SEARCH · bioRxiv Science

Results for “Pathology”

Search indexed bioRxiv preprints in genomics, neuroscience, cell biology and bioinformatics. Read source abstracts and check manuscript versions; preprints are not peer reviewed.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 505 records · Page 28Linked to original sources

bioRxiv2022

Density-Dependent Color Scanning Electron Microscopy (DDC-SEM) for calcified tissue and pathological calcification.

Scanning electron microscopy (SEM) is widely used for materials characterization. It has also been successfully applied to the imaging of biological samples, providing invaluable insights into the topography, morphology and composition of biological structures, including pathological minerals, in diseases affecting cardiovascular, kidney and ocular tissues. Here we provide a comprehensive and detailed guide on how to use colored SEM to aid the visualization and characterization of pathological calcification, and identify the effects of different sample preparation protocols for the visualisation of these minerals.

pathology↗

bioRxiv2022

Pathology of African Swine Fever in wild boar naturally infected with German virus variants

In 2020, African swine fever (ASF) was first notified in German wild boar, reaching a case number of about 4200 to date. Upon experimental infection, pathology is well-documented, however, data on field infections are very scarce in domestic pigs and not available from wild boar, respectively. Although ASF viral genome is considered exceptionally stable, a total of five lineages with 10 distinct virus variants of genotype II have emerged in Eastern Germany. To investigate the pathology in naturally infected wild boar and to evaluate virus variants II, III and IV for their virulence, wild boar carcasses were obtained from three different outbreak areas. The cadavers underwent virological and pathomorphological investigation. Regardless of the virus variant all wild boar revealed characteristic lesions of highest severity indicative for ASF. However, wild boar infected with variant IV from Spree-Nei{beta}e (SN) district showed lower viral genome loads and a lower total viral antigen score, but simultaneously revealed more chronic lesions. Our findings indicate a protracted course of the disease at least after infection with variant IV, but need confirmation under standardized experimental conditions. There is a strong need to monitor differences in the virulence among variants to identify potential attenuation that might complicate diagnosis.

pathology↗

bioRxiv2023

Age-linked lung pathology is reduced by immunotherapeutic targeting of isoDGR protein damage

Advancing age is the primary risk factor for pulmonary diseases. Our investigation revealed an 8-fold increase in aging induced isoDGR-damaged proteins in lung tissue from human pulmonary fibrosis patients compared to healthy tissues, accompanied by elevated frequencies of CD68+/CD11b+ macrophages, indicating lung tissue is susceptible to time-dependent accumulation of isoDGR-proteins. To elucidate the mechanisms through which isoDGR-proteins may exacerbate aging lung disorders for potential therapeutic targeting, we assessed the functional role of this isoDGR-motif in naturally-aged mice and mice lacking the corresponding isoDGR repair enzyme (Pcmt1-/-). IsoDGR-protein accumulation in mouse lung tissue and blood vessels correlated with chronic low-grade inflammation, pulmonary edema, and hypoxemia. IsoDGR accretion induced mitochondrial and ribosomal dysfunctions, cellular senescence, and apoptosis, contributing to progressive lung damage over time. Treatment with anti-isoDGR antibodies suppressed TLR pathway activity, mitigated cytokine-driven inflammation, restored mtDNA expression, and significantly reduced lung pathology in-vivo. Similarly, exposure of lung endothelial cells to isoDGR-modified fibronectin impaired oxygen consumption, increased reactive oxygen species levels, and disrupted acidification, but these effects were efficiently reversed by target-specific antibody therapy. Collectively, our findings underscore the significant contribution of isoDGR-damaged proteins to age-linked lung pathology. IsoDGR-specific therapy emerges as a promising treatment approach for pulmonary disorders in older patients.

pathology↗

bioRxiv2024

Digital pathology assessment of kidney glomerular filtration barrier ultrastructure in an animal model of podocytopathy

