- Research article
- Open Access
- Open Peer Review
Bacterial vaginosis is associated with uterine cervical human papillomavirus infection: a meta-analysis
BMC Infectious Diseasesvolume 11, Article number: 10 (2011)
Bacterial vaginosis (BV), an alteration of vaginal flora involving a decrease in Lactobacilli and predominance of anaerobic bacteria, is among the most common cause of vaginal complaints for women of childbearing age. It is well known that BV has an influence in acquisition of certain genital infections. However, association between BV and cervical human papillomavirus (HPV) infection has been inconsistent among studies. The objective of this meta-analysis of published studies is to clarify and summarize published literature on the extent to which BV is associated with cervical HPV infection.
Medline and Web of Science were systematically searched for eligible publications until December 2009. Articles were selected based on inclusion and exclusion criteria. After testing heterogeneity of studies, meta-analysis was performed using random effect model.
Twelve eligible studies were selected to review the association between BV and HPV, including a total of 6,372 women. The pooled prevalence of BV was 32%. The overall estimated odds ratio (OR) showed a positive association between BV and cervical HPV infection (OR, 1.43; 95% confidence interval, 1.11-1.84).
This meta-analysis of available literature resulted in a positive association between BV and uterine cervical HPV infection.
Bacterial vaginosis (BV) is the most prevalent cause of abnormal vaginal discharge, affecting women of reproductive age . This infestation is characterized by a loss of indigenous (hydrogen peroxide-producing) Lactobacillus-predominant vaginal microflora, and a concurrent massive overgrowth of anaerobic bacteria. The most common include Gardnerella vaginalis, Mobiluncus species, Prevotella species, Mycoplasma hominis and Atopobium vaginae . At least 50% of patients have no symptoms . In the other half, it most often manifests clinically as a thin homogenous vaginal discharge, a vaginal pH of more than 4.5, presence of 'clue cells', and an amine odour after addition of 10% of potassium hydroxide [1, 2].
The etiopathogenesis of this condition remains subject of debate. Some risk factors have been associated with BV, including cigarette smoking, use of intrauterine devices, frequent vaginal douches, multiple sexual partners, early age at first intercourse, and black ethnicity [4, 5]. BV has been shown to increase the risk of obstetric and gynaecologic complications such as preterm labour and delivery, chorioamnionitis, post-caesarean endometritis, postabortion pelvic inflammatory disease, and cervicitis [6, 7]. Moreover, BV has been associated with many sexually transmitted infections (STIs), including infection with Chlamydia trachomatis, Neisseria gonorrhoeae, HSV-1 and 2, and an increased risk of HIV acquisition [4, 8, 9]. The leading hypothesis concerning these associations is that absence of protective lactobacilli increases biological susceptibility of acquiring an STI upon exposure. However, the temporal nature of the association between BV and acquisition of STIs remains an ongoing discussion. Although there is a large bulk of evidence favouring the plausibility that BV also incurs an elevated risk for human papillomavirus (HPV) acquisition, this remains a matter of debate.
It is well known that infection with oncogenic HPV, a sexually transmitted DNA virus, is the central etiological agent in the development of cervical cancer. Persistent HPV infection is a prerequisite for progression to high-grade lesions . However, few HPV infections persist and progress to cervical cancer . The vast majority cause no or only mild cytological abnormalities that may go undetected and regress to normalcy . It is unknown why high risk HPV infection is cancerous in some women whereas in others it is eradicated. Individual differences in immunological defence may be one explanation . Local cervical factors may determine the outcome of HPV. For this reason, there is a lot of interest in studying factors predisposing towards acquisition and persistence of this infection.
In contrast with cervical HPV infection, BV is associated with major changes in the vaginal environment. Because women with BV possess a Lactobacillus-poor flora, their changes in the vaginal ecosystem may provide biological plausibility for an increased risk or reactivation of HPV infection. Little is known about how the changed vaginal milieu in BV influences mucosal susceptibility for HPV, or vice versa, how infection with STIs in general influences the vaginal environment. The magnitude of association between BV and HPV has varied in epidemiological studies and remains controversial, yielding conflicting results and ranging from absence of any association  to a clear positive relationship .
