The combination of alpha-fetoprotein (AFP), human chorionic gonadotropin (hCG), unconjugated estriol (uE3) and inhibin A, known as the quad screen or quadruple test, is the most effective multiple-marker screening test for Down syndrome in the second trimester. This approach yields an 81% detection rate at a false positive rate of 5% (Malone FD, Canick JA, Ball RH, Nyberg DA, Comstock CH, Bukowski R, et al. First and Second Trimester Evaluation of Risk for Fetal Aneuploidy [FASTER]: principal results of the NICHD Multicenter Down Syndrome Screening Study. Submitted for publication, 2004). Elevated AFP levels can be used to detect approximately 85% of all open neural tube defects.1
Maternal serum levels of AFP, hCG, uE3, and inhibin A have been shown to be associated with adverse obstetric outcomes in the absence of aneuploidy or neural tube defects.2–13 However, the majority of the previous work is comprised of small, retrospective, case-controlled cohort studies of only one or two of the quad screen markers. Furthermore, there are few previously published large studies that evaluated combinations of quad screen markers. The purpose of our study was to estimate the predictive relationship between second-trimester levels of maternal serum AFP, hCG, uE3, and inhibin A and obstetrical complications by using a large, prospectively collected database. The size of our screened patient population allowed us to evaluate the performance characteristics of all 4 of these markers, both individually and in combination.
MATERIALS AND METHODS
All women participating in this study were enrolled in the FASTER trial (First and Second Trimester Evaluation of Risk), a multicenter study sponsored by the National Institutes of Health, the goal of which was to compare the performance of first- and second-trimester screening methods for Down syndrome. After obtaining institutional review board–approved informed consent, all potential subjects underwent an ultrasound examination in one of the 14 study centers. Women (aged 16 or greater) confirmed to have a singleton gestation between 10 3/7 and 13 6/7 weeks of gestation, as defined by the Hadlock criteria,14 were eligible to be included in the FASTER trial. If they elected to participate, nuchal translucency was measured and serum was obtained for measurement of pregnancy-associated plasma protein A (PAPP-A) and free β-hCG. Women whose fetuses were diagnosed with anencephaly or a septated cystic hygroma at this initial ultrasound examination were excluded. Blood samples for measurement of maternal serum AFP, hCG, uE3, and inhibin A were obtained between 15 and 18 6/7 weeks of gestation. All blood specimens were centrifuged and sent overnight to Women and Infants Hospital Laboratory, Providence, Rhode Island. Alpha-fetoprotein and hCG levels were measured with chemiluminescent immunoassays (Diagnostic Products Corporation, Los Angeles, CA). Unconjugated E3 levels were measured with a radioimmunoassay (Diagnostic Systems Labs, Webster, TX), and inhibin A levels were measured with an enzyme-linked immunosorbent assay (ELISA) (Serotec, UK and Diagnostic Systems Labs, Webster, TX).
The analysis presented here includes data from 33,145 women enrolled in the FASTER trial between 1999 and 2002. We have excluded women whose fetus had a chromosomal or structural abnormality (n = 221 and 462, respectively). Relevant patient history, demographic data, and obstetric history were collected at the time of enrollment to the FASTER trial. Complete pregnancy and pediatric outcome data were available on all participants in this analysis. A software tracking program, with up to 10 contact options per patient, was used to optimize our data collection. A perinatologist and a pediatric geneticist reviewed the maternal and pediatric medical records for all subjects who had abnormal first- or second-trimester screening or adverse pediatric outcome. Finally, for quality control, the same individuals reviewed the records of 10% of normal subjects selected randomly from each trial site.
