2656 - Immune Landscape of Radiation Necrosis vs. Local Failure in Melanoma Brain Metastases Treated with Stereotactic Radiosurgery
Presenter(s)
M. Qazi1, R. Higazy1, P. A. Jablonska1, V. Sotov2, A. Gao3, M. O. Butler2, and D. Shultz1; 1Radiation Medicine Program, Princess Margaret Cancer Centre, Toronto, ON, Canada, 2Department of Medical Oncology and Hematology, Princess Margaret Cancer Centre, Toronto, ON, Canada, 3Department of Pathology, Princess Margaret Cancer Centre, Toronto, ON, Canada
Purpose/Objective(s): Stereotactic radiosurgery (SRS) is the preferred treatment modality for patients with limited numbers of melanoma brain metastases (MBrMets). Radiation necrosis (RN) and local failure (LF) are two common potential adverse outcomes of SRS for which the underlying pathobiological mechanisms remain poorly characterized. We hypothesize that specific patterns of immune activity, broadly characterized as inflammatory versus suppressive, are responsible for the pathogenesis of RN and LF, respectively.
Materials/Methods: In this study, we performed imaging mass cytometry (IMC) using 38 metal-tagged antibodies to assess the phenotypes and activation states of various immune cell types in MBrMets FFPE samples resected at the time of LF or RN development following SRS (n=3 each). Each tissue slide was segmented into tumor, stroma, necrosis, gliotic brain, and hemorrhage regions to facilitate comprehensive immune characterization. Cell segmentation was conducted using DeepCell to distinguish individual cells based on nuclear and membrane markers. Following segmentation, single-cell data were extracted, and marker intensities were normalized across all samples. Clustering was performed using PhenoGraph to identify distinct immune cell populations, followed by UMAP for dimensionality reduction and visualization. Manual annotation was performed based on established marker expression patterns to classify immune subsets.
Results: Our preliminary analysis reveals distinct immune landscapes in RN and LF. RN is characterized by reactive astrocytes and significant neutrophil infiltration, indicative of an inflammatory microenvironment. In contrast, LF is associated with an immunosuppressive milieu, marked by exhausted T cells and high expression of immunosuppressive macrophage markers, suggesting a tumor-promoting microenvironment. Additionally, FAP-positive stromal cells are enriched in LF, supporting a pro-tumorigenic stroma. These findings highlight key differences in immune composition between RN and LF, providing insights into their distinct pathological mechanisms.
Conclusion: Our data shows differential immune makeup of RN vs LF in MBrMets treated with SRS with unique immune populations present in RN samples. Further analysis will focus on determining the inflammatory vs suppressive propensity of these immune populations as well as spatial localization to delineate possible cell-cell interactions.