Functional dyspepsia (FD) is among the most prevalent functional gastrointestinal (GI) disorders, featuring complex chronic upper abdominal symptoms.¹ With its high prevalence and limited treatment options, FD detrimentally impacts quality of life and imposes substantial healthcare expenses.² FD has long been considered a syndrome caused by a psychopathological cause owing to the absence of macroscopic abnormalities during clinical investigations. However, the latest reports provide evidence that the duodenum or proximal small intestine may be key players in FD. A landmark study has brought about a paradigm shift in the etiology of FD, suggesting a significant association between duodenal eosinophilia and FD.³ Duodenal mast cell hyperplasia has also been identified in subjects with FD and other functional disorders.⁴ Patients with FD had impaired duodenal mucosal integrity with abnormally related gene expression compared to controls.⁵ These microscopic inflammatory and permeability changes in the duodenum may provoke localized motor-sensory abnormalities and systemic neuroimmune dysregulation via the gut-brain axis.

Despite increasing knowledge of duodenal inflammation in FD, few studies have investigated the causative factors (diet or microbiota) that cause intestinal inflammation and impaired duodenal mucosal integrity. We can assume that a certain type of diet (or nutrients) or human microbiota environment is associated with the symptoms of FD. Therefore, comprehensive studies are needed that investigate various luminal factors, such as dietary intake (nutrients), from the oral to stool microbiota, and FD symptoms (Supplementary Fig. 1). This prospective case-control study with a network analysis aimed to determine the interconnection of FD symptoms, dietary nutrition, duodenal inflammation and permeability, and the oral and gut microbiota.

1. Patients and controls

Patients aged 20 to 75 years who were indicated for gastroduodenoscopy with duodenal biopsies were prospectively recruited at the outpatient clinic of the Department of Gastroenterology, Dongguk University Ilsan Hospital, a tertiary referral hospital.

Exclusion criteria included a history of peptic ulcer or major abdominal surgery, Helicobacter pylori infection (both past and present), intestinal metaplasia or gastric cancer, abnormal biochemistry laboratory results (aspartate transaminase and alanine transferase 1.5 times the upper limit of normal, total bilirubin level >1.5 mg/dL, renal function test–creatinine clearance <30 mL/min, uncontrolled hypertension/diabetes, platelet count <10⁵/mm³, prothrombin time and activated partial thromboplastin time increased by more than 20% of upper normal), chronic alcoholics, taking antiplatelet, anticoagulant, or nonsteroidal anti-inflammatory drugs, corticosteroids, antibiotics, probiotics or other immunosuppressive drugs. Drugs that potentially affect gastric acid secretion (proton pump inhibitors or potassium competitive acid blockers) were discontinued at least 4 weeks before the start of the study and forbidden for the entire study course. Before enrollment, upper abdominal ultrasound or computed tomography findings within the last 6 months were reviewed for all patients to confirm the absence of underlying organic or systemic disease.

The control group included healthy participants without any upper GI symptoms. This group consisted of outpatient individuals who were willing to take diagnostic esophagogastroduodenoscopy (without colonoscopy) despite the absence of significant GI symptoms.

2. Study design

This prospective case-control study included patients diagnosed with FD-postprandial distress syndrome (PDS) (FD with predominant PDS rather than epigastric pain syndrome symptoms) according to the Rome IV criteria,⁶ and healthy controls. Patients with FD-epigastric pain syndrome were excluded from this study because the symptoms could overlap with non-erosive reflux esophagitis.

On visit 1, eligible patients completed the food frequency questionnaire (FFQ) of the 7th version of the Korea National Health and Nutrition Examination Survey,⁷ which is used to investigate patients’ dietary nutrients, use of nutritional supplements, and physical activity (in metabolic equivalents of task). The patients’ current medications were reviewed, and medications that had to be discontinued for the study were identified. A run-in period of 4 weeks was given prior to the gastroduodenoscopy (Fig. 1). In addition, a stool sampling kit was provided to each participant so that feces could be collected and submitted to the endoscopy room within 2 days before the endoscopy. At visit 2, gastroduodenoscopy was performed to rule out any structural abnormality of the upper GI tract, a rapid urease test (CLO) to detect H. pylori infection, and duodenal mucosal biopsies to analyze the degree of duodenal inflammation, permeability, and mucosa-associated microbiota. While stool samples were collected by the patients themselves, a single participating researcher (Y.C.) performed oral microbiota sampling through a swab of the patient’s tongue base, immediately before sedation for the endoscopy. At the last visit (visit 3), clinicians explained the test results to the participants, and any presence of adverse reactions according to the study was evaluated.

