Gene Expression Profiles for Molecular Classification of Multiple Myeloma in Newly Diagnosed Patients Included in a Prospective, Multicenter, Phase 3 Trial.

Multiple Myeloma (MM) is characterized by the presence of a monoclonal protein, immunodeficiency, anemia, renal failure and bone lesions. New agents like Bortezomib (Bor) and Thalidomide (Thal) have shown efficacy in 40% of patients with relapsed/refractory MM, and up to 75% in first-line treatment. Classical prognostic classifications such as serum b2-microglobulin, albumin and chromosomal aberrations, have insufficient predictive power in estimating long-term outcome with these targeted agents. We have performed gene expression profiling (GEP) in newly diagnosed MM patients who were included in a large multicenter, prospective phase III trial (HOVON65) comparing Bor with standard induction prior to high dose therapy (HDT). The aim of this profiling was to gain new insights in the pathogenesis of MM and to design a prognostic index for treatment based on molecular profiling. In 258 patients, CD138 magnetic cell selected (MACS) myeloma plasma cells (PC) obtained at diagnosis with PC purity > 80% were used for RNA extraction and applied to the Affymetrix GeneChip U133 plus 2.0 arrays. Data from the gene expression arrays were pre-processed using GCRMA (Bioconductor). Unsupervised gene clustering (Omniviz) using a correlation visualisation matrix could determine the clusters based on the top variably expressed genes, and differential gene expression within these clusters could be determined (BRB-array tools, p<0.001, FDR<1%). Enriched genes and pathways were determined using Ingenuity systems®. Based on GEP, the translocations t(4;14), t(14;16), t(14;20) or t(11;14) could be shown to drive the clusters using 1.25% of the most variable genes (675 genes). Increasing the variable genes within the analysis (2.5%, 5%, 10% or 20%) identifies 10 different clusters of myeloma, with the translocations being distributed differentially. This could indicate that other factors, besides the chromosomal aberrations have an important role in biologically driving the clustering of myeloma GEP. Using 10% of most variable genes, 4 out of 10 clusters were predominantly driven by genes involved in the translocations; for example: up-regulation of MMSET/FGFR3 (cluster 1), MAF downstream targets (cluster 9), CCND1/MS4A1/VPREB3 (cluster 5) and INHBE (cluster 10). Three of the larger clusters show an expression of genes involved in protein biosynthesis and/or interferon signalling; one of which showed expression of XRCC4, FRZB and HGF (cluster 4), one showing expression of IFIT27, IFIT3 and TRAIL (cluster 6) and the third showed expression of EGFR and CHES1 (cluster 8). Another cluster was identified with a high NF-κB index (cluster 3), in which NF-κB transcriptional activity is calculated from the mean expression level of four probe sets corresponding to CD74, IL2RG, and TNFAIP3 (2×) as described by Keats et al. 2 out of 10 clusters, each containing 6 samples, were shown to be non-descriptive (clusters 2 and 7). When combining GEP with clinical symptoms at presentation, 2 clusters (4 & 6) were identified which had a high frequency of trisomies, IgG isotype and a significantly higher incidence of bone lesions. In addition, clusters 1, 8, 9 and 10 presented more frequently with renal failure, thrombocytopenia and anemia, and clusters 5, 9 and 10 showed less bone involvement. These data indicate that GEP may identify clusters of patients, who have a distinct clinical expression of organ impairment (ROTI). In conclusion, we describe molecular subgroups of MM showing distinct gene expression signatures in combination with specific chromosomal aberrations and clinical characteristics. Awaiting the clinical analysis of the trial, the response and survival data of the molecular subgroups will be included in the analysis.