ASPO Abstracts
Can we use DNA from tumor samples for GWAS of cancer risk?
Category: Molecular Epidemiology & Environment
Conference Year: 2019
Abstract Body:
Purpose: We sought to determine if tumor samples could be utilized as a source of germline DNA to perform genome-wide association studies for rare cancers with high mortality earlyafter diagnosis, using adult glioblastoma as an example.Methods: Using data from a GWAS of chemotherapy-related toxicities in adults with glioblastoma treated on a cooperative group front-line therapy trial, we assessed the concordance of genotypes derived from DNA from blood samples vs tumor in 143 subjects. Genotypes were generated using Illumina Infinium-based SNP arrays. Tumor DNA was extracted from FFPE tissue either near the time of diagnosis or at the end of the trial.Buffy coat from blood collected at a clinic visit was frozen until extraction at the end of thetrial. Overall call rates were determined per sample and chromosome. Concordance of genotype calls was determined per sample and chromosome between blood- and tissue-based genotypes. Concordance was also compared based on tumor copy number variation and loss of heterozygosity.Results: The mean call rate for individual samples derived from tumor was 94.7 (range: 65.4-99.7). The mean concordance among tumor samples compared to blood for autosomal chromosomes in individual samples was 91.4% (range: 27.9-99.9). Samples that were genotyped using DNA extracted from FFPE closer to diagnosis had a consistently higher concordance (mean=93.82) compared to those extracted at the end of the trial (mean=84.72). Concordance was consistently lower for chromosomes 10, 13, and 21. These chromosomes also showed more areas of loss of heterozygosity and copy number alterations.Conclusions: In general, genotypes generated from glioblastoma tumor tissues had fairly good concordance with genotypes generated from blood samples. However, some individual tumors performed poorly with low call rates and high percent of missing genotypes. Interestingly,the performance varied by chromosome, indicating that somatic genomic alterations, especially loss of heterozygosity and copy number variations, contribute to poorer genotyping performance when using tumor DNA as the source. In particular, the areas that we identified with poor performance have been shown to exhibit high levels of copy number deletions in The Cancer Genome Atlas project.
Keywords: GWASmethodsrare cancers