Figure 1. Study design. LPDS, Leuven Postprandial Distress Scale.

All study procedures were approved by the Institutional Review Board of Dongguk University Ilsan Hospital, South Korea (IRB number: DUIH 2021-06-016-005) and performed in accordance with the Declaration of Helsinki. This study protocol is publicly available from the Clinical Research Information Service and the World Health Organization International Clinical Trials Registry Platform. All study participants provided informed consent prior to their inclusion in the study. To recruit voluntary patients, both groups of participants were provided with individualized dietary intake and nutrient status information collected during the study process, and free nutritional counseling sessions were conducted accordingly. All authors reviewed and approved the final manuscript.

3. Evaluation of symptom severity

The severity of symptoms associated with FD-PDS and the control group were assessed using a set of questionnaires at visit 1. The questionnaire used was the Leuven Postprandial Distress Scale (LPDS), which is a validated patient-reported outcome instrument for FD-PDS.⁸ The LPDS consists of eight questions, each of which is answered on a 5-point (0–4) scale (Supplementary Fig. 2). These questions can be further categorized into three cardinal PDS symptoms: early satiation, postprandial fullness, and upper abdominal bloating. We also assessed symptom scores for each subgroup of PDS symptoms.

4. Duodenal biopsies and brushing

During gastroduodenoscopy, biopsy and brushing samples were obtained from the second portion of the duodenum. The sequence of these two tests was first, duodenal mucosal brushing was performed repeatedly at the duodenal 2nd portion, and then mucosal biopsy was performed at a location where brushing was not affected (opposite side of the ampulla of Vater). Duodenal mucosal brushing was performed using cytology brushes (BC-24Q; Olympus, Tokyo, Japan). Swab samples for 16S rRNA sequencing analysis were stored at –20°C. Two biopsy samples were taken using biopsy forceps (EndoJaw biopsy forceps FB-231KA; Olympus) from the opposite side of the ampulla of Vater with specimen diameter of more than 5 mm. One of the biopsies for immunohistochemical analysis was kept in 4% formalin solution, and the other for RNA isolation was immediately stored in liquid nitrogen and then stored at –80°C until use.

During the endoscopic procedure, several efforts have been made to minimize contact with the oral cavity and saliva. This has been achieved by using sterile mouthpieces and employing the chin lift maneuver. Additionally, tongue microbiota sampling was conducted separately from the endoscopic procedures, and continuous staff education has been implemented.

5. Oral and stool sampling

Participants were guided in advance so as not to brush their teeth 8 hours prior to oral swabbing. Oral samples were collected from the deep base of the tongue by swabbing from a single investigator (Y.C.). The samples were then stored at –20°C until for bacterial DNA isolation. Stool samples were individually collected and submitted by the patient according to the guidance of the research nurse using a stool collection and stabilization kit with toilet paper (OMNIgene GUT, OM-200; DNA Genotek, Ottawa, ON, Canada). Since this kit maintains stable microbiota profiles for 60 days even at room temperature without a cold chain requirement, samples submitted within 2 days prior to the endoscopy were accepted.

6. Experimental methods

1) Immunohistochemistry

Biopsy samples were fixed in 4% formalin buffer for 1 day and dehydrated using a series of increasing concentrations of alcohol followed by xylene. The tissues were embedded in paraffin blocks and sectioned into 4-μm slices using a Leica RM2235 microtome (Leica, Nussloch, Germany). The sections were placed on glass slides (Muto, Tokyo, Japan), deparaffinized with xylene, and rehydrated using a series of decreasing concentrations of alcohol. The slides were boiled in 10 mM sodium citrate buffer (pH 6.0) for 10 minutes. After being blocked with 3% goat serum for 20 minutes at room temperature to minimize nonspecific staining, the slides were incubated with mouse anti-mast cell tryptase (sc-33676; Biotechnology, Santa Cruz, CA, USA), mouse anti-MBP (PA-78397; Invitrogen, Waltham, MA, USA), mouse anti-CD8 (IR62361; Agilent Technologies, Santa Clara, CA, USA), and rabbit anti-CD3 (MA1-90582; Invitrogen) for 1 hour at room temperature. After being washed, the slides were incubated with the secondary antibodies for 1 hour at room temperature. After being washed, the slides were treated with horseradish peroxidase–streptavidin (PK-7800; Vector Laboratories, Burlingame, CA, USA) at room temperature for 1 hour. The signals were visualized using a diaminobenzidine peroxidase substrate kit (SK-4100; Vector Laboratories, Newark, CA, USA) and hematoxylin staining. Images were captured using a DP70 digital camera (Olympus) at ×200 magnification.

