Bacterial vaginosis

Another condition whose treatment might make a difference to HIV incidence is bacterial vaginosis (BV). This, the most common vaginal condition in women of childbearing age, might double a woman's susceptibility to HIV infection, according to the results of a South African study.¹

BV is a condition in women where the normal balance of bacteria in the vagina is disrupted, resulting in a change from an acidic to an alkaline environment. Although it is sometimes accompanied by symptoms such as discharge, odour, pain, itching, or burning, it is often asymptomatic. The British Association for Sexual Health and HIV (BASHH) states that “Whilst BV is not regarded as a sexually transmitted disease, the prevalence is generally higher amongst sexually active than non-sexually active women”.² No single organism has been isolated as the cause, but it is characterised by an overgrowth of bacteria, such as Gardnerella vaginalis and Mycoplasma hominis, that favour a neutral or alkaline environment, replacing acid-loving bacteria such as Lactobacillus.

BV is very common, affecting between 10 and 30% of women attending sexual health clinics in the UK.²⁴ BV has been implicated in premature delivery in pregnant women and in the development of pelvic inflammatory disease.

Several - but not all - epidemiological and prospective studies have found an association between BV and HIV infection. In addition, in vitro studies suggest that BV has the potential to increase susceptibility to HIV infection.

Investigators at the University of Cape Town, South Africa, sought to ascertain whether there was indeed an association between BV and HIV acquisition in 5110 women enrolled in a cervical-cancer screening trial. BV was assessed during the enrolment visit, by gynaecological examination (so-called Amsel criteria). Blinded microbiologic assessment (Nugent scoring) of Gram-stained slides also assessed BV status.

During the three-year follow-up period, 324 of the women became HIV-positive (annual incidence, 2.1%). Their baseline characteristics were compared with 324 women who stayed HIV-negative.¹

Using Amsel criteria, 20% of the HIV-positive women were found to have had BV at baseline and 16% of the controls: not a significant difference. However, using Nugent scoring, 74% of women who acquired HIV had a BV diagnosis at baseline compared with 62% of controls. Seroconverting to HIV was thus associated with a 20% greater likelihood of having had BV at the start of the study.

The causation was much stronger the other way round, looking ‘forward’ from BV diagnosis rather than ‘backward’ from HIV diagnosis. Women with BV at baseline, as determined by Nugent scoring, were twice as likely (adjusted odds ratio (OR), 2.01 [95% confidence interval (CI), 1.12 to 3.62]) as women with normal vaginal flora to seroconvert to HIV over the subsequent three years. In other words, because BV is so much more common than HIV infection, HIV infection was only mildly retrospectively associated with having had BV at baseline. However, women with BV at baseline were, prospectively, twice as likely to subsequently acquire HIV as women without it.

The investigators point out several limitations to their study, including the fact that BV was assessed only once, whereas HIV seroconversions were identified over three years. During this time, vaginal flora may have changed, and a baseline BV assessment may not accurately reflect presence of BV at the time of seroconversion. Also, the investigators did not assess the presence of ulcerative STIs, including HSV-2.

Treating BV may not be a feasible way of reducing HIV susceptibility, however. In an accompanying editorial, Dr Jane Schwebke, of the University of Alabama at Birmingham, pointed out that "present achievable BV cure rates, combined with high recurrence rates make this solution impracticable."³

Another multicentre, prospective cohort study of 4531 women in Zimbabwe and Uganda also found that women newly diagnosed with HIV were 67% more likely to also have bacterial vaginosis at the same visit as that in which HIV was diagnosed, or the previous visit, than non-seroconverters.⁴ They were also 44% more likely to have yeast infections such as candida.

A 2008 meta-analysis⁵ of the association between BV and HIV infection identified 21 cross-sectional studies. All but one of these found that HIV prevalence was higher in women with bacterial vaginosis. The review also identified four prospective studies of new HIV infections in four African populations. Pooling their data, bacterial vaginosis was associated with an increased risk of HIV acquisition (risk ratio: 1.61; 95% confidence interval 1.21 to 2.13).

Although BV tends to be particularly prevalent in high-risk groups, such as sex workers, the review found that the association between BV and HIV infection was strongest for lower-risk women.