Clinical Toxicology (2015), 53, 807–814 Copyright © 2015 Informa Healthcare USA, Inc. ISSN: 1556-3650 print / 1556-9519 online DOI: 10.3109/15563650.2015.1066507
CRITICAL CARE
External validation of the paracetamol-aminotransferase multiplication product to predict hepatotoxicity from paracetamol overdose Anselm Wong,1,5 Marco L. A. Sivilotti,2 Paul I. Dargan,3,4 David M. Wood,3,4 and Shaun L. Greene1
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
1Emergency
Physician and Clinical Toxicologist, Victorian Poisons Information Centre and Austin Toxicology Service, Austin Hospital, Victoria, Australia 2Departments of Emergency Medicine, and of Biomedical & Molecular Sciences, Queen’s University, Kingston, Canada 3Guy’s and St Thomas’ NHS Foundation Trust and King’s Health Partners, London, UK 4Kings College London, London, UK 5Department
of Medicine and School of Clinical Sciences, Faculty of Medicine, Nursing and Health Sciences, Monash Medical Centre, Monash University, Victoria, Australia. Context. Risk prediction in paracetamol (acetaminophen, or APAP) poisoning treated with acetylcysteine helps guide initial patient management and disposition. The paracetamol-aminotransferase multiplication product may be a useful and less time-sensitive risk predictor. Objective. The aim of this study was to validate this multiplication product in an independent cohort of patients with paracetamol overdose. Materials and methods. Using an existing toxicology dataset of poisoned patients from two large inner-city United Kingdom teaching hospitals, we retrospectively identified by electronic search all paracetamol overdoses from February 2005 to March 2013. We assessed the diagnostic accuracy of the multiplication product (serum APAP concentration alanine transaminase [ALT] activity), especially at the pre-specified cut-off points of 1 500 mg/L IU/L (10 000 micromol/L IU/L) and 10 000 mg/L IU/L (66 000 micromol/L IU/L). The primary outcome was hepatotoxicity defined by a peak ALT 1000 IU/L. Results. Of 3823 total paracetamol overdose presentations, there were 2743 acute single, 452 delayed single ( 24 h post overdose), 426 staggered (ingestion over 1 h), and 202 supratherapeutic ingestions. Altogether, 34 patients developed hepatotoxicity. Among the acute single-ingestion patients, a multiplication product 10 000 mg/L IU/L had a sensitivity of 80% (95% confidence interval [CI]: 44%, 96%) and specificity of 99.6% [99.3%, 99.8%], while a product 1 500 mg/L IU/L had a sensitivity of 100% [66%, 100%] and specificity of 92% [91%, 93%]. Overall, 16 patients with a multiplication product 10 000 mg/L IU/L developed hepatotoxicity (likelihood ratio: 250, 95% CI: 130, 480), and 4 patients with a multiplication product between 1 500 and 10 000 (likelihood ratio: 2.5, 95% CI: 1.0, 6.0). No patient with a product 1 500 mg/L IU/L who received acetylcysteine developed hepatotoxicity. Conclusions. Regardless of ingestion type, a product 10 000 mg/L IU/L was associated with a very high likelihood, and 1 500 mg/L IU/L with a very low likelihood, of developing hepatotoxicity in patients treated with acetylcysteine. Keywords Poisoning; Risk; Score; Acetaminophen; Liver
Introduction
can lead to both under and over treatment with the antidote acetylcysteine and may increase the risk of hepatotoxicity or prolong hospitalisation, respectively. A new risk-stratification tool to help predict hepatotoxicity has been proposed recently to address these shortcomings.13 The tool consists of the multiplication product of serum aminotransferase activity and paracetamol concentration (APAP) as a predictor of subsequent hepatotoxicity. Importantly, the method either does not require knowledge of the exact time of paracetamol exposure, information that may often be inaccurate or unknown, or does it require that the overdose be taken at a single point in time. The original study by Sivilotti et al. involved 94 patients, all of whom had hepatotoxicity following paracetamol overdose.13 The aim of this study was to validate this reported method of predicting hepatotoxicity using a larger independent cohort of
Paracetamol is amongst the most common pharmaceutical products taken in overdose in the developed world and a leading cause of fulminant liver failure.1–7 A risk-stratification nomogram based on a timed serum paracetamol concentration is used to determine the need for treatment following a single acute overdose at a known time of ingestion.3 However, such an approach cannot be used in patients with delayed presentations, staggered ingestions, or time-unknown paracetamol overdoses.8–12 Imprecise risk assessment Received 11 May 2015; accepted 22 June 2015. Address correspondence to Anselm Wong, MBBS FACEM, Emergency Physician and Clinical Toxicologist, Victorian Poisons Information Centre and Austin Toxicology Service, Austin Hospital, Studley Road, Heidelberg, Victoria 3084, Australia. Tel: 61394964509. E-mail: [email protected]
807
808 A. Wong et al. paracetamol overdose patients recruited from a different clinical and geographical setting.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Methods Study design A retrospective analysis of patients presenting to the Emergency Department of two large inner city hospitals in the United Kingdom following paracetamol overdose was conducted. Data on all poisoned patients presenting to these Emergency Departments are collected prospectively on a purpose-designed clinical toxicology electronic database as described elsewhere.14 Database specialists enter information systematically using routine clinical notes and laboratory records/results. Approval for creating and accessing the database has previously been obtained from the local Caldicott guardian and research ethics committee (IRB). Study population and definitions Patients were included in the primary analysis if they presented following paracetamol overdose from February 2005 to March 2013, and if both serum paracetamol concentration and alanine transaminase (ALT, the routinely measured aminotransferase in the study hospitals) activity had been measured. Laboratory testing and the decision to administer acetylcysteine were based on national recommendations at the time the patient presented to the hospital.15 These guidelines recommend testing the paracetamol concentration at least 4 h after ingestion in acute overdoses, and measuring ALT, international normalised ratio (INR), creatinine and perhaps paracetamol at the end of the standard 20.25–21-h acetylcysteine infusion. In uncertain, delayed or staggered overdoses, these guidelines recommend treatment initially based on paracetamol dose ingested (mg/kg) for early presenters and on renal and hepatic function testing in late presenters. Despite the guidelines in some late presentations serum paracetamol concentrations had been measured. While routinely classified as such in the source database, for this study exposures were explicitly classified into the following mutually exclusive categories after being independently reviewed by two authors and disagreements resolved by consensus: acute (ingestion over a time period of less than 1 h), delayed (presentation greater than 24 h post ingestion), staggered (exposure over a period of greater than 1 h), and supratherapeutic (ingestion of 75 mg/kg/day paracetamol with therapeutic intent). The risk assessment multiplication product (APAP ALT) was calculated as per Sivilotti et al.13 by multiplying the first measured ALT activity (in IU/L) by the simultaneously measured paracetamol concentration (mg/L), which in almost all cases were done at presentation to hospital. Neither hospital laboratory routinely measures aspartate aminotransferase. In a secondary analysis, for patients in whom the serum paracetamol was undetectable, we imputed an arbitrary value of 5 mg/L to allow calculation of the product. This value was chosen because it was 50% below the laboratory limit of detection of 10 mg/L, as per the original study.13
Outcome measures The primary outcome was hepatotoxicity, defined to be a peak ALT 1000 IU/L attributed by the treating clinicians to paracetamol exposure.5,16 The secondary outcome was coagulopathy, defined as a peak INR 2. We were primarily interested in the diagnostic accuracy of the multiplication product for the outcome of hepatotoxicity, especially at arbitrarily pre-specified cut-off points of 1 500 mg/L IU/L (10 000 micromol/L IU/L) as identified by Sivilotti et al.13 as well as 10 000 mg/L IU/L (66 000 micromol/L IU/L). We also wished to describe the variation over time of the multiplication product in patients who developed hepatotoxicity and who had more than one detectable paracetamol concentration. Case search strategy The clinical toxicology database was interrogated for presentations meeting the inclusion/exclusion criteria. Patients with laboratory evidence of paracetamol overdose were recorded as such in the database. The narrative description of each case identified was then accessed. Electronic preformatted case reports were generated for each case, organised into a study database and analysed by the investigators. SPSS (v23) was used to analyse these results.
