Abstract
Information on the developmental changes in biliary excretion (BE) of drugs is sparse. The aims of this study were to collate literature data on the pharmacokinetics of biliary excretion of drugs used in pediatrics and to apply a physiologically based pharmacokinetic (PBPK) model to predict their systemic clearance (CL) with a view to elucidating age-related changes in biliary excretion. Drug parameters for azithromycin, ceftriaxone, and digoxin administered intravenously and buprenorphine (intravenous and sublingual) were collated from the literature and used in the Simcyp Simulator to predict adult CL values, which were then validated against observed data. The change in CL with age was simulated in the pediatric model and compared with observed data; where necessary, the ontogeny function associated with BE was applied to recover the age-related CL. For azithromycin a fraction of adult BE activity of 15% was necessary to predict the CL in neonates (26 weeks gestational age) and 100% activity was apparent by 7 months. For ceftriaxone and digoxin full BE activity appeared to be present at term birth; for digoxin, an adult BE activity of 10% was needed to predict the CL in premature neonates (30 weeks gestational age). The CL of buprenorphine with age was described by the ontogeny of the major elimination pathways (CYP3A4 and UGT1A1) with no ontogeny assumed for the biliary component. Thus, the ontogeny of BE for all four drugs appears to be rapid and they attain adult levels at birth or within the first few months of postnatal age.
Introduction
Biliary excretion is a major elimination pathway for many drugs in humans either as the intact parent drug, e.g., pravastatin (Hatanaka, 2000) or via glucuronidated metabolites, e.g., mycophenolate mofetil, which may then undergo enterohepatic recirculation (Bullingham et al., 1998). Measuring biliary excretion of drugs in humans is difficult and can be done using a variety of techniques, including mass balance fecal recovery of radiolabeled drug and duodenal aspirates. Owing to the invasive nature of the experimental procedures, much of the data are limited to patients having a cholecystectomy, those with a temporary bile shunt (T-tube), or those with bile duct stenosis treated with nasobiliary drainage (Ghibellini et al., 2006). For the aforementioned reasons, it is improbable, using currently available techniques, that the age-related change in biliary elimination of drugs will be measured directly in children in the near future.
Consequently, there is very little known about the ontogeny of the biliary excretion of drugs in humans and much of the available information is inferred from animal data or indirect measures in human infants. Gut intraluminal bile acid concentrations are markedly reduced at birth especially in preterm compared with term neonates. However, the latter have around 60% of the adult value and 1-year-olds, approximately 80% (Balistreri, 1983). The total bile acid pool increases progressively from fetus to older infants and reaches adult levels of 3–4 g by 2 years of age, but when corrected for body surface area is similar for adults and infants 2 months of age (Heubi et al., 1982). Although biliary elimination appears to develop relatively early, more information on key systems parameters, including the ontogeny of transporters, are required to substantiate this. The key transporters involved in the active transport of drugs from hepatocytes into the bile canaliculus [P-glycoprotein (P-gp), multidrug resistance–associated protein 2 (MRP2), and breast cancer resistance protein (BCRP)] are shown in Fig. 1.
The main human hepatic uptake and bile efflux transporters and their typical substrates. BSEP (ABCB11): bile salt export pump; MATE (SLC47A): multidrug and toxin extrusion protein; MDR1, P-glycoprotein (P-gp; ABCB1): multi-drug resistance 1; MRP (ABCC): multidrug resistance–associated protein; NTCP (SLC10A1): Na+-taurocholate co-transporting polypeptide; OAT (SLC22A): organic anion transporter; OATP (SLCO): organic anion–transporting polypeptide; OCT (SLC22A): organic cation transporter; OSTα/β: organic solute transporter.
