Abstract
Dogs treated with AR-H047108, an imidazopyridine potassium competitive acid blocker (P-CAB), developed clinical signs of hepatic dysfunction as well as morphologically manifest hepatotoxicity in repeat-dose toxicity studies. An investigative one-month study was performed, with interim euthanasia after one and two weeks. A detailed histopathological and immunohistochemical characterization of the liver lesions was conducted, including markers for fibrosis, Kupffer cell activation, apoptosis, and endothelial injury. In addition, hepatic retinoid and procollagen 1α2 mRNA levels in livers of dogs treated with AR-H047108 were analyzed. The results showed an early inflammatory process in central veins and centrilobular areas, present after one week of treatment. This inflammatory reaction was paralleled by activation of stellate/Ito cells to myofibroblasts and was associated with sinusoidal and centrivenular fibrosis. The early activation of stellate cells coincided with a significant decrease in retinyl ester levels, and a significant increase in procollagen 1α2 mRNA levels, in the liver. At later time points (three and six months), there was marked sinusoidal fibrosis in centrilobular areas, as well as occlusion of central veins resulting from a combination of fibrosis and increased thickness of smooth muscle bundles in the vessel wall. The pattern of lesions suggests a veno-occlusive-disease (VOD)–like scenario, possibly linked to the imidazopyridine chemical structure of the compound facilitated by specific morphological features of the dog liver.
Introduction
Several proton pump inhibitors for treatment of gastroesophageal reflux disease and other acid-related diseases have been developed since the 1970s. Omeprazole was the first proton pump inhibitor used in clinical practice. Omeprazole acts through specific inhibition of the gastric H+K+-ATPase located in the apical membrane of the parietal cell in the gastric mucosa. Since this enzyme represents the final step of acid secretion, inhibition of it reduces acid secretion regardless of how secretion is stimulated (Olbe et al. 2003). Esomeprazole is the S-isomer of omeprazole and has an identical mechanism of action, although its bioavailability is higher. Both omeprazole and esomeprazole belong chemically to the substituted benzimidazole class.
A second-generation class of proton pump inhibitors with a different mode of action, the so-called potassium competitive acid blockers (P-CABs), was developed as follow-up compounds to omeprazole and esomeprazole. The P-CABs act through a reversible, K+-competitive inhibition of the H+K+-ATPase in the parietal cells. They give a faster onset of gastric acid inhibition, and in contrast to omeprazole and esomeprazole, the effect is directly correlated to the plasma concentration (Andersson and Carlsson 2005).
Although the P-CABs have shown excellent pharmacological results in both animal studies and in humans, a recurring problem with a few of these drug candidates has been a specific hepatotoxicity restricted to dogs. In rats, none of the compounds has caused any liver changes apart from centrilobular hepatocytic hypertrophy associated with induction of cytochrome P450 enzymes. Monkeys have shown no liver effects whatsoever.
We here report on the histopathological and molecular characterization of drug-induced liver lesions in dogs treated with the imidazopyridine P-CAB AR-H047108, using a wide range of immunohistochemical markers as well as analysis of hepatic retinoid and procollagen 1α2 mRNA levels to elucidate possible mechanistic aspects. The pattern of lesions points in the direction of a veno-occlusive disease-(VOD)–like scenario with a very early activation of stellate cells, possibly linked to the imidazopyridine chemical structure of the compound facilitated by specific morphological features of the dog liver.
Materials and Methods
Test Substance
AR-H047108 (8-[(2-ethyl-6-methylbenzyl)amino]-2,3-dimethylimidazo[1,2-a]pyridine-6-carboxamide) was synthesized by AstraZeneca. The test substance was supplied as tablets packed in gelatin capsules. The chemical structure of AR-H047108 is shown in Figure 1.
Animals
Male and female beagle dogs from two different breeders were used in three studies: two toxicity studies (three months with a three-month recovery period and six-month duration, respectively) and one investigative study (one month). The dogs were acclimatized to laboratory conditions for at least one month before study start. Each animal received 200–350 g once daily of a dog laboratory diet (Specific CXD, Lövens Kemiske Fabrik, Denmark). Municipal tap water for human consumption was available at all times via an automatic watering system. Approximately twenty-four hours after the last dose, the animals were terminated. Approval from the animal research ethics committee had been obtained for using these animals in preclinical safety studies.
Three- and Six-month Toxicity Studies
Beagle dogs were treated perorally with AR-H047108 for three and six months, respectively. Control dogs were given empty capsules.
