Perfluorooctanoic acid inhibits cell proliferation through mitochondrial damage.

Grown evidence has shown that the liver and reproductive organs were the main target organs of perfluorooctanoic acid (PFOA). Herein, we studied a toxic mechanism of PFOA using HeLa Chang liver epithelial cells. When incubated with PFOA for 24 h or 48 h, cell proliferation was inhibited in a concentration- and time-dependent fashion, but interestingly, the feature of dead cells was not notable. Mitochondrial volume was increased with concentration and time, whereas the mitochondrial membrane potential and produced ATP amounts were significantly reduced. Autophagosome-like vacuoles and contraction of the mitochondrial inner membrane were observed in PFOA-treated cells. The expression of acetyl CoA carboxylase (ACC) and p-ACC proteins rapidly decreased, and that of mitochondrial dynamics-related proteins increased. The expression of solute carrier family 7 genes, ChaC glutathione-specific gamma-glutamylcyclotransferase 1, and 5S ribosomal RNA gene was up-regulated the most in cells exposed to PFOA for 24 h, and the KEGG pathway analysis revealed that PFOA the most affected metabolic pathways and olfactory transduction. More importantly, PPAR alpha, fatty acid binding protein 1, and CYP450 family 1 subfamily A member 1 were identified as the target proteins for binding between PFOA and cells. Taken together, we suggest that disruption of mitochondrial integrity and function may contribute closely to PFOA-induced cell proliferation inhibition.

Perfluorooctanoic acid induces cardiac dysfunction in human induced pluripotent stem cell-derived cardiomyocytes.

Perfluorooctanoic acid (PFOA), commonly found in drinking water, leads to widespread exposure through skin contact, inhalation, and ingestion, resulting in detectable levels of PFOA in the bloodstream. In this study, we found that exposure to PFOA disrupts cardiac function in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). We observed reductions in field and action potentials in hiPSC-CMs exposed to PFOA. Furthermore, PFOA demonstrated a dose-dependent inhibitory effect on various ion channels, including the calcium, sodium, and potassium channels. Additionally, we noted dose-dependent inhibition of the expression of these ion channels in hiPSC-CMs following exposure to PFOA. These findings suggest that PFOA exposure can impair cardiac ion channel function and decrease the transcription of genes associated with these channels, potentially contributing to cardiac dysfunction such as arrhythmias. Our study sheds light on the electrophysiological and epigenetic consequences of PFOA-induced cardiac dysfunction, underscoring the importance of further research on the cardiovascular effects of perfluorinated compounds (PFCs).

Differential regulations of neural activity and survival in primary cortical neurons by PFOA or PFHpA.

Perfluorinated compounds (PFCs), organofluoride compounds comprising carbon-fluorine and carbon-carbon bonds, are used as water and oil repellents in textiles and pharmaceutical tablets; however, they are associated with potential neurotoxic effects. Moreover, the impact of PFCs on neuronal survival, activity, and regulation within the brain remains unclear. Additionally, the mechanisms through which PFCs induce neuronal toxicity are not well-understood because of the paucity of data. This study elucidates that perfluorooctanoic acid (PFOA) and perfluoroheptanoic acid (PFHpA) exert differential effects on the survival and activity of primary cortical neurons. Although PFOA triggers apoptosis in cortical neurons, PFHpA does not exhibit this effect. Instead, PFHpA modifies dendritic spine morphogenesis and synapse formation in primary cortical neuronal cultures, additionally enhancing neural activity and synaptic transmission. This research uncovers a novel mechanism through which PFCs (PFHpA and PFOA) cause distinct alterations in dendritic spine morphogenesis and synaptic activity, shedding light on the molecular basis for the atypical behaviors noted following PFC exposure. Understanding the distinct effects of PFHpA and PFOA could guide regulatory policies on PFC usage and inform clinical approaches to mitigate their neurotoxic effects, especially in vulnerable population.

