Interpretation
Our finding and conclusions are in agreement with a recent study by Maser Zumaeta and colleagues, which has concluded that PAPP-A is the preferred marker for trisomy 21 screening, and that replacing PAPP-A with PlGF increases the false positive and screen positive rates.11Mazer Zumaeta A, Wright A, Syngelaki A, Maritsa VA, Bardani E, et al. Screening for trisomies at 11-13 weeks’ gestation: use of PAPP-A, PlGF or both. Ultrasound Obstet Gynecol 2020 doi: 10.1002/uog.22140 In a parallel study the same authors have reported that PlGF is the preferred biochemical marker for preeclampsia screening at 11-13 weeks of gestation, and that using PAPP-A instead of PlGF reduces preeclampsia screening sensitivity.22Mazer Zumaeta A, Wright A, Syngelaki A, Maritsa VA, Da Silva AB et al. Screening for pre-eclampsia at 11-13 weeks’ gestation: use of pregnancy-associated plasma protein-A, placental growth factor or both. Ultrasound Obstet Gynecol 2020 doi: 10.1002/uog.22093. Our data and analysis indicate that previous studies reporting increased or similar levels of PlGF relative to unaffected pregnancies, as opposed to reduced levels of PlGF in trisomy 21, were probably due to the impact of case-selection, potential sample degradation or underlying differences between earlier PlGF assays and current assays. Our earlier study indicated that PlGF in stored samples remained stable for at least 3 years 33Law LW, Sahota DS, Chan LW, Chen M, Lau TK, et al. Effect of long-term storage on placental growth factor and fms-like tyrosine kinase 1 measurements in samples from pregnant women.J Matern Fetal Neonatal Med 2010;23:1475-80, however its stability when serum samples were stored for longer periods remained unknown. It has been shown that duration of storage time accounts for up to 35% of plasma protein concentration variation in frozen biobank samples of healthy women. 44Enroth S, Hallmans G, Grankvist K, Gyllensten U. Effects of Long-Term Storage Time and Original Sampling Month on Biobank Plasma Protein Concentrations. EBioMedicine 2016;12:309-14 Sensitivity of reported PlGF concentrations to specific PlGF isoforms has also been reported7, 55Nucci M, Poon LC, Demirdjian G, Darbouret B, Nicolaides KH. Maternal serum placental growth factor (PlGF) isoforms 1 and 2 at 11-13 weeks’ gestation in normal and pathological pregnancies. Fetal Diagn Ther. 2014;36:106-16 Nucci et al. reported that the PlGF-2 isoform of PlGF is more abundant than the PlGF-1 isoform throughout pregnancy,24 whilst Cheng et al. reported that current assays had cross reactivity to PlGF-2 ranging from 10 to 21%, whilst PlGF-1 isomer recovery ranged from 38 to 60%.7
In the vast majority of cases adding PlGF or replacing PAPP-A with PlGF would not have changed whether women were screened high risk (≥1:250) or low risk (<1:250) for trisomy 21. Women whose status did change, particularly when PlGF was added to the combined test, were those in whom PAPP-A and PlGF MoMs were discordant and with an atypical screening marker pattern for trisomy 21, either PAPP-A MoM was reduced (<1 MoM) and PlGF MoM was increased (>1 MoM) or PlGF MoM was reduced and PAPP-A MoM was increased. Of the 29 women having a trisomy 21 affected pregnancy, 24 were screened as high risk by the combined test, the combined test plus PlGF and after replacing PAPP-A by PlGF. Each test identified 2 of the 5 remaining trisomy 21 affected pregnancies, and hence there was effectively a zero sum gain. One alternative would be to use non-probability based pattern recognition such as machine learning based on multi-layer neural networks to see if such an approach would allow increased flexibility and recognition of atypical trisomy 21 screening marker patterns. 66Koivu A, Korpimäki T, Kivelä P, Pahikkala T, Sairanen M. Evaluation of machine learning algorithms for improved risk assessment for Down’s syndrome. Comput Biol Med 2018;98:1-7
The current screening model for trisomy 21 is based on probability based pattern recognition, namely that trisomy 21 is associated with increasing maternal age, increased fetal NT and free β-hCG MoMs, reduced levels of PAPP-A and PlGF MoMs relative to unaffected fetuses. Our current and earlier study, as well as studies reporting reduced levels of PlGF in trisomy 21 affected pregnancies all indicated significant and strong correlations between PlGF and PAPP-MoMs. This is in contrast to other screening markers used to estimate risk for trisomy 21 as the inter-marker correlations are negligible, indicating that each marker is providing additional as opposed to effectively the same information when used to estimate risks. Inclusion of an additional biomarker which is significantly correlated with an existing marker is thus, as our study has demonstrated, unlikely to provide additional screening benefit. Replacing PAPP-A with PlGF or adding PlGF would only increase costs as the cost of the PlGF assay is higher than that of the PAPP-A assay.
Our study highlights also the difference in determining screening performance based on modelling versus that observed empirically when the test is used on a day to day basis. The former presumes collected data is in full agreement with the risk estimation model assumptions. Our earlier case-control study indicated that adding PlGF to the combined test would be expected to achieve a detection rate of 96% for the same 5% false positive rate based on modelling even though the population in our earlier study was 2 years younger than in our current study.17 Modelling, whilst useful remains only as a guide, and as our data indicate, models reporting expected detection rates in future should incorporate a random percentage of cases having measurement discordance between included screening markers.