Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jun 4.
Published in final edited form as: J Clin Pharmacol. 2019 Dec 4;60(5):595–604. doi: 10.1002/jcph.1555

Population Pharmacokinetic Modeling of Acetaminophen Protein Adducts in Adults and Children

Sibo Jiang 1, Kumpal Madrasi 1, Tanay Samant 1, Chakradhar Lagishetty 1, Valvanera Vozmediano 1, Angela Chiew 2,3, Susan M Abdel-Rahman 4, Laura P James 5, Stephan Schmidt 1
PMCID: PMC7643159  NIHMSID: NIHMS1637242  PMID: 31802503

Abstract

Acetaminophen protein adducts (adducts) are a well-established biomarker to diagnose acetaminophen toxicity. To date, the quantitative relationship between acetaminophen exposure, which drives the adduct formation, and adduct exposure remains to be established. Our study characterized the adduct formation and disposition in adults using the approach of population parent-metabolite modeling. It demonstrated formation-limited pharmacokinetics (PK) for adducts in healthy subjects. This finding expands the existing knowledge on adduct PK that showed an apparent long half-life. We then allometrically scaled the adduct PK model to children, simulated the adduct profiles, and compared these simulated profiles to observed ones in an independent cohort of children. The scaled model significantly over-predicted the adduct concentrations in children early on in treatment and under-predicted concentrations following repeated acetaminophen doses. These results suggest that children demonstrate different adduct PK behavior from that of adults, most likely due to increased reactive metabolite detoxification in children. In summary, we described the first PK model linking acetaminophen and acetaminophen protein adduct concentrations, which provides a semi-mechanistic understanding of varying profiles of adduct exposure in adults and children.

Keywords: Acetaminophen protein adducts, parent-metabolite model, biomarkers, hepatotoxicity, pediatrics

Introduction

Acetaminophen (APAP, paracetamol) is a commonly used analgesic and antipyretic small molecule (151 Da) agent, which is available in both over-the-counter and prescription medications. APAP has the potential to cause liver injury with therapeutic administration when dosed at or above the upper end of the therapeutic range.1 After administration, APAP is primarily metabolized by glucuronidation and sulfation in the liver,2 though a small portion (~5%) undergoes metabolism to form the toxic intermediate N-acetyl-p-benzoquinone imine (NAPQI), which can bind to cysteine groups on proteins to form APAP protein adducts (adducts). The reactive metabolite NAPQI is primarily detoxified by hepatic glutathione, but higher doses deplete the glutathione pool resulting in liver injury. Hepatocyte rupture releases adducts into the systemic circulation where they can be quantified by sensitive high-performance liquid chromatography with electrochemical (HPLC-EC) detection.3 To date, more than 20 different hepatic proteins have been reported to form adducts with a putative molecular mass of ≥ 50 kDa.46

These adducts represent a specific biomarker for APAP-induced liver injury. Current available tests for liver injury, aspartate aminotransferase (AST) and alanine aminotransferase (ALT), are non-specific to APAP. They can be elevated due to a number of conditions including but not limited to ischemic liver injury, alcoholic hepatitis, and viral hepatitis.7 AST and ALT are consequently limited in their role in identifying the underlying etiology of APAP toxicity. In contrast, the detection of adducts in serum is causally linked to the APAP ingestion history. Although the dose-exposure relationship of adducts has been described in animal models and clinical studies811, the pharmacokinetic (PK) behavior of adducts is still not yet fully understood. Furthermore, APAP exposure, which drives the dynamics of adducts, is subject to considerable inter-individual variability as a result of intrinsic (e.g., genetic polymorphisms and varied abundance of metabolic enzymes) and extrinsic factors (e.g., formulations and dosing frequency), which subsequently impact the adduct concentration-time profile. Bridging between APAP exposure and adduct exposure is critically important to fully understand the contributions of ontogenic and pathophysiologic factors to APAP detoxification. By extension, knowledge of adduct PK behavior offers the potential for improving the medical management of APAP hepatotoxicity.

The objective of our current study was to establish a combined APAP and APAP protein adduct model in order to explore the impact of dosing frequency and enzyme ontogeny on the adduct kinetics in children (0 to < 18 years unless otherwise noted). To this end, we reanalyzed the clinical APAP and adduct data, which were previously reported by our group8, using a semi-mechanistic population modeling approach. First, we developed a parent-metabolite PK model to describe APAP and adduct concentration-time profiles in healthy adult subjects. Once qualified, the adult model was scaled to children in order to assess the adduct PK and its impact on APAP detoxification in this special population.

Materials and Methods

Study design and analytical methods

Clinical data were obtained from two published clinical studies enrolling both adults8,12 and children11, respectively. Both studies were reviewed and approved by institutional review boards. Informed consent was obtained from each subject or the subject’s parents/legal guardians prior to participation.

