The Dog as a Second Species for Toxicology Testing Provides Value to Drug Development

Abstract

The objective of the pharmaceutical industry is to develop new drugs that are safe for human use. In many cases, the accepted approach codified in guidance from regulatory authorities to assess the nonclinical safety profile of potential pharmaceuticals is to perform toxicity testing in two species. However, the use of a second species to establish the safety of new pharmaceuticals has been the subject of much scrutiny in recent years and the industry has been repeatedly challenged to reduce, refine, or replace some or all of the animals used to establish the safety of these pharmaceutical candidates. Specifically, the value of the dog in this testing paradigm has been questioned. Publications reviewing available data for marketed drugs suggest that for many drugs, the dog does not identify unique toxicities critical to human safety. The weakness of this approach, however, is that many of the cases where the dog (or any other species) has the greatest impact on drug development are cases for which development decisions based on safety concerns are not shared publicly. The European Federation of Pharmaceutical Industries and Associations (EFPIA) Preclinical Development Expert Group (PDEG) decided to share case studies collected from its membership and the literature to illustrate the value of the dog in drug development decision-making and clinical monitoring practices to protect the safety of trial subjects.

Introduction

The objective of the pharmaceutical industry is to develop new drugs that are efficacious and safe for human use. The international regulatory guidelines specify that nonclinical toxicology studies with new small molecule drugs be conducted in two mammalian species to assess the safety risks for the intended human patient or volunteer populations.^1,2 Typically, studies must be conducted in a rodent (typically rat) and non-rodent (typically dog or nonhuman primate) species and must be of sufficient duration to support the proposed duration of clinical use as specified in the study protocol or marketing application (up to the maximum duration specified in the guidelines). In recent years, the pharmaceutical industry has been challenged to reduce, refine, or replace some or all of the animals used to establish the safety of a new pharmaceutical, and the pharmaceutical industry recognizes the value of implementing these practices in drug development. Although alternative assays can help to identify some risks associated with new therapeutics, to date their use remains limited as they are unable to adequately model the biologic complexities and integrated pathophysiology of in vivo systems. As such, these alternative assays cannot completely replace safety assessment studies in animals to evaluate the safety of drug development candidates.

The choice of which non-rodent species to use for the nonclinical testing of a new drug is driven by science-based decision-making. Pharmaceutical companies rely on an understanding of the target homology, pharmacology, bioavailability, and metabolism of the molecule to justify the selection of the most relevant species for the nonclinical toxicity testing program for a given compound. Despite these considerations, the value of the dog to the regulatory testing paradigm has been questioned. Based on a review of animal and human data for 2366 marketed drugs, Bailey et al ³ argued that the dog provided little predictive value for adverse effects on humans. However, the analysis by Bailey et al ³ is biased by the lack of data for pharmaceuticals for which development was stopped because of findings observed in the animal toxicity studies, as this information is often not available publicly. In addition, it is highly unlikely that compounds for which significant toxicity has been observed in animal studies would be tested in humans, ⁴ and the availability of data in both the dog and human was necessary for a compound to be included in the analysis conducted by Bailey et al. ³

Retrospective studies which focus only on approved therapeutics such as the one conducted by Bailey et al ³ have certain shortcomings as they do not reflect how dog studies may have impacted drug development and protected the safety of clinical trial subjects. For example, retrospective analysis of approved therapeutics are unable to determine when animal studies have impacted clinical trial design, clinical dose selection and dosing regimen, safety monitoring of subjects and, in some cases, led to termination of development.

The European Federation of Pharmaceutical Industries and Associations (EFPIA) includes a preclinical development expert group (PDEG) focused on nonclinical assessments of drug safety. This group gathered case studies to illustrate the information the dog adds for decision-making in drug development, establishing the safety profile of new pharmaceuticals, and helping to determine monitoring practices during clinical development. The approach of providing case studies to illustrate the impact of conducting toxicology studies in a particular species has been used previously by the BioSafe organization to illustrate the benefit of studies in nonhuman primates for the safety evaluation of monoclonal antibodies. ⁵ These examples are not intended to be comprehensive, but instead to provide a cross-section of studies to illustrate the contribution of dog studies to the detection of relevant safety signals, and thus guiding safety measures and monitoring resulting in the protection of clinical trial participants and patients.

Methods

EFPIA PDEG conducted a survey of its members to collect cases from drug development programs where nonclinical safety studies in the dog provided information important to drug development. Members were asked for up to three cases where data from dog studies influenced one of these 4 categories:

1. early project development decisions

2. clinical monitoring and safety measures

3. late project termination decisions

4. product approval and labeling.

Member companies were asked to provide as much information as possible about each case study. Cases provided by the member companies were anonymized by personnel at the EFPIA office and the anonymized list of cases was provided to the subteam of EFPIA PDEG members analyzing the role of the dog as a second species for regulatory nonclinical safety testing programs (Supplemental Table 1).