Prenatal Di-methoxyethyl phthalate exposure impairs cortical neurogenesis and synaptic activity in the mice.

Di-methoxyethyl phthalate (DMEP) is a well-known environmentally prevalent endocrine disruptor and may be associated with neurodevelopmental disorders including attention deficit/hyperactivity disorder and intellectual disability. However, the regulatory mechanisms leading to these neurodevelopmental disorders are still poorly understood. Here, we demonstrate that prenatal DMEP exposure causes abnormal brain morphology and function in the mice. DMEP (50 mg/kg) was chronically administered to pregnant mice orally once a day starting on embryonic day 0 (E0) to breast-feeding cessation for the fetus. We found that prenatal DMEP exposure significantly reduced the number of neurons in the parietal cortex by impairing neurogenesis and gliogenesis during the developing cortex. Moreover, we found that prenatal DMEP exposure impaired dendritic spine architectures and synaptic activity in the parietal cortex. Finally, prenatal DMEP exposure in mice induces hyperactivity and reduces anxiety behaviors. Altogether, our study demonstrates that prenatal DMEP exposure leads to abnormal behaviors via impairment of neurogenesis and synaptic activity.

Application of multiple-finding segmentation utilizing Mask R-CNN-based deep learning in a rat model of drug-induced liver injury.

Drug-induced liver injury (DILI) presents significant diagnostic challenges, and recently artificial intelligence-based deep learning technology has been used to predict various hepatic findings. In this study, we trained a set of Mask R-CNN-based deep algorithms to learn and quantify typical toxicant induced-histopathological lesions, portal area, and connective tissue in Sprague Dawley rats. We compared a set of single-finding models (SFMs) and a combined multiple-finding model (MFM) for their ability to simultaneously detect, classify, and quantify multiple hepatic findings on rat liver slide images. All of the SFMs yielded mean average precision (mAP) values above 85%, suggesting that the models had been successfully established. The MFM showed better performance than the SFMs, with a total mAP value of 92.46%. We compared the model predictions for slide images with ground-truth annotations generated by an accredited pathologist. For the MFM, the overall and individual finding predictions were highly correlated with the annotated areas, with R-squared values of 0.852, 0.952, 0.999, 0.990, and 0.958 being obtained for portal area, infiltration, necrosis, vacuolation, and connective tissue (including fibrosis), respectively. Our results indicate that the proposed MFM could be a useful tool for detecting and predicting multiple hepatic findings in basic non-clinical study settings.

Bisphenol A is more potent than bisphenol S in influencing the physiological and pathological functions of lungs via inducing lung fibrosis and stimulating metastasis.

Bisphenol A (BPA) is widely used in the production of plastics, food containers, and receipt ink globally. However, research has identified it as an endocrine disruptor, affecting the hormonal balance in living organisms. Bisphenol S (BPS), one of the alternative substances, was developed, but its effects on human health and the underlying mechanisms remain unclarified. Specifically, research on the effects of oral exposure to bisphenol on the lungs is lacking. We examined the potential differences in toxicity between these compounds in lung cells in vitro and in vivo. Our toxicity mechanism studies on MRC5 and A549 cells exposed to BPA or BPS revealed that BPA induced actin filament abnormalities and activated epithelial-mesenchymal transition (EMT). This finding suggests an increased potential for lung fibrosis and metastasis in lung cancer. However, given that BPS was not detected at the administered dose and under the specific experimental conditions, the probability of these occurrences is considered minimal. Additionally, animal experiments confirmed that oral exposure to BPA activates EMT in the lungs. Our study provides evidence that prolonged oral exposure to BPA can lead to EMT activation in lung tissue, similar to that observed in cell experiments, suggesting the potential to induce lung fibrosis. This research emphasizes the importance of regulating the use of BPA to mitigate its associated pulmonary toxicity. Furthermore, it is significant that within the parameters of our experimental conditions, BPS did not exhibit the toxicological pathways clearly evident in BPA.

