Karissa Cottier, PhD Research Scientist, BioIVT, ADME-Tox, examines the ongoing development and advances to ensure drug efficacy and safety
Researchers evaluate the efficacy and safety profiles of drug candidates, including their potential to cause liver damage, when considering whether to advance them in development programs. Drug-induced liver injury (DILI) is a significant contributor to safety-related clinical and post-marketing drug failure and continues to be a barrier in drug development. Despite continued in vitro and in vivo preclinical model advancements, the susceptibility of a drug to cause DILI can remain undiscovered until later stages in clinical development, an outcome that is both disappointing and costly. Clinical trials aim to enroll enough participants so that safety concerns, including DILI, can be identified; however, late-stage failures most often occur due to unpredictable, idiosyncratic responses occurring in a small and/or specific population of patients. Lead selection programs need to screen drug candidates for their propensity to cause DILI and are reliant on effective in vitro and in vivo models. Established pre-clinical models of DILI are useful in uncovering direct, or “intrinsic”, hepatotoxic responses, defined as damage that is predictable and dose dependent. Most of these tests include short-term drug treatment in primary hepatocyte monocultures. Unfortunately, these models cannot reliably predict idiosyncratic DILI.1 Recent in vitro innovations have sought to bridge this gap by addressing genetic or mechanism-specific causes of DILI. Likewise, the creation of long-term hepatocyte cultures and complex models that combine primary human hepatocytes (PHH) and nonparenchymal cells (NPC) allow researchers to perform in vitro examinations of longer-term consequences of drug treatment.
One key challenge when dealing with idiosyncratic DILI is its heterogeneity in clinical presentation and underlying mechanisms.Idiosyncratic DILI can mimic the phenotype of any liver disease, so DILI should only be diagnosed after excluding other causes.2 Non-DILI liver disorders are quite common, further complicating DILI diagnosis. For example, approximately 25% of the population is estimated to be affected by non-alcoholic fatty liver disease (NAFLD) and between 4.5% and 9.5% of all livers are found to be cirrhotic at autopsy.3,4 Many of these liver disorders show similar changes in blood markers to DILI, including elevated alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), alkaline phosphatase (ALP), and gamma glutamyl transferase (GGT).5 Therefore, it can be difficult to distinguish between damage occurring due to chronic liver disease and damage arising from drug-induced hepatotoxicity. During clinical workup for DILI, these biomarkers can indicate the type of liver injury presenting in the patient. DILI can be broadly classified based on the following categories of liver damage: hepatocellular, cholestatic, or mixed.6 While elevated serum levels of ALT and AST primarily detect hepatocellular injury, TBIL, ALP, and GGT typically indicate cholestatic injuries. However, elevated ALT and AST do not necessarily indicate liver damage as they are also found in response to injury to other tissue types.5 Development of biomarkers that are more specific and amenable to early detection is essential to both reduce serious DILI cases and determine the mechanisms leading to DILI.
DILI biomarkers in vitro
Although clinical investigations have uncovered several promising DILI biomarkers, many are not predictive in in vitro DILI studies. However, there have been focused efforts to develop suitable markers for cell culture experiments. For example, using a micropatterned primary hepatocyte co-culture system (MPCC), researchers investigated NRF1/NRF2 transcription mediated DILI biomarkers, which were first described in in vivo studies. In rodent studies, these biomarkers are highly predictive of the capability of a drug to generate large quantities of reactive metabolites. Previous attempts to link in vivo biomarkers to in vitro studies produced mixed results. One problem may be that conventional PHH cultures rapidly de-differentiate. Conversely, MPCC models retain in vivo-like function for weeks, allowing researchers to examine the transcriptional responses to reactive metabolites in functional hepatocytes.7 Hepatic spheroids have also demonstrated functionality of in vitro biomarkers in detecting DILI. Concentration and time dependent changes in the experimental liver injury biomarkers α-GST, HMGB1, and miR-122 were seen in hepatic spheroids during a broad screen of known DILI positive, but not negative, compounds.8 Determining how to reliably use these biomarkers to screen new drugs is an important next step in more accurately identifying drugs with the potential to cause DILI.
