Abstract
The hepatic extraction ratio (EH) is commonly considered an “inherent attribute” of drug. It determines the main physiological and biological elements of the system (patient attributes) that are most significant in interindividual variability of clearance. The EH consists of three age-dependent parameters: fraction of unbound drug in blood (fu.B), hepatic intrinsic clearance of unbound drug (CLu.int,H), and hepatic blood flow (QH). When the age-effects on these elements are not proportional, a given drug may shift from so-called high extraction status to low extraction. To demonstrate the impact of age-related changes on fu.B, CLu.int,H, and QH, the EH of midazolam and two hypothetical drugs with 10-fold higher and 10-fold lower CLu.int,H than midazolam were investigated in pediatrics based on known ontogeny functions. The EH was simulated using Simcyp software, version 14. This was then complemented by a comprehensive literature survey to identify the commonly applied covariates in pediatric population pharmacokinetic (PopPK) studies. Midazolam EH decreased from 0.6 in adults to 0.02 at birth, making its clearance much more susceptible to changes in CLu.int,H and fu.B than in adults and reducing the impact of QH on clearance. The drug with 10-fold higher CLu.int,H was categorized as high extraction from 4 days old onward whereas the drug with 10-fold lower CLu.int,H remained low extraction from birth to adulthood. Approximately 50% of collected PopPK studies (n = 120) did not consider interaction between age and other covariates. Interaction between covariates and age should be considered as part of studies involving younger pediatric patients. The EH cannot be considered an inherent drug property without considering the effect of age.
Introduction
Hepatic metabolic clearance (CLH,B) of intravenously administered drugs is determined by hepatic blood flow (QH) and their hepatic extraction ratio (EH), according to eq. 1:(1)The EH is calculated from the fraction of drug unbound in blood (fu.B), the hepatic intrinsic clearance of unbound drug (CLu.int,H), and QH, according to eq. 2:
(2)The extraction ratio of the drug is generally classified as high (>0.7), intermediate (0.3–0.7), or low (<0.3) according to the fraction of drug removed during one pass through the liver.
Commonly, the EH of a drug is considered an inherent attribute of the drug and is presented with a fixed value. However, this classification does not consider that the parameters in eq. 2 are age dependent or that changes in these parameters will affect the EH. For example, a rise in fu.B for low-EH drugs increases the hepatic metabolic clearance, whereas for high-extraction drugs this does not affect metabolic clearance. Unless the age-related physiologic changes in fu.B, CLu.int,H, and QH occur in parallel, it is expected that the EH of drugs varies with age. Therefore, a high-extraction drug in adults will not necessarily remain a high-extraction drug in neonates.
Age-varying EH can potentially be used as a covariate in clearance models when analyzing population pharmacokinetic (PopPK) studies. However, because applying the extraction ratio directly in the model might not be straightforward, this concept is considered in PopPK models through the interaction between covariate terms in the model. For example, age and body weight are commonly used as covariates in PopPK clearance models where body weight is also affected by age. The interaction between these two covariates should be considered in the model.
We investigated the relative differences in EH with age by using in vivo midazolam data and two hypothetical high- and low-extraction drugs through modeling and simulation techniques. We also use the concept of the age varying the EH to examine whether the interaction between covariate terms in modeling clearance has been considered in the PopPK studies.
Materials and Methods
Literature Data Collection.
Data on midazolam systemic clearance in pediatrics from birth to 17 years were collected from the literature. The literature search strategy and methodology for deconvolution of clearance to arrive at the CLH,B from midazolam systemic clearance (using the blood-to-plasma ratio) and QH based on cardiac output were explained previously (Salem et al., 2014).
Simulations.
A drug with 10-fold higher and 10-fold lower CLu.int,H than midazolam was designed by multiplying and dividing the deconvoluted midazolam CLu.int,H by 10 as proposed by Salem et al. (2014) to mimic a high- and low-extraction drug, respectively. Then, using the relevant CLu.int,H, QH, and fu.B in eq. 2, EH was calculated.
