Characteristics of Idiosyncratic Toxins

Reactive intermediate formation

It is generally accepted that formation of one or more reactive intermediates is required before an idiosyncratic reaction can be triggered. Avoidance of "toxicophores" or chemical structures that are likely to undergo bioactivation is a logical initial approach to lead optimization. A number of readily recognizable chemical structures have been associated with idiosyncratic reactions and avoiding these structures will likely reduce the risk of reactive intermediate formation associated with idiosyncratic toxins. The reader is referred to the chapter by John Walsh in this book for a detailed discussion of reactive intermediate formation and toxicophores associated with bioactivation. There is not, however, a direct correlation between the formation of reactive intermediates and occurrence of idiosyncratic reactions. That is, there are many compounds that form reactive intermediates that are either not associated with idiosyncratic reactions or cause a predictable dose dependent toxicity. Formation of free radicals is not generally associated with compounds that cause idiosyncratic reactions, presumably because they are not as commonly associated with covalent binding.

Many methods have been proposed to identify reactive intermediates: covalent binding studies (see below), screening for suicide inhibition of drug metabolizing enzymes, or screening for glutathione conjugates. These approaches do not identify all possible reactive intermediates either because of complex metabolic pathways (i.e. secondary bioactivation of a metabolite) or because of bioactivation by non-cytochrome P450 pathways such as myeloper-oxidase. They do, however, provide a substantive starting point if reactive intermediate formation is a concern from a structural viewpoint.

Eliminating any and all compounds that form or have the potential to form reactive intermediates would be expected to reduce the over-all likelihood of idiosyncratic reactions. However, this approach will also likely result in the loss of potentially useful compounds.

Should we optimize simply to minimize the quantitative formation of reactive intermediates or is there a qualitative difference between reactive intermediates that can be used to aid in lead optimization? A simple quantitative assessment of bioactivation is not a good indicator of likelihood of causing idiosyncratic reactions when comparing across groups of structurally unrelated

Compound groups

Risk of idiosyncratic reactions

Extent of bioactivation

Method of assessment

volatile anesthetics

halothane >>enflurane >isoflurane

halothane >>enflurane >isoflurane

Total metabolism; Covalent binding


slow acetylator >fast acetylator

slow acetylator >fast acetylator

Phenotyping studies; theoretical assessment of metabolism only




procainamide>> acetyl-procainamide

In vivo in humans and rats

chloramphenicol series

chloramphenicol>> thiamphenicol

chloramphenicol>> thiamphenicol

Block of metabolic pathway


clozapine >>olanzapine

clozapine =olanzapine

% conversion similar but total daily dose of olanzapine much lower

Table 5. Extent of bioactivation/metabolism and risk of idiosyncratic reactions.

Table 5. Extent of bioactivation/metabolism and risk of idiosyncratic reactions.

compounds. However, there are several examples that suggest quantity remains an influence. First, it is now recognized that idiosyncratic reactions are rarely associated with total daily doses less than 100 mg and practically non-existent at dose less than 10 mg a day (Uetrecht 1999). Within a structurally related series of compounds, there is evidence that reducing turn-over via metabolic pathways associated with bioactivation is associated with a reduced risk of idiosyncratic reactions (Christ et al. 1988; Freeman et al. 1981; Gardner et al. 1998; Gardner et al. 1998; Rieder et al. 1991; Shear et al. 1986). Reduction of metabolic turnover has been achieved either through chemical modification or through individuals that have genetic polymorphisms that influence the rate of bioacti-vation directly or indirectly (Table 5). These examples support the concept that within a structurally-related series, a reduction in the total formation of reactive intermediates will reduce the risk of idiosyncratic reactions. There are also examples where there are not large differences in bioactivation rates for compounds with large differences in idiosyncratic reaction rates (clozapine and olanzapine) or where the rate of metabolite turnover measured does not correlate with risk of IDR (ibuprofen and ibufenac) (Bolze et al. 2002; Castillo and Smith 1995; Gardner et al. 1998). Therefore, we must look at further properties of the compounds and their associated reactive intermediates, specifically there reactivity with cellular proteins and their propensity to cause direct cytotoxicity.

