The Effect of Probiotics Supplementation on Gut Microbiota After Helicobacter pylori Eradication: A Multicenter Randomized Controlled Trial

This multicenter randomized clinical trial was performed at seven hospitals in China from March 2019 to November 2019. Inclusion criteria: patients aged between 18 and 65 years, at least two positive tests of rapid urease test and ¹³C-urea breath test (¹³C-UBT). Patients with any one of the following criteria were excluded from the study: history of gastrectomy, previous eradication therapy for H. pylori, peptic ulcer or other upper gastrointestinal lesions, gastrointestinal malignant tumor, contraindication or previous allergic reactions to the study drugs, severe concurrent diseases or malignancy, pregnant or lactating women, the use of antiacids or gastric mucosal protective drugs or antibiotics or probiotics in the past month, and patients who could not give informed consent. Written informed consent was obtained from all patients before enrollment, and this trial was approved by Ethics Committee of Xinqiao Hospital, Third Military Medical University and also approved by the institutional review board (IRB) of each participating hospital (please see supplementary material for list of IRB names). This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments. The trial was registered at Chinese Clinical Trial Registry (Chictr.org.cn, ChiCTR1900022116).

On the basis of the inclusion and exclusion criteria, a total of 162 patients receiving eradication therapy were randomly assigned to the probiotics or placebo group. An independent statistician who was not involved in the enrollment generated a random number. In total, 83 patients received the 14-day BQT (esomeprazole 20 mg, amoxicillin 1000 mg, furazolidone 100 mg, bismuth potassium citrate 220 mg, all given twice daily) supplemented with probiotics (Medilac-S; Enterococcus faecium 4.5 × 10⁸ and Bacillus subtilis 5.0 × 10⁷, Hanmi, Beijing, China) three times a day for 4 weeks. In total, 79 patients received 14-day BQT supplemented with placebo (maltodextrin) three times a day. Gastrointestinal symptoms were assessed at baseline for all patients and on weeks 2, 4, 6, and 8 after therapy. ¹³C-UBT was used to evaluate the H. pylori eradication effect at 6 weeks after completion of treatment.

Fresh stool samples were collected from all patients at baseline (before treatment) and at weeks 2, 4, 6, and 8 after treatment. Participants were asked to return the fecal specimen to the research assistant in the hospital on the day of sample collection. All stool samples were immediately frozen and stored at − 80 °C.

Total DNA from the fecal samples was isolated by using TIANamp Stool DNA Kit (TIANGEN Biotech Co. Ltd., Beijing, China) according to the manufacturer's instructions. DNA concentration was quantified using a Nanodrop (Thermo Scientific, Wilmington, USA), and its integrity was assessed by 1% agarose gel electrophoresis. DNA was stored at − 20 °C until use. The V3–V4 hypervariable regions of the 16S rRNA were amplified using 341F and 785R primers: forward primer 5′-CCTACGGGNGGCWGCAG-3′ and reverse primer 5′-GACTACHVGGGTATCTAATCC-3′. The amplifications were performed employing a step cycling protocol consisting of 95 °C for 30 min, 25 cycles of 98 °C for 15 s, 55 °C for 30 s, and 72 °C for 45 s, ending with the final elongation at 72 °C for 10 min. PCR amplicons were purified using Agencourt AMPure Beads (Beckman Coulter, Indianapolis, IN) and quantified using the PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). The purified amplicons were then sequenced on an Illumina Miseq platform (Illumina, San Diego, USA) by Longsee Biomedical Corporation (Guangzhou, China).

