Incorporating Ontogeny in Physiologically Based Pharmacokinetic Modeling to Improve Pediatric Drug Development: What We Know About Developmental Changes in Membrane Transporters

Abstract

Developmental changes in the biological processes involved in the disposition of drugs, such as membrane transporter expression and activity, may alter the drug exposure and clearance in pediatric patients. Physiologically based pharmacokinetic (PBPK) models take these age-dependent changes into account and may be used to predict drug exposure in children. As a result, this mechanistic-based tool has increasingly been applied to improve pediatric drug development. Under the Prescription Drug User Fee Act VI, the US Food and Drug Administration has committed to facilitate the advancement of PBPK modeling in the drug application review process. Yet, significant knowledge gaps on developmental biology still exist, which must be addressed to increase the confidence of prediction. Recently, more data on ontogeny of transporters have emerged and supplied a missing piece of the puzzle. This article highlights the recent findings on the ontogeny of transporters specifically in the intestine, liver, and kidney. It also provides a case study that illustrates the utility of incorporating this information in predicting drug exposure in children using a PBPK approach. Collaborative work has greatly improved the understanding of the interplay between developmental physiology and drug disposition. Such efforts will continue to be needed to address the remaining knowledge gaps to enhance the application of PBPK modeling in drug development for children.

The off-label use of drugs in doses that are insufficiently studied is extensive in pediatric medicine.¹ This is mainly because drug development for treatment in pediatric patients is challenged by ethical concerns and logistical issues.² Because children widely differ from adults due to developmental changes in the biological processes involved in the disposition of drugs, this leaves them at risk for subtherapeutic or toxic exposures.³ The establishment of the Best Pharmaceuticals for Children Act in 2002 and the Pediatric Research Equity Act in 2003, which were made permanent under the Food and Drug Administration Safety and Innovation Act in 2012, and the European Paediatric Regulation (regulation no. 1901–2/2006) in 2006 have highlighted the commitment of the US Food and Drug Administration (FDA) and the European parliament and council to conduct studies in pediatric patients and thereby fill the pediatric gaps in drug development to increase the safety and efficacy of pediatric drug therapy.^4–6

With the advancement of in silico technologies, novel methodologies such as model-informed drug development (MIDD) can leverage our existing understanding of pediatric physiology, disease states, and pharmacology. This provides quantitative information to streamline decision making in drug development, such as clinical trial design and dose optimization, which can increase the success of pediatric clinical trials.⁷ To support this, the FDA has committed to advance MIDD under the Prescription Drug User Fee Act VI, with approaches that include convening a series of workshops to identify best practices for MIDD, conducting a pilot meeting program for MIDD approaches, publishing or revising an existing draft guidance on MIDD, and engaging in regulatory science research to develop expertise and capacity in MIDD approaches.^7,8

Physiologically based pharmacokinetic (PBPK) modeling is 1 of the mechanistic-based MIDD tools that has been increasingly incorporated into drug development programs to support submissions to the FDA and the European Medicines Agency.^9,10 Of all the PBPK analyses that were included in the New Drug Application submissions to the FDA between 2008 and 2017, 60% were utilized to assess enzyme-mediated drug-drug interactions (DDIs). This was followed by 15% of the submissions that supported the evaluation of pediatric-related issues such as initial dose recommendation for clinical trials and 7% that analyzed transporter-mediated DDIs.⁹ During the FDA Advisory Committee for Pharmaceutical Science and Clinical Pharmacology Meeting in March 2012, some experts expressed concerns regarding the routine use of PBPK modeling in pediatric drug development because pediatric PBPK models still had significant knowledge gaps in areas such as the ontogeny of membrane transporters and thereby may not predict drug exposure well.¹¹

New data on the ontogeny of membrane transporters have emerged since 2012, and the objective of this article is to review findings from recent studies that have evaluated pediatric developmental changes in the membrane transporters.

Ontogeny of Membrane Transporters

Membrane transporters facilitate the active movement of drug molecules and endogenous compounds into and out of cells of various organs, affecting drug absorption, distribution, and excretion.¹² Hence, they have a critical role in impacting pharmacokinetics (PK) and pharmacodynamics of drugs and should be considered and assessed carefully during drug development. In the FDA 2017 draft on in vitro DDI guidance, the FDA recommended the evaluation of DDI potential by studying whether a new drug is a potential substrate or inhibitor of the following 9 transporters (see Table 1 for full, protein, and gene names): P-glycoprotein (P-gp), breast cancer resistance protein (BCRP), multidrug and toxin extrusion 1 (MATE1), MATE2-K, organic-anion-transporting polypeptide (OATP)1B1, OATP1B3, organic anion transporter (OAT)1, OAT3, and organic cation transporter (OCT)2.¹³

Table 1.

