Breast Cancer Pathology Research Articles

Multi-classification of breast cancer pathology images based on a two-stage hybrid network

Multi-classification of breast cancer pathology images based on a two-stage hybrid network

Shared genetics between breast cancer and predisposing diseases identifies novel breast cancer treatment candidates.

Current effective breast cancer treatment options have severe side effects, highlighting a need for new therapies. Drug repurposing can accelerate improvements to care, as FDA-approved drugs have known safety and pharmacological profiles. Some drugs for other conditions, such as metformin, an antidiabetic, have been tested in clinical trials for repurposing for breast cancer. Here, we exploit the genetics of breast cancer and linked predisposing diseases to propose novel drug repurposing opportunities. We hypothesize that if a predisposing disease contributes to breast cancer pathology, identifying the pleiotropic genes related to the risk of cancer could prioritize drugs, among all drugs treating a predisposing disease. We aim to develop a method to not only prioritize drugs for repurposing, but also to highlight shared etiology explaining repurposing. We compile breast cancer's predisposing diseases from literature. For each predisposing disease, we use GWAS summary statistics data to identify genes in loci showing genetic correlation with breast cancer. Then, we use a network approach to link these shared genes to canonical pathways. Similarly, for all drugs treating the predisposing disease, we link their targets to pathways. In this manner, we are able to prioritize a list of drugs based on each predisposing disease, with each drug linked to a set of implicating pathways. Finally, we evaluate our recommendations against drugs currently under investigation for breast cancer. We identify 84 loci harboring mutations with positively correlated effects between breast cancer and its predisposing diseases; these contain 194 identified shared genes. Out of the 112 drugs indicated for the predisposing diseases, 74 drugs can be linked to shared genes via pathways (candidate drugs for repurposing). Fifteen out of these candidate drugs are already in advanced clinical trial phases or approved for breast cancer (OR = 9.28, p = 7.99e-03, one-sided Fisher's exact test), highlighting the ability of our approach to identify likely successful candidate drugs for repurposing. Our novel approach accelerates drug repurposing for breast cancer by leveraging shared genetics with its known predisposing diseases. The result provides 59 novel candidate drugs alongside biological insights supporting each recommendation.

Research progress of breast pathology image diagnosis based on deep learning

Breast cancer is a malignancy caused by the abnormal proliferation of breast epithelial cells, predominantly affecting female patients, and it is commonly diagnosed using histopathological images. Currently, deep learning techniques have made significant breakthroughs in medical image processing, outperforming traditional detection methods in breast cancer pathology classification tasks. This paper first reviewed the advances in applying deep learning to breast pathology images, focusing on three key areas: multi-scale feature extraction, cellular feature analysis, and classification. Next, it summarized the advantages of multimodal data fusion methods for breast pathology images. Finally, the study discussed the challenges and future prospects of deep learning in breast cancer pathology image diagnosis, providing important guidance for advancing the use of deep learning in breast diagnosis.

7993 Thienopyrimidine-Based Estrogen Receptor Modulators Adopt Unique Ligand Binding Poses to Elicit Anti-Proliferative Activities in ER+ Breast Cancer Cells

Abstract Disclosure: E.C. Fink: None. S.L. Mateus: None. N. Meganathan: None. V.K. Sammeta: None. J.D. Norris: None. D.P. McDonnell: None. T. Willson: None. S.W. Fanning: None. One in eight women will be diagnosed with breast cancer in their lifetimes. Estrogen receptor alpha (ER) drives breast cancer pathology and is expressed in approximately seventy percent of tumors. Hormone therapies are used to treat and prevent the metastasis of ER+ breast cancers and these patients have favorable 5-year survivals. However, most breast cancer patient deaths are ER+. While the next generation of antiestrogens have shown promise in the advanced setting, more work needs to be done to fully address these mortalities. ER is a ligand-dependent master transcriptional regulator, whereby ligand-specific ER conformational ensembles redirect multiple programs to produce oncogenic or therapeutic endpoints. To better understand ER structure-activity relationships, we have developed new antiestrogenic small molecules based on a thienopyrimidine scaffold. Comprehensive structural, ER activity, and cancer endpoint profiling shows that these new molecules adopt a unique ligand binding pose, which perturbs new structural elements within the orthosteric hormone binding pocket. These molecules downregulate ER target genes and halt the proliferation of ER+ breast cancer cells. Presentation: 6/3/2024