To examine this controversial literature in more detail, a meta-analysis of available literature on the association between BV and cervical HPV infection was conducted. Estimates of association between BV and HPV are presented for HPV prevalence studies and analyzed for publication bias and heterogeneity.
Relevant studies on association between BV and HPV infection were identified through an extensive search of Medline, based on the following keywords: 'bacterial vaginosis', 'bacterial infections or vaginitis', 'BV', 'Gardnerella', and 'dysbacteriosis', in combination with 'human papillomavirus', 'papillomavirus infections', 'HPV' or 'cervical screening'. This search yielded 349 different published articles. Web of Science was further searched using the same strings, and yielding a total of 115 different publications. Only one additional eligible article was found beyond the Medline search . Studies that addressed the relationship between BV and cervical HPV infection were reviewed for predefined eligibility criteria. Two authors independently reviewed and evaluated critically all studies for inclusion (EG and DVB). Figure 1 summarizes the study selection process.
Eligible studies needed a clear description of diagnostic methods used for detecting both BV and HPV. There was no restriction in study design. Articles were included if they either reported odds ratios and corresponding 95% confidence intervals (CI) representing the magnitude of association between BV and cervical HPV infection, or presented data for calculation.
Reference lists of relevant papers and reviews were examined to identify further articles. Studies were limited to those written in English. We stopped our literature search in December 2009, but there was no publication starting-date limitation. The meta-analysis was restricted to original articles (no expert opinions, editorials or reviews). Conference abstracts and other unpublished articles were excluded, as these could not be systematically reviewed and data could not be verified. This meta-analysis was based on the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines .
Data abstraction and selection criteria
For each study, the following data were extracted: first author, year of publication, country where the study was conducted, study design, number of cases enrolled, study population, age range of participants, method of HPV diagnosis and HPV prevalence, BV diagnostic criteria, and BV prevalence.
Participants were categorized in four groups: women referred to colposcopy clinic because of an abnormal Pap-smear (referred), women attending family planning/obstetrics and gynaecology clinics (attendees), screening population, and mixed patient groups (referred, attendees and screened). In most studies, pregnancy was an exclusion criterion. Only one study included pregnant women . Another study enrolled HIV positive and high-risk HIV uninfected women .
BV prevalence was recorded as an estimate of BV in the study population. Diagnostic criteria for BV included Nugent's scoring system, Amsel clinical criteria, modified Amsel criteria, and presence of clue cells. In Nugent's scoring system (BV when score ≥ 7), the most accurate method, Gram-stained vaginal smears are assessed for average number of bacterial morphotypes seen per oil immersion field (magnification 100 times). Briefly, large gram-positive rods (Lactobacilli) were scored inversely from 0 to 4, small Gram-variable or gram-negative rods (Gardnerella and Bacteroides spp) from 0 to 4, and curved gram-variable rods (typically Mobiluncus spp) scored from 0 to 2 . Amsel criteria define BV as presence of any three of the following characteristics:  homogeneous white grey discharge that sticks to the vaginal walls;  vaginal fluid pH > 4.5;  release of fishy amine odour from vaginal fluid when mixed with 10% potassium hydroxide (positive whiff test); and  clue cells visible on wet mount microscopy . Modification of Amsel criteria included diagnosing BV when only two of these four elements were present (Peters et al.  used clue cells and positive whiff test). Diagnosing BV only through presence of clue cells on wet smear or more than 20% clue cells on Papanicolaou smear was also considered an inclusion criteria, since this is confirmed by previous studies to be an accurate method and good predictor for BV .
Studies eligible for inclusion detected cervical HPV infection with Polymerase Chain Reaction (PCR) or Fluorescent In Situ Hybridization (FISH). Koilocytosis was not considered specific enough for HPV detection. A list of studies included in the analysis and a digest of information extracted is given in Table 1.
