227

Drug-Induced Liver Injury and Drug Development: Industry Perspective Arie Regev, MD1 1 Division of Gastroenterology and Hepatology, Indiana University

School of Medicine, Indianapolis, Indiana

Address for correspondence Arie Regev, MD, Division of Gastroenterology and Hepatology, Indiana University School of Medicine, Indianapolis, IN 46285 (e-mail: [email protected]).

Abstract

Keywords

► Drug-induced liver injury ► DILI prediction ► Hy’s law ► causality assessment ► drug development

Despite intensive ongoing research, drug-induced live injury (DILI) remains a serious issue for care providers and patients, and has been a major cause of drug withdrawal and non-approval by regulatory authorities in the past 50 years. Consequently, DILI remains a major concern for the pharmaceutical industry and a leading cause for attrition during drug development. In most instances, severe DILI is an uncommon idiosyncratic reaction, which typically does not present during preclinical phases or early clinical phases of drug development. In the majority of cases, drugs that caused severe DILI in humans have not shown clear and consistent hepatotoxic signals in preclinical assessment including animal studies, cell cultures, or other methods. Despite intensive efforts to develop better biomarkers that would help in predicting DILI risk in earlier phases of drug development, such biomarkers are currently not supported by sufficient evidence and are not yet available for routine use by drug makers. Due to the lack of effective and accurate methods for prediction of idiosyncratic DILI during preclinical phases of drug development, different drug makers have adopted different approaches, which are often not supported by strong systematic evidence. Based on growing experience, it is becoming increasingly evident that milder forms of liver injury occurring during clinical development, when assessed correctly, may significantly enhance our ability to predict the drug’s potential to cause more severe liver injury postmarketing. Strategies based on this concept have been adopted by many drug makers, and are being increasingly implemented during drug development. Meticulous causality assessment of individual hepatic cases and adherence to strict hepatic discontinuation rules are critical components of this approach and have to rely on thorough clinical evaluation and occasionally on assessment by liver experts experienced with DILI and drug development.

Drug-induced liver injury (DILI) has been a major cause of drug withdrawal, nonapproval, and regulatory action in the past 50 years. According to a recent review, of 47 drugs withdrawn during the period 1975 to 2007, 15 (32%) involved hepatotoxicity.1 As a result, DILI has been a major concern for drug makers and has remained a leading cause for attrition during drug development. In the vast majority of cases, drugs that caused severe DILI did so infrequently and unexpectedly, in a manner consistent with idiosyncratic DILI (IDILI). Most of the drugs withdrawn from the market for hepatotoxicity,

Issue Theme Drug-Induced Liver Injury; Guest Editors, Naga Chalasani, MD, and Paul H. Hayashi, MD, MPH

have caused liver failure leading to death or transplantation at frequencies lower than 1 per 10,000, and in most instances, such cases were identified only after drug approval and marketing.2,3 Most drug development databases, which typically include up to a few thousand subjects, have not shown any cases of severe liver injury, even when the drug was later found to have a high potential of causing DILI. Despite intensive research and development efforts, there are currently no specific biomarkers that can accurately predict IDILI in preclinical or clinical phases of drug development.4 In

Copyright © 2014 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662.

DOI http://dx.doi.org/ 10.1055/s-0034-1375962. ISSN 0272-8087.

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Semin Liver Dis 2014;34:227–239.

Drug-Induced Liver Injury and Drug Development: Industry Perspective addition, most drugs that were later found to cause severe DILI, did not exhibit any unequivocal signals in preclinical studies.3 Preclinical studies have been more useful in detecting overtly hepatotoxic agents, which cause dose-related on-target liver injury (e.g., carbon tetrachloride, chloroform, methylene chloride). As a result, drugs that cause such predictable and doserelated injury often (although not always) are discovered and rejected in preclinical testing.3 In contrast, the vast majority of drugs that caused severe DILI in humans have not shown hepatotoxicity in animal studies, and have typically not exhibited clear predictive signals in studies using liver cell cultures. Identifying drugs that cause IDILI during drug development remains a major challenge because this type of DILI largely depends on individual susceptibility of affected patients, which in most cases has not yet been well characterized. Despite this well-accepted notion, some drug makers take various measures during preclinical phases of drug development that are intended to decrease the risk of IDILI, by steering away from drug candidates that are perceived to be “high risk.” In addition, drug makers are increasingly bombarded with innovative experimental preclinical methods that claim to have high predictive value for IDILI. Unfortunately, in the vast majority of cases these methods rely on little prospective evidence and are almost never supported by controlled clinical studies. In the last three decades, it has become increasingly clear that meticulous collection and evaluation of clinical trial data during drug development may uncover evidence and signals of a drug’s potential for severe hepatotoxicity, even in the absence of cases of severe DILI. Current approaches to assessment and prediction of IDILI during drug development are therefore largely based on carful data collection and close monitoring of clinical trial subjects, during exposure to a new drug. Drug-induced liver injury may mimic almost any known type of liver disease.5,6 Of the many clinicopathologic types of DILI, acute hepatocellular injury has been associated with most of the significant hepatotoxic drugs.3 Although there are a few exceptions, hepatocellular DILI typically tends to progress more rapidly and has a higher tendency to lead to liver failure and death compared with other types. Furthermore, most (although not all) of the cases of drug withdrawal and nonapproval in the past 50 years have been related to hepatocellular DILI. As a result, efforts to identify and predict DILI have largely been directed at this type of injury. Still other forms of DILI, including cholestatic liver injury,7 steatosis,8 steatohepatitis, and hepatic fibrosis present increasing challenges to drug makers, and caregivers. This review will focus on the current dilemmas and evolving approaches of drug makers pertaining to prediction, monitoring, assessment, and management of IDILI during drug development.

Historical Perspective The awareness of DILI and approach toward identification and prediction by drug makers and regulators have changed significantly in the last six decades. Although historically it took several years and occasionally decades before IDILI was recognized, the time to identification and risk-management Seminars in Liver Disease

Vol. 34

No. 2/2014

Regev

has been shortened considerably in today’s drug development environment.9,10 One of the first well-studied examples was iproniazid, which was initially developed for treatment of tuberculosis and was subsequently approved by the U.S. Food and Drug Administration (FDA) for treatment of depression in 1958. In the mid- and late 1950s several cases of hepatic injury in patients taking iproniazid were reported to the manufacturer and published in the medical literature.11 In 1959, Zimmerman et al12 reported 86 patients that developed jaundice while receiving iproniazid. Most of these patients had hepatocellular injury and mortality rate was
20%. In the majority of cases alanine aminotransferase (ALT) or aspartate aminotransferase (AST) levels were above 7 times () the upper limit of normal (ULN), and in nearly 60% alkaline phosphatase (ALP) levels were not higher than 2 ULN. It was not until 1961 that iproniazid was withdrawn from the market due to hepatotoxicity. During the ensuing years, the time to DILI-identification has decreased significantly; however, in several well-documented cases hepatotoxic-drugs were identified as such only after being approved by regulators and marketed for several months or years. Ticrynafen (Tienilic acid), the third loop diuretic to be approved by the FDA was approved by the FDA in May 1979 for the treatment of hypertension. During the months after its approval, several hundreds of liver injury cases were reported to the manufacturer.13,14 A retrospective analysis of 340 of these cases showed that most of them presented with hepatocellular injury and jaundice. Jaundice was present in 246 of 287 patients with sufficient clinical information, and
10% of the patients who had jaundice died.14 Ticrynafen was withdrawn from the U.S. market in 1980. Other loop diuretics such as furosemide, ethacrynic acid, bumetanide, and torsemide approved in the United States in 1966, 1967, 1983, and 1993, respectively, have not been associated with clinically apparent liver injury, and DILI from these loop diuretics seems to be exceedingly rare, if it occurs at all.15 Interestingly, although it is very safe to the human liver, furosemide causes direct hepatotoxicity in mice and has been used as an animal model of DILI.16 Benoxaprofen is a nonsteroidal anti-inflammatory drug (NSAID) that was approved for marketing in the United Kingdom (UK) in 1980 and in the United States in 1982. In 1982, several cases of jaundice with fatal outcome in association with benoxaprofen were reported in the medical literature.17 The majority of these events occurred in elderly women and consisted of predominantly cholestatic injury, which was often accompanied by kidney dysfunction. A striking feature of most of these cases was the rapidly fatal course after the onset of jaundice despite the immediate withdrawal of the drug.17 It was shown that when used in higher doses, benoxaprofen had a much longer half-life in elderly patients than in younger patients.18 These reports together with reports from the British government and the FDA led to the suspension of sales and withdrawal of the drug from the market in 1982. Bromfenac is an NSAID, which was approved by the FDA in July 1997 for short-term use (up to 10 days). During the

