Endoscopic swab sampling as a novel method for gastric Microbiome profiling: A pilot study compared with tissue biopsy

Abstract

The gastric microbiome is increasingly recognized in gastric carcinogenesis, extending beyond Helicobacter pylori to oral-associated bacteria. Tissue biopsy is the current standard but is limited in spatial coverage. This pilot study evaluated an endoscopic swab-based approach as an alternative. Paired swab and biopsy samples were obtained from the gastric antrum and body of 16 patients undergoing esophagogastroduodenoscopy (32 swabs, 32 biopsies). Microbiome profiling was performed using 16 S rRNA V3–V4 amplicon sequencing. Swabs demonstrated higher alpha diversity than biopsies (observed ASVs: 123.5 vs. 48.5, p < 0.001; Shannon index: 4.73 vs. 3.99, p = 0.003). Overall community structures did not differ significantly on robust principal component analysis based on Aitchison distance (p = 0.16 and 0.17, permutational multivariate analysis). Helicobacter was significantly enriched in tissue samples (94.0% vs. 30.7%, log₂ fold change [LFC] = 1.521, q = 0.020), whereas Fusobacterium showed higher abundance in swab samples (LFC = -1.514, q = 0.006). These findings demonstrate that gastric swabs yield microbial profiles comparable to biopsies, with greater diversity and practical advantages. The swab method may provide a less invasive and reliable alternative for gastric and gastrointestinal microbiome research.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32028-4.

Introduction

The gastric microbiome has gained increasing attention for its potential role in gastric carcinogenesis. While Helicobacter pylori (H. pylori) has long been established as the primary pathogen associated with gastric cancer^1,2, recent studies suggest that other microbial communities within the stomach, including oral-origin bacteria, may also contribute to its pathogenesis^3–7. As H. pylori is actively diagnosed and treated due to its well-established role in gastric carcinogenesis, its incidence and prevalence have declined, particularly in developed countries and younger generations^1,8,9. This epidemiologic shift highlights the growing relative importance of non-H. pylori gastric microbiota in gastric cancer pathogenesis. Gastric acidity plays a key role in shaping the gastric microbial environment, as bacterial load and composition vary depending on intragastric pH¹⁰. Hypoacidic conditions, such as those observed in chronic atrophic gastritis or after long-term mucosal inflammation, may permit colonization by bacteria of oral origin. Therefore, intragastric pH is a relevant contextual factor when evaluating the role of non–H. pylori microbiota in gastric carcinogenesis.

Various sampling methods have been used to study the gastric microbiome, including endoscopic gastric fluid aspiration, endoscopic brushing, surgical tissue sampling, and endoscopic biopsy^4,6,11–13. Among these, endoscopic tissue biopsy remains the most widely used method. However, this method is invasive and samples only a limited area of the gastric mucosa, which may not adequately reflect the overall gastric microbial landscape. Our previous study demonstrated that endoscopic gastric mucosal swabs could serve as a viable alternative to tissue biopsy for diagnosing H. pylori infection. Notably, the study also reported that swab samples contained more than 20 times the amount of H. pylori bacterial DNA compared to tissue samples, suggesting a potential advantage in the success rate of microbiome analysis¹⁴.

In this context, the present study was conducted to address the limitations of tissue biopsy and to evaluate the feasibility of an endoscopic swab technique for gastric microbiome assessment. The aim was to compare the swab method with the conventional tissue biopsy in terms of microbial comparability and to assess its possibility as an alternative.

Methods

Study subjects and ethics statement

Participants were recruited from an existing longitudinal cohort at Seoul National University Hospital that was established for long-term follow-up of patients who had undergone endoscopic submucosal dissection (ESD) for gastric adenoma, early gastric cancer, or early esophageal cancer. Patients with a history of gastrectomy for gastric cancer or other upper gastrointestinal malignancies were excluded. Eligible cohort members who voluntarily agreed to participate were prospectively enrolled at the time of scheduled endoscopy, between July 2024 and April 2025. Informed consent was obtained from all participants, and the study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. 1906-083-1040). The study was conducted in accordance with the Declaration of Helsinki.

