Abstract 1978: Reference materials for measurable residual disease (MRD) monitoring

Here we describe reference materials for measurable residual disease (MRD) assay design and validation as well as initial data from testing. The analytical validation of liquid biopsy-based assays that attempt to monitor for the disappearance and reemergence of cancer can be challenging due to the need for reference materials that allow for the assessment of sensitivity and specificity at variant allele frequencies (VAFs) that can be over an order of magnitude below those that can be detected reliably by typical circulating tumor DNA (ctDNA) assays. At such low VAFs, a given plasma sample may only contain limited somatic variants - and only at a few copies - which is why some MRD assays focus on monitoring only patient-specific somatic variants and do not analyze other parts of the genome. However, this also means that validation should include the steps needed to monitor patient-specific somatic variants. Therefore, we designed our reference materials for a tumor/normal workflow as a set of three components. First, one cell line is a source of normal DNA that can be used to assess specificity. Second, a germline SNP-matched cell line provides hundreds to thousands of additional somatic variants. Third, blends of fragmented and sized DNA are used to mimic circulating cell-free DNA (ccfDNA) and serve as the input for MRD assays at VAFs from 0.1% to 0.01% (one mutant copy per 3.5 to 35 ng of ccfDNA) and at 0%. As a proof of concept, we analyzed three cell lines for somatic variants by Whole Exome Sequencing (WES). Using the resulting data, digital PCR assays were designed and two different error corrected NGS assays were customized to measure some of the identified variants. Based on past experiments, A>G (T>C) and C>T (G>A) changes had the highest background in NGS data, so these were generally avoided. The tumor and normal samples were used to assess the performance of a given customized assay to measure a variant. Digital PCR was used to verify VAFs in the ccfDNA format, which proved challenging for 0.01% given the expectation of approximately one positive droplet per assay well. NGS was then used to look for the variants using around 50 ng of input per reaction. While the detection of 0.1% VAFs appeared to be feasible with customized off-the-shelf assays, 0.01% was difficult given the need to measure a variant-containing region over 10,000 times to observe a somatic variant - if it was included in the library - and to measure each starting molecule around 10 times for error correction, which required a depth of greater than 100,000 for 20 of several hundred identified candidate variants. While we did not evaluate the reference materials with emerging clinical MRD assays and focused our initial testing on customized off-the-shelf assays, these reference materials should enable the design, validation, and evaluation of MRD assays. Citation Format: Yves Konigshofer, Matthew G. Butler, Jessica Dickens, Omoshile Clement, Andrew Anfora, Dan Brudzewsky, Bharathi Anekella, Russell K. Garlick. Reference materials for measurable residual disease (MRD) monitoring [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1978.