Information is relatively sparse on how specific hepatic uptake and canalicular transporters develop with age in pediatric subjects. However, there is active ongoing research in this area. The ontogeny of hepatic drug transporters and their relevance to drug use in pediatric subjects has been reviewed previously (Elmorsi et al., 2015). Much of the current data contains a high degree of uncertainty. Taking P-gp as an example, on the basis of recent data, P-gp mRNA expression is reported to be reduced in neonates, infants, and children compared with adults (Mooij et al., 2014), which is in agreement with previous studies (van Kalken et al., 1992; Miki et al., 2005; Fakhoury et al., 2009). However, the same group report no change in protein expression by age (Mooij et al., 2015), which is in agreement with a previous study that excluded neonates (Tang et al., 2007). Another group reports decreased protein expression in neonates, infants, and children (Prasad et al., 2015). On the basis of the same information sources, the results for MRP2 are even less certain.
Application of physiologically based pharmacokinetic (PBPK) modeling in support of drug development and regulatory review for both adult and pediatric medicines has expanded significantly in recent years (Leong et al., 2012; Zhao et al., 2011). By accounting for differences in absorption and transit rate, organ size, blood flow, tissue composition, and metabolic capacity, the effects of age and formulation on drug pharmacokinetics (PK) can be estimated and quantified by PBPK modeling (Johnson and Rostami-Hodjegan, 2011; Laer et al., 2009). Such models rely on robust systems data and given the uncertainty in the canalicular transporter ontogeny, this information could not at present be incorporated reliably into a pediatric-PBPK (p-PBPK) model to mechanistically predict the PK of biliary excretion of drugs in children. However, the flexible design of models that allow for different ontogeny profiles for the global biliary clearance element of pediatric drugs would allow them to be used as a research tool to investigate this issue. Accordingly, the aims of this study were to collate literature data on the pharmacokinetics of biliary excretion of drugs used in pediatrics and to apply a physiologically based pharmacokinetic model to predict their systemic clearance (CL) with a view to elucidating the age-related changes in biliary excretion.
Methods
Data Collection
Input drug parameters including physicochemical, protein-binding, and kinetic data for azithromycin and ceftriaxone were collated from the literature after Medline or PubMed searches. Other data were obtained from Drugbank and PubChem. The input parameters are summarized in Table 1. For buprenorphine, drug-related data cited previously were applied (Rowland Yeo et al., 2015), and the default parameters in the Simcyp Sim-vivo digoxin compound file were used for digoxin with modifications. Clinical studies on age-related changes in pharmacokinetics and related parameters were obtained from MEDLINE using the search terms “drug” plus “pharmacokinetics” or “clearance”. Searches were limited to “humans”, “premature neonates”, “neonates”, “child birth–18 years”. Article titles and abstracts were screened to restrict the focus of the search to relevant articles, and the reference lists of retrieved studies were also scanned to ensure completeness. A summary of the clinical studies used in the analysis is shown in Table 2. All clinical data were either taken directly from the research papers or were extracted from graphs using GetData Graph Digitizer 2.26.
Summary of the PBPK model drug input data for azithromycin, ceftriaxone, digoxin, and buprenorphine
Summary of clinical studies of drugs administered intravenously and compared to simulated data
PBPK Modeling of Biliary Elimination
Compound File Development.
A PBPK model for all compounds was compiled in Simcyp v14.1. The adult renal clearance values for all compounds were entered into the Simulator, and for intravenous azithromycin and ceftriaxone the remainder of the clearance was assigned to biliary elimination using the retrograde calculator to match the adult clinical data on intravenous clearance. For intravenous digoxin, the Simcyp Sim-vivo digoxin file within Simcyp was used with minor modification where the biliary elimination was optimized to 30% total elimination on the basis of literature reports (Caldwell and Cline, 1976; Hedman et al., 1990). The buprenorphine compound file contains metabolic information on elimination by CYP3A4, UGT1A1, as well as renal and biliary elimination. To mimic sublingual administration of buprenorphine, the inhaled route was used in the Simulator assuming 80% of the dose was swallowed. This allowed recovery of the observed bioavailability in adults, which is reported to be between 16 and 29% (Elkader and Sproule, 2005). The percentages of elimination assigned to each route for the different drugs are shown in Table 3.