In the three-month study, three males and three females, aged 5.5–8.5 months at the start of the study, were included at each dose level. The dose levels were 20, 80, and 320 μmol/kg (6.73, 26.9, or 108 mg/kg, expressed as the base form AR-H047108). After two months of dosing, the high dose was reduced to 240 μmol/kg (81 mg/kg base form) because of low food consumption and affected general condition among dogs at this exposure level. Recovery dogs (three months without dosing) were included for assessment.
In the six-month study, five males and five females, aged 7.5–8.5 months at the start of the study, were included at each dose level. The dose levels were 20, 70, and 249 μmol/kg (6.7, 24, or 81 mg/kg base form). After three and a half months, the high dose was reduced to 180 μmol/kg (61 mg/kg base form) because of low food consumption and affected general condition among dogs at this exposure level. To maintain a sufficient separation in the exposure levels, the intermediate group was reduced to 60 μmol/kg (20 mg/kg as free base).
Clinical signs were recorded daily. Body weight was recorded before dosing started, and then once weekly during the study. Individual food consumption was recorded daily.
Blood sampling for determination of the concentration of AR-H047108 and its pharmacologically active metabolite AR-H047116 in plasma was performed on all animals.
Blood samples for hematology and clinical chemistry analyses were obtained before dosing started and after one, three, and six months of dosing, and after one and three months of recovery.
All main-study dogs were subjected to necropsy. A full range of tissues from all dogs was evaluated microscopically on hematoxylin and eosin (H&E)-stained slides. Liver sections were also stained with Masson-Trichrome (MT) and Oil Red. Immunohistochemistry was performed on liver sections using a panel of markers covering various aspects of liver pathology (see below under Histopathology and Immunohistochemistry).
One-month Investigative Study
This study was designed to investigate the mechanism behind the liver toxicity observed in the three- and six-month toxicity studies. Groups of two or three male dogs, seven to eight months old, were given AR-H047108 orally as tablets in gelatin capsules. The dose levels were 160 and 240 μmol/kg (53.8 and 81 mg/kg base form). Control dogs were given empty capsules. Dogs were euthanized after eight days (160 μmol/kg), fourteen days (160 μmol/kg) and twenty-nine days (160 and 240 μmol/kg) of dosing. Clinical signs, body weight, and food consumption were monitored in the same way as described for the three- and six-month studies. Blood samples for hematology and clinical chemistry were collected on two occasions from all animals before dosing started and on days 5, 12, and 26 of dosing, respectively. Blood sampling for determination of the concentration of AR-H047108 and its pharmacologically active metabolite AR-H047116 in plasma was performed on all animals on days 5, 12, and 26. Finally, a single blood sample was taken at necropsy three hours after the last dose. Liver samples were taken at necropsy for histopathology, in formalin, and for molecular and biochemical analyses, snap-frozen in liquid nitrogen.
Histopathology and Immunohistochemistry
A detailed histopathological characterization of liver lesions was performed using H&E-, MT-, and Oil Red-stained sections from dogs treated with AR-H047108 and vehicle-treated control dogs. Immunohistochemistry was performed with a panel of markers chosen to address specific patterns of injury (e.g., apoptosis, endothelial damage, Kupffer-cell–mediated toxicity) (Table 1).
Immunohistochemical staining was in most cases performed using the Dakocytomation Techmate 500+, TMS20288. Some stainings were performed using the Ventana Discovery XT automated slide-processing system, according to the manufacturer’s instruction. The avidin-biotin complex (ABC) was used as detection system, with diaminobenzidine (DAB) as the chromogen. The staining of canine IgG was performed manually with an alkaline phosphatase-conjugated rabbit F(ab)2, using Fast Red as the chromogen. To further characterize transforming growth factor beta (TGFβ)-positive cells, double staining with TGFβ and tryptase was performed. After immunostaining, the slides were counterstained in hematoxylin and mounted with Pertex (except the Fast Red-stained slides, which were mounted in aqueous mounting medium).
Detection of Apoptotic Cells
Apoptotic cells in the liver sections were detected using the DeadEnd colorimetric apoptosis detection system (Promega), according to the manufacturer’s instructions. The DeadEnd system end-labels the fragmented DNA of apoptotic cells using a modified TUNEL (TdT-mediated dUTP nick-end labeling) assay. TUNEL staining was performed using the Ventana Discovery XT automated slide-processing system.
Evaluation of Immunohistochemical Stainings
Following immunohistochemistry and TUNEL staining, the number of positively labeled cells in each liver section was estimated semiquantitatively by light microscopy as follows: − = no positively labeled cells, + = a few, ++ = scattered, +++ = numerous positively labeled cells. Staining of vessel walls (smooth muscle actin, von Willebrand factor, fibrinogen) was graded as follows: − = none, + = minimal, ++ = slight, +++ = moderate.
