Preterm birth complicates approximately 11% of all pregnancies in the United States and remains a major cause of perinatal morbidity and mortality.1 Although the pathophysiologic mechanisms that lead to preterm delivery have not been completely elucidated, a substantial body of evidence has emerged linking both lower and upper genital tract infection and/or inflammation with subsequent spontaneous preterm birth.1–7 Bacterial colonization and/or infection of the lower and upper genital tract may result in local inflammation prompting a cascade of events that ultimately leads to spontaneous labor and delivery.5–7 The source of this intrauterine infection, however, has not been completely defined. The leading hypothesis linking lower and upper genital tract infection in this process is that the presence of altered vaginal flora, such as bacterial vaginosis, and inflammation facilitates a permissive condition to allow ascension of pathogenic bacteria from the vagina and cervix to reach the upper genital tract. Indeed, a large number of investigations have confirmed an association between bacterial vaginosis and an array of cervicovaginal inflammatory markers, with subsequent development of spontaneous preterm birth (Ramsey PS for the NICHD MFMU Network, Bethesda, MD. Cervical markers at 22–24 weeks’ gestation improve the prediction of subsequent spontaneous preterm birth [abstract]. J Soc Gynecol Investig 2001; 8:64A).8–12
Recent investigations have established an association between vaginal polymorphonuclear cells and increased risk for adverse reproductive outcomes in women, including asymptomatic upper genital tract histologic endometritis, intra-amniotic infection, histologic chorioamnionitis, and preterm birth.13–19 Additional evidence has established a link between the presence of vaginal polymorphonuclear cells (PMNs) with elevated levels of inflammatory and proinflammatory cytokines.19–21 The source of these cervicovaginal cytokines has not been well defined to date. Because PMNs represent one of the primary cellular host immune responses, it is compelling to hypothesize that these cells are a major source of this inflammatory reaction.
Given the suggested association between vaginal PMNs and preterm birth, refinement of strategies to more precisely quantitate these immune cells in vaginal secretions is needed. Previous investigations have used widely divergent methods, which make comparisons between studies difficult.13–21 Assessment of vaginal PMN counts on Gram-stain preparations is also subject to field selection bias, significant field variations, and preparation artifacts that may impact validity and reproducibility. Because epithelial cells are typically present on most fields of vaginal Gram-stain preparations and the density of these cells is related to thickness of the preparation, quantitation of epithelial cells (EPIs) along with PMNs may provide an intrafield marker of cellular density. Thus, the use of a polymorphonuclear cell to epithelial cell count (PMN/EPI) ratio may improve reliability of PNM assessment on Gram-stain preparations. In light of these issues, we sought to evaluate a standardized technique to quantify PNM counts and a PMN/EPI ratio on vaginal Gram-stain preparations. We also sought to compare the efficacy of these markers to predict subsequent spontaneous preterm birth in asymptomatic women.
MATERIALS AND METHODS
With approval from the University of Alabama Institutional Review Board, we conducted a nested case-control study to assess the potential association between vaginal PMNs and risk for subsequent spontaneous preterm birth. The study population from which women for this investigation were derived was a prospective longitudinal cohort study (March 1997 through March 2001, N = 3,160) designed to evaluate risk factors for spontaneous preterm delivery (National Institutes of Health Perinatal Emphasis Research Center grant 5P01HD33927-5).22 Asymptomatic women with intact membranes were prospectively enrolled between 20 and 25 weeks of gestation. Gestational age was assessed from the last menstrual period (LMP), which was corroborated by second-trimester ultrasonography. If dating by LMP was more than 14 days divergent from the second-trimester ultrasound dating, the ultrasound dating was used. If the date of the LMP was unavailable or uncertain, the first ultrasound assessment was used to establish gestational age. Extensive demographic and pregnancy risk assessments were made on all women at the initial study visit.
All women underwent an initial pelvic examination with an unlubricated sterile speculum, at which time a Dacron swab of the upper vagina and vaginal sidewalls was obtained for vaginal smear preparation. Vaginal pH was assessed using a pH indicator strip. Vaginal smears were Gram stained and evaluated for the presence of bacterial vaginosis using the methodology described by Nugent and colleagues.23 Bacterial vaginosis was defined using a Gram-stain score of 7–10. All participants were followed through delivery to collect pregnancy outcome data.
