Ultra-low coverage exome sequencing of FFPE tumor specimens identifies exposure to carcinogenic aristolochic acid

Dietary intake of Aristolochia plants containing aristolochic acid (AA, IARC Group 1 carcinogen), leads to renal disease and to recurrent urinary tract transitional cell carcinomas (TCC). Mutation screens identified frequent A>T transversions, alterations underrepresented in other cancer types, as the signature of AA exposure. AA- associated TCC is a preventable cancer yet millions are estimated at risk due to unregulated medicinal use of Aristolochia. Here we present an ultra-low coverage modification of whole-exome sequencing (ULC-WES) as a novel molecular epidemiology method that reliably detects the AA mutation signature in small amounts of DNA from FFPE tumor tissues and can identify new exposed populations. DNA was isolated from paraffin sections of 18 TCC tumors from endemic nephropathy (EN) patients from regions in Croatia and Bosnia linked to exposure to AA. Four non-EN upper tract TCCs were used as controls. 250 ng of DNA was used for sequencing libraries and exome enrichment (Nimblegen, Roche). Libraries were sequenced on Illumina HiSeq2500 at 10-12x coverage. After data processing, GATK was used to call single nucleotide variants (SNV). To assess platform independence, 7 samples were run on SOLID 5500XL (Life), using proprietary chemistry and pipeline for data processing and variant calling. Normal genetic variants were filtered out after SNV annotation. ULC-WES at 10x identified the A>T predominance in 17 of 18 EN tumors (94%), and at T pattern in 14 (78%) samples. In contrast, the A>T presence was T mutations in the genome (R^2=0.83 for mutation count correlated with the chromosome length) but also identified recurrent non-synonymous A>T mutations in tumor suppressors and oncogenes including TP53, ARID1A, BRCA1/2, AXIN1, GATA3 and NRAS, DNMT1, FLT3, CSF1R and PIK3CA. We next detected recurrent A>T mutations in a number of histone and DNA modifiers suggesting epigenetic deregulation of genome activity in the AA-associated TCCs. In 2 patients, pairs of distinct TCCs from the same surgery had markedly overlapping mutation patterns (624 SNVs shared between same-side pelvic and ureteric tumors in one case, and 191 shared between renal pelvic and bladder tumors in the other). Thus, ULC-WES provided evidence supporting the concept of tumor cell seeding along the urinary tract as a mechanism of TCC spread and recurrence. We conclude that ULC-WES is a sensitive and specific method to identify AA exposure in tumor tissue and is directly applicable to identification of new populations at risk. The method can ultimately facilitate evidence-based preventive measures in the affected regions.