Meta-analysis was conducted for twelve studies that fulfil the above-reported criteria [13–15, 17, 18, 20–26], using packages for STATA provided by Sterne et al. . Odds ratios and their respective standard errors were calculated from the provided raw data. For the remaining studies, crude odds ratios and standard errors as reported in the article were used [17, 24]. The resulting set of odds ratios were combined into a summary estimate of the association between BV and HPV using the random effects model of DerSimonian and Laird  and results were visualised in a forest plot. Evidence of publication bias was ruled out by funnel plot  and statistically evaluated for asymmetry using the Begg rank correlation . Homogeneity of effects across studies was assessed using Cochran's Q test  and quantified by Higgins and Thompson's I2 . Relative influence of different studies was evaluated by estimating the combined odds ratio after omitting one study at a time. Cumulative analysis, in which studies were added in order of descending variance on odds ratios, was done to rule out a potential small-study effect.
Study identification and description
Initial search gave rise to 406 unduplicated articles. Titles and abstracts were reviewed, and 51 out of 406 articles were considered of interest. These were retained for detailed evaluation, and 12 were finally retrieved for statistical analysis (Figure 1). Reasons for exclusion in the last step of our search strategy included: studies using koilocytosis as criterion for HPV detection [33–36], studies not describing their methodology of diagnosis [37, 38], and studies using presence of Gardnerella vaginalis [39, 40] or Grade II vaginal flora (according to Schröder et al.)  to diagnose BV.
Twelve eligible articles were identified, including a total of 6,372 women. These studies reported thirteen different estimates of association between BV and HPV prevalence for twelve study populations. One study reported estimates using two different methods of BV diagnosis, i.e. Amsel and presence of clue cells . The estimate based on the most stringent method (Amsel) was used for meta-analysis.
Most studies using adjusted odds ratios (AOR) did not describe clearly potential confounders and methods used. Consequently, the reported AOR could not be compared between studies. Therefore, where possible, raw data were retrieved for statistical analysis. Two studies did not mention raw data, hence only the reported crude odds ratios could be used [17, 24]. One study was excluded , because only crude and adjusted relative risks were described (crude RR, 1.20; 95% CI, 0.89-1.62; RR adjusted for ethnicity, sexual partners in past year and douching in past month, 1.08; 95% CI 0.82-1.42).
Studies included in the meta-analysis comprised ten cross-sectional studies [13–15, 17, 18, 20–22, 25, 26] and two follow up studies [23, 24]. One study measured additional incidence rates (defined as recruiting HPV-negative women and prospectively measuring incident HPV infection), but only baseline data was extracted for meta-analysis (odds ratio for incidence study, 1.41; 95% CI 1.25-1.59) .
Regarding geographical location, four studies were conducted in low-income [14, 15, 21, 25] and eight in developed countries [13, 17, 18, 20, 22–24, 26]). Five studies were conducted in Europe [13, 18, 20, 26, 43], three in the United States [17, 23, 24], three in South-America [14, 15, 21], and one in Asia . Eligible studies performed in Africa were not found.
Diagnosis and prevalence of bacterial vaginosis
BV was diagnosed either using clinical Amsel or modified Amsel criteria in seven out of twelve studies [14, 18, 20, 23, 25, 26, 43], Nugent's score in three out of twelve studies [17, 21, 24] and presence of clue cells in two out of twelve studies [13, 15]. BV prevalence ranged from 3.0% in sexually active university students ranged 18-24 years in the USA  to 47.2% in sexually active women ranged 13-19 years in the USA attending a primary care clinic . Large variation in reported prevalence figures can be attributed to differing recruitment strategies, inclusion of different patient populations, and variation in diagnostic criteria. Pooled BV prevalence was 31.2% (95% CI, 12.3%-51.6%).
Prevalence of BV using Nugent's criteria was consistently higher as opposed to studies using clinical Amsel criteria or presence of clue cells, ranging from 37.8 to 47.2%. Prevalence of BV using Amsel criteria and presence of clue cells ranged from 3.0 to 34.6% and from 15.2 to 20.2% respectively. Pooled prevalence of BV in low-income countries was 35.8% (95% CI, 20.8%-50.9%) while in developed countries it was 24.8% (95% CI, 12.4%-37.2%).