Specimen collection and sampling protocol

For swab sampling, commercially available, pre-packaged sterilized cotton balls (Dong-A Sterilized Cotton Balls; Dong-A Pharmaceutical Co., Ltd., Gwangju, Republic of Korea) approved for mucosal and dermal use were utilized. While wearing sterile gloves, a small portion (approximately 5 × 5 × 5 mm) was aseptically separated and shaped. Sterile, single-use biopsy forceps were opened immediately before sampling, used to grasp the swab, and introduced through the endoscope’s working channel. The endoscopist gently encircled the swab along the gastric mucosal surface several times—once in the antrum and once in the body—without applying excessive pressure. To minimize possible oral or esophageal microbial carryover due to endoscope passage, the endoscope was maintained as a closed system before sampling, avoiding suction and fluid instillation until sample collection was completed. Biopsy sampling was performed after swab collection to avoid biopsy-induced bleeding effects that could alter the result. Two biopsy specimens were obtained from each site (antrum and body – 4 tissues obtained in total) and pooled by site to yield one sample per location, enhancing sequencing success. All swab and biopsy samples were snap-frozen in liquid nitrogen immediately after collection and stored at − 80 °C. The procedures for obtaining swab and tissue samples are illustrated in Fig. 1, and further methodological details are described in our previous study¹⁴.

Fig. 1.

Gastric Swab and Biopsy Sampling Procedure. Swab (blue arrows) and biopsy (gray dots) sampling locations are indicated in the schematic.

To evaluate potential contamination originating from the swab material or endoscope channel, two control samples were collected: (1) a blank sample (cotton ball processed for Next Generation Sequencing [NGS] without any contact) and (2) a channel sample (cotton ball passed through the working channel without mucosal contact). Both controls were handled using the same sterile forceps handling and sample processing procedures as applied to the swab and biopsy samples.

Clinical data collection

Clinical information was obtained from electronic medical records, including age, sex, body mass index, smoking and alcohol history, and comorbidities. Family history of gastric cancer, rapid urease test results, and the Operative Link on Gastric Intestinal Metaplasia (OLGIM) stage were recorded, along with H. pylori serum IgG and histopathological evaluation using Giemsa stain. Serum pepsinogen I, pepsinogen II, and gastrin levels were measured after at least 8 h of fasting when available. Histopathologic diagnosis and the presence and grade of intestinal metaplasia were determined by board-certified gastrointestinal pathologists.

DNA Extraction, library Preparation, and sequencing

DNA was extracted using the DNeasy 96 PowerSoil Pro Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. The V3–V4 region of the 16 S rRNA gene was amplified using primers 341 F (5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′) and 805R (5′-TCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′)¹⁵. Library preparation followed the 16 S Metagenomic Sequencing Library Preparation protocol (Part #15044223, Rev. B; Illumina, San Diego, CA, USA) using Herculase II Fusion DNA Polymerase and the Nextera XT Index Kit v2 (Illumina)¹⁶. Sequencing was performed on the Illumina MiSeq platform using a paired-end 2 × 301 bp configuration. A minimum DNA concentration of 5 ng/µL at the post-amplification library quality control step was required for subsequent sequencing process. Samples that remained below this threshold after two repeat library preparation attempts with increased sample DNA input were considered to have low product yield or insufficient DNA concentration and were excluded from downstream sequencing.

Raw sequence Processing, Microbiome analysis, and statistical analysis

Continuous variables were presented as median with interquartile range (IQR) and compared using the Wilcoxon rank-sum test. Categorical variables were expressed as number and percentage, with comparisons made using the chi-square test or Fisher’s exact test. For microbiome analyses, the Wilcoxon test was used for pairwise comparison of continuous outcomes (e.g., alpha diversity), and Fisher’s exact test for categorical comparisons. The Holm method was used to control the false discovery rate. Considering the compositional nature of microbiome data, robust principal component analysis (RPCA) based on Aitchison distance was performed for dimensionality reduction, and group-wise differences were evaluated using permutational multivariate analysis of variance (PERMANOVA). Analysis of Composition of Microbiomes with Bias Correction 2 (ANCOM-BC2) was applied to identify differentially abundant taxa. PERMANOVA and ANCOM-BC2 were conducted as direct inter-group comparisons without covariate adjustment, as the limited sample size of this pilot study precluded stable multivariable modeling. Statistical significance was defined as a p-value and q-value (adjusted for multiple comparisons) less than 0.05. A sensitivity analysis was performed after excluding sequences assigned to Helicobacter to evaluate microbial patterns independent of Helicobacter dominance.

