Two crypt phenotypes (ACP): crypt rings in tandem (CRT) and crypts with lateral budding's (CLB's) were recently found interpolated amidst generally accepted crypt phenotypes (GACP, i.e., tubular, tubulovillous, villous and serrated). The frequency of cases with UC-PDG, with CRT and with CLB's in GACP and in UC-associated sporadic adenomas (UC-aspa) was recorded. The possibility that the presence of ACP could increase histological discrimination between UC-PDG and UC-aspa was also explored. Two hundred digitalized biopsy-cases were investigated: 100 with UC-PDG and 100 with UC-aspa. Six GACP and two types of cellular dysplasia: low-grade (LGD) and high-grade dysplasia (HGD) were found. CRT and CLB's were present in 66% of the 200 cases. A significantly higher number of cases with villous phenotype wihout or with HGD, with CLB's phenotype without or with HGD, and with tubulovillous phenotype with HGD were found in UC-PDG than in UC-asta. Several authors found no significant difference in the histological characteristics between UC-protruding dysplastic lesions and US-aspa. This work reveals, however, that the different number of cases with the above-mentioned histological parameters present in each group permitted to discriminate between UC-PDG and UC-asta.

Endometrial cancer (EC) is classified into four molecular subtypes with distinct prognosis and treatment implications. Despite this well-established molecular classification, morpho-molecular correlations remain understudied. Artificial intelligence (AI) enables biomarker prediction from H&E-stained whole-slide images (WSIs). However, real-world validation for EC molecular subtyping is lacking. In this study, we evaluated image-based molecular subtyping in a real-world cohort derived from routine diagnostic cases with heterogeneous image quality. We benchmarked the feature extraction with the CTransPath and UNI foundation models, demonstrating robust results across different scanner hardware and quantified performance variation by additional stain normalization. UNI-based features achieved a mean AUROC of 0.646 for POLEmut (n = 16), 0.700 for MMRd_MSI (n = 79), 0.684 for NSMP (n = 176), and 0.844 for p53abn (n = 18) on external real-world data. We provided human interpretations of subtype-specific morphological features. Our findings may thus lay the foundations of a more reliable framework for personalized treatment stratification based on morphologically informed molecular subtyping.

Abstract Spatial transcriptomics links gene expression to tissue architecture, providing a mechanistic view of cellular organization. Yet existing datasets cover few donors and miss the complexity of human disease. Experimental costs remain prohibitive, and large-scale profiling is impractically slow for population-level studies. Accurate computational methods are urgently needed. Predicting gene expression from standard histology, however, remains an open problem, as current approaches transfer poorly to unseen cohorts and diseases. Here, we present Phoenix, a (latent) flow matching generative model that infers pan-cancer spatially resolved single-cell gene expression with high accuracy. Phoenix analyzes treatment response in silico: Applied to 763 head and neck cancer patients, it identified three new spatial biomarkers that we validated across two cancers (breast cancer, n = 84; ovarian cancer, n = 157) and treatment regimens (platinum, trastuzumab). Phoenix generalizes beyond carcinomas: In a large sarcoma cohort (802 tissue microarray cores), it accurately predicted cell-type-specific signatures in held-out samples and captured chemotherapy-induced immune remodeling. Phoenix also extends across species: In a mouse model, it accurately predicted the expression of pancreatic cancer lineage markers and the mutant mKras^G12D allele in silico. In total, we evaluated Phoenix on over 10,000 patients. Our results establish virtual spatial transcriptomics as a scalable framework for studying tissue organization, therapeutic response, and disease mechanisms.

