Abstract 1212P
Background
We aim to uncover robust subtypes having distinct biological characteristics associated with clinical outcome and identify subtype-specific therapeutic targets.
Methods
Genomic data from 2527 gastric tumors are used to uncover clinically relevant molecular subtypes. Cluster of clusters assignment (COCA) approach was applied to integrate 8 genomic subtypes and uncover consensus subtypes. For validation, we identified 120 genes whose expression is highly specific to each subtype and used them to construct gastric cancer predictor of integrated consensus subtype with 120 genes (GPICS120).
Results
Analysis of genomic data revealed 6 consensus subtypes that showed marked interconnectivity among 8 independent classification systems. Consensus genomic subtype 1 (CGS1) is characterized by poorest prognosis, very high stem cell characteristics, and high IGF1 expression, but low genomic alterations. CGS2 showed canonical epithelial gene expression patterns. CGS3 and CGS4 are characterized by high copy number alterations and low immune activity. However, CGS3 and CGS4 are different in high HER2 activity (CGS3) and high lipid metabolic activity (CSG4). CGS5 has highest mutation rates and moderately high immune activity that is characteristics of MSI-high tumors. Most of CGS6 tumors are EBV-positive and shows extremely high methylation and high immune activity. Clinically, CGS1 and CGS4 are poor prognostic while prognosis of CGS2, CGS4, CGS5, and CGS6 is good. Interestingly, patients in CGS4 have very poor survival rate after relapse. By applying systematic analysis of genomic and proteomic data, we estimated potential response rate of each subtype to standard and experimental treatments such as chemoradiation therapy, target therapy, and immunotherapy. Importantly, association of subtype CGS3 with response to chemoradiation therapy was further validated with samples from phase III clinical trial and in experiments with cell lines.
Conclusions
Consensus subtype is robust classification system and can be the basis for future clinical investigation of subtype-based targeted interventions.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
CDMRP.
Disclosure
All authors have declared no conflicts of interest.