Preeclampsia is a common disorder affecting between 5% and 7% of all pregnancies.1 It remains a major contributor to maternal mortality1 and accounts for a substantial proportion of low-birth-weight neonates and iatrogenic preterm delivery.2 Prevalence and morbidity have remained unchanged over the past decade highlighting the need to improve diagnostic3,4 and prognostic5 testing facilitating appropriate resource allocation. Preeclampsia is unique to pregnancy and is characterized by poor placentation6 and abnormal inflammatory and vascular responses7 resulting in multiorgan dysfunction.
Presenting symptoms of preeclampsia are often subjective and nonspecific with clinical findings based on features of advanced disease or markers of end-organ involvement. High blood pressure and urinary protein excretion are typically used to diagnose the disease but both are secondary features of a primary placental problem and subject to measurement error and poor test accuracy.8 It is currently difficult to distinguish preeclampsia of a severity that requires early delivery from other less serious phenotypes.9,10 An accurate biomarker (or panel of biomarkers) to enable prognosis of perinatal complications could have a substantial effect on management strategies, with the aim of minimizing adverse maternal and fetal outcomes.
The aim of this study was to evaluate a wide panel of 47 candidate biomarkers (including those that are currently widely reported and reflect the heterogeneity of the disease) in women presenting preterm with suspected preeclampsia to optimize determination of an important clinical outcome, that of preeclampsia requiring delivery within 14 days.
MATERIALS AND METHODS
A prospective multicenter cohort study was undertaken between January 2011 and February 2012 in seven consultant-led maternity units in the United Kingdom and Ireland.4 Women were eligible for the study if they had been referred or presented with suspected preeclampsia (ie, signs or symptoms [or both] of preeclampsia), were between 20 0/7 and 36 6/7 weeks of gestation with a singleton or twin pregnancy, and were aged 16 years or older. Women with confirmed preeclampsia (or with any adverse outcome already present) were not eligible. We undertook a planned analysis reported here on two groups of women: group 1—presenting before 35 weeks of gestation, and group 2—presenting between 35 0/7 and 36 6/7 weeks of gestation. These gestational age groupings were prespecified based on known differences in pathophysiologic pathways associated with preterm preeclampsia and our prior knowledge of gestational changes of biomarker concentrations related to these pathways. Written informed consent was obtained and baseline demographic and pregnancy-specific information, including blood pressure readings, were entered onto the study database. Blood pressure was taken according to unit guidelines. Blood samples were drawn into ethylenediamine tetraacetic acid, with consent, at the time of enrollment. The samples were labeled, transported to the laboratory, and the plasma was stored until analysis at −80°C. Pregnancy outcomes were determined by case note review with independent adjudication (masked to all biomarker concentrations) for final maternal diagnosis. All hypertensive disorders of pregnancy were defined according to the American College of Obstetricians and Gynecologists' Practice Bulletin in use at the time of the study.11 Independent adjudication was undertaken by two senior physicians, masked to biomarker measurements, requiring documentation of endpoints required to fulfill the diagnostic criteria; disagreement was resolved by a third adjudicator. The predefined adverse maternal outcomes had been identified for a previous study in preeclampsia by iterative Delphi consensus10 and have been described in detail elsewhere.4 All sites managed women (including decision for delivery) in line with the Hypertension in Pregnancy recommendations from the National Institute for Health and Care Excellence.12
An initial panel of biomarkers was selected based on a priori knowledge of an association with preeclampsia, a biological role in placentation, or a role in cellular mechanisms involved in the pathogenesis of preeclampsia (eg, angiogenesis, inflammation, coagulation). The full list of 47 biomarkers, measured with 57 assays (where potentially biologically important assays of different epitope specificity were available), was generated following a review of the literature, appraisal of selected bibliographies, and consultation with medical experts (Appendix 1, available online at http://links.lww.com/AOG/A822).