Tumor detection in breast cancer pathology patches using a Multi-scale Multi-head Self-attention Ensemble Network on Whole Slide Images

Breast cancer (BC) is the most common type of cancer among women globally and is one of the leading causes of cancer-related deaths among women. In the diagnosis of BC, histopathological assessment is the gold standard, where automated tumor detection technologies play a pivotal role. Utilizing Convolutional Neural Networks (CNNs) for automated analysis of image patches from Whole Slide Images (WSIs) enhances detection accuracy and alleviates the workload of pathologists. However, CNNs often face limitations in handling pathological patches due to a lack of sufficient contextual information and limited feature generation capabilities. To address this, we propose a novel Multi-scale Multi-head Self-attention Ensemble Network (MMSEN), which integrates a multi-scale feature generation module, a convolutional self-attention module, and an adaptive feature integration and output module, effectively optimizing the performance of classical CNNs. The design of MMSEN optimizes the capture of key information and the comprehensive integration of features in WSIs pathological patches, significantly enhancing the precision of tumor detection. Validation results from a five-fold cross-validation experiment on the PatchCamelyon (PCam) dataset demonstrate that MMSEN achieves a ROC-AUC of 99.01%±0.02%, an F1 Score of 98.00%±0.08%, a Balanced Accuracy (B-Acc) of 98.00%±0.08%, and a Matthews Correlation Coefficient (MCC) of 96.00%±0.16% (p<0.05). These results demonstrate the effectiveness and potential of MMSEN in detecting tumors from pathological patches in WSIs for BC.

A Dedicated Menopausal After Cancer Clinic May Improve Adherence to Endocrine Therapy For Breast Cancer: A Population Based Study

PurposeTo examine utilization of a dedicated menopause symptoms after cancer clinic (MSAC) and to determine whether women referred to the MSAC for management of severe hot flush symptoms are more likely to adhere to endocrine therapy compare to those with severe symptoms not referred to MSAC. Patients and methodsBreast cancer patients prescribed endocrine therapy with a diagnosis of estrogen-receptor positive breast cancer between January 2003 and December 2011 were identified from the Royal Perth Hospital Breast Unit database. Details of breast cancer pathology, endocrine therapy, endocrine therapy related side effects, referral to MSAC and patient reported adherence to endocrine therapy for up to 4 years were ascertained from the database and medical records systems. For those with severe vasomotor symptoms, total duration of endocrine therapy was compared between women referred to MSAC and those who were not referred to MSAC. ResultsAbout 1275 women were identified from the database, with the cohort followed up until Dec 2016. Of these women, 120 (9.4%) were referred to MSAC and 1155 (90.1%) received usual care. In total, 147 reported severe vasomotor symptoms of whom almost half (71) were referred to MSAC. Women with severe vasomotor symptoms managed by MSAC were less likely to discontinue endocrine therapy (15.5%) compared with those managed with usual care (26.3%). However, this difference was not statistically significant (chi-square test statistic = 2.584, 1df, P = .1). ConclusionManagement of severe vasomotor symptoms at a dedicated menopause clinic may increase adherence to endocrine therapy for breast cancer.

2142: Clinical data extraction: Prompts for breast cancer pathology & ultrasound reports

2142: Clinical data extraction: Prompts for breast cancer pathology & ultrasound reports

Inflammation and Tumor Progression: The Differential Impact of SAA in Breast Cancer Models.

Background: Previous research has shown that the Serum Amyloid A (SAA) protein family is intricately involved in inflammatory signaling and various disease pathologies. We have previously demonstrated that SAA is associated with increased colitis disease severity and the promotion of tumorigenesis. However, the specific role of SAA proteins in breast cancer pathology remains unclear. Therefore, we investigated the role of systemic SAA1 and SAA2 (SAA1/2) in a triple-negative breast cancer mouse model. Methods: Syngeneic breast tumors were established in wild-type mice, and mice lacking the SAA1/2 (SAADKO). Subsequently, tumor volume was monitored, species survival determined, the inflammatory profiles of mice assessed with a multiplex assay, and tumor molecular biology and histology characterized with Western blotting and H&E histological staining. Results: WT tumor-bearing mice had increased levels of plasma SAA compared to wild-type control mice, while SAADKO control and tumor-bearing mice presented with lower levels of SAA in their plasma. SAADKO tumor-bearing mice also displayed significantly lower concentrations of systemic inflammatory markers. Tumors from SAADKO mice overall had lower levels of SAA compared to tumors from wild-type mice, decreased apoptosis and inflammasome signaling, and little to no tumor necrosis. Conclusions: We demonstrated that systemic SAA1/2 stimulates the activation of the NLRP3 inflammasome in breast tumors, leading to the production of pro-inflammatory cytokines. This, in turn, promoted apoptosis and tumor necrosis but did not significantly impact tumor growth or histological grading.

