3D Multi-Omics Tumour Atlases: From Technology to Clinical Translation

The convergence of spatial omics technologies and three-dimensional tissue imaging is reshaping how researchers understand tumour biology. A comprehensive review published in Nature Reviews Cancer charts the rapid evolution of 3D multi-omics tumour atlases, from foundational technologies to emerging clinical applications, and argues that volumetric molecular profiling is essential for capturing the true complexity of cancer.

The central premise is that tumours are inherently three-dimensional ecosystems. As Forjaz and colleagues demonstrated in 2025, "three-dimensional assessments are necessary to determine the true, spatially resolved composition of tissues." This finding underpins a growing consensus that 2D tissue sections, while historically transformative, provide an incomplete picture of tumour architecture, clonal dynamics, and immune interactions.

Technological Foundations for 3D Tumour Mapping

The technological backbone of 3D tumour atlases rests on advances in tissue processing, volumetric imaging, and spatial molecular profiling. Glaser and colleagues established slide-free, open-top light-sheet microscopy as "a practical route to rapid, non-destructive volumetric pathology of large clinical specimens," providing a technical foundation for 3D tumour atlas acquisition. Complementing this, Kiemen and colleagues introduced CODA, "a scalable framework for reconstructing large tissues from hundreds to thousands of serial H&E sections at cellular resolution, enabling quantitative 3D reconstruction from routine pathology material."

For molecular profiling in three dimensions, several key innovations have emerged. Kuett and colleagues extended imaging mass cytometry (IMC) into three dimensions, "enabling highly multiplexed single-cell mapping of tumour and microenvironmental architecture across tissue volumes." On the transcriptomics front, Sui and colleagues advanced imaging-based spatial molecular profiling "from thin sections to thick tissue blocks, enabling scalable 3D single-cell transcriptomic and translatomic mapping in intact tissues." Gandin and colleagues further pushed the frontier with "deep-tissue spatial transcriptomics by coupling high-resolution transcript detection with subcellular imaging in thick specimens, substantially expanding the depth and resolution accessible to 3D molecular atlas studies."

Landmark 3D Atlas Studies

Several landmark studies illustrate the biological insights gained through volumetric analysis. Lin and colleagues established "a landmark 3D colorectal cancer (搜索)搜索该疾病 atlas showing how tumour cell state transitions and immune interactions are organized across intact tumour volumes." In pancreatic cancer (搜索)搜索该疾病, Braxton and colleagues "revealed the multifocal architecture of human pancreatic precancers through 3D genomic mapping, illustrating how volumetric analysis can reshape the understanding of tumour initiation."

Perhaps most strikingly, Mo and colleagues demonstrated that "3D reconstruction refines inference in tumour evolution and microenvironmental interactions beyond what is possible in 2D analyses." This finding has profound implications for how researchers study clonal dynamics, immune evasion, and therapeutic resistance.

Multi-Omics Integration Across Scales

The review highlights the integration of diverse molecular layers. Spatial proteomics has been advanced by deep visual proteomics, which Rosenberger and colleagues showed "can resolve spatial proteotoxicity programmes in situ, highlighting the power of image-guided proteomics for mechanistic tissue analysis." Hu and colleagues introduced "a high-resolution spatial proteomics framework that combines microfluidics with transfer learning, markedly improving the scalability and anatomical precision of tissue proteome mapping."

For the 3D cancer genome, Liu and colleagues "generated single-cell 3D genome atlases of mutant KRAS (搜索)搜索该靶点-driven cancers, linking chromatin architecture to tumour evolution in the native tissue context across disease progression." Chang and colleagues established "a scalable single-cell chromatin conformation capture technology for heterogeneous tissues, providing a feasible technology for 3D cancer genome atlas construction."

Metabolomics adds yet another dimension. Techniques including MALDI-MSI, DESI-MSI, and stimulated Raman scattering microscopy enable label-free chemical imaging. Jang and colleagues provided "a super-resolution SRS framework for label-free chemical imaging at subcellular resolution, supporting nanoscale studies of cellular metabolism."

Computational Frameworks and AI

The scale and complexity of 3D multi-omics data demand sophisticated computational approaches. Song and colleagues established that "weakly supervised AI on volumetric 3D pathology can outperform 2D baselines for prognosis, thereby providing a computational framework for the analysis of 3D histopathology at scale."

For data integration, Qiu and colleagues introduced "a spatiotemporal modelling framework for molecular holograms that enables large-scale reconstruction, alignment and interrogation of whole-organism 3D cellular atlases." Klein and colleagues established "a scalable optimal-transport framework for mapping cells through time and space, offering a powerful computational foundation for reconstructing cellular trajectories in atlas-scale datasets."

Foundation models are increasingly being applied. Tejada-Lapuerta and colleagues developed Nicheformer, "a foundation model for single-cell and spatial omics," while Chen and colleagues created "a visual-omics foundation model to bridge histopathology with spatial transcriptomics."

Clinical Translation and Future Directions

The path to clinical translation is being paved by studies that connect 3D tumour architecture to patient outcomes. Bai and colleagues extended spatial transcriptomics in FFPE tissue "from gene-expression mapping to broader RNA biology, including RNA processing and variation, thereby increasing the clinical utility of archival specimens for spatial oncology."

Large-scale initiatives including the Human Tumor Atlas Network, HuBMAP, and the Human Cell Atlas are generating reference 3D atlases that will serve as benchmarks for clinical applications. As the review concludes, the integration of 3D multi-omics with artificial intelligence holds the promise of transforming cancer diagnosis, prognosis, and treatment selection by capturing the full spatial complexity of tumour biology.