Pharma Focus Europe
ThermoFisher Scientific - Custom and Bulks

Genome-wide Analyses Reveal the Contribution of Somatic Variants to the Immune Landscape of Multiple Cancer Types

Wenjian Bi, Zhiyu Xu, Feng Liu, Zhi Xie, Hao Liu, Xiaotian Zhu, Wenge Zhong, Peipei Zhang, Xing Tang 

Abstract

It has been well established that cancer cells can evade immune surveillance by mutating themselves. Understanding genetic alterations in cancer cells that contribute to immune regulation could lead to better immunotherapy patient stratification and identification of novel immune-oncology (IO) targets. In this report, we describe our effort of genome-wide association analyses across 22 TCGA cancer types to explore the associations between genetic alterations in cancer cells and 74 immune traits. Results showed that the tumor microenvironment (TME) is shaped by different gene mutations in different cancer types. Out of the key genes that drive multiple immune traits, top hit KEAP1 in lung adenocarcinoma (LUAD) was selected for validation. It was found that KEAP1 mutations can explain more than 10% of the variance for multiple immune traits in LUAD. Using public scRNA-seq data, further analysis confirmed that KEAP1 mutations activate the NRF2 pathway and promote a suppressive TME. The activation of the NRF2 pathway is negatively correlated with lower T cell infiltration and higher T cell exhaustion. Meanwhile, several immune check point genes, such as CD274 (PD-L1), are highly expressed in NRF2-activated cancer cells. By integrating multiple RNA-seq data, a NRF2 gene signature was curated, which predicts anti-PD1 therapy response better than CD274 gene alone in a mixed cohort of different subtypes of non-small cell lung cancer (NSCLC) including LUAD, highlighting the important role of KEAP1-NRF2 axis in shaping the TME in NSCLC.

Introduction

Over the last decades, the discovery of immune checkpoints and their applications in cancer therapy have revolutionized the treatment of various cancer types. [1] Immune checkpoint inhibition (ICI) therapies have been utilized as single agents or in combination with chemotherapies to treat over 50 types of cancer. Despite of these tremendous success, however, only a limited percentage of patients have achieved long-lasting benefits. [2,3] The ineffectiveness of immune-oncology (IO) therapies could be at least partially attributed to the imprecise selection of patients resulted from limited understanding of tumor microenvironment (TME). In the past, the study of cancer TME was largely restricted by the technology available for TME traits retrieval. Only a small number of TME traits can be derived from expensive and laborious experiments, such as flow cytometry [4] and immunohistochemistry [5]. Nowadays, with the advancements in omics technology and bioinformatics tools, various TME traits can be derived from RNA-seq data through de-convolution methods or gene signature enrichment analysis. [6,7] These bulk RNA-seq-derived traits have been shown to be highly consistent with immune traits obtained using cell flow cytometry or scRNA-seq. In a recent study by Sayaman et al., 139 TME traits were collected from multiple studies, mostly from bulk RNA-seq data. [8] 

Materials and Methods

TCGA source and data transformation

Genetic alteration data was downloaded from https://www.cbioportal.org/ for each cancer type. Clinical data was downloaded from https://portal.gdc.cancer.gov/. Immune traits were downloaded from Supplementary Tables 2–3 of Sayaman et al.[8] We first transform the TME traits to a quantitative or binary value depending on its distribution (S2 Table). For a TME trait, 1) if the raw trait values of more than 50% subjects are 0, then we dichotomize the trait to 0 and 1, depending on if the raw value is 0 or not; 2) if the raw trait values of more than 10% subjects are 0, we dichotomize the TME traits to 0 and 1, depending of if the raw value is less than median or not; 3) otherwise, we use inverse normalization transformation to calculate a quantitative value. Based on Variant Classification from cbioportal, we excluded somatic mutations annotated as 3’Flank, 3’UTR, 5’Flank, 5’UTR, Intron, and RNA. For each gene, if the sample is a somatic mutation carrier, the genotype was coded as 1, otherwise, the genotype was codes as 0. Genes with fewer than 5 somatic mutation carriers were excluded from analysis.