Long-chain perfluoroalkyl carboxylates induce cytoskeletal abnormalities and activate epithelial-mesenchymal transition in both renal cell carcinoma 3D cultures and Caki-1 xenografted mouse model.

Exposure to perfluorooctanoate (PFOA; a type of perfluoroalkyl carboxylates [PFACs]) may be correlated with the incidence of kidney cancer in individuals exposed to high levels of PFOA. However, mechanistic studies on the influence of PFACs on renal cell carcinoma (RCC) development are lacking. We explored the effects of five types of PFACs on RCC using in vitro and in vivo models to fill this knowledge gap and provide information for environmental/usage regulations. Using 2D/3D cultures of Caki-1 cells, a human clear cell RCC line, we examined the effects of short-chain (SC) PFACs and long-chain (LC) PFACs on RCC physio/pathological markers, including the cytoskeleton, epithelial-mesenchymal transition (EMT)-related proteins, and Na+/K+-ATPase. We also administered three different PFACs orally to mice harboring Caki-1 xenografts to assess the impact of these compounds on engrafted RCC in vivo. Compared with the effects of SCPFACs, mice with Caki-1 xenografts treated with LCPFACs showed increased EMT-related protein expression and exhibited liver toxicity. Therefore, LCPFACs induced EMT, influencing cancer metastasis activity, and displayed higher toxicity in vivo compared with SCPFACs. These findings improve our understanding of the effects of PFACs on RCC development and their corresponding in vivo toxicity, which is crucial for regulating these substances to protect public health.

Genotoxicity assessment of root extracts of Paeonia lactiflora Pall.

The roots of Paeonia lactiflora Pall., (Paeoniae Radix, PL) are a well-known herbal remedy used to treat fever, rheumatoid arthritis, systemic lupus erythematosus, hepatitis, and gynecological disorders in East Asia. Here we evaluated the genetic toxicity of PL extracts (as a powder [PL-P] and hot-water extract [PL-W]) in accordance with the Organization for Economic Co-operation and Development guidelines. The Ames test revealed that PL-W was not toxic to S. typhimurium strains and E. coli in absence and presence of the S9 metabolic activation system at concentrations up to 5000 µg/plate, but PL-P produced a mutagenic response to TA100 in the absence of S9 mix. PL-P was cytotoxic in in vitro chromosomal aberrations (more than a 50 % decrease in cell population doubling time), and it increased the frequency of structural and numerical aberrations in absence and presence of S9 mix in a concentration-dependent manner. PL-W was cytotoxic in the in vitro chromosomal aberration tests (more than a 50 % decrease in cell population doubling time) only in the absence of S9 mix, and it induced structural aberrations only in the presence of S9 mix. PL-P and PL-W did not produce toxic response during the in vivo micronucleus test after oral administration to ICR mice and did not induce positive results in the in vivo Pig-a gene mutation and comet assays after oral administration to SD rats. Although PL-P showed genotoxic in two in vitro tests, the results from physiologically relevant in vivo Pig-a gene mutation and comet assays illustrated that PL-P and PL-W does not cause genotoxic effects in rodents.

A general toxicity and biodistribution study of human natural killer cells by single or repeated intravenous dose in severe combined immune deficient mice.

Natural killer (NK) cells are a part of the innate immune system and represent the first line of defense against infections and tumors. NK cells can eliminate tumor cells without major histocompatibility restriction and are independent of the expression of tumor-associated antigens. Therefore, they are considered an emerging tool for cancer immunotherapy. However, the general toxicity and biodistribution of NK cells after transplantation remain to be understood. This study was conducted to evaluate the general toxicity and biodistribution of human NK cells after single or repeated intravenous dosing in severely combined immunodeficient (SCID) mice. There were no test item-related toxicological changes in single and repeated administration groups. The no observed adverse effect level of human NK cells was 2 × 107 cells/head for both male and female SCID mice. Results from the biodistribution study showed that human NK cells were mainly distributed in the lungs, and a small number of the cells were detected in the liver, heart, spleen, and kidney of SCID mice, in both the single and repeated dose groups. Additionally, human NK cells were completely eliminated from all organs of the mice in the single dose group on day 7, while the cells persisted in mice in the repeated dose group until day 64. In conclusion, transplantation of human NK cells in SCID mice had no toxic effects. The cells were mainly distributed in the lungs and completely disappeared from the body over time after single or repeated intravenous administration.

