https://orcid.org/0000-0002-8822-3617
Abstract
Vaginal dysbiosis characterized by depleted lactobacilli is usually correlated with human papillomavirus (HPV) infection and cervical carcinogenesis, but the effect of the Lactobacillus genus and represented species on this process remains unclear.
PubMed, EMBASE, and CENTRAL databases were searched up to February 15, 2019. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using a fixed-effect model and Review Manager (version 5.3) for Mac.
Eleven studies comprising 1230 cases were included. Lactobacillus spp. was associated with the decreased detection of high-risk subtype (hr)HPV infection (OR = 0.64, 95% CI = 0.48–0.87, I2 = 6%), cervical intraepithelial neoplasia (CIN) (OR = 0.53, 95% CI = 0.34–0.83, I2 = 0%), and cervical cancer (CC) (OR = 0.12, 95% CI = 0.04–0.36, I2 = 0%). At the level of Lactobacillus species, Lactobacillus crispatus, but not Lactobacillus iners, was correlated with the decreased detection of hrHPV infection (OR = 0.49, 95% CI = 0.31–0.79, I2 = 10%) and CIN (OR = 0.50, 95% CI = 0.29–0.88, I2 = 0%).
Cervicovaginal Lactobacillus spp. is associated with the decreased detection of hrHPV infection, CIN, and CC; L. crispatus may be the critical protective factor.
Cervical cancer (CC) is the fourth most common cancer for both incidence and mortality in women worldwide, with an estimated 570 000 new cases and 311 000 related deaths in 2018 [1]. It has been well established that high-risk subtypes of the human papillomavirus (hrHPV) are a necessary but not complete cause of CC and its premalignant precursor stages (cervical intraepithelial neoplasia [CIN]) [2]. In addition, there are other factors, such as immunodeficiency, smoking, and oral contraceptives, that could impact viral infection and subsequent disease [1]. In recent studies, there has been growing evidence that cervicovaginal microbiota could be an important cofactor in the etiology of CC [3–6].
The female genital tract harbors a diverse range of microorganisms that have been confirmed by emerging studies based on next-generation sequencing (NGS) [7]. In a single individual, microbial composition in the cervix and vagina shares an evident similarity, and cervicovaginal microbiota can be clustered into 5 community state types (CSTs) according to the predominant species [7–9]. Community state types I, II, III, and V are characterized by a dominance of Lactobacillus crispatus, Lactobacillus gasseri, Lactobacillus iners, and Lactobacillus jensenii, respectively [7]. Among them, L. crispatus and L. iners are the most prevalent Lactobacillus species in the female genital tract, and CST I and III are much more common than CST II and V in Asian, African, European, and North American women [7, 10–14]. Unlike these 4 CSTs that are dominated by lactobacilli, CST IV is characterized by the depletion of lactobacilli and enrichment with obligate anaerobic bacterial vaginosis (BV)-associated bacteria [7].
It is suggested that lactobacilli could play an important protective role in maintaining the homeostasis of the female genital tract, by blocking the adherence of pathogens, producing antimicrobial substances, and modulating inflammatory processes [15]. Lack of lactobacilli and concomitant dysbiosis of cervicovaginal microbiota increase the risk of genital infections, and previous meta-analyses already suggest a positive correlation between BV and CC along prevalent HPV infection, CIN, and cancer development pathways [3–6]. However, it still remains unclear what role the Lactobacillus genus and its represented species play in hrHPV infection and subsequent disease progression (see Discussion). Therefore, the current study aims to conduct a systematic review of the available literature and a meta-analysis of the results to statistically assess the associations of cervicovaginal lactobacilli with hrHPV infection, CIN, and CC, at both genus and species levels.
MATERIALS AND METHODS
Information Sources and Search Strategy
The electronic databases PubMed, EMBASE, and CENTRAL (Cochrane Central Register of Controlled Trials) were searched from inception to February 15, 2019 by using search terms in combination with both MeSH terms and free text for (Papillomaviridae OR uterine cervical neoplasms OR cervical intraepithelial neoplasia OR uterine cervical dysplasia) and (lactobacilli OR Lactobacillus OR Lactobacillales OR Lactobacillaceae). The full search strategy is described in Supplementary Tables 1–3. The reference lists of included studies and relevant reviews were also checked manually to identify eligible studies. The searches were not limited by publication date, but the included studies were limited to those published in the English language. The review protocol has been registered on the International Prospective Register of Systematic Reviews (PROSPERO) with the registration number CRD42017076281 and can be accessed at http://www.crd.york.ac.uk/prospero/.
Eligibility Criteria
Studies were included if they had assessed the association between cervicovaginal lactobacilli and the detection of hrHPV infection, CIN, and/or CC in women. Studies were excluded if the presence and percentage of lactobacilli were distinguished solely by microscopic examination. For quantitative synthesis, we only included studies in which cervicovaginal lactobacilli could be evaluated quantitatively and classified into CSTs. Studies in human cell lines or animal models were excluded. Non-English articles, published abstracts, reviews, and duplicated publications were excluded. Studies missing the numerical value of raw data were also excluded.
Study Selection and Data Extraction
Eligible studies were selected according to the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines [16]. Title and/or abstract screening and full-text review were conducted independently by 2 authors (H. W. and Y. M.), and all discrepancies were resolved by discussion or consultation with a senior researcher (W. Z.). The data were extracted by 2 other authors independently (X. C. and L. W.). Study characteristics included study sites, population, human immunodeficiency virus (HIV) status and age, study size and design, risk and protective factors, primary outcomes, assessment methods, and main conclusions.
Assessment of Risk of Bias
Two authors (X. C. and L. W.) independently assessed the quality of the study and risk of bias by using the Newcastle-Ottawa scale, a tool for assessing the quality of observational studies in meta-analyses [17]. The scale comprises 3 assessment categories: the selection of the study groups, the comparability between the groups, and the ascertainment of exposure. In this review, the Newcastle-Ottawa scale included the following quality-assessment criteria: (1) Case definition: Were detailed eligibility criteria with independent validation provided? (2) Case representativeness: Were the series of cases consecutive or obviously representative? (3) Selection of controls: Were these individuals selected from the same study population? (4) Definition of controls: Were the controls healthy subjects without hrHPV infection/CIN/CC? (5) Case-control comparability: Did analyses measure key population characteristics and possible confounding effects including age, race, HIV, and other genital pathogens? (6) Ascertainment of exposure: Were cervicovaginal lactobacilli assessed using NGS or quantitative microarray? (7) Assessment of cases and controls: Were cases and controls assessed using the same method? (8) Nonresponse rate: Did at least 80% of participants return for at least 1 follow-up visit? Answers with regard to bias were categorized to low risk, high risk, and unclear risk. Stars were given once the corresponding item was addressed satisfactorily in the study in question [17].
