Aging is not just about molecular changes; it also involves a remodeling of tissue architecture that underlies age-related diseases. Existing molecular atlases, such as the Human Cell Atlas and HuBMAP, are limited in sample size, adult age coverage, and systematic organ comparison. This work fills that gap by using the GTEx population histopathology archive, which includes slides from 16 organs from 980 donors aged 20 to 70. This makes it possible for the first time to quantitatively characterize the cellular and architectural landscape of aging at the scale of billions of cells. The work stands somewhat apart: there is almost nothing else quite like it nearby, but thematically it is close to research on tissue evolution, horizontal gene transfer, and cellular organization.
The researchers used 14,788 whole-slide histopathology images (WSIs) from the GTEx cohort. Cells were segmented and classified with CellViT, yielding about 3.5 billion cells of four types: epithelial, immune, connective, and neoplastic. For finer phenotyping, they combined nuclear features (DINOv2), cell morphology (CellViT), and text-image embeddings of the surrounding tissue (PLIP). Epithelial cells were divided into nine subtypes using Leiden clustering and manual annotation. Blood vessels were identified by classifying endothelial cells and building spatial proximity graphs. Functional tissue units (FTUs) were detected through spatial cell graphs and community clustering. Statistical models (linear regression, GLM) were used to assess age-related changes with adjustment for sex and ischemia time.
With age, cell density declined in 8 of 16 organs, most strongly in the ovary (a 47% decrease), lung (35%), and liver (27%). Epithelial cells were hit hardest. A more detailed analysis of epithelial subtypes revealed organ-specific trajectories: in the ovary, granulosa cells almost completely disappeared (a 45% decrease per decade), in the lung, alveolar cells were lost (25-30% per decade), while in the pancreas, the number of glandular columnar cells actually increased (+17% per decade). Cell morphology also changed: in the uterus, all three major populations underwent significant remodeling, and in the thyroid, despite stable density, the greatest remodeling was observed (25-28% of features). The vascular network of the gastrointestinal tract thinned: in the small intestine, arteries lost 20% of their density per decade, while veins and lymphatic vessels lost 15%. An analysis of 236 functional tissue units showed that 62 of them changed significantly with age: dense, specialized structures were replaced by sparse and heterogeneous ones enriched in stromal and immune cells. The key result: cellular composition explained only 7% of the variation in changes in FTU abundance, which points to an independent architectural layer of aging.
The work demonstrates that architectural aging of tissues is largely not reducible to changes in cellular composition. This opens a new layer of aging biology that can only be studied with spatial methods such as histopathology. Unlike neighboring studies that focus on molecular or cellular aspects, this research is the first to systematically link the morphology of individual cells to functional tissue units at the scale of a population and the entire adult age range. This provides a quantitative benchmark against which future longitudinal and interventional studies can be compared.
The topic could develop toward integration with spatial transcriptomics for molecular confirmation of morphological cell types. Applying the developed approach to other cohorts and diseases, such as cancer or chronic inflammatory conditions, is also promising. Longitudinal studies will help establish causal links between architectural changes and functional decline of organs.
The results will influence gerontology, pathology, and computational biology by providing tools for quantifying tissue aging. They may also be useful in the clinical diagnosis of age-related diseases and in developing interventions aimed at preserving tissue architecture.
Next steps include validating morphological cell types with spatial transcriptomics and extending the analysis to longitudinal cohorts. It is also important to investigate the mechanisms underlying the decoupling of architectural changes from cellular composition.
The work directly addresses an unsolved problem in aging biology: how molecular damage translates into functional decline of organs. It shows that architectural remodeling is an independent and quantitatively measurable aspect of aging that is not captured by dissociative methods. This raises the question of the mechanisms that maintain tissue organization and why they break down with age.
Cell density in the spleen is more than 8 times higher than in the skin: over 8,000 cells per square millimeter versus less than 1,000.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.