Cervical microbiome is altered in cervical intraepithelial neoplasia after loop electrosurgical excision procedure in china

Although human papillomavirus (HPV) infection is a major cause leading to the development of cervical intraepithelial neoplasia (CIN), the relationship between genital microbiome and HPV persistence/clearance is not well established. Loop electrosurgical excision procedure (LEEP) is one of standard treatments of CIN 2/3 globally, yet little is known about how the LEEP influence genital microbiota. We conducted a prospective study of 26 patients with CIN2/3 who underwent analysis of cervical microbiome before and after 3 months of LEEP treatment. Cervical swabs were collected, and microbiomes were analyzed by 16S ribosomal RNA gene sequencing. A decrease of cervical microbial diversity was observed after 3 months of LEEP treatment. Notably, a significant shift from community type of a Prevotella-containing and lack of a consistent dominant species to lactobacillus iners dominated microbiome correlated with LEEP. Particularly, Leptotrichia and clostridium were further decreased after LEEP treatment (P = 0.049 and P = 0.002, respectively). Our results suggest that the cervical microbiome is altered after LEEP treatment in patients with CIN2/3. Further studies with larger sample sizes are needed to validate these findings.


Results
LEEP's impact on the cervical microbiome. In total 1,914,331 reads were obtained from 52 samples with an average number of reads per sample of 36,814 and the mean and medium read lengths of 445 and 453 bp respectively. Using bacterial genera sequence data, samples were classified according to their bacterial communities consistent with previously described cervical microbiome as cervicotypes (CTs) 12 ; CT1: primarily composed of non-iners Lactobacillus (i.e., higher percentage of sequencing reads from non-iners Lactobacillus than L. iners, Gardnerella, or Prevotella), CT2: Lactobacillus iners dominated, CT3: Gardnerella dominated, CT4: lack of a consistent dominant species but communities all included Prevotella (Fig. 1A). To further explore the relationship between 4 distinct CTs, a PCA analysis was performed using the species-level taxonomic profiles. As shown in Fig. 1B and Supplementary Fig. S1, the first two principal components, representing 71.08% of the variance, classified the 52 samples into three groups, though CT3 and CT4 are a continuum. The analysis of similarity (ANOSIM) test using Bray-Curtis dissimilarity showed that the observed cluster patterns were significant (R 2 = 0.5676, P = 0.001). Although the clusters were not clearly separated by LEEP surgery, most women with LEEP surgery carried L. iners as the dominant members.
The rates and frequency of the different CTs were compared with and without LEEP surgery (Table 1). We found that 53.9% patients without consistent predominant bacterial taxon, though each community was found to have prevotella (CT4) before LEEP. Reduced rate of CT4 from 53.9% to 26.9% were observed in patients with LEEP surgery (P = 0.048). Conversely, frequency of CT2 in patients with LEEP increased from 23.1 to 53.9% (P = 0.023). These analyses are suggestive of association between Prevotella and cervical disease 13 .
Species richness ( Fig. 2A) were higher in patients without LEEP compared to those with LEEP (P = 0.049). Microbiota diversity was also found to be higher in patients without LEEP in compared with those with LEEP but this was not statistically significant ( Fig. 2B,C). Consistent with increased rates of CT4 in patients without LEEP, species alpha-diversity ( Supplementary Fig. S2B,C) were higher in CT4, compared to other CTs particularly CT1 (P < 0.001) and CT3 (P = 0.035 and P = 0.005), while there was no different in species richness between CTs ( Supplementary Fig. S2A).
Furthermore, to investigate whether the pre-and post-menopausal status altered the microbial communities across the LEEP treatments, we performed beta diversity analysis by using Principal Coordinates Analysis (PCoA) based on Bray-Curtis dissimilarities. Using Bray-Curtis dissimilarity significant differences were observed between pre-and post-menopausal groups before LEEP treatment (ANOSIM, R 2 = 0.224, P = 0.023) (Fig. 3A). In pre-menopausal group before LEEP, four major taxa at genus level were observed, Lactobacillus (40.2%), Prevotella (17%), Gardnerella (14.6%), and Sneathia (8.7%) (Fig. 3B). In post-menopausal group before LEEP, a similar composition of major cervical bacteria was observed, but with altered proportions, involving the replacement of Lactobacillus (20.7%), Enterococcus (19.4%), Bacillus (11.7%), and Sneathia (6.7%). In contrast, microbial diversity was not observed between pre and postmenopausal groups after treated with LEEP ( Supplementary Fig. S3A). After receiving LEEP therapy, premenopausal women exhibited increase in Lactobacillus up to 66.8%. Similar to the effect of LEEP on postmenopausal women, increase of Lactobacillus up to 35% was observed ( Supplementary Fig. S3B).

