EP18.18: Development of a multivariate predictive model for Trisomy 21 risk assessment at 20–25 weeks of gestation

This study aims to construct a multivariate predictive model for trisomy 21 (T21) risk assessment between 20 to 25 weeks of gestation, utilising a combination of categorical and quantitative markers. Conducted at the Fetal Medicine Unit of Hospital Clínico de la Universidad de Chile between January 2004 and December 2017, this case-control study focused on singleton pregnancies with a 20-25 weeks screening ultrasound. Included markers were maternal age (MA), nasal bone length (NBL), prenasal thickness (PNT), biparietal diameter (BPD), nuchal fold thickness (NFT), atrium diameter, cerebellar length, cisterna magna diameter, femur/humerus length, and presence of mild hydronephrosis. Quantitative markers were standardised using percentiles from reference curves. Exclusion criteria comprised multiple pregnancies, absence of antenatal/postnatal karyotype or neonatal phenotype, and other aneuploidies. Multivariate logistic regression analysis with a backward stepwise method was employed, with a significance level set at p < 0.1. Analysis included 3,125 singleton pregnancies. Notable markers in the multivariate model included MA (coef= 0.239; p < 0.001), humerus percentile (coef= -0.047; p = 0.018), mild hydronephrosis (coef= 3.185; p < 0.001), cerebellar length percentile (coef= -0.028; p = 0.056), NFT (coef= 0.803; p < 0.001), NBL z-score (coef= -0.396; p = 0.029), and PNT percentile (coef= 0.039; p = 0.01), with an intercept of -18.443. The multivariate model demonstrated a sensitivity of 70.6%, false positive rate of 2.5%, LHR(+) of 28.1, and LHR(-) of 0.3. The area under the curve (AUC) was 0.975 (95% CI: 0.952 – 0.999). Integration of second trimester markers with a quantitative approach facilitates personalised T21 risk prediction at 20-25 weeks gestation.