NEW YORK – An international team led by researchers at the University of Oxford has evaluated dozens of tumor hypoxia gene expression signatures and summary scores, using their findings to come up with recommendations for detecting low oxygen conditions related to tumor adaptations.
"This work gives much-needed clarity to the field and provides an important reference to laboratory and clinician scientists who seek validation of hypoxic status and/or are considering orchestration of prospective trials," co-senior and corresponding author Benjamin Harris of the University of Oxford and his colleagues wrote in a paper published in Cell Genomics on Friday.
For their analyses, the researchers reviewed and compared 70 proposed hypoxia gene expression signatures — each based on the expression of between three and 759 genes — and 14 expression summary scores in tumor samples and cell lines assessed by bulk RNA sequencing or single-cell RNA-seq following exposure to low oxygen or typical oxygen conditions.
Prior research suggests that a dip in tissue oxygen in the tumor microenvironment can boost drug or radiotherapy resistance, genomic instability, and the risk of metastases, the team explained, noting that past efforts to come up with hypoxia-targeting or hypoxia-activated treatments have been met with limited success, potentially owing in part to a lack of consistent strategies for identifying cases most apt to benefit from a hypoxia-targeted approach.
"Currently, there is no agreement on which signature to use in which context, nor a systematic evaluation of differential signature performance with different summarization methods (scores)," the authors explained, noting that "lack of [patient] stratification has been cited to be sufficient to account for the failure of Phase III trials for hypoxia-activated prodrugs."
When they brought together published hypoxia gene expression signature data, the researchers found that they included an average of 55 genes, with a median signature size of 24 genes. Though several genes and pathways were overrepresented in these signatures, they did not identify any genes that turned up across all of them.
Along with enrichment for HIF-1 signaling genes and related conditions, for example, the team found that the expression signatures tended to contain genes from endoplasmic reticulum (ER) stress response and ER-based protein processing pathways and genes linked to several non-cancer conditions.
In a pan-cancer analysis that included data on 104 cell lines, the investigators used a random gene set analytical approach to assess hypoxia scores and signatures in bulk RNA-sequenced cell line pairs exposed to low oxygen or typical oxygen conditions. They profiled these with a range of sequencing platforms and compared the signatures and scores to one another and to the hypoxia predictive performance of random gene sets.
After bringing in single-cell RNA-seq analyses on two breast cancer cell lines, and available data for 5,401 tumor samples profiled for the Cancer Genome Atlas project, the team highlighted several expression signatures that appeared to perform well in either cell lines or clinical samples.
While a signature known as Tardon detected hypoxia with an estimated accuracy of 94 percent in cell lines profiled with bulk or single-cell RNA-seq, the researchers noted that signatures such as Buffa or Ragnum appeared to perform well for finding tumor hypoxia in clinical tumor samples.
The authors cautioned that the commonly used cell lines they included "may not fully capture the heterogeneity of hypoxic responses observed across diverse tissue and cancer subtypes."
"Expanding validations to underrepresented cell lines, particularly those reflective of less-studied malignancies such as colorectal or kidney cancers, would enhance the robustness of these findings," they wrote, noting that a "concerted effort by the scientific community to contribute more … scRNA-seq data will be valuable."