In addition to the cases submitted by EFPIA PDEG member companies, the subteam searched PubMed Central^® and PharmaPendium^® for additional cases in the public domain to supplement the anonymized, proprietary information provided in the EFPIA PDEG member survey. Two cases were identified and added to the cases provided by EFPIA PDEG members.

The subteam sorted the cases according the 4 categories listed above and after reviewing the data selected 6 cases representative of these categories for a more detailed presentation in Section 4 (Illustrative Case Studies) of this review. Short narratives were prepared for the remaining cases and are presented in Section 5 (Brief Case Summaries) of this review. A significant challenge is that all of the case studies were blinded before being sent to the team, so some information that would have made the case studies more complete were unavailable to the investigators.

Results

The EFPIA PDEG member companies submitted 19 cases encompassing 9 therapeutic areas where the dog provided important information in drug development. Eighteen of these cases were based on toxicology data and one case was based on safety pharmacology data. Two additional cases were identified after a review of the literature and marketed product labeling and both of these cases were based on dog toxicology data. All cases were for small molecules, and the study durations for the selected studies ranged from 2 to 52 weeks with the majority of the cases being studies of 4 weeks or less duration. The cases include study results that influenced all stages of clinical development and are summarized in Supplemental Table 1. Member companies provided varying amounts of information on the cases they provided, with more information generally available for compounds in the later stages of development.

Five of the cases were selected for presentation of a more detailed case summary in this paper. The cases were selected to represent the following scenarios:

• The dog was the only pharmacologically relevant species;

• Notable effects in the dog predicted the observed human toxicity in early clinical trials;

• Metabolite safety qualification could only be achieved in the dog;

• Notable effects were observed in the long-term dog study resulting in clinical dose reduction and additional clinical monitoring;

• Notable effects not predicted from the pharmacologic basis of action were observed only in the dog toxicity studies and impacted clinical development and product labeling;

In addition, one detailed case summary was prepared for a class of marketed drugs. In this case, the findings in the dog toxicology studies for drugs in this class identified potential clinical adverse effects and influenced product labeling. The results of these six cases are presented below.

Illustrative Case Studies

Case 1

In this case study, the dog was the only pharmacologically-relevant toxicology species identified. Compound 2-4 is a small molecule targeting the 5-lipoxygenase (5-LO) pathway. 5-LO is expressed primarily in polymorphonuclear leukocytes, monocytes/macrophages, dendritic cells, mast cells, and B-lymphocytes and performs the first two catalytic reactions in the biosynthesis of leukotrienes. ⁶ The 5-LO pathway plays a role in the development of asthma, rheumatoid arthritis, and cardiovascular disease, ⁷ and Compound 2-4 was developed for a cardiovascular indication.

Compound 2-4 neither had affinity to the 5-LO pathway in rodents, nor significant pharmacological effect/potency on the target in rodents. The dog was the only pharmacologically responsive species appropriate for toxicity testing and was therefore used in the assessment of general toxicity. Four- and 13-week repeat-dose toxicity studies in the dog employed daily oral dosing and did not identify any findings of concern for administration of the Compound 2-4 to humans, and the results of these studies supported the decision to advance the clinical development of Compound 2-4.

Case 2

In this case study, the toxicity observed in the 4-week dog toxicity study predicted the toxicity seen in the initial Phase 1 clinical trial. Compound 7-2 is a small molecule HDM2 inhibitor. HDM2 binds and inhibits wild-type p53, and HDM2 and p53 regulate each other through an autoregulatory feedback loop. ⁸ Overexpression of HDM2 is oncogenic, and examples of tumors where HDM2 is commonly over-expressed include breast cancer, prostate, and sarcoma. ⁹ Compound 7-2 is an inhibitor of the HDM2:p53 protein-protein interaction being developed as a novel cancer therapy.

In a 4-week repeated dose toxicology study in dogs, Compound 7-2 produced several findings, including hemorrhagic syndrome and severe thrombocytopenia, which were seen in mid- and high dose animals (actual dose levels tested were not provided). Although bone marrow suppression was seen in rats, the dog was a much more sensitive species and reacted with severe thrombocytopenia which was not observed rats. In the clinical Phase 1 trial, prominent thrombocytopenia was observed in humans at exposure levels required for PD effects. Based on this, it was decided to terminate the project. In this instance the dog appeared to more adequately predict the severity of thrombocytopenia that was observed in the clinical trials.

Case 3

In this case study, toxicity testing in dogs provided critical information to support late-stage clinical development and market authorization for a new drug. Compound 4-1 is an orally administered small molecule currently marketed for treatment of multiple cardiovascular indications. Rat and dog were selected for the pivotal toxicology studies as Compound 4-1 is pharmacologically active in these species and adequate exposures to drug were achievable in both. Pharmacokinetic evaluations in humans revealed a major circulating metabolite (designated M1) after oral dosing that represented ∼20% of total drug exposure. The FDA guidance on safety testing of drug metabolites states that the toxicity of metabolites should be assessed at levels comparable to or greater than human exposures prior to starting large Phase 3 clinical trials ¹⁰ while the Q & A document accompanying the ICH M3(R2) guidance indicates that animal exposures to a metabolite at least 50% of the human exposure are generally considered adequate for safety assessment. ¹ In the case of Compound 4-1, dog was the only toxicology species in which these criteria could be met to ensure adequate safety evaluation of the M1 metabolite.

