Early cancer detection using multi-omic approach including epigenetic signal from ultra-small fragments.

e13561 Background: Genetic and epigenetic signals from plasma cell-free DNA (cfDNA) as well as proteins have been shown to detect cancer early. We have previously developed Circular Ligation and Amplification (CLAmp-seq) to detect mutations very sensitively and demonstrated higher performance than molecular barcode-based methods. The same single strand based technology yielded a much larger proportion of small fragments ( < 100 bp) than traditional methods typically relying on double stranded ligation, allowing us to investigate epigenetic signature of cancer in ultra-short fragments. We sought to measure the performance using a multi-omics approach of these genetic and epigenetic changes as well as proteins in detecting colorectal cancer (CRC), ovarian cancer (OC) and hepatocellular carcinoma (HCC). Methods: A healthy and a late stage cancer sample were assessed by whole genome sequencing (WGS) using CLAmp-seq. Then we analyzed cfDNA from plasma samples of 731 patients, including 69 CRC, 57 HCC, 49 OC patients and 556 age-matched healthy individuals. Out of the diseased samples, the numbers for stages I-IV are 49, 39, 71, and 16, respectively. CLAmp-seq WGS was performed on 58 healthy and 66 cancer samples to discover cancer epigenetic signature. In addition, all the samples were analyzed for a panel of proteins and a CLAmp-seq targeted panel that includes known mutation sites. Results: Using CLAmp-Seq in late stage cancer showed 33% of its fragments as smaller than 100 bp compared to 15% in healthy and < 1% in late stage by double stranded library prep. In addition, the difference in fragment size between late stage cancer and healthy was 29bp using CLAmp-Seq and 12bp using traditional double stranded prep. This focused our attention to detect epigenetic signature specific to cancer on the small fragments using CLAmp-Seq. Using data from whole genome analysis we demonstrated a performance using the epigenetic signature alone of 50% sensitivity at 97% specificity. Combined with mutations and proteins, we obtained at specificity of 97% sensitivities of 50%, 88%, 88%, and 100% in stage I, II, III, and IV, respectively. At the same 97% specificity we obtained the sensitivities of 73%, 100%, and 85% in CRC, OC, and HCC, respectively. Conclusions: We have demonstrated that CLAmp-Seq detects small fragments that are enriched in cancer. We have found predictive epigenetic signature in these small fragments. When combined with mutations and proteins we obtained a performance of 80% sensitivity at 97% specificity.