Modeling Down syndrome screening performance using first‐trimester serum markers

Objective To evaluate the modeled predictive value of three current screening markers (pregnancy‐associated plasma protein‐A (PAPP‐A), free β‐human chorionic gonadotropin (free β‐hCG), and nuchal translucency (NT)) and four potential screening markers (a disintegrin and metalloprotease 12 (ADAM12), total hCG, placental protein 13 (PP13), and placental growth factor (PlGF)) for Down syndrome using different screening strategies. Methods All markers were measured in stored first‐trimester serum of 151 Down syndrome cases and 847 controls. All marker levels were expressed as gestational age‐specific multiples of the median (MoMs) and comparisons were made using the Mann‐Whitney U‐test. Detection rates (DRs) for fixed false‐positive rates (FPRs) were modeled using different screening strategies. Results Significantly different median MoMs for Down syndrome cases compared to controls were found for PAPP‐A (0.49 vs. 1.00; P < 0.0001), free β‐hCG (1.70 vs. 1.01; P < 0.0001), ADAM12 (0.89 vs. 1.00; P < 0.0001), total hCG (1.28 vs. 1.00; P < 0.0001), PlGF (0.80 vs. 1.00; P < 0.0001) and NT (1.74 vs. 1.01; P < 0.0001). The lower PP13 MoM in Down syndrome cases (0.91 vs. 1.00) was not statistically significant (P = 0.061). Adding the four new markers to the current screening strategy (i.e. first‐trimester combined test) led to an increase in DR from 77% to 80% at a 5% FPR. The modeled application of a two‐sample screening strategy (with some markers assessed early and others later in the first trimester) increased the DR to 89%. In a two‐step contingent screening model, using an intermediate risk range of 1 in 100 to 1 in 2000 at biochemical screening (using all markers), the overall DR was 77%, but it was predicted that only 33% of women would require referral for NT measurement. Conclusions First‐trimester Down syndrome screening may be improved by adding new markers to the current screening test and by applying different screening strategies. The application of a two‐sample screening model resulted in the highest predicted DR, but this should be confirmed in population‐based prospective studies. Copyright © 2011 ISUOG. Published by John Wiley & Sons, Ltd.