Bisphenol A impairs renal function by reducing Na+/K+-ATPase and F-actin expression, kidney tubule formation in vitro and in vivo.

The kidney proximal tubule is responsible for reabsorbing water and NaCl to maintain the homeostasis of the body fluids, electrolytes, and nutrients. Thus, abnormal functioning of the renal proximal tubule can lead to life-threatening imbalances. Bisphenol A (BPA) has been used for decades as a representative chemical in household plastic products, but studies on its effects on the kidney proximal tubule are insufficient. In this study, immunocytochemical and cytotoxicity tests were performed using two- and three-dimensional human renal proximal tubular epithelial cell (hRPTEC) cultures to investigate the impact of low-dose BPA (1-10 µM) exposure. BPA was found to interfere with straight tubule formation as observed by low filamentous actin formation and reduced Na+/K+-ATPase expression in the tubules of hRPTEC 3D cultures. Similar results were observed in rat pup kidneys following oral administration of 250 mg/kg BPA. Moreover, the expression of HO-1 and 8-OHdG, key markers for oxidative stress, was increased in vitro and in vivo following BPA administration, whereas that of OAT1 and OAT, important transporters of the renal proximal tubules, was not altered. Overall, no-observed-adverse-effect-level (NOAEL)-dose BPA exposure can decrease renal function by promoting abnormal tubular formation both in vitro and in vivo. Therefore, we propose that although it does not exhibit life-threatening toxicity, exposure to low levels of BPA can negatively affect homeostasis in the body by means of long-term deterioration of renal proximal tubular function in humans.

Tumorigenicity Assessment of Human Cancer Cell Lines Xenografted on Immunodeficient Mice as Positive Controls of Tumorigenicity Testing.

Recent advances in human pluripotent stem cell (hPSC)-derived cell therapies and genome editing technologies such as CRISPR/Cas9 make regenerative medicines promising for curing diseases previously thought to be incurable. However, the possibility of off-target effects during genome editing and the nature of hPSCs, which can differentiate into any cell type and infinitely proliferate, inevitably raises concerns about tumorigenicity. Tumorigenicity acts as a major obstacle to the application of hPSC-derived and gene therapy products in clinical practice. Thus, regulatory authorities demand mandatory tumorigenicity testing as a key pre-clinical safety step for the products. In the tumorigenicity testing, regulatory guidelines request to include human cancer cell line injected positive control group (PC) animals, which must form tumors. As the validity of the whole test is determined by the tumor-forming rates (typically above 90%) of PC animals, establishing the stable tumorigenic condition of PC animals is critical for successful testing. We conducted several studies to establish the proper positive control conditions, including dose, administration routes, and the selection of cell lines, in compliance with Good Laboratory Practice (GLP) regulations and/or guidelines, which are essential for pre-clinical safety tests of therapeutic materials. We expect that our findings provide insights and practical information to create a successful tumorigenicity test and its guidelines.

Hyaluronic microparticle-based biomimetic artificial neighbors of cells for three-dimensional cell culture.

