Translational Relevance of Rodent Models to Predict Human Liver Disease

Abstract

Animals models are essential to understand the complex pathobiology of human diseases. George Box’s aphorism based on statistics “All models are wrong, but some are useful” certainly applies to animal models of disease. In this session, the translational relevance of various animal models applicable to human liver disease was explored starting with a historic overview of the rodent cancer bioassay with emphasis on hepatocarcinogenesis from early work at the National Cancer Institute, refinement by the National Toxicology Program and contemporary efforts to identify potential mechanisms and their relevance to human cancer risk. Subsequently, recently elucidated understanding of the molecular drivers and signaling mechanisms of liver pathophysiology and liver cancer, including factors associated with liver regeneration, metabolic hepatocellular zonation, and the role of macrophages and their crosstalk with stellate cells in understanding human liver disease was discussed. Next, our contemporary understanding of the role of nuclear receptors in hepatic homeostasis and drug response highlighting nuclear receptor activation and crosstalk in modulating biological responses associated with liver damage and neoplastic response were discussed. Finally, an overview and translational relevance of different drug-induced liver injury (DILI) rodent model systems focused on pathology and mechanisms with commentary on current relevant Food and Drug Administration (FDA) perspective were summarized with closing remarks.

Rodent Liver Tumors: National Cancer Institute/National Toxicology Program Historic Perspective

R. Maronpot, Raleigh, North Carolina

Documentation of environmental factors causing cancer dating back to the 18th and 19th century include Bernardino Ramazzini’s identification of breast cancer in nuns in 1713, John Hill’s 1761 documentation of snuff causing oral and nasal cancer, Percival Pot’s 1775 documentation of scrotal cancer attributed to tar and soot in chimney sweeps, William Jackson Elmslie’s identification in 1866 of abdominal and thigh epithelioma in Kashmir men associated with use of heated clay pots held against the skin for warmth, and Ludwig Rehn’s 1895 identification of bladder cancer in aniline dye workers ³¹ (Some Classics of Experimental Oncology. NIH Publication No. 80-2150). The first experimental demonstration of specific exposures leading to cancer induction was demonstrated in a series of papers initially published^2,42 starting in 1914. Subsequent early publications documented induction of lung tumors in tar-painted mice, ²⁷ skin cancer in mice painted with polycyclic aromatic hydrocarbons, ⁵ and liver cancer in rats following dietary exposure to o-amidoazotoluene. ³⁰ Subsequent experiments demonstrated the concepts of co-carcinogenesis, initiation, and promotion ¹ and identification and nitrosamines exposure causing liver cancer in rats ²² among early investigations related to cancer induction.

Following the documentation of occupational, environmental and life-style factors along with early experimental evidence of laboratory induction of cancer, the concept of a cancer bioassay was born and established as a mandate of the National Cancer Institute (NCI) in the early 1960s. Cancer induction studies in 18 month duration mouse studies involving 127 chemicals lead to the selection of the B6C3F1 mouse hybrid as the animal of choice for subsequent cancer bioassays. ¹⁵ This was followed by the 1971 US National Cancer Act to establish the NCI cancer bioassay testing program with the 1975 addition of the F344 rat to that program. The NCI rodent cancer bioassay program was standardized to include 50 male and 50 female rats and an equal number of male and female mice each allocated to a control group, a high maximum tolerated dose (MTD) and an ½ MTD dose for a test duration of 18 months (B6C3F1 mice) or 2 years (F344 rats). This standardized bioassay was originally designed for screening with the option for subsequent more in-depth studies and involved standardized statistical evaluation and extensive pathology peer review. Following transfer of the bioassay testing program to the National Toxicology Program (NTP), efforts were initiated to identify and establish alternative short-term animal models as there was initial concern regarding the resource intensive bioassay program utilizing high doses of questionable relevance to human exposures.

