ruminococcus, Blautia, and Dorea are prominent in the USA, whereas Ruminococcus, Roseburia, and Veillonellaceae dominate in the Netherlands. In Germany, Bifidobacterium, Clostridium, and Veillonella are prevalent, whereas in Japan, Bifidobacterium is dominant; in China, Bacteroides is dominant; and in Tanzania, Prevotella, Treponema, and Clostridiales are dominant [1]. Our study revealed that the microbiota composition and dominant bacterial species of people in our region differ from those in other countries. Specifically, S. copri was the most common dominant species found among our participants, followed by R. torques, F. prausnitzii, and F. cylindroides.

In studies conducted in Türkiye, the gut microbiota of individuals with conditions such as polycystic ovary syndrome, amyotrophic lateral sclerosis, and Alzheimer’s disease has been compared with that of healthy controls, and associations with microbial composition have been identified [10–12]. When comparing the present findings with those reported for healthy controls in previous studies, bacterial taxa such as Prevotella (also known as Segatella), Bacteroides, Ruminococcus, Faecalibacterium, Alistipes, Blautia, Roseburia, and Dialister were frequently detected [11, 12]. In contrast, although taxa such as Succinivibrio, Muribaculaceae, and Oscillospiraceae were commonly reported in some studies, these bacteria were rarely identified among the participants in the present study [12].

Obesity, a major global health issue, has doubled in adults and quadrupled in school-aged children since 1990 [13]. While multiple factors—including genetics, socioeconomic status, and environmental influences—contribute to obesity, the gut microbiota is increasingly recognized as a key player. Several studies have reported differences in the composition and diversity of the gut microbiota between obese and non-obese individuals [14, 15]. For example, a cohort study reported a decrease in the ratio of bacteria such as Akkermansia, Faecalibacterium, and Alistipes in obese individuals [14]. However, we did not detect a statistically significant difference in these bacteria. A systematic review and meta-analysis showed inconsistent findings regarding genera such as Lachnospira, Lactobacillus, Coprococcus, and Ruminococcus. In some studies, Coprococcus and Ruminococcus were significantly more abundant in obese individuals, while others reported the opposite. Notably, Lachnospira and Lactobacillus were more consistently associated with obesity [15]. In our initial unadjusted analyses using the Kruskal-Wallis, S. copri, Coprococcus, Vescimonas, and Lactobacillaceae showed significant differences in relative abundance across BMI groups. However, when linear regression models were applied, adjusting for age and sex, only Ruminococcus and Lachnospira were significantly and negatively associated with BMI. In our study, the Kruskal-Wallis test revealed that S. copri was more abundant in obese individuals, consistent with the findings of Duan et al., [16].

Previous studies examining the F/B ratio in obese individuals and healthy weight individuals have yielded inconsistent findings; for instance, a study conducted in Ukraine reported an elevated F/B ratio among obese participants [17], whereas a study in China reported a decreased F/B ratio [16]. A meta-analysis indicated no significant difference in this ratio [18]. Our study revealed that the F/B ratio and alpha diversity decreased with increasing BMI. Specifically, the F/B ratio (p < 0.004) and alpha diversity (p < 0.005) were significantly greater in healthy weight participants than in obese participants. However, although the F/B ratio is often cited in obesity-related microbiota research, it remains a controversial and inconsistent marker and should therefore be interpreted carefully in light of its limitations.

S. copri, previously identified as Prevotella copri, is a Gram-negative bacterium capable of fermenting various carbohydrates [19, 20]. It was the dominant species in 51.2% of our study participants and is generally found in the gut microbiota of populations with traditional diets in rural Mediterranean and African regions [21, 22]. S. copri produces short-chain fatty acids and detoxifies superoxide radicals, which help protect the mucosal barrier and prevent inflammation [22]. Medina-Vera et al. reported that the abundance of S. copri significantly increased in individuals with type 2 diabetes [23].

Ruminococcus, a Gram-negative bacterium, breaks down carbohydrates such as galactose, glucose, fructose, sucrose, maltose, and lactose. Previous studies have reported higher Ruminococcus ratios in individuals with carbohydrate-dominant or vegan/vegetarian diets [24–26]. Ruminococcus genus has been associated with various health conditions, including non-alcoholic fatty liver disease (NAFLD) and coronary artery disease (CAD). Ruminococcus species are found in lower amounts in NAFLD patients, and this has been linked to hepatic gene expression associated with inflammation and metabolic pathways [27].