This investigation was conducted as an ancillary evaluation to the above study. Of the 3,160 women in the cohort, 3,107 had complete outcome data available. There were 144 cases of preterm birth (≤ 35 weeks of gestation) in the cohort, with 95 of these cases spontaneously occurring secondary to either preterm labor or preterm premature rupture of membranes. Of these 95 cases of spontaneous preterm birth at 35 weeks or less, vaginal Gram-stain preparations were available on 83 women for use in this investigation. For controls, 120 randomly selected women who had a term delivery (≥ 37 weeks of gestation) were identified for use in this investigation. Only 108 of these identified controls had vaginal Gram-stain preparations available for this investigation. Random selection of controls was accomplished by assigning a random number from a uniform distribution to each potential cohort control, sequentially sorting renumbered specimens, and sequentially selecting the controls for this ancillary investigation. Vaginal smear preparations for the selected women were identified and reanalyzed. Polymorphonuclear and vaginal epithelial cells were quantitated by using a standardized protocol as outlined here (see box). Vaginal PNM/EPI ratios for each evaluated slide field were calculated to control for intraslide variation in cellular density. All slides were evaluated by laboratory personnel who were masked to the initial slide interpretation and pregnancy outcome. Statistical analysis included the Wilcoxon rank sum and χ2 tests, Pearson and Spearman correlation, and logistic regression analyses, where appropriate.
Standardization Protocol for Polymorphonuclear Cell Assessment on Gram-Stained Vaginal Smears
1. Slides were prepared from a vaginal swab obtained from the upper vaginal and side walls.
2. Air dried smears were stained using established Gram-stain methodology.
3. Five randomly selected, adequate nonadjacent fields were evaluated under oil immersion light microscopy (magnification, ×1,000). (Adequate fields were those without cervical mucous and a single cellular layer.)
4. Polymorphonuclear cells (PNM), identified by discrete lobar nuclear pattern, were counted and recorded for each field evaluated. (Polymorphonuclear cell counts >100 were coded as an oil immersion field count of 100.)
5. Epithelial cells (EPI), identified by cellular size and homogeneous nuclear material, were counted and recorded for each field evaluated.
6. Smudge cells (cells with ill-defined cellular and/or nuclear borders) were not included in the counts.
7. Vaginal polymorphonuclear cell to epithelial cell count ratios for each evaluated slide field were calculated to control for intraslide variation in cellular density. (To allow for calculation of the PMN/EPI ratio, EPI field counts that were 0 were reassigned a value of 1.)
As expected, mean delivery gestational age for the case group was significantly earlier than for the control group (Table 1). Maternal age, gestational age at initial study visit, race, marital status, maternal education level, and parity were similar between the groups (Table 1). The incidence of both previous preterm birth at less than 37 weeks of gestation and previous spontaneous preterm birth at less than 37 weeks of gestation was significantly higher among the cases than among the controls. Mean PMN counts were higher in the cases (13 ± 20 cells per oil immersions field) than in the controls (10 ± 14 cells/oil immersion field, P = .17), but this difference was not significant. The mean PMN/EPI ratio was significantly higher among the cases (3.4 ± 6.0) than among the controls (1.8 ± 2.4, P = .01).
To evaluate the potential use of vaginal PMN counts for the identification of women at risk for spontaneous preterm birth, we dichotomized PMN counts and PMN/EPI ratios by the 95th percentile cutoff to evaluate the association with spontaneous preterm birth (PMN cutoff based on control values, 32.4 cells/oil immersion field; PMN/EPI ratio cutoff based on control values, 7.0/oil immersion field). Although PMN counts greater than the 95th percentile were not significantly associated with subsequent spontaneous preterm birth at less than 35 weeks of gestation (odds ratio [OR] 2.1, 95% confidence interval [CI] 0.7–6.1, P = .18), a PMN/EPI ratio greater than the 95th percentile was significantly associated with spontaneous preterm birth (OR 3.8, 95% CI 1.3–11.2, P = .009). In a logistic regression model controlling for the presence of bacterial vaginosis defined as a Nugent score 7–10, a PMN/EPI ratio greater than the 95th percentile remained significantly associated with spontaneous preterm birth (OR 3.8, 95% CI 1.3–11.2, P = .01), and PMN counts greater than the 95th percentile were not significantly associated with subsequent spontaneous preterm birth at less than 35 weeks of gestation (OR 2.1 95% CI 0.7–6.1, P = .18).