Results Of a total of 3823 paracetamol ingestions (Table 1), there were 2743 acute single ingestions, 452 delayed ingestions, 426 staggered and 202 supratherapeutic ingestions. 34 patients developed hepatotoxicity: 13 staggered, 10 acute single, 10 delayed and 1 supratherapeutic. Altogether, 1304 patients received acetylcysteine, which was administered intravenously in all cases. The entire 300 mg/kg UK acetylcysteine regimen was administered to all patients who developed hepatotoxicity, and extended in 25 of these until INR and ALT were improving. Two patients died in hospital, neither attributed to paracetamol toxicity, and 3 patients were transferred to a specialised liver unit for further management (none of whom received transplant). Serum ALT activity was measured in almost all cases (3783 or 99%), on average 3.5 2.0 h after presentation. However, for one-third of the patients who developed hepatotoxicity, the serum paracetamol concentration was either undetectable (12 cases) or not done (2 cases), leaving 3112 patients for whom a multiplication product could be calculated in the primary analysis. The overall diagnostic accuracy of the multiplication product is shown in Table 2, and as a receiver operator curve in Fig. 1 (area under the curve (AUC): 0.994 (95% confidence interval [CI]: 0.988–1.00), departure from 100% sensitivity at 3961 mg/L IU/L). Using the upper cutpoint of 10 000 mg/L IU/L, the first measured product had a positive likelihood ratio over 200 for hepatotoxicity, identifying 16 of the 20 patients (sensitivity 80%, specificity 99.7%). None of the 2831 patients with an initial product below 1 500 mg/L IU/L developed hepatotoxicity (likelihood ratio near zero). Clinical Toxicology vol. 53 no. 8 2015
APAP product to predict hepatotoxicity in OD 809 Table 1. Comparison of demographics and clinical variables between patients with and without hepatotoxicity.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Female count (%) Age, years Initial Paracetamol concentration, mg/L Initial ALT all ingestions, IU/L Ingestion type, count (%) Single acute Delayed Staggered Supratherapeutic Acetylcysteine treatment, count (%) within 8 h of overdose more than 8 h (%) None
Hepatotoxicity (n 34)
No hepatotoxicity (n 3789)
30 (88%) 29 (26–32) 140 (95–147) 258 (86–1993)
2206 (58%) 30 (26–34) 99 (66–131) 25 (16–36)
10 (29.5%) 10 (29.5%) 13 (38%) 1 (3%)
2733 (72%) 442 (12%) 413 (11%) 201 (5%)
1 (3%) 33 (97%) 0
660 (18%) 610 (16%) 2519 (66%)
p value 0.001 0.7 0.001 0.001 0.0001
0.0001
LT, alanine aminotransferase. A Hepatotoxicity defined as ALT 1000 IU/L. Values in tables are reported as median (interquartile range) or count (percentage).
In the secondary analysis imputing a paracetamol concentration of 5 mg/L when undetectable, again none of the patients with hepatotoxicity had an initial product 1 500, four were between 1 500 and 10,000 and eight presentations were 10 000 mg/L IU/L. None of these 12 patients were transferred to a liver unit, and the highest peak INR was 2.8.
typically 160 IU/L and multiplications products range: 11 543–52 800 mg/L IU/L). The receiver operating characteristic (ROC) curve for all single acute ingestions (0–24 h) (Fig. 3) yielded an AUC of 0.995 (95% CI, 0.99–1.00), and departed from 100% sensitivity at 4053 mg/L IU/L. Amongst single acute ingestions presenting 8 or more hours post ingestion, the sensitivity and specificity of a product 10,000 mg/L IU/L in the 8–24 h increased to 88.8% (50.1,99.4%) and 100% (99.8,100%), respectively (ROC AUC: 0.999; 95% CI, 0.998–1.00), departing 100% sensitivity at 6519 mg/L IU/L. Only three of the patients who developed hepatotoxicity had multiple detectable paracetamol concentrations. All three had received acetylcysteine within 4 h of presentation, and all received extended treatment with acetylcysteine. The individual elimination half-life of paracetamol ranged from 5.0 to 7.8 h. The multiplication product remained similar over several hours in these three patients, always above 10 000 mg/L IU/L (Fig. 4). Turning to the 10 acute single ingestions (0.4% of total) with a multiplication product greater than 10 000 mg/L IU/L on presentation yet did not go on to develop hepatotoxicity,
Acute single ingestions Of the 2703 acute single ingestions (255 of whom received acetylcysteine), 40 did not have both a paracetamol concentration and ALT measured, and 10 developed hepatotoxicity. 18 (0.74%) had an initial multiplication product over 10 000 mg/L IU/L; 8 of these went on to develop hepatotoxicity (sensitivity 80%, specificity 99.6%; Fig. 2). The peak INR ranged from 1.3 to 3.0 in these 8 patients. All but 1 patient (7 h) of the 8 had their first multiplication product 8 h post ingestion. As shown in Fig. 2, all 8 had an abnormal ALT activity when first measured (range: 103–792 IU/L), unlike the others in whom the elevated product was primarily due to a high serum paracetamol concentration (initial ALT
Table 2. Diagnostic accuracy of paracetamol aminotransferase multiplication product regardless of paracetamol overdose presentation. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not obtained Total
Hepatotoxicity
No hepatotoxicity
Likelihood ratio (95% CI)
16 4 0 20
10 251 2831 3092
250 (130,480) 2.5 (1,6) 0 3112
12
380
3.52 (2.21,5.60)
2 34
317 3789
0.70 (0.18,2.71) 3823
t the pre-specified cutpoints, the interval likelihood ratio is shown after excluding patients without a measurable paracetaA mol concentration at the time of the first available aminotransferase. For these excluded patients, a positive likelihood ratio is calculated for having the exclusion criterion present. Copyright © Informa Healthcare USA, Inc. 2015
810 A. Wong et al.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
rise above 100 IU/L. One patient had a normal ALT of 22 IU/L and paracetamol concentration of 186 mg/L 7 h post overdose (product of 4092 mg/L IU/L) but received acetylcysteine 11 h post overdose and developed hepatotoxicity. None of the 2479 patients with a product below 1 500 mg/L IU/L developed hepatotoxicity following acute overdose.
Fig. 1. ROC curve of the paracetamol-aminotransferase multiplication product (APAP ALT) in predicting hepatotoxicity for all classifications of paracetamol overdose (colour version of this figure can be found in the online version at www.informahealthcare.com/ctx).
the multiplication products in each case were calculated using laboratory tests taken less than 8 h post ingestion (Fig. 5). Nine of these had acetylcysteine given within 8 h of ingestion and all by 12 h. Eight had an ALT 2–3 times the upper limit of normal at the time of presentation. The peak ALT ranged from 38 to 293 IU/L measured 3–56 h post ingestion. None had a peak INR 2 (mean: 1.1, 95% CI, 1.0–1.3). 216 (8.0%) of the 2703 acute single ingestion presentations had a multiplication product 1 500 mg/L IU/L but below 10 000 mg/L IU/L, and only 2 (1%) developed hepatotoxicity (Table 3). The peak INR from this subset was 1.3 (range, 0.9–1.3) and ALT ranged from 8 to 954 (median, 24) IU/L. All but 7 had acetylcysteine started within 8 h of ingestion, in which case ALT did not
Delayed paracetamol ingestions Of the 452 delayed paracetamol ingestions (acetylcysteine administered to 433), the paracetamol concentration was never measured (272, 60.2%) or was undetectable (140, 30.9%) in the vast majority, reflecting usual recommended practice in the United Kingdom and the delay to care (average time to presentation 53 h post ingestion; 95% CI, 65–85). Ten (2%) patients had or developed hepatotoxicity; of these, 8 had an undetectable paracetamol concentration at the time of presentation. The other two cases had multiplications products of 52 690 and 96 380 mg/L IU/L (presentation at 46 and 51 h post ingestion, and peak INR: 1.6 and 3.0, respectively). Five of the ten presented with established hepatotoxicity (initial ALT range: 2366–10116 IU/L, INR range: 1.4–2.6, 46 h to 70 h post ingestion). The diagnostic accuracy is shown in Table 4. All of the patients who presented with undetectable paracetamol concentrations survived. Staggered paracetamol ingestions There were 426 staggered overdoses, of whom 425 received acetylcysteine. 51 had undetectable (46, 10.8%) or no (5, 1.2%) paracetamol concentrations at time of presentation (Table 5). There were 13 cases of hepatotoxicity from staggered overdose and all received acetylcysteine at the time of initial medical assessment. Peak INR ranged from 1.5 to 4.4 and ALT ranged from 12 to 10 956 IU/L.
Fig. 2. Initial serum ALT concentrations matched with simultaneous serum paracetamol concentrations in single acute overdose patients with multiplication product 10 000 mg/L IU/L. Patients are classified based on hepatotoxicity () (peak ALT 1000 IU/L; n 8) versus no hepatotoxicity () (n 10). Any point under the dotted diagonal line represents a multiplication product 10 000 mg/L IU/L. Clinical Toxicology vol. 53 no. 8 2015
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
APAP product to predict hepatotoxicity in OD 811
Fig. 3. ROC curve of the paracetamol-aminotransferase multiplication product (APAP ALT) in predicting hepatotoxicity secondary to acute single ingestion paracetamol overdose for patients presenting 0–24 h after ingestion (colour version of this figure can be found in the online version at www.informahealthcare.com/ctx).
