Year: 2024
Type: review
Abstract: WHAT IS DILI, ITS INCIDENCE, AND ITS MORBIDITY AND MORTALITY? DILI is a rare adverse reaction to drugs or herbal supplements that often has toxic, metabolic, and immunological contributors. It is the leading cause of acute liver failure in the United States.1 Its incidence is estimated to be 13.9 per 100,000 individuals.2 It has high morbidity and mortality, with 64% of patients with DILI-induced acute liver failure undergoing liver transplantation or dying.3 Most cases of DILI are associated with small molecule drugs; however, 15%–20% of cases are associated with herbal supplements, sometimes known as herbal-induced liver injury.4 Less is known about DILI associated with biologics and large molecules, although immunotherapies such as immune checkpoint inhibitors and other biologics such as TNF receptor antagonists have been implicated.5 Diagnosing DILI and assigning drug causality is often a challenge as patients may be receiving multiple medications concurrently, and presentation may occur significantly after discontinuation of the offending medication. WHAT IS THE MOST WELL-KNOWN GENETIC RISK FACTOR FOR DILI? Although there can be many contributors, including polymorphisms in drug metabolizing enzymes, human leukocyte antigens (HLA) are the most well-known genetic risk factor for DILI.6 HLA is responsible for presenting self and foreign antigens to T cells for tolerance and sensitization, respectively. HLA class I (eg, HLA-A, HLA-B, and HLA-C) present intracellular source peptides, whereas HLA class II (eg, HLA-DR, HLA-DP, HLA-DQ) present extracellular source peptides to T-cell receptors (TCRs). HLA class I is expressed on all nucleated cells. HLA class II is expressed only on professional antigen-presenting cells, including B cells, macrophages, and dendritic cells. HLA class I and class II are highly polymorphic with >25,000 alleles. HLA-B is the most diverse gene in the human genome. Provided that a large number of controls are being compared against, an HLA association with DILI, if present, can be detected with as few as 6–15 cases because of the high effect size of HLA on T-cell–mediated immune-mediated adverse drug reactions.7 WHAT DRUG HAS THE STRONGEST ASSOCIATION WITH AN HLA RISK ALLELE FOR DILI? WHAT ARE SOME OTHER TOP DRUG OR HERBAL SUPPLEMENT HLA ASSOCIATIONS WITH DILI? The drug with the strongest HLA association for DILI is flucloxacillin, a beta-lactam antibiotic, available in Europe and Australia but not the United States with HLA-B*57:01 (OR 36.6 [26.1–51.3], p=2.6×10−97).8 Other top drug-specific HLA associations with DILI include amoxicillin-clavulanate with HLA-A*02:01 and HLA-DRB1*15:01,9 green tea with HLA-B*35:01,10 and terbinafine with HLA-A*33:01.11Table 1 includes a comprehensive list of drug and herbal supplement HLA associations with DILI. TABLE 1 - Reported genetic associations with DILI Compound Allele Population/geographic region in which the association described References Antibiotics/anti-inflammatory Dapsone HLA-B*13:01a East Indian 12,13 Infliximab HLA-B*39:01 HLA-DQB1*02:01HLA-DRB1*03:01 Caucasian/European 13–15 Lumiracoxib HLA-DRB1*15:01 Caucasian/European 13 Antibiotics, sulfonamides Sulfamethoxazole trimethoprim HLA-A*34:02 HLA-B*14:01 HLA-B*27:02 HLA-C*08:02 Caucasian/European 13,16 Sulfamethoxazole trimethoprim HLA-B*35:01 HLA-B*53:01 African American 13,16 Antibiotics, beta-lactam Amoxicillin-clavulanate HLA-B*15:18HLA-B*18:01 HLA-DRB1*15:01 HLA-DRB1*15:03a Caucasian/European 9,13,17–21 Amoxicillin-clavulanate HLA-A*02:01/HLA-DRB1*15:01 HLA-DQB1*06:02 Northern European 13,17,20 Flucloxacillinb HLA-B*57:01 HLA-B*57:03 Caucasian/European 9,13,17–20,22 Antibiotics, other Minocycline HLA-B*35:02 Caucasian/European 13,17,23 Nitrofurantoin HLA-A*30:02HLA-A*33:01 HLA-A*33:03 HLA-C*06:02HLA-DPB1*16:01HLA-DRB1*07:01 HLA-DRB1*11:04 HLA-DQA1*02:01HLA-DQB1*02:02 Caucasian/European 13,24,25 Vancomycin HLA-A*32:01a Caucasian/European 26 Antifungals Terbinafine HLA-A*33:01 Caucasian/European 13,17,27 Anticonvulsants Carbamazepine HLA-A*31:01aHLA-A*32:01 Caucasian/EuropeanAfrican American 13,28 Ethosuximidec HLA-A*32:01 African American 28 Lamotrigined HLA-A*32:01 African American 28 Phenytoin HLA-A*03:01HLA-A*32:01 African American 28 Phenytoine HLA-B*53:01a African American 28 Pregabalin HLA-A*32:01 