Spatial-Temporal Genomic Analyses Reveal a Component of "Big Bang" Kinetics in Multiple Myeloma Evolution

INTRODUCTION

Data available to date has been interpreted to show that Multiple myeloma (MM) is the end result of a multistep transformation process from normal to malignant plasma cells (PC) with molecular aberrations being attained through branching pathways. According to this Darwinian model, sequential mutations lead to the outgrowth of fitter clones, which outcompete previously dominant clones. This model does not fully explain the enormous intra-clonal heterogeneity at presentation and the extreme alternating clonal dominance during treatment. Also this model has not be adapted to take Focal Lesions (FL) into account, which are hallmarks of MM seen in imaging studies.

An alternative model has been discussed that stresses the importance of early events. In this "Big Bang" model most of the heterogeneity appears at early stages of a malignancy when clonal growth is not environmentally constrained. Transferring this model to the early evolution of MM, ancestor clones containing initiating mutations occupy the available BM niches. Subsequently, sub-clones acquire additional mutations and expand locally with some eventually forming FLs.

In order to understand the processes underlying the evolution of MM and the pathologic basis of FLs we analyzed a set of paired FL and random iliac crest samples including 42 newly diagnosed and 11 treated MM patients with 10 of these patients also being studied longitudinally.

METHODS

Paired-end whole exome sequencing of CD138-enriched MM PCs was performed on an Illumina HiSeq 2500. Somatic single nucleotide variants were called from BWA aligned sequencing reads using MuTect. Copy number aberrations were derived from Illumina HumanOmni 2.5 bead chip data using ASCAT. Subclonal reconstruction was performed using SciClone.

RESULTS

In newly diagnosed patients we observed three main sub-clonal patterns. In the first pattern, detected in 18 patients, the clonal composition was very similar between different sites. The second pattern, seen in 5 patients, was associated with site-specific dominant clones. In one of these patients we found four site-specific dominant clones in four different samples with each of them containing private mutations in known cancer driver genes, e.g. BRAF and KRAS. This type of regional clonal dominance strongly supports "Big Bang" dynamics rather than late metastasis in MM.

The third pattern, found in 19 patients, was characterized by regionally dominant clones that were also detectable at other sites at low frequencies. In one patient a clone that dominated in two distant FLs infiltrated the third FL and was borderline detectable on the mutational level at the iliac crest. This clone contained a bi-allelic deletion of the tumor suppressor CDKN2C, a gain(1q21) and a del(17p13), a combination of three events that potentially provided a massive fitness advantage, resulting in a selective sweep through the BM.

In "Big Bang" type behavior treatment which eradicates clones obviates the advantage of previously dominant clones of being first, restarting clonal competition. As a consequence we would expect to see reduced spatial heterogeneity in treated patients and longitudinally collected samples. Surprisingly, only one patient lacked spatial and temporal heterogeneity. In the majority of patients, however, we observed the opposite with a dramatic spatial heterogeneity which was further enhanced by the site-specific appearance of additional mutations. It is still not clear whether additional site-specific mutations in FLs represent randomly acquired events during treatment contributing to branching evolution or whether treatment led to a regional clonal collapse allowing descendants to catch up with the most-recent common ancestor in FLs.

CONCLUSIONS

Frequent pronounced spatial heterogeneity supports a component of "Big Bang" kinetics in the evolution of MM. This is an unexpected finding as MM is a BM cancer where easy movement of the MM propagating cells through the BM would be expected. It does, however, explain the massive heterogeneity found at diagnosis. It also highlights the important of FLs in the MM ecosystem as hotspot of evolution which could be the source of units of selection that underlie selective sweeps. It further suggests that aberrations found to be enriched in relapse patients already existed in regionally separated areas at baseline, supporting the recently introduced concept of some tumors being "born bad".