Conclusion
Our results highlight the presence of transcriptionally and histologically distinct immune subtypes among ovarian tumors and emphasize the potential benefit of immune subtyping as a clinical tool for treatment tailoring.
Results
Using clinical and gene expression data of 489 ovarian cancer patients in the publicly available dataset, we identified three transcriptionally distinct clusters, representing inflamed, excluded, and desert subtypes. We developed a two-step subtyping algorithm with COL5A2 serving as a marker for separating excluded tumors, and CD2, TAP1, and ICOS for distinguishing between inflamed and desert tumors. The accuracy of gene expression-based subtyping algorithm in a real-world cohort was 75%. Additionally, we confirmed that patients bearing inflamed tumors are more likely to survive longer.