Cell-Free DNA Whole Genome Sequencing for Non-Invasive MRD Detection in Multiple Myeloma

Introduction Accurate detection of minimal residual disease (MRD) is crucial for evaluating treatment efficacy in multiple myeloma (MM), yet current methods are invasive and often limited by bone marrow (BM) sample quality. We therefore compared standard MRD detection methods to whole-genome sequencing (WGS) of peripheral blood plasma cell-free DNA (cfDNA) to provide less invasive monitoring options. Methods The MM Molecular Monitoring (M4) prospective cohort study included 45 newly diagnosed transplant-eligible MM patients uniformly treated with standard of care frontline therapy at 8 Canadian sites. MRD testing was performed at 100 days post-autologous stem cell transplant (ASCT) (n=39) and/or after one year of lenalidomide (len) maintenance (n=33). We analyzed 43 patients by multiparameter flow cytometry (MFC) (71 samples, CytoQuest Technologies), 39 patients using EasyM (57 samples, Rapid Novor), 28 patients using clonoSEQ (Adaptive Technologies), and 18 patients using PET/CT imaging. We also performed 30-40X WGS on CD138+ selected BM cells pre-treatment initiation to identify somatic mutations (n=11) and tracked these mutations by 30-40X WGS in longitudinal peripheral blood cfDNA samples (cfWGS, n=12). Results MRD-negative rates at 100 days post-ASCT were lowest for EasyM at 0% (n=30), compared to 20% for cfWGS (n=5), 45% for clonoSEQ (n=11), and 49% for MFC (n=39). This trend persisted after one year of len maintenance, with MRD-negative rates of 22% for EasyM (n=27), 29% for cfWGS (n=7), 41% for clonoSEQ (n=17), 59% for MFC (n=32), and 83% for PET (n=18). Notably, among the EasyM-positive samples at 100 days post-transplant, only 21/27 remained positive after one year of maintenance therapy, likely due to delayed clearance of the M-protein. As of July 2024, 12/45 patients had relapsed, with an average time to relapse of 714 days (SD=375) after initiating len. At the 100 days post-ASCT timepoint, all samples from patients who later relapsed were positive by EasyM (n=6) and clonoSEQ (n=2), with 67% of samples (n=6/9) positive by MFC (mean proportion aberrant cells 0.006%, limit of detection (LOD) range 0.00038%-3.4%) and 50% (n=1/2) by cfWGS. After one year of len maintenance, all samples from relapsed patients were positive by EasyM (n=8), with 67% of samples positive by clonoSEQ (n=4/6), 78% by MFC (n=7/9, LOD 0.00035%-0.38%), 100% by cfWGS (n=3), and 17% by PET (n=1/6). All samples from patients who later relapsed which were negative by MFC and cfWGS showed detectable disease below the LOD, indicating the presence of potential subclinical residual disease. None of the patients who were EasyM-negative relapsed within two years of their sample collection (n=6). In contrast, relapse within two years occurred in 17% of clonoSEQ-negative (n=2/12), 5% of MFC-negative (n=2/38), 33% of cfWGS-negative (n=1/3), and 33% of PET-negative (n=5/15) samples. cfWGS demonstrated 82% concordance with EasyM (n=11), 67% with PET (n=6), 50% with MFC (n=12), and 25% with clonoSEQ (n=8). The lower concordance rates with MFC and clonoSEQ were primarily due to cfWGS detecting additional positive cases missed by these methods but identified as positive by EasyM. For example, cfWGS identified four clonoSEQ-negative cases as positive, with one case leading to relapse within a year. Conversely, two clonoSEQ-positive cases were below the LOD for cfWGS, highlighting the need for further method optimization. Conclusions Overall, cfWGS is a promising MRD testing alternative which offers less invasive monitoring than MFC and clonoSEQ. It demonstrated superior sensitivity to MFC and clonoSEQ, identifying residual disease missed by these methods but detected by EasyM. cfWGS may be particularly beneficial for non-secretory and some light chain only patients where EasyM is not currently feasible. Additionally, it provides information on clonal dynamics at progression that is not offered by any of the other technologies. However, these findings are based on preliminary data from a small cohort, requiring further validation in larger studies. Future research will focus on enhancing sensitivity and validating these findings in a broader patient population.