BackgroundTransmission electron microscopy (TEM) images can visualize kidney glomerular filtration barrier ultrastructure, including the glomerular basement membrane (GBM) and podocyte foot processes (PFP). Podocytopathy is associated with glomerular filtration barrier morphological changes observed experimentally and clinically by measuring GBM or PFP width. However, these measurements are currently performed manually. This limits research on podocytopathy disease mechanisms and therapeutics due to labor intensiveness and inter-operator variability. MethodsWe developed a deep learning-based digital pathology computational method to measure GBM and PFP width in TEM images from the kidneys of Integrin-Linked Kinase (ILK) podocyte-specific conditional knockout (cKO) mouse, an animal model of podocytopathy, compared to wild-type (WT) control mouse. We obtained TEM images from WT and ILK cKO littermate mice at 4 weeks old. Our automated method was composed of two stages: a U-Net model for GBM segmentation, followed by an image processing algorithm for GBM and PFP width measurement. We evaluated its performance with a 4-fold cross-validation study on WT and ILK cKO mouse kidney pairs. ResultsMean (95% confidence interval) GBM segmentation accuracy, calculated as Jaccard index, was 0.73 (0.70-0.76) for WT and 0.85 (0.83-0.87) for ILK cKO TEM images. Automated and manual GBM width measurements were similar for both WT (p=0.49) and ILK cKO (p=0.06) specimens. While automated and manual PFP width measurements were similar for WT (p=0.89), they differed for ILK cKO (p<0.05) specimens. WT and ILK cKO specimens were morphologically distinguishable by manual GBM (p<0.05) and PFP (p<0.05) width measurements. This phenotypic difference was reflected in the automated GBM (p<0.05) more than PFP (p=0.06) widths. ConclusionsThese results suggest that certain automated measurements enabled via deep learning-based digital pathology tools could distinguish healthy kidneys from those with podocytopathy. Our proposed method provides high-throughput, objective morphological analysis and could facilitate podocytopathy research and translate into clinical diagnosis. Key pointsO_LIWe leveraged U-Net architecture in an algorithm to measure the widths of glomerular basement membrane and podocyte foot processes. C_LIO_LIDeep learning-based automated measurement of glomerular filtration barrier morphology has promise in podocytopathy research and diagnosis. C_LI

pathology↗

bioRxiv2024

Exercise and disease state influence the beneficial effects of Fn14-depletion on survival and muscle pathology in the SOD1G93A amyotrophic lateral sclerosis (ALS) mouse model

BackgroundAmyotrophic lateral sclerosis (ALS) is a devastating and incurable neurodegenerative disease. Accumulating evidence strongly suggests that intrinsic muscle defects exist and contribute to disease progression, including imbalances in whole-body metabolic homeostasis. We have previously reported that tumour necrosis factor (TNF)-like weak inducer of apoptosis (TWEAK) and fibroblast growth factor inducible 14 (Fn14) are significantly upregulated in skeletal muscle of the SOD1G93A ALS mouse model. While antagonising TWEAK did not impact survival, we did observe positive effects in skeletal muscle. Given that Fn14 has been proposed as the main effector of the TWEAK/Fn14 activity and that Fn14 can act independently from TWEAK in muscle, we suggest that manipulating Fn14 instead of TWEAK in the SOD1G93A ALS mice could lead to differential and potentially improved benefits. MethodsWe thus investigated the contribution of Fn14 to disease phenotypes in the SOD1G93A ALS mice. To do so, Fn14 knockout mice (Fn14-/-) were crossed onto the SOD1G93A background to generate SOD1G93A;Fn14-/- mice. Investigations were performed on both unexercised and exercised (rotarod and/or grid test) animals (wild type (WT), Fn14-/-, SOD1G93A and SOD1G93A;Fn14-/-). ResultsHere, we firstly confirm that the TWEAK/Fn14 pathway is dysregulated in skeletal muscle of SOD1G93A mice. We then show that Fn14-depleted SOD1G93A mice display an increased lifespan and decreased muscle pathology, without an impact on motor function, and that this is dependent on exposure to exercise. Indeed, we observe that endurance (rotarod) and resistance (grid test) exercises influence the positive effects of Fn14 deletion on survival and muscle phenotypes in SOD1G93A mice, which may be further influenced by genotype and disease state. ConclusionsOur study provides further insights on the different roles of the TWEAK/Fn14 pathway in pathological skeletal muscle and how they can be influenced by age, disease and metabolic state. This is particularly relevant in the ALS field, where combinatorial therapies that include exercise regimens are currently being explored. As such, a better understanding and consideration of the interactions between treatments, muscle metabolism and exercise will be of importance in future studies.

