Prospective head-to-head comparison of accuracy of two sequencing platforms for screening for fetal aneuploidy by cell-free DNA: the PEGASUS study

Abstract

We compared clinical validity of two non-invasive prenatal screening (NIPS) methods for fetal trisomies 13, 18, 21, and monosomy X. We recruited prospectively 2203 women at high risk of fetal aneuploidy and 1807 at baseline risk. Three-hundred and twenty-nine euploid samples were randomly removed. The remaining 1933 high risk and 1660 baseline-risk plasma aliquots were assigned randomly between four laboratories and tested with two index NIPS tests, blind to maternal variables and pregnancy outcomes. The two index tests used massively parallel shotgun sequencing (semiconductor-based and optical-based). The reference standard for all fetuses was invasive cytogenetic analysis or clinical examination at birth and postnatal follow-up. For each chromosome of interest, chromosomal ratios were calculated (number of reads for chromosome/total number of reads). Euploid samples' mean chromosomal ratio coefficients of variation were 0.48 (T21), 0.34 (T18), and 0.31 (T13). According to the reference standard, there were 155 cases of T21, 49 T18, 8 T13 and 22 45,X. Using a fetal fraction ≥4% to call results and a chromosomal ratio z-score of ≥3 to report a positive result, detection rates (DR), and false positive rates (FPR) were not statistically different between platforms: mean DR 99% (T21), 100%(T18, T13); 79%(45,X); FPR < 0.3% for T21, T18, T13, and <0.6% for 45,X. Both methods' negative predictive values in high-risk pregnancies were >99.8%, except for 45,X(>99.6%). Threshold analysis in high-risk pregnancies with different fetal fractions and z-score cut-offs suggested that a z-score cutoff to 3.5 for positive results improved test accuracy. Both sequencing platforms showed equivalent and excellent clinical validity.

Trial registration: ClinicalTrials.gov NCT01925742.

Conflict of interest statement

The authors declare that they have no conflict of interest.

Figures

Fig. 1

Participants flow chart

Fig. 2

Individual z-scores and reference standard result for all fetuses tested with both index tests. Scatter plot of z-scores for same 3593 plasma samples comparing each index test (Proton (Y-axis) vs HiSeq (X-axis)) for chromosome 13, chromosome 18, chromosome 21, and chromosome X. Green triangles represent fetuses which, according to the reference standard, do not have the specific aneuploidy identified in the plot, while red circles represent fetuses that have the specific chromosome aneuploidy. Blue squares correspond to mosaic fetuses for the specific chromosome of interest, while yellow circles are triploid fetuses according to the reference standard. The lower left dashed green box in each plot shows the decision limit: z-score = 3