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

228

months after its approval, the manufacturer and FDA received several reports of liver injury and hepatic failure, including fatal cases and cases requiring liver transplantation. In most cases, patients who suffered from DILI had taken the medication for longer than the recommended 10-day period. A few of these cases were reported in the medical literature.19–22 Bromfenac was eventually withdrawn from the market in June 1998, 11 months after its approval. Troglitazone was the first thiazolidinedione approved in the United States and was licensed for use in type 2 diabetes in January 1997. After marketing approval, the drug maker and the FDA received numerous reports of severe liver injury associated with troglitazone use and several reports of acute liver failure caused by troglitazone were published in the medical literature.23–25 It was estimated that clinically significant liver injury occurred in
1:10,000 recipients.15 Troglitazone was eventually withdrawn from the U.S. market in March 2000, more than 3 years after its approval. A retrospective analysis of the combined North American clinical trials of troglitazone revealed a few interesting findings: while the total incidence of ALT elevations of any magnitude was higher in patients receiving placebo than in those receiving troglitazone (possibly due to a favorable effect on steatosis), elevation in ALT concentrations to more than 3x ULN occurred significantly more frequently in troglitazone-treated patients (1.9%) compared to patients given placebo (0.6%) (►Table 1).26 Elevations of more than 10 ULN occurred in 0.5% of patients (but none in placebo recipients) and a proportion of these developed symptoms of liver injury and jaundice. Five troglitazone treated patients (and no placebo-treated patients) had ALT levels of more than 20 ULN. The onset of the serum ALT elevations was typically delayed; in most patients, the peak values occurred between the third and seventh months. Treatment was discontinued because of abnormal hepatic biochemical tests in 20 troglitazone-treated patients (0.8%) and in no placebo-treated patients. In all 20 patients in whom therapy was discontinued, serum ALT levels returned to baseline. Two troglitazone-treated patients (and no placebotreated patients) had ALT elevation to more than 3 ULN and concomitant total bilirubin (TBL) elevation to more than 2 ULN, which was thought to be drug-related, consistent with the definition of Hy’s law (►Table 1).26 In other instances DILI has led to nonapproval or termination of drug development during late phases of drug develTable 1 Combined results of North American clinical studies with troglitazone14 Troglitazone

Placebo

Total number of subjects

2,510

475

Any ALT increase

""

"""

ALT> 3 ULN

48 (1.9%)

3 (0.6%)

ALT> 15 ULN

10

0

ALT> 20 ULN

5

0

Hy’s law levels

2

0

Abbreviations: ALT, Alanine aminotransferase; ULN, Upper limit of normal.

Regev

opment or final stages of new-drug application (NDA). Dilevalol is a nonselective β  adreno-receptor antagonist that was developed for treatment of hypertension. Clinical trials with dilevalol showed two Hy’s law cases3 in
1,000 exposures. The NDA was filed in 1986, but the manufacturer subsequently withdrew the application and discontinued the registration and marketing worldwide due to these cases, as well as other clinical trial DILI cases, and cases discovered postapproval in Portugal and Japan.27 Ximelagatran is an oral direct thrombin inhibitor developed for the prevention and treatment of thromboembolic conditions. In February 2006, the manufacturers of ximelagatran withdrew a pending applications to the FDA and the European Medicines Agency (EMA) for marketing approval due to reports of abnormal liver tests and DILI occurring during clinical trials.28,29 Review of the clinical trial data showed that 7.9% of patients receiving ximelagatran had ALT elevation to more than 3 ULN, compared with 1.2% of the comparator group. Combined elevations of ALT level of more than 3 ULN and TBL level of more than 2 ULN, regardless of etiology, were observed in 37 patients (0.5%) treated with ximelagatran. Eleven of these patients were found to have no alternative cause for liver injury. One of these patients sustained severe liver damage with liver failure and eventually died.29 At the time of withdrawal of the pending application to the FDA, the drug had already been approved in several European and South American countries. Animal studies and studies using primary human hepatocyte cultures did not show any hepatotoxic effects including with concentrations that were considerably higher than those found in plasma following therapeutic dosing.30 Despite intensive investigations, the cellular mechanism behind the ximelagatran-related hepatic effects has remained unknown.31

DILI Prediction during Preclinical Phases of Drug Development The need to identify a drug’s potential to cause DILI during the process of drug development is a critical issue for drug makers. Minimizing exposure of clinical trial subjects and patients to drugs with hepatotoxic potential is a top priority. Moreover, the need to steer away from such drugs as early as possible during drug development is becoming increasingly important at a time when the cost and time required to develop a new drug are at an all-time high.1 The process of drug development is complex, expensive, and inherently hit and miss, drawing on multiple disciplines.1,9 Balancing risk against the expected clinical benefit is the essence of preclinical and clinical testing. The current preclinical testing paradigm (►Fig. 1), established over 30 years ago, has improved drug safety markedly; however, some areas, including the area of DILI prediction, still present substantial challenges to the pharmaceutical industry. In general, preclinical testing is designed to determine the dose-limiting toxicity, reversibility of toxicity, and if toxicity can be monitored clinically.1,32 Preclinical findings are used to develop a margin of safety based on the difference between the efficacy dose and the highest dose that does not cause Seminars in Liver Disease

Vol. 34

No. 2/2014

229

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Regev

Fig. 1 The drug development process: an emphasis on drug safety during early phases of drug development. ADME, absorption, distribution, metabolism, and excretion; Cmpd, compound; H2L, hit-to-lead; ID identify, IND, investigational new drug; LO, lead optimization; NDA, new drug application. (Reproduced from Stevens and Baker1 with permission.)

toxicity, also called the “no observed adverse effect level” (NOAEL). Understanding the mechanisms of toxicity may be helpful in extrapolating the preclinical data to humans. Although there is some evidence to suggest that as much as 70% of the human toxicity seen during clinical trials may be predicted by preclinical studies,33 this is clearly not the case when it comes to IDILI. It seems that the current preclinical testing paradigm, which may be useful for identifying drugs with direct hepatotoxic effect (sometimes called “true,” “intrinsic,” or “dose-dependent” hepatotoxins), often falls short of identifying drugs associated with IDILI. As a result, IDILI is only rarely identified during preclinical phases of drug development,2,3 and new approaches and prediction methodologies are sorely needed. This limitation in our ability to predict IDILI during these early phases, using currently available methods, is well recognized by most regulatory authorities3,34 and by most clinicians and preclinical scientists.35 Nevertheless, it is still a matter of some debate; as a result, drug makers may not always agree on the type of approaches they choose to use for predicting or assessing the risk of IDILI during preclinical phases of drug development.1,32

Animal Studies Although helpful in steering away from overtly hepatotoxic drugs with dose-dependent/ on-target hepatotoxic effect, the role of animal studies in predicting IDILI in humans remains questionable. In the majority of cases, drugs that caused severe DILI in humans did not cause significant hepatotoxic effects in animals.3 Conversely, there are well known examSeminars in Liver Disease

Vol. 34

No. 2/2014

ples of drugs (e.g., hydroxyl-methylglutaryl coenzyme Areductase inhibitors [statins], and furosemide) that cause significant hepatic injury in animals,16,36,37 but are considered very safe to the human liver.15 Numerous attempts at development of animal models for IDILI based on hypothesized mechanisms have met with serious challenges.38,39 One of the major impediments to the development of reproducible animal models has been incomplete understanding of the mechanisms underlying IDILI. Another obstacle is the marked species differences between humans and animals in response to, and in the metabolism of xenobiotics. As a result, there is currently no universally accepted animal model for IDILI, and such models have not been recognized or approved by regulatory agencies.3 In general, because many drugs that cause significant hepatotoxicity in animals do not progress to clinical phases of drug development, there is limited data for a systematic assessment of the predictive value of such findings for DILI in humans.9 Earlier retrospective studies suggesting some concordance between clinical hepatic safety signals and hepatic findings in animals largely focused on elevated levels of aminotransferases,33 which are recognized today to be of low specificity and poor predictive value for IDILI. Whether or not abnormal hepatic findings in animals predict elevated aminotransferase levels in clinical trials, the significance of such prediction is unclear because many safe and effective drugs are associated with benign self-limiting aminotransferase elevation, with no significant hepatic injury.2,3 Typically, animal studies during drug development involve the use of at least two species including rodents and nonrodents for acute,

