Cigarette smoking is associated with an altered vaginal tract metabolomic profile

Abstract

Cigarette smoking has been associated with both the diagnosis of bacterial vaginosis (BV) and a vaginal microbiota lacking protective Lactobacillus spp. As the mechanism linking smoking with vaginal microbiota and BV is unclear, we sought to compare the vaginal metabolomes of smokers and non-smokers (17 smokers/19 non-smokers). Metabolomic profiles were determined by gas and liquid chromatography mass spectrometry in a cross-sectional study. Analysis of the 16S rRNA gene populations revealed samples clustered into three community state types (CSTs) ---- CST-I (L. crispatus-dominated), CST-III (L. iners-dominated) or CST-IV (low-Lactobacillus). We identified 607 metabolites, including 12 that differed significantly (q-value < 0.05) between smokers and non-smokers. Nicotine, and the breakdown metabolites cotinine and hydroxycotinine were substantially higher in smokers, as expected. Among women categorized to CST-IV, biogenic amines, including agmatine, cadaverine, putrescine, tryptamine and tyramine were substantially higher in smokers, while dipeptides were lower in smokers. These biogenic amines are known to affect the virulence of infective pathogens and contribute to vaginal malodor. Our data suggest that cigarette smoking is associated with differences in important vaginal metabolites, and women who smoke, and particularly women who are also depauperate for Lactobacillus spp., may have increased susceptibilities to urogenital infections and increased malodor.

Figure 1

Compounds differing between smokers and non-smokers with and without adjustment for CST. Volcano plots display –log₁₀ (p-value) and the median difference in concentration between smokers and non-smokers when unadjusted (A) and adjusted for community state type (B). Quantile regression was conducted on centered and scaled metabolite concentrations. Significance testing was conducted with Wilcoxon rank sum test and corrected for multiple comparisons. Metabolites that differed significantly where q-value < = 0.05 between smokers and non-smokers are shown above the line in each plot.

Figure 2

Vaginal metabolites that differ between smokers and non-smokers. Boxplots display metabolites identified as significantly (q-value = <0.05) different in the vagina of smokers and non-smokers when unadjusted and adjusted for the impact of bacterial community state type (CST). Samples categorized as CST-I (L. crispatus-dominated) and CST-III (L. iners-dominated) were grouped and compared with CST-IV (low-Lactobacillus spp.). Quantile regression was conducted on centered and scaled metabolite concentrations. Significance testing was conducted with Wilcoxon rank sum test and corrected for multiple comparisons.

Figure 3

Participant predictor variables fitted to vaginal metabolites. Distance-based linear modelling (DISTLM) was conducted on log-transformed Euclidean distance matrix of vaginal metabolites. All variables were included in the final model that when applied to the data cloud of vaginal metabolites (Table 1). Distance-based redundancy analysis (dbRDA) plot (A) displays the metabolite composition fitted to the four variables labeled with the strongest variable, CST. Variable vectors indicating strength and direction are identified (B).

Figure 4

Metabolites differ between CST. Heatmap displays selection of metabolites with a high fold change identified as significantly (q-value = <0.05) different in the vagina of bacterial community state type (CST) when unadjusted and adjusted for the impact of smoking status. Low-Lactobacillus CST-IV and Lactobacillus-dominated CST-I and CST-III are grouped. Quantile regression was conducted on centered and scaled metabolite concentrations. Significance testing was conducted with Wilcoxon rank sum test and corrected for multiple comparisons.