Evaluation of the Associations Between Cervical Microbiota and HPV Infection, Clearance, and Persistence in Cytologically Normal

Published OnlineFirst November 21, 2018; DOI: 10.1158/1940-6207.CAPR-18-0233

Research Article Cancer Prevention Research Evaluation of the Associations Between Cervical Microbiota and HPV Infection, Clearance, and Persistence in Cytologically Normal Women Wu Ritu1, Wu Enqi1, Siriguleng Zheng2, Jiandong Wang3, Yaqin Ling1, and Yan Wang3

Abstract

The aim of this study was to investigate the associa- Pseudomonas aeruginosa were associated with baseline tions between cervical microbiota and different human HPV negativity. For HPV-positive women, the base- papillomavirus (HPV) infection statuses in cytological- line abundance of Actinomyces was negatively associ- ly normal women. The cervical microbiota of HPV- ated with new HPV infection, Alloprevotella tannerae, positive or -negative women with a normal cytologic Prevotella nigrescens,andPrevotella oulorum;and diagnosis was characterized and compared using 16S Dialister invisus were positively associated with new rDNA-based high-throughput sequencing, and the HPV-type infection within the year of follow-up. differences in cervical microbiota associated with new Lactobacillus delbrueckii was found to be negatively acquisition, persistence, and clearances of HPV geno- associated with persistent HPV infection and 9 taxa types were analyzed via one-year follow-up. The results belonging to Prevotella, Dialister,andLachnospiraceae showed that the cervical microbial richness of HPV- were found to be positively associated with persis- positive women was lower than for HPV-negative tence, and/or negatively associated with clearance of women, and the difference was more signiﬁcant in the HPV types. We also observed 10 novel taxa associated postmenopausal group relative to the premenopausal with the clearance/persistence of HPV that had not group. Ureaplasma parvum and related taxa were been reported elsewhere. Those taxa associated with associated with baseline HPV positivity, while different infection statuses of HPV could be used as a Brochothrix, Diplorickettsia, Ezakiella, Faecalibacterium, biomarker to help predict the risk of developing and Fusobacterium genera and their related taxa and persistent HPV infection.

Introduction malignancies (1). It is known that HPV infection is very common in sexually active women, and the infecting HPVs Cervical cancer is one of the most common cancers can usually be spontaneously eliminated from individuals among women worldwide, and persistent infection within 6 to 18 months. Only a small proportion of infected with high-risk human papillomavirus (HPV) types has women retain the virus, and this could lead to the devel- been confirmed to be the necessary factor for developing opment of cervical intraepithelial neoplasia (CIN) and cervical carcinoma (2). The mechanism by which some 1Key Laboratory of Ethnomedicine (Minzu University of China), Ministry of individuals develop a persistent HPV infection that goes on Education, Beijing, China. 2Department of Information Technology, Polytechnic to develop into clinically significant disease, however, College, Beijing, China. 3Department of Gynecological Oncology, Beijing Obstet- remains largely unclear. rics and Gynecology Hospital of Capital Medical University, Beijing, China. Emerging evidence shows that the human microbiome Note: Supplementary data for this article are available at Cancer Prevention can mirror the host's physiology and that it plays an Research Online (http://cancerprevres.aacrjournals.org/). important role in human health (3). Several cross-sectional W. Ritu and W. Enqi contributed equally to this article. studies addressing the association between the microbiota Corresponding Authors: Yan Wang, Beijing Obstetrics and Gynecology Hospital in the female reproductive tract and HPV infection and of Capital Medical University, Beijing 100006, China. Phone: 186-1065-3059; Fax: 010-6512-1602; E-mail: [email protected]; and Yaqin Ling, Key Laboratory related diseases have been undertaken, and they clearly of Ethnomedicine (Minzu University of China), Ministry of Education, Beijing show that there are significant differences between HPV- 100081, P. R. China. Phone: 188-1022-2674; E-mail: [email protected] negative and -positive women and between healthy wom- doi: 10.1158/1940-6207.CAPR-18-0233 en and women with HPV-related diseases with respect 2018 American Association for Cancer Research. to microbial structure, diversity, and species composition

www.aacrjournals.org OF1

Downloaded from cancerpreventionresearch.aacrjournals.org on October 1, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst November 21, 2018; DOI: 10.1158/1940-6207.CAPR-18-0233

Ritu et al.

(4–6). Regarding the comparison between HPV-positive in the Beijing Obstetrics and Gynecology Hospital (Beijing, and -negative women, the studies in a community of China), from August to September 2016. Written and Korean and Chinese women both found HPV-positive verbal consent were obtained from each participant. The women had significantly higher microbial diversity than exclusive criteria included current pregnancy, sexual inter- HPV-negative women (4, 5). Lee and colleagues also course or washout within 3 days, oral or vaginal usage of observed the HPV-positive women have significantly less antibiotics within 2 weeks, previous history of cervical or Lactobacillus spp. than their counterparts, and identified other lower genital cancer, previous hysterectomy, or Sneathia spp. to be a microbiological marker of HPV destructive therapy of the cervix. To eliminate the impact infection (4). In the comparison between healthy women of the potential change in microbiota that is driven by and women with HPV-related diseases, the studies of Oh cervical lesions, only women with normal cytologic results and colleagues (6) and Audirac-Chalifour and collea- were included in this study. All of the participating women gues (7) conclude that women with CIN had a higher were asked to attend the 1-year follow-up visit for repeated vaginal diversity than healthy controls. The presence of HPV testing. Anaerococcus vaginae, Garderella vaginalis, and L. iners in the Risk factor information of lifestyle profile (such as age, absence of L. crispatus were identified to be the most high- cigarette smoking, parity, menstrual status, and use of risk combination for development of CIN. In addition, hormonal/oral contraceptives) was collected with a ques- Brotman and coworkers performed a longitudinal tionnaire while waiting for the physical examination. study in a North American cohort of 32 sexually active, Three cervical swab specimens were collected from premenopausal women over the course of 16 weeks using each participant by cyto-brush (Qiagen) during routine a twice-weekly interval of evaluation to describe the pelvic examination. The first specimens were used for temporal relationship between vaginal microbiota and cytologic diagnosis (SurePath Liquid System, BD Diag- HPV detection. Using hierarchical taxonomic clustering, nostics TriPath); the second specimens were stored in the vaginal microbial profile of each woman was classified standard transport medium (STM) for HPV testing; and into a total of five community state types (CST). The the third specimens were stored in 4 mol/L guanidine authors found that the CST was associated with changes thiocyanate solution for further microbiota analysis. The in HPV status and a low lactobacillus community with high specimens for microbiota analysis were frozen at 20C proportions of the genera Atopobium was found to have the immediately, and then transferred to the laboratory slowest rate of HPV clearance (8). within 24 hours and stored at 80CbeforeDNA The aforementioned evidence indicates that the cervi- extraction. covaginal microbiota plays a substantial role in the infec- DNA was extracted from the residual STM-stored cervical tion and clearance of HPV virus in the reproductive tract specimens using the QiAmp Mini DNA Kit (Qiagen) in and constitutes a new biomarker reservoir to predict the accordance with the manufacturer's recommendations. persistence or regression of HPV. So far, however, only a Total bacterial DNA was extracted from samples using few longitudinal studies concerning associations between the Power Soil DNA Isolation Kit (MO BIO Laboratories) cervicovaginal microbiota and HPV infection/clearance according to the manufacturer's protocol. DNA quality and have been conducted, and they were of very limited sample quantity were assessed using the ratios of 260 nm/280 nm size (9, 10). and 260 nm/230 nm. The DNA was then stored at 80C To investigate the correlation between cervicovaginal until further processing. microbiota and HPV infection status, particularly in asymptomatic women, we herein characterized and com- HPV assay and genotyping pared the composition of the cervical microbiota of 144 Cervical swab specimens stored in STM were tested by HPV-positive or -negative women with a normal cytologic Hybrid Capture 2 (HC2;Qiagen), using both the high- diagnosis, using 16S rDNA-based high-throughput risk (HR) panel of 13 pooled types (HPV16, 18, 31, 33, sequencing; and the differences in cervical microbiota 35, 39, 45, 51, 52, 56, 58, 59, and 68) and the low-risk associated with new acquisition, persistence, and clear- (LR) panel (types 6, 11, 42, 43, and 44). Samples were ances of HPV were analyzed throughout one year of processed according to the manufacturer's instructions follow-up. and considered positive if they reached the threshold value of at least 1.0 pg/mL of HPV DNA. All of the Materials and Methods specimens with positive HC2 (HR and/or LR HPV) Sample collection and processing results were further tested by Linear Array (Roche) for With the approval from the Institutional Review Board HPV genotyping. of the Department of Gynecologic Oncology of Beijing Obstetrics and Gynecology Hospital of Capital Medical Microbiota sequencing University (Beijing, China), the subjects of this study were The V4 region of the bacterial 16S rRNA gene was enrolled from women participating in health examinations amplified with the common primer pair (forward primer,

