An Integrated Pharmacogenomic Strategy for the Definition of Thalidomide Response Signatures in Presenting Cases of Multiple Myeloma.

Whilst gene expression signatures have been defined that correspond to poor overall survival, the mechanism for deregulation of such genes is often elusive. We and others have described acquired copy number change as one potential mechanism of gene deregulation in myeloma. Other potential mechanisms exist that may influence the expression of myeloma-associated genes such as inherited SNPs and copy number variation (CNV). We have therefore embarked upon an integrated pharmacogenomic strategy to determine the importance of acquired and inherited genetic changes in determining response to therapy. We have carried out gene expression analysis on CD 138 selected bone marrow plasma cells from 231 newly diagnosed myeloma cases using Affymetrix U133 Plus 2.0 expression arrays and copy number analysis using 500K Gene Mapping arrays on a subset of 90 cases. Peripheral blood DNA has been genotyped using Affymetrix 500K Gene Mapping arrays and the BOAC chips. Cytogenetics was available in the majority of cases. Younger, fitter patients received either cyclophosphamide, thalidomide and dexamethasone (CTD) or cyclohosphamide-VAD (C-VAD), followed by high dose melphalan (HDM). Older, less fit patients received attenuated dose CTD or MP. Response was assessed before and after HDM in the intensive group and on completion of therapy in the non-intensive group using EBMT criteria plus the category of VGPR. We used a supervised approach to define a gene expression signature corresponding to high level response (CR, VGPR or PR) against poor response (NC, PD or MR) overall and for each of the three induction strategies, CTD/CTDA, CVAD and MP. We have combined the data from expression arrays together with mapping data from tumor DNA and 2 different SNP arrays performed on germline DNA. We defined a poor response expression signature initially and then identified the genomic loci of these genes and how they were affected by acquired copy number change. For each candidate gene we also examined the constitutional DNA to see if each fell within a region of inherited CNV and how this could be affected by acquired copy number change. In a similar fashion, we used the BOAC chip to define genes and SNPs associated with response. This is different as it utilized mostly functional cSNPs in candidate genes. We then looked at how CNV affected these genes. Although not all genes in which functional cSNPs are present would necessarily be expected to be expressed in plasma cells, this approach is a vital step in identifying the clinical relevance of such cSNPs in myeloma. We also took the alternate approach and designed an algorithm able to correlate acquired copy number change with paraprotein response. We then identified differentially expressed genes in these loci and their impact on response, narrowing the candidate genes down to define a signature which could be validated. Using this approach has allowed us to identify genes important in determining response and their relation to tumor-associated copy number change and inherited CNV. Overall, this methodology provides significant insight in to the factors that predict response to different chemotherapy regimens. Preliminary data will be presented.