Fraction of each compound assigned on the basis of literature evidence to be eliminated by direct biliary, renal, and hepatic metabolic elimination in adult healthy volunteers and performance of the model in these subjects
Simulations were run to ensure that the PBPK models developed for each compound were able to predict accurately the PK in adults before any pediatric simulations are undertaken.
Pediatric PBPK Modeling
The Simcyp Pediatric Simulator contains information on the ontogeny of renal function, which is scaled on the basis of predicted glomerular filtration rate (GFR) in pediatric subjects divided by typical GFR in adults (120 ml/min). This renal value is then used to scale the adult renal clearance value for each compound to that expected for a particular age (Johnson et al., 2006). In all cases no ontogeny was applied to the biliary clearance in the first instance. For buprenorphine an ontogeny is applied both to the CYP3A4 (Upreti and Wahlstrom, 2016) and the UGT1A1 (Abduljalil et al., 2014) components.
For initial intravenous simulations in azithromycin, ceftriaxone, digoxin, and buprenorphine, 1000 virtual subjects (10 trials of 100 subjects) were selected covering a period from full-term birth to the maximum age covered by the clinical studies (Table 2). The systemic clearance was plotted against age (y), which was converted to postmenstrual age (PMA) in the Microsoft Excel output. The clinical data were then overlaid to allow direct visual comparison of the results.
In the case of azithromycin, the possible ontogeny of biliary elimination in the premature neonates was further investigated using the Simcyp Pediatric Simulator to replicate the concentration-time data in this group (Hassan et al., 2011) by applying a user-defined ontogeny to the biliary elimination. For these simulations a full-term newborn population 0–0.0027y (1 day) was used with a correction applied to account for the reduced GFR in the 26-week–gestational age (GA) premature group. On the basis of the reference values of Vieux et al. (2010), a 50% reduction in azithromycin renal CL was input into the model to achieve this, all other physiologic parameters were left the same. Ten trials of 12 subjects, with a proportion of females of 0.5, were set up as the trial design and a number of “what-if” scenarios were run using the user-defined ontogeny corresponding to 10, 15, 20 and 50% of the adult biliary CL value. The simulated mean CL (l/kg per hour) data and 5th and 95th percentiles were compared with the clinical data.
To further investigate the ontogeny of biliary elimination of digoxin in premature neonates, a number of simulations were undertaken by applying the user-defined ontogeny to this pathway to recover the CL values reported clinically by Hastreiter et al. (1985). Because the GA was not reported and only the weight ranges (<1.5 kg and 1.5–2.5 kg), the first two groups were combined and a GA of 30 weeks assumed on the basis of the demographic data of Cole et al. (2014). Again a newborn population 0–0.0027y was used in the simulations with a reduction of 40% in the renal clearance of digoxin applied to correct for the lower GFR in this group on the basis of the data from Vieux et al. (2010). Ten trials of six subjects, proportion of females 0.5, were run and the mean observed and predicted CL (l/kg per hour) were compared assuming 5, 10, 20, and 50% of adult biliary elimination.
For the buprenorphine study by Kraft et al. (2008), 10 trials of 12 subjects 0–0.0027y, proportion of females 0.5, were run to replicate the clinical study, initially assuming average multiple sublingual doses of 0.0087 mg/kg, with the lower and upper reported doses being simulated subsequently. Because of the lack of bioavailability data in neonates no change was made to the adult assumption that 80% of the sublingual dose was swallowed.
Because the drug was administered over a prolonged time period of 47 days in newborn term neonates, the time-varying physiology model in the Simulator was used to account for changes in physiology occurring during the time course of the study (Abduljalil et al., 2014). For this buprenorphine study the concentration-time data were compared rather than clearance values.