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

230

subacute, and chronic duration studies. Although supported by scarce published evidence33 it is widely believed that nonrodent studies are generally more predictive of human toxicity than rodent studies. Still, the predictive value of nonrodent studies for IDILI is also a matter of debate. Despite these limitations, many drug manufacturers, for lack of better predictive methods for IDILI, continue to rely on animal studies, which may increase the risk of unnecessary termination of potentially safe and effective medications due to adverse animal findings.9 To improve the predictive value of animals for DILI, several investigators have proposed the use of humanized animals. For example, mice expressing human cytochrome P450 enzymes or harboring transplanted human hepatocytes could be useful for the investigation of the initiation of hepatotoxicity owing to reactive metabolite formation.40 Nevertheless, considering the multicellular nature and physiological aspects of DILI, these models will require substantial refinement.41

Reactive Metabolites The occurrence of hepatotoxicity has been often associated with formation of chemically reactive metabolites that have the potential to bind irreversibly through covalent binding (CB) and modify cellular macromolecules.41 Reactive metabolite formation is thought to be a key initiation mechanism in the development of idiosyncratic hepatotoxicity, as well as on-target reactions, and certain substructures are considered structural alerts and a potential hazard. Consequently, several potential toxicophores have been identified, which medicinal chemists try to avoid during the design of a new chemical entity, such as furans and thiophenes.42 Drug manufacturers have made large investments in the screening of chemically reactive metabolites in an attempt to offer a pragmatic approach to produce safer drugs and reduce drug attrition. High-throughput screens have been developed to identify reactive metabolites early in lead optimization (►Fig. 1)

Regev

based on covalent binding potential and glutathione (GSH) adduct formation. That said, questions surrounding the importance of chemically reactive metabolites for predicting DILI still remain.41 Clearly, the relationship between bioactivation and the occurrence of DILI is not simple. It is possible for chemicals to undergo bioactivation in the liver without causing hepatotoxicity, as seen with therapeutic doses of acetaminophen. Ultimately, it is the balance among bioactivation, detoxification, and defense mechanisms that determine whether a reactive metabolite elicits a toxic effect.43 Furthermore, increasing evidence now exists for the multifactorial nature of DILI, in particular the possible roles played by the host immune system and specific metabolic pathways in the pathogenesis of DILI. Evidently, not all cases of DILI are associated with reactive metabolite formation. Among the well-known examples are flutamide44 and ximelagatran.30,31 Studies attempting to examine the predictive value of reactive metabolites and CB have shown conflicting results.45–47 Although a few studies found some predictive value,45 others found very limited predictability.46,47 A comparison of
100 Merck drug candidates indicated that there was no correlation between incidence of liver toxicity observed in vivo in preclinical safety studies and level of CB (unpublished data reported by Park et al) (►Table 2).47 Unfortunately, these studies have often suffered from methodological limitations. One consistent drawback pertains to the choice of drugs used to test a new predictive method. In some studies, drugs chosen as reference for hepatotoxic drugs were in reality very safe from the liver standpoint. For example, a recent study found that while in vitro CB levels in human hepatocytes failed to predict DILI, multiplying the CB amount by the maximum daily dose was able to discriminate between drugs that were considered hepatotoxic and those that were not.45 However, drugs used as “positive for hepatotoxicity” in this study included drugs that have only very rarely been associated with clinically significant DILI (e.

Table 2 Merck development candidates: CB data and liver findings of concern from rat studies Study In vitro (LM)

In vitro (LM þ GSH) In vivo

Observed pmola

n

Indication of toxicity (%) Positive

Negative

< 50

54

17

83

50–200

33

15

85

> 200b

13

0

100

< 50

51

17

83

50–220

16

14

86

< 50

78

14

86

50–150

9

22

78

Abbreviations: CB, covalent binding; GSH, glutathione; LM, liver microsomes. Source: Unpublished observations reported by Park BK, Boobis A, Clarke S, et al.47 a pmol refers to pmol equivalent per (mg protein at 1 h) for in vitro studies and pmol equivalent per mg protein for in vivo investigations. b Out of the 13 compounds in this group, 12 exhibited values ranging from 204 pmol per mg protein to 621 pmol per mg protein (mean ¼ 374), whereas one compound gave a value of 1,680 pmol per mg protein.

Seminars in Liver Disease

Vol. 34

No. 2/2014

231

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Drug-Induced Liver Injury and Drug Development: Industry Perspective g., tacrine15,48 and ticlopidine),15 which renders the significance of this study unclear. Based on these and other publications, Kevin Park,47 Thomas Baillie, and others from the University of Liverpool, UK, made the following observations and recommendations regarding the predictive powers of CB: (1) CB should be regarded as a marker of bioactivation and not toxicity and therefore must be discussed in the context of drug metabolism and chemical structure. (2) In vitro studies of drug metabolism cannot predict risk in either experimental animals or humans, and can only identify potential hazards. (3) It is not possible to dismiss the potential hazard associated with drug bioactivation based on current science. (4) If the potential liability associated with a particular bioactivation pathway is removed without evaluation of toxicity, we will never be able to measure the success of the process; continued investment is required in approaches and methodologies that can evaluate the contribution of bioactivation to toxicity. (5) There is a need to inform not only those involved in the drug development process, but also those involved in drug regulation, as well as the public, that preclinical scientists do not yet have the tools to define, let alone predict, the role of reactive metabolites in serious adverse drug reactions.

Regev

contrast, they did observe dose-dependent cytotoxicity with troglitazone (a known hepatotoxic drug), but also with Tacrine, which although associated with frequent aminotransferase elevation has very rarely been shown to cause clinically significant DILI. Another group of investigators recently described a novel high-content screening assay based on six parameters: nuclei counts, nuclear area, plasma membrane integrity, lysosomal activity, mitochondrial membrane potential, and mitochondrial area.51 This approach was tested using 100 drugs of which the clinical hepatotoxicity profile was known, and seemed to separate toxic from nontoxic compounds. Unfortunately, the authors included among the group of “hepatotoxicans” drugs like lamivudine, furosemide, nimodipine, simvastatin, and others, which are considered by clinicians to be very safe to the human liver and are very rarely, if ever, associated with clinically significant hepatotoxicity.15 Based on these and other data, it would seem that the use of current cell culture techniques for predicting IDILI is not yet supported by strong enough evidence, and there is a high risk that safe drugs would be identified and terminated for no good reason.

Other Methods Cell Cultures The use of various techniques involving liver cell cultures for IDILI prediction during early phases of drug development is highly controversial for several reasons: Although it is well accepted that drug metabolism can play an important initiating step in the development of DILI, numerous mechanistic studies have emphasized the fact that DILI may be a multicellular event.49 Whether the liver sustains mild or severe injury likely depends on the activation of and the balance between protective and deleterious intracellular and extracellular signaling processes between the hepatocytes and nonparenchymal cells, particularly Kupffer, natural killer (NK/NKT) cells, and neutrophils.49,50 These findings place a severe limitation on the use of in-vitro systems based on hepatocytes for DILI risk assessment.41 In addition, it is well accepted that IDILI is highly dependent on host susceptibility, and typically manifests in a small minority of hosts. The likelihood of a randomly chosen hepatocyte culture to exhibit these susceptibilities is extremely low. Studies assessing the utility of cell cultures techniques for DILI prediction have shown conflicting and often confusing results.30,51 In several cases, sophisticated testing based on cell cultures have yielded negative results for drugs with known hepatotoxic effects while exhibiting positive results in drugs that are very safe to the human liver. Investigators from the maker of ximelagatran30 (which was withdrawn from the market due to hepatotoxicity) examined the effect of this drug on various cellular components including hepatocyte cytoskeleton and mitochondrial function using an automated imaging technique. They found that ximelagatran did not cause a significant decrease in cell viability even when incubated for 24 hours at considerably higher concentrations than are found in plasma following therapeutic dosing.30 In Seminars in Liver Disease