Data Synthesis
Study results, including the association of cervicovaginal lactobacilli with the primary outcomes (hrHPV infection, CIN, and CC), were pooled in meta-analyses, respectively. Analysis of other factors was not performed. The meta-analyses were conducted by computing odds ratios (ORs) and the corresponding 95% confidence intervals (95% CIs) using a fixed-effect model. Between-study heterogeneity was calculated using the Higgins I2 statistic and interpreted as nonimportant (<30%), moderate (30%–60%), or substantial (>60%), with the level of statistical significance set at 5% (P < .05) according to the Cochrane handbook [18]. Results were presented through forest plots. Potential publication bias was not investigated because the power of the tests was too low when the number of included studies was <10 [19]. Sensitivity analyses were performed by omitting 1 study at a time and recalculating the pooled ORs and 95% CI for the remaining studies. All statistical analyses were conducted by 2 authors (H. W. and Y. M.) independently using Review Manager (version 5.3) for Mac (The Nordic Cochrane Centre, Cochrane Collaboration 2014 [Copenhagen, Denmark]).
RESULTS
Study Selection
The procedure for study selection is presented in a flow diagram (Figure 1) as per the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [20]. As shown in Figure 1, 666 studies were identified through database searches, with 10 additional studies included manually. After removing the duplicates, titles, and abstracts, the 590 remaining studies were screened. Of these, 487 studies were excluded and 103 were retained for full-text review. Finally, 11 studies met the inclusion criteria for systematic review and meta-analysis [10–12, 21–28].
PRISMA flow chart for literature search and selection. CC, cervical cancer, CIN, cervical intraepithelial neoplasia; hrHPV, high-risk subtypes of the human papillomavirus.
Study Characteristics
All of the included studies, involving 1230 cases, were published from 2013 to 2018. Of the 11 studies, 4 were conducted in North America [11, 22, 25, 28], 3 in Asia [12, 24, 26], 2 in Africa [10, 21], and 2 in Europe [23, 27]. Cervicovaginal lactobacilli were assessed quantitatively and classified into CSTs in all 11 studies: 1 study by quantitative microarray [21] and the other 10 by NGS [10–12, 22–28]. Among the studies, 9 reported hrHPV outcomes [10, 12, 21–25, 27, 28], 6 reported CIN [11, 12, 23, 25–27], and 3 reported CC outcomes [11, 23, 25]. All of them were observational studies and 8 of the 11 studies were cross-sectional. For the 3 prospective studies, the association between cervicovaginal lactobacilli and the detection of hrHPV infection was evaluated cross-sectionally at baseline [22, 27, 28]. Additional information regarding the characteristics of the eligible studies is provided in Table 1.
Characteristics of Eligible Studies
| Study | Country | Population | HIV Status; Age | Study Size | Study Design | Risk Factors | Protective Factors | Primary Outcomes | HPV Assessment | HPV Subtypes Detecteda | Cytology/ Pathology Assessment | Microbial Sampling | Microbial Assessment | Main Conclusions |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lee et al [12] | Korea | Female twins, siblings and their mothers | Not tested; 31–73 years | 68 | Cross- sectional | Sneathia spp. | Lactobacillus spp. | hrHPV infection; CIN | MY09/MY11, nested GP5+/GP6+ PCR; DNA sequencing | A wide spectrum of HPV subtypes with both high and low risk | Pap smear | Vagina | PCR amplification of the V2–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | HPV+ women have higher microbial diversity with a lower proportion of Lactobacillus spp than HPV− women. |
| Borgdorff et al [21] | Rwanda | Female sex workers | Negative and positive; 18–47 years | 174 | Cross- sectional | - | Lactobacillus crispatus | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics, Pleasanton, CA) | 37 subtypes with both high and low risk | Conventional cytology | Cervix | Microarray with probes targeting 16S, 18S rRNA, and groEL genes | Women in L crispatus predominant CST I are less likely to have HIV, HSV-2, hrHPV, and bacterial STIs. |
| Brotman et al [22] | USA | Nonpregnant reproductive women with regular menstrual cycle and douching in 2 months before screening | Not tested; 22–53 years | 32 | Longitudinalb | Atopobium spp. | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Vagina | PCR amplification of the V1–V2 regions from 16S rRNA genes; Roche 454 pyrosequencing | Vaginal microbiota dominated by L gasseri is associated with increased clearance of detectable HPV. |
| Mitra et al [23] | UK | Premenopausal nonpregnant women | Negative; 23–45 years | 169 | Cross- sectional | Sneathia sanguinegens; Anaerococcus tetradius; Peptostreptococcus anaerobius | Lactobacillus spp. | hrHPV infection; CIN; cervical cancer | RealTime High Risk (HR) HPV Assay (Abbott Molecular, Des Plaines, IL) | 14 high-risk subtypes | Histology; cytology | Posterior vaginal fornix | PCR amplification of the V1–V2 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Advancing CIN disease severity is associated with decreasing abundance of Lactobacillus spp. |
| Oh et al [24] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 120 | Cross- sectional | Atopobium vaginae; Gardnerella vaginalis; Lactobacillus iners | Lactobacillus crispatus | hrHPV infection; CINc | Digene Hybrid Capture II DNA Test (QIAGEN, Gaithersburg, MD) | 13 high-risk subtypes | Pap smear | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | The synergistic predominance of A. vaginae, G. vaginalis, and L. iners with a concomitant paucity of L. crispatus in the cervical microbiota is associated with CIN risk. |
| Audirac-Chalifour et al [25] | Mexico | Mexican women with Mexican parents and grandparents | Not tested; 22–61 years | 29 | Cross-sectional | Sneathia spp.; Fusobacterium spp. | Lactobacillus crispatus; Lactobacillus iners | hrHPV infection; CIN; cervical cancer | Anyplex II HPV HR Detecion Assay (Seegene, Seoul, Korea) | 14 high-risk subtypes | Pathology examination; Pap smear; colposcopy | Cervix | PCR amplification of the V3–V4 regions from 16S rRNA genes; Roche 454 pyrosequencing | The predominant bacteria in healthy women are L. crispatus and L. iners, whereas for SIL, it is Sneathia spp. and for CC, Fusobacterium spp. |
