OBJECTIVE: To investigate whether biomarkers from different pathways of spontaneous preterm birth (cervical membrane degradation [fetal fibronectin], cervical remodeling [soluble E-cadherin], and inflammation (elafin, surfactant protein-D, interleukin-6 [IL-6]) were superior to one biomarker alone in predicting preterm birth. Our secondary objective was to examine the association of these biomarkers with cervical length in predicting preterm birth.
METHODS: We performed a single-center, prospective cohort study from August 2011 to November 2012 of asymptomatic women at risk for spontaneous preterm birth as a result of obstetric and gynecologic history. Cervicovaginal fluid and cervical length measurements were collected at two time points (20–23 6/7 weeks and 24–27 6/7 weeks of gestation).
RESULTS: Among the 104 women with complete data, the preterm birth rate was 24.5%. Prior preterm birth (P=.006) and cervical length at visit 1 (P=.003) were significantly associated with preterm birth, whereas fetal fibronectin and median biomarker levels (elafin, soluble E-cadherin, IL-6) were not. Median surfactant protein-D levels at visit 1 by preterm birth status were statistically but not clinically different (0.44 ng/mL compared with 0.40 ng/mL, P<.001). Analyses of biomarkers from more than one pathway were not superior to single biomarker analyses in predicting prematurity. Neither inclusion of biomarkers nor fetal fibronectin improved the predictive ability of cervical length alone.
CONCLUSION: Cervical length assessment and obstetric history but not fetal fibronectin or biomarkers were useful in the risk stratification of women identified to be at greatest risk for spontaneous preterm birth.
LEVEL OF EVIDENCE: II
The ability to use cervical length to risk-stratify asymptomatic women at risk for spontaneous preterm birth is not improved by fetal fibronectin or other cervicovaginal biomarkers.
Maternal and Child Health Research Program, Department of Obstetrics & Gynecology, Center for Research on Reproduction and Women's Health, University of Pennsylvania, Perelman School of Medicine, Philadelphia, Pennsylvania.
Corresponding author: Jamie A. Bastek, MD, MSCE, Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Hospital of the University of Pennsylvania, 3400 Spruce Street, 585 Dulles Building, Philadelphia, PA 19104; e-mail: email@example.com.
Supported by the 2011–2012 American College of Obstetricians and Gynecologists/Hologic Research Award in Preterm Birth (J.A.B., principal investigator).
Financial Disclosure The authors did not report any potential conflicts of interest.
Presented at the Society for Maternal-Fetal Medicine Annual Meeting, February 11–16, 2013, San Francisco, California, and at the Society for Gynecologic Investigation Annual Meeting, March 20–23, 2013, Orlando, Florida.
Preterm birth is a leading cause of perinatal morbidity and mortality. Women with an obstetric history of preterm birth have a 30–55% risk of a recurrent preterm birth.1 Current strategies involving either use of 17-α hydroxyprogesterone caproate or cerclage may reduce the risk of recurrent preterm birth by approximately one-third.1,2 Although these interventions help reduce the rate of preterm birth, the burden of prematurity in this cohort remains high with many patients having a recurrent preterm birth. As a result, strategies are urgently needed to identify women at greatest risk so that further therapeutic interventions can be explored.
Despite our knowing the complexity of the preterm birth syndrome and the different pathways involved, there is a paucity of clinical studies investigating whether detection of more than one of these pathways in a single patient might enhance the identification of those at greatest risk for preterm birth.
Our primary objective, therefore, was to investigate whether the predictive value of a panel of cervicovaginal fluid biomarkers associated with different, biologically plausible pathways of preterm birth, specifically, mechanical or inflammatory-mediated membrane breakdown3,4 and cervical remodeling, has superior ability to predict preterm birth than any one biomarker alone. Although fetal fibronectin has been studied fairly extensively as a biomarker of membrane degradation,5–7 the use of cervicovaginal biomarkers to indicate the presence of cervical remodeling is in its infancy. Therefore, in addition to fetal fibronectin, based on work from our laboratory and emerging data, we chose to study soluble E-cadherin as a biomarker of cervical remodeling, and elafin, surfactant protein-D, and interleukin-6 (IL-6) as biomarkers of cervical vaginal inflammation.8–13
Therefore, our secondary objective was to investigate whether the incorporation of the aforementioned biomarkers could improve the test characteristics of transvaginal cervical length assessment in the risk stratification of preterm birth.
