Prevalence and predictors of bacterial vaginosis in HIV-infected women in Maharashtra, India.

We evaluated the prevalence and determinants of bacterial vaginosis (BV) in HIV-infected women from Maharashtra, India. Among 912 HIV-infected women enrolled, BV was diagnosed in 191 (20.9%) and intermediate BV was diagnosed in 258 (28.3%) women. Women with more than two pregnancies had 1.6 times increased risk of BV (95% CI 1.0, 2.5, p-value 0.038), women who were menopausal had 6.2 times increased risk of BV (95% CI 2.4, 15.6, p-value <0.001) and women who were human papillomavirus (HPV) positive had 2.3 times increased risk of BV (95% CI 1.4, 3.9, p-value 0.001). Although we observed significantly increased risk of BV among women diagnosed with cervical intraepithelial neoplasia or worse disease in the univariate analysis (odds ratio 3.5, 95% CI 1.5, 8.1, p-value 0.004), it did not reach statistical significance in the multivariate analysis. Women who had the first sexual intercourse after the age of 18 had significantly lower risk of BV. To conclude, we observed high prevalence of BV in HIV-infected women and increased risk of BV in HPV positive, HIV-infected women.


Introduction
Among the women of reproductive age, bacterial vaginosis (BV) is the most common cause of vulvovaginitis. 1 The aetiopathology of BV is still being debated and the epidemiologic data support sexual transmission; however, disagreement exists. 2 BV may be considered a sexually enhanced disease, with frequency of intercourse being a critical factor. 3 Associations between BV and factors such as ethnicity, intravaginal douching, new sexual partners, multiple sex partners and unprotected vaginal intercourse have been reported. 4,5 Globally, the prevalence of BV varies considerably between countries, regions within countries and ethnic groups, and it can be as high as 60% in certain regions. 6 This variation is because of the differences in the biological, behavioural, medical, social and economic factors. 2 BV is characterized by an imbalance in the normal vaginal flora. The normal vaginal flora consists of lactic acid-and hydrogen peroxide (H 2 O 2 )-producing lactobacilli. BV is associated with reduction in the normal vaginal flora and concurrent increase in vaginal anaerobic flora. Clinically, BV is characterized by the presence of three of the four criteria which include discharge (thin, homogenous, uniformly adherent, white discharge), vaginal pH > 4.5, fishy odour on addition of 10% potassium hydroxide (KOH) and 20% clue cells (epithelial cell margins obscured by bacteria) on microscopic examination of vaginal smear. 7 However, about 50% of the women with BV do not have any symptoms, 8 hence its control is difficult and it remains untreated. Therefore, a standardized scoring system for the interpretation of Gram-stained vaginal smears called as Nugent's scoring has been introduced. 9 Nugent's scoring is the gold standard for the laboratory diagnosis of BV.
BV has been associated with an increased risk of human immunodeficiency virus (HIV) acquisition 10 and transmission. 11 But the studies that concluded increased risk of HIV transmission to the HIVuninfected partner were done when women living with HIV did not receive highly active antiretroviral therapy (HAART) which reduces the HIV transmission significantly. 12 A recent study from Denmark has shown that in women who are on HAART, presence of BV, human papillomavirus (HPV) and herpes viridae does not predict vaginal HIV RNA shedding implying that HIV shedding is not increased by BV in women on HAART. 12 But women with BV are at an increased risk of acquiring various other sexually transmitted infections (STIs) such as herpes simplex virus type 2, 13 Trichomonas vaginalis, Neisseria gonorrhoeae [14][15][16] and Chlamydia trachomatis. 16 This increased risk is possibly because of the inflammation of the genital mucosa. 17 BV also increases the risk of adverse obstetric and gynaecological outcomes such as preterm delivery, complications of gynaecological surgeries and pelvic inflammatory disease. 18,19 A high prevalence of BV between 35 and 49% has been reported from a large study conducted among more than 37,000 women in sub-Saharan Africa. 20 A meta-analysis of studies for the risk factors of BV has reported that BV is significantly associated with sexual contact with a new partner or multiple sexual partners and that decreasing the number of unprotected sexual encounters may reduce incident and recurrent infection. 4 Very few studies in India have reported BV prevalence among women and it was 8.6 and 20.5% among pregnant women attending two different antenatal clinics in North India, respectively. 21,22 The prevalence of BV ranged from 17 to 19% among women from the general population 23,24 and among female sex workers, it was 45%. 25 To our knowledge, only one study in India has reported the prevalence of BV in HIVinfected women and it was 47.7%. 26 Having highlighted the negative effects of BV on the reproductive health and the evidence suggesting BV and HIV/STI control plus a dearth of data on this aspect in India, we are reporting the prevalence and risk factors of BV in HIV-infected women from a cross-sectional study conducted in Maharashtra, India.

