Feasibility of low-throughput next generation sequencing for germline DNA screening (Clin Chem, Dec 2014)

Nur Sabrina Sapari1, Eiram Elahi1, Mengchu Wu1, Marie Loh2, Hong Kiat Ng1, Xiao Han1, Hui Ling Yap3, Thomas P. Klemm4, Brendan Pang5, Touati Benoukraf1, Yik Ying Teo6, Barry Iacopetta7, Soo Chin Lee3 and Richie Soong1,5

1 Cancer Science Institute of Singapore, National University of Singapore, Singapore;
2 Department of Epidemiology and Biostatistics, Imperial College London, London, United Kingdom;
3 Department of Haematology Oncology, National University Cancer Institute, National University Health System, Singapore;
4 Illumina Inc., San Diego, CA;
5 Department of Pathology, National University Health System, Singapore;
6 School of Public Health, National University Health System, Singapore;
7 School of Surgery, The University of Western Australia, Perth, Australia.

Next generation sequencing (NGS) promises many benefits for clinical diagnostics. However, current barriers to its adoption include suboptimal amenability for low clinical throughputs and uncertainty over data accuracy and analytical procedures. We assessed the feasibility and performance of low-throughput NGS for detecting germline mutations for Lynch syndrome (LS).

Sequencing depth, time, and cost of 6 formats on the MiSeq and Personal Genome Machine platforms at 1-12 samples/run were calculated. Analytical performance was assessed from 3 runs of 3 DNA samples annotated for 7500 nucleotides by BeadChip arrays. The clinical performance of low-throughput NGS and 9 analytical processes were assessed through blinded analysis of DNA samples from 12 LS cases confirmed by Sanger sequencing, and 3 control cases.

The feasibility analysis revealed different formats were optimal at different throughputs. Detection was reproducible for 2619/2635 (99.39%) replicate variants, and sensitivity and specificity to array annotation were 99.42% and 99.99% respectively. Eleven of 16 inconsistently detected variants could be specifically identified by having allele frequencies ≤ 0.15, strand biases >-35, or genotype quality scores ≤ 80. Positive selection for variants in the Human Genome Mutation Database (colorectal cancer, nonpolyposis) and variants with ≤ 5% frequency in the Asian population gave the best clinical performance (92% sensitivity, 67% specificity).

Low-throughput NGS can be a cost-efficient and reliable approach for screening germline variants; however, its clinical utility is subject to the quality of annotation of clinically relevant variants.