Raw sequence processing was performed using QIIME 2 (version 2025.4; QIIME 2 Development Team)¹⁷, including adapter trimming with Cutadapt (version 5.1; Marcel Martin, Max Planck Institute, Tübingen, Germany)¹⁸, denoising and chimera removal with DADA2 (version 1.26.0; Callahan Lab, University of Oregon, Eugene, OR, USA)¹⁹, and multiple sequence alignment with MAFFT (version 7.526; Kazutaka Katoh, Osaka University, Osaka, Japan)²⁰. Each ASV was assigned to the organism with the highest similarity in the NCBI 16 S reference database (version NCBI_16S_20241203) using the Bayesian Naive Classifier implemented in DADA2 with a confidence threshold of 50%²¹. R (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria) was used for downstream statistical analyses and data visualization.

Results

Clinical features of study participants

A total of 64 gastric samples (32 paired swab and tissue specimens) were collected and analyzed from 16 patients enrolled in the study. Final pathological diagnoses included early gastric cancer in 10 patients (62.5%), gastric adenoma in 3 (18.8%), and superficial esophageal squamous cell carcinoma in 3 (18.8%). Operative Link on Gastric Intestinal Metaplasia (OLGIM) staging showed stage 0 in 5 patients (31.3%), stage 3 in 2 (12.5%), and stage 4 in 6 (37.5%); 3 patients (18.8%) were not assessed. H. pylori IgG serology was positive in 5 patients (31.3%), negative in 4 (25%), indeterminate in 2 (12.5%), and not performed in 5 (31.3%). Median serum pepsinogen I and II levels were 61.6 ng/mL (IQR 47.2–76.2) and 13.9 ng/mL (IQR 11.4–24.5), respectively, with a median pepsinogen I/II ratio of 3.4 (IQR 2.1–5.0); these measurements were not performed in 6 patients. Serum gastrin levels were measured in 13 patients, with a median of 56.0 pg/mL (IQR 43.7–146.0; institutional reference range 13.0–115 pg/mL), which did not suggest hypoacidity. All clinical features of the study participants are summarized in Table 1. As the cohort was not initially established solely for this study, certain clinical assessments were not performed for all participants, leading to missing data.

Table 1.

Clinical features of study Participants.

	N = 16
Sex
Female	2 (12.5%)
Male	14 (87.5%)
Age	68.0 (60.0, 70.5)
Body Mass Index	25.0 (22.2, 27.3)
Drinking	5 (31.3%)
Smoking	1 (6.3%)
Family history of gastric cancer	1 (6.3%)
Diagnosis
Early gastric cancer	10 (62.5%)
Gastric adenoma	3 (18.8%)
Superficial esophageal squamous cell carcinoma	3 (18.8%)
Rapid Urease Test
Negative	8 (50.0%)
Positive	5 (31.3%)
Not performed	3 (18.8%)
OLGIM
0	5 (31.3%)
3	2 (12.5%)
4	6 (37.5%)
Not performed	3 (18.8%)
Helicobacter pylori IgG
Negative	4 (25.0%)
Indeterminate	2 (12.5%)
Positive	5 (31.3%)
Not performed	5 (31.3%)
Serum Pepsinogen
I	61.6 (47.2, 76.2)
II	13.9 (11.4, 24.5)
I/II	3.4 (2.1, 5.0)
Not performed	6 (37.5%)
Serum Gastrin	56.0 (43.7, 146.0)
Not performed	3 (18.8%)
Helicobacter pylori Eradication regimen
Fexuprazan + Amoxicillin + Clarithryomcin	2 (12.5%)
Proton pump inhibitor + Amoxicillin + Clarithryomcin	2 (12.5%)
NA	12 (75.0%)
Hypertension	10 (62.5%)
Diabetes	5 (31.3%)
Cardiac disease	3 (19.3%)
Liver disease	3 (19.3%)
Chronic kidney disease	1 (6.3%)
Other cancer	2 (12.5%)
Others
Gout	1 (6.3%)
Meningioma	1 (6.3%)
Myasthenia gravis	1 (6.3%)
Panic disorder	1 (6.3%)
Parathyroid adenoma	1 (6.3%)

Data are presented as median (Q1, Q3) for continuous variables and number (percentage) for categorical variables. OLGIM, Operative Link on Gastric Intestinal Metaplasia.

