Candidate Attrition Research Articles

Consideration of the Root Causes in Candidate Attrition During Oncology Drug Development.

Consideration of the Root Causes in Candidate Attrition During Oncology Drug Development.

The use of PD-1 functional knockout rats to study idiosyncratic adverse reactions to nevirapine.

Idiosyncratic drug reactions (IDRs) are associated with significant patient morbidity/mortality and lead to considerable drug candidate attrition in drug development. Their idiosyncratic nature makes the study of IDRs difficult. In particular, nevirapine is associated with a relatively high risk of serious skin rash and liver injury. We previously found that nevirapine causes a similar skin rash in female Brown Norway rats, but these animals do not develop significant liver injury. Programmed cell death protein-1 (PD-1) is an immune checkpoint involved in immune tolerance, and anti-PD-1 antibodies have been used to treat cancer. However, they increase the risk of liver injury caused by co-administered drugs. We found that PD-1-/- mice are more susceptible to drug-induced liver injury, but PD-1-/- mice are not a good model for all drugs. In particular, they do not develop a skin rash when treated with nevirapine, at least in part because they lack the sulfotransferase in their skin that forms the reactive metabolite responsible for the rash. Therefore, we developed a PD-1 mutant (PD-1m/m) rat, with an excision in the ligand-binding domain of PD-1, to test whether nevirapine would cause a more serious skin rash in these animals. The PD-1m/m rat was based on a Sprague Dawley background, which has a lower incidence of skin rash than Brown Norway rats. The treated PD-1m/m rats developed more severe liver injury than PD-1-/- mice, but in contrast to expectations, they did not develop a skin rash. Functional knockouts provide a unique tool to study the mechanisms of IDRs.

Heart Rate Variability Assessment of Land Navigation and Load Carriage Activities in Specialist Police Selection.

Police tactical group (PTG) personnel are exposed to physical, mental, and emotional stressors. Consequently, PTG selection courses (SCs) impart similar challenges, often resulting in candidate attrition. Holistic assessment may provide additional support to stakeholders given these risks. Heart Rate Variability (HRV) is an objective holistic stress measure that may be applicable in PTG SCs but has not been thoroughly researched. Therefore, this study aimed to report HRV data in an end-user accessible format and determine the relationship between HRV and attrition. A total of 18 qualified Australian State law enforcement officers completed a 1-day physical readiness assessment. Of those, six males progressed to an additional two-day course, on which this study is focused. This two-day selection consisted of additional physical challenges and occupational assessments. HRV was obtained from 2-lead ECGs and defined as the percentage of R-R intervals that varied by ≥50 ms (pRR50). Data were summarized in a heat map of consecutive short-term analyses. Three candidates withdrew. A logistic regression based on heat map data found high HRV was significant for predicting attrition, χ2 (6) = 8.318, p = 0.004. HRV may provide insight for PTG stakeholders monitoring attrition. While the sample size was limited and replication is needed, this study tentatively establishes value for HRV monitoring in PTG SCs.

Assessing seizure liability in vitro with voltage-sensitive dye imaging in mouse hippocampal slices.

Non-clinical toxicology is a major cause of drug candidate attrition during development. In particular, drug-induced seizures are the most common finding in central nervous system (CNS) toxicity. Current safety pharmacology tests for assessing CNS functions are often inadequate in detecting seizure-inducing compounds early in drug development, leading to significant delays. This paper presents an in vitro seizure liability assay using voltage-sensitive dye (VSD) imaging techniques in hippocampal brain slices, offering a powerful alternative to traditional electrophysiological methods. Hippocampal slices were isolated from mice, and VSD optical responses evoked by stimulating the Schaffer collateral pathway were recorded and analyzed in the stratum radiatum (SR) and stratum pyramidale (SP). VSDs allow for the comprehensive visualization of neuronal action potentials and postsynaptic potentials on a millisecond timescale. By employing this approach, we investigated the in vitro drug-induced seizure liability of representative pro-convulsant compounds. Picrotoxin (PiTX; 1-100 μM), gabazine (GZ; 0.1-10 μM), and 4-aminopyridine (4AP; 10-100 μM) exhibited seizure-like responses in the hippocampus, but pilocarpine hydrochloride (Pilo; 10-100 μM) did not. Our findings demonstrate the potential of VSD-based assays in identifying seizurogenic compounds during early drug discovery, thereby reducing delays in drug development and providing insights into the mechanisms underlying seizure induction and the associated risks of pro-convulsant compounds.

