Validation of prostate and breast cancer detection artificial intelligence algorithms for accurate histopathological diagnosis and grading: a retrospective study with a Japanese cohort.

Prostate and breast cancer incidence rates have been on the rise in Japan, emphasising the need for precise histopathological diagnosis to determine patient prognosis and guide treatment decisions. However, existing diagnostic methods face numerous challenges and are susceptible to inconsistencies between observers. To tackle these issues, artificial intelligence (AI) algorithms have been developed to aid in the diagnosis of prostate and breast cancer. This study focuses on validating the performance of two such algorithms, Galen Prostate and Galen Breast, in a Japanese cohort, with a particular focus on the grading accuracy and the ability to differentiate between invasive and non-invasive tumours. The research entailed a retrospective examination of 100 consecutive prostate and 100 consecutive breast biopsy cases obtained from a Japanese institution. Our findings demonstrated that the AI algorithms showed accurate cancer detection, with AUCs of 0.969 and 0.997 for the Galen Prostate and Galen Breast, respectively. The Galen Prostate was able to detect a higher Gleason score in four adenocarcinoma cases and detect a previously unreported cancer. The two algorithms successfully identified relevant pathological features, such as perineural invasions and lymphovascular invasions. Although further improvements are required to accurately differentiate rare cancer subtypes, these findings highlight the potential of these algorithms to enhance the precision and efficiency of prostate and breast cancer diagnosis in Japan. Furthermore, this validation paves the way for broader adoption of these algorithms as decision support tools within the Asian population.

Antimicrobial Resistance Patterns of Common Gram-Negative Microorganisms Isolated from Patients with Lower Respiratory Tract Infections in a Teaching Hospital in Vietnam.

This cross-sectional study investigated the antimicrobial resistance (AMR) patterns of gram-negative pathogens isolated from 4,789 hospitalized patients with lower respiratory tract infections (LRTIs). Of the collected specimens, 1,325 (27.7%) tested positive for gram-negative bacteria. Acinetobacter baumannii (38.6%), Pseudomonas aeruginosa (33.5%), Klebsiella pneumoniae (18.7%), Escherichia coli (5.6%), and Klebsiella aerogenes (3.5%) were the most prevalent isolates. AMR analysis revealed high resistance rates (79.9%-100%) of A. baumannii isolates to multiple classes of antibiotics except amikacin, trimethoprim/sulfamethoxazole, and colistin. P. aeruginosa displayed low resistance to colistin (< 10%) but high resistance to other antibiotics. K. pneumoniae displayed high resistance rates of 90.0%-100.0% to most penicillins, whereas resistance rates were notably lower for colistin (7.1%) and amikacin (16.7%). K. aerogenes exhibited high resistance to various antibiotics and sensitivity to amikacin (95.1%), ampicillin (100.0%), and colistin (100.0%). E. coli isolates exhibited resistance to ampicillin (96.9%) and maximum sensitivity to several antibiotics. Our study identified significant AMR trends and highlighted the prevalence of multidrug-resistant strains (93.6% for K. aerogenes and 69.1%-92.4% for other isolates). These findings emphasize the urgent need for appropriate antibiotic management practices to combat AMR in gram-negative pathogens associated with LRTIs.

Comparative Analysis of Transcriptome and Proteome Revealed the Common Metabolic Pathways Induced by Prevalent ESBL Plasmids in Escherichia coli.

Antibiotic resistance has emerged as one of the most significant threats to global public health. Plasmids, which are highly efficient self-replicating genetic vehicles, play a critical role in the dissemination of drug-resistant genes. Previous studies have mainly focused on drug-resistant genes only, often neglecting the complete functional role of multidrug-resistant (MDR) plasmids in bacteria. In this study, we conducted a comprehensive investigation of the transcriptomes and proteomes of Escherichia coli J53 transconjugants harboring six major MDR plasmids of different incompatibility (Inc) groups, which were clinically isolated from patients. The RNA-seq analysis revealed that MDR plasmids influenced the gene expression in the bacterial host, in particular, the genes related to metabolic pathways. A proteomic analysis demonstrated the plasmid-induced regulation of several metabolic pathways including anaerobic respiration and the utilization of various carbon sources such as serine, threonine, sialic acid, and galactarate. These findings suggested that MDR plasmids confer a growth advantage to bacterial hosts in the gut, leading to the expansion of plasmid-carrying bacteria over competitors without plasmids. Moreover, this study provided insights into the versatility of prevalent MDR plasmids in moderating the cellular gene network of bacteria, which could potentially be utilized in therapeutics development for bacteria carrying MDR plasmids.