Supratherapeutic paracetamol ingestions There were 202 supratherapeutic ingestions (199 treated with acetylcysteine), of whom only 8 (4%) had a detectable paracetamol concentration. Seven of these had a paracetamol multiplication product 1,500 mg/L IU/L. Only one patient (ingestion over 4 days) had an initial multiplication product 10 000 mg/L IU/L (initial APAP, 58 mg/L; ALT, 3042 IU/L), but subsequent aminotransferase concentrations fell and peak INR was only 1.7 with a standard course of acetylcysteine started 1 h post presentation.
Discussion A risk prediction calculation of developing hepatotoxicity in paracetamol overdose treated with an antidote is a much needed tool, whether trying to identify high-risk patients who
Fig. 4. Multiplication product relative to initiation of acetylcysteine in the three patients who developed hepatotoxicity following single acute ingestion in whom serial paracetamol concentrations were available. Copyright © Informa Healthcare USA, Inc. 2015
might benefit from more intensive therapy or abbreviating unnecessary treatment and reducing uncertainty for low-risk patients. Since liver function test abnormalities may occur only a day or more post ingestion, there can be considerable uncertainty for the patient and treating clinicians following excess acetaminophen exposure. We report the diagnostic accuracy of a novel risk instrument, the paracetamol-aminotransferase multiplication product, in a very large cohort of patients across the spectrum of categories of paracetamol overdose. We observed that a paracetamol-aminotransferase multiplication product has high diagnostic accuracy for hepatotoxicity whether the overdose is acute, staggered, delayed or supratherapeutic. An initial multiplication product 10 000 mg/L IU/L was strongly associated with hepatotoxicity, unless calculated within 8 h of a single acute overdose, while patients with a product 1 500 mg/L IU/L were highly unlikely to develop hepatotoxicity. A multiplication product between 1 500 and 10 000 mg/L IU/L on presentation had only a modestly increased likelihood of developing hepatotoxicity. It should be emphasised that the paracetamol nomogram remains useful for predicting which single acute overdoses are unlikely to develop hepatotoxicity in the absence of treatment with acetylcysteine.8–12 However, the nomogram cannot be used for many patients because of time unknown, staggered, delayed or supratherapeutic ingestions of paracetamol. The nomogram is also not used to predict risk in patients who receive treatment with acetylcysteine. Instead, our findings represent the risk of hepatotoxicity in treated patients. Furthermore, the paracetamol-aminotransferase product can be calculated without certain knowledge of the time of ingestion, or even the requirement that the ingestion be taken at a single point in time. For example, in a rural setting a paracetamol-aminotransferase product may lead to earlier identification, closer monitoring and discussion about transferring patients to facilities with an intensive care unit or specialist service. This is especially useful in rural hospitals with minimal inpatient resources and availability of laboratory facilities on predicting the time course of treatment for a patient. Also it may help decide direct ward admissions for prolonged acetylcysteine treatment rather than a short stay observation unit. In addition to identifying patients who have a very high likelihood of developing hepatotoxicity, the paracetamolaminotransferase multiplication product appears able to identify patients at a very low risk. This result is noteworthy in the context of recent efforts to further abbreviate acetylcysteine treatment duration,17 and also the controversial decision to lower the treatment threshold substantially in the United Kingdom.18 Yet a risk prediction instrument with a sensitivity near 100% will be needed to reduce unnecessary treatment, given the relatively low cost of an individual course of treatment and the high stakes of preventing avoidable fulminant hepatic failure. Reducing unnecessary treatment is particularly important for paracetamol exposure, given the frequency of acetylcysteine administration for this indication. We believe that the multiplication product offers such a high sensitivity in addition to simplicity and
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
812 A. Wong et al.
Fig. 5. Paracetamol-Aminotransferase Multiplication product relative to interval from ingestion to initiation of acetylcysteine following single acute overdose. Only patients with initial multiplication products 10 000 mg/L IU/L are shown, classified based on those who developed hepatotoxicity () and did not () (n 10).
availability. It is worth also noting that we have characterised the outcome of hepatotoxicity, considered to be a necessary but not sufficient criterion of hepatic failure. The outcomes of coagulopathy and hepatic failure were too infrequent in this cohort to characterise the diagnostic accuracy of the product for these more serious but less common outcomes. Moreover, while a threshold of 1 500 mg/L IU/L enjoyed 100% sensitivity, more work is needed to identify the optimal cutpoint and further reduce the lower confidence band of this estimate. Dichotomising the product around a cutpoint also fails to take advantage of the full spectrum of information given the continuous nature of the predictor, whereas customising treatment intensity (e.g. cumulative dose of antidote administered) based on the product value would also be clinically useful. Several single acute ingestion patients who had paracetamol multiplication products greater than 10 000 (mg/L IU/L) did not go on to develop hepatotoxicity. The majority of this group presented early, had mostly elevated paracetamol concentrations contributing to the elevated product and also received acetylcysteine within 8 h of acute overdose. Meanwhile the early administration of acetylcysteine in acute overdose is known to be protective.19 Chomchai et al. have previously reported in an independent validation study that the multiplication product had higher specificity in patients presenting at least 8 h after acute overdose.20 Our study supports this finding and suggests that a multiplication product calculated within 8 h of single ingestion could be repeated 12 h later to allow optimal specificity. Hence in the
small cohort where the paracetamol concentrations are still high and the ALT is mildly elevated the standard regimen of acetylcysteine may not be enough to prevent hepatotoxicity. Other modalities such as haemodialysis or increased acetylcysteine regimens may help in these situations but further investigation is needed.21 A 12-h interval between paracetamol measurements would be expected to allow the paracetamol concentration to fall by an order of magnitude ( 3 half-lives if absorption complete and elimination is normal), and the product could also fall by the same degree if the aminotransferases remained unchanged. We had initially hoped to use only multiplication products calculated at least 8–12 h post ingestion in this study, but discovered that too few patients had serial measurements of the falling paracetamol concentration. While repeat testing of post-peak paracetamol concentrations has historically been discouraged in part for cost considerations, we believe serial testing to be essential for any patient being treated with acetylcysteine until the paracetamol concentration is undetectable. In addition to testing ‘liver function’ (i.e. paracetamol elimination), serial testing is inexpensive and would avoid the pitfall of premature discontinuation of acetylcysteine after 21 h, especially given reports of persistently elevated paracetamol concentrations.21 Of the three single acute overdose presentations that developed hepatotoxicity and had serial paracetamol concentrations, elimination half-life was greater than 4 h, and the multiplication product remained relatively stable, consistent with earlier studies (mean, 7.6/ 0.8 h).8,9,13
Table 3. Diagnostic accuracy for hepatotoxicity in single acute overdose patients presenting within 24 h. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total
Hepatotoxicity 8 2 0 10
No hepatotoxicity 10 214 2469 2693
Likelihood ratio (95% CI)
Sensitivity % (95% CI)
Specificity% (95%CI)
PPV, NPV%
215 (108,430) 2.52 (0.7,8.7) 0 (0,0.001)
80 (44,96)
99.6 (99.3,99.8)
44,99.9
100 (66,100)
92 (91,93)
4.4,100
PPV, positive predictive value; NPV, negative predictive value. Clinical Toxicology vol. 53 no. 8 2015
APAP product to predict hepatotoxicity in OD 813 Table 4. Diagnostic accuracy of multiplication product in delayed paracetamol ingestions. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not done Total Secondary analysis: products for imputed 5 mg/L value for 10 mg/L concentrations 10 000 1 500–10 000 1 500 Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Likelihood ratio (95% CI)
Hepatotoxicity No hepatotoxicity 2 0 0 2
0 9 29 38
infinity 0 0
8 0 10
132 272 442
2.68(1.9,3.7) 452
7 1 0
0 2 130
infinity (2.3,inf) 8.25 (0.83,81.63) 0 (0,0.02)
secondary analysis was performed for presentations with paracetamol concentrations below the detectable limit A ( 10 mg/L). A value of 5 mg/L was imputed for paracetamol to obtain a multiplication product.