Message Passing Neural Networks Improve Prediction of Metabolite Authenticity.

Cytochrome P450 enzymes aid in the elimination of a preponderance of small molecule drugs, but can generate reactive metabolites that may adversely react with protein and DNA and prompt drug candidate attrition or market withdrawal. Previously developed models help understand how these enzymes modify molecule structure by predicting sites of metabolism or characterizing formation of metabolite-biomolecule adducts. However, the majority of reactive metabolites are formed by multiple metabolic steps, and understanding the progenitor molecule's network-level behavior necessitates an integrative approach that blends multiple site of metabolism and structure inference models. Our previously developed tool, XenoNet 1.0, generates metabolic networks, where nodes are molecules and weighted edges are metabolic transformations. We extend XenoNet with a bidirectional message passing neural network that integrates edge feature information and local network structure using edge-conditioned graph convolutions and jumping knowledge to predict the authenticity of inferred Phase I metabolite structures. Our model significantly outperformed prior work and algorithmic baselines on a data set of 311 networks and 6606 intermediates annotated using a chemically diverse set of 20 736 individual in vitro and in vivo reaction records accounting for 92.3% of all human Phase I metabolism in the Accelrys Metabolite Database. Cross-validated predictions resulted in area under the receiver operating characteristic curves of 88.5% and 87.6% for separating experimentally observed and unobserved metabolites at global and network levels, respectively. Further analysis verified robustness to networks of varying depth and breadth, accurate detection of metabolites, such as d,l-methamphetamine, that are experimentally observed or unobserved in different network contexts, extraction of important metabolic subnetworks, and identification of known bioactivation pathways, such as for nimesulide and terbinafine. By exploiting network structures, our approach accurately suggests unreported metabolites for experimental study and may rationalize modifications for avoiding deleterious pathways antecedent to reactive metabolite formation.

Nonclinical Safety Signals in PharmaPendium Improve the Predictability of Human Drug-Induced Liver Injury.

Drug-induced liver injury (DILI) is a leading cause of candidate attrition during drug development in the pharmaceutical industry. This study evaluated liver toxicity signals for 249 approved drugs (114 of "most-DILI concern" and 135 of "no-DILI concern") using PharmaPendium and assessed the association between nonclinical and clinical injuries using contingency table analysis. All animal liver findings were combined into eight toxicity categories based on nature and severity. Together, these analyses revealed that cholestasis [odds ratio (OR): 5.02; 95% confidence interval (CI) 1.04-24.03] or liver aminotransferase increases (OR: 1.86; 95% CI 1.09-3.09) in rats and steatosis (OR-1.9; 95% CI 1.03-3.49) or liver aminotransferase increases (OR-2.57; 95% CI 1.4-4.7) in dogs were significant predictors of human liver injury. The predictive value further improved when the liver injury categories were combined into less severe (steatosis, cholestasis, liver aminotransferase increase, hyperbilirubinemia, or jaundice) and more-severe (liver necrosis, acute liver failure, or hepatotoxicity) injuries. In particular, less-severe liver injuries in the following pairs of species predicted human hepatotoxicity {[dog and mouse] (OR: 2.70; 95% CI 1.25-5.84), [dog and rat] (OR-2.61; 95% CI 1.48-4.59), [monkey and mouse] (OR-4.22; 95% CI 1.33-13.32), and [monkey and rat] (OR-2.45; 95% CI 1.15-5.21)} were predictive of human hepatotoxicity. Meanwhile, severe liver injuries in both [dog and rat] (OR-1.9; 95% CI 1.04-3.49) were significant predictors of human liver toxicity. Therefore, we concluded that the occurrence of DILI in humans is highly likely if liver injuries are observed in one rodent and one nonrodent species and that liver aminotransferase increases in dogs and rats can predict DILI in humans. Together, these findings indicate that the liver safety signals observed in animal toxicity studies indicate potential DILI risk in humans and could therefore be used to prioritize small molecules with less potential to cause DILI in humans.

