High-Dose Melphalan Significantly Increases Mutational Burden in Multiple Myeloma Cells at Relapse: Results from a Randomized Study in Multiple Myeloma

We recently shown that high-dose melphalan (HDM) followed by autologous stem cell transplant (ASCT) as first line therapy in young (<66 yrs) multiple myeloma (MM) patients significantly improves progression-free survival (IFM/DFCI 2009 study). However, the impact of alkylating agent melphalan inducing N-alkylpurine-monoadducts forming interstrand crosslinks (ICLs) in surviving myeloma cells remains an important biological question. We here profiled samples from the IFM/DFCI 2009 study, where patients were randomized to RVD+HDM vs RVD alone, to identify genomic changes induced by HDM and observed at relapse. We analyzed paired purified MM cells collected at diagnosis and at relapse from 68 patients using deep (75X) whole genome sequencing. Forty-five patients were treated with RVD only, while 23 patients received RVD followed by HDM. There was no significant difference between the 2 groups in regard to disease characteristics including sex, age, cytogenetic risk, and best response. Median follow-up was similar (29 vs 31 months, respectively), removing longer follow up as a confounding variable. The number of mutations at diagnosis was similar on both arms (7137 [IQR=3742] vs. 7230 [IQR=3702], p value = 0.67). Although mutational load increased in both arms; there was a significantly higher increase in number of mutations and indels in the HDM arm compared to RVD alone (mutations 5686 vs 1745, p=1.4e-5; and indels 467 vs 360, p= 0.02, respectively ). Using a model incorporating number of new mutations, depth, and purity, we found that HDM causes a 4.1 fold higher mutation accumulation rate per month than RVD only (158.3 vs 38.3 mutations/ month; p=0.003). Importantly, newly acquired mutations were localized to regions which overlap with transcribed regions, and accumulated at significantly higher rate in the HDM group (p=0.009). In contrast, we did not observe any significant changes in copy number alterations (CNAs) and structural variants, including translocations, between both arms.

A significant change in frequency of driver mutations including RAS/RAF, FAM46C, TP53, and DIS3 was not observed at the time of relapse. Clonality level was increased only for KRAS (p=0.054), while all other specific driver genes had similar clonality level at diagnosis and relapse. Interestingly, a significant increase in mutations involving MYO16 and SLC7A8 genes was observed at relapse in both arms, implicating components of the induction regimen (RVD). Investigating the mutational signature utilization in only newly acquired mutations identified 4 signatures: APOBEC, HR Double Strand Repair, clock-like signature, and unknown. k-means clustering analysis of samples based on signature utilization showed four distinct clusters. All patients clustering with high DNA repair signature utilization were in the HDM arm (65% HDM patients), the majority of whom achieved CR or sCR (74%); these patients acquired 8308 (range 3302-19107) new mutations between diagnosis and relapse. None of the RVD only treated patients were in this cluster. The remaining 35% HDM group patients were clustered with RVD samples and showed unknown signature utilization. Furthermore, motif enrichment analysis identified CYWR and ATGAGATV (p < 1e-130) as enriched motifs around the new mutations in HDM compared to RVD cohort. Importantly and as expected, DNA damage repair pathway genes were frequently targeted in the HDM group: 72% HDM samples accumulated DDR gene mutations vs. only 17% in the RVD alone arm (p < 0.001). At the time of relapse, 100% HDM arm patients had at least one DDR gene mutation and 80% had two or more, while only 37% RVD only group had one or more such mutation. Finally, we have reconstructed phylogenetic and evolutionary trajectories based on mutation and copy-number data from samples at diagnosis and relapse. The clonal composition in both arms was similar at diagnosis; however, HDM caused a significant shift to more subclonal mutations at relapse. chromothripsis and chromoplexy events were detected in 30% patients at diagnosis, which remained constant at relapse regardless of treatment.

In summary, we describe significant accumulation of mutations following high dose melphalan. This fundamental molecular change in the disease at relapse, suggests the need for reappraisal of the optimal use and sequencing of high dose melphalan in the era of novel agents.