Availability and geographic access to breast cancer pathology services in Ghana.

Breast cancer poses a significant health challenge in Sub-Saharan Africa, particularly in Ghana, where late-stage diagnoses and limited healthcare access contribute to elevated mortality rates. This study focuses on the crucial role of pathology and laboratory medical (PALM) services in the timely diagnosis of breast cancer within Ghana. A cross-sectional survey of hospitals was completed from November 2020 to October 2021, with 94.8% of identified in-country hospitals participating. Pathology service-related parameters assessed included whether pathology was available for the diagnosis of breast cancer on-site or via external referral, the number of pathology personnel, additional breast cancer diagnostic capabilities including estrogen and progesterone and/or HER2 testing, and the time from biopsy to patients receiving their results. Geospatial mapping was used to identify areas of limited access. Of the 328 participating hospitals, 136 (41%) reported breast cancer pathology services, with only 6 having on-site capabilities. Pathology personnel, comprising 15 consultants and 15 specialists, were concentrated in major referral centers, particularly in Greater Accra and Kumasi. An assessment of referral patterns suggested that 75% of the population reside within an hour of breast cancer pathology services. Among the 136 hospitals with access to breast cancer pathology, only a limited number reported that results included ER/PR (38%) and HER2 testing (33%). Ghana has been able to ensure significant pathology service availability through robust referral pathways with centralized labs. Despite this, difficulties persist with the majority of pathology results not including hormone receptor testing which is important in providing tumor specific treatment.

Three-Dimensional Visualization of Breast Cancer Pathology Evolution in Clinical Patient Tissues with NIR-II Imaging.

Breast cancer (BC) is the most common tumor worldwide and requires crucial molecular typing for treatment and prognosis assessment. Currently, approaches like pathological staining, immunohistochemistry (IHC), and immunofluorescence (IF) face limitations due to the low signal-to-background ratio (SBR) and high tumor heterogeneity, resulting in a high misdiagnosis rate. Fluorescent assay in the second near-infrared region (NIR-II, 1000-1700 nm) exhibits ultrahigh SBR owing to diminished scattering and tissue autofluorescence. Here, we present a NIR-II strategy for accurate BC molecular typing and three-dimensional (3D) visualization based on the atomically precise fluorescent Au24Pr1 clusters. Single-atom Pr doping results in 3.9-fold fluorescence enhancement and long-term photostability. The Au24Pr1 clusters possess high fluorescence centered at ∼1100 nm and the SBR on pathological section diagnosis was 4 times higher than that of NIR-I imaging. This enables high spatial resolution 3D visualization of biopsy specimens, which can surmount tissue heterogeneity for clinical diagnosis of BC.

Role of WAVE3 as an of actin binding protein in the pathology of triple negative breast cancer.

Breast cancer, a prevalent global health concern, has sparked extensive research efforts, particularly focusing on triple negative breast cancer (TNBC), a subtype lacking estrogen receptor (ER), progesterone receptor, and epidermal growth factor receptor. TNBC's aggressive nature and resistance to hormone-based therapies heightens the risk of tumor progression and recurrence. Actin-binding proteins, specifically WAVE3 from the Wiskott-Aldrich syndrome protein (WASP) family, have emerged as major drivers in understanding TNBC biology. This review delves into the intricate molecular makeup of TNBC, shedding light on actin's fundamental role in cellular processes. Actin, a structural element in the cytoskeleton, regulates various cellular pathways essential for homeostasis. Its dynamic nature enables functions such as cell migration, motility, intracellular transport, cell division, and signal transduction. Actin-binding proteins, including WAVE3, play pivotal roles in these processes. WAVE3, a member of the WASP family, remains the focus of this review due to its potential involvement in TNBC progression. While actin-binding proteins are studied for their roles in healthy cellular cycles, their significance in TNBC remains underexplored. This review aims to discuss WAVE3's impact on TNBC, exploring its molecular makeup, functions, and significance in tumor progression. The intricate structure of WAVE3, featuring elements like the verprolin-cofilin-acidic domain and regulatory elements, plays a crucial role in regulating actin dynamics. Dysregulation of WAVE3 in TNBC has been linked to enhanced cell migration, invasion, extracellular matrix remodeling, epithelial-mesenchymal transition, tumor proliferation, and therapeutic resistance. Understanding the role of actin-binding proteins in cancer biology has potential clinical implications, making them potential prognostic biomarkers and promising therapeutic targets. The review emphasizes the need for further research into actin-binding proteins' clinical applications, diagnostic value, and therapeutic interventions. In conclusion, this comprehensive review explores the complex interplay between actin and actin-binding proteins, with special emphasis on WAVE3, in the context of TNBC. By unraveling the molecular intricacies, structural characteristics, and functional significance, the review paves the way for future research directions, clinical applications, and potential therapeutic strategies in the challenging landscape of TNBC.