Material and methods
The study was reviewed and approved by the ethical review committees of Hirabai Cowasji Jehangir Medical Research Institute (HCJMRI) and Prayas Health Group, Pune, India and that of the International Agency for Research on Cancer of the WHO, Lyon, France. Screening was initiated on 9 September 2010 and completed on 3 November 2011. Our study procedures have been described earlier. 27 The study was conducted at a designated study clinic in Pune, India where consecutive, serologically confirmed HIV-infected women were enrolled in the study.
We contacted physicians treating HIV-infected women as well as various non-governmental organizations (NGOs) working for HIV-infected individuals in high HIV prevalent districts in Maharashtra state, and they were informed about the need for screening HIVinfected women for cervical cancer and the study procedures. Group meetings of HIV-infected women were arranged by the NGOs; a study specific social worker attended these meetings and educated women with the help of a user friendly, pictorial flip chart. Women willing to get enrolled were referred to the study clinic in groups by the NGOs and their travel expenses were reimbursed.
HIV-infected women in the age group of 21-60, having an intact uterus, who were not pregnant and had not received any prior treatment for cervical intraepithelial neoplasia (CIN) or cervical cancer, were eligible for the study. They were explained the study objectives and procedures, and their written informed consent was obtained. They were interviewed for sociodemographic, sexual, reproductive, medical and HIV infection-related characteristics using a structured questionnaire by a female social worker.
Cervical cell samples were collected following the manufacturer's instructions (Qiagen) using a cervical brush provided by the manufacturer after exposing the cervix with a cusco vaginal speculum and the brush was placed in a specimen transport medium (STM) for HPV DNA detection by the Hybrid Capture 2 (HC2) test. A vaginal sample was collected from the lateral vaginal wall and posterior fornix with a sterile cotton swab for smear preparation and rolled on a glass slide for Nugent's scoring.
Women presenting with any reproductive tract infection (RTI) or STI were offered treatment following the WHO guidelines for STIs. 28 Procedures for cervical cancer screening, colposcopy, biopsy, treatment for CIN and histopathology reporting of cervical biopsy specimens have been described in our previous publications of this study. 27 The air dried slides for Nugent's scoring were transported to the laboratory at room temperature where they were stained by Gram stain, and reporting was done following Nugent's criteria by a microbiologist who was blinded to the study data. Each Gramstained smear was evaluated for the following morphotypes under oil immersion (1000Â magnification): large Gram-positive rods (Lactobacillus morphotypes), small Gram-negative to Gram-variable rods (Bacteroides spp. and Gardnerella vaginalis morphotypes) and curved Gram variable rods (Mobiluncus spp. morphotypes). Each morphotype was quantitated from 1 to 4þ with regard to the number of morphotypes per oil immersion field (0, no morphotypes; 1þ, less than 1 morphotype; 2þ, 1-4 morphotypes; 3þ, 5-30 morphotypes; 4þ, 30 or more morphotypes) by a microbiologist. The scoring criteria summed the weighted quantitation (0, 1 to 4þ) of the three morphotypes to yield a score of 0-10 for each person. The criterion for BV was a score of 7 or higher; a score of 4-6 was considered intermediate, and a score of 0-3 was considered normal.
The Gram stain reagents were checked weekly and whenever a new lot of stain was put into use. The reagents were evaluated by staining the recommended bacterial strains: Staphylococcus aureus ATCC 25923 and Escherichia coli ATTC 25922. Gram stain quality control was performed with every run of samples with the ATCC controls. The Gram stain slides were reported by an independent microbiologist who was trained using the training set of slides provided by the Microbicide Trials Network and was blinded to any other data including HPV test report or CIN status. In case of any doubt, the slides were seen by two other expert microbiologists for a consensus in reporting.
The STM containing cervical cells was tested by the HC2 assay for 13 high-risk HPV types (16,18,31,33,35,39,45, 51, 52, 56, 58, 59 and 68) as per the manufacturer's instructions (Digene Corporation, USA) at the Nargis Dutt Memorial Cancer Hospital, Barshi. A positive result was recorded for specimens with a ratio of relative light unit to a positive control of 1 or more, corresponding to 5000 or more viral copies.
Data were entered using Access 2000 software and statistical analysis was carried out using STATA software, version 14