3D spheroids, which have the potential to bridge the gap between 2D cell culture and native tissue, are used as tissue models in many applications, particularly in cancer, stem cell, and pharmaceutical research. A considerable amount of effort has focused on the development of more relevant physiological models. However, spheroids still have limitations in that they cannot replicate the components and structure of the ECM in the natural environment. In this study, we proposed new concept of scaffold-based techniques for the generation of spheroids. Spheroids were successfully generated by single cell or small number of aggregated cells between HA particles. The size of each spheroid was uniform, a necrotic core didn't form, and the system showed high viability. The expression levels of the proteins and genes required to maintain cell-specific functions increased. Thus, our system provides more physiologically relevant models and could be applied to regenerative medicine or drug screening.

Bisphenol-A impairs synaptic formation and function by RGS4-mediated regulation of BDNF signaling in the cerebral cortex.

Bisphenol-A (BPA) is a representative endocrine disruptor, widely used in a variety of products including plastics, medical equipment and receipts. Hence, most people are exposed to BPA via the skin, digestive system or inhalation in everyday life. Furthermore, BPA crosses the blood-brain barrier and is linked to multiple neurological dysfunctions found in neurodegenerative and neuropsychological disorders. However, the mechanisms underlying BPA-associated neurological dysfunctions remain poorly understood. Here, we report that BPA exposure alters synapse morphology and function in the cerebral cortex. Cortical pyramidal neurons treated with BPA showed reduced size and number of dendrites and spines. The density of excitatory synapses was also decreased by BPA treatment. More importantly, we found that BPA disrupted normal synaptic transmission and cognitive behavior. RGS4 and its downstream BDNF/NTRK2 pathway appeared to mediate the effect of BPA on synaptic and neurological function. Our findings provide molecular mechanistic insights into anatomical and physiological neurotoxic consequences related to a potent endocrine modifier.

Artificial Intelligence-Assisted Image Analysis of Acetaminophen-Induced Acute Hepatic Injury in Sprague-Dawley Rats.

Although drug-induced liver injury (DILI) is a major target of the pharmaceutical industry, we currently lack an efficient model for evaluating liver toxicity in the early stage of its development. Recent progress in artificial intelligence-based deep learning technology promises to improve the accuracy and robustness of current toxicity prediction models. Mask region-based CNN (Mask R-CNN) is a detection-based segmentation model that has been used for developing algorithms. In the present study, we applied a Mask R-CNN algorithm to detect and predict acute hepatic injury lesions induced by acetaminophen (APAP) in Sprague-Dawley rats. To accomplish this, we trained, validated, and tested the model for various hepatic lesions, including necrosis, inflammation, infiltration, and portal triad. We confirmed the model performance at the whole-slide image (WSI) level. The training, validating, and testing processes, which were performed using tile images, yielded an overall model accuracy of 96.44%. For confirmation, we compared the model's predictions for 25 WSIs at 20× magnification with annotated lesion areas determined by an accredited toxicologic pathologist. In individual WSIs, the expert-annotated lesion areas of necrosis, inflammation, and infiltration tended to be comparable with the values predicted by the algorithm. The overall predictions showed a high correlation with the annotated area. The R square values were 0.9953, 0.9610, and 0.9445 for necrosis, inflammation plus infiltration, and portal triad, respectively. The present study shows that the Mask R-CNN algorithm is a useful tool for detecting and predicting hepatic lesions in non-clinical studies. This new algorithm might be widely useful for predicting liver lesions in non-clinical and clinical settings.

Development of human pluripotent stem cell-derived hepatic organoids as an alternative model for drug safety assessment.

Human in vitro hepatic models that faithfully recapitulate liver function are essential for successful basic and translational research. A limitation of current in vitro models, which are extensively used for drug discovery and toxicity testing, is the loss of drug metabolic function due to the low expression and activity of cytochrome P450 (CYP450) enzymes. Here, we aimed to generate human pluripotent stem cell-derived hepatic organoids (hHOs) with a high drug metabolic ability. We established a two-step protocol to produce hHOs from human pluripotent stem cells for long-term expansion and drug testing. Fully differentiated hHOs had multicellular composition and exhibited cellular polarity and hepatobiliary structures. They also displayed remarkable CYP450 activity and recapitulated the metabolic clearance, CYP450-mediated drug toxicity, and metabolism. Furthermore, hHOs successfully modeled Wilson's disease in terms of Cu metabolism, drug responses, and diagnostic marker expression and secretion. In conclusion, hHOs exhibit high capacity for drug testing and disease modeling. Hence, this hepatic model system provides an advanced tool for studying hepatic drug metabolism and diseases.

