INTEGRATIVE GENOMIC APPROACH TO THE MOLECULAR BIOLOGY OF MULTIPLE MYELOMA

marrowplasma cells that accounts for 10% of all hematological malignancies.The broad clinical spectrum of plasma cell dyscrasias range from apre-malignant condition termed monoclonal gammopathy of undeterminedsignificance (MGUS) to smouldering MM (SMM), truly overt andsymptomatic MM, and extra-medullary myeloma/plasma cell leukemia(PCL).1,2 MM is characterised by a profound genomic instability thatinvolves both ploidy and structural rearrangements.3 Nearly half of MMtumours are hyperdiploid (H-MM) characterised by recurrent trisomies,particularly of chromosomes 3, 5, 7, 9, 11, 15, 19 and 21; the non-hyperdiploid(NH-MM) tumours are cases with a hypodiploid, pseudodiploidor near tetraploid chromosome number, and are frequently associatedwith chromosome 13 deletions and immunoglobulin heavy chain (IGH)locus translocations involving a promiscuous number of partner loci,mainly CCND1 (11q13), FGFR3/MMSET(4p16.3), MAF (16q24), MAFB(20q) or CCND3 (6q).4-7 It has also been demonstrated that almost allMM patients are affected by deregulation of one of the cyclin D genes(CCND1, 2 or 3), which may therefore play an important role in themolecular pathogenesis of the disease.MM patients can be stratified into five molecular groups on the basisof the presence of known IGH translocations and cyclin D deregulation(TC classification): TC1 is characterised by t(11;14) or t(6;14); TC2 isassociated with hyperdiploidy and low-moderate levels of CCND1 inthe absence of IGH translocations; TC3 includes tumours not fallinginto any of the other groups, most of which express CCND2; TC4 isassociated with t(4;14) and high CCND2 levels; and TC5 expresses thehighest levels of CCND2 in association with either t(14;16) or t(14;20).8,9We have combined fluorescence in situ hybridization (FISH) analysesand global gene expression profiling (GEP) in a series of studies aimedat elucidating the transcriptional profiles associated with plasma celldyscrasias in a panel of newly diagnosed patients including 11 withMGUS, 132 with MM and nine with PCL. The unsupervised analysis ofthe gene expression data profiled on high-density oligonucleotidemicroarrays identified two major groups: one including the majority ofMGUS patients and normal controls, and the other all the PCL and mostof the MM cases. Therefore, neither the MGUS, nor the PCL and MMsamples could be identified as distinct entities. A multi-class analysisrevealed probe sets specifically distinguishing MGUS from PCL, with theMM cases showing their progressively modulated expression. TheMGUS cases showed the up-regulation of immune response genes,whereas the PCL cases showed the positive modulation of primarymetabolism, and cell cycle and apoptosis induction. The hierarchicalclustering generated in the 132 MM database was mainly driven bygroups reflecting the TC classification.We also analysed the GEP data in the context of distinct genetic lesionsinvolving chromosomal gains or losses. All of the del(13) cases showed67 down-regulated genes involved in protein biosynthesis, ubiquitinationor transcriptional regulation, most of which (44/67) mapped alongthe whole chromosome 13.10 In terms of 1q gain, the differential expressionof 61 genes mainly localised on chromosome 1q12-1q44 distinguishedMM patients with or without 1q extra copies. Functional analysisof the identified genes revealed their involvement in energy productionpathways, intracellular protein transport, and stress-induced endoplasmicreticulum responses.11Finally, the differential expression of 225 genes mainly involved inprotein biosynthesis, transcriptional machinery and oxidative phosphorylationdistinguished H-MM from NH-MM. Most of the up-regulatedgenes in H-MM mapped to the chromosomes involved in hyperdiploidy,whereas a significant fraction of the genes in NH-MM mapped to 16q.12Overall, the GEP data suggested a widespread gene-dose effect as theimbalances in expression closely correlated with the genomic structuralabnormalities.These results prompted us to use novel high-throughput approaches,such as high-density single nucleotide polymorphism (SNP) arrays, in anattempt to define the allelic imbalances that may contribute to thegenomic instability and bio-clinical heterogeneity of MM. Furthermore,the integration of data derived from genome-wide DNA microarrayanalysis with transcriptional profiles may identify differentiallyexpressed genes related to underlying chromosomal alterations, as beingcandidate tumour genes. We therefore used integrated FISH, GEP andwhole-genome DNA SNP analyses to study a panel of 23 human myelomacell lines (HMCLs) and identified some novel genetic imbalances.13Subsequently, we used the same approach in a study of 45 MM primarytumours included in the GEP dataset and, by means of a self-developedcomputational model based on combined FISH and genome-wide profilinganalyses, found that marked aneuploidy characterised a significantfraction of the patients. In particular, an unsupervised analysis of the 45genome profiles showed the presence of at least five main clusters ofpatients with different characteristics: an altered number of odd chromosomessuggesting hyperdiploidy; 1q gain and chromosome 13 deletion;deletions involving chromosomes 1p, 8p, 13, 22 and 14; aneuploidy(mostly near tetraploidy); and limited alterations. Non-parametric analysesusing both the genomic and gene expression data identified a largenumber of genes whose expression closely correlated with copy numbervariations, thus further suggesting a marked gene-dose effect associatedwith allelic imbalances.These findings provide a focus for further studies aimed at identifyingand characterising genes that are involved in the pathogenesis ofmyeloma. To this end, we have started studies aimed at integratingFISH, GEP and SNP data with those derived from global microRNA(miRNA) gene expression analysis on microarrays. miRNA are smallnon-coding sequences that are thought to play important roles in regulatingthe genes involved in controlling cell cycle, survival and differentiationprogrammes, and are frequently located in hot spots for chromosomalabnormalities. Altered patterns of miRNA expression have alreadybeen demonstrated in a number of solid and hematological tumours.14,15The integration of multiple high-throughput approaches should increasethe reliability and significance of our investigations, and provide synergisticinformation allowing the discovery of new pathogenetic networksand therapeutic treatments for MM.