MATERIALS AND METHODS
We performed a prospective cohort study at a single, urban tertiary care center between August 2011 and November 2012. The cohort consisted of women carrying a singleton pregnancy between 20 weeks and 23 6/7 weeks of gestation who were at high risk for preterm birth as a result of a history of prior spontaneous preterm birth or second-trimester loss, prior cervical surgery without a subsequent full-term birth, or uterine anomaly. We excluded women who used systemic steroids or immunosuppressive therapy; had an immunologic disease such as human immunodeficiency virus, lupus, pregestational diabetes, rheumatoid arthritis, Crohn's disease, ulcerative colitis, cancer, or organ transplant; had anything in the vagina in the 24 hours preceding cervicovaginal specimen collection; active vaginal bleeding; evidence of rupture of membranes; or cervical dilatation of 3 cm or more.
Cervicovaginal fluid and transvaginal cervical length measurement data were collected at two time points: between 20 weeks and 23 6/7 weeks of gestation (visit 1) and again between 24 weeks and 27 6/7 weeks of gestation (visit 2).
Cervicovaginal fluid was collected using a sterile speculum and a sterile cotton-tipped swab. Samples were collected in phosphate-buffered saline, immediately flash-frozen in liquid nitrogen, and stored at −80°C. Before biomarker analyses were performed, the sample was thawed and centrifuged at 4°C at 12,000 rotations per minute for 10 minutes. The resulting supernatant was collected and subsequently analyzed for elafin, soluble E-cadherin, surfactant protein-D, and IL-6 levels using standard enzyme-linked immunosorbent assays. For efficiency, samples were batched for processing. Analyses were performed to ensure there were no significant differences in analyte levels by processing date. Each sample was run in duplicate, and the results for each biomarker were the average of these two values. If the values between the duplicates were discrepant, the sample was not used in the analyses. Intra-assay and interassay coefficients of variation, respectively, for each biomarker were as follows: elafin (less than 10%; less than 12%), soluble E-cadherin (less than 9%; less than 11%), surfactant protein-D (less than 9%; less than 10%), and IL-6 (less than 8%; less than 10%). Potential risks to protein content through cryopreservation were minimized through immediate placement of the samples into liquid nitrogen after collection and the storage of multiple aliquots per sample to avoid repeat freeze and thaw cycles. Samples were analyzed for fetal fibronectin using a system analyzer .
Cervical length measurement was performed per the protocol established by Iams14 and the shortest measurement was used for each patient. Because cervical length assessment was performed at the time of a clinical ultrasonographic appointment, the physicians performing these ultrasonograms were not blinded. Patients with a cervical length less than 15 mm and a history of prior spontaneous preterm birth were recommended to have a cerclage placed,2 whereas patients with a cervical length less than 20 mm and no such history were recommended vaginal progesterone.15 All patients with a prior spontaneous preterm birth or second-trimester loss are also routinely offered 17-α hydroxyprogesterone caproate, and patients had either initiated or declined this intervention by the time of their visit 1 appointment for this study. Information regarding 17-α hydroxyprogesterone caproate, vaginal progesterone use, or cerclage placement was recorded for all study participants.
Pertinent obstetric and gynecologic history data were initially abstracted from electronic medical records and verified with the patient by the principal investigator (J.A.B.). For pregnancy outcome data, trained clinical research coordinators (B.A.A., M.A.M., M.E.R.) abstracted medical information, which was confirmed by physician co-authors (A.H., C.M.O.). Data were finally verified and transferred to an electronic database by the principal investigator. All women provided written consent for specimen collection and use of their deidentified health information for research purposes.
Pregnancy was dated using standard obstetric estimates of gestational age. Specifically, patients with a sure last menstrual period were dated by their period if this gestational age was confirmed by ultrasonographic dating within 7 days of a first-trimester ultrasonogram or 10 days of a second-trimester ultrasonogram between 14 weeks and 20 weeks of gestation. Patients whose last menstrual period dating was inconsistent with their ultrasonographic dating or who were unsure of their last menstrual period were dated by their earliest ultrasonogram. Preterm birth was defined as spontaneous delivery at less than 37 weeks of gestation.
