Deep Clustering-Based Immunotherapy Prediction for Gastric Cancer mRNA Vaccine Development.

Area of Science:

• Oncology
• Immunology
• Bioinformatics

Background:

• Immunotherapy shows promise for cancer treatment but benefits limited gastric cancer (GC) patients due to heterogeneous immunosuppressive tumor microenvironments.
• Sophisticated immunological subclassification of GC is crucial for advancing mRNA vaccine therapy.

Purpose of the Study:

• To establish a novel immunological subclassification for GC patients.
• To identify specific tumor antigens for the development of mRNA vaccines.

Main Methods:

• Deep autoencoder (AE)-based clustering to define four GC immune subtypes.
• Machine learning for accurate classification (97.6% average accuracy).
• Immunoinformatic tools to identify tumor-specific antigens and develop a customized mRNA vaccine for Subtype 4.

Main Results:

• Four distinct GC immune subtypes (Subtypes 1-4) were identified, showing significant clinicopathological, molecular, and immune differences.
• Subtype 4 exhibited poor prognosis, lower tumor purity, and increased immune cell infiltration and activity.
• Tumor-specific antigens for Subtype 4 were identified, leading to the development of a targeted mRNA vaccine.

Conclusions:

• The novel subclassification provides new insights into GC heterogeneity.
• Targeted mRNA vaccine development for specific immune subtypes, like Subtype 4, may improve immunotherapy efficacy.
• Understanding the tumor microenvironment (TME) is key to optimizing treatment strategies for GC patients.