Levels of Apolar Retinoids in Liver Samples
Liver samples from dogs in the one-month investigative study were analyzed for vitamin A levels as described previously (Nilsson et al. 2000). Briefly, apolar retinoids were extracted from liver homogenates (20% w/v in water) using di-isopropyl ether. Retinol and retinyl esters were separated on a Nucleosil C18 5-μm HPLC column (Macherey-Nagel, GmbH, Germany) using an ethanol:water (90:10 v/v) mobile phase. The retinoids were detected with a JASCO 821-FP fluorescence detector (ex = 325 nm, em = 475 nm) and quantified individually using internal (retinyl acetate) and external (retinol and retinyl palmitate) standards. All analyses were carried out in duplicate. Statistical analysis of total retinoid levels was performed using the nonparametric Kruskal-Wallis test, followed by the Mann-Whitney U-test. A p value of < .05 was considered to be statistically significant.
Procollagen 1α 2 mRNA Levels in Liver Samples
Quantitative RT-PCR was performed using the Lightcycler method. Liver samples from dogs in the one-month investigative study were stored at −80°C prior to use. Samples from control animals and animals dosed with 240 μmol/kg for eight, fourteen, or twenty-nine days were included in the analyses.
Total RNA was prepared with the QIAGEN RNeasy Midi Prep Kit. The frozen tissue samples were transferred to Lysing MatrixD tubes (BI0101, cat. no. 6913-100) containing 1 mL RLT buffer (1 mL lysis buffer + 10 μ1 14,3M β-ME) and homogenized in the Fastprep FP120 instrument (BIO 101, SAVANT, cat. no. 6000-120). Liver samples were homogenized three times for ten seconds at level 6 and chilled on ice in between. The homogenates were transferred to Qiashredder tubes (QIAGEN, cat. no. 79656) and centrifuged at 8000 rpm (6800xg) for three minutes in an Eppendorf microcentrifuge. The eluates were mixed with 1 mL EtOH and vortexed for five seconds. The eluates were then immediately transferred to an RNeasy midi spin column, and total RNA preparation was performed according to the manual provided with the RNeasy Midi Kit (QIAGEN, cat. no. 75144). The total RNA was eluted in 2×100 μL DEPC water and subsequently quantified using the Agilent 2100 Bioanalyzer and the RNA 6000 Nano Kit (Agilent Cat. no. 5065-4476). The procedure was performed according to the manufacturer’s manual (Reagent Kit Guide, RNA 6000 Nano Assay, Edition 07/01). After quantification, the total RNA was stored at −70°C.
DNase1 treatment was performed using RQ1 RNase-Free DNase from Promega (cat. no. M6101). The reaction was incubated at 37°C for thirty minutes, and then RQ1 DNase Stop Solution was added to terminate the reaction. The samples were incubated at 65°C for ten minutes to inactivate the DNase. This reaction was divided in two samples, and the cDNA synthesis was performed +/− reverse transcription enzyme.
Total RNA was transcribed to cDNA using the TaqMan Reverse Transcription Kit (cat. no. N808-0234). Tubes were incubated at 25°C for ten minutes, 48°C for thirty minutes, and 95°C for five minutes. cDNA samples were stored at −20°C until required for analysis.
A reference pool cDNA was generated by mixing 15 μL of each +RT sample, and this reference cDNA was used to generate a standard in the PCR reactions. SYBRgreen mix (Qiagen Quantitect) was used in the Lightcycler PCR. The optimal PCR protocol for measuring the PCR product while minimizing detection of primer dimers was determined empirically on trial runs for the primer pairs. The 5′ to 3′ primer sequences for dog procollagen (EMBL:AF035120) mRNA were: GGATTCCCTGGACCCAAAGG (forward) and ACCCTGGAAGCCTGGAGGAC (reverse). The 5′ to 3′ primer sequences for dog β-actin (EMBL: AF021873) mRNA were: TCCGTAAGGACCTGTATGCC (forward) and ACATCTGCTGGAAGGTGGAC (reverse). The melting temperature for both primer pairs was 82°C. Procollagen 1α2 mRNA levels were normalized to β-actin, and the results represent fold induction of procollagen 1α2 mRNA compared to vehicle controls. Statistical analysis of differences between control dogs and drug-treated dogs was performed using the nonparametric Kruskal-Wallis test, followed by the Mann-Whitney U-test. A p value of <.05 was considered to be statistically significant.