African American 28 Topiramate HLA-A*32:01 African American 28 Valproate HLA-A*32:01 African American 28 Antithyroid agents Methimazole HLA-C*03:02 Chinese 13 Aminopyridines Flupirtine HLA-DRB1*16:01 Caucasian/European 13 Antihemintics Albendazole HLA-B*13:01HLA-B*15:02 — 29 Anti-HIV and anti-TB drug combinations Various anti-HIV and anti-TB drug combinations HLA-B*57:02HLA-B*57:03 African American 13 Isoniazid HLA-B*52:01 — 30,31 Direct thrombin inhibitors Ximelagatran HLA-DRB1*07:01 Caucasian/European 13 Fibric acid derivatives Fenofibrate HLA-A*33:01 Caucasian/European 13 Nonsteroidal anti-inflammatory agents Diclofenac HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Etodolac HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Ibuprofen HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Meloxicam HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Nimesulide HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Sulindac HLA-A*02:01HLA-B*35:03HLA-B*57:03HLA-DPB1*01:01 HLA-DRB1*04:03 — 32 Thienopyridine agents Ticlopidine HLA-A*33:01HLA-A*33:03 Caucasian/EuropeanJapanese 13 Tyrosine kinase inhibitors Pazopanib HLA-B*57:01 Caucasian/European 13 Lapatinib HLA-DQA1*02:01 HLA-DRB1*07:01 Caucasian/European 13,33 Xanthine oxidase inhibitors Allopurinol HLA-A*34:02 HLA-B*53:01aHLA-B*58:01a European, African, and Hispanic American 13,28,34 Vaccinations COVID-19 vaccine HLA-DRB1*11 — 35 Herbal drugs Polygonum multiflorum HLA-B*35:01 Chinese 13,36,37 Green tea extract HLA-B*35:01 Multiple ethnicities 10,13 Garcinia cambogia + green tea HLA-B*35:01 European American 10,38 Hydroxycut + green tea HLA-B*35:01 — 39 Turmeric + curcumin HLA-B*35:01 European American 40 Kampo medicines HLA-B*35:01 HLA-C*03:03 Japanese 36,41 Note: This table represents a comprehensive summary of various drugs associated with DILI and their commonly linked alleles. The sensitivity and specificity are listed for some of these alleles, along with the PPV and NPV per the general population. The bolded alleles in this table represent primary associations.aDenotes association with both DILI and DRESS.bPPV: 0.12%; NPV: 99.99%.cSensitivity: 53%; Specificity: 100%; PPV: 0.7%; NPV: 100%.dSensitivity: 50%; Specificity: 100%; PPV: 1%; NPV: 100%.eSensitivity: 89%; Specificity: 100%; PPV: 0.6%; NPV: 100%.Abbreviations: DRESS, drug reaction with eosinophilia and systemic symptoms; PPV, positive predictive values; NPV, negative predictive values. Some important lessons learned about HLA and DILI are that there is no single HLA associated with general DILI risk. While HLA alleles are drug-specific, DILI risk for a particular drug may extend across several HLA alleles, such as those with shared peptide specificities. For example, flucloxacillin is associated not only with HLA-B*57:01 but also HLA-B*57:03, which differ by only 2 amino acids and hence have significant shared peptide binding specificities.8 In addition, a specific HLA risk allele may be associated with multiple drugs with varying phenotypes but very different clinical implications, as in the case of HLA-B*57:01 with flucloxacillin DILI and abacavir hypersensitivity syndrome, with a positive predictive value (PPV) of 0.12% and 55%, respectively.42 WHAT IS DRUG-INDUCED AUTOIMMUNE-LIKE HEPATITIS, AND HOW DOES IT DIFFER FROM AUTOIMMUNE HEPATITIS? WHAT DRUGS ARE COMMONLY IMPLICATED WITH DI-ALH? Drug-induced autoimmune-like hepatitis (DI-ALH) is a subtype of DILI that shares both laboratory and histologic similarities with autoimmune hepatitis (AIH).43 These include the presence of serum liver autoantibodies (eg, anti-nuclear antibody, anti-smooth muscle antibody, and anti-liver-kidney microsomal antibody) and prominent lympho-plasmocytic infiltrates in portal spaces and interface hepatitis. The onset of DI-ALH is usually insidious. Although typically longer than 2 months, the latency of DI-ALH can vary significantly, and it is largely dependent on the drug. For instance, it is notably shorter for drugs such as infliximab but can be much longer for a drug such as nitrofurantoin, which is used chronically for urinary tract infection prophylaxis, where it can be prevented even years after initiation. While there is no single biomarker able to differentiate between DI-ALH and AIH, DI-ALH typically resolves with the withdrawal of the culprit drug alone, whereas