Sample characteristics and NGS library quality control outcomes

A total of 95 samples were collected, including 16 blanks, 15 endoscopic channel samples, 32 swabs, and 32 tissue biopsies. All samples underwent DNA extraction and library preparation for NGS. After library quality control (QC), 74 samples proceeded to sequencing: 5 blanks, 5 channel samples, 32 swabs, and 32 tissue samples. The remaining 21 samples (11 blanks and 10 channel samples) were excluded due to low product yield or insufficient DNA concentration, defined as < 5 ng/µL at the post-amplification library QC step, based on the Illumina library preparation protocol¹⁶. DNA concentrations at the QC stage was significantly lower in blank and channel samples compared to swab and tissue samples (all p < 0.001). However, no significant difference was observed between blank and channel samples (blank: median 0.0 [0.0–1.5]; channel: 1.2 [0.0–7.2]; p = 0.17) (Fig. 2A). Among the 64 gastric samples that passed quality control, 32 were tissue biopsies and 32 were swab samples, with 16 obtained from the antrum and 16 from the body in each group. Table 2 summarizes histological characteristics by sampling method and location. H. pylori (by NGS result) was detected in 25% of all sample groups, with no significant differences by gastric location (all p >0.9). Intestinal metaplasia and Giemsa staining grades also showed no significant differences between antrum and body tissue samples (p = 0.4 and >0.9, respectively). As the cohort was not initially established solely for this study, certain histopathologic assessments were not performed for all participants, leading to missing data.

Fig. 2.

DNA concentration, alpha diversity, and bacterial composition across sample types. (A) DNA concentrations at the library quality control stage prior to sequencing process. (B) Alpha diversity indices (ASVs, Shannon, Simpson) by sample type. (C) Relative abundance of the top 20 bacterial genera across sample types. Each bar represents an individual sample. Only the top 20 genera are shown; others were excluded for visual clarity. Asterisks indicate statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant.

Table 2.

Sample characteristics by sampling method and gastric Location.

	Tissue			Swab
	Antrum N = 16	Body N = 16	p-value	Antrum N = 16	Body N = 16	p-value
Helicobacter pylori positive (Next Generation Sequencing)			> 0.9			> 0.9
Negative	12 (75.0%)	12 (75.0%)		12 (75.0%)	12 (75.0%)
Positive	4 (25.0%)	4 (25.0%)		4 (25.0%)	4 (25.0%)
Intestinal metaplasia			0.4
Absent	6 (37.5%)	5 (31.3%)
Mild	3 (18.8%)	3 (18.8%)
Moderate	2 (12.5%)	0 (0.0%)
Marked	2 (12.5%)	5 (31.3%)
Not performed	3 (18.8%)	3 (18.8%)
Giemsa stain			> 0.9
Absent	9 (56.3%)	10 (62.5%)
Mild	3 (18.8%)	2 (12.5%)
Moderate	0 (0.0%)	1 (6.3%)
Marked	1 (6.3%)	0 (0.0%)
Not performed	3 (18.8%)	3 (18.8%)

Data are presented as number (percentage). P-values were calculated using Fisher’s exact test.

Alpha diversity comparison between swab and tissue samples

In the comparison of alpha diversity metrics between swab and tissue samples, the swab group demonstrated higher diversity. The number of observed Amplicon Sequencing Variants (ASVs) was greater in swab samples (median: 123.5, IQR: 100.0–147.5) compared to tissue samples (median: 48.5, IQR: 29.0–65.5), with a statistically significant difference (p < 0.001). Shannon diversity was also higher in swabs (median: 4.7, IQR: 4.2–5.4) versus tissues (median: 4.0, IQR: 3.2–4.7), reaching significance (p = 0.003). However, Simpson diversity did not differ significantly between swab and tissue samples (median: 0.1, IQR: 0.1–0.2 vs. median: 0.1, IQR 0.1–0.2, respectively; p = 0.20). (Fig. 2B).

