Gastric cancer is among the leading causes of cancer-related death worldwide. Based on Global Cancer Statistics (GLOBOCAN), it is the fifth most common neoplasm and the fourth most deadly cancer (CA: Cancer J Clin 2020; https://doi.org/10.3322/caac.21660). The prognosis of this type of cancer is poor as evidenced by the 5-year survival rate of 32.4 percent, according to the National Cancer Institute. This is due to most cases typically being metastatic when diagnosed and current pretreatment stratification modalities being inadequate (Front Med 2021; https://doi.org/10.3389/fmed.2021.744839)
Most patients with gastric cancer are treated with chemotherapy, and sometimes immunotherapy, as part of their treatment regimen (CA: Cancer J Clin 2021; https://doi.org/10.3322/caac.21657). And although adjuvant chemotherapy after surgery improves survival of resectable gastric cancer, many patients do not derive benefit from the potentially toxic therapies. In addition, biomarkers that predict patient response to chemotherapy and immune checkpoint inhibition in gastric cancer are lacking.
The use of next-generation sequencing has revealed the genomic landscape of a wide variety of cancers, including gastric cancer (Bioinformatics 2016; https://doi.org/10.1093/bioinformatics/btv692). However, the mutational cataloging of cancer genomes neglects to integrate functional analyses of gene-gene interaction and signaling pathway dynamics. As a result, a potentially wealthy source of prognostic and predictive information has generally been ignored. Genomic profiling can provide prognostic and predictive information to guide treatment choices for patients with gastric cancer at an advanced stage and in the palliative setting, such as immune checkpoint inhibitors.
A new study conducted by scientists at the Mayo Clinic Cancer Center in Florida examined how combination of artificial intelligence (AI) machine learning with genomic sequencing may predict the likelihood that patients with gastric cancer will derive benefit from chemotherapy or from immunotherapy (Nat Commun 2022; https://doi.org/10.1038/s41467-022-28437-y).
Study Details
To address this unmet need, a machine learning algorithm called NTriPath was employed that they had previously been developed to integrate pan-cancer somatic mutation data, gene-gene interaction networks, and pathway databases to distinguish prognostic cancer-associated molecular pathways. Using the algorithm, the researchers had identified prognostic gene signatures for various cancers, such as renal cell carcinoma, bladder carcinoma, melanoma, and head and neck squamous cell carcinoma.
In this retrospective analysis, NTriPath was utilized to identify gastric adenocarcinoma-specific pathways that are clinically relevant as an assistive tool to guide clinical care for patients with gastric cancer.
To build this model, Tae Hyun Hwang, PhD, senior author and principal investigator, along with his team applied NTriPath to perform a pan-cancer analysis of somatic mutations of 6,681 patients spanning 19 cancer types from The Cancer Genome Atlas Project and identified the pathways that were altered in gastric cancer. The researchers determined that the top three stomach cancer pathways consisted of a 32-gene gastric cancer signature that could be used to guide patient care decisions.
To determine clinical utility of this signature, the team used expression levels from a cohort of 567 patients-where 89 percent had either Stage II or Stage III gastric cancer-to define four distinct molecular subtypes that are prognostic for survival. Next, the team tested the prognostic utility of these genetic signatures and built a molecular subtype-based risk scoring model to predict overall survival and response to chemotherapy and immune checkpoint blockade. The risk score was validated using three large independent cohorts of patients with gastric cancer.
The findings revealed that the 32-gene gastric cancer signature are prognostic and predictive of response to adjuvant 5-fluorouracil and therapy after gastrectomy and to immune checkpoint inhibitors in patients with metastatic or recurrent disease.
"We were pleased that our 32-gene signature provided not only prognostic information, but also predicted patient benefit from chemotherapy and immunotherapy," Hwang stated. "In particular, we were surprised that the 32-gene signature we identified 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."
The team believes their discoveries represent a significant step forward in recognizing cancer patients that are likely to benefit from chemotherapy and/or immunotherapy.
"Similarly, we would also be able to identify patients who are unlikely to benefit from chemotherapy and immunotherapy, thereby sparing them the potential side effects of these therapies," Hwang noted.
For follow-up studies, the researchers proposed that the 32-gene signature still requires prospective validation. Additionally, the Hwang lab is working to develop new assays based on the expression level of a single or several genes to make biomarkers more available and easily implemented in the clinical setting.
"We are working on artificial intelligence algorithms that utilize diagnostic histopathology images to identify patients most likely to derive benefit from immunotherapy. We are also studying the molecular mechanisms of immunotherapy resistance made available by the machine learning and artificial intelligence approaches that we have developed in our lab," Hwang concluded.
Dibash Kumar Das is a contributing writer.