An elevated PMN count (> 95th percentile) had low sensitivity (60.0%) and specificity (58.0%) for the prediction of women at risk for spontaneous preterm birth at less than 35 weeks of gestation. An elevated PMN/EPI ratio (> 95th percentile) had low sensitivity (15.7%) for the prediction of women at risk for spontaneous preterm birth at less than 35 weeks of gestation but had high specificity (95.4%). The intraobserver coefficient of variation for PMN counts was 65.3%. Use of the PMN/EPI ratio, however, resulted in an improvement in the intraobserver coefficient of variation (38.6%).
Because we speculated that the presence of bacterial vaginosis might be related to the presence of elevated vaginal PMN counts and PMN/EPI ratios, we calculated Pearson correlation coefficients with bacterial vaginosis as well as several of the key components of the Nugent score (Table 2). The only significant association that was noted was a modest correlation between the Gardnerella sp score and the PMN/EPI ratio.
A substantial body of evidence has established a strong link between upper and lower genital tract infections and/or inflammations and early spontaneous preterm deliveries.2–9 Polymorphonuclear cells are the primary cellular mediator of the innate host immune response to infectious organisms. These cells likely play a pivotal primal role in the inflammatory process, which culminates in a cascade of events that results in spontaneous preterm birth. Several recent investigations have reported a significant association between elevated vaginal PMN counts, preterm labor, spontaneous preterm birth.16–20 These investigations, however, have used widely divergent methodologies, which make comparisons between studies difficult (Table 3). 13,15–21,24 Variations in technique include disparate specimen collection (vaginal lavage or washing versus smear), staining used (Gram stain versus other), magnification used (range × 400–1,000), and variable cutoffs used (range ≥ 1 to ≥ 13 per field examined; Table 3).13,15–21,24 In addition, assessment of vaginal PMN counts on Gram-stain preparations is subject to field selection bias, significant field variations, and preparation artifacts that may impact validity and reproducibility. Given the potential importance of assessment of vaginal PMN counts as a marker of risk for subsequent preterm birth, development of a standardized technique for assessment is needed. In this investigation, we describe a standardized methodology for the assessment of PMN counts on vaginal smear Gram-stain preparations. Elevated PMN counts were associated with a 2-fold increased risk of spontaneous preterm birth, but this association was not statistically significant. Alternatively, the PNM/EPI count ratio, which provides internal standardization of variation in slide cellular density, was significantly associated with subsequent spontaneous preterm birth at less than 35 weeks of gestation. In addition, use of the PMN/EPI ratio resulted in a markedly lower intraobserver coefficient of variation than did use of PMN counts alone.
Although we did not note an association between crude PMN counts and preterm birth, other investigators have. Simhan and colleagues16 recently reported a significant association between elevated PMN cell counts (> 90th percentile; ≥ 5 cells per oil immersion field) on vaginal smear Gram-stain preparations and spontaneous preterm birth. Yamada and colleagues,17 using vaginal lavage for PMN assessment, have also reported a significant association between elevated PMN counts and preterm labor. The disparate findings between our study and those of others may be in part secondary to variable nonstandardized methods used.
Although vaginal lavage with appropriate cellular staining (ie, Wright or Giemsa stains) would be the ideal methodology for vaginal PMN assessment, this technique is labor intensive and requires additional specimen preparation. Use of vaginal smear preparations has the appeal of an established procedure used clinically, with readily available supplies for specimen collection. Because Gram-stained vaginal smear preparations are commonly used for the assessment of bacterial vaginosis, especially in the research setting, there is obvious interest in using such preparations to also evaluate PMN counts. While the Gram-stain technique is the gold standard for prokaryotic cellular assessment, it is not the ideal method for eukaryotic cell preparation or staining. Heat fixation associated with the Gram preparation introduces cellular artifacts, including loss of cellular borders and disruption of nuclear material, thus impairing the ability to adequately evaluate cellular differentials. Additional confounding problems with vaginal smear preparations include the presence of cervical mucus and variability of cellular density. Despite these limitations, Gram-stained vaginal preparations are readily available for concurrent evaluation of PMN counts at the time of evaluation for bacterial vaginosis.25 These factors further support the need for the use of a standardized method, such as the one used in our investigation, to conduct vaginal PMN counts on Gram-stain preparations.
