Ontogeny of Pediatric Pharmacogenetics: Celebrating the Past and Vision for the Future.

Abstract

It is, indeed, an honor to be selected as the 2020 recipient of the Sumner J. Yaffe Lifetime Achievement Award in Pediatric Pharmacology and Therapeutics. Dr. Yaffe's role as the Father of Pediatric Clinical Pharmacology1 was well-known to me during my formative years in pediatric clinical pharmacology at The Hospital for Sick Children in Toronto. However, it was not until my move to join Drs. Greg Kearns (Yaffe Award, 2008) and Ralph Kauffman at Children's Mercy Hospital in Kansas City in 1996 and the opportunity to build a new pediatric clinical pharmacology program at that institution that I first met Dr. Yaffe personally in the context of meetings of the National Institute of Child Health and Human Development (now Eunice Kennedy Shriver National Institute of Child Health and Human Development) Network of Pediatric Pharmacology Research Units (PPRUs). His enthusiasm for pediatric clinical pharmacology was infectious, and his commitment to the continued development of the field is a legacy to which we can all aspire.As the title of my presentation implies, my career in pediatric clinical pharmacology has focused on the relative contributions of ontogeny, or stage of development, and genetic variation as major factors contributing to observed variability in drug disposition and response in children. The path I followed in getting to this point was quite circuitous and more a consequence of discovering what I did NOT want to do, rather than any clear understanding of where my true passions lay. In fact, I distinctly recall declaring that I had no interest in research (at my interview for acceptance into the PharmD program at SUNY Buffalo; not surprisingly, I was not accepted) and that I had no interest in pediatrics after my clinical rotation in neonatology while in the PharmD program at the University of Minnesota. Approximately six months later in July, 1982, I was beginning a fellowship in pediatric clinical pharmacology at The Hospital for Sick Children in Toronto, in the same class as Dr. Gideon Koren (Yaffe Award, 2012). Exposure to pharmacogenetics occurred approximately two years later when I began my PhD in Pharmacology training under the mentorship of Dr. Stephen Spielberg (Yaffe Award, 2009) and with the participation of Prof. Werner Kalow – considered to be one of the founding fathers of pharmacogenetics2 as a member of my thesis advisory committee. In fact, one of my prized possessions is a copy of Dr. Kalow's book, Pharmacogenetics. Heredity and the Response to Drugs,3 the first book written on the subject and that he graciously personalized for me (Image 1).In my search for data characterizing the developmental trajectory of the field of pediatric pharmacogenetics I came upon two particularly surprising findings. The first I attribute to Ralph Kauffman, MD, an accomplished pediatric clinical pharmacologist who had moved to Children's Mercy Hospital in 1995 to establish the Department of Medical Research. Dr. Kauffman directed me to Harry C. Shirkey's Pediatric Drugs 1966–1967, 2nd Edition, and specifically to Chapter15, “Pharmacogenetics”, authored by William L. Nyhan. My interest piqued, a search of Google Books revealed that “pharmacogenetics” was mentioned six times in the 1964 1stEdition, mostly referencing Professor Kalow's text. The first “Pharmacogenetics” chapter in the 2ndEdition essentially acknowledged the growing interest in the study of genetic variation as a factor contributing to variability in drug response, with the content limited to a discussion of G6PD deficiency and succinylcholine sensitivity and little discussion of pediatric applications. By the 4thEdition, also authored by Dr. Nyhan, the Pharmacogenetics chapter (Chapter 13) had expanded to include a section on drug-induced hemolysis in the newborn, the phenylthiourea bitter taste polymorphism, isoniazid pharmacogenetics and hemoglobinopathies associated with sulfonamides. Although the therapeutic applications of pharmacogenetic principles to pediatrics was minimal in the first four editions of Pediatric Drugs, Dr. Shirkey's foresight of the potential role of genetic variation as a determinant of drug response in the pediatric population is quite remarkable.The second surprising finding was the contribution of Dr. Yaffe toward an understanding of the interplay between ontogeny and genetic variation during development. In this regard, Dr. Yaffe and colleagues4 presented a paper at the first international conference on Pharmacogenetics organized by Bert LaDu and Werner Kalow, with the proceedings published as an issue of the Annals of the New York