Full Names, Protein Names, and Gene Names of Membrane Transporters Discussed in This Article

Full Name	Protein Name	Gene Name
P-glycoprotein	P-gP	ABCB1
Breast cancer resistance protein	BCRP	A8CG2
Multidrug and toxin extrusion 1	MATE 1	SLC47A1
Multidrug and toxin extrusion 2-K	MATE2-K	SLC47A2
Organic anion transporting polypeptide 1A2	OATP1A2	SLCO1A2
Organic anion transporting polypeptide 1B1	OATP1B1	SLCO1B1
Organic anion transporting polypeptide 1B3	OATP1B3	SLCO1B3
Organic anion transporting polypeptide 2B1	OATP2B1	SLCO2B1
Organic anion transporter 1	OAT1	SLC22A6
Organic anion transporter 3	OAT3	SLC22A8
Organic cation transporter 1	OCT1	SLC22A1
Organic cation transporter 2	OCT2	SLC22A2
Multidrug resistance-associated protein 1	MPR1	ABCC1
Multidrug resistance-associated protein 2	MRP2	ABCC2
Multidrug resistance-associated protein 3	MRP3	ABCC3
Multidrug resistance-associated protein 4	MRP4	ABCC4
Multidrug resistance-associated protein 6	MRP6	ABCC6
Peptide transporter 1	PEPT1	SLCI5A1
Sodium/taurocholate cotransporting polypeptide	NTCP	SLCIOA1
Bile salt export pump	BSEP	ABCB11
Glucose transporter 1	GLUTI	SLC2A1
Glucose transporter 2	GLUT2	SLC2A2
Monocarboxylate transporter 1	MCT1	SLCI6A1
Uric acid transporter 1	URAT1	SLC22A12

There is a wealth of information on how alterations in the transporter activity, mainly due to genetic polymorphisms and DDIs, can lead to variability in drug safety and efficacy in adults. However, less is known about age-related changes in transporter expression levels and activities and how that relates to the safety and efficacy of pediatric drug use. In 2015 the Pediatric Transporter Working Group performed a comprehensive review on the data available for the ontogeny of clinically relevant membrane transporters.¹⁴ Further, the working group also provided recommendations to address and overcome some of the challenges in filling the pediatric knowledge.¹⁴ These include building multidisciplinary and international collaborative networks to facilitate data sharing, increasing awareness of clinicians about the importance of transporters in pediatric drug disposition, and identifying biomarkers for transporter activity in children. In the discussion that follows and in Table 2, human data presented in that review are highlighted, and updated information from recent literature is provided. Figure 1 also depicts the human membrane transporters in the intestine, liver, and kidneys that are mentioned in this article.

Table 2.

Human Ontogeny Data of Membrane Transporters in Intestine, Liver, and Kidney Highlighted in This Article GA indicates gestational age; TM₅₀, age at which half of the adult value is reached.