Briefly, the adult study was an open-label, crossover trial to evaluate the PK of two APAP formulations: immediate release (IR, 500 mg tablets) and extended release (ER, 665 mg tablets, 69% slow release and 31% fast release) formulation.8,12 Data obtained from 15 adults (9 from ER formulation, 6 from IR formulation) were used for model development. These participants aged 27–46 years (body weight: 62 – 84 kg) received a single dose of APAP (80 mg/kg) either as IR or ER formulation. Blood samples immediately before APAP dose and at 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, and 12 h post-dose were collected for all participants. Additional blood samples at 16 and 24 h post-dose were available in 4 of the 6 subjects (IR) and 6 of 9 subjects (ER). Liver function was normal for all participants.

The pediatric study was a prospective, observational study, which enrolled 181 hospitalized patients who received repeated APAP administration (mostly 2–19 doses, 5–20 mg/kg) up to 26 days.11 APAP formulations administered consisted of injections, suppositories, elixirs, solutions, and tablets. Scheduled blood samples were collected 8 and 24 hours after the first dose of APAP and 48 hours after discontinuation of APAP. Additional samples (2–5 per patient) were collected as scavenged blood samples during the course of administration. A total of 1034 adduct measurements were obtained with a median of 4 samples (5th −95th percentiles: 1–14) collected from each patient. APAP plasma concentrations were not measured in this cohort.

Samples were stored at −80°C and analyzed using the previously reported HPLC-EC adduct assay13. Assay modifications included centrifugal gel filtration and high-efficiency proteolytic digestion of APAP residuals from adducts. The lower limit of quantification (LLOQ) for this assay was 0.03 nmol/mL with a coefficient of variation (CV) of less than 15%.

Population pharmacokinetic modeling

The model was developed in a stepwise fashion: 1) characterization of adduct PK in adults for which APAP and adduct measurements were both available; 2) extrapolation of the model to children using an allometric scaling approach; and 3) assessment of the adduct PK in children using the scaled adult model and the observed adduct levels in pediatrics. The structural PK model was selected first, followed by statistical model and covariate model analysis.

Characterization of adduct pharmacokinetics in adults

The parent model was parameterized in terms of oral PK parameters. For example, CL/FIR represents the oral clearance for IR formulation; the oral clearance for ER formulation was subsequently derived from the estimates of CL/FIR and FER/FIR (Table 1). Specifically, APAP serum profiles of the two formulations were first used to develop a single parent model including two oral bioavailabilities (FIR and FER). Next, the parent model was linked to a metabolite model to describe the adduct PK profile. A sensitivity analysis of the fraction of the extended-release component of the ER formulation was conducted to obtain the best goodness of model fitting. In addition, one-, two-, and three-compartment models, with or without lag time were tested to characterize the time-course of APAP in serum.

Table 1.

Parameter estimates of the parent-metabolite model in adults.

Parent Modela Metabolite Modelb
Parameters Value RSE (%) Parameters Value RSE (%)
Fixed effect
CLAPAP/FIR (L/h) 11.1 7 Vmax (µmol/h) 0.248 13
 k for CL/F 1.5 23 Km (mg/L) 3.56 15
VC/FIR (L) 30.7 14 CLadduct (L/h) 2.21 16
 k for VC 1.1 14 Vadduct (L) 7 (Fixed) -
VP/FIR (L) 19.9 12
Q/FIR (L/h) 22.2 33
Ka1-order in IR & ER (1/h) 4.33 38
Fraction0-order in ER 62% (Fixed) -
Lag time0-order in ER (h) 0.29 24
Duration0-order in ER (h) 3.66 4
FER/FIRc 0.61 10
Random effect
ω_CLAPAP 17% 12 ω_Vmax 21% 26
ω_Vc 39% 11 ω_Km 63% 34
ω_Q 70% 30 ω_CLadduct 29% 21
ω_Ka 84% 30
Prop Error 14% 21 Prop Error 10% 12
Add Error (mg/L) 1.23 28 Add Error (µmol/L) 0.005 17
a:

This parent model solely served to provide APAP input guiding development of the metabolite model in adults. This parent model was not scaled to children for simulation.

b:

This metabolite model was scaled to children to simulate the adduct profile in conjunction with the literature parent model.

c:

FER/FIR * CLAPAP/FIR serves to derive CLAPAP/FER

For unique identification of the metabolite model, adducts (≥ 50 kDa) were assumed to distribute within the plasma and interstitial fluid, which generally applies to large molecules. Accordingly, their volume of distribution at steady state (Vss) was fixed to 7 L, the typical Vss for large molecule drugs (mostly monoclonal antibodies) in 70 kg adults14,15. In this way, the estimation of other adduct PK parameters (e.g., CLadduct) was neither confounded by an unmeasurable total amount of adducts nor uncertain oral bioavailability of APAP in the present study. In other words, with a pre-determined Vss, the estimates of the best-fit metabolite model are uniquely identified given certain model structure. With this assumption, linear and non-linear adduct formation and elimination were sequentially tested to describe the adduct profile.

