Gene expression profiling of multiple myeloma patients included in the HOVON65/GMMG-HD4 trial

In order to identify relevant, molecularly defined subgroups in Multiple Myeloma (MM), gene expression profiling (GEP) was performed on purified CD138+ plasma cells of 320 newly diagnosed myeloma patients included in the Dutch-Belgian/German HOVON-65/ GMMG-HD4 trial using Affymetrix GeneChip U133 plus 2.0 arrays. Hierarchical clustering identified 10 distinct subgroups. Using this dataset as training data, a prognostic signature was built. The dataset consists of 282 CEL files previously used in the hierarchical clustering study of Broyl et al (Blood, 116(14):2543-53, 2010) outlined above. To this set 8 CEL-files/gene expression profiles were added. Using this set of 290 CEL-files, a prognostic signature of 92 genes (EMC-92-genesignature) was generated by supervised principal components analysis combined with simulated annealing (Kuiper et al.). Bone marrow plasma cell samples were obtained from 320 newly diagnosed multiple myeloma patients included in a large multicenter, prospective, randomized phase III trial (HOVON65/GMMG-HD4). Purified myeloma plasma cells samples with a monoclonal plasma cell purity > 80% were used for analysis. To the original set of 320 samples which were used for the hierarchical clustering analysis, 8 CEL-files were added. Out of 328 CEL-files, 290 CEL-files were linked to survival data and were used to generate the 92 gene survival signature. GCRMA was performed on 290 CEL-files to give the values given for the additional 8 CEL-files. The values of the original 320 CEL-files are left unchanged (i.e. GCRMA on the original 320 CEL-files) to allow optimal comparison to the original paper by Broyl et al.