AIH requires long-term immunosuppressive treatment. HLA testing may be helpful in the diagnosis of DI-ALH versus AIH if a drug-specific HLA risk allele, such as HLA-B*35:02 with minocycline exposure, or idiopathic AIH risk alleles, such HLA-DRB1*03:01 or HLA-DRB1*04:01 in European populations, are present.44 Of interest, HLA-B*35:01, which has been frequently associated with DILI and herbal-induced liver injury, was recently found to be associated with AIH in a Han Chinese population.45 The most implicated drugs with DI-ALH include minocycline, nitrofurantoin, methyldopa, and infliximab. WHAT ARE THE PROPOSED MODELS OF HLA-ASSOCIATED IMMUNE-MEDIATED DRUG REACTIONS? The models of HLA-associated immune-mediated drug reactions, as shown in Figure 1, include the hapten/pro-hapten, where the self-protein and drug-antigen are bound and presented, the pharmacological interaction (ie, HLA or TCR) where T cells are stimulated by such bound drug antigens. Lastly, the altered peptide (ie, HLA or TCR) model shows how immunologically labeled peptides are modified through the binding of drug antigens to HLA and TCR.46 The heterologous immunity model is nonmutually exclusive to all of the other models and proposes that an early-life HLA-restricted response to a pathogen (eg, virus) is followed later in life by a cross-reactive memory T-cell response (restricted by the same HLA allele) between this early-life response and the drug-altered self-peptide.47 The heterologous immunity model, which is not mutually exclusive to the models above, could explain why some drug reactions can occur within a single dose or a few days.FIGURE 1: Models of drug-induced HLA-restricted T-cell activation and heterologous immunity. (A) The drug-induced HLA-restricted T-cell activation models include (i) the hapten and pro-hapten model, (ii) HLA and (iii) T-cell receptors (TCR) pharmacological interactions models (PI), and the altered (iv) HLA and (v) TCR peptide models. This model of T-cell activation is not specific to drug-induced liver injury but expected among several mechanisms of T-cell–mediated liver injury, including immune and chemically induced injury. (i) In the hapten/pro-hapten model, an immunogenic drug-modified peptide, created through the binding of drug-antigen to the self-protein, is presented by HLA to the TCR. (ii, iii) The PI models display how T-cell activation occurs by a process of drug-antigen binding with either the risk HLA or TCR. (iv, v) The altered peptide models display how peptides are modified to present as immunogenic to T cells through the binding of drug antigens to HLA and TCR. Reprinted with permission from Deshpande et al46 (B) The heterologous immunity model. (i) Early-life exposure to a virus leads to (ii) priming of naive T cells with HLA-presented viral peptides. (iii) Future exposure to structurally similar drug antigens (iv) triggers a reactivation of these viral-primed T cells, (v) eliciting an immune-mediated hypersensitivity reaction. Created with BioRender.com. Abbreviations: APC, antigen-presenting cells; HLA, human leukocyte antigens; TCR, T-cell receptors.WHAT MODELS EXIST TO EXPLAIN WHY MOST INDIVIDUALS WITH AN HLA RISK ALLELE WOULD BE DRUG TOLERANT? In a Caucasian/European population, while HLA-B*57:01 has a 99.9% negative predictive value (NPV) for flucloxacillin DILI, it only has a 0.12% PPV.8 This underscores that most individuals with the HLA-B*57:01 risk allele are tolerant. Using transgenic murine models expressing HLA-B*57:01, Ananthula et al48 studied the effects of depleting CD4+ T cells with anti-CD4+ antibody and knocking out programmed cell death 1 expression. CD4+ T cells were targeted to deplete T regulatory cells. Knockout of programmed cell death 1 expression by CD8+ T cells, which engages with programmed cell death ligand 1 expressed by liver sinusoidal epithelial cells and KCs, intended to remove cytotoxicity inhibition. By removing the tolerogenic environment, including T regulatory cells and programmed cell death 1 expression, there was increased liver infiltration in mice by CD8+ T cells, but the degree of liver damage was not consistent with the severity of flucloxacillin DILI in humans. This likely reflects additional unaccounted-for tolerizing mechanisms or differences in drug bioavailability, antibiotic metabolism, and