Bacterial vaginosis - cervical human papillomavirus association
Analysis of the association between BV and cervical HPV infection shows that HPV prevalence is significantly higher in BV positive women in only three out of twelve studies compared to women without BV [14, 17, 23]. Figure 2 represents reported odds ratios with their 95% CI for the likelihood of detecting cervical HPV in presence of BV, weight given to each study in random effects model, and combined odds ratio with 95% CI. Odds ratios in different studies ranged from 0.60  to 6.42 . The combined odds ratio for included cross-sectional studies was 1.43 (95% CI, 1.11-1.84, p = 0.005), indicating a positive association between BV and cervical HPV infection.
A funnel plot confirmed lack of obvious publication bias as no clear asymmetry could be detected (Figure 3). Also Begg's rank correlation test could not detect a significant publication bias (z = 0.82, p > 0.05). Included studies showed clear heterogeneity according to Cochran's Q test (χ2 = 28.8, p < 0.01). About 60% of the total variation could be explained by heterogeneity between samples (I2 = 60.8). Two studies [14, 15] reported higher odds ratios than can be expected in a homogeneous set of studies (figure 2). Substantial differences in reported odds ratios among other studies form an extra indication for existing heterogeneity.
Cumulative meta-analysis showed that small-study effects are unlikely to have an impact on the combined odds ratio (Figure 4). Studies with the largest standard error on their odds ratio have also a higher combined estimate [14, 15]. However, this effect is only visible for two studies. In general, evolution of the combined odds ratio is stable. Moreover, two studies with an odds ratio lower than one are among the smallest in this analysis.
To our knowledge, this review and meta-analysis with over 6,000 women is the first systematic evaluation of association between BV and cervical HPV infection. Although BV enhances acquisition of certain STIs, its relationship to cervical HPV infection is still an issue of controversy. Our results show evidence of a positive association between these two very common conditions, with an overall estimated odds ratio of 1.43.
Several hypotheses have been postulated, supporting this association. In BV-negative women, hydrogen peroxide-producing lactobacilli dominate the vaginal microflora and are part of the main defence mechanisms . Loss of these protective micro-organisms and other changes in the vaginal milieu, related to BV, could facilitate survival of other sexually transmitted agents and are risk factors for developing vaginal infections. It is well recognised that BV renders women vulnerable to acquisition of Neisseria gonorrhoeae, Chlamydia trachomatis, HSV-1 and 2, and HIV [8, 9, 42]. Moreover, BV has been associated with a reduction in vaginal fluid levels of secretory leukocyte protease inhibitor (SLPI), able to block HIV infection in vitro . It has been documented that BV propagates viral replication and vaginal shedding of HSV, thereby further enhancing spread of this STI .
Another hypothesis proposes that mucin-degrading enzymes are increased in vaginal fluid of women with BV. These enzymes, like sialidases, play a role in degradation of the gel layer coating the cervical epithelium, causing micro-abrasions or alterations of epithelial cells. The team of Briselden demonstrated positivity for sialidases in 84% of BV-positive women . Such enzymes may promote virulence through destroying the protective mucosa barrier and hence increase susceptibility to cervical HPV infection by facilitating adherence, invasion and eventually incorporation of HPV oncogenes into the genome of cells of the transformation zone. Abnormal vaginal microflora could also be implicated in maintenance of subclinical HPV. Furthermore, changes in cervico-vaginal milieu resulting from co-infections may exert an influence on the natural history of cervical HPV infection.
It is also possible that BV is a cofactor involved in acquisition or reactivation of HPV infection by affecting immunological balance within the cervical tissue as a result of changes in production of factors, such as cytokines (interleukin-1ß, interleukin-10) . Mucosal immune system activation represents a critical response against micro-organisms colonizing the reproductive tract. Neutrophil recruitment and activation is considered the main innate immune response against microbial and viral infections of vaginal mucosa . Women harbouring clue cells show no inflammatory signs and neutrophils are typically relatively absent in BV smears subjected to microscopy . Enzymes produced by anaerobic bacteria involved in the pathogenesis of BV can potentially alter immune signals and promote degradation of host factors, rendering women more susceptible of acquiring HPV.