Academy of Science in 1968 (Image 2). In the first paragraph of this paper, the authors acknowledged the existing knowledge at the time: 1) the developing fetus and newborn infant were more sensitive to drug effects relative to adults; 2) drug metabolic processes were observed to be lower in newborns compared to adults; and 3) that variations in drug metabolism appear to be genetically determined, with apparent species and strain differences. On the basis of this background, the stated purpose of their paper was: “… to determine how soon after birth one could detect and differentiate the contribution to drug metabolism made by genetic endowment”.First, to assess the genetic contribution to drug effect, hexobarbital sleeping time in minutes was compared in three strains of adult mice – C57BL, Balb/C and I29J. Sleeping time was shortest in the C57BL mice and longest in I29J, and intermediate in Balb/C for both males and females, but for each strain sleeping time was shorter in female mice compared to male mice. A subsequent study then compared hexobarbital sleeping time in male Balb/C and I29J mice at four different post-natal time points between one day and three weeks of age. Similar strain differences were observed (sleeping time in I29J mice longer than in Balb/C mice), and for each strain sleeping time decreased with increasing postnatal age, being at least 40-fold greater at one day of age compared to three weeks of age. A comparison of the hexobarbital concentrations in plasma upon wakening in each group confirmed that the observed age-dependent differences in sleeping time were due to developmental changes in drug clearance rather than age-related differences in sensitivity to the drug. Additional studies reported in this paper revealed that distinct developmental trajectories were present for glucuronidation of bilirubin and p-nitrophenol as well as unusual variation in phenolsulfotransferase activity with increasing postnatal age. Overall, this work illustrated the importance of characterizing the developmental trajectories of drug disposition pathways as an important first step in informing rational drug dosing throughout infancy and childhood and was the first to draw our attention to the possibility that genetic variation may impact those developmental trajectories.Dr. Yaffe's contribution to pediatric pharmacogenetics also included his role as editor of the first issue of Pediatric Clinics of North America to be devoted solely to pharmacology. Of interest, the second article in the symposium was a paper on Pharmacogenetics co-authored by Drs. Sandy Cohen and Wendell Weber,5 which covered much of the same material as the chapters in Pediatric Drugs. However, Cohen and Weber also described several phenomena relating to differences in genetically mediated drug responses between children and adults. For example, they noted that peripheral neuropathy associated with isoniazid was relatively common in adults and associated with the slow acetylator phenotype; in contrast, this genetically mediated side effect was observed to be quite rare in children and could not be explained. Malignant hyperthermia had not been reported in children less than three years of age, but children between the ages of three and 10 years of age appeared to be particularly susceptible. Finally, studies of glutathione stability, a marker of G6PD deficiency revealed that tests are abnormal in infants less than 78 hours of age, implying that phenotypic G6PD deficiency does not become clinically apparent until later after birth. As I will discuss later, interest in identifying the age at which drug metabolism/clearance phenotype is concordant with genotype was an important driver for our own studies related to cytochrome P450 2D6 (CYP2D6). The Cohen and Weber article also described reports of resistance to treatment of coumarin anticoagulants in children. The apparent resistance was attributed to genetic variation in the “receptor site for anticoagulants” several decades before discovery of vitamin K oxidoreductase complex 1 (VKORC1) as the target of warfarin action and the role of genetic variation in VKORC1 as a determinant of warfarin dosing.Collectively, the early days of pediatric pharmacogenetics can be characterized by an increasing awareness of the role for genetic factors contributing to variability in drug response, largely involving adverse drug reactions affecting newborns, infants and children, including speculation regarding the risk of birth defects. The state of knowledge was essentially that the consequences of the same genes and same drug could also affect children, but the adverse drug reaction phenotypic traits could be different in