P. vulgatus, formerly classified under the Bacteroides genus, is a Gram-negative, anaerobic bacterium that utilizes carbohydrates and has anti-inflammatory effects through short-chain fatty acid production and the production of capsular polysaccharides [28, 29]. A previous study identified Phocaeicola as a dominant species in children at high risk for type 1 diabetes [30].

F. prausnitzii, known for fermenting prebiotics and producing important inflammatory modulators such as butyrate and sialic acid, constitutes approximately 5% of the healthy microbiota [31–33]. In our study, the percentage of F. prausnitzii ranged from 1 to 23% among the participants. Despite previous findings of low F. prausnitzii levels in diabetic patients, no significant difference was observed in our study between F. prausnitzii levels and glucose risk, insulin resistance, or HbA1c levels [34].

Although S. copri, F. prausnitzii, Ruminococcus, and Phocaeicola were frequently identified in our study, none showed statistically significant associations with cardiometabolic parameters. Further research is needed to clarify the potential roles of these genera in metabolic health.

In our study, the genera Ruthenibacterium, Lachnospira, and Vescimonas were found to be associated with cardiometabolic parameters such as HDL, HbA1c, and HOMA-IR, respectively. Ruthenibacterium is an obligate anaerobic, Gram-negative bacterium that produces fermentation products such as D-lactate and succinate [35]. Although a previous study reported that an increased abundance of Ruthenibacterium lactatiformans may contribute to elevated cardiovascular risk [36], our findings revealed a positive association between Ruthenibacterium and HDL levels, suggesting a potential protective effect against cardiovascular risk. Lachnospira was found to be negatively associated with HbA1c, indicating that it may play a beneficial role in glycaemic regulation. Additionally, Vescimonas showed a negative association with HOMA-IR, implying that a higher abundance of this genus may be linked to lower insulin resistance.

Lactobacillaceae, comprising approximately 0.01–1.8% of the intestinal microbiota, are probiotics that protect against pathogens and modulate the immune system [37]. In our study, the most frequently identified Lactobacillus species in the gut microbiota of the participants were Ligilactobacillus ruminis, Ligilactobacillus salivarius, Limosilactobacillus reuteri, and Lactobacillus gasseri. Consistent with the findings Armougom et al., [38] Lactobacillus genera were found to be more abundant in the gut microbiota of obese individuals than in healthy-weight individuals. These results could be caused by frequent consumption of fermented food in the population.

The eating habits of Türkiye’s Aegean region are shaped by its geographical and cultural characteristics, emphasizing local ingredients and traditional cooking methods. The Mediterranean diet, rich in vegetables, fruits, and wild greens, aligns with the region’s biodiversity and cultural history. Key dietary elements include monounsaturated fats, legumes, and cereals, linked to health benefits [39]. Influences from ancient civilizations, including the Persians and Byzantines, have further enriched the region’s cuisine [40]. However, studies show a decline in the Mediterranean diet among young adults in Izmir, with only 13% adhering to it, highlighting the challenges modern lifestyles pose to traditional practices [41]. This trend emphasizes the need for initiatives to preserve the region’s culinary heritage and promote its health benefits. Traditional foods in Türkiye, such as fermented beverages and dairy products, are rich in nutrients and play a role in cultural heritage. However, some of these foods are becoming less common, highlighting the need to preserve traditional dietary practices [42]. As we supported with the nutrition anamnesis from our participants legumes, nuts, whole grains are also largely consumed and aromatic herbs were used extensively in cooking. Our participants were also consuming high amount (one portion daily) of dairy products mainly in form of fermented products (kefir, cheese or yogurt). Higher portion of protein was consumed from plant-based sources and red meat consumed once a week in average as we learned from our nutrition anamnesis. However, dietary intake was assessed using a frequency-based questionnaire that captured how often participants consumed specific food items but did not quantify actual nutrient intakes such as total energy, carbohydrates, protein, or fat, which limits our ability to evaluate specific dietary contributions to microbiota or metabolic outcomes. We found no significant differences in dietary patterns between BMI groups, and thus did not include diet as a covariate in regression models; however, we acknowledge the lack of detailed nutrient data as a limitation.