Covalent Binding

Covalent binding appears to be an integral step in the pathogenesis of idiosyncratic reactions, whether you adhere to the traditional hapten hypothesis or to the danger hypothesis. This would suggest that eliminating covalent binding would eliminate idiosyncratic reactions. However, we know that many effective and safe compounds are associated with covalent binding in in vitro systems and in vivo in animals or humans. Therefore, for covalent binding to be a more useful marker of risk and hence a useful tool in lead optimization, additional characteristics of covalent binding must be considered.

The same arguments that apply to bioactivation to reactive intermediates apply to covalent binding. Compounds that are associated with increased reactive metabolite formation are often also associated with increased covalent binding in vitro and in vivo. Within a structurally related series, compounds with the lowest covalent binding are likely to have the lowest likelihood of triggering idiosyncratic reactions (Christ et al. 1988; Freeman et al. 1981; Gardner et al. 1998). However, this does not help resolve comparisons across structurally unrelated compounds. It is clear that an absolute level of covalent binding has little relevance to prediction of idiosyncratic reactions. Three properties of covalent binding have therefore received attention: targets of binding, association with cytotoxicity, and in vivo binding.

Considerable effort has been devoted to identifying specific targets or patterns of covalent binding that may be linked to idiosyncratic reactions. The majority of protein targets identified to date have been either endoplasmic reticulum proteins or cell surface proteins (Table 6). As discussed in the preceding section on immunopathogenesis, the endoplasmic reticulum (ER) proteins that have been identified as targets of the immune response are also found expressed on the cell surface, albeit not in large quantities. This observation also holds for targets of covalent binding. Most of the work has been performed in animals (usually rats) but some comparisons have been performed with either human hepatocytes or in samples obtained from humans receiving the drug. For several compounds, there have been comparisons of the patterns of covalent bindings based on molecular weights on gel electrophoresis, without attempts to identify specific targets. These studies have not been able to clearly identify any pattern or unique targets associated with idiosyncratic toxins other than a tendency to adduct membrane proteins.

It is instructive to view a few specific examples. Tienilic acid and halothane are two compounds that are associated with idiosyncratic hepatitis with limited systemic signs (e.g. dermatopathy is not a major feature of the reaction). Tienilic acid displays a relatively specific binding to predominantly one target, a cytochrome P450: CYP2C11 in rats and CYP2C9 in humans (Beaune et al. 1987; Pons et al. 1991). In contrast, halothane binds to a whole range of ER proteins, including CYP2E1 and a number of ER stress proteins (Eliasson and Kenna 1996; Kenna et al. 1987; Pumford et al. 1997a). Thus, the range from specific binding to wide ranging covalent binding is associated with idiosyncratic reactions, and similar observations have been made regarding compounds that are intrinsically



Covalent binding quantitatively/ qualitatively linked to direct cytotoxicity

Method of assessment

Volatile anesthetics: halothane

ER proteins: GRP78, GRP94, microsomal carboxylesterase, protein disulfide isomerase, 58 kDa PDI-related protein, CYP2E1


In vivo rat and human studies In vitro

(Gut et al. 1993)


ER proteins: GRP78, GRP94, PDI, unidentified P450s Cell surface proteins Unidentified serum protein


In vitro only for ER and cell surface proteins; Human in vivo for serum protein

(Cribb et al. 1996; Cribb et al. 1997; Manchanda et al. 2002)

Tienilic acid



in vitro/in vivo

(Lecoeur et al. 1994)


Cell surface ER/nuclear membrane proteins


in vitro/in vivo

(Aithal et al. 2004; KretzRommel and

Boelsterli 1994; Wade et al. 1997)

Table 6. Examples of targets of covalent binding of idiosyncratic toxins. The following table provides examples of targets of covalent binding of idiosyncratic toxins that have been identified. Many other studies have been performed in which specific targets have not been identified. Examples of covalent binding to cytosolic proteins have been found, but they are less common.