An exponential error model of η (assuming η is normally distributed with a mean of 0 and a variance of ω2) was used to describe the inter-individual variability (IIV) of PK parameters. For PK parameters that show IIV, body weight and age were assessed as potential covariates by examining the η distribution in the base model. For a significant covariate, the effect of continuous covariate was modeled as follows:

θtyp=θref×(CoviCovref)k (1)

Where θtyp is the typical PK value for the subjects sharing the same covariate characteristics (denoted as Covi); θref is the typical PK value for the subjects with a reference covariate value denoted as Covref. Additionally, the residual variability of both APAP and adducts was assumed to follow a combined additive and proportional error model. The performance of the final metabolite model was qualified (internal qualification) by visual predictive checks (VPCs, 500 replicates).

Assessment of adduct pharmacokinetics in children

Since the adduct formation is directly driven by APAP plasma concentrations, accurate APAP concentrations are the prerequisite to unbiasedly assess the adduct PK behavior in children. For this purpose, a literature parent model (by Wang et al.16) was firstly tested for its predictive power in children before it was used to inform the metabolite model for assessment of adduct PK.

Compared the in-house parent model, Wang et al.’s parent model holds greater generalizability to the pediatric study since it was developed using a wide range of APAP formulations (e.g., intravenous and rectal administrations) in a large pediatric cohort across ages. Specifically, it consists of an APAP-specific maturation function as well as the general allometric equation. Scaling of various PK parameters from adults to children was performed using the following body weight-based function:

θtyp=θref×(BWi70kg)k (2)

where k is the scaling-exponent for PK parameter θ; BWi is the individual body weight; 70 kg is the typical body weight in the reference adult population. As defined by the maturation function proposed by Wang et al.16, the k for the linear clearance (CLAPAP) is a variable as a function of body weight:

k=1.20.45×BWi1.433.2+BWi1.4 (3)

For other PK parameters, k was set to a constant: 1) 1 for VdAPAP; 2) 0.75 for Q (Inter-compartmental clearance of APAP).

The scaled parent model across ages was externally validated using data from the literature (Zuppa et al).17 The literature data consist of mixed APAP concentrations collected from both children and adolescents (combined, 2 years to < 14 years17) who received either 12.5 or 15 mg/kg 15-minute IV infusion of APAP. Since the ages (29 days to < 14 years) of individual patients in this cohort were not reported, an infant aged 1 year was simulated for the APAP profile in the infant group (29 days to 2 years). Similarly, a representative subject aged 8 years was simulated for the APAP profile in a mixed children and adolescent group. Their respective body weights (9 and 25 kg, respectively) used for simulation were collected from the reported clinical growth charts.18

Once validated by external dataset, the scaled literature parent model was used to provide APAP input to the in-house metabolite model as we assume that adduct PK behavior is APAP-formulation dependent (the in-house metabolite model was built on tablet formulation data). Following the typical allometry, k was set to: 1) 1 for Vdadduct; 2) 0.75 for Vmax (as partial clearance of APAP) and CLadduct. Km was assumed to be age-independent and hence not scaled because: 1) Km represents the inherent characteristics of the relevant metabolic enzymes; and 2) same enzymes are implicated in adduct formation between adults and children. This new joint parent-metabolite was validated using the adult adduct data that were utilized to develop the metabolite model. Specifically, the observed adduct concentrations were simulated using the reported clinical dosing regimens and the body weight of the adult subjects (n=15).

After having been validated in adults, the joint parent-metabolite was scaled to children (n=181) to simulate adduct concentrations. The typical input absorption parameters for simulation are listed in Supplemental Table S1. In addition, a local sensitivity analysis was conducted using the plausible PK values (Table S1) to evaluate the influence of varied absorption processes on the concentration-time profile of adducts. To assess potential model mis-specification, each observed adduct concentration was compared to the corresponding simulated adduct concentration (median) by computing their ratio (obs./sim.). To facilitate data visualization at the early sampling times, the ratios were plotted against time on a log-log scale, which were further stratified by age and admission diagnosis. Notably, additional records of APAP formulation were not labeled in this plot as 35% of patients (mostly 2–12 year groups) received more than one formulation, which created plotting challenges.

Software and model selection criteria

Population PK analyses were performed in NONMEM (version 7.3, Icon Development Solutions) using the first-order conditional estimation method with interaction (FOCE-I). Pirana (version 2.9.2), Perl-speaks-NONMEM (PsN, version 4.2.0), and R (version 3.3.0) were used for graphic analysis and processing of NONMEM output for model evaluation. A decrease in value of more than 3.84 (p ≤ 0.05 for one degree of freedom assuming a Chi-square distribution) in objective function value (OFV) was considered statistically significant between two nested models. Other criteria to guide model selection included goodness-of-fit plots, shrinkage, the certainty of parameter estimates, and pharmacological knowledge.

Results

Characterization of adduct pharmacokinetics in adults

A two-compartment model with parallel linear and trace non-linear (adduct formation) clearances accurately described the APAP data (Figure 1). The lowest OFV was reached when the zero-order-release components of ER formulation was fixed to 62%, which is close to the reported fraction (69%)19. The flat plateaus of APAP concentrations between 2–5 hours post-dose were well captured with this final model. The bioavailability of ER formulation relative to IR formulation was estimated to be 61%, which may be attributed to the saturation of hepatic metabolism after the IR formulation and/or incomplete absorption in ER formulation12.

