This was a clinical trial conducted between January 2020 to March 2020 at the Department of Obstetrics and Gynecology of Kunming Tongren Hospital. Patients with BV were selected for a one-week intervention of a postbiotic gel to measure changes in the properties of vaginal discharge and vaginal microbiota before (day 0) and after (day 7) intervention. All (n = 50) patients diagnosed with BV at this hospital were considered for participation in the study and were screened by hospital professionals for eligibility. The inclusion criteria were: (1) women clinically diagnosed with diagnostic criteria for BV, aged 18-55 years; (2) body mass index between 19-24; (3) no history of heart, liver, lung, kidney, digestive tract, nervous system, and metabolic abnormalities; (4) not taken other vaginal preparation drugs 30 days before this trial; (5) voluntarily signed the informed consent. Subjects that were allergic to the postbiotics gel were excluded from this study ( Figure 1 ).

The postbiotic gel used in this work was produced by Qingyitang Industrial Co., Ltd. (Yunnan, China). Probiotic strains were provided by the Lactic Acid Bacteria Collection Center of Inner Mongolia Agricultural University (Hohhot, China). The gel was prepared by mixing the raw materials ( Table 1 ) with distilled water at 60°C, homogenize, sterilize, cool and ferment to pH 4.60 at 37°C. Citric acid was adjusted to pH 4.50 with deionized water. Kabo 940 was added and stirred until completely dissolved, followed by adding and mixing with triethanolamine, propylene glycol, PEG-90M, moisturizing gel, phenoxyethanol, and the fermentation solution in turn. The well-mixed gel was packaged as 3-gram tubes.

Participants was registered to participate in this trial by providing basic demographic information (age: 35.8 ± 8.96) before the start of the trial. The gel was applied every night and avoided the menstrual period. To apply the gel, subjects first cleaned their hands and vulva with warm water, took a product, slightly lifted the hips in a supine position, slowly inserted the gel catheter containing the gel into the deep part of the vagina, pushed it into the pubic area with a pusher, and maintained the posture for about 15 minutes. The product was applied every night for seven days.

Sample collection is carried out by professional doctors. Fresh vaginal secretions were collected with sterile cottons before the start of medication and the next day of the last gel application. In order to obtain fresh samples, we collect the secretions of volunteers according to strict methods and submit them for examination as soon as possible. First of all, sterile cotton swabs were gently rotated in the subjects’ posterior vaginal fornix to collect fresh vaginal secretions. After sampling, cottons with the samples were placed in test tubes containing physiological saline and stored at -80°C until further processing. At the same time, symptoms of subjects, including severity of vulvar itching and leucorrhea, color and odor of vaginal secretions were recorded. The biochemical and microbiological properties of vaginal secretion were determined in the laboratory within 30 minutes of sample collection by a medical professional. Hydrogen peroxide test was performed using a combined aerobic vaginitis/bacterial vaginosis five-item qualitative test kit (based on an enzyme-chemical reaction method; CSC Goldfield Diagnostics, Beijing, China). The presence of lactobacilli, fungi, and the cleanliness of collected samples were observed using a microscope (Olympus Corporation of Japan). The pH of the vaginal secretion was determined using pH test strips. All assessments were operated by professional doctors or personnel, and the data were entered and confirmed by more than two medical staff to ensure data consistency and accuracy. A total of 50 subjects completed the clinical assessment in the trial, and 47 subjects provided a complete sample (two vaginal secretions).

The metagenomic DNA in the collected vaginal secretions was extracted by QIAamp kit (Qiagen, Hilden, Germany), and the purity and concentration of the extracted DNA were detected by Nanodrop (Thermo Fisher Scientific, USA) and agarose gel electrophoresis. Qualified DNA samples were amplified, targeting to the 16S rRNA V4 region using barcoded region-specific primers, 515F (GTGYCAGCMGCCGCGGTA) and 806R (GGACTACHVGGGTWTCTAAT). The amplification conditions were pre-denaturation at 95°C for 1 min; denaturation at 95°C for 30 s; annealing at 60°C for 40 s; extension at 72°C for 1 min, for a total of 30 cycles; terminal extension at 72°C for 7 min and termination at 4°C. Amplified samples were checked by agarose gel electrophoresis for the amplicon product size and purity. Follow-up analysis was only performed on amplicon products appearing as a single and bright band.

