NEW YORK – A research team from the University of Miami and other centers in the US and China has identified a tumor gene expression signature that appears to correspond with treatment response in colon cancer patients receiving chemotherapy after cancer surgery.
"Our study lays the groundwork for improving adjuvant chemotherapy and potentially expanding into immunotherapy decision-making in colon cancer," co-senior and co-corresponding author Steven Chen, a biostatistician at the University of Miami Miller School of Medicine and the institute's Sylvester Comprehensive Cancer Center, and his colleagues wrote in Cell Reports Medicine on Thursday.
With the help of a semi-supervised gene network analysis and machine learning, the researchers scrutinized array-based gene expression data for 933 stage II or stage III colon cancer samples collected during surgery, prior to adjuvant chemotherapy (ACT) with 5-fluorouracil (5-FU)-based drugs.
The team's initial analyses suggested the expression of 18 genes reflected chemotherapy response in stage II or III colon cancer patients — a potentially prognostic expression signature that was subsequently whittled down to 10 genes.
The 18-gene signature was further tested using data from five additional colon cancer datasets, the researchers explained, while the 10-gene signature was assessed with expression profiles from two other datasets. The validation sets included NanoString nCounter-based profiles from formalin-fixed, paraffin-embedded samples.
"[W]e demonstrated both 18- and 10-gene signatures to have independent prognostic and predictive value beyond the clinical variables, including tumor stage," the authors reported.
When the team took a closer look at the pathways and processes associated with genes in the 10-gene prognostic signature, meanwhile, it identified pathways involved in everything from the tumor microenvironment to the cell cycle and cell surface interactions.
Because expression of genes in the 10-gene signature seemed to coincide with high microsatellite instability cases that responded to ACT, the investigators went on to explore possible ties between the ACT-related prognostic signature and a "tumor immune dysfunction and exclusion" (TIDE) score used to help predict response to immune checkpoint blockade immunotherapy.
In a set of 442 colon cancer samples, the authors saw a "significant, although moderate, correlation" between the 10-gene expression signature and the immunotherapy-related TIDE score. Even so, they cautioned that that association "should be interpreted with caution," as the 10-gene signature has not been directly linked to checkpoint immunotherapy response prediction itself.
The investigators also emphasized that further research, including clinical trials, will be needed to explore the ACT response predictability and prognostic applicability of the proposed 10-gene signature in a clinical setting. In addition, they noted that such expression patterns may provide insights into anticipated response to other treatment types, including immune checkpoint immunotherapy.
"[O]ur research introduces a 10-gene signature that could play a significant role in predicting ACT benefits," the authors wrote, noting that "its value in clinical decision-making could greatly enhance how we tailor ACT and immunotherapy for patients with colon cancer."
In the meantime, the team performed simulation analyses aimed at spelling out the cohort sizes needed to prospectively validate the proposed prognostic signature.
While larger cohort sizes are expected to improve predictive precision of the approach, for example, the authors suggested that "even smaller sample sizes in our study provide valuable insights, contributing significantly to the nuanced understanding of gene signature validations in oncological research."