Genetic Predisposition to Type B Adverse Drug Reactions

GENETIC PREDISPOSITION TO TYPE B ADVERSE DRUG REACTIONS

Type B ADRs have typically been defined to be host-dependent (Rawlins and Thompson, 1991). However, the nature of this host dependency has not been defined for most drugs, although genetic factors have long been suspected. Indeed, genetic factors are also important for type A reactions as discussed above. It is becoming clear that the genetic basis of ADRs, in most cases, is going to be multi-genic (dependent on a combination of genes) and multi-factorial (depen-dent on an interaction between genetic and environ-mental factors). This is going to make it difficult to unravel the genetic basis of adverse reactions and will require a concerted effort to collect suitable cases and controls as part of multi-centre international collabo-rations (Pirmohamed and Park, 2001a).

The nature of the polygenic predisposition is unclear but in general could be divided into several areas (Figure 8.4) as follows (Park and Pirmohamed, 2001;  Pirmohamed  et al.,  1998;  Pirmohamed  and Park, 2001a):

·    Activation: Involves the activation of drug to CRMs. The bioactivation of drugs is largely medi-ated by cytochrome P450 enzymes, many of which have now been shown to be polymorphically expressed (Park, Pirmohamed and Kitteringham, 1995). Importantly, a deficiency of an enzyme will lead to reduced bioactivation of a drug and will act as a protective factor. No good exam-ples have been identified to date. By contrast, the amplification of a P450 isoform, as seen with

·        CYP2D6 2D6^∗2xN (Ingelman-Sundberg, Oscarson and McLellan, 1999), would increase bioactivation, but again no good example has yet been identified.

·        Detoxification: Absence or reduced activity of a detoxification enzyme would lead to a decrease in bioinactivation of the reactive metabolite (Pirmohamed and Park, 1999) and hence increase the possibility of the reactive metabolite interacting with important cellular macromolecules result-ing in different forms of toxicity. The best characterised example of this is the slow acety-lator phenotype predisposing to hypersensitivity with co-trimoxaole in HIV-negative patients (Rieder et al., 1991) and SLE with hydralazine and procainamide (Park, Pirmohamed and Kitteringham, 1992). There has also been inter-est in the role of the glutathione-S-transferase genes, many of which have been shown to be polymorphically expressed. However, although these gene polymorphisms may be important with respect to certain cancers, studies to date have not shown any association of the GST gene polymor-phisms with idiosyncratic drug reactions observed with co-trimoxazole (Pirmohamed et al., 2000), carbamazepine (CBZ) (Leeder, 1998) and tacrine (De Sousa et al., 1998; Green et al., 1995b).

·    Immune-response genes: The process by which the body’s immune system recognises a drug/drug metabolite as being foreign or antigenic and thereby mounts an immune response was conceived to be protective, but, perversely, this may lead to clinical manifestations typical of hypersensitivity. The genes encoding for immune responsiveness include MHC, T-cell receptors and co-stimulatory molecules.

·    Tissue-injury genes: The process by which an immune response is translated into tissue injury, the nature and extent of which can be counteracted by repair mechanisms that limit any tissue damage. Typical candidates include cytokines, chemokines and prostaglandins.

Since the completion of the human genome project, there have been some striking findings in the MHC with respect to its role in the genetic predisposition to drug hypersensitivity. These are illustrated below with reference to two compounds, abacavir and CBZ. However, it is important to bear in mind two important issues with reference to the MHC, which means that much more work is required in this area of the human genome. First, it is the most polymorphic region of the genome and exhibits a high degree of linkage disequi-librium. Therefore an association with one polymor-phism does not necessarily mean that this is a causal association. Second, the MHC has been sequenced and initial findings suggest that over 60% of the genes in this area are of unknown function, with only 40% being involved in the immune response (The MHC Sequencing Consortium, 1999).

Abacavir Hypersensitivity

Abacavir, an HIV-1 reverse transcriptase inhibitor, causes hypersensitivity, characterised by skin rash, gastrointestinal and respiratory manifestations, in about 5% of patients (Hetherington et al., 2001).

