REVIEW pubs.acs.org/crt
Multiple Factors Govern the Association between Pharmacology and Toxicity in a Class of Drugs: Toward a Unification of Class Effect Terminology Dennis A. Smith, Anthony Harrison, and Paul Morgan* Department of Pharmacokinetics, Dynamics and Metabolism, Pfizer Worldwide Research and Development, Sandwich, Kent, CT13 9NJ, United Kingdom ABSTRACT: The term class effect has gained in use to describe a side effect including toxicity common to a series of drugs. There is no definition of what constitutes a class effect, and it is not applied against a rigid set of criteria.Thus, the finding of toxicity in one of a series of drugs can raise the concern of a class effect, especially if one or more of the others shows findings even slightly related or at very much lower incidence. This is particularly problematic when the term is used loosely or speculatively on initial events that are themselves of low incidence and serious. This speculation exaggerates and distorts the scientific process in establishing the true benefit risk of the individual drugs and can lead to lengthy development times, or highly restrictive labeling, to the detriment of patient welfare. To provide better definition and application of the term, we suggest that the term class effect toxicity is only used when a clear mechanistic link has been established between a safety concern and drug class based on (I) where the primary pharmacology delivers a clear rationale for the observed findings and toxicities; and (II) where the secondary pharmacology is obligate to the class of the molecule and not subject to variation of structure, and the selectivity cannot be impacted significantly by variations in potency introduced by structural manipulation. With these categorizations, we believe class effect toxicity will be mainly confined to I with examples such as the tetracycline class of antibacterials which inhibit protein synthesis both as a mechanism of antibacterial activity and to produce hepatic injury by mitochondrial injury in the liver.
’ CONTENTS 1. 2. 3. 4. 5.
Introduction Class Effect and the Scientific Literature Class Effects and Public Perception Toward a Unification of Class Effect Terminology Class Effects Associated with Primary Pharmacology 5.1. Obligatory Mechanism Liabilities 5.2. Hyper-Response in Individuals to Primary Pharmacology in Target Tissue 5.3. Hyper-Response in Individuals to Primary Pharmacology in Nontarget Tissue 5.4. Eventual Resolution of Primary Pharmacology Effects 6. Class Effects Associated with Secondary Pharmacology 6.1. Inhibition of Related Receptors 6.2. Inhibition of Related Enzymes 6.3. Obligatory Structural Liabilities 7. Impact of Variations in Primary Pharmacology Potency on Selectivity r 2011 American Chemical Society
8. Impact of Structural Variation on Toxicity 9. Dilemma of the Liver and Plausible Relationship to Class Effects: A Case Study with 5-Lipoxygenase (5LO) Inhibitors 10. Hepatobiliary Transport Proteins As a Key Determinant in Liver Effects: Important Secondary Pharmacological Effect of Drugs 11. Conclusions Author Information Abbreviations References
463 464 464 465 465 465 465 465 466
467
468
470 471 472 472 472
1. INTRODUCTION Drugs are often categorized by class. The term is normally reserved for compounds acting at the same target(s), for instance, β-adrenoceptor antagonists (β-blockers), although it can be used
466 466 466 466
Received: November 26, 2010 Published: March 10, 2011
467 463
dx.doi.org/10.1021/tx100408v | Chem. Res. Toxicol. 2011, 24, 463–474
Chemical Research in Toxicology
REVIEW
to describe a similar pharmacological outcome across a broad range of targets. The term class effect is used to describe a pharmacological effect including toxicity of a series of drugs. Difficulties arise in the use and interpretation of the term because its use is not applied against a rigid set of criteria. When a class of drugs gives rise to a specific unusual side effect that is not seen to any large degree in the untreated population, the term class effect can be applied with some confidence. Where the untoward side effect is also seen relatively frequently in the general population, then there is great difficulty and confusion in applying the term. Thus, the finding of toxicity in one of a series of drugs can raise the concern of a class effect, especially if one or more of the others shows findings even slightly related or at very much lower incidence. This is particularly problematic when the term is used loosely or speculatively on initial events that are themselves of low incidence and serious. This speculation exaggerates and distorts the scientific process in establishing the true benefit risk of the individual drugs and can lead to lengthy development times, or highly restrictive labeling, to the detriment of patient welfare. When the side effect is triggered by occupation of the primary target and this is accessed by passive diffusion into extracellular water, e.g., ion channels, G-coupled protein receptors (GCPR), etc., then it is likely that all drugs in a class will show similar side effect profiles. Where the receptor is present in a discrete biophase such as the brain or liver and both passive and active processes differ, it is possible to mitigate the effects even though the primary pharmacology is unchanged. Considerable concern exists relating to hepatotoxicity. During the course of drug development and marketing, incidences of raised liver function tests (LFTs) and even frank hepatocellular injury will occur. These may or may not be drug related. Only when considerable data is accumulated, in most cases, can true correlations be made with the use of the drug.