OF2 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women

50-GTGCCAGCMGCCGCGGTAA-30; reverse primer, 50- Alpha diversity, which represents the taxonomic diver- GGACTACHVGGGTWTCTAAT-30) combined with adapt- sity of microbial populations within one sample, was er sequences and barcode sequences (11). PCR amplifica- evaluated by Chao1, Observed_otus, PD_whole_tree, tion was performed with a modified protocol in a total Shannon and Simpson indices using QIIME pipeline. volume of 50 mL that contained 10 mL of buffer, 0.2 mLof Of them, the Observed_otus, Chao1, and PD_whole_tree Q5 High-Fidelity DNA Polymerase, 10 mL of High GC values reflect the richness aspect of microorganisms in Enhancer, 1 mL of dNTP, and 10 mmol/L of each primer, the sample. The Shannon and Simpson are diversity and 60 ng of genomic DNA. Thermal cycling conditions indices, combining the abundance and evenness of were as follows: an initial denaturation at 95C for OTUs. Beta diversity represents the distance or dissim- 5minutes,followedby15cyclesat95C for 1 minute, ilarity between microbial composition among different 50C for 1 minute, and 72C for 1 minute, with a final samples. The Jaccard distance matrix, the Bray–Curtis primer template extension at 72Cfor7minutes.The distance matrix, the unweighted Unifrac distance matrix, PCR products from the first-step PCR were purified with the weighted Unifrac distance matrix, and the Janson– VAHTSTM DNA Clean Beads. A second round of PCR ShannondistancematrixwerecalculatedbyQIIMEand was then performed in a 40-mL reaction volume that used for evaluating the beta diversity of the bacterial contained 20 mLof2 Phmsion HF MM, 8 mL of ddH2O, communities. 10 mmol/L of each primer, and 10 mL of PCR products from the first step. Thermal cycling conditions were as Statistical analysis follows: an initial denaturation at 98C for 30 seconds, All statistical analyses were performed using SPSS ver- followed by 10 cycles at 98C for 10 seconds, 65Cfor30 sion 22 (IBM) and the statistical package R (R Foundation seconds, and 72C for 30 seconds, with a final extension for Statistical Computing). at 72C for 5 minutes. Finally, all PCR products were Categorical variables were presented as frequencies quantified with Quant-iT dsDNA HS Reagent and pooled and percentages. We used c2 and Fisher exact tests to together. We performed high-throughput sequencing assess statistical associations between/among variables. analysis of bacterial rRNA genes on the purified, pooled Numerical variables were expressed as mean SD. We sample using the Illumina Hiseq 2500 platform (2 250 used one-way ANOVA (one-way analysis of variance) to paired ends). compare the differences among different groups. Spearman correlation analysis was performed for each of Bioinformatics analysis the alpha diversity indices (including Chao1, PD_whole_ The raw reads were demultiplexed and then trimmed, tree, observed_otus, and the Shannon and Simpson) to merged, and filtered by program Usearch9.0.2132_ explore their correlations with taxa at various levels of i86linux32 (available at https://www.drive5.com) fol- taxonomy and other variables. lowing the UPARSE pipeline. We trimmed all reads to Principal coordinate analysis (PCoA) and Adonis tests the position of the first base with a quality score <¼ 2, db-RDA were performed on Jaccard distance matrix, the and discarded the sequences shorter than 64 after trim- Bray–Curtis distance matrix, the unweighted Unifrac ming. Paired reads with a number of expected errors > distance matrix, the weighted Unifrac distance matrix, 1.00 were further filtered out during the filtering step. and the Janson–Shannon distance matrix, respectively, Sequences were dereplicated and clustered with a to investigate the differences of the beta diversity threshold of 97% similarity for picking operational between different characteristic variables of the partici- taxonomic unit (OTU) representatives after Chimera pants. The CST of the cervical microbiota was assigned by checking. We then mapped all of the sequences back hierarchical clustering analysis based on the Jensen– to the representative sequences resulting in the OTU Shannon distance matrix and Ward linkage described table for all of the samples. OTU taxonomies (from previously (12, 13). Phylum to Genus) were determined using the RDP The LEfSe difference analysis with LDA cutoff of 2 and Classifier (available at https://pyro.cme.msu.edu/classifi Wilcoxon rank-sum test were performed on abundance er/form.spr) with a confidence threshold of 80%; the data to explore the taxa that were significantly different species level taxonomies of the OTUs were further among the groups. We performed a pairwise Spearman determined using EzBioCloud Identify Service correlation with a 2-tailed probability of t for each (https://www.ezbiocloud.net/identify). OTU representa- correlation. tive sequences were aligned and further filtered to create a phylogenetic tree using QIIME1.90 pipeline (available Bayesian network at http://qiime.org/). The OTU table was randomly To assess the potential associations between/among subsampled down to the size of the smallest sample different relevant variables, a Bayesian network was con- to obtain equal sequencing depth. Finally, a total of structed using bnlearn R package, V.3.6.45 (available on 61,096 reads per sample were used for further analysis. http://www.bnlearn.com/), applying the Hill Climbing