These results, however, should be interpreted in light of a number of methodological limitations. The analysis suffers from the fact that most included studies had a cross-sectional design, where data on prevalence of BV and HPV infection were gathered simultaneously, instead of over time. Therefore this analysis is liable to reverse causation bias that would result from HPV infected women being more likely to acquire BV. This disadvantage prohibits concluding that BV increases risk of HPV acquisition or that there is a causal relationship. The sequence of infection is unknown and only a follow up study can determine which condition facilitates the other. In an incidence study by Watts et al., BV was significantly associated with detection of new HPV infection at follow-up visit (OR, 1.41; 95% CI 1.25-1.59) . Association between BV and HPV persisted even after adjustment for number of sexual partners, suggesting that women with BV may be more susceptible for HPV and not simply because of shared risk factors. In contrast, another longitudinal study performed a time-lag analysis to evaluate which condition preceded the other . The result suggested a temporal relationship, where BV was found to occur simultaneously with or after HPV infection, rather than ante-dating acquisition of HPV. Perhaps cervical HPV infection may favour changes in the vaginal milieu that enhances development of BV.
The question remains whether BV and cervical HPV infection are simply related because there is a biologic interaction between them, or because both occur frequently in sexually active women. A positive correlation between BV and HPV might be explained by the fact that sexual risk behaviour and promiscuity are found more often in women with BV than in comparison groups. Role of sexual transmission in causing or promoting BV continues to be a topic of debate, as e.g. highlighted by data in lesbians, who have a high prevalence of BV . Although not considered an STI in its usual sense (e.g. treatment of the sexual partner has no effect on frequency or relapses), the epidemiological profile of BV mirrors an STI . HPV is known to be one of the most common STIs, thus concerns regarding confounding by sexual behaviour certainly remain.
A number of variables are contributing to observed heterogeneity. Most prominent, prevalence of BV varied according to the population studied. Various social habits and ethno-geographical risk factors may explain the wide BV prevalence range observed (3%-47.2%). It is well recognized that prevalence of BV in African women is among the highest worldwide . This meta-analysis did not include studies conducted in Africa. Considering the high prevalence of BV in this continent, it would be very interesting to evaluate the association between BV and cervical HPV infection in African women, since we may expect a more pronounced effect. Our unpublished data of a cross-sectional study including 820 HIV-negative female sex workers in Mombasa (Kenya) confirms this. In multivariate logistic regression, controlled for other STIs and behavioural characteristics, borderline significance was found between BV and high-risk HPV infection (AOR, 1.72; p = 0.06).
Technical biases (e.g. collection of specimen), subjectivity, sensitivity and specificity of diagnostic methods are also attributing to detected heterogeneity. HPV detection methods varied among included studies (e.g. FISH is less sensitive compared to PCR) and also distribution of HPV viral genotypes differed largely. However, high-risk genotypes 16 and 18 present in prophylactic vaccines were (when mentioned) always included.
Further, this meta-analysis was limited to that of published studies, which could have caused publication bias, resulting from tendency to selectively publish results that are statistically significant. However, this had probably little impact as there was no evidence of funnel plot asymmetry. In addition, most studies reported a non-significant effect, which makes publication bias highly unlikely.
Currently available vaccines targeting HPV types 16 and 18, accounting for 70% of cervical cancers worldwide, opened up new avenues in prevention of this important public health problem. If a longitudinal prospective study shows a cause - effect model, than it is clear that greater attention needs to be given to BV in the global fight against HPV infection and women with BV should be considered a priority group for prophylactic vaccination. Cervical screening remains of course a major preventive focus for the cancer control program. If BV is a risk factor for cervical HPV acquisition, it is clear that screening guidelines must adapt and implement a sensitive tool like HPV DNA testing in primary screening in BV-positive women, instead of cytological testing. Closer follow-up of these patients should be considered. Restoring the vaginal microflora should in that case be a promising answer to the high prevalence of HPV infections. Randomized clinical trials to determine effect of BV control measures on HPV acquisition may then be worth considering. In addition to the need to evaluate the potential of BV treatment to prevent HPV acquisition and transmission, a better understanding of its risk factors and determinants of recurrence is required.
