Drug-induced liver injury (DILI) can mimic all forms of acute and chronic liver disease but the rare occurrence of acute liver failure particularly has put the onus on industry and regulatory bodies to protect the general public. Consequently, identifying hepatotoxic potential prior to approval and marketing has assumed a critical role. The traditional approach has been to monitor for alanine aminotransferase (ALT) increases in clinical trials. Although this has proven to be effective in identifying a toxic potential, considerable financial investment already has been made in reaching this stage of drug development. Therefore, it would be beneficial in compound selection or for heightened vigilance in developing specific agents to identify which chemical entities are entirely safe and which have potential liability before ever reaching a human subject. Administration of drugs at high doses to several animal species for varying durations, seeking pathologic changes, is a requirement but as often as not fails to identify the risk of liver injury for drugs that reach man. The reasons include differences in metabolic pathways of drug handling and the current lack of suitable animal models that reproduce the human risk factors. Nevertheless, animal testing does successfully identify many highly toxic chemicals that are similar to acetaminophen in directly injuring the liver.

With the exception of acetaminophen, most drugs that are known to cause DILI in man do so in an unpredictable fashion with a relatively long latency in a small proportion of exposed individuals and this is referred to as idiosyncratic DILI (IDILI), reflecting the unique susceptibility of certain individuals. Recent evidence has strongly implicated adaptive immunity determined by genetic polymorphisms in major histocompatibility complex (MHC) Class I and II genes in the pathogenesis of IDILI.1 These associations have been described with flucloxacillin, ximelagatran, ticoplidine, lumiracoxib, lapatanib, and amoxiclillin-clavulanate. The absence of the specific haplotype strongly predicts that no DILI will occur, whereas the presence of the common genetic variant is a poor predictor, indicating that it is necessary but not sufficient to lead to IDILI. The big question currently facing this field is what else other than human leukocyte antigen (HLA) contributes to the pathogenesis of IDILI and is it related to the drugs' chemistry or how the patient handles the drug, or both.

In the current issue, Chen et al.2 provide evidence that the chemical entity itself, particularly with respect to dose and its physiochemical nature (lipophilicity), is a key factor. They assessed the predictive value of dose (≥100 mg per day) and calculated octanol-water partition coefficient (logP > 3) in two independent databases of Food and Drug Administration (FDA)-approved drugs labeled for the presence or absence of liver injury. The present study confirms previous studies that suggested both factors separately could predict hepatotoxicity3-7 and suggests that a “rule-of-two” which combines both dose and lipophilicity performs better than dose alone, increasing the positive predictive value of dose alone from 85% to 96% (“rule-of-two”) while decreasing the negative predictive value from 55% to 39%. Whereas only 8 of 114 drugs in the two databases with no DILI concern exhibited positive “rule-of-two,” only half of the hepatotoxic drugs were positive. Thus, false-positives were low but false-negatives were substantial. Clearly, the “rule-of-two” is far from perfect and cannot replace preclinical testing but could be useful as an additional guide in compound selection during drug development. One interesting clinical application of the “rule-of-two” was illustrated by performance in six cases of DILI from LiverTox who received multiple medications; in five cases the implicated drug was the only one exhibiting a positive “rule-of-two.” This suggests that application of “rule-of-two” or a facsimile may improve causality assessment in the setting of multiple medications.

Why should dose and lipophilicity be of predictive value? Likely this is because of the need for the liver to be exposed to a threshold level of the parent drug and/or reactive metabolite. Lipophilic drugs are cleared by the liver and generally require biotransformation to be eliminated. As noted by the authors, a significant relationship was observed between the extent of hepatic metabolism and logP (calculated-water partition coefficient). Therefore, logP may simply be a surrogate for extensive biotransformation and hepatic exposure to a reactive metabolite. Indeed, others have shown the increased predictive value of combining dose with information on hepatic metabolism.8