In vitro models that address mechanisms contributing to idiosyncratic DILI
Cell death and innate immune system activation
Generally, liver injuries produced by drugs or their metabolites will lead to cell death, via necrosis or apoptosis. Numerous mechanisms are involved, causing either direct cellular damage or reduced ability to overcome damaging stimuli (ie oxidative stress). In some cases, drug-induced cell death can be detected in conventional PHH monocultures, particularly if it occurs due to an acute reaction to the parent compound. In other cases, detecting DILI may require long-term functional culture. Many of the long-term models have exhibited superiority in correctly identifying DILI positive compounds in vitro.9 This improvement in testing accuracy is probably due to the extended hepatocyte-specific function in these cultures. It is known, however, that often a drug will only induce DILI in the presence of multiple contributing factors, which can be tested using more specialised models.
Inflammation is a key response to tissue damage, including damage occurring due to liver disease or DILI. Usually, inflammation is instrumental in cell survival and repair in response to an acute injury. However, as inflammation becomes excessive or unresolved it becomes self-destructive rather than helpful. Under certain circumstances, a drug may cause an initial injury which then leads to inflammation and cell death. Other drugs have shown a much stronger hepatotoxic effect when innate immune cells are activated.10 For example, trovafloxacin, a broad-spectrum antibiotic, was withdrawn from the market due to its potential to cause severe hepatoxicity in a small subset of patients. Post-withdrawal in vitro studies were able to show notable hepatotoxicity only when Kupffer cells, the liver’s resident macrophages, were activated by a strong inflammatory response.11 In this case, a drug that was normally nontoxic became highly damaging due to immune cell activation. Unfortunately, this effect was only discovered after its market release, when trovafloxacin was linked to multiple cases of liver failure in patients. This underscores the necessity of adopting new in vitro models to detect idiosyncratic DILI before a drug reaches the clinic.
The immune-mediated trovafloxacin hepatotoxicity finding has been recapitulated in MPCC and hepatic spheroids.12,13 For these studies, the long-term culture models have been supplemented with Kupffer cells, which are activated using bacterial lipopolysaccharide (LPS). Newer studies have investigated whether additional compounds that cause DILI might do so via inflammatory signaling. Studies using hepatic spheroids to screen drugs suggest a dual role for immune cell activation. While some compounds became more toxic with inflammation, others seemed to show less toxicity, suggesting a protective effect.8 Other investigations using MPCC have uncovered compounds that show greater toxicity under inflammatory conditions, including clozapine and chlorpromazine. Additionally, these studies suggested that immune-mediated toxicity was time dependent. With these two compounds, the difference in toxicity due to inflammation was not seen until six days after dosing started.14 This example further highlights the importance of using long-term in vitro DILI models in assessing drug candidates.
Genetic factors
Idiosyncratic DILI is usually a rare occurrence, dependent on the specific characteristics of individuals. Amongst other factors, their genomes may contribute to susceptibility to DILI for specific medicines. Polymorphisms in genes involved in drug metabolism such as certain cytochrome P450s (CYPs) or drug transporters may predispose someone to DILI. Other variants, including those that affect immune function or the antioxidant capacity of cells could also increase the risk of DILI by blunting the ability of cells to respond to a mildly injurious event.15 Identifying genes that influence patient vulnerability to DILI can be difficult because these variants are often rare and may only lead to DILI when other conditions, such as existing inflammation or liver disease, are present. The rarity of these variants may make examining their DILI effect in vitro particularly challenging since these cells may not always be available in the quantity necessary for studies. Therefore, if primary samples are not available, alternative strategies may be needed.