χ2 analyses were performed to determine associations between categorical data. Continuous cervical length and biomarker data were found to be nonparametric. Therefore, nonparametric tests were performed to determine the association between preterm birth and median biomarker values and cervical lengths, and nonparametric correlation coefficients were calculated between biomarkers and cervical length as well as biomarkers to each other.
We theorized that the median biomarker values could be used as cut points between normal and pathologic states and performed receiver operating characteristic curve calculations to verify that this assumption was statistically sound. Thus, we defined elevated biomarker values as values greater than the median.
Univariable logistic regression was performed to compute odds ratios (ORs) as well as 95% confidence intervals (CIs) to estimate the association between preterm birth and each biomarker and risk factor. Variables identified as potential risk factors in unadjusted logistic analyses (P<.2) were used to create separate multivariable logistic models using biomarker and cervical length data from visit 1 and 2.16 A backward selection method was performed to determine which combination of risk factors generated the most parsimonious yet predictive model for each outcome.17 Areas under the receiver operating characteristic curve (AUC) were calculated and compared. STATA 10.1 was used to perform all analyses.
A priori sample size calculations were performed and the study was powered to determine the number of patients necessary to demonstrate that positive screening for two pathways (membrane degradation and cervical remodeling) was associated with a twofold increase in preterm birth. We assumed that the baseline prevalence of preterm birth in our cohort would be 30%.1 We assumed a type I error rate of 5%, 80% power, and a one-to-one enrollment ratio between those with and without both a positive fetal fibronectin and an elevated biomarker of cervical remodeling. We determined we would need a sample size of 98 women to detect a twofold difference in the odds of preterm birth between exposure groups.
This study was approved by the institutional review board at the University of Pennsylvania.
Patient enrollment is summarized in Figure 1. Complete visit 1 and visit 2 data were available for 104 and 47 patients, respectively. One-fourth of patients experienced spontaneous preterm delivery, secondary to either preterm labor or preterm premature rupture of membranes, before 37 weeks of gestation (24.5%). We compared demographic variables, including diagnosis of chorioamnionitis or preeclampsia in the current pregnancy, among women who delivered at term and preterm. History of prior spontaneous preterm birth or second-trimester loss was the only demographic variable significantly associated with prematurity. There were no other significant differences between the two groups (Table 1).
The association between fetal fibronectin and other biomarkers, cervical length, and preterm birth was also explored. Fetal fibronectin was not associated with elafin, surfactant protein-D, cervical length, or preterm birth at visit 1 or with elafin, soluble E-cadherin, surfactant protein-D, cervical length, or preterm birth at visit 2. However, positive fetal fibronectin was associated with significantly higher median levels of soluble E-cadherin at visit 1 (3.54 micrograms/mL compared with 1.59 micrograms/mL, P=.03) and IL-6 at both visit 1 (16.18 pg/mL compared with 1.06 pg/mL, P<.001) and visit 2 (30.35 pg/mL compared with 0.82 pg/mL, P=.004).
Median values of biomarkers were compared to determine their association with preterm birth. There was no association between preterm birth and elafin, soluble E-cadherin, or IL-6. Median surfactant protein-D values were significantly greater at visit 1 in preterm compared with term deliveries (0.44 ng/mL compared with 0.40 ng/mL, P<.001); however, there was no association between preterm birth and median surfactant protein-D at visit 2. There was also no association between 17-α hydroxyprogesterone caproate use and median biomarker values.
Median cervical length was significantly shorter at both visit 1 and visit 2 in patients with preterm rather than full term birth (visit 1: 3.38 cm compared with 3.02 cm, P=.002; visit 2: 3.29 cm compared with 2.88 cm, P=.046). Using a definition of either less than 2 cm or less than 3 cm was not associated with preterm birth at (visit 1, 2 cm: P=.055, 3 cm: P=.055) or (visit 2, 2 cm: P=.44, 3 cm: P=.09).
Differences between visit 1 and visit 2 biomarker and cervical length measurements were compared to evaluate whether longitudinal changes within a given participant were associated with preterm birth. There was no association between visit 1 and visit 2 levels of biomarkers or cervical length. Only a change in fetal fibronectin from negative at visit 1 to positive visit 2 was significantly associated with preterm birth (P=.04).
Nonparametric correlation coefficients between biomarkers and cervical length as well as biomarkers to each other were also calculated. All correlations were either weak with correlation coefficients less than 20% or were statistically not significant (Table 2).