Table 6. Examples of targets of covalent binding of idiosyncratic toxins. The following table provides examples of targets of covalent binding of idiosyncratic toxins that have been identified. Many other studies have been performed in which specific targets have not been identified. Examples of covalent binding to cytosolic proteins have been found, but they are less common.

toxic (Pumford et al. 1997a). The most consistent feature of the covalent binding of compounds that are associated with idiosyncratic reactions is that covalent binding is heavily weighted towards membrane-associated proteins. The endoplasmic reticulum, nuclear envelope, and cell membrane proteins are the most common targets of idiosyncratic toxins. Non-idiosyncratic toxins also bind to membrane proteins, but there appears to be higher binding to cytosolic proteins (Cohen and Khairallah 1997; Cohen et al. 1997; Pumford et al. 1997b). Our failure to identify one or more critical targets of covalent binding has three possible explanations. The first is that such common, critical targets do not exist. The second is that scientists as a group have inadvertently studied outlier compounds that do not share critical important targets. The third possibility is that the common, critical targets exist but that we have not been able to identify them because the experimental approaches used have not been sensitive enough or are simply not an appropriate approach. Only further experimentation will resolve this issue.

Another way of looking at covalent binding is to assess its link with cytotoxicity. For compounds that are associated with intrinsic toxicity, there is generally a good correlation between covalent binding and intrinsic toxicity (Gibson et al. 1996; Pumford et al. 1997a). That is, cytotoxicity is apparent shortly after covalent binding appears and increases in association with total covalent binding. For compounds that are associated with idiosyncratic toxicity, there is generally extensive covalent binding in cells (in vitro systems) or in tissues (in vivo) at concentrations below that causing overt cellular toxicityl, or the covalent binding is dissociated from cytotoxicity (Kenna et al. 1988; Kretz-Rommel and Boelsterli 1993; Naisbitt et al. 2002; Reilly et al. 2000; Summan and Cribb 2002). For example, in vitro studies have shown that covalent binding of sulfamethoxazole can occur without toxicity. In lymphoid cells, surviving cells after exposure to sulfamethoxazole hydroxylamine have extensive covalent adducts (Summan and Cribb 2002). Keratinocytes can bioactive sulfamethoxazole to protein reactive compounds without cytotoxicity and circulating adducts of sulfamethoxazole have been observed in people (Reilly et al. 2000). Studies using isolated cells have shown that the concentration of sulfamethoxazole reactive metabolites required to cause covalent binding sufficient to trigger lymphocyte activation through immune recognition are approximately one-tenth the concentration associated with cytotoxicity (Naisbitt et al. 2002).

The reactive intermediates responsible for covalent binding and for cytotoxicity are not necessarily the same. Diclofenac is a good example where it has been shown that the covalent binding is primarily related to acylglucuronide formation while cytotoxicity is primarily the result of cytochrome P450-dependent bioactivation (Kretz-Rommel and Boelsterli 1993). Early studies on the molecular basis of liver toxicity from acetaminophen established that it was possible to create compounds that were bioactivated to protein-reactive reactive intermediates, but did not cause cell death (Roberts et al. 1990). The partial or complete dissociation between covalent binding and cytotoxicity may allow the accumulation of covalent adducts such that when they are released or a danger signal is received, a relatively strong immunogen exists within the body as haptenic density is important in inducing an immune response against haptens.

In summary, the current data suggest that covalent binding is a risk factor for idiosyncratic reactions particularly when it occurs prior to the onset of intrinsic cellular toxicity. The current data suggest that this property is likely of greater importance than binding to specific cellular targets or the absolute extent of covalent binding in fractionated in vitro systems.


All the idiosyncratic toxins examined to date have been shown to be bioactivated to compounds that cause small but measurable cytotoxicity. As was discussed in the previous section, the cytotoxicity occurs at concentrations higher than those that result in covalent binding or may result from a different metabolite than that responsible for the majority of covalent binding. The ability to measure cytotoxicity from idiosyncratic toxins is dependent on the assay system used and the compound involved. For some compounds, toxicity can be demonstrated in isolated human hepatocytes with intact cytochrome P450 systems while in other cases a bioactivation system (e.g. hepatic microsomes or in some cases a peroxidase-based system) must be combined with a sensitive cell type such as lymphocytes (Riley and Leeder 1995; Riley et al. 1988; Shear et al. 1986). The utility of such systems early in lead optimization is limited, but they may be used in the late phase of lead optimization to assist in further characterization of compounds or in comparisons of compounds. The ability of idiosyncratic toxins to cause cytotoxicity suggests that cell damage or stress is a required part of the pathogenesis, whether through the release of immunogens or the generation of a danger signal.

