Identification of specific transcriptional patterns associated with hyperdiploidy in multiple myeloma.

Karyotypic instability is strongly associated with multiple myeloma (MM). According to the chromosome number pattern, two major groups are recognized: hyperdiploid (H) tumors, associated with recurrent trisomies involving non-random chromosomes (3, 5, 7, 9, 11, 15, 19 and 21); and non hyperdiploid (NH) tumors associated with hypodiploid, pseudodiploid or near-tetraploid karyotypes. MM patients are approximately equally distributed between the two categories; notably, the most recurrent IGH translocations and chromosome 13 deletion appear to be prevalently associated with NH-MM, whereas recent evidences have suggested that H-MM correlates with a favorable prognosis. To molecularly characterize these two genetic categories, we performed a gene expression profiling analysis on 66 newly-diagnosed MM, characterized by FISH analyses for IGH translocations, 13q14 deletions and additional copies of chromosomes 1, 11 and 19. The ploidy status was investigated by combining two recently proposed FISH approaches (Wuilleme S. et al., 2004; Chng W.J. et al., 2005). The gene expression profiles of highly purified MM plasma cells have been generated by means of high-density oligonucleotide arrays (Affymetrix GeneChip U133A) and subsequently analyzed using unsupervised and supervised approaches (two-dimensional hierarchical clustering and SAM, respectively). The differential expression of 229 genes distinguished the 28 H-MM from the 38 NH-MM cases. The 208 upregulated genes in H-MM mapped mainly on the chromosomes involved in hyperdiploidy, while a significant percentage (29%) of the 21 genes upregulated in NH-MM were localized on 16q. The identified transcripts have been further validated on a publicly available gene expression dataset of an independent cohort of 64 MM patients (Carrasco et al., 2005). Notably, the global classification rate for the 64 cases resulted of 81%, confirming the validity of the identified transcriptional fingerprint. A functional analysis revealed a significant fraction of genes involved in protein biosynthesis (38%), transcriptional machinery and oxidative phosphorylation. Furthermore, an integrative genomic approach using a model-free statistical method (LAP, locally adaptive statistical procedure) supported these findings, allowing the identification in H-MM of globally upregulated regions on the chromosomes 3, 5, 9, 15 and 19, along with the downregulation of a region on 16q arm. Remarkably, two sub-groups are clearly distinguishable within H-MM group, one associated with chromosome 11 gain and the other showing 1q gain and chromosome 13 deletion. A supervised analysis of the H-11 vs H-13/1 patients identified 57 differentially expressed genes. Eleven of the 18 genes up-regulated in the H-11 group mapped to chromosome 11, whereas 21 of the 39 genes up-regulated in the H-13/1 group mapped to the 1q region. Notably, CCND2 resulted the most significantly upregulated gene in H-13/1 group. Our data reinforce the importance of combining cytogenetics and gene expression approaches for a better definition of the genetic alterations in MM and provide a molecular and genomic framework for dissection of disease pathogenesis and clinical management.