Tumor immunogenicity status in high-grade serous ovarian cancer

Abstract

Objectives: High Grade Serous Ovarian Cancer (HGSOC) patients experience an immune-mediated survival benefit, defined, in part, by the significantly prolonged survival of patients whose tumors exhibit a measure of abundant effector immune cell infiltration. Gene signatures that quantify the abundance of these cell types could reflect clinically relevant anti-tumor immunity. Our primary aim is to demonstrate the existence of immune-mediated survival benefit in HGSOC utilizing a novel method to estimate the proportion of immune cell types in HGSOC tumors. Methods: The Cell type Identification By Estimating Relative Subsets of known RNA Transcripts (CIBERSORT) method was developed by a group at Stanford to analyze cellular heterogeneity in RNA-seq data. CIBERSORT employs a novel support vector regression to deconvolve proportions of distinct cell types found within complex mixtures. Proportions of immune cell types in tumors have been computed for the Cancer Genome Atlas (TCGA) ovarian cancer (OVCAR) data set according to the CIBERSORT method and its associated LM-22 gene matrix (which comprises 547 genes that distinguish 22 human hematopoietic cell types). We utilized 376 serous ovarian cancer samples whose RNAseq expression profiles and clinical data were publically available. CIBERSORT derived cell proportion estimates were examined by Multivariable Cox Proportional Hazards Regression for significant associations between immune cell proportion estimates and overall survival (OS) of patients. Covariates included in the model were patient age and cancer stage. Kaplan-Meier plots were then created based on significant survival associations. Results: Multivariable cox proportional hazards model showed no significant association between age, stage, OS and cell proportion estimates of B cells (na&iuml;ve, memory), plasma cells, T cells (CD8, CD4 naive, CD4 memory resting, CD4 memory activated, regulatory (Tregs), natural killer cells (resting, activated), monocytes, macrophages (M0, M2), dendritic cells (resting, activated), mast cells (resting, activated), eosinophils, or neutrophils. The model revealed a significant association between patient age, tumor stage, OS and cell proportion estimates of follicular helper T cells (TFH HR 0.02 95% CI 0.00-0.37, p=0.008), gamma delta T cells (T&Gamma;&Delta; HR 0, 95% CI 0.00-0.07, p=0.022), and M1 macrophages (HR 0.002, 95% CI 0.00-0.08, p<0.001). Higher proportions of CD8+ CTLs (CD8) were not significant in the model (HR 0.8, 95% CI 0.09-6.67, p=0.834). Kaplan-Meier plots were consistent with these findings as were their log-rank P-values (Figure 1). Conclusions: Tumors with higher estimated proportions of follicular helper T cells (TFH), gamma delta T cells (T&Gamma;&Delta;) and M1 macrophages were significantly and independently associated with long-term survival. A model where anti-tumor immunity in HGSOC, as it relates to these immunological functions, provides survival-protective benefit for patients and should be a focus on future research in high-grade ovarian cancer. High Grade Serous Ovarian Cancer (HGSOC) patients experience an immune-mediated survival benefit, defined, in part, by the significantly prolonged survival of patients whose tumors exhibit a measure of abundant effector immune cell infiltration. Gene signatures that quantify the abundance of these cell types could reflect clinically relevant anti-tumor immunity. Our primary aim is to demonstrate the existence of immune-mediated survival benefit in HGSOC utilizing a novel method to estimate the proportion of immune cell types in HGSOC tumors. The Cell type Identification By Estimating Relative Subsets of known RNA Transcripts (CIBERSORT) method was developed by a group at Stanford to analyze cellular heterogeneity in RNA-seq data. CIBERSORT employs a novel support vector regression to deconvolve proportions of distinct cell types found within complex mixtures. Proportions of immune cell types in tumors have been computed for the Cancer Genome Atlas (TCGA) ovarian cancer (OVCAR) data set according to the CIBERSORT method and its associated LM-22 gene matrix (which comprises 547 genes that distinguish 22 human hematopoietic cell types). We utilized 376 serous ovarian cancer samples whose RNAseq expression profiles and clinical data were publically available. CIBERSORT derived cell proportion estimates were examined by Multivariable Cox Proportional Hazards Regression for significant associations between immune cell proportion estimates and overall survival (OS) of patients. Covariates included in the model were patient age and cancer stage. Kaplan-Meier plots were then created based on significant survival associations. Multivariable cox proportional hazards model showed no significant association between age, stage, OS and cell proportion estimates of B cells (na&iuml;ve, memory), plasma cells, T cells (CD8, CD4 naive, CD4 memory resting, CD4 memory activated, regulatory (Tregs), natural killer cells (resting, activated), monocytes, macrophages (M0, M2), dendritic cells (resting, activated), mast cells (resting, activated), eosinophils, or neutrophils. The model revealed a significant association between patient age, tumor stage, OS and cell proportion estimates of follicular helper T cells (TFH HR 0.02 95% CI 0.00-0.37, p=0.008), gamma delta T cells (T&Gamma;&Delta; HR 0, 95% CI 0.00-0.07, p=0.022), and M1 macrophages (HR 0.002, 95% CI 0.00-0.08, p<0.001). Higher proportions of CD8+ CTLs (CD8) were not significant in the model (HR 0.8, 95% CI 0.09-6.67, p=0.834). Kaplan-Meier plots were consistent with these findings as were their log-rank P-values (Figure 1). Tumors with higher estimated proportions of follicular helper T cells (TFH), gamma delta T cells (T&Gamma;&Delta;) and M1 macrophages were significantly and independently associated with long-term survival. A model where anti-tumor immunity in HGSOC, as it relates to these immunological functions, provides survival-protective benefit for patients and should be a focus on future research in high-grade ovarian cancer.