Acetaminophen is a widely used analgesic that has the propensity to be hepatotoxic, with a very small percentage of affected patients exhibiting the most severe manifestation in the form of acute liver failure. Acetaminophen itself is not toxic, and cellular injury is caused by its unstable metabolite, N-acetyl-p-benzoquinone imine (NAPQI), which normally is present in small amounts and is rapidly neutralized by conjugation with glutathione. Toxic levels of NAPQI accumulate when large amounts of substrate are available for metabolism or the metabolism is accelerated by enzyme induction, as in individuals who regularly consume alcohol or medications that cause enzyme induction, e.g., antiepileptic therapy.1 With these scenarios, the hepatic pool of glutathione is depleted, permitting the accumulation of NAPQI and subsequent hepatotoxicity. Possible additional risk factors that lower the threshold for hepatotoxicity have been identified and include fasting and malnutrition.2
The classical model of acetaminophen hepatotoxicity clearly was temporally related to the ingestion of amounts of acetaminophen that exceeded the maximum recommended daily dose of 4 g in adults. In the United Kingdom this was most frequently encountered in the context of self-harm, and only 8% of cases were linked to the therapeutic use of acetaminophen.3 Data from the United States suggested that a higher proportion of patients with acetaminophen-induced hepatotoxicity did not intend to inflict self-harm.4 Liver injury was attributed to ignorance of, or failure to comply with, the dosing limitation, or lack of realization that acetaminophen was a constituent of different proprietary preparations.
In the current issue of HEPATOLOGY, the US Acute Liver Failure group presents the findings of a prospective study suggesting that at least 48%, and possibly as many as 56%, of patients with acetaminophen-induced acute liver failure did not intend self-harm.5 The design of the study did not restrict the diagnosis of acetaminophen-induced acute liver failure to the classical models described above but sought to define the role of acetaminophen by adopting a broader set of diagnostic criteria. Thus, patients with any detectable acetaminophen levels in blood or an aminotransferase level greater than 1,000 IU/L in the context of using acetaminophen within the preceding 7 days were considered to have acetaminophen-induced acute liver failure if no other candidate etiology was identified. Only 40% of the patients fulfilled a more conventional definition of having a clear history of taking acetaminophen in excess, having acetaminophen detectable in blood and having markedly raised aminotransferases.
The adoption of these broader criteria almost certainly resulted in the inclusion of some cases that were not truly related to acetaminophen. It is stretching credibility to suggest that a patient developed acute liver failure after taking 1.2 g of acetaminophen, which is barely above a single therapeutic dose. Patients with seronegative or indeterminate hepatitis who took some acetaminophen for symptom relief would be inappropriately designated as having acetaminophen-induced acute liver failure with this approach. In this context, at least one patient in this study had symptoms for 32 days before presenting with liver failure. Furthermore, the findings in this study are not supported by any histological evidence; for example, from the explanted livers of transplant recipients. Despite these concerns, the credibility of these findings is significantly enhanced by the remarkable similarity in many key respects between the typical patients with acknowledged overdoses and the group whose disease was attributed to the therapeutic use or misuse of acetaminophen. The latter group was a little older, but only by about 6 years, and had lower acetaminophen levels and serum aminotransferases. However, in terms of other demographics, natural history, and outcome there were no discernible differences between the two groups.
This study establishes the need for greater circumspection with regard to the possible role of acetaminophen in the causation of acute liver failure. The data suggest the burden of proof should be lower than previously required and it is appropriate to loosen diagnostic criteria, although exactly by how much is a moot point. Interestingly, the authors concluded that the patients with a history of therapeutic use are not representing a variant of acetaminophen liver injury with a more protracted pathogenesis. Instead, they felt that for currently unknown reasons a point is reached where these patients acutely become susceptible to liver injury. The possibility of dual pathology was not addressed but should be considered given the likelihood that acetaminophen will be used therapeutically in patients with viral illnesses. Acetaminophen has been implicated as a significant co-factor in the pathogenesis of acute liver failure in patients with acute hepatitis B and in patients taking antituberculous therapy.6, 7 The authors of the former study stressed the importance of glutathione depletion, whereas the authors of the latter study concluded that the routine use of n-acetylcysteine in that setting was legitimate. It is probably reasonable to extend this conclusion to all etiologies of acute liver failure when there is any evidence of acetaminophen exposure pending the conclusion of the trial currently being conducted by the US Acute Liver Failure group.
