Background Better prediction models are required to identify patients with high-risk smoldering multiple myeloma (SMM) who can benefit from early intervention therapy before progression. Several studies have reported heterogeneity of genetic abnormalities found at the SMM stage, but whether any mutations are sufficient to cause, or instead prevent progression, remains an important open question. We leveraged whole-genome sequencing (WGS) in a large cohort of precursor patients, including repeated assessment over time, to discover drivers of disease growth and monitor malignant transformation to multiple myeloma (MM). Methods WGS was performed on tumor and normal cells from 141 untreated patients with SMM or Monoclonal Gammopathy of Undetermined Significance (MGUS) from the PCROWD study. Moreover, this dataset was combined with genomic data from the literature totaling 1,034 patients with WGS and/or exomes sampled across MM stages for comparative analysis and biomarker discovery. MutSig2CV, GISTIC2, and a novel method for structural variants were used to establish the list of candidate drivers across disease stages, with statistical power to detect myeloma drivers from points mutations found in ≥2% of patients. Longitudinal WGS data were analyzed with the PhylogicNDT suite of tools to characterize clonal competition in treated and untreated patients. Results Patients were followed for a median time of 20 months from the sample date (range: 3 to 78 months), during which 11 progressed. Patients evenly distributed across 20/2/20 stages: 44 Low-Risk (37%), 29 Intermediate-Risk (24%), 47 High-Risk (39%) SMM were combined with 20 MGUS, showing our method is reliable to perform WGS in lower disease burden settings. A validation cohort comprising 61 SMM with a median follow-up time of 36 months (range: 5 to 121), with 22 progressions was used to corroborate our findings. As significant drivers could be specifically enriched among indolent disease (i.e., protective alleles), we ran de novo driver mutation discovery across MM stages. In addition to well-characterized MM drivers ( KRAS, NRAS, etc.), 16 new candidate genes were found significantly mutated, including IKFZ3 (Aiolos), a transcription factor and direct target of degradation with lenalidomide therapy, harboring frameshift and stop-gain mutations in the protein dimerization domain which could affect complete differentiation of plasma cells. Mutants in KRAS found primarily at codons Q61, G12, G13, and A146, were found in all disease risk groups and associated with progression to overt myeloma (Odds Ratio: 6.1, CI95%: [1.4, 27], p=0.01, q<0.1). Next, we investigated genes targeted by copy-number abnormalities and found candidates in four regions with focal gains, including on chr(1q21), chr(8q24)( MYC), chr(3q26), and chr(16p13) encompassing BCMA. These genes have higher expression in patients with broad gain (e.g. BCMA: P=8.8E-3) or focal amplification (e.g. BCMA: P=6.2E-5), and therefore are likely to be underlying targets of the rearrangements driving cancer growth. Of note, BCMA antigen point mutations remained undetected in SMM, therefore patients with gain(16p) represent an attractive population for targeted therapy with anti-BCMA agents, in particular when the gain is clonal (n=3/8, 38%). In 15 patients, serial sampling of tumor cells allowed longitudinal WGS analyses and reconstruction of phylogenetic trees of disease clones (2 to 3 timepoints, 4 progressors, median 14 months). Maintenance of the major clone with subclonal structure evolving over time was confirmed in all patients, and we observed a steady increase in the cancer cell fraction of known MM mutations (del(13q), KRAS, DIS3 loss of function) in untreated patients. In one patient receiving early interventional treatment, a 70% reduction of M-spike concentration was achieved after four cycles of therapy; however, partial response was found in tandem with the selection of a high-risk del(1p) and of a KRAS activating G12S mutation, before increase and clinical progression to MM. Conclusion These results highlight the power of genomic profiling in MM for early detection, discovery of novel drivers, monitoring of clonal selection and transformation to malignant disease. We show SMM is not a simple genomically-mature disorder, but rather a dynamic state with competing subclones, which could be leveraged for therapeutic interventions.
Read full abstract