| Dareng et al [10] | Nigeria | Nonpregnant women having intact uterus and prior vaginal sexual intercourse experience | Negative and positive; ≥18 years | 278 | Cross-sectional | Prevotella spp.; Leptotrichia spp. | Lactobacillus spp. | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | In HIV-negative women, hrHPV infection is associated with decreased abundance of Lactobacillus spp. but increased abundance of anaerobes. |
| Seo et al [26] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 137 | Cross-sectional | Lactobacillus iners; Atopobium vaginae | Lactobacillus crispatus | CIN | Digene Hybrid Capture II DNA Test (QIAGEN) | 13 high-risk subtypes | Pap smear; histology | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | Women in L. iners predominant CST III and A. vaginae predominant CST IV have higher CIN risk compared with L. crispatus predominant CST I. |
| Di Paola et al [27] | Italy | Nonpregnant Caucasian women without immunodeficiency-related diseases or HPV vaccine | Negative; 26–64 years | 72 | Longitudinalb | Atopobium spp.; Gardnerella vaginalis | - | hrHPV infection, CIN | Hybrid Capture II (QIAGEN, Milano, Italy); PGMY09/11 and HLA primers PCR | 13 high-risk subtypes | Cytology; colposcopy | Cervicovagina | PCR amplification of the V3–V5 regions from 16S rRNA genes; Roche 454 pyrosequencing | CST IV with depleted Lactobacillus spp. is the most represented CST in HPV+ women. |
| Shannon et al [28] | Canada | Nonpregnant African/Caribbean women without Neisseria gonorrhoeae or Chlamydia trachomatis | Negative; ≥16 years | 51d | Longitudinalb | - | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Cervicovaginal secretion | PCR amplification of the V3–V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Women with HPV infection more frequently have a high diversity cervicovaginal microbiome and are less likely to have a L. gasseri predominant microbiome. |
| Łaniewski et al [11] | USA | Nonpregnant premenopausal women | Not tested; NA | 100 | Cross- sectional | Sneathia spp. | Lcatobacillus spp.; Lactobacillus crispatus | HPV infectione; CIN; cervical cancer | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | Histology of colposcopy-directed biopsy samples; cytology | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Sneathia spp. is enriched, whereas Lactobacillus spp. is underrepresented in CC, CIN, and HPV+ controls. |
| Study | Country | Population | HIV Status; Age | Study Size | Study Design | Risk Factors | Protective Factors | Primary Outcomes | HPV Assessment | HPV Subtypes Detecteda | Cytology/ Pathology Assessment | Microbial Sampling | Microbial Assessment | Main Conclusions |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lee et al [12] | Korea | Female twins, siblings and their mothers | Not tested; 31–73 years | 68 | Cross- sectional | Sneathia spp. | Lactobacillus spp. | hrHPV infection; CIN | MY09/MY11, nested GP5+/GP6+ PCR; DNA sequencing | A wide spectrum of HPV subtypes with both high and low risk | Pap smear | Vagina | PCR amplification of the V2–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | HPV+ women have higher microbial diversity with a lower proportion of Lactobacillus spp than HPV− women. |
| Borgdorff et al [21] | Rwanda | Female sex workers | Negative and positive; 18–47 years | 174 | Cross- sectional | - | Lactobacillus crispatus | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics, Pleasanton, CA) | 37 subtypes with both high and low risk | Conventional cytology | Cervix | Microarray with probes targeting 16S, 18S rRNA, and groEL genes | Women in L crispatus predominant CST I are less likely to have HIV, HSV-2, hrHPV, and bacterial STIs. |
| Brotman et al [22] | USA | Nonpregnant reproductive women with regular menstrual cycle and douching in 2 months before screening | Not tested; 22–53 years | 32 | Longitudinalb | Atopobium spp. | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Vagina | PCR amplification of the V1–V2 regions from 16S rRNA genes; Roche 454 pyrosequencing | Vaginal microbiota dominated by L gasseri is associated with increased clearance of detectable HPV. |
| Mitra et al [23] | UK | Premenopausal nonpregnant women | Negative; 23–45 years | 169 | Cross- sectional | Sneathia sanguinegens; Anaerococcus tetradius; Peptostreptococcus anaerobius | Lactobacillus spp. | hrHPV infection; CIN; cervical cancer | RealTime High Risk (HR) HPV Assay (Abbott Molecular, Des Plaines, IL) | 14 high-risk subtypes | Histology; cytology | Posterior vaginal fornix | PCR amplification of the V1–V2 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Advancing CIN disease severity is associated with decreasing abundance of Lactobacillus spp. |
| Oh et al [24] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 120 | Cross- sectional | Atopobium vaginae; Gardnerella vaginalis; Lactobacillus iners | Lactobacillus crispatus | hrHPV infection; CINc | Digene Hybrid Capture II DNA Test (QIAGEN, Gaithersburg, MD) | 13 high-risk subtypes | Pap smear | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | The synergistic predominance of A. vaginae, G. vaginalis, and L. iners with a concomitant paucity of L. crispatus in the cervical microbiota is associated with CIN risk. |
| Audirac-Chalifour et al [25] | Mexico | Mexican women with Mexican parents and grandparents | Not tested; 22–61 years | 29 | Cross-sectional | Sneathia spp.; Fusobacterium spp. | Lactobacillus crispatus; Lactobacillus iners | hrHPV infection; CIN; cervical cancer | Anyplex II HPV HR Detecion Assay (Seegene, Seoul, Korea) | 14 high-risk subtypes | Pathology examination; Pap smear; colposcopy | Cervix | PCR amplification of the V3–V4 regions from 16S rRNA genes; Roche 454 pyrosequencing | The predominant bacteria in healthy women are L. crispatus and L. iners, whereas for SIL, it is Sneathia spp. and for CC, Fusobacterium spp. |
| Dareng et al [10] | Nigeria | Nonpregnant women having intact uterus and prior vaginal sexual intercourse experience | Negative and positive; ≥18 years | 278 | Cross-sectional | Prevotella spp.; Leptotrichia spp. | Lactobacillus spp. | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | In HIV-negative women, hrHPV infection is associated with decreased abundance of Lactobacillus spp. but increased abundance of anaerobes. |