Compound 4-1 was given by oral gavage once daily to rats at doses up to 600 mg/kg for up to 6 months and to dogs at doses up to 600 mg/kg for up to 12 months. The highest doses tested in the chronic toxicity studies in both species were considered the no adverse effect levels (NOAELs). While adequate exposure margins to Compound 4-1 were achieved in both species, exposures to the M1 metabolite comparable to those in humans were only seen in dogs. At the maximal feasible dose in rats, the highest achievable exposures to M1 in rats were only ≤0.2× of human exposures at the highest recommended daily dose. By comparison, M1 exposures at the highest achievable dose in dogs were comparable (1.3×) to human exposures at the maximum recommended daily dose. Additional studies in which Compound 4-1 was administered to rats in the diet did not increase exposures to M1. In the end, toxicity testing in dogs was critical for achieving two key milestones during development of Compound 4-1. First, having adequate M1 exposures in dogs allowed the prevailing FDA requirement for safety assessment of major human metabolites prior to starting the Phase 3 clinical studies to be met. Secondly, data from the dog study was used to successfully address questions about the safety of M1 that arose during review of the Marketing Authorization Application (MAA).

In the absence of sufficient M1 exposures in at least one of the toxicology species, the only alternative would have been to attempt to synthesize the M1 metabolite and dose it directly to animals. Assuming M1 could have been synthesized, there would have been additional challenges to overcome in order to conduct a toxicology study of sufficient duration (likely 3 months) with the metabolite. The FDA guidance indicates that the dose route for the parent compound is preferred. However, given the low bioavailability of M1 in rodents and the similarity in structure between M1 and Compound 4-1, there were no guarantees that sufficient exposures to M1 could have been achieved after oral dosing due to poor absorption, lack of stability in the acid environment of the stomach, or subsequent metabolism not representative of the human situation after administration of Compound 4-1. Another option would have been intravenous administration of M1, but this dose route proved problematic with Compound 4-1 due to the excipients needed to achieve a solution.

In conclusion, the toxicology studies in the dog were critical for adequately assessing the safety of the M1 metabolite and, ultimately, supporting the marketing registration of Compound 4-1. If the dog were not available for use, the development of Compound 4-1, which has become an important medication for patients, would likely have been delayed.

Case 4

In this case study, toxicity testing in dogs identified a new potential target organ of toxicity, resulting in a reconsideration of the clinical dose and additional clinical monitoring. Compound 6-1 is a potent modulator of survival of motor neuron 2 (SMN2) messenger ribonucleic acid (mRNA) splicing by which it increases exposure to the functional SMN protein and was intended for oral use in patients with Spinal Muscular Atrophy (SMA Type 1). SMA is a rare (incidence between 1 in 6000 to 1 in 10 000 live births), debilitating and frequently life-threatening motor neuron disease. SMA has a predominantly pediatric onset, with the exception of Type 4 SMA which occurs in adulthood. Type 1 SMA is the most common genetic cause of death in infants, with more than 50% of affected patients dying before 2 years of age. The disease is inherited in an autosomal recessive manner.

Pharmacology studies were conducted in a mouse model of the disease and nonclinical safety profiling was conducted in juvenile rats and dogs. While rat and dog do not express SMN2, they still were considered adequate for safety profiling since the target organs of toxicity were comparable between the mouse PD model and the species used in safety profiling.

In a 52-week juvenile dog study, axonal degeneration was observed in dogs at all dose levels. No such changes were observed in a juvenile rat study at comparable exposure. Consequently, the company voluntarily issued an urgent safety measure consisting of lowering the clinical dose levels and implementing additional safety monitoring (eg, nerve conduction measurements) to minimize the risk to the pediatric patients enrolled in the clinical trial at the time of the finding was reported. Some patients showed worsening of motor skills and increasing respiratory muscle weakness about 6 to 7 weeks after dose reduction suggestive of a loss of efficacy, and as a result, the company then decided to offer patients’ families the option of returning to the original dose. The clinical development of the drug has been stopped meanwhile since other treatment options have become available to SMA patients.

The 52-week toxicology study in juvenile dogs identified target organs of toxicity that were not observed in a 26-week juvenile rat study at comparable exposures. ¹¹ Based on the findings a voluntary reduction of the clinical dose and additional monitoring endpoints of patients were implemented. Although the voluntary reduction of the clinical dose was ultimately rescinded, the additional clinical monitoring continued for the remainder of the ongoing clinical trial in an effort to protect patient safety.