Results
The characteristics of the 912 women enrolled (mean age 34.6, SD: 6.3; range: 21-62) and the distribution of intermediate BV and BV are given in Table 1 Women who reported use of tobacco had 1.6 times increased risk of BV (95% CI 1.0, 2.5, p-value 0.037) as compared to those who did not report use of tobacco. Women who reported more than two pregnancies had 1.8 times increased risk of BV (95% CI 1.3, 2.7, p-value <0.001). Women who were menopausal had significantly increased risk of BV (Crude OR 3.1, 95% CI 1.7, 5.7, p-value <0.001). Women who had history of opportunistic infections in the past had 1.6   Table 3 presents the determinants of BV and intermediate BV in multivariate multinomial logistic regression analysis. Women who had their first sexual intercourse after the age of 18 had significantly lower risk of BV (AOR 0.4, 95% CI 0.3, 0.7, p-value <0.001) as compared to those who had their age at first sex below 18 years of age. Women with more than two pregnancies had 1.6 times increased risk of BV (95% CI 1.0, 2.5, p-value 0.038), and women who had their last menstruation more than 12 months ago had 6.2 times increased risk of BV (95% CI 2.4, 15.6, p-value <0.001). Women who tested positive by HC2 test had 2.3 times increased risk of BV (95% CI 1.4, 3.9, p-value 0.001).
We did not find statistically significant increased risk of intermediate BV among women with any of the  Conversely, we also analysed data to see if women with BV had an increased risk of HPV infection (data not shown). Of the 912 women assessed, 77/191 (40.3%) women with BV were infected with high-risk HPV as assessed by the HC2 test. Women with BV had 2.7 times (95% CI 1.9-3.9, p-value <0.0001) increased risk of HPV infection in the crude multinomial multivariate analysis and 3.1 (95% CI 2.1-4.6, p-value <0.0001) times increased risk of HPV infection in the adjusted multinomial multivariate analysis (data not shown).

Discussion
In our study, about a fifth of the HIV-infected women were diagnosed with BV. When the 'clinical iceberg' concept of adverse health outcomes is applied to BV, the burden of BV is likely to be much more with molecular methods than what is diagnosed by Amsel's or Nugent's criteria. 29 Since BV increases the risk of other STIs, it is important that at least symptomatic women are treated appropriately following syndromic management guidelines 28 in order to reduce their risk of acquiring other STIs and associated co-morbidities.
Factors found independently associated with BV and that increased the risk of BV in HIV-infected women were more than two pregnancies, menopause and HPV infection. Women who reported their age at first sex after the age of 18 had significantly lower risk of BV as compared to those who had first sex prior to 18 years of age. Younger age at sexual intercourse has been previously shown as a risk factor for BV. 30 The risk of BV was significantly higher in menopausal women suggesting that reproductive hormones might be playing a protective role against BV. There have been previous reports of early menopause in HIV-infected women due to HIV-related immunodeficiency. 3 Increasing prevalence of BV as age advances has also been reported previously among general population. 31 Another important finding of our study is about the intermediate BV being present in about one-fourth of the women. It is estimated that about one-third of women with intermediate BV are likely to proceed to BV, 32   About one-third of our study participants were diagnosed with HPV infection 27 and women with HPV infection had 2.3 times increased risk of having BV. A previous meta-analysis has confirmed that women with BV have an increased risk of HPV infection (OR 1.43; 95% CI 1.11-1.84) 36 but this meta-analysis included only one study that was done among HIV-infected women. Because of the cross-sectional nature of our study, we do not know whether women were infected with HPV first or developed BV first. We have previously reported that women with no abortions and women with diagnosis of HIV infection in the past five years had significantly high risk of having multiple HPV infections. We have not found any common factors associated with determinants of BV and determinants of HPV infection. 37 Previous studies and meta-analysis have reported a positive association between BV and CIN. 38 Although we observed increased risk of BV among women with CIN 3 or worse diagnosis in the univariate analysis, it was not significant in the multivariate analysis. Previous meta-analysis has shown a positive association between BV and CIN (OR 1.51 (95% CI 1.24-1.83)). 38 Our study has lack of availability of data on recent sexual behaviour. A previous systematic review and meta-analysis has reported that BV is transmitted to a woman by sexual contact with either a man or woman. 4 Another limitation of our study is availability of Nugent's scoring reports for 912/1153 participants enrolled in the main study as described previously 27 because the Gram stain slides of the first 232/1153 (20.12%) participants enrolled in the beginning of the study could not be found. We have however confirmed that there are no differences in the characteristics of the women who had and who did not have a slide for Gram stain and subsequent Nugent's scoring.
The prevalence of STIs is more in HIV-infected women than women who are not infected with HIV because of the same risk factors, the same route of transmission and impaired immunity among HIVinfected women. 39 Persistence of HPV infection is necessary for the development of cervical cancer. 40 BV can increase the risk of acquisition or reactivation of HPV infection 41 and is also reported to be conducive to the persistence of HPV infection. 42 To conclude, we observed high prevalence of BV as well as intermediate BV among HIV-infected women. Since BV increases the risk of acquiring STIs, persistence of HPV infection and increased risk of gynaecological outcomes, BV control among HIV-infected women should not be neglected.