Abstract P2121: Modeling Genetic Dilated Cardiomyopathies With Engineered Heart Tissues From Patient-derived And Isogenic Mutant Induced Pluripotent Stem Cells

The availability of human induced pluripotent stem cells (hiPSCs) has offered the possibility to study human-derived models of different genetic strata of cardiomyopathy for mechanistic discovery and therapeutic development. However, the phenotypes derived from cardiomyocytes differentiated from hiPSCs in conventional 2-dimensional culture systems often fail to reproducibly model clinical presentations, thereby reducing the translatability of readouts from these assays. Here we report 3-dimensional engineered heart tissues (EHTs) produced with a variety of cell lines harboring patient-derived mutations (TTN, RBM20, BAG3) known to cause dilated cardiomyopathy (DCM) which recapitulate contractile deficits associated with DCM. Internal development of EHTs also recapitulate contractile defects with small interfering RNA (siRNA) knockdown models of haploinsufficiency across multiple disease states. These observed findings with EHTs suggests improved maturity in cardiomyocytes through the presentation of microenvironmental cues akin to those seen in vivo . With continued advances in EHT technologies and capabilities, this platform may serve to significantly reduce drug candidate attrition, as well as provide new insights into pathology, prototyping of therapeutic approaches, and mechanism of action that were previously difficult to obtain.

Novel regulators of PrPC biosynthesis revealed by genome-wide RNA interference.

The cellular prion protein PrPC is necessary for prion replication, and its reduction greatly increases life expectancy in animal models of prion infection. Hence the factors controlling the levels of PrPC may represent therapeutic targets against human prion diseases. Here we performed an arrayed whole-transcriptome RNA interference screen to identify modulators of PrPC expression. We cultured human U251-MG glioblastoma cells in the presence of 64’752 unique siRNAs targeting 21’584 annotated human genes, and measured PrPC using a one-pot fluorescence-resonance energy transfer immunoassay in 51’128 individual microplate wells. This screen yielded 743 candidate regulators of PrPC. When downregulated, 563 of these candidates reduced and 180 enhanced PrPC expression. Recursive candidate attrition through multiple secondary screens yielded 54 novel regulators of PrPC, 9 of which were confirmed by CRISPR interference as robust regulators of PrPC biosynthesis and degradation. The phenotypes of 6 of the 9 candidates were inverted in response to transcriptional activation using CRISPRa. The RNA-binding post-transcriptional repressor Pumilio-1 was identified as a potent limiter of PrPC expression through the degradation of PRNP mRNA. Because of its hypothesis-free design, this comprehensive genetic-perturbation screen delivers an unbiased landscape of the genes regulating PrPC levels in cells, most of which were unanticipated, and some of which may be amenable to pharmacological targeting in the context of antiprion therapies.

Discarding Livers From Deceased Donors: Is It Ever OK?

Organ scarcity in liver transplantation (LT) underlies candidate attrition on the waiting list. In the most recent OPTN report, of waitlisted adults followed over 3 years, only 58.7% were transplanted, while 11.8% died and 19.4% were removed from the list for other reasons.1 Improvements in donation, recovery, and allocation have contributed to an increase in LT activity from 6494 to 8896 over a decade with living donors accounting for under 5% of the volume over this time (https://optn.transplant.hrsa.gov/data/).