Plasma samples were tested for placental growth factor (PlGF) using the Triage PlGF Test by trained laboratory staff at the study site where the sample was taken (as previously published). Samples were labeled and transported to the laboratory where they were spun at 3,000 rotations per minute for 10 minutes. The additional 56 biomarker assays were analyzed in a central laboratory facility (Alere, San Diego, California) and full details of assay methods given in Appendix 2 (available online at http://links.lww.com/AOG/A822) and Appendix 3 (available online at http://links.lww.com/AOG/A822). All participants had delivered and pregnancy outcomes recorded before biomarker concentrations were analyzed and revealed and all laboratory staff were masked to clinical outcomes.
Standard distributional checks showed high levels of skewness for all 57 assays, consistent with underlying log normal distributions. Logged values of these biomarkers were therefore used. Before considering the pregnancy outcomes, statistical factor analysis of biomarker data were undertaken, reducing the 47 biomarkers into a smaller group of factors. Factor analysis sorted the biomarkers into a small number of highly correlated groups, without reference to outcome, containing the majority of the information in the full data set.13 Factor summary scores were then calculated for all women. Consideration of scree plots and eigenvalues (greater than two) identified the most important factors for further analysis.14 These factors were rotated (orthogonal varimax method) so that each factor related strongly (correlation greater than 0.6) to a small number of biomarkers only (factor analysis is displayed in Appendix 4, available online at http://links.lww.com/AOG/A822). Significant factors (and their biomarkers) were identified for further investigation (Appendix 5, available online at http://links.lww.com/AOG/A822). For the multiple logistic regression model, the principal outcome was preeclampsia requiring delivery within 14 days (prespecified by consensus of clinical investigators). Stepwise logistic regression was used to determine which biomarkers or factors appeared to provide additional information beyond that derived from PlGF, and prediction scores were extracted for the best combinations. A comparison of receiver operating curve (ROC) areas of individual biomarkers and combinations was made to see whether any of the additional information was both consistent and large enough to be clinically useful. Significance was assessed through use of a nonparametric test, which allowed for nonindependence of observations on the same participant with Bonferroni correction for multiple testing.15
Some biomarkers, with high uniqueness scores, were not strongly associated with any factor. To investigate whether any of these biomarkers had diagnostic power in addition to that provided by PlGF and biomarkers identified earlier, stepwise logistic regression was undertaken. To avoid excluding a biomarker that may be of potential value, it had to pass a series of tests so that the chance of a false-positive was greatly reduced (rather than using a standard multiple-testing correction to P values such as Bonferroni). The biomarker had to be a component of a significant factor, a significant predictor in logistic regression both alone and after allowing for PlGF and have a ROC area for the combined score significantly greater than PlGF alone. For biomarkers with a substantial proportion of measurements outside the limits of detection, we used a nonparametric test (ROC area) to determine whether the biomarkers had useful predictive power. Where the biomarker measurement (whether resulting from censoring or lack of predictive ability) was noninformative, it was excluded from further analysis.
Statistical analysis was carried out in Stata 11.2. Clinical variables and outcomes were compared using a Wilcoxon rank-sum nonparametric test. The prespecified sample size was calculated for accurate estimation of the sensitivity (within 10%) and specificity (within 6%) of a biomarker, assumed a sensitivity of 0.90, specificity 0.90, and 95% confidence intervals (CIs, two-tailed), for determining the primary endpoint; this required 62 patients with preeclampsia and 150 women not meeting the primary endpoint. The study is reported in accordance with Strengthening the Reporting of Observational Studies in Epidemiology guidelines.
The study was approved by East London Research Ethics Committee (reference 10/H0701/117). Participants gave informed consent and the study followed institutional guidelines.
Between January 2011 and February 2012, 423 women with enrollment samples and outcome data available were recruited to the study in seven centers across the United Kingdom and Ireland. There were 286 women in group 1 (presenting at 20 0/7 to 34 6/7 weeks of gestation) and 137 women in group 2 (presenting at 35 0/7 to 36 6/7 weeks of gestation) (Fig. 1).
