Gene Expression Profiling (GEP) in Multiple Myeloma (MM): Comparison of Purified MM Cells (PMM), Random Bone Marrow Biopsies (RBX) and Fine Needle Biopsies from Focal Lesions (FNBX).

Background: GEP has become a useful tool for the distinction of genetic subgroups of patients with malignant disease of a common pathologic diagnosis, including MM. The aim of this study was to determine whether whole bone marrow biopsies could replace the laborious magnetic bead purification of CD 138-positive plasma cells, with the potential added advantages of: 1. less sample manipulation and thus facilitating shipping to a central processing and analysis facility; 2. encompassing the bone marrow micro-environment as an important contributor to MM survival and drug resistance.

Patients and Methods: Samples included paired PMM and RBX from 190 newly diagnosed patients treated on Total Therapy protocols TT2 (n = 85) and TT3 (n = 105); 29 of these patients also had CT-guided fine needle biopsies (FNBX); 5 subjects with MGUS and 5 normal donors were used for comparison. Total RNA was applied to Affymetrix U133 Plus2 microarray chips. Unsupervised hierarchical clustering (UHC) was performed on Log2 transformed GEP intensity values of probe sets with >95% present detection calls and >0.5 standard deviation.

Results: UHC analysis of RBX (MM, MGUS, normal) identified 2 major subgroups: one containing both normal and MGUS samples (“normal-like”, N-L) and the other comprising only MM samples (“myeloma-like”, M-L) (Figure 1). In the context of previously defined 7 GEP subgroups (Zhan et al, Blood 2004), the N-L group contained higher proportions of hyperdiploid and normal karyotype cases, whereas the M-L group comprised the majority of MAF/MAFB, FGFR3/MMSET and proliferation cases. Good-risk standard laboratory features (non-IgA isotype, lower B2M and CRP levels, higher Hb concentrations) were over-represented in the N-L group. When applied to TT2 patients with a median follow-up of 3 years, N-L patients had superior 3-yr EFS (85% vs 55% for M-L group; p=.02). A further UHC analysis addressed similarities/differences among PMM, RBX and FNBX samples available in 29 patients. The dendrogram had 2 major branches; one comprising all PMM samples and the other all BX with sub-branches separating RBX and FNBX subgroups. Finally, 90% of all MAF, MAFB, FGFR3, MMSET and CCND1 “spikes” identified in PMM (with well established prognostic implications) were also detectable in RBX and FNBX samples.

Conclusion: GEP analysis of RBX distinguished 2 subgroups, M-L and N-L, in the context of comparison with normal subjects and MGUS. Compared to M-L, N-L patients (comprising more favorable genetic and standard laboratory features) had superior EFS. Both RBX and FNBX were adequate tissue sources for the detection of prognosis- and possibly therapy-relevant translocation-associated “spiked” genes. Genes differentiating PMM, RBX and FNBX will be reported at the meeting. BX as tissue source should make GEP studies of MM feasible in the context of multi-institutional trials.