Adverse obstetric outcomes examined included spontaneous loss at 24 weeks or less, fetal loss at more than 24 weeks, preterm delivery at 32 weeks of gestation or less, preeclampsia, and birth weight less than or equal to the 5th and less than the 10th percentile for gestational age. Preeclampsia was defined as gestational hypertension in the setting of significant proteinuria. Significant proteinuria was defined as a minimum of 300 mg/24 hours or 0.1g/L (more than 2+ on a dipstick) in at least 2 random samples collected 6 or more hours apart. Birth weight at less than the 10th and less than or equal to the 5th percentile was established by using a growth curve described by Alexander et al.15
The first step in the analysis involved the creation of indicators to reflect abnormal second-trimester serum levels. For each pairwise combination of outcome and marker, we performed a receiver operating characteristic analysis examining the tradeoff between sensitivity and specificity at every cutoff of the given marker. Using this information in conjunction with clinical judgment, we determined the most appropriate thresholds. We classified values as abnormally low if they were at or below 0.5 multiples of the median (MoM) and as abnormally high if they were at or above 2.0 MoMs. We specifically analyzed the effects of AFP, hCG, and inhibin A at or above 2.0 MoMs, and the effect of uE3 at or below 0.5 MoMs. The incidence of adverse obstetric outcomes, considered separately, was estimated for each marker, and we evaluated the effect of the marker, independent of other markers on each outcome by using analysis of variance.
We then assessed the impact of having single or multiple abnormal markers simultaneously. We were restricted in the specific combinations that could be examined due to low representation in many groups. Specifically, we classified women as having no abnormal markers, exactly 1 (AFP ≥ 2.0, hCG ≥ 2.0, uE3 ≤ 0.5, or inhibin A ≥ 2.0 MoMs, respectively), exactly 2, or 3 or more abnormal markers. We estimated the effects of abnormal marker combinations on each outcome using multiple logistic regression analysis. Crude and adjusted odds ratios, along with 95% confidence intervals, were generated for each multiple marker group. Where possible, we then explored which specific combinations of abnormal markers were associated with most elevated risks.
Confounding variables for the adjusted models were selected in a 2-stage process. Potential confounders were initially considered based on statistical significance (P < .05) at the bivariate level with either markers or outcomes. Tests included analysis of variance for continuous confounders and Fisher exact test for categorical confounders. Final confounders for the adjusted models were then selected by using backwards elimination stepwise logistic regression analysis, retaining only those variables that were significant at a level of .05.
The confounders controlled for in the final models included maternal age and weight, parity, prior preterm pregnancy, current threatened abortion, smoking in the current pregnancy, diabetes, and use of antihypertensive medications at the time of enrollment. Preeclampsia and low birth weight were also controlled for in the models with preterm birth or fetal demise at 24 or more weeks as an outcome, and preeclampsia was controlled for in the models with low birth weight as an outcome. Models were also investigated excluding preeclampsia and low birth weight cases to isolate a spontaneous preterm birth or fetal demise at 24 weeks or more.
Performance characteristics, including estimates and 95% confidence intervals for sensitivity, specificity, false positive rate, positive and negative predictive value, and positive and negative likelihood ratios, were then generated for women with 2 or more abnormal markers. All analyses were conducted in SAS 8.2 (SAS Institute Inc, Cary, NC). To adjust for the 6 outcomes tested in the primary analysis, we used P < .008 as our significance criterion, based on a Bonferroni correction for 6 comparisons (0.5/6).
The mean maternal age at the estimated date of delivery (determined at the first-trimester visit) was 30.2 ± 5.71 years, with a range of 16–53 years. There were 15,127 (45.7%) nulliparous women. The majority of the women were white (68.7%) or Hispanic (21.7%). African Americans represented 4.7% of the study population, 4.0% were Asian/Pacific Islanders, and the remaining 0.9% identified themselves as “other.” The distribution of abnormal markers in the study sample is shown in Table 1. The overall incidence of adverse obstetric outcomes is presented in Table 2. Fewer than 4% of the study subjects had a fetus with a birth weight less than the 5th percentile for gestational age, and fewer than 3% had preeclampsia. Table 2 also provides the incidence of the adverse obstetric outcome for each marker and the crude results evaluating the effect of the marker on each outcome independent of the other markers.
