Search

The benefits of AI in pathology are demonstrated by the growing potential of diagnostic accuracy and efficiency in tissue segmentation, cancer detection and grading. In this scoping review, we studied how artificial intelligence (AI) technologies have been developed for usage in pathology. In the study, a literature search on databases and publisher portals PubMed, ScienceDirect, ProQuest, and CINAHL was conducted to collect peer-reviewed studies published between 2019 and 2023. Then after a thorough database search, 57 articles covering the topic were included out of 19 185 potential ones and were categorised into groups by their main purposes, such as pre-processing, segmentation, analysis, and prediction. The findings demonstrate that AI technologies like deep learning and convolutional neural networks (CNNs) enhance medical imaging for tissue segmentation and cancer detection by reducing human error and variability.

IntroductionDeep learning (DL) models predicting biomarker expression in images of hematoxylin and eosin (H&E)-stained tissues can improve access to multi-marker immunophenotyping, crucial for therapeutic monitoring, biomarker discovery, and personalized treatment development. Conventionally, these models are trained on ground truth cell labels derived from IHC-stained tissue sections adjacent to H&E-stained ones, which might be less accurate than labels from the same section. Although many such DL models have been developed, the impact of ground truth cell label derivation methods on their performance has not been studied.MethodologyIn this study, we assess the impact of cell label derivation on H&E model performance, with CD3+ T-cells in lung cancer tissues as a proof-of-concept. We compare two Pix2Pix generative adversarial network (P2P-GAN)-based virtual staining models: one trained with cell labels obtained from the same tissue section as the H&E-stained section (the ‘same-section’ model) and one trained on cell labels from an adjacent tissue section (the ‘serial-section’ model).ResultsWe show that the same-section model exhibited significantly improved prediction performance compared to the ‘serial-section’ model. Furthermore, the same-section model outperformed the serial-section model in stratifying lung cancer patients within a public lung cancer cohort based on survival outcomes, demonstrating its potential clinical utility.DiscussionCollectively, our findings suggest that employing ground truth cell labels obtained through the same-section approach boosts immunophenotyping DL solutions.

Inflammasome is linked to many inflammatory diseases, including COVID-19 and autoimmune liver diseases. While severe COVID-19 was reported to exacerbate liver failure, we report a fatal acute-on-chronic liver failure (ACLF) in a stable primary biliary cholangitis-autoimmune hepatitis overlap syndrome patient triggered by a mild COVID-19 infection. Postmortem liver biopsy showed sparse SARS-CoV-2-infected macrophages with extensive ASC (apoptosis-associated speck-like protein containing a CARD) speck-positive hepatocytes, correlating with elevated circulating ASC specks and inflammatory cytokines, and depleted blood monocyte subsets, indicating widespread liver inflammasome activation. This first report of a fatal inflammatory cascade in an autoimmune liver disease triggered by a mild remote viral infection hopes to elucidate a less-described pathophysiology of ACLF that could prompt consideration of new diagnostic and therapeutic options.

Abstract

Abstract Angiosarcomas are rare malignant tumors of the endothelium, arising commonly from the head and neck region (AS-HN) and recently associated with ultraviolet (UV) exposure and human herpesvirus-7 infection. We examined 81 cases of angiosarcomas, including 47 cases of AS-HN, integrating information from whole genome sequencing, gene expression profiling and spatial transcriptomics (10X Visium). In the AS-HN cohort, we observed recurrent somatic mutations in CSMD3 (18%), LRP1B (18%), MUC16 (18%), POT1 (16%) and TP53 (16%). UV-positive AS-HN harbored significantly higher tumor mutation burden than UV-negative cases (p = 0.0294). NanoString profiling identified three clusters with distinct tumor inflammation signature scores (p

Abstract Spatially resolved transcriptomics involves a set of emerging technologies that enable the transcriptomic profiling of tissues with the physical location of expressions. Although a variety of methods have been developed for data integration, most of them are for single-cell RNA-seq datasets without consideration of spatial information. Thus, methods that can integrate spatial transcriptomics data from multiple tissue slides, possibly from multiple individuals, are needed. Here, we present PRECAST, a data integration method for multiple spatial transcriptomics datasets with complex batch effects and/or biological effects between slides. PRECAST unifies spatial factor analysis simultaneously with spatial clustering and embedding alignment, while requiring only partially shared cell/domain clusters across datasets. Using both simulated and four real datasets, we show improved cell/domain detection with outstanding visualization, and the estimated aligned embeddings and cell/domain labels facilitate many downstream analyses. We demonstrate that PRECAST is computationally scalable and applicable to spatial transcriptomics datasets from different platforms.

Background Combination therapy with radioembolization (yttrium-90)-resin microspheres) followed by nivolumab has shown a promising response rate of 30.6% in a Phase II trial (CA209-678) for advanced hepatocellular carcinoma (HCC); however, the response mechanisms and relevant biomarkers remain unknown.Methods By collecting both pretreatment and on-treatment samples, we performed multimodal profiling of tissue and blood samples and investigated molecular changes associated with favorable responses in 33 patients from the trial.Results We found that higher tumor mutation burden, NCOR1 mutations and higher expression of interferon gamma pathways occurred more frequently in responders. Meanwhile, non-responders tended to be enriched for a novel Asian-specific transcriptomic subtype (Kaya_P2) with a high frequency of chromosome 16 deletions and upregulated cell cycle pathways. Strikingly, unlike other cancer types, we did not observe any association between T-cell populations and treatment response, but tumors from responders had a higher proportion of CXCL9+/CXCR3+ macrophages. Moreover, biomarkers discovered in previous immunotherapy trials were not predictive in the current cohort, suggesting a distinctive molecular landscape associated with differential responses to the combination therapy.Conclusions This study unraveled extensive molecular changes underlying distinctive responses to the novel treatment and pinpointed new directions for harnessing combination therapy in patients with advanced HCC.

IntroductionImmune checkpoint blockade (ICB) is a systemic therapeutic option for advanced hepatocellular carcinoma (HCC). However, low patient response rates necessitate the development of robust predictive biomarkers that identify individuals who will benefit from ICB. A 4-gene inflammatory signature, comprising CD8, PD-L1, LAG-3, and STAT1, was recently shown to be associated with a better overall response to ICB in various cancer types. Here, we examined whether tissue protein expression of CD8, PD-L1, LAG-3, and STAT1 predicts response to ICB in HCC.MethodsHCC samples from 191 Asian patients, comprising resection specimens from 124 patients (ICB-naïve) and pre-treatment specimens from 67 advanced HCC patients treated with ICB (ICB-treated), were analyzed for CD8, PD-L1, LAG-3, and STAT1 tissue expression using multiplex immunohistochemistry followed by statistical and survival analyses.ResultsImmunohistochemical and survival analyses of ICB-naïve samples showed that high LAG-3 expression was associated with shorter median progression-free survival (mPFS) and overall survival (mOS). Analysis of ICB-treated samples revealed that high proportions of LAG-3+ and LAG-3+CD8+ cells pre-treatment were most closely associated with longer mPFS and mOS. Using a log-likelihood model, adding the total LAG-3+ cell proportion to the total CD8+ cell proportion significantly increased the predictive values for mPFS and mOS, compared with the total CD8+ cell proportion alone. Moreover, levels of CD8 and STAT1, but not PD-L1, were significantly correlated with better responses to ICB. After analyzing viral-related and non-viral HCC samples separately, only the LAG3+CD8+ cell proportion was significantly associated with responses to ICB regardless of viral status.ConclusionImmunohistochemical scoring of pre-treatment levels of LAG-3 and CD8 in the tumor microenvironment may help predict ICB benefits in HCC patients. Furthermore, immunohistochemistry-based techniques offer the advantage of being readily translatable in the clinical setting.

In order to design cancer immune therapies, it is important to understand how tumours evade the immune response that is mounted against them. Authors here analyse the distribution and properties of immune cells in hepatocellular carcinoma and describe a progressive tumour-immune co-evolution programme from early to late stage cancer.

Abstract Angiosarcomas are a rare subtype of soft-tissue sarcomas which exhibit aggressive clinical phenotypes with limited treatment options and poor outcomes. In this study, we investigated the clinical relevance of the peripheral blood neutrophil-to-lymphocyte ratio (NLR) as a marker of systemic immune response, as well as its correlation with intra-tumoral immune profiles in a subgroup of cases (n = 35) using the NanoString PanCancer IO360 panel and multiplex immunohistochemistry. In the overall cohort (n = 150), angiosarcomas of the head and neck (AS-HN) comprised most cases (58.7%) and median overall survival (OS) was 1.1 year. NLR, classified as high in 78 of 112 (70%) evaluable patients, was independently correlated with worse OS (HR 1.84, 95%CI 1.18–2.87, p = 0.0073). Peripheral blood NLR was positively correlated with intra-tumoral NLR (tNLR) (Spearman’s rho 0.450, p = 0.0067). Visualization of tumor-infiltrating immune cells confirmed that tNLR scores correlated directly with both neutrophil (CD15+ cells, rho 0.398, p = 0.0198) and macrophage (CD68+ cells, rho 0.515, p = 0.0018) cell counts. Interestingly, tNLR correlated positively with oncogenic pathway scores including angiogenesis, matrix remodeling and metastasis, and cytokine and chemokine signaling, as well as myeloid compartment scores (all p

Nasopharyngeal carcinoma (NPC) lacks effective diagnostic and therapeutic strategies, in particular at advanced stages. Here, the authors show that expression of the somatostatin receptor 2 is induced by Epstein-Barr virus in NPC and has a key role in the diagnosis, imaging, targeted therapies and prognosis of NPC.

Abstract Background Stromal and collagen biology has a significant impact on tumorigenesis and metastasis. Collagen is a major structural extracellular matrix component in breast cancer, but its role in cancer progression is the subject of historical debate. Collagen may represent a protective layer that prevents cancer cell migration, while increased stromal collagen has been demonstrated to facilitate breast cancer metastasis. Methods Stromal remodeling is characterized by collagen fiber restructuring and realignment in stromal and tumoral areas. The patients in our study were diagnosed with triple-negative breast cancer in Singapore General Hospital from 2003 to 2015. We designed novel image processing and quantification pipelines to profile collagen structures using numerical imaging parameters. Our solution differentiated the collagen into two distinct modes: aggregated thick collagen (ATC) and dispersed thin collagen (DTC). Results Extracted parameters were significantly associated with bigger tumor size and DCIS association. Of numerical parameters, ATC collagen fiber density (CFD) and DTC collagen fiber length (CFL) were of significant prognostic value for disease-free survival and overall survival for the TNBC patient cohort. Using these two parameters, we built a predictive model to stratify the patients into four groups. Conclusions Our study provides a novel insight for the quantitation of collagen in the tumor microenvironment and will help predict clinical outcomes for TNBC patients. The identified collagen parameters, ATC CFD and DTC CFL, represent a new direction for clinical prognosis and precision medicine. We also compared our result with benign samples and DICS samples to get novel insight about the TNBC heterogeneity. The improved understanding of collagen compartment of TNBC may provide insights into novel targets for better patient stratification and treatment.

Coronavirus disease-19 (COVID-19) is caused by the newly discovered coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). While the lung remains the primary target site of COVID-19 injury, damage to myocardium, and other organs also contribute to the morbidity and mortality of this disease. There is also increasing demand to visualize viral components within tissue specimens. Here we discuss the cardiac autopsy findings of 12 intensive care unit (ICU) naïve and PCR-positive COVID-19 cases using a combination of histological, Immunohistochemical/immunofluorescent and molecular techniques. We performed SARS-CoV-2 qRT-PCR on fresh tissue from all cases; RNA-ISH and IHC for SARS-CoV-2 were performed on selected cases using FFPE tissue from heart. Eight of these patients also had positive post-mortem serology for SARS-CoV-2. Histopathologic changes in the coronary vessels and inflammation of the myocardium as well as in the endocardium were documented which support the reports of a cardiac component to the viral infection. As in the pulmonary reports, widespread platelet and fibrin thrombi were also identified in the cardiac tissue. In keeping with vaccine-induced activation of virus-specific CD4+ and CD8+ T cells, and release of cytokines such as interferon-gamma (IFNγ), we observed similar immune cellular distribution and cytokines in these patients. Immunohistochemical and immunofluorescent localisation for the viral Spike (S-protein) protein and the nucleocapsid protein (NP) were performed; presence of these aggregates may possibly contribute to cardiac ischemia and even remodelling.

Abstract The role of programmed cell death protein-1 (PD-1)/programmed cell death ligand 1 (PD-L1) in triple negative breast cancer (TNBC) remains to be fully understood. In this study, we investigated the role of PD-1 as a prognostic marker for TNBC in an Asian cohort (n = 269). Samples from patients with TNBC were labeled with antibodies against PD-L1 and PD-1, and subjected to NanoString assays to measure the expression of immune-related genes. Associations between disease-free survival (DFS), overall survival (OS) and biomarker expression were investigated. Multivariate analysis showed that tumors with high PD-1+ immune infiltrates harbored significantly increased DFS, and this increase was significant even after controlling for clinicopathological parameters (HR 0.95; P = 0.030). In addition, the density of cells expressing both CD8 and PD-1, but not the density of CD8−PD-1+ immune infiltrates, was associated with improved DFS. Notably, this prognostic significance was independent of clinicopathological parameters and the densities of total CD8+ cell (HR 0.43, P = 0.011). At the transcriptional level, high expression of PDCD1 within the tumor was significantly associated with improved DFS (HR 0.38; P = 0.027). In line with these findings, high expression of IFNG (HR 0.38; P = 0.001) and IFN signaling genes (HR 0.46; p = 0.027) was also associated with favorable DFS. Inclusion of PD-1 immune infiltrates and PDCD1 gene expression added significant prognostic value for DFS (ΔLRχ2 = 6.35; P = 0.041) and OS (ΔLRχ2 = 9.53; P = 0.008), beyond that provided by classical clinicopathological variables. Thus, PD-1 mRNA and protein expression status represent a promising, independent indicator of prognosis in TNBC.

Intravesical Bacillus Calmette-Guerin (BCG) is an effective immunotherapy for non-muscle invasive bladder cancer (NMIBC). However, recurrence and progression remain frequent warranting deeper insights into its mechanism. We herein comprehensively profiled blood and tissues obtained from NMIBC patients before, during and after BCG treatment using cytometry by time-of-flight (CyTOF) and RNA sequencing to identify the key immune subsets crucial for anti-tumor activity. We observed the temporal changes of peripheral immune subsets including NKT cells, central memory CD4+ T cells, CD8+ T cells and regulatory T cells (Treg) during the course of BCG. Gene expression analysis revealed enriched immune pathways involving in T cell activation and chemotaxis, as well as a more diversified T cell receptor repertoire in post-BCG tissues. Moreover, tissue multiplexed-immunofluorescence (mIF) showed baseline densities of non-Treg and CD8+PD-1+ T cells were predictive of response and better recurrence-free survival after BCG. Remarkably, post-BCG tissues from responders were found to be infiltrated with more active CD8+PD-1- T cells and non-Treg CD4+FOXP3- T cells; but increased exhausted CD8+PD-1+ T cells were found in non-responders. Taken together, we identified predictive biomarkers for response and uncovered the post-treatment expansion of exhausted PD-1+CD8+ T cells as key to BCG resistance, which could potentially be restored by combining with anti-PD-1 immunotherapy.

Introduction Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer-associated mortality globally. Immune-checkpoint blockade (ICB) is one of the systemic therapy options for HCC. However, response rates remain low, necessitating robust predictive biomarkers. In the present study, we examined the expression of CD38, a molecule involved in the immunosuppressive adenosinergic pathway, on immune cells present in the tumor microenvironment. We then investigated the association between CD38 and ICB treatment outcomes in advanced HCC.Methods Clinically annotated samples from 49 patients with advanced HCC treated with ICB were analyzed for CD38 expression using immunohistochemistry (IHC), multiplex immunohistochemistry/immunofluorescence (mIHC/IF) and multiplex cytokine analysis.Results IHC and mIHC/IF analyses revealed that higher intratumoral CD38+ cell proportion was strongly associated with improved response to ICB. The overall response rates to ICB was significantly higher among patients with high proportion of total CD38+cells compared with patients with low proportion (43.5% vs 3.9%, p=0.019). Higher responses seen among patients with a high intratumoral CD38+cell proportion translated to a longer median progression-free survival (mPFS, 8.21 months vs 1.64 months, p=0.0065) and median overall survival (mOS, 19.06 months vs 9.59 months, p=0.0295). Patients with high CD38+CD68+macrophage density had a better mOS of 34.43 months compared with 9.66 months in patients with low CD38+CD68+ macrophage density. CD38hi macrophages produce more interferon γ (IFN-γ) and related cytokines, which may explain its predictive value when treated with ICB.Conclusions A high proportion of CD38+ cells, determined by IHC, predicts response to ICB and is associated with superior mPFS and OS in advanced HCC.

Abstract Background Expression quantitative trait loci (eQTL) databases represent a valuable resource to link disease-associated SNPs to specific candidate genes whose gene expression is significantly modulated by the SNP under investigation. We previously identified signal inhibitory receptor on leukocytes-1 (SIRL-1) as a powerful regulator of human innate immune cell function. While it is constitutively high expressed on neutrophils, on monocytes the SIRL-1 surface expression varies strongly between individuals. The underlying mechanism of regulation, its genetic control as well as potential clinical implications had not been explored yet. Methods Whole blood eQTL data of a Chinese cohort was used to identify SNPs regulating the expression of VSTM1, the gene encoding SIRL-1. The genotype effect was validated by flow cytometry (cell surface expression), correlated with electrophoretic mobility shift assay (EMSA), chromatin immunoprecipitation (ChIP) and bisulfite sequencing (C-methylation) and its functional impact studied the inhibition of reactive oxygen species (ROS). Results We found a significant association of a single CpG-SNP, rs612529T/C, located in the promoter of VSTM1. Through flow cytometry analysis we confirmed that primarily in the monocytes the protein level of SIRL-1 is strongly associated with genotype of this SNP. In monocytes, the T allele of this SNP facilitates binding of the transcription factors YY1 and PU.1, of which the latter has been recently shown to act as docking site for modifiers of DNA methylation. In line with this notion rs612529T associates with a complete demethylation of the VSTM1 promoter correlating with the allele-specific upregulation of SIRL-1 expression. In monocytes, this upregulation strongly impacts the IgA-induced production of ROS by these cells. Through targeted association analysis we found a significant Meta P value of 1.14 × 10–6 for rs612529 for association to atopic dermatitis (AD). Conclusion Low expression of SIRL-1 on monocytes is associated with an increased risk for the manifestation of an inflammatory skin disease. It thus underlines the role of both the cell subset and this inhibitory immune receptor in maintaining immune homeostasis in the skin. Notably, the genetic regulation is achieved by a single CpG-SNP, which controls the overall methylation state of the promoter gene segment.

Background: CD38 is involved in the adenosine pathway, which represents one of the immunosuppressive mechanisms in cancer. CD38 is broadly expressed across immune cell subsets, including human macrophages differentiated in vitro from monocytes, but expression by tissue-resident macrophages remains to be demonstrated.Methods: Tissue samples were obtained from 66 patients with hepatocellular carcinoma (HCC) from Singapore and analyzed using immunohistochemistry. Tumor-infiltrating leukocytes (TILs) were further examined using DEPArray™, and the phenotype of freshly isolated TILs was determined using flow cytometry.Results: CD38 was frequently co-expressed with the macrophage-specific marker CD68. CD38+CD68+ macrophage density was associated with improved prognosis after surgery, while total CD68+ macrophage density was associated with poor prognosis. DEPArray™ analysis revealed the presence of large (>10 μm), irregularly shaped CD45+CD14+ cells that resembled macrophages, with concurrent CD38+ expression. Flow cytometry also revealed that majority of CD14+HLA-DR+ cells expressed CD38.Conclusion: CD38 expression was clearly demonstrated on human macrophages in an in vivo setting. The positive association identified between CD38+ macrophage density and prognosis may have implications for routine diagnostic work.

Breast cancer is the most common malignancy affecting women, but the heterogeneity of the condition is a significant obstacle to effective treatment. Triple negative breast cancers (TNBCs) do not express HER2 or the receptors for estrogen or progesterone, and so often have a poor prognosis. Tumor-infiltrating T cells have been well-characterized in TNBC, and increased numbers are associated with better outcomes; however, the potential roles of B cells and plasma cells have been large. Here, we conducted a retrospective correlative study on the expression of B cell/plasma cell-related genes, and the abundance and localization of B cells and plasma cells within TNBCs, and clinical outcome. We analyzed 269 TNBC samples and used immunohistochemistry to quantify tumor-infiltrating B cells and plasma cells, coupled with NanoString measurement of expression of immunoglobulin metagenes. Multivariate analysis revealed that patients bearing TNBCs with above-median densities of CD38+ plasma cells had significantly better disease-free survival (DFS) (HR = 0.44; 95% CI 0.26–0.77; p = 0.004) but not overall survival (OS), after adjusting for the effects of known prognostic factors. In contrast, TNBCs with higher immunoglobulin gene expression exhibited improved prognosis (OS p = 0.029 and DFS p = 0.005). The presence of B cells and plasma cells was positively correlated (p

Introduction: Deep learning (DL) models predicting biomarker expression in images of hematoxylin and eosin (H&E)-stained tissues can improve access to multi-marker immunophenotyping, crucial for therapeutic monitoring, biomarker discovery, and personalized treatment development. Conventionally, these models are trained on ground truth cell labels derived from IHC-stained tissue sections adjacent to H&Estained ones, which might be less accurate than labels from the same section. Although many such DL models have been developed, the impact of ground truth cell label derivation methods on their performance has not been studied. Methodology: In this study, we assess the impact of cell label derivation on H&E model performance, with CD3+ T-cells in lung cancer tissues as a proof-ofconcept. We compare two Pix2Pix generative adversarial network (P2P-GAN)- based virtual staining models: one trained with cell labels obtained from the same tissue section as the H&E-stained section (the 'same-section' model) and one trained on cell labels from an adjacent tissue section (the 'serial-section' model). Results: We show that the same-section model exhibited significantly improved prediction performance compared to the 'serial-section' model. Furthermore, the same-section model outperformed the serial-section model in stratifying lung cancer patients within a public lung cancer cohort based on survival outcomes, demonstrating its potential clinical utility. Discussion: Collectively, our findings suggest that employing ground truth cell labels obtained through the same-section approach boosts immunophenotyping DL solutions. [ABSTRACT FROM AUTHOR]