OC04.02: A novel counting algorithm to detect common fetal trisomies in non‐invasive prenatal testing by maximal parallel sequencing and cross‐platform validation results

Abstract

To report the performance of our self-designed counting algorithm (genome-wide normalised score, namely GWNS) by comparison with the two existing methodologies (z-score and the normalised chromosomal value, namely NCV) and its cross-platform validations. We used theoretical approximations, computer simulations and real experimental data to compare the performance of GWNS with z-score and NCV in our own dataset, and we also did cross-platform comparisons on the different MPSS sequencer platforms (Miseq and GAIIx) manufactured by Illumina. We also looked into the minimal cell-free fetal DNA fraction as well as the lowest read numbers that can achieve a reliable clinical utility. With a fixed level of significance and power, our method requires consistently smaller fetal DNA proportions and total reads compared to the two popular methods – z-scores and NCV. This superiority sustains in theoretic approximations, simulated reads and experimental data collected from maternal plasmas of normal and T21/T18/T13 fetuses. Cross-platform interchangeability between Illumina's GAIIX and Miseq is also demonstrated in our setting. Approximately 4% and 4M reads are the safe lowest limit for safe clinical utility in our setting. Our results demonstrate the benefits of incorporating whole-genome information and can potentially be utilised to detect aneuploidies of other autosomes. GWNS is comparable to the two prevalent counting algorithms in terms of test performance.