Emerging trends and converging evidence in tumor evolution: A comprehensive review.

[BACKGROUND] Tumor evolution is a spatiotemporal dynamic process orchestrated by the interplay of genetic mutations, epigenetic reprogramming, and bidirectional microenvironmental interactions, which collectively generate the phenotypic diversity necessary for cancer progression, metastasis, and therapeutic resistance. Foundational models - linear, branched, neutral, and parallel evolution - provide complementary, albeit incomplete, frameworks to illustrate how tumors diversify through the accumulation of gradual mutations or catastrophic genomic events (e.g., chromosomal instability, disruptions in topologically associating chromatin domains). The applicability of each model is context-dependent, shaped by the specific selective pressures present across space and time. These evolutionary processes are fundamental to clonal heterogeneity, immune evasion, and the establishment of adaptive cellular ecosystems.

[CONTENT] Somatic mutations, including single-base substitutions and structural variations, function as evolutionary barcodes that trace tumor lineage. Beyond the genetic code, epigenetic dysregulation-encompassing DNA hypermethylation that silences tumor suppressors and dynamic RNA modifications (e.g., m6A) that fine-tune mRNA stability-confers a layer of phenotypic plasticity. This allows for rapid, often reversible, adaptation to therapeutic and microenvironmental stresses without altering the underlying DNA sequence, thereby generating non-genetic heterogeneity. Non-coding RNAs, including microRNAs that post-transcriptionally fine-tune gene expression and circular RNAs that can function as miRNA sponges or encode peptides, comprise a critical regulatory network. They orchestrate oncogenic signaling, metastasis, and immune suppression, often in response to signals from the tumor microenvironment, thereby integrating diverse cues to shape evolutionary trajectories. The tumor microenvironment transcends a passive supportive role to act as a dynamic and decisive orchestrator of evolution: hypoxia stabilizes HIFs to drive angiogenic and metabolic reprogramming; lactate accumulation in acidic niches polarizes immunosuppressive macrophages; and neural-tumor crosstalk promotes perineural invasion. These bi-directional interactions create distinct ecological niches that exert spatially heterogeneous selection pressures, fundamentally shaping the clonal landscape. Treatment pressures (e.g., chemotherapy, radiotherapy, immunotherapy, etc.) impose evolutionary bottlenecks, selecting resistant clones and fostering cross-resistance through shared pathways. Emerging technologies - single-cell sequencing, spatial multi-omics, and liquid biopsies - now decode intra-tumoral heterogeneity, map cellular ecosystems, and monitor clonal dynamics in real time and multiple dimensions.

[CONCLUSION] Integrating evolutionary models with multi-omics data reveals the complexity of tumor adaptation, emphasizing the need for temporally adaptive therapeutic strategies. Current preclinical models inadequately recapitulate human tumor-microenvironment interactions, necessitating advanced systems to bridge this translational gap. Looking forward, the convergence of artificial intelligence and dense, longitudinal biomarker profiling holds the potential to move personalized oncology beyond static genomic matching. The future lies in refining dynamic interventions that simultaneously target the dual pillars of malignancy: the molecular hallmarks of cancer cells and the ecological hallmarks of the tumor ecosystem, thereby aiming to control the process of evolution itself.