Statistical analysis
PlGF concentration levels were converted to MoMs by performing a regression analysis to allow for change with gestational age, weight, smoking and method of conception (spontaneous or in vitrofertilization (IVF)) in the first 500 women screened. The PlGF MoM in these 500 women were then retrospectively calculated and recorded in our screening database. PlGF MoMs in subsequent screened pregnancies were prospectively calculated along with the estimated risks for trisomy 21 as described below.
Trisomy 21 risks using PlGF alone or in conjunction with other screening markers were determined using the multivariate Gaussian model approach currently used as a standard method in risk calculation software.11Reynolds TM, Penney MD. The mathematical basis of multivariate risk screening: with special reference to screening for Down’s syndrome associated pregnancy. Ann Clin Biochem 1990;27:452-8 The expected PlGF MoM distributions in unaffected and trisomy 21 affected pregnancies were based on an earlier collaborative study. 22Han J, Liu H, Xu ZP, Cuckle H, Sahota D, et al. Maternal serum PlGF (placental growth factor) in Chinese women in the first trimester undergoing screening for Down syndrome. Eur J Obstet Gynecol Reprod Biol 2016;201:166-70 This study indicated that log10 PlGF MoM distribution means and standard deviation in unaffected pregnancies were 0 and 0.1638, respectively, whilst corresponding figures in trisomy 21 affected pregnancies were 0.1979 and 0.1511.17 The assumed correlation between log10 PlGF MoM and log10 PAPP-A MoM, log10 free β-hCG and log10 NT MoMs in unaffected pregnancies were 0.285, −0.019 and −0.027, respectively, whilst that in trisomy 21 affected pregnancies were respectively 0.0435, −0.121 and −0.337.17 Modeling in our earlier study indicated a screening test based on maternal age, fetal NT, PAPP-A, free β-hCG and PlGF MoMs was expected to have a 96.2% detection rate for trisomy 21 for a 5% false positive rate. 17
Three risks for trisomy 21 were prospectively estimated, one based on the conventional combined test, a second based on age, fetal NT, free β-hCG and PlGF and lastly one based on age, fetal NT, free β-hCG, PAPP-A and PlGF.
Area under the receiver operating characteristic (ROC) curve (AUC), detection rates and false positive rates were calculated for the combined risk as well as estimated risks based on replacement of PAPP-A by PlGF as well as the addition of PlGF to the conventional combined test. AUCs among the different first trimester combined tests were compared using the Delong test. 33DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44: 837-45 McNemar test was used to determine if replacement of PAPP-A by PlGF or the addition of PlGF resulted in a significant number of pregnancies being reallocated with regard to risk status for trisomy 21.
All trisomy 21 risks were estimated using our in house laboratory risk calculation software whilst all other analyses were performed using SPSS for Windows version 20 (SPSS, Illinois, USA) and MedCalc Statistical Software version 18.10.2 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2018).