Performance of single‐gene noninvasive prenatal testing for autosomal recessive conditions in a general population setting

Abstract

AbstractObjectiveCarrier screening with reflex to single‐gene noninvasive prenatal testing (sgNIPT) is an alternative approach for identifying pregnancies at risk for inherited autosomal recessive conditions without the need for a sample from the reproductive partner. This study is the largest clinical validation of this approach in a general population setting.MethodsThe clinical performance of carrier screening with reflex to sgNIPT for cystic fibrosis, spinal muscular atrophy, alpha thalassemias, and beta hemoglobinopathies was assessed by collecting pregnancy outcome data on patients who underwent this testing and comparing the neonatal outcome to the assay‐predicted fetal risk.ResultsOf 42,067 pregnant individuals who underwent screening, 7538 carriers (17.9%) had reflex sgNIPT, and neonatal or fetal outcomes were obtained for 528 cases, including 25 affected pregnancies. Outcomes demonstrated high concordance with sgNIPT, for example, all pregnancies with 9 in 10 personalized fetal risk results were affected (positive predictive value (PPV) of 100% for the sub‐group) and the sgNIPT assay showed a sensitivity of 96.0% (95% CI: 79.65%–99.90%), specificity of 95.2% (95% CI: 92.98%–96.92%), average PPV of 50.0% (95% CI: 35.23%–64.77%), and negative predictive value (NPV) of 99.8% (95% CI: 98.84%–99.99%). The end‐to‐end performance of carrier screening with reflex to sgNIPT was calculated to have a sensitivity of 92.4% and specificity of 99.9%, which are unaffected by partner carrier screening or misattributed paternity unlike a traditional carrier screening workflow, which has a 35% sensitivity and a maximum of 25% PPV (1 in 4) in a real‐life setting.ConclusionThis study builds upon earlier findings to confirm that carrier testing with reflex to sgNIPT is highly accurate for general population screening. Given this high accuracy and an NPV of 99.8%, this workflow should be considered as an option for most of the general pregnant population. When the biological partner sample is unavailable, this workflow should be recommended as the first‐line approach.