The weight of evidence therefore suggests that a small and normally clinically insignificant degree of cytotoxicity is a characteristic of idiosyncratic toxins. Compounds that are intrinsically highly cytotoxic or bioactivated to highly cytotoxic intermediates will more likely be associated with intrinsic than idiosyncratic toxicity. If we accept that a low, but measurable, cytotoxic potential exists for idiosyncratic toxins, how can be exploit this information? The degree of cytotoxicity often parallels the over-all extent of bioactivation, so it may move in parallel with bioactivation and covalent binding. Within the few groups of structurally related compounds studied to date, there appears to be a correlation between the frequency of idiosyncratic reactions and the degree of cytotoxicity. The volatile anesthetics are a good example of this. Similarly, it has been observed for olanzepine and clozapine that while metabolic bioactivation is similar, cytotoxicity is lower with olanzepine, as is the frequency of idiosyncratic reactions. In vitro studies suggest that clozapine bioactivation can result in cytotoxicity at clinically relevant concentrations, while higher concentrations of olanzepine are required (Gardner et al. 1998). These observations suggest that to a certain extent, increased cytotoxicity increases risk but that eventually a compound will become an intrinsically toxic compound. Cytotoxicity is a graded response, so one would predict that some compounds will be in a grey zone. This is indeed that case and there are some compounds, such as diclofenac, where clinical presentation is more consistent with intrinsic toxicity in some patients and idiosyncratic immunemediated toxicity in others (Boelsterli 2003).

The requirement for an underlying cytotoxic potential may explain the unusual but consistent observation that isolated mononuclear leukocytes from patients are more susceptible to the toxic effects of bioactivation products of drugs associated with idiosyncratic reactions than are controls (Shear and Spielberg 1988; Shear et al. 1986). The assay is based on exposure of isolated mononuclear leukocytes from patients and controls to the drugs in the presence of a microsomal bioactivating system to generate reactive metabolites (variously referred to as the MNL toxicity or lymphocyte toxicity assay). Drugs studied include sulfonamides (Rieder et al. 1989; Shear et al. 1986), cefaclor (Kearns et al. 1994), aromatic anticonvulsants (Shear and Spielberg 1988), sorbinil (Shear and Spielberg 1988), and amineptine (Larrey et al. 1989). The MNL toxicity assay may be used to characterize and screen compounds in development to help optimize leads but only after idiosyncratic or some other unusual toxicity has been observed. It is not an assay that is appropriate for routine use during lead optimization.

Other Characteristics

The hope remains that array screening will allow us to identify particular patterns of gene activation that will be linked with idiosyncratic reactions. Using proteomics, we may be able to identify critical targets or responses that can be used to guide lead optimization. However, the major limitation is that the animal models that develop the complete idiosyncratic reaction remain limited and not widely applicable, often for practical reasons. It is unclear whether animal models that do not develop the complete clinical profile will ever be predictive. In vitro studies with isolated human cells are limited by the lack of cell-to-cell interactions that are likely required for the full manifestation of the response.

Paradigms that may identify highly immunogenic compounds have been proposed. These include the popliteal lymph node assay, co-administration of drugs with Freund's adjuvant as a danger signal, and in vitro T-cell priming. However, these assays all have limitations and no assays have been developed or validated to the point that they may be useful in lead optimization.

How To Bolster Your Immune System

How To Bolster Your Immune System

All Natural Immune Boosters Proven To Fight Infection, Disease And More. Discover A Natural, Safe Effective Way To Boost Your Immune System Using Ingredients From Your Kitchen Cupboard. The only common sense, no holds barred guide to hit the market today no gimmicks, no pills, just old fashioned common sense remedies to cure colds, influenza, viral infections and more.

Get My Free Audio Book

Post a comment