When examining the statistical relationships of each marker without accounting for the impact of other markers that may be abnormal, each of the 4 markers was significantly associated with a number of adverse pregnancy outcomes. Elevated hCG and inhibin A each showed a significant association with every adverse outcome we evaluated (Table 2). Elevated AFP was significantly associated with every adverse outcome except for fetal demise at 24 weeks or more (P = .047). Low uE3 values were significantly associated with birth weight less than or equal to the 5th and less than the 10th percentiles and fetal loss at less than 24 weeks. However, and very importantly, when we examined the relationships of isolated abnormalities of a single marker (ie, all other markers normal), there were far fewer statistically significant associations (Table 3). This analysis dramatically altered the statistical relationships. For example, as seen in this table, elevated hCG levels, when occurring alone, were not significantly associated with any of the adverse perinatal outcomes we studied. When considered alone, elevated AFP and low uE3 were both significantly associated with birth weight less than or equal to the 5th and less than the 10th percentiles and with fetal loss at less than 24 weeks. An isolated elevation of inhibin A was significantly associated with preterm birth at 32 weeks or less, birth weight less than or equal to the 5th and less than the 10th percentiles, and preeclampsia, and was marginally associated with fetal demise at 24 weeks or more (P = .040).
For all of the outcomes, combinations of at least 2 markers were more strongly associated with all of the adverse outcomes than any single isolated marker (Table 3). As the number of abnormal markers increased, the association with adverse outcomes became stronger.
We reanalyzed our data excluding preeclamptics because preeclampsia can itself cause fetal death, low birth weight, and preterm birth for maternal or fetal indications. The same relationships described in Table 3 held true for low birth weight (≤ 5th and < 10th percentiles). The same relationships also maintained for the outcome fetal demise at 24 weeks or more when preeclamptic and low birth weight subjects were excluded. However, isolated elevated inhibin A levels were no longer significantly associated with preterm birth at 32 weeks or less when we excluded preeclampsia and low birth weight cases (odds ratio [OR] 1.69, 95% confidence interval [CI] 0.79–3.65, P = .179). Isolated elevated AFP was only marginally associated with preterm birth when preeclamptics and low birth weight cases were excluded (OR 2.58, 95% CI 1.19–5.60, P = .017). Results remained consistent when 2 or more abnormal markers were considered.
The adjusted odds ratios for specific combinations of markers with sufficient representation to analyze are noted in Table 4. Elevated inhibin A in conjunction with either elevated AFP or elevated hCG, or both, was significantly associated with preterm birth at 32 weeks or less, birth weight less than or equal to the 5th and less than the 10th percentiles, preeclampsia, and fetal demise at 24 weeks or more. Preeclampsia was also significantly associated with the combination of elevated AFP and hCG. Spontaneous loss at less than 24 weeks was significantly associated with the combination of elevated AFP and inhibin A.
Several relationships did not occur sufficiently frequently to be well described by our statistical model. For example, when all markers were normal, the incidence of birth weight of 5th percentile or lower was 3.47%. This number rose to 7.7% when uE3 was low in conjunction with elevated hCG, and to 22.2% when low uE3 was seen in conjunction with high AFP or inhibin A. Low uE3 levels in combination with other abnormal markers also impacted the incidence of birth weight less than 10th percentile (8.48% for normal markers versus 11.5%, 22.2%, and 27.8% for low uE3 paired with high hCG, AFP, and inhibin A, respectively, and 41.8% when combined with elevation of both hCG and inhibin A).
As mentioned, there was a marked increase in the incidence of adverse obstetric outcome in the presence of 2 or more abnormal markers. Table 5 shows estimates and 95% confidence intervals for the sensitivity, predictive values, and likelihood ratios for the test of 2 or more abnormal markers versus single or no abnormal markers for each of the adverse outcomes. Sensitivity, predictive values, and likelihood ratios, however, were fairly low for all adverse pregnancy outcomes. Nevertheless, the performance characteristics for the test of 2 or more abnormal markers performs better than the single-marker test for every outcome.
We present data from a large, prospective study of the 4 components of the quad screen and the association with adverse pregnancy outcomes. When we looked at individual markers without taking into account the potential effect of other abnormal markers, we found the same associations previously reported by other authors. However, our finding changed significantly when we used a new and novel methodology that allowed us to account for each marker in isolation from all others, and also combinations of all 4 markers. This technique allowed us to better understand the relationship between each individual marker and adverse outcomes. Additionally, we were able to evaluate combinations of adverse markers to hone our ability to identify associations that would predict adverse outcomes. For example, where previous studies that we cited earlier showed an association between elevated hCG and a number of adverse outcomes, our new methodology showed that there was no such association when effects of additional abnormal markers were accounted for in the analysis. We observed similar findings for the other markers.