Figure 1:

Figure 1:

Structure of the parent-metabolite model in adults following oral APAP administration. †: Two parallel pathways (first-order and zero-order) were used to describe the mixed absorption of extended release formulation. §: lumped model to describe the adduct formation (see discussion for details).

Identical Cmax (~0.10 nmol/mL) of adduct was observed for both formulations at a similar Tmax (~9 hours) irrespective of the 3-fold difference of the peak APAP concentration indicating saturable adduct formation. This observation was adequately described by the Michaelis-Menten model (Figure 2). The Vmax for adduct formation was estimated to be 0.248 µmol/h (0.89 mg/day for APAP) demonstrating the negligible contribution of adduct formation to the total clearance of APAP. According to the parsimony principle, a one-compartment model with linear clearance was selected as the final model. The estimates of both the parent model and metabolite model are summarized in Table 1.

Figure 2.

Figure 2.

Individual fitting of the parent-metabolite model for APAP protein adducts in adults. The results were stratified by molecules (A,B and C,D) and formulations (A,C and B,D). Blue dots: observed APAP or adduct concentrations. Red line: individual prediction. Black line: population prediction.

Overall, the individual predictions (Figure 2) and VPCs (Supplementary Figure S1) plots showed good agreement between the simulated and observed adduct concentrations. The metabolite PK parameters showed varied degrees of inter-individual variability (Table 1), which was marked for Km (63%) and moderate for Vmax (21%) and CL (29%). The individual CLadduct of ID 1 in ER formulation was exceptionally low (1.48 L/h, i.e., 67% of the mean CL), which led to a remarkable deviation of individual-predicted from population-predicted adduct level (Figure 2D). No demographic characteristics were found to be associated with this observation. In addition, a minor mismatch of adduct concentrations was noted between 8 and 10 hours post-dose (Figure 2C and 2D), which is attributed to the fluctuation of adduct concentrations observed in a few patients (e.g., ID2 (IR) and ID5 (ER)). Additionally, the conditional weighted residuals (CWRES, Supplementary Figure S2) showed a slight trend after 10 hours post-dose, which is most likely associated with the missing sampling in some individuals as mentioned in the section of study design. The parameter estimates of the final model are listed in Table 1.

Assessment of adduct pharmacokinetics in children

The literature parent model predicted APAP plasma concentrations in the different pediatric groups reasonably well (Figure 3). The observed adduct concentrations were close to the predicted median value and included in the 5th-95th percentile for infant and mixed children/adolescent groups. This demonstrates that the parent model is able to provide actuate input into the metabolite model. The joint parent-metabolite model showed adequate agreement between the simulated and observed adduct concentrations (Figure S3). Particularly, the observed adduct data were included within 95th and 5th percentiles of predicted APAP concentrations. The medians of the simulation and the observations are mostly within 2- and 1/2-fold difference.

Figure 3.

Figure 3.

Qualification of the literature parent model (Wang et al.) in a pediatric group receiving a single dose of APAP. The simulations were stratified by infants (29 days to 2 years) and children & adolescents (2 years to < 14 years) with the age ranges defined by Zuppa et al. Blue dots: observed APAP concentrations. Middle solid, upper and lower dotted red line: medians, 95th and 5th percentiles of predicted APAP concentrations, respectively.

In children receiving repeated dosing (1–4 doses) of APAP for one day or less, the ratios of observed to simulated adduct concentrations were mostly less than 1 irrespective of age. This suggests that our scaled parent-metabolite model either over-predicts adduct formation or under-predicts adduct clearance or both. The situation completely reversed with multi-day APAP dosing (median: 3 days; Figure 4), whereby more than half of the computed observed vs. predicted rations exceeded 1. In addition to this bias, the ratios demonstrated slightly greater variability in children (2 to <12 years), particularly on day 1 and 10, compared with adolescents (12 to <18 years) in Figure 4A. Additionally, adduct concentrations were consistently under-predicted in patients with infection. These findings are in accordance with the previously reported significant covariates:11 highly varied adduct levels in children compared to adolescents, and higher adduct levels with infection compared to absence of infection. A similar trend of early over-prediction and later under-prediction of adducts over time was also observed in the sensitivity analysis that perturbed absorption parameters, although the magnitude of mis-prediction varied given different input values (figure not shown).

Figure 4.

Figure 4.

Simulation of the adduct concentrations in the pediatric group by the scaled parent-metabolite model. (A) ratios stratified by infants (1 year to <2 years), children (2 to <12 years), and adolescents (12 to <18 years); (B) ratios stratified suspected infection and non-infection. The median time of sampling is approximately 3.0 days after the first dose. The rising ratio over time reflects the increasing adduct concentrations following repeated dosing. The solid line represents the line of identity (ratio = 1.0); the upper and lower dotted line represent the ratios of 2.0 and 0.5, respectively.