Following completion of 490 mouse and rat bioassay, liver was identified as the most frequently occurring tumor response with a high background hepatocellular tumor response in B6C3F1. Tumor progression and stop studies carried out over the years from the 1980s to 2010 identified the morphological and molecular stages of hepatocellular tumor development in mice and rats to better assess their human relevance. Even with the 50 plus years of conducting these rodent cancer bioassays, there was a consistent concern about the significance and relevance of rodent cancer bioassays to human health risk. Since approximately 2010 to the present time, the NTP has gradually stopped doing conventional rodent cancer bioassays and initiated and focused resources on other investigative approaches including cell-based assays related to cancer induction, use of cell culture and organoid testing systems, use of high-throughput screening assays, and other mechanistic studies to identify potential human carcinogens.

Recapitulating Human HCC Subsets in Mice for Understanding Biology and Precision Therapy

Satdarshan (Paul) Singh Monga, University of Pittsburgh and UPMC, Pittsburgh

According to GLOPBOCAN 2020, liver cancer accounts for close to 5% of new cancer cases worldwide with liver cancer responsible for over 830 000 associated deaths. Liver cancer is the 13th most common tumor in the United States with over 30 000 estimated deaths in 2022. Hepatocellular carcinoma occurs in a setting of chronic liver insult that occurs over decades. Wnt-β-catenin signaling, known for its role in embryogenesis as well as carcinogenesis, plays a critical role in hepatic development and regeneration, and its dysregulation is evident in liver tumor growth. Wnt/β-catenin hepatic functional roles include regulation of cell proliferation, apoptosis, oxidative stress, and differentiation, all critical to hepatic metabolic processes and subject to temporal and tissue-specific modulation.

Activation of the Wnt-β-catenin pathway is evident in 35% to 40% of hepatocellular cancers (HCCs) in patients. This is mostly due to activating gain-of-function missense mutations in CTNNB1, the gene encoding for β-catenin (25%-35% of cases) or loss of function mutations in AXIN1/2, encoding Axin protein which is essential for β-catenin degradation. ²⁶ Activation of β-catenin has a major impact on the biology of HCC which has been determined by modeling this tumor subset in mice using innovative models. Using hydrodynamic tail vein injection (HDTVI) to deliver mutant-CTNNB1 in a sleeping beauty transposon/transposase-based expression system (SBTT), scientists have been able to recapitulate patient HCC subsets in mouse models. The credibility of these models has been shown by notable overlap of gene expression evident in mouse HCC arising in these models and in patient HCC subsets. Co-expression of mutant β-catenin and active Nrf2 or Met in hepatocytes yields HCC that represent 9% to 12% and 11% of all human HCC subsets, respectively. Using various such models, two major mechanisms of how β-catenin contributes to HCC ²⁵ have been demonstrated. Because β-catenin regulates expression of glutamine synthetase and in turn increases intratumoral glutamine levels, there is evidence of mTORC1 activation in mutant-CTNNB1 harboring HCCs. Indeed, β-catenin-mutated HCCs respond to mammalian target of rapamycin (mTOR) inhibitors preclinically and in proof of concept (POC) studies in patients. Secondly, β-catenin activation leads to immune cell exclusion in tumors, and hence, there is both preclinical and clinical evidence that these tumors do not respond to immune checkpoint inhibitors. To induce immune cell surveillance in β-catenin mutated HCCs, innovative bioinformatic approaches to identify underlying mechanisms have been employed in addition to using SBTT-HDTVI models to modulate these mechanisms. Finally, evidence shows that in the preclinical models of mutant CTNNB1, irrespective of additional mutations, these tumors are highly susceptible to β-catenin inhibition. Hence, anti-β-catenin therapies could have notable therapeutic implications and pave the way for individualized treatment in HCC.