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF3

Ritu et al.

algorithm, which is a greedy search algorithm based upon 9.6) 64)

scoring metrics (14). ¼ N 65 (40.0 – Results 31 Characteristics of participants 9.8) A total of 144 women were recruited to this study at 29) No ( ¼

baseline, and after 1 year 133 women attended the N 65 – follow-up visit and were enrolled in the study. Of these (42.4 women, 90 were HPV positive and 43 were HPV negative 28 at baseline; 55 women who were positive at baseline became HPV negative after 1 year, and 1 woman who 10.6) 15) Yes ( was HPV negative at baseline became positive at the ¼ N

follow-up visit. 65 – (44.4

Based upon the HPV genotyping results of 2 visits, 29 2 binary categorical variables were created for all participants: the baseline HPV positivity and the acquisition of 8.7) 75) No (

new HPV genotypes at the follow-up visit. For the baseline ¼

HPV-positive participants, 2 more binary categorical vari- N 64 – ables were created according to whether any type of HPV (42.6 63 (84.0%)6 (8.0%) 10 (66.7%) 2 (13.3%) 21 (72.4%) 3 (10.3%) 52 (81.3%) 5 (7.8%) 6 (8.0%) 3 (20.0%) 5 (17.2%) 4 (6.3%) 27 was cleared in the follow-up visit, and whether a patient was positive for any type of HPV at both visits. The risk-

factor characteristics available for the participants for each 8.7) 109) Yes ( b ¼ HPV status are summarized in Table 1. The detailed infor- a N 65 – mation on HPV genotyping before and after follow-up was (44.6 shown in Supplementary Table S1. 27 The age of the women with no new HPV genotype ﬁ

acquisition within a year were signi cantly higher than 8.6) 24) No (

their counterparts, while the proportion of "no parity" was ¼

ﬁ ﬁ N signi cantly higher. There were no other signi cant differ- 60 – (39.9 28 5 (20.8%) 5 (4.6%) 1 (4.2%) 11 (10.1%) ences among different groups of HPV statuses in terms of 18 (75.0%) 93 (85.3%) age, parity, delivery type, menstrual status, cigarette smoking, or use of hormonal contraceptives. 8.3) 43) Yes ( ¼

Overall assessment of intestinal microbiota N 60 – A total of 10,103K PE-reads of the 16S rDNA gene V4 (45.7 30 region were generated from the 133 specimens, with an average of 75,965.3 [1,168.9 (SD)] reads for each spec- 9.0) imen, ranging from 65,379 to 78,172. A total of 10,016 K 90) No ( high-quality PE-reads were obtained after trimming and ¼ N

ﬁ 65 – ltering. In the OTU clustering process, a total of 2,085 (42.9 sequences of chimeras were ﬁltered, with a yield of 1,528 27

OTUs. After alignment of the OTU representative 0.05). < P sequences using QIIME pipeline, a total of 1,494 OTUs 8.9)

were included for further data analysis. 133 Yes (

ﬁ 65 In the taxonomic assignment process with a con dence values ( – (43.8 ¼ P threshold of 0.8, 1,494 OTUs were identiﬁed and anno- OverallN Baseline HPV positive New HPV-type acquisition Any HPV-type cleared Any HPV-type persisted

tated. Among these, 1,342 OTUs at the phylum level had an cant ﬁ annotation reliability over 0.8 and covered 26 phyla; 1,283 OTUs at the class level had an annotation reliability over

0.8 and covered 48 classes; 1,220 OTUs at the order level SD) 27 had an annotation reliability over 0.8 and covered 85 orders; 1,113 OTUs at the family level had an annotation

reliability over 0.8 and covered 183 families; and 792 . Summary of the risk factor characteristics available for the participants in each HPV status 0.019. OTUs at the genus level had an annotation reliability over 0.018. ¼ ¼ 0 10 (7.5%) 9 (10.0%) 1 (2.3%) 2 12 (9.0%) 8 (8.9%) 4 (9.3%) Vaginal deliveryCaesarean sectionNot available 77 (57.9%) 28 (21.1%) 28 (21.1%) 52 (57.8%) 22 (24.4%) 16 25 (17.8%) (58.1%) 6 (14.0%) 10 12 (41.7%) (27.9%) 5 (20.8%) 9 67 (37.5%) (61.5%) 23 (21.1%) 44 19 (58.7%) (17.4%) 19 (25.3%) 8 (53.3%) 12 (16.0%) 3 (20.0%) 15 4 (51.7%) (26.7%) 6 (20.7%) 37 (57.8%) 8 (27.6%) 16 (25.0%) 8 (12.5%) 1 111 (83.5%) 73 (81.1%) 38 (88.4%) P P Age range (mean Parity number Table 1 a b 0.8 and covered 386 genera. Delivery type Postmenopausal [n (%)]Currently smoking [n (%)]Hormonalcontraception[n (%)]NOTE: Bold text 11 denotes (8.3%) 32 signi (24.1%) 11 (8.3%) 19 4 (21.1%) (4.4%) 8 (8.9%) 13 (30.2%) 4 (9.3%) 3 (7.0%) 4 (16.7%) 3 (12.5%) 3 (12.5%) 28 (25.7%) 8 (7.3%) 8 (7.3%) 15 (20.0%) 3 (4.0%) 8 (10.7%) 4 (26.7%) 0 (0.0%) 0 (0.0%) 6 (20.7%) 2 (6.9%) 1 13 (3.4%) (20.3%) 1 (1.6%) 7 (10.9%)

