Background: The development of sequencing-based noninvasive prenatal testing (NIPT) has been largely focused on whole-chromosome aneuploidies (chromosomes 13, 18, 21, X, and Y). Collectively, they account for only 30% of all live births with a chromosome abnormality. Various structural chromosome changes, such as microdeletion/microduplication (MD) syndromes are more common but more challenging to detect. Recently, several publications have shown results on noninvasive detection of MDs by deep sequencing. These approaches demonstrated the proof of concept but are not economically feasible for large-scale clinical applications.
Methods: We present a novel approach that uses low-coverage whole genome sequencing (approximately 0.2×) to detect MDs genome wide without requiring prior knowledge of the event's location. We developed a normalization method to reduce sequencing noise. We then applied a statistical method to search for consistently increased or decreased regions. A decision tree was used to differentiate whole-chromosome events from MDs.
Results: We demonstrated via a simulation study that the sensitivity difference between our method and the theoretical limit was <5% for MDs ≥9 Mb. We tested the performance in a blinded study in which the MDs ranged from 3 to 40 Mb. In this study, our algorithm correctly identified 17 of 18 cases with MDs and 156 of 157 unaffected cases.
Conclusions: The limit of detection for any given MD syndrome is constrained by 4 factors: fetal fraction, MD size, coverage, and biological and technical variability of the event region. Our algorithm takes these factors into account and achieved 94.4% sensitivity and 99.4% specificity.
© 2015 American Association for Clinical Chemistry.