DILI: An Overview and Model Systems to Study DILI

Frederic Moulin, US Food and Drug Administration, Silver Spring, MD

DILI (drug-induced liver injury) is a rare, multifaceted, yet potentially severe life-threatening adverse drug reaction of the liver to xenobiotics exposure. It captures multiple safety events associated with multiple mechanisms and is estimated to be responsible for ~25% of human safety failures in the clinic. Studies aimed at identifying major etiologies of acute liver failure leading to liver transplant or death in the United States reveals acetaminophen (APAP) overdose as the number one cause accounting for 50% of cases amounting to around 70 000 hospitalizations each year which has grown to be a significant public health concern.^3,12,19

The clinical manifestations of DILI are diverse, reflecting a heterogeneity of toxicity mechanisms, and can resemble common hepatic diseases. Clinical manifestations of DILI can be life-threatening and typically associated with one or more of the clinical findings: acute liver failure with fulminant hepatitis and loss of liver function occurring in individuals with or without pre-existing liver disease; chronic liver injury with hepatic dysfunction leading to end stage liver disease; hepatic decompensation in patients with underlying cirrhosis; vanishing bile duct syndrome – a rare complication of DILI characterized by sever bile duct destruction and loss in the interlobular and portal areas.^12,38 Clinical signs and symptoms are poorly correlated between animals and humans and as such animal models predict only ~50% of human DILI events. The ambiguous symptoms, inconsistent injury timelines and potentially severe outcome of DILI have made regulatory authorities cautious and sporadic elevations of plasma liver enzymes and bilirubin during clinical studies pose a significant challenge to the marketing approval of new therapeutic agents.^16,28,39

DILI is classified into Direct (Intrinsic), Indirect, and Idiosyncratic categories based on dose dependency, latency, rate of occurrence, clinico-pathologic phenotype, underlying mechanism, and non-clinical prediction of outcome. The “intrinsic” hepatotoxins exert their effects through mechanisms that largely depend on the chemical structure of the molecule and thus produce adverse reactions that are strongly related to the dose administered and in close temporal proximity to the time of administration (ie, acetaminophen, nicotinic acid, aspirin, etc). By contrast, the “idiosyncratic” hepatotoxins require the presence of genetic and/or physiologic predispositions in the patients to fully manifest their toxic potentials (isoniazid, statins, antibiotics, etc). As such combinations of host sensitivity and drug exposure are uncommon, idiosyncratic DILI is often described as sporadic and rare, without obvious relation to dose, and occurring at random moments during drug treatment. ²⁰ Indirect DILI, on the contrary, is intermediate with respect to toxicity and rate of occurrence and is dose independent with delayed latency and occasional non-clinical outcome ⁸ (corticosteroids, kinase inhibitors, monoclonal antibodies, etc). The severity and outcome of DILI coincide with the observation (Hy’s law) that drug-induced hepatocellular injury (parenchymal damage with high alanine transferase values 10-15 times normal) accompanied by jaundice (total bilirubin values greater than twice the normal) has a poor prognosis with 10% to 50% chances of liver transplantation and or mortality. ⁴⁶ A recent analysis of drug-induced potential of different drug classes identified four pharmacological subgroups (ie, antidepressants, antiepileptics, anti-inflammatory and alimentary tract metabolism drugs). Of these, antineoplastic agents, kinase inhibitors, interferons, angiotensin-converting enzyme (ACE) inhibitors, and monoamine oxidase inhibitors showed moderate to high risk for DILI. ³⁷

Based on injury of the target cell type, DILI can be grouped into two major types: hepatocellular and cholestatic. Liver injury from hepatocellular DILI entails a multitude of complex underlying molecular mechanisms that closely allies with mechanisms involved with the phases of drug metabolism in general. ²¹ A prime example is the CYP2E1-mediated biotransformation of APAP in the liver that leads to the generation of N-acetyl-p-benzoquinone imine (NAPQI), a highly reactive metabolite that ultimately causes cell death via oxidative stress and mitochondrial dysfunction, the hepatocellular mitochondria being a key target. ²³ Further evidence of mitochondrial dysfunction was obtained from a study in which rats and mice treated with APAP. Although rats were highly resistant to APAP toxicity, mice showed significant oxidative stress, mitochondrial protein adducts formation and hepatocyte necrosis. ²⁴ Liver injury from cholestatic DILI, on the contrary, can among other things result from inhibition of transporter proteins mainly MDR (Multidrug Resistance) and BSEP (Bile Salt Export Pump) leading to cholestatic hepatitis. This blockage can occur following intake of medications primarily antibiotics (ie, amoxicillin/clavulanate) in susceptible individuals. ³⁴ Cholestatic injury also emanates from bile-induced cytotoxicity from non-detergent hydrophobic properties of human bile acids leading to oxidative stress and apoptosis. In contrast, rodent bile acids do not reproduce biliary injury owing to its hydrophilic properties that in fact imparts cytoprotection. A double deficient (DKO) mice model, however, shows promise for investigating roles of hydrophobic bile acids in human diseases as it produces bile acids profiles similar to that of humans. ¹¹

