Study population

This prospective cohort study recruited pregnant and nonpregnant women at the various clinics serving Women & Infants Hospital in Providence, RI, from 2007 through 2010. Inclusion criteria included: (1) pregnancy documented by urine human chorionic gonadotropin, serum human chorionic gonadotropin, or ultrasound with low-risk pregnancy status and a gestational age <14 weeks; (2) healthy nonpregnant women age 18-35 years, not planning pregnancy within the next year; and (3) willingness to avoid the use of intravaginal products and willingness to use condoms during sexual intercourse 48 hours prior to each examination. Exclusion criteria included acute systemic illness; chronic illness (hypertension, preexisting diabetes mellitus, autoimmune disease, history of thromboembolic disease); use of systemic steroids in past 3 months; immunization in past 1 month; active alcohol, drug, or tobacco use; immunocompromised state; known active infection; symptomatic vaginal discharge; current urinary tract infection; antibiotic use within 1 month of enrollment; prior preterm birth (<36 weeks’); current or planned cerclage; history of preeclampsia <36 weeks’ gestation; and multiple gestation. The Women & Infants Hospital Institutional Review Board approved this study. Written informed consent was obtained from each participant before enrollment. Women considered low risk after their first prenatal appointments were recruited by study staff to participate. Nonpregnant women were recruited from physician offices as well as through advertisements on local college campuses.

Study visit procedures

A total of 4 study visits were planned, 1 enrollment visit and 3 follow-up visits. In pregnant patients, visits occurred at <14, 14-28, and >28 weeks’ gestation, and again at the postpartum visit, approximately 4-6 weeks after delivery. In nonpregnant patients, study appointments were scheduled at approximately 12-week intervals to mimic the time duration between visits for the pregnant participants, excluding menses. In all study participants, each visit was standardized to include a brief questionnaire about recent exposures such as use of antibiotics or unprotected intercourse. A pelvic examination with collection of cervical and vaginal swabs, wet mount, vaginal pH, and a single blood sample was conducted.

A swab was used for collection of cervical and vaginal fluid as previously described. All cervical swabs were placed in 100 μL of phosphate-buffered saline and stored at −70°C until the assays were performed. Concentrations of IL-1β, IL-4, IL-6, IL-10, interferon (IFN)-γ, tumor necrosis factor-α, granulocyte macrophage colony-stimulating factor, and macrophage inflammatory protein-1α were measured using the Luminex (Luminex Corp, Austin, TX) multiplex bead assay via a previously validated method. Serum concentrations of these mediators were consistently too low to measure and were discontinued during the course of the study.

Vaginal samples were used for the measurement of the antimicrobial peptide secretory leukocyte protease inhibitor (SLPI). Specimens were collected from the posterior fornix using a swab and stored at −70°C. The concentration of SLPI was determined using commercially available Quantikine human SLPI immunoassay kits (R&D Systems Inc, Minneapolis, MN) per manufacturer’s instructions. C-reactive protein (CRP) was measured from serum samples using a highly sensitive assay. Serum samples collected from each participant during each visit were frozen and stored at −70°C. The CRP assay employed a simple sandwich enzyme-linked immunosorbent assay adapted from that of Erhardt et al.

Specimens for vaginal cultures were collected via swab placed in Port-A-Cul transport gel (Becton-Dickinson, Sparks, MD) and transported via overnight shipping to the reference laboratory in Pittsburgh, PA, within 24 hours of collection for the following organisms: Mycoplasma hominis, Ureaplasma species, G vaginalis, Lactobacillus species, Staphylococcus aureus, Escherichia coli, Enterococcus species, Candida species, and anaerobic gram-negative rods, both pigmented and nonpigmented. The swabs were removed from the transport gel and inoculated onto Columbia agar supplemented with 5% sheep blood, 2 plates of human bilayer Tween agar, Rogosa agar, A-8 agar, Ureaplasma broth, and 1 plate of prereduced laked blood kanamycin agar. The Columbia agar, 1 set of human bilayer Tween agar plates, the A-8 agar plate, and the Ureaplasma broth were incubated at 36°C in 5-7% carbon dioxide for a minimum of 48 hours. The remaining plates were incubated within an anaerobic glove box at 36°C for a minimum of 5 days. Biochemical tests were used to identify the Enterococcus species, Escherichia coli, and Group B streptococcus. Lactobacillus species were tested for production of hydrogen peroxide using a qualitative assay. Anaerobic gram-negative rods were identified by their characteristic colony morphology on selective media and Gram stain morphology.

Statistical analysis

Statistical power was calculated based on previously reported SLPI concentrations in pregnancy. We performed a sample size calculation using a 2-group repeated measures design. The present study with 47 pregnant and 16 nonpregnant patients was able to detect an 85% difference in SLPI concentrations at the enrollment visit with 80% power. The study lacked sufficient statistical power to assess differences in adverse pregnancy outcomes. Cytokines, chemokines, and SLPI concentrations were not normally distributed. Therefore, comparisons were made between study arms using Wilcoxon rank sum test. Adjustment for variables such as white race and overweight (body mass index [BMI] ≥ 25) was performed by the van Elteren test. Organism growth was compared between groups by the χ ² test or the Cochran-Mantel-Haenszel test (adjusted analysis). Statistical analysis was performed using SAS 9.2 (SAS Institute, Cary, NC). We conducted 2-sided hypothesis tests with a P value < .05 considered statistically significant. To account for multiple testing, we adjusted P values by the Benjamini-Hochberg method with a false discovery rate of 0.05. This method controls the proportion of type I errors among all significant hypothesis tests at 5%. The sample size did not afford adequate power to make meaningful comparisons between individual microflora and mediators.