Acetaminophen misconceptions

Abstract

Examination of the pharmacokinetics of acetaminophen can decrease misconceptions involved in clinical evaluation. Enzyme patterns and acetaminophen levels must be related to time and known metabolic phenomena. A careful look at ethanol and nutrition, especially fasting demonstrates that therapeutic doses of acetaminophen do not place patients at a greater risk in either of these instances. An overdose of acetaminophen in a chronic alcohol abuser may result in more severe hepatotoxicity than in the nonalcoholic. CYP2E1 and glutathione must be evaluated simultaneously rather than in isolation. Glucuronidation capacity in humans is not a factor except in massively overdosed patients. (HEPATOLOGY 2004;40:10–15.)

Acetaminophen (APAP) has been the subject of more than 30,000 articles in the medical literature since 1966. The metabolism of this drug to a toxic intermediate was described in 1973 and has been reported in detail.1 Our report in 1975 indicated that overdose toxicity was not unique to the United Kingdom but also occurred in the United States.2 Despite continued evolution over the last 35 years in our understanding of APAP toxicity and its interaction in the human, misconceptions continue to appear in reviews, case reports, and articles.

APAP, acetaminophen; NAC, N-acetylcysteine; GSH, glutathione.

Pharmacokinetics

As with most drugs the dose related to time along with calculated or known metabolic constants are needed to interpret individual levels. APAP has an approximate absorption T ½ of 30 minutes and a normal metabolic T ½ of about 2 hours and a volume of distribution of approximately 1 L/kg. Given a dosing pattern of 1 g every 4 hours for 4 doses in a 50-kg and 70-kg patient with 8 hours of sleep, the levels attained are shown in Figure 1. If the doses are spread out over 24 hours rather than including a sleep period, the peak levels attained are a little lower. Levels reported as therapeutic range from 10 to 30 μg/mL, although most laboratories do not specify a time relationship, and stand-alone levels may be difficult to interpret without pharmokinetic calculations.

Figure 1. Pharmacokinetics plot by Barry H. Rumack, M.D. and Daniel A. Spyker, M.D., Ph.D. (see text for details).

Following ingestion of an overdose, the initial pharmacokinetics remain the same as therapeutic until such time as the toxic events change the metabolic patterns. Interpretation of levels related to time has become the standard method of determining which patients are at risk for hepatotoxicity and therefore require treatment in advance of evidence of toxicity.2 A second treatment line 25% lower than the original line was added in 1976 at the request of the FDA for use in the national multicenter study.3 A treatment outcome nomogram utilizing both the original “200” and 25% lower study safety “150” lines (Fig. 2) was recently published and shows the likelihood of developing hepatotoxicity following a single acute overdose (all APAP consumed in 8 hours or less) while receiving N-acetylcysteine (NAC) treatment in 2,540 patients.1 The most important observation from this outcome nomogram is that a change in historical timing of an hour or two can make a huge difference in which segment the patient is classified. If there is any question as to the history, and the patient is anywhere close to the treatment line, then therapy with NAC is indicated.

Figure 2. Outcome Nomogram with the original nomogram line and the 25% safety line added during the nationwide NAC study. Patients were plotted at the point of their initial APAP level. Percent is the number of cases with AST greater than 1,000 IU/L at any time during their course.

The recent availability of intravenous NAC (Acetadote) may make early treatment more available than the less-than-palatable oral NAC. The incidence of side effects with the intravenous NAC and the potential ineffectiveness of the shortened protocol may be an issue with this preparation. The product insert suggests the 20 hour protocol rather than the 72 hour protocol, since the safety and efficacy studies were done outside of the United States (http://www.fda.gov/cder/foi/label/2004/21539_N-acetylcysteine_lbl.pdf). As in all cases of overdose, patients should be treated with the longer U.S. protocol as long as there is an elevation of aspartate aminotransferase or APAP. One small study showed that patients in whom NAC was terminated early developed a rise in their enzymes.4

As toxicity develops, the T ½ becomes prolonged and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) enzymes rise and fall as shown (Fig. 3).5 The change in T ½ occurs gradually so that measurement of the T ½ is not as useful as a predictor in the early period following ingestion but does provide confirmatory data later in the course.

Figure 3. Pattern of Enzyme Elevation following a single APAP overdose at time zero. Solid line represents AST in untreated cases. Dotted line represents AST in cases treated with NAC. Dashed line represents AST in non-overdose normal patients. Dot and dash line represents AST in cases progressing to fulminant hepatic failure. Single dot indicates a spurious AST as it cannot rise that rapidly after overdose.