Our study focused on the 20 most abundant bacterial genera, which were analysed individually as independent predictors concerning cardiometabolic health parameters. This targeted approach allowed us to explore associations between specific taxa and clinical markers. Similar strategies have been adopted in previous microbiome research investigating links with metabolic outcomes, such as the study by Galie et al. [43], which used linear regression to examine associations between microbiome-derived metrics and cardiometabolic parameters. Our use of standard linear regression, combined with adjustment for key covariates and False Discovery Rate correction, provided a robust and interpretable framework to identify biologically relevant associations.

Despite these strengths, several limitations should be considered. First, the sample size was relatively small, and all participants were recruited from a single urban area (Izmir province), which may limit the generalizability of our findings. Additionally, there was an imbalance in sex distribution and participant numbers across BMI categories (healthy weight, overweight, and obese). Another important methodological limitation is the use of BMI to define obesity. Although BMI is commonly used in epidemiological studies, waist circumference is considered a more accurate predictor of metabolic dysfunction and is recommended in the NCEP-ATP III criteria. Although statistically significant associations were observed between certain bacterial taxa and blood metabolic health markers—including HbA1c, HOMA-IR, and HDL levels—several potential confounding factors were not fully accounted for. These include detailed nutrient intake (e.g., total caloric and macronutrient consumption), physical activity levels, and the use of medications other than antimicrobials, proton pump inhibitors, or probiotic supplements—all of which may influence gut microbiota composition. However, a small number of participants were taking medications for other conditions due to the low and heterogeneous usage, which limited the feasibility of meaningful adjustment in the statistical models. Lastly, this study employed 16 S rRNA gene sequencing, which provides taxonomic resolution but lacks insight into the functional potential of microbial communities. Future research incorporating metagenomic or metabolomic approaches may help elucidate the biological mechanisms underlying the observed associations.

Conclusion

This study provides valuable insight into the composition of the gut microbiota in Turkish adults without diagnosed chronic diseases and its association with cardiometabolic parameters. The results indicated that the dominant bacterial taxa in the gut microbiota of individuals in our study population differ from those reported in other countries, suggesting potential regional or population-specific microbial signatures.

The findings demonstrate that key bacterial genera such as Lachnospira and Ruminococcus are negatively associated with BMI, suggesting a potential protective role against obesity. Additionally, Lachnospira abundance was inversely associated with HbA1c levels, highlighting its possible link to glycemic regulation. The relative abundance of Vescimonas was significantly predicted by HOMA-IR, indicating a relationship with insulin resistance, while Ruthenibacterium was positively associated with HDL cholesterol, suggesting a potential role in lipid metabolism.

These results contribute to the growing body of evidence highlighting the gut microbiota’s influence on metabolic health. They underscore the importance of considering microbiota profiles in diverse populations and support further research into the functional and therapeutic implications of microbial composition in cardiometabolic risk.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

BMI

Body mass index

HDL

High-density lipoprotein

LDL

Low-density lipoprotein

CRP

C-reactive protein

HOMA-IR

Homeostatic Model Assessment for Insulin Resistance

AIC

Akaike information criterion

Author contributions

[SSO, NA, EDG, MA, NDK, PA and AAO] actively participated in the acquisition of data, analysis, and interpretation, and also contributed to revising the manuscript and providing final approval of the version to be published. [FDOD, DA, PA and AAO] was involved in the conception and design of the study, drafted the article, and provided final approval of the version to be published.

Funding

This research was supported by the Dokuz Eylul University Research Foundation grant TSG-2023-3174. The funders had no role in the study design, data collection, or analysis.

Data availability

The sequencing datasets from this study have been deposited in the NCBI Sequence Read Archive (SRA). The accession number for these data is PRJNA1120247.

Declarations

Ethics approval and consent to participate

The study was approved by the Dokuz Eylul University Noninterventional Research Ethics Committee (ethics number 22.03.2023, 2023/09–33) and was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants included in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.