Given the unpredictability surrounding accurate diagnosis and prognosis of DILI, the search for new models to predict and study mechanisms of DILI has been on the rise. Primary human hepatocytes (PHHs) are considered the gold standard in vitro model as they are capable of evaluating a variety of end points; however, because of their short viability, PHHs in sandwich configuration are popular as it lengthens hepatocyte viability and thus has been used extensively. ¹⁰ Other cell types and platforms include Hepa RG cells, Upcyte human hepatocytes, hepatocyte-like cells derived from pluripotent stem cells, 3D spheroids, human liver organoids, microfluidic liver-on-a-chip systems, and multi-organ-on-a-chip platforms. Each of these systems and platforms confer a host of advantages and limitations and has been reviewed by Donatao and colleagues. ⁷ Among the new and emerging biomarkers, keratin 18 (K18), caspase-cleaved keratin 18 (ccK18), macrophage colony-stimulating factor receptor-1 show promise. Others such as the high mobility group box1 and microRNA-122, although showed potential initially, were later on dismissed due to lack of scientific credibility. ¹² On the contrary, monocyte-derived hepatocytes generated from patients with DILI could be useful in identifying potential DILI causing agents from among a panel of co-prescribed drugs. ¹⁸ Until more robust and specific biomarkers are available in clinical practice, a systematic case specific approach using combination of liver biopsy and standardized causality assessment scales is recommended for early diagnosis and management of DILI. ¹²

Nuclear Receptor MOA-Induced Hepatocarcinogenesis: Human Relevance

Samuel M. Cohen, Department of Pathology and Microbiology, University of Nebraska Medical Center.

Among the liver tumors, hepatocellular carcinoma (HCC) is the most common form of primary liver cancer in humans and the fifth leading cause of cancer-related death worldwide. Occurrences of HCC are attributed to several risk factors including hepatic viral infections, alcohol consumption, and other chronic liver diseases. ³⁵ Earliest experiments of liver cancer bioassays in mice entailed the use of Azo dyes.^31,45 Eventually, more refined bioassays were conducted in rodents with the aim of addressing human risk assessment by following two basic assumptions: (1) carcinogenic effects observed at high doses in rodents will also occur at low doses in humans (dose extrapolations) and (2) chemicals causing cancer in rodents will cause cancer in humans (species extrapolation). Later, multiple studies challenged the validity of these assumptions as species-specific responses and dose extrapolations differed with chemicals and the estimation of human risk from rodent data was affected. Consequently, a human relevance framework was developed jointly by the International Program on Chemical Safety (IPCS), International Life Sciences Institute (ILSI), and Risk Science Institute (RSI) based on the identification of a mode of action (MOA). This approach allows for establishment of key cellular and molecular events which can then be used to bridge species and dose differences for a given MOA. Using this framework however, MOA for phenobarbital could not be determined in humans unlike that for rodents; conversely, MOA for metal overload for rodents could be established. Interestingly, the plausibility of estrogen induced tumors in humans was given consideration.^2,13,33

