Spontaneous and Drug-Induced Hepatic Pathology of the Laboratory Beagle Dog,the Cynomolgus Macaque and the Marmoset

Abstract

This review focuses on the background hepatic pathology present in three of the most commonly used species in the safety assessment of drugs, namely the beagle dog, the marmoset and the cynomolgus macaque. Both the nonneoplastic and neoplastic pathology are reviewed with a discussion on the potential impact that significant background pathology might have on the interpretation of any drug-induced pathology during subsequent testing. Although specific instances, such as parasitological infection in wild-caught primates can pose problems of interpretation, in general the background pathology in both the dog and the nonhuman primates, is not significantly different from that seen in the liver of laboratory rodents and with experience should not pose significant problems for the experienced pathologist. The relative merits of the primate versus the dog as a choice of second species are also considered in some detail. Although there is an inbuilt prejudice that the primate will more closely mimic subsequent effects that might occur in man in the clinic, insofar as the liver is concerned, there are many instances where the dog has been more representative of human exposure and metabolism and there is little evidence to show that the nonhuman primate is consistently better than dog in predicting human liver toxicity. As with most areas of science, comparative toxicology would dictate that the more information gained, from as wide a range of species as is practical, will give the best assessment for any subsequent problems in the clinic. This pragmatic approach should prove to be more successful than one based entirely upon an assumption, and in many cases the assumption is incorrect, that the primate always predicts human toxicity better than the nonprimate, including the dog.

Introduction

Toxicity to the liver is reported to be the second most common cause of drug failure through adverse effects in clinical trials of potential drugs (Lumley and Walker, 1990; Olsen et al., 2000). Surveys of the literature suggests that preclinical toxicity testing in rodent and nonrodent species can predict subsequent adverse hepatic effects in the clinic in approximately 50% of the time although in some cases toxicity revealed in preclinical studies was ignored and compounds were progressed into clinical trials in man in spite of adverse hepatic effects in preclinical studies (Ballet, 1997). Evaluation of the relative predictivity of the rodent versus nonrodent studies for the prediction of human hepatotoxicity suggests that nonrodent species had a concordance of 63% between laboratory results and those obtained in the clinic, while rodent studies alone were only able to predict approximately 43% of subsequent hepatotoxic events in the clinic (Olsen et al., 2000). However, together studies conducted in both nonrodent and rodent species were able to predict approximately 71% of subsequent hepatotoxic events in man. It should be borne in mind however that considerable differences exist, between the calculated predictivity of the preclinical studies for clinical hepatotoxicity, dependent upon the therapeutic class under investigation, with anticancer therapy consistently outperforming other therapy areas in this respect (Schein, 1977).

Regulatory authorities throughout the world require preclinical toxicity testing in both a rodent and a nonrodent species before the potential drug is administered to man (Smith et al., 2001; Greaves et al., 2004). These studies establish the expected dose-limiting toxicity in man, and give therapeutic ratios between the toxicity seen in these species with reference to the expected therapeutic levels needed to elicit the desired pharmacological action of the drug (Bjornson, 2004). As such they are an essential part of the drug development process in ensuring the best method of limiting predictable toxicity in the initial clinical studies in man. Pharmaceutical companies have different policies with respect to which second species they routinely use as the nonrodent, but the beagle dog, the marmoset, and the Cynomolgus macaque are the 3 most common, nonrodent species, used for this purpose (Mansfield, 2003; Greaves et al., 2004). Various considerations determine the choice of second species to use and these include similarity of biochemical response to man, similar pharmacological distribution and behaviour of the target protein, similar absorption, metabolic, pharmacokinetic, and distribution profile for the chemical (Lowenstine, 2003).

