Study design and population

Individuals with positive H. pylori test results (urea breath test, fecal H. pylori antigen, blood/urine H. pylori IgG antibody, and rapid urease tests) evaluated at Kobe University Hospital between February 2018 and December 2021 were recruited. A total of 76 participants were enrolled. In this randomized controlled trial, eligible participants were randomly assigned to receive either E. faecium 129 BIO 3B-R or placebo in a 1:1 ratio, based on a computer-generated randomization list. The randomization was conducted by an allocation specialist independent of the main research team. Participants were equally divided between the two groups, with each participant receiving medication packs numbered in advance by the allocator. E. faecium 129 BIO 3B-R was acquired over the counter as the commercially available Biofermin-R. Both E. faecium 129 BIO 3B-R and placebo were in powder form, and each dose was carefully prepared and packaged by the hospital’s Pharmacy Department. The study employed a double-blind method, where both participants and researchers were blinded to the treatment allocation, ensuring transparency and fairness in the intervention. Data on primary and secondary endpoints were collected throughout the study’s follow-up period.

The study conformed with the principles set forth in the Declaration of Helsinki and was approved by the Clinical Research Ethical Committee of Kobe University Hospital (approval number C180029). The study was registered with the Japan Registry of Clinical Trials (jRCT) and the University Hospital Medical Information Network Clinical Trials Registry (UMIN). The initial trial registration date for jRCT is 25/03/2019 (jRCTs051180155; https://jrct.niph.go.jp/latest-detail/jRCTs051180155/), and that for UMIN is 18/12/2017 (UMIN000030015; https://center6.umin.ac.jp/cgi-open-bin/ctr/ctr_view.cgi?recptno=R000034265).

Eligibility criteria

The inclusion criteria were as follows: (1) Diagnosis of H. pylori infection based on both upper-gastrointestinal endoscopic findings suggestive of infection and at least one additional positive H. pylori test (e.g., urea breath test; serum or urine antibody test; stool antigen test; culture; or rapid urease test), in accordance with Japanese Helicobacter Study Group guidelines; (2) patients for whom H. pylori eradication therapy is clinically beneficial (3) provided informed consent; and (4) absence of allergies to disinfectants.

The exclusion criteria were as follows: (1) history of H. pylori eradication; (2) use of antibacterial drugs or proton pump inhibitors within 4 weeks before the start of eradication therapy; (3) age < 20 years; (4) history of surgery for gastric cancer (excluding endoscopic treatment); (5) severe renal or liver disorder, or heart disease; (6) allergy to the drug used; and (7) inappropriateness for treatment as determined by the investigator.

Treatment

In accordance with Japanese clinical guidelines, all participants received a standard 7-day first-line H. pylori eradication therapy with oral vonoprazan (20 mg twice daily), amoxicillin (750 mg twice daily), and clarithromycin (200 mg twice daily). Participants in the E. faecium 129 BIO 3B-R group (3B-R group) received probiotic strain E. faecium 129 BIO 3B-R (Biofermin-R) at 3 g/day and those in the placebo group received placebo at 3 g/day as oral therapy for 14 days, beginning on day 1 of the 7-day H. pylori eradication regimen and continuing for an additional 7 days after completion of antibiotic therapy. E. faecium 129 BIO 3B-R is a multi-antibiotic-resistant lactic acid bacterium; therefore, it was administered concurrently with the antibiotic treatment without adjusting the timing of intake. Because it is not acid-resistant, participants were instructed to take it after each meal.

The ingredients used to compound the placebo and E. faecium 129 BIO 3B-R tablets were purchased commercially and prepared at the Pharmacy Department of Kobe University Hospital. The placebo was formulated using components found in Biofermin-R, specifically, barley starch, glucose, lactose, and precipitated calcium carbonate.