Univariable logistic regression was performed to determine the odds of preterm birth associated with each biomarker, cervical length, and obstetric history. Prior preterm birth was associated with an 11-fold increase (95% CI 1.37–83.66) in the risk of preterm birth compared with prior cervical surgery or uterine anomaly. For every 1-cm increase in cervical length at visit 1, the odds of preterm birth decreased 61% (95% CI 0.21–0.72), whereas there was no significant relationship between cervical length at visit 2 and preterm birth (OR 0.41, 95% CI 0.15–1.14) (Table 3).
Elevated biomarker values were defined as values greater than the median. The presence of both a biomarker for membrane breakdown (fetal fibronectin) and an elevated biomarker for inflammation (surfactant protein-D) was associated with a twofold increase in the risk of preterm birth at both visit 1 (OR 2.57, 95% CI 1.32–4.98) and visit 2 (OR 2.93, 95% CI 1.02–8.36). The predictive ability of both a positive fetal fibronectin and surfactant protein-D was stronger at visit 1 (AUC=0.68) and visit 2 (AUC=0.63) than that of fetal fibronectin, elafin, soluble E-cadherin, or IL-6 alone; however, it was no stronger than surfactant protein-D alone (Table 3). A positive fetal fibronectin was not associated with an increased risk of preterm birth among women with other elevated biomarkers of cervical remodeling, including elafin, soluble E-cadherin, or IL-6 at visit 1 or visit 2.
After identifying potential candidate variables through unadjusted analyses as described previously, the initial multivariable model to predict preterm birth using visit 1 data included prior preterm birth, cervical length, and surfactant protein-D. Although the predictive value of this model was not significantly better than the model without surfactant protein-D (AUC=0.79 compared with 0.76, P=.28), removal of cervical length from the model significantly reduced the predictive value (AUC=0.76 compared with 0.63, P=.003) (Fig. 2). Thus, the most parsimonious yet predictive model for preterm birth using visit 1 data included cervical length and prior preterm birth history with a strong AUC (0.76) to predict preterm birth.
Finally, the initial multivariable model to predict preterm birth using visit 2 data included prior preterm birth, cervical length, and fetal fibronectin. The predictive value of this model was not significantly better than the model without fetal fibronectin (AUC=0.72 compared with 0.71, P=.69); however, removal of cervical length from the model again significantly reduced the predictive value (AUC=0.73 compared with 0.60, P=.03) (Fig. 3). Thus, the most parsimonious yet predictive model for preterm birth using visit 2 data included cervical length and prior preterm birth history (AUC=0.73).
Consistent with our primary hypothesis and sample size calculation, we did observe that women with positive screening for two pathways, membrane degradation (fetal fibronectin) and inflammation (surfactant protein-D), were at twofold greater risk of preterm birth than patients without this dual exposure. However, inclusion of biomarkers from two pathways did not significantly increase the ability to predict preterm birth compared with analysis of one biomarker alone nor did it significantly improve the predictive ability of transvaginal ultrasonography to identify asymptomatic high-risk patients at greatest risk of preterm birth.
Prior authors have studied the ability of transvaginal ultrasonography with fetal fibronectin or other biomarkers to predict preterm birth in both symptomatic (fetal fibronectin,18,19 other biomarkers20,21) and asymptomatic (fetal fibronectin,22,23 other biomarkers24,25) patients with varying degrees of success. Furthermore, previous studies have also examined the ability of a panel of biomarkers alone to predict preterm birth in both symptomatic26 and asymptomatic25,27 patients without corroboration by meta-analysis.28 Two previously published studies are somewhat similar to ours in their incorporation of cervical length, fetal fibronectin, and a biomarker to predict preterm birth: the 2001 Preterm Prediction Study, which examined the combined role of maternal serum alpha-fetoprotein, alkaline phosphatase, and granulocyte colony-stimulating factor along with cervical length and fetal fibronectin to predict risk of preterm birth in asymptomatic women25 and a 2007 study by Eroglu examining the predictive role of cervical length, fetal fibronectin, and phosphorylated insulin-like growth factor binding protein-1 to predict preterm birth in symptomatic women.29 To our knowledge, however, based on a May 1, 2013, PubMed search of articles published in the English language using medical subject headings “preterm birth,” “biomarker,” “fetal fibronectin,” and “cervical length,” we are the first authors to study whether the predictive value of a panel of these novel cervicovaginal fluid biomarkers associated with different, biologically plausible pathways of preterm birth were superior both to a single biomarker from an isolated pathway and to transvaginal cervical length assessment alone in an asymptomatic high-risk population.
