arxiv:2603.16587

HistoAtlas: A Pan-Cancer Morphology Atlas Linking Histomics to Molecular Programs and Clinical Outcomes

Published on Mar 17

· Submitted by

Authors:

Abstract

HistoAtlas creates a comprehensive computational map linking histomic features from H&E slides to clinical outcomes and molecular profiles across multiple cancer types.

AI-generated summary

We present HistoAtlas, a pan-cancer computational atlas that extracts 38 interpretable histomic features from 6,745 diagnostic H&E slides across 21 TCGA cancer types and systematically links every feature to survival, gene expression, somatic mutations, and immune subtypes. All associations are covariate-adjusted, multiple-testing corrected, and classified into evidence-strength tiers. The atlas recovers known biology, from immune infiltration and prognosis to proliferation and kinase signaling, while uncovering compartment-specific immune signals and morphological subtypes with divergent outcomes. Every result is spatially traceable to tissue compartments and individual cells, statistically calibrated, and openly queryable. HistoAtlas enables systematic, large-scale biomarker discovery from routine H&E without specialized staining or sequencing. Data and an interactive web atlas are freely available at https://histoatlas.com .

View arXiv page View PDF Project page GitHub 2 Add to collection

Community

PABannier

Paper submitter about 7 hours ago

Hi!

HistoAtlas is a pan-cancer computational histopathology atlas that extracts 38 interpretable morphology features from 6,745 H&E diagnostic slides across 21 TCGA cancer types, then systematically links every feature to survival, gene expression, somatic mutations, pathways, and immune subtypes.

Key results

4.4 billion cells detected and classified into 9 morphological types using Owkin's HistoPLUS model
88,920 significant molecular correlations (FDR < 0.05) between morphology and transcriptomics/genomics
Intratumoral lymphocyte density is protective pan-cancer (HR = 0.84, p < 0.0001, n = 5,957), while stromal lymphocyte density has a weaker effect. Where immune cells sit matters more than how many there are.
10 unsupervised morphological clusters recover canonical biology (Wnt in CRC, estrogen response in BRCA/PRAD, immune rejection in THYM) without any molecular input
Two "immune-cold" clusters have opposite survival outcomes, suggesting binary hot/cold classification misses biologically distinct subtypes

What you can explore

The full interactive atlas is live at https://histoatlas.com: UMAP embeddings, Kaplan-Meier curves, molecular correlations, cluster characterization, and individual slide deep-dives. No login, no backend, everything loads instantly from pre-computed static JSON.

All associations are covariate-adjusted, BH-corrected, and classified into evidence tiers. We explicitly report what the atlas cannot detect (minimum detectable effect sizes) and which cancer types are out-of-distribution for the segmentation model.

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2603.16587 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2603.16587 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2603.16587 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.