Case 5

In this case study, notable effects were observed in the dog toxicity studies and similar effects were not identified in the corresponding rat toxicity studies, nor were they predicted from the pharmacologic basis of action. These effects warranted additional nonclinical and clinical investigation to understand the potential human relevance of the dog finding and the dog findings were ultimately included in product labeling.

Compound 8-1 is an orally administered small molecule that is marketed globally for a variety of urogenital and cardiovascular diseases. A comprehensive toxicology program was conducted to support clinical development and product registration. The rat and dog were the species selected for the repeat-dose toxicity studies. In the dog repeat-dose toxicity studies of 3-12 months duration, nonreversible degeneration and atrophy of the seminiferous tubular epithelium indicating reduced spermatogenesis was noted during histologic examination. In the 12-month dog toxicity study, the findings occurred at all dose levels tested. Based on area under the curve (AUC), the lowest dose in dogs produced systemic exposures that were similar to human exposures at the maximum recommended human dose (MRHD). There were no adverse effects on fertility in rats or on the male reproductive system in rats or mice at systemic exposures up to 14 times greater than those that occur at the MRHD. This included a lack of testicular findings in the rodent carcinogenicity studies which represent near lifetime treatment. Based on the findings in dogs, clinical trials were conducted to assess the risk to male reproduction in humans. A series of three, randomized, placebo-controlled clinical trials were conducted that included healthy males as well as male patients. The endpoints included sperm concentration and hormone levels. These clinical studies did not detect adverse effects on either spermatogenesis or hormone concentrations. The clinical data indicated that the dog findings were not predictive of adverse effects on the male reproductive systems of either healthy subjects or patients.

The approved product labeling by the United States Food and Drug Administration (FDA) notes the lack of rat or mouse findings and describes the effects on dog testes in the section on Impairment of Fertility. The clinical data noting a lack of effects on male reproductive parameters is described in the Pharmacodynamics section of the label. There are no warnings or precautions regarding male reproductive effects.

Case 6

In addition to having an impact on drug development decisions, findings in dog studies have had an impact on product labeling for marketed drugs. Statins are orally administered HMG-CoA reductase inhibitors first approved in the 1990’s to lower low-density lipoprotein-C (LDL-C) and triglycerides in the plasma by inhibiting HMG-CoA reductase in the liver, leading to a reduction of very low density lipoprotein (VLDL) synthesis as well as by increasing clearance of LDL-C via the LDL receptor in all lipoprotein fractions. ¹² The toxicology program conducted to support marketing of all statins included (at a minimum) studies in mice, rats, and dogs, and the FDA summary basis of approval for atorvastatin states that for statins, the dog is the best model for humans in terms of metabolism, pharmacodynamics effects, and toxicity. ¹²

As a class, the primary toxicity information contained in the Nonclinical Toxicology section of the labeling (USPI) for statins describes the CNS effects observed in dog toxicity studies conducted with these agents. CNS vascular lesions in dogs are considered a class effect of statins, and CNS vascular lesions characterized by perivascular hemorrhages, edema and mononuclear cell infiltration of perivascular spaces are noted in the labeling of most agents in this class. In addition, class labeling for statins contains a description of dose-dependent optic nerve degeneration (Wallerian degeneration of retinogeniculate fibers in dogs) which was observed in dogs given lovastatin at a dose (180 mg/kg/day) that produced plasma drug levels about 30 times higher than the mean drug level in humans taking the highest recommended dose. These findings were originally observed with lovastatin, and became class labeling for all statins. ¹³ Examples of agent-specific labeling reflecting additional CNS effects noted in dog toxicity studies can be found in the labeling of atorvastatin, lovastatin, fluvastatin, pravastatin, simvastatin, cerivastatin, and rosuvastatin.

In general, no CNS lesions were observed in mice or rats after oral administration of statins for up to 2 years. However, for fluvastatin, CNS effects, such as decreased activity, ataxia, loss of righting reflex, and ptosis were seen in the 18-month carcinogenicity study in mice (50 mg/kg/day), the 6-month hamster study (40 mg/kg/day), and in acute high-dose studies in rats and hamsters (50 mg/kg), rabbits (300 mg/kg), and mice (1500 mg/kg). In mice, the CNS toxicity in the high-dose acute studies with fluvastatin was characterized by vacuolation in the ventral white columns of the spinal cord (mice; 5000 mg/kg) or edema with separation of myelinated fibers of the ventral spinal tracts and sciatic nerve (rat; 1500 mg/kg). ¹⁴

Ocular findings are often noted in dogs treated with statins but are not seen in other species. Cataracts were observed in dogs after chronic treatment with cerivastatin. ¹⁵ Cataracts and/or lens opacities were seen in dogs treated with pitavastatin for 52 weeks a dose level of 1 mg/kg/day (9 times the human exposure based on AUC at the highest human dose of 4 mg/day. ¹⁶ Cataracts were also noted in dogs treated with simvastatin for 3 months (90 mg/kg/day) or 2 years (50 mg/kg/day), resulting in respective mean drug plasma levels 19 or 5 times higher than the mean plasma levels in humans taking the highest dose, 80 mg/day (FDA, 2015). Cataracts were seen in dogs treated with lovastatin for 11 and 28 weeks (180 mg/kg/day) and 1 year (60 mg/kg/day). ¹⁷