Membrane Transporter, Protein Name (Gene Name)	Types of Ontogeny Data Available	Reported Ontogeny Pattern	Reference
Intestinal transportersa
P-gp (ABCB1)	Gene expression	Overall, the mRNA levels in neonates and infants were comparable to those in adults	Mizuno et al,20 Fakhouryetal,21,22 Miki et al,24 van Kalken et al,25 Mooij et al26
BCRP (ABCG2)	Immunohistochemistry	BCRP distribution was similar in fetal (5.5–28 wk of gestation) and adult samples	Konieczna et al27
MRPI (ABCC1)	Immunohistochemistry	MRPI distribution was similar in adults and fetal samples (9–28 wk of gestation)	Konieczna et al27
MRP2 (ABCC2)	Gene expression	mRNA level was stable from neonates to adults	Mooij et al26
OATP2B1 (SLCO2B1)	Gene expression	mRNA level was higher in neonates than in adults	Mooij et al26
PEPT1 (SLCI5A1)	Gene expression	mRNA was slightly lower in neonates than in older counterparts	Mooij et al28
Liver transporters	Immunohistochemistry	Tissue distribution was relatively stable from preterm neonates to adolescents	Mooij et al28
OCT1 (SLC22A1)	Gene expression	Transcript levels in pediatric livers were comparable with those in adults	Hayashi et al36
	Western blot	Age-dependent increase in OCT1 protein expression from birth up to 8–12 y	Hahn et al37
	Quantitative proteomics	Age-dependent increase in protein expression level; TM50 was approximately 6 mo	van Groen et al,33 Prasad et al34
OATP1B1 (SLCO1B1)	Gene expression	mRNA expression of OATP1B1 in fetal liver was 20-fold lower than that in adults. Neonates and infants have even lower levels than fetus (500-fold and 90-fold lower than adults, respectively)	Mooij et al26
	Quantitative proteomics	van Groen et al reported higher protein expression in fetal livers compared with that in term neonates. The protein expressions in infants to adults were similar. Genetic polymorphism was not associated with expression levels in this study. Prasad et al reported that when all samples were considered, no age-dependent changes in the protein expression was found. Protein levels were higher in *IAI*IA > 1-y-old cohort than the 0- to 12-mo group	van Groen et al,33 Prasad et al34
OATP1B3 (SLCO1B3)	Gene expression	mRNA levels in fetus, neonates and infants were lower than that in adults	Mooij et al26
	Quantitative proteomics	No age-dependent changes were found in van Groen et al; age-dependent increase reported in Prasad et al with TM50 approximately 6 mo	van Groen et al,33 Prasad et al34
OATP2B1 (SLCO2B1)	Gene expression	mRNA level was significantly higher in adult liver compared with that in fetus (GA 18–23 weeks)	Sharma et al39
	Quantitative proteomics	Comparable protein expression levels in livers from fetus to adults	van Groen et al,33 Prasad et al3440
NTCP (SLCIOAI)	Gene expression	mRNA level was low in fetal liver compared with adults	Sharma et al,39 Chen et al41
	Western blot	Relative expression was stable in livers samples from neonates and adults	Yanni et al42
	Quantitative proteomics	Prasad et al reported stable protein expression from neonates to adults, van Groen et al showed that protein expression was significantly lower in fetuses than in term neonates, infants, children, and adults	van Groen et al,33 Prasad et al34
P-gp (ABCB1)	Gene expression	Detected in fetal liver; mRNA level increased rapidly during first 12 mo of life in infants	Mooij et al26
	Western blot	No significant differences in the relative protein expression from 0.3 to 12 y old	Tang et al43
	Quantitative proteomics	Protein level increase from fetus to adults with TM50 approximately 2.9 y old	van Groen et al,33 Prasad et al34
MRP2 (ARCC2)	Gene expression	mRNA level increased; levels in fetal, neonatal, and infant livers were substantially lower than that in older children up to 12 y old	Mooij et al,26 Klaassen and Aleksunes44
	Quantitative proteomics	van Groen et al reported that MRP2 level was much lower in fetal and term newborn livers than that in adults. Prasad et al found no age-dependent changes	van Groen et al,33 Prasad et al34
MRP3 (ABCC3)	Gene expression	mRNA level was lower in fetal livers than that in adults	Sharma et al39
	Quantitative proteomics	van Groen et al reported that protein abundance was lower in fetus and term neonates than in adults. In Prasad et al, lower protein abundance was found in infants and adolescents than adults.	van Groen et al,33 Prasad et al34
MRP1 (ABCC1)	Quantitative proteomics	Protein levels were lower in fetus and term neonates than adults	van Groen et al33
MRP4 {ABCC4)	Gene expression	No age-dependent changes in mRNA level	Sharma et al39
MRP6 (ABCC6)	Gene expression	mRNA level increase from neonates to older children and adults	Klaassen and Aleksunes44
BCRP (ABCG2)	Immunohistochemistry Gene expression	Detected in fetus as young as GA 5.5 weeks mRNA level was lower in fetal livers than in adults	Konieczna et al27 Sharma et al,39 Klaassen and Aleksunes44
	Quantitative proteomics	Stable across age groups from fetus to adults but age-dependent decrease observed in fetal and newborn cohorts	van Groen et al,33 Prasad et al34
BSEP (ABCB11)	Quantitative proteomics	Significantly lower in fetal livers compared with that in adults; no age-dependent changes after birth	van Groen et al,33 Prasad et al34
MATE1 (SLC47A1)	Gene expression	mRNA showed age-dependent increase	Klaassen and Aleksunes44
	Quantitative proteomics	No age-dependent changes in protein abundance	Prasad et al34
GLUTI (SLC2A1)	Quantitative proteomics	Protein abundance was high in fetus and lower in other age groups	van Groen et al33
MCTI (SLC16A1)	Quantitative proteomics	No age-dependent changes in protein abundance	van Groen et al33
Kidney transporters
BCRP (ABCG2)	Gene expression	mRNA level was higher in term neonates than older counterparts	Cheung et al46
	Quantitative proteomics	No age-dependent changes in protein abundance	Cheung et al46
MATE1 (SLC47A1)	Gene expression	No age-dependent changes in mRNA level	Cheung et al46
	Quantitative proteomics	No age-dependent changes in protein abundance	Cheung et al46
MATE2-K (SLC47A2)	Gene expression	mRNA level was lower in term newborns than in adults	Cheung et al46
	Quantitative proteomics	No age-dependent changes in protein abundance	Cheung et al46
MRP2 (ABCC2)	Gene expression	No age-dependent changes in mRNA level	Cheung et al46
MRP4 (ABCC4)	Gene expression	No age-dependent changes in mRNA level	Cheung et al46
	Immunohistochemistry	Proper localization observed in renal cortical fetal sample as early as GA 27 wk	Cheung et al46
URAT1 (SLC22A12)	Gene expression	mRNA levels increased with age from term newborn to adults	Cheung et al46
	Quantitative proteomics	Protein levels increased with age from term newborn to adults	Cheung et al46
P-gp (ABCB1)	Immunohistochemistry	Localization detected as early as end of first trimester of fetal life	van Kalken et al25
	Gene expression	mRNA levels were lower in preterm newborns, term newborns, and infants compared with older counterparts	Cheung et al46
	Quantitative proteomics	Protein levels increased with age with TM50 approximately 1 mo	Cheung et al46
GLUT2 (SLC2A2)	Gene expression	No age-dependent changes in mRNA level	Cheung et al46
	Quantitative proteomics	No age-dependent changes in protein abundance	Cheung et al46
OATI (SLC22A6)	Gene expression	mRNA level increased with age	Cheung et al46
	Quantitative proteomics	Protein abundance increased with age. TM50 was approximately 5 mo	Cheung et al46
OAT3 (SLC22A8)	Gene expression	mRNA level increased with age	Cheung et al46
	Quantitative proteomics	Protein abundance increased with age. TM50 was approximately 8 mo	Cheung et al46
OCT2 (SLC22A2)	Gene expression Quantitative proteomics	mRNA level increased with age Protein expression increased with age. TM50 was approximately 1 mo	Cheung et al46 Cheung et al46

^aSee Brouwer et al for a more detailed review.¹⁴

Figure 1.

