Variability in Fetal Fraction Estimation: Comparing Fetal Fractions Reported by Noninvasive Prenatal Testing Providers Globally

Background Fetal fraction (FF) measurement is considered important for reliable noninvasive prenatal testing (NIPT). Using minimal FF threshold as a quality parameter is under debate. We evaluated the variability in reported FFs of individual samples between providers and laboratories and within a single laboratory. Methods Genomic quality assessment and European Molecular Genetics Quality Network provide joint proficiency testing for NIPT. We compared reported FFs across all laboratories and stratified according to test methodologies. A single sample was sequenced repeatedly and FF estimated by 2 bioinformatics methods: Veriseq2 and SeqFF. Finally, we compared FFs by Veriseq and SeqFF in 87 351 NIPT samples. Results For each proficiency test sample we observed a large variability in reported FF, SDs and CVs ranging from 1.7% to 3.6% and 17.0% to 35.8%, respectively. FF measurements reported by single nucleotide polymorphism-based methods had smaller SDs (0.5% to 2.4%) compared to whole genome sequencing-based methods (1.8% to 2.9%). In the internal quality assessment, SDs were similar between SeqFF (SD = 1.0%) and Veriseq v2 (SD = 0.9%), but mean FF by Veriseq v2 was higher compared to SeqFF (9.0% vs 6.4%, P < 0.001). In patient samples, reported FFs were on average 1.12%-points higher in Veriseq than in SeqFF (P < 0.001). Conclusions Current methods do not allow for a reliable and consistent FF estimation. Our data show estimated FF should be regarded as a laboratory-specific range, rather than a precise number. Applying strict universal minimum thresholds might result in unnecessary test failures and should be used with caution.