Stephens, Michelle D.

Genome-wide fragmentation patterns in cell-free DNA (cfDNA) in plasma are strongly influenced by cellular origin due to variation in chromatin accessibility across cell types. Such differences between healthy and cancer cells provide the opportunity for development of novel cancer diagnostics. Here, we investigated whether analysis of cfDNA fragment end positions and their surrounding DNA sequences reveals the presence of tumor-derived DNA in blood. We performed genome-wide analysis of cfDNA from 521 samples and analyzed sequencing data from an additional 2147 samples, including healthy individuals and patients with 11 different cancer types. We developed a metric based on genome-wide differences in fragment positioning, weighted by fragment length and GC content [information-weighted fraction of aberrant fragments (iwFAF)]. We observed that iwFAF strongly correlated with tumor fraction, was higher for DNA fragments carrying somatic mutations, and was higher within genomic regions affected by copy number amplifications. We also calculated sample-level means of nucleotide frequencies observed at genomic positions spanning fragment ends. Using a combination of iwFAF and nine nucleotide frequencies from three positions surrounding fragment ends, we developed a machine learning model to differentiate healthy individuals from patients with cancer. We observed an area under the receiver operative characteristic curve (AUC) of 0.91 for detection of cancer at any stage and an AUC of 0.87 for detection of stage I cancer. Our findings remained robust with as few as 1 million fragments analyzed per sample, demonstrating that analysis of fragment ends can become a cost-effective and accessible approach for cancer detection and monitoring.

Abstract Background: Accurate circulating biomarkers for detecting residual disease can help guide therapy decisions, particularly in early-stage cancer patients. However, currently available methods lack the sensitivity required to confidently assess the presence of residual disease in patients with low tumor burden. To address this need, we have developed TARDIS (Targeted Digital Sequencing), a personalized, multiplexed amplicon sequencing method capable of tracking as many as 100 or more mutations simultaneously. Methods: We obtained tumor biopsies and longitudinal plasma samples from patients with early-stage breast cancer, glioblastoma, and pancreatic cancer. Each tumor biopsy was analyzed whole-exome sequencing. Founder mutations were selected, accounting for copy number alterations (analyzed using sequenza) and a consensus allele fraction approach that combined pyclone and custom in-house methods. Patient-specific TARDIS primers were designed to detect these mutations in plasma cfDNA. Error suppression in TARDIS was achieved using a combination of unique molecular identifiers and fragment sizes to group sequencing reads into read families. Results: In 33 patients with early-stage breast cancer treated with neoadjuvant therapy, we targeted between 3 and 116 (mean 30) mutations per patient and analyzed between 1 and 4 longitudinal plasma samples using TARDIS. Prior to treatment, we detected ctDNA in 100% patients with Stage I-III breast cancer (n=32, 95% CI= 89%-100%). We detected tumor-specific mutations in 100% of baseline breast cancer plasma samples. After completion of neoadjuvant therapy and before surgery, ctDNA levels were significantly lower in patients with pathologic complete response (pathCR, no evidence of disease at surgery) compared to patients with residual disease (median tumor fractions 0.003% and 0.017%, respectively, p=0.0058, AUC=0.83). Conclusions: TARDIS enables highly sensitive detection of ctDNA in patients with nonmetastatic cancers. Analysis of longitudinal plasma samples using TARDIS holds promise for personalizing the extent of treatment in patients with curable disease. Multiple clinical validation studies across cancer types are ongoing to define quantitative thresholds for changes in ctDNA levels that could improve clinical decision making. Citation Format: Bradon R. McDonald, Tania Contente-Cuomo, Stephen-John Sammut, Michelle D. Stephens, Ahuva Odenheimer-Bergman, Brenda Ernst, Nieves Perdigones, Suet-Feung Chin, Maria Farooq, Rosa Mejia, Patricia A. Cronin, Karen S. Anderson, Heidi E. Kosiorek, Donald W. Northfelt, Ann E. McCullough, Bhavika K. Patel, Jeffrey N. Weitzel, Thomas P. Slavin, Carlos Caldas, Barbara A. Pockaj, Muhammed Murtaza. Personalized monitoring of treatment response using Targeted Digital Sequencing of circulating tumor DNA [abstract]. In: Proceedings of the AACR Special Conference on Advances in Liquid Biopsies; Jan 13-16, 2020; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(11_Suppl):Abstract nr A51.