Mutational profiling of cancer candidate genes in glioblastoma, melanoma and pancreatic carcinoma reveals a snapshot of their genomic landscapes

Abstract

A recent systematic analysis of 18.191 well annotated coding sequences (RefSeq) in breast and colorectal cancers has led to the identification of somatic mutations in 1.718 genes (Wood et al., 2007). Based on statistical parameters 280 of these have been denominated candidate cancer (CAN) genes. This analysis has provided an interesting snapshot of the landscape of tumor genomes by showing that they contain a few frequently mutated genes (denominated 'mountains'). On the contrary, the large majority of CAN genes are altered at low frequency (designated 'hills'). Whether 'hill' type CAN genes are tumor specific is largely unknown. To address this question we evaluated the mutational profiles of 27 'hill' CAN genes in glioblastoma, melanoma and pancreatic carcinoma by sequencing the exons previously found mutated by Wood and colleagues. Only 4 of the breast/colorectal 'hill' type CAN genes (SMAD4, MYO18B, NAV3 and MMP2) were also mutated in melanoma and pancreatic carcinoma, while none was altered in glioblastoma. These results suggest that 'hill' type CAN genes are not frequently shared by different tumor types and that their mutation patterns are tissue specific. Tumor-specific genome wide mutational profiling will be required to identify 'hill' type CAN genes that characterize the genomic landscapes of each cancer lineage.