Abstract LB-048: Copy number and loss of heterozygosity (LOH) analysis in 52 breast cancer FFPE samples using molecular Inversion probe array: detailed analysis of reproducibility and performance compared to NGS platforms

Abstract

Abstract Introduction: Somatic mutations are routinely identified using NGS cancer panels but these panels lack genome-wide coverage for copy number (CN) and LOH analysis. To investigate mutation, CN, and LOH in late-stage breast cancer we tested >50 samples using the Oncoscan molecular inversion probe (MIP) array and evaluated its reproducibility and performance compared to NGS platforms. Methods: 52 breast cancer samples (stage IIIA - IV) were analyzed using MIP array (Oncoscan, Affymetrix). Four samples were tested in technical triplicates to determine assay reproducibility. In addition, 28 samples were sequenced by amplicon-based NGS and five of these samples were also tested using a capture-based NGS platform for mutation, CN, and LOH comparison. Results: MIP array provided highly reproducible results for CN and LOH, with >98% of calls showing CN range in the technical triplicates of < 0.5 copy for 891 cancer genes analyzed. Variability in CN seems to be proportional to absolute copy number at the tested locus, with CN range in technical triplicates of >2 copies seen in only two cases, once for ERBB2 (CN range 32 - 35 copies) and once for FLOT2 (CN range 22 - 25 copies). Gene level results were then categorized in five groups: homozygous deletion, single copy loss, diploid, low grade amplification (≤6 copies), or high grade amplification (>6 copies). Using these predetermined cut points, we saw >99% concordance rate among the technical replicates in the MIP array. We found a 93% concordance rate between MIP array and CN/LOH calls by capture-based NGS. Discordant calls between NGS and MIP array were either LOH calls or single copy number change (diploid vs. single copy loss or gain). MIP array mutation analysis of 28 samples showed good sensitivity, correctly detecting the 17 PIK3CA mutations and one TP53 mutation identified by NGS in this cohort. There were seven false positive calls by MIP array, five of them occurring in two genotypes (2x NRAS G12S/C, and 3x EGFR L858R). The other two false positives occurred in PIK3CA, with one false positive (H1047L) occurring in association with a high-grade PIK3CA amplification (7 copies). Increasing CN at the mutation locus was associated with a higher mutation score provided by MIP array (p<0.0001), which may explain some false positive calls. Conclusions: MIP array platform provides a great alternative for assessing CN and LOH in FFPE samples at lower cost and using less input DNA than NGS (80ng vs. 250ng). There was good correlation between CN and LOH results from MIP array and capture-based NGS, with discordant results limited to small CN differences or LOH calls. Mutation analysis by MIP array showed no false negatives when compared to NGS, while false positives seem to occur either due to probe-specific issues or in association with amplifications at the genotyping locus. Citation Format: Candice L. Horn, Fabio Nunes, John Calley, Steven Bray, Isabella Wulur, Mark Farmen, Robert Gallavan, Iris Halfpenny, Paul Medlow, Keith McGreeghan-Crosby, Gera Jellema. Copy number and loss of heterozygosity (LOH) analysis in 52 breast cancer FFPE samples using molecular Inversion probe array: detailed analysis of reproducibility and performance compared to NGS platforms. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-048. doi:10.1158/1538-7445.AM2015-LB-048