Prenatal developmental toxicity study of 2,4-dichlorobenzyl alcohol in rats.

Sore throat lozenges, which are over-the-counter drugs, contain 2,4-dichlorobenzyl alcohol (DCBA) as the primary ingredient. However, comprehensive data on the prenatal developmental toxicity of DCBA is limited. Therefore, this study was conducted to determine the effects of DCBA on pregnant rats and prenatal development. Sprague-Dawley rats were administered different doses of DCBA (0, 25, 100, 400, and 800 mg/kg/day) daily via an oral gavage from gestation day (GD) 6-19. Thereafter, all the live dams were sacrificed on GD 20, and caesarean sections were conducted. Live fetuses and their placenta were weighed and then examined for external, visceral, and skeletal malformations and variations. Based on the results obtained, dams at 800 mg/kg/day showed systemic toxicities, including a decrease in body weight and food consumption, and liver changes. Additionally, this treatment induced decreases in fetal and placental weights, as well as the increased incidence of retarded ossifications and full supernumery rib, and the decreased number of ossification centers. Therefore, based on these findings, the no-observed-adverse-effect level of DCBA was determined to be 400 mg/kg/day for dams and prenatal development.

Disruption of Membrane Integrity as a Molecular Initiating Event Determines the Toxicity of Polyhexamethylene Guanidine Phosphate Depending on the Routes of Exposure.

Polyhexamethylene guanidine phosphate (PHMG-P), a cationic biocide, is widely used in household products due to its strong bactericidal activity and low toxicity. However, it causes fatal lung damage when inhaled. In this study, we investigated why PHMG-P causes fatal lung injury when inhaled, and demonstrated that the disruption of membrane integrity through ionic interaction-a molecular initiating event of PHMG-P-determines toxicity. Mice were injected intravenously with 0.9 or 7.2 mg/kg PHMG-P (IV group), or instilled intratracheally with 0.9 mg/kg PHMG-P (ITI group); they were euthanatized at 4 h and on days 1 and 7 after treatment. Increased total BAL cell count and proinflammatory cytokine production, along with fibrotic changes in the lungs, were detected in the ITI group only. Levels of hepatic enzymes and hepatic serum amyloid A mRNA expression were markedly upregulated in the 7.2 mg/kg IV and ITI groups at 4 h or day 1 after treatment, but returned to baseline. No pathological findings were detected in the heart, liver, or kidneys. To simulate the IV injection, A549, THP-1, and HepG2 cells were treated with PHMG-P in cell culture media supplemented with different serum concentrations. Increased serum concentration was associated with an increase in cell viability. These results support the idea that direct contact between PHMG-P and cell membranes is necessary for PHMG-induced toxicity.

Adipose stem cell-derived extracellular vesicles ameliorates corticosterone-induced apoptosis in the cortical neurons via inhibition of ER stress.