If PHH with specific genotypes are not available, it may be easier to perform genetic manipulations in immortalized cell lines or induced pluripotent stem cell (iPSC) derived hepatocytes. Although abundant, human hepatocyte cell lines do not have adequate in vivo like characteristics or a diversity of genetic backgrounds. One novel solution is the use of expanded primary cells. To create expanded primary hepatocytes, proliferation is induced in PHH to generate a larger quantity of cells. Replication is stopped prior to experimentation so that the cells can retain a primary-like phenotype. The resulting cells have intermediate DME activity, are able to identify known DILI-inducing compounds, and retain the original donor’s genetic background, making them especially useful for studies involving rare genotypes.16
Another strategy for examining the role of a particular DME or transporter in DILI is to genetically modify PHH. SiRNA knockdown of CYP450s has been performed in human MPCC as a long-term model for investigating the contribution of specific enzymes to drug toxicity. Knockdown of a particular CYP allows researchers to screen drugs for their potential toxicity in high- and low-metabolising conditions while keeping the rest of the hepatocyte gene expression and function the same. These tools can help alleviate the burden of having to find and screen many PHH donors with different genetic signatures to determine individual responses to a drug.17
Metabolic dysfunction
Drugs can cause metabolic dysfunction by acutely or chronically compromising mitochondrial function. In the liver, this leads to fat accumulation in hepatocytes as the cell becomes less capable of metabolising fatty acids. Lipid accumulation itself is an undesirable outcome; however, in many patients with existing fatty liver disease the additional strain on fat metabolism can lead to worsening of their existing disease. This can cause a cascade of effects associated with NAFLD progression including immune cell infiltration and activation which can help promote liver fibrosis. NAFLD has also been shown in some clinical studies to alter the activity or expression of DME. Together, these factors suggest the possibility for increased DILI sensitivity in patients with fatty liver disorders.18 Conventional in vitro models containing PHH can demonstrate the potential of a drug to cause acute lipid accumulation. More complex, long-term models that incorporate co-cultures of immune cells and hepatic stellate cells (HSC) are needed to identify the risk of a drug initiating inflammation or fibrosis because of metabolic dysfunction. Currently, many in vitro NAFLD/NASH models are in development in long-term culture models including MPCC, hepatic spheroids, and organ on a chip.19–21 The focus for most of these models has been on establishing in vivo like disease characteristics, including fat loading, metabolic dysfunction, inflammation, and fibrosis induction. Once functional for disease modeling and therapeutic screening, they may also provide utility for hepatoxicity screening within a disease background.
Bile duct injury
Obstruction of bile flow (cholestasis) due to bile duct injury is another serious drug effect that can lead to DILI. Drugs that are removed from the liver via bile excretion have the potential to produce cholestatic liver injury in certain individuals. Cholestatic DILI can be difficult to diagnose due to potentially vague symptoms that can develop gradually. Often the damage and resulting symptoms will persist for some time after concluding the treatment, making it difficult to pinpoint the drug as the causative agent. Several factors determine whether a drug will induce cholestatic injury. For instance, reduced activity of DME or transporters due to inflammation or genetic variants can lead to build-up of a reactive drug or metabolite. Some drugs are removed from hepatocytes by excretion into bile. This excretion is dependent on several transporters, especially the bile salt export pump (BSEP) and multidrug resistance proteins (MRPs). Therefore, a combination of excess drug and/or metabolite with reduced transporter activity can prevent these transporters regulating bile salt transport into bile and / or bile flow, two of their primary functions.22 Simple in vitro PHH monocultures lack the complexity to examine the intricate interactions between therapeutic compounds, bile salts, and transporters in a susceptible environment. Specific models designed to examine cholestatic DILI are necessary to investigate whether a drug may induce cholestatic injury.
Human clinical studies have found a strong correlation between BSEP inhibition and cholestatic DILI. In vitro inhibition of BSEP alone, however, has not provided a strong link to in vivo cholestatic DILI severity. The latest research findings suggest that the failure to incorporate farnesyl X receptor (FXR) mediated compensatory mechanisms into the model may be responsible for the poor in vitro in vivo correlation. These studies describe a culture system that combines BSEP and FXR inhibition to provide a more accurate prediction of cholestatic DILI.23 This model has been refined to include sensitisation with bile acids and lipids. When all of these parameters were included in in vitro models, researchers were able to recapitulate rodent cholestatic DILI findings for four preclinical drug candidates, a result that is very promising for the future of in vitro identification of cholestatic DILI.24
DILI remains a serious issue in drug development. Established in vitro models can identify drugs that will produce direct DILI, however idiosyncratic DILI is still difficult to predict using pre-clinical models. Numerous mechanisms are proposed to contribute to idiosyncratic DILI, and new in vitro models have been developed to address these specific contributors. Adoption of some of these new screening systems will help to identify drugs with DILI potential before they reach clinical trials. Future studies should continue to look for mechanisms that contribute to idiosyncratic DILI in the pursuit of better, more predictive in vitro models.
Volume 22, Issue 1 – Winter 2020/21
About the author
Dr. Karissa Cottier received her PhD in Medical Pharmacology from the University of Arizona in 2018, where she focused on drug delivery to the brain during episodic headache disorders. Dr. Cottier is currently a Product and Applications Development Scientist in the ADME-Tox division at BioIVT.
