Abstract 3900: Genomic landscape of human cancer reveals dysregulated TGF-β signaling with prognostic significance

Objective: Genome-wide analysis enables predictive modeling of genetic pathways driving many cancers. While somatic mutations and patterns reflecting key pathways have been identified for many cancers, an integrated analysis of driver mutations identified through mouse/human genetics have yet to be comprehensively defined for hepatocellular cancer (HCC). Previously our group and others have identified that loss of TGF-β signaling leads to spontaneous HCC development, through mouse models and human genetics. Patients with hepatocellular cancer have a poor survival of 9-11 months. Recent clinical studies show that targeting TGF-β improves survival up to 21 months, yet prognostic significances are undefined. The relationships between patterns of mutations and transcriptomic phenotypes for the TGF-β pathway are unclear. Methods: (1) We analyzed the transcriptome of 488 hepatocellular cancers and screened for mutations in the TGF-β pathway in 202 HCCs from The Cancer Genome Atlas (TCGA). (2) Increased levels of TGF-β-related genes were designated as an “activated” signature that is associated with hepatic fibrosis. Conversely, decreased levels of TGF-β-related genes were defined as an “inactivated” signature, which was associated with the loss of TGF-β tumor suppressor function. (3) We further performed high-fidelity (80x) whole-genome sequence analysis and transcriptome sequencing analysis of eight additional HCCs to define the role of TGF-β in their development and characterize a potential novel “driver mutations” in HCV- and alcohol-associated hepatocellular cancer. (4) We validated the clinical relevance of β2SP alterations in 22 human liver specimens. Results: (1) Transcriptomic analyses revealed aberrant TGF-β superfamily profiles in 72% of hepatocellular cancers, with mutations in 38% of patients. (2) HCCs characterized by the “inactivated” TGF-β signature were associated with a significantly poorer survival particularly in early stage HCCs, compared to HCCs with the “activated” TGF-β signature (p = 0.0027). (3) We observed the greatest number of functional mutations in the SPTBN1 gene (6%), which encodes a tumor suppressor TGF-β/Smad3 adaptor protein. (4) Furthermore, we found a strong association between DNA damage response genes and the TGF-β pathway at both transcriptomic and genomic levels. Conclusions: The TGF-β pathway plays a pivotal role in liver tumorigenesis and the molecular signatures we characterize here appear to have prognostic significan