We conducted a nested case-control study of subjects from the Calcium for Preeclampsia Prevention (CPEP) Study. CPEP was a randomized, double-blind clinical trial performed by the National Institute of Child Health and Human Development during 1992–1996 to evaluate the effects of daily supplementation with calcium or placebo on the incidence and severity of preeclampsia. Healthy, normotensive, nulliparous women with singleton pregnancies were enrolled between 13 and 21 weeks gestation at 5 participating U.S. medical centers and followed until 24 hours postpartum, using a common protocol and identical data collection forms. Gestational age was determined by ultrasound examination. Serum specimens were requested from participants at 13–21 wks, at 26–29 wks, at 36 wks, and when hypertension or proteinuria was noted. Many women actually provided samples outside the requested gestational periods; all specimens were included.
Enrollment details for the original study are described elsewhere (Levine et al., 1996
). Of 4589 CPEP participants, we excluded 253 who were lost to follow-up, 21 whose pregnancy terminated prior to 20 weeks, 13 who were missing maternal or perinatal outcome data, 4 without smoking information, 9 with hypertension not verified by team chart review, 32 with stillbirths, and one woman whose infant had a chromosomal abnormality, leaving 4257 women with adequate information and a live birth. Of these, 102 were excluded because they lacked a baseline serum specimen; another 524 were excluded because their baseline specimens had label dates more than 2 days following the date recorded by the research nurse on the laboratory specimen forms.
Of the remaining 3631 women, 2469 had remained normotensive throughout pregnancy and had delivered an appropriate- or large-for-gestational age infant. We randomly selected 120 of these women to serve as a pool from which to select normal controls. We identified all 32 women within the CPEP cohort who experienced abruptio placentae (including women with live births and stillbirths), but excluded one woman since her only serum specimen could not be located in the specimen repository. Ten of the 31 women with abruption developed hypertension during their pregnancies – seven, preeclampsia, and three, gestational hypertension. From the pool of normal controls we selected one control subject for each abruption case, matched for clinical center, current smoking status (current smoker or quit after the last menstrual period vs. never smoked or quit before the last menstrual period), and specimen collection pattern. Matching for specimen collection pattern meant that the control must have had at least as many specimens as the case in the following gestational age intervals: <20 weeks, 21–32 weeks, and ≥33 weeks. If more than one control met all matching criteria for a given case, a random selection was made. Once a control was matched to a case, it could not be used as a match for any other case. Because controls had to be matched on several factors and had to have at least as many samples as the cases, we were able to match only one control to each case. All serum specimens collected from subjects prior to the onset of hypertension or proteinuria and before labor or delivery were included in the present study.
Abruptio placentae was clinically diagnosed by physicians participating in the CPEP study, according to the study protocol and manual of operations, and was defined as the presence at delivery after 24 weeks gestation of a retroplacental blood clot and/or antecedent vaginal hemorrhage not associated with vasa previa, placenta previa, or uterine rupture. Evidence of fetal distress and/or sonographic evidence of abruption were present in some cases, but these findings were not required for diagnosis. Hypertension was defined as a diastolic blood pressure of at least 90 mm Hg on two occasions 4 to 168 hours apart in a woman with blood pressure <135/85 prior to CPEP enrollment at 13–21 weeks of gestation. Preeclampsia required, in addition, proteinuria of at least 1+ (30 mg per deciliter) on dipstick testing, each on two occasions 4 to 168 hours apart, a single dipstick indicating ≥2+ proteinuria, a protein / creatinine ratio of 0.35 or greater, or a 24-hour urine collection with >300 mg protein. Gestational hypertension was the de novo onset of hypertension after 20 weeks of gestation in the absence of proteinuria. A small for gestational age (SGA) infant was defined as an infant whose birth weight was below the 10th
percentile according to U.S. tables of birth weight for gestational age that account for race, parity, and the sex of the infant (Zhang & Bowes, Jr. 1995
Never-thawed serum specimens were randomly allocated to batches and analyzed, using an enzyme-linked immunosorbent assay (ELISA) for endoglin according to the manufacturer's instructions (R & D Systems, Minneapolis, MN). The measured interassay coefficient of variation in our laboratory was 12%. All samples were run in duplicate by technicians blinded to pregnancy outcome, and the mean values of the duplicate samples were reported.
Chi-square or t-tests were used in analyses of maternal or infant characteristics to compare categorical or continuous variables, respectively. Although arithmetic mean concentrations of sEng are reported in the text and tables, statistical tests for differences in levels of this factor were performed after logarithmic transformation. Because the number of samples in each group was relatively small, we performed both a matched and an unmatched analysis. In the matched analysis, the unit of analysis was the case-control pair; as such, in any gestational age interval in which one member of the pair had no serum specimen (for example, because she had delivered prior to that interval), no sample(s) from the other member of the pair was/were considered in the analysis. In the unmatched analysis, the unit of analysis was an individual specimen; specimens from each subject were retained in the analysis regardless of the availability of a specimen from the same gestational interval from the matched subject. Since the results of both these analytic methods were similar; the unmatched analysis is presented because the numbers were greater, giving it greater statistical power. All P values are 2-sided; a value of P<0.05 was considered significant. Analyses were conducted using SAS v 9.1 (SAS Institute, Cary, NC).
The Office of Human Subjects Research at the National Institutes of Health granted an exemption from institutional review board approval because data and specimens could not be linked to identifiable women.