Endpoints

The primary endpoint was the cumulative incidence of diarrhea (Bristol stool form scale score²⁰ of 6 or 7 points) in the E. faecium 129 BIO 3B-R and placebo groups during 7 days of antibiotic treatment (day 0–7).

Secondary endpoints included the following: (1) changes in gut flora diversity at one day, one week, and at the eradication judgment (day 64 (± 7) after the end of antibiotic treatment); (2) occurrence of adverse events other than diarrhea; (3) frequency scale for the symptoms of gastroesophageal reflux disease scores (FSSG score)²¹ after treatment; (4) diamine oxidase activity (DAO) in the serum at one day after the end of antibiotic treatment; (5) cumulative incidence of diarrhea (defined by the Bristol stool form scale score of 6 or 7 points) during 7 days after antibiotic treatment completion (day 8–14); (6) frequency of bowel movements during and during 7 days after antibiotic treatment; and (7) proportion of stool properties according to the Bristol stool form scale during and during 7 days after antibiotic treatment. The evaluation of the intervention’s safety was based on the monitoring of adverse events other than diarrhea and the exacerbation of gastroesophageal reflux symptoms. As a post-hoc analysis, subgroup analyses based on age and sex were conducted on the primary and secondary endpoints. Secondary outcomes (6) and (7) were added during the study before the key opening and data analysis. These changes were made to provide a more comprehensive assessment of the intervention’s impact on gastrointestinal health and safety.

Sample collection

Fecal samples were collected at the following four time points: before antibiotic treatment, on day 7(± 2), day 14(± 2) of eradication therapy, and day 64 (± 7) (at the time of eradication judgment). Fecal samples were collected using a stool collection kit (TechnoSuruga Laboratory Co., Ltd, Japan, FS-0002®) and were stored at − 80 °C until use.

DNA extraction and 16S rDNA sequencing

The QIAamp DNA Stool Mini Kit (Qiagen, Manchester, UK) was used in the DNA extraction. Bacterial DNA was extracted from 200 mg of stool sample according to the manufacturer’s protocol and stored at − 80 °C.

Two-step thermal asymmetric interlaced polymerase chain reaction (PCR) method was used to amplify the V3-V4 region of the gene encoding 16S rDNA, and 16S rDNA sequencing was performed using the MiSeq™ system (Illumina, San Diego, CA, USA). The protocol was modified slightly from that of the manufacturer. In the first PCR step, DNA extracted from stool samples was amplified under PCR conditions (95 °C for 3 min, followed by 25 cycles of denaturation at 95 °C for 30 s, primer annealing at 55 °C for 30 s, extension at 72 °C for 30 s, and final extension at 72 °C for 5 min) using 2 × KAPA HiFi HotStart ReadyMix (KAPA Biosystems, KK2602, Woburn, MA, USA) and Amplicon PCR primers. Purification was performed using AMPure XP Beads (Beckman Coulter, A63881, San Diego, CA, USA) in 80% ethanol. The forward and reverse primer sequences with overhang adapters were 5-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG-3 and 5-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG-3, respectively.

In a second PCR step, purified amplicons were subjected to PCR conditions (95 °C for 3 min, followed by 8 cycles of denaturation at 95 °C for 30 s, primer annealing at 55 °C for 30 s, extension at 72 °C for 30 s, and final extension at 72 °C for 5 min) using Nextera XT Index Primer (Illumina, FC-131-2001, San Diego, CA, USA), 2 × KAPA HiFi HotStart ReadyMix, and sterile, purified water. DNA was purified using AMPure XP Beads and 80% ethanol. The DNA concentration was measured using a Qubit® 4 Fluorometer (Invitrogen, Q33238) and TapeStation (Agilent Technologies, 5067-5584/5585), and approximately equal amounts of DNA from each sample were pooled. Pooled samples were sequenced using a MiSeq Reagent Kit v2 (500 cycles) (Illumina, San Diego, CA, USA) on a MiSeq System.