Vaginal PMNs are a common finding on vaginal Gram-stain preparations. Evidence linking the presence of these cells in the vagina with increased vaginal levels of proinflammatory cytokines, as well as increased risk for both intra-amniotic infection and/or histologic chorioamnionitis, supports the potential inflammatory role of vaginal polymorphonuclear cells in the pathophysiologic pathway that culminates in preterm labor and delivery.15,19–21 Cauci et al21 noted a strong correlation between vaginal PMN counts and both vaginal interleukin (IL)-8 (r = 0.60, P < .001), a potent chemotactic and activating factor for polymorphonuclear cells, and IL-1β (r = 0.56, P < .001). Vaginal levels of both IL-8 and IL-1β were significantly higher among women with a high PMN count (≥ 13.9 cells per oil immersion field) than among women with lower counts.21 Yamada and colleagues20 similarly reported a strong association between vaginal PMN counts and vaginal levels of IL-8 (r = 0.46). Hitti and colleagues15 also noted a strong association between vaginal PMN counts and IL-8 levels (r = 0.50, P < .001), but not vaginal IL-6 levels (r = 0.07, P = not significant). Importantly, these investigators noted that vaginal PMN counts were strongly associated with both amniotic fluid levels of IL-8 (r = 0.27, P = .001) and IL-6 (r = 0.30, P < .001). Women with an elevated PMN count (> 5 per × 400 high-power field), had a 1.6-fold increased risk to have either evidence of amniotic fluid infection or elevated amniotic fluid IL-6 levels, after adjusting for gestational age. These associations highlight the potential importance of vaginal PMN evaluation as part of both research and clinical risk assessment for preterm birth and upper genital tract infection/inflammation.
Use of a standardized approach for the PMN assessment on vaginal smear Gram-stain preparations is needed as we further our understanding of the relationship of activation of the innate immune system with preterm labor and subsequent delivery (see box). Because epithelial cells are typically present on most fields of vaginal Gram-stain preparations and the density of these cells is related to thickness of the preparation, coquantitation of epithelial cells with PNM cells may provide an intrafield marker of field cellular density. Thus, the use of PMN/EPI ratio may improve reliability of PMN assessment on Gram-stain preparations. Further investigation, using alternative preparations and staining methods (eg, vaginal smears with Wright or Giemsa staining for eukaryotic cells), is needed to more rigorously evaluate the relationship and predictive value of PMN counts and PMN/EPI ratios with preterm birth and upper and lower genital tract infection and/or inflammation. At the present time, however, use of these markers for the identification of women at increased risk for preterm birth remains limited to the research arena.
1.Goldenberg RL. The management of preterm labor. Obstet Gynecol 2002;100:1020–37.
2.Hillier SL, Nugent RP, Eschenbach DA, Krohn MA, Gibbs RS, Martin DH, et al, for the Vaginal Infections and Prematurity Group. Association between bacterial vaginosis and preterm delivery of a low-weight infant. N Engl J Med 1995;333:1737–42.
3.Gravett MG, Hummel D, Eschenbach DA, Holmes KK. Preterm labor associated with subclinical amniotic fluid infection and with bacterial vaginosis. Obstet Gynecol 1986;67:229–37.
4.Hillier SL, Martius J, Krohn M, Kiviat N, Holmes KK, Eschenbach DA. A case-control study of chorioamniotic infection and histologic chorioamnionitis in prematurity. N Engl J Med 1988;319:972–8.
5.Andrews WW, Hauth JC, Goldenberg RL, Gomez R, Romero R, Cassell GH. Amniotic fluid interleukin-6: correlation with upper genital tract microbial colonization and gestational age in women delivered after spontaneous labor versus indicated delivery. Am J Obstet Gynecol 1995;173:606–12.
6.Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection and preterm delivery. N Engl J Med 2000;342:1500–7.
7.Hauth JC, Andrews WW, Goldenberg RL. Infection-related risk factors predictive of spontaneous preterm labor and birth. Prenat Neonat Med 1998;3:86–90.
8.Silver HM, Sperling RS, St. Clair PJ, Gibbs RS. Evidence relating bacterial vaginosis to intra-amniotic infection. Am J Obstet Gynecol 1989;161:808–12.