An international multi-institutional validation study of the algorithm for prostate cancer detection and Gleason grading.

Pathologic examination of prostate biopsies is time consuming due to the large number of slides per case. In this retrospective study, we validate a deep learning-based classifier for prostate cancer (PCA) detection and Gleason grading (AI tool) in biopsy samples. Five external cohorts of patients with multifocal prostate biopsy were analyzed from high-volume pathology institutes. A total of 5922 H&E sections representing 7473 biopsy cores from 423 patient cases (digitized using three scanners) were assessed concerning tumor detection. Two tumor-bearing datasets (core n = 227 and 159) were graded by an international group of pathologists including expert urologic pathologists (n = 11) to validate the Gleason grading classifier. The sensitivity, specificity, and NPV for the detection of tumor-bearing biopsies was in a range of 0.971-1.000, 0.875-0.976, and 0.988-1.000, respectively, across the different test cohorts. In several biopsy slides tumor tissue was correctly detected by the AI tool that was initially missed by pathologists. Most false positive misclassifications represented lesions suspicious for carcinoma or cancer mimickers. The quadratically weighted kappa levels for Gleason grading agreement for single pathologists was 0.62-0.80 (0.77 for AI tool) and 0.64-0.76 (0.72 for AI tool) for the two grading datasets, respectively. In cases where consensus for grading was reached among pathologists, kappa levels for AI tool were 0.903 and 0.855. The PCA detection classifier showed high accuracy for PCA detection in biopsy cases during external validation, independent of the institute and scanner used. High levels of agreement for Gleason grading were indistinguishable between experienced genitourinary pathologists and the AI tool.

The effect of carboxymethyl cellulose and ß-cyclodextrin as debittering agents on bitterness and physicochemical properties of bitter gourd extract.

Bitter gourd extract (BGE) is rich in antioxidants and anti-diabetic components that promote good human health; however, its bitter taste makes it challenging to use in food. In this study, the effect of carboxymethyl cellulose and ß-cyclodextrin (ß-CD) on the bitterness and properties of BGE were investigated. The bitterness intensity was evaluated by the trained sensory panel, and the physicochemical properties were also determined, including viscosity, total saponin, polyphenol content, antioxidant capacity, and α-amylase inhibition activity. It was found that the bitterness of BGE with 0.75%, w/v ß-cyclodextrin decreased significantly by more than 90%. Additionally, FTIR, 1 H-NMR, and thermogravimetric analysis of BGE supplemented with ß-CD confirmed the formation of a complex between ß-CD and components of BGE. The findings of the current study also reveal that debittering agents did not inhibit the bioactivities of BGE.

Comparative analysis of outer membrane vesicles from uropathogenic Escherichia coli reveal the role of aromatic amino acids synthesis proteins in motility.

Uropathogenic Escherichia coli (UPEC) are causative agent that causes urinary tract infections (UTIs) and the recent emergence of multidrug resistance (MDR) of UPEC increases the burden on the community. Recent studies of bacterial outer membrane vesicles (OMV) identified various factors including proteins, nucleic acids, and small molecules which provided inter-cellular communication within the bacterial population. However, the components of UPEC-specific OMVs and their functional role remain unclear. Here, we systematically determined the proteomes of UPEC-OMVs and identified the specific components that provide functions to the recipient bacteria. Based on the functional network of OMVs' proteomes, a group of signaling peptides was found in all OMVs which provide communication among bacteria. Moreover, we demonstrated that treatment with UPEC-OMVs affected the motility and biofilm formation of the recipient bacteria, and further identified aromatic amino acid (AAA) biosynthesis proteins as the key factors to provide their movement.