| Seo et al [26] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 137 | Cross-sectional | Lactobacillus iners; Atopobium vaginae | Lactobacillus crispatus | CIN | Digene Hybrid Capture II DNA Test (QIAGEN) | 13 high-risk subtypes | Pap smear; histology | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | Women in L. iners predominant CST III and A. vaginae predominant CST IV have higher CIN risk compared with L. crispatus predominant CST I. |
| Di Paola et al [27] | Italy | Nonpregnant Caucasian women without immunodeficiency-related diseases or HPV vaccine | Negative; 26–64 years | 72 | Longitudinalb | Atopobium spp.; Gardnerella vaginalis | - | hrHPV infection, CIN | Hybrid Capture II (QIAGEN, Milano, Italy); PGMY09/11 and HLA primers PCR | 13 high-risk subtypes | Cytology; colposcopy | Cervicovagina | PCR amplification of the V3–V5 regions from 16S rRNA genes; Roche 454 pyrosequencing | CST IV with depleted Lactobacillus spp. is the most represented CST in HPV+ women. |
| Shannon et al [28] | Canada | Nonpregnant African/Caribbean women without Neisseria gonorrhoeae or Chlamydia trachomatis | Negative; ≥16 years | 51d | Longitudinalb | - | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Cervicovaginal secretion | PCR amplification of the V3–V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Women with HPV infection more frequently have a high diversity cervicovaginal microbiome and are less likely to have a L. gasseri predominant microbiome. |
| Łaniewski et al [11] | USA | Nonpregnant premenopausal women | Not tested; NA | 100 | Cross- sectional | Sneathia spp. | Lcatobacillus spp.; Lactobacillus crispatus | HPV infectione; CIN; cervical cancer | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | Histology of colposcopy-directed biopsy samples; cytology | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Sneathia spp. is enriched, whereas Lactobacillus spp. is underrepresented in CC, CIN, and HPV+ controls. |
Abbreviations: CC, cervical cancer; CIN, cervical intraepithelial neoplasia; CST, community state type; DNA, deoxyribonucleic acid; HIV, human immunodeficiency virus; HLA, human leukocyte antigen; HPV, human papillomavirus; hrHPV, high-risk subtype of HPV; HSV, herpes simplex virus; NA, not available; PCR, polymerase chain reaction; rRNA, ribosomal ribonucleic acid; SIL, squamous intraepithelial lesion; STI, sexually transmitted infection.
aWhen both high- and low-risk subtypes of HPV were assessed in one research, only high-risk ones were considered and analyzed in our study.
bThis is a prospective longitudinal study. However, the association between cervicovaginal microbiota and hrHPV status was evaluated cross-sectionally here.
cAll participants in the study had been included to evaluate the association between cervical microbiota and CIN in another research conducted by Seo et al[26] in 2016, so CIN is no longer set as the primary outcome for analysis here.
dThe total number of subjects in the study was 65, but only 51 had information regarding hrHPV and cervical microbiota.
eHPV subtype for each participant was not available in the study, so the association between cervical microbiota and hrHPV status could not be assessed here.
Characteristics of Eligible Studies
| Study | Country | Population | HIV Status; Age | Study Size | Study Design | Risk Factors | Protective Factors | Primary Outcomes | HPV Assessment | HPV Subtypes Detecteda | Cytology/ Pathology Assessment | Microbial Sampling | Microbial Assessment | Main Conclusions |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lee et al [12] | Korea | Female twins, siblings and their mothers | Not tested; 31–73 years | 68 | Cross- sectional | Sneathia spp. | Lactobacillus spp. | hrHPV infection; CIN | MY09/MY11, nested GP5+/GP6+ PCR; DNA sequencing | A wide spectrum of HPV subtypes with both high and low risk | Pap smear | Vagina | PCR amplification of the V2–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | HPV+ women have higher microbial diversity with a lower proportion of Lactobacillus spp than HPV− women. |
| Borgdorff et al [21] | Rwanda | Female sex workers | Negative and positive; 18–47 years | 174 | Cross- sectional | - | Lactobacillus crispatus | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics, Pleasanton, CA) | 37 subtypes with both high and low risk | Conventional cytology | Cervix | Microarray with probes targeting 16S, 18S rRNA, and groEL genes | Women in L crispatus predominant CST I are less likely to have HIV, HSV-2, hrHPV, and bacterial STIs. |
| Brotman et al [22] | USA | Nonpregnant reproductive women with regular menstrual cycle and douching in 2 months before screening | Not tested; 22–53 years | 32 | Longitudinalb | Atopobium spp. | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Vagina | PCR amplification of the V1–V2 regions from 16S rRNA genes; Roche 454 pyrosequencing | Vaginal microbiota dominated by L gasseri is associated with increased clearance of detectable HPV. |
| Mitra et al [23] | UK | Premenopausal nonpregnant women | Negative; 23–45 years | 169 | Cross- sectional | Sneathia sanguinegens; Anaerococcus tetradius; Peptostreptococcus anaerobius | Lactobacillus spp. | hrHPV infection; CIN; cervical cancer | RealTime High Risk (HR) HPV Assay (Abbott Molecular, Des Plaines, IL) | 14 high-risk subtypes | Histology; cytology | Posterior vaginal fornix | PCR amplification of the V1–V2 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Advancing CIN disease severity is associated with decreasing abundance of Lactobacillus spp. |
| Oh et al [24] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 120 | Cross- sectional | Atopobium vaginae; Gardnerella vaginalis; Lactobacillus iners | Lactobacillus crispatus | hrHPV infection; CINc | Digene Hybrid Capture II DNA Test (QIAGEN, Gaithersburg, MD) | 13 high-risk subtypes | Pap smear | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | The synergistic predominance of A. vaginae, G. vaginalis, and L. iners with a concomitant paucity of L. crispatus in the cervical microbiota is associated with CIN risk. |