The samples that met the quality requirement were used for DNA library construction and sequenced using the Illumina novaseq PE250 platform. The original sequences were quality controlled and grouped by the sample nucleotide barcode. Microbial diversity analysis was performed using the QIIME platform (Caporaso et al., 2010) as follows: sequences aligned using PyNAST (Y. Zhang & Alekseyenko, 2017) to establish operational taxonomic units (OTUs) (Vu et al., 2018) according to the UCLUST two-step method. Selected representative sequences were taxonomically assigned by comparing against the SILVA database (Majaneva et al., 2015), the Greengene database (DeSantis et al., 2006), and the Ribosomal Database Project (Release 11.5) database (Maidak et al., 1996).

For α diversity analysis, QIIME (version 1.9.1) was used to calculate the Shannon index, chao1 index, and Simpson index. The Shannon curve and the observed species number curve were plotted using R (version 4.2.1), which were used to assess the bacterial diversity of each sequenced sample and the sequencing depth. Wilcoxon tests were used to evaluate the α diversity between samples collected at the two time points. β diversity (weighted and unweighted Unifrac) was calculated by R packages (vegan), and principal coordinate analysis (PCoA) was used for descending presentation. Association between β diversity and study groups was assessed using a non-parametric analysis of similarities (Adonis, vegan R package) with 999 permutations. Followed by comparative analysis to identify significant differential marker bacteria of treatment (cut-off: P < 0.05; Wilcoxon test). PICRUST2 (Douglas et al., 2020) was used for functional annotation and STAMP (Parks et al., 2014) was used for data visualization.

In our study, 50 people completed clinical indicators, including 45 people who completed vaginal discharge collection before the intervention and 42 people after the intervention. After matching, the per-protocol population was 42, who completed the process of clinical information collection and provision of vaginal secretion for microbiota sequencing analysis before and after using the postbiotic gel. The vaginal sample of a healthy woman is transparent or milky white, with no odor and in a small amount. A large amount of vaginal secretion that is thin, homogeneous, and of fishy odor is indicative of vaginal inflammation (Amsel et al., 1983). This study compared these properties of subjects’ vaginal secretions before and after use of the postbiotic gel, and found that properties of subjects’ vaginal secretions (e.g., color, clarity, odor) improved significantly after using the postbiotic gel ( Table 2 ).

Vulvar pruritus is one of the reference indicators for evaluating vaginitis. Three among the 50 patients in this study complaint about vaginal itching at the start of the trial, and the symptom remained only in one subject after using the postbiotic gel. A high level of vaginal secretions cleanliness (the cleanliness rating is closer to IV) and positive reaction in hydrogen peroxide test indicate poor BV or at least poor vaginal health. Initially, 47 patients showed a positive result for hydrogen peroxide, and 18 (remission rate = 38.3%) of them showed a negative hydrogen peroxide result after using the gel. Moreover, 29 people had a cleanliness grade of III or IV, and most (18) of them showed significant improvement after using the postbiotic gel. These data suggested that applying the postbiotic gel for a week could improve BV and vaginal health.

A total of 5,479,608 high-quality sequences were obtained from 87 (Before: n = 45; After: n = 42) sequencing samples. The rarefaction curves of the Shannon diversity ( Figure 2A ), but not the number of observed species ( Figure 2B ) leveled off, suggesting that most of the bacterial diversity was already captured although new species could still be found. Therefore, this sequencing depth was adequate to reflect a representative diversity of the vaginal microbiota.

In order to study the difference between the two groups of samples, the samples were screened, and the samples that could not be paired were excluded, and a total of 42 pairs of samples were obtained. The chao1, number of observed species, Shannon, and Simpson indexes of the vaginal microbiota of subjects were analyzed, and decreasing trends were observed in all four measured diversity indexes after using the postbiotic gel, though the differences were not statistically significant ( Figure 2C ). The decrease in species diversity and abundance could be attributed to the effect of application of the postbiotic gel in shaping the vaginal environment for bacterial growth.

