Acetaminophen hepatoxicity: What do we know, what don't we know, and what do we do next?


  • Neil Kaplowitz M.D.

    Corresponding author
    1. Keck School of Medicine, University of Southern California, Los Angeles, CA
    • Keck School of Medicine, GI/Liver Division, 2011 Zonal Ave., HMR 101, Los Angeles, CA 90033
    Search for more papers by this author
    • fax: 323-442-5425

  • See Articles on Pages 6 and 10.

Acetaminophen (APAP) is a widely used analgesic that can be found alone and in combination with other medications in more than 100 over-the-counter preparations and numerous prescription medications, particularly in combination with opiates. When taken in overdose, either as a suicidal gesture, or with therapeutic intent, severe liver injury and acute liver failure can occur. Despite a massive experimental literature, the detailed understanding of the mechanism of acetaminophen's toxic effect remains in some aspects controversial and in other aspects unclear. A small fraction of a dose of acetaminophen is converted, mainly by CYP2E1, to the electrophilic metabolite N-acetyl-p-quinoneimine. This metabolite is detoxified by preferential binding to glutathione (GSH) in hepatocytes despite the greater abundance of protein-SH, which suggests that the reaction with GSH is enzymatically catalyzed by GSH S-transferases. When sufficient toxic metabolite is generated to deplete nearly completely all the cytosol and mitochondrial GSH, liver injury follows.1 However, controversy exists as to whether the profound GSH depletion itself is toxic by unmasking oxidative stress (i.e., loss of defense against endogenous reactive species even normally produced by mitochondria) or whether the reactive metabolite attacks critical protein thiols (particularly in mitochondria), which leads to loss of function and cell death. Clearly, at the very least, profound GSH depletion and covalent binding are signs of a toxic exposure level of N-acetyl-p-quinoneimine, and this situation precedes and is a necessary prerequisite for the development of toxicity. But GSH depletion and covalent binding may not be sufficient, at least on the basis of studies in animal models, which suggest that the progression and severity of the injury in vivo depends upon the interplay of other factors, both within and outside of hepatocytes.


APAP, acetaminophen; GSH, glutathione; ALF, acute liver failure.

A number of factors determine the dose of APAP required to reach the toxic threshold (full GSH depletion and protein adducts) and include the levels of CYPE1, the baseline levels of GSH in cytosol and mitochondria, the capacity to upregulate the generation of GSH once its consumption begins, and the capacity to sulfate and glucuronidate APAP.1, 2 These factors are under the control of known or potential genetic and environmental influences. Therefore, the dose of acetaminophen that will lead to an individual patient crossing the threshold, as evidenced by covalent binding, is highly variable and unpredictable. Furthermore, adding to the complexity of our understanding of APAP toxicity is the response of the regenerative/repair process as well as the participation of the innate immune system. Table 1 provides a list of experimental circumstances that have lead to increased or decreased susceptibility to APAP in vivo in rodent models. In most of the studies involving cytokines, chemokines, and the innate immune system, susceptibility to liver injury was altered without a change in APAP metabolism, as reflected in the extent of GSH depletion and covalent binding. Thus, aside from a massive bioenergetic catastrophe at huge doses of APAP, the metabolism of APAP in hepatocytes simultaneously leads to sensitization to the toxicity of the innate immune system and activates a “battle” between injurious and protective factors external to hepatocytes. The outcome of the battle depends on genetic and environmental influences on this delicate balance.

Table 1. Biological Modulation of Susceptibility to APAP Toxicity in Experimental Models
Decreased SusceptibilityRef. No.Increased SusceptibilityRef. No.
  1. NOTE. Role of TNF and NO are very controversial with studies suggesting either a protective or aggravating role.

  2. Abbreviations: IFN, interferon; IP, inducible protein; MIP, macrophage inducible protein; IL, interleukin; NK, natural killer; NKT, natural killer T cells.

