Reassessment of BV causation

Sir: The reassessment of the causation of bacterial vaginosis (BV) by Josey and Schwebke ¹ does the ‘polymicrobial hypothesis’ a disservice. We agree that the recent description ² of the biofilm consisting predominantly of Gardnerella vaginalis, with a significant proportion of Atopobium vaginae in some cases, provides further evidence for G. vaginalis being very important and possibly key in the pathogenesis of BV. However, the existence of the biofilm does not prove that it is sufficient to cause BV without other organisms or prove that BV is sexually transmitted. Finding BV in the absence of G. vaginalis is rare and may merely reflect an insensitivity of the detection method used. Indeed, in a recent study using molecular techniques, Fredricks et al. ³ detected G. vaginalis in 96% of women with BV and in 70% of those without BV, together with three novel bacteria that were more specifically associated with this condition. Josey and Schwebke ¹ also take issue with the accuracy of the diagnosis of BV made by using Amsel's criteria or Gram stain reading. However, the Gram stain scoring systems, while imperfect, do recognize the spectrum from plentiful lactobacilli (healthy vagina) through to ‘severe BV’ without visible lactobacilli. Longitudinal studies have demonstrated that BV is most common around the time of menstruation ⁴ and it is likely that in any population, the cumulative prevalence of BV over a time period will be much greater than the point prevalence.^4,5

Vaginal inoculation experiments were performed by Gardner and Dukes ⁶ in an attempt to fulfil Koch's postulates for the then Haemophilus vaginalis (later termed as G. vaginalis). Inoculation of a pure culture of H. vaginalis produced apparent colonization in three of 13 women, only one of whom developed symptoms. A further 15 women were inoculated directly with material obtained from infected cases. Eleven (73%) of these women developed clinical signs and the organism was recovered from all of them, eight developing the syndrome within one week of inoculation. This does not support G. vaginalis as the single aetiological agent of BV, but does support the case that an inoculum of many bacterial species, with organisms in large numbers, may overwhelm the vaginal ecosystem leading to the development of BV. As such, this mechanism, which may operate in women who have sex with women, and in women whose male partner has multiple concurrent partners, explains some of the associations of BV with sexual risk-taking behaviour. The concept that BV might be sexually associated rather than sexually transmitted is well-established.

We also take issue with the logic of discounting studies of virgin and non-virgin women based on the belief that self-completed questionnaires or face-to-face interviews are unreliable. Even if there is some ascertainment bias, a trend in favour of an association of BV with sexual activity might be expected. The results of a recent study of adolescent girls in rural Ecuador, ⁷ in which data on risk factors for BV were obtained by questionnaire, replicated the findings of Bump and Buesching ⁸ with an almost identical prevalence of BV (31% and 33%) in the virgin and non-virgin girls, respectively. If questionnaires are regarded as unreliable, logically the same standard should be applied by the authors to the studies of sexual behaviour on which they base their hypothesis that BV has the characteristics of a sexually transmitted disease.

Trichomonas vaginalis frequently coexists with intermediate flora or BV. ⁹ It is possible that T. vaginalis might induce BV or that BV might favour the growth of T. vaginalis and increase the likelihood of the woman developing symptoms. Either way, it has no bearing on whether or not BV is a sexually transmitted disease.

The hypothesis that BV is an imbalance in the endogenous vaginal bacteria in which the growth conditions, particularly the elevation of vaginal pH, favour the rapid multiplication of anaerobic organisms, appears adequate to explain the reported epidemiological associations of BV with sexual and other behaviours. ¹⁰ The presence of infections causing cervicitis and purulent discharge, such as chlamydial ones, are also likely to alter the vaginal environment. Unprotected intercourse with exposure to relatively alkaline pH has an effect on the vaginal environment, and concordance within lesbian couples is likely to be replicating the early inoculation experiments of Gardner and Dukes, ⁶ without calling a sexually transmitted disease into play. Searching for a single aetiology or external infectious cause for BV is likely to prove fruitless. Particularly noteworthy is the existence of BV in adolescent girls, including virgins, ⁷ a situation where it is unrealistic to think in terms of a sexually transmitted disease. More appropriate is the idea that some noninfectious internal or external trigger initiates the floral imbalance. The effect on the vaginal flora of the menstrual cycle and of pregnancy, as well as the effect of sex hormones on the vaginal microflora of animals, suggests that this trigger could be hormonal, as suggested previously. ¹¹ The frequent failure to treat BV effectively, ¹² probably reflects the failure to deal with this underlying reason for the microbial change occurring.