children compared to adults – the lower risk of isoniazid-induced peripheral neuropathy and increased risk of malignant hyperthermia being two examples. However, there do not appear to have been any investigations specifically targeting the genetic basis of a drug-related phenotype in children. In their concluding remarks Cohen and Weber state: “The recent advent of systematic surveillance of hospitalized patients for adverse drug reactions, coupled with technical advances that provide more sensitive methods to study drug disposition in the body, will provide indicators of individual differences in drug response that are sensitive enough for rapid detection of new inherited traits. These advances will undoubtedly bring the importance of pharmacogenetics into better focus in the future.”It would take almost 10 years for this final statement to begin to bear fruit for children.Although many key pharmacogenetic discoveries (acetylation of isoniazid, primaquine and G6PD deficiency, succinylcholine and cholinesterase deficiency) were made in the 1950s, more clinically relevant discoveries, at least with respect to commonly used medications used today, were reported later (1970s and 1980s). Genes, such as CYP2D6 and CYP2C19, are certainly involved in the clearance of medications used by children as well as adults, but the discoveries of genetically variable phenotypes and the subsequent elucidation of the genetic basis for those phenotypes were made following observations of unexpected phenomena in response to drugs used to treat adult patients (e.g., debrisoquine, sparteine, mephenytoin). A major exception to this general paradigm was discovery of the thiopurine S-methyltransferase (TPMT) polymorphism reported in 1980 by Weinshilboum and Sladek6 in what I believe is the first phenotyping study that included children and involved a medication (substrate; 6-mercaptopurine (6-MP)) used to treat a primarily pediatric condition (acute lymphoblastic anemia, or ALL).Although the initial population distribution of TPMT activity was measured in erythrocyte lysates from 298 randomly selected blood donors at the Mayo Clinic, this study also reported TPMT activity in 115 children with a mean age of 13.0 ± 0.4 years.6 In adults, wide variation in erythrocyte TPMT activity segregated as a monogenic trait consistent with autosomal codominant inheritance, and the distribution of TPMT activity (including the frequency of the low activity phenotype) was similar in the pediatric group. In 1983, Lennard et al7 provided evidence for a relationship between high intracellular 6-thioguanine (6-TGN) concentrations and the presence of neutropenia in children with ALL treated with 6-MP, and a subsequent collaboration with Dr. Weinshilboum established the relationship between genetically controlled TPMT activity and intracellular 6-TGN concentrations in 40 children with ALL receiving long-term 6-MP treatment.8 This study also reported some interesting observations that imply the presence of potential developmental factors, such as higher TPMT activity in children with ALL treated with 6-MP relative to healthy adult controls, as well as higher TPMT activity in children with ALL treated with 6-MP relative to children with ALL in remission and not receiving drug, implying potential inducibility of TPMT activity; the issues of ontogeny and inducibility have been the focus of additional investigations.The observation of a genetically driven trait with an important clinical impact in a vulnerable patient population naturally led to efforts to elucidate the mechanisms involved. The group at St. Jude Children's Research Hospital cloned TPMT cDNA from a 6-MP-treated TPMT-deficient patient who developed severe hematopoietic toxicity and identified the TPMT*2 allele,9 whereas the Mayo Clinic group cloned the TPMT gene and subsequently characterized the TPMT*3A and TPMT*3B alleles.10 It is fair to say that TPMT has been the subject of extensive investigation in ALL and pediatric inflammatory bowel disease throughout the 1980s, 1990s and 2000s, and these efforts continue to the present time culminating in the publication of TPMT genotype-guided dosing recommendations in 2011,11 updated in 2013,12 and expanded to include NUDT15 in 2019.13 Publication of these guidelines represents a tangible product of the expectations of investigations to “bring the importance of pharmacogenetics into better focus” expressed by Cohen and Weber in 1972.5The formative stages of my academic career were characterized more by learning what I did NOT want to do after completion of a degree or training program, rather than having a clear idea of what I wanted to do from a long-term career perspective. In