antagonists do not meet those criteria, each compound’s safety against low incidence severe toxicities can only be evaluated after careful study of data from very large patient numbers. Speculation about class effects are not confined to unrefereed web reports. This is exemplified by articles such as “Hepatotoxicity with Thiazolidinediones: Is it a Class Effect?” appearing in refereed and established journals.2 Thiazolidinediones, more commonly termed glitazones, were the first drugs to specifically target muscular insulin resistance, reducing plasma glucose levels in patients with type 2 diabetes mellitus. When this report was authored, troglitazone, the first compound approved by the Food and Drug Administration in the US, proved to be hepatotoxic and was withdrawn from the market after the report of several dozen deaths or cases of severe hepatic failure requiring liver transplantation. Rosiglitazone and pioglitazone were developed to follow troglitazone. In contrast, at the time of the report the incidence of significant (3 times ULN) increases in liver enzyme levels was similar for rosiglitazone or pioglitazone and placebo. Comparative data demonstrated that troglitazone was associated with a 3-fold greater incidence of 3 times ULN compared to placebo. Moreover, in contrast to the numerous case reports of acute liver failure in patients receiving troglitzone, at the time of the paper few case reports of hepatotoxicity had been reported in patients treated with rosiglitazone, with the causal relationship remaining uncertain. Furthermore, no single case of severe hepatotoxicity had been reported with pioglitazone. Notwithstanding this data, the published title begs a question suggesting that class effects should be suspected (and all drugs in the class treated with suspicion and maybe avoided by prescribers and patients), perhaps without thought for mechanism and dose.
3. CLASS EFFECTS AND PUBLIC PERCEPTION There has been a long-standing perception in the general public that the consumption of alcohol while taking antibiotics should be avoided due to unpleasant side effects and/or attenuation of antibiotic efficacy.3 There have been a number of proposals for the origin of this myth, which include (i) prohibition of alcohol consumption in patients being treated with antibiotics for sexually transmitted diseases, (ii) interference of alcohol in penicillin recovery from urine of patients in the second world war, and (iii) consumption of alcohol while taking the antiparasitic disulfiram resulting in unpleasant side effects due to the inhibiton of alcohol detoxication.4 The perceived link between alcohol and antibiotics was probably more fimly established with metronidazole (Flagyl). and its close structural analogue tinidazole (Tindamax), which have very similar side effects following alcohol consumption as was observed with disulfiram, resulting in labeling statements that alcohol should be avoided while taking these nitro-imidazole antibiotics.5,6 The United Kingdom National Health Service only specifies metronidazole and tinidazole as antibiotics for which it is necessary to completely avoid alcohol consumption due to a specific interaction.7 There is a link between alcohol consumption and side effects of the oxazolidinone antibiotic linezolid (Zyvox), but this is specific to tyraminecontaining alcoholic beverages (e.g., red wine) and a mechanistic link to the drug’s properties as a monoamine oxidase inhibitor.8 While a mechanistic link with alcohol has been proposed for these 3 antibiotics,8,9 there is little evidence for a general alcohol interaction with antibiotics as a class of drugs. Over time, these separate observations, where the coincidence of alcohol and antibiotic therapy has had undesirable effects, have