For the 286 women who were enrolled before 35 0/7 weeks of gestation, characteristics of the study population at antenatal booking are shown in Table 1 subdivided into those that met the primary outcome (preeclampsia requiring delivery within 14 days) and all others. Table 2 shows characteristics of delivery and maternal and neonatal outcomes. Table 3 shows the test performance for the most promising individual biomarkers, depicted by ROC areas. Placental growth factor had the highest ROC area (0.87) for determining preeclampsia requiring delivery within 14 days; the ROC areas for soluble fms-like tyrosine kinase-1 (sFlt-1) (0.83) and endoglin (0.83) were not significantly different from that for PlGF. Addition of further biomarkers to PlGF increased the ROC area by a small, nonsignificant increment only. The highest test performance for preeclampsia requiring delivery within 14 days was found using a combination of PlGF, podocalyxin, soluble endoglin, and procalcitonin with a ROC area of 0.90, not significantly greater than the ROC area for PlGF alone (0.87; P=.43). Appendix 6 (available online at http://links.lww.com/AOG/A822) shows ROC areas for all 47 biomarkers analyzed and individual median biomarker concentrations in all women sampled are shown in Appendix 7 (available online at http://links.lww.com/AOG/A822). Sensitivity analysis demonstrated that excluding twin pregnancies altered PlGF test performance by less than 1%.
For women presenting between 35 0/7 and 36 6/7 weeks of gestation (n=137), the characteristics at booking and enrollment are shown in Appendix 8 (available online at http://links.lww.com/AOG/A822) and those for delivery and pregnancy outcomes in Appendix 9 (available online at http://links.lww.com/AOG/A822). Receiver operating characteristic areas and individual median biomarker concentrations for the individual biomarkers are given in Appendix 10 (available online at http://links.lww.com/AOG/A822) and Appendix 11 (available online at http://links.lww.com/AOG/A822), respectively. The results follow a similar pattern as for women presenting at earlier gestations. The ROC area for PlGF alone (0.75; 95% CI 0.67–0.83) in determining need for delivery for preeclampsia within 14 days was lower than that achieved in earlier gestations and other angiogenesis-related biomarkers were not significantly different to that for PlGF alone. Integration of sFlt-1 with PlGF (as a ratio) increased the ROC to 0.77 (95% CI 0.69–0.84). The combination of PlGF, pregnancy-associated plasma protein A, and cystatin yielded the highest ROC area of 0.81 (95% CI 0.74–0.88) (Table 4). Both increments were small and not significant.
This prospective multicenter study is a comprehensive direct comparison of diagnostic biomarkers for preeclampsia. The results demonstrate that in women with suspected preeclampsia presenting preterm, use of a single angiogenesis-related biomarker (PlGF, sFlt-1, or endoglin) alone represents a useful diagnostic test for determining preeclampsia requiring delivery within 14 days, a relevant endpoint indicating that a clinician has considered that the risks of adverse outcomes associated with ongoing expectant management are outweighed by the risks of delivery.
Suspected hypertensive disorders in pregnancy are the most common reason for presentation for obstetric assessment in the third trimester of pregnancy. Diagnostic uncertainty is common when women present to obstetric assessment units with one or more signs suggestive of preeclampsia. Women undergo a series of investigations, many of which are poor predictors of the need for delivery or a likely adverse outcome. In practice, obstetricians require a test that enables a woman to be triaged, to determine those that require increased surveillance, and those where the likelihood of needing delivery for preeclampsia within 14 days is very low and outpatient care may be appropriate. Such a test would enable development of safe clinical algorithms and avoid inappropriate intervention or unnecessary maternal anxiety.
Placental growth factor is an angiogenic factor synthesized by the trophoblast, a marker of associated placental dysfunction in preeclampsia, with known low plasma concentrations in the disease.16 Although combining PlGF with some of the other 46 biologically plausible biomarkers marginally improved the ROC area, the combinations added little to the diagnostic performance of a single biomarker alone. This important negative result demonstrates the diagnostic option of using a single biomarker (over and above a combination of biomarkers) in preterm preeclampsia. These findings are more marked in women presenting before 35 weeks of gestation and are similar, with lesser diagnostic efficacy, in women presenting between 35 0/7 and 36 6/7 weeks of gestation. This probably reflects the inclusion of women who meet the primary outcome definition (preeclampsia with delivery within 14 days) who were delivered routinely at 37 weeks of gestation following national guideline recommendations and not because of a clinician concern over a potential placentally mediated adverse event.