Results
Three-month Toxicity Study
Exposure, Clinical Observations, and Clinical Chemistry
There was no obvious difference in Cmax and AUC between male and female dogs. The time for Cmax varied and was typically observed between one and eight hours after dosing. At all dose levels, there was a large inter-individual as well as intra-individual variability in Cmax and AUC (AUC on day 1 and AUC(0–24h) on days 15 and 85). However, based on the geometric mean values at each dose level, the exposure to AR-H047108 increased approximately in proportion to the increase in dose (Table 2).
Dose-related gastrointestinal disturbances were seen during the entire exposure period. Episodic low food consumption was seen among most animals in the high-dose group. Slight to moderate body weight loss was observed during the exposure period in four out of twelve high-dose–treated animals.
There was a moderate decrease in the levels of hemoglobin, erythrocytes, and hematocrit in both sexes in the groups given the high dose after three months of dosing. There was also a tendency toward a decrease in these variables in the group given the intermediate dose. A slight decrease was also observed after one month of dosing in both sexes in the high-dose groups. A slight increase in the activated partial thromboplastin time (APTT) values was noted in one male given the intermediate dose and in two females given the high dose (one during the dosing period and one after one month of recovery). Furthermore, after one month of high-dose treatment, a slight increase was noted in the mean activity of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase (ALP), and a slight decrease was noted in the mean plasma concentration of total protein and albumin. After three months of dosing, the changes were more pronounced. A marked decrease in albumin levels was noted after three months in two high-dose females.
Partial recovery was noted in most variables after one month without dosing. After three months, full recovery was obtained in both sexes in the APTT values and in the total protein concentration. The red cell variables and the albumin levels were still slightly decreased, and the liver enzyme activities in general slightly increased after the three-month recovery period.
Liver Histopathology and Immunohistochemistry
Moderate to marked fibrotic narrowing or occlusion of central veins and fibrosis of centrilobular sinusoids was observed in six of six dogs treated with AR-H047108 at the high dose level (240 μmol/kg), and in one of three males treated with AR-H047108 at the intermediate dose level (80 μmol/kg) (Figure 2B). The sinusoidal fibrosis was accompanied by strong immunoreactivity for smooth muscle actin (SMA) in the stellate cells, indicating activation to a myofibroblastic phenotype (Figure 2D). Moderate to marked centrivenular fibrosis/occlusion and sinusoidal fibrosis were still present after the three-month recovery period in six of six dogs treated with AR-H047108 at 240 μmol/kg. In one dog, the occlusion of central veins was associated with the presence of intraluminal clusters of cells, presumably dislocated endothelial cells. Prominent smooth muscle bundles surrounding hepatic sublobular veins, and an increase in thickness of the smooth muscle layer in central veins, were seen mainly in the recovery animals (Figure 2F).
In one dog, occasional necrotic hepatocytes were observed surrounding occluded, or partially occluded, central veins. Moderate to marked cytoplasmic vacuolar degeneration and ballooning were present in centrilobular hepatocytes in several drug-treated dogs, including recovery animals (Figure 2G). As shown by Oil Red staining, the empty vacuoles contained fat (Figure 2H). Deposition of fibrinogen, von Willebrand factor and tissue factor sometimes occurred around sublobular and central veins, and along centrilobular sinusoids, in the drug-treated dogs (Figure 3B).
There was also an increased staining of TGFβ in the centrilobular area, and a decreased staining of CD14 in the same region (Figures 3D, 3F). This effect was more pronounced in the main study of dogs than in the recovery group, indicating at least partial reversibility. There was only minor, if any, periportal involvement, and bile ducts and ductules were not affected.
Centrivenular inflammation, although present, was minimal at the three-month time point. TUNEL staining for detection of apoptotic cells revealed a slight increase in the number of apoptotic hepatocytes in dogs treated with AR-H047108. However, individual variation was large. Other immunohistochemical markers did not show any obvious differences between control dogs and dogs treated with AR-H047108.
Six-month Toxicity Study
Exposure, Clinical Observations, and Clinical Chemistry
The time for Cmax varied and was most commonly observed between one and three hours for AR-H047108 and between three and five hours for the metabolite on the different sampling days. At all dose levels on the different sampling days, there was a large inter-individual as well as intra-individual variability in Cmax and AUC(0–24h). The exposure to AR-H047108 increased with increasing doses. However, owing to the large variability and the dose reductions it was difficult to draw any further conclusions regarding time and dose dependency or regarding any sex-related difference in the exposure during the dosing period. At the lowest dose level (20 μmol/kg), the exposure to the metabolite was approximately two to six times higher than to the parent compound, whereas the exposure was similar at the highest dose level (240/180 μmol/kg). Thus, exposure to the metabolite relative to the test compound decreased with increasing doses (Table 2).
