Materials and Methods

The study population consisted of 99 women who attended the Division of Prenatal Medicine, University of Bologna, Bologna, Italy. Gestational age was calculated by ultrasound measurements at 11-14 weeks’ gestation. A second ultrasound examination was performed at 22-24 weeks’ gestation for measurement of fetal growth and examination for fetal defects. Those cases in which no major fetal defects were detected were included in the study. Eleven women with preeclampsia but who were asymptomatic at the time of blood drawing were matched with 88 control subjects (1:8 match for fetal sex and gestational age expressed in weeks + days). All women were informed and agreed to participate in the study, which was approved by the local Hospital Ethics Committee.

Preeclampsia was defined as gestational hypertension (systolic pressure of >140 mm Hg or diastolic blood pressure of >90 mm Hg on at least 2 occasions after 20 weeks’ gestation) with proteinuria (>0.3 g/d). Severe preeclampsia was defined by the presence of 1 or both of the following events: (1) severe gestational hypertension (systolic pressure of >160 mm Hg or diastolic pressure of >110 mm Hg on at least 2 occasions after 20 weeks’ gestation) and (2) severe proteinuria (≥5g protein in a 24-hour urine specimen or ≥3 g on 2 random urine samples that were collected at least 4 hours apart). Intrauterine growth restriction was defined as the estimated fetal weight being 2.0 standard deviations below the mean expected weight for the gestational age, as determined by ultrasound evaluation.

Blood samples were taken at the time of first examination (11-14 weeks’ gestation) from subjects who had been scheduled for chorionic villous sampling or amniocentesis. The blood samples (2.5 mL) were collected in PAXgene blood RNA tubes (PreAnalytic, Hombrechtikon, Switzerland), kept at room temperature for 3 hours, and then stored at –20°C until being transported to Japan. Molecular analysis was performed in the Department of Obstetrics and Gynecology at Showa University School of Medicine, Tokyo, Japan. RNA extraction and real-time polymerase chain reaction (PCR) were performed according to protocols described elsewhere. In brief, cellular component samples were centrifuged twice at 4000 g for 10 minutes at room temperature to remove the entire supernatant and any mRNA that was present in the residual plasma. The pellet was then washed, resuspended, and incubated in optimized buffer solution that contained proteinase K to digest protein. A second round of centrifugation was performed to remove any residual cell debris, and the resulting supernatant was transferred to a fresh microcentrifuge tube. We added 100% ethanol to the supernatant to adjust the binding conditions; the resultant lysate was then applied to a PAXgene spin column (PreAnalytiX; PreAnalytic), which resulted in selective binding of RNA to the silica-gel membrane of the spin column. After the column was washed 3 times, pure RNA was eluted in 80 μL of RNase-free water.

Reverse transcription of the mRNA was performed using an Omniscript RT Kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Complementary DNA products were amplified by real-time quantitative PCR according to the manufacturer’s instructions (QuantiTect Probe PCR kit; Qiagen) with a 2-μL aliquot of complementary DNA and the kit’s components in a reaction volume of 20 μL. QuantiTect PCR analyses for vascular endothelial growth factor, FLT1, endoglin, PlGF, and TGFβ1 were performed with predeveloped and commercially available primers and probe sets (catalog no. Hs00900054_m1 for vascular endothelial growth factor; catalog no. Hs01052936_m1 for FLT1; catalog no. Hs00923997_g1 for endoglin; catalog no. Hs00182176_m1 for PlGF; and catalog no. Hs0000171257_m1 for TGFβ1; Applied Biosystems, Foster City, CA). The following thermal cycling protocol was used for PCR: initial denaturation at 95°C for 15 minutes, 40 cycles of denaturation at 94°C for 15 seconds, and annealing at 60°C for 1 minute. As an initial step, we verified that each PCR assay was specific to mRNA and not to genomic DNA. Amplification data were collected and analyzed with an ABI Prism 7900T Sequence Detector (Applied Biosystems). Each sample was analyzed in duplicate, and multiple negative water blanks were included in every analysis. Quantification of gene expression was performed with investigators blinded to the outcome of pregnancy. Amounts of mRNA samples were expressed in term of copies per milliliter. To quantify mRNA concentrations, we prepared plasmid DNA for calibration curves as previously described.

Data analysis was performed by nonparametric statistics because of the small sample size. A 1:8 match (1 case with 8 control subjects) for gestational age expressed in days, and fetal sex was determined. Such a match would guarantee a proper comparison, even in the presence of a relatively low number of cases. The median mRNA concentration of each available marker (FLT1, endoglin, PlGF, TGFβ, and PP13) as a function of increasing gestational age was measured initially and calculated by weighted log10–linear regression. All data were expressed as multiples of the median. We used a logistic regression to calculate the posterior risk of preeclampsia in both control subjects and affected cases, taking the panel of available mRNA values that were expressed in multiples of the median and parity as predictors of the disease. The detection and false-positive rates were calculated for each available marker with the use of a univariable receiver operating characteristic (ROC) curve. Multivariable analysis was performed with logistic regression to calculate the risk for each patient for classification as a control or preeclampsia case. The logistic output was adjusted for the incidence of preeclampsia in the general population (2%) by the calculation of the sampling fraction as described by Collett. Again, quantitative data were transformed into an ordinal scale (in 4 categories <25th percentile, 25-50th percentile, 50-75th percentile, and >75th percentile).

To obtain a more robust risk estimation, a multivariable ROC curve for the calculation of multivariable detection rate was built with the use of the calculated risk for preeclampsia by logistic regression analysis as the test variable, for each patient in the series.