Therapy options for women with vulvar squamous cell carcinoma (VSCC) are limited. New strategies such as immune checkpoint inhibitors (ICI) or antibody-drug conjugates (ADC) are lacking. One potential target for the ADC Enfortumab Vedotin (EV) is Nectin-4. EV has been successfully used to treat metastatic urothelial carcinoma (mUC). Currently, there is limited data on the expression of Nectin-4 for VSCC. Our aim was therefore to investigate the expression of Nectin-4 in women with VSCC. In total, 219 formalin-fixed paraffin-embedded tissue (FFPE) samples of primary VSCC were collected between 2000 and 2021. A next generation tissue microarray (ngTMA) was constructed. The assessment of Nectin-4 expression was performed through the implementation of immunohistochemistry (IHC). We also performed NECTIN-4 fluorescence in situ hybridization (FISH) to evaluate amplification of Nectin-4. The median Nectin-4 H-score for membranous expression was 22.5, with 32.4

The microscopic observation of blood cells is a crucial step in diagnosing pathologies such as leukemia. DINOv2 models have been employed to extract features from blood cell images, but they do not include biological knowledge, nor do they allow multi-granular labels. To enhance the representation of these cells, we propose leveraging a biologically informed hierarchy of white blood cell types. We train a DINOv2-based foundation model with a semi-supervised framework that uses hierarchical supervision. It enables using datasets with varying levels of label precision within a structure that represents the process of cell differentiation. To support multi-level label precision, we modify the original hierarchical loss function, allowing any hierarchy level to serve as a ground truth class. We evaluate our model on three external datasets, including an out-of-domain set of cervical cells. Our approach improves generalization of the model to new datasets, improving by 1 percentage point the balanced accuracy on the two blood cell external datasets, and by 2.5 percentage point the balanced accuracy on the out-of-domain dataset. In addition the proposed strategy better aligns the model’s latent space with biological properties, leading to more acceptable misclassifications

Timely and accurate lymphoma diagnosis is essential for guiding cancer treatment. Standard diagnostic practice combines hematoxylin and eosin (HE)-stained whole slide images with immunohistochemistry, flow cytometry, and molecular genetic tests to determine lymphoma subtypes, a process requiring costly equipment, skilled personnel, and causing treatment delays. Deep learning methods could assist pathologists by extracting diagnostic information from routinely available HE-stained slides, yet comprehensive benchmarks for lymphoma subtyping on multicenter data are lacking. In this work, we present the first multicenter lymphoma benchmarking dataset covering four common lymphoma subtypes and healthy control tissue. We systematically evaluate five publicly available pathology foundation models (H-optimus-1, H0-mini, Virchow2, UNI2, Titan) combined with attention-based (AB-MIL) and transformer-based (TransMIL) multiple instance learning aggregators across three magnifications (10x, 20x, 40x). On in-distribution test sets, models achieve multiclass balanced accuracies exceeding 80

BACKGROUND:Branching of colon crypts represents a histological hallmark of inflammatory bowel disease (IBD). The branching of the crypt has been observed to occur both symmetrically and asymmetrically, suggesting two distinct reaction patterns of the colon mucosa. Accurate classification of these two patterns can contribute to improved quantitative description and histologic characterization of IBD subtypes. METHODS:We describe the morphology of branching crypts using manually crafted morphological features. Using a dataset annotated by an expert, we developed and implemented an machine learning model capable of classifying individual crypts based on these features. A multirater survey was conducted to compare interrater agreement between experts and our model. RESULTS:A classic ensemble model utilizing our manually crafted features achieved a mean balanced accuracy of 0.80, while a deep learning-based model using the segmentation masks achieved a value of 0.79. The survey also showed moderate agreement between the classic ensemble model and senior pathologists. CONCLUSIONS:We present a machine learning model capable of distinguishing both modes of crypt branching patterns. Furthermore, using a hand-crafted feature approach allowed us to directly interpret the classification criteria of our algorithm, rendering it more transparent and interpretable than black box classification models.

Head and neck cancer is a common disease and is associated with a poor prognosis. A promising approach to improving patient outcomes is personalized treatment, which uses information from a variety of modalities. However, only little progress has been made due to the lack of large public datasets. We present a multimodal dataset, HANCOCK, that comprises monocentric, real-world data of 763 head and neck cancer patients. Our dataset contains demographical, pathological, and blood data as well as surgery reports and histologic images, that can be explored in a low-dimensional representation. We can show that combining these modalities using machine learning is superior to a single modality and the integration of imaging data using foundation models helps in endpoint prediction. We believe that HANCOCK will not only open new insights into head and neck cancer pathology but also serve as a major source for researching multimodal machine-learning methodologies in precision oncology.