In addition to presenting data for isolated abnormal markers, it is optimal to present risks associated with specific combinations and numbers of abnormal markers. This provides the clinician with more specific information regarding risks for specific adverse outcomes. Although there are a number of significant associations between an isolated abnormal marker and adverse outcomes, the risk for an adverse outcome increases with the number of abnormal markers.
Although many of the associations between the quad screen markers and adverse obstetric outcomes are statistically highly significant in this as well as previous studies,2–13 the sensitivity and positive predictive values for the individual outcomes are relatively low. The presence of 2 or more abnormal markers, although strongly associated with a number of adverse outcomes, was only a modest predictor of these outcomes in our population and does not support the use of quad screen markers to screen for adverse pregnancy outcomes in a general population. We had a relatively low incidence of adverse outcome in our study sample. Although the FASTER trial patient population was unselected, meaning that patients were not excluded based on any potentially confounding factors such as race, parity, marital status, education, smoking, and pre-existing medical conditions, there may have been significant provider or patient self-selection. In addition, patients could only enroll in the study if they started antepartum care in the first trimester. It is possible that the markers would perform better in a high-risk population.
If we are to be successful in using these markers to identify pregnancies at high risk for adverse outcomes, it is likely that we will need to develop a panel that includes additional serum or other types of markers. We have previously reported that low levels of PAPP-A in the first trimester are associated with pregnancy loss at or before 24 weeks, preterm birth, preeclampsia, and low birth weight.16 Similarly, abnormal second-trimester uterine artery Doppler has been associated with an increased risk of preeclampsia and intrauterine growth restriction.17–20 Perhaps the use of PAPP-A, Doppler studies, or other factors, such as maternal demographic characteristics, will enhance the screening efficacy of the combined quad screen markers. We plan to perform future investigations to evaluate these and other potential markers.
1. American College of Obstetricians and Gynecologists. Maternal serum screening. ACOG Educational Bulletin 228. Washington, DC: ACOG; 1996.
2. Yaron Y, Cherry M, Kramer RL, O'Brien JE, Hallak M, Johnson MP, et al. Second-trimester maternal serum marker screening: Maternal serum α-fetoprotein, β-human chorionic gonadotropin, estriol, and their various combinations as predictors of pregnancy outcome. Am J Obstet Gynecol 1999;181:968–74.
3. Chandra S, Scott H, Dodds L, Watts C, Blight C, Van den Hof M. Unexplained elevated maternal serum α-fetoprotein and/or human chorionic gonadotropin and the risk of adverse outcomes. Am J Obstet Gynecol 2003;189:775–81.
4. Ilagan JG, Stamilio DM, Ural SH, Macones GA, Odibo AO. Abnormal multiple marker screens are associated with adverse perinatal outcomes in cases of intrauterine growth restriction. Am J Obstet Gynecol 2004;191:1465–9.
5. Aquilina J, Maplethorpe R, Ellis P, Harrington K. Correlation between second trimester maternal serum inhibin-A and human chorionic gonadotrophin for the prediction of pre-eclampsia. Placenta 2000;21:487–92.
6. Jauniaux E, Gulbis B, Tunkel S, Ramsay B, Campbell S, Meuris S. Maternal serum testing for alpha-fetoprotein and human chorionic gonadotropin in high-risk pregnancies. Prenat Diagn 1996;16:1129–35.
7. Wenstrom KD, Owen J, Boots LR, DuBard MB. Elevated second-trimester human chorionic gonadotropin levels in association with poor pregnancy outcome. Am J Obstet Gynecol 1994;171:1038–41.
8. Lepage N, Chitayat D, Kingdom J, Huang T. Association between second-trimester isolated high maternal serum maternal serum human chorionic gonadotropin levels and obstetric complications in singleton and twin pregnancies. Am J Obstet Gynecol 2003;188:1354–9.