| Audirac-Chalifour et al [25] | Mexico | Mexican women with Mexican parents and grandparents | Not tested; 22–61 years | 29 | Cross-sectional | Sneathia spp.; Fusobacterium spp. | Lactobacillus crispatus; Lactobacillus iners | hrHPV infection; CIN; cervical cancer | Anyplex II HPV HR Detecion Assay (Seegene, Seoul, Korea) | 14 high-risk subtypes | Pathology examination; Pap smear; colposcopy | Cervix | PCR amplification of the V3–V4 regions from 16S rRNA genes; Roche 454 pyrosequencing | The predominant bacteria in healthy women are L. crispatus and L. iners, whereas for SIL, it is Sneathia spp. and for CC, Fusobacterium spp. |
| Dareng et al [10] | Nigeria | Nonpregnant women having intact uterus and prior vaginal sexual intercourse experience | Negative and positive; ≥18 years | 278 | Cross-sectional | Prevotella spp.; Leptotrichia spp. | Lactobacillus spp. | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | In HIV-negative women, hrHPV infection is associated with decreased abundance of Lactobacillus spp. but increased abundance of anaerobes. |
| Seo et al [26] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 137 | Cross-sectional | Lactobacillus iners; Atopobium vaginae | Lactobacillus crispatus | CIN | Digene Hybrid Capture II DNA Test (QIAGEN) | 13 high-risk subtypes | Pap smear; histology | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | Women in L. iners predominant CST III and A. vaginae predominant CST IV have higher CIN risk compared with L. crispatus predominant CST I. |
| Di Paola et al [27] | Italy | Nonpregnant Caucasian women without immunodeficiency-related diseases or HPV vaccine | Negative; 26–64 years | 72 | Longitudinalb | Atopobium spp.; Gardnerella vaginalis | - | hrHPV infection, CIN | Hybrid Capture II (QIAGEN, Milano, Italy); PGMY09/11 and HLA primers PCR | 13 high-risk subtypes | Cytology; colposcopy | Cervicovagina | PCR amplification of the V3–V5 regions from 16S rRNA genes; Roche 454 pyrosequencing | CST IV with depleted Lactobacillus spp. is the most represented CST in HPV+ women. |
| Shannon et al [28] | Canada | Nonpregnant African/Caribbean women without Neisseria gonorrhoeae or Chlamydia trachomatis | Negative; ≥16 years | 51d | Longitudinalb | - | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Cervicovaginal secretion | PCR amplification of the V3–V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Women with HPV infection more frequently have a high diversity cervicovaginal microbiome and are less likely to have a L. gasseri predominant microbiome. |
| Łaniewski et al [11] | USA | Nonpregnant premenopausal women | Not tested; NA | 100 | Cross- sectional | Sneathia spp. | Lcatobacillus spp.; Lactobacillus crispatus | HPV infectione; CIN; cervical cancer | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | Histology of colposcopy-directed biopsy samples; cytology | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Sneathia spp. is enriched, whereas Lactobacillus spp. is underrepresented in CC, CIN, and HPV+ controls. |
| Study | Country | Population | HIV Status; Age | Study Size | Study Design | Risk Factors | Protective Factors | Primary Outcomes | HPV Assessment | HPV Subtypes Detecteda | Cytology/ Pathology Assessment | Microbial Sampling | Microbial Assessment | Main Conclusions |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lee et al [12] | Korea | Female twins, siblings and their mothers | Not tested; 31–73 years | 68 | Cross- sectional | Sneathia spp. | Lactobacillus spp. | hrHPV infection; CIN | MY09/MY11, nested GP5+/GP6+ PCR; DNA sequencing | A wide spectrum of HPV subtypes with both high and low risk | Pap smear | Vagina | PCR amplification of the V2–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | HPV+ women have higher microbial diversity with a lower proportion of Lactobacillus spp than HPV− women. |
| Borgdorff et al [21] | Rwanda | Female sex workers | Negative and positive; 18–47 years | 174 | Cross- sectional | - | Lactobacillus crispatus | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics, Pleasanton, CA) | 37 subtypes with both high and low risk | Conventional cytology | Cervix | Microarray with probes targeting 16S, 18S rRNA, and groEL genes | Women in L crispatus predominant CST I are less likely to have HIV, HSV-2, hrHPV, and bacterial STIs. |
| Brotman et al [22] | USA | Nonpregnant reproductive women with regular menstrual cycle and douching in 2 months before screening | Not tested; 22–53 years | 32 | Longitudinalb | Atopobium spp. | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Vagina | PCR amplification of the V1–V2 regions from 16S rRNA genes; Roche 454 pyrosequencing | Vaginal microbiota dominated by L gasseri is associated with increased clearance of detectable HPV. |
| Mitra et al [23] | UK | Premenopausal nonpregnant women | Negative; 23–45 years | 169 | Cross- sectional | Sneathia sanguinegens; Anaerococcus tetradius; Peptostreptococcus anaerobius | Lactobacillus spp. | hrHPV infection; CIN; cervical cancer | RealTime High Risk (HR) HPV Assay (Abbott Molecular, Des Plaines, IL) | 14 high-risk subtypes | Histology; cytology | Posterior vaginal fornix | PCR amplification of the V1–V2 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Advancing CIN disease severity is associated with decreasing abundance of Lactobacillus spp. |
| Oh et al [24] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 120 | Cross- sectional | Atopobium vaginae; Gardnerella vaginalis; Lactobacillus iners | Lactobacillus crispatus | hrHPV infection; CINc | Digene Hybrid Capture II DNA Test (QIAGEN, Gaithersburg, MD) | 13 high-risk subtypes | Pap smear | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | The synergistic predominance of A. vaginae, G. vaginalis, and L. iners with a concomitant paucity of L. crispatus in the cervical microbiota is associated with CIN risk. |
| Audirac-Chalifour et al [25] | Mexico | Mexican women with Mexican parents and grandparents | Not tested; 22–61 years | 29 | Cross-sectional | Sneathia spp.; Fusobacterium spp. | Lactobacillus crispatus; Lactobacillus iners | hrHPV infection; CIN; cervical cancer | Anyplex II HPV HR Detecion Assay (Seegene, Seoul, Korea) | 14 high-risk subtypes | Pathology examination; Pap smear; colposcopy | Cervix | PCR amplification of the V3–V4 regions from 16S rRNA genes; Roche 454 pyrosequencing | The predominant bacteria in healthy women are L. crispatus and L. iners, whereas for SIL, it is Sneathia spp. and for CC, Fusobacterium spp. |