A Robust, Mechanistically Based In Silico Structural Profiler for Hepatic Cholestasis.

Owing to the primary role which it holds within metabolism of xenobiotics, the liver stands at heightened risk of exposure to, and injury from, potentially hazardous substances. A principal manifestation of liver dysfunction is cholestasis-the impairment of physiological bile circulation from its point of origin within the organ to the site of action in the small intestine. The capacity for early identification of compounds liable to exert cholestatic effects is of particular utility within the field of pharmaceutical development, where contribution toward candidate attrition is great. Shortcomings associated with the present in vitro methodologies forecasting cholestasis render their predictivity questionable, permitting scope for the adoption of computational toxicology techniques. As such, the intention of this study has been to construct an in silico profiler, founded upon clinical data, highlighting structural motifs most reliably associated with the end point. Drawing upon a list of >1500 small molecular drugs, compiled and annotated by Kotsampasakou, E. and Ecker, G. F. (J. Chem. Inf. Model. 2017, 57, 608-615), we have formulated a series of 15 structural alerts. These describe fragments intrinsic within distinct pharmaceutical classes including psychoactive tricyclics, β-lactam antimicrobials, and estrogenic/androgenic steroids. Description of the coverage and selectivity of each are provided, alongside consideration of the underlying reactive mechanisms and relevant structure-activity concerns. Provision of mechanistic anchoring ensures that potential exists for framing within the adverse outcome pathway paradigm-the chemistry conveyed through the alert, in particular enabling rationalization at the level of the molecular initiating event.

Application of a Rat Liver Drug Bioactivation Transcriptional Response Assay Early in Drug Development That Informs Chemically Reactive Metabolite Formation and Potential for Drug-induced Liver Injury.

Drug-induced liver injury is a major reason for drug candidate attrition from development, denied commercialization, market withdrawal, and restricted prescribing of pharmaceuticals. The metabolic bioactivation of drugs to chemically reactive metabolites (CRMs) contribute to liver-associated adverse drug reactions in humans that often goes undetected in conventional animal toxicology studies. A challenge for pharmaceutical drug discovery has been reliably selecting drug candidates with a low liability of forming CRM and reduced drug-induced liver injury potential, at projected therapeutic doses, without falsely restricting the development of safe drugs. We have developed an in vivo rat liver transcriptional signature biomarker reflecting the cellular response to drug bioactivation. Measurement of transcriptional activation of integrated nuclear factor erythroid 2-related factor 2 (NRF2)/Kelch-like ECH-associated protein 1 (KEAP1) electrophilic stress, and nuclear factor erythroid 2-related factor 1 (NRF1) proteasomal endoplasmic reticulum (ER) stress responses, is described for discerning estimated clinical doses of drugs with potential for bioactivation-mediated hepatotoxicity. The approach was established using well benchmarked CRM forming test agents from our company. This was subsequently tested using curated lists of commercial drugs and internal compounds, anchored in the clinical experience with human hepatotoxicity, while agnostic to mechanism. Based on results with 116 compounds in short-term rat studies, with consideration of the maximum recommended daily clinical dose, this CRM mechanism-based approach yielded 32% sensitivity and 92% specificity for discriminating safe from hepatotoxic drugs. The approach adds new information for guiding early candidate selection and informs structure activity relationships (SAR) thus enabling lead optimization and mechanistic problem solving. Additional refinement of the model is ongoing. Case examples are provided describing the strengths and limitations of the approach.

Most Influential Physicochemical and In Vitro Assay Descriptors for Hepatotoxicity and Nephrotoxicity Prediction.

