Screening of Homozygous Deletions Identifies Key Deregulated Genes and Pathways in Multiple Myeloma.

Homozygous deletions (HD) are important in cancer cell lines and have a non-random distribution in the genome. We have determined the frequency and distribution of HD, along with the genes affected, in 84 presenting multiple myeloma cases using the Affymetrix 500K SNP GeneChip mapping arrays. Initially we took a highly sensitive approach identifying regions with a copy number (CN) <1.0 in at least 4 adjacent SNPs. The analysis was done both manually, in dChip, and using an in-house algorithm. To reduce the complexity of the data generated we removed deletions that only occurred in one sample and further filtered the data using U133 Plus 2.0 expression array analysis. Genes passed expression filtering criteria by having a lower expression than the median expression of all samples. HD in plasma cells occur during IgH and IgL chain rearrangements and were used as internal controls to validate our approach. We found that regions of homozygous deletion were not randomly distributed. At the CN<1.0 level, we identified 1761 regions containing 6650 genes. After expression filtering 129 genes were identified (with an additional 116 genes which did not have expression array probesets) which fell in chromosomal regions 1p, 6q, 8p, 11q, 12p, 13q, 14q, 16q, 20q, and 22q, which are also the regions that most frequently have loss of heterozygosity (LOH). The distribution of deletions was even between different myeloma molecular groups. We then applied a more stringent copy number cut-off of 0.65 and identified 18/84 cases with HD (21%, median size 18 kb) and most cases had only one homozygous deletion. In this analysis the number of genomic regions falls to 29 spread over 10 chromosomes, including 241 genes of which only 20 genes survived expression filtering. Of these regions 3 are in known fragile sites and only 1p, 11q, 13q, and 22q showed HD containing genes with a decreased expression. Although deletions were seen on chromosome 13, none involved the RB1 locus. Several microRNA were also noted to be deleted on chromosomes 6, 13, and 22 indicating that these may also be important in myeloma and are being explored further. The NF-kB pathway inhibitors BIRC2 and BIRC3, on 11q, were homozygously deleted in 4 cases as were the neighbouring loci ANGPTL5 and TMEM123. Interestingly, the 11q region was the exception to the rule that HD occur only in regions of LOH. The most frequent regions of homozygous deletion involved CDKN2C (p18) and FAF1 on 1p32.3, which were deleted in 6 cases (CN<1.0). In addition we have demonstrated that deletion of this region is an important prognostic marker (median survival 19 months vs 33 months). We and others have suggested that p18 is the important event at this locus but in this study when stringent criteria (CN<0.65) were applied more frequent deletions were found at the FAF1 locus. To define the critical lesion we carried out mutation and methylation analysis on both these genes and did not find any abnormalities at p18 and are using this approach to analyse FAF1. FAF1 is potentially relevant to myeloma as it is a regulator of NF-kB which also induces apoptosis via the FAS pathway. Loss of FAF1 could enable cells to tolerate DNA damage better than when it is intact. We have attempted to differentiate these two roles further through correlation with downstream deregulated genes using global expression arrays.