Non-Invasive Prenatal Diagnosis (NIPD) of Sickle-Cell Disease By Massively Parallel Sequencing of Cell-Free Fetal DNA in Maternal Serum

Sickle cell disease (SCD) is the most common genetic haematological disorder worldwide. Around 300.000 affected infants are born every year, including at least 1000 in the United States. Prenatal diagnosis is currently carried out using amniotic fluid or chorionic villus sampling. These invasive procedures are perceived to have a small risk of miscarriage. The availability of non-invasive prenatal diagnosis (NIPD) is predicted to increase uptake of prenatal diagnosis for SCD, as it has no perceived miscarriage risk. NIPD may also be more readily implemented than invasive prenatal diagnosis in the low-resource countries in which SCD is the most prevalent. However, accurate NIPD of autosomal recessive disorders such as sickle cell disease has proven challenging as this requires detection of fetal inheritance of a maternal allele from a mixed maternal-fetal pool of cell-free DNA.

We report the development of a targeted massively parallel sequencing assay for the NIPD of fetal SCD using cell-free fetal DNA from maternal plasma. No paternal or previous offspring samples were required. 44 clinical samples were analysed, including 37 plasma samples from pregnant SCD carriers and 7 plasma samples from women with SCD due to Hb SC. We used a relative mutation dosage based approach for the 37 samples from maternal SCD carriers (Hb AS or Hb AC), integrating Unique Molecular Identifiers (UMIs) into the analysis to improve the accuracy of wildtype and mutant allele counts. We used a separate wildtype allele detection approach for the 7 samples from women with compound heterozygous SCD, in whom the detection of wildtype cell-free DNA indicates the presence of a carrier fetus. The success of the assay was evaluated by comparing results with the established fetal sickle status as determined through either invasive prenatal diagnosis or newborn screening.

During development, two key factors improved the accuracy of the results: i) Selective analysis of only smaller cell-free DNA fragments enhanced the fetal fraction for all samples, with greater effects observed in samples from earlier gestations. This approach improved diagnostic accuracy: for 3 out of 44 samples, the genotype was inconclusive or incorrect before size selection, but correct after size selection. ii) Modifications to DNA fragment hybridisation capture optimised the diversity of Unique Molecular Identifier-tagged molecules analysed. This led to improvements in the results obtained for 5 samples, with 3 previously inconclusive samples correctly called and 2 previously discrepant results moved into the inconclusive range.

In total, 37 results were concordant with the established fetal sickle status; this included 30/37 samples from carrier women and 7/7 samples from women with sickle cell disease due to Hb SC. The remaining 7 carrier samples gave an inconclusive result, which for 3 samples was attributed to a low fetal fraction.

Samples from as early as 8 weeks gestation were successfully genotyped. There were no false positive or false negative results.

This study is the largest to use NGS-based NIPD on clinical plasma samples from pregnancies at risk of SCD. Efforts to validate the assay on a larger sample cohort and to reduce the inconclusive rate are warranted. This study shows that NIPD for SCD is approaching clinical utility and has the potential to provide increased choice to women with pregnancies at risk of sickle cell disease.