This meta-analysis suggests a positive association between BV and cervical HPV infection. Considering that these conditions are very common among women worldwide, further research in this field is imperative. More data from prospective studies are needed to accurately evaluate temporal sequence of acquisition of both conditions in any attempt to determine a causal relationship and to identify specific sub-populations with a stronger association between BV and HPV.
Morris M, Nicoll A, Simms I, et al: Bacterial vaginosis: a public health review. BJOG. 2001, 108: 439-450. 10.1016/S0306-5456(00)00124-8.
Livengood CH: Bacterial vaginosis: an overview for 2009. Rev Obstet Gynecol. 2009, 2: 28-37.
Verstraelen H, Verhelst R: Bacterial vaginosis: an update on diagnosis and treatment. Expert Rev Anti Infect Ther. 2009, 7: 1109-1124. 10.1586/eri.09.87.
Cherpes TL, Hillier SL, Meyn LA, et al: A delicate balance: risk factors for acquisition of bacterial vaginosis include sexual activity, absence of hydrogen peroxide-producing lactobacilli, black race, and positive herpes simplex virus type 2 serology. Sex Transm Dis. 2008, 35: 78-83. 10.1097/OLQ.0b013e318156a5d0.
Fethers KA, Fairley CK, Hocking JS, et al: Sexual risk factors and bacterial vaginosis: a systematic review and meta-analysis. Clin Infect Dis. 2008, 47: 1426-1435. 10.1086/592974.
Sweet RL: Gynecologic conditions and bacterial vaginosis: implications for the non-pregnant patient. Infect Dis Obstet Gynecol. 2000, 8: 184-190. 10.1002/1098-0997(2000)8:3/4<184::AID-IDOG16>3.0.CO;2-P.
Ugwumadu AH: Bacterial vaginosis in pregnancy. Curr Opin Obstet Gynecol. 2002, 14: 115-118. 10.1097/00001703-200204000-00003.
Atashili J, Poole C, Ndumbe PM, et al: Bacterial vaginosis and HIV acquisition: a meta-analysis of published studies. AIDS. 2008, 22: 1493-1501. 10.1097/QAD.0b013e3283021a37.
Wiesenfeld HC, Hillier SL, Krohn MA, et al: Bacterial vaginosis is a strong predictor of Neisseria gonorrhoeae and Chlamydia trachomatis infection. Clin Infect Dis. 2003, 36: 663-668. 10.1086/367658.
Koshiol J, Lindsay L, Pimenta JM, et al: Persistent human papillomavirus infection and cervical neoplasia: a systematic review and meta-analysis. Am J Epidemiol. 2008, 168: 123-137. 10.1093/aje/kwn036.
Castellsague X: Natural history and epidemiology of HPV infection and cervical cancer. Gynecol Oncol. 2008, 110: S4-7. 10.1016/j.ygyno.2008.07.045.
Huh WK: Human papillomavirus infection: a concise review of natural history. Obstet Gynecol. 2009, 114: 139-143.
Rahkola P, Mikkola TS, Ylikorkala O, et al: Association between high risk papillomavirus DNA and nitric oxide release in the human uterine cervix. Gynecol Oncol. 2009, 114: 323-326. 10.1016/j.ygyno.2009.05.003.
da Silva CS, Adad SJ, Hazarabedian de Souza MA, et al: Increased frequency of bacterial vaginosis and Chlamydia trachomatis in pregnant women with human papillomavirus infection. Gynecol Obstet Invest. 2004, 58: 189-193. 10.1159/000079822.