pathology↗

bioRxiv2024

Endomucin knockout leads to delayed retinal vascular development and reduced ocular pathological neovascularization

Endomucin (EMCN), an endothelial-specific glycocalyx component highly expressed in capillary and venous endothelium, plays a critical role in regulating VEGF receptor 2 (VEGFR2) endocytosis and downstream VEGF signaling. Using the first global EMCN knockout mouse model, we investigated the effects of EMCN deficiency on retinal vascularization during development and pathological angiogenesis. We found relatively high expression of EMCN in choroidal capillaries and retinal vasculature. Emcn-/- mice exhibited delayed retinal vascularization at postnatal day 5, with fewer tip cells and reduced vessel density. Ultrastructural examination revealed disrupted and reduced fenestrations in choroidal capillary endothelium. In an oxygen-induced retinopathy model, while Emcn-/- mice showed no significant difference in avascular area compared to Emcn+/+ mice at postnatal day 12, there was a significant reduction in neovascular tufts in Emcn-/- mice at postnatal day 17. Similarly, in a laser-induced choroidal neovascularization model, Emcn-/- mice showed a significant reduction in vascular leakage and lesion size. These findings suggest that EMCN plays a critical role in both vascular development and pathological neovascularization, highlighting its potential as a target for anti-angiogenic therapies.

pathology↗

bioRxiv2025

Gut bacterial metabolite imidazole propionate potentiates Alzheimer's disease pathology

The gut microbiome modulates metabolic and neurovascular processes implicated in Alzheimers disease and related dementias (ADRD), but the underlying mechanisms remain unclear. Here, we identify the bacterial metabolite imidazole propionate (ImP) as a modifier of ADRD pathology. In a cohort of 1,196 cognitively unimpaired adults, higher plasma ImP levels were associated with lower preclinical cognitive scores and biomarkers of ADRD, both cross-sectionally and longitudinally. Fecal metagenomic analysis linked putative ImP producers to ADRD phenotypes. Genome-wide integrative analysis revealed a locus on chromosome 12 associated with both plasma ImP levels and AD risk in humans, supporting a host genetic contribution to ImP regulation and a causal role of this metabolite in AD. In mice, chronic ImP administration exacerbated AD-like pathology. Mechanistically, ImP impaired brain endothelial barrier and promoted tau hyperphosphorylation in primary neurons, an effect blocked by glycogen synthase kinase-3{beta} inhibition. Together, our study links ImP to hallmarks of neurodegeneration and suggest that targeting ImP may represent a potential strategy to modify ADRD risk.

pathology↗

bioRxiv2025

Understanding Foundation Models in Digital Pathology: Performance, Trade-offs, and Model-Selection Recommendations

The rapid proliferation of digital pathology foundation models (FMs), spanning widely in architectural scales and pre-training datasets, poses a significant challenge in selecting the optimal model for a specific clinical application. To systematically evaluate these performance trade-offs, we conducted a comprehensive benchmark of five FMs, stratified from small to huge scales, across a diverse suite of whole slide image (WSI) and region of interest (ROI) tasks. Our findings demonstrate that model superiority is strongly task-specific, challenging the assumption that a larger scale universally offers an advantage. For instance, while the huge-scale Virchow2 model excelled at WSI metastasis detection, the small-scale Lunit model was superior for fine-grained lung subtype classification. This trend was even more pronounced in survival analysis, where smaller models outperformed their massive-scale counterparts, a phenomenon we hypothesize is linked to potential information bottlenecks within downstream aggregation models. Notably, the base-scale Kaiko model consistently provided a compelling balance, delivering competitive accuracy, superior stability in prognostic tasks, and higher computational efficiency. Our analysis suggests that the optimal FM is not necessarily the largest, but one whose scale, data composition, and training strategy are best aligned with the specific task. This work offers a practical, evidence-based framework for balancing performance, stability, and real-world deployment costs in computational pathology.