Our study had several strengths. Our patients were identified prospectively, thus limiting misclassification bias. Eligibility criteria for enrollment were determined before the start of the study by the primary and senior investigators and not by treating physicians, which minimized potential enrollment biases. Finally, there was very good follow-up on all of the patients enrolled, because only one patient was lost to follow-up after deciding not to continue her care at our institution.
Our study was not without weaknesses. Although we exceeded our a priori sample size, our study was still somewhat small with complete data (ie, biomarker, cervical length, and delivery information) from only 104 patients. Furthermore, despite having over 100 biospecimens and cervical length measurements to analyze from visit 1, we were able to capture visit 2 biomarker levels and repeat cervical length data on significantly fewer patients (69.8% and 44.3%, respectively). This limited our statistical power for visit 2 analyses. Furthermore, women were eligible for enrollment based on a history of preterm birth after either spontaneous labor or preterm premature rupture of membranes. These conditions were considered together in our analyses when in fact their underlying pathophysiology may be different. Finally, based on the population we chose to study, it is not clear whether our findings can be extrapolated to women at risk for preterm birth for reasons other than obstetric and gynecologic history.
In conclusion, our findings do not support the use of biomarkers to predict the risk of spontaneous preterm birth in asymptomatic high-risk women because cervical length and obstetric history are more predictive.
1. Meis PJ, Klebanoff M, Thom E, Dombrowski MP, Sibai B, Moawad AH, et al.. Prevention of recurrent preterm delivery by 17 alpha-hydroxyprogesterone caproate. N Engl J Med 2003;348:2379–85.
2. Owen J, Hankins G, Iams JD, Berghella V, Sheffield JS, Perez-Delboy A, et al.. Multicenter randomized trial of cerclage for preterm birth prevention in high-risk women with shortened midtrimester cervical length. Am J Obstet Gynecol 2009;201:375.e1–8.
3. Matsuura H, Greene T, Hakomori S. An alpha-N-acetylgalactosaminylation at the threonine residue of a defined peptide sequence creates the oncofetal peptide epitope in human fibronectin. J Biol Chem 1989;264:10472–6.
4. Sibille Y, Lwebuga-Mukasa JS, Polomski L, Merrill WW, Ingbar DH, Gee JB. An in vitro model for polymorphonuclear-leukocyte-induced injury to an extracellular matrix. Relative contribution of oxidants and elastase to fibronectin release from amnionic membranes. Am Rev Respir Dis 1986;134:134–40.
5. Lockwood CJ, Senyei AE, Dische MR, Casal D, Shah KD, Thung SN, et al.. Fetal fibronectin in cervical and vaginal secretions as a predictor of preterm delivery. N Engl J Med 1991;325:669–74.
6. Peaceman AM, Andrews WW, Thorp JM, Cliver SP, Lukes A, Iams JD, et al.. Fetal fibronectin as a predictor of preterm birth in patients with symptoms: a multicenter trial. Am J Obstet Gynecol 1997;177:13–8.
7. Sanchez-Ramos L, Delke I, Zamora J, Kaunitz AM. Fetal fibronectin as a short-term predictor of preterm birth in symptomatic patients: a meta-analysis. Obstet Gynecol 2009;114:631–40.
8. Kyongo JK, Jespers V, Goovaerts O, Michiels J, Menten J, Fichorova RN, et al.. Searching for lower female genital tract soluble and cellular biomarkers: defining levels and predictors in a cohort of healthy Caucasian women. PLoS One 2012;7:e43951.
9. Stock SJ, Duthie L, Tremaine T, Calder AA, Kelly RW, Riley SC. Elafin (SKALP/Trappin-2/proteinase inhibitor-3) is produced by the cervix in pregnancy and cervicovaginal levels are diminished in bacterial vaginosis. Reprod Sci 2009;16:1125–34.
10. Nold C, Anton L, Brown A, Elovitz M. Inflammation promotes a cytokine response and disrupts the cervical epithelial barrier: a possible mechanism of premature cervical remodeling and preterm birth. Am J Obstet Gynecol 2012;206:208.e1–7.