Two male high-dose animals were preterminally sacrificed after five months of dosing as a result of gradually deteriorated general condition and development of marked ascites. Dose-related gastrointestinal disturbances were seen during the entire exposure period. A gradual deterioration of the general condition, including episodic low food consumption, was seen among animals in the high-dose group. One intermediate- and four high-dose male dogs developed ascites during the last month of dosing. A dose-related decrease in body weight gain was seen in all dose groups, as compared to the control group.
After three and six months of dosing, there was a moderate decrease in the mean erythrocyte count, hemoglobin concentration, and hematocrit value in both sexes in the group given the high dose. After six months of dosing, the changes were slightly more pronounced and also noted in males given the intermediate dose. In addition, after six months, an increase in the mean leucocyte, neutrophil, and monocyte count was seen in the high-dose group.
After three and six months of dosing, a slight to moderate decrease in the mean plasma concentration of total protein was noted in males and females given the high dose. The albumin concentration decreased moderately to markedly in both sexes. After six months, a slight decrease in protein concentration and a moderate decrease in albumin were noted in both sexes in the intermediate-dose group, although of statistical significance only in the males. A decrease was also observed in the α-1-globulin fraction in both sexes given the high dose and in males given the intermediate dose. A slight to moderate increase in the mean AST activity was observed after three and six months in both sexes given the high dose, and after six months in the group given the intermediate dose. There was a small increase in the mean ALT activity in the intermediate- and high-dose males after three and six months. In the females there was a tendency toward an increase in all groups given the compound. ALP activity was slightly higher in both sexes in the intermediate- and high-dose groups than in the control group.
Liver Histopathology and Immunohistochemistry
After six months of treatment with AR-H047108, the histopathological picture showed many similarities with the three-month time point. Moderate to marked fibrous narrowing and occlusion of central veins accompanied by sinusoidal fibrosis (Figure 4B) and SMA-positive stellate cells of myofibroblast phenotype were present also at the six-month time point, in ten of ten high-dose (180 μmol/kg) and ten of ten intermediate-dose (70 μmol/kg) dogs.
Increased thickness of the smooth muscle layer in sublobular and central veins, congestion, and dilation of sinusoids and lymph vessels were observed in several drug-treated dogs. Scattered necrotic centrilobular hepatocytes occurred in one dog (Figure 4A), and slight centrilobular vacuolar degeneration/ballooning in two dogs. Fat deposition in hepatocytes, as shown by positive Oil Red staining, was present in three of five examined dogs.
The periportal and subcapsular regions were more involved at this time point than at three months, with minimal to slight periportal inflammation in three of ten dogs and subcapsular fibrosis in two of ten dogs. Some dogs also showed proliferation of biliary epithelial cells, in the form of a mixed typical/atypical reaction, in the subcapsular areas.
Deposition of von Willebrand factor along centrilobular sinusoids was more pronounced at the six-month time point compared to the three-month time point. In one dog, there was also an increased staining for fibrinogen within the centrilobular areas (Figure 4C–D). The effect on TGFβ staining (increase) and CD14 staining (decrease) observed at the three-month time point was not evident at the six-month time point. Other immunohistochemical markers did not show any obvious differences between control dogs and dogs treated with AR-H047108.
One-month Investigative Study
Exposure, Clinical Observations, and Clinical Chemistry
Cmax of AR-H047108 and its metabolite AR-H047116 was most commonly observed three and five hours after dose, respectively. There was a large individual variation and no obvious difference in exposure between the two dose groups. The exposure to the metabolite was between 1 and 1.6 times the exposure to the parent compound (Table 2).
Dose-related gastrointestinal disturbances were seen during the entire exposure period.
Hematology showed a tendency toward a decrease in hemoglobin, erythrocytes, and hematocrit in treated animals. In the group given the low dose (160 μmol/kg), a transient increase (two- to three-fold) was noted in the plasma activity of ALT after five days of dosing and an approximately two-fold increase after twelve and twenty-six days. There was a large individual variability in the effects on ALT. In the group given the high dose (240 μmol/kg), a three- to six-fold increase in the plasma activity of ALT was observed in two dogs after five days and in one dog after twelve days of treatment. An increase in the plasma activity of ALP was observed at both dose levels after twelve or twenty-six days of dosing. A slight, but consistent, decrease was noted in the plasma concentration of total protein and albumin during the dosing period. The albumin levels were lowest after twenty-six days (80%).
Liver Histopathology and Immunohistochemistry
Profile after One Week
All three dogs had a moderate centrivenular inflammation consisting of a mixed infiltrate (neutrophils, mononuclear cells, plasma cells). Many of the mononuclear cells were T lymphocytes, as evidenced by positive staining for the CD3 marker. There was discrete damage to the endothelial cells and minimal hemorrhage into the vessel wall (Figure 5A). Owing to inflammation, the central veins were minimally narrowed in one case, but there was no fibrotic occlusion of central veins in any of the dogs. A few necrotic centrilobular hepatocytes were observed in regions with affected central veins. In two dogs there were also scattered foci of inflammation in the parenchyma, involving a few necrotic hepatocytes. Minimal to slight centrivenular and sinusoidal fibrosis was present in the dogs, and there was an increased SMA staining of stellate cells with myofibroblast morphology, predominantly in the centrilobular areas (Figure 5D). Slight subcapsular inflammation and fibrosis was observed in one dog. Minimal to slight periportal inflammation was also present in all three dogs. Apart from the increased SMA staining of stellate cells, there were no other differences in the staining pattern of immunohistochemical markers between control dogs and dogs treated with AR-H047108.
Profile after Two Weeks
The histopathological picture was comparable to the one-week time point (Figure 5B), the only major difference being an increased deposition of von Willebrand factor along the centrilobular sinusoids in all three dogs.
Profile after One Month
In these dogs, inflammatory changes were either absent or minimal. Sinusoidal fibrosis was seen, and SMA-positive stellate cells with myofibroblast morphology were present to the same extent as at the earlier time points. One dog had a slight fibrotic narrowing of the central veins. Two dogs had slightly increased deposition of the von Willebrand factor along the centrilobular sinusoids. No obvious differences in the staining pattern of immunohistochemical markers between control dogs and dogs treated with AR-H047108 were observed.
Levels of Apolar Retinoids in Liver Samples
Liver retinoid concentrations, expressed as the sum of retinol and retinyl esters (nmol/g liver), were decreased in the treated groups at different time points as compared to vehicle-treated control animals. At day 8, high-dose dogs showed markedly decreased liver retinoid levels (83%; Figure 6A). Similar results were seen at days 14 and 29. No dose-response relationship was apparent, as both the low-dose and high-dose animals exhibited similar effects on the retinoid levels. The decrease in total retinoid levels in the drug-treated dogs was statistically significant (p =.0238) at all time points. The predominant reason for the decrease in retinoid levels was a decrease in retinyl palmitate and retinyl stearate (Figure 6B) concentrations, whereas there was only a slight decrease in retinol levels (Figure 6C).
Procollagen 1α2 mRNA Levels in Liver Samples
The levels of procollagen1α2 mRNA were increased from day 8 and throughout the study in dogs treated with the high dose (240 μmol/kg) of AR-H047108 (Figure 7). However, statistical significance (p =.0476) was reached only at the twenty-nine-day time point.
Discussion
The aim of this study was to investigate and characterize histopathological lesions in the liver of dogs treated with AR-H047108, a proton pump inhibitor acting through a reversible, K+-competitive inhibition of the H+K+-ATPase in the parietal cells. It is unlikely that the primary pharmacological action of proton pump inhibitors is involved in the pathogenesis of this canine hepatotoxicity. In toxicity studies with proton pump inhibitors such as omeprazole and esomeprazole, reversible changes in the gastric mucosa were observed in both dogs and rats (Carlsson et al. 1986; Carlsson 1989). In dogs, the changes are characterized by rugal hypertrophy, hyperplasia of oxyntic mucosal cells, and slight chief cell atrophy. The stomach weights are usually increased. The effects on the gastric mucosa result from the pronounced hypergastrinemia produced as a secondary effect of almost complete inhibition of acid secretion by the large doses of compound used in the toxicity studies (Carlsson et al. 1986). Similar changes in the gastric mucosa are seen in toxicity studies with P-CABs and were also present in dogs treated with AR-H047108 (data not shown).
In contrast to the well-documented gastric changes in toxicity studies with proton pump inhibitors, the liver has not been a target organ for toxicity with omeprazole or esomeprazole, neither in dogs nor in other animal species (Ekman et al. 1985). Accordingly, it seems that the hepatotoxicity of AR-H047108 more likely may be related to its chemical structure, that is, a substituted imidazopyridine.
