Collagen fiber orientation disorder from H&E images is prognostic for early stage breast cancer: clinical trial validation

Publicly available datasets of breast histopathology H&E whole-slide images: A scoping review

Advancements in digital pathology and computing resources have made a significant impact in the field of computational pathology for breast cancer diagnosis and treatment. However, access to high-quality labeled histopathological images of breast cancer is a big challenge that limits the development of accurate and robust deep learning models. In this scoping review, we identified the publicly available datasets of breast H&E-stained whole-slide images (WSIs) that can be used to develop deep learning algorithms. We systematically searched 9 scientific literature databases and 9 research data repositories and found 17 publicly available datasets containing 10 385 H&E WSIs of breast cancer. Moreover, we reported image metadata and characteristics for each dataset to assist researchers in selecting proper datasets for specific tasks in breast cancer computational pathology. In addition, we compiled 2 lists of breast H&E patches and private datasets as supplementary resources for researchers. Notably, only 28% of the included articles utilized multiple datasets, and only 14% used an external validation set, suggesting that the performance of other developed models may be susceptible to overestimation. The TCGA-BRCA was used in 52% of the selected studies. This dataset has a considerable selection bias that can impact the robustness and generalizability of the trained algorithms. There is also a lack of consistent metadata reporting of breast WSI datasets that can be an issue in developing accurate deep learning models, indicating the necessity of establishing explicit guidelines for documenting breast WSI dataset characteristics and metadata.

Emulation and evaluation of tumor cell combined chemotherapy in isotropic/anisotropic collagen fiber microenvironments

The rapid emergence of anisotropic collagen fibers in the tissue microenvironment is a critical transition point in late-stage breast cancer. Specifically, the fiber orientation facilitates the likelihood of high-speed tumor cell invasion and metastasis, which pose lethal threats to patients. Thus, based on this transition point, one key issue is how to determine and evaluate efficient combination chemotherapy treatments in late-stage cancer. In this study, we designed a collagen microarray chip containing 241 high-throughput microchambers with embedded metastatic breast cancer cell MDA-MB-231-RFP. By utilizing collagen's unique structure and hydromechanical properties, the chip constructed three-dimensional isotropic and anisotropic collagen fiber structures to emulate the tumor cell microenvironment at early and late stages. We injected different chemotherapeutic drugs into its four channels and obtained composite biochemical concentration profiles. Our results demonstrate that anisotropic collagen fibers promote cell proliferation and migration more than isotropic collagen fibers, suggesting that the geometric arrangement of fibers plays an important role in regulating cell behavior. Moreover, the presence of anisotropic collagen fibers may be a potential factor leading to the poor efficacy of combined chemotherapy in late-stage breast cancer. We investigated the efficacy of various chemotherapy drugs using cell proliferation inhibitors paclitaxel and gemcitabine and tumor cell migration inhibitors 7rh and PP2. To ensure the validity of our findings, we followed a systematic approach that involved testing the inhibitory effects of these drugs. According to our results, the drug combinations' effectiveness could be ordered as follows: paclitaxel + gemcitabine > gemcitabine + 7rh > PP2 + paclitaxel > 7rh + PP2. This study shows that the biomimetic chip system not only facilitates the creation of a realistic in vitro model for examining the cell migration mechanism in late-stage breast cancer but also has the potential to function as an effective tool for future chemotherapy assessment and personalized medicine.

Intraoperative Imaging Techniques in Oncology

Imaging-based procedures have become well integrated into the diagnosis and management of oncological patients and play a significant role in reducing morbidity and mortality rates. Here we describe the established and upcoming surgical oncological imaging techniques and their impact on cancer management.

Computational pathology identifies immune-mediated collagen disruption to predict clinical outcomes in gynecologic malignancies

BACKGROUND

The role of immune cells in collagen degradation within the tumor microenvironment (TME) is unclear. Immune cells, particularly tumor-infiltrating lymphocytes (TILs), are known to alter the extracellular matrix, affecting cancer progression and patient survival. However, the quantitative evaluation of the immune modulatory impact on collagen architecture within the TME remains limited.

METHODS

We introduce CollaTIL, a computational pathology method that quantitatively characterizes the immune-collagen relationship within the TME of gynecologic cancers, including high-grade serous ovarian (HGSOC), cervical squamous cell carcinoma (CSCC), and endometrial carcinomas. CollaTIL aims to investigate immune modulatory impact on collagen architecture within the TME, aiming to uncover the interplay between the immune system and tumor progression.