Discussion
In the current study, we applied a 16S rRNA sequencing approach to a longitudinal group of patients analyzed before and after LEEP therapy. An alteration of cervical microbiome was observed in patients of CIN2/3 follow-up three months of LEEP therapy. The cervical microbiome after LEEP reflected reduced species richness as well as a significant shift from community type of a Prevotella-containing and lack of a consistent dominant species to L. iners-dominated microbiome.
It is well established that healthy vaginal microbial communities are dominated with low bacterial diversity, while vaginal dysbiosis is often characterized by high diversity bacterial populations with increased anaerobic bacteria 14 . The increased diversity of vaginal microbiota has been associated with bacterial vaginosis 15 , preterm birth 16 , acquisition of sexually transmitted infections 14 , and persistence and development of cervical cancer 6,17 . There are also several reports regarding the association between cervical microbiomes and genital diseases. A significant difference in microbial diversity was found between non-cervical lesions with HPV negative women and those with squamous intraepithelial lesions and cervical cancer in a recent study 7 . The cervical mucosal community dominated by L. iners and unclassified Lactobacillus spp. was associated with higher grade cervical intraepithelial neoplasia in women infected with HR HPV 4 . As LEEP is one of effective therapies to remove of cervical dysplasia, we observed that the cervical community type dominated by L. iners in patients with CIN2/3 after LEEP.
Ravel et al. 18 demonstrated that vaginal microbiota can be categorized into five community state types; four are dominated by different species of Lactobacillus species (L. iners, L. crispatus, L. gasseri, L. jensenii), but one is devoid of Lactobacillus spp. and instead enriched with strict anaerobic species including Gardnerella, Megasphera, Sneathia and Prevotella, which have subsequently been used by numerous other studies 6,19,20 . However, the cervical microbiota profiles have been categorized to four distinct cervicotypes 12 . Consistently, we identified CT4 (mixed microbial community containing Prevotella) prominent in patients with CIN2/3 before LEEP, whereas CT2 (dominated by Lactobacillus iners) was enriched in patients after LEEP.
It has been proved that the bacterial community within the female genital tract has a profound impact in human's health, since the cervical microbiome is involved in the modulation of the inflammatory immune  12,14 . The predominant cervical microbiome in women with normal cytology was Lactobacillus crispatus and Lactobacillus iners, whereas with squamous intraepithelial lesions was Sneathia spp 7 . Lactobacillus crispatus-rich cervical microbiota were found to manifest the lowest inflammation, while Prevotella-containing and to a lesser extent Gardnerella-dominant microbiota, correlated with multiple pro-inflammatory mucosal cytokines even in the absence of overt sexually transmitted diseases 12 . Prevotella is a key vaginal species has negative impacts on women's health 13 .
Lactobacillus crispatus is associated with a relatively stable state in the cervico-vaginal environment, whereas L. iners is associated with increased risk of transiting from normal to bacterial vaginosis (BV) 3,21,22 . On the other hand, L. iners has been reported to become a predominant part of the microbial community when the vaginal microbiota transitions between abnormal and normal states 23 . In our study, L. iners was the most abundant species in the cervix of women with LEEP. Thus, LEEP is not sufficient to restore a healthy cervical bacterial community. Recent advances in the knowledge of human microbiome have implications for the potential use of probiotics and prebiotics to restore human healthy microbial communities in promoting genital health in women 24,25 .    Sneathia spp. (L. amnionii) was found to be correlated with HPV infection in a Korean twin cohort 26 . Sneathia amnii has been identified as a possible microbiological marker associated cervical cancer in HPV-positive subjects 27 . In our study, we found that Sneathia amnii was enriched in patients with CIN2/3 before LEEP.
Although our results revealed the alterations of cervical microbiome after LEEP in patients with CIN2/3, there are several limitations needed to be addressed in the future studies. First, we were not able to observe microbiome variations among patients with different outcome after LEEP due to its small number of patients and relative short follow up period. Second, the impact of mucosal microbiota alteration on the inflammation cytokine profiles should be observed.

