Abstract Chronic lymphocytic leukemia (CLL) is characterized by multiple copy number abnormalities (CNVs) with prognostic value. Identifying these structural variations is central to defining CLL pathogenesis, risk stratification, and therapeutic approaches. Fluorescence in situ hybridization (FISH) is the clinical gold standard in detecting prognostic CNVs in CLL. However, next-generation sequencing (NGS) techniques have become more readily available for clinical genomic applications and can also be used to identify CNVs. Here we present bioinformatic methods to accurately identify CNVs in CLL using NGS data. We used the CNV-calling algorithm PatternCNV to detect clinically relevant CNVs: deletion 17p13 [del(17p)], deletion 11q23 [del(11q)], deletion 13q14 [del(13q)], and trisomy 12. PatternCNV was run on 2274 samples (1500 somatic and 774 germline samples) from six different sequencing batches, screened using a targeted sequencing panel that covers all exons of 59 recurrently CLL mutated genes and additional amplicons covering the minimal affected regions of relevant CNVs. To correct for potential batch effects, PatternCNV was initially run to quantify exon coverage behavior without the chromosomes containing recurrent CNV events, 11, 12, 13, and 17. Principal component analyses and correlation matrices were analyzed, grouping the samples into four distinct clusters that contain similar exon coverage patterns. Samples in each of the four clusters were then independently re-run through PatternCNV using all chromosomes. Visual analysis of CNV plots revealed a bias in normalization. To correct this, the log2ratios were corrected to center the log ratio on the median coverage. Sample noisiness was calculated from the difference in the median absolute deviation (DiffMAD) and samples with a DiffMAD score greater than 0.3 were excluded. All CNV analyses were blinded to clinical FISH results. The effectiveness of our CNV calling was evaluated in 522 CLL patients who had FISH conducted within three months of the sample date. We excluded samples with low tumor metrics identified by FISH (less than 20% of cells with either del(17p), del(11q), trisomy 12 or del(13q)). When we compared our CNV analyses with the FISH data, we found high concordance 99.6% for del(17p), 97.5% for del(11p), 99.1% for trisomy 12, and 93.7% for del(13q). N=46 total discordant pairs were identified, with the highest discordance for del(13q), N=28, followed by del(11q), N=12. These novel bioinformatic methods allow for accurate detection of CNVs across NGS sequencing batches. The high concordance in detecting CNVs between targeted NGS and the gold standard of FISH, suggest NGS is an accurate tool for calling CNVs in CLL. Further, NGS can infer clinically relevant CNVs in genomic locations not targeted by FISH. Citation Format: Chantal E. McCabe, Erik Jessen, Daniel R. O'Brien, Julia E. Wiedmeier-Nutor, Susan L. Slager, Esteban Braggio. Identifying copy number variations in chronic lymphocytic leukemia using targeted next generation sequencing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3352.
65