CAR−/−9GSH depletion21
Gst-Pi−/−12Induction of CYP2E122
IFNγ−/− and anti-IFNγ14IL-10−/−23
FasL deficiency (gld)15IL-6−/−19
Fas deficiency (1 pr and Fas antisense)15, 16Cox 2−/−23
IP-10 and MIP-2 administration17, 18Kupffer cell depletion24
IL-6 administration19CCR2−/− (mcp-1 target)25
IL-11 administration20Anti-MIP-226
NK/NKT cell depletion15Agonistic anti-Fas pretreatment27

Two accompanying “Sounding Board” articles deserve comment. Dr. Lee finds that >50% of acute liver failure(ALF) in his multicenter surveillance study is due to APAP.3 The majority of cases appear to be “unintentional” with a median dose of 34 g (taken over 3 days). Many were taking opiate combinations. Incidentally, older work has suggested that central effects of opiate agonists promote hepatic glutathione depletion4; this issue has not been addressed in humans. Dr. Lee's group has employed a simple and logical approach to confirm APAP toxicity, namely the detection of N-acetyl-p-quinoneimine-cysteine protein adducts derived from plasma proteins after dialysis and proteolysis of the plasma polypeptides.5 The detection of these protein adducts released into the circulation from the injured liver indicates unequivocally that the toxic threshold has been crossed (at least in a subpopulation of hepatocytes) and is therefore a prerequisite for toxicity. This technique can reliably verify the plausibility of APAP toxicity. Interestingly, preliminary results indicated that 20% of indeterminate cases of ALF had these adducts, showing that history is often unreliable (patient or physician?) and that APAP toxicity is probably being significantly underestimated. Also detectable blood levels of APAP were found in some ALF patients with acute viral hepatitis; these patients had worse biochemical parameters. This preliminary finding is extremely provocative in suggesting additive effects and perhaps increased susceptibility to APAP toxicity in viral hepatitis. N-acetyl-p-quinoneimine -protein adduct determination would add credibility to this hypothesis. However, even without detectable adducts, one wonders whether moderate APAP-induced GSH depletion will influence susceptibility of hepatocytes to viral or immune-mediated apoptosis. This area urgently needs more investigation.

Dr. Rumack provides a number of useful points and clarifies a number of misconceptions.6 He notes that the peak APAP levels after therapeutic doses do not exceed 20 ug/mL and it does not accumulate with multiple daily dosing, so when higher levels are seen, the patient must have taken more than the recommended dose. However, one issue is that patients who develop severe injury with “chronic” dosing and continue to ingest multiple doses of APAP in theory could accumulate higher levels due to prolongation of APAP half-life. Dr. Rumack raises the issue of attribution of causality and the fact that the very high aminotransferases characteristic of APAP toxicity can be seen in ischemic and viral hepatitis. However, the first tenet of hepatotoxicity causality assessment is to exclude other causes.7 Therefore, faced with towering aminotransferases and, coagulopathy, initially mild hyperbilirubinemia, a history of APAP use, and exclusion of other causes, one is lead to the inevitable strong suspicion of APAP toxicity. At present this is not an exact science, and causality assessment in the clinical setting is aimed at identifying the most probable etiology. The ability to identify APAP protein adducts in blood promises to have an enormous impact in clarifying this dilemma in cases of suspected APAP hepatotoxicity.

Dr. Rumack addresses the controversy concerning chronic ethanol as a risk factor. He points out that the effect of ethanol on CYP2E1 expression is small and short-lived, and the cases in the literature of ethanol increasing susceptibility to APAP at recommended doses are not well validated, raising doubts as to how often, if ever, ethanol alone leads to toxicity at therapeutic doses. He does not dispute that chronic ethanol worsens toxicity at excessive doses. However, the modest effects of ethanol on CYP2E1 is not the whole story in determining how ethanol may affect the risk or susceptibility to acetaminophen. Dr. Rumack similarly discusses the weaknesses in the evidence that fasting increases susceptibility to acetaminophen. Although I find his comments to be generally valid, one must recognize that the liver injury observed with chronic “overdose” situation appears to have an idiosyncratic nature, and therefore the effects of ethanol or fasting may be of more significance in certain individuals when added to other factors that determine risk.

There is debate about what is unintentional, accidental, and misadventure. This debate to me is a fruitless semantic argument. Many patients develop ALF from APAP without suicidal intent. There seems no doubt that there are many such patients, although we do not know the denominator of patients who consume excess acetaminophen but do not get into trouble. One assumes they are in great excess of those who develop ALF. There certainly are anecdotes of individuals consuming truly massive quantities of APAP without liver injury.8 One wonders whether the situation is analogous to other idiosyncratic toxins. As a rule these agents (e.g., isoniazid and troglitazone) cause mild liver injury in a much larger number of patients than actually develop symptomatic liver disease. The injury is usually transient and resolves with continued use, a phenomenon referred to as “tolerance” or adaptation. Therefore, in some individuals there may be a failure to adapt which permits injury to progress. As an example, in the case of acetaminophen, it is conceivable that the initial GSH depletion and oxidative stress upregulate detoxification. Failure to adapt is known to occur in CAR−/−9 and Nrf-2−/− mice.10 Could polymorphisms in pathways such as these affect the outcome and idiosyncratic nature of the problem?

