Beyond M1 M2: single-cell insights into how glioblastoma reshapes tumour-associated macrophages.

Glioblastoma (GBM) is an immunologically cold tumour in which glioma-associated microglia and macrophages constitute the predominant immune cell population, comprising brain-resident tumour-associated microglia (Mg) and infiltrating monocyte-derived (Mo) tumour-associated macrophages (TAMs). Recent single-cell and spatial transcriptomic studies have fundamentally revised the traditional M1/M2 activation paradigm, revealing a dynamic continuum of niche-dependent myeloid states rather than discrete polarisation phenotypes. In this review, we discuss how tumour-derived signals reshape Mg and Mo-TAMs through coordinated transcriptional, metabolic, and spatial reprogramming, with particular emphasis on interferon-responsive and antigen-presenting cell (APC)-like microglia, lipid-laden Mo-TAMs, and immunosuppressive myeloid populations. We critically compare myeloid states identified across independent mouse-derived and human GBM datasets, highlighting both convergent programmes and unresolved discrepancies in how axes such as interferon response, hypoxia and lipid handling are resolved at different taxonomic resolutions across studies. We further examine how metabolic constraints, including lipid accumulation, hypoxia and tryptophan catabolism, together with receptor-mediated inhibitory pathways and cytokine circuits, uncouple transcriptional identity from functional immune competence, creating a central paradox of GBM myeloid biology. Finally, we discuss emerging therapeutic strategies aimed at reversing tumour-driven myeloid reprogramming, including modulation of immunosuppressive circuits, disruption of metabolic symbiosis, inhibition of extracellular vesicles (EVs)-mediated communication, and spatially informed combinatorial immunotherapies. Collectively, we argue that while single-cell and spatial profiling have substantially redefined Mg and Mo-TAM biology in GBM, reconciling mouse-derived and human-derived myeloid taxonomies remains an open challenge with direct implications for translating myeloid-directed immunotherapeutic strategies from preclinical models to patients.