pathology↗

bioRxiv2025

AnnotateAnyCell: Open-Source AI Framework for Efficient Annotation in Digital Pathology

AO_SCPLOWBSTRACTC_SCPLOWManual annotation of histopathological whole slide images remains a critical bottleneck for computational pathology and clinical AI deployment, requiring prohibitive expert time at scale. Here we present an open-source semi-supervised framework combining active contrastive learning with iterative human-in-the-loop feedback for efficient cellular annotation and classification. The pipeline integrates Cellpose segmentation, UMAP-based latent space visualization, and contrastive learning with pseudolabel propagation, evaluated on five whole slide images of canine invasive urothelial carcinoma across low, intermediate, and high histological grades at 40x magnification. Latent space clustering-guided annotation required 47 minutes compared to 63 minutes for sequential annotation, a 25% reduction (95% CI 18-32%). Classification accuracy reached 96.3% {+/-} 1.2% for mitotic figures and 98.3% {+/-} 1.4% for nucleoli using 1,075 labeled samples, with nucleoli classification achieving 95.5% {+/-} 1.5% accuracy from only 215 samples. Inter-annotator agreement was high for chromatin ({kappa} = 1.00) and nucleoli ({kappa} = 0.95) but moderate for mitotic figures ({kappa} = 0.58) and nuclear shape ({kappa} = 0.36), reflecting intrinsic morphological ambiguity in these categories. This framework substantially reduces annotation burden while achieving expert-level accuracy for well-defined morphological features, providing a scalable path toward AI-assisted diagnostics in resource-constrained pathology settings.

pathology↗

bioRxiv2018

A conserved role of the insulin-like signaling pathway in uric acid pathologies revealed in Drosophila melanogaster

Elevated uric acid (UA) is a key factor for disorders, including gout or kidney stones and result from abrogated expression of Urate Oxidase (Uro) and diet. To understand the genetic pathways influencing UA metabolism we established a Drosophila melanogaster model with elevated UA using Uro knockdown. Reduced Uro expression resulted in the accumulation of UA concretions and diet-dependent shortening of lifespan. Inhibition of insulin-like signaling (ILS) pathway genes reduced UA and concretion load. In humans, SNPs in the ILS genes AKT2 and FOXO3 were associated with UA levels or gout, supporting a conserved role for ILS in modulating UA metabolism. Downstream of the ILS pathway UA pathogenicity was mediated partly by NADPH Oxidase, whose inhibition attenuated the reduced lifespan and concretion accumulation. Thus, genes in the ILS pathway represent potential therapeutic targets for treating UA associated pathologies, including gout and kidney stones.\n\nHighlightsO_LIIn Drosophila high uric acid (UA) levels shorten lifespan and cause UA aggregation\nC_LIO_LIConserved in flies and humans, the ILS pathway associates with UA pathologies\nC_LIO_LIFoxO dampens concretion formation by reducing UA levels and ROS formation\nC_LIO_LIInhibition of NOX alleviates the lifespan attenuation and UA aggregation\nC_LI

pathology↗

bioRxiv2018

Green autofluorescence intensity of skin and fingernails: A novel biomarker for non-invasive evaluation of pathological state of blood vessels

Stroke and myocardial infarction (MI) are two leading causes of death around the world. It is of great significance to establish novel and non-invasive approaches for evaluating pathological state of blood vessels, so that early interventions may be carried out to prevent incidence of stroke or MI. Our recent studies have suggested that altered Pattern of Autofluorescence (AF) of skin and fingernails are novel biomarkers of acute ischemic stroke (AIS) and MI. In particular, our studies have shown characteristic increases in the green AF intensity of the fingernails and certain regions of the skin of AIS patients and MI patients. By determining the skins green AF of the Healthy Group, the Low-Risk Group for Developing AIS, and the High-Risk Group for Developing AIS, our current study has indicated that the green AF intensity in the fingernails and certain regions of the skin, including the right and left Dorsal Index Fingers, Ventroforefingers, Dorsal Antebrachium and Index Fingernails as well as right Dorsal Centremetacarpus, is highly correlated with the risk to develop AIS. There is also evidence suggesting that increased oxidative stress may account for the increased AF intensity in the Low-Risk Group for Developing AIS and the High-Risk Group for Developing AIS. These findings have suggested that the green AF intensity of the fingernails and certain regions of the skin is a novel biomarker for non-invasive evaluation of the pathological state of blood vessels and the risk for developing AIS or MI.