Statistics
For meta-analysis of adult clinical studies the weighted mean and standard deviation from multiple reports was derived using the following equations:
Overall means (WX) were calculated using eq. 1:
(1)
where nj is the number of subjects in the jth study and xj is the mean value from that study. Overall S.D. was calculated using eqs. 2 and 3.
(2)
where sdj is the standard deviation from the jth study and N is the number of subjects in all studies.
(3)For concentration-time data the predicted 5th and 95th percentiles for the populations were included to ease visual comparison.
Results
The performance of each drug model in adults is shown in Table 3. For all drugs, the predicted clearances were consistent with those reported clinically.
The clearance predictions with PMA for all four drugs are shown in Fig. 2. In all of the simulations, no ontogeny was applied for the fraction eliminated directly by biliary excretion. For azithromycin (Fig. 2A) the ontogeny of biliary elimination appears to be complete before 8 months of age. However, there are no clinical data on the CL of intravenous azithromycin from term birth to this age. The CL of azithromycin in premature neonates around 26 weeks PMA is around 0.18 l/h compared with 0.98 l/h in infants aged 6 months. The possible ontogeny of azithromycin biliary elimination in the premature neonates was further investigated using various ontogeny functions within the Simcyp Pediatric Simulator to recover the concentration-time data from the study of Hassan et al., 2011. The results from these “what-if” simulations are shown in Fig. 3, in which application of an ontogeny to biliary CL corresponding to 15% of the adult value appears to give the best fit of simulated-compared-with-observed data in the 26-week-PMA neonates, assuming a 50% reduction in renal elimination compared with term.
Change in weight-normalized clearance with PMA for (A) azithromycin, (B) ceftriaxone, (C) digoxin, and (D) buprenorphine. The light gray open circles represent the simulated individual subjects and the open black symbols are clinical studies identifiable in the individual legends. The size of the symbols reflects the size of the clinical study. Error bars are ±S.D. (where reported). The dashed vertical black line represents full-term birth at 40 weeks PMA.
Predicted concentration-time profile for azithromycin in 26-week-old preterm neonates assuming (A) 50%, (B) 20%, (C) 15%, and (D) 10% of adult biliary elimination activity. The black line is the mean predicted profile and dashed gray lines the predicted 5th and 95th percentiles. Open black circles are the clinical data.
The results for ceftriaxone (Fig. 2B) show a lot of variability in the clinical data, with mean observed CL values between 0.017 and 0.032 l/kg per hour at 32 weeks GA and between 0.015 and 0.051 l/kg per hour at 40 weeks GA. All of these values fall within the range of values predicted, assuming no biliary ontogeny, and suggest that for this drug the biliary CL is at or near adult values by the time of term birth.
The results for digoxin are shown in Fig. 2C. Some of the CL values in the preterm infant populations show values below those predicted, assuming no ontogeny for biliary elimination of this drug. However, by the time of term birth most of the CL values fall within the range of predicted values. Again there is a lot of variability in the clinical data, with values ranging from 0.12 to 0.45 l/kg per hour in the term neonatal population. To replicate the results in the preterm neonates, an ontogeny of biliary elimination corresponding to 10% of the adult level had to be applied in the 30-week GA group to recover the CL value of 0.06 l/h per kilogram seen clinically. This was after applying a 40% reduction to the renal elimination to allow for prematurity; full results are shown in Table 4.
“What if” scenarios for the ontogeny of biliary elimination of digoxin in premature neonates 30 weeks GA
The results for buprenorphine are shown in Fig. 2D. The results in children and adults are captured reasonably well. The observed results in the premature neonatal population fall at the bottom end of the predicted results in term newborns with no biliary ontogeny applied, again suggesting that biliary elimination is reasonably close to adult values by this stage. Further evaluation of the buprenorphine drug model was undertaken to predict the concentration-time data from the study by Kraft et al. (2008) with no biliary ontogeny applied (Fig. 4, A and B).
Predicted concentration-time profile of multiple sublingual doses of buprenorphine in term neonates (A) assuming median dose and (B) assuming upper dose. The black line is the mean predicted profile and dashed gray lines the predicted 5th and 95th percentiles. The filled gray circles are the clinical data.
