Circulating Tumor DNA: Less Invasive, More Representative Method to Unveil the Genomic Landscape of Newly Diagnosed Multiple Myeloma Than Bone Marrow Aspirates

Abstract

Multiple myeloma (MM) is highly heterogenous and dynamic in its genomic abnormalities. Capturing a representative image of these alterations is essential in understanding the molecular pathogenesis and progression of the disease but was limited by single-site invasive bone marrow (BM) biopsy-based genomics studies. We compared the mutational landscapes of circulating tumor DNA (ctDNA) and BM in 82 patients with newly diagnosed MM. A 413-gene panel was used in the sequencing. Our results showed that more than 70% of MM patients showed one or more genes with somatic mutations and at least half of the mutated genes were shared between ctDNA and BM samples. Compared to the BM samples, ctDNA exhibited more types of driver mutations in the shared driver genes, higher numbers of uniquely mutated genes and subclonal clusters, more translocation-associated mutations, and higher frequencies of mutated genes enriched in the transcriptional regulation pathway. Multivariate Cox analysis showed that age, ctDNA mutations in the transcriptional regulation pathway and DNA repair pathway were independent predictors of progression-free survival (PFS). Our results demonstrated sequencing of ctDNA provides more thorough information on the genomic instability and is a potential representative biomarker for risk stratification and in newly diagnosed MM than bone marrow.

Keywords: circulating tumor DNA; multiple myeloma; prognosis.

Figure 1

Comparison of the genomic landscape of ctDNA and Bone marrow (BM) DNA: (A) Somatic mutation profiles of newly diagnosed multiple myeloma (NDMM) patients from pre-treatment ctDNA sequencing of 413 cancer genes. Eighty-two patients were arranged along the x-axis. Age, sex, R-ISS stage, genes with somatic mutations were shown; (B) Somatic mutation profiles of NDMM patients from pre-treatment BM sequencing of 413 cancer genes; (C) Comparison of somatic mutation detected in ctDNA and paired BM DNA; (D) The relationship of variant allele frequencies (VAFs) between ctDNA and paired BM DNA; (E) Number of mutated genes in ctDNA and paired BM DNA.

Figure 2

Representative genomic architecture derived from paired ctDNA versus BM DNA. Cancer Cell Fraction (CCF) of mutations was calculated in ctDNA and BM DNA. Each dot represents one mutation, and the color of each dot indicates the subclone that the given mutation was clustered to.

Figure 3

Genomic landscape under different IgH translocations. Somatic mutation profiles of NDMM patients grouping by IgH translocations. Different colors represent different IgH translocations and mutation types. (A) ctDNA; (B) BM.

Figure 4

Correlation heatmap for hot somatic mutations and cytogenetic abnormalities. Correlation between mutations and recurrent cytogenetic abnormalities. Intensity of color shade represents the degree of correlation (blue, negative; red, positive) as per the scale. *, **, *** represent p value less than 0.05, 0.01, 0.001, respectively. (A) ctDNA; (B) BM.

Figure 5

Clinical correlation of the molecular tumor burden index (mTBI) AND clinical parameters in 82 multiple myeloma patients. The patients were divided into mTBI-hi and mTBI-lo groups based on the median value. Compared to the MM patients in mTBI-lo group, those in mTBI-hi group had higher concentrations of serum lactate dehydrogenase (LDH) and higher percentages of bone plasma cells (BM-PC) and peripheral blood circulating plasma cells (PB-PC).

Figure 6

Multidimensional analysis of prognostic factors for PFS: (A–F) Kaplan–Meier curves for PFS according to the different factors; (G) Independent risk factors for PFS with different Hazard Ratios; (H) Nomogram models to predict PFS; (I) Kaplan–Meier curves for PFS according to risk factors.

Figure 7

Clonal evolution before and after therapy according to different therapeutic responses. PyClone was employed to analyze the clonal structure using a Bayesian clustering method. For serial ctDNA, multiple inputs of each sample were used to analyze the serial clonal population. Each mutation’s CCF (cancer cell fraction) was calculated in ctDNA samples before and after therapy. The cluster with the highest CCF was identified as the clonal cluster, and mutations in this cluster were clonal mutations. Meanwhile, other clusters and mutations were considered subclonal. Each dot indicated one cluster, and a load of clusters was calculated with the mean VAF of each mutation that was clustered.

Figure 8

Involved pathways enriched in the clonal and subclonal mutations in ctDNA before and after treatment for (A) CR + VGPR; (B) PR + PD. PyClone was used to analyze the clonal population structures in baseline ctDNA and post-treatment ctDNA samples. The copy number information of each SNV was used as input. For each clustering process, a PyClone algorithm was run for 20,000 iterations with a burn-in of 2000, using a beta-binomial model with the “total_copy_number” option (Murtaza M, Dawson SJ, Pogrebniak K, Rueda OM, Provenzano E, Grant J et al. Multifocal clonal evolution characterized using circulating tumor DNA in a case of metastatic breast cancer. Nature communications. 2015; 6: 8760.) [25]. To clarify the functional role of clonal and subclonal genes, we performed pathway enrichment analysis in the mutations detected in before and after treatment samples.