NEW YORK – Researchers are hoping to validate a gene signature for identifying cancer patients who are likely to respond to the commonly prescribed chemotherapy paclitaxel.
In Oncotarget this month, researchers described their efforts to discover and validate a set of genes whose differential expression is associated with better survival outcomes in patients with different tumor types receiving paclitaxel, a taxane chemotherapy. Before they can advance a predictive gene signature based on their findings, the researchers are first conducting a Phase II trial to explore how the overexpression of one of the genes identified, SSR3, as well as how the combination of differentially expressed genes, might impact glioblastoma patients' outcomes on paclitaxel.
Crismita Dmello, senior author and a research assistant professor of neurological surgery at Northwestern University's Feinberg School of Medicine, said the recently published research was spurred by a Phase I trial evaluating a novel way to deliver paclitaxel to the brains of patients with glioblastoma. Although the device — an implantable ultrasound device designed to open the blood-brain barrier — demonstrated that it could deliver paclitaxel to the brain in the study, a proportion of patients simply did not respond to the treatment, she explained.
"If there is a way we could preemptively tell who is going to respond depending on the expression profile of the tumor of individual patients, it would be beneficial for patient selection in the future and also in our ongoing [Phase II] trial to find a signal of response [to paclitaxel]," she said.
Paclitaxel is a widely used anti-cancer treatment, often prescribed for breast, pancreatic, ovarian, and non-small cell lung cancers. However, previous research has estimated that about half of the patients who receive paclitaxel do not benefit from it. The drug is also unable to cross the blood-brain barrier on its own, and despite preclinical models suggesting potential efficacy in patients with brain tumors or against cancers that have spread to the brain, the treatment has largely been unusable in these settings, according to the researchers.
After that initial Phase I glioblastoma study showing that some glioblastoma patients still didn't respond to paclitaxel even after it had crossed the blood-brain barrier, Dmello and her colleagues began exploring biomarkers associated with response to paclitaxel. They used CRISPR knockout models to turn certain genes off in glioblastoma cell lines and confirmed their initial insights in large datasets, such as The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO).
The CRISPR analysis identified 51 genes that were statistically significantly associated with paclitaxel response. The researchers then explored the expression of these 51 genes alongside overall survival outcomes in a dataset of breast cancer patients who received taxane-based chemotherapy including paclitaxel. That dataset narrowed the gene-expression signature to a smaller set of genes that were associated with paclitaxel response in glioblastoma and breast cancer: SSR3, CEP63, IRAK4, TMEM131, MBNL1, ZBTB20, and TDRD1. That study also found that SSR3 was overexpressed in glioblastoma cells as compared with the normal brain tissue, while SSR3 showed a spectrum of expression in breast cancer cells compared with the noncancerous breast tissue.
The researchers have further examined the correlation between SSR3 expression and paclitaxel outcomes using tumor samples from two studies of AngioChem's peptide antibody conjugate ANG1005, which combines paclitaxel with an LRP-1-targeted antibody to deliver the drug across the blood-brain barrier. By retrospectively analyzing patients' samples from two ANG1005 trials in glioma and brain metastasized solid tumors, Dmello and colleagues found "positive signals" of response prediction using SSR3 expression, she said.
The Phase II glioblastoma trial Dmello and colleagues are now conducting is designed to primarily evaluate the same implantable ultrasound device from the Phase I trial that allows paclitaxel to cross the blood-brain barrier. But in this study, researchers are also evaluating the relationship between SSR3 overexpression and overall survival on paclitaxel as an endpoint. Dmello's team also hopes to learn more about how the expression of the other genes they identified in the CRISPR screen may be associated with paclitaxel response in this trial.
"We have the data where we show that each of these genes has a contribution to [a patient's] susceptibility to the drug," Dmello explained, adding that they are now exploring the expression of different combinations of the genes in this study. "The idea that we are proposing and that we are working on is to find the right combination of genes that will have the best prediction of which patient is going to respond to this drug."
Once the current Phase II trial is complete, which is expected next year, the researchers will have the largest prospective validation cohort yet on SSR3 expression as a potentially predictive biomarker of response to paclitaxel.
She noted that there have been previous efforts to identify biomarkers of response or resistance to paclitaxel, but none have made it to clinical practice. One example of a biomarker of resistance identified was the overexpression of class III beta-tubulin, which is associated with limited response to taxane-based chemo in several cancers. Dmello noted that some of the research into biomarkers of response for paclitaxel stalled because they failed validation in larger studies or the researchers did not pursue the larger studies needed for validation.
Researchers, more recently, have also explored ways of predicting which patients will respond to standard chemo. One study published last month identified mutations in ERCC2 as a potential predictor of response to cisplatin-based chemo and radiation treatment in bladder cancer. Additionally, in a 2022 paper, researchers reported on a 32-gene signature that could be used to predict chemo responses in gastric cancer.
The type of technology necessary to gauge a biomarker certainly plays a role in its real-world adoption. To gauge SSR3 expression in tumor samples, Dmello and her colleagues used TissueGnostics' HistoQuest, an immunohistochemistry analysis software, which automates the analysis process for pathologists and aims to remove subjectivity in the analysis of SSR3 expression. Such a test may help broaden access to this type of gene expression testing, in Dmello's view.
While SSR3 is the first gene being validated from the set identified in the study, Dmello's team's ultimate goal is to develop a gene signature to predict response to paclitaxel. Toward that end, her team is continuing to explore the 51 original biomarkers identified by the CRISPR screen to determine if there is a DNA methylation signature that could be used to predict response, too.
"One gene like SSR3 cannot predict a response to an untargeted therapeutic drug like paclitaxel," Dmello said. "But if you use a set of genes that make a gene signature that regulate different pathways that are affected by treatment or that are regulated by the drug metabolism, then perhaps that is the way to go to find to find the response signal in these patients."