2. CLASS EFFECT AND THE SCIENTIFIC LITERATURE There is an increasing trend to discuss the evidence of hepatotoxicty in one or more drugs in a class as possible evidence of a class effect. An example of this is discussions at the 10th European AIDS Conference (EACS) held in 2005, Dublin. Discussion of the decision to halt aplaviroc, a new CCR5 antagonist, raised concerns of a class effect that included other CCR5 antagonists such as maraviroc. Four subjects in the course of development trials with aplaviroc had raised serum alanine transaminase (ALT, up to 70 times the upper limit of normal (ULN)) and bilirubin (up to 5 times ULN) after taking aplaviroc. In developing maraviroc, Pfizer had recently found a single case of hepatotoxicity (following over a thousand treated subjects at this time). The patient was also simultaneously receiving several drugs (isoniazid, trimethoprin-sulfamethoxazole, and high dose acetaminophen) which had reported hepatic toxicities making any causality unclear. A summary of this meeting1 states: “will hepatotoxicity prove to be a CCR5 antagonist class effect? The optimist would say toxic class effects are the exception rather than the rule. The pessimist would say don’t rule anything out until you have lots of data from lots of people”. This statement immediately distorts the reality of assessing drug safety by the mention of class effects and optimizts and pessimists. As we will attempt to demonstrate in this review, severe drug toxicities only occur as class effects under strict criteria. Even though CCR5 464
dx.doi.org/10.1021/tx100408v |Chem. Res. Toxicol. 2011, 24, 463–474
Chemical Research in Toxicology
REVIEW
Table 1. Summary of Drug Classes Discussed in This Review, with Adverse Effects That Have Been Historically Ascribed to Each Class (By the Medical/Scientific Community and/or the Public, Not by the Authors) drug class
a
ascribed class effect (historical)a
indication
5HT2c agonists
obesity
valvular heart disease
5HT3 antagonists 5LO inhibitors
IBS asthma
ischemic colitis liver injury
antibiotics (general)
bacterial infection
alcohol interaction
antibiotics (quinolones)
bacterial infection
chondrotoxicity
antibiotics (sulphonamides)
bacterial infection
skin toxicities
antibiotics (tetracyclines)
bacterial infection
liver injury
azole antifungals
fungal infection
CYP3A4/2C9 inhibiton
CCR5 inhibitors
HIV
liver injury
ETRAs glitazones
PAH diabetes
liver injury liver injury
H1 antagonists
allergy
sedation
statins
cardiovascular disease
rhabdomyolysis and liver injury
Confirmed or perceived.
evolved into a class effect generalization that alcohol should not be consumed while taking antibiotics. However, from a mechanistic point of view, given the broad range of antibiotic classes, mechanisms of action and individual drugs used, such a generalization does not make sense from a scientific perspective (and yet the myth perpetuates even in the scientific/medical community). While the alcohol/antibiotic myth is not thought to lead to noncompliance of antibiotic therapy, this example serves to demonstrate the power of suggesting a class effect without establishing a solid link between mechanism of action and effect.