Drug-induced organ injury is a major reason for drug candidate attrition in preclinical and clinical drug development. The liver, kidneys, and heart have been recognized as the most common organ systems affected in safety-related attrition or the subject of black box warnings and postmarket drug withdrawals. In silico physicochemical property calculations and in vitro assays have been utilized separately in the early stages of the drug discovery and development process to predict drug safety. In this study, we combined physicochemical properties and in vitro cytotoxicity assays including mitochondrial dysfunction to build organ-specific univariate and multivariable logistic regression models to achieve odds ratios for the prediction of clinical hepatotoxicity, nephrotoxicity, and cardiotoxicity using 215 marketed drugs. The multivariable hepatotoxic predictive model showed an odds ratio of 6.2 (95% confidence interval (CI) 1.7-22.8) or 7.5 (95% CI 3.2-17.8) for mitochondrial inhibition or drug plasma Cmax >1 μM for drugs associated with liver injury, respectively. The multivariable nephrotoxicity predictive model showed an odds ratio of 5.8 (95% CI 2.0-16.9), 6.4 (95% CI 1.1-39.3), or 15.9 (95% CI 2.8-89.0) for drug plasma Cmax >1 μM, mitochondrial inhibition, or hydrogen-bond-acceptor atoms >7 for drugs associated with kidney injury, respectively. Conversely, drugs with a total polar surface area ≥75 Å were 79% (odds ratio 0.21, 95% CI 0.061-0.74) less likely to be associated with kidney injury. Drugs belonging to the extended clearance classification system (ECCS) class 4, where renal secretion is the primary clearance mechanism (low permeability drugs that are bases/neutrals), were 4 (95% CI 1.8-9.5) times more likely to to be associated withkidney injury with this data set. Alternatively, ECCS class 2 drugs, where hepatic metabolism is the primary clearance (high permeability drugs that are bases/neutrals) were 77% less likely (odds ratio 0.23 95% CI 0.095-0.54) to to be associated with kidney injury. A cardiotoxicity model was poorly defined using any of these drug physicochemical attributes. Combining in silico physicochemical properties descriptors along with in vitro toxicity assays can be used to build predictive toxicity models to select small molecule therapeutics with less potential to cause liver and kidney organ toxicity.

Improving Drug Design: An Update on Recent Applications of Efficiency Metrics, Strategies for Replacing Problematic Elements, and Compounds in Nontraditional Drug Space.

Drug discovery and development is a complex and lengthy enterprise that suffers from high rates of candidate attrition at all stages of the process. The physical, biological, and toxicological properties of a drug candidate are inextricably linked to its structure, and once a molecule has been synthesized, all subsequent studies along the development path are focused only on assessing and understanding its properties in greater detail. Unfortunately, a full prediction of the biological properties of a molecule from an analysis of its 2- or 3-dimensional structure is currently beyond our expertise. This backdrop mandates that considerable care be taken at the design stage if a molecule is to be successful in testing a mechanistic concept underlying a disease process and to progress into late stage clinical trials and, ultimately, marketing approval. While there are multiple potential causes of candidate attrition, an introspective analysis of drug design practices over the past decade has focused attention on the perception that contemporary molecules are unnecessarily obese, burdened by high molecular weight and excessive lipophilicity. This practice is believed to have its roots in the singular pursuit of enhancing potency during lead optimization rather than adopting a more holistic approach to drug design that gives broader consideration to how structural features affect developability properties. In an effort to provide the medicinal chemistry community with practical guideposts to enhancing compound quality in the drug design phase and which can readily be applied, a series of efficiency indices have been proposed that attempt to define aspects of compound quality in the context of a series of physicochemical parameters. Of these metrics, lipophilic ligand efficiency (LLE or LipE), which provides an index of the dependence of the potency of a molecule on its intrinsic lipophilicity, has been characterized as the most robust metric that has potential for broad-based application. In this review, after describing the background literature behind the derivation of efficiency metrics and approaches to assessing compound aesthetics, synopses of some recent practical application in lead optimization campaigns are presented. However, molecules that fall into space beyond that associated with traditional drug-like properties are an important part of the current and future landscape, exemplified by the summary of direct acting hepatitis C virus NS3 and NS5A inhibitors that have transformed clinical therapy for this chronic disease. While drug development in nontraditional drug-like space is more challenging and the rules for compound quality will be different with much still to be understood, careful and disciplined drug design practices will be an essential element of success.

