Patient population

We performed a nested case-control study using the samples in the Massachusetts General Hospital Obstetrical Maternal Study (MOMS). The MOMS cohort was established in 1998 for the prospective study of risk factors for adverse outcomes that occur later in pregnancy. Serum samples were collected at the first prenatal visit (11-13 weeks of gestation) and again in the second trimester (17-20 weeks of gestation). For this study, women who had their first prenatal visit from November 2001 through May 2005, who enrolled in the MOMS cohort at or before 12 weeks of gestation, and who delivered after 20 weeks of gestation were eligible. All subjects provided written informed consent, and this study was approved by the institutional review board of the Massachusetts General Hospital.

An electronic medical record was used to record demographic and clinical data, which included gestational age, which was estimated from the last menstrual period and verified by ultrasound scan. All subjects had no history of preexisting hypertension (blood pressure >140/90 mm Hg or need for antihypertensive medications before pregnancy or <20 weeks of gestation), remained nonproteinuric throughout pregnancy, initiated and completed their prenatal care and pregnancy within the Massachusetts General Hospital network, and delivered a singleton live infant. Women with a history of diabetes mellitus or thyroid, liver, or chronic renal disease were excluded. Outcomes were determined prospectively and verified by examination of medical records. SGA newborn infants were defined by birthweight <10th percentile of the US population that was specific for gestational age and sex. Control subjects were selected randomly from women who participated in the MOMS cohort within the same time period as the cases and delivered appropriate for gestational age infants.

Assays

Samples were stored on ice for <3 hours and frozen at −80°C for future analysis. We measured serum sFas and PlGF levels using commercial colorimetric sandwich enzyme-linked immunosorbent assay kits (ELISA; R&D Systems, Minneapolis, MN). The intraassay coefficients of variation for sFas and PlGF were 4.6% and 7.0%, respectively. The interassay coefficients of variation for sFas and PlGF were 2.9 % and 11.8%, respectively. The laboratory was blinded to case status, and all samples were ordered randomly. The sFas ELISA measures total free sFas because there was no interference when we added exogenous Fas Ligand. The PlGF ELISA detects free PLGF only.

Statistical analysis

All analyses were conducted with the Statistical Analysis System (SAS 9.2; SAS Institute Inc, Cary, NC). Baseline characteristics are reported as proportions or the mean and SD, as appropriate. Levels of sFas and PlGF are reported as means and SE. Because of small sample sizes and data that are not normally distributed, the robust sandwich estimator was used to calculate the variance and probability values. Comparisons between groups were conducted with a t test for continuous variables and a χ ² test or Fisher’s exact test for categoric variables. Multivariate linear regression was used to adjust first-trimester mean sFas and PlGF levels for significantly different baseline covariates. Because of the strong correlation between gravidity and parity, gravidity alone was included as a covariate. Adjusted means could not be calculated for the second trimester, given the small number of available samples and the limited ability to detect statistically significant differences in baseline covariates. The change in crude means from the first to second trimester was compared within the SGA and control groups. In each case, there were 1 or 2 women who contributed samples in both the first and second trimester; for these women, only the second trimester samples were used for the comparisons over time. A probability value < .05 was considered statistically significant, and all probability values were generated from 2-sided tests.

The sample size for this nested case-control study was constrained by the quantity of samples available from the MOMS cohort. We calculated the power for this study using all of the available samples, an α score of .05, and previously published data. In addition, the power calculation includes an inflation factor of 15% to account for the expected nonnormal distribution of the data. Based on first trimester sFas levels in normal pregnancies as reported by Malamitsi-Puchner et al, our sample size allowed for 99% power to detect a 1 SD (750 pg/mL) difference in the first trimester and 49% power to detect a difference in the second trimester. Based on the difference in first-trimester PlGF values between normal and preeclamptic pregnancies (40 pg/mL), our sample size allowed for 99% power to detect the same difference in the first trimester and 12% power to detect this difference in the second trimester. Even though our second-trimester sample size was underpowered, we decided to proceed with the measurement of sFas and PlGF in these samples because our result could serve as preliminary data that can be used for sample-size calculations for future studies. With the number of second-trimester samples available, we had 80% power to detect an approximately 25% difference in levels of sFas and a 35% difference in PlGF.