Discussion

To date, formal adduct PK studies in liver injured patients have been typically hindered by the inaccuracy of patient self-reported dosing history, late admission to hospitals post-dose, and altered adduct formation due to use of an antidote. The clinical data used in the current study were previously analyzed using both non-compartmental analysis and one-compartment model fitting without linking to the parent drug8. However, the true adduct kinetics cannot be reliably estimated due to the possibility of flip-flop kinetics for the metabolite.20 Therefore, the objective of the current study was to develop a semi-mechanistic population PK model that links APAP to APAP protein adduct kinetics.

To our knowledge, the proposed model represents the first PK model developed to quantitatively characterize the adduct formation and disposition in a semi-mechanistic manner. The “lumped” Michaelis-Menten kinetics of adduct formation combined the following two steps together: 1) production of NAPQI mediated by cytochrome P450 enzymes, and 2) formation of adducts through conjugation of NAPQI to hepatic proteins and subsequent release of adducts to blood. It has been reported that therapeutic doses of APAP markedly stimulates the turnover rate of the cysteine pool available for the synthesis of glutathione;21 hence, the Michaelis-Menten kinetics most likely reflects a saturable adduct formation when glutathione detoxification is enhanced in response to higher APAP dose ingestion. Apparently, this homeostatic capacity of glutathione pool must hold in a certain context of APAP dose. Moreover, given the low Km (3.56 mg/L) compared with the range of APAP concentrations, the adduct formation was saturated and followed zero-order kinetics most of the time.

A one-compartment model with linear clearance adequately described the mono-exponential decline of adducts. This profile is commonly observed for large molecules with a slow input (e.g., subcutaneous injection of monoclonal antibodies).22,23 The estimated CLadduct (2.21 L/h) is much lower than that of small molecules24 but higher compared with other large molecules (e.g., factor VIII). Neither adducts nor B domain deleted recombinant Factor VIII (170 kDa) undergo neonatal Fc receptor (FcRn) recycling due to lack of binding site. However, Factor VIII showed much lower CL (0.276 L/h) than that of adducts in adults.25 Therefore, unspecific proteolysis alone seems insufficient to explain the high elimination capacity for adducts, which suggests additional elimination pathways involved (e.g., immune response). Earlier studies showed that antibodies against drug-protein adducts were formed following multiple doses of diclofenac.26 Collectively, considering the estimated low Vmax and relative large CL, the adduct PK demonstrated formation-limited pharmacokinetics/elimination at the dose level used in our study. In other words, the adduct levels will drop instantly in case the constant formation of adduct discontinues. This new finding expands the previous report that observed apparent fast formation and long half-life (~20 hours) for adducts.8

In order to assess the adduct PK behavior in children, we developed a joint parent-metabolite model in this population using the scaling functions of the literature parent model (Wang et al16) that was derived from rich clinical data. The parent model was also successfully qualified using the external data reported by Zuppa et al.,17 confirming the predictive power in children. For adducts, since no PK model or related scaling function has not been reported, the PK parameters were extrapolated to children using empirical allometry27. The allometry provides a starting point to account for the empirical relationship between PK parameters and body weight based on Kleiber’s law28,29. Consequently, mis-prediction of the adduct levels in children indicates differences in the adduct PK behavior between adults and children, which cannot solely be explained by the difference in body weight. Notably, NAPQI formation accounts for only ~5% of the total APAP elimination in healthy subjects.3033 The generated NAPQI is mostly detoxified by hepatic glutathione, leaving a trace amount of adducts formed. Therefore, the addition of a direct link between APAP and adducts (not NAPQI) makes a negligible contribution to the total APAP elimination. In this way, accurate scaling of the linear clearance (CLAPAP) by the maturation function16 is presumably sufficient to characterize the total elimination of APAP across ages.

Since the scaled model was developed from single-dose data, the adduct PK with short-term dosing is able to be compared directly between adults and children using the scaled model. The lower observed adduct concentrations in children could be attributed to low adduct formation and/or higher adduct elimination compared with adults. In animal models, young mice showed to be less susceptible to APAP toxicity due to a four-fold higher rate of glutathione turnover compared with old mice.34 Therefore, a plausible explanation is a reduced adduct formation as the result of increased NAPQI detoxification by glutathione in children. Furthermore, since adduct formation is the key event in APAP hepatotoxicity,35 this result is consistent with the lower morbidity of APAP toxicity observed in children.36,37 In addition, after day 1, the ratio of observation versus simulation shifted towards above 1, which is likely the artifact of the scaled model that was based on single-dose data; it also indicates the elevation of adduct levels following repeated dosing. While the accumulation of APAP is unlikely based on the recent data,38 an explanation could be the depletion of the glutathione pool, which led to more NAPQI converted to adducts. The inadequacy of allometry to capture the adduct profile implicate the maturation of revenant metabolic enzymes. However, neither the molecular mechanism nor the maturation of adduct formation and elimination has been fully characterized yet. For example, the glutathione synthesis and detoxification involved a very complex network and more than ten enzymes.39 Currently, the reports on the maturation of CYP2E1 spanning fetal (8–37 weeks) age are inconsistent,40,41 but it seems relatively certain that CYP2E1 > 2 years of age reached more than half of the abundance in adults.40