Serious ethical considerations are taken into account when considering the use of the second species and will not only be applicable to the nonhuman primate since the use of the dog also needs careful consideration to avoid unnecessary experimental procedures. The ethics of such studies is beyond the scope of this review but the Fund for the Replacement of Animals in Medical Experiments (FRAME), amongst many, together with the industry are working together in refining the approach to the problem (Carlsson et al., 2004). Of consideration when selecting a suitable species as the nonrodent is the similarity to man in terms of metabolism, pharmacology etc., cost, experience in terms of familiarity with the foibles of the particular species chosen, and the existence of background disease, are important considerations if interpretation of the results of such studies is not to be compromised (Lowenstine, 2003). There is no such thing as an ideal surrogate species for man and while nonhuman primates are phylogenetically closer to man than is the dog, numerous examples exist where the latter has proven to be more predictive than have been subsequent primate studies (Bogaards et al., 2000; Tibbets, 2003), particularly in terms of pharmacokinetic parameters of absorption and metabolism. Part of the problem pertinent to primate studies is that there are a number of species available, all of which have particular advantages and disadvantages, the details of which have been reviewed in detail by Lowenstine (2003).

This current review therefore examines the available scientific literature with a view to describing the spontaneous hepatic pathology that the nonrodent species, the beagle dog, the cynomolgus macaque and the marmoset, exhibit and if, and how, this can affect the sensitivity and specificity of these preclinical studies for extrapolation of the data to man and will explore the relative merits of each species in their role in the development of novel drugs for human use. Lesions described will focus on those that could potentially compromise the interpretation of a potential drug-induced change and hence for the beagle dog the review will limit itself to those generally describe pathologies seen in dogs less than 3 years of age. For the nonhuman primates, the age range is wider, but at least for pharmaceutical studies, will still generally refer to primates less than 5 years of age.

Incidence of Spontaneous Nonneoplastic Lesions in the Beagle Dog

The general histology of the control dog liver is similar to that of other common laboratory species with the possible exception of the amount of fibrous tissue present around the central vein, which increases with increasing size of the blood vessel (Figure 1a). A review of the spontaneous lesions exceeding 1% in the Beagle dog from laboratory studies (Figure 2) has shown that the liver has the highest incidence of spontaneous microscopic lesions of all the organs in both the male and female (Hottendorf and Hirth, 1974; Maita et al., 1977; Oghiso et al., 1982; Morishima et al., 1990). In terms of the characteristic pathology observed in the dog liver, inflammatory lesions, including focal phlebitis, granuloma, and the inflammation accompanying local necrotic lesions, make up >95% of all hepatic lesions in both sexes of laboratory Beagles of less than 2.5 years of age (Figure 3). Generally speaking, parasitological disease in the beagle dog is not a significant problem since the majority of laboratory beagles are specifically bred and housed under superior conditions.

Incidence of Spontaneous Nonneoplastic Lesions in the Marmoset (Callithrix jacchus)

The general histology of the control marmoset liver is similar to that of other common laboratory species with the possible exception of the amount of glycogen present in the hepatocytes (Figure 1b). This can vary considerably dependent upon the diet given the primates, and in the photographs shown in Figure 1b the hepatocytes show considerable vacuolation distributed in a diffuse way across the liver lobule (Okazaki et al., 1996). On a less carbohydrate rich diet, or in older animals, the liver has an appearance similar to that of the laboratory rodents (Kurata et al., 1998). A review of the spontaneous lesions exceeding 1% in the marmoset (Figure 4) has shown that the liver has the second highest incidence of spontaneous microscopic lesions after the adrenal gland in the male and the third highest incidence after the adrenal and thymus gland in the female (Okazaki et al., 1996). The overall incidence of hepatic lesions was reported to be similar in both sexes. In this study, spontaneous pathology was described, and differentiated, in marmosets described as “healthy” (16–47 weeks), “weak” (12–67 weeks) and “dead” (12–50 weeks). In female marmosets, the most common spontaneous pathology observed in the “healthy” marmoset liver was extramedullary haematopoiesis, microgranuloma, necrosis, and sinusoidal vacuolation. In the so-called “weak” individuals, additional findings included cystic degeneration and inflammatory cell infiltration were seen, while in the “dead” animals, congestion and fatty change were apparent in addition to the findings described for the “healthy” individuals (Figure 5). Extramedullary haematopoiesis (EMH) has been described as the most common change observed in a second survey by Tucker (1984) where the pathology of 567 laboratory-bred cotton-eared marmosets was reviewed. EMH was reported to be extremely sensitive to frequent bleeds, of the kind that can occur in toxicity studies. The author emphasised the importance in differentiating chemical induction of haematopoiesis from that induced by the frequent blood sampling. In the condition referred to as “fatal wasting” syndrome, severe fatty change of the liver can be observed, together with atrophy of the gastrointestinal tract, salivary glands and gonads, haemosiderosis, and osteoporosis (Tucker, 1984). Haemosiderosis has also been described as a common finding in marmosets with diet being sited as an important determinant of the incidence and severity (Miller et al., 1997).