Microbial community composition by sample type and H. pylori status

RPCA based on Aitchison distance revealed no significant differences in microbial community composition between swab and tissue samples, regardless of H. pylori infection status (H. pylori-negative: PERMANOVA p = 0.16; H. pylori-positive: p = 0.17; Fig. 3A). Relative abundance analysis showed compositional differences between environmental samples (blank and channel), which were enriched in genera such as Sphingomonas, Acinetobacter, and Roseateles, and biological samples (swab and tissue), which showed higher abundance of genera such as Streptococcus, Prevotella, Veillonella, and Helicobacter (Fig. 2C). To evaluate these differences, PERMANOVA based on Aitchison distance was performed comparing environmental (blank and channel) and biological (swab and biopsy) samples, demonstrating a statistically significant separation between the two groups (p = 0.039; Supplementary Figure S1). In H. pylori-negative individuals, microbial profiles of swab and tissue samples were comparable (Fig. 3C). In contrast, H. pylori-positive individuals demonstrated a significantly higher relative abundance of Helicobacter in tissue samples compared to swab samples (median: 94.0%, IQR: 68.0–99.6% vs. median: 30.7%, IQR: 15.6–44.5%, p = 0.008) (Fig. 3B and C). Five biopsy samples showed markedly lower Shannon and Simpson diversity values, appearing as outliers, and these corresponded to cases with very high H. pylori dominance (median: 99.5%, IQR: 94.8–99.6%).

Fig. 3.

Gastric microbiome composition by sample type and Helicobacter pylori (H. pylori) status. (A) Robust principal component analysis (RPCA) of the gastric microbiome, stratified by sample type and H. pylori status. NGS, Next Generation Sequencing; PC, Principal Component; PERMANOVA, Permutational Multivariate Analysis of Variance. (B) Relative abundance of H. pylori by sampling method. Statistical significance was compared using the Wilcoxon test. Asterisks indicate the following significance level: ***p < 0.001. (C) Mean genus-level microbial composition by sampling method and H. pylori status. Only the top 20 genera are shown; others are grouped as “Others” for visual clarity. H. pylori status was determined by NGS results. Each bar represents the average value across sample types and locations.

Differentially abundant taxa between swab and tissue samples

Differential abundance analysis using ANCOM-BC2 identified genera with statistically significant differences between swab and tissue samples. Helicobacter was significantly enriched in tissue samples (log₂ fold change [LFC] = 1.521, q = 0.020), while Fusobacterium showed higher abundance in swab samples (LFC = −1.514, q = 0.006). Although other genera demonstrated directional differences—such as enrichment of Streptococcus, Prevotella, Peptostreptococcus, and Veillonella in swabs, and Staphylococcus and Lautropia in tissues—these did not reach statistical significance after multiple testing correction (q > 0.05) (Fig. 4).

Fig. 4.

Differentially abundant genera between swab and tissue samples identified by Analysis of Composition of Microbiomes with Bias Correction 2. Asterisks indicate statistical significance after multiple testing correction (*q < 0.05, **q < 0.01). False discovery rate was adjusted using the Holm method.

Sensitivity analysis excluding Helicobacter sequences

A sensitivity analysis excluding all sequences assigned to Helicobacter was performed to determine whether the observed results reflected true methodological and biological variation rather than a dilutional effect of Helicobacter dominance. After exclusion, swab samples continued to show higher alpha diversity than tissue samples, with greater ASV counts (swab: 121.5, IQR 102.0–144.5; tissue: 48.5, IQR 27.8–63.0; p < 0.001) and higher Shannon index values (swab: 3.44, IQR 3.07–3.81; tissue: 2.98, IQR 2.49–3.27; p < 0.001). Simpson diversity remained comparable (swab: 0.93, IQR 0.89–0.96; tissue: 0.92, IQR 0.87–0.94; p = 0.118) (Supplementary Figure S2). To assess community-level structure, a robust PCA–based PERMANOVA using Aitchison distance was performed in the H. pylori–positive samples after excluding all Helicobacter sequences. No significant difference in microbial composition was observed between swab and tissue samples (p = 0.304), consistent with the main analysis (Supplementary Figure S3).

These findings indicate that the greater diversity observed in swab samples reflects broader microbial niche capture rather than a dilutional effect of Helicobacter dominance, and that the microbial community composition remains comparable between swab and tissue samples in PERMANOVA even after excluding all Helicobacter sequences.