chemotherapy. Dolasetron, ondansetron, palosetron, and zatosetron12 all cause a mixture of diarrhea and constipation in these indications, but their manner of use does not lead to further complications. Several compounds have been developed for the treatment of irritable bowel syndrome (IBS). These compounds suppress the reflex activation of colonic motor function in response to food ingestion in health and disease states. This reflex is part of postprandial function, but in diarrhea-predominant IBS, it is exaggerated resulting in cramping and diarrhea. Alosetron was in the forefront of this class of drugs and gave improvement in pain and discomfort in diarrhea-predominant IBS patients.13 Constipation occurred in 28% of treated patients (placebo around 3%). In a subgroup of patients, the fecal impaction of constipation resulted in increased colonic intraluminal pressure, which compressed the mucosal vessels and impeded mucosal circulation14 and led to ischemic colitis. Like alosetron, cilansetron is a potent and selective 5-HT3 receptor antagonist which improves global and specific IBS symptoms but with constipation the most frequent adverse event. Treatment with cilansetron,15 similar to that with alosetron, is associated with a similar low incidence of ischemic colitis events. The close association of severe side effect with the primary pharmacology indicates that ischemic colitis in IBS treatment with 5HT3 antagonists is a class effect and also not amenable to manipulation by changes in structure or pharmacokinetics. 5.3. Hyper-Response in Individuals to Primary Pharmacology in Nontarget Tissue. Statins represent the main therapeutic class of lipid lowering drugs that lower total and lowdensity lipoprotein (LDL) cholesterol. Statins inhibit 3-hydroxy3-methylglutarylCoA (HMG-CoA) reductase, the rate-limiting enzyme in cholesterol biosynthesis, which converts HMG-CoA to mevalonate. Inhibition of this lowers intracellular cholesterol thereby upregulating the expression of LDL receptors. Mevalonate is also the substrate for the synthesis of nonsteroid isoprenoids including farnesylpyrophosphate, geranylgeranylpyrophosphate (both attached to small GTP-binding proteins by protein prenyltransferases), coenzyme Q, dolichol, and isopentenyladenosine.16 The effects resulting from lowering the concentration of these isporenoids are termed pleiotropic. Muscle toxicity (myopathy) is the most common, but still a low incidence adverse effect of HMG-CoA reductase inhibitors, the extreme form of
4. TOWARD A UNIFICATION OF CLASS EFFECT TERMINOLOGY In this review, we critically examine various mechanisms (Table 1) which could cause toxicities in a pharmacological class, and we extend this examination to hepatoxicity and then provide some conclusions including how the terminology should be applied. We have used the term primary pharmacology to relate to the drug’s activity against its intended target and secondary pharmacology to include all unintended pharmacological action, whether due to reversible or irreversible binding of the drug or its metabolites. 5. CLASS EFFECTS ASSOCIATED WITH PRIMARY PHARMACOLOGY 5.1. Obligatory Mechanism Liabilities. In many early approaches to the chemotherapy of infection, mechanisms common to both mammalian cells and bacterial cells were the target, often with fairly low selectivity. For instance, the tetracycline class of antibacterials inhibit protein synthesis10 to provide antibacterial activity. This inhibition also occurs in the mammalian liver producing mitochondrial injury, thus interfering with triglyceride metabolism11 and resulting in microvesicular steatosis. Both antibacterial activity and hepatic toxicity are dose related. 5.2. Hyper-Response in Individuals to Primary Pharmacology in Target Tissue. 5-HT3 antagonists illustrate a class effect which is not attenuated by a biophase. The receptor, a GPCR, resides on the cell surface and is freely accessible. 5HT3 antagonists have been used in a number of indications such as the control of nausea and vomiting in cancer patients undergoing 465
dx.doi.org/10.1021/tx100408v |Chem. Res. Toxicol. 2011, 24, 463–474
Chemical Research in Toxicology
REVIEW
