Deep learning-based image-analysis algorithm for classification and quantification of multiple histopathological lesions in rat liver.

Artificial intelligence (AI)-based image analysis is increasingly being used for preclinical safety-assessment studies in the pharmaceutical industry. In this paper, we present an AI-based solution for preclinical toxicology studies. We trained a set of algorithms to learn and quantify multiple typical histopathological findings in whole slide images (WSIs) of the livers of young Sprague Dawley rats by using a U-Net-based deep learning network. The trained algorithms were validated using 255 liver WSIs to detect, classify, and quantify seven types of histopathological findings (including vacuolation, bile duct hyperplasia, and single-cell necrosis) in the liver. The algorithms showed consistently good performance in detecting abnormal areas. Approximately 75% of all specimens could be classified as true positive or true negative. In general, findings with clear boundaries with the surrounding normal structures, such as vacuolation and single-cell necrosis, were accurately detected with high statistical scores. The results of quantitative analyses and classification of the diagnosis based on the threshold values between "no findings" and "abnormal findings" correlated well with diagnoses made by professional pathologists. However, the scores for findings ambiguous boundaries, such as hepatocellular hypertrophy, were poor. These results suggest that deep learning-based algorithms can detect, classify, and quantify multiple findings simultaneously on rat liver WSIs. Thus, it can be a useful supportive tool for a histopathological evaluation, especially for primary screening in rat toxicity studies.

Phosphatidylcholine (18:0/20:4), a potential biomarker to predict ethionamide-induced hepatic steatosis in rats.

Ethionamide (ETH), a second-line drug for multidrug-resistant tuberculosis, is known to cause hepatic steatosis in rats and humans. To investigate predictive biomarkers for ETH-induced steatosis, we performed lipidomics analysis using plasma and liver samples collected from rats treated orally with ETH at 30 and 100 mg/kg for 14 days. The ETH-treated rats developed hepatic steatosis with Oil Red O staining-positive vacuolation in the centrilobular hepatocytes accompanied by increased hepatic contents of triglycerides (TG) and decreased plasma TG and total cholesterol levels. A multivariate analysis for lipid profiles revealed differences in each of the 35 lipid species in the plasma and liver between the control and the ETH-treated rats. Of those lipids, phosphatidylcholine (PC) (18:0/20:4) decreased dose-dependently in both the plasma and liver. Moreover, serum TG-rich very low-density lipoprotein (VLDL) levels, especially the large particle fraction of VLDL composed of PC containing arachidonic acid (20:4) involved in hepatic secretion of TG, were decreased dose-dependently. In conclusion, the decreased PC (18:0/20:4) in the liver, possibly leading to suppression of hepatic TG secretion, was considered to be involved in the pathogenesis of the ETH-induced hepatic steatosis. Therefore, plasma PC (18:0/20:4) levels are proposed as mechanism-related biomarkers for ETH-induced hepatic steatosis.

Lack of toxicological influences by microsampling (50 µL) from jugular vein of rats in a collaborative 28-day study.

Due to finalization of the ICH S3A Q&A focusing on microsampling, application of microsampling technique to regular non-clinical animal studies is expected for non-clinical safety assessment of pharmaceuticals. In Europe, microsampling from the tail vein or saphenous vein has often been used, whereas sampling from the jugular vein is thought to be more common for non-clinical studies in Japan. Therefore, we assessed the toxicological effects of serial microsampling from the jugular vein of SD rats in a common 28-day study at 4 independent organizations. Fifty microliter sampling was performed at 6 timepoints on day 1 to 2 and 7 timepoints on day 27 to 28 and its toxicological influences on body weight, food consumption, hematological and clinical chemistry parameters, and organ weights (on day 29 for 3 and day 28 for 1 organizations) were evaluated. The serial microsampling was shown to have no or minimal influences on the assessed parameters. The observed statistical differences for the 18 parameters were sporadic and did not appear to be systemically associated with microsampling. However, the sporadic changes were more often observed in females (14/18 parameters) than in males (6/18), suggesting the possibility that female rats were more susceptible to treatment-based influences. The current results indicate that serial 50 µL sampling from the jugular vein of SD rats had no or very slight toxicological effects, suggesting that this microsampling condition is applicable for toxicokinetic evaluation of non-clinical rat toxicity studies.