pathology↗

bioRxiv2026

Neural Networks as Entropic Systems: Applications in Digital Pathology

Deep learning systems in digital pathology are widely regarded as opaque, limiting clinical trust and interpretability. We present a framework for empirically characterizing training-time learning dynamics in neural networks by directly measuring activation structure, weight evolution, and spectral organization during optimization. Using TCGA-BRCA whole-slide images with replication-timing-derived methylation proxies as regression targets, we trained a Vision Transformer and tracked its intra-epoch behavior across 20 epochs. We observed reproducible structural signatures during training. Correlated groups of neurons formed stable activation modules whose modularity increased as training progressed, accompanied by a reduction in representation entropy of up to 60%. Weight trajectories exhibited bounded diffusion with progressively reduced variance, consistent with a damped stochastic process, and converged toward a stable stationary regime in later epochs. In image space, model attention systematically shifted from collagen-rich stromal regions in early epochs to basophilic, proliferative nuclear regions in later epochs, aligning with known histologic correlates of replicative stress. These findings demonstrate that neural networks develop predictable, quantifiable internal structure during training that can be directly visualized and measured. Framing learning dynamics in terms of entropy, modular organization, and stochastic stabilization provides a practical, mechanistic lens for interpreting how pathology AI models acquire biologically meaningful representations.

pathology↗

bioRxiv2026

NAD boosting mediated by CD38 inhibition drives reversal of a pathological vicious cycle of intracrine activity and inflammation in eyelid meibomian gland dysfunction

Vicious cycles--reciprocal cause-and-effect loops that progressively intensify dysfunction--are implicated in many diseases. Here, we examined whether "intracrine" steroidogenesis, a tissue-local mode of hormone production, participates in a pathological vicious cycle in eyelid meibomian gland dysfunction (MGD), the most common cause of evaporative dry eye disease with incidence increasing with age. We show that loss of meibomian intracrine androgen production triggers inflammatory remodeling of the eyelid tarsal plate. Conversely, inflammation suppresses nicotinamide adenine dinucleotide (NAD)-dependent 3{beta}-hydroxysteroid dehydrogenase activity in the meibomian gland, thereby further weakening intracrine steroidogenesis. Mechanistically, we identified inflammation-associated accumulation of the NAD-degrading enzyme CD38 as a key down-regulator for the NAD-dependent intracrine activity. Accordingly, pharmacological inhibition of CD38 (using 78c) in aged mice, restores meibomian cellular NAD levels, rescues intracrine activity, suppresses inflammatory signatures, and promotes recovery of gland size. Thus, our data suggest that boosting NAD availability not only ameliorates meibomian gland dysfunction but also induces a reciprocal "virtuous" cycle, in which suppression of inflammation and restoration of intracrine activity mutually reinforce each other. These findings constitute a previously unknown example in which the reversal of a pathological vicious cycle into a beneficial virtuous cycle underlies the therapeutic effects of NAD boosting and may help combat age-associated meibomian gland disorder.

pathology↗

bioRxiv2026

Proteomics-Enhanced AI-Digital Pathology in Metastatic Mucinous Colorectal Carcinoma: A Case Report

Mucinous colorectal carcinoma (CRC) is a distinct histomorphological subtype characterized by abundant extracellular mucin that may promote immune evasion and chemoresistance. We describe a metastatic mucinous CRC case integrating digital pathology and proteomics to investigate disease progression and therapy resistance. Formalin-fixed paraffin-embedded samples from the primary tumor, peritoneal metastasis, and hepatoduodenal ligament metastasis of a 56-year-old patient were analyzed. Whole-slide imaging with QuPath-based AI enabled detailed histological annotation, while data-independent acquisition mass spectrometry identified over 6,000 proteins. Digital pathology revealed extensive mucin pools, architectural evolution from heterogeneous glandular patterns in the primary tumor to cribriform morphology in advanced metastases, and immune cell exclusion from mucin-rich regions. Proteomics revealed metabolic reprogramming, suppressed antigen presentation, and stage-specific activation of inflammatory, angiogenic, EMT, and PI3K/AKT/mTOR-MYC signaling pathways, consistent with proliferative and therapy-resistant phenotypes.Integration of AI-assisted histopathology with spatial proteomics highlighted the mucin barrier as a key mediator of immune evasion and chemoresistance. These findings support a personalized therapeutic framework targeting mucin-associated mechanisms alongside pathway-directed inhibitors, suggesting that spatial multi-omics may guide precision management strategies for aggressive mucinous colorectal cancer.