9.Hillier SL, Krohn MA, Cassen E, Easterling TR, Rabe LK, Eschenbach DA. The role of bacterial vaginosis and vaginal bacteria in amniotic fluid infection in women in preterm labor with intact fetal membranes. Clin Infect Dis 1995;20(suppl):S276–8.
10.Korn AP, Bolan G, Padian N, Ohm-Smith M, Schachter J, Landers DV. Plasma cell endometritis in women with symptomatic bacterial vaginosis. Obstet Gynecol 1995;85:387–90.
11.Goepfert AR, Goldenberg RL, Andrews WW, Hauth JC, Mercer B, Iams J, et al, for the NICHD MFMU Network. The Preterm Prediction Study: association between cervical interleukin 6 concentration and spontaneous preterm birth. Am J Obstet Gynecol 2001;184:483–8.
12.Goldenberg RL, Andrews WW, Guerrant RL, Newman M, Mercer B, Iams J, et al. The Preterm Prediction Study: cervical lactoferrin concentration, other markers of lower genital tract infection, and preterm birth. Am J Obstet Gynecol 2000;182:631–5.
13.Yudin MH, Hillier SL, Wiesenfeld HC, Krohn MA, Amortegui AA, Sweet RL. Vaginal polymorphonuclear leukocytes and bacterial vaginosis as markers for histologic endometritis among women without symptoms of pelvic inflammatory disease. Am J Obstet Gynecol 2003;188:318–23.
14.Peipert JF, Ness RB, Soper DE, Bass D. Association of lower genital tract inflammation with objective evidence of endometritis. Infect Dis Obstet Gynecol 2000;8:83–7.
15.Hitti J, Hillier SL, Agnew KJ, Krohn MA, Reisner DP, Eschenbach DA. Vaginal indicators of amniotic fluid infection in preterm labor. Obstet Gynecol 2001;97:211–9.
16.Simhan HN, Caritis SN, Krohn MA, Hillier SL. Elevated vaginal pH and neutrophils are associated with early spontaneous preterm birth. Am J Obstet Gynecol 2003;189:1150–4.
17.Yamada T, Minakami H, Matsubara S, Yatsuda T, Sato I. Changes in polymorphonuclear leukocytes in the vagina of patients with preterm labor. Gynecol Obstet Invest 1998;45:32–4.
18.Matsubara S, Yamada T, Minakami H, Sato I. Stimulated polymorphonuclear leukocytes in vaginal secretions from patients with preterm labor. Gynecol Obstet Invest 1998;45:35–40.
19.Yamada T, Matsubara S, Minakami H, Ohkuchi A, Hiratsuka M, Sato I. Relation between viability of vaginal polymorphonuclear leukocytes and presence of histologic chorioamnionitis. Acta Obstet Gynecol Scand 2000;79:818–23.
20.Yamada T, Minakami H, Matsubara S, Kohmura Y, Aoya T, Sato I. Changes in the number of polymorphonuclear leukocytes and concentrations of IL-8 and granulocyte elastase in the vaginas of normal pregnant women. Am J Reprod Immunol 2002;47:98–103.
21.Cauci S, Guaschino S, DeAloysio D, Driussi S, DeSanto D, Penacchioni P, et al. Interrelationships of interleukin-8 and interleukin-1β and neutrophils in vaginal fluid of healthy and bacterial vaginosis positive women. Mol Hum Reprod 2003;9:53–8.
22.Neggers Y, Goldenberg R, Cliver S, Hauth J. Effects of domestic violence on preterm birth and low birth weight. Acta Obstet Gynecol Scand 2004;83:455–60
23.Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of Gram stain interpretation. J Clin Microbiol 1991;29:297–301.
24.Buchvald D, Demes P, Gombosova A, Mraz P, Valent M, Stefanovic J. Vaginal leucocyte characteristics in urogenital trichomoniasis. APMIS 1992;100:393–400.
25.Hauth JC, Macpherson C, Carey JC, Klebanoff MA, Hillier SL, Ernest JM, Leveno KJ, Wapner R, Varner M, Trout W, Moawad A, Sibai B. Early pregnancy threshold vaginal pH and Gram stain scores predictive of subsequent preterm birth in asymptomatic women. Am J Obstet Gynecol 2003;188:831–5.