OP16.03: First trimester ultrasound markers for the prediction of adverse pregnancy outcomes in monochorionic diamniotic pregnancies

To investigate the role of intertwin discordance of Crown–rump length (CRL) and nuchal translucency (NT) in predicting Twin–twin transfusion syndrome (TTTS) or selective growth restriction (sIUGR) in unselected monochorionic diamniotic (MCDA) twin pregnancies. MCDA pregnancies undergoing first trimester ultrasound between 11+0 and 13+6 weeks from January 2014 to January 2023 were retrospectively selected. Those referred for second opinion, with fetal anomalies or lost to follow up were excluded. Descriptive statistics were calculated by median and interquartile range (IQR). The Chi-square test and the Fisher's test were used for the analysis of contingency tables of proportions. The accuracy of CRL and NT discordance in predicting pregnancy complications was examined by the area under the ROC curve (AUC). 183 MCDA pregnancy were included. The median NT discordance was 14% (IQR 8-23.5%) while the median CRL discordance was 4% (IQR 2–7%). In 67/183 pregnancies (36.6%) NT discordance was ≥20% and in 23/183 (12.6%) pregnancies CRL discordance was ≥10%. Overall 40/183 (21.8%) pregnancies developed sIUGR while 29/183 (15.8%) developed TTTS. Compared to MCDA pregnancies with NT discordance < 20%, MCDA pregnancies with NT discordance ≥20% had a significantly higher incidence of TTTS (26.9% versus 9.5%; p = 0.002) but no significantly different incidence of sIUGR (28.4% versus 18.1%; p = 0.106). The AUC of NT discordance was 0.66 for TTTS. Compared to MCDA pregnancies with CRL discordance < 10%, in MCDA pregnancies with CRL discordance ≥10% there was a significantly higher incidence of sIUGR (47.8% versus 18,1%; p = 0.001) but no difference in TTTS (13% versus 16.2%; p = 1). The AUC of CRL discordance for sIUGR was 0.59. In MCDA pregnancies, NT discordance ≥20% is associated with higher incidence of TTTS while a CRL discordance ≥10% is associated with higher incidence of sIUGR, however their predictive value is low.