Multimodal histopathologic models stratify hormone receptor-positive early breast cancer

Boehm, K. M.; El Nahhas, O. S. M.; Marra, A.; Selenica, P.; Wen, H. Y.; Weigelt, B.; Paul, E. D.; Cekan, P.; Erber, R.; Loeffler, C. M. L.; Guerini-Rocco, E.; Fusco, N.; Frascarelli, C.; Mane, E.; Munzone, E.; Dellapasqua, S.; Zagami, P.; Curigliano, G.; Razavi, P.; Reis-Filho, J. S.; Pareja, F.; Chandarlapaty, S.; Shah, S. P.; Kather, J. N.

doi:10.1101/2024.02.23.581806

Home/Archive/bioRxiv

bioRxiv · 10.1101/2024.02.23.581806

Multimodal histopathologic models stratify hormone receptor-positive early breast cancer

View original source ↗Export metadata ↓

Abstract

For patients with hormone receptor-positive, early breast cancer without HER2 amplification, multigene expression assays including Oncotype DX (R) recurrence score (RS) have been clinically validated to identify patients who stand to derive added benefit from adjuvant cytotoxic chemotherapy. However, cost and turnaround time have limited its global adoption despite recommendation by practice guidelines. We investigated if routinely available hematoxylin and eosin (H&E)-stained pathology slides could act as a surrogate triaging data substrate by predicting RS using machine learning methods. We trained and validated a multimodal transformer model, Orpheus, using 6,203 patients across three independent cohorts, taking both H&E images and their corresponding synoptic text reports as input. We showed accurate inference of recurrence score from whole-slide images (r = 0.63 (95% C.I. 0.58 - 0.68); n = 1,029), the raw text of their corresponding reports (r = 0.58 (95% C.I. 0.51 - 0.64); n = 972), and their combination (r = 0.68 (95% C.I. 0.64 - 0.73); n = 964) as measured by Pearsons correlation. To predict high-risk disease (RS>25), our model achieved an area under the receiver operating characteristic curve (AUROC) of 0.89 (95% C.I. 0.83 - 0.94), and area under the precision recall curve (AUPRC) of 0.64 (95% C.I. 0.60 - 0.82), compared to 0.49 (95% C.I. 0.36 - 0.64) for an existing nomogram based on clinical and pathologic features. Moreover, our model generalizes well to external international cohorts, effectively identifying recurrence risk (r = 0.61, p < 10-4, n = 452; r = 0.60, p < 10-4, n = 575) and high-risk status (AUROC = 0.80, p < 10-4, AUPRC = 0.68, p < 10-4, n = 452; AUROC = 0.83, p < 10-4, AUPRC = 0.73, p < 10-4, n = 575) from whole-slide images. Probing the biologic underpinnings of the model decisions uncovered tumor cell size heterogeneity, immune cell infiltration, a proliferative transcription program, and stromal fraction as correlates of higher-risk predictions. We conclude that at an operating point of 94.4% precision and 33.3% recall, this model could help increase global adoption and shorten lag between resection and adjuvant therapy.

Source connections

Published article ↗

Explore related subjects

pathology Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Boehm, K. M., El Nahhas, O. S. M., Marra, A., Selenica, P., Wen, H. Y., Weigelt, B., Paul, E. D., Cekan, P., Erber, R., Loeffler, C. M. L., Guerini-Rocco, E., Fusco, N., Frascarelli, C., Mane, E., Munzone, E., Dellapasqua, S., Zagami, P., Curigliano, G., Razavi, P., Reis-Filho, J. S., Pareja, F., Chandarlapaty, S., Shah, S. P., Kather, J. N.. 2024-02-28. Multimodal histopathologic models stratify hormone receptor-positive early breast cancer. https://doi.org/10.1101/2024.02.23.581806

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

NAE1-Dependent Protein Neddylation Preserves Endothelial Identity and Vascular Integrity

Background: Endothelial dysfunction is a central driver of cardiovascular and inflammatory diseases, yet the post-translational mechanisms that preserve endothelial homeostasis remain incompletely understood. Protein neddylation, the covalent conjugation of a ubiquitin-like modifier, regulates diverse cellular processes, yet its physiological role in the vascular endothelium remains unknown. This study investigated whether protein neddylation is required to preserve endothelial identity and vascular homeostasis. Methods: We generated tamoxifen-inducible endothelial-specific Nae1 knockout mice to inhibit neddylation and combined bulk RNA sequencing, single-cell and single-nucleus transcriptomics, quantitative proteomics, biochemical analyses, and gain- and loss-of-function approaches to define the role of endothelial neddylation in vascular homeostasis and inflammatory injury. Results: Endothelial-specific Nae1 deletion caused rapid mortality associated with vascular leakage, platelet accumulation, inflammation, and multi-organ injury. Multi-omics analyses demonstrated profound loss of endothelial identity, characterized by suppression of core endothelial programs and activation of inflammatory, procoagulant, and pyroptotic pathways. Single-cell analyses revealed progressive endothelial dysfunction culminating in depletion of the endothelial population and remodeling of the vascular niche. Mechanistically, endothelial neddylation deficiency activated gasdermin D (GSDMD)- and gasdermin E (GSDME)-dependent pyroptosis, whereas dual inhibition of GSDMD and GSDME markedly attenuated inflammatory transcriptomic remodeling, vascular injury, hepatocyte death, immune cell infiltration, and platelet accumulation. Translational analyses demonstrated reduced endothelial neddylation in experimental endotoxemia and decreased expression of neddylation pathway components in human atherosclerosis and COVID-19 datasets. Conversely, restoration of endothelial neddylation partially reversed inflammatory endothelial transcriptomic reprogramming in vivo. Conclusions: NAE1-dependent protein neddylation is an essential regulator of endothelial identity and vascular integrity. Loss of endothelial neddylation promotes gasdermin-dependent pyroptosis and thrombo-inflammatory vascular injury, whereas restoration of the neddylation pathway mitigates inflammatory endothelial dysfunction. These findings identify endothelial neddylation as a fundamental mechanism maintaining vascular homeostasis and a potential therapeutic target for cardiovascular and inflammatory diseases.