| Dareng et al [10] | Nigeria | Nonpregnant women having intact uterus and prior vaginal sexual intercourse experience | Negative and positive; ≥18 years | 278 | Cross-sectional | Prevotella spp.; Leptotrichia spp. | Lactobacillus spp. | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | In HIV-negative women, hrHPV infection is associated with decreased abundance of Lactobacillus spp. but increased abundance of anaerobes. |
| Seo et al [26] | Korea | Nonpregnant, sexually active women with intact uterus, no history of gynecological cancers or treatment for CIN | Not tested; 18–65 years | 137 | Cross-sectional | Lactobacillus iners; Atopobium vaginae | Lactobacillus crispatus | CIN | Digene Hybrid Capture II DNA Test (QIAGEN) | 13 high-risk subtypes | Pap smear; histology | Cervix | PCR amplification of the V1–V3 regions from 16S rRNA genes; Roche 454 pyrosequencing | Women in L. iners predominant CST III and A. vaginae predominant CST IV have higher CIN risk compared with L. crispatus predominant CST I. |
| Di Paola et al [27] | Italy | Nonpregnant Caucasian women without immunodeficiency-related diseases or HPV vaccine | Negative; 26–64 years | 72 | Longitudinalb | Atopobium spp.; Gardnerella vaginalis | - | hrHPV infection, CIN | Hybrid Capture II (QIAGEN, Milano, Italy); PGMY09/11 and HLA primers PCR | 13 high-risk subtypes | Cytology; colposcopy | Cervicovagina | PCR amplification of the V3–V5 regions from 16S rRNA genes; Roche 454 pyrosequencing | CST IV with depleted Lactobacillus spp. is the most represented CST in HPV+ women. |
| Shannon et al [28] | Canada | Nonpregnant African/Caribbean women without Neisseria gonorrhoeae or Chlamydia trachomatis | Negative; ≥16 years | 51d | Longitudinalb | - | Lactobacillus gasseri | hrHPV infection | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | - | Cervicovaginal secretion | PCR amplification of the V3–V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Women with HPV infection more frequently have a high diversity cervicovaginal microbiome and are less likely to have a L. gasseri predominant microbiome. |
| Łaniewski et al [11] | USA | Nonpregnant premenopausal women | Not tested; NA | 100 | Cross- sectional | Sneathia spp. | Lcatobacillus spp.; Lactobacillus crispatus | HPV infectione; CIN; cervical cancer | Linear Array HPV Genotyping Test (Roche Molecular Diagnostics) | 37 subtypes with both high and low risk | Histology of colposcopy-directed biopsy samples; cytology | Mid-vagina | PCR amplification of the V4 regions from 16S rRNA genes; Ilumina MiSeq sequencing | Sneathia spp. is enriched, whereas Lactobacillus spp. is underrepresented in CC, CIN, and HPV+ controls. |
Abbreviations: CC, cervical cancer; CIN, cervical intraepithelial neoplasia; CST, community state type; DNA, deoxyribonucleic acid; HIV, human immunodeficiency virus; HLA, human leukocyte antigen; HPV, human papillomavirus; hrHPV, high-risk subtype of HPV; HSV, herpes simplex virus; NA, not available; PCR, polymerase chain reaction; rRNA, ribosomal ribonucleic acid; SIL, squamous intraepithelial lesion; STI, sexually transmitted infection.
aWhen both high- and low-risk subtypes of HPV were assessed in one research, only high-risk ones were considered and analyzed in our study.
bThis is a prospective longitudinal study. However, the association between cervicovaginal microbiota and hrHPV status was evaluated cross-sectionally here.
cAll participants in the study had been included to evaluate the association between cervical microbiota and CIN in another research conducted by Seo et al[26] in 2016, so CIN is no longer set as the primary outcome for analysis here.
dThe total number of subjects in the study was 65, but only 51 had information regarding hrHPV and cervical microbiota.
eHPV subtype for each participant was not available in the study, so the association between cervical microbiota and hrHPV status could not be assessed here.
Quality Assessment
The study quality was generally moderate to high, according to the Newcastle-Ottawa scale (Table 2). Eight studies were assigned at least 7 stars and were considered as having a low risk of bias. Three studies had medium risk because of inadequate case-control comparability and lack of case representativeness. A summary of bias assessment for the included studies can be found in Table 2.
Newcastle-Ottawa Scale for Risk of Bias Assessment for Observational Studies
| Checklist | Lee 2013 | Brogdorff 2014 | Brotman 2014 | Mitra 2015 | Oh 2015 | Audirac-Chalifour 2016 | Dareng 2016 | Seo 2016 | Di Paola 2017 | Shannon 2017 | Łaniewski 2018 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Selection | |||||||||||
| Case definition | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Case representativeness | - | ★ | ★ | ★ | ★ | - | ★ | ★ | ★ | ★ | - |
| Selection of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Definition of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Comparability | |||||||||||
| Case-control comparability | ★ | ★★ | ★ | ★★ | ★ | ★ | ★ | ★ | ★★ | ★★ | ★ |
| Exposure | |||||||||||
| Ascertainment of exposure | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Same ascertainment for cases and controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Nonresponse rate | N/A | N/A | ★ | N/A | N/A | N/A | N/A | N/A | ★ | ★ | N/A |
| Stars | 6 | 8 | 8 | 8 | 7 | 6 | 7 | 7 | 9 | 9 | 6 |
| Checklist | Lee 2013 | Brogdorff 2014 | Brotman 2014 | Mitra 2015 | Oh 2015 | Audirac-Chalifour 2016 | Dareng 2016 | Seo 2016 | Di Paola 2017 | Shannon 2017 | Łaniewski 2018 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Selection | |||||||||||
| Case definition | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Case representativeness | - | ★ | ★ | ★ | ★ | - | ★ | ★ | ★ | ★ | - |
| Selection of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Definition of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Comparability | |||||||||||
| Case-control comparability | ★ | ★★ | ★ | ★★ | ★ | ★ | ★ | ★ | ★★ | ★★ | ★ |
| Exposure | |||||||||||
| Ascertainment of exposure | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Same ascertainment for cases and controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Nonresponse rate | N/A | N/A | ★ | N/A | N/A | N/A | N/A | N/A | ★ | ★ | N/A |
| Stars | 6 | 8 | 8 | 8 | 7 | 6 | 7 | 7 | 9 | 9 | 6 |
Abbreviations: N/A, not applicable.