Although dose and lipophilicity both may play roles as determinants of hepatic risk through exposure of hepatocytes to some critical threshold of drug and/or metabolite, how can we link this to the ultimate development of IDILI? Across the pharmaceutical industry, systems of screening drug candidates have emerged that include transcriptomic profiling of animals in addition to animal pathology, assessment of covalent binding and glutathione (GSH) adducts in microsomal test systems and in vivo, inhibition of bile salt export pump (BSEP) in vitro, impairment of function of isolated animal mitochondria, and cell stress responses and viability in human hepatoma and hepatocyte culture systems. Several common themes emerge in all these test systems especially involving oxidative stress, mitochondrial impairment, covalent binding, and endoplasmic reticulum (ER) stress. Remarkably, all of these test systems seem to have moderately strong predictive value for IDILI.9-13 A unifying picture emerges in which IDILI drugs at high doses used in preclinical studies seem to generate a hazard in hepatocytes and their organelles, which itself may or may not be lethal (Fig. 1). One could argue that all these changes simply reflect the fact that drugs which cause IDILI are likely to be metabolized in hepatocytes and induce covalent-bound haptens that elicit an adaptive immune response in genetically susceptible individuals with the relevant HLA haplotype. The hazards observed in various test systems may simply be a surrogate for reactive immunogenic metabolites. The alternative view is that the hazards may actually contribute to the development of IDILI. Certainly in situations in which IDILI is not mediated by adaptive immunity, one can envision the gradual accumulation of hazard-induced damage to mitochondria reaching some critical threshold for injury, as exemplified by the rapid development of damage due to acetaminophen or the progressively longer latency seen with valproate, nucleosides, and amiodarone. However, few other examples can be identified and, as noted above, the preponderance of evidence has recently redirected the field towards the adaptive immune system. The hazards induced by exposure of hepatocytes may still be extremely relevant. Hepatocyte stress may generate danger signals that costimulate the development of an adaptive immune response directed at haptenized peptides14 or neoantigens such as hapten-free peptides misdirected to the wrong HLA molecule.15 Furthermore, the stress induced by hepatocyte exposure to the hazard of reactive metabolites may sensitize hepatocytes to T-cell or cytokine-mediated killing, unmasking or worsening immune-mediated killing. This is exemplified by drug-induced redox perturbations and oxidative stress sensitizing to tumor necrosis factor (TNF)-induced nuclear factor kappa B (NF-κB) survival pathways at various steps from IKK to p50 p65-mediated transcription.16, 17 Another key feature of IDILI is the phenomenon of adaptation (Fig. 1). Nearly all IDILI drugs that rarely induce severe acute liver injury much more frequently induce asymptomatic anicteric injury that disappears with continued drug administration. Again, there are two ways to look at this: first, adaptation reflects the development of immune tolerance. The example of lumiracoxib hepatotoxicity demonstrated that even the subgroup with the mild transient ALT increases (adaptors) exhibited the same HLA haplotype risk factor as those with severe injury,18 so that one could speculate that those subjects who progressed to overt liver injury (Hy's law) failed to develop immune tolerance. Second, adaptation may reflect the dampening of the hazard induced by drugs. A number of possibilities can be considered including decreased exposure to the toxic moiety, enhanced phase 2 or 3 defense (e.g., GSH, antioxidant enzymes, nontoxic metabolism, or MRP2 and other adenosine triphosphate [ATP] dependent export pump expression), unfolded protein responses in the ER or mitochondria, mitochondrial adaptive responses such as mitophagy or biogenesis, and regeneration with proliferation of resistant hepatocytes. Further adding to the complexity of susceptibly to IDILI is the modulation of these adaptive responses by genetic variations and environmental factors.

thumbnail image

Figure 1. Model of the pathogenesis of IDILI. Hepatocyte exposure to critical levels of a reactive metabolite concomitantly leads to hapten formation and a hazardous effect on hepatocytes. Both may be needed for IDILI to occur. The adaptive responses to the hazard as well as the development of immune tolerance may lead to dissipation of the injury or its not occurring at all. Conversely, impaired adaptation may contribute to progression to overt injury.

Download figure to PowerPoint

In conclusion, the “rule-of-two” seems to offer some added value in identifying the potential for IDILI. Aside from its application in candidate selection in drug development by industry, it may be worth exploring as a component of future causality assessment tools especially when weighing the contribution of multiple concomitant medications. The strong relationship between dose, lipophilicity, and IDILI underscores the key role of exposure of the liver to a threshold level of hazardous chemicals.