pathology↗

bioRxiv2026

First report on Chicken Pediculosis in Modern Battery-Cage Layer Farms in Bangladesh: Behavioral, Pathological and Production Performance Impacts

The recent shift in Bangladesh from tradition backyard rearing system to modern commercial layer farming has made the birds immune to infectious diseases. Pediculosis, however, continues to pose a challenge in modern production system due to its invasive nature, often going unnoticed and neglected as it is typically non-lethal, yet capable of causing significant production losses. Lice infestation is a persistent threat in poultry production; however, its implications in battery-caged commercial layer hens in Bangladesh remain insufficiently characterized. The study aimed to identify the causative louse species and evaluate its associations with clinical pathology, hematological alteration and productive performance in 30 white-feathered (15 infested + 15 non-infested) and 30 brown-feathered (15 infested + 15 non-infested) laying birds from two commercial farms in Tangail. Morphological characterization confirmed the parasite as Menacanthus stamineus, distinguished by a dorsoventrally flattened body, parabolicallly rounded head wider than long, concealed club-shaped antennae, an oblong-oval abdomen with fine setae and three pairs of short legs each bearing paired claws. Infested birds exhibited consistent clinical pathology, including pale combs, petechial hemorrhages around the vent, severe feather damage with alopecic and exudative areas and incidence of irregular and broken-shelled eggs. Production performance analysis revealed significant reduction in hen-day egg production, egg weight, and feed intake, accompanied by significantly increased feed conversion ratios. Hematological evaluation demonstrated significantly reduced hemoglobin concentration, hematocrit and erythrocyte counts in infested hens, indicating mild anemia and compromised oxygen-carrying capacity. Collectively, pediculosis was strongly associated with lice-induced self-inflicted injury and cannibalism, systemic physiological stress, impaired erythropoiesis, reduced production efficiency and compromised welfare in caged laying hens. To best of our knowledge, it was the first integrative reports from Bangladesh documenting M. stramineus infestation in battery-caged commercial layer system with concurrent evidence of hematological disruption and measurable productivity losses, underscoring its epidemiological and economic significance and urgent need for targeted, evidence-based ectoparasite control strategies.

pathology↗

bioRxiv2026

Spatial Glyco-Codes Define Human Liver Pathology and Progression

Glycosylation is a fundamental process regulating cellular function, tissue organization, and disease progression. However, comprehensive glycan profiling at single-cell spatial resolution remains largely inaccessible, particularly in clinical archival tissues. Here we develop spatial-GPT, a multimodal platform for simultaneous profiling of glycans, proteins, and/or transcripts in archival formalin-fixed paraffin-embedded (FFPE) tissues. Using a panel of 30 DNA-encoded lectins recognizing major mammalian glycan motifs and structural classes, sequencing-based spatial-GPT (DBiT-GPT) mapped the spatial glycome, proteome, and transcriptome across 16 human liver specimens encompassing steatosis, fibrosis, cirrhosis, and hepatocellular carcinoma (HCC), leading to identification of spatial glyco-codes - combinatorial glycan states associated with distinct cellular identities, tissue features, and pathological processes. Unexpectedly, glyco-codes alone were sufficient to resolve major cell types, disease states, and HCC subtypes, revealing a previously unappreciated level of biological information encoded within the tissue glycome. Spatial glycomics uncovered tumor-like glyco-codes in premalignant regions, suggesting that glycan reprogramming may precede overt malignant transformation. Using imaging-based single-cell spatial glycan-protein profiling (CODEX-GP), we track glyco-codes across the whole-tissue architecture of 3 representative HCC samples. We further examined the glyco-codes across more than 300 patient specimens and quantified cell-type- and disease-specific glyco-codes as well as glycan-defined immune-evasion, T-cell-exhaustion, and steato-fibrotic niches. Together, these findings establish spatial glyco-codes as a previously unrecognized layer of tissue organization that encodes cellular identity, tissue function, and disease progression. The ability of glyco-codes to distinguish major liver pathologies across independent patient cohorts further highlights their potential as a new class of molecular histopathology biomarkers.