In our study, the majority of delayed ( 24 h after ingestion) paracetamol presentations did not have a paracetamol concentration measured on arrival, consistent with national recommendations. Instead acetylcysteine was typically started based on history and pending the results of other tests taken on arrival. More than half of these patients presented in established hepatotoxicity, hence limiting the utility of calculating the multiplication product to identify this outcome as described in this study. Nevertheless, we wished to characterise the product across all presentations. We assigned a low, non-zero value for the paracetamol concentration when it was undetectable, although it may well have been much lower. We acknowledge that as time passes, other factors such as pH, encephalopathy and coagulopathy become more useful to identify the most serious outcomes. 22,23 In regards to the staggered ingestions of paracetamol that developed hepatotoxicity, serum paracetamol concentrations were not measured in the majority (again consistent with national recommendations) and therefore products were not calculated in all patients. For those in whom a
product could be calculated, as well as in the secondary analysis, no patients with a product 1 500 mg/L IU/L developed hepatotoxicity. The multiplication product may be particularly helpful in staggered overdoses that receive acetylcysteine as there is often an opportunity to administer a much shorter course of acetylcysteine once a favourable outcome is certain. In a study of 410 patients of whom 264 were acute overdoses presenting within 8 h, Al-Hourani et al.24 reported that an abnormal serum ALT activity at presentation was present in all 16 patients who developed hepatotoxicity, for an estimated sensitivity between 81% and 100% (95% CI). This high sensitivity is consistent with Canadian data involving 94 acute overdose patients in which ‘almost all aminotransferase concentrations measured 12 or more hours post ingestion exceeded 50 IU/L and exceeded 100 IU/L when measured more than 24 h post ingestion’.25 One of the uncertainties involving the clinical application of the multiplication product is therefore the interpretation of a product calculated too soon after overdose, or the closely related issue of
Table 5. Multiplication product in staggered paracetamol ingestions with primary and secondary analyses. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not obtained Total Secondary analysis: products for imputed 5 mg/L value for 10 mg/L concentrations 10 000 1 500–10 000 1 500
Hepatotoxicity No hepatotoxicity
Likelihood ratio (95% CI)
5 2 0 7
0 17 345 352
infinity 5.75(1.63,20.3) 0
4 2 13
42 3 413
3.03 (1.27,7.18) 21.2 (3.86,116) 426
2 2 0
0 1 41
infinity (2.4,inf) 21.0 (2.40,184) 0 (0,0.07)
secondary analysis was performed for presentations with paracetamol concentrations below the detectable limit ( 10 mg/L). A A value of 5 mg/L was imputed for paracetamol to obtain a multiplication product. Copyright © Informa Healthcare USA, Inc. 2015
814 A. Wong et al. being high despite a normal aminotransferase concentration. Notwithstanding boundary conditions which define a very low risk (e.g. normal ALT at initiation of acetylcysteine, or timed serum paracetamol concentration below the nomogram treatment line), the product yields a continuous risk measure for all patients at some risk. Other tests such as protein adducts and microRNA are not readily available yet but may help in the future in predicting development of hepatotoxicity secondary to paracetamol misuse.26,27
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Conclusion In a large cohort of patients presenting to hospital for paracetamol overdose of any kind, the initial paracetamol-aminotransferase multiplication product has a very high diagnostic accuracy for predicting hepatotoxicity in patients treated with acetylcysteine. Regardless of ingestion type, a product 10 000 mg/L IU/L was associated with a very high likelihood, and 1 500 mg/L IU/L with a very low likelihood, of developing hepatotoxicity. We urge caution using the product within 8 h of an acute overdose, as the serum paracetamol concentration itself may contribute to a high product, but recommend instead repeating the product approximately 12 h later for optimal specificity.
Declaration of interest The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.
References 1. National Poisons Information Service UK. Annual Reports. Available from http://legacytools.hpa.org.uk/Publications/ChemicalsPoisons/ NationalPoisonsInformationServiceAnnualReports/1308NPISannual report20122013/.2013; (accessed 8 July 2015). 2. Bronstein AC, Spyker DA, Cantilena LR Jr, Rumack BH, Dart RC. 2011 Annual report of the American Association of Poison Control Centers’ National Poison Data System (NPDS): 29th Annual Report. Clin Toxicol (Phila) 2012; 10:911–1164. 3. Buckley N, Eddleston M. Paracetamol (acetaminophen) poisoning. Clin Evid 2005; 14: 1738–1744. 4. Linden CH, Rumack BH. Acetaminophen overdose. Emerg Med Clin North Am 1984; 2:103–119. 5. Ostapowicz G, Fontana RJ, Schiødt FV, Larson A, Davern TJ, Han SH, et al. Results of a prospective study of acute liver failure at 17 tertiary care centers in the United States. Ann Intern Med 2002; 137:947–954. 6. Sood S, Howell J, Sundararajan V, Angus PW, Gow PJ. Paracetamol overdose in Victoria remains a significant health burden. J Gastroenterol Hepatol 2013; 28:1356–1360. 7. Smilkstein MJ, Knapp GL, Kulig KW, Rumack BH. Efficacy of oral N-acetylcysteine in the treatment of acetaminophen overdose. Analysis of the national multicenter study (1976 to 1985). N Engl J Med 1988; 319:1557–1562.
8. Rumack BH, Peterson RC, Koch GG, Amara IA. Acetaminophen overdose. 662 Cases with evaluation of oral acetylcysteine treatment. Arch Intern Med 1981; 141:380–385. 9. Prescott LF, Roscoe P, Wright N, Brown SS. Plasma-paracetamol half-life and hepatic necrosis in patients with paracetamol overdosage. Lancet 1971; 1:519–522. 10. Dart RC, Rumack BH. Patient-tailored acetylcysteine administration. Ann Emerg Med 2007; 50:280–281. 11. Woo OF, Mueller PD, Olson KR, Anderson IB, Kim SY. Shorter duration of oral N-acetylcysteine therapy for acute acetaminophen overdose. Ann Emerg Med 2000; 35:363–368. 12. Heard KJ. Acetylcysteine for acetaminophen poisoning. N Engl J Med 2008; 359:285–292. 13. Sivilotti ML, Green TJ, Langmann C, Yarema M, Juurlink D, Johnson D. Multiplying the serum aminotransferase by the acetaminophen concentration to predict toxicity following overdose. Clin Toxicol 2010; 48:793–799. 14. Green SL, Wood DM, Gawarammana IB, Warren-Gash C, Drake N, Jones AL, Dargan PI. Improvement in the management of acutely poisoned patients using an electronic database, prospective audit and targeted educational intervention. Postgrad Med J 2008; 84:603–608. 15. Ferner RE, Dear JW, Bateman DN. Management of paracetamol poisoning. BMJ 2011; 342:d2218. doi: 10.1136/bmj.2218 16. Makin AJ, Wendon J, Williams R. A 7-year experience of severe acetaminophen-induced hepatotoxicity (1987–1993). Gastroenterology 1995; 109:1907–1916. 17. Bateman DN, Dear JW, Thanacoody HK, Thomas SH, Eddleston M, Sandilands EA, et al. Reduction of adverse effects from intravenous acetylcysteine treatment for paracetamol poisoning: a randomised controlled trial. Lancet 2014; 383:697–704. 18. Bateman DN, Carroll R, Pettie J, Yamamoto T, Elamin ME, Peart L, et al. Effect of the UK’s revised paracetamol poisoning management guidelines on admissions, adverse reactions and costs of treatment. Br J Clin Pharmacol 2014; 78:610–618. 19. Prescott LF, Illingworth RN, Critchley JA, Stewart MJ, Adam RD, Proudfoot AT. Intravenous N-acetylcysteine: the treatment of choice for paracetamol poisoning. Br Med J 1979; 2:1097–1100. 20. Chomchai S, Chomchai C. Predicting acute acetaminophen hepatotoxicity with acetaminophen-aminotransferase multiplication product and the Psi parameter. Clin Toxicol 2014; 52: 506–511. 21. Sivilotti MLA, Juurlink DN, Garland JS, Lenga I, Poley R, Hanly N, Thompson N. Antidote removal during haemodialysis for massive acetaminophen overdose. Clin Toxicol 2013; 51:855–863. 22. Schiødt FV, Bondesen S, Tygstrup N, Christensen E. Prediction of hepatic encephalopathy in paracetamol overdose: a prospective and validated study. Scand J Gastroenterol 1999; 34:723–728. 23. O’Grady JG, Alexander GJ, Hayllar KM, Williams R. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 1989; 97: 439–445. 24. Al-Hournai K, Mansi R, Pettie J, Dow M, Bateman DN, Dear JW. The predictive value of hospital admission serum alanine transaminase activity in patients treated for paracetamol overdose. Q J Med 2013; 106:541–546. 25. Green TJ, Sivilotti ML, Langmann C, Yarema M, Juurlink D, Burns MJ, Johnson DW. When do the aminotransferases rise after acute acetaminophen overdose? Clin Toxicol 2010; 48:787–792. 26. Davern TJ, James LP, Hinson JA, Polson J, Larson AM, Fontana RJ, et al. Measurement of serum acetaminophen-protein adducts in patients with acute liver failure. Gastroenterology 2006; 130:687–694. 27. Hornby RJ, Starkey LP, Dear J, Goldring C, Park BK. MicroRNAs as potential circulating biomarkers of drug-induced liver injury: key current and future issues for translation to humans. Expert Rev Clin Pharmacol 2014; 7: 349–362.