Evaluation of Three Feature Dimension Reduction Techniques for Machine Learning-Based Crop Yield Prediction Models.

Machine learning (ML) has been widely used worldwide to develop crop yield forecasting models. However, it is still challenging to identify the most critical features from a dataset. Although either feature selection (FS) or feature extraction (FX) techniques have been employed, no research compares their performances and, more importantly, the benefits of combining both methods. Therefore, this paper proposes a framework that uses non-feature reduction (All-F) as a baseline to investigate the performance of FS, FX, and a combination of both (FSX). The case study employs the vegetation condition index (VCI)/temperature condition index (TCI) to develop 21 rice yield forecasting models for eight sub-regions in Vietnam based on ML methods, namely linear, support vector machine (SVM), decision tree (Tree), artificial neural network (ANN), and Ensemble. The results reveal that FSX takes full advantage of the FS and FX, leading FSX-based models to perform the best in 18 out of 21 models, while 2 (1) for FS-based (FX-based) models. These FXS-, FS-, and FX-based models improve All-F-based models at an average level of 21% and up to 60% in terms of RMSE. Furthermore, 21 of the best models are developed based on Ensemble (13 models), Tree (6 models), linear (1 model), and ANN (1 model). These findings highlight the significant role of FS, FX, and specially FSX coupled with a wide range of ML algorithms (especially Ensemble) for enhancing the accuracy of predicting crop yield.

A collaborative workflow between pathologists and deep learning for the evaluation of tumour cellularity in lung adenocarcinoma.

AIMS: The reporting of tumour cellularity in cancer samples has become a mandatory task for pathologists. However, the estimation of tumour cellularity is often inaccurate. Therefore, we propose a collaborative workflow between pathologists and artificial intelligence (AI) models to evaluate tumour cellularity in lung cancer samples and propose a protocol to apply it to routine practice. METHODS AND RESULTS: We developed a quantitative model of lung adenocarcinoma that was validated and tested on 50 cases, and a collaborative workflow where pathologists could access the AI results and adjust their original tumour cellularity scores (adjusted-score) that we tested on 151 cases. The adjusted-score was validated by comparing them with a ground truth established by manual annotation of haematoxylin and eosin slides with reference to immunostains with thyroid transcription factor-1 and napsin A. For training, validation, testing the AI and testing the collaborative workflow, we used 40, 10, 50 and 151 whole slide images of lung adenocarcinoma, respectively. The sensitivity and specificity of tumour segmentation were 97 and 87%, respectively, and the accuracy of nuclei recognition was 99%. One pathologist's visually estimated scores were compared to the adjusted-score, and the pathologist's scores were altered in 87% of cases. Comparison with the ground truth revealed that the adjusted-score was more precise than the pathologists' scores (P < 0.05). CONCLUSION: We proposed a collaborative workflow between AI and pathologists as a model to improve daily practice and enhance the prediction of tumour cellularity for genetic tests.

Effect of 3-caffeoyl, 4-dihydrocaffeoylquinic acid from Salicornia herbacea on endothelial nitric oxide synthase activation via calcium signaling pathway.

3-Caffeoyl-4-dicaffeoylquinic acid (CDCQ) is a natural chlorogenic acid isolated from Salicornia herbacea that protects against oxidative stress, inflammation, and cancer. Nitric oxide (NO) plays a physiologically beneficial role in the cardiovascular system, including vasodilation, protection of endothelial cell function, and anti-inflammation. However, the effect of CDCQ on NO production and eNOS phosphorylation in endothelial cells is unclear. We investigated the effect of CDCQ on eNOS phosphorylation and NO production in human endothelial cells, and the underlying signaling pathway. CDCQ significantly increased NO production and the phosphorylation of eNOS at Ser1177. Additionally, CDCQ induced phosphorylation of PKA, CaMKII, CaMKKß, and AMPK. Interestingly, CDCQ increased the intracellular Ca2+ level, and L-type Ca2+ channel (LTCC) blockade significantly attenuated CDCQ-induced eNOS activity and NO production by inhibiting PKA, CaMKII, CaMKKß, and AMPK phosphorylation. These results suggest that CDCQ increased eNOS phosphorylation and NO production by Ca2+-dependent phosphorylation of PKA, CaMKII, CaMKKß, and AMPK. Our findings provide evidence that CDCQ plays a pivotal role in the activity of eNOS and NO production, which is involved in the protection of endothelial dysfunction.