As there was some uncertainty regarding the final doses that individual subjects were titrated to in this study, three scenarios were run: median dose, lower dose, and upper dose. Only results for median and upper dose are shown. The different cases show that overall the clinical concentration-time data were captured in the simulations assuming no ontogeny for the biliary elimination.
Discussion
In this study we have used a combined “bottom up” (PBPK) and “top down” (PK) approach to assess the development of biliary elimination of specific drugs in humans. The biliary ontogeny for each drug is dependent on the transporters involved in their canalicular efflux. Key transporters contributing to the disposition of each drug are shown in Table 5. In general, biliary excretion appears to develop rapidly and be at adult equivalent capacity at or soon after birth, which is in line with some of the emerging data on canalicular transporter ontogeny in humans. As mentioned previously, the findings of one recent study (Mooij et al., 2015) indicated that there was no ontogeny for P-gp protein expression in the liver. However, it should be noted that there are conflicting data emerging that show reduced expression in neonates and infants and compared with adults (Prasad et al., 2015).
Summary of the transporters involved in the drug canalicular efflux
In the current study, the clinical data and PBPK modeling for azithromycin and digoxin suggest that P-gp as a canalicular transporter is at or near adult activity at or just after birth. The same conclusion can be made for MRP2 on the basis of the clinical data from ceftriaxone and azithromycin. This is supported by the protein expression data of Mooij et al. (2015), whereas that of Prasad et al. (2015) showed significantly (P < 0.05) reduced MRP2 expression in infants compared with adults. The data on BCRP ontogeny also suggests no correlation of protein expression between neonates and adults (Yanni et al., 2011; Mooij et al., 2015; Prasad et al., 2015) and this is reflected in the clinical data for ceftriaxone. The specific transporters involved in buprenorphine biliary excretion have not yet been identified.
To further explore the ontogeny of canalicular transporters in premature neonates a number of “what if” scenarios were run for azithromycin and digoxin to capture either the concentration-time or CL data. The ontogeny tool within the Pediatric Simulator allows for a user-defined fractional ontogeny for biliary excretion relative to adult. Because the p-PBPK model was built on the basis of data from term neonates onwards, it was necessary to include a correction for renal function and thus renal clearance before simulating in premature neonates. All other factors, including body size metrics, liver blood flows, and protein expression, were left the same in the model, so although it cannot be regarded as a true physiologic representation of a premature neonate it is close enough for the purpose of these simulations, especially because the clearance results are normalized to body weight. For digoxin, a P-gp substrate, an ontogeny representing around 10% of adult activity had to be applied to capture the clinical data, for azithromycin a 15% adult activity was applied, which may reflect the additional MRP2 transport of this drug.
Although it is an interesting example because of the sublingual as well as intravenous administration in the examples shown, buprenorphine is perhaps the least useful of the compounds in terms of biliary ontogeny because its fraction eliminated by this route is relatively low. The ontogeny of the other main pathways (CYP3A4 and UGT1A1) is relatively well defined in p-PBPK models (Abduljalil et al., 2014; Salem et al., 2014; Upreti and Wahlstrom, 2015), and assuming no ontogeny for the biliary component, the current drug model was not able to capture the clinical data in the 27- to 32-week PMA group. The ontogeny of both of these enzymes prior to birth is relatively unknown, but assuming they are expressed at a lower level than at birth, then this would be enough to explain the clinical data without applying additional ontogeny to biliary elimination. To simulate the data of Kraft et al. (2008) for treatment of neonatal abstinence syndrome, assumptions had to be made regarding the dose given, as only a range was reported in the clinical study. To cover all eventualities three dose levels were simulated, starting with the median dose of 8.7 μg/kg three times a day (tid) but also 4.4 and 13 μg/kg tid representing lower and upper doses. The clinical data were captured by the median and upper dose simulations assuming no ontogeny for the biliary elimination component. Because of the long term nature of the clinical study (up to 46 days) relative to the age of the neonates, it was necessary to use the time-based changing physiology option in the Simulator, which allowed for the “growth” of the system parameters with time, as described in more detail by Abduljalil et al. (2014). The assumption that 80% of the sublingual dose of buprenorphine was swallowed in an adult may not apply to neonates, whose bioavailability is unknown but may be lower owing to drooling. This could also explain some of the overprediction of the simulated concentration-time profile following the median dose, but more probably this is a result of the misspecification of dose.
