Zheng Yuqi, Fuse Haneya, Alzoubi Islam, Graeber Manuel B
Ken Parker Brain Tumour Research Laboratories, Brain and Mind Centre, Faculty of Medicine and Health, University of Sydney, Camperdown, Sydney, NSW 2050, Australia.
School of Medicine, Sydney Campus, University of Notre Dame, 160 Oxford Street, Darlinghurst, Sydney, NSW 2010, Australia.
Cells. 2025 Mar 11;14(6):413. doi: 10.3390/cells14060413.
Malignant gliomas, and notably glioblastoma, are highly aggressive brain tumors. Understanding the mechanisms underlying their progression is crucial for developing more effective treatments. Recent studies have highlighted the role of microglia and brain macrophages in glioblastoma development, but the specific interactions between these immune cells and glioblastoma stem cells (GSCs) remain unclear. To address this question, we have utilized AI-assisted cell recognition to investigate the spatial relationship between GSCs expressing high levels of CD276 (B7-H3) and microglia- and bone marrow-derived brain macrophages, respectively. Using PathoFusion, our previously developed open-source AI framework, we were able to map specific immunohistochemical phenotypes at the single-cell level within whole-slide images. This approach enabled us to selectively identify Iba1+ and CD163+ macrophages as well as CD276+ GSCs with high specificity and to study their co-localization. Our analysis suggests a closer association of Iba1+ macrophages with GSCs than between CD163+ macrophages and GSCs in glioblastoma. Our findings provide novel insights into the spatial context of tumor immunity in glioblastoma and point to microglia-GSC interactions as a potential mechanism for tumor progression, especially during diffuse tissue infiltration. These findings have significant implications for our understanding of glioblastoma biology, providing a foundation for a comprehensive analysis of microglia activation phenotypes during glioma development. This, in turn, may lead to new therapeutic strategies targeting the early stages of the immune microenvironment of glioblastoma.
恶性胶质瘤，尤其是胶质母细胞瘤，是极具侵袭性的脑肿瘤。了解其进展的潜在机制对于开发更有效的治疗方法至关重要。最近的研究强调了小胶质细胞和脑巨噬细胞在胶质母细胞瘤发展中的作用，但这些免疫细胞与胶质母细胞瘤干细胞（GSCs）之间的具体相互作用仍不清楚。为了解决这个问题，我们利用人工智能辅助细胞识别来研究分别表达高水平CD276（B7-H3）的GSCs与小胶质细胞和骨髓来源的脑巨噬细胞之间的空间关系。使用我们之前开发的开源人工智能框架PathoFusion，我们能够在全切片图像的单细胞水平上绘制特定的免疫组织化学表型。这种方法使我们能够以高特异性选择性地识别Iba1+和CD163+巨噬细胞以及CD276+GSCs，并研究它们的共定位。我们的分析表明，在胶质母细胞瘤中，Iba1+巨噬细胞与GSCs的关联比CD163+巨噬细胞与GSCs的关联更紧密。我们的研究结果为胶质母细胞瘤肿瘤免疫的空间背景提供了新的见解，并指出小胶质细胞与GSCs的相互作用是肿瘤进展的潜在机制，特别是在弥漫性组织浸润过程中。这些发现对我们理解胶质母细胞瘤生物学具有重要意义，为全面分析胶质瘤发展过程中小胶质细胞激活表型提供了基础。反过来，这可能会导致针对胶质母细胞瘤免疫微环境早期阶段的新治疗策略。