Summary of the human membrane transporters in the intestine, liver, and kidneys that are mentioned in this review. Transporters with only mRNA or limited data are depicted in brown circles, whereas those that have both gene expression and protein abundance data are depicted in green circles. (Adapted/modified from Brouwer et al¹⁴ and Chu et al.⁶⁷) Noteworthy, the localization of OATP2B1 remains questionable. Future investigation is needed to ascertain if its localization is subject to developmental changes. BCRP, breast cancer resistance protein; BSEP, bile salt export pump; GLUT, glucose transporter; MATE, multidrug and toxin extrusion; MCT, monocarboxylate transporter; MRP, multidrug resistance-associated protein; NTCP, sodium/taurocholate cotransporting polypeptide; OAT, organic anion transporter; OATP, organic anion transporting polypeptide; OCT, organic cation transporter; PEPT1, peptide transporter 1; P-gp, P-glycoprotein; URAT1, uric acid transporter 1.

Ontogeny of Intestinal Transporters

Most drugs prescribed to children are administered orally.¹⁵ The intestine is a major absorption site of drugs that are administered via the oral route. Transporters that are present in the enterocytes on the gut wall mucosa govern the initial access into the systemic circulation of molecules such as sugars, amino acids, vitamins, but also of drug substrates.^16,17 P-gp, multidrug resistance-associated protein 2(MRP2), and BCRP, for instance, are major efflux transporters that are responsible for limiting drug absorption. OATP1A2 and OATP2B1 have been suggested to participate in the intestinal absorption of drugs in humans.¹⁸ Further, peptide transporter (PEPT)1 is a major uptake transporter that facilitates absorption of peptide-like drugs in the systemic circulation such as β-lactam antibiotics.^16,19 Therefore, drug absorption in children will be highly dependent on the expression and activity of these intestinal transporters.

P-gp, BCRP, MRPs, OATP2B1, and PEPT1.

In their review, Brouwer et al noted that ontogeny of intestinal transporters was mainly revealed by mRNA expression and localization data using immunohistochemistry.¹⁴ P-gp and MRP2 mRNA expression levels in neonates and infants appeared to be comparable to those in adults.^20–27 Localization data suggested that BCRP and MRP1 distributions were similar in adult and fetal samples (5.5–28 weeks and 9–28 weeks of gestation, respectively).²⁷ In contrast to the other intestinal transporters, OATP2B1 gene expression levels were much higher in neonates than in adults.²⁶ Interestingly, the localization of OATP2B1 remains questionable; 2 studies observed localization of the transporter to the apical membrane of human enterocytes, but another research group, which studied mainly pediatric intestinal tissue samples, detected OATP2B1 in the basolateral membrane.^28–31 This basolateral localization was also reported by another independent group using 6 healthy human adult jejunal tissue samples.³² Future studies are warranted to elucidate the localization of OATP2B1 and to ascertain if it is subject to developmental changes. Using a total of 26 intestinal tissues samples, which included 19 preterm and term neonates, 1 infant 13.9 weeks old, 2 children, and 4 adolescents, Mooij and colleagues studied the developmental changes in PEPT1 mRNA expression and localization.²⁸ Although PEPT1 expression appeared to be slightly lower in neonates than in their older counterparts, the tissue distribution was relatively stable among all the samples studied.

Although changes in the gene expression and localization of these transporters during development were stressed in various published studies, data on their protein expression levels are still missing. In addition, the ontogeny patterns of other human intestinal transporters, such as OATP1A2 and monocarboxylate transporter (MCT)1, remain uncertain. Because many drugs are administered orally, it is crucial to fill this knowledge gap in intestinal transporter ontogeny.

Ontogeny of Liver Transporters

In comparison to intestinal transporters, data on developmental changes in hepatic transporters have grown quite rapidly recently. Classic analytical approaches include quantitative real-time polymerase chain reaction, which measures gene expression levels, immunohistochemistry, which visualizes localization, and Western blot, which measures the relative protein expression. In addition, quantitative proteomics via liquid chromatography/tandem mass spectrometry has been increasingly utilized to measure the absolute protein abundance of these transporters, allowing the quantification of many transporters in only a small amount of tissue. Proteomics data generated from 2 independent laboratories complemented each other in terms of age range of the samples and provided a more complete picture of the developmental patterns of hepatic transporters with higher confidence than what was known previously.^33–35 In 1 study the protein abundance of 11 hepatic transporters was measured in approximately 69 postmortem tissue samples that covered the whole pediatric age range (4 neonates, 19 infants, 32 children, and 14 adolescents) and in 41 adult samples (>16 years old).³⁴ In another study the absolute protein expression of 13 liver transporters was quantified in a pediatric cohort with a focus on the fetus and newborn up to postnatal 18 weeks of age that consisted of 62 pediatric tissue samples (36 fetuses, 12 premature newborns, 10 term newborns, 4 pediatric patients, and 8 tissue samples from adults).³³ The findings in these 2 studies and other previous studies are discussed below.

