Accurate genome-wide germline profiling from decade-old archival tissue DNA reveals the contribution of common variants to precancer disease outcome

BackgroundInherited variants have been shown to contribute to cancer risk, disease progression, and response to treatment. Such studies are, however, arduous to conduct, requiring large sample sizes, cohorts or families, and more importantly, a long follow-up to measure a relevant outcome such as disease onset or progression. Unless collected for a dedicated study, germline DNA from blood or saliva are typically not available retrospectively, in contrast to surgical tissue specimens which are systematically archived. ResultsWe evaluated the feasibility of using DNA extracted from low amounts of fixed-formalin paraffin-embedded (FFPE) tumor tissue to obtain accurate germline genetic profiles. Using matching blood and archival tissue DNA from 10 individuals, we benchmarked low-coverage whole-genome sequencing (lc-WGS) combined with genotype imputation and measured genome-wide concordance of genotypes, polygenic risk scores (PRS), and HLA haplotypes. Concordance between blood and tissue was high (r2>0.94) for common genome-wide single nucleotide polymorphisms (SNPs) and across 22 disease-related PRS (mean r=0.93). HLA haplotypes imputed from tissue DNA were 96.7% (Class I genes) and 82.5% (Class II genes) concordant with deep targeted sequencing of HLA from blood DNA. Using the validated methodology, we estimated breast cancer PRS in 36 patients diagnosed with breast ductal carcinoma in situ (11.7 years median follow-up time) including 22 who were diagnosed with breast cancer subsequent event (BSCE). PRS was significantly associated with BCSE (HR=2.5, 95%CI: 1.4-4.5) and the top decile patients were modeled to have a 24% chance of BCSE at 10 years, hence suggesting the addition of PRS could improve prognostic models which are currently inadequate. ConclusionsThe abundance and broad availability of archival tissue specimens in oncology clinics, paired with the effectiveness of germline profiling using lc-WGS and imputation, represents an alternative cost and resource-effective alternative in the design of long-term disease progression studies.