Vol. 34

No. 2/2014

Although the use of other in vitro tools, such as human liver microsomes, liver slices, and recombinant enzymes aid in the assessment of human drug metabolism, these alone cannot predict the risk of DILI in humans.39 The practical value and utility of approaches such as in silico models,52 chemoinformatics, structural alert systems, and toxicogenomics53,54 for the prediction of DILI risk during drug development are still highly controversial. Over the past decade, attempts have been made to compile hepatotoxicity data and develop in silico models to be used as a first-line screening of drug candidates. Although some of these models were considered helpful in improving drug makers understanding of mechanisms of dose-dependent hepatotoxicity in animals, they have not yet been able to significantly enhance the pharmaceutical industry’s ability to predict or assess the risk of IDILI, mostly because of lack of specific and sensitive biomarkers, scarcity of toxicological data, and lack of concrete unequivocal understanding of underlying mechanisms of most known hepatotoxic drugs.52 In a recent review, Przybylak and Cronin52 explored the current status of existing in silico models designed to predict hepatotoxicity. They concluded that most of the predictive methods discussed in their review were based on the structural properties of chemicals and did not take into account genetic and environmental factors. As a result, the authors felt that the predictions of these models were still uncertain. They suggested that to improve the predictability of in silico models for DILI, it is essential to better understand its mechanisms as well as to develop sensitive toxicogenomics biomarkers. Although of some interest for drug manufacturers, in silico models are not yet in common use for routine DILI screening, and most drug makers are still waiting for further evidence of their utility. Furthermore, the FDA and

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

232

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Regev

233

100

ALTx ULN ASTx ULN ALPx ULN TBLx ULN

Value x Upper Limit Normal

start drug 10

stop drug

1

0.1 0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

40

42

44

46

48

Months Since Exposed to Drug Fig. 2 Time course of liver tests in an “adaptation,” during exposure to a new drug. Reversible increase in ALT and AST starting on month 2 of drug treatment and returning to baseline 2 to 3 months later, despite continuation of the drug in a clinical trial patient. In the absence of another cause of liver injury or accompanying changes in TBL, these changes may be consistent with an adaptation. ALT, alanine aminotransferase; AST, aspartate aminotransferase, ALP, alkaline phosphatase, TBL, total bilirubin.

European regulatory agencies currently preclude basing regulatory decisions on toxicogenomics data alone. Although some regulators encourage voluntary submission of quality toxocogenomic data in parallel with traditional toxicologic test results, it is currently unlikely that these data will have a significant role in regulatory decisions pertaining to DILI.55 The use of toxicogenomic approaches to improve assessment of the risk of DILI and DILI prediction has been published by several groups.53,54 In some cases, toxicogenomic techniques were felt to be informative and relevant to the mechanisms of DILI.54 However, in most cases the predictive powers of toxicogenomics for IDILI have not been supported by strong reliable clinical evidence. In some instances, the use of historical hepatotoxicity data has been limited by relying on drug labels as a gold standard of their hepatotoxic properties.53 The use of a hybrid approach of quantitative structure activity relationship (QSAR) modeling combined with toxicogenomic profiles has also been reported, although it was not found to be more accurate than models based on toxicogenomic data alone.54

DILI Prediction during Clinical Phases of Drug Development The basic approach of drug makers towards DILI prediction during clinical phases has evolved considerably. One of the major advancements has been in our understanding of the predictive significance of common changes in basic hepatic biomarkers. This concept has been formed by DILI experts from the US FDA and academia, and was documented in the FDA’s guidance to industry which was finalized in July 2009.3 Based on this concept, individuals who are exposed to a new drug or chemical may generally be divided into three main groups according to the pattern of their hepatic response.56

Tolerators In most cases of exposure to a new drug, the patient shows no significant changes in hepatic biochemical tests. The ALT and

AST levels remain within the patient’s baseline levels or close to them, and the patient remains asymptomatic. This group, which typically makes up the majority of the patients, has been called “tolerators.” Typically, this group makes the majority of the population and the majority of clinical trial subjects during drug development. These subjects will tolerate the drug well throughout the treatment and will typically not develop DILI related to this specific drug.

Adapters A smaller but significant group may exhibit transient increases in ALT and AST, which eventually return to baseline levels despite continuation of the drug (►Fig. 2). These ALT and AST increases have been a source of confusion and debate among drug makers and regulatory authorities. However, during the last three decades it has become increasingly apparent that transient ALT and AST elevation occurs relatively commonly during the first few months of exposure to a new drug and, in most cases, constitutes a poor predictor of the drug’s potential to cause DILI. This phenomenon has been called “adaptation,” and patient exhibiting these changes have been dubbed “adapters.”56 Although adaptation has been described with some drugs (e.g., isoniazid) even after significant liver injury had developed56–58 in its clean form it typically involves transient elevations in ALT and AST, which are usually asymptomatic and not accompanied by signs of decreased liver function (such as jaundice, elevated direct bilirubin, prolonged prothrombin time (PT), or increased international normalized ratio (INR)). The mechanism (or mechanisms) underlying adaptation are still poorly understood. It may represent true mild liver injury with spontaneous resolution59 or may have another underlying mechanism, but it does not represent a clinically important liver injury. Changes in aminotransferases consistent with adaptation occur with numerous drugs, many of which are rarely if ever associated with clinically significant DILI, such as statins,60 aspirin,3 heparin,61–63 tacrine,48 and others.3 A recent study in healthy volunteers demonstrated Seminars in Liver Disease

Vol. 34

No. 2/2014

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

-2

Drug-Induced Liver Injury and Drug Development: Industry Perspective that at least in the case of heparin, adaptation may involve some degree of liver injury.60 According to this study elevated ALT and AST levels in individuals exposed to heparin were accompanied by changes in other biomarkers, such as sorbitol dehydrogenase (SDH), glutamate dehydrogenase (GLDH), and micro RNA-122 (miR-122), indicating high likelihood of true liver injury59; however, these patients invariably adapted to the drug and did not develop any signs or symptoms of clinically significant hepatic injury. Based on growing experience with “adapters,” it looks like they typically do not experience DILI due to the specific drug to which they adapted. The frequency and magnitude of ALT and AST changes occurring during adaptation vary between individual drugs. Higher frequency of aminotransferase elevation, although deserving of full attention during drug development, does not rule-out the possibility of adaptation. Of the patients receiving heparin,
15% would develop ALT elevations to more than 3 ULN; however, there have been no reliable reports of serious DILI in association with this drug. In patients receiving tacrine,
25% exhibited ALT elevation to more than 3 ULN, while
2% exhibited elevation to more than 20 ULN. Nevertheless, there are rare, if any, unequivocal reports of patients developing liver failure due to tacrine.48,63 Drug makers still vary in their approaches toward ALT and AST elevations, occurring during clinical trials, but owing to increasingly consistent message from FDA3 and several other regulatory agencies,34 most have accepted the concept of adaptation and its implications.

Susceptibles The third group of patients plays a leading role in identifying adrug’s potential to cause severe DILI. After receiving a new drug these patients will typically experience a progressive increase in aminotransferases that will continue increasing while on the drug, and will subsequently evolve into clinically significant liver injury. This injury may be accompanied by symptoms such as fatigue, nausea, vomiting, right upper quadrant pain or tenderness, and in some cases fever, rash, or eosinophilia. Occasionally, they may exhibit signs of decreased liver function such as jaundice, elevated direct bilirubin levels, or prolonged PT, and some will progress to overt liver failure. This group has been called “susceptibles,” and is the main focus of endeavors to identify and predict IDILI during drug development. For some drugs, only one or two of the three groups (tolerators, adaptors, susceptibles) may exist. For example, only tolerators and adapters have been well described for heparin and tacrine.48,62,63 Conversely, for many drugs, all three groups may be encountered, and the difference between safe drugs and those that are considered less safe to the liver may depend on the relative frequency of tolerators, adapters, and susceptibles. For example, for statins, tolerators and adaptors are, by far, the main groups. Susceptibles may exist, but are extremely rare, making these drugs very safe from the hepatic standpoint.60 On the other hand, isoniazid is considered “hepatotoxic” because it causes clinically significant DILI, with jaundice, in
0.5% to 1% of the patients and severe DILI, with acute liver failure, in
1 in 1,000.15 Therefore, in many cases, drugs that are considered “safe to the Seminars in Liver Disease

