Next Generation Sequencing Identifies Smoldering Multiple Myeloma Patients with a High Risk of Disease Progression

Introduction: Multiple myeloma (MM) is an incurable plasma cell malignancy that arises from the precursor states: Monoclonal Gammopathy of Undetermined Significance (MGUS) and Smoldering Multiple Myeloma (SMM). Some patients rapidly progress from MGUS/SMM to MM, while others remain indolent with minimal progression over their lifetime. The genetic and molecular factors that underlie disease evolution and distinguish progressors from non-progressors are not well elucidated. To address these questions, we examined clinically-annotated samples from SMM patients, including sequential samples at the SMM stage and at time of progression.

Methods: We utilized next generation sequencing methods to study 186 SMM patient samples. We performed whole exome sequencing (WES) of 70 tumor-germline matched samples (mean target coverage 50X/100X for germline/tumor) and targeted deep sequencing on 116 samples of progressor and non-progressor SMM (mean target coverage ~500X). We also performed ultra-low pass whole-genome sequencing (ULP-WGS) of 116 SMM samples as well as 20 available cell-free DNA (cfDNA) samples to detect the tumor fraction before proceeding with targeted sequencing. For WES, libraries were constructed with Agilent SureSelect XT2 library prep kit, and hybridized to Agilent's whole exome V5+UTR capture probes and then sequenced on HiSeq 4000 (Illumina). For targeted sequencing, libraries were enriched by hybridizing to the customized target bait (60 MM related genes and 56 pan-cancer genes). For ULP-WGS, libraries were prepared with Kapa Hyper Prep kit and custom adapters. Sequencing data were analyzed using previously established analytic pipelines including ichorCNA, MuTect, RecapSeg, MutSig, GISTIC, ABSOLUTE, and PHYLOGIC. High-risk SMM (HRSMM) and low-risk SMM (LRSMM) were defined according to the criteria proposed by Rajkumar et al. in Blood 2014.

Results: Higher mutation load (average 1.44 mutations/Mb) was observed in HRSMM patients, which is close to the average of 1.6 mutations/Mb in symptomatic MM, compared to LRSMM patients (average 0.73 mutations/Mb, p -value = 0.001). Somatic mutations in known signaling pathways were observed four times more frequently (43.8% vs. 9.5%, p- value = 0.015) in HRSMM patients compared to LRSMM patients. Mutations in MAPK pathway genes (KRAS, NRAS, BRAF) were detected in 21.7% of HRSMM vs. 0% in LRSMM (p- value = 0.004). Mutations in NFkB pathway were present in 10% of patients and were also associated with HRSMM (17.4% vs. 0%, p- value = 0.018). MM somatic copy number aberrations (SCNAs) were detected in 64% of all SMM samples, and were two times more frequently in HRSMM than LRSMM patients (78.1% vs. 42.9%, p- value = 0.020). Phylogenetic studies of sequential patient samples, who progressed from SMM to MM, revealed different patterns of clonal evolution. Furthermore, we have performed ULP-WGS of available cfDNA samples from 20 patients of this cohort. Interestingly, tumor fraction of cfDNA was higher in HRSMM compared to LRSMM patients (p- value = 0.03). Together, this data suggests that specific genomic alterations are associated with HRSMM and could potentially be predictive of the risk to progression in SMM patients.

Conclusion: This study demonstrates that next generation sequencing (WES, targeted-sequencing, and ULP-WGS) of SMM patients can identify genomic alterations associated with HRSMM vs LRSMM through different parameters such as mutation load, somatic mutations, and SCNAs.