A number of simulations in Simcyp Pediatric version 14 (Certara, Princeton, NJ) were performed for midazolam, a drug with 10-fold higher and 10-fold lower CLu.int,H than midazolam, to show the age-related changes in the magnitude of EH. We simulated 100 subjects, consisting of an equal proportion of males and females and a combination of age bands (1 day, 1 month, 2 years, and 12 years as well as adult). The EH was calculated using eq. 3 from the output data. The mean values of EH at each age band were plotted against age for each of the simulated drugs.
Calculation of the Hepatic Extraction Ratio.
The hepatic extraction ratio was calculated from CLH and QH for midazolam and the other two hypothetical drugs assuming the well-stirred model, as seen in eq. 3:

Sensitivity Analysis.
Sensitivity analysis was performed with a view to identify which component of the extraction ratio (CLu.int,H, QH, or fu.B) plays the most dominant role in the variation of EH from adult values at any given age. The impact of age-dependent QH was evaluated by fixing fu.B and CLu.int,H (l/h/g of liver) to the adult values for all age ranges. This involved assumptions about the lack of any ontogeny for the abundance of the enzymes (pmol per mg of microsomal protein) and no age-related changes in the level of microsomal protein per gram of liver (MPPGL). The value of CLu.int,H (l/h/g of liver) was used to calculated the pediatric CLu.int,H values per whole liver by applying age-related liver weight. The EH was plotted against age, and the patterns were compared.
In another set, only fu.B values were fixed to adult values to demonstrate the sensitivity of the EH to age-related changes in QH and CLu.int,H (l/h) without the impact of age-related changes in binding. The EH was calculated and plotted against age and compared with the original set of results (where all age-related parameters had been considered).
To separate the size-related effects (i.e., liver mass and hepatic blood flow) from ontogeny-related factors on EH, a graphical representation was devised to demonstrate the pediatric values of enzyme abundance relative to adults at a given age (in this case, the cytochrome P450 enzyme 3A4 [CYP3A4]) alongside relative values for liver volume, hepatic blood flow, and MPPGL.
Population Pharmacokinetic Studies (PopPK).
A comprehensive literature survey using PubMed was performed to identify commonly used covariates in pediatric PopPK studies for drugs after intravenous administration. No year, journal, or language restriction applied to the search process. Collated publications were carefully checked for modeling covariates and the form of the covariates–clearance relationship in the reported model. We identified the studies that considered the interaction between covariates and clearance. Interaction between covariate terms was also considered if the presence of a covariate modified the impact of another covariate in a multiplicative or exponential way. Where there were different clearance models for different pediatric age ranges, the interaction with age was also considered. Interaction between covariate terms was not considered if only one covariate was considered in the final clearance model or if the covariates were in linear additive relationship to the clearance. When the modeling section was not clear, we contacted the corresponding authors.
Results
Midazolam Hepatic Extraction Ratio.
The hepatic extraction ratio of midazolam, after deconvolution of clinical systemic clearance, increased with age. Available data in Figure 1 illustrates that midazolam is a low-extraction drug until about the age of 10 months. However, in some individuals it remained low at the age of 9 years.
Hepatic extraction for intravenous midazolam calculated from reports of clinical studies in the literature using ontogeny functions in pediatric subjects and healthy adult volunteers (n = 523).
Figure 2 shows that the EH increases with age for midazolam and two other hypothetical compounds. The degree of change in the EH with age depends on magnitude of CLu.int,H against a given enzyme. As shown in the figure, a 10-fold reduction in CLu.int,H results in a drug with low hepatic extraction across the pediatric and adult age ranges whereas a 10-fold increase in CLu.int,H shifts the drug from so-called intermediate to high-extraction status.
Simulated hepatic extraction in Simcyp version 14 shows changes with age for midazolam, a drug with a 10-fold higher CLu.int,H, and a drug with 10-fold lower CLu.int,H. A high- or intermediate-extraction drug in adults is not necessarily a high- or intermediate-extraction drug in pediatric subjects. Dashed profiles are the same EH values with age when fu.B remains unchanged (fu.B = 0.05). Dotted horizontal lines show the limits for high (>0.7) and low (0.3>) extraction.