Newcastle-Ottawa Scale for Risk of Bias Assessment for Observational Studies
| Checklist | Lee 2013 | Brogdorff 2014 | Brotman 2014 | Mitra 2015 | Oh 2015 | Audirac-Chalifour 2016 | Dareng 2016 | Seo 2016 | Di Paola 2017 | Shannon 2017 | Łaniewski 2018 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Selection | |||||||||||
| Case definition | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Case representativeness | - | ★ | ★ | ★ | ★ | - | ★ | ★ | ★ | ★ | - |
| Selection of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Definition of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Comparability | |||||||||||
| Case-control comparability | ★ | ★★ | ★ | ★★ | ★ | ★ | ★ | ★ | ★★ | ★★ | ★ |
| Exposure | |||||||||||
| Ascertainment of exposure | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Same ascertainment for cases and controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Nonresponse rate | N/A | N/A | ★ | N/A | N/A | N/A | N/A | N/A | ★ | ★ | N/A |
| Stars | 6 | 8 | 8 | 8 | 7 | 6 | 7 | 7 | 9 | 9 | 6 |
| Checklist | Lee 2013 | Brogdorff 2014 | Brotman 2014 | Mitra 2015 | Oh 2015 | Audirac-Chalifour 2016 | Dareng 2016 | Seo 2016 | Di Paola 2017 | Shannon 2017 | Łaniewski 2018 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Selection | |||||||||||
| Case definition | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Case representativeness | - | ★ | ★ | ★ | ★ | - | ★ | ★ | ★ | ★ | - |
| Selection of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Definition of controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Comparability | |||||||||||
| Case-control comparability | ★ | ★★ | ★ | ★★ | ★ | ★ | ★ | ★ | ★★ | ★★ | ★ |
| Exposure | |||||||||||
| Ascertainment of exposure | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Same ascertainment for cases and controls | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ | ★ |
| Nonresponse rate | N/A | N/A | ★ | N/A | N/A | N/A | N/A | N/A | ★ | ★ | N/A |
| Stars | 6 | 8 | 8 | 8 | 7 | 6 | 7 | 7 | 9 | 9 | 6 |
Abbreviations: N/A, not applicable.
Synthesis of Results
First, cervicovaginal lactobacilli were considered as a whole (CST I, II, III, and V) and assessed quantitatively compared with nonlactobacilli-predominant CST IV. As shown in Figure 2A, the pooled analysis of 9 studies involving 941 cases yielded a summary OR of 0.64 (95% CI = 0.48–0.87), indicating that lactobacilli-predominant CSTs were associated with a reduced detection of hrHPV infection and the between-study heterogeneity was very low (I2 = 6%). The sensitivity analysis suggested that none of the studies was excessively influential, because no significant change to the pooled OR (range: minimum 0.59, maximum 0.70) was found when any one was excluded (Supplementary Table 4). According to the pooled analyses in Figure 2B and C, lactobacilli-predominant CSTs were also correlated with a reduced detection of CIN (OR = 0.53, 95% CI = 0.34–0.83) and CC (OR = 0.12, 95% CI = 0.04–0.36). No statistically significant heterogeneity was found in either analysis. The sensitivity analyses in Supplementary Table 4 show that none of the included studies was found to significantly affect the combined OR for CIN (range: minimum 0.43, maximum 0.57) or CC (range: minimum 0.07, maximum 0.18).
Forest plots of the cross-sectional associations between cervicovaginal Lactobacillus spp-predominant community state types and high-risk subtypes of the human papillomavirus (hrHPV) infection (A), cervical intraepithelial neoplasia (CIN) (B), and cervical cancer (CC) (C). CI, confidence interval.
Subsequently, the associations between L. iners, 1 of the 2 major cervicovaginal Lactobacillus species, and the outcomes were assessed, respectively. As shown in Figure 3A and B, the pooled analyses did not suggest a significant association between L. iners-predominant CST III and hrHPV infection (OR = 0.96, 95% CI = 0.69–1.34) or CIN (OR = 0.99, 95% CI = 0.60–1.64). None of the included studies was influential on the pooled results, as shown in the sensitivity analysis in Supplementary Table 5. However, an unexpected association between CST III and a decreased detection of CC (OR = 0.13, 95% CI = 0.02–1.13) can be observed in Figure 3C, whereas only 50 cases were involved, and no CST III event was found in the CC group.
Forest plots of the cross-sectional associations between cervicovaginal Lactobacillus iners-predominant community state type III and high-risk subtypes of the human papillomavirus (hrHPV) infection (A), cervical intraepithelial neoplasia (CIN) (B) and cervical cancer (CC) (C). CI, confidence interval.
Unlike L. iners-predominant CST III, the CST I dominated by L. crispatus was found to be significantly correlated with the decreased detection of both hrHPV infection (OR = 0.49, 95% CI = 0.31–0.79) and CIN (OR = 0.50, 95% CI = 0.29–0.88), which is shown in Figure 4A and B. For each of these analyses, between-study heterogeneity was very low and pooled results were robust, as suggested by the sensitivity analyses (Supplementary Table 6). Moreover, the pooling analysis of 2 studies involving only the 50 cases in Figure 4C also suggested a possible negative correlation between CST I and CC (OR = 0.17, 95% CI = 0.03–1.05).
Forest plots of the cross-sectional associations between cervicovaginal Lactobacillus crispatus-predominant community state type I and high-risk subtypes of the human papillomavirus (hrHPV) infection (A), cervical intraepithelial neoplasia (CIN) (B) and cervical cancer (CC) (C). CI, confidence interval.
Discussion
In the present systematic review and meta-analysis of the association between cervicovaginal lactobacilli and the detection of hrHPV infection, CIN, and CC, we found that women with Lactobacillus spp.-predominant CSTs were correlated with a decreased detection of hrHPV infection, CIN, and CC. At the level of Lactobacillus species, women with L. crispatus-predominant CST I were still correlated with the decreased detection of hrHPV infection and CIN. However, a similar pattern was not observed when L. iners was evaluated with hrHPV infection and CIN, because nonsignificant combined ORs were obtained in meta-analyses. Existing data were not adequate to obtain reliable results for the detection of CC at the level of Lactobacillus species.
A previous systematic review and meta-analysis undertaken by Tamarelle et al [6] correlates a low-Lactobacillus vaginal microbiota to HPV infection with a reported pooled OR of 1.53 (95% CI, 1.23–1.82) when compared with a high-Lactobacillus vaginal microbiota. The authors categorize the vaginal microbiota into low Lactobacillus or high Lactobacillus through 4 different methods: Nugent score, Amsel’s criteria, presence of clue cells, and 16s ribosomal ribonucleic acid (rRNA) gene amplicon sequencing. Among these methods, 3 depend on microscopic examination, which has intrinsic technical limitations including low resolution and specificity. Although microscopic examination can distinguish the large Gram-positive rod-shaped lactobacilli from small Gram-negative or Gram-variable rods, it is unable to distinguish different Lactobacillus species. In addition, the authors declare that the characterization of vaginal microbiota based on the 3 methods is not consistent. Therefore, they advocate the use of molecular techniques to provide a more in-depth characterization of vaginal microbiota and to evaluate the association between microbiota and HPV more precisely [6]. Moreover, this study does not address the high-risk subtypes of HPV, which are actually the prerequisite for tumorigenesis [2].