In some case reports and series, investigators misunderstand basic pharmacokinetics, as they assume that the T ½ would become prolonged more rapidly than is pharmacologically possible. It is critical to examine a level in relationship to time, and if the history is unknown, then other factors must be used to determine ingested quantities. In some reports the total body burden (× volume of distribution × weight) is greater than the reported dose. Other errors include the lack of time between an ingestion of APAP and peak enzyme levels, which is impossible. Enzymes generally follow the pattern shown in Figure 3.

Calculation of body burden is sometimes helpful to the clinician when there is a paucity of other data and/or to determine the veracity of the history. A level of 75 μg/mL in an 80-kg patient indicates that there is a 6-g body burden. If it has been 12 hours since the last dose, then the amount ingested is clearly much higher than 6 g. If the half-life is prolonged, then toxicity could not have started 12 hours earlier but must date back a considerable amount of time. If enzymes are elevated, then once again the toxic event must have occurred considerably earlier than 12 hours prior. Thus, in addition to looking at the data in a patient, kinetic and metabolic factors must be considered prior to drawing any conclusion as to the accuracy of the history.

Hepatic Enzymes

High levels of aspartate aminotransferase and alanine aminotransferase are not themselves pathognomonic of APAP toxicity, although this has been reported in some literature. Similar elevations of enzymes are seen in viral hepatitis, hypoxia, obstructive sleep apnea, heat stroke, anorexia nervosa, Epstein-Barr virus, etc.6–10 It does not meet any of the basic tenets of cause and effect to label a patient APAP hepatotoxic with elevated enzymes and no corresponding APAP level or other confirmatory evidence. Patients repeatedly demonstrate a pattern shown in Figure 3.

Induction

Ethanol.

A question to be addressed in this brief review is whether therapeutic doses of APAP in a patient consuming ethanol leads to any increased risk in toxicity. A review of over 2,000 cases in the literature indicates that the only reports linking ethanol and enhanced toxicity were retrospective reviews and case reports with no such link found in any patients involved in a prospective evaluation or class I literature.11 The only prospective double-blind study demonstrated no toxicity, even when APAP was administered in the most susceptible time.12

Data from the largest series of APAP cases (11,195) showed that toxicity was related only to those patients utilizing ethanol who were in the high-risk group with APAP levels above the 300-mg/mL study line.13 To put this information into relative perspective, one alcohol calculator indicates that a 160-pound patient will reach 97 mg/dL of ethanol after 5 standard drinks in one hour. (http://www2.uchsc.edu/pharm/arc_misc/etoh_show.asp#). Although over-the-counter product labeling contains a warning about more than 3 drinks' being consumed in a day, there is no evidence that this level of consumption produces any significant induction of CYP2E1. It is well known that acetaminophen ingested while there is still ethanol in the tissue is effectively blocked from producing the toxic metabolite.14

In humans it has been shown that the administration of APAP during the most susceptible time period requires a significant and continuous dose of ethanol well in excess of 3 drinks per day.14 The effect is dose related, with levels of ethanol below 250 mg/dL producing an increase in activity through ligand formation and enzyme stabilization. At levels above 250 mg/dL there appears to be increased synthesis of CYP2E1.15–18

When APAP is administered during the most susceptible time to alcoholics, there can be an increase of 22% in the mean formation of the toxic metabolite.14 This effect would result in a reduction of the overdose amount required for toxicity from 15.9 g to 13 g. Even if there were a twofold increase, the dose of APAP required for threshold toxicity would be about 8 g at a single time, well above a total daily dose of 4 g in divided doses.19–20

The only appropriate conclusions are that chronic heavy drinkers are at greater risk following overdose of APAP than nonheavy drinkers and that there is no change in risk in either group when therapeutic doses of APAP are consumed.

Phenytoin.

This drug does not enhance acetaminophen toxicity, although it is reported to do so erroneously even in recent literature. Over 20 years ago it was shown that phenytoin increases the glucuronidation component of APAP metabolism without increasing the mercapturate (toxic) metabolic component. Phenytoin is primarily metabolized by CYP3A4, which provides an insignificant amount of APAP metabolism. Confusion in the literature appears related to earlier studies done in vitro versus more recent studies done in vivo.21 Phenytoin does not induce CYP2E1 but may in fact be hepato-protective in APAP overdose due to enhanced glucuronidation. Such human marker studies provide far more convincing evidence than in vitro or animal data.