Figueiredo PG, Sarian LO, Tambascia JK, et al: Increased detection of clue cells in smears from cervical intraepithelial lesions with reduced expression of COX-2. Diagn Cytopathol. 2008, 36: 705-709. 10.1002/dc.20900.
Stroup DF, Berlin JA, Morton SC, et al: Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000, 283: 2008-2012. 10.1001/jama.283.15.2008.
Watts DH, Fazzari M, Minkoff H, et al: Effects of bacterial vaginosis and other genital infections on the natural history of human papillomavirus infection in HIV-1-infected and high-risk HIV-1-uninfected women. J Infect Dis. 2005, 191: 1129-1139. 10.1086/427777.
Peters N, Van Leeuwen AM, Pieters WJ, et al: Bacterial vaginosis is not important in the etiology of cervical neoplasia: a survey on women with dyskaryotic smears. Sex Transm Dis. 1995, 22: 296-302. 10.1097/00007435-199509000-00005.
Discacciati MG, Simoes JA, Amaral RG, et al: Presence of 20% or more clue cells: an accurate criterion for the diagnosis of bacterial vaginosis in Papanicolaou cervical smears. Diagn Cytopathol. 2006, 34: 272-276. 10.1002/dc.20418.
Sikstrom B, Hellberg D, Nilsson S, et al: Gynecological symptoms and vaginal wet smear findings in women with cervical human papillomavirus infection. Gynecol Obstet Invest. 1997, 43: 49-52. 10.1159/000291818.
Castle PE, Hillier SL, Rabe LK, et al: An association of cervical inflammation with high-grade cervical neoplasia in women infected with oncogenic human papillomavirus (HPV). Cancer Epidemiol Biomarkers Prev. 2001, 10: 1021-1027.
Boyle DC, Barton SE, Uthayakumar S, et al: Is bacterial vaginosis associated with cervical intraepithelial neoplasia?. Int J Gynecol Cancer. 2003, 13: 159-163. 10.1046/j.1525-1438.2003.13007.x.
Mao C, Hughes JP, Kiviat N, et al: Clinical findings among young women with genital human papillomavirus infection. American Journal of Obstetrics and Gynecology. 2003, 188: 677-684. 10.1067/mob.2003.164.
Samoff E, Koumans EH, Markowitz LE, et al: Association of Chlamydia trachomatis with persistence of high-risk types of human papillomavirus in a cohort of female adolescents. Am J Epidemiol. 2005, 162: 668-675. 10.1093/aje/kwi262.
Nam KH, Kim YT, Kim SR, et al: Association between bacterial vaginosis and cervical intraepithelial neoplasia. J Gynecol Oncol. 2009, 20: 39-43. 10.3802/jgo.2009.20.1.39.
Verteramo R, Pierangeli A, Mancini E, et al: Human Papillomaviruses and genital co-infections in gynaecological outpatients. Bmc Infectious Diseases. 2009, 9: 10.1186/1471-2334-9-16.
Sterne J: Meta-analysis in Stata: an updated collection from the Stata Journal. College Station. 2009, TX: Stata Press
DerSimonian R, Laird N: Meta-analysis in clinical trials. Control Clin Trials. 1986, 7: 177-188. 10.1016/0197-2456(86)90046-2.
Egger M, Davey Smith G, Schneider M, et al: Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997, 315: 629-634.
Begg CB, Mazumdar M: Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994, 50: 1088-1101. 10.2307/2533446.
Cochran W: The combination of estimates from different experiments. Biometrics. 1954, 10: 101-129. 10.2307/3001666.
Higgins JP, Thompson SG: Quantifying heterogeneity in a meta-analysis. Stat Med. 2002, 21: 1539-1558. 10.1002/sim.1186.
Verbruggen BS, Boon ME, Boon LM: Dysbacteriosis and squamous (pre)neoplasia of immigrants and Dutch women as established in population-based cervical screening. Diagn Cytopathol. 2006, 34: 377-381. 10.1002/dc.20374.
Engberts MK, Verbruggen BS, Boon ME, et al: Candida and dysbacteriosis: a cytologic, population-based study of 100,605 asymptomatic women concerning cervical carcinogenesis. Cancer. 2007, 111: 269-274. 10.1002/cncr.22947.