Incidence of Spontaneous Nonneoplastic Lesions in the Cynomolgus Macaque (Macaca fascicularis)

The general histology of the liver from a control Cynomolgus monkey is similar to that of other common laboratory species (Figure 1c). As with the marmoset the appearance can vary dependent upon the diet given, with variations in glycogen and fat being the 2 most common confounders observed. A review of the more common spontaneous lesions, exceeding a 1% incidence, in wild caught Cynomolgus monkeys (Figure 6) has shown that the liver has the highest incidence of spontaneous microscopic lesions of all of the tissues surveyed in both the male and female (Ito et al., 1992). Mononuclear cell infiltration was by far the most common spontaneous pathology observed, with lower incidences of hepatocyte vacuolation, parasitic granuloma, necrosis, and sinusoidal pigmentation (Figure 7). Multinucleated hepatocytes are a not uncommon finding in control macaques (Lowenstine, 2003) and severe fatty liver occurs in fatal fasting syndrome of overweight individuals of both the Cynomolgus and other macaques (Bronson et al., 1982).

Sporadic case reports of hepatic diseases such as herpes virus B infection and various parasitic and bacterial infections exist in the literature (Simon et al., 1993) and on occasions lesions may be observed in the livers of monkeys of various species. In particular wild-caught Cynomolgus macaques have been shown to have pseudotuberculosis (Yersiniosis) and Echinococcus cysts, amongst others (MacArthur and Wood, 1983; Abbott and Majeed, 1984; Bacciarini et al., 2004) and in the study by MacArthur and Wood, (1983) approximately 7% of clinically healthy monkeys were found to be excreting Yersinia pseudotuberculosis with a further 5% excreting Y. enterocolitica.

Incidence of Spontaneous Neoplastic Lesions in the Beagle Dog, the Marmoset and the Cynomolgus Macaque

The incidence of spontaneous neoplastic disease in the Beagle dogs used routinely in laboratory animal studies is extremely low. It should be borne in mind that this statement refers to those animals routinely used in laboratory studies whose age is generally less than 3 years of age (Hottendorf and Hirth, 1974), when it is well recognised that neoplasia in general, and in the liver in particular, is a disease of old age, even in the dog (Patnaik et al., 1980; Dobson et al., 2002). In a survey of 110 primary hepatic neoplasms in dogs, 55 hepatocellular carcinomas and 2 combined hepatocellular and cholangiocarcinomas were diagnosed (Patnaik et al., 1981a). In this series 24 adenocarcinomas of the hepatobiliary system were found among the 110 primary hepatic neoplasms: 22 of these were intrahepatic, 1 involved the extrahepatic bile duct and 1 the gall bladder. Histologically, 10 intrahepatic neoplasms were classified as cholangiocarcinoma, and 12 as bile duct cystadenocarcinoma (Patnaik et al., 1981a). Metastasis was found in 61% of the hepatocellular carcinomas (35 of 57), in contrast to the reported higher percent in man, with the authors suggesting the possibility of a different pathogenesis for these tumours in the dog as opposed to man. In addition 15 carcinoids were diagnosed on the basis of 3 distinct histological patterns: solid nests, cords or ribbons, and an alveolar pattern with rosettes (Patnaik et al., 1981b). Diffuse involvement of all liver lobes with severe hemorrhage and necrosis was seen in all of these cases.

