Frequent LOH of CYP2D6 in ER+ breast cancer determined by next-generation sequencing (NGS)

• Published in: Journal of clinical oncology Vol. 31; no. 15_suppl; p. 534
• Main Authors: Ratain, Mark J., Sun, James, Nakamura, Yusuke, Cox, Nancy J, Sahmoud, Tarek, Stephens, Phil, Chen, David, Miller, Vincent A., Ross, Jeffrey S., Yelensky, Roman
• Format: Journal Article
• Language: English
• Published: 20.05.2013
• ISSN: 0732-183X; 1527-7755
• DOI: 10.1200/jco.2013.31.15_suppl.534

Abstract only 534 Background: The role of CYP2D6 genetic variation in predicting response to tamoxifen in ER+ breast cancer is a subject of ongoing debate. There has been great variability in approaches to both genotyping and phenotyping, and in particular many investigators have extracted DNA from breast cancer samples rather than peripheral blood. We hypothesized that CYP2D6 gene copy number alterations are common in ER+ breast cancer, affecting genotype results, and used NGS to characterize CYP2D6 in patients with ER+ disease. Methods: CYP2D6 sequencing was performed as part of a comprehensive NGS profile of cancer-related genes for 261 predominantly relapsed and metastatic ER+ breast cancer FFPE specimens. Sequence data were resolved into genotypes according to the * allele nomenclature. Tumor LOH was determined at CYP2D6, and its error impact on genotyping methods was estimated. To assess biological significance, the prevalence of CYP2D6 alleles and LOH in ER+ disease was compared against a control set of 99 ER- tumors. Results: CYP2D6 allele frequencies in our full cohort (ER+, 261; ER-, 99) were consistent with prior studies; 64.4%, 16.8%, 9.0% vs. 63.1%, 17.2%, 7.0% for *1/*2, *4, and *41 respectively, and 1%-2% for the rarer alleles *9, *10, and *5. The rate of CYP2D6 LOH was higher in ER+ disease (41% vs. 26%, p<0.01), with all excess arising from copy-loss (as opposed to copy-neutral) changes (22% vs. 7%, p<0.002). The estimated impact of LOH on germline genotype assessment from tumor was considerable; an assay sensitive at >20% minor allele frequency (e.g., Sanger sequencing) can misclassify >10% of heterozygotes, leading to significant Hardy-Weinberg disequilibrium (e.g., p=8.3x10 -8 for *4). Interestingly, an enrichment of reduced or non-functional CYP2D6 alleles in ER+ samples was observed (61% vs. 47%, p<0.03). Conclusions: Our results demonstrate the distorting effect of extensive LOH on genotype assessment of CYP2D6 in breast cancer. Therefore, tumor DNA should not be routinely used for determination of germline 2D6 genotype, although it appears possible to use NGS. The apparent association between reduced function CYP2D6 alleles and ER+ breast cancer in our dataset requires further investigation.