Quantitative Systems Toxicology Approaches to Understand and Predict Drug-Induced Liver Injury

Quantitative Systems Toxicology Modeling Informed Safe Dose Selection of Emvododstat in Acute Myeloid Leukemia Patients

Clinical investigation of emvododstat for the treatment of solid tumors was halted after two patients who were heavily treated with other anticancer therapies experienced drug-induced liver failure. However, preclinical investigations supported that emvododstat at lower doses might be effective in treating acute myeloid leukemia (AML) and against severe acute respiratory syndrome-coronavirus 2 as a dihydroorotate dehydrogenase inhibitor. Therefore, a quantitative systems toxicology model, DILIsym, was used to predict liver safety of the proposed dosing of emvododstat in AML clinical trials. In vitro mechanistic toxicity data of emvododstat and its desmethyl metabolite were integrated with in vivo exposure within DILIsym to predict hepatotoxicity responses in a simulated human population. DILIsym simulations predicted alanine aminotransferase elevations observed in prior emvododstat clinical trials in patients with solid tumors, but not in the prospective AML clinical trial with the proposed dosing regimens. Exposure predictions based on physiologically-based pharmacokinetic modeling suggested that reduced doses of emvododstat would produce clinical exposures that would be efficacious to treat AML. In the AML clinical trial, only eight patients experienced aminotransferase elevations, all of which were mild (grade 1), all resolving within a short period of time, and no patient showed symptoms of hepatotoxicity, confirming the prospective prediction of liver safety. Overall, retrospective DILIsym simulations adequately predicted the liver safety liabilities of emvododstat in solid tumor trials and prospective simulations predicted the liver safety of reduced doses in an AML clinical trial. The modeling was critical to enabling regulatory approval to proceed with the AML clinical trial wherein the predicted liver safety was confirmed.

An In Silico Platform to Predict Cardiotoxicity Risk of Anti-tumor Drug Combination with hiPSC-CMs Based In Vitro Study

OBJECTIVE

Antineoplastic agent-induced systolic dysfunction is a major reason for interruption of anticancer treatment. Although targeted anticancer agents infrequently cause systolic dysfunction, their combinations with chemotherapies remarkably increase the incidence. Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) provide a potent in vitro model to assess cardiovascular safety. However, quantitatively predicting the reduction of ejection fraction based on hiPSC-CMs is challenging due to the absence of the body's regulatory response to cardiomyocyte injury.

METHODS

Here, we developed and validated an in vitro-in vivo translational platform to assess the reduction of ejection fraction induced by antineoplastic drugs based on hiPSC-CMs. The translational platform integrates drug exposure, drug-cardiomyocyte interaction, and systemic response. The drug-cardiomyocyte interaction was implemented as a mechanism-based toxicodynamic (TD) model, which was then integrated into a quantitative system pharmacology-physiological-based pharmacokinetics (QSP-PBPK) model to form a complete translational platform. The platform was validated by comparing the model-predicted and clinically observed incidence of doxorubicin and trastuzumab-induced systolic dysfunction.

RESULTS

A total of 33,418 virtual patients were incorporated to receive doxorubicin and trastuzumab alone or in combination. For doxorubicin, the QSP-PBPK-TD model successfully captured the overall trend of systolic dysfunction incidences against the cumulative doses. For trastuzumab, the predicted incidence interval was 0.31-2.7% for single-agent treatment and 0.15-10% for trastuzumab-doxorubicin sequential treatment, covering the observations in clinical reports (0.50-1.0% and 1.5-8.3%, respectively).

CONCLUSIONS

In conclusion, the in vitro-in vivo translational platform is capable of predicting systolic dysfunction incidence almost merely depend on hiPSC-CMs, which could facilitate optimizing the treatment protocol of antineoplastic agents.

Quantitative Systems Toxicology Identifies Independent Mechanisms for Hepatotoxicity and Bilirubin Elevations Due to AKR1C3 Inhibitor BAY1128688

BAY1128688 is a selective inhibitor of AKR1C3, investigated recently in a trial that was prematurely terminated due to drug-induced liver injury. These unexpected observations prompted use of the quantitative systems toxicology model, DILIsym, to determine possible mechanisms of hepatotoxicity. Using mechanistic in vitro toxicity data as well as clinical exposure data, DILIsym predicted the potential for BAY1128688 to cause liver toxicity (elevations in serum alanine aminotransferase (ALT)) and elevations in serum bilirubin. Initial simulations overpredicted hepatotoxicity and bilirubin elevations, so the BAY1128688 representation within DILIsym underwent optimization. The liver partition coefficient Kp was altered to align simulated bilirubin elevations with those observed clinically. Altering the mode of bile acid canalicular and basolateral efflux inhibition was necessary to accurately predict ALT elevations. Optimization results support that bilirubin elevations observed early during treatment are due to altered bilirubin metabolism and transporter inhibition, which is independent of liver injury. The modeling further supports that on-treatment ALT elevations result from inhibition of bile acid transporters, particularly the bile salt excretory pump, leading to accumulation of toxic bile acids. The predicted dose-dependent intrinsic hepatotoxicity may increase patient susceptibility to an adaptive immune response, accounting for ALT elevations observed after completion of treatment. These BAY1128688 simulations provide insight into the mechanisms behind hepatotoxicity and bilirubin elevations and may inform the potential risk posed by future compounds.