Selenium and Glutathione-Depleted Rats as a Sensitive Animal Model to Predict Drug-Induced Liver Injury in Humans.

Drug-induced liver injury (DILI) is one of the most serious and frequent drug-related adverse events in humans. Selenium (Se) and glutathione (GSH) have a crucial role for the hepatoprotective effect against reactive metabolites or oxidative damage leading to DILI. The hepatoprotective capacity related to Se and GSH in rodents is considered to be superior compared to the capacity in humans. Therefore, we hypothesize that Se/GSH-depleted rats could be a sensitive animal model to predict DILI in humans. In this study, Se-deficiency is induced by feeding a Se-deficient diet and GSH-deficiency is induced by l-buthionine-S,R-sulfoxinine treatment via drinking water. The usefulness of this animal model is validated using flutamide, which is known to cause DILI in humans but not in intact rats. In the Se/GSH-depleted rats from the present study, decreases in glutathione peroxidase-1 protein expression and GSH levels and an increase in malondialdehyde levels in the liver are observed without any increase in plasma liver function parameters. Five-day repeated dosing of flutamide at 150 mg/kg causes hepatotoxicity in the Se/GSH-depleted rats but not in normal rats. In conclusion, Se/GSH-depleted rats are the most sensitive for detecting flutamide-induced hepatotoxicity in all the reported animal models.

Ether-phosphatidylcholine characterized by consolidated plasma and liver lipidomics is a predictive biomarker for valproic acid-induced hepatic steatosis.

Valproic acid (VPA) is known to induce hepatic steatosis due to mitochondrial toxicity in rodents and humans. In the present study, we administered VPA to SD rats for 3 or 14 days at 250 and 500 mg/kg and then performed lipidomics analysis to reveal VPA-induced alteration of the hepatic lipid profile and its association with the plasma lipid profile. VPA induced hepatic steatosis at the high dose level without any degenerative changes in the liver on day 4 (after 3 days dosing) and at the low dose level on day 15 (after 14 days dosing). We compared the plasma and hepatic lipid profiles obtained on day 4 between the VPA-treated and control rats using a multivariate analysis to determine differences between the two groups. In total, 36 species of plasma lipids and 24 species of hepatic lipids were identified as altered in the VPA-treated group. Of these lipid species, ether-phosphatidylcholines (ePCs), including PC(16:0e/22:4) and PC(16:0e/22:6), were decreased in both the plasma and liver from the low dose level on day 4, however, neither an increase in hepatic TG level nor histopathological hepatic steatosis was observed at either dose level on day 4. Hepatic mRNA levels of glycerone-phosphate O-acyltransferase (Gnpat), which is a key enzyme for biosynthesis of ePC, was also decreased by treatment with VPA along with the decrease in ePCs. In conclusion, the changes in ePCs, (PC[16:0e/22:4] and PC[16:0e/22:6]), have potential utility as predictive biomarkers for VPA-induced hepatic steatosis.

High fructose diet feeding accelerates diabetic nephropathy in Spontaneously Diabetic Torii (SDT) rats.

Diabetic nephropathy (DN) is one of the complications of diabetes and is now the most common cause of end-stage renal disease. Fructose is a simple carbohydrate that is present in fruits and honey and is used as a sweetener because of its sweet taste. Fructose has been reported to have the potential to progress diabetes and DN in humans even though fructose itself does not increase postprandial plasma glucose levels. In this study, we investigated the effects of high fructose intake on the kidney of the Spontaneously Diabetic Torii (SDT) rats which have renal lesions similar to those in DN patients and compared these with the effects in normal SD rats. This study revealed that a 4-week feeding of the high fructose diet increased urinary excretion of kidney injury makers for tubular injury and accelerated mainly renal tubular and interstitial lesions in the SDT rats but not in normal rats. The progression of the nephropathy in the SDT rats was considered to be related to increased internal uric acid and blood glucose levels due to the high fructose intake. In conclusion, high fructose intake exaggerated the renal lesions in the SDT rats probably due to effects on the tubules and interstitium through metabolic implications for uric acid and glucose.