RESULTS

We observe that an increased immune infiltrate is associated with chaotic collagen architecture and higher entropy, while immune sparse TME exhibits ordered collagen and lower entropy. Importantly, CollaTIL-associated features that stratify disease risk are linked with gene signatures corresponding to TCA-Cycle in CSCC, and amino acid metabolism, and macrophages in HGSOC.

CONCLUSIONS

CollaTIL uncovers a relationship between immune infiltration and collagen structure in the TME of gynecologic cancers. Integrating CollaTIL with genomic analysis offers promising opportunities for future therapeutic strategies and enhanced prognostic assessments in gynecologic oncology.

A population-level digital histologic biomarker for enhanced prognosis of invasive breast cancer

Breast cancer is a heterogeneous disease with variable survival outcomes. Pathologists grade the microscopic appearance of breast tissue using the Nottingham criteria, which are qualitative and do not account for noncancerous elements within the tumor microenvironment. Here we present the Histomic Prognostic Signature (HiPS), a comprehensive, interpretable scoring of the survival risk incurred by breast tumor microenvironment morphology. HiPS uses deep learning to accurately map cellular and tissue structures to measure epithelial, stromal, immune, and spatial interaction features. It was developed using a population-level cohort from the Cancer Prevention Study-II and validated using data from three independent cohorts, including the Prostate, Lung, Colorectal, and Ovarian Cancer trial, Cancer Prevention Study-3, and The Cancer Genome Atlas. HiPS consistently outperformed pathologists in predicting survival outcomes, independent of tumor-node-metastasis stage and pertinent variables. This was largely driven by stromal and immune features. In conclusion, HiPS is a robustly validated biomarker to support pathologists and improve patient prognosis.

Identifying primary tumor site of origin for liver metastases via a combination of handcrafted and deep learning features

Liver is one of the most common sites for metastases, which can occur on account of primary tumors from multiple sites of origin. Identifying the primary site of origin (PSO) of a metastasis can help in guiding therapeutic options for liver metastases. In this pilot study, we hypothesized that computer extracted handcrafted (HC) histomorphometric features can be utilized to identify the PSO of liver metastases. Cellular features, including tumor nuclei morphological and graph features as well as cytoplasm texture features, were extracted by computer algorithms from 175 slides (114 patients). The study comprised three experiments: (1) comparing and (2) fusing a machine learning (ML) model trained with HC pathomic features and deep learning (DL)-based classifiers to predict site of origin; (3) identifying the section of the primary tumor from which metastases were derived. For experiment 1, we divided the cohort into training sets composed of primary and matched liver metastases [60 patients, 121 whole slide images (WSIs)], and a hold-out validation set (54 patients, 54 WSIs) composed solely of liver metastases of known site of origin. Using the extracted HC features of the training set, a combination of supervised machine classifiers and unsupervised clustering was applied to identify the PSO. A random forest classifier achieved areas under the curve (AUCs) of 0.83, 0.64, 0.82, and 0.64 in classifying the metastatic tumor from colon, esophagus, breast, and pancreas on the validation set. The top features related to nuclear and peri-nuclear shape and textural attributes. We also trained a DL network to serve as a direct comparison to our method. The DL model achieved AUCs for colon: 0.94, esophagus: 0.66, breast: 0.79, and pancreas: 0.67 in identifying PSO. A decision fusion-based strategy was deployed to fuse the trained ML and DL classifiers and achieved slightly better results than ML or DL classifier alone (colon: 0.93, esophagus: 0.68, breast: 0.81, and pancreas: 0.69). For the third experiment, WSI-level attention maps were also generated using a trained DL network to generate a composite feature similarity heat map between paired primaries and their associated metastases. Our experiments revealed that epithelium-rich and moderately differentiated tumor regions of primary tumors were quantitatively similar to paired metastatic tumors. Our findings suggest that a combination of HC and DL features could potentially help identify the PSO for liver metastases while at the same time also potentially identify the spatial sites of origin for the metastases within primary tumors.

A Review of AI-Based Radiomics and Computational Pathology Approaches in Triple-Negative Breast Cancer: Current Applications and Perspectives

Breast cancer is one of the most common and deadly cancers worldwide. Approximately, 20% of all breast cancers are characterized as triple negative (TNBC). TNBC typically is associated with a poorer prognosis relative to other breast cancer subtypes. Due to its aggressiveness and lack of response to hormonal therapy, conventional cytotoxic chemotherapy is the usual treatment; however, this treatment is not always effective, and an important percentage of patients develop recurrence. More recently, immunotherapy has started to be used on some populations with TNBC showing promising results. Unfortunately, immunotherapy is only applicable to a minority of patients and responses in metastatic TNBC have overall been modest in comparison to other cancer types. This situation evidences the need for developing effective biomarkers that help to stratify and personalize patient management. Thanks to recent advances in artificial intelligence (AI), there has been an increasing interest in its use for medical applications aiming at supporting clinical decision making. Several works have used AI in combination with diagnostic medical imaging, more specifically radiology and digitized histopathological tissue samples, aiming to extract disease-specific information that is difficult to quantify by the human eye. These works have demonstrated that analysis of such images in the context of TNBC has great potential for (1) risk-stratifying patients to identify those patients who are more likely to experience disease recurrence or die from the disease and (2) predicting pathologic complete response. In this manuscript, we present an overview on AI and its integration with radiology and histopathological images for developing prognostic and predictive approaches for TNBC. We present state of the art approaches in the literature and discuss the opportunities and challenges with developing AI algorithms regarding further development and clinical deployment, including identifying those patients who may benefit from certain treatments (e.g., adjuvant chemotherapy) from those who may not and thereby should be directed toward other therapies, discovering potential differences between populations, and identifying disease subtypes.

Multimodal imaging of metabolic activities for distinguishing subtypes of breast cancer

Triple negative breast cancer (TNBC) is a highly aggressive form of cancer. Detecting TNBC early is crucial for improving disease prognosis and optimizing treatment. Unfortunately, conventional imaging techniques fall short in providing a comprehensive differentiation of TNBC subtypes due to their limited sensitivity and inability to capture subcellular details. In this study, we present a multimodal imaging platform that integrates heavy water (D2O)-probed stimulated Raman scattering (DO-SRS), two-photon fluorescence (TPF), and second harmonic generation (SHG) imaging. This platform allows us to directly visualize and quantify the metabolic activities of TNBC subtypes at a subcellular level. By utilizing DO-SRS imaging, we were able to identify distinct levels of de novo lipogenesis, protein synthesis, cytochrome c metabolic heterogeneity, and lipid unsaturation rates in various TNBC subtype tissues. Simultaneously, TPF imaging provided spatial distribution mapping of NAD[P]H and flavin signals in TNBC tissues, revealing a high redox ratio and significant lipid turnover rate in TNBC BL2 (HCC1806) samples. Furthermore, SHG imaging enabled us to observe diverse orientations of collagen fibers in TNBC tissues, with higher anisotropy at the tissue boundary compared to the center. Our multimodal imaging platform offers a highly sensitive and subcellular approach to characterizing not only TNBC, but also other tissue subtypes and cancers.

Linearly Polarized Infrared Spectroscopy for the Analysis of Biological Materials

The analysis of biological samples with polarized infrared spectroscopy (p-IR) has long been a widely practiced method for the determination of sample orientation and structural properties. In contrast to earlier works, which employed this method to investigate the fundamental chemistry of biological systems, recent interests are moving toward "real-world" applications for the evaluation and diagnosis of pathological states. This focal point review provides an up-to-date synopsis of the knowledge of biological materials garnered through linearly p-IR on biomolecules, cells, and tissues. An overview of the theory with special consideration to biological samples is provided. Different modalities which can be employed along with their capabilities and limitations are outlined. Furthermore, an in-depth discussion of factors regarding sample preparation, sample properties, and instrumentation, which can affect p-IR analysis is provided. Additionally, attention is drawn to the potential impacts of analysis of biological samples with inherently polarized light sources, such as synchrotron light and quantum cascade lasers. The vast applications of p-IR for the determination of the structure and orientation of biological samples are given. In conclusion, with considerations to emerging instrumentation, findings by other techniques, and the shift of focus toward clinical applications, we speculate on the future directions of this methodology.

Extracellular matrix-derived mechanical force governs breast cancer cell stemness and quiescence transition through integrin-DDR signaling

The extracellular matrix (ECM) serves as signals that regulate specific cell states in tumor tissues. Increasing evidence suggests that extracellular biomechanical force signals are critical in tumor progression. In this study, we aimed to explore the influence of ECM-derived biomechanical force on breast cancer cell status. Experiments were conducted using 3D collagen, fibrinogen, and Matrigel matrices to investigate the role of mechanical force in tumor development. Integrin-cytoskeleton-AIRE and DDR-STAT signals were examined using RNA sequencing and western blotting. Data from 1358 patients and 86 clinical specimens were used for ECM signature-prognosis analysis. Our findings revealed that ECM-derived mechanical force regulated tumor stemness and cell quiescence in breast cancer cells. A mechanical force of ~45 Pa derived from the extracellular substrate activated integrin β1/3 receptors, stimulating stem cell signaling pathways through the cytoskeleton/AIRE axis and promoting tumorigenic potential and stem-like phenotypes. However, excessive mechanical force (450 Pa) could drive stem-like cancer cells into a quiescent state, with the removal of mechanical forces leading to vigorous proliferation in quiescent cancer stem cells. Mechanical force facilitated cell cycle arrest to induce quiescence, dependent on DDR2/STAT1/P27 signaling. Therefore, ECM-derived mechanical force governs breast cancer cell status and proliferative characteristics through stiffness alterations. We further established an ECM signature based on the fibrinogen/fibronectin/vitronectin/elastin axis, which efficiently predicts patient prognosis in breast cancer. Our findings highlight the vital role of ECM-derived mechanical force in governing breast cancer cell stemness/quiescence transition and suggest the novel use of ECM signature in predicting the clinical prognosis of breast cancer.

Multilayer outperforms single-layer slide scanning in AI-based classification of whole slide images with low-burden acid-fast mycobacteria (AFB)

Manual screening of Ziehl-Neelsen (ZN)-stained slides that are negative or contain rare acid-fast mycobacteria (AFB) is labor-intensive and requires repetitive refocusing to visualize AFB candidates under the microscope. Whole slide image (WSI) scanners have enabled implementation of AI to classify digital ZN-stained slides as AFB+ or AFB-. By default, these scanners acquire a single-layer WSI. However, some scanners can acquire a multilayer WSI with a z-stack and an extended focus image layer embedded. We developed a parameterized WSI classification pipeline to assess whether multilayer imaging improves ZN-stained slide classification accuracy. A CNN built into the pipeline classified tiles in each image layer to form an AFB probability score heatmap. Features extracted from the heatmap were then entered into a WSI classifier. 46 AFB+ and 88 AFB- single-layer WSIs were used for the classifier training. 15 AFB+ (with rare microorganisms) and 5 AFB- multilayer WSIs comprised the test set. Parameters in the pipeline included: (a) a WSI representation: z-stack of image layers, middle image layer (a single image layer equivalent) or an extended focus image layer, (b) 4 methods of aggregating AFB probability scores across the z-stack, (c) 3 classifiers, (d) 3 AFB probability thresholds, and (e) 9 feature vector types extracted from the aggregated AFB probability heatmaps. Balanced accuracy (BACC) was used to measure the pipeline performance for all parameter combinations. Analysis of Covariance (ANCOVA) was used to statistically evaluate the effect of each parameter on the BACC. After adjusting for other factors, a significant effect of the WSI representation (p-value < 1.99E-76), classifier type (p-value < 1.73E-21), and AFB threshold (p-value = 0.003) was observed on the BACC. The feature type had no significant effect (p-value = 0.459) on the BACC. WSIs represented by the middle layer, extended focus layer and the z-stack followed by the weighted averaging of AFB probability scores were classified with the average BACC of 58.80%, 68.64%, and 77.28%, respectively. The multilayer WSIs represented by the z-stack with the weighted averaging of AFB probability scores were classified by a Random Forest classifier with the average BACC of 83.32%. Low classification accuracy of WSIs represented by the middle layer suggests that they contain fewer features permitting identification of AFB than the multilayer WSIs. Our results indicate that single-layer acquisition can introduce a bias (sampling error) into the WSI. This bias can be mitigated by the multilayer or the extended focus acquisitions.

Artificial Intelligence in Clinical Oncology: From Data to Digital Pathology and Treatment

Recently, a wide spectrum of artificial intelligence (AI)-based applications in the broader categories of digital pathology, biomarker development, and treatment have been explored. In the domain of digital pathology, these have included novel analytical strategies for realizing new information derived from standard histology to guide treatment selection and biomarker development to predict treatment selection and response. In therapeutics, these have included AI-driven drug target discovery, drug design and repurposing, combination regimen optimization, modulated dosing, and beyond. Given the continued advances that are emerging, it is important to develop workflows that seamlessly combine the various segments of AI innovation to comprehensively augment the diagnostic and interventional arsenal of the clinical oncology community. To overcome challenges that remain with regard to the ideation, validation, and deployment of AI in clinical oncology, recommendations toward bringing this workflow to fruition are also provided from clinical, engineering, implementation, and health care economics considerations. Ultimately, this work proposes frameworks that can potentially integrate these domains toward the sustainable adoption of practice-changing AI by the clinical oncology community to drive improved patient outcomes.

Biological insights and novel biomarker discovery through deep learning approaches in breast cancer histopathology

Breast cancer remains a highly prevalent disease with considerable inter- and intra-tumoral heterogeneity complicating prognostication and treatment decisions. The utilization and depth of genomic, transcriptomic and proteomic data for cancer has exploded over recent times and the addition of spatial context to this information, by understanding the correlating morphologic and spatial patterns of cells in tissue samples, has created an exciting frontier of research, histo-genomics. At the same time, deep learning (DL), a class of machine learning algorithms employing artificial neural networks, has rapidly progressed in the last decade with a confluence of technical developments - including the advent of modern graphic processing units (GPU), allowing efficient implementation of increasingly complex architectures at scale; advances in the theoretical and practical design of network architectures; and access to larger datasets for training - all leading to sweeping advances in image classification and object detection. In this review, we examine recent developments in the application of DL in breast cancer histology with particular emphasis of those producing biologic insights or novel biomarkers, spanning the extraction of genomic information to the use of stroma to predict cancer recurrence, with the aim of suggesting avenues for further advancing this exciting field.

Molecular breast cancer subtype identification using photoacoustic spectral analysis and machine learning at the biomacromolecular level

Breast cancer threatens the health of women worldwide, and its molecular subtypes largely determine the therapy and prognosis of patients. However, an uncomplicated and accurate method to identify subtypes is currently lacking. This study utilized photoacoustic spectral analysis (PASA) based on the partial least squares discriminant algorithm (PLS-DA) to identify molecular breast cancer subtypes at the biomacromolecular level in vivo. The area of power spectrum density (APSD) was extracted to semi-quantify the biomacromolecule content. The feature wavelengths were obtained via the variable importance in projection (VIP) score and the selectivity ratio (Sratio), to identify the biomarkers. The PASA achieved an accuracy of 84%. Most of the feature wavelengths fell into the collagen-dominated absorption waveband, which was consistent with the histopathological results. This paper proposes a successful method for identifying molecular breast cancer subtypes and proves that collagen can be treated as a biomarker for molecular breast cancer subtyping.

Identification of molecularly unique tumor-associated mesenchymal stromal cells in breast cancer patients

The tumor microenvironment is a complex mixture of cell types that bi-directionally interact and influence tumor initiation, progression, recurrence, and patient survival. Mesenchymal stromal cells (MSCs) of the tumor microenvironment engage in crosstalk with cancer cells to mediate epigenetic control of gene expression. We identified CD90+ MSCs residing in the tumor microenvironment of patients with invasive breast cancer that exhibit a unique gene expression signature. Single-cell transcriptional analysis of these MSCs in tumor-associated stroma identified a distinct subpopulation characterized by increased expression of genes functionally related to extracellular matrix signaling. Blocking the TGFβ pathway reveals that these cells directly contribute to cancer cell proliferation. Our findings provide novel insight into communication between breast cancer cells and MSCs that are consistent with an epithelial to mesenchymal transition and acquisition of competency for compromised control of proliferation, mobility, motility, and phenotype.

Extracellular matrix remodeling in tumor progression and immune escape: from mechanisms to treatments

The malignant tumor is a multi-etiological, systemic and complex disease characterized by uncontrolled cell proliferation and distant metastasis. Anticancer treatments including adjuvant therapies and targeted therapies are effective in eliminating cancer cells but in a limited number of patients. Increasing evidence suggests that the extracellular matrix (ECM) plays an important role in tumor development through changes in macromolecule components, degradation enzymes and stiffness. These variations are under the control of cellular components in tumor tissue via the aberrant activation of signaling pathways, the interaction of the ECM components to multiple surface receptors, and mechanical impact. Additionally, the ECM shaped by cancer regulates immune cells which results in an immune suppressive microenvironment and hinders the efficacy of immunotherapies. Thus, the ECM acts as a barrier to protect cancer from treatments and supports tumor progression. Nevertheless, the profound regulatory network of the ECM remodeling hampers the design of individualized antitumor treatment. Here, we elaborate on the composition of the malignant ECM, and discuss the specific mechanisms of the ECM remodeling. Precisely, we highlight the impact of the ECM remodeling on tumor development, including proliferation, anoikis, metastasis, angiogenesis, lymphangiogenesis, and immune escape. Finally, we emphasize ECM "normalization" as a potential strategy for anti-malignant treatment.

Collagen-Based Biomimetic Systems to Study the Biophysical Tumour Microenvironment

The extracellular matrix (ECM) is a pericellular network of proteins and other molecules that provides mechanical support to organs and tissues. ECM biophysical properties such as topography, elasticity and porosity strongly influence cell proliferation, differentiation and migration. The cell's perception of the biophysical microenvironment (mechanosensing) leads to altered gene expression or contractility status (mechanotransduction). Mechanosensing and mechanotransduction have profound implications in both tissue homeostasis and cancer. Many solid tumours are surrounded by a dense and aberrant ECM that disturbs normal cell functions and makes certain areas of the tumour inaccessible to therapeutic drugs. Understanding the cell-ECM interplay may therefore lead to novel and more effective therapies. Controllable and reproducible cell culturing systems mimicking the ECM enable detailed investigation of mechanosensing and mechanotransduction pathways. Here, we discuss ECM biomimetic systems. Mainly focusing on collagen, we compare and contrast structural and molecular complexity as well as biophysical properties of simple 2D substrates, 3D fibrillar collagen gels, cell-derived matrices and complex decellularized organs. Finally, we emphasize how the integration of advanced methodologies and computational methods with collagen-based biomimetics will improve the design of novel therapies aimed at targeting the biophysical and mechanical features of the tumour ECM to increase therapy efficacy.

Dynamic Changes in the Extracellular Matrix in Primary, Metastatic, and Recurrent Ovarian Cancers

Cancer-associated fibroblasts (CAFs) and their extracellular matrix are active participants in cancer progression. While it is known that functionally different subpopulations of CAFs co-exist in ovarian cancer, it is unclear whether certain CAF subsets are enriched during metastatic progression and/or chemotherapy. Using computational image analyses of patient-matched primary high-grade serous ovarian carcinomas, synchronous pre-chemotherapy metastases, and metachronous post-chemotherapy metastases from 42 patients, we documented the dynamic spatiotemporal changes in the extracellular matrix, fibroblasts, epithelial cells, immune cells, and CAF subsets expressing different extracellular matrix components. Among the different CAF subsets, COL11A1+ CAFs were associated with linearized collagen fibers and exhibited the greatest enrichment in pre- and post-chemotherapy metastases compared to matched primary tumors. Although pre- and post-chemotherapy metastases were associated with increased CD8+ T cell infiltration, the infiltrate was not always evenly distributed between the stroma and cancer cells, leading to an increased frequency of the immune-excluded phenotype where the majority of CD8+ T cells are present in the tumor stroma but absent from the tumor parenchyma. Overall, most of the differences in the tumor microenvironment were observed between primary tumors and metastases, while fewer differences were observed between pre- and post-treatment metastases. These data suggest that the tumor microenvironment is largely determined by the primary vs. metastatic location of the tumor while chemotherapy does not have a significant impact on the host microenvironment.

Artificial intelligence for multimodal data integration in oncology

In oncology, the patient state is characterized by a whole spectrum of modalities, ranging from radiology, histology, and genomics to electronic health records. Current artificial intelligence (AI) models operate mainly in the realm of a single modality, neglecting the broader clinical context, which inevitably diminishes their potential. Integration of different data modalities provides opportunities to increase robustness and accuracy of diagnostic and prognostic models, bringing AI closer to clinical practice. AI models are also capable of discovering novel patterns within and across modalities suitable for explaining differences in patient outcomes or treatment resistance. The insights gleaned from such models can guide exploration studies and contribute to the discovery of novel biomarkers and therapeutic targets. To support these advances, here we present a synopsis of AI methods and strategies for multimodal data fusion and association discovery. We outline approaches for AI interpretability and directions for AI-driven exploration through multimodal data interconnections. We examine challenges in clinical adoption and discuss emerging solutions.

Computational tumor stroma reaction evaluation led to novel prognosis-associated fibrosis and molecular signature discoveries in high-grade serous ovarian carcinoma

Background

As one of the key criteria to differentiate benign vs. malignant tumors in ovarian and other solid cancers, tumor-stroma reaction (TSR) is long observed by pathologists and has been found correlated with patient prognosis. However, paucity of study aims to overcome subjective bias or automate TSR evaluation for enabling association analysis to a large cohort.

Materials and methods

Serving as positive and negative sets of TSR studies, H&E slides of primary tumors of high-grade serous ovarian carcinoma (HGSOC) (n = 291) and serous borderline ovarian tumor (SBOT) (n = 15) were digitally scanned. Three pathologist-defined quantification criteria were used to characterize the extents of TSR. Scores for each criterion were annotated (0/1/2 as none-low/intermediate/high) in the training set consisting of 18,265 H&E patches. Serial of deep learning (DL) models were trained to identify tumor vs. stroma regions and predict TSR scores. After cross-validation and independent validations, the trained models were generalized to the entire HGSOC cohort and correlated with clinical characteristics. In a subset of cases tumor transcriptomes were available, gene- and pathway-level association studies were conducted with TSR scores.

Results

The trained models accurately identified the tumor stroma tissue regions and predicted TSR scores. Within tumor stroma interface region, TSR fibrosis scores were strongly associated with patient prognosis. Cancer signaling aberrations associated 14 KEGG pathways were also found positively correlated with TSR-fibrosis score.

Conclusion

With the aid of DL, TSR evaluation could be generalized to large cohort to enable prognostic association analysis and facilitate discovering novel gene and pathways associated with disease progress.

Computational pathology in ovarian cancer

Histopathologic evaluations of tissue sections are key to diagnosing and managing ovarian cancer. Pathologists empirically assess and integrate visual information, such as cellular density, nuclear atypia, mitotic figures, architectural growth patterns, and higher-order patterns, to determine the tumor type and grade, which guides oncologists in selecting appropriate treatment options. Latent data embedded in pathology slides can be extracted using computational imaging. Computers can analyze digital slide images to simultaneously quantify thousands of features, some of which are visible with a manual microscope, such as nuclear size and shape, while others, such as entropy, eccentricity, and fractal dimensions, are quantitatively beyond the grasp of the human mind. Applications of artificial intelligence and machine learning tools to interpret digital image data provide new opportunities to explore and quantify the spatial organization of tissues, cells, and subcellular structures. In comparison to genomic, epigenomic, transcriptomic, and proteomic patterns, morphologic and spatial patterns are expected to be more informative as quantitative biomarkers of complex and dynamic tumor biology. As computational pathology is not limited to visual data, nuanced subvisual alterations that occur in the seemingly “normal” pre-cancer microenvironment could facilitate research in early cancer detection and prevention. Currently, efforts to maximize the utility of computational pathology are focused on integrating image data with other -omics platforms that lack spatial information, thereby providing a new way to relate the molecular, spatial, and microenvironmental characteristics of cancer. Despite a dire need for improvements in ovarian cancer prevention, early detection, and treatment, the ovarian cancer field has lagged behind other cancers in the application of computational pathology. The intent of this review is to encourage ovarian cancer research teams to apply existing and/or develop additional tools in computational pathology for ovarian cancer and actively contribute to advancing this important field.

The state of the art for artificial intelligence in lung digital pathology

Lung diseases carry a significant burden of morbidity and mortality worldwide. The advent of digital pathology (DP) and an increase in computational power have led to the development of artificial intelligence (AI)-based tools that can assist pathologists and pulmonologists in improving clinical workflow and patient management. While previous works have explored the advances in computational approaches for breast, prostate, and head and neck cancers, there has been a growing interest in applying these technologies to lung diseases as well. The application of AI tools on radiology images for better characterization of indeterminate lung nodules, fibrotic lung disease, and lung cancer risk stratification has been well documented. In this article, we discuss methodologies used to build AI tools in lung DP, describing the various hand-crafted and deep learning-based unsupervised feature approaches. Next, we review AI tools across a wide spectrum of lung diseases including cancer, tuberculosis, idiopathic pulmonary fibrosis, and COVID-19. We discuss the utility of novel imaging biomarkers for different types of clinical problems including quantification of biomarkers like PD-L1, lung disease diagnosis, risk stratification, and prediction of response to treatments such as immune checkpoint inhibitors. We also look briefly at some emerging applications of AI tools in lung DP such as multimodal data analysis, 3D pathology, and transplant rejection. Lastly, we discuss the future of DP-based AI tools, describing the challenges with regulatory approval, developing reimbursement models, planning clinical deployment, and addressing AI biases. © 2022 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Poly (L-Lactic Acid) Cell-Laden Scaffolds Applied on Swine Model of Tracheal Fistula

INTRODUCTION

Tracheal fistula (TF) treatments may involve temporary orthosis and further ablative procedures, which can lead to infection. Thus, TF requires other therapy alternatives development. The hypothesis of this work was to demonstrate the feasibility of a tissue-engineered alternative for small TF in a preclinical model. Also, its association with suture filaments enriched with adipose tissue-derived mesenchymal stromal stem cells (AT-MSCs) was assessed to determine whether it could optimize the regenerative process.

METHODS

Poly (L-Lactic acid) (PLLA) membranes were manufactured by electrospinning and had morphology analyzed by scanning electron microscopy. AT-MSCs were cultured in these scaffolds and in vitro assays were performed (cytotoxicity, cellular adhesion, and viability). Subsequently, these cellular constructs were implanted in an animal small TF model. The association with suture filaments containing attached AT-MSCs was present in one animal group. After 30 d, animals were sacrificed and regenerative potential was evaluated, mainly related to the extracellular matrix remodeling, by performing histopathological (Hematoxylin-Eosin and trichrome Masson) and immunohistochemistry (Collagen I/II/III, matrix metalloproteinases-2, matrix metalloproteinases-9, vascular endothelial growth factor, and interleukin-10) analyses.

RESULTS

PLLA membranes presented porous fibers, randomly oriented. In vitro assays results showed that AT-MSCs attached were viable and maintained an active metabolism. Swine implanted with AT-MSCs attached to membranes and suture filaments showed aligned collagen fibers and a better regenerative progress in 30 d.

CONCLUSIONS

PLLA membranes with AT-MSCs attached were useful to the extracellular matrix restoration and have a high potential for small TF treatment. Also, their association with suture filaments enriched with AT-MSCs was advantageous.

Second-Harmonic Generation Imaging Reveals Changes in Breast Tumor Collagen Induced by Neoadjuvant Chemotherapy

Breast cancer is the most common invasive cancer in women, with most deaths attributed to metastases. Neoadjuvant chemotherapy (NACT) may be prescribed prior to surgical removal of the tumor for subsets of breast cancer patients but can have diverse undesired and off-target effects, including the increased appearance of the 'tumor microenvironment of metastasis', image-based multicellular signatures that are prognostic of breast tumor metastasis. To assess whether NACT can induce changes in two other image-based prognostic/predictive signatures derived from tumor collagen, we quantified second-harmonic generation (SHG) directionality and fiber alignment in formalin-fixed, paraffin-embedded sections of core needle biopsies and primary tumor excisions from 22 human epidermal growth factor receptor 2-overexpressing (HER2+) and 22 triple-negative breast cancers. In both subtypes, we found that SHG directionality (i.e., the forward-to-backward scattering ratio, or F/B) is increased by NACT in the bulk of the tumor, but not the adjacent tumor-stroma interface. Overall collagen fiber alignment is increased by NACT in triple-negative but not HER2+ breast tumors. These results suggest that NACT impacts the collagenous extracellular matrix in a complex and subtype-specific manner, with some prognostic features being unchanged while others are altered in a manner suggestive of a more metastatic phenotype.

Collagen fiber orientation disorder from H&E images is prognostic for early stage breast cancer: clinical trial validation

Collagen fiber organization has been found to be implicated in breast cancer prognosis. In this study, we evaluated whether computerized features of Collagen Fiber Orientation Disorder in Tumor-associated Stroma (CFOD-TS) on Hematoxylin & Eosin (H&E) slide images were prognostic of Disease Free Survival (DFS) in early stage Estrogen Receptor Positive (ER+) Invasive Breast Cancers (IBC). A Cox regression model named M_CFOD-TS, was constructed using cohort S_t (N = 78) to predict DFS based on CFOD-TS features. The prognostic performance of M_CFOD-TS was validated on cohort S_v (N = 219), a prospective clinical trial dataset (ECOG 2197). M_CFOD-TS was prognostic of DFS in both S_t and S_v, independent of clinicopathological variables. Additionally, the molecular pathways regarding cell cycle regulation were identified as being significantly associated with M_CFOD-TS derived risk scores. Our results also found that collagen fiber organization was more ordered in patients with short DFS. Our study provided a H&E image-based pipeline to derive a potential prognostic biomarker for early stage ER+ IBC without the need of special collagen staining or advanced microscopy techniques.