The Combination of a Human Biomimetic Liver Microphysiology System with BIOLOGXsym, a Quantitative Systems Toxicology (QST) Modeling Platform for Macromolecules, Provides Mechanistic Understanding of Tocilizumab- and GGF2-Induced Liver Injury

Biologics address a range of unmet clinical needs, but the occurrence of biologics-induced liver injury remains a major challenge. Development of cimaglermin alfa (GGF2) was terminated due to transient elevations in serum aminotransferases and total bilirubin. Tocilizumab has been reported to induce transient aminotransferase elevations, requiring frequent monitoring. To evaluate the clinical risk of biologics-induced liver injury, a novel quantitative systems toxicology modeling platform, BIOLOGXsym™, representing relevant liver biochemistry and the mechanistic effects of biologics on liver pathophysiology, was developed in conjunction with clinically relevant data from a human biomimetic liver microphysiology system. Phenotypic and mechanistic toxicity data and metabolomics analysis from the Liver Acinus Microphysiology System showed that tocilizumab and GGF2 increased high mobility group box 1, indicating hepatic injury and stress. Tocilizumab exposure was associated with increased oxidative stress and extracellular/tissue remodeling, and GGF2 decreased bile acid secretion. BIOLOGXsym simulations, leveraging the in vivo exposure predicted by physiologically-based pharmacokinetic modeling and mechanistic toxicity data from the Liver Acinus Microphysiology System, reproduced the clinically observed liver signals of tocilizumab and GGF2, demonstrating that mechanistic toxicity data from microphysiology systems can be successfully integrated into a quantitative systems toxicology model to identify liabilities of biologics-induced liver injury and provide mechanistic insights into observed liver safety signals.

Investigating bile acid-mediated cholestatic drug-induced liver injury using a mechanistic model of multidrug resistance protein 3 (MDR3) inhibition

Inhibition of the canalicular phospholipid floppase multidrug resistance protein 3 (MDR3) has been implicated in cholestatic drug-induced liver injury (DILI), which is clinically characterized by disrupted bile flow and damage to the biliary epithelium. Reduction in phospholipid excretion, as a consequence of MDR3 inhibition, decreases the formation of mixed micelles consisting of bile acids and phospholipids in the bile duct, resulting in a surplus of free bile acids that can damage the bile duct epithelial cells, i.e., cholangiocytes. Cholangiocytes may compensate for biliary increases in bile acid monomers via the cholehepatic shunt pathway or bicarbonate secretion, thereby influencing viability or progression to toxicity. To address the unmet need to predict drug-induced bile duct injury in humans, DILIsym, a quantitative systems toxicology model of DILI, was extended by representing key features of the bile duct, cholangiocyte functionality, bile acid and phospholipid disposition, and cholestatic hepatotoxicity. A virtual, healthy representative subject and population (n = 285) were calibrated and validated utilizing a variety of clinical data. Sensitivity analyses were performed for 1) the cholehepatic shunt pathway, 2) biliary bicarbonate concentrations and 3) modes of MDR3 inhibition. Simulations showed that an increase in shunting may decrease the biliary bile acid burden, but raise the hepatocellular concentrations of bile acids. Elevating the biliary concentration of bicarbonate may decrease bile acid shunting, but increase bile flow rate. In contrast to competitive inhibition, simulations demonstrated that non-competitive and mixed inhibition of MDR3 had a profound impact on phospholipid efflux, elevations in the biliary bile acid-to-phospholipid ratio, cholangiocyte toxicity, and adaptation pathways. The model with its extended bile acid homeostasis representation was furthermore able to predict DILI liability for compounds with previously studied interactions with bile acid transport. The cholestatic liver injury submodel in DILIsym accounts for several processes pertinent to bile duct viability and toxicity and hence, is useful for predictions of MDR3 inhibition-mediated cholestatic DILI in humans.

