Direct on-swab metabolic profiling of vaginal microbiome host interactions during pregnancy and preterm birth

Abstract

The pregnancy vaginal microbiome contributes to risk of preterm birth, the primary cause of death in children under 5 years of age. Here we describe direct on-swab metabolic profiling by Desorption Electrospray Ionization Mass Spectrometry (DESI-MS) for sample preparation-free characterisation of the cervicovaginal metabolome in two independent pregnancy cohorts (VMET, n = 160; 455 swabs; VMET II, n = 205; 573 swabs). By integrating metataxonomics and immune profiling data from matched samples, we show that specific metabolome signatures can be used to robustly predict simultaneously both the composition of the vaginal microbiome and host inflammatory status. In these patients, vaginal microbiota instability and innate immune activation, as predicted using DESI-MS, associated with preterm birth, including in women receiving cervical cerclage for preterm birth prevention. These findings highlight direct on-swab metabolic profiling by DESI-MS as an innovative approach for preterm birth risk stratification through rapid assessment of vaginal microbiota-host dynamics.

Fig. 1. Direct on-swab desorption electrospray ionization mass spectrometry (DESI-MS) metabolic profiling of cervicovaginal fluid enables robust prediction of vaginal microbiome compositions.

A Study design and longitudinal multi-omic sampling and analysis workflow of cervicovaginal swab samples collected from two independent pregnancy cohorts (VMET: n = 165, 455 swabs; VMET2: n = 205, 573 swabs). Data from each cohort were analysed independently, and features selected only if their Benjamini–Hochberg q-value was smaller than 0.05 in both datasets, after matching the hits from each analysis by their m/z values. B Heatmaps representing relative concentrations of DESI-MS (negative mode)-derived metabolic features (n = 88) significantly differing between Lactobacillus spp. dominated (L-dominated, green) and Lactobacillus spp. depleted (L-depleted, red) states in both independent patient cohorts (see Table S2). C Boxplots of representative discriminatory metabolic features with corresponding Benjamini–Hochberg q-values identified including thiomalic acid, leucyl-serine, docosanoic acid (C22:0), lignoceric acid (C24:0) with calculated z-score measured in the two patient cohorts VMET2 (left, n = 203, 539 swabs) and VMET (right, n = 160, 428 swabs). The lower and upper bounds of the box represent the 25th and 75th percentile values, respectively, and the interior horizontal line the median value. Whiskers are drawn from the corresponding box boundary to the most extreme data point located within the box bound ± 1.5 × IQR (interquartile range). m/z mass-to-charge ratio, CST community state type.

Fig. 2. Comparison of DESI-MS classification performance between different vaginal microbiome compositions.

A ROC-curve analysis showing performance of direct swab analysis by DESI-MS operating in both negative and positive ion polarity modes to predict Lactobacillus-depleted vaginal microbiome compositions in both the VMET (blue; AUC: 94.1, sensitivity: 62.0, specificity: 97.8) and VMET2 (red; AUC: 90.6, sensitivity: 54.5, specificity: 96.4) patient cohorts. Discrimination between the major vaginal community state types (CST) could also be readily achieved using DESI-MS across both patient cohorts, including B CST I vs IV, C CST III vs IV and D CST I vs III. Overall, predictive performance of DESI-MS was comparable to that of models constructed from LC-MS assays (Supplementary Table 3 and Supplementary Fig. 4).

Fig. 3. Detection of in vivo discriminatory metabolite features in bacterial biomasses by DESI-MS.

Discriminatory metabolites identified in the in vivo DESI-MS analyses were detected by DESI-MS in bacterial isolates (n = 25) of species recognized as being predominant members of major vaginal community state types (CST). A total of 27 metabolites were detected at levels lower or higher than that observed in media background controls, where the mean log₂ fold change (FC) was estimated as the ratio of the mean intensity in the bacterial biomass samples to the mean intensity in the background culture media samples.

Fig. 4. Assessment of host response at the mucosal interface using direct on-swab DESI-MS profiling.

A Cross-validated R² value for all 22 corresponding measured immune mediator concentrations. B Heatmap of top 23 significantly correlated metabolite features with top 10 immune mediators (IL-1β, IL-8, MBL, C5, C3b/C3bi, IgE, IgG2, IgG3, IgG4, IgM). t-ratio ranges from +10 (red) to −7.5 (blue). C Association between predicted log-transformed value of immune marker by DESI-MS and measured log-transformed values by multiplexed immune-assay for IL-1β (CV R² = 0.51), IL-8 (CV R² = 0.37), C3b/iC3b (CV R² = 0.33), IgG3 (R² = 0.31), IgG2 (CV R² = 0.27), MBL (CV R² = 0.26). A linear regression line was fitted to the log-transformed values and their corresponding prediction. A box plot of predicted immune marker levels for LDEP (red) and LDOM (green) samples is also presented (n = 136 pregnancies, 369 swabs). The lower, interior horizontal line, and upper bounds of the box represent the 25th, median and 75th percentile values, respectively. Whiskers are drawn from the corresponding box boundary to the most extreme data point located within the box bound ± 1.5 × IQR (interquartile range). P values are reported for a two-tailed Welch t-test for the difference in mean predicted immune markers between LDEP and LDOM.

Fig. 5. Vaginal microbiome instability and immune activation associates with preterm birth risk and poor outcomes following cervical cerclage.

A Increased risk of PTB (red) was associated with vaginal microbiome instability (defined by shifts between Lactobacillus spp.-dominated (LDOM) and Lactobacillus spp.-depleted (LDEPL) compositions) measured by 16S rRNA-based metataxonomics (OR 1.97, 95% CI 1.03–3.66, p = 0.04, two-sided mid-p exact test) or predicted using DESI-MS profiles (OR 1.47, 95% CI: 0.75–2.78, p = 0.25, two-sided mid-p exact test). B LDEPL vaginal composition was associated with increased IL-1β levels compared to LDOM; however, highest levels were observed in LDEPL women subsequently having preterm delivery. This relationship was also observed when IL-1β levels and vaginal microbiota composition were predicted using direct swab profiling by DESI-MS (n = 103 pregnancies, 103 swabs). C A relationship between LDEPL, increased MBL and subsequent preterm birth was also detected by DESI-MS profiling (n = 103 pregnancies, 103 swabs). D Elevated MBL and E elevated IL-1β levels were observed in response to cervical cerclage performed with braided cerclage material, but not monofilament material (n = 34 pregnancies, 68 swabs). F Preterm birth in women treated with cervical cerclage using braided cerclage material was associated with higher IL-1β levels compared to term birth outcomes (n = 13 pregnancies, 13 swabs), whereas no relationship between IL-1β levels measured or DESI-MS-predicted were observed with pregnancy outcome following cervical cerclage using monofilament material (n = 21 pregnancies, 21 swabs). All box and whisker plots are drawn with the lower, horizontal interior line, and upper bounds of the box representing the 25th percentile, median and 75th percentile values, and whiskers extending from the lower or upper box bonds to the position of the most extreme data point within ± 1.5 × IQR (interquartile range).