The results, published in Science Advances on Wednesday, suggest that it may eventually be possible to better detect and manage those at risk of progression after chemotherapy treatment.
“Our findings help to identify already, prior to treatment, the tumors that are the most likely to have a poor response to therapy,” co-senior and co-corresponding author Anna Vähärautio, a researcher at the University of Helsinki, said in a statement. “Our results also suggest that therapies could be targeted at the inflammatory microenvironment of cancer cells and the surrounding tissue to improve treatment outcomes with the help of combination therapy.”
Vähärautio and her colleagues searched for transcriptome features corresponding with resistance to neoadjuvant chemotherapy using single-cell RNA sequence profiles for nearly 51,800 tumor, stromal, or immune cells from high-grade serous ovarian cancer samples collected prospectively from 11 patients at Turku University Hospital. They also considered RNA in situ hybridization data for 10 of the matched pre- and post-treatment tumor sets, along with bulk RNA-seq data for dozens more paired pre- and post-treatment or pre-treatment and relapse samples.
The team analyzed RNA-seq profiles in pre-treatment diagnostic laparoscopy samples and in post-chemotherapy samples obtained after debulking surgery using a computational method called PRIMUS that was designed to dial down patient-specific features to find shared transcriptional signals linked to chemotherapy response.
In ovarian cancer patients with chemotherapy-resistant tumors primed to progress to metastasis, they found a rise in stress-associated cell features that had been present in baseline samples.
“The stress-associated state exists before chemotherapy, is subclonally enriched during the treatment, and associates with poor progression-free survival,” the authors reported. “Co-occurrence with an inflammatory cancer-associated fibroblast subtype in tumors implies that chemotherapy is associated with stress response in both cancer cells and stroma, driving a paracrine feed-forward loop.”
Those results lined up with patterns found in another 271 individuals with ovarian cancer who had their pre-treatment tumors tested by bulk RNA-seq through the Cancer Genome Atlas project, the researchers reported. Along with available gene expression profiles, they used reverse-phase protein array testing to distinguish between cases with stress-high or stress-low tumors.
That group included 86 individuals with high-grade serous ovarian tumors classified as stress-high and 144 individuals with stress-low tumors, the team noted. While the median progression-free survival time was 21.2 months for patients with stress-low tumors, it dropped to just shy of 15 months in the subset of patients with ovarian tumors showing high levels of stress-associated features.
With expression profiles for more than a dozen immune or stromal cell types that turned up in the tumor microenvironment samples, meanwhile, the researchers found related inflammation and paracrine signaling in stromal cells in the microenvironment of tumors showing stress-associated features, along with enhanced representation of cancer-associated fibroblast cells.
Based on these and other findings, the authors suggested that “a combination of induced and selective processes” contribute to the transcriptomic changes detected following chemotherapy — shifts that may be influenced by everything from tumor microenvironment features to differences in an individual’s underlying genetics and biology.
“[T]he identification of [a] stress signature opens avenues for combinatorial drug testing in preclinical models that maintain both subclonal heterogeneity and paracrine tumor-stromal signaling,” they concluded. “As many drugs targeting inflammatory effectors are already in clinical use for other indications, they may offer a realistic option for safe combinatorial therapies with a wide array of currently used oncological drugs to restrain the broadly adaptive stress response of tumors.”
This article was published by Genome Web.