Clinical Significance of Four Molecular Subtypes of Gastric Cancer Identified by The Cancer Genome Atlas Project

Abstract

Purpose: The Cancer Genome Atlas (TCGA) project recently uncovered four molecular subtypes of gastric cancer: Epstein-Barr virus (EBV), microsatellite instability (MSI), genomically stable (GS), and chromosomal instability (CIN). However, their clinical significances are currently unknown. We aimed to investigate the relationship between subtypes and prognosis of patients with gastric cancer.Experimental Design: Gene expression data from a TCGA cohort (n = 262) were used to develop a subtype prediction model, and the association of each subtype with survival and benefit from adjuvant chemotherapy was tested in 2 other cohorts (n = 267 and 432). An integrated risk assessment model (TCGA risk score) was also developed.Results: EBV subtype was associated with the best prognosis, and GS subtype was associated with the worst prognosis. Patients with MSI and CIN subtypes had poorer overall survival than those with EBV subtype but better overall survival than those with GS subtype (P = 0.004 and 0.03 in two cohorts, respectively). In multivariate Cox regression analyses, TCGA risk score was an independent prognostic factor [HR, 1.5; 95% confidence interval (CI), 1.2-1.9; P = 0.001]. Patients with the CIN subtype experienced the greatest benefit from adjuvant chemotherapy (HR, 0.39; 95% CI, 0.16-0.94; P = 0.03) and those with the GS subtype had the least benefit from adjuvant chemotherapy (HR, 0.83; 95% CI, 0.36-1.89; P = 0.65).Conclusions: Our prediction model successfully stratified patients by survival and adjuvant chemotherapy outcomes. Further development of the prediction model is warranted. Clin Cancer Res; 23(15); 4441-9. ©2017 AACR.

Figure 1. Prediction signatures for 4 molecular subtypes of gastric cancer in The Cancer Genome Atlas project (TCGA) cohort

Subtype-specific gene expression signatures were identified by applying multiple t tests (P < .001). Among the significant genes in each subtype, the top 200 genes were selected for development of prediction models (all 143 significant genes were used for the CIN subtype). Data are presented in matrix format in which each row represents an individual gene and each column represents a tissue sample. Each cell in the matrix represents the expression level of a gene feature in an individual tissue sample. The coloring in the cells reflects relatively high (red) and low (green) expression levels, as indicated in the scale bar (log₂ transformed scale). EBV, Epstein-Barr virus; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability.

Figure 2. Scatter plot matrix of Bayesian probability for each gastric cancer subtype predictor in the TCGA cohort

EBV, Epstein-Barr virus; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability.

Figure 3. Schematic diagram of The Cancer Genome Atlas project (TCGA) prediction model

A decision tree approach was employed for categorizing patients in test cohorts into the 4 subtypes of gastric cancer according to the Bayesian probability of each predictor. EBV, Epstein-Barr virus; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability.

Figure 4. Prognosis associated with each of the 4 subtypes of gastric cancer in 2 independent patient cohorts

Patients in the MDACC cohort (A) and SMC cohort (B) were stratified by subtype recurrence-free survival (RFS) and overall survival (OS) were plotted for each subtype. EBV, Epstein-Barr virus; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability.

Figure 5. Benefit of chemotherapy among patients with each subtype of gastric cancer

(a) Kaplan-Meier plots of recurrence-free survival (RFS) among patients who received adjuvant chemotherapy (CTX) and those who did not (no CTX) for each subtype. P values were obtained using the log-rank test. (b) Cox proportional hazards regression analysis estimating the benefit of adjuvant chemotherapy for patients with each subtype. The dotted line represents the 95% confidence intervals of the hazard ratios. EBV, Epstein-Barr virus; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability.

Figure 6. The Cancer Genome Atlas project (TCGA) risk score (TRS)

(a) Kaplan-Meier plots of recurrence-free survival (RFS) and overall survival (OS) among patients stratified by TRS. P values were obtained using the log-rank test. (b) Relationship between continuous TRS values and 5-year recurrence risk estimated by a Cox proportional hazards model. The dotted lines represent the 95% confidence intervals of the hazard ratios.