Distinguishing cell types in a peripheral blood smear is critical for diagnosing blood diseases, such as leukemia subtypes. Artificial intelligence can assist in automating cell classification. For training robust machine learning algorithms, however, large and well-annotated single-cell datasets are pivotal. Here, we introduce a large, publicly available, annotated peripheral blood dataset comprising >40,000 single-cell images classified into 18 classes by cytomorphology experts from the Munich Leukemia Laboratory, the largest European laboratory for blood disease diagnostics. By making our dataset publicly available, we provide a valuable resource for medical and machine learning researchers and support the development of reliable and clinically relevant diagnostic tools for diagnosing hematological diseases.

Transcriptome-based tumor classification has enhanced the molecular characterization of muscle-invasive bladder cancer (MIBC) subtypes. However, the degraded nature of formalin-fixed paraffin-embedded (FFPE) material and the expensive sequencing costs for routine use have limited the use of subtypes in clinical and trial settings. Here, we present an optimized analysis workflow for MIBC molecular subtype prediction from FFPE samples. FFPE material from 240 MIBC samples was sequenced using QuantSeq 3' mRNA sequencing with unique molecular identifiers (UMIs) and analyzed via a customized RNA-Seq pipeline. The association of consensus subtypes with histology and immunohistochemical expression of core basal/luminal protein markers was assessed. In addition, subtype robustness was explored by simulating scenarios at lower sequencing depths and without UMIs. Five MIBC consensus subtypes were identified in the cohort. The basal/squamous group showed higher expression of KRT14, KRT5, and CD44, and was mainly divergent squamous. Vice versa, luminal, and stroma-rich subtypes had conventional urothelial or urothelial subtype histology, with higher expression of KRT20, FOXA1, and GATA3. The neuroendocrine-like samples had small cell neuroendocrine histology and were negative for luminal/basal markers. Subtype calling from 24 matched fresh-frozen samples analyzed with full-length RNA-Seq showed 87.5% agreement. Furthermore, the subtypes were robust to decreasing sequencing depths and to the absence of UMIs. Taken together, we provide a robust and cost-effective workflow for MIBC consensus molecular subtyping from FFPE-derived RNA. This workflow can be easily implemented as a molecular pathological assay for patient care, clinical trials, and translational research.

Building deep learning models that can rapidly segment whole slide images (WSIs) using only a handful of training samples remains an open challenge in computational pathology. The difficulty lies in the histological images themselves: many morphological structures within a slide are closely related and very similar in appearance, making it difficult to distinguish between them. However, a skilled pathologist can quickly identify the relevant phenotypes. Through years of training, they have learned to organize visual features into a hierarchical taxonomy (e.g., identifying carcinoma versus healthy tissue, or distinguishing regions within a tumor as cancer cells, the microenvironment, ...). Thus, each region is associated with multiple labels representing different tissue types. Pathologists typically deal with this by analyzing the specimen at multiple scales and comparing visual features between different magnifications. Inspired by this multi-scale diagnostic workflow, we introduce the Navigator, a vision model that navigates through WSIs like a domain expert: it searches for the region of interest at a low scale, zooms in gradually, and localizes ever finer microanatomical classes. As a result, the Navigator can detect coarse-grained patterns at lower resolution and fine-grained features at higher resolution. In addition, to deal with sparsely annotated samples, we train the Navigator with a novel semi-supervised framework called S5CL v2. The proposed model improves the F1 score by up to 8% on various datasets including our challenging new TCGA-COAD-30CLS and Erlangen cohorts.

