Extensive Regional Intra-Clonal Heterogeneity in Multiple Myeloma - Implications for Diagnostics, Risk Stratification and Targeted Treatment

INTRODUCTION

In multiple myeloma (MM) samples for diagnostics, prognostication and response evaluation are most commonly obtained from the patients' posterior iliac crest due to its accessibility and safety, assuming a homogenous spread throughout the bone marrow. However, imaging studies revealed a highly imbalanced distribution of the disease in the majority of the patients, presenting with accumulations of malignant plasma cells (PC) in restricted areas in the bone marrow (BM), so called focal lesions (FL). In line with this pattern, our recently reported preliminary results of paired FL and random BM (RBM) samples strongly indicate an unequal distribution of sub-clones in the BM.

Spatial genomic heterogeneity has not been systematically analyzed in MM thus far, although its existence would have a high impact on interpretation of drug resistance studies, risk stratification and personalized treatment based on genomic markers. Here we report on an extended genomic analysis of regional heterogeneity in paired FL and RBM samples including 42 newly diagnosed and 11 extensively treated MM patients with 10 of these patients also being studied longitudinally.

MATERIAL & METHODS

MM PCs were CD138-enriched. Leukapheresis products were used as controls. For whole exome sequencing (WES) we applied the qXT kit and the SureSelect Clinical Research Exome bait design (Agilent). Paired-End sequencing was performed on an Illumina HiSeq 2500. Sequencing data were aligned to the GRCh37/hg19 reference using BWA. Somatic single nucleotide variants (SNV) were identified using MuTect. Copy number aberrations (CNA) were derived from Illumina HumanOmni 2.5 bead chip data using ASCAT. Subclonal reconstruction was performed using SciClone. Gene expression profiles (GEP) were generated using Affymetrix U133plus2 microarrays. Statistical analyses were carried out using the R software package 3.1.1.

RESULTS

In 42 newly diagnosed patients we detected a median number of 86 (34 to 807) mutations per patient with up to 42% (median 5%) of them being unique to a specific site (non-ubiquitous). Among known MM driver genes, BRAF (n=2) and KRAS (n=4) were the genes that most often showed non-ubiquitous mutations at baseline. In treated patients mutations in KRAS, NRAS and RB1 contributed to regional heterogeneity in one patient each. Furthermore, we found temporal heterogeneity in mutations affecting ATRin two patients, aberrations recently associated with poor outcome.

Analyzing chromosomal aberrations with known prognostic value we observed three newly diagnosed patients with a site-specific del(1p) affecting CDKN2C and/or FAM46Cwith two of these patients also showing regional heterogeneity in del(17p13). Non-ubiquitous gain(1q) or amp(1q) was seen in two patients at baseline. Of note, in all of these cases the unique event was detected in a FL and one case with a unique gain(1q) at baseline presented with this aberration in subsequent samples. These observations strongly support the concept of FLs being sites of resistant clones able to cause relapse.

In four patients a MYC translocation was seen at only one site. In the longitudinal analysis we found one patient in whom a MYC translocation clone was replaced by a clone with a different MYC translocation, indicating that events at the MYC locus are secondary and can be sub-clonal. In contrast, primary IgH translocations were always shared, confirming that they are initiating events.

Paired samples from RBM and FLs derived from three newly diagnosed patients showed discordant GEP risk profiles, further supporting the existence of site-specific high risk (HR) clones. To investigate the clinical relevance of this finding we analyzed outcome data of 263 newly diagnosed patients with paired GEP data. The 34 cases with discordant GEP based risk scores showed no significant differences in outcome compared to cases with HR at both sites, suggesting that HR sub-clones drive prognosis even if they are not ubiquitously present.

CONCLUSIONS

We show that spatial genomic heterogeneity is common in MM. The existence of site-specific sub-clones highlights the importance of heterogeneity analyses for accurate risk prediction, detection of suitable targets for precision medicine and identification of aberrations contributing to treatment resistance. As a result we strongly recommend to include FL examinations into routine diagnostics and follow-up analyses in MM.