The Quantitative Insights into Microbial Ecology 2 (QIIME2, version 2019. 7) platform within a conda environment was used to process the sequencing data on our Linux server (i7-8700K, 64 Gb RAM). The bioinformatic analysis processes were performed according to the official “Moving Pictures” tutorial provided by the QIIME2 website (https://​docs.​qiime2.​org/​2019.​7/​tutorials). Firstly, the pair-end fastq format sequence files (spanning the entire 16S rRNA gene V3–V4 region) were imported into QIIME2 by using the “qiime tools import” command. Then, sequence quality control and feature table construction were performed by using the “qiime dada2 denoise-paired” command of QIIME2 with the following parameters: p-trim-left-f = 9, p-trim-left-r = 9, p-trunc-len-f = 250, p-trunc-len-r = 250. DADA2 is a pipeline for detecting and correcting (where possible) Illumina amplicon sequence data, which group unique sequences to construct amplicon sequence variant (the equivalent of 100% OTU in QIIME1 and also called “feature” in QIIME2). As implemented in the q2-dada2 plugin, this quality control process will additionally filter any phiX reads (commonly present in marker gene Illumina sequence data) that are identified in the sequencing data, and will filter chimeric sequences. After the quality filtering step had completed, the “qiime feature-table summarize” command was used to generate a feature-table listing how many sequences are associated with each sample and with each feature. A sampling-depth of 12,200 was chosen on the basis of the sample with the lowest number of sequences, which is a decent amount of sequences and also allowed us to retain all of our samples. Then the “qiime feature-table rarefy” command was used to subsample frequencies from all samples so that the sum of frequencies in each sample is equal to sampling-depth and generated a rarefied feature table. After rarefaction, a rooted tree and an unrooted tree were constructed for further phylogenetic diversity analyses by using the “qiime phylogeny align-to-tree-mafft-fasttree” command. Then, all the alpha diversity and beta diversity were calculated by “qiime diversity alpha” and “qiime diversity beta” commands from the rarefied feature-table. After diversity metrics were computed, the PCoA results were calculated by the “qiime diversity pcoa” command and visualized by the “qiime emperor plot” command. In the context of the sample metadata, a pre-trained naive Bayes classifier (silva-132-99-nb-classifier.qza) and the “qiime feature-classifier classify-sklearn” command were used to explore the taxonomic composition of the samples. This classifier was trained from the Silva database release 132 with 99% similarity, where the sequences have been trimmed to only include the V3–V4 region of the 16S that was sequenced in this analysis. Furthermore, linear discriminant analysis effect size (LEfSe) was used to select significant candidates at the genus level. We compare relative abundance of taxa between the two groups and different time periods within each group using a nonparametric Mann–Whitney U test, followed by a linear discriminant analysis (LDA) to estimate the effect size of each microbial feature with differential abundance. Taxa with an LDA score greater than 2.0 and p < 0.05 were considered significantly enriched.

The 16S rRNA functional prediction by used amplicon sequence variants (ASVs) is performed by PICRUSt2 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States, version 2). Then ASVs were categorized into Clusters of Orthologous Groups (COG) and into Kyoto Encyclopedia of Genes and Genome (KEGG) orthology (KO). According to the COG database, the descriptive information of each COG and its functional information were parsed from the eggNOG database to obtain the functional abundance spectrum. KO, Pathway, and Enzyme (EC) information were obtained according to the KEGG database while the abundance of each functional category was calculated according to OTU abundance.

Data are presented as mean ± standard deviation (SD). Statistical analyses were performed using SPSS (IBM, Armonk, NY) and R. The χ² test or Fisher’s exact test were used for analysis of categorical data and Student’s t test or the analysis of variance (ANOVA) test for analysis of continuous data. Eradication efficacy was performed on an intention-to-treat (ITT) population where patients who dropped out were considered as treatment failures. Secondary per-protocol (PP) analyses were performed which excluded patients lost to follow-up or prematurely withdrew before completion of the study. The Wilcoxon signed-rank test was used to evaluate the ecological similarity between and within groups. Analysis of similarities (ANOSIM), a nonparametric statistical test widely used in the field of ecology, was used to test whether the beta-diversity between groups (two or more) was significantly greater than the differences within groups. Best subsets regression and backward stepwise regression were used to pick variables capable of predicting the result of H. pylori eradication. The leaps performs an exhaustive search for the best subsets of the variables for predicting the result of H. pylori eradication by using an efficient branch-and-bound algorithm. The picked variables were then loaded into the stepAIC function of the MASS package (version 7.3-51.4, http://​www.​stats.​ox.​ac.​uk/​pub/​MASS4/​) in R for backward stepwise regression. Backward stepwise regression starts with the model including all predictors and deletes one variable at a time until the model quality is reduced, which was used to validate the robustness of the final variables. All statistical tests were two-tailed. P values less than 0.05 were considered significant.