pathology↗

bioRxiv2026

Multicenter self-supervised computational pathology identifies prognostic histomorphological phenotypes in colorectal cancer

H&E whole-slide images capture prognostic information encoded in tumor morphology and the surrounding microenvironment, but these signals remain difficult to extract and interpret at scale. Here, we developed a self-supervised computational pathology framework to predict disease-free survival in colorectal cancer and link model-derived risk to interpretable histomorphology and spatial tumor biology. Using a multicenter developmental cohort spanning colorectal adenomas and invasive colorectal cancer, we trained HPL-PanColon, a self-supervised representation model, to extract tile-level embeddings and identify recurrent histomorphological phenotype clusters across the adenoma-carcinoma spectrum. Compared with general-purpose pathology foundation models, HPL-PanColon yielded representations with reduced institution- and dataset-specific batch effects. We then applied HPL-PanColon to a global survival cohort of 1,024 colorectal cancer patients in a leave-one-institution-out framework, using tile embeddings to train an attention-based survival model and derive the Colon Histomorphology Prognostic Score (CHiPS). CHiPS stratified patients by disease-free survival and provided complementary prognostic information to a UICC TNM-informed clinicopathological model, increasing the c-index from 0.683 to 0.706. Integrating model attention with phenotype assignments traced CHiPS-associated risk to pathologist-recognizable tissue patterns, with high-risk regions enriched for desmoplastic, stromal, and fibroinflammatory morphologies and low-risk regions reflecting tumor-rich epithelial glandular patterns. Spatial transcriptomic analysis further linked high-risk morphologies to fibroblastic, perivascular, myofibroblastic, and immune-reactive tumor microenvironment programs, while low-risk morphologies mapped to epithelial and tumor-enriched regions. These findings establish a scalable framework for interpretable histology-based prognosis and spatial biological discovery in colorectal cancer.

pathology↗

bioRxiv2026

Explainable machine learning relates histological to genomic pathology

Background & AimsHaematoxylin and eosin (H&E) staining remains the diagnostic gold standard for solid cancers, including hepatocellular carcinoma, and is increasingly complemented by genomic profiling for precision medicine. Inferring genomic alterations directly from H&E images could streamline testing, but heterogeneity and biases in human training data limit interpretation of genotype-phenotype associations. Here, we aimed to relate histologic to genomic pathology to provide biological explainability for mutation prediction models and assess the impact of germline variation on model performance. MethodsWe analysed 597 murine liver tumours with matched whole-genome sequencing and histopathology (163,835 image tiles; 22.9 million nuclei). Our controlled in vivo design accounted for germline variation, biological sex, and causal mutagen (N-diethylnitrosamine), removing confounding factors present in human cohorts. We trained and evaluated deep learning and supervised machine learning models to predict germline variation and cancer driver alterations from H&E. ResultsModelling accurately predicted germline and somatic alterations from histology, at both locus-specific and genome-wide scales. Quantitative image analysis revealed an unexpected association between Egfr driver mutations and hepatic steatosis, linking genotype to an interpretable morphological phenotype. While model performance declined when applied to tumours from unrepresented genetic backgrounds, this limitation was biologically informative, revealing strain-dependent differences in tumour evolution, notably the prevalence of whole-genome duplication. ConclusionsMachine learning integration of histological and genomic pathology enables accurate, interpretable inference of genetic alterations from H&E, potentially reducing reliance on costly ancillary molecular assays. Our predictions are supported by human-interpretable biological features, addressing concerns around "black-box" technologies. However, caution is required when applying such methods to samples with a genetic background that, even if closely related, is beyond the genetic horizon of training data.

pathology↗