Background Squamous cell vulvar carcinoma is a rare malignant disease of women. In higher tumor stages survival rates are poor. Therapy options are limited. Immunoncology plays an increasing role in the treatment of gynecology cancers. Data on the expression of PD-L1 in vulvar cancer are rare and contradictory. We sought to describe the expression of PD-L1 in VSCC in respect to the clinicopathologic characteristics of the tumor. Study design We conducted a retrospective analysis including women with primary and recurrent vulvar cancer between 2000 and 2021. A next generation tissue micro array (ngTMA) was constructed for the analysis of PD-L1 expression. Results In total 238 women with primary VSCC and 66 cases of local or distant recurrent vulvar cancer were included. 80 women with primary VSCC (33.6 %) had tumors with common positive score (CPS) <1 and 63 women (26.5 %) had tumors with CPS 1- < 10 and 95 women with CPS ≥10 (39.9 %). In the PD-L1 positive group the rates of p53+, groin metastasis, lymphatic invasion and tumor infiltration lymphocytes were higher as compared to PD-L1 negative (CPS <1). There was no significant influence of CPS in overall survival in addition to other prognostic factors (P = 0.13, likelihood ratio test). Conclusion PD-L1 expression in primary vulvar cancer is associated with poorer prognosis. Hence, PD-L1 is a possible target for immune checkpoint inhibitors and women might benefit from special treatment options.

Pathogenic activating mutations in the fibroblast growth factor receptor 3 (FGFR3) drive disease maintenance and progression in urothelial cancer. 10-15% of muscle-invasive and metastatic urothelial cancer (MIBC/mUC) are FGFR3-mutant. Selective targeting of FGFR3 hotspot mutations with tyrosine kinase inhibitors (e.g., erdafitinib) is approved for mUC and requires FGFR3 mutational testing. However, current testing assays (polymerase chain reaction or next-generation sequencing) necessitate high tissue quality, have long turnover time, and are expensive. To overcome these limitations, we develop a deep-learning model that detects FGFR3 mutations using routine hematoxylin-eosin slides. Encompassing 1222 cases, our study is a large-scale validation of a model prescreening FGFR3 mutations for MIBC and mUC patients. In this work, we demonstrate that our model achieves high sensitivity (>93%) on advanced and metastatic cases while reducing molecular testing by 40% on average, thereby offering a cost-effective and rapid pre-screening tool for identifying patients eligible for FGFR3 targeted therapies.

AbstractUpper tract urothelial carcinoma (UTUC) is a rare and aggressive, yet understudied, urothelial carcinoma (UC). The more frequent UC of the bladder comprises several molecular subtypes, associated with different targeted therapies and overlapping with protein‐based subtypes. However, if and how these findings extend to UTUC remains unclear. Artificial intelligence‐based approaches could help elucidate UTUC's biology and extend access to targeted treatments to a wider patient audience. Here, UTUC protein‐based subtypes were identified, and a deep‐learning (DL) workflow was developed to predict them directly from routine histopathological H&E slides. Protein‐based subtypes in a retrospective cohort of 163 invasive tumors were assigned by hierarchical clustering of the immunohistochemical expression of three luminal (FOXA1, GATA3, and CK20) and three basal (CD44, CK5, and CK14) markers. Cluster analysis identified distinctive luminal (N = 80) and basal (N = 42) subtypes. The luminal subtype mostly included pushing, papillary tumors, whereas the basal subtype diffusely infiltrating, non‐papillary tumors. DL model building relied on a transfer‐learning approach by fine‐tuning a pre‐trained ResNet50. Classification performance was measured via three‐fold repeated cross‐validation. A mean area under the receiver operating characteristic curve of 0.83 (95% CI: 0.67–0.99), 0.8 (95% CI: 0.62–0.99), and 0.81 (95% CI: 0.65–0.96) was reached in the three repetitions. High‐confidence DL‐based predicted subtypes showed significant associations (p < 0.001) with morphological features, i.e. tumor type, histological subtypes, and infiltration type. Furthermore, a significant association was found with programmed cell death ligand 1 (PD‐L1) combined positive score (p < 0.001) and FGFR3 mutational status (p = 0.002), with high‐confidence basal predictions containing a higher proportion of PD‐L1 positive samples and high‐confidence luminal predictions a higher proportion of FGFR3‐mutated samples. Testing of the DL model on an independent cohort highlighted the importance to accommodate histological subtypes. Taken together, our DL workflow can predict protein‐based UTUC subtypes, associated with the presence of targetable alterations, directly from H&E slides.