Phenobarbital.

This drug also does not induce CYP2E1 but is rather a pleiotropic inducer of both phase I and phase II reactions. Unfortunately, when work was done in this area in the past it was not possible to look at individual isoenzymes. Consequently it was assumed that such an increase in metabolism meant that all drugs metabolized via P450 were induced. While phenobarbital does induce CYP2B and CYP2C, as well as a wide variety of other enzymes, it does not have anything to do with any of the steps of APAP-toxic metabolites.

Nutrition and Metabolic Components

Glucuronide, sulfate, CYP2E1, and glutathione (GSH) all play a role in the metabolism of acetaminophen. While some experimental and observational work has looked at each one, little has been done to observe them simultaneously. In terms of the toxic metabolic pathway and detoxification mechanisms, induction or inhibition of CYP2E1 or decrease or increase in GSH and its turnover must be examined simultaneously. Most conditions that result in diminution of GSH also result in changes in CYP2E1. This effect is not unexpected, considering the synthesis of GSH containing three amino acids versus CYP2E1 protein containing 493 amino acids. It is the relationship of this tripeptide and protein that must be examined before interpretation of either one can be made. An assumption in some literature has been that if a patient has poor metabolic status, then only glutathione will decrease and therefore there will be an increase in hepatotoxicity. In fact, in true protein calorie malnutrition, for example, both CYP2E1 and glutathione decrease, resulting in no change in toxicity.

Seemingly discrepant data from animal and in vitro and in vivo human work will likely be resolved in the near future primarily through the use of markers in intact humans that should decrease the difficulty of extrapolating data from animals to humans.22 The changes in knowledge regarding phenytoin, reviewed above, are a good example, and until resolution, greater weight should be placed on those studies with adequate design done in humans.

Earlier human observations indicated that in a variety of hepatic diseases, P450 either stayed the same or decreased, and levels of GSH primarily increased.1, 23–26 These observations treated all of the P450 isoenzymes together and did not account for changes in individual isoenzymes. More recent work has been able to demonstrate differences in some isoenzymes and provide greater specificity.

GSH and CYP2E1.

In a recent human study in patients with alcoholic hepatitis, both GSH and CYP2E1 messenger RNA were reduced by 70% to 80% apparently due to malnutrition.22 CYP2E1 messenger RNA was reduced by 18% and correlated with CYP2E1 protein in patients with hepatocellular disease.27 In contrast to the human, the rat shows induction of CYP2E1 and depletion of mitochondrial GSH after 6 weeks of ethanol treatment but only with APAP doses in overdose quantities.28 Maximum induction in rats showed a doubling of CYP2E1 activity that resolved in 17 hours after ethanol withdrawal and reduction in mitochondrial GSH of 51% that recovered to normal in 17 hours, leading the authors to conclude that the time window for both mechanisms to act in concert is narrow.29 Rats demonstrate increased levels of CYP2E1 following short-term fasting and obesity, but this increase does not appear to occur humans.30–31 Humans with alcoholic liver disease show an increase in CYP2E1 metabolic rates in ethanol-induced patients and a decrease in CYP2E1 with progressive severity of alcoholic liver disease as measured by chlorzoxazone.32 Some of the discrepancies in human data may be explained by the reduction of CYP2E1 in patients with cholestatic forms of cirrhosis compared to unchanged and not increased levels of CYP2E1 in patients with noncholestatic (e.g., ethanol) related cirrhosis.33 Further work by this same group on human liver demonstrates reduction of hepatic CYP2E1 of 13% to 26% depending upon the degree of nutritional status but independent of underlying liver disease.34 In human in vitro cells that overexpress CYP2E1, elevated GSH levels were also demonstrated as an apparent protective mechanism.35

Early work in humans indicated that chronic alcoholics might be more susceptible to APAP because of decreased availability of GSH.36 However, 14 years later the same group has concluded that low GSH in patients with human immunodeficiency virus, patients with chronic hepatitis C, malnourished patients, and patients with cirrhosis does not put these patients at a higher risk with APAP.

Patients with anorexia nervosa demonstrate a 30% reduction in GSH compared to controls.38 Although CYP2E1 was not measured simultaneously with GSH in these patients, other studies in humans and rats have indicated reduced metabolic rates of drug metabolism in malnutrition,.39–42 which may explain why fasting and malnutrition do not demonstrate increased toxicity with APAP.