Discussion

In this study, we demonstrated that the novel swab-based method for gastric microbiome sampling is comparable to the conventional invasive tissue biopsy technique. Analysis of blank swabs and swabs passed through the endoscopic working channel indicated that contamination from the swab material or the channel did not substantially affect the microbial profiles of swab samples, although a minor concern regarding potential channel-related contamination still exists. The swab method yielded significantly higher alpha diversity indices than tissue samples, while overall microbial community composition remained comparable. These findings were consistent in a sensitivity analysis excluding all Helicobacter sequences, further supporting that the result reflects an intrinsic characteristic of the swab sampling method rather than a dilutional effect of Helicobacter dominance. Tissue samples showed a higher relative abundance of Helicobacter, whereas swab samples more frequently captured genera associated with the oral microbiome, such as Fusobacterium, Streptococcus, Peptostreptococcus, Veillonella, and Prevotella, which have been reported to be linked to gastric cancer development^5–7.

These findings suggest that the swab technique may serve as a lesser invasive and informative alternative to tissue biopsy for gastric microbiome profiling. Although the swab method still requires endoscopic procedure, the increasing prevalence of cardiovascular comorbidities has led to a growing proportion of patients receiving antithrombotic therapy, for whom repeated tissue biopsies may pose a bleeding risk. In this context, a swab method that does not require multiple biopsies from different gastric sites offers methodological advantages in selected patients. The higher alpha diversity observed in swab samples compared to tissue samples may reflect broader microbial niche coverage by swab, aligning with previous studies. Certain bacteria, such as H. pylori, preferentially adhere to the gastric mucosal epithelium through mechanisms including the type IV secretion system (T4SS), resulting in strong localization to the epithelial surface rather than the lumen or outer mucus layers. This epithelial attachment has been demonstrated by immunohistochemical visualization of H. pylori adherent to gastric epithelial cells^1,22–25. Similarly, studies of the colon using immunohistochemical staining have demonstrated that distinct bacterial communities occupy different spatial niches along the luminal and mucosal layers. These researches support the concept of spatial stratification of the microbiota within the gastrointestinal tract and reinforce the rationale that swab can capture a broader and more diverse microbial community^26,27. Tissue biopsies, which sample a limited portion of the mucosa, are more likely to capture mucosa-associated bacteria such as H. pylori, potentially underrepresenting microbes residing in other regions. This is supported by compositional differences observed between H. pylori-positive and -negative individuals. In H. pylori-positive samples, the relative abundance of Helicobacter was significantly higher in tissue samples compared to swabs, indicating a mucosal surface–oriented sampling tendency. Additionally, differentially abundant taxa analysis showed enrichment of oral-associated genera in swab samples, further suggesting that swabs capture a broader and more diverse microbial niche than tissue biopsies.

Recent studies have emphasized the role of non-H. pylori bacteria—particularly oral microbiota—in the pathogenesis of gastric cancer. In this context, the swab method may serve as a valuable tool for assessing microbial shifts during gastric cancer development, especially given the declining prevalence of H. pylori infection among younger populations in developed countries^1,8,9. Evidence in truly H. pylori–negative gastric cancer remains limited, as most existing studies have not rigorously excluded prior infection or corpus atrophy^28–30. Further investigation in strictly defined, genuinely H. pylori–negative populations will be necessary to clarify the causal role of non–H. pylori microbiota in gastric carcinogenesis.

Swab is also a more practical and simpler sampling method compared to gastric juice aspiration or surgical specimen collection. A similar issue may arise in colorectal microbiome studies, where sampling across different colonic regions typically involves multiple biopsies. In these situations, swab-based sampling may offer a lesser invasive option, though its applicability and reliability would require further investigation.

Another potential advantage of the swab method is its lower material cost and wider accessibility. Unlike the relatively expensive commercialized endoscopic luminal brush (costing approximately $4.90–88.60 per use)^31–35, the swab method requires only a commercially available sterile cotton swab, typically costing less than $1 per use^13,36,37. Although this study was not designed to evaluate cost-effectiveness, the lower per-use cost suggests that swab-based sampling may be more economically efficient for routine sampling. Future systematic studies should directly evaluate this aspect.

The swab method may also help address a commonly cited limitation in human microbiome research—high host DNA content. This issue is particularly prominent in tissue samples, which contain more human DNA than other sample types such as gastric fluid^38,39. In contrast, the swab method involves only gentle contact with the mucosal surface rather than obtaining human tissue, it is theoretically possible that swab samples contain lower levels of host DNA contamination. However, this study did not quantitatively measure host DNA content, and therefore this consideration remains speculative and requires further investigation.