A proposed mechanism for the adverse effects of acebutolol: CES2 and CYP2C19-mediated metabolism and antinuclear antibody production.

Acebutolol, a ß-adrenergic receptor-blocker, occasionally causes drug-induced lupus erythematosus (DILE). Acebutolol is mainly metabolized to diacetolol. Because metabolic activation has been considered to be related to acebutolol-induced toxicity, we sought to identify the enzymes that are responsible for acebutolol metabolism and investigate their involvement in acebutolol-induced toxicity. By using human liver microsomes (HLM) or intestinal microsomes and recombinant enzymes, we found that diacetolol was produced via hydrolysis by carboxylesterase 2 (CES2) and subsequent acetylation by N-acetyltransferase 2 (NAT2). When acetolol, a hydrolytic metabolite of acebutolol, was incubated with HLM and an NADPH-generating system, a metabolite conjugated with N-acetylcystein was generated. This metabolite was found to be formed by CYP2C19 based on studies with a panel of recombinant cytochrome P450 enzymes and an inhibition study using HLM with tranylcypromine, a CYP2C19 inhibitor. Because antinuclear antibody (ANA) production is associated with DILE, we investigated whether ANA was detected in plasma from mice treated with acebutolol. Administration of acebutolol (100mg/kg, p.o.) to female C57BL/6 mice for 30 days resulted in ANA production in plasma in seven of thirteen mice. The number of mice that showed ANA production was larger in mice co-treated with pregnenolone 16α-carbonitrile, an inducer of P450s, whereas it was lower in mice co-treated with tri-o-tolylphosphate or 1-aminobenzotriazole, which are inhibitors of esterases or P450s, respectively. These results suggested that the hydrolysis and oxidation of acebutolol was associated with ANA production. In summary, this study demonstrated that metabolic activation may be a causal factor of adverse reactions of acebutolol.

N-Glycosylation during translation is essential for human arylacetamide deacetylase enzyme activity.

Human arylacetamide deacetylase (AADAC) can hydrolyze clinical drugs such as flutamide, phenacetin, and rifamycins. AADAC is a glycoprotein, but the role of glycosylation remains unclear. In the present study, we investigated the effect of glycosylation on AADAC enzyme activity. Immunoblot analysis of mutant AADACs that contained an asparagine (N, Asn) to glutamine (Q, Gln) substitution at either residue 78 or 282 (N78Q or N282Q) showed a different migration compared with the wild-type protein. A mutant AADAC that contained N to Q substitutions at both residue 78 and 282 (N78Q/N282Q) showed a similar migration to AADAC in human liver microsomes (HLM) treated with endoglycosidase H (Endo H), which produces deglycosylated proteins. This result indicated that AADAC was glycosylated at both N78 and N282. Mutant types of AADAC with the N282Q and the N78Q/N282Q substitutions showed dramatically lower phenacetin hydrolase activity than did the wild-type protein. The treatment of wild-type AADAC-expressing HuH-7 cells with tunicamycin, which produces unglycosylated protein, decreased AADAC enzyme activity. However, the treatment of the HLM with Endo H caused no decrease of AADAC activity. Thus, the oligosaccharide chain, per se, was not important for AADAC activity in the mature form. The mutant types of AADAC containing the N282Q and the N78Q/N282Q substitutions were not detected by immunoblotting analysis after non-reducing SDS-PAGE, suggesting that the glycosylation of AADAC at N282 was important for proper protein folding. Overall, this study found that the translational, but not post-translational, N-glycosylation of AADAC plays a crucial role in regulating AADAC enzyme activity.