Even in surveys of pet dogs the incidences of hepatic neoplasia remained extremely rare (Dobson et al., 2002). Of the hepatic tumours found spontaneously, cholangiocarcinomas were reported to the commonest phenotypic form of the tumours appearing in the liver (Bastianello, 1983) although the database is small. Case reports describing individual hepatic tumours of varying sorts do appear in the literature but they tend to be in older dogs of varying strains (Ivoghli and Strafuss, 1974; Trigo et al., 1982; Shiga et al., 2001).

The incidence of spontaneous neoplastic disease in the liver of nonhuman primates maintained under laboratory conditions is similarly very low (O’Gara and Adamson, 1972; Lapin, 1982; Thorgiersson et al., 1994; Reindel et al., 2000; Porter et al., 2004). In the most extensive study of its kind, which screened 373 animals, Thorgiersson et al. (1994) reported considerable difference between the incidences of neoplastic disease of all kinds between the African green monkey, the Cynomolgus, and the Rhesus monkey, with the African Green having the highest incidence at 8% (Figure 8). Amongst these neoplasms, 5 were found within the liver, and of these 5 hepatic tumours, 3 were lymphomas.

As with the dog, individual case reports of hepatic neoplasms are present in the literature, generally in older monkeys of various different species (Clark and Olsen, 1973; Borda et al., 1996). In a survey of 1,065 monkeys (32 species), only a single incidence of hepatocholangiocellular carcinoma was recorded in an adult African Green Monkey (Cercopithecus aethiops) (Siebold and Wolf, 1973) while in a second survey of 2000 Cynomolgus monkeys only 2 cases of spontaneous hepatic neoplasia were described (Reindel et al., 2000). The commonest hepatic tumour found spontaneously in the marmoset and cynomolgus macaque appears to be the hepatocellular carcinoma (Table 1) although the numbers involved mean that any additional case could change this situation. The histological features of the hepatic neoplasms are consistent with that defining the same neoplasms in other species, and altered hepatocellular foci (foci of cellular alteration), analogous to those described in rodents, have been reported in the liver of Cynomolgus monkeys (Reindel et al., 2000).

Drug-Induced Nonneoplastic Hepatic Changes in the Beagle and the Nonhuman Primate

Throughout the last 25 or so years considerable effort has been aimed at studying the species differences observed commonly in the response to any single chemical agent (McClellan, 1995; Holden and Tugwood, 1999; Collins, 2001; Ballatori and Villalobos, 2002). Such differences can be due to differing metabolic, pharmacokinetic and pharmacodynamic properties of the compound in question together with intrinsic differences in the target molecules for which the drug or chemical is directed. The peroxisome proliferating class of chemicals are one such important group that has engaged numerous investigators in innumerable hours of speculation and experimentation (Doull et al., 1999). The major reason for this effort is to discover the perfect surrogate species, which mimics the reaction that will occur when the chemical in investigation enters man (Smith et al., 2001). Experience suggests that no single species can perfectly mirror the human experience even though some reflect responses better, on average, than do others for certain types of toxicity. Hence a survey of the correlation between responses in experimental animal species and man indicates that for pharmaceuticals the dog on average will better predict toxicity than will the rodent although the sparseness of data on primates probably biases the result in favour of the canine over the primate (Olsen et al., 2000). The best predictor was a combination of studies in multiple species including the rodent, and for certain classes of drug, such as cytotoxic anticancer agents, the mouse alone was reported to best predict for human side effects (Newell et al., 2004).