Non-genetic heterogeneity and immune subtyping in breast cancer: Implications for immunotherapy and targeted therapeutics

Breast cancer (BC) is a complex and multifactorial disease, driven by genetic alterations that promote tumor growth and progression. However, recent research has highlighted the importance of non-genetic factors in shaping cancer evolution and influencing therapeutic outcomes. Non-genetic heterogeneity refers to diverse subpopulations of cancer cells within breast tumors, exhibiting distinct phenotypic and functional properties. These subpopulations can arise through various mechanisms, including clonal evolution, genetic changes, epigenetic changes, and reversible phenotypic transitions. Although genetic and epigenetic changes are important points of the pathology of breast cancer yet, the immune system also plays a crucial role in its progression. In clinical management, histologic and molecular classification of BC are used. Immunological subtyping of BC has gained attention in recent years as compared to traditional techniques. Intratumoral heterogeneity revealed by immunological microenvironment (IME) has opened novel opportunities for immunotherapy research. This systematic review is focused on non-genetic variability to identify and interlink immunological subgroups in breast cancer. This review provides a deep understanding of adaptive methods adopted by tumor cells to withstand changes in the tumor microenvironment and selective pressure imposed by medications. These adaptive methods include alterations in drug targets, immune system evasion, activation of survival pathways, and alterations in metabolism. Understanding non-genetic heterogeneity is essential for the development of targeted therapies.

Analysis of the Clinical Value of Immunohistochemical Testing in the Pathological Diagnosis of Breast Cancer

Objective: To investigate the clinical value of immunohistochemistry (IHC) detection in the pathological diagnosis of breast cancer. Methods: Eighty breast cancer patients admitted to Baoding No. 1 Central Hospital from June 2022 to June 2023 were selected as study subjects. The samples were divided into a positive group (40 cases) and a negative group (40 cases) according to ER and PR test results. Immunohistochemistry was performed on all patients to compare the differences between the two groups in C-erbB-2 positive expression and axillary lymph node metastasis. Results: The positive expression rate of C-erbB-2 in the positive group (35.00%) was significantly lower than that in the negative group (80.00%), with a highly significant difference (P &lt; 0.001). The axillary lymph node metastasis rate in the positive group (40.00%) was significantly lower than that in the negative group (75.00%), with a significant difference (P &lt; 0.05). Conclusion: Immunohistochemical detection in breast cancer pathology enhances diagnostic accuracy, predicts prognosis, and supports personalized treatment by identifying ER, PR, and C-erbB-2. It is worth being widely adopted in clinical practice.

Assessment of Breast Pathology Reporting Needs and Development of Tumor Synoptic Templates in Sub-Saharan Africa.

Breast pathology reports include many important details to guide clinical management. Reports with missing critical data elements are commonly seen in non-subspecialized pathology practices. The use of synoptic templates has been shown to improve pathology reports. Although synoptic templates are readily available by professional societies, many are not tailored to low-resource settings. To perform an assessment of current breast pathology reporting at 3 referral hospitals in sub-Saharan Africa and design a locally adapted breast cancer synoptic template. We conducted semi-structured interviews with key stakeholders involved in breast cancer care including pathologists, radiologists, oncologists, and surgeons from Nigeria, Tanzania, and Mozambique. Moreover, each stakeholder reviewed a preliminary synoptic template that was compiled by using templates from the College of American Pathologists, Royal College of Pathologists, and International Collaboration on Cancer Reporting, and was asked to score each data element as essential, optional, or exclude. A locally adapted synoptic template was then designed from the needs assessment. Using the adapted templates, a retrospective review of breast cancer pathology reports from 2020 to 2022 was conducted to determine the completeness of reports at the 3 institutions. A total of 17 physicians were interviewed. Review of pathology reports revealed that none of the reports across all 3 sites contained all data elements considered essential by local physicians. There is an urgent need to improve breast pathology reporting in sub-Saharan Africa. Development and implementation of synoptic templates in collaboration with key stakeholders has the potential to improve pathology reporting practices in low-resource settings.