There are only a few compounds with imidazopyridine structure currently on the market. Hepatotoxicity of varying degree has been reported for two imidazopyridines that act as anxiolytic drugs, targeting the central benzodiazepine receptor. Alpidem (Ananxyl), which also has a strong affinity for the peripheral benzodiazepine receptor (PBR), was withdrawn in 1995 because of severe hepatotoxicity in humans (Ausset et al. 1995; Baty et al. 1994). The development of ocinaplon, another imidazopyridine anxiolytic drug, was discontinued in 2005 because of an unacceptable rate of liver enzyme elevations in the phase III trial (http://www.dovpharm.com). These data could possibly indicate that the imidazopyridine structure carries a potential for hepatotoxicity. Our findings in dogs treated with AR-H047108 might serve to support this hypothesis. However, the absence of liver lesions in rats and monkeys treated with the same compound could indicate that additional factors are involved, or that the dog liver is more susceptible to certain types of toxic insults.
Gastrointestinal disturbances occurred frequently in dogs treated with AR-H047108 and may have contributed to the large inter- and intra-individual variability in Cmax and AUC observed in all three studies. Despite the large variability, the dogs were clearly exposed to the drug and moderate to marked histopathological changes in the liver were present in the vast majority of dogs treated at high doses. Although gastrointestinal disturbances may to some extent have affected a few of the clinical pathology parameters (i.e., albumin and total protein), the clinical pathology picture as a whole (decreased red cell variables and decreased total protein and albumin in combination with increased levels of ALT, AST, and ALP, and a slight increase in APTT) indicates an impaired liver function with a decrease in the synthesis of plasma proteins and damage to hepatocytes. It should be noted that hepatic fibrosis at the six-month time point was of sufficient severity to produce ascites in the majority of male high-dose dogs.
Overall, the histopathology of AR-H047108-induced liver toxicity is that of early inflammation (present after one week of dosing) in the central veins and centrilobular areas, paralleled by activation of stellate/Ito cells to myofibroblasts, and associated with sinusoidal and centrivenular fibrosis. At later time points (three and six months), the fibrotic changes predominate, with the addition of increased thickness of the smooth muscle layer in central and sublobular veins. Occlusion of central veins appears at the later time points and seems to include a combination of fibrosis and increased thickness of smooth muscular bundles with a sphincterlike appearance.
Early activation of stellate cells and transformation into collagen-producing myofibroblasts was confirmed by the observed increase in procollagen 1α2 mRNA and decrease in liver retinyl ester levels already at one week. Stellate cells represent approximately 5%–10% of the cells in the liver and are the principal storage site for retinoids (mainly as retinyl esters) in the body. Upon liver injury, stellate cells are known to be activated and undergo transformation, a process in which they lose their lipid content and become collagen-producing myofibroblasts (Wells 2005). The dramatic decrease of liver retinyl ester levels observed in the present study suggests an activation of the majority of retinoid-storing stellate cells. There were no histopathological signs of injury to the stellate cells, discarding the alternative possibility that the decrease of liver retinoids might be a result of cytotoxicity.
The observed early activation of hepatic stellate cells in these dogs, together with the venosinusoidal lesion, is reminiscent of so-called veno-occlusive disease (VOD), a toxic lesion involving obstruction of small intrahepatic venules and damage to the surrounding centrilobular hepatocytes and sinusoids (Shulman and Hinterberger 1992). It was originally described in Jamaicans who had consumed local herbal teas containing pyrrolizidine alkaloids. Nowadays, it is seen mainly as a complication following treatment of malignancy and bone marrow transplantation. Pathognomonic features include concentric narrowing or occlusion of terminal hepatic venules and sublobular veins, dilatation and fibrosis of centrilobular sinusoids, and hepatocyte necrosis in zone 3 (Bearman 2000).
VOD can be reproduced experimentally in rats and dogs with monocrotaline, a pyrrolizidine alkaloid. In the rat, the initial injury has been localized to the sinusoidal endothelial cells (SECs) through a metabolite of monocrotaline binding to actin in the endothelial cells, which causes a “rounding up” of the SECs, permitting blood to penetrate into the Disse’s spaces (DeLeve et al. 2003). The flow of blood within the Disse’s spaces dissects the sinusoidal lining away from the parenchyma, resulting in an embolism of sinusoidal lining cells that obstructs the microcirculation. There is also an aggregation of monocytes that contributes to the sinusoidal obstruction. Endothelial injury causes activation of the coagulation system, as evidenced by the immunolocalization of fibrinogen and factor VIII/von Willebrand factor in the perivenular zone (DeLeve et al. 2002).
