Comparison of the New PRC Software with the Established Algorithm of the FMF UK for the Detection of Trisomy 21 and 18/13

• Published in: Fetal diagnosis and therapy Vol. 24; no. 4; pp. 376 - 384
• Main Author: Lüthgens, K.
• Format: Journal Article
• Language: English
• Published: Basel, Switzerland Karger 01.01.2008; S. Karger AG
• Subjects: Adult; Algorithms; Biological and medical sciences; Body Weight; Chorionic Gonadotropin, beta Subunit, Human - metabolism; Chromosome aberrations; Chromosomes, Human, Pair 13; Chromosomes, Human, Pair 18; Delivery. Postpartum. Lactation; Down Syndrome - diagnosis; Down Syndrome - epidemiology; European Continental Ancestry Group - statistics & numerical data; False Positive Reactions; Female; General aspects; Gynecology. Andrology. Obstetrics; Humans; Maternal Age; Medical genetics; Medical sciences; Nuchal Translucency Measurement - standards; Pregnancy; Pregnancy-Associated Plasma Protein-A - metabolism; Reproducibility of Results; Risk Factors; ROC Curve; Sensitivity and Specificity; Smoking - epidemiology; Software - standards; Trisomy - diagnosis; United Kingdom; Europe; Prenatal risk calculation PRC; ASTRAIA; Trisomy 18/13; Nuchal translucency; PAPP-A; Free beta-hCG; Screening for trisomy 21; Chromosomal aberration; Trisomy; Chromosome D13; Neonatology; Aneuploidy; Pregnancy associated plasma protein A; Prenatal; Human chorionic gonadotrophin; Edwards syndrome; Patau syndrome; Down syndrome; Medical screening; Algorithm; Medicine; Chromosome E18; Risk factor; Software; Detection; Comparative study
• ISSN: 1015-3837; 1421-9964; 1421-9964
• DOI: 10.1159/000165116

Objective: The aim of this study was to compare the newly developed Prenatal Risk Calculation (PRC) software and the established Fetal Medicine Foundation (FMF) UK algorithm for their detection and false-positive rates in prenatal screening of trisomy 21 or 18/13. Methods: Nuchal translucency was measured by FMF-certified sonographers in 39,004 pregnancies. Risks for trisomy 21 and 18/13 were calculated together with serum PAPPA and free beta-hCG in all cases. Overall, 109 cases of trisomy 21 and 39 cases of trisomy 18 or 13 occurred. The detection rates were calculated for all 109 trisomy 21 cases using both PRC and ASTRAIA, software based on the official algorithm of the FMF UK. The false-positive rate was calculated on the basis of all 39,004 cases for ASTRAIA and on the basis of 3,620 additional cases for PRC. The comparison of the algorithms was performed by the calculation of receiver-operating characteristic (ROC) curves. Results: The detection rates for trisomy 21 for a 5% false-positive rate were 91% (99/109) for PRC and 90% (98/109) for ASTRAIA. The difference was not significant. ROC curve analysis showed an area under the curve (AUC) of 0.976 for PRC and 0.975 for ASTRAIA (p = 0.80). At a cutoff of 1:300, PRC showed a slightly (nonsignificantly) higher, but nonsignificant false-positive rate (3.8% for ASTRAIA, version since 2005, 5.1% for PRC) associated with a slightly (nonsignificantly) higher, but nonsignificant detection rate of 91% (PRC) versus 88% (ASTRAIA). The false-positive rate of the biochemical risk (without NT) in PRC was significantly higher (17.7%) than with the FMF UK algorithm (9.0%) at a common risk cutoff of 1:300. A reason for the higher false-positive rates in PRC may be the missing correction for maternal weight, smoking status, and ethnicity. The detection rates for trisomy 18/13 at a risk cutoff of 1:150 were 79% for PRC and 77% for ASTRAIA. False-positive rates for trisomy 18/13 at a cutoff of 1:150 were 1.5% for PRC and 0.6% for ASTRAIA (differences were nonsignificant). Conclusion: For a 5% false-positive rate, the new PRC software detects as many trisomy 21 and trisomy 18/13 cases as the established algorithms of the FMF UK. In order to reduce the false-positive rate for the biochemical risk, the algorithm of the PRC software should be redesigned to include the maternal weight, smoking status of the mother, and the ethnicity.