Chronic treatment (2 years) of dogs with fluvastatin (1, 8, or 16 mg/kg/day) was associated with the development of prominent posterior Y suture lines in the ocular lens. ¹⁴ Corneal opacity was seen in dogs treated with rosuvastatin for 12 (30 mg/kg/day) or 52 weeks (at 6 mg/kg/day), resulting in systemic exposures 60 and 20 times the systemic exposure at the maximum human dose of 40 mg/day based on AUC. Additional ocular findings noted in dogs treated with rosuvastatin for 4 weeks included retinal dysplasia and retinal loss. These findings were observed at 90 mg/kg/day with systemic exposures 100 times the human exposure at 50 mg. Similar findings were not noted at doses ≤30 mg/kg/day (systemic doses of ≤60 times the human exposure at the highest dose. ¹⁸

Two findings observed in dogs, cerebral hemorrhage and cataracts have been reported in humans taking various statins. In some cases, these findings occurred in clinical trials but many were reported in postmarketing safety reports. ¹⁹

The toxicology studies in the dog were critical for characterizing the potential adverse effects associated with the use of statins that could not be easily detected in humans during clinical development. These risks are described in the labels for the various agents, and serve to alert prescribing physicians of these potential adverse effects.

Brief Case Summaries

Short summaries of the additional cases submitted by member companies or identified in the literature are also included in this publication, because each case contributes to our understanding of the value that studies conducted in dogs provide in drug development.

Case 7

Compound 1-1 is a small molecule inhibitor of a pathway for adenosine triphosphate (ATP) production intended for an oncology indication. The clinical route of administration is oral, and the compound was administered orally in the repeated-dose toxicity studies conducted. In a 14-day toxicity study in dogs, irreversible adverse bilateral retinal degeneration was seen at the mid- and high-dose levels corresponding to the targeted human exposure. No such findings were seen in the corresponding toxicity rat study. The toxicity identified in the dog toxicity study resulted in termination of the development of the compound prior to the initiation of clinical trials.

Case 8

Compound 1-2 is a small molecule protease inhibitor intended to be given orally for treatment of viral infections. The intended clinical route of administration was oral, and the compound was administered orally in the repeated-dose toxicity studies conducted to enable FIH studies. In the 2-week toxicity study in dogs, neutrophilic portal hepatitis, increased liver enzymes and increased total bilirubin were seen at the mid- and high-dose levels, at a dose level corresponding to 13× the expected clinical exposure level. Similar findings were not seen in the 2-week toxicity study in rats. The occurrence of this toxicity resulted in termination of the development of the compound prior to the initiation of clinical trials.

Case 9

Compound 3-1 is a small molecule soluble guanylate cyclase (sGC) stimulator intended for treatment of an unspecified cardiovascular indication. The intended clinical route of administration was oral, and the compound was administered orally in the repeated-dose toxicity studies conducted to enable FIH. While no adverse effects were seen in the rat, strong gastric side effects (eg, reflux, vomitus) were seen in the 2-week toxicity study in dogs at the highest tested dose. The toxicity identified in the dog toxicity study resulted in clinical monitoring for similar effects during clinical development.

Case 10

Compound 2-3 is a small molecule phosphodiesterase 4 (PDE4)-inhibitor, intended as a treatment for an unspecified respiratory condition. The intended clinical route of administration was inhalation, and the compound was administered by inhalation in the toxicity studies conducted prior to FIH. In a non-pivotal 2-week toxicity study in dogs, 1.2 mg/kg/day (inhaled dose) caused respiratory signs, acute rhinitis and purulent bronchopneumonia. No findings were seen in the respiratory tract in the 7-day rat studies. The findings in the dog study resulted in termination of the compound prior to the conduct of pivotal toxicology studies or clinical trials.

Case 11

Compound 1-3 is a small molecule kinase inhibitor intended for treatment of an unspecified rheumatologic condition. The intended clinical route of administration was oral, and Compound 1-3 was administered orally in the repeated-dose toxicity studies conducted. In the 4-week toxicity study in dogs, expansion of the paracortical region of the mesenteric and popliteal lymph nodes by larger, immature lymphocytes and adverse decreased red cell mass were seen at all dose levels, including the lowest dose level, which corresponded to the intended clinical dose level. No such findings were seen in the corresponding toxicity rat study. The toxicity identified in the dog toxicity study resulted in termination of the compound prior to the conduct of clinical trials.