OF4 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women

Firmicutes (60.9%), Actinobacteria (14.4%), Proteobac- The Adonis test and db-RDA analysis were performed on teria (9.9%), and Bacteroidetes (8.5%) were found to be the all 5 distance matrices to investigate the association of the 4 most dominant taxa at the phylum level; as were Bacilli beta diversity and the variables of risk factor and HPV (57.5%), Actinobacteria (14.5%), Bacteroidia (8.4%), status. A significant association was found between men- and Gammaproteobacteria (7.3%) at the class level; strual status and the bacterial community structure based and Lactobacillales (57.0%), Bifidobacteriales (10.0%), on Bray-Curtis distance matrix, Janson-Shannon distance Bacteroidales (8.4%), and Pseudomonadales (4.7%) at the matrix, and weighted Unifrac distance matrix in Adonis order level. At the family level, Lactobacillaceae (53.8%), analysis and db-RDA analysis; and in the analysis based on Bifidobacteriaceae (10.0%), and Prevotellaceae (8.0%) Unweighted Unifrac distance matrix, the factors of age and were the most dominant taxa, covering more than 10% currently smoking were found to be significantly associated of the overall reads. At the genus level, a total of 30 genera with the bacterial community structure. No significant had an abundance of greater than 0.1%; and of these, association was found between the bacterial community Lactobacillus (53.8%), Prevotella (7.9%), Atopobium structure and other factors (Supplementary Table S2). The (4.0%), Pseudomonas (3.3%), Streptococcus (1.9%), and variable of menstrual status was associated with all 3 Acinetobacter (1.1%) were the most dominant taxa covering distance matrices based on abundance data, indicating the more than 1% of the overall reads. importance of abundance characters of the bacterial community on the difference between premenopausal and Comparison of the diversity of microbial communities postmenopausal women. Differences among taxa with among different HPV statuses higher abundance might play a significant role in the To compare the alpha diversity, the Chao1, PD-whole differences observed in the community structure among tree, observed-otu, and Shannon and Simpson indices different menstrual statuses. were calculated after we randomly subsampled the OTU The cervical communities were classified into 5 CSTs table down to 61,096 reads per sample, the size of the in hierarchical clustering analyses based on the Jensen– smallest sample to obtain equal sequencing depth. The 3 Shannon distance matrix and Ward linkage. As shown diversity indices regarding the richness (Chao1, PD-whole in Fig. 2, the CST1 is dominated by Lactobacillus iners tree, and observed-otu indices) were all lower for the HPV- (81.0%); CST2 by Lactobacillus crispatus (84.2%); CST3 by positive group compared with the HPV-negative women Leptotrichia amnionii (17.2%), Gardnerella ADEV_s (Fig. 1a, b and c). Although the variabilities were high (11.0%), and a variety of Pseudomonas spps that sum up among samples, the difference for the Chao1 index was to 36.3% of relative abundance; CST4 by Gardnerella significant in postmenopausal women (Fig. 1 a5). For the ADEV_s (41.8%), Atopobium vaginae (22.0%), and diversity indices regarding evenness, no differences were Lactobacillus iners (10.6%); and CST5 by Salmonella enterica found for both Shannon and Simpson indices between (10.2%), a variety of Lactobacillus spps, Pseudomonas spps, HPV-positive and -negative women (Fig. 1d and e). How- Prevotella spps, and Streptococcus spps, summing up to ever, after stratification by menstrual status, the evenness 19.5%, 9.5%, 8.7%, and 7.8% of relative abundance, diversity of HPV-positive women was found to be higher respectively. Among these, the CST1 and CST2 were similar than for the HPV-negative group in the premenopausal to the CST III (dominated by L. crispatus) and CST I women (Fig. 1d and e); whereas an opposite trend was (dominated by L. iners), respectively, previously described found in the postmenopausal group (Fig. 1d and 5). by Ravel and colleagues (13); while CST IV in a previous Similar tendency was observed for all 5 diversity indices study was further clustered into 3 groups (CST 3, CST 4, and relative to women with new HPV infections in a year CST 5) in this study. The proportion of each CST in and for women without new HPV infections, but the different groups of HPV status is shown in Supplementary differences were not statistically significant. Among Table S3. No significant differences were observed in the HPV-positive women, no significant differences were evaluation of the association between cervical CST and observed in the comparison of the cervical microbiota different HPV statuses or menstrual statuses. diversity between women with or without HPV persistence, or with or without HPV clearance within a year. Identification of cervical microbiota composition markers with respect to different HPV statuses Comparison of beta diversity of microbial communities Linear discriminant analysis (LDA) effect size (LEfSe) among different HPV statuses analysis was performed to identify differences in micro- To examine the bacterial community structure, 5 dis- biota composition that may be related to different statuses tance matrices were created on the basis of the 61,096 reads of HPV infection (Fig. 3). We identified 17 bacterial taxa after we randomly subsampled the OTU table, the Jaccard that have different abundancies between the baseline distance matrix, the Bray-Curtis distance matrix, the Jan- HPV-positive and -negative women. Among them, genus son-Shannon distance matrix, the unweighted Unifrac Ureaplasma and the family, order, class, and the phylum to distance matrix, and the weighted Unifrac distance matrix. which it belongs, were more abundant in the HPV-positive

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF5

Ritu et al.

Figure 1. Comparison of Chao 1, observed OTUs, PD_whole_tree, and the Shannon and Simpson index in different groups of women.

group; whereas Deinococcus and Thermus at the phylum and Fusobacteriaceae at the family level; and Faecalibacter- level; Deinococci and Acidobacteria Gp4 at the class ium, Ezakiella, Brochothrix, Diplorickettsia, and Fusobacterium level; Listeriaceae, Ectothiorhodospiraceae, Coxiellaceae, at the genus level were more abundant in the baseline

OF6 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women

Figure 2. 2D PCoA plots of the cervical microbiota performed on the Janson–Shannon distance metric correlated with CSTs.

HPV-negative group. In the comparison of women with or year, Rhodobacterales at the order level and Rhodobacter- without new acquisition of HPV within a year, only one aceae at the family level were associated with the clearing taxon, the genus Actinomyces was found to manifest a group. No taxon was found have a different abundance different abundance between the 2 groups. In a compar- between women with a HPV type persisting for a year and ison of women with or without any HPV type cleared in a women with no HPV type persisting for a year.

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF7

Ritu et al.

Figure 3. Linear discriminant effect size (LEfSe) analysis comparing differentially abundant taxa according to HPV status.