hapten/pro-hapten formation. In addition, the murine model did not account for ongoing bacterial infection, as is the case for humans receiving flucloxacillin, which leads to increased expression of co-stimulatory molecules through pathogen-associated and danger-associated molecular patterns. A relative decrease in co-stimulatory molecules would be expected to skew toward tolerance. HAS DRUG-SPECIFIC PRE-PRESCRIPTION HLA SCREENING BEEN IMPLEMENTED IN THE UNITED STATES FOR IMMUNE-MEDIATED DRUG REACTIONS? Since 2008, it has been the standard of care to test for the HLA-B*57:01 risk allele before treatment with abacavir for HIV. The success of this discovery through translation into clinical practice for HLA-B*57:01 pre-prescription screening for abacavir was multifaceted and included characteristics of the test, the evidence and generalizability to support pre-prescription testing, and the widespread availability of quality assured and cost-effective single-allele HLA-B*57:01 testing with a rapid turn-around-time.42,49 HLA-B*57:01 is associated with a 55% PPV and a 100% NPV for abacavir hypersensitivity syndrome as demonstrated by Prospective Randomized Evaluation of DNA Screening in a Clinical Trial-1.42 In Prospective Randomized Evaluation of DNA Screening in a Clinical Trial-1, individuals were randomized to receive real-time HLA-B*57:01 screening versus standard of care. Importantly, the PPV and NPV were consistent across diverse populations.49 As a result of this pre-emptive screening strategy, abacavir hypersensitivity syndrome, and false positive clinical diagnosis have been successfully eliminated. While other drug-specific HLA associations have been described, they are hampered by incomplete NPV and PPV, such as for flucloxacillin DILI. While HLA-DRB1*15:01 is the most commonly HLA allele associated with amoxicillin-clavulanate DILI, there are other HLA alleles associated that differ across populations.9 WHAT NON-HLA GENETIC RISK FACTORS HAVE BEEN REPORTED FOR DILI? Non-HLA genetic risk factors that have been reported include protein tyrosine phosphatase nonreceptor type 22 in a genome-wide association study for all-cause DILI, including a large contribution of amoxicillin-clavulanate DILI cases, versus controls.17 Protein tyrosine phosphatase nonreceptor type 22, a phosphatase enzyme that contributes to B-cell and T-cell responses, has also been implicated in the risk of autoimmune diseases, such as rheumatoid arthritis and systemic lupus erythematosus. Endoplasmic aminopeptidase (ERAP) 2, an enzyme that trims peptides before loading onto HLA class I, was shown to have reduced expression in the liver in a transcriptome-wide association study of individuals with amoxicillin-clavulanate DILI compared to controls. Variation in ERAP is not only a risk for drug hypersensitivity reactions but is one factor contributing to the PPV.50 N-acetyltransferase 2 and ATP7B co-occurrence have also been reported to be associated with isoniazid-induced DILI.51 Specifically, the addition of the copper transporter gene ATP7B increases the risk of DILI in the presence of ultraslow acetylator and isoniazid-metabolizing enzyme N-acetyltransferase 2. There is also some evidence to suggest that one underlying mechanism involved includes mitochondrial dysfunction by copper in the presence of isoniazid. IS HLA PRE-PRESCRIPTION TESTING READY FOR USE FOR DILI? WHAT ABOUT COMBINING HLA AND NON-HLA GENETIC RISK FACTORS? At present, due to low NPV and PPV, there is no clinical utility for population-level screening of HLA risk alleles for high-risk DILI medications. In cases where the NPV is 100%, meaning that the specific HLA risk allele is necessary but not sufficient, the number needed to test is prohibitively high. The main factors influencing the number needed to test are the PPV, the frequency of the risk allele in the target population, and the prevalence of the DILI reaction. Other important considerations for the implementation of HLA screening include the cost-effectiveness and ease of screening in clinical practice, the economic consequences of the disease, and the availability of alternative drugs with a more favorable safety margin or that do not have an associated genetic risk. In a study that used a genetic risk score (GRS) for amoxicillin-clavulanate DILI by combining HLA