Strengths of this study include use of seven study sites and a large participant cohort encompassing a wide demographic and ethnic profile including women with underlying maternal disease. Plasma testing was carried out in a central laboratory ensuring that results were obtained with rigorous quality control. Progressive statistical analysis explored single biomarker predictive power and compared the effect of combining groups of markers or using biomarker ratios. A limitation was that test results were not validated in a repeat sample or by comparative testing at a second laboratory.
Previous studies have described other pathophysiologically relevant third-trimester markers, including soluble endoglin,17 or measurement of a ratio such as PlGF–sFlt-1.3,5 However, some of these studies have been small or from a single center, often using a case–control design. Such study design can result in overfitting and does not provide data indicative of how a biomarker may perform if introduced into clinical practice.
Systematic reviews have indicated that currently utilized tests such as proteinuria,8 transaminases,18 and uric acid19 are not good predictors of maternal or fetal complications in women with suspected preeclampsia. The lack of reliable diagnostic tests results in poorly targeted antenatal monitoring and hospitalisation.20 Development of an improved diagnostic test using pathophysiologically relevant biomarkers may have advantages over traditional diagnostic measures.21 A test performed at presentation that enables targeted surveillance for those at increased risk of maternal or fetal complications and provides appropriate reassurance to those who test negative has the potential to assist in the allocation of health resources.22 Further work is also needed on prognosis of multiorgan maternal complications in established preeclampsia.
Improved detection of placental disease remains a global health priority. Growing evidence suggests the use of angiogenic factors as biomarkers across a range of demographic settings in the prediction of preeclampsia,4 adverse outcome,23 and placentally related stillbirth.24 Previous work has shown that women with low or very low PlGF concentrations experienced adverse perinatal outcomes4 and our findings suggest that increased surveillance should be considered for these women.
We have previously reported that PlGF outperforms disease markers currently in use4; this study confirms that use of a single angiogenesis-related biomarker may be clinically useful as a diagnostic test without the need for combinations (which entail additional cost and complexity). Biomarkers such as PlGF can be analyzed quickly, representing a test that could aid risk stratification of women with suspected preterm preeclampsia. Further research, through randomized controlled trials, is essential to assess how these biomarker measurements can assist in determining (or refuting) diagnosis in preeclampsia and how this can improve outcomes for the mother and neonate through optimal tailored clinical management.
1. Steegers EA, von Dadelszen P, Duvekot JJ, Pijnenborg R. Pre-eclampsia. Lancet 2010;376:631–44.
2. Meis PJ, Goldenberg RL, Mercer BM, Iams JD, Moawad AH, Miodovnik M, et al. The preterm prediction study: risk factors for indicated preterm births. Maternal-Fetal Medicine Units Network of the National Institute of Child Health and Human Development. Am J Obstet Gynecol 1998;178:562–7.
3. Rana S, Powe CE, Salahuddin S, Verlohren S, Perschel FH, Levine RJ, et al. Angiogenic factors and the risk of adverse outcomes in women with suspected preeclampsia. Circulation 2012;125:911–9.
4. Chappell LC, Duckworth S, Seed PT, Griffin M, Myers J, Mackillop L, et al. Diagnostic accuracy of placental growth factor in women with suspected preeclampsia: a prospective multicenter study. Circulation 2013;128:2121–31.
5. Chaiworapongsa T, Romero R, Savasan ZA, Kusanovic JP, Ogge G, Soto E, et al. Maternal plasma concentrations of angiogenic/anti-angiogenic factors are of prognostic value in patients presenting to the obstetrical triage area with the suspicion of preeclampsia. J Matern Fetal Neonatal Med 2011;24:1187–207.
