Abstract 783: Epigenomic detection of multiple cancers in plasma derived cell free DNA

Background: Our feasibility study employed a novel genomic detection methodology that enriches 5-hydroxymethylcytosine (5hmC) loci in cell free DNA (cfDNA) from the plasma of cancer patients using click chemistry coupled with sequencing and machine learning based classification methods. These classification methods were developed to detect the presence of disease in the plasma of cancer and control subjects. Cancer and control patient cfDNA cohorts were accrued from multiple sites consisting of 48 breast, 55 lung, 32 prostate and 2 pancreatic datasets consisting of 41 and 53 cancer subjects (Set 1 and 2). In addition, a control cohort of 260 subjects (non-cancer) was employed to match cancer patient demographics (age, sex and smoking status) in a case-control study design. Methods: Machine learning methods, applied to each cancer case cohort individually, with a balancing non-cancer cohort, were able to classify cancer and control samples. Measures of predictive performance using 5-fold cross validation coupled with out-of-fold Area Under the Receiver Operating Characteristic Curve (AUROC) measures were employed. Gene sets selected as part of biomarker discovery were further analyzed for disease relevance using pathway analysis tools (GSEA, mSigDB). Results: 260 controls and 229 cancers from four disease types (breast, lung, pancreas and prostate) were analyzed; more than 60% of cancer patients had early stage disease (I or II). Predictive performance, employing AUROC measures, was established for breast (0.89), lung (0.84), pancreas (set 1 - 0.95 and 2 - 0.93) and prostate (0.83). The genes defining each of these predictive models were enriched for pathways relevant to disease specific etiology, notably in the control of gene regulation in these same pathways. The breast cancer cohort consisted primarily of stage I and II patients including tumors < 2 cm and these samples exhibited a higher prediction probability score. The prostate cancer cohort consisted of both indolent and aggressive disease sample and prediction performance was equally high for both (AUROC for indolent vs aggressive was 0.81 and 0.77, respectively). Conclusions: These findings suggest that 5hmC changes in cfDNA enable non-invasive detection of early stage breast, pancreatic, prostate, and lung cancers. Furthermore, 5hmC profiling in cfDNA may enable the prediction of clinically relevant features such as tumor size in breast adenocarcinoma or indolent disease in prostate cancer. Final