BACKGROUND: Corticosterone (CORT) can induce neuronal damage in various brain regions, including the cerebral cortex, the region implicated in depression. However, the underlying mechanisms of these CORT-induced effects remain poorly understood. Recently, many studies have suggested that adipose stem cell-derived extracellular vesicles (A-EVs) protect neurons in the brain. METHODS: To investigated neuroprotection effects of A-EVs in the CORT-induced cortical neurons, we cultured cortical neurons from E15 mice for 7 days, and the cultured cortical neurons were pretreated with different numbers (5 × 105-107 per mL) of A-EVs (A-EVs5, A-EVs6, A-EVs7) for 30 min followed by administration of 200 µM CORT for 24 h. RESULTS: Here, we show that A-EVs exert antiapoptotic effects by inhibiting endoplasmic reticulum (ER) stress in CORT-induced cortical neurons. We found that A-EVs prevented neuronal cell death induced by CORT in cultured cortical neurons. More importantly, we found that CORT exposure in cortical neurons resulted in increased levels of apoptosis-related proteins such as cleaved caspase-3. However, pretreatment with A-EVs rescued the levels of caspase-3. Intriguingly, CORT-induced apoptosis involved upstream activation of ER stress proteins such as GRP78, CHOP and ATF4. However, pretreatment with A-EVs inhibited ER stress-related protein expression. CONCLUSION: Our findings reveal that A-EVs exert antiapoptotic effects via inhibition of ER stress in CORT-induced cell death.

Comparison of toxicity and cellular responses following pulmonary exposure to different types of nanofibers.

Pulmonary effects of inhaled microfibers are an emerging public health concern. In this study, we investigated toxicity following pulmonary exposure to synthetic polyethylene oxide fibroin (PEONF) and silk fibroin (SFNF) nanofibers and the cellular responses. When instilled intratracheally weekly for four weeks, body weight gain was significantly reduced in female mice exposed to the higher dose of SFNF when compared with the control group. The total number of cells in the lungs was more significant in all treated groups than in the control, whereas the relative portion of neutrophils and eosinophils increased significantly only in female mice exposed to SFNF. Both types of nanofibers induced notable pathological changes and increased pulmonary expression of MCP-1α, CXCL1, and TGF-ß. More importantly, blood calcium, creatinine kinase, sodium, and chloride concentration were affected significantly, showing sex- and material-dependent differences. The relative portion of eosinophils increased only in SFNF-treated mice. In addition, both types of nanofibers induced necrotic and late apoptotic cell death in alveolar macrophages after 24 h of exposure, with accompanying oxidative stress, increased NO production, cell membrane rupture, intracellular organelle damage, and intracellular calcium accumulation. Additionally, multinucleated giant cells were formed in cells exposed to PEONF or SFNF. Taken together, the findings indicate that inhaled PEONF and SFNF may cause systemic adverse health effects with lung tissue damage, showing differences by sex- and material. Furthermore, PEONF- and SFNF-induced inflammatory response may be partly due to the low clearance of dead (or damaged) pulmonary cells and the excellent durability of PEONF and SFNF.

Implementation and Practice of Deep Learning-Based Instance Segmentation Algorithm for Quantification of Hepatic Fibrosis at Whole Slide Level in Sprague-Dawley Rats.

Exponential development in artificial intelligence or deep learning technology has resulted in more trials to systematically determine the pathological diagnoses using whole slide images (WSIs) in clinical and nonclinical studies. In this study, we applied Mask Regions with Convolution Neural Network (Mask R-CNN), a deep learning model that uses instance segmentation, to detect hepatic fibrosis induced by N-nitrosodimethylamine (NDMA) in Sprague-Dawley rats. From 51 WSIs, we collected 2011 cropped images with hepatic fibrosis annotations. Training and detection of hepatic fibrosis via artificial intelligence methods was performed using Tensorflow 2.1.0, powered by an NVIDIA 2080 Ti GPU. From the test process using tile images, 95% of model accuracy was verified. In addition, we validated the model to determine whether the predictions by the trained model can reflect the scoring system by the pathologists at the WSI level. The validation was conducted by comparing the model predictions in 18 WSIs at 20× and 10× magnifications with ground truth annotations and board-certified pathologists. Predictions at 20× showed a high correlation with ground truth (R2 = 0.9660) and a good correlation with the average fibrosis rank by pathologists (R2 = 0.8887). Therefore, the Mask R-CNN algorithm is a useful tool for detecting and quantifying pathological findings in nonclinical studies.