• OCT1: As previously reported, OCT1 mRNA levels in pediatric livers appeared to be comparable to those in adult livers.^14,36 Nonetheless, OCT1 protein levels have been shown to undergo an age-dependent increase.^33,34,37 This was supported by a recently published clinical study in neonates who were admitted to the neonatal intensive care unit where postmenstrual age as well as OCT1 genotype impacted the PK of the OCT1 substrate morphine.³⁸ Further, the age at which half of adult level is reached (TM₅₀) was also estimated using a sigmoidal maximum effect (E_max) model and was reported to be about 6 months.

• OATP1B1: mRNA expression of OATP1B1 in fetal liver was 20-fold lower than that in adults, and that in neonates and infants was even lower (500-fold and 90-fold, respectively).^14,26 Recent quantification of protein expression, nonetheless, revealed different findings. In their sample set van Groen et al found that the OATP1B1 expression was significantly higher in the fetal livers compared with that in term neonatal livers. The protein expressions in infants, children, and adults were similar.³³ OATP1B1 is highly polymorphic. The impact of genetic variants on developmental changes in OATP1B1 expression was investigated in this cohort, but no association was identified for the studied genotypes. When all tissue samples were considered, Prasad et al reported that OAPT1B1 did not show age-dependent changes in the protein expression.³⁴ Yet, when the analysis was performed on samples from donors with the OATP1B1 reference allele, *1A/*1A, samples from >1 year old were found to have higher protein expression than the 0-to 12-months group. Notably, in the >1-year-old cohort, OATP1B1 expression was about 2.5-fold higher in samples from donors with *14/*1A than from those with *15/*1A.

• OATP1B3: Similar to OATP1B1, mRNA expression of OATP1B3 was reported to be much lower in fetuses, neonates, and infants compared with adults.^14,26 Although proteomics data in 1 study showed that OATP1B3 expression was not associated with age, another illustrated that the expression of the transporter was subjected to age-dependent increase, and by 6 months of age, similar to OCT1, the protein expression would have reached 50% of the adult level.^33,34

• OATP2B1: mRNA levels of OAT2B1 were significantly higher in adult livers compared with those in livers from fetus at gestational age 18–23 weeks.^14,39 However, quantitative proteomics suggested that OATP2B1 expression in liver from fetus of median 23.4 (range 15.2–41.3) weeks was comparable to that from preterm and term neonates, children, and adults.³³ This lack of correlation with age was supported by 2 other analyses.^34,40

• Sodium taurocholate cotransporting polypeptide (NTCP): Various studies suggested that maturation of NTCP starts during perinatal stage and that the expression reaches adult levels at birth.^{14,34,39,41,42} Protein expression of NTCP revealed a similar trend in which NTCP expression was significantly lower in fetuses than in term neonates, infants, children, and adults and was lower in preterm neonates than in adults.³³

• P-gp: Previously it has been reported that P-gp is subject to developmental changes in the mRNA expression.¹⁴ The transcript level of P-gp was detected as early as 14 weeks gestational age, and the level increased rapidly during the first 12 months of life in infants and then reached a level comparable to that in adults.²⁶ Despite the developmental changes in gene expression, 1 study reported no age-related differences in the relative protein expression in patients from 0.3 to 12 years old.⁴³ Interestingly, however, the results from the 2 recent proteomic studies were in agreement with the mRNA data: P-gp protein expression was low in fetal liver tissues but increased with age.^33,34 Further, TM₅₀ was also estimated to be 2.94 years of age, suggesting that the P-gp expression continued to increase postnatally and would achieve adult level later on in children.³⁴

• MRP2: Using gene expression analysis, previous studies have shown that MRP2 mRNA levels were substantially lower in fetal, neonatal, and infant livers compared with older children up to 12 years of age.^14,26,44 The result reported in 1 of the recent proteomic studies was in agreement with these findings in that MRP2 protein expression was approximately 3-fold lower in fetal and term newborn livers compared with adults.³³ Yet, in another study, it was reported that MRP2 expression was not age dependent in their cohort.³⁴

• MRP3: MRP3 mRNA was detected in fetal hepatocytes as early as 18 weeks gestational age and was significantly lower than that found in adult livers.³⁹ Proteomic data from recent studies agree with this observation. The fetal MRP3 protein level was approximately 3-fold lower than the adult level.³³ Interestingly it was found in 1 study that the transporter expression appeared to be lower in adolescents than in adults.³⁴

• MRP1, MRP4, MRP6: Developmental information on these 3 MRPs is scarce. In their study, van Groen et al showed that MRP1 levels in livers from fetus and term neonates were about 2-fold lower than those in adults.³³ MRP4 mRNA did not change with age.^14,39 Although MRP6 mRNA expression was shown to increase from neonates to older children and adults, no proteomic data are currently available to determine if the actual protein expression shows a similar age-dependent change.⁴⁴

• BCRP: Localization of BCRP in the hepatocytes was detected in fetus as young as 5.5 weeks gestational age.²⁷ BCRP mRNA expression was lower in fetal samples compared with adults.^39,44 BCRP protein levels appeared to be comparable in fetus and after birth in all age groups.^33,34 However, when the data set was analyzed as continuous data by postnatal age and postmenstrual age within the fetal and newborn cohort, BCRP expression interestingly showed an age-dependent decrease with Spearman correlation coefficients of −0.345 and −0.421, respectively.³³