The present study leveraged all available data to infer the adduct PK behavior and gain knowledge on the mechanisms of APAP induced hepatotoxicity. The limitations of the present research include: 1) The metabolite model was built on a single dose (80 mg/kg) data in two formulations. The extended formulation showed a similar Cmax of APAP as that of low dose42, and the parent-metabolite model links the APAP profile (instead of dose) to adduct profile. Therefore, the developed metabolite model is highly likely applicable to the low APAP dose, but future clinical data are need to verify this assumption; 2) the present pediatric study lacks APAP measurements that are needed to rule out the possibility of APAP accumulation after multiple doses, whereas the historical findings are controversial;38,43,44 3) constant absorption parameters (F and Ka) were used for simulation of adduct levels across ages, which may not accurately reflect the related physiological changes in children; and 4) the influence of various demographic and clinical characteristics on APAP and adduct PK behaviors remain to be elucidated in children. For example, it has been reported that cardiac surgery with cardiopulmonary bypass decreases APAP clearance in children.45 Despite these limitations, altered adduct PK behavior in children compared with adults can be concluded from our analysis since this finding was drawn from the abundant adduct concentrations from a large cohort of children (n=181).

In conclusion, our study represents the first attempt to gain a semi-mechanistic understanding of the adduct PK profile and associated APAP hepatotoxicity using a quantitative modeling approach. The developed parent-metabolite model demonstrated a formation-limited elimination of adducts in healthy subjects. The lower adduct formation in children as suggested by the simulation approach agrees with the low morbidity observed in children with APAP overdose. In addition, the present study identified multiple knowledge gaps in adduct PK (e.g., mechanism of high CLadduct), which warrant future studies to further characterize adduct formation and disposition as well as to confirm some assumptions made in model building and result interpretation of the present study.

Supplementary Material

sup 01

Acknowledgments

This work was supported through a grant from the National Institute of Digestive, Diabetes, and Kidney Diseases (DK081406). The pediatric study was a collaborative effort with contributions from multiple institutions. The authors would like to acknowledge: Michael Reed, Pharm D, Akron Children’s Hospital, Akron, OH; Janice E Sullivan, MD, Kosair Children’s Hospital, Louisville, KY; Gregory L Kearns, PharmD, PhD, Mercy Children’s Hospital, Kansas City, MO; James Marshall, MD, Cook Children’s Hospital, Fort Worth, TX; Johannes van den Anker, MD, PhD, Children’s National Medical Center, Washington, DC; and Jeffery L Blumer, PhD, MD, University of Toledo, Toledo, OH. The authors would also like to acknowledge the clinical research staff who assisted in this study.

Funding source

This work was supported through a grant from the National Institute of Digestive, Diabetes, and Kidney Diseases (DK081406).

Footnotes

Conflict of Interest

Dr. Laura James is a part owner of Acetaminophen Toxicity Diagnostics (ATD), LLC. ATD is developing a rapid assay for the measurement of acetaminophen protein adducts in human blood samples.

Data Sharing

The data supporting the findings of this study may be shared on request from the co-author Laura James at JamesLauraP@uams.edu.