Initial assessment of the spatial learning, reversal, and sequencing task capabilities of knock-in rats with humanizing mutations in the Aß-coding region of App.

Model organisms mimicking the pathogenesis of human diseases are useful for identifying pathogenic mechanisms and testing therapeutic efficacy of compounds targeting them. Models of Alzheimer's disease (AD) and related dementias (ADRD) aim to reproduce the brain pathology associated with these neurodegenerative disorders. Transgenic models, which involve random insertion of disease-causing genes under the control of artificial promoters, are efficient means of doing so. There are confounding factors associated with transgenic approaches, however, including target gene overexpression, dysregulation of endogenous gene expression at transgenes' integration sites, and limitations in mimicking loss-of-function mechanisms. Furthermore, the choice of species is important, and there are anatomical, physiological, and cognitive reasons for favoring the rat over the mouse, which has been the standard for models of neurodegeneration and dementia. We report an initial assessment of the spatial learning, reversal, and sequencing task capabilities of knock-in (KI) Long-Evans rats with humanizing mutations in the Aß-coding region of App, which encodes amyloid precursor protein (Apph/h rats), using the IntelliCage, an automated operant social home cage system, at 6-8 weeks of age, then again at 4-5 months of age. These rats were previously generated as control organisms for studies on neurodegeneration involving other knock-in rat models from our lab. Apph/h rats of either sex can acquire place learning and reversal tasks. They can also acquire a diagonal sequencing task by 6-8 weeks of age, but not a more advanced serial reversal task involving alternating diagonals, even by 4-5 months of age. Thus, longitudinal behavioral analysis with the IntelliCage system can be useful to determine, in follow-up studies, whether KI rat models of Familial AD (FAD), sporadic late onset AD (LOAD), and of ADRD develop aging-dependent learning and memory deficits.

Enhancing Crop Yield Prediction Utilizing Machine Learning on Satellite-Based Vegetation Health Indices.

Accurate crop yield forecasting is essential in the food industry's decision-making process, where vegetation condition index (VCI) and thermal condition index (TCI) coupled with machine learning (ML) algorithms play crucial roles. The drawback, however, is that a one-fits-all prediction model is often employed over an entire region without considering subregional VCI and TCI's spatial variability resulting from environmental and climatic factors. Furthermore, when using nonlinear ML, redundant VCI/TCI data present additional challenges that adversely affect the models' output. This study proposes a framework that (i) employs higher-order spatial independent component analysis (sICA), and (ii), exploits a combination of the principal component analysis (PCA) and ML (i.e., PCA-ML combination) to deal with the two challenges in order to enhance crop yield prediction accuracy. The proposed framework consolidates common VCI/TCI spatial variability into their respective subregions, using Vietnam as an example. Compared to the one-fits-all approach, subregional rice yield forecasting models over Vietnam improved by an average level of 20% up to 60%. PCA-ML combination outperformed ML-only by an average of 18.5% up to 45%. The framework generates rice yield predictions 1 to 2 months ahead of the harvest with an average of 5% error, displaying its reliability.

Amyotrophic Lateral Sclerosis and Noninvasive Positive Pressure Ventilatory Support: "Nasal Noninvasive Ventilation" or "Noninvasive Ventilatory Support"?