• Bile salt export pump (BSEP): A functional study was conducted using sandwich-cultured fetal and adult hepatocytes, which showed that the biliary excretion index for taurocholate, an endogenous BSEP substrate, was lower in the fetal liver cells than in those of adults.³⁹ Results from quantitative proteomics studies coincide with this observation; the fetal liver tissues expressed significantly lower BSEP compared with term newborns and adults.³³ Maturation of BSEP appeared to occur mainly during the perinatal period, as no significant age-dependent changes were seen from neonates onward.^33,34

• MATE1: In contrast to the age-dependent increase in mRNA reported previously, protein expression of MATE1 appeared to be independent of age.^14,34,44

• Glucose transporter 1 (GLUT1): Developmental information for GLUT1 was previously lacking, but a recent proteomic study indicated that GLUT1 expression showed an age-dependent decrease with fetal liver tissues expressing the highest protein abundance and lower expression in the other age groups.³³ This age-dependent decrease was more apparent in analyses of the expression levels in the youngest cohorts, fetus and newborn, based on the postnatal age and postmenstrual age with Spearman correlation coefficients of −0.51 and −0.59, respectively.

• MCT1: Similarto GLUT1, the ontogeny of MCT1 was missing. The absolute protein abundance of this transporter was found to be comparable in fetal liver and in other age groups after birth.³³

Recent Knowledge Gain on Liver Transporters.

Recent proteomics studies provided valuable ontogeny information for the liver transporters. Although gaps in the developmental changes in various liver transporters such as OAT2 and OAT7 still exist, the understanding in the association between transporter expression and age has been improved substantially, particularly for those transporters that have been shown to be clinically important: BCRP, P-gp, MATE1, and OATP1B1/3.¹³

Ontogeny of Renal Transporters

The kidney is the major site for elimination of many drugs. Three major processes are involved in drug disposition: glomerular filtration, active secretion, and reabsorption. Maturation of glomerular filtration has been studied quite extensively, but information on ontogeny of renal membrane transporters, which are key players in the active secretion, has been relatively scarce.^14,45 Yet, information on the developmental changes in renal membrane transporters has emerged recently. Gene expression of 11 transporters was analyzed from a total of 184 frozen human renal cortical samples from preterm newborn to 75 years of age. The protein expression of 9 transporters and localization of MRP4 using immunohistochemistry were also studied using a subset of the kidney samples.⁴⁶

• BCRP: The mRNA level of BCRP was significantly higher in term neonates compared with other age groups, but the protein abundance appeared to be comparable across all age groups from term neonates to adults. Further studies are warranted to investigate this lack of gene-protein correlation, as only 1 term neonate was included for the proteomic analysis in that study.⁴⁶

• MATE1 and MATE2-K: mRNA and protein levels of MATE1 were independent of age.⁴⁶ Although the transcript level of MATE2-K in term newborns was significantly lower than that in adults, the protein level was found to be comparable across all the age groups studied from term newborn to adults. However, similar to BCRP, the cohort of term neonates for proteomic analysis would need to be expanded in order to better characterize the correlation between gene and protein expression.

• MRP2 and MRP4: mRNA levels of MRP2 and MRP4 appeared to be stable in preterm newborn, term newborn, infants, children, and their older counterparts. Interestingly, proper MRP4 localization was detected as early as gestational age 27 weeks, postnatal 9 days old. This result appeared to accompany the stable gene expression during development.⁴⁶

• Uric acid transporter 1 (URAT1): The mRNA and protein abundance of URAT1 increased with age from term newborn to adults.⁴⁶

• P-gp: Similarly to those of the liver and intestine, the ontogeny of renal P-gp was studied relatively extensively. P-gp localization was detected as early as the end of first trimester of fetal life.²⁵ Results from gene expression analysis and quantitative proteomics expanded the understanding of the developmental changes of P-gp in kidney. P-gp mRNA levels were significantly lower in preterm newborns, newborns, and infants as compared with children, adolescents, and adults. This observation appeared to be translated well to protein expressions. A sigmoidal E_max model described this age-dependent increase, and the TM₅₀ was approximately 1 month.⁴⁶

• GLUT2: An efficient carrier of glucose, GLUT2 did not show age-dependent changes in its mRNA expression and protein abundance.⁴⁶

• OAT1 and OAT3: The ontogeny of these 2 organic anion transporters was reflected in clinical data.⁴⁵ For instance, 1 study showed that the secretion capacity of p-aminohippurate, an OAT1/3 substrate, appeared to be about one fifth of the adult level at birth.^47–49 These observed age-related changes in the PK of transporter substrates are likely due to a combination of maturation in both transporter expression and glomerular filtration. Yet, the changes in transcript levels and protein abundance aligned with the clinical observations. mRNA and protein expressions for both OAT1 and OAT3 increased with age with a TM₅₀ of approximately 5 months and 8 months, respectively. Further, intertransporter correlation analysis also demonstrated that these 2 transporters were highly correlated in their gene and protein expressions.⁴⁶

• OCT2: Similar to OATs and P-gp, the OCT2 mRNA levels and protein abundance are age dependent with the levels in newborns being significantly lower than those in children and adults. As with P-gp, OCT2 would reach half of the adult level about 1 month after birth.⁴⁶

Recent Knowledge Gain on Renal Transporters.