Campos AC, Freitas-Junior R, Ribeiro LF, et al: Prevalence of vulvovaginitis and bacterial vaginosis in patients with koilocytosis. Sao Paulo Med J. 2008, 126: 333-336.
Roeters AM, Boon ME, van Haaften M, et al: Inflammatory events as detected in cervical smears and squamous intraepithelial lesions. Diagn Cytopathol. 2009, 38: 85-93.
Moi H: Prevalence of bacterial vaginosis and its association with genital infections, inflammation, and contraceptive methods in women attending sexually transmitted disease and primary health clinics. Int J STD AIDS. 1990, 1: 86-94.
Watt A, Garwood D, Jackson M, et al: High-risk and multiple human papillomavirus (HPV) infections in cancer-free Jamaican women. Infect Agent Cancer. 2009, 4 (Suppl 1): S11-
Klomp JM, Boon ME, Van Haaften M, et al: Cytologically diagnosed Gardnerella vaginalis infection and cervical (pre)neoplasia as established in population-based cervical screening. Am J Obstet Gynecol. 2008, 199: 480-485. 10.1016/j.ajog.2008.04.036.
Murta EF, Souza MA, Araujo Junior E, et al: Incidence of Gardnerella vaginalis, Candida sp and human papilloma virus in cytological smears. Sao Paulo Med J. 2000, 118: 105-108. 10.1590/S1516-31802000000400006.
McNicol P, Paraskevas M, Guijon F: Variability of polymerase chain reaction-based detection of human papillomavirus DNA is associated with the composition of vaginal microbial flora. J Med Virol. 1994, 43: 194-200. 10.1002/jmv.1890430218.
Allsworth JE, Lewis VA, Peipert JF: Viral sexually transmitted infections and bacterial vaginosis: 2001-2004 National Health and Nutrition Examination Survey data. Sex Transm Dis. 2008, 35: 791-796. 10.1097/OLQ.0b013e3181788301.
Boyle DC, Smith JR: Infection and cervical intraepithelial neoplasia. Int J Gynecol Cancer. 1999, 9: 177-186. 10.1046/j.1525-1438.1999.99007.x.
Wahl SM, McNeely TB, Janoff EN, et al: Secretory leukocyte protease inhibitor (SLPI) in mucosal fluids inhibits HIV-I. Oral Dis. 1997, 3 (Suppl 1): S64-69.
Cherpes TL, Melan MA, Kant JA, et al: Genital tract shedding of herpes simplex virus type 2 in women: effects of hormonal contraception, bacterial vaginosis, and vaginal group B Streptococcus colonization. Clin Infect Dis. 2005, 40: 1422-1428. 10.1086/429622.
Briselden AM, Moncla BJ, Stevens CE, et al: Sialidases (neuraminidases) in bacterial vaginosis and bacterial vaginosis-associated microflora. J Clin Microbiol. 1992, 30: 663-666.
Cauci S: Vaginal Immunity in Bacterial Vaginosis. Curr Infect Dis Rep. 2004, 6: 450-456. 10.1007/s11908-004-0064-8.
Bailey JV, Farquhar C, Owen C: Bacterial vaginosis in lesbians and bisexual women. Sex Transm Dis. 2004, 31: 691-694. 10.1097/01.olq.0000143093.70899.68.
Verstraelen H: Bacterial vaginosis: a sexually enhanced disease. Int J STD AIDS. 2008, 19: 575-576. 10.1258/ijsa.2008.008189.
The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2334/11/10/prepub
DVB has a post-doctoral fellowship of the Research Foundation - Flanders (FWO). Part of this work was supported by the FWO.
The authors declare that they have no competing interests.
DVB and MT created the concept and design of this study. EG and DVB were responsible for literature search and extraction of data. JM carried out the statistical analysis. EG and DVB drafted the manuscript, which was critically revised and edited by HV, CB, PDS and MT. All authors read and approved the final manuscript.