Reactive oxygen species (ROS) production in cancer cells is elevated by oncogenic signalling, tumour suppressor loss, metabolic reprogramming, and nutrient stress within the tumour microenvironment. Hydrogen peroxide can promote cellular states that favour cancer initiation and a metastatic phenotype. By contrast, excessive ROS production, particularly the uncontrolled generation of lipid peroxides, can trigger cell death pathways such as ferroptosis, thereby limiting cancer progression and metastasis. Cancer cells, therefore, increase their antioxidant capacity to maintain ROS levels high enough to support signalling, but below toxic levels. Antioxidant systems can suppress early neoplastic transformation in some tissues by limiting ROS during precancerous lesion formation. However, after transformation, these pathways can support tumour maintenance, metastatic dissemination and therapy resistance. For decades, strategies to modulate oxidative stress focused primarily on administering dietary antioxidants. However, the lack of clinical benefit from broad dietary antioxidant supplementation and even evidence of harm in some settings, together with growing evidence that tumours are selectively dependent on specific antioxidant pathways at distinct stages of tumorigenesis and metastasis, has shifted attention towards targeting antioxidant defences in established cancers. The key question is whether these pathways can be targeted with sufficient selectivity to spare normal tissues. In this Perspective, I examine how tumour stage-specific dependencies on antioxidant pathways create therapeutic vulnerabilities that can be exploited to treat tumours, prevent metastasis and overcome therapy resistance. Antioxidant pathways support tumour growth, metastasis and therapy resistance while protecting cancer cells from ferroptosis. In this Perspective, Navdeep Chandel explores how stage-specific antioxidant dependencies can be exploited for therapy.
Tumour progression fundamentally disrupts extracellular matrix (ECM) homeostasis, generating circulating molecular signatures that reflect the integrated activities of cancer cells, stromal fibroblasts and infiltrating immune populations. These tumour ECM-derived fragments comprise specific protease-generated neo-epitopes of collagens, proteoglycans and glycoproteins, and provide dynamic, real-time readouts of microenvironmental remodelling that distinguish them from other classes of biomarker, such as genomic biomarkers. Recent technical advances in neo-epitope detection platforms have enabled measurement of specific ECM fragments with exceptional sensitivity and specificity, transforming these signatures from research tools into clinically viable biomarkers. ECM fragment fingerprints are demonstrating utility across the cancer continuum, including tracking disease activity; differentiating malignancy from benign conditions during early detection; stratifying patients by biological aggressiveness of the disease independent of anatomical staging; monitoring the treatment response with pharmacodynamic precision; and predicting immunotherapeutic efficacy through immune–ECM interaction signatures. Established regulatory pathways through the US Food and Drug Administration Biomarker Qualification Program and European Union In Vitro Diagnostic Regulation frameworks provide clear routes to clinical implementation. Thus, the convergence of robust analytical platforms, mechanistic understanding and regulatory readiness is positioning ECM fragment fingerprinting as a potential emerging component of precision oncology, complementing genomic profiling to enable comprehensive tumour ecosystem characterization and rational therapy selection. Through the action of proteases, the tumour-derived extracellular matrix (ECM) generates circulating ECM fragments that reflect dynamic tumour microenvironment remodelling in real time. In this Perspective article, the authors propose that advances in neo-epitope detection have enabled the use of ECM fragments as sensitive and specific biomarkers in clinical applications ranging from early detection and prognostication to treatment monitoring and immunotherapy prediction, complementing genomic profiling in precision oncology.
Tumour heterogeneity is a defining feature of cancer and a major obstacle to effective therapy as it drives metastasis, resistance and relapse. Experimental tumour models have largely exploited simplified, homogeneous systems that fail to capture the full biological and spatial diversity of real tumours. Recent advances in 3D models have considerably expanded our capacity to model different dimensions of tumour heterogeneity. These platforms also enable controlled perturbations of cancer-cell-intrinsic and microenvironmental features, including genetic programmes, therapy responses and transport, extracellular matrix composition and mechanics, oxygenation, perfusion, stromal and immune interactions, and tissue architecture. Such perturbations could reveal how different aspects of tumour heterogeneity shape disease progression and therapy response dynamics. However, how close we are to generating 3D models that not only recapitulate the dynamic, multiscale nature of human cancers but also enable systematic perturbations for mechanistic insights remains a key question. In this Review, we summarize the current landscape of 3D models and how effectively they capture tumour heterogeneity, what perturbations they enable, and how these capabilities can be matched to specific biological or therapeutic questions. We also discuss outstanding challenges in model fidelity, standardization, scalability, longitudinal analysis and clinical predictability. Tumour heterogeneity drives disease progression yet remains difficult to capture in experimental systems. In this Review, Fomina and McGuigan discuss how emerging 3D tumour models, including organoids, assembloids and microfluidic platforms, recapitulate distinct dimensions of tumour heterogeneity to enable mechanistic studies of cancer biology and drug response. They also outline future directions for improving model fidelity, scalability and translational utility.
The undulating topography of the intestinal epithelium creates regionalized homeostatic niches at the base of each crypt, with each niche supporting neutrally evolving oligoclonal stem cell populations. At the crypt base, the fate of stem cell progeny is tightly regulated by intercompartmental polarized morphogenetic gradients, which converge to instruct continuous, but controlled, epithelial-cell turnover. To initiate tumorigenesis, a neoplastic clone must defy or escape such morphogenetic constraints either by outcompeting neighbouring stem cells or by establishing an ectopic stem cell population de novo. The position of the cell-of-origin within the crypt thus determines the oncogenic mechanisms required to subvert local homeostatic control. In turn, this defines the route of carcinogenesis and underpins the pervasive heterogeneity of colorectal cancers. Here, we explore the relationship between the position of the cell-of-origin on the vertical crypt axis and the cell plasticity mechanisms co-opted and corrupted to escape homeostatic control. We consider how this influences clonal dynamics at the pre-cancer stage and initiates distinct carcinogenesis pathways engendering morphologically discrete precursor lesions. Finally, we discuss how such lesions progress to molecularly heterogeneous subtypes with distinct clinicopathological features and therapeutic outcomes, and how phenotypic, rather than just genotypic, heterogeneity promotes tumour-cell adaptation, therapeutic resistance and metastasis. This Review by Sphyris et al. discusses how the cell-of-origin determines the route of colorectal carcinogenesis and how phenotypic plasticity and heterogeneity shape tumour evolution, disease recurrence, metastatic competence and response to therapy.
Ye et al. show that ischaemic stroke can drive glioma progression through microenvironmental remodelling of tumour-associated astrocytes.
