Letter to the Editor, BJOG Exchange

Re: The vaginal metabolome and microbiota of cervical HPV-positive and HPV-negative women: a cross-sectional analysis

Dear Sir,

We thank Zhang et al.¹ for their interest in our recent research article². The resident vaginal microbiota has a considerable effect on the overall vaginal metabolome, as was first shown in 2013 by Yeoman et al. who demonstrated that the metabolome was associated with the low-Lactobacillus state of bacterial vaginosis (BV)³, a finding also supported by Srinivasan et al.⁴ In two subsequent papers, our group has demonstrated that the microbiota contributes between 23–31% of the total variation in the vaginal metabolome^2,5. While it is impossible to eliminate the effect of the vaginal microbiota in observational studies of the metabolome, we controlled in multivariable models for bacterial community state types (CSTs), as well as smoking⁵ in our analyses. We also stratified by CST, another approach to holding the bacterial composition constant, which may also partially adjust for unmeasured confounding factors that influence the metabolome through microbiota. Under these statistical model conditions, we still observed metabolomic differences between HPV+ and HPV- women in each CST stratum, indicating a direct metabolic signal associated with HPV.

Certainly, it is plausible that other unmeasured variables might also affect the metabolome irrespective of microbiota and consequently our results; this was the motivation for our HPV study. Herein, we were limited by our relatively small sample size (39 women). The national prevalence of Trichomonas vaginalis, Chlamydia trachomatis and Neisseria gonorrhoeae range between 1–7%;^6–8 therefore, in the present study of 39 women, we expect that few participants would have one of these sexually transmitted infections (STIs); thus, it is unlikely that further adjustment or exclusion for a STI would dramatically affect our results and doing so would add to statistical issues of data sparsity. Further, study participants reported relatively few sexual risk-taking behaviors, suggesting low STI risk. Most participants (>90%) reported either none, or one, sexual partner within the two months prior to the study and number of partners was not significantly associated with HPV detection status (See Supplemental Table S1). Assessing Candida albicans in the context of the vaginal metabolome is of interest, unfortunately, Candida data are not available for this study.

The vaginal metabolome has the potential to provide novel insights into female sexual and reproductive health. However, as in all research, the application and generalizability of results are limited by study design, sample size, and clinical feasibility. Utilizing multi-omics-based approaches in the context of well-designed epidemiologic studies provides additional tools to enable the identification of potential biomarkers associated with infections and diseases and may lead to clinically pertinent innovative therapies and preventative measures.