Applications of In Silico Models to Predict Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is a major cause of the withdrawal of pre-marketed drugs, typically attributed to oxidative stress, mitochondrial damage, disrupted bile acid homeostasis, and innate immune-related inflammation. DILI can be divided into intrinsic and idiosyncratic DILI with cholestatic liver injury as an important manifestation. The diagnosis of DILI remains a challenge today and relies on clinical judgment and knowledge of the insulting agent. Early prediction of hepatotoxicity is an important but still unfulfilled component of drug development. In response, in silico modeling has shown good potential to fill the missing puzzle. Computer algorithms, with machine learning and artificial intelligence as a representative, can be established to initiate a reaction on the given condition to predict DILI. DILIsym is a mechanistic approach that integrates physiologically based pharmacokinetic modeling with the mechanisms of hepatoxicity and has gained increasing popularity for DILI prediction. This article reviews existing in silico approaches utilized to predict DILI risks in clinical medication and provides an overview of the underlying principles and related practical applications.

Challenges and Future of Drug-Induced Liver Injury Research-Laboratory Tests

Drug-induced liver injury (DILI) is a rare but potentially severe adverse drug event, which is also a major cause of study cessation and market withdrawal during drug development. Since no acknowledged diagnostic tests are available, DILI diagnosis poses a major challenge both in clinical practice as well as in pharmacovigilance. Differentiation from other liver diseases and the identification of the causative agent in the case of polymedication are the main issues that clinicians and drug developers face in this regard. Thus, efforts have been made to establish diagnostic testing methods and biomarkers in order to safely diagnose DILI and ensure a distinguishment from alternative liver pathologies. This review provides an overview of the diagnostic methods used in differential diagnosis, especially with regards to autoimmune hepatitis (AIH) and drug-induced autoimmune hepatitis (DI-AIH), in vitro causality methods using individual blood samples, biomarkers for diagnosis and severity prediction, as well as experimental predictive models utilized in pre-clinical settings during drug development regimes.

Quantitative Systems Toxicology and Drug Development: The DILIsym Experience

DILIsym^® is a Quantitative Systems Toxicology (QST) model that has been developed over the last decade by a public-private partnership to predict the liver safety liability in new drug candidates. DILIsym integrates the quantitative abilities of parent and relevant metabolites to cause oxidative stress, mitochondrial dysfunction, and alter bile acid homeostasis. Like the prediction of drug-drug interactions, the data entered into DILIsym are assessed in the laboratory in human experimental systems, and combined with estimates of liver exposure to predict the outcome. DILIsym is now frequently used in decision-making within the pharmaceutical industry and its modeling results are increasingly included in regulatory communications and NDA submissions. DILIsym can be used to identify dominant mechanisms underlying liver toxicity and this information is increasingly being used to identify patient-specific risk factors, including certain disease states. DILIsym is also increasingly used to optimize the interpretation of liver injury biomarkers. DILIsym provides an example of how QST modeling can help speed the delivery of safer new drugs to the patients who need them.

Navigating Between Right, Wrong, and Relevant: The Use of Mathematical Modeling in Preclinical Decision Making

The goal of this mini-review is to summarize the collective experience of the authors for how modeling and simulation approaches have been used to inform various decision points from discovery to First-In-Human clinical trials. The article is divided into a high-level overview of the types of problems that are being aided by modeling and simulation approaches, followed by detailed case studies around drug design (Nektar Therapeutics, Genentech), feasibility analysis (Novartis Pharmaceuticals), improvement of preclinical drug design (Pfizer), and preclinical to clinical extrapolation (Merck, Takeda, and Amgen).

Joint Decision-Making Model Based on Consensus Modeling Technology for the Prediction of Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is the major cause of clinical trial failure and postmarketing withdrawals of approved drugs. It is very expensive and time-consuming to evaluate hepatotoxicity using animal or cell-based experiments in the early stage of drug development. In this study, an in silico model based on the joint decision-making strategy was developed for DILI assessment using a relatively large dataset of 2608 compounds. Five consensus models were developed with PaDEL descriptors and PubChem, Substructure, Estate, and Klekota–Roth fingerprints, respectively. Submodels for each consensus model were obtained through joint optimization. The parameters and features of each submodel were optimized jointly based on the hybrid quantum particle swarm optimization (HQPSO) algorithm. The application domain (AD) based on the frequency-weighted and distance (FWD)-based method and Tanimoto similarity index showed the wide AD of the qualified consensus models. A joint decision-making model was integrated by the qualified consensus models, and the overwhelming majority principle was used to improve the performance of consensus models. The application scope narrowing caused by the overwhelming majority principle was successfully solved by joint decision-making. The proposed model successfully predicted 99.2% of the compounds in the test set, with an accuracy of 80.0%, a sensitivity of 83.9, and a specificity of 73.3%. For an external validation set containing 390 compounds collected from DILIrank, 98.2% of the compounds were successfully predicted with an accuracy of 79.9%, a sensitivity of 97.1%, and a specificity of 66.0%. Furthermore, 25 privileged substructures responsible for DILI were identified from Substructure, PubChem, and Klekota–Roth fingerprints. These privileged substructures can be regarded as structural alerts in hepatotoxicity evaluation. Compared with the main published studies, our method exhibits certain advantage in data size, transparency, and standardization of the modeling process and accuracy and credibility of prediction results. It is a promising tool for virtual screening in the early stage of drug development.

