Replication stress, defined as impaired DNA replication leading to genomic instability, is a central feature of tumour development and progression. In cancer cells, it arises from oncogene activation, transcription-replication conflicts, altered nucleotide metabolism and DNA secondary structures that hinder DNA synthesis. These pressures compromise accurate genome duplication and promote chromosomal instability and mutagenesis. Recent advances have clarified the molecular mechanisms that stabilize, remodel and restart stressed replication forks, and highlighted the essential role of the ataxia telangiectasia and Rad3-related protein (ATR)-checkpoint kinase 1 (CHK1) pathway in preventing excessive single-stranded DNA accumulation and catastrophic genome fragmentation. Although replication stress fuels tumour evolution and heterogeneity, it also creates dependencies on replication stress response pathways that can be therapeutically exploited. Inhibitors targeting ATR, CHK1 and WEE1 kinases, which override the G2 checkpoint, are in clinical development, whereas synthetic lethal strategies, including poly(ADP-ribose) polymerase (PARP) inhibition in BRCA1-deficient or BRCA2-deficient tumours and inhibition of the DNA helicase WRN in microsatellite-unstable cancers, illustrate how replication stress-associated vulnerabilities can be targeted. In this Review, we position replication stress as both a driver of cancer genome evolution and a tractable framework for precision oncology. We emphasize that replication stress in tumours is often chronic and sublethal, shaping genome evolution. We further highlight its integration with oncogenic and metabolic alterations and broader impact on tumour-microenvironment interactions, which create dynamic and context-dependent therapeutic vulnerabilities.

Malignant transformation is driven by disruption of pathways regulating proliferation and cell fate, but these same disruptions can create a collateral vulnerability: loss of transcriptional and epigenetic control over transposable elements and other normally silenced genomic regions. Consequently, emerging cancer cells can accumulate transposable element-derived and other endogenous immunogenic nucleic acids capable of triggering antiviral responses, a process termed viral mimicry. Increasing evidence indicates that viral mimicry can eliminate precancerous cells and shape tumour evolution, positioning it as an intrinsic tumour-suppressive mechanism. Here we highlight how cancer-associated changes in DNA methylation, histone modifications, splicing and RNA processing can lead to the presence of immunogenic nucleic acids that can activate viral mimicry pathways. We outline how cancer cells suppress viral mimicry, including compensatory epigenetic repression, RNA editing, nucleic acid decay and dampening of interferon signalling to enable cancer cell growth. Finally, we highlight the evidence suggesting that escaping viral mimicry is a fundamental process for cancer initiation and progression, and suggest that viral mimicry escape is necessary for cancer transformation and a therapeutic target in combination with immunotherapies. By framing viral mimicry escape as a necessary part of cancer transformation, this Review provides a unifying conceptual model for its translational exploitation.

Every cancer carries the history of its own evolution, hidden in its genome. Modern DNA sequencing can catalogue millions of mutations and profile tumours across space and time, but sequencing alone struggles to answer the questions that matter most: when did key adaptations emerge, how strongly were they selected, why do some tumours relapse whereas others do not, and how will the cancer evolve next? The reason is fundamental: sequencing is a snapshot, whereas evolution is a dynamic process. Bridging this gap requires moving beyond descriptive cancer genomics towards quantitative evolutionary inference. In this Review, we argue that population genetics provides the mathematical framework needed to extract evolutionary dynamics from cancer genomes. We show how models of mutation, selection and drift transform allele frequencies from descriptive measurements into quantitative estimates of clonal fitness and evolutionary timings. We discuss how these principles extend to epigenetic inheritance, plasticity and ecological interactions within the tumour ecosystem, and examine the assumptions and limitations for their application to modern sequencing data. By reframing cancer genomes as quantitative records of evolutionary processes rather than catalogues of mutations, researchers have used population genetics to provide a foundation for understanding - and ultimately predicting - the trajectories of cancer evolution.

Genetic heterogeneity and clonal outgrowths are observed even in otherwise healthy human tissues, shaping the genetic composition of cell populations in non-malignant disease and during physiological ageing. This clonal mosaicism likely provides the pre-cancerous seeds for malignant transformation. Once a tumour arises, clonal evolution poses a major challenge to achieving cure, as clonal diversification provides an expanded number of substrates upon which therapy can act as a selective pressure, leading to the selection of resistant clones that ultimately fuel disease recurrence. Understanding somatic clonal evolution requires not only mapping genetic diversity but also defining the resulting phenotypes that provide a fitness advantage to mutated clones. This Review discusses multimodal single-cell technologies that enable the measurement of genotypes and additional molecular features from the same cell. These technologies unveil mutant-specific phenotypic traits, often show cell-state specificity in genotype-phenotype effects and can define therapeutic vulnerabilities for precision elimination of disease-propagating mutant cells. Furthermore, the combination of phylogenetic reconstruction with phenotypic measurements allows for the temporal mapping of clonal evolution and phenotypic plasticity. These breakthroughs have created a unique opportunity to define, directly in primary human samples, the mechanisms underlying clonal expansion in both healthy and malignant tissues.