References
1 J. G. Kenna and J. Uetrecht, Drug Metab. Dispos., 2018, 46, 1658–1669.
2 R. J. Fontana, Gastroenterology, 2013, 146, 914–28.
3 K. Wijarnpreecha, M. Scribani, C. Thongprayoon, P. Ungprasert and W. Cheungpasitporn, Gastroenterology, 2017, 152.
4 Y. S. Lim and W. R. Kim, Clinics in Liver Disease, 2008, 12, 733–46, vii.
5 S. E. Roth, M. I. Avigan, D. Bourdet, D. Brott, R. Church, A. Dash, D. Keller, P. Sherratt, P. B. Watkins, L. W. Baker, S. Lentini, M. Merz, L. Ramaiah, S. K. Ramaiah, A. M. Stanley and J. Marcinak, Clinical Pharmacology & Therapeutics, 2019, 107, 333–346.
6 S. H. Hussaini and E. A. Farrington, Expert Opinion on Drug Safety, 2013, 13, 67–81.
7 W. Kang, A. A. Podtelezhnikov, K. Q. Tanis, S. Pacchione, M. Su, K. B. Bleicher, Z. Wang, G. M. Laws, T. G. Griffiths, M. C. Kuhls, Q. Chen, I. Knemeyer, D. J. Marsh, K. Mitra, J. Lebron and F. D. Sistare, Toxicol. Sci., 2020, 177, 121–139.
8 W. R. Proctor, A. J. Foster, J. Vogt, C. Summers, B. Middleton, M. A. Pilling, D. Shienson, M. Kijanska, S. Strobel, J. M. Kelm, P. Morgan, S. Messner and D. Williams, Archives of Toxicology, 2017, 91, 2849–2863.
9 S. R. Khetani, C. Kanchagar, O. Ukairo, S. Krzyzewski, A. Moore, J. Shi, S. Aoyama, M. Aleo and Y. Will, Toxicol. Sci., 2012, 132, 107–17.
10 W. BL and et al., Curr Pharmacol Rep., 2018, 4,5, 346-357.
11 P. J. Shaw, A. C. Ditewig, J. F. Waring, M. J. Liguori, E. A. Blomme, P. E. Ganey and R. A. Roth, Toxicol. Sci.,2009, 107, 270-280.
12 M. McVay, C. Kanchagar, S. R. Khetani and O. Ukairo, Toxicology Letters, 2013, 221.
13 S. Messner, I. Agarkova, W. Moritz and J. M. Kelm, Arch Toxicol., 2013, 87, 209-13.
14 O. Irrechukwu, M. McVay and O. Ukairo, Poster Presentation, Society of Toxicology (SOT) Conference, 2015.
15 T. J. Urban, D. B. Goldstein and P. B. Watkins, Pharmacogenomics, 2012, 13, 735–8.
16 L. Tolosa, N. Jimenez, M. Pelecha, J. V. Castell, M. J. G. Lechon and M. T. Donato, Archives of Toxicology, 2018, 93, 519–532.
17 L. M. Silva, H. E. Fleming, A. B. Miller, S. Milstein, A. Liebow, P. Haslett, L. S. Lorenzino and S. N. Bhatia, Cell Metabolism, 2019, 29, 727-735.e3.
18 J. Massart, K. Begriche, C. Moreau and B. Fromenty, Journal of Clinical and Translational Research, 2017, 3, 212–232.
19 M. Gori, M. C. Simonelli, S. M. Giannitelli, L. Businaro, M. Trombetta and A. Rainer, PLoS ONE, 2016, 11.
20 M. D. Davidson, D. A. Kukla and S. R. Khetani, Integrative Biology, 2017, 9, 662–677.
21 M. Kozyra, I. Johansson, A. Nordling, S. Ullah, V. M. Lauschke and M. I. Sundberg, Sci. Reports, 2018, 8.
22 M. S. Padda, M. Sanchez, A. J. Akhtar and J. L. Boyer, Hepatology, 2011, 53, 1377-87.
23 J. P. Jackson, K. M. Freeman, M. K. Palmer, R. L. St. Clarie III, C. B. Black and K. R. Brouwer, Poster Presentation, AAPS Joint Workshop on Drug Transporters in ADME, 2018.
24 S. Lamore, C. Rowbottom, S. Kapadnis, A. Easter, E. Tien, J. Fikes, K. Brouwer and J. Jackson, Poster Presentation, 39th Annual Meeting of the American College of Toxicology (ACT), 2018.
