Genomic Discovery and Clonal Tracking in Multiple Myeloma By Cell Free DNA Sequencing

Summary: Development of drug resistance and clonal evolution is a major problem in multiple myeloma (MM) and may be due to genetic heterogeneity and clonal evolution but its longitudinal genetic characterization has been impractical, thus far. We have developed a genomic discovery method to monitor genetic evolution of the disease through whole exome sequencing and low-pass whole genome sequencing of myeloma-derived circulating cell-free DNA (cfDNA). Low-pass whole genome sequencing of 110 cfDNA samples from 75 randomly selected MM patients was used to develop an approach to predict the utility of further whole exome sequencing based on cfDNA concentration as a quick and cost-effective marker. Whole exome sequencing of cfDNA, matched normal blood cells, and bone marrow MM cells obtained from five myeloma patients at the same time point demonstrated that identification of clonal somatic mutations was robust and highly concordant between cfDNA and bone marrow, and provided sufficient resolution to define the subclonal composition by cfDNA. cfDNA allowed longitudinal tracking of disease burden and captured clonal evolution of MM. We provide a framework for effectively utilizing comprehensive discovery-oriented cfDNA sequencing in MM.

Methods and Results: In order to estimate the subclonal composition of MM from "blood biopsy",we collected 110 plasma samples from 75 randomly selected patients with multiple myeloma, who were undergoing treatment, and isolated cfDNA. As a control, we also isolated cfDNA from the plasma of 12 healthy blood donors. We prepared next-generation sequencing libraries and performed low-pass whole genome sequencing to a sequencing depth of 0.1-1 fold coverage across the entire genome for all of these samples. We deployed a computational approach that enables assessment of the fraction of myeloma-derived DNA in cfDNA by utilizing the relative ratio of somatic copy number alterations (SCNAs) from low pass whole genome sequencing data, after correcting for GC bias. Using a detection threshold cutoff of ~5%, we detected myeloma-derived cfDNA in 22% of total MM samples. In two patients from whom multiple serial samples were available, we found strong concordance between the myeloma DNA amount and serum free light chains, suggesting that low pass whole genome sequencing of cfDNA allows tracking of progression of MM longitudinally in patients.

We then asked if whole exome sequencing (WES) and robust calling of somatic mutations is possible from cfDNA in myeloma patients. We performed deep whole-exome sequencing of cfDNA from five patients with multiple myeloma in whom we detected the presence of myeloma-derived circulating DNA by low-pass whole genome sequencing of cfDNA. We also performed whole exome sequencing of CD138⁺ cells that were flow-sorted from the bone marrow of the same patients, as well as CD45^- cells as a matched normal sequencing control. cfDNA was obtained on the same day as the bone marrow aspirate, to rule out ongoing clonal evolution between the time of blood and bone marrow harvest. We first determined if canonical mutations that are typically seen in MM can be detected in cfDNA. We identified mutations in NRAS, KRAS, DIS3, TP53, and RB1, all known to be recurrently mutated in MM. These somatic mutations were identified in both the cfDNA as well as in the CD138+ sorted MM cells from the bone marrow aspirate.

We then determined that somatic mutations in the BM are well represented in the cfDNA in general, and that the allelic fractions of somatic mutations correlated well between cfDNA and matched bone marrow biopsies (Pearson's r=0.704; 5 patients with a tumor fraction >16%). We found overall that 87% of 840 mutations in the bone marrow were detected in cfDNA and 92% of clonal mutations were also identified in cfDNA. Some distinct subclones could be identified either only in the cfDNA or only in the bone marrow. In one case, extensive clonal evolution occurred during treatment and individual subclones likely accounting for the development of drug resistant disease could be distinguished enabling the reconstruction of evolutionary trees.

These results show that low-pass whole genome as well as whole-exome sequencing of cfDNA is feasible in patients with MM, can recapitulate the genetic evolution of MM in the bone marrow and reveals biologically and therapeutically relevant data when detectable.