There are three important questions:

  • Does APAP cause liver injury and ALF when doses do not exceed 4g/day? There are very few well-documented cases in the literature. As pointed out by Dr. Rumack, many of the reported cases have some inaccuracies or inconsistencies. Furthermore, can we confidently rely on the history, particularly in alcoholics and addicts? Nevertheless, there are enough cases out there that it would seem unwise to deny that this never happens, particularly given the myriad risk factors that theoretically can modify APAP metabolism and the response to injury. These very rare cases most likely represent idiosyncratic circumstances. Perhaps alcoholics represent one group in which idiosyncrasy is somewhat potentiated—the final nail in the coffin. Obviously, most alcoholics get away without problems at the maximum therapeutic dose level. However, some may not.

  • What is a maximum safe dose of acetaminophen? The answer to this question has to be influenced by the idiosyncrasy hypothesis. At 4 to 8 g/day most individuals do not seem to get into trouble, but some do, certainly more often than at therapeutic doses. An important point that emerges from the ALF surveillance data is that although many of the patients were taking very large amounts of APAP, half the cases were taking less than 10 to12 g/day. It leaves me with the uncomfortable feeling that exceeding 4 g/day for more than a day or so is simply dangerous because of the idiosyncratic nature of a severe injury at these doses. These data lead to the conclusion that everything possible should be done to educate the public about avoiding excessive dosing under all circumstances. This education should include media advertising; warning labels; making it clear that Tylenol = acetaminophen; clearly identifying that combination preparations contain acetaminophen; and stating explicitly in advertising and warnings in an unambiguous fashion that there is a danger of severe liver injury when the daily dose of acetaminophen exceeds 4 g/day.

  • Should the FDA withdraw APAP from the market? Given what we know about the nonsuicidal chronic overdosing leading to acute liver failure, this is a valid question. Unfortunately the alternative, nonsteroidal anti-inflammatory drugs, have their own inherent problems unrelated to liver failure, such as ulcers, gastrointestinal bleeding, and renal damage.

So we're left with a “pick your poison” dilemma until a safer analgesic/antipyretic is available. In the meantime, rare cases of acute liver failure probably will occur at therapeutic doses, and many cases will occur in an idiosyncratic fashion in the nonsuicidal setting due to moderate-to-severe excess dosing. Because of the sheer number of individuals consuming excess acetaminophen, the actual number of cases are substantial, alone representing the major cause of ALF.

It is interesting to contrast the APAP situation to troglitazone. About 2 million patients took troglitazone, and about 100 of them developed ALF (i.e., 1:20,000).11 This effect was grounds for withdrawal of the drug, based upon risk-benefit analysis. Far more cases, in absolute number, occur due to nonsuicidal “chronic” APAP, which, however, has an undefined denominator, so we do not know how high or low is the risk. The fact that a handful, scores, or even hundreds of study subjects withstand 4 to 8 g/day does not address the idiosyncratic situation. APAP is a drug that has the propensity to damage the liver, and even though this damage happens in a small percentage of users who exceed recommended doses, every effort needs to be implemented for it to be avoided. Certainly public education and labeling are well worth trying. If this action fails to decrease the incidence of ALF due to nonsuicidal toxicity, the next step, in my opinion, is to eliminate acetaminophen from all combinations, including over-the-counter preparations and opiates; doing so will allow the more controlled, defined use of acetaminophen by patients without limiting the benefits of adding acetaminophen separately to other drugs. If all these measures fail, even more drastic action may have to be considered.

Nonsuicidal APAP-induced hepatotoxicity has emerged as the major cause of acute liver failure. Most cases appear to be associated with variable level of overdose. However, the safety of this drug and its toxic threshold are influenced by a large number of factors that are potentially modified by genetics and the environment. More effort is required by the manufacturers, the FDA, and physicians to prevent overdosing, and more work is required in humans to identify the risk factors for what appears to be in idiosyncratic toxicity in the nonsuicidal setting. At present, I am not certain as to what is a perfectly safe dose of acetaminophen, especially when taken for more than a day or two.