Case identification

All early-onset preeclampsia cases (n = 197) evaluated in this study were identified by the California Very Preterm Birth Study from prenatal screening records and validated via chart review at the hospital of delivery. Detailed study design and methods have been published elsewhere. Briefly, very preterm births of white, Hispanic, or black origin were identified from a population-based cohort created through record linkage of 346,456 mother-baby pairs with stored maternal serum specimens linked to certificates of live births in California from January 2000 through April 2007. White and Hispanic births were defined as both parents being non-Hispanic white and Mexican Hispanic white, respectively, based on birth records. Mexican Hispanic white referred to self-reported white with Hispanic ethnicity origin of Mexican, Mexican-American, or Chicano. Black births were defined as black mothers of any ethnicity, regardless of father’s race-ethnicity. All of these women participated in routine prenatal screening for aneuploidies and neural tube defects between 15-20 weeks through the Genetic Disease Screening Program within the California Department of Public Health and had serum banked after screening. Study samples were limited to singleton live births delivered in 5 southern California counties (Los Angeles, Orange, Riverside, San Bernardino, and San Diego). Within this large cohort, the medical records of 1117 chronologically identified very preterm births of <32 weeks gestation (429 whites, 440 Hispanics, and 248 blacks) were abstracted to distinguish very preterm births with known causes including preeclampsia, which formed the case group for this study. Cases with a birthweight >2500 g were excluded based on suspected misdating. Pregnancies with structural abnormalities reported on delivery records or identified by the California Birth Defects Monitoring Program registry were also excluded.

Preeclampsia was defined as hypertension accompanied by proteinuria with onset >20 weeks’ gestation in the California Very Preterm Birth Study. Hypertension was defined as blood pressure ≥140 mm Hg systolic or ≥90 mm Hg diastolic on 2 occasions 6 hours apart. Proteinuria was defined as ≥0.3 g of protein in a 24-hour urine collection or ≥1+ dipstick on a urine test on 2 occasions 6 hours apart. Early-onset preeclampsia was defined as preeclampsia with delivery <32 weeks’ gestation. Gestational age estimates for cases were mostly based on ultrasound (61.5%) and date of last menstrual period (29.5%) from the California Prenatal Screening Program with a small part (9.0%) based on maternal hospital chart review.

Hospital chart review identified 207 cases of early-onset preeclampsia with a maternal midtrimester specimen available for laboratory testing. Ten of the 207 cases were excluded due to maternal specimens collected beyond the gestational window of 15-20 weeks. The final sample included 90 white, 67 Hispanic, and 40 black cases. Details of the sample selection are listed in the Figure .

Identification flowchart of early-onset preeclampsia cases and controls

Step-by-step sample selection. Exclusion rates in each step vary between race-ethnicity groups due to different population distributions and exclusion details by race-ethnicity groups.

Yang et al. Racial-ethnic differences in angiogenic and antiangiogenic factors. Am J Obstet Gynecol 2016 .

Control identification

Controls were noncases selected at an original sample size of ≥10 times the number of cases before specimen pulling using simple random sampling within the same race-ethnicity-specific singleton live-birth cohort as cases. Births with adverse outcomes such as late-onset preeclampsia, preterm birth, and low birthweight were also eligible controls. Controls of white, Hispanic, and black origin were drawn separately for each racial-ethnic group. For women with multiple pregnancies during the study period, only the first pregnancy was chosen. Control births without banked serum specimens or with structural malformations according to the California Birth Defects Monitoring Program registry were excluded. The final sample included 916 white, 709 Hispanic, and 738 black controls ( Figure ). Gestational age estimates for controls were based on ultrasound (55.8%), date of last menstrual period (34.9%), and physical exam (9.3%) from the California Prenatal Screening Program.

Biological specimens

Banked maternal blood specimens were retrieved from the California Biobank Program. As part of the state prenatal screening program, obstetrical care service providers at doctors’ offices or drawing stations collected venous blood from pregnant women who voluntarily enrolled at 15-20 weeks’ gestation. The blood, which was collected in serum separator tubes, was sent to 1 of the 7 newborn and prenatal screening laboratories by mail under ambient conditions. Specimens arrived for triple-marker prenatal testing within 7 days of collection (median time 3 days). After routine testing, leftover specimens were refrigerated and then processed for long-term storage at –20°C. Women consented to the routine prenatal screening test in writing and were notified that the blood becomes the property of the state, which allows for a variety of approved research uses of the leftover blood under the highest level of security and confidentiality standards. While women have the option to request the specimen not be used for research purposes, very few assert this right. Specimen vials from cases and controls were intermingled, with vial positions randomly assigned during shipping and laboratory testing. Laboratory personnel were unaware of subjects’ case-control status.