2) Quantitative real-time PCR

Total RNA was extracted from the duodenal biopsy tissue using TRIsure™ reagent (Bioline Reagent, London, UK) according to the manufacturer’s instructions. The quantity and quality of the extracted RNA were verified by measuring the optical densities at 280 and 260 nm using a NanoDrop spectrophotometer (Implen, Munich, Germany). Circular DNA was synthesized by the reverse transcription of 1 μg of extracted RNA using Oligo(dT) 18 primer (Thermo Fisher Scientific, Waltham, MA, USA) and an RT PreMix kit (Bioneer, Daejeon, Korea). Real-time polymerase chain reaction (PCR) was performed on a LightCycler 96 Instrument (Roche Diagnostics, Rotkreuz, Switzerland) in a 96-well plate using SYBR Green real-time PCR Master Mix (Toyobo, Tokyo, Japan). The amplification reactions were performed according to the manufacturer’s instructions in a total volume of 50 μL of PCR mixture containing 2 μL of circular DNA, 10 pmol of specific primer sets, 20 μL of SYBR Green master mix, and 26 μL of nuclease-free water. The following conditions were applied for the PCR amplification: an initial denaturation step at 95°C for 10 minutes followed by 45 cycles of amplification involving denaturation at 95°C for 10 seconds, annealing at 55°C to 62°C for 5 seconds, and extension at 72°C for 10 seconds. The obtained PCR data were processed and analyzed using a dedicated Light Cycler software (version 1.1; Roche, Basel, Switzerland) and normalized using glyceraldehyde-3-phosphatase dehydrogenase (GAPDH) as the housekeeping gene. Relative gene expression levels were quantified using the standard 2-ΔCt estimation method, in which Ct represents the crossing threshold value derived by the software with Ct=(Ct-target gene–Ct-GAPDH).

3) 16S rRNA sequencing

Bacterial DNA from the oral, duodenal, and stool samples was isolated using a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany) by applying the method described in our previous report.⁹ PCR of the V1–V2 region of the 16S rRNA gene sequences was performed using a C1000 Touch thermal cycler with a 96-deep well reaction module (Bio-Rad, Hercules, CA, USA). The primer set used for this reaction contained a unique 8-base barcode to tag the PCR products of the samples. The PCR products were purified using a QIAquick PCR Purification Kit (QIAGEN). Amplicons from each sample were pooled in equimolar amounts. Amplicon sequencing libraries were constructed using an Ion Plus Fragment Library kit (Thermo Fisher Scientific) and quantified using high-sensitivity DNA chips and a Bioanalyzer 2100 (Agilent Technologies). Sequencing reactions were performed using an Ion Torrent Personal Genome Machine (Ion PGM, Thermo Fisher Scientific) according to the manufacturer’s instructions. Raw sequence reads were quality-filtered, and taxonomy assignment was conducted using the Silva database using the Quantitative Insights into Microbial Ecology 2 software package.¹⁰ Alpha diversity was calculated using the observed features and the Chao1 estimator. The total structural changes in the gut microbial composition were analyzed using UniFac-based principal coordinate analysis to reveal the clustering pattern of the microbial composition in each experimental group. The enterotypes were generated using mathematical methods.¹¹

4) Nutrition assessment and network analysis

Participants’ dietary information was obtained by using a FFQ.¹² The FFQ was validated against dietary reports collected over a 9-month period.¹³ Through this, it is possible to investigate the average nutrient intake of an individual for a day, a week, and a year. Nutrient intakes were energy-adjusted by using the residual method.¹⁴

The Pearson correlations between microbiota, genes, symptoms, nutrition, and other markers were analyzed using the Scipy library of the Python programming language¹⁵ with statistical significance at the level of 0.05. The correlation network was visualized using Cytoscape 3.9.1.¹⁶ A heatmap was generated to demonstrate the correlation between all markers using the Pearson correlation test. A network analysis was generated to demonstrate the same as a heatmap and considered significant at values of p<0.05.