Case 12

Compound 2-2 is a small molecule PDE4 inhibitor intended for treatment of an unspecified respiratory indication. The intended clinical route of administration was by inhalation, and the compound was administered by inhalation in the toxicity studies conducted prior to FIH. In the 4-week toxicity study in dogs, nasal and pulmonary findings (fibropurulent bronchopneumonia and acute rhinitis) were seen at all doses levels of 0.05, 0.3, 1/0.6 mg/kg/day (inhaled dose). Clinical respiratory signs developing during the first or second week of dosing leading to termination of dosing. Rhinitis, but no pulmonary findings, was observed in the 4-week rat toxicity study, but only at high dose (10 mg/kg/day), and the NOAEL in rats was determined to be 1 mg/kg/day. The toxicity identified in the dog toxicity study resulted in termination of the compound prior to the initiation of clinical trials.

Case 13

Compound 5-2 targeted a G protein-coupled receptor as a potential treatment to provide glycemic control in diabetic patients. Compound 5-2 was administered by oral daily gavage in 1-month repeat-dose toxicity studies in rats and dogs. The primary target organ of toxicity in dogs was the testis, where microscopic evaluation revealed seminiferous tubule degeneration, syncytial cell formation, and slight increases in Sertoli cell vacuolation at the high dose. In addition, sperm granulomas and increased cell debris were observed in the epididymis. The projected exposure margin at the NOAEL in dogs (based on AUC) was 15-fold relative to the proposed starting dose for Phase I, while the exposure multiple based on testicular changes was 24-fold. There were no testicular findings observed in the 1-month rat study, indicating dog was the most sensitive nonclinical species for this effect. Based on Compound 5-2 exposures at the dog NOAEL, there appeared to be an adequate safety margin to support the start of human dosing. However, there were insufficient safety margins to allow for adequate exploration of the pharmacologic properties in the clinic, so development was terminated prior to the start of first-in-human (FIH) studies.

Case 14

Compound 7-3 is a small molecule specific tyrosine kinase inhibitor against vascular endothelial growth factor receptor-3 (VEGFR-3) intended for treatment of an unspecified oncology indication. The clinical route of administration was intended to be oral, and Compound 7-3 was administered orally in the repeated-dose toxicology studies conducted. In a 28-day toxicity study in dogs, severe neurological clinical signs including incoordination and stereotypic movements, tremors, hypomotility, rigidity, convulsions, and opisthotonos were seen at the mid- and high-dose levels, at exposures below the anticipated pharmacologically active exposures based on data from a mouse pharmacology model. No histopathologic finding correlated to the clinical signs. Similar findings were not observed in the corresponding 28-day rat toxicity study. The toxicity identified in the dog toxicity study resulted in termination of the compound prior to Phase 1 because of the absence of a mechanistic explanation for the central nervous system effects seen in the dog toxicity study.

Case 15

Compound 7-4 is a small molecule multikinase inhibitor optimized against human spleen tyrosine kinase (Syk) and was intended for treatment of an unspecified rheumatology indication. The clinical route of administration was intended to be oral, and Compound 7-4 was administered orally in the repeated-dose toxicity studies conducted. In a 28-day toxicity study in dogs, mortality occurred at the high dose of 300 mg/kg/day. Upon microscopic evaluation, alveolar inflammation was observed in the lungs. The exposure (based on AUC) at the anticipated clinically efficacious dose was 8760 ng*h/mL, based on an in vivo rat efficacy model. The NOAEL in the 28-day dog study was determined to be 100 mg/kg/day, corresponding to AUC of 7090 ng*h/mL, while the AUC at the lowest observed adverse effect level (LOAEL) was 16800 ng*h/mL. The project was terminated prior to Phase 1 because (a) the margin of exposure between NOAEL and the dose at which mortality was observed was very narrow; (b) the pulmonary finding was not monitorable; and (c) the mechanism of toxicity was not determined.

Case 16

Compound 7-6 is a small molecule selective adenosine A1 agonist intended for treatment of an unspecified cardiovascular indication. The clinical route of administration was oral, and Compound 7-6 was administered orally in the repeated-dose toxicity studies conducted prior to Phase 1. In a 14-day toxicity study in dogs, leukoencephalomalacia, decreased food consumption, reduced motor activity, incoordination, and decreased platelet counts were seen at the dose level of 150 mg/kg/day. No such findings were seen in the corresponding toxicity rat study. These findings were considered for the selection of the starting dose, the maximum clinical dose, and clinical monitoring for the Phase 1 FIH study.

Case 17

Compound 7-5 was an orally-active inhibitor of the Kv1.5 potassium channel being developed for treatment of an unspecified cardiovascular indication. The proposed clinical route of administration was oral. To support the FIH studies, 1-month oral toxicity studies were conducted in rats and dogs. In dogs, hepatic centrilobular necrosis and inflammatory infiltrates were observed in the liver at all doses in an initial study, and at the mid- and high doses in a second study. While a no-effect level for the liver toxicity was identified in the second study, the exposure multiple at the NOAEL was only 1.5× the anticipated pharmacologically-active exposure in humans. There was no evidence of liver toxicity in the companion rodent study, and it was hypothesized that the observed liver toxicity might be a dog-specific finding related to a species-specific anatomical feature. To test this hypothesis, a subsequent toxicity study was conducted in monkeys in which similar findings were observed at exposures that, like the dog, did not provide an adequate safety margin. The replication of the finding in a second large animal species indicated that liver toxicity, a finding not detected in the rodent toxicity studies, was likely to occur in humans. Therefore, the decision was made to stop development Compound 7-5 prior to initiation of human clinical testing.