Laboratory methods

Maternal serum levels of PlGF, sEng, and sVEGFR-1 were determined by immunoassays (R&D Systems Inc, Minneapolis, MN) utilizing the quantitative sandwich enzyme immunoassay technique. Laboratory testing was conducted at the research core laboratory of Perinatology Research Branch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Department of Health and Human Services. Standard serum specimens were used in each laboratory run and analyte value was determined by interpolation from standard curves. The assays utilized 120, 25, and 25 μL of serum for the PlGF, sEng, and VEGFR-1 assays, respectively. The interassay and intraassay coefficients of variation obtained in the laboratory were 4.6% and 3.6% for PlGF, 5.3% and 4.3% for sEng, and 2.8% and 2.7% for sVEGFR-1, respectively. The sensitivity of the assays was 5.7 pg/mL for PlGF, 40.0 pg/mL for sEng, and 12.9 pg/mL for sVEGFR-1.

Data analysis

The laboratory levels of PlGF, sEng, and sVEGFR-1 were linked to birth records, prenatal screening program data, and maternal hospital chart data (cases only) for analysis. Maternal characteristics were compared between cases and controls. Gestational weeks at delivery and at preeclampsia diagnosis for cases were compared among the 3 racial-ethnic groups. The χ ² test and Student t test were used to estimate the statistical significance of the comparisons for categorical and continuous variables, respectively. Natural logarithmic transformed circulating levels of PlGF, sEng, and sVEGFR-1 were normally distributed in cases and controls. Means (log pg/mL) and SE were calculated for cases and controls of each racial-ethnic group and were adjusted by maternal age, education, parity, specimen storage year, gestational age at specimen collection, and maternal weight at or near specimen collection. The adjusted logarithmic means were compared using Tukey honestly significant difference test between cases and controls in each racial-ethnic group and between racial-ethnic groups among cases and controls, respectively. Odds ratios (OR) and 95% confidence intervals (CI) derived from binomial logistic regressions were used to measure effect magnitude of analyte levels in association with case-control status. The logarithmic transformed levels (log pg/mL) of the 3 analytes were used in the logistic regression to reduce both variance and outlier influence. Race-ethnicity is a significant effect modifier for PlGF and sEng in association with early-onset preeclampsia and therefore association results were presented separately for each race-ethnicity group. Maternal age, education, parity, specimen storage year, gestational weeks at specimen collection, and maternal weight at specimen collection were included as potential confounders. Standard Z test was conducted to compare OR between racial-ethnic groups. Given the above-mentioned final sample sizes, this study had adequate power of ≥80% to detect association of moderate or higher effect estimates (OR ≥2.6) between cases and controls in each racial-ethnic group under a logistic model with a 2-sided test.

All of the statistical analyses were conducted using software (SAS, Version 9.2 for Windows; SAS Institute, Cary, NC). The study protocol was approved by the California Health and Human Services Agency Committee for the Protection of Human Subjects (project no. 10-10-52).

Case identification

All early-onset preeclampsia cases (n = 197) evaluated in this study were identified by the California Very Preterm Birth Study from prenatal screening records and validated via chart review at the hospital of delivery. Detailed study design and methods have been published elsewhere. Briefly, very preterm births of white, Hispanic, or black origin were identified from a population-based cohort created through record linkage of 346,456 mother-baby pairs with stored maternal serum specimens linked to certificates of live births in California from January 2000 through April 2007. White and Hispanic births were defined as both parents being non-Hispanic white and Mexican Hispanic white, respectively, based on birth records. Mexican Hispanic white referred to self-reported white with Hispanic ethnicity origin of Mexican, Mexican-American, or Chicano. Black births were defined as black mothers of any ethnicity, regardless of father’s race-ethnicity. All of these women participated in routine prenatal screening for aneuploidies and neural tube defects between 15-20 weeks through the Genetic Disease Screening Program within the California Department of Public Health and had serum banked after screening. Study samples were limited to singleton live births delivered in 5 southern California counties (Los Angeles, Orange, Riverside, San Bernardino, and San Diego). Within this large cohort, the medical records of 1117 chronologically identified very preterm births of <32 weeks gestation (429 whites, 440 Hispanics, and 248 blacks) were abstracted to distinguish very preterm births with known causes including preeclampsia, which formed the case group for this study. Cases with a birthweight >2500 g were excluded based on suspected misdating. Pregnancies with structural abnormalities reported on delivery records or identified by the California Birth Defects Monitoring Program registry were also excluded.

Preeclampsia was defined as hypertension accompanied by proteinuria with onset >20 weeks’ gestation in the California Very Preterm Birth Study. Hypertension was defined as blood pressure ≥140 mm Hg systolic or ≥90 mm Hg diastolic on 2 occasions 6 hours apart. Proteinuria was defined as ≥0.3 g of protein in a 24-hour urine collection or ≥1+ dipstick on a urine test on 2 occasions 6 hours apart. Early-onset preeclampsia was defined as preeclampsia with delivery <32 weeks’ gestation. Gestational age estimates for cases were mostly based on ultrasound (61.5%) and date of last menstrual period (29.5%) from the California Prenatal Screening Program with a small part (9.0%) based on maternal hospital chart review.

Hospital chart review identified 207 cases of early-onset preeclampsia with a maternal midtrimester specimen available for laboratory testing. Ten of the 207 cases were excluded due to maternal specimens collected beyond the gestational window of 15-20 weeks. The final sample included 90 white, 67 Hispanic, and 40 black cases. Details of the sample selection are listed in the Figure .