1. Study population

Twenty-eight participants (12 patients who fulfilled the Rome IV criteria and 16 healthy controls) were enrolled. One patient in the control group was excluded from the study because severe gastric atrophy was observed during gastroduodenoscopy. As a result, 12 FD patients and 15 healthy controls were included in the analysis. All 27 patients had negative results on the CLO test. The participants’ demographic and clinical characteristics are presented in Table 1. Demographic variables, such as age and sex, showed no significant intergroup differences. Other indicators, such as body mass index, family history of gastric cancer, education level, and food allergies, were not significantly different. Analysis of the LPDS in the three different subgroups according to symptom categories, postprandial fullness, epigastric pain, and nausea were significantly severe in the FD group.

Table 1 . Demographic and Clinical Characteristics of the Participants.

Characteristic	Functional dyspepsia (n=12)	Control (n=15)	p-value
Female sex	9 (75.0)	11 (73.3)	0.922
Age, yr	43.1±3.5	35.6±2.7	0.671
Range	27–61	24–59
BMI, kg/m2	23.1±0.7	24.3±1.1	0.298
Family history of gastric cancer	3 (25.0)	4 (26.7)	0.460
Education (above university graduation)	7 (58.3)	13 (86.6)	0.224
Food allergy	1 (8.3)	1 (6.7)	0.701
Physical activity, METs	14.7±4.1	21.5±9.8	0.283
LPDS (postprandial fullness)	10.0±1.0	3.2±0.1	<0.001*
LPDS (epigastric pain)	4.9±0.7	2.0±0.1	0.003*
LPDS (nausea)	1.9±0.4	1.0±0.1	0.008*

Data are presented as number (%) or mean±standard error of mean..

BMI, body mass index; MET, metabolic equivalents, LPDS, Leuven Postprandial Distress Scale..

*p-value <0.05 is considered significant..

2. Increased duodenal inflammation with eosinophil and mast cell infiltrations in FD

Low-grade inflammation of the duodenal mucosa was evaluated by staining for intraepithelial lymphocytes, eosinophils (major basic protein), and mast cells (CD117) in the tissue. Compared to healthy controls, patients with FD showed increased CD8, eosinophil, and mast cell counts (Fig. 2). The number of mast cells correlated well with that of CD8 cells. In contrast, no significant difference in CD3 cell counts was observed between patients and controls.

Figure 2. Increased intraepithelial lymphocytes infiltration supporting the low-grade inflammation in the duodenal mucosa in functional dyspepsia (FD) patients. (A) Images of CD3, CD8, major basic protein (MBP), and mast cell immunohistochemistry were performed on a biopsy specimen. The upper images are from a representative healthy control individual, and the lower aisle images are from a representative FD patient. (B) Comparison of immune cell counts per area (mm²) between FD and control groups. The count of eosinophils and mast cells in the duodenum per square millimeter (mm²) was determined using hematoxylin and eosin and c-kit staining, respectively. The counting was conducted across three distinct regions, and subsequently, the mean value was calculated for each subject, providing a comprehensive analysis. The bold bar on the left is the immune cell count/area of the control group, while the shaded bar on the right represents that of the FD group. While CD3 counts/area did not differ significantly between the two groups, CD8, eosinophil, and mast cell counts were significantly increased in the FD group. *p<0.001, ^†p<0.0001.

3. Impaired expression of the intercellular adhesion proteins

The mRNA expressions of ZO-1, claudin-2, and occludin, which are tight junction proteins related to the gut barrier integrity of the duodenal intestinal mucosa, were investigated (Fig. 3). As a result, expressions of proteins such as ZO-1, claudin-2, and occludin were decreased in FD. These results suggest impaired duodenal mucosal integrity in FD patients due to reduced expression of tight junction proteins.

Figure 3. mRNA expression levels of ZO-1, claudin-2 and occludin, the major tight junction proteins. The mRNA expression level of the control group was used as the reference value, and the relative expression in the functional dyspepsia (FD) group was investigated. Occludin expression was significantly decreased in FD (p<0.001), and other tight junction proteins (ZO-1 and claudin-2) also showed a trend towards reduction in FD.

