Liquid biopsy for non-invasive molecular characterization and measurable residual disease analysis in multiple myeloma

Abstract

Abstract Introduction: Molecular profiling and measurable residual disease (MRD) assessment have become essential in multiple myeloma (MM) decision-making. Although bone marrow (BM) aspirates are the standard for MRD measurement and molecular characterization, focal lesions and clonal heterogeneity can render BM analyses unrepresentative of the total tumor burden, particularly in relapsed or refractory settings. Plasma-derived circulating tumor DNA (ctDNA) offers a minimally invasive alternative that may better capture the total disease burden. Methods: 20 MM patients (3 patients in complete response (CR), 2 in very good partial response (VGPR), and 15 in progressive disease (PD)) were included in this proof-of-concept study. Biomarkers were analyzed in baseline samples by next generation flow cytometry (NGF), and additionally by targeted capture next generation sequencing (NGS) using the EuroClonality NDC panel (ECNDC) of genomic DNA from CD138-enriched BM cells and ctDNA from peripheral blood plasma to detect clonal immunoglobulin (IG) rearrangements, single nucleotide variants (SNVs), and structural variants (SVs). A follow-up ctDNA ECNDC analysis was conducted up to 9 months after initiation of salvage therapy. Data were analyzed by ARResT/Interrogate – an interactive immunoprofiler for IG and T-cell receptor NGS data – with an adapted pipeline for targeted capture in cell-free DNA (cfDNA).Targets for initial disease characterization were called with at least ≥3 unique reads at ≥0.5% variant allele frequency (VAF) and for MRD analysis with at least ≥2 unique reads. In addition, NMR metabolomics, glycoproteomics, mass spectrometry-based proteomics, and Olink® technology, i.e. the use of paired antibodies with DNA tags that generate a quantifiable DNA signature when binding to the target protein, were conducted on plasma samples taken at screening and during the 4 follow-ups from both therapy responders and non-responders to explore changes in protein and cytokine composition across disease states. Results: NGF and NGS from both BM and cfDNA were available for 10 of the 20 patients, with MM detectable by NGF in 6 and undetectable in 4 patients. All NGF-positive patients showed concordant BM molecular markers, whereas none were detected in NGF-negative BM samples. At baseline, ctDNA identified markers in 4 of 6 NGF-positive patients, but also in 3 of 4 NGF-negative cases. Furthermore, ctDNA analysis mirrored BM clonal IG rearrangements and variants while also detecting mutations (e.g., NRAS, KRAS, IKZF1) that were absent in the BM samples. Longitudinal ctDNA tracking showed persistent positivity in PD and clearance in responding patients, with 1 case of molecular relapse preceding clinical progression. Plasma proteomics and Olink® revealed distinct profiles between active and controlled disease, with intra-individual expression differences (e.g., IL-5, IL-6, MSLN). Metabolomic analyses discovered decreased creatine levels and increased HDL levels along the response gradients (PD → VGPR → CR). Glycoproteomics enabled separation of response groups in partial least squares-discriminant analysis (PLS-DA) machine learning (cross-validated area under the ROC curve(AUROCcross-val) = 0.87), suggesting potential for predictive modeling. Conclusion: ctDNA analysis provides a minimally invasive, comprehensive molecular snapshot of MM across a diverse clinical cohort, complementing BM analyses by overcoming sampling limitations. Integration of ctDNA with plasma-based omics may enable minimally invasive, comprehensive disease monitoring and support personalized treatment strategies.