Clinical Toxicology vol. 53 no. 8 2015
CRITICAL CARE
External validation of the paracetamol-aminotransferase multiplication product to predict hepatotoxicity from paracetamol overdose Anselm Wong,1,5 Marco L. A. Sivilotti,2 Paul I. Dargan,3,4 David M. Wood,3,4 and Shaun L. Greene1
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
1Emergency
Physician and Clinical Toxicologist, Victorian Poisons Information Centre and Austin Toxicology Service, Austin Hospital, Victoria, Australia 2Departments of Emergency Medicine, and of Biomedical & Molecular Sciences, Queen’s University, Kingston, Canada 3Guy’s and St Thomas’ NHS Foundation Trust and King’s Health Partners, London, UK 4Kings College London, London, UK 5Department
of Medicine and School of Clinical Sciences, Faculty of Medicine, Nursing and Health Sciences, Monash Medical Centre, Monash University, Victoria, Australia. Context. Risk prediction in paracetamol (acetaminophen, or APAP) poisoning treated with acetylcysteine helps guide initial patient management and disposition. The paracetamol-aminotransferase multiplication product may be a useful and less time-sensitive risk predictor. Objective. The aim of this study was to validate this multiplication product in an independent cohort of patients with paracetamol overdose. Materials and methods. Using an existing toxicology dataset of poisoned patients from two large inner-city United Kingdom teaching hospitals, we retrospectively identified by electronic search all paracetamol overdoses from February 2005 to March 2013. We assessed the diagnostic accuracy of the multiplication product (serum APAP concentration alanine transaminase [ALT] activity), especially at the pre-specified cut-off points of 1 500 mg/L IU/L (10 000 micromol/L IU/L) and 10 000 mg/L IU/L (66 000 micromol/L IU/L). The primary outcome was hepatotoxicity defined by a peak ALT 1000 IU/L. Results. Of 3823 total paracetamol overdose presentations, there were 2743 acute single, 452 delayed single ( 24 h post overdose), 426 staggered (ingestion over 1 h), and 202 supratherapeutic ingestions. Altogether, 34 patients developed hepatotoxicity. Among the acute single-ingestion patients, a multiplication product 10 000 mg/L IU/L had a sensitivity of 80% (95% confidence interval [CI]: 44%, 96%) and specificity of 99.6% [99.3%, 99.8%], while a product 1 500 mg/L IU/L had a sensitivity of 100% [66%, 100%] and specificity of 92% [91%, 93%]. Overall, 16 patients with a multiplication product 10 000 mg/L IU/L developed hepatotoxicity (likelihood ratio: 250, 95% CI: 130, 480), and 4 patients with a multiplication product between 1 500 and 10 000 (likelihood ratio: 2.5, 95% CI: 1.0, 6.0). No patient with a product 1 500 mg/L IU/L who received acetylcysteine developed hepatotoxicity. Conclusions. Regardless of ingestion type, a product 10 000 mg/L IU/L was associated with a very high likelihood, and 1 500 mg/L IU/L with a very low likelihood, of developing hepatotoxicity in patients treated with acetylcysteine. Keywords Poisoning; Risk; Score; Acetaminophen; Liver
Introduction
can lead to both under and over treatment with the antidote acetylcysteine and may increase the risk of hepatotoxicity or prolong hospitalisation, respectively. A new risk-stratification tool to help predict hepatotoxicity has been proposed recently to address these shortcomings.13 The tool consists of the multiplication product of serum aminotransferase activity and paracetamol concentration (APAP) as a predictor of subsequent hepatotoxicity. Importantly, the method either does not require knowledge of the exact time of paracetamol exposure, information that may often be inaccurate or unknown, or does it require that the overdose be taken at a single point in time. The original study by Sivilotti et al. involved 94 patients, all of whom had hepatotoxicity following paracetamol overdose.13 The aim of this study was to validate this reported method of predicting hepatotoxicity using a larger independent cohort of
Paracetamol is amongst the most common pharmaceutical products taken in overdose in the developed world and a leading cause of fulminant liver failure.1–7 A risk-stratification nomogram based on a timed serum paracetamol concentration is used to determine the need for treatment following a single acute overdose at a known time of ingestion.3 However, such an approach cannot be used in patients with delayed presentations, staggered ingestions, or time-unknown paracetamol overdoses.8–12 Imprecise risk assessment Received 11 May 2015; accepted 22 June 2015. Address correspondence to Anselm Wong, MBBS FACEM, Emergency Physician and Clinical Toxicologist, Victorian Poisons Information Centre and Austin Toxicology Service, Austin Hospital, Studley Road, Heidelberg, Victoria 3084, Australia. Tel: 61394964509. E-mail: [email protected]
807
808 A. Wong et al. paracetamol overdose patients recruited from a different clinical and geographical setting.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Methods Study design A retrospective analysis of patients presenting to the Emergency Department of two large inner city hospitals in the United Kingdom following paracetamol overdose was conducted. Data on all poisoned patients presenting to these Emergency Departments are collected prospectively on a purpose-designed clinical toxicology electronic database as described elsewhere.14 Database specialists enter information systematically using routine clinical notes and laboratory records/results. Approval for creating and accessing the database has previously been obtained from the local Caldicott guardian and research ethics committee (IRB). Study population and definitions Patients were included in the primary analysis if they presented following paracetamol overdose from February 2005 to March 2013, and if both serum paracetamol concentration and alanine transaminase (ALT, the routinely measured aminotransferase in the study hospitals) activity had been measured. Laboratory testing and the decision to administer acetylcysteine were based on national recommendations at the time the patient presented to the hospital.15 These guidelines recommend testing the paracetamol concentration at least 4 h after ingestion in acute overdoses, and measuring ALT, international normalised ratio (INR), creatinine and perhaps paracetamol at the end of the standard 20.25–21-h acetylcysteine infusion. In uncertain, delayed or staggered overdoses, these guidelines recommend treatment initially based on paracetamol dose ingested (mg/kg) for early presenters and on renal and hepatic function testing in late presenters. Despite the guidelines in some late presentations serum paracetamol concentrations had been measured. While routinely classified as such in the source database, for this study exposures were explicitly classified into the following mutually exclusive categories after being independently reviewed by two authors and disagreements resolved by consensus: acute (ingestion over a time period of less than 1 h), delayed (presentation greater than 24 h post ingestion), staggered (exposure over a period of greater than 1 h), and supratherapeutic (ingestion of 75 mg/kg/day paracetamol with therapeutic intent). The risk assessment multiplication product (APAP ALT) was calculated as per Sivilotti et al.13 by multiplying the first measured ALT activity (in IU/L) by the simultaneously measured paracetamol concentration (mg/L), which in almost all cases were done at presentation to hospital. Neither hospital laboratory routinely measures aspartate aminotransferase. In a secondary analysis, for patients in whom the serum paracetamol was undetectable, we imputed an arbitrary value of 5 mg/L to allow calculation of the product. This value was chosen because it was 50% below the laboratory limit of detection of 10 mg/L, as per the original study.13
Outcome measures The primary outcome was hepatotoxicity, defined to be a peak ALT 1000 IU/L attributed by the treating clinicians to paracetamol exposure.5,16 The secondary outcome was coagulopathy, defined as a peak INR 2. We were primarily interested in the diagnostic accuracy of the multiplication product for the outcome of hepatotoxicity, especially at arbitrarily pre-specified cut-off points of 1 500 mg/L IU/L (10 000 micromol/L IU/L) as identified by Sivilotti et al.13 as well as 10 000 mg/L IU/L (66 000 micromol/L IU/L). We also wished to describe the variation over time of the multiplication product in patients who developed hepatotoxicity and who had more than one detectable paracetamol concentration. Case search strategy The clinical toxicology database was interrogated for presentations meeting the inclusion/exclusion criteria. Patients with laboratory evidence of paracetamol overdose were recorded as such in the database. The narrative description of each case identified was then accessed. Electronic preformatted case reports were generated for each case, organised into a study database and analysed by the investigators. SPSS (v23) was used to analyse these results.