OBJECTIVES:Serum protein electrophoresis (SPE) in combination with immunotyping (IMT) is the diagnostic standard for detecting monoclonal proteins (M-proteins). However, interpretation of SPE and IMT is weakly standardized, time consuming and investigator dependent. Here, we present five machine learning (ML) approaches for automated detection of M-proteins on SPE on an unprecedented large and well-curated data set and compare the performance with that of laboratory experts. METHODS:SPE and IMT were performed in serum samples from 69,722 individuals from Norway. IMT results were used to label the samples as M-protein present (positive, n=4,273) or absent (negative n=65,449). Four feature-based ML algorithms and one convolutional neural network (CNN) were trained on 68,722 randomly selected SPE patterns to detect M-proteins. Algorithm performance was compared to that of an expert group of clinical pathologists and laboratory technicians (n=10) on a test set of 1,000 samples. RESULTS:The random forest classifier showed the best performance (F1-Score 93.2 %, accuracy 99.1 %, sensitivity 89.9 %, specificity 99.8 %, positive predictive value 96.9 %, negative predictive value 99.3 %) and outperformed the experts (F1-Score 61.2 ± 16.0 %, accuracy 89.2 ± 10.2 %, sensitivity 94.3 ± 2.8 %, specificity 88.9 ± 10.9 %, positive predictive value 47.3 ± 16.2 %, negative predictive value 99.5 ± 0.2 %) on the test set. Interestingly the performance of the RFC saturated, the CNN performance increased steadily within our training set (n=68,722). CONCLUSIONS:Feature-based ML systems are capable of automated detection of M-proteins on SPE beyond expert-level and show potential for use in the clinical laboratory.

IntroductionUrothelial bladder cancer is frequent and exhibits diverse prognoses influenced by molecular subtypes, urothelial subtype histology, and immune microenvironments. HLA-G, known for immune regulation, displays significant membranous expression in tumor tissues.MethodsWe studied the protein expression of Human Leucocyte Antigen G (HLA-G) in 241 Muscle-Invasive Bladder Cancer (MIBC) patients, elucidating its potential clinical and biological significance. Protein expression levels were evaluated and correlated with molecular subtypes, histological characteristics, immune microenvironment markers, and survival outcomes.ResultsHigh HLA-G expression associates with poor overall survival (OS) and diseasespecific survival (DSS), independent of clinicopathological parameters. HLA-G expression varies among molecular subtypes and Urothelial Subtype Histology, e.g., elevated expression levels in basal/squamous MIBC and those with sarcomatoid differentiation. Notably, HLA-G is increased in MIBC with an immune evasive microenvironment (high PD-L1 tumor cell expression, NK cell depletion, granzyme B (GZMB)/CD8 ratio reduction, MHC class I (MHCI) expression reduction) that are characterized by immunosuppressive features and poor prognosis. Furthermore, HLA-G correlates with elevated levels of other immune checkpoint proteins (TIGIT, LAG3, CTLA-4), indicating its role in immune evasion.DiscussionOur findings underscore HLA-G’s role as a potential prognostic marker and interesting immunotherapeutic target in MIBC. Its impact on immune evasion mechanisms and broad expression, coupled with associations withpoor survival and distinct tumor phenotypes, positions HLA-G as a promising protein for further exploration in developing targeted immunotherapies for MIBC patients.

Due to the progress of image analysis and classification systems in recent years, algorithms have been developed that support morphologic examination of both single cells and tissue samples. These algorithms are typically developed using data-driven strategies, which require comprehensive, large-scale datasets. In the diagnostic workup of hematopoietic malignancies, cytomorphologic examination and differentiation represents a key first step. In recent years, the availability of large-scale, high-quality datasets of single leukocytes from peripheral blood and bone marrow has led to the development of diagnostic support algorithms for this modality. These methods not only allow a faster and more consistent classification of diagnostically relevant cell types, but also pave the way for integrated analysis of cytomorphologic and molecular findings.