ABSTRACT Next-generation sequencing (NGS) has become increasingly integral to the practice of clinical oncology, where its ability to scalably examine hundreds to thousands of targets now routinely enables identification of prognostic and therapeutically actionable markers that support the practice of precision medicine. There are many applications for which it would be useful to detect and quantitate cancer-associated genotypes at ultra- low levels (<1 in 10,000 or more), such as identifying drug-resistance mutations in tumors, detecting residual cancer cells after therapy, or early cancer detection. Nevertheless, standard NGS technologies are hampered by a relatively high error rate (~1 in 100bp), below which true biological variation cannot be distinguished from noise. Various methods have been proposed to bypass this issue by allowing error correction of NGS sequence reads, but such techniques require redundantly sequencing individual template molecules at high depth such that an error-corrected consensus sequence can be produced. As a result, those methods require a large amount of sequencing power, are costly, and are limited in the number of specimens and genomic targets that can be examined. They have consequently seen little uptake in clinical use. There remains an unmet need for highly accurate sequencing methods that are cost-effective, scalable, and allow interrogation of enough gene targets for meaningful use in clinical practice. We have recently developed a new experimental paradigm, termed “Linked Duplex Sequencing”, that addresses these limitations. In our approach, we join the two strands of DNA from an initial template fragment into a single, covalently linked molecule. Error correction of the duplex can be performed by comparing separate reads from the two linked strands, thereby eliminating the need for redundant sequencing of template molecules. This innovative technology provides robust error correction with scalability, cost-effectiveness, efficiency, and quantitative precision, and is compatible with low-to-mid output short read sequencing platforms (ie, Illumina) that are already in widespread clinical use. In our first Aim, we will develop workflows to support Linked Duplex Sequencing, will develop supportive bioinformatic analysis pipelines, and will characterize the cardinal performance metrics of the approach using fresh and formalin-fixed reference material. In our second Aim, we will develop protocols for the targeted enrichment of genes or variants of interest for Linked Duplex Sequencing, and will evaluate performance using a variety of clinical materials. This work will provide information and deliverables with immediate, direct, and transformative benefit to cancer patients by substantially improving the quality of oncology sequencing assays while imbuing them with enhanced diagnostic capabilities for the ultrasensitive detection of cancer associated mutations relevant to disease emergence, relapse, and therapy resistance in routine clinical practice. Our goal is to make ultrasensitive, error corrected sequencing so inexpensive and straightforward that it will be used as standard operating procedure for NGS clinical oncology assays and cancer research studies.

PROJECT NARRATIVE Next-Generation DNA sequencing is now widely used in clinical oncology to detect mutations in cancers that inform prognosis and therapeutic options, but is limited in its sensitivity for low-prevalence mutations. This proposal outlines the development and validation of inexpensive, efficient, and ultrasensitive methods for detecting cancer-relevant mutations in clinical specimens. These technologies will be immediately translatable and practical for clinical diagnostics, and will provide enhanced capabilities in a broad range of applications, including the identification of emerging cancer subclones and ultrasensitive screening for the emergence or recurrence of malignancy.