At the OTU level, 5 OTUs were identified to have associated with the HPV-persisting group in the compar- different abundancies between the baseline HPV-positive ison of women with or without any HPV type persisting for and -negative women. Of these, 2 OTUs, annotated to a year. The OTUs were annotated to Prevotella timonensis, Ureaplasma parvum and Fusobacterium nucleatum, respec- Lactobacillus delbrueckii, Eubacterium_g23 HQ746544_s, tively, had a relatively higher abundance in the baseline Microvirga lupini, Deinococcus swuensis, HM748650_g HPV-positive group; whereas, the other 3 OTUs, annotated FJ936969_s (belonging to the family Haliangiaceae), Dial- to Faecalibacterium GL538271_s, Ezakiella massiliensis, and ister invisus, Anaerobacterium chartisolvens, and Coprococcus Pseudomonas aeruginosa, respectively, had a relatively higher comes, respectively. abundance in the baseline HPV- negative group. In the comparison of women with or without new-acquisition Evaluation of the association of cervical microbiota HPV within a year, 4 OTUs, annotated to Alloprevotella composition and alpha diversity indices tannerae, Prevotella nigrescens, Prevotella oulorum, and Dial- To identify the cervical microbiota composition that ister invisus, respectively, possessed a relatively higher abun- correlated with each of the alpha diversity indices, we dance in women with new-acquisition HPV within a year. performed a pairwise Spearman correlation on all the In a comparison of women with or without any HPV type identified genus and 5 alpha diversity indices, respectively. cleared in a year, 11 OTUs were associated with the non- In the results, 79 taxa at the genus level were associated with cleared group. The OTUs were annotated to EU728721_g Chao1 index, the rho ranging from 0.17 to 0.54. For the AY959069_s (belonging to the family Lachnospiraceae), PD_whole_tree index, the correlations of 106 genera were Gemella asaccharolytica, Prevotella amnii, Prevotella shown to be statistically significant, the rho ranging GQ037615_s, Saccharimonas spps, Prevotella KQ959344_s, between 0.27 and 0.62. A total of 110 genera were Prevotella DQ815942_s, Solirubrobacter EU289476_s, Labili- statistically correlated with the observed_otu index with thrix AB286604_s, and Lachnotalea glycerin, respectively; the rho ranging from 0.20 to 0.56. For the Shannon and the remaining 1 (sharing 90.12% of similarity with index, the correlations of 52 genera were shown to be a strain of HG-B01269 in the EzBioCloud 16S database), statistically significant, the rho ranging from 0.75 to 0.69. failed to be annotated to a known taxon. Nine OTUs were For the Simpson index, the correlations of 37 genera were

OF8 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women 19 07 22 06 08 06 06 10 06 – – 13

ﬁ – – – – – – shown to be statistically signi cant, the rho ranging from – – 0.71 to 0.67. The top 10 genera correlated with each of the alpha indices are shown in Table 2. Campylobacter was the P genus most correlated with the Chao1 and PD_whole_tree indices, and it correlated third most with the observed_otu 0.71184 7.67E 0.406304 1.22E 0.397986 2.10E 0.515356 2.19E 0.408083 1.08E 0.672265 8.02E 0.586784 1.15E 0.470603 1.09E 0.415728 6.48E 0.397001 2.23E index. Genus Acinetobacter was the genus most correlated with the observed_otu index; and Lactobacillus and Prevotella were the 2 genera exhibiting the most negative and positive correlations, respectively, for both Shannon and Simpson indices. Porphyromonas Sneathia Gardnerella Anaerococcus Prevotella Dialister GenusLactobacillus Rho Peptoniphilus Parvimonas Atopobium 06 08 25 08 20 10 10 14 06 06 – – – – – – – – – Bayesian network – A Bayesian network was constructed on the variables of P 5 alpha diversity indices, HPV status, and menstrual status to assess potential associations (Fig. 4). Figure 4A was built on overall datasets, and it was shown 0.74636 6.36E 0.401343 1.69E 0.383539 5.19E 0.513237 2.67E 0.449137 5.88E 0.692605 2.59E 0.599151 2.57E 0.460197 2.50E 0.500198 8.77E 0.375381 8.50E that the observed_otu index was the only alpha diversity index directly associated with baseline HPV positivity; and the new acquisition of HPV type in a year was also directly associated with baseline HPV positivity. No variable tested was associated with menstrual status. Sneathia Parvimonas Peptoniphilus Porphyromonas Prevotella Lactobacillus Dialister Genus Rho Anaerococcus Gardnerella Figure 4B was constructed from data on HPV-positive Peptostreptococcus women to assess the potential associations of different 12 10 12 10 09 09 09 12 12 – 10 – – – – – – – – variables with HPV clearance and persistence. Our results – P showed that none of the 5 alpha diversity indices was 1.19E associated with either menstrual status or HPV clearance or persistence. To investigate the composition background of the cer- 0.48501 3.29E 0.487084 2.76E 0.550015 7.02E 0.505036 5.68E 0.562595 1.82E 0.564771 1.43E 0.553301 4.96E 0.510321 3.50E 0.521811 0.472572 9.29E vical microbiota that might underlie the association between baseline HPV positivity and observed_otu index, we created Fig. 4c on abundance data from 11 genera that were both statistically correlated with baseline HPV positivity (by the Wilcoxon rank-sum test, P < 0.05) and Sphingomonas Methylobacterium Pseudomonas Halorientalis Burkholderia Acinetobacter Campylobacter Genus Rho Halorubrum Corynebacterium observed_otu index (by Spearman correlation, P < 0.05). Ph 08 07 yllobacterium 09 08 15 09 09 08 09 07 – – – – – – – – –

The results showed that 4 genera were directly associated — with observed_otu index and 2 genera were directly associated with baseline HPV positivity. Of these, the genus P Ezakiella was the only microbiota composition that was directly linked to both baseline HPV positivity and obser- 0.438629 1.29E 0.432506 2.00E 0.483513 3.74E 0.454923 3.77E 0.483968 3.60E 0.487162 2.74E 0.457682 3.05E 0.491421 1.90E 0.617303 2.51E ved_otu index, indicating that it might play a critical role in 0.457876 3.00E the association of these 2 variables.

Discussion Peptoniphilus Methylobacterium Corynebacterium Sphingomonas Acinetobacter Pseudomonas Halorientalis Burkholderia Campylobacter Genus Rho In this study, we investigated the association between Halorubrum cervical microbiota and variables that included HPV pos- 05 05 06 05 06 05 05 11 05 05 – – – – – – – – – itivity and other risk factors (such as age, cigarette smoking, –

parity, menstrual status, and use of hormonal/oral contra- P ceptives) in women with a negative cytologic screening diagnosis. One-year follow-up was conducted on the par- 0.347797 4.10E 0.342055 5.58E 0.353552 2.99E 0.368356 1.29E 0.370631 1.13E 0.386527 4.32E 0.366248 1.45E 0.357015 2.46E ticipants to further investigate the association of baseline 0.543236 1.42E 0.393037 2.87E cervical microbiota and the acquisition of new HPV-type Chao1 PD_whole_tree Observed_otu Shannon Simpson infection, persistent infection of any HPV type during a year, and the clearance of any HPV type during a year. The top 10 genera correlated to alpha diversity indexes The results indicated that the cervical microbial richness Ezakiella Ruminococcus Varibaculum Porphyromonas Corynebacterium Bacteroides Roseburia Peptoniphilus GenusCampylobacter Finegoldia Rho of HPV-positive women was lower than HPV-negative Table 2.

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF9

Ritu et al.

Figure 4. Bayesian network showing the dependencies between/among variables selected on the basis of their association.

women, and the difference was more significant in the of any HPV type during a year, and the clearance of any postmenopausal group than in the premenopausal group. HPV type during a year) and the beta diversity of the Although no significant associations were found cervical microbiota, several taxa whose abundance between HPV status (namely, baseline HPV positivity, the differed significantly in subgroups with different HPV acquisition of new HPV type infection, persistent infection statuses were identified.