ChemInform Abstract: Metabolic Engineering: The Ultimate Paradigm for Continuous Pharmaceutical Manufacturing

AbstractReview: [32 refs.

Biosynthonics: Charting the Future Role of Biocatalysis and Metabolic Engineering in Drug Discovery

Ironic as it sounds, the pharmaceutical industry is ailing. Despite doubling their R&D expenditures over the past decade, innovator drug companies have struggled to increase their annual drug approval rates. Worse still, candidate attrition rates and development times actually rose during this period. It now costs well in excess of a billion dollars and takes over a decade to discover and develop a new drug. A variety of prognostications have been made regarding the future course that the pharmaceutical industry might take, but none has been as controversial as the suggestion that small-molecule drugs could disappear by 2030. Rest assured that these claims could not be further from the truth. Herein, I outline how biocatalysis and metabolic engineering could drive the next wave of innovation in small-molecule drug discovery. The recent past has witnessed staggering developments in analytical chemistry, genome sequencing and assembly, flow chemistry, chemi- and bioinformatics, metabolic engineering and synthetic chemistry, as well as exponential improvements in the raw computing power that is available for drug research; integrating these parallel innovations into a single platform could translate a vastly higher number of small-molecule drugs to the bedside. I term this integrated approach as “biosynthonics”. Many of its individual elements are already in use drug discovery and development, albeit in a scattered manner, and despite the substantial progress that has occurred in each individual domain, a lot still remains to be achieved if biosynthonics is to become a broad platform for generating focused libraries of natural product-like molecules. It is hoped that this article spurs debate and future collaborations toward achieving these goals, which, if successful, would undoubtedly improve small-molecule R&D productivity.

Metabolic Engineering: The Ultimate Paradigm for Continuous Pharmaceutical Manufacturing

Research and development (R&D) expenditures by pharmaceutical companies doubled over the past decade, yet candidate attrition rates and development times rose markedly during this period. Understandably, companies have begun downsizing their pipelines and diverting investments away from R&D in favor of manufacturing. It is estimated that transitioning to continuous manufacturing could enable companies to compete for a share in emerging markets. Accordingly, the model for continuous manufacturing that has emerged commences with the conversion of late-stage intermediates into the active pharmaceutical ingredient (API) in a series of continuous flow reactors, followed by continuous solid processing to form finished tablets. The use of flow reactions for API synthesis will certainly generate purer products at higher yields in shorter times compared to equivalent batch reactions. However, transitioning from batch to flow configuration simply alleviates transport limitations within the reaction milieu. As the catalogue of reactions used in flow syntheses is a subset of batch-based chemistries, molecules such as natural products will continue to evade drug prospectors. Also, it is uncertain whether flow synthesis can deliver improvements in the atom and energy economies of API production at the scales that would achieve the levels of revenue growth targeted by companies. Instead, it is argued that implementing metabolic engineering for the production of oxidized scaffolds as gateway molecules for flow-based addition of electrophiles is a more effective and scalable strategy for accessing natural product chemical space. This new paradigm for manufacturing, with metabolic engineering as its engine, would also permit rapid optimization of production variables and allow facile scale-up from gram to ton scale to meet material requirements for clinical trials, thus recasting manufacturing as a tool for discovery.

Higher degree research candidates’ initial expectations in fine art

A range of research studies in the past decade have yielded a much fuller understanding of the nature of higher degree research candidature, student experiences and the changes in this population. Nonetheless candidate attrition continues to be a source of concern and there is still much territory to explore in connection with barriers to progress and their successful resolution. One aspect that is rarely explored is research candidates’ initial expectations at masters and doctoral level. Drawing on interviews with fine art masters and doctoral level candidates, this paper explores candidates’ recollections of their initial expectations of candidature in a practice-based discipline, and any ‘mismatch’ or incongruence between their expectations and academic requirements. We found initial expectations clustered under two main areas: expectations about the research degree and expectations about the institutional environment. Suggestions for supervision practice are presented from the findings.