Identification flowchart of early-onset preeclampsia cases and controls

Step-by-step sample selection. Exclusion rates in each step vary between race-ethnicity groups due to different population distributions and exclusion details by race-ethnicity groups.

Yang et al. Racial-ethnic differences in angiogenic and antiangiogenic factors. Am J Obstet Gynecol 2016 .

Control identification

Controls were noncases selected at an original sample size of ≥10 times the number of cases before specimen pulling using simple random sampling within the same race-ethnicity-specific singleton live-birth cohort as cases. Births with adverse outcomes such as late-onset preeclampsia, preterm birth, and low birthweight were also eligible controls. Controls of white, Hispanic, and black origin were drawn separately for each racial-ethnic group. For women with multiple pregnancies during the study period, only the first pregnancy was chosen. Control births without banked serum specimens or with structural malformations according to the California Birth Defects Monitoring Program registry were excluded. The final sample included 916 white, 709 Hispanic, and 738 black controls ( Figure ). Gestational age estimates for controls were based on ultrasound (55.8%), date of last menstrual period (34.9%), and physical exam (9.3%) from the California Prenatal Screening Program.

Biological specimens

Banked maternal blood specimens were retrieved from the California Biobank Program. As part of the state prenatal screening program, obstetrical care service providers at doctors’ offices or drawing stations collected venous blood from pregnant women who voluntarily enrolled at 15-20 weeks’ gestation. The blood, which was collected in serum separator tubes, was sent to 1 of the 7 newborn and prenatal screening laboratories by mail under ambient conditions. Specimens arrived for triple-marker prenatal testing within 7 days of collection (median time 3 days). After routine testing, leftover specimens were refrigerated and then processed for long-term storage at –20°C. Women consented to the routine prenatal screening test in writing and were notified that the blood becomes the property of the state, which allows for a variety of approved research uses of the leftover blood under the highest level of security and confidentiality standards. While women have the option to request the specimen not be used for research purposes, very few assert this right. Specimen vials from cases and controls were intermingled, with vial positions randomly assigned during shipping and laboratory testing. Laboratory personnel were unaware of subjects’ case-control status.

Laboratory methods

Maternal serum levels of PlGF, sEng, and sVEGFR-1 were determined by immunoassays (R&D Systems Inc, Minneapolis, MN) utilizing the quantitative sandwich enzyme immunoassay technique. Laboratory testing was conducted at the research core laboratory of Perinatology Research Branch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Department of Health and Human Services. Standard serum specimens were used in each laboratory run and analyte value was determined by interpolation from standard curves. The assays utilized 120, 25, and 25 μL of serum for the PlGF, sEng, and VEGFR-1 assays, respectively. The interassay and intraassay coefficients of variation obtained in the laboratory were 4.6% and 3.6% for PlGF, 5.3% and 4.3% for sEng, and 2.8% and 2.7% for sVEGFR-1, respectively. The sensitivity of the assays was 5.7 pg/mL for PlGF, 40.0 pg/mL for sEng, and 12.9 pg/mL for sVEGFR-1.

Data analysis

The laboratory levels of PlGF, sEng, and sVEGFR-1 were linked to birth records, prenatal screening program data, and maternal hospital chart data (cases only) for analysis. Maternal characteristics were compared between cases and controls. Gestational weeks at delivery and at preeclampsia diagnosis for cases were compared among the 3 racial-ethnic groups. The χ ² test and Student t test were used to estimate the statistical significance of the comparisons for categorical and continuous variables, respectively. Natural logarithmic transformed circulating levels of PlGF, sEng, and sVEGFR-1 were normally distributed in cases and controls. Means (log pg/mL) and SE were calculated for cases and controls of each racial-ethnic group and were adjusted by maternal age, education, parity, specimen storage year, gestational age at specimen collection, and maternal weight at or near specimen collection. The adjusted logarithmic means were compared using Tukey honestly significant difference test between cases and controls in each racial-ethnic group and between racial-ethnic groups among cases and controls, respectively. Odds ratios (OR) and 95% confidence intervals (CI) derived from binomial logistic regressions were used to measure effect magnitude of analyte levels in association with case-control status. The logarithmic transformed levels (log pg/mL) of the 3 analytes were used in the logistic regression to reduce both variance and outlier influence. Race-ethnicity is a significant effect modifier for PlGF and sEng in association with early-onset preeclampsia and therefore association results were presented separately for each race-ethnicity group. Maternal age, education, parity, specimen storage year, gestational weeks at specimen collection, and maternal weight at specimen collection were included as potential confounders. Standard Z test was conducted to compare OR between racial-ethnic groups. Given the above-mentioned final sample sizes, this study had adequate power of ≥80% to detect association of moderate or higher effect estimates (OR ≥2.6) between cases and controls in each racial-ethnic group under a logistic model with a 2-sided test.

All of the statistical analyses were conducted using software (SAS, Version 9.2 for Windows; SAS Institute, Cary, NC). The study protocol was approved by the California Health and Human Services Agency Committee for the Protection of Human Subjects (project no. 10-10-52).