9. Lambert-Messerlian GM, Silver HM, Petraglia F, Luisi S, Pezzanti I, Maybruck WM, et al. Second-trimester levels of maternal serum human chorionic gonadotropin and inhibin A as predictors of pre-eclampsia in the third trimester of pregnancy. J Soc Gynecol Investig 2000;7:170–4.
10. Kowalczyk TD, Cabaniss ML, Cusmano L. Association of low unconjugated estriol in the second trimester and adverse pregnancy outcome. Obstet Gynecol 1998;91:396–400.
11. Williams MA, Hickok DE, Zingheim RW, Luthy DA, Kimelman J, Nyberg DA, et al. Elevated maternal serum α-fetoprotein levels and midtrimester placental abnormalities in relation to subsequent adverse pregnancy outcomes. Am J Obstet Gynecol 1992;167:1032–7.
12. Krause TG, Christens P, Wohlfahrt J, Lei U, Westergaard T, Norgaard-Pedersen B, et al. Second-trimester maternal serum alpha-fetoprotein and risk of adverse pregnancy outcome. Obstet Gynecol 2001;97:277–82.
13. Butler EL, Dashe JS, Ramus RM. Association between maternal serum alpha-fetoprotein and adverse outcomes in pregnancies with placenta previa. Obstet Gynecol 2001;97:35–8.
14. Hadlock FP, Shah YP, Kanon DJ, Lindsey JV. Fetal crown rump length: reevaluation of relation to menstrual age (5–18 weeks) with high-resolution real-time ultrasound. Radiology 1992;182:501–5.
15. Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United States national reference for fetal growth. Obstet Gynecol 1996;87:163–8.
16. Dugoff L, Hobbins JC, Malone FD, Porter TF, Luthy D, Comstock CH, et al. First-trimester maternal serum PAPP-A and free-beta subunit human chorionic gonadotropin concentrations and nuchal translucency are associated with obstetric complications: a population-based screening study (The FASTER Trial). Am J Obstet Gynecol 2004;191:1446–51.
17. Bewley S, Cooper D, Campbell S. Doppler investigation of uteroplacental blood flow resistance in the second trimester: a screening study for pre-eclampsia and intrauterine growth retardation. Br J Obstet Gynaecol 1991;98:871–9.
18. North RA, Ferrier C, Long D, Townend K, Kincaid-Smith P. Uterine artery Doppler flow velocity waveforms in the second trimester for the prediction of pre-eclampsia and fetal growth retardation. Obstet Gynecol 1994;83:378–86.
19. Albaiges G, Missfelder-Lobos H, Parra M, Lees C, Cooper D, Nicolaides KH. Comparison of color Doppler uterine artery indices in a population at high risk for adverse outcome at 24 weeks' gestation. Ultrasound Obstet Gynecol 2003;21:170–3.
20. Papageorghiou AT, Yu CK, Bindra R, Pandis G, Nicolaides KH; Fetal Medicine Foundation Second Trimester Screening Group. Multicenter screening for pre-eclampsia and fetal growth restriction by transvaginal uterine artery Doppler at 23 weeks of gestation. Ultrasound Obstet Gynecol 2001;18:441–9.
The members of the FASTER Trial Research Consortium include K. Welch, R. Denchy (Columbia University); R. Ball, L. Cannon, K. Nelson, C. Loucks, A. Yoshimura (University of Utah); D. Nyberg, S. Coe (Swedish Hospital); D. Schmidt, J. Esler (William Beaumont Medical Center); G. Hankins, R. Bukowski, J. Lee (University of Texas Medical Branch); R. Berkowitz, Y. Kharbutli (Mount Sinai School of Medicine); S. Gross, S. Carter (Albert Einstein College of Medicine); L. Schultz (University of Colorado Health Sciences Center); D. Bianchi, B. MacKinnon, B. Isquith, B. Berlin (Tufts University); M. Paidas, J. Borsuk (New York University); C. Duquette (Brown University); R. Baughman (UNC Medical Center); K. Dukes, D. Emig, T. Tripp, and P. Folan (DM-STAT).
© 2005 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.