Vol. 34

No. 2/2014

Regev

liver” and those that are considered “hepatotoxic drugs” are actually on a continuum, where the former are associated with rare cases of susceptibles with severe DILI (the safer drugs are associated with DILI in less than 1 per a million cases) and the latter are associated with higher frequency of susceptibles with severe DILI. Because benign reversible ALT and AST elevation occurs relatively frequently during the first months of exposure to a new drug it is critical for the drug maker to differentiate between self-limiting aminotransferase elevations that represent adaptation and significant hepatic injury, which may indicate a potential of the drug to cause severe liver injury. The following are considered warning signs that may suggest that the abnormality in aminotransferases is associated with clinically significant hepatic injury3: 1. Symptoms suggestive of liver injury such as a. Right upper abdominal pain, right upper abdominal discomfort or tenderness b. Worsening fatigue or malaise c. Jaundice d. Nausea or vomiting 2. Symptoms or laboratory changes suggestive of hypersensitivity-type reaction such as a. Rash b. Fever c. Lymphadenopathy d. Eosinophilia 3. Signs or laboratory changes suggestive of decreased liver function or liver failure, such as a. A significant increase in direct bilirubin level, especially when not accompanied by cholestasis (i.e., substantially elevated ALP levels) b. Development of ascites c. Hepatic encephalopathy d. Increased prothrombin time or international normalized ratio The appearance of any one of these signs or symptoms in a clinical trial subject with ALT elevation to more than 3 ULN, is considered a significant signal by most drug makers, and would typically lead to a discussion on whether or not the study drug should be discontinued. An important part of this discussion should focus on the likelihood that the study drug is the cause of the observed manifestations (i.e., thorough causality assessment). This discussion should optimally take place before the decision on discontinuation of the drug has been finalized; however, this is not always possible, and immediate discontinuation may occasionally be the preferred option. Some drug makers would bring into this discussion internal or external clinicians with expertise in hepatology and experience in DILI, to assess the severity of the liver injury and relatedness to the study drug. It should be noted that although ALT is a very sensitive marker for hepatocellular injury, for other types of DILI, such as drug-induced cholestasis (e.g., amoxicillin clavulanate),64 mitochondrial toxicity (e.g., fialuridine),65 or hepatic steatosis/ fibrosis (e.g., methotrexate),66 ALT is a much less sensitive marker than for hepatocellular DILI.

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

234

Employing Hy’s Law during Drug Development Hy’s law is currently the most specific tool available to the pharmaceutical industry for assessing a drug’s potential to cause severe DILI. Most drug makers recognize today that the finding of a higher rate of ALT elevations in drug-treated subjects than in a control group is a poor predictor of the drug’s potential to cause severe DILI. The ALT elevation is therefore a necessary, but not sufficient, signal.48 Very high levels of ALT (e.g., >10 ULN or >15 ULN) may be a somewhat better indicator, but still carry less than optimal specificity for DILI prediction.3 Therefore, the most specific indicator for a drug’spotential to cause severe DILI is occurrence of cases of drug inducedhepatocellular injury (ALT elevation) accompanied byevidence of altered liver function, such as increased serum TBL (predominantly direct), prolonged PT, or increased INR. The combination of elevated ALT equal to or higher than 3 the ULN and elevated serum TBL equal to higher than 2 the ULN, not explained by any other cause, in the absence of evidence of cholestasis has been dubbed Hy’s law after Hymann Zimmerman,67 and carries an alarming prognosis, with over 10% mortality from acute liver failure without liver transplantation.10 Drugs that cause such injury during drug development raise serious concerns about liver safety in the postmarketing phase, when a much larger number of patients are exposed to the drug.68 Furthermore, Hy’s law allows for a rough estimation of the number of serious liver injury cases likely to occur postmarketing, which will be roughly onetenth the rate of Hy’s law cases observed during clinical development.3 In the last decade, Hy’s law has gained support from two population-based studies in Europe and one by the US Drug-Induced Liver Injury Network (DILIN). In one study, Bjornsson and Olssone69 found a mortality or transplantation rate of 9.2% in patients with hepatocellular DILI who had concomitant TBL level greater than or equal to 2 the ULN. Similarly, Andrade et al70 found a mortality rate of 11.7% among patients with suspected DILI and concomitant jaundice.70 Both studies offer support to the notion that hepatocellular DILI with jaundice or hyperbilirubinemia is a serious entity, with a potentially severe outcome. To define a clinical trial subject as a Hy’s law case the following four components need to be present.3,67 1. ALT elevation to a level equal to or higher than 3 ULN 2. TBL elevation to a level equal to or higher than 2  ULN 3. No significant elevation of ALP (initial ALP value does not reach 2 ULN). 4. No other cause is found to explain the combination of increased ALT and TBL, such as viral hepatitis A, B, C, or E; autoimmune hepatitis; acute gallstone disease, alcoholic liver disease; congestive heart failure; or another drug that is known to cause the observed injury. Notably, the Hy’s law criteria outlined abovepertain to patients who had normal or near normal hepatic biochemicaltests at baseline. Because one of the criteria for Hy’s law is absence of an alternative cause for liver injury, there are

Regev

currently no universally accepted criteria equivalent to Hy’s law for patients who had significantly abnormal liver tests at baseline. As a result, drug makers still struggle with the definition of Hy’s law in the growing population of patients who suffer from liver diseases before entering the study (patients with fatty liver disease, chronic viral hepatitis, oncology patients with nonalcoholic hepatic involvement, patients with alcoholic liver disease, and others.) Hy’s law cases typically appear on a background of severalcases of ALT elevations to more than 3 ULN (seediscussion on eDISH below). Generally, a drug that causeshepatocellular DILI is likely to exhibit a higher incidence of > 3 ALT elevations above the ULN than placebo or a comparator, assuming that the comparator does not itself cause ALT elevations. A case, that meets the four Hy’s law criteria, but is not accompanied by increased incidence of ALT elevation, should be viewed with suspicion, as it is unlikely to be drug induced. Hy’s law cases were observed in clinical trials with dilevalol, tasosartan, ximelagatran, and troglitazone, which were either abandoned or withdrawn from the market due to DILI.3,26,29

Current Approaches to Monitoring, Organization, and Analysis of Hepatic Safety Data Due to increasing awareness and understanding of current approaches for DILI prediction, most drug makers have made considerable changes in the processes of monitoring and analysis of hepatic safety data in the last decade. Most companies use hepatic monitoring algorithms based on the FDA’s guidance3 to facilitate close observation and assessment of patients who cross the thresholds of ALT or AST of 3 ULN, TBL of 2 ULN or ALP of 2 ULN during a clinical trial. The trial investigators are often instructed to contact the clinical trial physician when hepatic biochemical tests cross these thresholds (thresholds may change according to baseline values.) The use of Data Monitoring Committees (DMCs) also called Drug Safety Monitoring Boards (DSMBs) is almost universal, based on the FDA’s guidance for clinical trial sponsors on establishment and operation of clinical trial data monitoring committees (March 2006).71 The importance and function of these committees has been discussed extensively elsewhere.72 They are commonly composed of biostatisticians, scientists, bioethicists, and clinicians who are knowledgeable about the disease state being studied. Their principal role is to ensure the safety of patients during the clinical trial. Many drug makers also use internal or external hepatic assessment committees for adjudication of clinically significant hepatic cases during clinical trials.6

Evaluation of Drug-Induced Serious Hepatotoxicity (eDISH) Because the most reliable and specific biomarker for DILI prediction is the concomitant elevation of ALT and bilirubin levels, there is a great advantage to an efficient and Seminars in Liver Disease

Vol. 34

No. 2/2014

235

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Drug-Induced Liver Injury and Drug Development: Industry Perspective

100.0 hyperbilirubinemia

Hy's Law range Control Drug X

10.0

2x

1.0

normal range 3x

0.1 0.1

1.0

Temple's Corollary range

10.0

100.0

Peak ALT, xULRR

Fig. 3 eDISH, Evaluation of Drug-Induced Serious Hepatotoxicity) in a phase 3 clinical trial. Overrepresentation of points on study drug in the Hy’s law range quadrant and the Temple’s corollary range quadrant suggests a possible increased risk of severe DILI. ALT, alanine aminotransferase; TBL, total bilirubin; ULRR, upper limit of reference range.