Results Of a total of 3823 paracetamol ingestions (Table 1), there were 2743 acute single ingestions, 452 delayed ingestions, 426 staggered and 202 supratherapeutic ingestions. 34 patients developed hepatotoxicity: 13 staggered, 10 acute single, 10 delayed and 1 supratherapeutic. Altogether, 1304 patients received acetylcysteine, which was administered intravenously in all cases. The entire 300 mg/kg UK acetylcysteine regimen was administered to all patients who developed hepatotoxicity, and extended in 25 of these until INR and ALT were improving. Two patients died in hospital, neither attributed to paracetamol toxicity, and 3 patients were transferred to a specialised liver unit for further management (none of whom received transplant). Serum ALT activity was measured in almost all cases (3783 or 99%), on average 3.5 2.0 h after presentation. However, for one-third of the patients who developed hepatotoxicity, the serum paracetamol concentration was either undetectable (12 cases) or not done (2 cases), leaving 3112 patients for whom a multiplication product could be calculated in the primary analysis. The overall diagnostic accuracy of the multiplication product is shown in Table 2, and as a receiver operator curve in Fig. 1 (area under the curve (AUC): 0.994 (95% confidence interval [CI]: 0.988–1.00), departure from 100% sensitivity at 3961 mg/L IU/L). Using the upper cutpoint of 10 000 mg/L IU/L, the first measured product had a positive likelihood ratio over 200 for hepatotoxicity, identifying 16 of the 20 patients (sensitivity 80%, specificity 99.7%). None of the 2831 patients with an initial product below 1 500 mg/L IU/L developed hepatotoxicity (likelihood ratio near zero). Clinical Toxicology vol. 53 no. 8 2015
APAP product to predict hepatotoxicity in OD 809 Table 1. Comparison of demographics and clinical variables between patients with and without hepatotoxicity.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Female count (%) Age, years Initial Paracetamol concentration, mg/L Initial ALT all ingestions, IU/L Ingestion type, count (%) Single acute Delayed Staggered Supratherapeutic Acetylcysteine treatment, count (%) within 8 h of overdose more than 8 h (%) None
Hepatotoxicity (n 34)
No hepatotoxicity (n 3789)
30 (88%) 29 (26–32) 140 (95–147) 258 (86–1993)
2206 (58%) 30 (26–34) 99 (66–131) 25 (16–36)
10 (29.5%) 10 (29.5%) 13 (38%) 1 (3%)
2733 (72%) 442 (12%) 413 (11%) 201 (5%)
1 (3%) 33 (97%) 0
660 (18%) 610 (16%) 2519 (66%)
p value 0.001 0.7 0.001 0.001 0.0001
0.0001
LT, alanine aminotransferase. A Hepatotoxicity defined as ALT 1000 IU/L. Values in tables are reported as median (interquartile range) or count (percentage).
In the secondary analysis imputing a paracetamol concentration of 5 mg/L when undetectable, again none of the patients with hepatotoxicity had an initial product 1 500, four were between 1 500 and 10,000 and eight presentations were 10 000 mg/L IU/L. None of these 12 patients were transferred to a liver unit, and the highest peak INR was 2.8.
typically 160 IU/L and multiplications products range: 11 543–52 800 mg/L IU/L). The receiver operating characteristic (ROC) curve for all single acute ingestions (0–24 h) (Fig. 3) yielded an AUC of 0.995 (95% CI, 0.99–1.00), and departed from 100% sensitivity at 4053 mg/L IU/L. Amongst single acute ingestions presenting 8 or more hours post ingestion, the sensitivity and specificity of a product 10,000 mg/L IU/L in the 8–24 h increased to 88.8% (50.1,99.4%) and 100% (99.8,100%), respectively (ROC AUC: 0.999; 95% CI, 0.998–1.00), departing 100% sensitivity at 6519 mg/L IU/L. Only three of the patients who developed hepatotoxicity had multiple detectable paracetamol concentrations. All three had received acetylcysteine within 4 h of presentation, and all received extended treatment with acetylcysteine. The individual elimination half-life of paracetamol ranged from 5.0 to 7.8 h. The multiplication product remained similar over several hours in these three patients, always above 10 000 mg/L IU/L (Fig. 4). Turning to the 10 acute single ingestions (0.4% of total) with a multiplication product greater than 10 000 mg/L IU/L on presentation yet did not go on to develop hepatotoxicity,
Acute single ingestions Of the 2703 acute single ingestions (255 of whom received acetylcysteine), 40 did not have both a paracetamol concentration and ALT measured, and 10 developed hepatotoxicity. 18 (0.74%) had an initial multiplication product over 10 000 mg/L IU/L; 8 of these went on to develop hepatotoxicity (sensitivity 80%, specificity 99.6%; Fig. 2). The peak INR ranged from 1.3 to 3.0 in these 8 patients. All but 1 patient (7 h) of the 8 had their first multiplication product 8 h post ingestion. As shown in Fig. 2, all 8 had an abnormal ALT activity when first measured (range: 103–792 IU/L), unlike the others in whom the elevated product was primarily due to a high serum paracetamol concentration (initial ALT
Table 2. Diagnostic accuracy of paracetamol aminotransferase multiplication product regardless of paracetamol overdose presentation. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not obtained Total
Hepatotoxicity
No hepatotoxicity
Likelihood ratio (95% CI)
16 4 0 20
10 251 2831 3092
250 (130,480) 2.5 (1,6) 0 3112
12
380
3.52 (2.21,5.60)
2 34
317 3789
0.70 (0.18,2.71) 3823
t the pre-specified cutpoints, the interval likelihood ratio is shown after excluding patients without a measurable paracetaA mol concentration at the time of the first available aminotransferase. For these excluded patients, a positive likelihood ratio is calculated for having the exclusion criterion present. Copyright © Informa Healthcare USA, Inc. 2015
810 A. Wong et al.
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
rise above 100 IU/L. One patient had a normal ALT of 22 IU/L and paracetamol concentration of 186 mg/L 7 h post overdose (product of 4092 mg/L IU/L) but received acetylcysteine 11 h post overdose and developed hepatotoxicity. None of the 2479 patients with a product below 1 500 mg/L IU/L developed hepatotoxicity following acute overdose.
Fig. 1. ROC curve of the paracetamol-aminotransferase multiplication product (APAP ALT) in predicting hepatotoxicity for all classifications of paracetamol overdose (colour version of this figure can be found in the online version at www.informahealthcare.com/ctx).
the multiplication products in each case were calculated using laboratory tests taken less than 8 h post ingestion (Fig. 5). Nine of these had acetylcysteine given within 8 h of ingestion and all by 12 h. Eight had an ALT 2–3 times the upper limit of normal at the time of presentation. The peak ALT ranged from 38 to 293 IU/L measured 3–56 h post ingestion. None had a peak INR 2 (mean: 1.1, 95% CI, 1.0–1.3). 216 (8.0%) of the 2703 acute single ingestion presentations had a multiplication product 1 500 mg/L IU/L but below 10 000 mg/L IU/L, and only 2 (1%) developed hepatotoxicity (Table 3). The peak INR from this subset was 1.3 (range, 0.9–1.3) and ALT ranged from 8 to 954 (median, 24) IU/L. All but 7 had acetylcysteine started within 8 h of ingestion, in which case ALT did not
Delayed paracetamol ingestions Of the 452 delayed paracetamol ingestions (acetylcysteine administered to 433), the paracetamol concentration was never measured (272, 60.2%) or was undetectable (140, 30.9%) in the vast majority, reflecting usual recommended practice in the United Kingdom and the delay to care (average time to presentation 53 h post ingestion; 95% CI, 65–85). Ten (2%) patients had or developed hepatotoxicity; of these, 8 had an undetectable paracetamol concentration at the time of presentation. The other two cases had multiplications products of 52 690 and 96 380 mg/L IU/L (presentation at 46 and 51 h post ingestion, and peak INR: 1.6 and 3.0, respectively). Five of the ten presented with established hepatotoxicity (initial ALT range: 2366–10116 IU/L, INR range: 1.4–2.6, 46 h to 70 h post ingestion). The diagnostic accuracy is shown in Table 4. All of the patients who presented with undetectable paracetamol concentrations survived. Staggered paracetamol ingestions There were 426 staggered overdoses, of whom 425 received acetylcysteine. 51 had undetectable (46, 10.8%) or no (5, 1.2%) paracetamol concentrations at time of presentation (Table 5). There were 13 cases of hepatotoxicity from staggered overdose and all received acetylcysteine at the time of initial medical assessment. Peak INR ranged from 1.5 to 4.4 and ALT ranged from 12 to 10 956 IU/L.
Fig. 2. Initial serum ALT concentrations matched with simultaneous serum paracetamol concentrations in single acute overdose patients with multiplication product 10 000 mg/L IU/L. Patients are classified based on hepatotoxicity () (peak ALT 1000 IU/L; n 8) versus no hepatotoxicity () (n 10). Any point under the dotted diagonal line represents a multiplication product 10 000 mg/L IU/L. Clinical Toxicology vol. 53 no. 8 2015
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
APAP product to predict hepatotoxicity in OD 811
Fig. 3. ROC curve of the paracetamol-aminotransferase multiplication product (APAP ALT) in predicting hepatotoxicity secondary to acute single ingestion paracetamol overdose for patients presenting 0–24 h after ingestion (colour version of this figure can be found in the online version at www.informahealthcare.com/ctx).