Changes in the β-diversity of the vaginal microbiota was analyzed by PCoA (weighted and un-weighted Unifrac), which revealed no significant differences before/after the gel intervention (P > 0.05; Figures 3A, B ). The taxonomic analysis of the vaginal microbiota composition revealed five dominant phyla. After applying the gel, the vaginal microbiota showed increased levels in Actinobacteria and Fusobacteria, and decreased levels in Proteobacteria and Bacteroidetes, though the differences were not statistically significant (P > 0.05; Figure 3C ). At the genus level ( Figure 3D ), the major genera (those > 1% average relative abundance before and after using the postbiotic gel) included Lactobacillus (71.56%), Gardnerella (6.02%), Prevotella (3.30%), Streptococcus (1.62%), and Atopobium (2.89%). After using the postbiotic gels, the relative abundances of some pathogenic bacteria, including Gardnerella ( Figure 3E ), Streptococcus, and Prevotella, decreased, while the relative abundance of Atopobium increased. The relative abundance of Streptococcus decreased significantly after the gel use (P = 0.0001) ( Figure 3F ), while no significant changes were observed in the genus Atopobium (P = 0.0596; Table S1 ).

We used microscopy to analyze Lactobacillus in the secretions and found that Lactobacillus was not observed before the gel intervention and 10 samples in which Lactobacillus was found after the intervention, but only 5 of these 10 cases were sequenced by Illumina both before and after the use of the gel. Through the analysis of the relative content of Lactobacillus in 5 cases, it was found that the relative content of Lactobacillus in 4 samples increased in varying degrees. It is worth mentioning that among all sequenced samples (Before: n = 42; After: n = 42), the content of Lactobacillus in vaginal secretions of 24 patients increased ( Figure 3G ), accounting for 57% of the total number of patients, it can be seen that exogenous gel can increase the number of lactobacilli vaginal.

We then used PICRUSt2 to predict functional changes in the vaginal microbiota after using the postbiotic gel ( Figure 4 ), and significant changes were found in the encoded function of the vaginal microbiota, including pathways of metabolism, genetic information processing, and organic system ( Table S2 ). On the secondary level of KEGG pathway, the vaginal microbiota was significantly enriched in the pathways of biosynthesis of other secondary metabolites, metabolism of terpenoids and polyketides, and endocrine system and digestive system, which were significantly decreased after using the gel (P < 0.05; Table S2 ). On the tertiary level of KEGG pathway, eight pathways exhibited significant decreases after using the postbiotic gel (P < 0.05; Figure 4 ; Table S2 ). These data supported that the application of the postbiotic gel could modulate the predicted functional pathways of the vaginal microbiota, which might be associated with the symptom improvement.

According to our clinical observation, we selected the common clinical symptoms and signs, and referred to the symptom scoring standard in the Guidelines for Clinical Research on New Chinese Medicines (for Trial Implementation). For evaluating the amount of secretion: a small amount was scored as 2, a medium amount was scored as 1, and a large amount was scored as 0. For evaluating the nature of secretion: a thin amount was scored as 1, and a thick amount was scored as 0. For evaluating the odor of secretion: an odorless amount was scored as 1, and an odorous amount was scored as 0. For evaluating the color of secretion: a clear color was scored as 3, a light yellow color was scored as 2, a yellow color was scored as 1, and a green color was scored as 0. Then we used Spearman’s correlation analysis to analyze the association between the dominant bacterial genera and clinical features of the collected vaginal samples ( Figure 5 ; Table S3 ). Interestingly, we found significant negative correlations between Lactobacillus and a number of (potential) pathogens, i.e., Gardnerella, Streptococcus, Prevotella, Atopobium (P < 0.001, r = -0.4937; P < 0.001, r = -0.4165; P < 0.001, r = -0.6267; P = 0.0043, r = -0.3082, respectively). Streptococcus showed a weak positive correlation with Prevotella (P = 0.045, r = 0.22), but negative correlations with the amount and characteristics of vaginal secretion (P = 0.0061, r = -0.2968; P = 0.0178, r = -0.2579, respectively; Figure 5 ).