comprehensive approach to monitoring and assessment of concomitant changes in ALT and TBL levels during a clinical trial. Modern phase III clinical trials may have many thousands of data points related to liver tests. Such data are often difficult to visualize and assess. The potential difficulty is increased by the need for an in-depth case-by-case analysis of patients with treatment emergent hepatic abnormalities. A simple and efficient method to visualize, assess and summarize ALT and TBL levels in a clinical trial is the recently introduced evaluation of Drug-Induced Serious Hepatotoxicity (eDISH), which was developed by Guo and Senior at the FDA to allow medical reviewers to display the peak serum ALT and TBL values for every clinical trial subject (►Fig. 3).73 eDISH, allows each point in the display to be directly linked to each subject’s clinical and laboratory data for a full evaluation of the clinical picture and history related to the hepatic abnormality. It is currently being used by FDA medical reviewers, when evaluating liver safety databases from phase III clinical trials, and is gradually being adopted by drug makers in the United States and Europe. It may be useful in presenting liver safety data to FDA-appointed advisory committees.74 On the eDISH plot, the peak serum ALT is shown along the x-axis,and the peak serum TBL is shown along the y-axis as multiples of theULN on log scales (►Fig. 3). Four quadrants are defined by a line corresponding to 3 ULN for serum ALT and a line corresponding to 2 the ULN for TBL. The right upper quadrant includes all subjects with serumALT > 3 ULN and serumTBL > 2 ULN. This quadrant is referredto as “Hy’s law range.” It is important to notethat ALT > 3 ULN and TBL > 2 ULN are only two of the four criteria for the definition of a Hy’s law case. The ability to use the specific point on the eDISH plot as a link to the rest of the subject’s clinical data is helpful in allowing assessment of the other two criteria (initial ALP level < 2 ULN, and no other cause explaining the hepatic injury.) Furthermore, when true cases of Hy’s law are present, there should be a higher incidence of ALT elevations greater than 3 ULN in the drug-treated subjects relative to subjects treated with the comparator, assuming that the comparator does not itself cause ALT elevation (“Temple’s Corolary”). Seminars in Liver Disease

Vol. 34

No. 2/2014

Regev

The right lower quadrant shows subjects who had ALT elevations to more than 3 ULN without elevations of TBL to >2 ULN. This quadrant has been labeled “Temple’s Corollary,” reflecting an observation first made by Dr. Robert Temple of the FDA that an imbalance in this quadrant between those subjects treated with study drug and those treated with placebo (or a nonhepatotoxic comparator) has been reliably present with drugs capable of causing serious liver injury.3 As discussed above, an imbalance of ALT elevation between treatment and comparator does not in itself constitute a clear liver safety signal, as ALT is often elevated in drugs that rarely if ever cause severe liver injury. The left upper quadrant has been labeled“hyperbilirubinemia” or “cholestasis’quadrant. It represents subjects with TBL levels > 2 ULN along with ALT < 3 ULN. In most cases patients in this quadrant would have a benign elevation of TBL due to Gilbert syndrome. An overrepresentation of points on study drug would be expected in this quadrant for drugs that are associated with cholestatic liver injury. The left lower quadrant represents subjects who never showed ALT levels > 3 ULN or TBL levels > 2x the ULN. In summary, eDISH appears to be an effective method to visualize, summarize, and assess hepatic safety data during randomized controlled clinical trials. It may be instrumental in identifying drugs with a potential to cause severe DILI postmarketing.

Applying Hepatic Discontinuation Rules During Drug Development Avoiding injury to clinical trial subjects and patients is a top priority for the drug industry. However, IDILI is typically first encountered in clinical phases of drug development. In most DILI cases, the only effective treatment is discontinuation of the causal agent. Delayed discontinuation may occasionally lead to worsening of the liver injury, which may result in liver failure and death. On the other hand, automatic discontinuation of a trial drug upon finding of a mild abnormality in liver enzymes (e.g., elevation of ALT or AST to > 3 ULN) is not only unnecessary, but may render it impossible to differentiate between drugs that cause benign self-limiting elevations in aminotransferase and those that may cause a clinically significant liver injury.3 The correct timing of discontinuation of an investigational drug when faced with abnormal liver tests has been a matter of debate, and a source of great anxiety to clinical investigators and safety physicians in the pharmaceutical industry. In the absence of systematic and prospective studies addressing this question, drug makers and regulators have suggested somewhat arbitrary hepatic stopping rules. In general, though early discontinuation of the drug may better protect study subject from a rare DILI event, it may lead to failure to identify the drug’s potential to cause severe DILI, and consequently may lead to a potentially increased risk of DILI postmarketing. Clearly, stopping a drug at the first hint of ALT or AST elevation will preclude the drug maker from learning whether adaptation occurs. Furthermore, regulators may justifiably be reluctant to approve marketing of a drug which was

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Peak TBL, xULRR

236

Drug-Induced Liver Injury and Drug Development: Industry Perspective

• • • •

ALT or AST > 8 ULN ALT or AST > 5 ULN for more than 2 weeks ALT or AST > 3 ULN and (TBL > 2xULN or INR > 1.5) ALT or AST > 3 ULN with the appearance of fatigue, nausea, vomiting, right upper quadrant pain or tenderness, fever, rash, and/or eosinophilia (>5%)

The decision as to whether a study drug should bediscontinued, is a complicated one, and generally hinges on theinvestigator, who is the physician closest to the patient. Optimally,this decision should be made after discussion with the clinical trial physician. Several considerations may influence the decision to discontinue a drug. First, when applying these discontinuation rules, it is assumed that the study drug is the cause of abnormal hepatic tests. Discontinuation of the study drug may not be necessary if the abnormality is caused by an intercurrent viral hepatitis, a concomitantly administered drug, or gallstones. Consequently, a timely and comprehensive causality assessment may help prevent an unnecessary early discontinuation of the study drug. Second, when recommending these specific discontinuation rules the FDA assumed that the patient’s baseline hepatic tests levels were either within or close to normal levels. For study subjects who are enrolled with abnormal tests due to an underlying hepatic or systemic disorder, discontinuation rules may need to be adjusted. Third, the decision on whether to discontinue the drug may be influenced by our knowledge of the drug’s probability of effect, and the potential importance of that effect to the patient’s health. Obviously, an investigator will be less inclined to discontinue a drug that has a high likelihood of saving the patient’s life (e. g., an anticancer drug) than a drug which has no proven benefit. Clinical trial investigators, who are not experienced with hepatic disorders or DILI, may tend to discontinue the drug long before the hepatic abnormalities met the FDA’s recommended discontinuation rules. Drug makers often need to utilize intensive guidance and training, and maintain close communication with their clinical trial investigators to prevent unnecessary early discontinuation. The importance of such training and close communication cannot be overemphasized.

Causality Assessment for Suspected DILI during Drug Development Causality assessment is a critical component in the process of understanding the drug’s hepatic safety profile, and predicting its potential to cause severe DILI. Unfortunately, causality assessment for suspected DILI is still a major challenge in clinical practice and even more so during drug development. In contrast to many other liver disorders, there is presently no specific biomarker or a combination of tests that can establish

237

the diagnosis of DILI and differentiate it from other causes of liver injury. Furthermore, DILI may resemble almost any type of liver disease.75 See further discussion on causality assessment for suspected DILI elsewhere in this publication.

Postmarketing Safety Surveillance and Risk Management Despite significant improvement in DILI prediction during clinical studies, there are still significant limitations in our ability to identify and predict IDILI, and it is well recognized that there might still be cases in which IDILI may not be identified during drug development. It is therefore critical that drug makers remain vigilant and continue safety surveillance postapproval. This includes close monitoring of spontaneous adverse events reports as well as postmarketing data mining. It is conceivable that for some drugs the benefit–risk profile may continue to evolve over the postapproval phase of their life cycle.1,9 However, postmarketing safety surveillance is currently limited by the rarity of the cases, quantity and quality of information provided to drug makers and regulators, the lack of a control group, and the lack of a reliable method to confirm relationship to the drug.39 It has been shown that almost half of the cases reported by health care providers as being caused by DILI are unrelated to the suspected drug.76 These limitations may result in a severely under- or overestimated DILI incidence. Methods to identify patients with DILI include patient registries that enroll identified DILI cases, such as the U.S.-based DILI Network.77 In addition, drug makers are increasingly engaged in efforts to identify adverse drug events in large population based cohorts based on networks of health care databases (e.g., through medical billing claims or electronic health records).

Summary and Future Prospects DILI remains one of the major challenges faced by drug makers. Despite ongoing efforts to develop preclinical approaches and biomarkers, which would help in predicting the risk of IDILI in earlier phases of drug development, such approaches and biomarkers are not yet ready for prime time. Nevertheless, clinical trial databases may show evidence of a drug’s potential for severe DILI if clinical and laboratory data are properly collected, organized, and evaluated for evidence of milder forms of liver injury. Data pertaining to milder forms of liver injury occurring during clinical development, when assessed correctly, may enhance our ability to predict the drug’s potential to cause severe liver injury postmarketing. Although elevation of ALT is a sensitive marker for hepatocellular DILI, the specificity of such elevation is low, and it is generally considered a poor predictor of severe hepatotoxicity. On the other hand, the occurrence of Hy’s law cases may be a very significant and specific predictor of a drug’s ability to cause severe DILI postmarketing. To maximize the success of this approach, drug developers are increasingly using meticulous causality assessment in clinical-trial subjects with treatment-emergent abnormal liver Seminars in Liver Disease

Vol. 34

No. 2/2014

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

repeatedly discontinued at low ALT elevations during clinical development. Based on these considerations, the FDA has suggested the following discontinuation rules for drugs in clinical phases of development.3

Regev

Drug-Induced Liver Injury and Drug Development: Industry Perspective tests, and strict hepatic discontinuation rules to prevent unnecessary early discontinuation of the study drug. The use of internal and/or external hepatic assessment committees is becoming more prevalent and is supplementing the work of data monitoring committees. A few potentially useful biomarkers, such as micro RNA,78 are emerging that may be helpful in the future in diagnosing and assessing the severity of DILI during clinical trials. Although not yet qualified for clinical development, some of these biomarkers may facilitate real progress in our ability to understand and predict DILI. Advances in the development of genomic biomarkers may also improve our understanding of DILI mechanisms and our predictive ability. In the last two decades chimeric mice with humanized livers, via transplantation of human hepatocytes, have been generated.40,79 Such models may provide another approach to in vivo evaluation of human genetic polymorphisms, and may improve our understanding of the mechanisms underlying DILI.