11. Brik M, Antonio P, Perales-Puchalt A, Diago V, Perales A. Cervical interleukin-6 as a predictive test for preterm delivery in symptomatic women: preliminary results. Eur J Obstet Gynecol Reprod Biol 2011;155:14–8.
12. Oberley RE, Goss KL, Ault KA, Crouch EC, Snyder JM. Surfactant protein D is present in the human female reproductive tract and inhibits Chlamydia trachomatis infection. Mol Hum Reprod 2004;10:861–70.
13. Oberley RE, Goss KL, Hoffmann DS, Ault KA, Neff TL, Ramsey KH, et al.. Regulation of surfactant protein D in the mouse female reproductive tract in vivo. Mol Hum Reprod 2007;13:863–8.
14. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das A, et al.. The length of the cervix and the risk of spontaneous premature delivery. National Institute of Child Health and Human Development Maternal Fetal Medicine Unit Network. N Engl J Med 1996;334:567–72.
15. Hassan SS, Romero R, Vidyadhari D, Fusey S, Baxter JK, Khandelwal M, et al.. Vaginal progesterone reduces the rate of preterm birth in women with a sonographic short cervix: a multicenter, randomized, double-blind, placebo-controlled trial. Ultrasound Obstet Gynecol 2011;38:18–31.
16. Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York (NY): Wiley; 2000. p. xii, 375.
17. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:839–43.
18. Gomez R, Romero R, Medina L, Nien JK, Chaiworapongsa T, Carstens M, et al.. Cervicovaginal fibronectin improves the prediction of preterm delivery based on sonographic cervical length in patients with preterm uterine contractions and intact membranes. Am J Obstet Gynecol 2005;192:350–9.
19. Hincz P, Wilczynski J, Kozarzewski M, Szaflik K. Two-step test: the combined use of fetal fibronectin and sonographic examination of the uterine cervix for prediction of preterm delivery in symptomatic patients. Acta Obstet Gynecol Scand 2002;81:58–63.
20. Tsiartas P, Holst RM, Wennerholm UB, Hagberg H, Hougaard D, Skogstrand K, et al.. Prediction of spontaneous preterm delivery in women with threatened preterm labour: a prospective cohort study of multiple proteins in maternal serum. BJOG 2012;119:866–73.
21. Kim MA, Lee BS, Park YW, Seo K. Serum markers for prediction of spontaneous preterm delivery in preterm labour. Eur J Clin Invest 2011;41:773–80.
22. Goldenberg RL, Iams JD, Das A, Mercer BM, Meis PJ, Moawad AH, et al.. The Preterm Prediction Study: sequential cervical length and fetal fibronectin testing for the prediction of spontaneous preterm birth. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 2000;182:636–43.
23. Iams JD, Goldenberg RL, Mercer BM, Moawad A, Thom E, Meis PJ, et al.. The Preterm Prediction Study: recurrence risk of spontaneous preterm birth. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 1998;178:1035–40.
24. Hee L. Likelihood ratios for the prediction of preterm delivery with biomarkers. Acta Obstet Gynecol Scand 2011;90:1189–99.
25. Goldenberg RL, Iams JD, Mercer BM, Meis PJ, Moawad A, Das A, et al.. The Preterm Prediction Study: toward a multiple-marker test for spontaneous preterm birth. Am J Obstet Gynecol 2001;185:643–51.
26. Campbell MK, Challis JR, DaSilva O, Bocking AD. A cohort study found that white blood cell count and endocrine markers predicted preterm birth in symptomatic women. J Clin Epidemiol 2005;58:304–10.
27. Smith GC, Crossley JA, Aitken DA, Jenkins N, Lyall F, Cameron AD, et al.. Circulating angiogenic factors in early pregnancy and the risk of preeclampsia, intrauterine growth restriction, spontaneous preterm birth, and stillbirth. Obstet Gynecol 2007;109:1316–24.
28. Menon R, Torloni MR, Voltolini C, Torricelli M, Merialdi M, Betran AP, et al.. Biomarkers of spontaneous preterm birth: an overview of the literature in the last four decades. Reprod Sci 2011;18:1046–70.
© 2013 by The American College of Obstetricians and Gynecologists.
29. Eroglu D, Yanik F, Oktem M, Zeyneloglu HB, Kuscu E. Prediction of preterm delivery among women with threatened preterm labor. Gynecol Obstet Invest 2007;64:109–16.