4. Microbial alpha and beta diversity in FD and controls

We conducted a principal coordinate analysis to confirm the distinct clustering pattern of microbial population among each sample (oral, duodenal, and stool). By analyzing the clustering pattern of microbial population among samples, we confirmed that the duodenal and oral samples had similar microbial ecosystem diversity, while the stool samples had an independent pattern (Supplementary Fig. 3).

Alpha diversity is used to check microbial diversity in local samples. Chao1 is a verified alpha diversity measure that estimates species richness based on confirmed species information. We screened for a difference in alpha diversity between the FD and control groups and found no significant difference in species diversity in the oral, duodenal, and stool samples (Supplementary Fig. 4A). Even in the beta diversity analysis through principal coordinate analysis, no clear distance was observed between the ecosystem of FD and the controls at any location in the GI tract (Supplementary Fig. 4B). However, there was a significant overlap and similarity in microbial composition between the oral and duodenal ecosystems.

5. GI microbial changes in FD

We compared the abundance of bacterial genera in the tongue base, duodenal mucosa, and stool of FD patients and healthy controls between the two groups (Fig. 4). First, in the duodenal mucosa, genus Streptococcus, a well-known oral symbiont, increased significantly compared with that in the control group (p=0.014). In locations other than the duodenum, oral Neisseria genus were significantly increased in FD patients compared to controls (p=0.005), and Faecalibacterium genus (p=0.047) and butyrate-producing bacteria Butyricicoccus (p=0.028) were observed less frequently in the stool samples.

Figure 4. Characteristic microbiota genera of functional dyspepsia (FD) patients determined by 16S rRNA sequencing. From left to right, an increase in oral Neisseria (p=0.005), increase in duodenal Streptococcus (p=0.014), and decrease in stool Faecalibacterium (p=0.047) and Butyricicoccus (p=0.028) were notably found in the FD patients. *p<0.05, ^†p<0.01.

We also analyzed the association of these microbiota with FD symptoms and markers of duodenal mucosal integrity. A strong inverse relationship (p=0.004) was found between stool Butyricicoccus and the severity of FD symptoms and their subcategories (Fig. 5A). Oral Neisseria also showed a relatively strong positive correlation with postprandial fullness; however, this was not statistically significant. In addition, the relative abundance of stool Butyricicoccus showed the most significant relationship with intercellular proteins (ZO-1, occludin, and claudin-2) (Fig. 5B), demonstrating that patients with elevated stool Butyricicoccus showed higher intercellular tight junction protein expression. On the other hand, there was a trend toward a leaky gut as oral Neisseria and duodenal Streptococcus genera increased.

Figure 5. Heatmap analysis of the gut microbiota. (A) Gut microbiota and symptom correlation. Stool Butyricicoccus was negatively correlated with functional dyspepsia symptom severity. (B) Relationship between the gut microbiota and tight junction proteins. Stool Butyricicoccus was significantly correlated with increased expression levels of ZO-1, occludin, and claudin-2, which are essential proteins for duodenal epithelial integrity. LPDS, Leuven Postprandial Distress Scale.

6. Dietary nutrition and microbial relationships in FD

According to a quantitative analysis of age- and energy-adjusted dietary nutrition of FD patients and healthy controls using the data obtained from the FFQ (Fig. 6A), FD patients showed lower protein (p=0.002), fat (p=0.007), polyunsaturated fatty acids (PUFA; p=0.022), and vitamin B₂ (p=0.014). The other dietary nutrients were similar between the two groups. Meanwhile, a heatmap analysis of each dietary nutrient with the relative abundance of GI microbiota (Fig. 6B) showed that increased dietary calcium was related to reduced oral Neisseria (p=0.035), while dietary carbohydrate was negatively correlated with duodenal Streptococcus (p=0.030). In contrast, the nutrient significantly related to the increase in stool Butyricicoccus was valine (p=0.045).

Figure 6. Dietary nutrition and gastrointestinal microbial relationships. (A) Functional dyspepsia (FD) patients showed lower dietary intakes of protein (p=0.002), fat (p=0.007), PUFA (p=0.021), and vitamin B₂ (p=0.014) than healthy control individuals. (B) The relationship between dietary nutrition and gastrointestinal tract microbiota schematized through a heatmap analysis. Bright red indicates a strong positive correlation, while bright green suggests a strong negative correlation. Notably, dietary calcium, carbohydrate, and valium were correlated with microbiota associated with FD (p<0.05). PUFA, polyunsaturated fatty acids. *p<0.05, ^†p<0.01.

