Search

Circulating cell-free DNA (ccfDNA) sequencing for low-burden cancer monitoring is limited by sparsity of circulating tumor DNA (ctDNA), the abundance of genomic material within a plasma sample, and pre-analytical error rates due to library preparation, and sequencing errors. Sequencing costs have historically favored the development of deep targeted sequencing approaches for overcoming sparsity in ctDNA detection, but these techniques are limited by the abundance of ccfDNA in samples, which imposes a ceiling on the maximal depth of coverage in targeted panels. Whole genome sequencing (WGS) is an orthogonal approach to ctDNA detection that can overcome the low abundance of ccfDNA by supplanting sequencing depth with breadth, integrating signal across the entire tumor mutation landscape. However, the higher cost of WGS limits the practical depth of coverage and hinders broad adoption. Lower sequencing costs may thus allow for enhanced ctDNA cancer monitoring via WGS. We therefore applied emerging lower-cost WGS (Ultima Genomics, 1USD/Gb) to plasma samples at ∼120x coverage. Copy number and single nucleotide variation profiles were comparable between matched Ultima and Illumina datasets, and the deeper WGS coverage enabled ctDNA detection at the parts per million range. We further harnessed these lower sequencing costs to implement duplex error-corrected sequencing at the scale of the entire genome, demonstrating a ∼1,500x decrease in errors in the plasma of patient-derived xenograft mouse models, and error rates of ∼10−7in patient plasma samples. We leveraged this highly de-noised plasma WGS to undertake cancer monitoring in the more challenging context of resectable melanoma without matched tumor sequencing. In this context, duplex-corrected WGS allowed us to harness known mutational signature patterns for disease monitoring without matched tumors, paving the way for de novo cancer monitoring.

In many areas of oncology, we lack sensitive tumor-burden monitoring to guide critical decision making. While circulating tumor DNA (ctDNA) promises to enable disease monitoring, this approach is limited by the sparsity of ctDNA in the plasma. To overcome this challenge, error-corrected deep targeted sequencing has been proposed. Nonetheless, this framework is limited by the low number of genomic equivalents (GEs, ~103/mL of plasma), imposing a ceiling on effective sequencing depth. We have previously shown that genome-wide mutational integration through plasma whole genome sequencing (WGS) can sever the dependency between available GEs and assay sensitivity (Zviran et al, 2020). In this approach, tumor-informed mutational profiles are applied to plasma WGS, allowing detection of tumor fractions as low as 10−5. However, the higher cost of WGS limits practical depth of coverage (20-30X) and may limit broad adoption. Lower costs may thus allow for enhanced ctDNA cancer monitoring via WGS. We therefore applied emerging lower-cost WGS (1USD/Gb, Almogy et al, 2022) to plasma from 7 patients with metastatic cancer at ~115x coverage depth. Read depth profiling and error rates were comparable between matched Ultima and standard platform datasets. Integration of deep learning architectures for signal to noise enrichment (Widman et al, biorxiv, 2022) with deeper WGS coverage enabled ctDNA detection at the parts per million range. We reasoned that lower sequencing cost can be harnessed for duplex error-corrected WGS. Proof-of-concept experiments in mouse PDX samples showed ~1,500x decrease in errors. Applied to the plasma of stage IV melanoma patients (n=5), we obtained error rates ~10−7. We used this approach to tackle the challenging context of cancer monitoring in early-stage melanoma without matched tumor sequencing. While in uncorrected WGS, de novo mutation calling yielded limited ability to detect melanoma specific mutations, duplex-corrected WGS allowed us to harness melanoma mutational signatures for disease monitoring without matched tumor profiling. Analytic validation of our assay showed sensitive and specific cancer detection when the concentration of ctDNA was at 10−4 concentrations. Applied to a cohort of stage III melanoma patients with negative ctDNA detection using previously described methods, we detected ctDNA in all cases (n=4), demonstrating enhanced sensitivity using duplex WGS. These data demonstrate the exciting potential of low cost WGS for ultra-sensitive ctDNA cancer monitoring. In the tumor-informed setting, deeper sequencing increased sensitivity for mutational profile detection. Moreover, the application of duplex error-correction at genome scale allowed for sensitive cancer monitoring without matched tumor profiles. We envision that the era of low-cost sequencing will empower ultra-sensitive cancer monitoring via WGS, with transformative impact on cancer care. Citation Format: Alexandre P. Cheng, Adam J. Widman, Anushri Arora, Itai Rusinek, William F. Hooper, Rebecca Murray, Daniel Halmos, Theophile Langanay, Giorgio Inghirami, Soren Germer, Melissa Marton, Adrienne Helland, Rob Furatero, Jaime McClintock, Lara Winterkorn, Zoe Steinsnyder, Yohyoh Wang, Srinivas Rajagopalan, Asrar I. Alimohamed, Murtaza S. Malbari, Ashish Saxena, Margaret K. Callahan, Dennie T. Frederick, Lavinia Spain, Ariel Jaimovich, Doron Lipson, Samra Turajlic, Michael C. Zody, Nasser K. Altorki, Jedd D. Wolchok, Michael A. Postow, Nicolas Robine, Genevieve Boland, Dan A. Landau. Whole genome error-corrected sequencing for sensitive circulating tumor DNA cancer monitoring. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5709.

In solid tumor oncology, circulating tumor DNA (ctDNA) is poised to transform care through accurate assessment of minimal residual disease (MRD) and therapeutic response monitoring. To overcome the sparsity of ctDNA fragments in low tumor fraction (TF) settings and increase MRD sensitivity, we previously leveraged genome-wide mutational integration through plasma whole genome sequencing (WGS). We now introduce MRD-EDGE, a composite machine learning-guided WGS ctDNA single nucleotide variant (SNV) and copy number variant (CNV) detection platform designed to increase signal enrichment. MRD-EDGE uses deep learning and a ctDNA-specific feature space to increase SNV signal to noise enrichment in WGS by 300X compared to our previous noise suppression platform MRDetect. MRD-EDGE also reduces the degree of aneuploidy needed for ultrasensitive CNV detection through WGS from 1Gb to 200Mb, thereby expanding its applicability to a wider range of solid tumors. We harness the improved performance to track changes in tumor burden in response to neoadjuvant immunotherapy in non-small cell lung cancer and demonstrate ctDNA shedding in precancerous colorectal adenomas. Finally, the radical signal to noise enrichment in MRD-EDGE enables de novo mutation calling in melanoma without matched tumor, yielding clinically informative TF monitoring for patients on immune checkpoint inhibition.