A recent systematic review and meta-analysis performed by Brusselaers et al [3] highlights the inconsistency between the results of microscopy-based and molecular techniques, and it pools the data separately. The analyses of microscopy studies suggest that vaginal dysbiosis (defined as deviation from the lactobacilli-dominated microbiota) is correlated with increased detection of HPV infection and persistence, as well as high-grade lesions and cancer. According to the analyses of molecular studies, non-L. crispatus-dominated vaginal microbiota is associated with an increased detection of HPV infection and persistence compared with the microbiota dominated by L. crispatus. However, only 3 studies were included for these analyses, and the total number of cases in each analysis was <100. Moreover, the association has not been evaluated at the level of the Lactobacillus genus. Two of the 3 molecular studies mentioned above [22, 27] were included in our analyses; one study [29] did not provide sufficient original numerical data and so it was finally excluded (Figure 1).
In addition to the above-mentioned systematic meta-analysis reviews, there were also 8 relevant reviews conducted without meta-analysis [30–37]. These reviews also support the notion that a decreased level of cervicovaginal lactobacilli is associated with increased detection of HPV infection and CC development [30–37]. We assessed the full texts of all the included references related to this topic and found that cervicovaginal lactobacilli were evaluated quantitatively by molecular techniques in most references. Therefore, the present systematic review and meta-analysis was conducted with more targeted inclusion criteria, in that (1) cervicovaginal microbiota was assessed quantitatively by molecular techniques and classified into CSTs and (2) high-risk subtypes of HPV were assessed or could be distinguished from low-risk ones.
Our main finding was that Lactobacillus spp.-predominant CSTs correlated to a decreased detection of hrHPV infection, CIN, and CC. This notion is in accordance with the conclusions of previous reviews and meta-analyses [3, 6, 30–37], and it is supported by the results of certain original research that could not be included in our meta-analysis for the following reasons: lack of original results [29], the limits in microbial identification [38–40], and the limits in microbial quantification [41]. As is generally accepted, cervicovaginal Lactobacillus spp. can produce various inhibitory substances, such as lactic acid, H2O2, and bacteriocin, to prevent pathogenic infection and maintain the integrity of genital epithelial surfaces [32]. They can also modulate host immune function and influence disease susceptibility [42]. In addition, cervicovaginal lactobacilli could also exert direct cytotoxic effects on CC cells, as reported by in vitro studies [43, 44]. Therefore, cervicovaginal Lactobacillus spp. may play a protective role in the process of hrHPV infection and subsequent CC development [32].
Although lactobacilli dominance in the female genital tract usually indicates a healthy and stable status, the manifestations were diverse both in vitro and in vivo among different Lactobacillus species [45, 46]. On the one hand, the prevalence and abundance varied considerably among the 4 major species of cervicovaginal lactobacilli [7, 10–14]. In the currently included cohort of studies, the L. jensenii-predominant CST V was observed in 2 of 117 cases by Mitra et al [23]. Lactobacillus gasseri-predominant CST II was observed in the following 4 studies: 2 of 32 cases in Brotman et al [22], 2 of 117 cases in Mitra et al [23], 8 of 72 cases in Di Paola et al [27], and 2 of 51 cases in Shannon et al [28]. Lactobacillus crispatus and L. iners were the 2 most prevalent lactobacilli in all of our included studies [10–12, 21–28]. This agrees with the results of other studies [7, 13, 14]. On the other hand, L. crispatus dominance was found to be associated with a lower pH and lower levels of inflammation in the female genital tract compared with other Lactobacillus species [7, 42, 47]. Reimers et al [29] observed that L. crispatus-predominant CST I, but not L. iners-predominant CST III or other CSTs, was also inversely associated with HPV infection. In the present review and meta-analysis, the detection of hrHPV infection and CIN was found to be inversely correlated with CST I but not III. This notion has also been supported by 2 studies, which were finally excluded from our meta-analysis because of the limit in their microbial quantification [48, 49].
The present review and meta-analysis had some limitations. First, the majority of the included studies were cross-sectional, and the data as analyzed were entirely cross-sectional. Cross-sectional studies have no dimension of time, because all variables, including assumed outcomes and exposures, are examined at the same point in time. Therefore, they do not provide a strong basis for establishing a causal relationship. When evaluating the correlation of a potential risk factor and a disease, the strength of cross-sectional studies is quite limited. This cross-sectional nature of the data and analyses was a primary limitation, not only in this study but also in the above-mentioned meta-analyses [3, 6] and reviews [30–37], which are all predominantly based on cross-sectional data. Second, the number of studies and the sample size of included studies were relatively small. On the one hand, this prevented a more comprehensive analysis of lactobacilli other than L. crispatus and L. iners. As mentioned above, among the 11 included studies involving a total of 1230 cases, only 2 cases of CST V had been detected in 1 study [23] and 14 cases of CST II from 4 studies [22, 23, 27, 28]. It is not satisfactory to extract data and conduct a meta-analysis based on so few cases and studies. Although it seems unlikely that these 2 CSTs with such a low occurrence could be the critical species contributing to the protective effects of the Lactobacillus genus, this was found to be a limitation of the current study. On the other hand, the number of studies and samples was insufficiently rich to provide a thorough separate analysis for women of every ethnicity. Therefore, any difference related to sample ethnicity could not be determined at the current time. Third, although we selected studies based on NGS or quantitative microarray, the target regions of the 16s rRNA gene amplified in the included studies were not identical (see Table 1), which might lead to bias in the results [50]. In addition, the main analyses in the current study were entirely based on CST, which has been proposed and widely used regarding cervicovaginal microbiota [7, 10, 12, 13, 21–29, 32–34, 37, 45]. However, there might be other types of data, such as relative abundance, that might strengthen results, especially analyses on a species-specific level.
Conclusions
In conclusion, our study suggests that cervicovaginal lactobacilli are associated with the reduced detection of hrHPV infection, CIN, and CC. Lactobacillus crispatus, but not L. iners, may be the specific protective factor in the process. These findings need to be examined using large prospective, multicenter studies in the future.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Author contributions. H. W. and W. Z. designed the study. H. W. and Y. M. performed the literature search and review. X. C. and L. W. assessed the study quality and risk of bias. X. C. and L. W. also extracted and collected the data. H. W. and Y. M. analyzed and interpreted the data. H. W. drafted the manuscript. R. L. critically revised the manuscript. All authors have read and approved the final manuscript.
Financial support. This work was funded by a grant from the National Natural Science Foundation (Grant Number 81702560), People’s Republic of China.
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.