Thus it seems that the toxins causing VOD are primarily targeting the SECs rather than the hepatocytes. Direct drug-induced toxicity to hepatocytes is overshadowed by the circulatory disruption caused by the damage to the SECs. The resulting ischemia, perhaps in conjunction with any direct effect on hepatocytes, leads to the ultimate parenchymal dysfunction (DeLeve et al. 1996).
If a VOD-like mechanism is responsible for the liver lesions in dogs treated with AR-H047108, the fact that similar lesions are absent in rats and monkeys remains to be explained. At present, available data do not support that differences in metabolism between the various species may be the crucial factor. A more plausible hypothesis might be that the special anatomic features of the dog liver play a role in the pathogenesis of drug-induced liver lesions. The first of these features is the presence of smooth muscle fibers in the sublobular and central liver veins (Arey and Simonds 1920). In the case of the sublobular veins, these muscle fibers act as sphincters, causing vasoconstriction and reduced venous drainage during conditions of experimental shock in dogs (Arey 1941). Smooth muscle sphincters in the sublobular veins are claimed to be unique for the domestic dog and a few other species (raccoons and seals). They are not present in humans, rats, guinea pigs, horses, cats, or rabbits (Arey 1941; Aharinejad et al. 1997). The sublobular venous sphincters could play a crucial role in regulating blood flow in the canine liver. They are reported to contract in response to endothelin-1 (ET-1) as well as to the compound 48/80, a histamine releaser (Aharinejad et al. 1997; Yamamoto 1998). Innervation of sublobular venous sphincters has not been demonstrated, which could suggest that instead, they respond to bloodborne substances or metabolites from nearby tissue (Aharinejad et al. 1997). It has been proposed that the action of ET-1 is a result of binding to ET receptors expressed on the smooth muscle cells (Zhang et al. 1994). In contrast, 48/80 does not stimulate smooth muscle cells directly but instead stimulates degranulation of endothelial cells and mast cells, causing the release of histamine and ET-1 (Yamamoto 1998).
Another special feature of the dog liver is the presence of mast cells along the endothelial lining of sublobular and central veins, as well as within the Disse’s spaces (Fujita 1964; Kobayashi et al. 1985). Mast cells in the Disse’s spaces have not been documented in any other vertebrate species than the dog (Kobayashi et al., 1985). Interestingly, the mast cells in the Disse’s spaces of dogs have been found to be in close contact with the stellate (Ito) cells, the primary source of collagen in the liver (Kobayashi et al. 1985). The relationship between mast cells and endothelial cells in the sublobular veins is remarkably close: it is not uncommon to find mast cells that seem to join the endothelial layer itself (Yamamoto 2000). The hepatic mast cells of dogs contain histamine, heparin and ET-1 (Yamamoto 2000). Release of histamine and ET-1 from mast cells in the canine liver is probably an important mechanism for regulation of hepatic blood flow, via constriction of the sublobular venous sphincters. ET-1 has been shown to exert a dual effect on hepatic stellate cells: stimulation of contraction and transformation into activated myofibroblasts (Rockey et al. 1998). Thus, release of ET-1 from mast cells and/or endothelial cells could be an early initiating event triggering the activation of stellate cells in dogs treated with AR-H047108.
Taken together, the histopathological findings and the results from immunohistochemical stainings and molecular analyses, as well as the published literature on VOD, are considered to support the following hypothesis for the development of liver lesions in dogs treated with AR-H047108:
The compound, or one of its metabolites, exerts a direct toxic effect on SECs, which causes the SECs to release ET-1 and thereby stimulates contraction of smooth muscle venous sphincters and causes congestion of blood in the sinusoids. In addition, ET-1 acts on stellate cells in the Disse’s spaces, causing them to transform into activated myofibroblasts. At the same time, ET-1 stimulates degranulation and release of TGFβ from mast cells in the Disse’s spaces, resulting in further activation of the stellate cells. Once activated, the stellate cells are likely to autonomously promote fibrosis, which is the predominant feature during later stages. These events also lead to disruption of the hepatic microcirculation, resulting in sinusoidal obstruction and secondary damage to centrilobular hepatocytes (fatty degeneration, necrosis).
In conclusion, this study provides support for the notion that a VOD-like mechanism involving early activation of stellate cells is responsible for the development of liver toxicity in dogs treated with AR-H047108. The absence of similar lesions in rats and monkeys treated with this compound could imply that the unique morphological features of the dog liver may be of importance in the pathogenesis. This implication could be relevant for humans in a broader sense, and especially highlights the problems in interpreting and translating results of dog toxicity studies where the liver is a target organ.