Sensitivity Analysis.
Figure 2 compares the EH when all age-related components (CLu.int,H, QH, and fu.B) are considered (solid lines) with a scenario involving no age-related changes in fu.B (dashed lines). As shown in Fig. 2, the EH is marginally lower in younger groups if age-related fu.B is not considered. However, this might be different for drugs with higher protein binding.
When the CLu.int,H (l/h/g of liver) and fu.B are fixed to adult values, the changes in the EH will be driven by age-related changes in QH and liver weight (Supplemental Fig. 1). In this scenario, there are no significant differences between the EH values across pediatric age groups for the three drugs because the low activity of CYP3A4 in younger age groups is not considered.
The rate of change with age for liver volume, hepatic blood flow, and MPPGL as a fraction of adult values are shown in Fig. 3. This figure shows the changes in the underlying parameters of the EH. The changes in blood flow and liver volume relative to adults occur almost in parallel to each other. Therefore, the discrepancy in QH and liver size alone cannot account for the observed differences in the EH; instead, changes in intrinsic activity to the level of enzyme abundance and to a lesser extent MPPGL are determinants of age-varying EH. Table 1 summarizes the contributing parameters to EH that are reported in Fig. 3. Needless to say, if the relative values to adult for all these elements had a similar rate of change with age, no age-related differences would have been anticipated in the EH.
Age-related variations in parameters defining the EH are shown as relative values to the corresponding adult level of each parameter. (A) Changes in liver size (Johnson et al., 2005), hepatic blood flow (Guyton, 1991), and MPPGL (Barter et al., 2008) that apply to all drugs. (B) Relative values of serum albumin (Johnson et al., 2003, 2006; Sethi et al., 2015), CYP3A4 abundance (of relevance to our study) (Salem et al., 2014) alongside age variation in serum alpha-acid glycoprotein (AAG) (Johnson et al., 2003, 2006), and abundance of CYP1A2 (Salem et al., 2014). The impact of the parameters shown in B will depend on the relative importance of the protein binding to each protein and the role of the specific enzyme to overall elimination.
Examples of age-related parameters defining the EH and prior knowledge on their age dependency
Analysis of Covariates in Population Pharmacokinetic Studies.
A total of 120 PopPK studies were retrieved in the pediatric age range (birth to 18 years) for intravenously administered drugs. The interaction between covariate terms was not considered in 50% of the studies (n = 60). Supplemental Table 1 summarizes the most commonly used covariates in the analyzed PopPK studies. Supplemental Table 2 shows all the analyzed PopPK studies with interaction between covariates.
Discussion and Conclusions
The EH of the drugs in this study increases with age due to the rapid physiologic changes in the parameters determining the EH after birth, including the ontogeny of enzyme abundance and to a lesser extent MPPGL. Although it is not relevant to the cases represented in this study, the ontogeny of plasma proteins can play a significant role in age-related changes of the EH for highly bound, low-extraction drugs. As shown previously by several investigators, the concentration of plasma proteins increases with age whereas the unbound fraction of drugs in plasma and thus in blood decreases with age (McNamara and Alcorn, 2002; Johnson et al., 2006; Sethi et al., 2015).
The ontogeny of plasma protein binding and enzyme abundance on any given compound depends on the extent of binding to a particular protein and the importance of that enzymatic pathway to the overall elimination of a drug. As a consequence, a drug that is coined a high, low, or intermediate hepatic extraction compound in adults is not necessarily going to carry the same extraction category in pediatrics. For the particular case studies in this report, where the binding was not a major factor and CYP3A4 was the main metabolizing enzyme, we demonstrated the switch from high extraction or intermediate extraction to low extraction in neonates and younger children. However, the results from this study should be generalized to other drugs metabolized by other pathways with caution. The CLu.int,H value is an interplay between the enzyme abundance and kinetic parameters (Vmax and Km).