ABSTRACT: Many studies suggest a brief statistical benefit on survival and quality of life by using nasal noninvasive ventilation for patients with amyotrophic lateral sclerosis and other neuromuscular conditions. Indeed, nasal noninvasive ventilation has become synonymous with continuous positive airway pressure and lo-span bilevel positive airway pressure. Nasal noninvasive ventilation, however, may not normalize CO2 levels and continuous positive airway pressure and O2 exacerbate hypercapnia and often lead to CO2 narcosis, intubation, and ultimately tracheostomy or palliative care death. However, a third option can be to offer up to continuous noninvasive ventilatory support and extubation to it. Noninvasive ventilatory support can be effective for full, definitive ventilatory support, even for people with no measurable vital capacity, and has maintained classic amyotrophic lateral sclerosis patients for up to 12 yrs without resort to tracheotomies. Nineteen centers have reported 335 amyotrophic lateral sclerosis patients using continuous noninvasive ventilatory support instead of tracheostomy mechanical ventilation for an average of 14 mos (6 mos to 14 yrs). The noninvasive ventilatory support must also be used in conjunction with mechanical insufflation-exsufflation to clear airway debris and normalize or renormalize ambient air oxyhemoglobin saturation, both to avoid intubation and to facilitate extubation. People with amyotrophic lateral sclerosis satisfying specific criteria, even when continuously dependent on tracheostomy mechanical ventilation, can be decannulated and placed on continuous noninvasive ventilatory support with mechanical insufflation-exsufflation.

Cold and heterogeneous T cell repertoire is associated with copy number aberrations and loss of immune genes in small-cell lung cancer.

Small-cell lung cancer (SCLC) is speculated to harbor complex genomic intratumor heterogeneity (ITH) associated with high recurrence rate and suboptimal response to immunotherapy. Here, using multi-region whole exome/T cell receptor (TCR) sequencing as well as immunohistochemistry, we reveal a rather homogeneous mutational landscape but extremely cold and heterogeneous TCR repertoire in limited-stage SCLC tumors (LS-SCLCs). Compared to localized non-small cell lung cancers, LS-SCLCs have similar predicted neoantigen burden and genomic ITH, but significantly colder and more heterogeneous TCR repertoire associated with higher chromosomal copy number aberration (CNA) burden. Furthermore, copy number loss of IFN-γ pathway genes is frequently observed and positively correlates with CNA burden. Higher mutational burden, higher T cell infiltration and positive PD-L1 expression are associated with longer overall survival (OS), while higher CNA burden is associated with shorter OS in patients with LS-SCLC.

Immune evolution from preneoplasia to invasive lung adenocarcinomas and underlying molecular features.

The mechanism by which anti-cancer immunity shapes early carcinogenesis of lung adenocarcinoma (ADC) is unknown. In this study, we characterize the immune contexture of invasive lung ADC and its precursors by transcriptomic immune profiling, T cell receptor (TCR) sequencing and multiplex immunofluorescence (mIF). Our results demonstrate that anti-tumor immunity evolved as a continuum from lung preneoplasia, to preinvasive ADC, minimally-invasive ADC and frankly invasive lung ADC with a gradually less effective and more intensively regulated immune response including down-regulation of immune-activation pathways, up-regulation of immunosuppressive pathways, lower infiltration of cytotoxic T cells (CTLs) and anti-tumor helper T cells (Th), higher infiltration of regulatory T cells (Tregs), decreased T cell clonality, and lower frequencies of top T cell clones in later-stages. Driver mutations, chromosomal copy number aberrations (CNAs) and aberrant DNA methylation may collectively impinge host immune responses and facilitate immune evasion, promoting the outgrowth of fit subclones in preneoplasia into dominant clones in invasive ADC.

All-biobased transparent-wood: A new approach and its environmental-friendly packaging application.

Transparent-wood (TW) is an emerging research topic that can be applied to biobased products. However, it is necessary to fill pores in TW with natural substances to prepare all-biobased TW. This paper reports an all-biobased TW by infiltrating cellulose nanofiber (CNF) and chitosan (CTS) suspensions into the bleached wood. CNF was isolated by combining the chemical and physical methods, and CTS was dissolved in acetic acid, and they were infiltrated into the pores of the bleached Fir veneer wood using a vacuum jar. The CNF and chitosan effects on the mechanical properties of the TW were studied, and the morphologies, crystallinity index, water contact angle, antioxidant, thermal degradation, and UV-shielding properties were investigated. The prepared TW showed 80 % total transmittance and 30-60 % haze, suitable for solar cell application. The all-biobased TW showed good thermal stability up to 315 °C and excellent UV shielding property for UV-B and UV-C. The antioxidant property of the CTS-TW significantly increased as compared to the original wood. The CNF-TW showed considerable tensile strength and yield strength of more than 200 % improved from the original wood. The potential for environment-friendly packaging applications was demonstrated by making a bag, medicine packaging, and straw for a drink.

Clinicopathological significance of the expression of PD-L1 in non-small cell lung cancer.

INTRODUCTION: PD1/PD-L1 pathway targeting therapies are nowadays an established treatment option for patients with NSCLC. We assessed whether PD-L1 expression in NSCLC tumor cells was associated with specific clinical features or overall survival using four different clones. METHODS AND RESULTS: A retrospective study included formalin-fixed paraffin embedded (FFPE) surgical tumors from 482 patients. PD-L1 status was assessed with immunohistochemistry in tumor cells on tissue microarrays using clones 28-8, 22C3, SP263 and SP142. Associations with OS were assessed by Kaplan-Meier and multivariate Cox's regression analysis. Patients' median age: 68 years (39-86); histology: adenocarcinoma (AdCa) 61%, squamous-cell carcinoma (SqCC) 33%, and large cell carcinoma (LCC) 6%; p-stage: IA (46%), IB (30%), IIA (10%), IIB (11,4%), IIIA (1,2%), IIIB - IV (0,4%). PD-L1 positivity (≥1%) in NSCLC for clones 28-8, 22C3, SP263, SP142 was 41.5%, 34.2%, 42.7%, 10.4%, respectively (Pearson Chi-square p < 0.0001). PD-L1 expression was correlated with histology, tumor size and grading. Statistically significant association between PD-L1 expression and OS in NSCLC and Non-AdCa was observed with clone SP142 (log-rank p = 0.045 and p = 0.05, respectively). Statistically significant association between PD-L1 expression and OS in LCC was observed with clones 22C3 (log-rank p = 0.009) and SP263 (log-rank p = 0.050). CONCLUSIONS: Overexpression of the PD-L1 clone SP142 was associated with poor overall survival in NSCLC and Non-AdCa. Clones 22C3 and SP263 were associated with poor prognosis in LCC. PD-L1 status might serve as a prognostic marker in NSCLC.

Evolution of DNA methylome from precancerous lesions to invasive lung adenocarcinomas.

The evolution of DNA methylome and methylation intra-tumor heterogeneity (ITH) during early carcinogenesis of lung adenocarcinoma has not been systematically studied. We perform reduced representation bisulfite sequencing of invasive lung adenocarcinoma and its precursors, atypical adenomatous hyperplasia, adenocarcinoma in situ and minimally invasive adenocarcinoma. We observe gradual increase of methylation aberrations and significantly higher level of methylation ITH in later-stage lesions. The phylogenetic patterns inferred from methylation aberrations resemble those based on somatic mutations suggesting parallel methylation and genetic evolution. De-convolution reveal higher ratio of T regulatory cells (Tregs) versus CD8 + T cells in later-stage diseases, implying progressive immunosuppression with neoplastic progression. Furthermore, increased global hypomethylation is associated with higher mutation burden, copy number variation burden and AI burden as well as higher Treg/CD8 ratio, highlighting the potential impact of methylation on chromosomal instability, mutagenesis and tumor immune microenvironment during early carcinogenesis of lung adenocarcinomas.

The mRNA Distribution of Cancer Stem Cell Marker CD90/Thy-1 Is Comparable in Hepatocellular Carcinoma of Eastern and Western Populations.