Microbiome analysis

QIIME 1.8.0 was used to perform sequence read processing, quality trimming, operational taxonomic unit definition, and taxonomic assignment. A total of 5000 paired-end reads randomly selected per sample were used for the analysis. Paired-end reads were merged using fastq-join. Chimeric sequences were excluded using USEARCH 6.1.544, and operational taxonomic clustering was performed using UCLUST with a threshold justification of 97%²². Based on 97% sequence similarity, operational taxonomic units were classified into bacterial groups at the genus level using the Greengenes Reference Database (version 13.8)²³. α-diversity analysis (observed species, PD whole tree, Shannon, and Chao1), weighted UniFrac principal coordinate analysis, and permutational multivariate analysis of variance were performed using QIIME 1.8.0. Galaxy version 1.0 (The Huttenhower Lab, Harvard T.H. Chan School of Public Health) was used for linear discriminant effect size analysis to estimate the effect size for each taxon with significantly different abundances. Functional profiling of the intestinal microbiota communities was predicted by phylogenetic investigation of communities using reconstruction of unobserved states (PICRUSt, version 1.0.0) based on the Greengenes database (version 13.5). PICRUSt can predict metagenomic functions using marker genes based on 16S rDNA sequencing and reference genome databases²⁴. The obtained sequence data were registered in the DNA Data Bank of Japan database under the accession number DRA016376.

DAO assay

DAO is used as an indicator of small intestinal mucosal maturity and integrity which inversely correlates with the degree of intestinal damage^25,26. We measured serum DAO activity as a noninvasive, indirect marker in order to detect any intestinal mucosal injury associated with post-eradication diarrhea. Previous human studies have shown that circulating DAO activity indirectly reflects intestinal mucosal DAO activity and that more severe tissue injury leads to proportionally lower serum DAO levels^27,28. The assays were conducted externally by Macrophi Inc. (Takamatsu city, Kagawa, Japan) by the Takagi method as previously described²⁹. Briefly, 1.5 ml of cadaverine solution (30 mmol/l in PIPES buffer) was incubated at 37 °C for 5 min. Then, 0.1 ml of serum sample was added, mixed, and incubated at 37 °C for 30 min. Following this, 1.5 ml of the color solution (DA-67, POD, ASOD in MES buffer) was added, mixed, and incubated for 60 min. The reaction was stopped by adding 0.05 ml of sodium diethyldithiocarbamate solution (30 mmol/l in water). The absorbance was measured at 668 nm using a spectrophotometer.

Sample size calculation

Assuming that past comparative trials with Lactobacillus and placebo groups during H. pylori eradication therapy are equivalent to this study, we calculated the required sample size using Pearson’s chi-squared test with a significance level of 5% and a power of 80%². The calculated required sample size per group was 36 participants. Taking dropouts into account, we set the sample size at 38 participants per group, for a total of 76 participants.

Statistical analysis

Comparison between the groups for the incidence of diarrhea was performed using the chi-squared test. 95% confidence intervals for proportions were calculated using the Wilson score method. Let z be the 97.5th percentile of the standard normal distribution (z = 1.96), and n the sample size. We first compute the adjusted proportion \(p=\frac{X+\frac{{z}^{2}}{2}}{n+{z}^{2}}\), and the adjusted standard error \({SE}_{wilson}=\sqrt{\frac{p(1-p)}{n+{z}^{2}}}\). The 95% confidence limits are then given by \({CI}_{95\%} :p\pm {z\times SE}_{wilson}\).

Likewise, 95% CIs for continuous data presented as mean ± SEM were calculated as \(95\%CI=mean\pm {t}_{0.975, n-1}\times SEM\). α-diversity was analyzed by Tukey’s multiple comparison test using GraphPad Prism software (version 9.1.0). β-diversity was estimated using the UniFrac distance metric and analyzed by permutational analysis of variance using QIIME 1.8.0. Statistical significance was set at p < 0.05.