References

  • 1.Heard KJ. Acetylcysteine for acetaminophen poisoning. N Engl J Med. 2008;359(3):285–292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Mazaleuskaya LL, Sangkuhl K, Thorn CF, FitzGerald GA, Altman RB, Klein TE. PharmGKB summary: pathways of acetaminophen metabolism at the therapeutic versus toxic doses. Pharmacogenetics and genomics. 2015;25(8):416–426. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Davern TJ. Indeterminate acute liver failure: a riddle wrapped in a mystery inside an enigma. Hepatology. 2006;44(3):765–768. [Google Scholar]
  • 4.Qiu Y, Benet LZ, Burlingame AL. Identification of the hepatic protein targets of reactive metabolites of acetaminophen in vivo in mice using two-dimensional gel electrophoresis and mass spectrometry. J Biol Chem. 1998;273(28):17940–17953. [DOI] [PubMed] [Google Scholar]
  • 5.James LP, Mayeux PR, Hinson JA. Acetaminophen-induced hepatotoxicity. Drug Metab Dispos. 2003;31(12):1499–1506. [DOI] [PubMed] [Google Scholar]
  • 6.Curry SC, Padilla-Jones A, O’Connor AD, et al. Prolonged Acetaminophen-Protein Adduct Elimination During Renal Failure, Lack of Adduct Removal by Hemodiafiltration, and Urinary Adduct Concentrations After Acetaminophen Overdose. J Med Toxicol. 2015;11(2):169–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kasarala G, Tillmann HL. Standard liver tests. Clin Liver Dis (Hoboken). 2016;8(1):13–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.James LP, Chiew A, Abdel-Rahman SM, et al. Acetaminophen protein adduct formation following low-dose acetaminophen exposure: comparison of immediate-release vs extended-release formulations. European journal of clinical pharmacology. 2013;69(4):851–857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.McGill MR, Lebofsky M, Norris HR, et al. Plasma and liver acetaminophen-protein adduct levels in mice after acetaminophen treatment: dose-response, mechanisms, and clinical implications. Toxicol Appl Pharmacol. 2013;269(3):240–249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Heard K, Anderson VE, Lavonas EJ, Dart RC, Green JL. Serum paracetamol-protein adducts in ambulatory subjects: Relationship to recent reported paracetamol use. Biomarkers. 2018;23(3):288–292. [DOI] [PubMed] [Google Scholar]
  • 11.Jiang S, Vozmediano V, Abdel-Rahman SM, Schmidt S, James LP. Acetaminophen protein adducts in hospitalized children receiving multiple doses of acetaminophen. The Journal of Clinical Pharmacology (In press). 2019. [DOI] [PMC free article] [PubMed]
  • 12.Chiew A, Day P, Salonikas C, Naidoo D, Graudins A, Thomas R. The comparative pharmacokinetics of modified-release and immediate-release paracetamol in a simulated overdose model. Emerg Med Australas. 2010;22(6):548–555. [DOI] [PubMed] [Google Scholar]
  • 13.James LP, Letzig L, Simpson PM, et al. Pharmacokinetics of acetaminophen-protein adducts in adults with acetaminophen overdose and acute liver failure. Drug Metab Dispos. 2009;37(8):1779–1784. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Ternant D, Azzopardi N, Raoul W, Bejan-Angoulvant T, Paintaud G. Influence of Antigen Mass on the Pharmacokinetics of Therapeutic Antibodies in Humans. Clin Pharmacokinet. 2019;58(2):169–187. [DOI] [PubMed] [Google Scholar]
  • 15.Dirks NL, Meibohm B. Population pharmacokinetics of therapeutic monoclonal antibodies. Clin Pharmacokinet. 2010;49(10):633–659. [DOI] [PubMed] [Google Scholar]
  • 16.Wang C, Allegaert K, Tibboel D, et al. Population pharmacokinetics of paracetamol across the human age-range from (pre)term neonates, infants, children to adults. J Clin Pharmacol. 2014;54(6):619–629. [DOI] [PubMed] [Google Scholar]
  • 17.Zuppa AF, Hammer GB, Barrett JS, et al. Safety and population pharmacokinetic analysis of intravenous acetaminophen in neonates, infants, children, and adolescents with pain or Fever. J Pediatr Pharmacol Ther. 2011;16(4):246–261. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.US Centers for Disease Control and Prevention. Clinical Growth Charts 2000; https://www.cdc.gov/growthcharts/clinical_charts.htm#Set1.
  • 19.SmithKline Beecham (New Zealand) Ltd. Submission to the Medicines Classification Committee for Reclassification of a Medicine. 2001; https://medsafe.govt.nz/profs/class/Agendas/agen25-paracetamol.pdf. Accessed Oct,02, 2019.
  • 20.Yanez JA, Remsberg CM, Sayre CL, Forrest ML, Davies NM. Flip-flop pharmacokinetics--delivering a reversal of disposition: challenges and opportunities during drug development. Ther Deliv. 2011;2(5):643–672. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Lauterburg BH, Mitchell JR. Therapeutic doses of acetaminophen stimulate the turnover of cysteine and glutathione in man. J Hepatol. 1987;4(2):206–211. [DOI] [PubMed] [Google Scholar]
  • 22.Xu Y, Adedokun OJ, Chan D, et al. Population Pharmacokinetics and Exposure-Response Modeling Analyses of Golimumab in Children With Moderately to Severely Active Ulcerative Colitis. J Clin Pharmacol. 2019;59(4):590–604. [DOI] [PubMed] [Google Scholar]