Case 18

Compound 5-1 targeted a G protein-coupled receptor and was intended to provide glycemic control in diabetic patients. The proposed clinical route of administration was oral. Compound 5-1 was administered daily by oral gavage in 4-week repeat-dose toxicity studies in rats and dogs. In the dog study, microscopic examination detected multiple foci of epicardial inflammation. The inflammation did not extend into the subjacent myocardium and was not accompanied by any correlative live phase, macroscopic or clinical pathology findings. The change was considered adverse and therefore a NOAEL was not determined. Similar findings did not occur in rats despite achieving similar systemic exposures. Because the change in dogs was limited to the epicardium, traditional cardiac biomarkers such as cardiac troponin I (cTnI) were not considered to provide a clinical monitoring strategy that would allow dosing into the range of exposures that produced epicardial inflammation. In a follow-up dog toxicity study with increased group size and revised dose levels, a no-effect level for epicardial inflammation in the dog was established. The NOAEL from this study supported conduct of a clinical trial; however, the dose range explored was relatively narrow as the epicardial finding limited the proposed top clinical dose. As such, these dog findings guided clinical trial design in terms of both the dose range which could be studied as well as the nature of cardiac monitoring.

Case 19

In this case study, the findings in the chronic (6-month) toxicity study in dogs, which were not observed in corresponding rat studies, together with clinical findings related to the dog findings contributed to the decision to stop the development of this compound prior to the initiation of Phase 2 testing.

Compound 7-1 is an orally administered small molecule that was being developed for a gastrointestinal indication. This compound retards gastric emptying, although the molecular mechanism of action is unknown. The repeated-dose toxicity studies were conducted in rats and dogs, and toxicity studies of 1-, 3-, and 6-months duration were conducted in the rat and dog to support conduct of Phase 1 and 2 clinical trials. Hepatotoxicity was observed in the 6-month toxicity study in the dog at systemic exposure multiples of less than one for the parent compound and approximately 20× for the main metabolite. Specific findings observed in this study included bile duct hyperplasia, periportal fibrosis, and scattered small foci of hepatocellular degeneration/necrosis (all minimal) associated with increases in alkaline phosphatase (ALP), alanine aminotransferase (ALT), and gamma-glutamyl transferase (GGT). Similar findings were not noted in the 1- and 3-month studies conducted in this species. Except for hepatocellular hypertrophy, there was no evidence of liver toxicity observed in any of the rat studies. Elevated liver enzymes were observed in the Phase 1 clinical trial. This finding, in conjunction with the hepatotoxicity observed in the 6-month dog study, led to the termination of the project prior to the initiation of Phase 2 clinical studies.

Case 20

Compound 3-2 is a small molecule phosphatidylinositol (PI3K) inhibitor intended for treatment of an unspecified oncology indication. The clinical route of administration was not specified, but in a cardiovascular safety pharmacology study conducted prior to the FIH study, dogs were administered single doses (cumulative in anesthetized dogs) of Compound 3-2 using the intravenous route of administration. At the lowest tested dose, peripheral vasoconstriction and an increase in diastolic pressure were recorded at an exposure that was 2× the exposure at the maximum tolerated dose in the Phase I study. As a result of the findings in the dog cardiovascular safety pharmacology study, a clinical cardiovascular monitoring plan was developed. Similar findings were observed in clinical trials, and the clinical monitoring plan developed based on the results of the dog safety pharmacology study were sufficient to detect these findings.

Case 21

This case is the second case from the literature; however, unlike Case 6, the findings in the dog chronic toxicity study resulted in the termination of clinical development of Compound L-1. Compound L-1 is an orally administered, small molecule liver-selective thyroid hormone receptor agonist that was intended for reducing the independent risk factors involved in the development of atherosclerotic cardiovascular disease, specifically dyslipidemia. Repeated-dose 6- and 12-month toxicity studies were conducted in dogs. Damage to cartilage was observed in dogs that were given Compound L-1 for 12 months. This cartilage damage was only seen after 12 months exposure. Damage was apparent in all animals treated with the high dose but was also observed in some dogs in the lower dose groups. No such findings were seen in the control group or in the dogs that were a part of a concurrent 6-month study. The findings from the 12-month study resulted in termination of the ongoing Phase 2 and Phase 3 studies.