Epidemiology of hepatocellular carcinoma (HCC) showed a correlation between incidence and geographical-relevant risk factors. This study aims to compare the distributions of cancer stem cells (CSC) in two distant populations in Asia and Europe. We analyzed 52 and 43 selected HCC patients undergoing hepatectomy in Ho Chi Minh City (Vietnam) and Trieste (Italy). Each patient sample consisted of HCC, peri-HCC, and non-tumoral (distal) tissue. Demographic data were recorded together with clinical findings. The protocol for the collection of tissue samples and RNA was standardized in both laboratories and gene expression analysis was performed in a single laboratory with identical PCR conditions. Baseline data showed comparable laboratory findings between the two cohorts. mRNA distribution showed a comparable pattern of all CSC markers analyzed with the expression of CD90 progressively increasing from distal and peri-HCC to be highest in HCC (p < 0.001), confirmed by immunofluorescence data. CD90 mRNA distribution was related to HBV-related HCC and a tumor diameter less than 5 cm. Patients with high tumoral CD90 mRNA had a shorter time (p < 0.05) to tumor recurrence compared to patients with lower CD90. This comparative study showed that CD90 mRNA expressions are comparable between Eastern and Western HCC cases.

A narrative review of digital pathology and artificial intelligence: focusing on lung cancer.

The emergence of whole slide imaging technology allows for pathology diagnosis on a computer screen. The applications of digital pathology are expanding, from supporting remote institutes suffering from a shortage of pathologists to routine use in daily diagnosis including that of lung cancer. Through practice and research large archival databases of digital pathology images have been developed that will facilitate the development of artificial intelligence (AI) methods for image analysis. Currently, several AI applications have been reported in the field of lung cancer; these include the segmentation of carcinoma foci, detection of lymph node metastasis, counting of tumor cells, and prediction of gene mutations. Although the integration of AI algorithms into clinical practice remains a significant challenge, we have implemented tumor cell count for genetic analysis, a helpful application for routine use. Our experience suggests that pathologists often overestimate the contents of tumor cells, and the use of AI-based analysis increases the accuracy and makes the tasks less tedious. However, there are several difficulties encountered in the practical use of AI in clinical diagnosis. These include the lack of sufficient annotated data for the development and validation of AI systems, the explainability of black box AI models, such as those based on deep learning that offer the most promising performance, and the difficulty in defining the ground truth data for training and validation owing to inherent ambiguity in most applications. All of these together present significant challenges in the development and clinical translation of AI methods in the practice of pathology. Additional research on these problems will help in resolving the barriers to the clinical use of AI. Helping pathologists in developing knowledge of the working and limitations of AI will benefit the use of AI in both diagnostics and research.

Differential capacity of CD90+ cells in autophagy activation following chemotherapy in hepatocellular carcinoma.

INTRODUCTION AND OBJECTIVES: Analysis of cancer biomarkers is an important tool in developing targeted-therapy and in modulating chemoresistance. Here, we analyze the relevance of CD90, a marker of cancer stem cells (CSC) in hepatocellular carcinoma (HCC) and its correlation with autophagy. MATERIALS AND METHODS: For in vivo study, 86 specimens were collected from 43 patients undergoing liver resections. In each patient, HCC nodule (HCC) and surrounding non-tumor (SNT) were collected. For in vitro study, HCC cells JHH6 subpopulations expressing CD90+ and CD90- were isolated using magnetic-sorter and confirmed by flow-cytometry. Upon doxorubicin treatment, autophagy turn-over was analyzed by RTqPCR for mRNA expression, Western blot for protein expression, and autophagosome staining for autophagy-flux. Cytotoxicity test was performed by MTT assay. Gene and protein analysis were performed in clinical samples together with immunohistostaining. RESULTS: CD90 mRNA expression was higher in HCC than in SNT for 8-fold (p < 0.001). LC3-II protein was up-regulated in the HCC in comparison with the SNT (p < 0.05). In vitro model showed that CD90+ and CD90- cells had diverse expressions of autophagy-related genes. Upon doxorubicin treatment, autophagy was activated in both cells by increasing LC3-II protein expression, autophagic vacuoles, and dysregulation of autophagy-related mRNAs. A differential autophagic capacity was noticed between two subpopulations and it was correlated with cellular toxicity assay. CONCLUSIONS: We demonstrated the relevance of differential autophagy capacity of CD90+ cells in HCC. Autophagy was involved in cancer-defense mechanism against doxorubicin. Cancer promoting function of autophagy in CD90+ cells was also related to cancer environment.