The data from gene expression analysis, quantitative proteomics, and immunohistochemistry have painted a more complete picture for the ontogeny of renal membrane transporters. Of the 6 transporters that are clinically important and should be carefully considered in drug development, 4 of them, P-gp, OAT1, OAT3, and OCT2, showed age-dependent increases in their expression levels. This implies that drug substrates of these transporters would also be subject to age-dependent changes and might impact the elimination of these drugs in pediatric patients. Despite this increase in knowledge, more studies on term and preterm neonates are needed to better capture the variability in the age-related changes of transporter expression and also the interplay with maturation of glomerular filtration during this rapid developmental phase of life.^50,51

Application of Ontogeny of Transporters to MIDD in Children

Overall, there has been a recent surge in data on the ontogeny of membrane transporters, which will greatly enhance our understanding in not only the disposition of drug substrates but the involvement of these transporters in developmental physiology in children. Although scarce data on the ontogeny of intestinal transporters still limit their application to modeling and simulation of oral drugs, the wealth of data in the domains of hepatic and renal transporter ontogeny present an opportunity to be leveraged for pediatric PBPK modeling, especially for intravenously administrated drugs, to assist the prediction in drug disposition and clearance in children.

The workflow of pediatric PBPK model development has been previously described (see Figure 2).^52–55 In most cases an adult PBPK model is first established, verified, and refined. This model is comprised of the drug profile as well as the virtual adult population in which transporter protein abundance data and kinetics parameters, such as the Michaelis constant and maximum rate, can be incorporated to predict the organ-specific clearance (CL).⁵⁶ Following the finalization and verification of the robust adult PBPK model, pediatric models could be generated by modifying the population-specific inputs (eg, blood flow to organs, organ weights, and protein abundance of drug-metabolizing enzymes and transporters) using algorithms or parameters such as ontogeny scaling factor for transporter abundance or intrinsic clearance. These are expressed as a function of age and can be derived from the developmental changes in the expression data described in previous section. Of note, although pediatric PBPK models can also be established based on drug physiological properties and preclinical data alone, this approach could lead to a lower confidence in the prediction compared with a model that is verified with adult clinical data.

Figure 2.

Workflow of the pediatric physiologically based pharmacokinetic model (PBPK) establishment to simulate drug exposure in children. An adult PBPK model was first established and verified by comparing the output from the simulations to that in observed data.After ensuring that the adult model was robust, the pediatric model was generated by scaling the anatomical and physiological parameters using default age-dependent algorithms and incorporating ontogeny information for the transporters that are pertinent to this study. The pediatric PBPK model was verified, once again, by comparing the output from the simulation input with observed data from literature. Predictive performance of allometry and PBPK in estimating the clearance in children was also evaluated. CL indicates clearance; ped, pediatric; vs, versus.

The success and confidence of PBPK modeling and simulation that involves transporter-mediated disposition using a bottom-up approach are critically dependent on factors such as the quality and availability of transporter kinetic data and understanding of in vitro–in vivo correlation. The following case study illustrates that the utility of leveraging transporter ontogeny data in PBPK modeling, with sufficient information gathered, can be useful to simulate drug exposure in pediatric patients.⁵⁷

PBPK Modeling With Integrated Transporter Ontogeny Reasonably Predicted the Exposure of an Actively Renally Secreted Drug in Children

Tazobactam is a β-lactamase inhibitor. Currently, it is formulated as an intravenously administered combination product with either piperacillin, a β-lactam, or ceftolozane, a cephalosporin, as a broad-spectrum antibiotic.^58,59 Tazobactam is prescribed for infections that could potentially be life threatening when left untreated. Hence, sufficient exposure is needed to assure therapeutic action without adverse events. As much as 80% of tazobactam is renally cleared in adults.^59,60 In addition to glomerular filtration, tazobactam undergoes active tubular secretion that is mediated by OAT1 and OAT3.^60,61 The remaining tazobactam is either converted to the inactive metabolite, M1, via hydrolysis and then eliminated renally, or it undergoes biliary excretion (Figure 3).^60,62

Figure 3.

The elimination pathway of tazobactam. After intravenous administration, approximately 80% of tazobactam would be cleared renally by glomerular filtration and active secretion via OAT1 and OAT3. The majority of the rest of tazobactam would undergo hydrolysis to form the inactive metabolite, tazo-M1, which, similar to the parent drug, will be eliminated renally. A small amount (<1%) of tazobactam would undergo biliary excretion. CL indicates clearance; CL_R, renal clearance; GFR, glomerular filtration rate; OAT, organic anion transporter.

The workflow of the adult and pediatric PBPK model establishment is summarized in Figure 2. OAT1/3 protein abundance and transporter kinetics from in vitro studies are obtained from the literature.^46,61 The ontogeny scaling factors, which are the sigmoidal E_max functions, of OAT1/3 were incorporated.⁴⁶ To address the argument on whether PBPK modeling is preferred over an allometric scaling approach in predicting the PK for pediatric patients <2 years old, the clearance in the pediatric cohorts was also estimated based on allometry and compared with that predicted using PBPK models.⁹ Noteworthy, previous studies have noted that the allometric exponent may be different for various pediatric age groups. However, since a validated exponent has not been established for neonates and infants, a fixed allometric exponent of 0.75 was applied to the pediatric cohorts in this study for simplicity.