Modeling approaches for reducing safety-related attrition in drug discovery and development: a review on myelotoxicity, immunotoxicity, cardiovascular toxicity, and liver toxicity

Introduction:Safety and tolerability is a critical area where improvements are needed to decrease the attrition rates during development of new drug candidates. Modeling approaches, when smartly implemented, can contribute to this aim.Areas covered:The focus of this review was on modeling approaches applied to four kinds of drug-induced toxicities: hematological, immunological, cardiovascular (CV) and liver toxicity. Papers, mainly published in the last 10 years, reporting models in three main methodological categories - computational models (e.g., quantitative structure-property relationships, machine learning approaches, neural networks, etc.), pharmacokinetic-pharmacodynamic (PK-PD) models, and quantitative system pharmacology (QSP) models - have been considered.Expert opinion:The picture observed in the four examined toxicity areas appears heterogeneous. Computational models are typically used in all areas as screening tools in the early stages of development for hematological, cardiovascular and liver toxicity, with accuracies in the range of 70-90%. A limited number of computational models, based on the analysis of drug protein sequence, was instead proposed for immunotoxicity. In the later stages of development, toxicities are quantitatively predicted with reasonably good accuracy using either semi-mechanistic PK-PD models (hematological and cardiovascular toxicity), or fully exploited QSP models (immuno-toxicity and liver toxicity).

Application of the DILIsym® Quantitative Systems Toxicology drug-induced liver injury model to evaluate the carcinogenic hazard potential of acetaminophen

In 2019, the California Office of Environmental Health Hazard Assessment (OEHHA) initiated a review of the carcinogenic hazard potential of acetaminophen. The objective of the analysis herein was to inform this review by assessing whether variability in patient baseline characteristics (e.g. baseline glutathione (GSH) levels, pharmacokinetics, and capacity of hepatic antioxidants) leads to potential differences in carcinogenic hazard potential at different dosing schemes: maximum labeled doses of 4 g/day, repeated doses above the maximum labeled dose (>4-12 g/day), and acute overdoses of acetaminophen (>15 g). This was achieved by performing simulations of acetaminophen exposure in thousands of diverse virtual patients scenarios using the DILIsym® Quantitative Systems Toxicology (QST) model. Simulations included assessments of the dose and exposure response for toxicity and mode of cell death based on evaluations of the kinetics of changes of: GSH, N-acetyl-p-benzoquinone-imine (NAPQI), protein adducts, mitochondrial dysfunction, and hepatic cell death. Results support that, at therapeutic doses, cellular GSH binds to NAPQI providing sufficient buffering capacity to limit protein adduct formation and subsequent oxidative stress. Simulations evaluating repeated high-level supratherapeutic exposures or acute overdoses indicate that cell death precedes DNA damage that could result in carcinogenicity and thus acetaminophen does not present a carcinogenicity hazard to humans at any dose.

Comparing Machine Learning Algorithms for Predicting Drug-Induced Liver Injury (DILI)

Drug-induced liver injury (DILI) is one the most unpredictable adverse reactions to xenobiotics in humans and the leading cause of postmarketing withdrawals of approved drugs. To date, these drugs have been collated by the FDA to form the DILIRank database, which classifies DILI severity and potential. These classifications have been used by various research groups in generating computational predictions for this type of liver injury. Recently, groups from Pfizer and AstraZeneca have collated DILI in vitro data and physicochemical properties for compounds that can be used along with data from the FDA to build machine learning models for DILI. In this study, we have used these data sets, as well as the Biopharmaceutics Drug Disposition Classification System data set, to generate Bayesian machine learning models with our in-house software, Assay Central. The performance of all machine learning models was assessed through both the internal 5-fold cross-validation metrics and prediction accuracy of an external test set of compounds with known hepatotoxicity. The best-performing Bayesian model was based on the DILI-concern category from the DILIRank database with an ROC of 0.814, a sensitivity of 0.741, a specificity of 0.755, and an accuracy of 0.746. A comparison of alternative machine learning algorithms, such as k-nearest neighbors, support vector classification, AdaBoosted decision trees, and deep learning methods, produced similar statistics to those generated with the Bayesian algorithm in Assay Central. This study demonstrates machine learning models grouped in a tool called MegaTox that can be used to predict early-stage clinical compounds, as well as recent FDA-approved drugs, to identify potential DILI.