which is rhabdomyolysis. To a lesser extent, hepatotoxicity, peripheral neuropathy, and impaired myocardial contractility have been detected. These are likely to result from impaired protein prenylation, deficiency of coenzyme Q involved in mitochondrial electron transport and antioxidant protection, abnormal protein glycosylation due to dolichol shortage, or deficiency of selenoproteins.16 The effects on muscle probably relate to systemic inhibition of the enzyme, in comparison to the beneficial effects and hepatotoxicity which relate to hepatic inhibition. Cerivastatin showed rates of rhabdomyolysis up to 10-fold greater than other HMG-CoA inhibitors.17 The bioavailability in terms of systemic concentrations was the highest of any statin drug, and metabolism gave rise to circulating active metabolites. Other statin drugs show lower bioavailabilities due to active transport into the liver and selective concentration of the drug and active metabolites in the biophase. Recently, genetic variants in the gene encoding the organic anion transporting polypeptide (SLCO1B1; also known as OATP1B1) were reported to be significantly associated with myopathy in patients receiving high doses (80 mg) of simvastatin.18 In this study, patients with definite myopathy, classified as having creatine kinase levels greater than 10 times the upper limit of normal, were found to have an odds ratio of 4.5 due to association with the rs4149056 single nucleotide polymorphism (c.521T > C) of SLCO1B1 transport protein. However, it does not appear to be appropriate to apply this association universally to other statins for several reasons. First, the incidence of myopathy with statins has been shown to be concentration-dependent, and therefore, it is noteworthy that SLCO1B1 521T > C single nucleotide polymorphism has a greater effect on simvastatin pharmacokinetics, where a 3.2-fold increase in plasma concentrations is observed,19 relative to other statins where a lower increase is observed, e.g., atorvastatin and rosuvastatin,20 or where no effect is observed, e.g., fluvastatin.21 Indeed the 421C > A SNP for ABCG2 transporter (also known as Breast Cancer Resistance Protein, BCRP) had a much greater impact on rosuvastatin pharmacokinetics than SLCO1B1.22 Second, it is likely that the incidence of myopathy is also linked to the ability of statins to enter skeletal muscle cells. Aside from the varying physiochemical properties of statins which leads to differing passive permeability across cell membranes, it has recently been shown that a number of uptake (OATP2B1) and efflux (MRP1, 4 and 5) transport proteins and are present in skeletal muscle cell membrane and that the intracellular concentrations of atorvastatin and rosuvastatin are impacted by the expression of these transport proteins.23 Further studies are required for other statins to fully elucidate the importance of skeletal muscle transporters to incidence of myopathy. In support of the argument that incidence of myopathy with statin treatment cannot be ascribed solely to SLCO1B1 affinity, a recent genetic study has confirmed the association between the rs149056 variant of SLCO1B1 and myopathy for simvastatin but shows that there is no association between this variant and incidence of myopathy with atorvastatin.24 5.4. Eventual Resolution of Primary Pharmacology Effects. With the passage of time and increased knowledge, or even serendipity, problems affecting a number of compounds in the series seemingly leading to a class effect may be overcome. Early antiallergy antihistamines were sedating due to the effects of inhibiting H1 receptors in the brain. Fortuitous discovery of terfenadine and its active zwitterionic metabolite fexofenadine allowed the concept of nonsedating antihistamines. These later analogues had physicochemical properties that accessed the
systemic receptors but did not penetrate the biophase of the brain.25
6. CLASS EFFECTS ASSOCIATED WITH SECONDARY PHARMACOLOGY This grouping represents drugs where the selectivity for receptors or biophase concentrations are fixed across the various drugs that make up the class. For a few of these, the selectivity is against specific secondary receptors with narrow structureactivity relationships. For many, the secondary receptors are promiscuous receptors, ion channels, and transporters. For a secondary pharmacology to be a class effect in a drug series then, the structural features that govern the lack of selectivity are fixed and not subject to normal improvements by medicinal chemistry. 6.1. Inhibition of Related Receptors. Agonism of mitogenic 5-HT2B receptors on heart valves and pulmonary artery interstitial cells lead to the formation of proliferative fibromyxoid plaques that cause the deleterious changes in tissue integrity and function. Drugs implicated in valvular heart disease which are active at 5-HT2B receptors include fenfluramine, dexfenfluramine, ergotamine, and methysergide. The appetite suppressants fenfluramine and dexfenfluramine (a single enantiomer of fenfluramine) act at 5-HT receptors principally through an N-deethylated metabolite, norfenfluramine (nordexfenfluramine), which is potent in activating 5-HT2C receptors. This leads to the appetite suppression. The metabolite is equi-active in the agonism of 5HT2B receptors.26,27 This effect led to the withdrawal of the drug. Attempts to separate the activities and produce a pure 5-HT2C agonist have proved very difficult due to the overlapping structureactivity relationships (SAR) around agonism of these receptors. 