risk alleles (eg, HLA-A*02:01, HLA-DRB1*15:01, HLA-B*15:18) and risk variants in protein tyrosine phosphatase nonreceptor type 22 and ERAP2, individuals in the highest GRS quintile had a predicted more than 1 in 100 risk of developing amoxicillin-clavulanate DILI.18 Individuals with the lowest GRS quintile were predicted to have a 1 in 5000 risk of developing amoxicillin-clavulanate DILI. The estimated baseline risk of amoxicillin-clavulanate DILI is estimated to be 1 in 2350. WHAT IS THE CURRENT CLINICAL UTILITY OF HLA TESTING IN DILI? While there may be a low clinical utility for pre-emptive HLA testing in DILI, drug causality, and adjudication are often confounded by the co-administration of multiple drugs. Therefore, HLA testing for drug-specific risk alleles can be used for diagnosis. Currently, inexpensive and accurate rapid single-allele screening assays are available for HLA-B*57:01, -B*15:02, -B*58:01, -B*13:01, and -A*32:01.46 These rapid single-allele assays can be leveraged to test for a drug-specific HLA risk allele to increase confidence in drug causality. For example, if a patient with DILI was exposed to both flucloxacillin and amoxicillin-clavulanate, the carriage of HLA-B*57:01 would provide strong evidence for flucloxacillin as the causative drug. Caution should be applied when an HLA allele is not well represented in a population. HLA-B*13:01 is predominantly associated with DILI and severe cutaneous adverse reactions in Asian populations,12 while HLA-A*32:01 is primarily associated with DILI26 and drug reaction with eosinophilia and systemic symptoms (DRESS)52 in European populations and the generalizability must be determined. Other important clues may include the presence of DRESS.47 For some medications, symptoms and signs of DRESS commonly accompany DILI and the HLA associated with the DILI reactions appears to be the same as the DRESS reaction (eg, vancomycin and HLA-A*32:01; carbamazepine and HLA-A*31:01; and allopurinol and HLA-B*58:01). WHAT ARE THE FUTURE DIRECTIONS TO BRING PRE-EMPTIVE GENETIC TESTING FOR DILI INTO CLINICAL PRACTICE? To bring pre-emptive genetic testing for DILI into clinical practice, the low PPV needs to be overcome with further study into genetic risk factors that confer tolerance or susceptibility in addition to HLA. Non-HLA genetic risk factors, such as ERAP2 with amoxicillin-clavulanate, may be important across other drug causes of DILI. Further, genetic variation across the antigen processing axis, in addition to ERAP, may be at play and worthy of future investigation for DILI, such as the killer cell immunoglobulin-like receptors. Killer cell immunoglobulin-like receptor, which with and has binding specificity for HLA, is expressed on both killer cells and T cells and Non-HLA genetic risk factors to be are likely to be those that are variants with low effect size or rare variants with high effect In addition, variation in polymorphic immune such as ERAP and killer cell immunoglobulin-like receptors that are linked to HLA may not be on genome-wide association By multiple polymorphic immune through models such as a it may be to PPV and bring the number needed to test to a These genetic must be in diverse populations to their generalizability and clinical utility across Prospective for DILI, such as the United States DILI and for DILI through DNA such as the of be in and both HLA and genetic risk factors across diverse populations to bring for DILI into clinical
Source: Clinical Liver Disease
Language: English
Subfield: Pharmacology
Domain: Life Sciences
Open Access status: green
Funders: AstraZeneca, Regeneron Pharmaceuticals, Novavax, BioCryst, Esperion Therapeutics, National Institutes of Health, Medical Research Council, National Health and Medical Research Council, National Heart, Lung, and Blood Institute, National Human Genome Research Institute, National Institute of Allergy and Infectious Diseases
Awards: G265873141, G4950536974, Genetic risk and long-term outcomes associated with Drug-induced Stevens-Johnson syndrome and toxic epidermal necrolysis in survivors, G6771029017, Clinical and Translational Training Program in Pulmonary Medicine, Immune-mediated adverse drug reactions to HIV and TB treatments in South Africa: predict, prevent and improve long-term outcomes (IMARI SA study)
OpenAlex ID: W4399576744