Added prognostic value of the fetal fraction in non-invasive prenatal testing in the prediction of adverse pregnancy outcomes: analysis of a nationwide cohort of 56 110 pregnant women participating in the TRIDENT-2 study.

Abstract

Background The proportion of fetal cell-free DNA (cfDNA) in the maternal circulation (i.e., the fetal fraction) is universally measured as a quality parameter in non-invasive prenatal testing (NIPT) for fetal aneuploidies. As fetal cfDNA originates from the placenta, the fetal fraction might reflect placental health. We assessed the added prognostic value of the fetal fraction in the prediction of adverse pregnancy outcomes.

Methods We performed a retrospective cohort study of women with singleton pregnancies that opted for NIPT over a 1-year period between June 2018 and June 2019 within the Dutch national implementation study on NIPT for fetal aneuploidies (TRIDENT-2 study). The TRIDENT-2 study data were linked to the Dutch registry of prenatal screening results (Peridos) and the Dutch registry of pregnancy outcomes (Perined). Outcomes included hypertensive disorders of pregnancy (HDP), birthweight <p10 and <p2.3, spontaneous preterm birth (sPTB), diabetes, congenital anomalies, and a combined poor neonatal outcome. The prognostic value of the fetal fraction was assessed by comparing logistic regression models based on clinical predictors without the fetal fraction (base model) and with the fetal fraction included (extended model) in terms of goodness of fit (likelihood ratio test (LRT)). In case of a statistically significant LRT (p<0.05), the amount of prognostic value was quantified by predictive measures including the area under the curve (AUC), integrated discrimination improvement (IDI), and the percentage of new predictive information based on difference in variance of predicted probabilities.

Results The analysis included 56 110 pregnant women. Based on the LRT, fetal fraction showed significant (p<0.05) added prognostic value for HDP, birthweight <p10, birthweight <p2.3, total sPTB, moderate to late sPTB, and diabetes, but not (p>0.05) for extremely sPTB, very sPTB, congenital anomalies, and combined poor neonatal outcome. For outcomes with a statistically significant LRT, the AUC showed marginal, but statistically significant (p<0.05) improvements for birthweight <p10, birthweight <p2.3, and all sPTB. The IDI was statistically significant (p<0.05) for HDP, birthweight <p10, birthweight <p2.3, all sPTB, moderate to late sPTB, and diabetes. The percentage of new predictive information based on variance was 5.4%, 7.5%, 10.1%, 2.3%, 2.9%, and -0.5% for HDP, birthweight <p10, birthweight <p2.3, all sPTB, moderate to late sPTB, and diabetes respectively.

Conclusion This study shows that the fetal fraction has added prognostic value in the prediction of HDP, birthweight<p10, birthweight<p2.3, all sPTB, moderate to late sPTB, and diabetes. The utility of the fetal fraction for the prediction of adverse pregnancy outcomes in clinical practice needs to be established in future research.