6.2. Inhibition of Related Enzymes. Azole antifungal drugs such as ketoconazole, voriconazole, posaconazole, miconazole, itraconazole, and fluconazole target the endogenous CYP450 sterol 14-R-demethylase (CYP51). A key part of their activity is the interaction of the lone pair of electrons of one of the nitrogens of the azole group with the heme of the CYP450 enzyme. This interaction takes place by release of a water molecule from the sixth binding site of the heme reactive site when the ligand approaches the enzyme in the resting state. The pentacoordinated reactive site formed transitions to the coordination of the azole nitrogen to the heme iron. The azole binding is significantly stronger than the displaced water leading to the inhibition of the catalytic cycle.28 Not surprisingly, the same interaction is observed with the CYP450s metabolizing exogenous compounds such as CYP3A4 and CYP2C9. Because much of the interactions are hydrophobic in both classes of CYP450, there is only limited selectivity, and all of the compounds inhibit the metabolism of other drugs (conmeds) to some degree.29,30 However, there are appreciable differences between the drugs. Table 2 contrasts some of the agents and their selectivity against the target lanosterol14R-demethylase (cCYP51) in Candida albicans, its human equivalent (hCYP51), and the exogenous substrate metabolizing CYP3A4.31 Because of the extremely high affinity of these agents for the CYP450s, these results may reflect in part the stoichiometry of binding. The data clearly shows that even though the interaction of the azole ligand with the heme is a large component of the total binding energy, the variation in protein structure leads to selectivities blurring the crisp definition of a class effect. 6.3. Obligatory Structural Liabilities. Sulfonamide antibacterials are one of the earliest examples of chemotherapy against infection. The compounds as a class are inhibitors of tetrahydropteroic 466
dx.doi.org/10.1021/tx100408v |Chem. Res. Toxicol. 2011, 24, 463–474
Chemical Research in Toxicology
REVIEW
the thiazolidinedione group has been implicated. Metabolism of this group may lead to an ultimate reactive sulfonium ion, which could be formed from an initial sulfoxide followed by a formal Pummerer rearrangement, or a C5 thiazolidinedione radical or a sulfur cation radical.38 Other reaction possibilies leading to redox cycling are referenced in the section on 5-lipoxygenase (5LO) inhibitors. Recent evidence has emerged that troglitazone causes mitochondrial dysfunction which is likely to be an additional factor in the hepatotoxicity observed with this glitazone.39 Principal follow-on glitazone drugs also incorporate this group, but there is a much improved safety profile (see Introduction). Although the grouping is obligate, the other modifications of the molecule give up to 100-fold greater activity as PPARγ agonists. These differences are reflected in the clinical doses with rosiglitazone, pioglitazone, and troglitazone dosed at 4-8, 15-45, and 400-600 mg/day.37 Daily dose size has been observed to be a major factor in hepatotoxicity.40 This latter example illustrates how even common pharmacology and common structural features should not be correlated into class effect toxicology since intrinsic potency against the target and pharmacokinetic factors may change clinical doses dramatically. Hepatotoxicty is clearly not a class effect of the glitazones (represented by a major decline in publications on the subject), and the major clinical concern in the leading drugs in the class, pioglitazone and rosiglitazone, is fluid retention in 5-15% of patients.41 The glitazones target the nuclear transcription factor peroxisome proliferator-activated receptor γ, which is expressed in many tissues apart from the liver including the cardiovascular and renal systems. Fluid retention can result in, or exacerbate, edema and congestive heart failure. Since fluid retention reflects aspects of the primary pharmacology, this side effect can be considered a true class effect. That the glitazones differ subtly in aspects of selectivity is possibly reflected in the results of metaanalyses of clinical trials that suggest that rosiglitazone may increase the risk of myocardial ischemic events. In contrast, the results of meta-analyses and an outcomes trial provide indications that pioglitazone does not similarly increase the risk of coronary events, and this aspect of cardiac risk cannot be considered a class effect.42
Table 2. Comparison of IC50 or Ki Values for Fungal, Candida albicans (c), and Human (h) CYP51 and CYP3A4 Indicating That Some Selectivity Is Possible Even though Systemic Antifungals Normally All Cause Some Form of DDI Reaction with CYPs cCYP51a
hCYP51a
CYP3A4a
fluconazole
0.05
>30
11.9
ketoconazole itraconazole
0.06 0.039
0.4 >30
0.042 0.028
drug
a
All data are in μM (data are from ref 31).