The PBPK model captured the exposure of tazobactam after a 500-mg 60-minute IV infusion in adults well with the predicted maximal concentration, area under the curve, and clearance (CL) between 1.02- to 1.2-fold of the observed data (Figure 4a).⁶³ After verifying and ensuring that the adult model was robust, 3 virtual pediatric populations were generated using the approach as outlined above: 0 to 3 months, 3 months to 2 years old, and 2 to 7 years old. Maximal concentration, area under the curve, and CL were all within 1.5-fold of observed data when the simulation was performed in these 3 cohorts, suggesting that the pediatric PBPK model predicted the exposure of tazobactam adequately in neonates, infants, and children (Figure 4b).⁶⁴ Significantly, allometric scaling approaches resulted in CL estimations that were comparable to those predicted using PBPK. However, for the youngest age group, 0–3 months old, the PBPK model performed slightly better, as allometry slightly overpredicted the CL (1.2-fold versus 1.8-fold of observed CL).

Figure 4.

Simulation of tazobactam exposure in adult population (a) and 3 pediatric cohorts (b-d) using PBPK (PK-Sim v7.3, Bayer, St. Louis, Missouri). Following a 500-mg 60-minute infusion in the virtual adult population, the predicted C_max, AUC, and CL were between 1.02- and 1.2-fold of the observed data.⁶³ Three pediatric cohorts were generated: 0–3 months old (b), 3 months to 2 years old (c), and 2–7 years old (d). By taking into account the physiological and anatomical changes during development, and the ontogeny of transporters that are pertinent to the disposition of tazobactam, the tazobactam exposure was predicted reasonably well with C_max, AUC, and CL were all within 1.5-fold of observed data.⁶⁴ Allometric scaling approach resulted in CL estimation that were comparable to that predicted using PBPK. However, for the youngest age group, 0–3 months old (b), the PBPK model performed slightly better, as allometry slightly overpredicted the CL (1.2-fold vs 1.8-fold of observed CL). AUC indicates area under the concentration-time curve; CL, clearance; C_max, peak plasma concentration; PBPK, physiologically based pharmacokinetics.

This case study illustrates the utility of a pediatric PBPK model with integrated renal transporter ontogeny function in simulating exposure of a drug that is actively renally secreted in pediatric patients. It exemplifies how this approach could be applied in pediatric drug development to support decision making on dosing to limit unnecessary exposure in pediatric patients. Further, it also highlights how a PBPK model can be used to complement the allometric scaling approach by predicting the whole PK (concentration-time) profile rather than just the drug clearance.

Considerations

The recent emergence of quantitative proteomics data on the expression and ontogeny of transporters substantially improves the predictive power of pediatric PBPK models for drug substrates. Nonetheless, there are important factors that need to be considered when incorporating protein abundance into PBPK models. Although liquid chromatography/tandem mass spectrometry-based quantitative proteomics provides a powerful tool by using peptide sequences to measure the absolute abundance of transporter proteins, it could not acknowledge if the transporters are successfully localized to the membrane, nor could it distinguish truncated protein and splice variants from properly formed proteins or glycosylated from nonglycosylated protein.⁶⁵ Consequently, these would undermine the assumed correlation between transporter expression and activity. Further, scaling from the protein abundance data per crude membrane protein or per gram tissue level to per organ level is the first step in integrating such data into PBPK. This should be done carefully, as parameters that are used to scale, such as membrane protein yield per gram of tissue, could also be subject to age-related changes.³³ Last, it is important to reiterate that the success of PBPK simulation in children using a bottom-up approach depends highly on knowledge of the drug disposition pathway and the data available for the ontogeny of the metabolic enzymes and transporters involved. For instance, with a good understanding of the maturational differences between UGT enzymes and sulfotransferase, 1 study successfully predicted the exposure, as well as metabolic formation and elimination, of acetaminophen, which is mainly glucuronidated in adults but almost exclusively sulfated in newborns due to age-dependent changes in the UGT enzymes expression and activity, in various pediatric age groups using PBPK.⁶⁶ Nonetheless, for drugs that are substrates of certain metabolic enzymes and transporters of which the developmental changes are not fully understood, results from the simulation should be interpreted carefully.

Conclusion/Future Direction

Collectively, international collaborative efforts have greatly improved the understanding of the role(s) of transporters in drug PK, pharmacodynamics, safety, and efficacy not only in adults but also in specific populations such as pediatrics. This understanding is supported by the expansion of knowledge in the ontogeny of membrane transporters, especially those in the liver and kidney. This increased knowledge has significant implications for PBPK modeling for drug substrates and therefore is of great importance for pediatric drug development. However, knowledge gaps in the ontogeny of transporters in the intestine and other important barrier tissues such as the blood-brain barrier remain and are waiting to be addressed through future collaborative work. Further investigation would also be required to elucidate how gene and protein expression relate to transporter activity. Last, as illustrated in the case study, in addition to ontogeny, a thorough understanding of the disposition of drugs and their interplay is critical in the application of PBPK to adequately predict drug exposure in children.