One major retrospective study looking at fasting and ethanol demonstrated that no patients developed toxicity with less than 4g/day of APAP and that all of the patients who developed hepatotoxicty after consuming more than 10g/day were ethanol abusers.43 A prospective trial with restricted caloric intake demonstrated no evidence of any change in disposition of APAP or toxicity.44 The authors concluded that APAP should be restricted to 2g/day, although they cite no evidence for such a recommendation, nor is there any such data in the literature.

Glucuronide.

The role of glucuronidation is an interesting one, since an overdose animal model has shown that there is a decrease in glucuronide availability following APAP overdose.45–49 However, glucuronidation is not saturable except in the most massively poisoned humans.50–51 Examination of patients with substantial overdosage of APAP demonstrates a rise in bilirubin but no diminution in its conjugation with glucuronide.

Sulfate.

Sulfate conjugation does appear to be saturable. This pathway is of great interest because it may be a significant protective mechanism once NAC is administered. While initially thought to act as a glutathione surrogate, NAC has been shown to be a sulfate donor as well.

Incidence

While the FDA-labeled doses are clear and concise, some patients may ignore the directions and consume quantities in excess of recommended doses. Data from studies of supratherapeutic ingestion indicate a requirement for 10 g/ day or greater for several days before hepatotoxicity occurs, although cases have been reported of greater than 12 g/day without toxicity.52–53

The issue of “intentional” versus “accidental” or “therapeutic misadventure” is complex, since the molecule has no knowledge of intent when consumed, and in fact all of these are pharmacologically supratherapeutic. We must remain skeptical of “accidental” or “unintentional” overdosages in adults. More than 30 years ago it was fashionable for those of us in medical toxicology to consider some barbiturate overdosages accidental. We opined that these overdosages occurred when a patient consumed a barbiturate for sleeping, became confused, and took another … and another. It was not long until we realized that these patients had actually taken an overdose, changed their mind about attempting suicide, and then wanted a face-saving way to deny their overdose. Asking the friends and relatives of an encephalopathic patient for the history and intent related to consumption of any drug is problematic. Further confusion has been introduced by some authors who utilize “side effects” in discussions of overdose when this term is and should be one of art, reserved for therapeutic doses.

Patients who do overdose and are seen early have a mortality rate of less than 0.5%.3 Patients who wait to seek treatment until they develop overt toxicity are usually beyond the time of most effective action of NAC and may have a higher morbidity rate, although case selection has been skewed.54 These patients are self-selected in terms of whether the incident was a suicide attempt or an unintentional overdose, and placement in these categories must be viewed skeptically. It is not possible to calculate the true rate, since the patients represent the numerator with an unknown denominator, as we do not know how many patients took overdosages and did not seek treatment nor how many misrepresented what they actually did.

Although there has been discrepant data regarding limitation of APAP in a package and in “blister” packaging, the most recent information from Scotland demonstrates that the effects of the restriction introduced in 1998 have been short-lived. Table 1 shows that the rate of overdose has once again increased in Scotland after an initial drop in 1999 (N.D. Bateman, personal communication, February 2004).

Table 1. Incidence of APAP Overdose in Scotland (Population 5.2 Million)

Paracetamol (APAP) Hospital Deaths and Discharges in Scotland, 1995–200255
	1995	1996	1997	1998	1999	2000	2001	2002
Male	2,374	2,583	2,953	2,705	2,421	2,733	2,808	2,672
Female	3,412	3,681	4,225	3,648	3,289	3,984	4,428	4,348
Total	5,786	6,264	7,178	6,353	5,710	6,717	7,236	7,020

In summary, aspects of each case that should be considered to avoid errors include: reliability of the patient history; agreement of the history and acetaminophen level in terms of time; evaluation of the half-life, with the understanding that a prolonged half-life takes time to develop; agreement of the history and aspartate aminotransferase levels relative to time; and careful evaluation of repeated supratherapeutic ingestion.

Aspects of the literature that should be considered when evaluating issues include: animal to human data validity and species differences; in vitro versus in vivo data; class I versus class II or III literature; and case selection bias including confounding variables.

Neither fasting nor maximal induction with ethanol are capable of causing increased toxicity with therapeutic doses of acetaminophen. Alcoholics, however, may be at an increased risk of toxicity following an overdose of acetaminophen. Short-term fasting does not produce enhanced toxicity, and longer-term fasting is unlikely to produce any changes. With these thoughts in mind, more rational case analysis and reporting will be likely.

For a more comprehensive review of the issues in this article, the reader is referred to Dart and Rumack.55