Results

Overview of genetic test to associate somatic variants with TME traits

We conducted genome-wide gene-level association analyses to identify genes in which somatic variants alter TME traits significantly. Of the TME traits Sayaman et al. analyzed [8], we selected 74 traits (S1 Table), most of which were derived from bulk RNA-seq data of TCGA tumor samples by scoring different gene signatures using ssGSEA or by deconvoluting bulk RNA-seq using CIBERSORT. [8,22] These immune traits were selected to represent proportion of different immune cell types, activity of immune pathways, and states of different immune cells. To ensure data processing workflow consistency across different cancer types, we used somatic mutations, log2 copy number alterations, and clinical data from TCGA panCancer project. [23] In total, 22 cancer types with > 100 samples were selected for analysis (S2 Table). Somatic mutations in non-coding regions (UTR or intron) except splicing changing variants were excluded from analysis. For each cancer, genes mutated in ≥ 5 tumor samples were fed into association test pipeline. 

Discussion

In this study, we systematically investigated the contribution of somatic variants to TME traits using TCGA cancer data. The results demonstrated that somatic variants play an important role in shaping cancer TME in tissue-specific and cancer-type-specific manner. A total of 451 significant gene-trait associations were identified across 22 cancer types, with 14 genes significantly associated with three or more TME traits, including IDH1 in KIRC, BRAF in THCA, CDH1 in BRCA, and TP53 in multiple cancer types. We select KEAP1-NRF2 in LUAD as an example to highlight the value of our study in patient stratification for IO therapies. Our results showed that KEAP1 mutated or NRF2 activated LUAD samples share unique TME characteristics, such as lower T cell infiltration, higher T cell exhaustion level, and higher expression of immune checkpoint ligands that are targeted by existing therapies. Real data analysis revealed that the NRF2 gene signature curated in our study could serve as a better biomarker than the currently used biomarker CD274 for selecting patients for anti-PD1/PD-L1 therapies. Moreover, we observed upregulation of the PVR gene (ligand for TIGIT) in NRF2 signature high LUAD tumors, which suggests that targeting NRF2 signature high population with anti-PD1 (or anti-PD-L1) and anti-TIGIT combo therapy may achieve synergistic effect. By May of 2022, the phase 3 SKYSCRAPER-01 study (NCT04294810) evaluating the addition of tiragolumab (anti-TIGIT) to atezolizumab (anti-PD-L1) as first-line treatment for people with PD-L1-high, locally advanced or metastatic non-small cell lung cancer (NSCLC) did not meet its co-primary end point of progression-free survival (PFS), according to Roche (https://bit.ly/37EbDJX). 

Acknowledgments

This research is supported by High-performance Computing Platform of Peking University. Jun Li gives great suggestions for scRNA-seq data analysis.

Citation: Bi W, Xu Z, Liu F, Xie Z, Liu H, Zhu X, et al. (2024) Genome-wide analyses reveal the contribution of somatic variants to the immune landscape of multiple cancer types. PLoS Genet 20(1): e1011134. https://doi.org/10.1371/journal.pgen.1011134

Editor: Peter Hammerman, MOMA Therapeutics, UNITED STATES

Received: September 6, 2023; Accepted: January 9, 2024; Published: January 19, 2024

Copyright: © 2024 Bi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting Information files.

Funding: This research was supported by National Natural Science Foundation of China (62273010, W. B.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: Zhiyu Xu., F.L., Zhi Xie, H.L., X.Z., W.Z., and X.T. are employees of Regor Pharmaceuticals Inc., Cambridge, Massachusetts, USA.

 

https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011134#abstract0

magazine-slider-img
Get instant
access to our latest e-book
patheon - Mastering API production at every scaleLab Indonesia 20247th Pharma Packaging and Labelling Forum 2024Future Labs Live - 2024World Orphan Drug Congress 2024World Vaccine Congress Europe 2024Sartorius Webinar - Pave Your Weigh to Accurate Analytical ResultsEUROPEAN PHARMA OUTSOURCING SUMMIT 2024patheon - Revolutionizing PharmaWorld Vaccine Congress Europe 2024