Neutrophils, long appreciated for their central role in host defences, have largely pro-tumorigenic effects in cancer owing to their sensitivity to signals in the tumour environment. Cancers regularly co-opt the functions of neutrophils to promote angiogenesis, immunosuppression, tumour growth and metastasis. However, depending on context, many of the same neutrophil-derived factors involved in these processes can mediate tumour cell killing. In this Review, we examine the functions of neutrophils and their impact on cancer, drawing parallels between their roles in cancer and non-cancer contexts, and highlighting their remarkable sensitivity to environmental cues. We call attention to promising strategies for targeting and manipulating neutrophils for the treatment of cancer.

Early-onset colorectal cancer (EOCRC) among individuals under age 50 is increasing globally. Lifestyle factors such as obesity, metabolic comorbid conditions, poor diet and sedentary behaviours are linked to EOCRC, but critical gaps remain in identifying additional risk factors, with growing evidence for early-life and gut microbial-related exposures. In addition, how or when the risk factors act, individually or collectively, to initiate or promote colorectal cancer at much younger ages remains unknown. Compounding the complexity are the unique challenges of developing and implementing effective prevention strategies among younger populations. In this Roadmap, we review the progress and challenges of risk factor discovery for EOCRC, highlight opportunities for prevention and propose a transdisciplinary framework integrating population, mechanistic, behavioural and implementation sciences to accelerate causal risk factor discovery and translate insights into strategies to reverse the rising EOCRC burden. This framework, exemplified by emerging global initiatives including the Cancer Grand Challenges team PROSPECT, is broadly adaptable and intended to inspire collaborative efforts across the field. We also emphasize the critical role of patient perspectives and public engagement in shaping these efforts. With the urgency in risk factor discovery, scientists, the public and policymakers must unite to transform knowledge into life-saving solutions for future generations.

The 3D structure of chromatin is altered in cancer. In healthy cells, the genome is folded into functional 3D structural elements, including chromosomes, chromatin compartments, domains and loops. Cancer-associated alterations can modify each of these structural elements and, sometimes, their oncogenic capacity can be explained only when accounting for their effect on chromatin 3D conformation. In this Review, we discuss multiple lines of evidence demonstrating the impact of genetic and epigenetic alterations on chromatin structural elements, and the potential for chromatin 3D architecture to favour the acquisition of specific alterations. As a result, chromatin 3D remodelling emerges as both a cause and a consequence of oncogenic alterations. Finally, we highlight outstanding questions and opportunities related to the role of chromatin 3D heterogeneity in cancer, and the possibility of exploiting chromatin 3D conformation for diagnosis and therapy. Overall, chromatin 3D conformation integrates information from multiple molecular features, and thus it may represent a key to decoding cancer phenotypes.

Immunotherapies have substantially advanced cancer treatment; however, their efficacy in gliomas remains limited. This observation cannot be fully explained by tumour-intrinsic factors and may rather be linked to the distinct relationship between the central nervous system (CNS) and the immune system, commonly described as CNS immune privilege. CNS immune privilege is maintained by specialized brain barriers that divide the CNS into compartments with distinct accessibility to immune mediators and immune cells. Although maintaining homeostasis of the CNS parenchyma, these brain barriers direct CNS immune surveillance to the subarachnoid and the perivascular spaces at the CNS borders. Consequently, tumours arising in the CNS parenchyma are shielded from effective immune detection, limiting the efficacy of immunotherapies such as immune checkpoint inhibitors, cancer vaccines and adoptive T cell therapies such as chimeric antigen receptor (CAR) T cells and T cell receptor (TCR) transgenic T cells by restricting their access. Importantly, emerging evidence also indicates that gliomas actively remodel brain barrier functions to reinforce immune evasion. Failure to adequately consider brain barrier function in the context of immunotherapy strategies and clinical trial design therefore represents a major gap in the field. Understanding the orchestrated function of the brain barriers as neuroimmunological interfaces is essential for enhancing immune surveillance and improving immunotherapy responses in patients with brain tumours.