Abstract 6694: Evaluation of a multi-omics approach to molecular residual disease detection

Abstract

Abstract Introduction: Advancements in DNA sequence detection technology and assay design have enabled molecular residual disease (MRD) assays to detect cancer earlier and with greater sensitivity. In Tumor-Informed (TI) strategies, mutations from the tumor are used to define a set of patient-specific markers that are tracked in the blood as circulating tumor DNA (ctDNA). Whereas tumor-naïve (TN) approaches do not require prior genomic information from the tumor and instead employ a fixed panel of marker(s). Methylated DNA Markers (MDMs) can be used to distinguish cancer from non-cancer and can be found in plasma. We have developed a proprietary TN MRD platform for colorectal cancer (CRC) capable of measuring as few as 1-5 strands of MDM target. Additionally, we are actively developing a TI MRD assay platform. Here we describe the evaluation of both TI and TN approaches for the detection of circulating tumor DNA (ctDNA) in blood across all stages of colorectal cancer and explore the complementarity of these two approaches in the context of MRD monitoring. Procedures: A 20-sample cohort of paired tumor/normal and pre-surgery plasma specimens from stage I (6/20), stage II (7/20), stage III (6/20), and stage IV (1/20) CRC patients and 20 demographically matched non-cancer controls were acquired for our study. All cancer plasma samples were collected prior to tumor resection. Neat and diluted plasma samples were tested for the presence of ctDNA containing short variants specifically derived from the corresponding primary tumor and for a pre-defined set of MDMs and protein cancer biomarkers. Tumor-specific short variants were characterized by whole exome sequencing of the primary tumors. MDMs were detected from bisulfite converted cfDNA by Target Enrichment Long probe Quantitative Amplified Signal (TELQAS) for 35 MDMs. A cancer detection classifier for the TN assay was trained and validated using a separate cohort of non-cancer and CRC plasma samples and employed for this study. Protein was detected by high throughput ELISA.Results: Tumor sequencing of the 20 CRC samples identified between 122 and 7146 (median = 230) somatic mutations per tumor sample. The combination of features from TI and TN assays achieved 95% accuracy, demonstrating an improvement over either assay in isolation. Indeed, 5 “positive” calls were made for cancer samples for which only one of the two methods met pre-defined RUO thresholds for ctDNA detection. Comparison of median variant allele fraction from the TI assay with median percent methylation of CRC markers for these CRC samples had a Pearson’s correlation of 0.967. Conclusion: TI and TN assays show high concordance in cancer detection, but also complement each other. Additionally, ctDNA quantifications produced from TI and TN assays are linearly correlated. The combination of data from both assays can be used to further differentiate cancer samples from non-cancer samples. Citation Format: Alanna Chan, Martin You, Oliver Limbo, Billyana Tsevtanova, Gerald Dodson, Brenda Curiel, Karma Farhat, Samantha Scott, Archana Ramesh, Jacquelyn Hennek, Viacheslav Katerov, Michael Kaiser, Hatim Allawi, Brady P. Culver, Michael Reshatoff, Jorge Garces, Gina Costa. Evaluation of a multi-omics approach to molecular residual disease detection. [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 6694.