OF10 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women

In the comparison of the alpha diversity, it is interesting Campylobacter, Pseudomonas, Corynebacterium, Halorubrum, to note that the richness of the cervical microbiota of HPV- and Halorientalis—rather than Lactobacillus—were found to positive women was lower than for HPV-negative women be correlated significantly. in our study. This result is controversial with respect to Regarding beta diversity, the db-RDA and adonis anal- previous studies that showed that for the microbiota in the ysis found no significant association between different female reproductive tract, low microbial diversity was HPV statuses and distant matrices in this study, which is more closely associated with a healthier condition, and similar to the results of Di Paola and colleagues (10) and Di HPV-positive women have a higher species diversity Pietro and colleagues (17). The unsupervised clustering (4, 10). A possible explanation for this controversy is that based on distance matrix was another analyzing method compared with the previous study, only cytologically nor- commonly used in reproductive tract microbiota data to mal women were included in the HPV-positive group in classify women into different CSTs. Several investigators this study, so as to eliminate the potential impact of the using hierarchical clustering analyses based on the Jensen– microbiologic changes derived from HPV-related cervical Shannon distance matrix and Ward linkage reported clas- lesions. It is known that the microbiota's diversity increases sifying the participants into one of 5 distinct CSTs: CST-I with the severity of the cervical lesion and is significantly (most often dominated by Lactobacillus crispatus), CST-II higher than in healthy women (7, 15). Therefore, compar- (most often dominated by Lactobacillus gasseri), CST-III isons between the HPV-positive and -negative groups, (most often dominated by Lactobacillus iners), CST-IV irrespective of histologic or cytologic status, might lead to (characterized by a paucity of Lactobacillus spp. and a wide an overestimation of the diversity of the HPV-positive array of facultative and strict anaerobes), and CST-V (most group. Audirac–Chalifour and colleagues also showed no often dominated by Lactobacillus jensenii; refs. 15, 18, 19). significant differences (P ¼ 1, Student t test) in the com- Among these, it was reported that CST-IV or a certain parison of pd and Shannon index between HPV-positive subgroup of CST-IV was more highly associated with a and -negative groups when only NCL-negative cases were high Nugent score, increasing the severity of cervical considered (7). Collectively, our results indicated that the lesions and positivity for HPV. alpha diversity indices of microbiota in the female repro- In this study, using the same clustering method, we ductive tract were not a stable indicator of HPV positivity, also classified the participants into 5 groups with CST1 especially when the infection was asymptomatic. dominated by Lactobacillus iners similar to CST-III in the Lactobacillus species are the most common inhabitants of aforementioned studies and CST 2 dominated by Lactoba- the female reproductive tract, and reduced relative abun- cillus crispatus similar to CST-I. CSTs 3, 4, and 5 were all dance of Lactobacillus is often considered to lead to similar to CST-IV in the previous studies in not being increased diversity of reproductive tract microbiota (16). dominated by any Lactobacillus spp., and having their In this study, by calculating the 2-by-2 Spearman rank distinct indicator species, respectively. CST-II (dominated correlation coefficient (Spearman rho) for all of the taxa by Lactobacillus gasseri) and CST-V (dominated by Lactoba- and 5 alpha diversity indices uncovered, we found that cillus jensenii) described in the previous studies were not Lactobacillus—the most abundant genus in our overall found in our study. A possible cause of the inconsistent microbiota data—is only significantly associated with clustering pattern in this study compared with previous evenness but not associated with richness diversity. It was studies is that with a very high sequencing depth, much the most highly correlated (negatively) genus in both more detailed information of the microbiota, which might Shannon and Simpson indices, whereas its correlations be ignored with a lower sequencing depth, was demon- with pd_whole_tree, Chao1, and observed_otu indices strated in this study. The minimal number of reads for each were all nonsignificant. This result supports the theory sample in this study was high (up to 61,096), and a total of that the predominance of Lactobacillus spp. prevents the 1,528 OTUs that were assigned to 386 genera were colonization of other bacterial species and could lead to a observed from the sequences. In the comparison of the decrease in evenness of the microbiota. Conversely, our proportion of each CST in different HPV statuses, we failed results also suggest that the dominance of Lactobacillus does to observe any statistical association between CST and HPV not diminish the other bacterial species thoroughly, but status, indicating that the association between CST and that the abundance of other bacterial species might remain HPV infection status might also be population-specific, but at a very low level and be detected with the increase this requires further confirmation in a wider range of in sequencing depth. The genera Prevotella, Dialister, studies with larger sample sizes. Peptoniphilus, and Gardnerella were found to be positively In this study, the taxonomic differences in cervical correlated with evenness diversity, indicating that as microbiota in groups with different HPV statuses were Lactobacillus diminishes, the abundance of these other investigated with LEfSe analysis. We found several taxa to genera will significantly increase together with the Shan- possess significantly different abundances with respect to non and Simpson indices. For richness diversity, the comparison of the baseline positive and negative several other genera, such as Acinetobacter, Burkholderia. groups. Among them, the Ureaplasma parvum and their

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF11

Ritu et al.