This study has inherent limitations as a proof-of-concept investigation. Although participants were prospectively enrolled, the study cohort was not predefined for this research purpose, resulting in a relatively small sample size and limited generalizability. The study was not powered to perform subgroup analyses or multivariable adjustments, making it difficult to incorporate clinical covariates into statistical analyses such as PERMANOVA and ANCOM-BC2. Future studies with larger, well-defined patient populations will be necessary to validate these findings and allow for covariate-adjusted modeling.

In addition, although blank and channel control samples indicated generally low bacterial biomass and distinct taxonomic profiles compared with biological samples, low-level channel-related contamination cannot be completely excluded, particularly given that swabs may contact a greater surface area of the endoscope working channel than biopsy forceps. This concern is supported by the alpha diversity results, in which channel and tissue samples did not differ significantly, indicating that a cautious interpretation of the findings is warranted. Likewise, because both swab and biopsy sampling require passage of the endoscope through the oropharynx and esophagus, the possibility of minor oral carry-over contamination should also be considered, even though procedural steps (e.g., closed-system insertion and avoidance of suction) were taken to minimize this risk. The swab size also presents an inherent limitation. Because the swab must pass through the endoscope working channel, its dimensions are restricted, and the 5 × 5 × 5 mm swab used in this study represents the maximum size that can be inserted. Larger swabs could potentially sample a broader mucosal area and collect greater bacterial biomass but are not feasible with standard endoscopic equipment. Additionally, the swab method may underrepresent pathogens that are strongly adherent to the mucosal surface. In H. pylori–positive cases, tissue biopsies showed higher H. pylori abundance than swabs, indicating greater sensitivity for detecting mucosa-tightly attached bacteria. Nevertheless, H. pylori still remained a dominant taxon in swab samples (median relative abundance ~ 30%), suggesting that major mucosa-associated taxa were still sufficiently captured.

Not including a positive control in the study design is also a limitation. Although blank and channel controls helped assess potential contamination, the use of a defined mock microbial community would have strengthened the study result. Incorporating such positive controls should be considered in future studies to enhance methodological robustness.

Another limitation is that intragastric pH was not assessed. Gastric acidity influences both bacterial load and community structure, and hypoacidic conditions may favor expansion of oral-associated taxa. For example, autoimmune gastritis, which represents a markedly hypoacidic state, has been associated with increased microbial diversity and enrichment of Firmicutes, particularly Streptococcus⁴⁰, and similar trends have been observed with the progression of atrophy and metaplasia⁴¹. Future studies evaluating non–H. pylori gastric microbiota should therefore incorporate assessments of gastric acidity.

Finally, while our findings suggest that swab sampling may capture a broader representation of the gastric luminal microbiota—including microbial communities of potential relevance to gastric carcinogenesis in H. pylori-negative settings—this study does not establish clinical applicability. The relevance and utility of swab sampling should be examined in larger, systematic studies, ideally with longitudinal follow-up.

Conclusions

This pilot study shows that endoscopic mucosal swab sampling yields gastric microbiome profiles comparable to conventional tissue biopsy, while demonstrating higher alpha diversity and broader microbial niche representation. Although the possibility of low-level channel-related contamination cannot be completely excluded, the swab technique may serve as a practical approach for gastric microbiome assessment, particularly in settings where multiple biopsies are less feasible. Further studies with larger and more diverse cohorts are needed to validate these observations and clarify the clinical relevance of swab-based gastric microbiome profiling.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

G.P. drafted the initial manuscript, prepared visual material, and performed the main analysis. H.C. edited the manuscript, supervised the study, and provided overall guidance. All authors reviewed and approved the final manuscript.

Funding

This work received no external funding.

Data availability

The DNA sequence datasets generated and analyzed during the current study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1309291 (https://www.ncbi.nlm.nih.gov/sra/?term=PRJNA1309291). “Subj000” indicates the subject number; “BL,” “C,” “T,” and “S” represent specimen types (blank, channel, tissue, and swab, respectively); and “A” and “B” denote sampling locations (antrum and body). Each sample name follows the format SubjXXX_[Type]_[Location].

Declarations

Competing interests

The authors declare no competing interests.