There are a number of limitations of the current study not least of which is the limited number of drugs used in the pediatric age range at which the drug is excreted to a significant degree by biliary elimination. However, the current study provides some evidence that biliary elimination appears to be reasonably well developed by the time of birth with a possible ranking of evidence of azithromycin > ceftriaxone > digoxin > buprenorphine. Even in the case of azithromycin the lack of available clinical data in the birth-to-6-months age group makes it difficult to conclude exactly when the biliary ontogeny reaches adult levels. The current study does illustrate the potential usefulness of the p-PBPK modeling approach as a research tool in pediatric subjects where performing direct clinical studies to assess the ontogeny of biliary excretion would not be feasible because of logistical and ethical reasons. Previously we have shown the utility of this methodology for other scenarios, where it was challenging to perform studies that assessed metabolic drug-drug interactions in children of different ages (Salem et al., 2013). The latter approach has been taken up in regulatory guidance on drug-drug interaction assessment in special populations, including pediatric subjects, by both the Food and Drug Administration and European Medicines Agency. Such use of PBPK models to extrapolate outside the study populations and experimental conditions required careful attention to a number of issues and confidence that key systems parameters within the PBPK model were correct (Tsamandouras et al., 2015).
Combining the “bottom up” and “top down” approaches with fitting of PBPK models to the clinical data is a useful approach to quantifying some “unknown” systems parameters, such as canalicular transporter ontogeny in this study, or to better quantify others such as CYP3A4 ontogeny (Salem et al., 2014). Because some of the drugs are substrates for more than one canalicular transporter, the information from them can only be interpreted in general terms in relation to their probable ontogeny. Although some information on the ontogeny of specific transporters in relation to protein expression is now emerging, some is contradictory, and there is a clear need for more information in this area using validated methodology. Clinical samples should be well characterized in terms of handling and general patient characteristics, including relevant genotyping.
Conclusion
On the basis of limited clinical data, the ontogeny of biliary elimination for all four drugs appears to be rapid and reaches adult levels at birth or in the first few months of postnatal age. More research is required in this area, particularly on the detailed ontogeny of specific canalicular transporters in humans.
Acknowledgments
The help of Eleanor Savill in the preparation and submission of this article and also that of Dr. Sibylle Neuhoff as the originator of Figure 1 is gratefully acknowledged.
Authorship Contributions
Participated in research design: Johnson, Jamei, Rowland-Yeo.
Performed data collection and analysis: Johnson, Rowland-Yeo.
Wrote or contributed to the writing of the manuscript: Johnson, Jamei, Rowland-Yeo.
Footnotes
- Received November 30, 2015.
- Accepted February 8, 2016.
This study was funded by the Treating Infections in Neonates 2 (TINN2) programme (Collaborative Project) supported by the European Commission under the Health Cooperation Work Programme of the 7th Framework Programme [Grant agreement no. 260908].
Abbreviations
- BCRP (ABCG2)
- breast cancer resistance protein
- CL
- systemic clearance
- GA
- gestational age
- GFR
- glomerular filtration rate
- MRP2 (ABCC)
- multidrug resistance–associated protein 2
- P-gp
- P-glycoprotein
- PBPK
- physiologically based pharmacokinetic
- PK
- pharmacokinetic
- PMA
- postmenstrual age
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics
References
In this issue
In Vivo Ontogeny of Biliary Elimination