The role of histamine and its receptors in breast cancer: from pathology to therapeutic targets.

Breast cancer is the most common malignancy in women, and despite the development of new treatment methods and the decreasing mortality rate in recent years, one of the clinical problems in breast cancer treatment is chronic inflammation in the tumor microenvironment. Histamine, an inflammatory mediator, is produced by tumor cells and can induce chronic inflammation and the growth of some tumors by recruiting inflammatory cells. It can also affect tumor physiopathology, antitumor treatment efficiency, and patient survival. Antihistamines, as histamine receptor antagonists, play a role in modulating the effects of these receptors in tumor cells and can affect some treatment methods for breast cancer therapy; in this review, we investigate the role of histamine, its receptors, and antihistamines in breast cancer pathology and treatment methods.

A lightweight spatially-aware classification model for breast cancer pathology images

Breast cancer is a prevalent malignant tumour with high global incidence. Its diagnosis relies primarily on the analysis of pathological breast images. Owing to the complex organisation of the tumour microenvironment, neural network models are essential as efficient classification tools in the field of pathological image analysis. This study introduced spatially-aware attention swift parallel convolution network (SPA-SPCNet), a lightweight and low-latency model for classifying breast pathologies. A novel module for multi-scale feature extraction was constructed using a depthwise separable convolution method. It focuses on the multi-scale features of pathological images to alleviate recognition problems caused by similar local features in breast cancer tissues. The module concatenates the convolutions of different kernels from three branches. Second, a lightweight dynamic spatially-aware attention module was introduced to integrate the visual graph convolutional architecture in a branch. This allowed the model to capture the spatial structure and relationships in image, enabling better handling of the unique spatial distribution relationship between breast cancer tissue structures. The other branch utilises a self-attention mechanism in the transformer. The module can dynamically adjust the attention of the model to different regions in the image, allowing it to focus on the key features of the complex spatial distribution of breast cancer tissue. This feature fusion method enabled the model to capture both global semantics and local details. Compared with existing lightweight models, the proposed model has advantages in terms of tissue structure classification accuracy, parameter quantity, floating-point operations, and real-time inference speed, providing a powerful tool for computer-aided breast pathological image classification.

Isophorone-based crystallization-induced-emission sensors detect proteome aggregation in live cells and tissues with breast cancer

BackgroundProtein misfolding and aggregation can lead to various diseases. Recent studies have shed light on the aggregated protein in breast cancer pathology, which suggests that it is crucial to design chemical sensors that visualize protein aggregates in breast cancer, especially in clinical patient-derived samples. However, most reported sensors are constrained in cultured cell lines. ResultsIn this work, we present the development of two isophorone-based crystallization-induced-emission fluorophores for detecting proteome aggregation in breast cancer cell line and tissues biopsied from diseased patients, designated as A1 and A2. These probes exhibited viscosity sensitivity and recovered their fluorescence strongly at crystalline state. Moreover, A1 and A2 exhibit selective binding capacity and strong fluorescence for various aggregated proteins. Utilizing these probes, we detect protein aggregation in stressed breast cancer cells, xenograft mouse model of human breast cancer and clinical patient-derived samples. Notably, the fluorescence intensity of both probes light up in tumor tissues. SignificanceThe synthesized isophorone-based crystallization-induced-emission fluorophores, A1 and A2, enable sensitive detection of protein aggregation in breast cancer cells and tissues. In the future, aggregated proteins are expected to become indicators for early diagnosis and clinical disease monitoring of breast cancer.

Shared genetics between breast cancer and predisposing diseases identifies novel breast cancer treatment candidates.

Current effective breast cancer treatment options have severe side effects, highlighting a need for new therapies. Drug repurposing can accelerate improvements to care, as FDA-approved drugs have known safety and pharmacological profiles. Some drugs for other conditions, such as metformin, an antidiabetic, have been tested in clinical trials for repurposing for breast cancer. Here, we exploit the genetics of breast cancer and linked predisposing diseases to propose novel drug repurposing. We hypothesize that if a predisposing disease contributes to breast cancer pathology, identifying the pleiotropic genes related to the risk of cancer could prioritize drug targets, among all drugs treating a predisposing disease. We aim to develop a method to not only prioritize drug repurposing, but also to highlight shared etiology explaining repurposing. We compile breast cancer's predisposing diseases from literature. For each predisposing disease, we use GWAS summary statistics to identify genes in loci showing genetic correlation with breast cancer. Then, we use a network approach to link these shared genes to canonical pathways, and similarly for all drugs treating the predisposing disease, we link their targets to pathways. In this manner, we are able to prioritize a list of drugs based on each predisposing disease, with each drug linked to a set of implicating pathways. Finally, we evaluate our recommendations against drugs currently under investigation for breast cancer. We identify 84 loci harboring mutations with positively correlated effects between breast cancer and its predisposing diseases; these contain 194 identified shared genes. Out of the 112 drugs indicated for the predisposing diseases, 76 drugs can be linked to shared genes via pathways (candidate drugs for repurposing). Fifteen out of these candidate drugs are already in advanced clinical trial phases or approved for breast cancer (OR = 9.28, p = 7.99e-03, one-sided Fisher's exact test), highlighting the ability of our approach to identify likely successful candidate drugs for repurposing. Our novel approach accelerates drug repurposing for breast cancer by leveraging shared genetics with its known risk factors. The result provides 59 novel candidate drugs alongside biological insights supporting each recommendation.

A comparative study of large language model-based zero-shot inference and task-specific supervised classification of breast cancer pathology reports.

Although supervised machine learning is popular for information extraction from clinical notes, creating large annotated datasets requires extensive domain expertise and is time-consuming. Meanwhile, large language models (LLMs) have demonstrated promising transfer learning capability. In this study, we explored whether recent LLMs could reduce the need for large-scale data annotations. We curated a dataset of 769 breast cancer pathology reports, manually labeled with 12 categories, to compare zero-shot classification capability of the following LLMs: GPT-4, GPT-3.5, Starling, and ClinicalCamel, with task-specific supervised classification performance of 3 models: random forests, long short-term memory networks with attention (LSTM-Att), and the UCSF-BERT model. Across all 12 tasks, the GPT-4 model performed either significantly better than or as well as the best supervised model, LSTM-Att (average macro F1-score of 0.86 vs 0.75), with advantage on tasks with high label imbalance. Other LLMs demonstrated poor performance. Frequent GPT-4 error categories included incorrect inferences from multiple samples and from history, and complex task design, and several LSTM-Att errors were related to poor generalization to the test set. On tasks where large annotated datasets cannot be easily collected, LLMs can reduce the burden of data labeling. However, if the use of LLMs is prohibitive, the use of simpler models with large annotated datasets can provide comparable results. GPT-4 demonstrated the potential to speed up the execution of clinical NLP studies by reducing the need for large annotated datasets. This may increase the utilization of NLP-based variables and outcomes in clinical studies.

Prediction of Tumor-Associated Macrophages and Immunotherapy Benefits Using Weakly Supervised Contrastive Learning in Breast Cancer Pathology Images.

The efficacy of immune checkpoint inhibitors is significantly influenced by the tumor immune microenvironment (TIME). RNA sequencing of tumor tissue can offer valuable insights into TIME, but its high cost and long turnaround time seriously restrict its utility in routine clinical examinations. Several recent studies have suggested that ultrahigh-resolution pathology images can infer cellular and molecular characteristics. However, few study pay attention to the quantitative estimation of various tumor infiltration immune cells from pathology images. In this paper, we integrated contrastive learning and weakly supervised learning to infer tumor-associated macrophages and potential immunotherapy benefit from whole slide images (WSIs) of H &E stained pathological sections. We split the high-resolution WSIs into tiles and then apply contrastive learning to extract features of each tile. After aggregating the features at the tile level, we employ weak supervisory signals to fine-tune the encoder for various downstream tasks. Comprehensive experiments on two independent breast cancer cohorts and spatial transcriptomics data demonstrate that the computational pathological features accurately predict the proportion of tumor-infiltrating immune cells, particularly the infiltration level of macrophages, as well as the immune subtypes and potential immunotherapy benefit. These findings demonstrate that our model effectively captures pathological features beyond human vision, establishing a mapping relationship between cellular compositions and histological morphology, thus expanding the clinical applications of digital pathology images.