acid synthetase. The natural substrate for this enzyme is paraaminobenzoic acid (PABA). Sulphonamide antibacterials structurally resemble the natural substrate PABA, being composed of para-aminobenzene (aniline) with the carboxylic acid being replaced by the isosteric sulphonamide. This grouping prevents further metabolism by the enzyme and also results in high affinity. The para-amino group is essential for antibacterial activity, and thus, this class of compounds contains an obligatory structural alert for reactive metabolites. The formation of reactive metabolites32 is of major importance in one of the principal side effects of sulphonamide antibacterials, that of skin rashes and more serious skin toxicities. The most serious skin toxicities are severe erythema multiforme, Stevens-Johnson syndrome, and toxic epidermal necrolysis. The initial rationale was that the N-4-hydroxylamine metabolites, which are formed by oxidation of the aniline nitrogen, bind covalently to proteins (because of their chemical reactivity), resulting in the induction of specific adverse immune responses; moreover, these metabolites are formed in the skin.33 Other metabolites of these drugs, particularly the nitroso, have gained more attention recently34 and shown to be potent antigenic determinants. T cell responses against the parent drug are also detected to a lesser extent suggesting a secondary low affinity complex.34 These T cells are low in proportion and responsive only to a particular drug (e.g., sufamethoxazole), whereas those generated from reactive metabolites can be stimulated34 by other structurally related drugs (e.g., sulfapyridine and sulfadiazine). Regardless of the exact mechanism, the aromatic amine group at the N4 position is essential in the cross reactivity35 and, as stated above, primary pharmacology. Chondrotoxicity of quinolone antibacterials is a common finding and can be deemed a class effect.The toxicity can affect articular cartilage and/or the epiphyseal growth plate, depending on the developmental stage. This toxicity is most likely related the magnesium-chelating properties of the quinolone drugs.36 Another manifestation of the chelating properties is on connective tissue leading to tendinitis and tendon rupture.
8. IMPACT OF STRUCTURAL VARIATION ON TOXICITY The case of CCR5 inhibitors referred to above has been resolved to the extent that the toxicity of aplaviroc was intrinsic to the molecule and not a class effect. The evidence for a possible class effect is limited to CCR5 knockout mice which are more sensitive to concanavalin A-induced immune-mediated hepatic toxicity.43 Contrary to this, human polymorphisms of CCR5 (the finding that led to the rationale for CCR5 antagonists and HIV treatment) are silent in terms of hepatic effects and additionally have shown no influence on the severity of hepatitis C virus cirrhosis and subsequent outcomes including death and hepatocellular carcinoma.44 A detailed description of the human toxicity of aplaviroc has been published detailing the cases of hyperbilirubinemia and hepatic cytolysis and higher than expected rate of liver enzyme elevations. This study also describes the absence of hepatic effects in macaque monkeys (a species in which apalaviroc inhibits the CCR5 receptor) treated at 2,000 mg/kg/day and hepatic findings of raised ALT and bilirubin levels in rat (a species whose CCR5 receptor does not bind aplaviroc) at 500 mg/kg/ day. The paper concludes (referencing to the absence of hepatic effects with maraviroc apart from the case detailed earlier) that
7. IMPACT OF VARIATIONS IN PRIMARY PHARMACOLOGY POTENCY ON SELECTIVITY Common structural features, even if a suspected toxicophore, will often show different outcomes with different drugs. The glitazone type drugs have been referred to in the Introduction. PPARR agonists, such as the fibrates, can correct dyslipidemia. PPARγ agonists, such as the glitazones, act as insulin sensitizers and improve insulin resistance. The key structural feature in these drugs is the acidic thiazolidinedione grouping which gives the glitazones unique potency as PPARγ agonists.37 Troglitazone, the prototype, was withdrawn due to hepatotoxicity, and 467
dx.doi.org/10.1021/tx100408v |Chem. Res. Toxicol. 2011, 24, 463–474
Chemical Research in Toxicology
REVIEW
Figure 1. Structures of maraviroc and aplaviroc showing their structural diversity. Maraviroc is a moderately lipophilic base (pKa 10.2), whereas aplaviroc is a zwitterion (pKa acid 4.3, pKa base 8.15).