7. Network analysis

A network analysis was performed to graphically confirm the correlation and the direction between the various comprehensive data collected (Supplementary Fig. 5). Duodenal Streptococcus, stool Butyricicoccus were associated with FD symptom severity. Of particular note was that the increase in fecal Butyricicoccus was more prominent in the healthy controls than in the FD patients. In contrast, the expression of duodenal mucosal integrity proteins such as occludin and ZO-1 was positively correlated with total protein, PUFAs, and valine.

Since Talley et al.³ and Vanheel et al.⁵ first reported an association between FD and duodenal eosinophilia and impaired mucosal integrity, the duodenum has attracted attention as a breakthrough target to overcome these treatment limitations.¹⁷ Historically, the duodenal ecosystem was considered to be sterile, with microbes present only because of cross-contamination. However, accumulating evidence suggests that intestinal dysbiosis is associated with FD.¹⁸ The causative factors (such as dietary nutrition and oral microbiota) related to changes in the duodenal ecosystem have not been elucidated. In addition, there is a need to identify an effective noninvasive biomarker that can represent the status of the duodenal ecosystem without endoscopy or duodenal biopsies. In our research, we found that the average eosinophil count in patients with FD was increased to 192.2/mm², which is approximately equivalent to 48 eosinophils per high-power field. This number is in close agreement with the high levels reported by Vanheel et al.,¹⁹ which showed an eosinophil count of 241/mm² in FD patients. Given the adaptive role of eosinophilic cells in the immune response to various microbial and environmental stimuli, we agree with the idea that the elevated eosinophil levels as a contributing factor for the symptoms experienced in patients with FD.²⁰ Therefore, the discovery of these unmet needs will lead to the potential correction of the patient’s lifestyle, earlier diagnosis and treatment, and more accurate tracking of FD.

In this study, we recognized a link between oral and gut microbiota, that oral and duodenal microbiota show similar microbial diversity in FD patients. We also investigated duodenal pro-inflammatory markers, expression of intercellular adhesion proteins, and related GI microbial changes in FD using real-time PCR and 16S-rRNA sequencing. Our results are consistent with those of previous studies, in which the increase in eosinophils and mast cells was significantly increased in the FD group compared to the controls, and an increase in duodenal Streptococcus.^21-23 The relationship between duodenal mucosal inflammation and FD symptoms, as demonstrated in previous studies, served as a means to validate the reliability of the samples obtained from the study cohort.

The two main new findings of our research are as follows: the identification of previously unidentified novel microbial genera associated with FD in stool samples, and the relationship between dietary micronutrition and FD. Our data suggest that duodenal Streptococcus appears to be associated with FD. This corresponds to the fact that oral pathobionts such as Streptococcus and Fusobacterium induce intestinal dysbiosis and leaky gut syndrome by suppressing the expression of ZO-1 and occludin.²⁴ However, our data demonstrated that the abundance of duodenal Streptococcus seemed more closely related to epigastric pain and nausea than postprandial fullness. Therefore, it was recognized that it may not be a microbial indicator that accurately reflect PDS symptoms. Meanwhile, this study recognized a decrease in stool Butyricicoccus as a characteristic of patients with FD. This butyrate producing Butyricicoccus also showed the strongest and most significant relationship with duodenal intestinal microbiota and FD symptom severity. Butyrate is believed to alleviate gut inflammation by coupling with cell surface G protein–coupled receptor 43.^25,26 In this regard, a reduction in fecal Butyricicoccus might be a significant FD-PDS predictive biomarker. This strain is also expected to be investigated as a potential next-generation probiotic strain. In previous literature, patients with inflammatory bowel disease had lower numbers of Butyricicoccus bacteria in their stools.²⁷ Besides, administration of Butyricicoccus pullicaecorum attenuated trinitrobenzenesulfonic-induced colitis in rats supporting the idea that Butyricicoccus are attractive as probiotics as they prevent cytokine-induced epithelial integrity loss in vitro cell culture model.²⁷