Supratherapeutic paracetamol ingestions There were 202 supratherapeutic ingestions (199 treated with acetylcysteine), of whom only 8 (4%) had a detectable paracetamol concentration. Seven of these had a paracetamol multiplication product 1,500 mg/L IU/L. Only one patient (ingestion over 4 days) had an initial multiplication product 10 000 mg/L IU/L (initial APAP, 58 mg/L; ALT, 3042 IU/L), but subsequent aminotransferase concentrations fell and peak INR was only 1.7 with a standard course of acetylcysteine started 1 h post presentation.
Discussion A risk prediction calculation of developing hepatotoxicity in paracetamol overdose treated with an antidote is a much needed tool, whether trying to identify high-risk patients who
Fig. 4. Multiplication product relative to initiation of acetylcysteine in the three patients who developed hepatotoxicity following single acute ingestion in whom serial paracetamol concentrations were available. Copyright © Informa Healthcare USA, Inc. 2015
might benefit from more intensive therapy or abbreviating unnecessary treatment and reducing uncertainty for low-risk patients. Since liver function test abnormalities may occur only a day or more post ingestion, there can be considerable uncertainty for the patient and treating clinicians following excess acetaminophen exposure. We report the diagnostic accuracy of a novel risk instrument, the paracetamol-aminotransferase multiplication product, in a very large cohort of patients across the spectrum of categories of paracetamol overdose. We observed that a paracetamol-aminotransferase multiplication product has high diagnostic accuracy for hepatotoxicity whether the overdose is acute, staggered, delayed or supratherapeutic. An initial multiplication product 10 000 mg/L IU/L was strongly associated with hepatotoxicity, unless calculated within 8 h of a single acute overdose, while patients with a product 1 500 mg/L IU/L were highly unlikely to develop hepatotoxicity. A multiplication product between 1 500 and 10 000 mg/L IU/L on presentation had only a modestly increased likelihood of developing hepatotoxicity. It should be emphasised that the paracetamol nomogram remains useful for predicting which single acute overdoses are unlikely to develop hepatotoxicity in the absence of treatment with acetylcysteine.8–12 However, the nomogram cannot be used for many patients because of time unknown, staggered, delayed or supratherapeutic ingestions of paracetamol. The nomogram is also not used to predict risk in patients who receive treatment with acetylcysteine. Instead, our findings represent the risk of hepatotoxicity in treated patients. Furthermore, the paracetamol-aminotransferase product can be calculated without certain knowledge of the time of ingestion, or even the requirement that the ingestion be taken at a single point in time. For example, in a rural setting a paracetamol-aminotransferase product may lead to earlier identification, closer monitoring and discussion about transferring patients to facilities with an intensive care unit or specialist service. This is especially useful in rural hospitals with minimal inpatient resources and availability of laboratory facilities on predicting the time course of treatment for a patient. Also it may help decide direct ward admissions for prolonged acetylcysteine treatment rather than a short stay observation unit. In addition to identifying patients who have a very high likelihood of developing hepatotoxicity, the paracetamolaminotransferase multiplication product appears able to identify patients at a very low risk. This result is noteworthy in the context of recent efforts to further abbreviate acetylcysteine treatment duration,17 and also the controversial decision to lower the treatment threshold substantially in the United Kingdom.18 Yet a risk prediction instrument with a sensitivity near 100% will be needed to reduce unnecessary treatment, given the relatively low cost of an individual course of treatment and the high stakes of preventing avoidable fulminant hepatic failure. Reducing unnecessary treatment is particularly important for paracetamol exposure, given the frequency of acetylcysteine administration for this indication. We believe that the multiplication product offers such a high sensitivity in addition to simplicity and
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
812 A. Wong et al.
Fig. 5. Paracetamol-Aminotransferase Multiplication product relative to interval from ingestion to initiation of acetylcysteine following single acute overdose. Only patients with initial multiplication products 10 000 mg/L IU/L are shown, classified based on those who developed hepatotoxicity () and did not () (n 10).
availability. It is worth also noting that we have characterised the outcome of hepatotoxicity, considered to be a necessary but not sufficient criterion of hepatic failure. The outcomes of coagulopathy and hepatic failure were too infrequent in this cohort to characterise the diagnostic accuracy of the product for these more serious but less common outcomes. Moreover, while a threshold of 1 500 mg/L IU/L enjoyed 100% sensitivity, more work is needed to identify the optimal cutpoint and further reduce the lower confidence band of this estimate. Dichotomising the product around a cutpoint also fails to take advantage of the full spectrum of information given the continuous nature of the predictor, whereas customising treatment intensity (e.g. cumulative dose of antidote administered) based on the product value would also be clinically useful. Several single acute ingestion patients who had paracetamol multiplication products greater than 10 000 (mg/L IU/L) did not go on to develop hepatotoxicity. The majority of this group presented early, had mostly elevated paracetamol concentrations contributing to the elevated product and also received acetylcysteine within 8 h of acute overdose. Meanwhile the early administration of acetylcysteine in acute overdose is known to be protective.19 Chomchai et al. have previously reported in an independent validation study that the multiplication product had higher specificity in patients presenting at least 8 h after acute overdose.20 Our study supports this finding and suggests that a multiplication product calculated within 8 h of single ingestion could be repeated 12 h later to allow optimal specificity. Hence in the
small cohort where the paracetamol concentrations are still high and the ALT is mildly elevated the standard regimen of acetylcysteine may not be enough to prevent hepatotoxicity. Other modalities such as haemodialysis or increased acetylcysteine regimens may help in these situations but further investigation is needed.21 A 12-h interval between paracetamol measurements would be expected to allow the paracetamol concentration to fall by an order of magnitude ( 3 half-lives if absorption complete and elimination is normal), and the product could also fall by the same degree if the aminotransferases remained unchanged. We had initially hoped to use only multiplication products calculated at least 8–12 h post ingestion in this study, but discovered that too few patients had serial measurements of the falling paracetamol concentration. While repeat testing of post-peak paracetamol concentrations has historically been discouraged in part for cost considerations, we believe serial testing to be essential for any patient being treated with acetylcysteine until the paracetamol concentration is undetectable. In addition to testing ‘liver function’ (i.e. paracetamol elimination), serial testing is inexpensive and would avoid the pitfall of premature discontinuation of acetylcysteine after 21 h, especially given reports of persistently elevated paracetamol concentrations.21 Of the three single acute overdose presentations that developed hepatotoxicity and had serial paracetamol concentrations, elimination half-life was greater than 4 h, and the multiplication product remained relatively stable, consistent with earlier studies (mean, 7.6/ 0.8 h).8,9,13
Table 3. Diagnostic accuracy for hepatotoxicity in single acute overdose patients presenting within 24 h. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total
Hepatotoxicity 8 2 0 10
No hepatotoxicity 10 214 2469 2693
Likelihood ratio (95% CI)
Sensitivity % (95% CI)
Specificity% (95%CI)
PPV, NPV%
215 (108,430) 2.52 (0.7,8.7) 0 (0,0.001)
80 (44,96)
99.6 (99.3,99.8)
44,99.9
100 (66,100)
92 (91,93)
4.4,100
PPV, positive predictive value; NPV, negative predictive value. Clinical Toxicology vol. 53 no. 8 2015
APAP product to predict hepatotoxicity in OD 813 Table 4. Diagnostic accuracy of multiplication product in delayed paracetamol ingestions. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not done Total Secondary analysis: products for imputed 5 mg/L value for 10 mg/L concentrations 10 000 1 500–10 000 1 500 Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Likelihood ratio (95% CI)
Hepatotoxicity No hepatotoxicity 2 0 0 2
0 9 29 38
infinity 0 0
8 0 10
132 272 442
2.68(1.9,3.7) 452
7 1 0
0 2 130
infinity (2.3,inf) 8.25 (0.83,81.63) 0 (0,0.02)
secondary analysis was performed for presentations with paracetamol concentrations below the detectable limit A ( 10 mg/L). A value of 5 mg/L was imputed for paracetamol to obtain a multiplication product.
