A study conducted by scientists at Yale Cancer Center demonstrates that spatial gene signatures can significantly enhance the prediction of immunotherapy outcomes in patients with melanoma. This research, whose first author is Thazin Nwe Aung, PhD, associate research scientist in pathology, was published June 5 in Clinical Cancer Research. Co-authors include Yale scientists David L. Rimm, MD, PhD, Anthony N. Brady Professor of Pathology and professor of medicine (medical oncology); Mark Gerstein, PhD, Albert L Williams Professor of Biomedical Informatics and professor of molecular biophysics & biochemistry, of computer science, and of statistics & data science; Jonathan H. Warrell, PhD, visiting fellow; and Harriet M. Kluger, MD, Harvey and Kate Cushing Professor of Medicine (Oncology) and professor of dermatology.
By focusing on the spatial aspects of gene expression within tumor tissue samples, the study provides oncologists with a refined method to forecast how patients will respond to immune checkpoint inhibitors (ICIs).
The investigation is based on the premise that the current gene signatures used to predict immunotherapy treatment outcomes lack accuracy due to their inability to account for the spatial distribution of cells within tumors and their surrounding microenvironment. To enhance predictive accuracy, the researchers utilized Digital Spatial Profiling of Whole Transcriptome Atlas (DSP-WTA) technology to collect gene expression data from three distinct cellular compartments (CD68+ macrophages, CD45+ leukocytes, and S100B+ tumor cells) across 55 melanoma specimens treated with ICIs.
Subsequently, the team developed a computational pipeline to derive compartment-specific gene signatures and evaluate whether the integration of spatial information could enhance patient stratification. The study found that their method performed well in predicting how patients would respond to ICIs. Among the three compartment-specific signatures that they developed; the tumor signature was the most accurate in an independent validation group of 45 patients. This signature included eight genes, with five genes indicating a positive response to treatment and three genes indicating resistance.
The researchers also showed that the S100B tumor spatial signature outperformed previously published gene signatures that did not consider spatial information. Their findings suggest that these spatially defined signatures, which use detailed information about the tumor and its microenvironment, can provide more accurate predictions to treatment outcomes.
The authors concluded that "the translational implications of our analysis pipeline could extend beyond melanoma, potentially optimizing immunotherapeutic approaches in other cancer types." They advocate for the prospective clinical assessment of these spatially defined melanoma compartment signatures to validate their utility in clinical practice.