Another notable finding of our study was that dietary micronutrients such as valine were significantly related to FD traits. Our study showed that increased valine intake is associated with an advantageous GI ecosystem, a finding that is compatible with previous reports that dietary supplementation of valine has a beneficial effect on intestinal barrier function and microbial homeostasis in fatty liver disease patients.²⁸ Furthermore, a decreased intake of carbohydrates showed a significant association with an increased abundance of Streptococcus in the duodenum, implying a potential link to the FD symptoms. Similarly, a recent large-scale cross-sectional study reported that patients with dyspepsia had higher dietary fat and lower carbohydrate intake than healthy controls.²⁹ Our heatmap analysis showed similar results, supporting previous evidence. PUFA, vitamin B₂, protein, and carbohydrates are primarily abundant in fish such as salmon and tuna, whole grains like rice and wheat, and green vegetables. Consequently, this study suggests that the dietary habits of FD patients are characterized by relatively lower consumption of these food sources. In our study, we also found that integrity proteins like ZO-1 and occludin do show strong correlations with most nutrients through the Pearson analysis. However, these correlations are statistically significant only for certain nutrients. We chose to emphasize the relationship with total protein, PUFA, and valine, as these are more closely aligned with the primary objectives of our paper.

To our knowledge, our study is the first to analyze comprehensive data from dietary nutrition to oral and duodenal microbiome, controlling for confounding variables in the FD and control groups. Efforts to restrict the administration of proton pump inhibitors and other drugs, selectively recruit FD patients who have PDS-oriented symptoms rather than epigastric pain syndrome, and match age and sex between groups were made to ensure the reliability of the study results. Nevertheless, this study had several limitations. First, it is difficult to prove causality between categorized data. These new suggestions made by our study should be verified in additional studies to reveal the pathophysiological mechanisms. In addition, the dietary pattern of patients with FD may be the result of conscious lifestyle changes. Therefore, the findings should be interpreted with caution. Patients may have already had many experiences in which a certain nutrient-rich diet worsened their symptoms, and therefore, they adapted by avoiding these foods. Hence, one should be careful when proposing changes to dietary patterns based on the results of this study. Another issue to consider is the possibility of the coexistence of FD and irritable bowel syndrome among the participants. FD and irritable bowel syndrome can have similar or overlapping symptoms, and in some cases, they can occur simultaneously. Therefore, it is important to acknowledge the potential influence of irritable bowel syndrome in interpreting the research results. In addition, despite the various preventive efforts taken, it is challenging to completely exclude the possibility of contamination from oral microbiota during the endoscopic biopsy process. During the mucosal biopsies and swabbing during standard endoscopic procedures, for technical reasons, it is potentially contaminated by GI luminal contents.³⁰ The use of a non-Korean language validated LPDS questionnaire, relatively small sample size of the FD group and possible recall bias from survey-based nutritional analysis are additional study limitations. In addition, the authors call for further research and consensus to clarify how the duodenal eosinophilia commonly seen in FD patients, including those in our study, relates to eosinophilic duodenitis.

In conclusion, our findings indicate that individual dietary nutrition and gut microbiota are associated with FD symptoms as well as impaired duodenal barrier function and inflammation. This study presents clear, comprehensive evidence of altered duodenal ecosystems in patients with FD, with the discovery of a novel stool biomarker that strongly represents the symptoms and deteriorated GI ecosystem of FD. We believe that our results will not only contribute to early diagnosis and more effective prevention of FD but also help develop therapeutics based on novel mechanisms in the future.

This work was supported by the National Research Foundation of Korea Grant funded by the Korean Government (NRF-2022M3A9E4017033 and NRF-2021R1I1A1A01054673), main Research Program of the Korea Food Research Institute (KFRI) funded by the Ministry of Science and ICT (grant number E0170600-06).

No potential conflict of interest relevant to this article was reported.

Study concept and design: S.H.K., H.K., Y.C. Data acquisition: S.H.K., Y.C. Data analysis and interpretation: Y.C., E.Y.L., J.E.L., E.J.S., J.O., Y.D.N. Drafting of the manuscript: S.H.K. Critical revision of the manuscript for important intellectual content: Y.C., H.K. Statistical analysis: S.H.K., Y.C. Obtained funding: S.H.K., H.K. Administrative, technical, or material support; study supervision: H.K. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl230130.

The raw sequencing data presented in this study are deposited in the DNA Data Bank of Japan (DDBJ) under accession number DRA015691.