Materials and Methods
Patients. We prospectively recruited 50 patients who underwent LEEP for CIN2/3 at our department of Cervical Disease of Obstetrics and Gynecology hospital of Fudan University from June 2016 to Jan 2017. Inclusion criteria were included, not pregnant, no previous history of cervical or other lower genital cancer, no previous hysterectomy, or destructive therapy of the cervix, had no current use of hormonal or barrier contraceptive products, vaginal douching, tobacco or alcohol abuse, not HIV or hepatitis B/C positive or autoimmune disorders or antibiotics/probiotics (oral or topical) within six months prior to sample collection, and had no sexual activity in the previous days of the sampling. Cervical mucus samples were collected just before and after three months of LEEP. Of the 50 patients, 2 (4%) underwent a hysterectomy during the study period, primarily due to invasive cancer, and 22 (44%) patients had no follow-up visit after LEEP. Finally, 26 patients enrolled in this study. All of the subjects were Chinese ethnicity whose age ranged between 25-68 years (including six post-menopausal women), and detailed patient characteristics are listed in Table 2 and Supplementary Table S1. Epidemiological Microbial analysis. Sequencing reads were demultiplexed and filtered by Trimmomatic and merged by FLASH with the following criteria: (1) The reads were truncated at any site receiving an average quality score <20 over a 50 bp sliding window. (2) Primers were exactly matched allowing 2 nucleotides mismatching, and reads containing ambiguous bases were removed. (3) Sequences whose overlap longer than 10 bp were merged according to their overlap sequence. Paired-end reads were overlapped using PANDAseq with a required overlap length of >300 bp 28 . Reads less than 100 nucleotides or lacking a correct primer were removed 29 . The 16S rRNA sequencing data were processed using the Quantitative Insights Into Microbial Ecology platform (QIIME, V.1.9.1) 30 . Operational taxonomic units (OTUs) were clustered with 97% similarity cutoff using UPARSE (version 7.1 http://drive5.com/uparse/) and chimeric sequences were identified and removed using UCHIME 31 . The taxonomy of each 16S rRNA gene sequence was analyzed by RDP Classifier algorithm (http://rdp.cme.msu.edu/) against the Silva (SSU123) 16S rRNA database using confidence threshold of 70%. A total of 8,600,000 sequences clustered to 652 OTUs were obtained after quality filtering. Alpha diversity was analyzed with mother 32 . Richness of each sample was calculated with the Sobs index and diversity accounting for both relative abundance and evenness was evaluated with Invsimpson and Shannon index. The Principal Component Analysis (PCA) was performed by the R package ade4. Each coordinate on the score plot represents an individual sample. Principal Coordinates Analysis (PCoA) based on Bray-Curtis dissimilarities were performed. Permutational Multivariate Analysis of Variance Using Distance Matrices (PERMANOVA) and Analysis of Similarities (ANOSIM) were carried out using the 'adonis' and 'anosim' functions in the 'vegan' package, respectively, with Bray-Curtis dissimilarities and 999 permutations.
Statistical Analysis. Examination of statistical differences between cervical microbiota was performed at bacterial genera and species levels using the Statistical Analysis of Metagenomic Profiles software package. Ward's linkage hierarchical clustering analysis (HCA) of bacterial genera was performed using a clustering density threshold of 0.75. Bacterial species data were classified into CTs as described by Anahtar et al. 12 : CT I (non-iners lactobacillus; high percentage of Lactobacillus crispatus), CT II (L. iners), CT III (Gardnerella), and CT IV (mixed bacterial species containing Prevotella). The effects of LEEP on bacterial genera, number of species observed, and α diversity were assessed using one-way ANOVA, Kruskal-Wallis test, and Dunn's multiple comparison test, where appropriate. The LEfSe method 33 characterized differentially abundant taxonomic features before and 3 months after LEEP. An α value of 0.05 was used for factorial Kruskal-Wallis test between classes, and a threshold of 2.0 was used for logarithmic LDA score for discriminative features. Fisher's exact test was used to comparing categorical data among two or more groups. P values are two-sided. The analyses were performed with R packages (V.2.15.3) and Prism (GraphPad).