These reactions can occasionally be fatal, particu-larly on rechallenge. Mallal et al. (2002) found a strong association between abacavir hypersensitivity and the haplotype comprising HLA-B^∗5701, HLA-DR7 and HLA-DQ3 with an odds ratio of over

This association has now been shown in two other cohorts (Hetherington et al., 2002; Hughes et al., 2004a,b). The same association however has not been shown in an African American population presumably because of ethnic differences in linkage disequilib-rium patterns in the MHC (Hughes et al., 2004a). The association with the MHC in Caucasians is consis-tent with the immune nature of the reaction and the identification of drug-specific T cells in abacavir hypersensitive patients (Dodd et al., 2003; Phillips et al., 2005). By contrast, no association has been found with polymorphisms in the genes coding for various abacavir-metabolising enzymes (Hetherington et al., 2002). Mallal et al. (2002) have proposed that in Caucasians genotyping for HLA-B^∗5701 should be performed before the prescription of abacavir, and indeed in their clinic, this has resulted in a reduction in the incidence of abacavir hypersensi-tivity (Martin et al., 2004). An analysis of the cost effectiveness of prospective HLA-B^∗5701 genotyp-ing before abacavir hypersensitivity based on a meta-analysis of three cohorts showed that in Caucasians this would be a cost-effective strategy (Hughes et al., 2004b).

Carbamazepine Hypersensitivity

Carbamazepine, a widely used anticonvulsant, causes rashes in up to 10% of patients, and in occasional cases, this may be the precursor to the develop-ment of a hypersensitivity syndrome characterised by systemic manifestations such as fever and eosinophilia (Leeder, 1998; Vittorio and Muglia, 1995). Rarely, CBZ can induce blistering skin reactions such as SJS and toxic epidermal necrolysis, two conditions associ-ated with a high fatality rate (Rzany et al., 1999). CBZ hypersensitivity is a T-cell-mediated disease (Mauri-Hellweg et al., 1995; Naisbitt et al., 2003). CBZ is metabolised to CRMs that have been implicated in the pathogenesis of hypersensitivity (Pirmohamed et al., 1992). To date, no polymorphisms in the drug-metabolising enzyme gene polymorphisms have been associated with susceptibility to CBZ hypersensitivity (Gaedigk, Spielberg and Grant, 1994; Green et al., 1995a). Analysis of the MHC has led to the find-ing that CBZ hypersensitivity syndrome, but not mild maculopapular eruptions, is associated with the haplo-type TNF2-DR3-DQ2 (Pirmohamed et al., 2001). This has also been borne out in more recent studies in an extensive analysis of the heat shock protein (HSP) locus, which has shown that severe but not mild CBZ hypersensitivity reactions are associated with three SNPs in the HSP-70 locus, two in HSP-70-1 and one in HSP-Hom (Alfirevic et al., 2006). These stud-ies suggest that in Caucasians the causal variant for CBZ hypersensitivity resides on the ancestral haplo-type 8.1 (Pirmohamed, 2006). In the Han Chinese, however, the susceptibility locus has been suggested to be different following the finding of a strong asso-ciation between CBZ-induced SJS and HLA-B^∗1502 (Chung et al., 2004).

In the future, it may be possible to use a comprehen-sive, densely spaced, genome-wide SNP map that may screen for pharmacogenetically active genes as whole genome, unbiased searches (Roses, 2000). SNPs are single-base differences in the DNA sequence, observed between individuals, which occur through-out the human genome. The International SNP Map Working Group (2001) has published a map of 1.42 million SNPs throughout the genome, occurring at an average density of one SNP every 1.9 kb; by the end of 2005, almost 10 million have been identified, of which 50 000 code for variants that can lead to amino acid changes. High-density SNP maps derived from this information will provide an opportunity to perform SNP profiling to identify genetic factors predispos-ing to ADRs. However, before this can become a reality, the cost of genotyping needs to come down. Furthermore, given the need to test for multiple mark-ers simultaneously, an issue that needs to be consid-ered is the sample size and the level of statistical significance required to prevent the detection of false-positive associations. A recent study has reported that for testing 100 000 loci in a genome-wide screen will require a 3-fold greater sample size at a significance level of 2 5 × 10⁻⁷ (Cardon et al., 2000). This does suggest that for pharmacogenomic detection of rare adverse events, testing in phases I–III is not likely be practical and will require prospective storage of samples and evaluation in phase IV when a problem has been identified.