6. Redman CW, Sargent IL, Staff AC. IFPA Senior Award Lecture: making sense of pre-eclampsia—two placental causes of preeclampsia? Placenta 2014;35(suppl):S20–5.
7. Staff AC, Johnsen GM, Dechend R, Redman CW. Preeclampsia and uteroplacental acute atherosis: immune and inflammatory factors. J Reprod Immunol 2014;101–102:120–6.
8. Thangaratinam S, Coomarasamy A, O'Mahony F, Sharp S, Zamora J, Khan KS, et al. Estimation of proteinuria as a predictor of complications of pre-eclampsia: a systematic review. BMC Med 2009;7:10.
9. Menzies J, Magee LA, Macnab YC, Ansermino JM, Li J, Douglas MJ, et al. Current CHS and NHBPEP criteria for severe preeclampsia do not uniformly predict adverse maternal or perinatal outcomes. Hypertens Pregnancy 2007;26:447–62.
10. von Dadelszen P, Payne B, Li J, Ansermino JM, Broughton Pipkin F, Côté AM, et al. Prediction of adverse maternal outcomes in pre-eclampsia: development and validation of the fullPIERS model. Lancet 2011;377:219–27.
11. Diagnosis and management of preeclampsia and eclampsia. ACOG Practice Bulletin No. 33. American College of Obstetricians and Gynecologists. Obstet Gynecol 2002;99:159–67.
12. National Institute for Health and Clinical Excellence. Hypertension in pregnancy: the management of hypertensive disorders during pregnancy. NICE guidelines [CG107]. London (UK): NICE; 2010.
13. van Belle G, Fisher L, Heagerty P, Lumley T. Biostatistics: a methodology for the health sciences. 2nd ed. New York (NY): Wiley; 2004.
14. Costello AB, Osborne JW. Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis. Pract Assess Res Eval 2005;10:1–9.
15. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45.
16. Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, et al. Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med 2004;350:672–83.
17. Rana S, Cerdeira AS, Wenger J, Salahuddin S, Lim KH, Ralston SJ, et al. Plasma concentrations of soluble endoglin versus standard evaluation in patients with suspected preeclampsia. PLoS One 2012;7:e48259.
18. Thangaratinam S, Koopmans CM, Iyengar S, Zamora J, Ismail KM, Mol BW, et al. Accuracy of liver function tests for predicting adverse maternal and fetal outcomes in women with preeclampsia: a systematic review. Acta Obstet Gynecol Scand 2011;90:574–85.
19. Thangaratinam S, Ismail KM, Sharp S, Coomarasamy A, Khan KS; Tests in Prediction of Pre-eclampsia Severity review group. Accuracy of serum uric acid in predicting complications of pre-eclampsia: a systematic review. BJOG 2006;113:369–78.
20. Anumba DO, Lincoln K, Robson SC. Predictive value of clinical and laboratory indices at first assessment in women referred with suspected gestational hypertension. Hypertens Pregnancy 2010;29:163–79.
21. Staff AC, Benton SJ, von Dadelszen P, Roberts JM, Taylor RN, Powers RW, et al. Redefining preeclampsia using placenta-derived biomarkers. Hypertension 2013;61:932–42.
22. Schnettler WT, Dukhovny D, Wenger J, Salahuddin S, Ralston SJ, Rana S. Cost and resource implications with serum angiogenic factor estimation in the triage of pre-eclampsia. Br J Obstet Gynaecol 2013;120:1224–32.
23. Moore AG, Young H, Keller JM, Ojo LR, Yan J, Simas TA, et al. Angiogenic biomarkers for prediction of maternal and neonatal complications in suspected preeclampsia. J Matern Fetal Neonatal Med 2012;25:2651–7.
24. Chaiworapongsa T, Romero R, Korzeniewski SJ, Kusanovic JP, Soto E, Lam J, et al. Maternal plasma concentrations of angiogenic/antiangiogenic factors in the third trimester of pregnancy to identify the patient at risk for stillbirth at or near term and severe late preeclampsia. Am J Obstet Gynecol 2013;208:287.e1–15.