  1. Top of page
  2. References
  • 1
    Kaplowitz N. Drug-Induced Liver Injury: Introduction and Overview in Drug-Induced Liver Disease. Kaplowitz N, DeLeve L, eds. New York: Elsevier; 2013, Chapter 1, p 3-14.
  • 2
    Chen M, Borlak J, Tong W. High lipophilicity and high daily dose of oral medication are associated with signicant risk for drug-induced liver injury. HEPATOLOGY 2013; 58: 388-396.
  • 3
    Moylan CA, Suzuki A, Papay JI, Yuen NA, Ames M, Hunt CM. A pre-marketing ALT signal predicts post-marketing liver safety. Regul Toxicol Pharmacol 2012; 63: 433-439.
  • 4
    Lammert C, Einarsson S, Saha C, Niklasson A, Bjornsson E, Chalasani N. Relationship between daily dose of oral medications and idiosyncratic drug-induced liver injury: search for signals. HEPATOLOGY 2008; 47: 2003-2009.
  • 5
    Hughes JD, Blagg J, Price DA, Bailey S, Decrescenzo GA, Devraj RV, et al. Physiochemical drug properties associated with in vivo toxicological outcomes. Bioorg Med Chem Lett 2008; 18: 4872-4875.
  • 6
    Peters JU, Schnider P, Mattei P, Kansy M. Pharmacological promiscuity: dependence on compound properties and target specificity in a set of recent Roche compounds. Chem Med Chem 2009; 4: 680-686.
  • 7
    Waring MJ. Defining optimum lipophilicity and molecular weight ranges for drug candidates. Molecular weight dependent lower logD limits based on permeability. Bioorg Med Chem Lett 2009; 19: 2844-2851.
  • 8
    Lammert C, Bjornsson E, Niklasson A, Chalasani N. Oral medications with significant hepatic metabolism at higher risk for hepatic adverse events. HEPATOLOGY 2010; 51: 615-20.
  • 9
    Khetani S, Kanchagar C, Krzyzewski C, Krzyzewski S, Moore A, Shi J, et al. Micropatterned co-cultures increase the ability to detect compounds that cause drug induced liver injury in humans over sandwich cultures. Toxicol Sci [published online] 2012.
  • 10
    Porceddu M, Buron N, Roussel C, Labbe G, Fromenty B, Borgne_Sanchez A. Prediction of liver injury induced by chemicals in human with a multiparametric assay on isolated mouse liver mitochondria. Toxicol Sci 2012; 129: 332-345.
  • 11
    Morgan RE, Trauner M, van Staden CJ, Lee PH, Ramachandran B, Eschenberg M, et al. Interference with bile salt export pump function is a susceptibility factor for human liver injury in drug development. Toxicol Sci 2010; 118: 485-500.
  • 12
    Dawson S, Stahl S, Paul N, Barber J, Kenna JG. In vitro inhibition of the bile salt export pump correlates with risk of cholestatic drug-induced liver injury in humans. Drug Metab Dispos 2012; 40: 130-138.
  • 13
    Park BK, Boobis A, Clarke S, Goldring CE, Jones D, Kenna JG, et al. Managing the challenge of chemically reactive metabolites in drug development. Nat Rev Drug Discov 2011; 10: 292-306.
  • 14
    Monshi M, Faulkner L, Gibson A, Jenkins RE, Farrell J, Earnshaw CJ, et al. HLA-B*57:01-restricted activation of drug-specific T-cells provides the immunological basis for flucloxacillin-induced liver injury. HEPATOLOGY 2012 [Epub ahead of print].
  • 15
    Norcross M, Luo S, Lu L, Boyne MT, Gomarteli M, Rennels AD, et al. Abacavir induces loading of novel self-peptides into HLA-B*57:01: an auto immune model for HLA-associated drug hypersensitivity. AIDS 2012; 26: F21-F29.
  • 16
    Han D, Hanawa N, Saberi B, Kaplowitz N. Hydrogen peroxide and redox modulation sensitize primary mouse hepatocytes to TNF-induced apoptosis. Free Radic Biol Med 2006; 41: 627-639.
  • 17
    Lou H, Kaplowitz N. Glutathione depletion down-regulates TNFα-induced NF-kB activity via IKK-dependent and independent mechanisms. J Biol Chem 2007; 282: 29470-29481.
  • 18
    Singer JB, Lewitzky S, Leroy E, Yang F, Zhao X, Klickstein L, et al. A genome-wide study identifies HLA alleles associated with lumiracoxib-related liver injury. Nat Genet 2010; 42: 711-714.