In our study, the majority of delayed ( 24 h after ingestion) paracetamol presentations did not have a paracetamol concentration measured on arrival, consistent with national recommendations. Instead acetylcysteine was typically started based on history and pending the results of other tests taken on arrival. More than half of these patients presented in established hepatotoxicity, hence limiting the utility of calculating the multiplication product to identify this outcome as described in this study. Nevertheless, we wished to characterise the product across all presentations. We assigned a low, non-zero value for the paracetamol concentration when it was undetectable, although it may well have been much lower. We acknowledge that as time passes, other factors such as pH, encephalopathy and coagulopathy become more useful to identify the most serious outcomes. 22,23 In regards to the staggered ingestions of paracetamol that developed hepatotoxicity, serum paracetamol concentrations were not measured in the majority (again consistent with national recommendations) and therefore products were not calculated in all patients. For those in whom a
product could be calculated, as well as in the secondary analysis, no patients with a product 1 500 mg/L IU/L developed hepatotoxicity. The multiplication product may be particularly helpful in staggered overdoses that receive acetylcysteine as there is often an opportunity to administer a much shorter course of acetylcysteine once a favourable outcome is certain. In a study of 410 patients of whom 264 were acute overdoses presenting within 8 h, Al-Hourani et al.24 reported that an abnormal serum ALT activity at presentation was present in all 16 patients who developed hepatotoxicity, for an estimated sensitivity between 81% and 100% (95% CI). This high sensitivity is consistent with Canadian data involving 94 acute overdose patients in which ‘almost all aminotransferase concentrations measured 12 or more hours post ingestion exceeded 50 IU/L and exceeded 100 IU/L when measured more than 24 h post ingestion’.25 One of the uncertainties involving the clinical application of the multiplication product is therefore the interpretation of a product calculated too soon after overdose, or the closely related issue of
Table 5. Multiplication product in staggered paracetamol ingestions with primary and secondary analyses. Multiplication product (mg/L IU/L) 10 000 1 500–10 000 1 500 Total Excluded: Paracetamol concentration 10 mg/L at presentation Paracetamol concentration not obtained Total Secondary analysis: products for imputed 5 mg/L value for 10 mg/L concentrations 10 000 1 500–10 000 1 500
Hepatotoxicity No hepatotoxicity
Likelihood ratio (95% CI)
5 2 0 7
0 17 345 352
infinity 5.75(1.63,20.3) 0
4 2 13
42 3 413
3.03 (1.27,7.18) 21.2 (3.86,116) 426
2 2 0
0 1 41
infinity (2.4,inf) 21.0 (2.40,184) 0 (0,0.07)
secondary analysis was performed for presentations with paracetamol concentrations below the detectable limit ( 10 mg/L). A A value of 5 mg/L was imputed for paracetamol to obtain a multiplication product. Copyright © Informa Healthcare USA, Inc. 2015
814 A. Wong et al. being high despite a normal aminotransferase concentration. Notwithstanding boundary conditions which define a very low risk (e.g. normal ALT at initiation of acetylcysteine, or timed serum paracetamol concentration below the nomogram treatment line), the product yields a continuous risk measure for all patients at some risk. Other tests such as protein adducts and microRNA are not readily available yet but may help in the future in predicting development of hepatotoxicity secondary to paracetamol misuse.26,27
Clinical Toxicology Downloaded from informahealthcare.com by RMIT University on 08/18/15 For personal use only.
Conclusion In a large cohort of patients presenting to hospital for paracetamol overdose of any kind, the initial paracetamol-aminotransferase multiplication product has a very high diagnostic accuracy for predicting hepatotoxicity in patients treated with acetylcysteine. Regardless of ingestion type, a product 10 000 mg/L IU/L was associated with a very high likelihood, and 1 500 mg/L IU/L with a very low likelihood, of developing hepatotoxicity. We urge caution using the product within 8 h of an acute overdose, as the serum paracetamol concentration itself may contribute to a high product, but recommend instead repeating the product approximately 12 h later for optimal specificity.
Declaration of interest The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.
References 1. National Poisons Information Service UK. Annual Reports. Available from http://legacytools.hpa.org.uk/Publications/ChemicalsPoisons/ NationalPoisonsInformationServiceAnnualReports/1308NPISannual report20122013/.2013; (accessed 8 July 2015). 2. Bronstein AC, Spyker DA, Cantilena LR Jr, Rumack BH, Dart RC. 2011 Annual report of the American Association of Poison Control Centers’ National Poison Data System (NPDS): 29th Annual Report. Clin Toxicol (Phila) 2012; 10:911–1164. 3. Buckley N, Eddleston M. Paracetamol (acetaminophen) poisoning. Clin Evid 2005; 14: 1738–1744. 4. Linden CH, Rumack BH. Acetaminophen overdose. Emerg Med Clin North Am 1984; 2:103–119. 5. Ostapowicz G, Fontana RJ, Schiødt FV, Larson A, Davern TJ, Han SH, et al. Results of a prospective study of acute liver failure at 17 tertiary care centers in the United States. Ann Intern Med 2002; 137:947–954. 6. Sood S, Howell J, Sundararajan V, Angus PW, Gow PJ. Paracetamol overdose in Victoria remains a significant health burden. J Gastroenterol Hepatol 2013; 28:1356–1360. 7. Smilkstein MJ, Knapp GL, Kulig KW, Rumack BH. Efficacy of oral N-acetylcysteine in the treatment of acetaminophen overdose. Analysis of the national multicenter study (1976 to 1985). N Engl J Med 1988; 319:1557–1562.
8. Rumack BH, Peterson RC, Koch GG, Amara IA. Acetaminophen overdose. 662 Cases with evaluation of oral acetylcysteine treatment. Arch Intern Med 1981; 141:380–385. 9. Prescott LF, Roscoe P, Wright N, Brown SS. Plasma-paracetamol half-life and hepatic necrosis in patients with paracetamol overdosage. Lancet 1971; 1:519–522. 10. Dart RC, Rumack BH. Patient-tailored acetylcysteine administration. Ann Emerg Med 2007; 50:280–281. 11. Woo OF, Mueller PD, Olson KR, Anderson IB, Kim SY. Shorter duration of oral N-acetylcysteine therapy for acute acetaminophen overdose. Ann Emerg Med 2000; 35:363–368. 12. Heard KJ. Acetylcysteine for acetaminophen poisoning. N Engl J Med 2008; 359:285–292. 13. Sivilotti ML, Green TJ, Langmann C, Yarema M, Juurlink D, Johnson D. Multiplying the serum aminotransferase by the acetaminophen concentration to predict toxicity following overdose. Clin Toxicol 2010; 48:793–799. 14. Green SL, Wood DM, Gawarammana IB, Warren-Gash C, Drake N, Jones AL, Dargan PI. Improvement in the management of acutely poisoned patients using an electronic database, prospective audit and targeted educational intervention. Postgrad Med J 2008; 84:603–608. 15. Ferner RE, Dear JW, Bateman DN. Management of paracetamol poisoning. BMJ 2011; 342:d2218. doi: 10.1136/bmj.2218 16. Makin AJ, Wendon J, Williams R. A 7-year experience of severe acetaminophen-induced hepatotoxicity (1987–1993). Gastroenterology 1995; 109:1907–1916. 17. Bateman DN, Dear JW, Thanacoody HK, Thomas SH, Eddleston M, Sandilands EA, et al. Reduction of adverse effects from intravenous acetylcysteine treatment for paracetamol poisoning: a randomised controlled trial. Lancet 2014; 383:697–704. 18. Bateman DN, Carroll R, Pettie J, Yamamoto T, Elamin ME, Peart L, et al. Effect of the UK’s revised paracetamol poisoning management guidelines on admissions, adverse reactions and costs of treatment. Br J Clin Pharmacol 2014; 78:610–618. 19. Prescott LF, Illingworth RN, Critchley JA, Stewart MJ, Adam RD, Proudfoot AT. Intravenous N-acetylcysteine: the treatment of choice for paracetamol poisoning. Br Med J 1979; 2:1097–1100. 20. Chomchai S, Chomchai C. Predicting acute acetaminophen hepatotoxicity with acetaminophen-aminotransferase multiplication product and the Psi parameter. Clin Toxicol 2014; 52: 506–511. 21. Sivilotti MLA, Juurlink DN, Garland JS, Lenga I, Poley R, Hanly N, Thompson N. Antidote removal during haemodialysis for massive acetaminophen overdose. Clin Toxicol 2013; 51:855–863. 22. Schiødt FV, Bondesen S, Tygstrup N, Christensen E. Prediction of hepatic encephalopathy in paracetamol overdose: a prospective and validated study. Scand J Gastroenterol 1999; 34:723–728. 23. O’Grady JG, Alexander GJ, Hayllar KM, Williams R. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 1989; 97: 439–445. 24. Al-Hournai K, Mansi R, Pettie J, Dow M, Bateman DN, Dear JW. The predictive value of hospital admission serum alanine transaminase activity in patients treated for paracetamol overdose. Q J Med 2013; 106:541–546. 25. Green TJ, Sivilotti ML, Langmann C, Yarema M, Juurlink D, Burns MJ, Johnson DW. When do the aminotransferases rise after acute acetaminophen overdose? Clin Toxicol 2010; 48:787–792. 26. Davern TJ, James LP, Hinson JA, Polson J, Larson AM, Fontana RJ, et al. Measurement of serum acetaminophen-protein adducts in patients with acute liver failure. Gastroenterology 2006; 130:687–694. 27. Hornby RJ, Starkey LP, Dear J, Goldring C, Park BK. MicroRNAs as potential circulating biomarkers of drug-induced liver injury: key current and future issues for translation to humans. Expert Rev Clin Pharmacol 2014; 7: 349–362.
Clinical Toxicology vol. 53 no. 8 2015
