Abstract P011: Clonal hematopoiesis and risk of lethal prostate cancer: a prospective cohort study with long-term follow-up

Background: Clonal hematopoiesis (CH), the presence of acquired mutations in leukemia driver genes, promotes systemic inflammation and is common among aging men. Prostate cancer, a subset of which is lethal, also develops in aging men. We hypothesized that CH contributes to development of lethal prostate cancer. Methods: We conducted nested case-cohort studies for metastatic prostate cancer and prostate cancer-specific death (lethal prostate cancer) within the prospective Health Professionals Follow-up Study. First, we followed 1155 men free of prostate cancer and cardiovascular disease at blood draw (1993-1995) for development of lethal prostate cancer over up to 26 years. Second, we followed 532 men with incident non-metastatic prostate cancer for development of lethal prostate cancer. We sequenced blood DNA from 1488 participants for putative CH driver mutations in the 9 most common CH-defining genes with a custom targeted panel (VariantPlex, Invitae, Inc.) at ultra-high depth (mean, 18,000x), employing unique molecular identifiers for error correction. CH variant calling used a novel ensemble calling approach, ArCCH, validated with in-silico tumor dilutions and blinded technical replicates, which had high accuracy for variant allele frequencies (VAFs) as low as 0.1%. We estimated hazard ratios (HRs) with 95% confidence intervals (CIs) in proportional hazards regression with Prentice case-cohort weights. Results: In a random sample of 968 men initially free of prostate cancer and cardiovascular disease (median age at blood draw 60 years, interquartile range 52 to 67), 80% of men had CH at a variant allele frequency (VAF) of >0.1%, 15% had VAFs >2%, and 3% had VAFs >10%; 75% of men had variants in epigenetic modifier genes (DNMT3A, TET2, ASXL1), and 21% had variants in DNA repair genes (PPM1D, TP53, CHEK2). CH burden was strongly age-associated, with approximately one additional CH variant per decade of age at blood draw (mean difference 1.07 variants, 95% CI 0.93 to 1.21). Among men initially free from prostate cancer, after adjusting for age at blood draw, CH clones between 0.1% and 10% VAF were not related to lethal prostate cancer (206 events total; HR 0.93, 95% CI 0.49-1.76 for VAFs 2-10% vs. no variants detected at >0.1% VAF). Results for epigenetic modifiers and DNA repair genes were similar. While inconclusive, data were compatible with positive associations among younger men (< 65 years) or for VAFs >10%. Among men initially diagnosed with non-m