An excellent example of a drug-induced hepatic change in the beagle dog is the vacuolation induced by excessive glucocorticoid administration also referred to as steroid hepatopathy (Dillon et al., 1983; Rutgers et al., 1995; Miller et al., 2000). Figure 9 shows the effect of chronic administration of prednisone to the beagle dog on the liver weight following both a subacute (14 days) and a chronic dosing regime (52 weeks). At the 14-day period a dose-dependent increase in liver weight was observed although this became less at the end of the 1-year study and the dose-response relationship was less marked. Figure 1d shows the histological appearance of hepatocyte vacuolation following chronic administration of prednisone to a dog. Staining of these hepatocytes for lipid or for glycogen showed the vacuolation to result from a marked accumulation of glycogen as a result of the pharmacological action of the glucocorticoid on glycogen synthesis. In addition to the hepatic changes shown other changes, classic of glucocorticoid overdose, can also be observed including adrenocortical atrophy, reduced bone mass with retention of epiphyseal growth plates in long bones, prominence of stromal adipose tissue in bone marrow, and atrophy of lymphoid tissues.

As with the situation in rodents, acute and chronic administration of sodium phenobarbitone to primates has been shown to induce liver growth and hepatocellular hypertrophy although the dose levels quoted in the literature tend to be higher than those used in rodents (Bullock et al., 1995; Ramana and Kohli 1996; Weaver et al., 1999). Similarly the dog has also been reported to be less sensitive to the barbiturate than are rodents (McKillop, 1985; Muller et al., 2000) and in general both the nonhuman primate and the dog have been reported to be less responsive to the induction of enzymes by so-called classic inducers of rodent hepatic enzymes. Figure 10 gives some data from the paper by Amacher et al. (2001) on the hepatic response of the dog liver to a number of rodent hepatic enzyme inducing chemicals. Results in this paper have shown that for some classical enzyme inducing chemicals similar, or identical, mixed function oxidases are induced by treatment of the beagle dog as are seen in rodent livers (Amacher et al., 2001). Although several papers suggest that the catalytic activity in the dog and primate are generally lower than that seen in the rodent liver (Orton et al., 1984; Souhaili-el Amri et al., 1986; Tanaka et al., 1999), several examples exist where the opposite was found to be true (Clark et al., 1979; Sharer et al., 1995; McKillop et al., 1998; Bogaards et al., 2000). In general it seems that for those chemicals that elicit a significant induction of metabolism in rodents, the responses in either the nonhuman primate or the dog will be similar.

The review by Shimada et al. (1997) suggests in general that oxidase activity from the liver of the cynomolgus monkey was more similar to human metabolism for Phenytoin (p-hydroxylation) than the dog, guinea pig, and rat, while for the metabolism of Phenacetin (o-deethylation), Coumarin (7-hydroxylation) and Nifedipine (oxidation) the dog had metabolic rates most similar to those observed with human microsomes, while the monkey microsomes generally had metabolic rates many fold higher than that seen in human microsomes (Shimada et al., 1997). The folly of simple comparisons based upon limited databases is illustrated by data in the same publication where oxidation rates between liver microsomes of the cynomolgus monkey and man were more similar than the other species included in the study if CYP oxidative activity was referred back to total CYP content in the livers. The guiding principle in the selection of the most suitable species for these types of studies would appear to be that each chemical/drug needs to be taken on a case-by-case basis if accurate representation of the human hepatic metabolism is the desired outcome (Tibbets, 2003).