The determination of prognosis in acetaminophen liver injury is challenging and must be done by a process substantially different from other etiologies of acute liver failure. Rigid extrapolation from the latter experience would result in a gross overestimate of the risk of mortality. Serum aminotransferase levels typically are very high and are of no prognostic value either in absolute levels or trends. Likewise, a coagulopathy in isolation, no matter how severe, has never been shown to be predictive of outcome in this setting. It is not uncommon to encounter prothrombin times greater than 100 seconds or international normalized ratios greater than 10 on the third day after an acetaminophen overdose, which then fall rapidly, with the patients recovering without showing any clinical signs of encephalopathy. The phenomenon of the rapid reversal of a coagulopathy induced by liver injury on this scale is unique to acetaminophen. An intriguing clue to the basis for early and effective regeneration was recently provided by the observation in a mouse model that not only was the oval cell, a putative liver progenitor cell, resistant to acetaminophen toxicity, but the first surge seen in oval cell populations was accelerated by higher levels of exposure to acetaminophen.8
Failure of coagulopathy to improve on the fourth day after drug ingestion was clearly associated with a poor prognosis.9 Prognosis worsens with the development of progressive encephalopathy, but even then outcome is good unless the additional complications of renal failure and cerebral edema develop.10 Conversely, the presence of a severe metabolic acidosis was associated with a dreadful prognosis even in the absence of clinical encephalopathy.10
The King's College Hospital criteria attempted to capture these prognostic indicators as quickly as possible and with the greatest possible level of discrimination between survivors and nonsurvivors.11 The first component of the criteria, an arterial pH < 7.30, achieved these objectives but identified only about one third of patients who died. The second component was less effective as a poor prognosis and was determined quite late when advanced encephalopathy, renal failure, and a severe coagulopathy co-existed. The combined criteria yielded a high level of specificity but suboptimal sensitivity, demonstrated by the fact that 17% of patients who had a poor prognosis were not identified as being at risk of death.11 However, tinkering with the model to improve sensitivity rapidly led to a reduction in its specificity. Subsequent experience has lowered the pH threshold determining a poor prognosis to 7.25 to reflect the clinical benefit of more aggressive early resuscitation and rehydration.
The King's College Hospital criteria were deemed to have performed poorly in this study. In defense of the criteria, they were inappropriately evaluated, as they were only applied at the point of admission of the study and serial re-evaluation of individual patients as they progressed through their illnesses was not assessed. One practical reason for this may have been the practice of giving these patients prophylactic fresh frozen plasma, which is common practice in the United States but which is avoided in the United Kingdom. Therefore, a more accurate comment on the performance of the King's College Hospital criteria would be that they performed poorly at the single timepoint of evaluation, before their utility was lost. Nevertheless, concerns about the validity of the rationale used to reach the conclusion that new prognostic models are required in acetaminophen-induced acute liver failure do not extend to the conclusion itself. It is remarkable how little progress has been made in almost two decades given the shortcomings of the King's College Hospital criteria in acetaminophen-induced acute liver failure, although the full impact of more recent observations relating to serum lactate and phosphate have yet to be determined.12–14
The relevance of effective prognostic criteria mainly is due to the application of liver transplantation in the management of the most severe cases of acetaminophen-induced liver failure. The need for accuracy is self-evident, given the desired objectives of avoiding unnecessary transplantation and offering all those in need the opportunity to benefit from a liver graft. Although the current study is not explicit about the criteria used for listing patients for transplantation, it illustrates the potential scale of the problem regarding unnecessary transplants, as 59% of patients listed for transplantation but who were not allocated an organ survived. While imperfect prognostication and limited access to organs remain highly pertinent issues, there will continue to be a debate about how best to apply this component of the management strategy. This current study potentially introduces an additional complexity: should the patient suffering from unintentional hepatotoxicity be salvaged at all costs while liver transplantation is selectively offered to patients with intentional liver injury? Let the wisdom of Solomon prevail!
Measures to minimize acetaminophen hepatotoxicity are important but need to be considered in the context that the apparent scale of the problem is a reflection of the huge number of patients taking acetaminophen with good effects and in the absence of any adverse event. Educational initiatives to highlight the range of preparations containing acetaminophen, together with reiteration of advice on maximum daily dosing, have potential benefits, especially with respect to unintentional overdosing. Restriction of access to acetaminophen is an alternative, and this approach was taken in the United Kingdom in 1998, when over-the-counter sales of acetaminophen were restricted to 16 g. In the 4 years following the change in legislation there was a 30% reduction in patients with severe acetaminophen-induced acute liver failure admitted to specialist liver units and liver transplant centers.15 More severe restrictions operate in France (8-g maximum packet size), and this measure is highly effective in minimizing severe acetaminophen hepatotoxicity.16 The required judgment from society, the medical profession, and other interested parties, is whether that level of restriction is too high a price to pay.