pathology↗

bioRxiv2025

Fasting-mimicking diet counteracts gut microbial dysbiosis in experimental Lynch syndrome

The development of colorectal cancer (CRC) is largely influenced by hereditary factors, with up to one-third of cases linked to genetic predisposition. In parallel, environmental factors such as diet and intestinal microbiota play a significant role. Lynch syndrome (LS), the most common form of hereditary CRC, is due to mutations in DNA mismatch repair genes. Diet interventions such as calorie restriction (CR) can modify the course of the disease, altering nutrient supply and promoting beneficial microbial populations. Fasting-mimicking diets (FMD) are plant-based CR regimens that showed promise in modulating the gut microbiota and suppressing CRC progression in pre-clinical ectopic cancer models. In this study, Villin-Cre/Msh2-floxed (VCM) mice, modelling LS, were subjected to periodic FMD cycles for 10 months. Although not impacting on macroscopic tumor development, FMD influenced animal weight in a sexually dimorphic manner. Moreover, shotgun metagenomic sequencing revealed that FMD mitigated the dysbiotic longitudinal changes associated with cancer onset, preserving beneficial species, such as Lactobacillus johnsonii, and reducing adverse species, such as Escherichia coli. Metabolic pathway analysis also showed significant differences, with FMD preventing the upregulation of pathways involved in amino acid and nucleotide synthesis, potentially promoting tumor growth. Overall, the findings suggest that periodic FMD may be adopted as an adjuvant therapy in LS management, counteracting gut microbiota alterations.

pathology↗

bioRxiv2025

Quantification of collagen and associated features from H&E-stained whole slide pathology images across cancer types using a physics-based deep learning model

BackgroundCollagen is the major component of the extracellular matrix (ECM). Collagen structural organization undergoes significant transformation during tumorigenesis. The visualization of collagen in histological tissue sections would aid in the study of tumor growth, encapsulation, and invasion. However, such visualization requires the use of special stains such as Picrosirius Red (PSR) or Massons Trichrome (MT), or more recently, second-harmonic generation imaging (SHG) in unstained tissue sections. However, PSR and MT both suffer from significant inter- (and intra-) lab stain variabilities, and SHG, while considered a ground truth by many, suffers from issues of system complexity/reliability, cost, and speed/throughput. These technical hurdles limit more widespread assessment of collagen in tissue samples. MethodsUsing high-contrast, high-throughput polarization imaging on PSR-stained slides to generate ground truth training polarization images, we developed a deep learning model (iQMAI) to infer the presence of collagen directly from hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) with high specificity. After iQMAI inference across WSIs, individual collagen fibers were extracted, and features describing overall collagen intensity and fiber morphology were computed. iQMAI pixel-and feature-wise outputs were compared to ground truth polarization imaging to assess model performance. The trained iQMAI model was deployed on H&E-stained WSI from the TCGA LUAD, LUSC, LIHC, and PAAD datasets for evaluation. iQMAI-derived collagen features were compared to tissue composition, gene expression, and overall survival. ResultsThe iQMAI model shows significant generalization across multiple indications. iQMAI collagen predictions were similar to polarization imaging measurements of the same sample, with a mean structural similarity index (SSIM) of 0.84 (95% CI 0.69-0.93), a mean patch-wise RMSE of 0.04 (95% CI 0.02-0.08), and a linear correlation (R2=0.93). Comparing features of the collagen fibers extracted from iQMAI vs. polarization images yielded similar linear correlations between computed fiber tortuosity, length, width, and relative angle. The relationship between collagen fiber density and fibroblast density was distinct in non-small cell lung cancer (LUAD and LUSC), hepatocellular carcinoma (LIHC), and pancreatic ductal adenocarcinoma (PAAD). In PAAD, fiber density and fiber width were both negatively associated with the LRRC-15 gene expression signature, and increased fiber width was associated with longer overall survival. ConclusionsiQMAI is a deep learning model that accurately predicts collagen from an H&E-stained WSI, allowing for spatially resolved quantification of collagen morphology and enabling investigation of the interplay between collagen and other TME components. We demonstrate an example of the utility of iQMAI-based collagen assessment in PAAD, where collagen features are associated with immunosuppressive cancer-associated fibroblasts and overall survival. Understanding the relationship between collagen, the tumor microenvironment composition, and disease progression may aid the development of effective immunotherapies in PAAD and other cancer types.

pathology↗