related taxa manifested higher richness in the baseline In a comparison between HPV clearance and persistence, HPV-positive group compared with the baseline HPV- we found 22 taxa to be associated with the clearance or negative group. This is consistent with previous studies persistence of HPV over a year, and of these, 5 taxa belong in different populations of the world indicating that to Prevotella and 1 taxon belongs to Dialister. These taxa Ureaplasma parvum was associated with a significantly were all negatively associated with the clearance or posi- increased risk of overall HPV infection and increased risk tively associated with persistence of HPV types. These of abnormal cervical cytopathology (20–22). It was outcomes are in accordance with the findings of Dareng reported that the infection of Ureaplasma could lead to and colleagues (24) and Lee and colleagues (4), indicating scarring and damage of the epithelium (23), suggesting that for women with a higher abundance of Prevotella and that infection of the reproductive tract with Ureaplasma Dialister, it is more difficult to clear HPV and they more parvum might also increase the risk of HPV infection by easily develop HPV persistence, thereby exhibiting a higher facilitating viral infections. probability of detection as being persistently HPV positive. In contrast, the Brochothrix, Diplorickettsia, Ezakiella, Three of the remaining taxa found in our study belong to Faecalibacterium, and Fusobacterium genera and their related Lachnospiraceae, one of the bacteria characteristic of micro- taxa and Pseudomonas aeruginosa showed significantly biota associated with bacterial vaginosis (BV; refs. 25, 26). higher abundance in the HPV-negative group compared Several studies have shown that BV was correlated with with the HPV-positive group. It is interesting to note that higher incidence, prevalence, and persistence of HPV infec- although the abundance of Fusobacterium on a genus level tion and with development of CIN (27–29). Lactobacillus was associated with HPV negativity, one of the 5 OTUs delbrueckii was found to be associated with the absence belonging to Fusobacterium—assigned as Fusobacterium of HPV persistence in the HPV-positive group, which is JQ465429_s—was more abundant in the HPV-positive consistent with a recent study by Zhang and colleagues group. It was also reported in a previous study that (30), and supports the theory that certain species of Fusobacterium nucleatum (another species of Fusobacterium) Lactobacillus promote HPV clearance (18). However, in the was more abundant in the HPV-negative group (9). The 12 species of Lactobacillus found in this study, only 1 species controversial association of HPV positivity and different was found to be significantly associated with HPV infection Fusobacterium-related taxa indicates that different species of status, suggesting great heterogenicity of different Fusobacterium might play different roles in the process of subgroups of Lactobacillus in their association with HPV. HPV infection. Species of the Faecalibacterium genus were It is known that the species of the genus Lactobacillus share previously reported to demonstrate a higher abundance less than 70% of nucleotide identity across their genomes in women who cleared HPV infection after one year (15), so further investigations using metagenomic techni- compared with women who developed persistent HPV ques at the gene level would help us to understand how infection (10). However, little is known about interactions Lactobacillus plays protective or pathogenic roles with between Brochothrix, Diplorickettsia, Ezakiella, and Pseudo- regard to HPV. The remaining 10 taxa associated with HPV monas aeruginosa and HPV. clearance or persistence were novel and have not been The comparison between the baseline HPV-positive and reported elsewhere, but their association with HPV status HPV-negative groups was cross-sectional; therefore, we need to be confirmed in further studies. could not provide information about cause-and-effect There are some limitations to our study. First, we were relationships. Thus, a one-year follow-up was conducted not able to obtain information regarding BV status of the in this study and the causal links between cervical micro- participants; therefore, the relationship between cervical biota and new infection, persistence, and clearance of HPV microbes, BV, and HPV could not be evaluated. Second, were investigated; and we found 5 taxa that might be linked we used 16 sRNA gene sequencing to analyze the cervical to new HPV infection. Of these, the baseline abundance of microbiome, and this method could introduce bias Actinomyces was negatively associated with new HPV infec- during several processes, including the selection of the tion; and 3 taxa, Alloprevotella tannerae, Prevotella nigrescens, gene amplification area, the gene amplification proce- and Prevotella oulorum (which belong to Prevotellaceae) dure, the selection of the gene sequence database, and and Dialister invisus were positively associated with new OTU clustering. The strengths of our work include its HPV infection within a year's time. The association of these longitudinal study design, the strict inclusion criteria of taxa with HPV that we found in our study was similar to cytologic diagnosis that we used for enrollment, the those of previous studies. Actinomyces was negatively asso- relatively large sample size, and the higher sequencing ciated with HPV positivity in the study by Di Paola and depth. colleagues (10); whereas, the species of Prevotellaceae were In conclusion, in this study, we investigated the associa- reported to show greater abundance in the HPV-positive tions of cervical microbiota with different HPV infection group relative to the HPV-negative group (24). Dialister was statuses in cytologically normal women. We observed reported to be associated with HPV positivity in studies in several taxa that could differentiate baseline HPV positivity Korea and Nigeria (4, 24). and predict the acquisition of new HPV infection and

OF12 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Cervical Microbiota and HPV Status in Asymptomatic Women

persistence and/or clearance of HPV within a year. In the Authors' Contributions analysis of associations among alpha diversity indices, Conception and design: W. Ritu, W. Enqi, Y. Ling, Y. Wang HPV status, and microbiota component, we found that Development of methodology: W. Enqi, S. Zheng, Y. Wang the richness in HPV-positive women was higher than in Acquisition of data (provided animals, acquired and managed HPV-negative women. The richness diversity was primarily patients, provided facilities, etc.): W. Ritu, Y. Wang Analysis and interpretation of data (e.g., statistical analysis, correlated with the abundance of genera other than biostatistics, computational analysis): W. Enqi, S. Zheng, Y. Ling, Lactobacillus, such as Acinetobacter, Burkholderia. Campylo- Y. Wang bacter, Pseudomonas, Corynebacterium, Halorubrum, and Writing, review, and/or revision of the manuscript: W. Ritu, W. Enqi, Halorientalis; although, Lactobacillus was shown to be S. Zheng, Y. Ling, Y. Wang the taxon most correlated with evenness diversity. No Administrative, technical, or material support (i.e., reporting or signiﬁcant difference was observed among various HPV organizing data, constructing databases): W. Ritu, J. Wang, Y. Ling Study supervision: W. Ritu, Y. Ling infection statuses with respect to evenness diversity. — With our results together with those of previous Acknowledgments — studies we can conclude that certain cervical microbiota This work was supported in part by Beijing Municipal Science and compositions associated with different HPV infection sta- technology Commission (D131100005313009), the foundation of tuses may constitute a biomarker to help discriminate Key Laboratory of Ethnomedicine (Minzu University of China), women who are at risk for developing persistent HPV Ministry of Education (KLEM-ZZ201802) and National natural sci- infection. Furthermore, investigations of the mechanisms ence foundation of China (81473451 81673769). underlying these associations might provide useful The costs of publication of this article were defrayed in part by the insights into the development of new therapeutic strategies payment of page charges. This article must therefore be hereby marked to boost the clearance of HPV infection by modifying advertisement in accordance with 18 U.S.C. Section 1734 solely to reproductive tract microbiota. indicate this fact.

Disclosure of Potential Conflicts of Interest Received July 1, 2018; revised September 7, 2018; accepted Novem- fi No potential conflicts of interest were disclosed. ber 5, 2018; published rst November 21, 2018.

