Written by Charlotte Harrison, Freelance Science writer
A potential new biomarker for gastric cancer
Choosing the best treatment for patients with gastric cancer can be tricky, due to the lack of available biomarkers that predict a patient’s response to therapy. Indeed, many patients do not get any benefit from treatments that are potentially toxic.
A study by researchers at Yonsei University College of Medicine in South Korea and the Mayo Clinic in the USA has used machine learning to identify a new genetic biomarker for gastric cancer. The biomarker could predict if a patient would respond to therapy and their prognosis.
A genetic signature
The researchers used their previously developed machine learning algorithm, known as NTriPath, to identify molecular pathways linked to gastric cancer. This algorithm integrates gene mutation data, gene–gene interaction networks, and cellular pathway databases.
Data on the genetic profile of over 6,500 patients with 19 cancer types from The Cancer Genome Atlas was inputted into NTriPath. The algorithm identified a signature of 32 genes specifically associated with gastric cancer.
The next step was to turn the 32-gene signature into a clinically useful tool. “We sought to use genomic sequencing to build a model that predicts the likelihood that a patient will derive benefit from chemotherapy or from immunotherapy,” said the study’s lead author Tae Hyun Hwang in a press release.
To this end, the researchers analyzed over 500 patient samples taken at the time of diagnosis, plus information on patient outcomes.
Predicting therapy response
Within the 32-gene signature, the researchers found four distinct molecular subtypes. For example, one subtype overexpressed genes associated with the cell cycle and DNA repair. These four molecular subtypes were able to predict patient responses to commonly used chemotherapy drugs, such as 5-fluorouracil and platinum chemotherapy.
Immunotherapy with checkpoint inhibitors is a treatment option for some types of gastric cancer. Samples from patients treated with checkpoint inhibitors were analyzed with RNA-sequencing and classified into one of the four molecular subtypes. Two of the subtypes had a much higher response to the checkpoint inhibitor pembrolizumab (Keytruda), showing that the gene signature can also predict the response to targeted therapies.
”We were surprised that the 32-gene signature…was able to predict a patient’s response to immunotherapy because identifying reliable biomarkers for immunotherapy response in patients with gastric cancer has been a challenge for the field,” said Hwang.
As well as predicting treatment response, the gene signature could also predict the five-year overall survival rate in patients with gastric cancer. So, this biomarker signature has the potential to optimize treatment choice for patients with gastric cancer. To realize this potential, the next steps of research involve validating the biomarker prospectively in larger groups of patients.
Image Credit: Canva