A specific bacterial DNA signature in the vagina of Australian women in midpregnancy predicts high risk of spontaneous preterm birth (the Predict1000 study)

Background: Intrauterine infection accounts for a quarter of the cases of spontaneous preterm birth; however, at present, it is not possible to efficiently identify pregnant women at risk to deliver preventative treatments. Objective: This study aimed to establish a vaginal microbial DNA test for Australian women in midpregnancy that will identify those at increased risk of spontaneous preterm birth. Study Design: A total of 1000 women with singleton pregnancies were recruited in Perth, Australia. Midvaginal swabs were collected between 12 and 23 weeks’ gestation. DNA was extracted for the detection of 23 risk-related microbial DNA targets by quantitative polymerase chain reaction. Obstetrical history, pregnancy outcome, and demographics were recorded. Results: After excluding 64 women owing to losses to follow-up and insufficient sample for microbial analyses, the final cohort consisted of 936 women of predominantly white race (74.3%). The overall preterm birth rate was 12.6% (118 births); the spontaneous preterm birth rate at <37 weeks’ gestation was 6.2% (2.9% at ≤34 weeks’ gestation), whereas the preterm premature rupture of the membranes rate was 4.2%. No single individual microbial target predicted increased spontaneous preterm birth risk. Conversely, women who subsequently delivered at term had higher amounts of Lactobacillus crispatus, Lactobacillus gasseri, or Lactobacillus jensenii DNA in their vaginal swabs (13.8% spontaneous preterm birth vs 31.2% term; P=.005). In the remaining women, a specific microbial DNA signature was identified that was strongly predictive of spontaneous preterm birth risk, consisting of DNA from Gardnerella vaginalis (clade 4), Lactobacillus iners, and Ureaplasma parvum (serovars 3 and 6). Risk prediction was improved if Fusobacterium nucleatum detection was included in the test algorithm. The final algorithm, which we called the Gardnerella Lactobacillus Ureaplasma (GLU) test, was able to detect women at risk of spontaneous preterm birth at <37 and ≤34 weeks’ gestation, with sensitivities of 37.9% and 44.4%, respectively, and likelihood ratios (plus or minus) of 2.22 per 0.75 and 2.52 per 0.67, respectively. Preterm premature rupture of the membranes was more than twice as common in GLU-positive women. Adjusting for maternal demographics, ethnicity, and clinical history did not improve prediction. Only a history of spontaneous preterm birth was more effective at predicting spontaneous preterm birth than a GLU-positive result (odds ratio, 3.6). Conclusion: We have identified a vaginal bacterial DNA signature that identifies women with a singleton pregnancy who are at increased risk of spontaneous preterm birth and may benefit from targeted antimicrobial therapy.