The difference in enzyme abundance, depending on the pathway and age, can be masked or stressed by enzyme kinetic parameters, resulting in similar or different CLu.int,H and EH values from those we have shown in our study. Changes in EH is not confined to age-varying parameters. Induction or inhibition of drug-metabolizing enzymes for a flow-limited drug can also change the extraction ratio of drugs.
In addition, changes to QH and fu.B resulting from hemodynamic changes occurring in clinical conditions or during the progression of disease may affect the extraction ratio of drugs. This consideration can be more important in preterm neonates because of the prematurity of metabolic pathways, in special populations such as elderly, and in pregnancy, which can affect the free fraction of the drug (fu), enzyme activity, and/or QH, which ultimately can alter the EH.
Because the determinants of CL (covariates) change with age, it is incorrect to assume no interaction between age and covariates. In some of the PopPK studies we have collected, the interaction between covariate terms was not identified by the investigators. The reason for the lack of such interactions can be the wide age range in some of these studies with a limited number of subjects at lower end of spectrum. In addition, several investigations only examined older subsets of children, where the ontogeny of enzymes responsible for metabolism is likely to be fully mature. These investigations are not likely to find that the addition of age into their clearance models provides a better fit.
Another reason for the lack of interaction originates from unbalanced blood sampling in early life after birth compared with older children. Some pharmacokinetic studies retrospectively analyzed the available samples of drug concentration in blood or plasma where the relevant covariates were not always available. Also some relevant covariates such as fu may not have been measured in newborns. Only one PopPK study on morphine concluded that an independent clearance model is required for newborns (Knibbe et al., 2009).
The results in this study suggest that CLu.int,H is the most important parameter that affects the EH of drugs. Due to rapid physiologic changes after birth and especially in the neonatal period, the EH of drugs can be significantly affected by changes in the CLu.int,H.
Hepatic extraction also contributes to the determination of the oral bioavailability of drugs. Currently, there is contradicting evidence as to whether the bioavailability of drugs is different between pediatrics and adults (Harper et al., 1988; Stratchunsky et al., 1991; Pinkerton et al., 1993; Hassan et al., 1994; Fujiwara et al., 1996; Anderson et al., 2002; Crill et al., 2006; Zane and Thakker, 2014). In clinical practice the bioavailability of drugs is assumed to be similar between pediatrics and adults, but our study supports that bioavailability also can be an age-dependant parameter and can change with age because the EH changes with age. Assuming a higher hepatic extraction and thus a lower bioavailability in neonates, oral clearance can be overestimated in this population, and unnecessarily higher doses can be given to neonates. However, the clinical significance of these underestimations and overestimations is not clear and requires further investigation.
In conclusion, a high-extraction drug in adults is not necessarily a high-extraction drug in pediatrics. Unless the age-related changes in factors determining the EH occur at the same rate, the extraction ratio will be different between pediatrics and adults. More attention should be given to the interaction terms of covariates during analysis of such data as the impact of certain physiologic covariates might change with age. Further clarification of the underlying mechanisms for the metabolism (and bioavailability) of drugs should heavily rely on modeling and simulation techniques.
Acknowledgments
The authors thank the reviewers of this manuscript for their considerable input and constructive comments which significantly expanded the scope of this article, and Eleanor Savill for assistance with preparation of the manuscript.
Authorship Contributions
Participated in research design: Salem, Abduljalil, Kamiyama, Rostami-Hodjegan.
Performed data analysis: Salem, Abduljalil, Kamiyama.
Wrote or contributed to the writing of the manuscript: Salem, Abduljalil, Rostami-Hodjegan.
Footnotes
- Received October 1, 2015.
- Accepted February 9, 2016.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- CLu.int,H
- hepatic intrinsic clearance of unbound drug
- CLH,B
- hepatic metabolic clearance
- CYP3A4
- cytochrome P450 3A4 enzyme
- EH
- hepatic extraction ratio
- fu, fraction of drug in plasma unbound
- fu.B
- unbound drug in blood
- MPPGL
- microsomal protein per gram of liver
- PopPK
- population pharmacokinetics
- QH
- hepatic blood flow
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics