High Detection Rate of Copy Number Variations Using Capture Sequencing Data: A Retrospective Study

Abstract

Abstract Background Capture sequencing (CS) is widely applied to detect small genetic variations such as single nucleotide variants or indels. Algorithms based on depth comparison are becoming available for detecting copy number variation (CNV) from CS data. However, a systematic evaluation with a large sample size has not been conducted to evaluate the efficacy of CS-based CNV detection in clinical diagnosis. Methods We retrospectively studied 3010 samples referred to our diagnostic laboratory for CS testing. We used 68 chromosomal microarray analysis–positive samples (true set [TS]) and 1520 reference samples to build a robust CS-CNV pipeline. The pipeline was used to detect candidate clinically relevant CNVs in 1422 undiagnosed samples (undiagnosed set [UDS]). The candidate CNVs were confirmed by an alternative method. Results The CS-CNV pipeline detected 78 of 79 clinically relevant CNVs in TS samples, with analytical sensitivity of 98.7% and positive predictive value of 49.4%. Candidate clinically relevant CNVs were identified in 106 UDS samples. CNVs were confirmed in 96 patients (90.6%). The diagnostic yield was 6.8%. The molecular etiology includes aneuploid (n = 7), microdeletion/microduplication syndrome (n = 40), and Mendelian disorders (n = 49). Conclusions These findings demonstrate the high yield of CS-based CNV. With further improvement of our CS-CNV pipeline, the method may have clinical utility for simultaneous evaluation of CNVs and small variations in samples referred for pre- or postnatal analysis.