Discussion

The use of two species to establish the safety of new pharmaceuticals has been the subject of much scrutiny in recent years. As a result, the pharmaceutical industry has been challenged to reduce, refine, or replace some or all of the animals used to establish the safety of these new pharmaceuticals. International regulatory guidelines specify that nonclinical toxicology studies with new small molecule drugs be conducted in two mammalian species to assess the safety risks for the intended human patient or volunteer populations.^1,2 For biotherapeutics, if two species are pharmacologically relevant then both species should be used for short-term general toxicity studies. ²⁰ The approach of using two species for repeat-dose toxicity testing is rooted in the fact that all models have limitations and by employing two different species the nonclinical safety data is more robust. In an analysis of nonclinical testing paradigms enabling First-In-Human clinical trials, a positive finding in both rodent and non-rodent species increased the Positive Predictive Value (PPV) of the test paradigm. While the rat is generally recognized as the primary species for toxicity testing, the beagle dog is considered the default second species for small molecule drugs for toxicology testing programs based on its size, tractability, and the availability of a wealth of historical control data. ²²

To reduce the number of animals used in the evaluation of new pharmaceuticals, the pharmaceutical industry has embraced the principles of the 3Rs––reduce, refine, replace. However, although significant progress has been made in the areas of ocular and dermal in vitro alternatives to animal testing, the development of in vitro models to replace in vivo studies in laboratory test species is slow and challenging due to the complex biologic network responses, metabolism and biologic pathways in a full organism. In addition, alternative test systems such as microphysiologic systems, often reflect only one single aspect or a limited set of pre-defined endpoints and cannot cover the full spectrum of responses in the complex biologic system. These systems, like all complex models, require extensive intra-laboratory standardization and validation efforts that take years to complete. ²² Significant progress has been made in a number of areas to refine or reduce animal including marked reduction in acute toxicity testing, ²³ refined approaches to safety pharmacology and developmental and reproductive toxicity testing, ²⁴ careful consideration of the use of recovery animals, ²⁵ and considerations of when testing in a single species may be appropriate. ²⁶

Over the past 8 years, the value of a second species, and in particular, the dog, to the prediction of the toxicity of new small molecule compounds to humans based on evaluation using the accepted regulatory testing paradigm has been questioned.^3,27 Publications reviewing available data for marketed drugs suggest that for many drugs, the dog does not identify unique toxicities critical to human safety, and that the absence of toxicity in rodents or a second species (dog, minipig or nonhuman primate) does not predict a similar lack of toxicity in humans. ²⁷ In contrast, Monticello et al ²¹ found that the current toxicity testing paradigms have effectively supported safety in Phase 1 clinical trials and identified that a lack of findings in animal studies, particularly dogs, strongly predicts safety in Phase 1 trials. This analysis also showed that identifying the same target organ in the rodent and nonrodent increased the positive predictive value for humans, while absence of target organ toxicity in two species increased the negative predictive value. In an analysis that assessed adverse drug reactions in clinical trials and the associated toxicity data, Tamaki et al ²⁸ determined that current nonclinical testing has a greater predictive value for what were termed “clinically-significant’ adverse drug reactions and certain adverse drug reactions such as hematologic, ocular, infection, and application site reactions had a relatively high correlation (<70%) between animals and humans. Nonrodent species have been shown to be invaluable for assessing human risk of QTc prolongation for drug candidates. An analysis by Vargas et al demonstrated that a negative nonrodent QTc assay for a drug candidate significantly lowered the probability of a QTc interval and/or ventricular arrhythmia liability occurring in clinical studies. ²⁹ Similarly, a retrospective evaluation of 113 small molecule compounds by Ewart et al showed strong concordance between drug-related QTc effects in dogs and humans, especially with respect to negative predictive value. ³⁰ In addition to these quantitative approaches to understanding the value of nonclinical safety testing, Brennan et al gathered a set of case studies to examine the value of studies in nonhuman primates for risk assessment of monoclonal antibodies. ³¹ Eighteen cases were collected via a survey and found that the nonhuman primate studies identified notable safety concerns in some cases and in other cases identified that a pathway that had notable theoretical safety concerns, but for which monkey studies helped define dose levels and dose frequency that could be administered safety. Similar to the collection of NHP case studies by Brennan et al, ⁵ the objective of these case reviews from dog studies was to understand if dog toxicity studies were providing relevant information for safety assessment. In a subset of the cases in this paper, the dog data identified what was considered an important safety concern and led to the termination of the compound under study; however, the cases also illustrate that findings from the dog study were impactful in terms of altering clinical dose range and/or safety monitoring, providing the only relevant model to assess the pharmacology or metabolite profile, or inform product labeling. While this collection of case studies does not provide a quantitative assessment of the predictivity of the dog model for human clinical trials, they provide relevant examples to further our understanding of the role of animal models in pharmaceutical safety assessment.

Supplemental Material

Supplemental Material - The Dog as a Second Species for Toxicology Testing Provides Value to Drug Development

Supplemental Material for The Dog as a Second Species for Toxicology Testing Provides Value to Drug Development by Nancy Bower, William E. Achanzar, Virginie Boulifard, Peter R. Brinck, Birgit Kittel, and John L. Vahle in International Journal of Toxicology