Unsupervised analyses reveal molecular subtypes associated to prognosis and response to therapy in colorectal cancer

Exome Sequencing Reveals AMER1 as a Frequently Mutated Gene in Colorectal Cancer

PURPOSE

Somatic mutations occur at early stages of adenoma and accumulate throughout colorectal cancer progression. The aim of this study was to characterize the mutational landscape of stage II tumors and to search for novel recurrent mutations likely implicated in colorectal cancer tumorigenesis.

EXPERIMENTAL DESIGN

The exomic DNA of 42 stage II, microsatellite-stable colon tumors and their paired mucosae were sequenced. Other molecular data available in the discovery dataset [gene expression, methylation, and copy number variations (CNV)] were used to further characterize these tumors. Additional datasets comprising 553 colorectal cancer samples were used to validate the discovered mutations.

RESULTS

As a result, 4,886 somatic single-nucleotide variants (SNV) were found. Almost all SNVs were private changes, with few mutations shared by more than one tumor, thus revealing tumor-specific mutational landscapes. Nevertheless, these diverse mutations converged into common cellular pathways, such as cell cycle or apoptosis. Among this mutational heterogeneity, variants resulting in early stop codons in the AMER1 (also known as FAM123B or WTX) gene emerged as recurrent mutations in colorectal cancer. Losses of AMER1 by other mechanisms apart from mutations such as methylation and copy number aberrations were also found. Tumors lacking this tumor suppressor gene exhibited a mesenchymal phenotype characterized by inhibition of the canonical Wnt pathway.

CONCLUSIONS

In silico and experimental validation in independent datasets confirmed the existence of functional mutations in AMER1 in approximately 10% of analyzed colorectal cancer tumors. Moreover, these tumors exhibited a characteristic phenotype.

Altered pathways and colorectal cancer prognosis

The identification of molecular markers with prognostic value in colorectal cancer is a challenging task that is needed to define therapeutic guidelines. Clinical factors are insufficient to identify those patients with stage II at risk of relapse or those patients with stage III at low risk. There is a current effort to define a consensus in molecular subtypes based on expression profiles, which are characterized by a distinctive prognostic outcome. Also several gene expression signatures based on individual genes have been proposed to predict prognosis, but they show low consistency and reproducibility. Slattery et al. describe a pathway-based approach to analyze gene expression differences between normal and colon cancer tissues. The most interesting finding is that having more deregulated pathways is associated with good prognosis. If these findings are properly validated, new insights into the mechanisms of colon carcinogenesis may be revealed. Please see related article: http://dx.doi.org/10.1186/s12916-015-0292-9 .