  • 23.Nader A, Beck D, Noertersheuser P, Williams D, Mostafa N. Population Pharmacokinetics and Immunogenicity of Adalimumab in Adult Patients with Moderate-to-Severe Hidradenitis Suppurativa. Clin Pharmacokinet. 2017;56(9):1091–1102. [DOI] [PubMed] [Google Scholar]
  • 24.Goodman LS, Brunton LL, Chabner B, Knollmann BrC. Goodman & Gilman’s pharmacological basis of therapeutics. 12th ed. New York: McGraw-Hill; 2011. [Google Scholar]
  • 25.Abrantes JA, Nielsen EI, Korth-Bradley J, Harnisch L, Jonsson S. Elucidation of Factor VIII Activity Pharmacokinetics: A Pooled Population Analysis in Patients With Hemophilia A Treated With Moroctocog Alfa. Clinical pharmacology and therapeutics. 2017;102(6):977–988. [DOI] [PubMed] [Google Scholar]
  • 26.Aithal GP, Ramsay L, Daly AK, et al. Hepatic adducts, circulating antibodies, and cytokine polymorphisms in patients with diclofenac hepatotoxicity. Hepatology. 2004;39(5):1430–1440. [DOI] [PubMed] [Google Scholar]
  • 27.Samant TS, Mangal N, Lukacova V, Schmidt S. Quantitative clinical pharmacology for size and age scaling in pediatric drug development: A systematic review. J Clin Pharmacol. 2015;55(11):1207–1217. [DOI] [PubMed] [Google Scholar]
  • 28.Kleiber M Body size and metabolic rate. Physiological reviews. 1947;27(4):511–541. [DOI] [PubMed] [Google Scholar]
  • 29.Bonate PL, Steimer J-L. Pharmacokinetic-pharmacodynamic modeling and simulation. Springer; 2011. [Google Scholar]
  • 30.de Morais SM, Uetrecht JP, Wells PG. Decreased glucuronidation and increased bioactivation of acetaminophen in Gilbert’s syndrome. Gastroenterology. 1992;102(2):577–586. [DOI] [PubMed] [Google Scholar]
  • 31.Prescott LF, Speirs GC, Critchley JA, Temple RM, Winney RJ. Paracetamol disposition and metabolite kinetics in patients with chronic renal failure. European journal of clinical pharmacology. 1989;36(3):291–297. [DOI] [PubMed] [Google Scholar]
  • 32.Clements JA, Critchley JA, Prescott LF. The role of sulphate conjugation in the metabolism and disposition of oral and intravenous paracetamol in man. Br J Clin Pharmacol. 1984;18(4):481–485. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Prescott LF, Critchley JA, Balali-Mood M, Pentland B. Effects of microsomal enzyme induction on paracetamol metabolism in man. Br J Clin Pharmacol. 1981;12(2):149–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Lauterburg BH, Vaishnav Y, Stillwell WG, Mitchell JR. The effects of age and glutathione depletion on hepatic glutathione turnover in vivo determined by acetaminophen probe analysis. J Pharmacol Exp Ther. 1980;213(1):54–58. [PubMed] [Google Scholar]
  • 35.McGill MR, Jaeschke H. Metabolism and disposition of acetaminophen: recent advances in relation to hepatotoxicity and diagnosis. Pharm Res. 2013;30(9):2174–2187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Kozer E, Greenberg R, Zimmerman DR, Berkovitch M. Repeated supratherapeutic doses of paracetamol in children--a literature review and suggested clinical approach. Acta Paediatr. 2006;95(10):1165–1171. [DOI] [PubMed] [Google Scholar]
  • 37.Hedeland RL, Christensen VB, Jorgensen MH, Teilmann G, Iskandar A, Andersen J. Early Risk Factors of Moderate/Severe Hepatotoxicity After Suicide Attempts With Acetaminophen in 11- to 15-Year-Old Children. Glob Pediatr Health. 2014;1:2333794X14552897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Temple AR, Lynch JM, Vena J, Auiler JF, Gelotte CK. Aminotransferase activities in healthy subjects receiving three-day dosing of 4, 6, or 8 grams per day of acetaminophen. Clin Toxicol (Phila). 2007;45(1):36–44. [DOI] [PubMed] [Google Scholar]
  • 39.Geenen S, du Preez FB, Snoep JL, et al. Glutathione metabolism modeling: a mechanism for liver drug-robustness and a new biomarker strategy. Biochim Biophys Acta. 2013;1830(10):4943–4959. [DOI] [PubMed] [Google Scholar]
  • 40.Johnsrud EK, Koukouritaki SB, Divakaran K, Brunengraber LL, Hines RN, McCarver DG. Human hepatic CYP2E1 expression during development. J Pharmacol Exp Ther. 2003;307(1):402–407. [DOI] [PubMed] [Google Scholar]
  • 41.Vieira I, Sonnier M, Cresteil T. Developmental expression of CYP2E1 in the human liver. Hypermethylation control of gene expression during the neonatal period. Eur J Biochem. 1996;238(2):476–483. [DOI] [PubMed] [Google Scholar]
  • 42.Shinoda S, Aoyama T, Aoyama Y, Tomioka S, Matsumoto Y, Ohe Y. Pharmacokinetics/pharmacodynamics of acetaminophen analgesia in Japanese patients with chronic pain. Biol Pharm Bull. 2007;30(1):157–161. [DOI] [PubMed] [Google Scholar]
  • 43.Nahata MC, Powell DA, Durrell DE, Miller MA. Acetaminophen accumulation in pediatric patients after repeated therapeutic doses. European journal of clinical pharmacology. 1984;27(1):57–59. [PubMed] [Google Scholar]
  • 44.Bannwarth B, Pehourcq F, Lagrange F, et al. Single and multiple dose pharmacokinetics of acetaminophen (paracetamol) in polymedicated very old patients with rheumatic pain. J Rheumatol. 2001;28(1):182–184. [PubMed] [Google Scholar]
  • 45.Mian P, Valkenburg AJ, Allegaert K, et al. Population Pharmacokinetic Modeling of Acetaminophen and Metabolites in Children After Cardiac Surgery With Cardiopulmonary Bypass. J Clin Pharmacol. 2019. [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

sup 01

RESOURCES