Among the many modes of actions that have been identified for hepatocellular carcinogenesis over the years, nuclear receptor mediated mechanisms have gained particular attention. Nuclear receptors are ligand-activated transcriptional regulators involved in hepatic physiology and pathophysiology. They control a variety of metabolic processes in the liver including lipid metabolism, drug metabolism and disposition, bile acid homeostasis, cell proliferation, inflammation, fibrosis, and tumor formation. Much study over the past several years has been devoted to defining the role of activation of nuclear receptors (including AhR [Aryl hydrocarbon Receptor], ER [Estrogen Receptor], CAR [Constitutive Androstane Receptor], PXR [Pregnane-X Receptor], and Peroxisome Proliferator Activator Receptor alpha [PPARα]) by liver carcinogens with a nongenotoxic MOA. The nuclear receptor MOA involves multiple steps including activation of the receptor, activation of downstream receptors or activation factors, increase in hepatic cell proliferation, selective clonal expansion of preneoplastic hepatocytes, and neoplasm formation. Each step is sequential and dose-dependent. Induction of human nuclear receptors has been particularly associated with development of HCC in humans. A recent study reported that eight Nuclear Receptor (NR) classes were effective in predicting the prognosis of HCC patients as an independent prognostic signature. Moreover, poor prognosis correlated with expression of key immune checkpoints indicating high risk associated with immunosuppressive microenvironments in liver cells. ³⁵ A number of nongenotoxic hepatic carcinogens (ie, phenobarbital) have been shown to activate specific nuclear receptors leading to downstream modification of multiple cellular pathways including hepatic growth. In the case of nongenotoxic rodent hepatocarcinogens that activate CAR and PPARα, their human relevance has been disputed based on the lack of fulfillment of the MOA stages coupled with lack of epidemiological evidence. These following paragraphs will focus on some aspects of mechanisms of action and human relevance of CAR and PPAR receptors.

CAR plays an important role in the regulation of xenobiotic response in hepatocytes including induction of hepatocyte proliferation and carcinogenesis in rodents. CAR activation is typically associated with induction of P450 isoenzymes (CYP2B), increased numbers of intracytoplasmic smooth endoplasmic reticulum and clinically observed increased liver weights. Phenobarbital is well known to promote hepatocellular proliferation through induction of CAR. However, that proliferative response is limited to rodents and CAR knockout mice fail to develop liver tumors with phenobarbital. Huang et al ¹⁴ also demonstrated the phenobarbital mediated activation of CAR and murine CAR ligand 1,4-bis-[2-(3,5,-dichloropyridyloxy)] benzene (TCPOBOP) for murine liver tumorigenesis.^9,36,41 Several studies in the recent years have indicated the involvement of key signaling pathways Wnt/B catenin and Hippo/YAP in CAR-mediated hepatocellular carcinogenesis. CAR induction likely activates both signaling pathways synergistically leading to CAR-mediated liver cancer development. In addition, low homology between the ligand binding domains of human and rodent CAR receptors may be responsible for species differences in the way proteins controlling the signaling pathways interact with CAR, thus leading to differences in liver cancer induction. ³²

Peroxisome Proliferator Activator Receptor alpha, in particular, is another important nuclear receptor which upon activation by a number of structurally diverse chemicals has been reported to cause hepatocellular tumors in rodents. ²⁹ Key events in the induction of liver tumors by PPARα agonists in rodents involve PPARα activation, altered cellular growth pathways, altered hepatocyte fate, increased cell proliferation, and clonal expansion leading to hepatocellular tumors. Associated events include intracytoplasmic peroxisome proliferation, centrilobular hypertrophy, increased liver weight, and oxidative damage. ¹⁷ The human relevance and MOA of liver tumors in rodents has been debated for some time and recently been disqualified. Crucial to this argument was understanding that key downstream events (ie, oxidative stress through nuclear factor [NF]-kappa B activation) was critical to the tumor responses in the livers of rats and mice. This was further supported by studies using a PPARα-null mouse model and a transgenic mouse model with viral transactivation domain from VP16 fused to PPARα.^6,43,44 In Addition, studies using PPARα humanized mice did not result in tumors, thus clearly establishing species-specific differences between human and rodent receptors.^4,44,45 Contrary to above findings, it has been recently reported that increased expression of PPARα correlated with longer survival times in patients with HCC and served as an independent factor for prognostically favorable outcome. ⁴⁰ In conclusion, CAR activators (ie, phenobarbital, permethrin) and PPARα activators (ie, fibrates) show no evidence of increased risk of cancer in humans.