References 1. Groves IJ, Coleman N. Pathogenesis of human papillomavirus- 10. Di Paola M, Sani C, Clemente AM, Iossa A, Perissi E, Castronovo associated mucosal disease. J Pathol 2015;235:527–38. G, et al. Characterization of cervico-vaginal microbiota in women 2. Wheeler CM. The natural history of cervical human papillo- developing persistent high-risk Human Papillomavirus infection. mavirus infections and cervical cancer: gaps in knowledge Sci Rep 2017;7:10200. and future horizons. Obstet Gynecol Clin North Am 2013;40: 11. Fu J, Bonder MJ, Cenit MC, Tigchelaar EF, Maatman A, Dekens JA, 165–76. et al. The gut microbiome contributes to a substantial proportion 3. Noecker C, McNally CP, Eng A, Borenstein E. High-resolution of the variation in blood lipids. Circ Res 2015;117:817–24. characterization of the human microbiome. Transl Res 2017; 12. Romero R, Hassan SS, Gajer P, Tarca AL, Fadrosh DW, Nikita L, 179:7–23. et al. The composition and stability of the vaginal microbiota of 4. Lee JE, Lee S, Lee H, Song YM, Lee K, Han MJ, et al. Association of normal pregnant women is different from that of non-pregnant the vaginal microbiota with human papillomavirus infection in a women. Microbiome 2014;2:4. Korean twin cohort. PLoS One 2013;8:e63514. 13. Ravel J, Gajer P, Abdo Z, Schneider GM, Koenig SS, McCulle SL, 5. Gao W, Weng J, Gao Y, Chen X. Comparison of the vaginal et al. Vaginal microbiome of reproductive-age women. Proc Natl microbiota diversity of women with and without human papil- Acad Sci U S A 2011;108:4680–7. lomavirus infection: a cross-sectional study. BMC Infect Dis 14. Scutari M. Learning Bayesian networks with the bnlearn R pack- 2013;13:271. age. J Stat Softw 2010;35:22. 6. Oh HY, Kim BS, Seo SS, Kong JS, Lee JK, Park SY, et al. The 15. Mitra A, MacIntyre DA, Lee YS, Smith A, Marchesi JR, Lehne B, association of uterine cervical microbiota with an increased risk et al. Cervical intraepithelial neoplasia disease progression is for cervical intraepithelial neoplasia in Korea. Clin Microbiol associated with increased vaginal microbiome diversity. Sci Rep Infect 2015;21:674. 2015;5:16865. 7. Audirac-Chalifour A, Torres-Poveda K, Bahena-Roman M, Tellez- 16. van de Wijgert JH, Borgdorff H, Verhelst R, Crucitti T, Francis S, Sosa J, Martinez-Barnetche J, Cortina-Ceballos B, et al. Cervical Verstraelen H, et al. The vaginal microbiota: what have we learned microbiome and cytokine proﬁle at various stages of cervical after a decade of molecular characterization? PLoS One 2014;9: cancer: a pilot study. PLoS One 2016;11:e0153274. e105998. 8. Brotman RM, Shardell MD, Gajer P, Tracy JK, Zenilman JM, Ravel 17. Di Pietro M, Filardo S, Porpora MG, Recine N, Latino MA, Sessa R. J, et al. Interplay between the temporal dynamics of the vaginal HPV/Chlamydia trachomatis co-infection: metagenomic analysis microbiota and human papillomavirus detection. J Infect Dis of cervical microbiota in asymptomatic women. New Microbiol 2014;210:1723–33. 2018;41:34–41. 9. Shannon B, Yi TJ, Perusini S, Gajer P, Ma B, Humphrys MS, et al. 18. Mitra A, MacIntyre DA, Marchesi JR, Lee YS, Bennett PR, Kyrgiou Association of HPV infection and clearance with cervicovaginal M. The vaginal microbiota, human papillomavirus infection and immunology and the vaginal microbiota. Mucosal Immunol cervical intraepithelial neoplasia: what do we know and where are 2017;10:1310–19. we going next? Microbiome 2016;4:58.

www.aacrjournals.org Cancer Prev Res; 12(1) January 2019 OF13

Ritu et al.

19. Jahic M, Mulavdic M, Hadzimehmedovic A, Jahic E. Association 25. Dols JA, Molenaar D, van der Helm JJ, Caspers MP, de Kat between aerobic vaginitis, bacterial vaginosis and squamous Angelino-Bart A, Schuren FH, et al. Molecular assessment of intraepithelial lesion of low grade. Med Arch 2013;67:94–6. bacterial vaginosis by Lactobacillus abundance and species diver- 20. Ye H, Song T, Zeng X, Li L, Hou M, Xi M. Association between sity. BMC Infect Dis 2016;16:180. genital mycoplasmas infection and human papillomavirus infec- 26. Muzny CA, Sunesara IR, Kumar R, Mena LA, Griswold ME, Martin tion, abnormal cervical cytopathology, and cervical cancer: a DH, et al. Characterization of the vaginal microbiota among systematic review and meta-analysis. Arch Gynecol Obstet sexual risk behavior groups of women with bacterial vaginosis. 2018;297:1377–87. PLoS One 2013;8:e80254. 21. Amorim AT, Marques LM, Campos GB, Lobao TN, de Souza Lino 27. King CC, Jamieson DJ, Wiener J, Cu-Uvin S, Klein RS, V, Cintra RC, et al. Co-infection of sexually transmitted pathogens Rompalo AM, et al. Bacterial vaginosis and the natural and human papillomavirus in cervical samples of women of history of human papillomavirus. Infect Dis Obstet Gynecol Brazil. BMC Infect Dis 2017;17:769. 2011;2011:319460. 22. Drago F, Herzum A, Ciccarese G, Dezzana M, Casazza S, Pastorino 28. Guo YL, You K, Qiao J, Zhao YM, Geng L. Bacterial vaginosis is A, et al. Ureaplasma parvum as a possible enhancer agent of HPV- conducive to the persistence of HPV infection. Int J STD AIDS induced cervical intraepithelial neoplasia: Preliminary results. 2012;23:581–4. J Med Virol 2016;88:2023–4. 29. Gillet E, Meys JF, Verstraelen H, Verhelst R, De Sutter P, Temmer- 23. Waites KB, Katz B, Schelonka RL. Mycoplasmas and ureaplasmas man M, et al. Association between bacterial vaginosis and cervical as neonatal pathogens. Clin Microbiol Rev 2005;18:757–89. intraepithelial neoplasia: systematic review and meta-analysis. 24. Dareng EO, Ma B, Famooto AO, Akarolo-Anthony SN, Ofﬁong PLoS One 2012;7:e45201. RA, Olaniyan O, et al. Prevalent high-risk HPV infection and 30. Zhang C, Liu Y, Gao W, Pan Y, Gao Y, Shen J, et al. The direct and vaginal microbiota in Nigerian women. Epidemiol Infect indirect association of cervical microbiota with the risk of cervical 2016;144:123–37. intraepithelial neoplasia. Cancer Med 2018;7:2172–9.

OF14 Cancer Prev Res; 12(1) January 2019 Cancer Prevention Research

Evaluation of the Associations Between Cervical Microbiota and HPV Infection, Clearance, and Persistence in Cytologically Normal Women

Wu Ritu, Wu Enqi, Siriguleng Zheng, et al.

Cancer Prev Res Published OnlineFirst November 21, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1940-6207.CAPR-18-0233

Supplementary Access the most recent supplemental material at: Material http://cancerpreventionresearch.aacrjournals.org/content/suppl/2018/11/21/1940-6207.CAPR -18-0233.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerpreventionresearch.aacrjournals.org/content/early/2018/12/18/1940-6207.CAPR- 18-0233. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.