It is increasingly clear that genetic factors can predict susceptibility to drug-induced cholestasis. Advances in gene cloning and development of PCR-based genotyping methods during the 1980s and 1990s have enabled specific genetic associations to be investigated in detail. Initially this involved a candidate gene approach where genetic polymorphisms in biologically plausible gene candidates were analyzed in liver injury cases and frequencies compared with those in healthy controls. Candidate genes examined included mainly human leukocyte antigen (HLA) and genes encoding enzymes relevant to drug metabolism and transport. The main limitation with genetic studies on all forms of drug-induced liver injury (DILI) is that case finding is difficult; so most studies have been small which limits the power to detect associations reliably. Replicating associations has also been difficult. The development of genome-wide association studies (GWAS) where the contribution of all genes in the human genome to disease susceptibility can be investigated [12] has been a useful development in understanding the genetics of DILI better, but lack of large numbers of cases continues to be a limitation.

The key advances on genetic susceptibility relate to cholestasis induced by the two widely prescribed antimicrobials, flucloxacillin and co-amoxiclav. The main genes showing associations are the HLA genes. The proteins encoded by HLA genes are important in presenting foreign peptides, usually derived from invading pathogens, to T lymphocytes. HLA genes are located in a large gene cluster on chromosome 6 in an area of the genome referred to as the major histocompatibility complex (MHC). More detailed descriptions of HLA genes and their biological roles are widely available [6, 27]. HLA genes are divided into two types, class I and II. The proteins encoded by these gene classes have overlapping but slightly different functions in the immune response. Class I genes tend to be expressed in most cells in the body and have a role in presenting peptides to cytotoxic T-cells which may cause local damage. On the other hand, class II genes tend to be expressed mainly by immune system cells and have a more important role in stimulating T helper cells to produce cytokines which may induce inflammation. HLA genes are subject to a large amount of genetic polymorphism. This probably arose originally to help protect the host from pathogens, but increasingly alternative forms of HLA genes (alleles) are considered to be risk factors for autoimmune disease (e.g., multiple sclerosis and primary biliary cirrhosis) and also for adverse drug reactions. There is increasing evidence that many cholestatic DILI reactions involve an inappropriate immune response to a drug, possibly after a covalent link between the drug and a cellular protein has been formed. HLA proteins are likely to contribute by presenting the complex to T-cells.

Well-established and replicated HLA associations have been described for hypersensitivity reactions affecting various tissues, particularly the skin, and for DILI. Those carrying certain HLA alleles have an increased risk of developing these reactions. For cholestatic DILI, some of the HLA-associated reactions do not show obvious features of a hypersensitivity reaction, so findings of strong HLA associations have been slightly unexpected. The first HLA genotyping studies were candidate gene association studies on co-amoxiclav-related DILI. Two independent candidate gene association studies reported an identical association with the HLA class II DRB1*15:01 allele [15, 19, 30]. Further studies have found additional HLA class I and II associations for cholestatic DILI (Table 7.2). Effect sizes vary depending on the drug with odds ratios for risk of DILI of between 2 and 80 reported.

The strongest HLA association reported to date for any form of DILI is for the predominantly cholestatic reactions due to the antimicrobial flucloxacillin, which is used in a number of countries to treat gram-positive infections. This association was detected in the first GWAS performed for any form of DILI. A very strong association (odds ratio 80) for the development of flucloxacillin DILI was seen with the class I HLA allele B*57:01 [13] (Fig. 7.1). This allele had been shown previously to be a strong risk factor for hypersensitivity reactions to the drug abacavir, but these reactions generally do not involve the liver. Importantly, the underlying mechanism involved in the flucloxacillin DILI reactions appears to be different to the mechanism reported for abacavir with flucloxacillin believed to form a covalent complex with protein [28], while abacavir appears to interact directly with the HLA protein, causing inappropriate presentation of cellular peptides to T-cells [21]. No associations with B*57:01 for any other form of DILI have been described to date.

Early reports of an association between HLA-DRB1*15:01 and co-amoxiclav DILI [15, 19, 30] have been extended by a relatively large GWAS involving 201 predominantly cholestatic and mixed DILI cases related to this drug from Europe and the USA [26] (Fig. 7.2). This GWAS confirmed the DRB1*15:01 association but also described a second novel association involving HLA-A*02:01. Though the HLA class I A*02:01 signal is independent of the class II DRB1*15:01, those positive for both these alleles have a greater than additive increased risk of DILI, suggesting that the two risk factors interact. Importantly, the odds ratios observed for co-amoxiclav-related DILI are much lower than for B*57:01 in flucloxacillin DILI. Individually, DRB1*15:01 and A*02:01 carriage give odds ratios of approximately 2.5 for DILI development (Table 7.2), while individuals who are positive for both risk alleles show an odds ratio of approximately 9. One further non-HLA risk factor, a non-synonymous polymorphism in the gene PTPN22, was also detected when additional analyses involving autoimmune disease-related risk factors were performed. This non-HLA gene codes for a phosphatase enzyme that is believed to contribute to T-cell responses. The variant is also a risk factor for type 1 diabetes and rheumatoid arthritis but not for other forms of cholestatic liver disease such as primary biliary cirrhosis.

A third GWAS involving the DILI cases due to flucloxacillin and co-amoxiclav in the two earlier GWAS studies described above [13, 26], together with approximately 500 additional DILI cases, has also been performed [40]. The study included new cholestatic cases relating to drugs such as azathioprine, sulfamethoxazole/trimethoprim, terbinafine, and tetracyclines, but the numbers for each were relatively small (typically 5–10 cases). No genome-wide signals other than a HLA signal were seen either for the entire study or cholestatic cases alone. When the flucloxacillin and co-amoxiclav cases were excluded from the analysis leaving 77 cholestatic DILI cases due to other drugs, the significant HLA signal was lost suggesting that HLA genotype might not be a universal risk factor for cholestatic DILI.

A separate candidate gene study based in Japan was performed in 22 cases of DILI induced by ticlopidine [20]. The majority of these cases (64%) had suffered cholestatic liver injury. A significant association with the HLA class I allele A*33:03 was seen (Table 7.1) with the association being strongest in cholestatic cases only (odds ratio 36.5). This study was followed up by further genotyping for the main cytochrome P450 that metabolizes ticlopidine, CYP2B6 [5]. A significant association with a CYP2B6 allele that results in high metabolic activity was obtained suggesting that both high levels of a metabolite and the presence of a particular HLA protein might be factors in cholestatic DILI in response to this drug.