Drug-Induced Hepatic Neoplastic Changes in the Beagle and the Nonhuman Primate

Although the number of carcinogenicity bioassays, in either the dog, or nonhuman primate, is limited, there are some excellent examples that illustrate that the species respond in the expected way to certain human and rodent hepatocarcinogens. The most comprehensive review of its kind refers to the 32-year history of primate carcinogenicity studies that were initiated by the National Cancer Institute in 1961, and that were first reported in 1994 (Thorgiersson et al., 1994). These studies looked at a range of carcinogenic compounds including a variety of food additives such as saccharin and 3-methyl diaminoazobenzene, environmental contaminants such as (DDT) and arsenic, N-nitroso compounds such as diethyl- and dimethylnitrosamine, and anti-neoplastic agents such as Cytoxan, Adriamycin, and melphalan. Figure 11 summarises the results for the food additives and environmental components. In summary, a number of the expected hepatic carcinogens delivered positive results in these assays, such as Aflatoxin B1 and the cooking-generated heterocyclic aromatic amine, 2-amino-3-methylimidazo[4,5- f ]quinoline (IQ) (Thorgiersson et al., 1996a), whereas other known hepatic carcinogens, such as 3-methyldiaminoazobenzene and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (MeIQX), failed to give any indication of an hepatocarcinogenic response in monkeys. Subsequent studies by Thorgiersson et al. (1996b) showed that the reason for the lack of a positive response in the monkey with MeIQX was due to the poor metabolising ability of monkey liver for this substrate, which requires activation to the proximate carcinogen. Of the rodent carcinogens tested, which included 2-acetyl aminofluorene (2-AAF), 3-methylcholanthrene and urethane, only the latter showed a positive result with the formation of adenocarcinomas and adenomas of the liver with additional tumours in the lungs, small intestine, and pancreas. Once again metabolic differences are expected to explain the apparent lack of activity of these known hepatic carcinogens in the monkey.

In a separate study, which looked at the carcinogenicity of cycasin and methylazoxymethanol (Sieber et al., 1980), in rhesus, cynomolgus and African green monkeys that received the compounds individually or together by the oral or intraperitoneal routes, for up to 11 years. The results showed that long-term administration of cycasin and methylazoxymethanol, known rodent carcinogens, to old-world monkeys induced the development of hepatocellular carcinomas, renal carcinomas, squamous cell carcinomas of the oesophagus, and adenocarcinomas of the small intestine.

Carcinogenicity studies in the dog are considerably more limited than those in the nonhuman primate but those that are available show that the dog reacts in the expected way to known hepatic carcinogens although firm comments on relative sensitivity to other species are inappropriate considering the paucity of data. Hence a 7-year carcinogenicity study in female beagle dogs (Stula et al., 1978) with 3,3′-dichlorobenzidine showed that 4 of the 5 dogs that survived to termination on the study developed carcinomas of the liver with all of the animals also developing transitional cell carcinomas of the urinary bladder. This compound has been shown to be an hepatocarcinogen in the hamster, rat, and mouse although the time course for development was, obviously, different in these 4 species. Similarly the hepatic carcinogen, diethylnitrosamine, also induced hepatic neoplasms in the liver of beagles given the compound chronically (Hirao et al., 1974).

Conclusions

Current guidelines for the safety assessment of drugs dictate that both rodent and nonrodent toxicity studies are required to be undertaken before any prospective drug is allowed to be administered to man (Newell et al., 2004). From a histopathology viewpoint the liver from both the dog and the nonhuman primate species commonly employed should pose little problems of interpretation for the experienced pathologist considering that the background pathology from laboratory bred, and raised, animals is limited and controlled. A constant question remains as to which is the most appropriate nonrodent species with the beagle dog and cynomolgus macaque being the most favoured at present. The factors pertaining to species choice have been outlined previously and I will not reiterate them at this point.

It is relevant to state that simple phylogenetic closeness is not the only consideration in determining the species that most closely mimics the expected effect in the clinic since this effect is multifactorial, occasionally including features that are species-specific and not mirrored by any of the available species. The pragmatic view is to take each new drug case separately, to gain as much information as is possible, particularly for interspecies differences in metabolism, and then to choose the most appropriate animal species to study, taking into account the entire myriad of factors involved.