hepatotoxicity is intrinsic to the molecule rather than its mechanism of action.45 Consideration of the structures (Figure 1) illustrates the structural diversity of the molecules. The zwitterionic nature of aplaviroc compared to the basic maraviroc potentially changes the permeability characteristics of the molecules as well as their potential interactions with transporter proteins. Aplaviroc has been characterized as a substrate and inhibitor of OATP1B1 and concentrations of radioactivity after radiolabeled aplaviroc are 70-100-fold greater in liver than blood.45
Figure 2. Structures of 5LO inhibitors; note the hydoxyurea function in zileuton, MLN-977, and atreleuton, and the thiophene in zileuton and altreluton.
and can be modulated through different physicochemical interactions with the enzyme (i.e., competitive, iron chelating, and electron transfer; Figure 3). Zileuton inhibits 5LO via the reduction of the ferric iron of the activated enzyme (and also via iron chelation), thus rendering the enzyme inactive, and is classed as a redox inhibitor.53 However, electron transfer from zileuton to the ferric iron of the enzyme does not occur in isolation, and it has been proposed that zileuton and analogous inhibitors undergo redox cycling via electron transfer through the N-hydroxyurea group,54 ultimately generating unstable intracellular lipid alkoxide and thiol radicals (Figure 4). Such redox cycling has the potential to generate reactive superoxide radicals, lipid peroxidation, and, ultimately, cytotoxicity. Indeed zileuton has been shown to increase reactive oxygen species in a human hepatocyte functional assay shown to be predictive of human hepatotoxicity, with a corresponding flag for drug-induced liver injury in these systems.55,56 There is perhaps a natural conclusion that the hepatotoxicity of zileuton is directly linked to its mechanism of action as a 5LO inhibitor, particularly when taking into consideration the serum liver enzyme elevations observed with the other 5LO inhibitors atreleuton and MLN-977. However, all three inhibitors contain the N-hydroxyurea moiety, which is a known toxicant via free radical-mediated lipid peroxidation of cell membranes57 and has been shown to take part in electron transfer reactions consistent with redox cycling.58 Electron transfer and reactive oxygen species have been implicated in the general toxicity of many drugs,59 but, importantly, redox cycling of drugs and/or their metabolites has been implicated in the hepatotoxicity of several other drug classes (e.g., doxorubicin,60 nilutamide,61 troglitazone,62 diclofenac,63 and geldanamycin64), albeit some are idiosyncratic in nature (whereas
9. DILEMMA OF THE LIVER AND PLAUSIBLE RELATIONSHIP TO CLASS EFFECTS: A CASE STUDY WITH 5-LIPOXYGENASE (5LO) INHIBITORS The link between hepatotoxicity and 5LO inhibition is not yet fully established, but a number of development compounds and the only marketed 5LO inhibitor zileuton (Figure 2) all show similar findings. In clinical trials, zileuton caused raised ALT greater than 3 times ULN in 2-4% of treated patients compared to 0.2-1.0% of placebo controls. Furthermore, there have been reports of patients suffering from hepatitis, jaundice, and mild reversible hyperbilirubinemia, with a rare occurrence (