Back to the future: can physical models of passive membrane permeability help reduce drug candidate attrition and move us beyond QSPR?

It is widely recognized that adsorption, distribution, metabolism, excretion, and toxicology liabilities kill the majority of drug candidates that progress to clinical trials. The development of computational models to predict small molecule membrane permeability is therefore of considerable scientific and public health interest. Empirical qualitative structure permeability relationship models of permeability have been a mainstay in industrial applications, but lack a deep understanding of the underlying biologic physics. Others and we have shown that implicit solvent models to predict passive permeability for small molecules exhibit mediocre predictive performance when validated across experimental test sets. Given the vast increase in computer power, more efficient parallelization schemes, and extension of current atomistic simulation codes to general use graphical processing units, the development and application of physical models based on all-atom simulations may now be feasible. Preliminary results from rigorous free energy calculations using all-atom simulations indicate that performance relative to implicit solvent models may be improved, but many outstanding questions remain. Here, we review the current state-of-the-art physical models for passive membrane permeability prediction and present a prospective look at promising new directions for all-atom approaches.

Building Bridges to the Future of Medicine: Recommendations for Boosting Development of Novel and Companion Diagnostics

Evaluating the prospects for 'personalized healthcare' has become topical at the 10-year anniversary of the first results from the Human Genome Project. The coincidence of this milestone with a period of global financial difficulty is unfortunate, but industry has signaled its willingness to invest in this new medical paradigm in order to address the problems of high drug candidate attrition as well as escalating healthcare and R&D costs. For diagnostics manufacturers, justifying the ever-growing costs of development without an appropriate framework to provide reasonable return on investment is extremely challenging. Regulators have made good progress in defining what will satisfy current regulatory mandates, but there is still far to go to create conditions that would stimulate investment. My personal perspective is that personalized healthcare represents a solution rather than a problem and herein I describe the bottlenecks slowing development of novel diagnostics from the perspective of a willing commercial stakeholder.

Preclinical Strategy to Reduce Clinical Hepatotoxicity Using in Vitro Bioactivation Data for >200 Compounds

Drug-induced liver injury is the most common cause of market withdrawal of pharmaceuticals, and thus, there is considerable need for better prediction models for DILI early in drug discovery. We present a study involving 223 marketed drugs (51% associated with clinical hepatotoxicity; 49% non-hepatotoxic) to assess the concordance of in vitro bioactivation data with clinical hepatotoxicity and have used these data to develop a decision tree to help reduce late-stage candidate attrition. Data to assess P450 metabolism-dependent inhibition (MDI) for all common drug-metabolizing P450 enzymes were generated for 179 of these compounds, GSH adduct data generated for 190 compounds, covalent binding data obtained for 53 compounds, and clinical dose data obtained for all compounds. Individual data for all 223 compounds are presented here and interrogated to determine what level of an alert to consider termination of a compound. The analysis showed that 76% of drugs with a daily dose of <100 mg were non-hepatotoxic (p < 0.0001). Drugs with a daily dose of ≥100 mg or with GSH adduct formation, marked P450 MDI, or covalent binding ≥200 pmol eq/mg protein tended to be hepatotoxic (∼ 65% in each case). Combining dose with each bioactivation assay increased this association significantly (80-100%, p < 0.0001). These analyses were then used to develop the decision tree and the tree tested using 196 of the compounds with sufficient data (49% hepatotoxic; 51% non-hepatotoxic). The results of these outcome analyses demonstrated the utility of the tree in selectively terminating hepatotoxic compounds early; 45% of the hepatotoxic compounds evaluated using the tree were recommended for termination before candidate selection, whereas only 10% of the non-hepatotoxic compounds were recommended for termination. An independent set of 10 GSK compounds with known clinical hepatotoxicity status were also assessed using the tree, with similar results.