16 Walker RM, McElligott TF. Furosemide induced hepatotoxicity. J

Pathol 1981;135(4):301–314 17 Taggart HM, Alderdice JM. Fatal cholestatic jaundice in elderly

18

19

20

21

22

23

24

References 1 Stevens JL, Baker TK. The future of drug safety testing: expanding

2

3

4 5 6 7 8 9

10

11 12

13

14

15

the view and narrowing the focus. Drug Discov Today 2009;14(34):162–167 Watkins PB, Seligman PJ, Pears JS, Avigan MI, Senior JR. Using controlled clinical trials to learn more about acute drug-induced liver injury. Hepatology 2008;48(5):1680–1689 FDA. Guidance for industry. Drug-induced livery injury: premarketing clinical evaluation. Available at: http://www.fda.gov/ downloads/Drugs/GuidanceComplianceRegulatoryInformation/ Guidances/UCM174090.pdf. Accessed December 17, 2013 Navarro VJ, Senior JR. Drug-related hepatotoxicity. N Engl J Med 2006;354(7):731–739 Goodman ZD. Drug hepatotoxicity. Clin Liver Dis 2002;6(2): 381–397 Lewis JH, Ahmed M, Shobassy A, Palese C. Drug-induced liver disease. Curr Opin Gastroenterol 2006;22(3):223–233 Bhamidimarri KR, Schiff E. Drug-induced cholestasis. Clin Liver Dis 2013;17(4):519–531, vii Patel V, Sanyal AJ. Drug-induced steatohepatitis. Clin Liver Dis 2013;17(4):533–546, vii Lewis JH. Drug-induced liver injury throughout the drug development life cycle: where we have been, where we are now, and where we are headed. Perspectives of a clinical hepatologist. Pharm Med 2013;27:165–191 Zimmerman HJ. Drug-induced liver disease. In: Hepatotoxicity, The Adverse Effects of Drugs and Other Chemicals on the Liver. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 1999: 428–433 Zetzel L, Kaplan H. Liver damage concurrent with iproniazid administration. N Engl J Med 1958;258(24):1209–1211 Zimmerman HJ, Rosenblum L, Korn RJ, Feldman PE. Some preliminary clinical and biochemical aspects of liver function in patients receiving iproniazid. Ann N Y Acad Sci 1959;80:915–927 Manier JW, Chang WW, Kirchner JP, Beltaos E. Hepatotoxicity associated with ticrynafen—a uricosuric diuretic. Am J Gastroenterol 1982;77(6):401–404 Zimmerman HJ, Lewis JH, Ishak KG, Maddrey WC. Ticrynafenassociated hepatic injury: analysis of 340 cases. Hepatology 1984; 4(2):315–323 NIDDK. LiverTox, Clinical and research information on druginduced liver injury. Available at: http://livertox.nlm.nih.gov/ Isoniazid.htm. Accessed January 7, 2014

Seminars in Liver Disease

Vol. 34

No. 2/2014

Regev

25 26 27

28

29 30

31

32

33

34

35 36

37

38

patients taking benoxaprofen. Br Med J (Clin Res Ed) 1982; 284(6326):1372 Hamdy RC, Murnane B, Perera N, Woodcock K, Koch IM. The pharmacokinetics of benoxaprofen in elderly subjects. Eur J Rheumatol Inflamm 1982;5(2):69–75 Moses PL, Schroeder B, Alkhatib O, Ferrentino N, Suppan T, Lidofsky SD. Severe hepatotoxicity associated with bromfenac sodium. Am J Gastroenterol 1999;94(5):1393–1396 Hunter EB, Johnston PE, Tanner G, Pinson CW, Awad JA. Bromfenac (Duract)-associated hepatic failure requiring liver transplantation. Am J Gastroenterol 1999;94(8):2299–2301 Rabkin JM, Smith MJ, Orloff SL, Corless CL, Stenzel P, Olyaei AJ. Fatal fulminant hepatitis associated with bromfenac use. Ann Pharmacother 1999;33(9):945–947 Fontana RJ, McCashland TM, Benner KG, et al; The Acute Liver Failure Study Group. Acute liver failure associated with prolonged use of bromfenac leading to liver transplantation. Liver Transpl Surg 1999;5(6):480–484 Gitlin N, Julie NL, Spurr CL, Lim KN, Juarbe HM. Two cases of severe clinical and histologic hepatotoxicity associated with troglitazone. Ann Intern Med 1998;129(1):36–38 Vella A, de Groen PC, Dinneen SF. Fatal hepatotoxicity associated with troglitazone. Ann Intern Med 1998;129(12):1080 Herrine SK, Choudhary C. Severe hepatotoxicity associated with troglitazone. Ann Intern Med 1999;130(2):163–164 Watkins PB, Whitcomb RW. Hepatic dysfunction associated with troglitazone. N Engl J Med 1998;338(13):916–917 Maria VA, Victorino RM. Hypersensitivity immune reaction as a mechanism for dilevalol-associated hepatitis. Ann Pharmacother 1992;26(7-8):924–926 He R. 2004, Clinical Review of Exanta (ximelagatran) Tablets, FDA Cardiovascular and Renal Drugs Advisory Committee Briefing Information. Available at: http://www.fda.gov/ohrms/dockets/ ac/04/briefing/2004-4069B1_01_AstraZeneca-Backgrounder.pdf. Accessed January 7, 2014 Lee WM, Larrey D, Olsson R, et al. Hepatic findings in long-term clinical trials of ximelagatran. Drug Saf 2005;28(4):351–370 Ainscow EK, Pilling JE, Brown NM, et al. Investigations into the liver effects of ximelagatran using high content screening of primary human hepatocyte cultures. Expert Opin Drug Saf 2008;7(4): 351–365 Kenne K, Skanberg I, Glinghammar B, et al. Prediction of druginduced liver injury in humans by using in vitro methods: the case of ximelagatran. Toxicol In Vitro 2008;22(3):730–746 Kramer JA, Sagartz JE, Morris DL. The application of discovery toxicology and pathology towards the design of safer pharmaceutical lead candidates. Nat Rev Drug Discov 2007;6(8):636–649 Olson H, Betton G, Robinson D, et al. Concordance of the toxicity of pharmaceuticals in humans and in animals. Regul Toxicol Pharmacol 2000;32(1):56–67 Release of Guidance Document. Pre-market evaluation of hepatotoxicity in health Products. Available at: http://www.hc-sc.gc.ca/ dhp-mps/alt_formats/pdf/prodpharma/applic-demande/guideld/hepatotox_guide_ld-eng.pdf Accessed January 10, 2014 Uetrecht J. Idiosyncratic drug reactions: past, present, and future. Chem Res Toxicol 2008;21(1):84–92 Horsmans Y, Desager JP, Harvengt C. Biochemical changes and morphological alterations of the liver in guinea-pigs after administration of simvastatin (HMG CoA reductase-inhibitor). Pharmacol Toxicol 1990;67(4):336–339 Kornbrust DJ, MacDonald JS, Peter CP, et al. Toxicity of the HMGcoenzyme A reductase inhibitor, lovastatin, to rabbits. J Pharmacol Exp Ther 1989;248(2):498–505 Roth RA, Ganey PE. Animal models of idiosyncratic drug-induced liver injury—current status. Crit Rev Toxicol 2011;41(9):723–739

