Back

Systematic mapping of human tissue microanatomy reveals age-associated remodeling and resilience

Buljan, I.; Bago-Horvath, Z.; Rendeiro, A. F.

2025-10-14 systems biology
10.1101/2025.10.13.682110 bioRxiv
Show abstract

Aging disrupts tissue structure at various scales, from cellular alterations to tissue and organ-level integrity. Microanatomical domains - recurrent cellular arrangements essential to organ-specific function, provide a highly physiologically relevant perspective on tissue homeostasis but are severely understudied in human aging. To address this gap, we developed H&E-UTAG, an unsupervised algorithm to detect microanatomical domains in whole slide histopathological images, which enables large-scale, label-free analysis of human microanatomy. Applying it to 24,945 whole-slide images from 40 human tissues of 983 individuals aged 20 to 70 years old, we identified 218 recurrent microanatomical domains categorized into 74 types across tissues. Domain types varied widely in tissue specificity, with 16% shared across 3 or more tissues and 69% restricted to a specific tissue. Age emerged as the dominant factor in influencing domain abundance, with 28% of domains changing significantly over the adult lifespan. By integrating tissue-level pathology annotations, we distinguished structural changes associated with healthy aging from those linked to subclinical disease, revealing that these processes often remodel distinct tissue compartments. Finally, mapping higher-order networks of domain-domain interactions uncovered age-associated reorganization of organ architecture, while a core framework of interactions remained resilient with age. Our novel analytical framework reveals fundamental principles of tissue organization and how they are restructured across the human lifespan, offering new insights into aging biology and tissue architecture in health and in the path to age-associated diseases. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=116 SRC="FIGDIR/small/682110v1_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@1d05840org.highwire.dtl.DTLVardef@151e211org.highwire.dtl.DTLVardef@d0de41org.highwire.dtl.DTLVardef@afb744_HPS_FORMAT_FIGEXP M_FIG C_FIG

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.