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

238

Regev

39 Corsini A, Ganey P, Ju C, et al. Current challenges and controversies

60 Cohen DE, Anania FA, Chalasani N; National Lipid Association

in drug-induced liver injury. Drug Saf 2012;35(12):1099–1117 Katoh M, Matsui T, Nakajima M, et al. Expression of human cytochromes P450 in chimeric mice with humanized liver. Drug Metab Dispos 2004;32(12):1402–1410 Antoine DJ, Williams DP, Park BK. Understanding the role of reactive metabolites in drug-induced hepatotoxicity: state of the science. Expert Opin Drug Metab Toxicol 2008;4(11):1415–1427 Kalgutkar AS, Gardner I, Obach RS, et al. A comprehensive listing of bioactivation pathways of organic functional groups. Curr Drug Metab 2005;6(3):161–225 Williams DP. Toxicophores: investigations in drug safety. Toxicology 2006;226(1):1–11 Iwanaga T, Nakakariya M, Yabuuchi H, Maeda T, Tamai I. Involvement of bile salt export pump in flutamide-induced cholestatic hepatitis. Biol Pharm Bull 2007;30(4):739–744 Usui T, Mise M, Hashizume T, Yabuki M, Komuro S. Evaluation of the potential for drug-induced liver injury based on in vitro covalent binding to human liver proteins. Drug Metab Dispos 2009;37(12):2383–2392 Obach RS, Kalgutkar AS, Soglia JR, Zhao SX. Can in vitro metabolism-dependent covalent binding data in liver microsomes distinguish hepatotoxic from nonhepatotoxic drugs? An analysis of 18 drugs with consideration of intrinsic clearance and daily dose. Chem Res Toxicol 2008;21(9):1814–1822 Park BK, Boobis A, Clarke S, et al. Managing the challenge of chemically reactive metabolites in drug development. Nat Rev Drug Discov 2011;10(4):292–306 Watkins PB, Zimmerman HJ, Knapp MJ, Gracon SI, Lewis KW. Hepatotoxic effects of tacrine administration in patients with Alzheimer’s disease. JAMA 1994;271(13):992–998 Liu ZX, Govindarajan S, Kaplowitz N. Innate immune system plays a critical role in determining the progression and severity of acetaminophen hepatotoxicity. Gastroenterology 2004;127(6): 1760–1774 Liu ZX, Han D, Gunawan B, Kaplowitz N. Neutrophil depletion protects against murine acetaminophen hepatotoxicity. Hepatology 2006;43(6):1220–1230 Persson M, Løye AF, Mow T, Hornberg JJ. A high content screening assay to predict human drug-induced liver injury during drug discovery. J Pharmacol Toxicol Methods 2013;68(3):302–313 Przybylak KR, Cronin MT. In silico models for drug-induced liver injury—current status. Expert Opin Drug Metab Toxicol 2012;8(2): 201–217 Zhang M, Chen M, Tong W. Is toxicogenomics a more reliable and sensitive biomarker than conventional indicators from rats to predict drug-induced liver injury in humans? Chem Res Toxicol 2012;25(1):122–129 Low Y, Uehara T, Minowa Y, et al. Predicting drug-induced hepatotoxicity using QSAR and toxicogenomics approaches. Chem Res Toxicol 2011;24(8):1251–1262 Corvi R, Ahr HJ, Albertini S, et al. Meeting report: Validation of toxicogenomics-based test systems: ECVAM-ICCVAM/NICEATM considerations for regulatory use. Environ Health Perspect 2006; 114(3):420–429 Senior John. . Lessons from isoniazid. Would it be approved today? Available at: www.fda.gov/downloads/Drugs/ScienceResearch/ .../ucm076755.pdf. Accessed January 7, 2014 Mitchell JR, Long MW, Thorgeirsson UP, Jollow DJ. Acetylation rates and monthly liver function tests during one year of isoniazid preventive therapy. Chest 1975;68(2):181–190 Nolan CM, Goldberg SV, Buskin SE. Hepatotoxicity associated with isoniazid preventive therapy: a 7-year survey from a public health tuberculosis clinic. JAMA 1999;281(11):1014–1018 Harrill AH, Roach J, Fier I, et al. The effects of heparins on the liver: application of mechanistic serum biomarkers in a randomized study in healthy volunteers. Clin Pharmacol Ther 2012;92(2): 214–220

Statin Safety Task Force Liver Expert Panel. An assessment of statin safety by hepatologists. Am J Cardiol 2006;97(8A):77C–81C Olsson R, Korsan-Bengtsen BM, Korsan-Bengtsen K, Lennartsson J, Waldenström J. Serum aminotransferases after low-dose heparin treatment. Short communication. Acta Med Scand 1978;204(3): 229–230 Dukes GE Jr, Sanders SW, Russo J Jr, et al. Transaminase elevations in patients receiving bovine or porcine heparin. Ann Intern Med 1984;100(5):646–650 Carlson MK, Gleason PP, Sen S. Elevation of hepatic transaminases after enoxaparin use: case report and review of unfractionated and low-molecular-weight heparin-induced hepatotoxicity. Pharmacotherapy 2001;21(1):108–113 Lucena MI, Molokhia M, Shen Y, et al; Spanish DILI Registry; EUDRAGENE; DILIN; DILIGEN; International SAEC. Susceptibility to amoxicillin-clavulanate-induced liver injury is influenced by multiple HLA class I and II alleles. Gastroenterology 2011;141(1): 338–347 Kleiner DE, Gaffey MJ, Sallie R, et al. Histopathologic changes associated with fialuridine hepatotoxicity. Mod Pathol 1997; 10(3):192–199 Berends MAM, Snoek J, de Jong EM, et al. Liver injury in long-term methotrexate treatment in psoriasis is relatively infrequent. Aliment Pharmacol Ther 2006;24(5):805–811 Temple R. Hy’s law: predicting serious hepatotoxicity. Pharmacoepidemiol Drug Saf 2006;15(4):241–243 Senior JR. Regulatory perspectives. In: Kaplowitz N, DeLeve LD, eds. Drug-Induced Liver Disease. New York, NY: Marcel Dekker; 2003:739–754 Björnsson E, Olsson R. Outcome and prognostic markers in severe drug-induced liver disease. Hepatology 2005;42(2): 481–489 Andrade RJ, Lucena MI, Fernández MC, et al; Spanish Group for the Study of Drug-Induced Liver Disease. Drug-induced liver injury: an analysis of 461 incidences submitted to the Spanish registry over a 10-year period. Gastroenterology 2005;129(2):512–521 Food and Drug Administration. Guidance for clinical trial sponsors on the establishment and operation of clinical trial data monitoring committees. Available at: http://www.fda.gov/downloads/ Regulatoryinformation/Guidances/ucm127073.pdf. Accessed January 7, 2014 Slutsky AS, Lavery JV. Data safety and monitoring boards. N Engl J Med 2004;350(11):1143–1147 Guo T, Gelperin K, Senior J. A tool to help you decide (detect potentially serious liver injury). . Available at: http://www.fda.gov/ downloads/Drugs/ScienceResearch/ResearchAreas/ucm076777. pdf. Accessed December 17, 2009 Watkins PB, Desai M, Berkowitz SD, et al. Evaluation of druginduced serious hepatotoxicity (eDISH): application of this data organization approach to phase III clinical trials of rivaroxaban after total hip or knee replacement surgery. Drug Saf 2011;34(3): 243–252 Chitturi S, Farrell GC. Drug induced liver disease. In: Schiff ER et al, eds. Schiff’s diseases of the liver. 11th ed. Hoboken, NJ: WileyBlackwell; 2012:703–782 Shapiro MA, Lewis JH. Causality assessment of drug-induced hepatotoxicity: promises and pitfalls. Clin Liver Dis 2007;11(3): 477–505, v Fontana RJ, Watkins PB, Bonkovsky HL, et al; DILIN Study Group. Drug-Induced Liver Injury Network (DILIN) prospective study: rationale, design and conduct. Drug Saf 2009;32(1):55–68 Starkey Lewis PJ, Dear J, Platt V, et al. Circulating microRNAs as potential markers of human drug-induced liver injury. Hepatology 2011;54(5):1767–1776 Locke JE, Sun Z, Warren DS, et al. Generation of humanized animal livers using embryoid body-derived stem cell transplant. Ann Surg 2008;248(3):487–493

40

41

42

43 44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

61

62

63

64

65

66

67 68

69

70

71

72 73

74

75

76

77

78

79

Seminars in Liver Disease

Vol. 34

No. 2/2014

239

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Drug-Induced Liver Injury and Drug Development: Industry Perspective

Copyright of Seminars in Liver Disease is the property of Thieme Medical Publishing Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.