fact, I distinctly recall declaring that I was not interested in research during my interview for acceptance into the PharmD program at SUNY-Buffalo (not surprisingly, I was not accepted), after completion of my neonatology rotation for the University of Minnesota PharmD program I knew that pediatrics was an area I wanted to avoid; it is safe to say that establishing a research career in “pediatric pharmacogenetics” definitely was not on the radar screen. Just a few months after the neonatology rotation I found myself in the Division of Clinical Pharmacology at The Hospital for Sick Children (SickKids) in Toronto, first completing a post-PharmD fellowship in pediatric clinical pharmacology and then a PhD in Pharmacology under the supervision of Stephen P. Spielberg, MD, PhD. My experience illustrates how important a stimulating and engaging research environment can be for trainees in the formative stages of their career development, something that I have tried to emulate for other developing clinician-scientists. Dr. Spielberg provided me with my first exposure to pharmacogenetics through his work on anticonvulsant and sulfonamide hypersensitivity reactions. A genetic basis for susceptibility to the idiosyncratic events involving a detoxication pathway was inferred from the results of an in vitro rechallenge in which peripheral blood mononuclear cells from patients experiencing a phenytoin hypersensitivity reaction, for example, were exposed to a standardized challenge of reactive metabolites generated from phenytoin by a murine liver microsomal drug oxidation system; patient cells demonstrated differential toxicity relative to cells from healthy controls, and intermediate cell death was observed in cells from parents.14 My PhD thesis project was to purify the cytochrome(s) P450 responsible for phenytoin bioactivation and substitute a purified, reconstituted system for mouse liver microsomes in the in vitro rechallenge assay. When Dr. Spielberg explained the project I was not deterred by the fact that I did not know what a “P450” was (we only learned about a “microsomal drug oxidizing system” in pharmacology lectures), but I soon learned that there were two types of P450s: aromatic hydrocarbon-inducible and phenobarbital-inducible; I set out to purify the latter. More important than the project, however, was the environment at SickKids and the connections that would be made through the Department of Pharmacology at the University of Toronto where I was enrolled for the PhD program. Specifically, Professor Kalow served as a member of my advisory committee, and I overlapped briefly with a graduate student of his, Denis Grant, who had completed his PhD on the hepatic N-acetyltransferase (NAT2) polymorphism and was on his way to Basel, Switzerland where he undertook a post-doctoral fellowship with Prof. Urs Meyer. While in Basel Dr. Grant met Andrea Gaedigk, a PhD student with Prof. Michel Eichelbaum in Stuttgart and completing some of the experimental work in Basel with Prof Meyer and his group for what would subsequently be called the CYP2D6*5 gene deletion event; of interest, Drs. Bill Evans (Yaffe Award, 2004) and Mary Relling (Yaffe Award, 2011) were also visiting Prof. Meyer's laboratory on sabbaticals. Upon completion of his post-doctoral training Dr. Grant returned to Toronto, and specifically to assume a faculty position in the Division of Clinical Pharmacology at SickKids where Dr. Spielberg was now Division director. I, too, assumed a faculty position upon completion of my training and shared an office with Dr. Grant for the next seven years. Once she completed her PhD with Prof. Eichelbaum, Dr. Gaedigk arrived at SickKids to continue her training with Dr. Grant, setting place the building blocks for development of a future pediatric pharmacogenetics program, albeit not at SickKids.In mid-1996 Dr. Gaedigk and I joined Drs. Ralph Kauffman and Gregory Kearns at Children's Mercy Hospital in Kansas City, where they had been recruited to lead the Department of Medical Research and build a new Division of Clinical Pharmacology in the Department of Pediatrics, respectively. The program quite rapidly developed a critical mass with Jim Marshall, MD, a pediatric intensivist who had trained with Dr. Kearns at Arkansas Children's Hospital, and Doug Blowey, MD, a pediatric nephrologist from Children's Mercy Hospital who received his clinical pharmacology training in our group at SickKids; it was not until a month or two before leaving SickKids that I realized that I was moving to the same institution where Dr. Blowey was returning. A couple of years later, Susan Abdel-Rahman, PharmD, also joined the growing program, further expanding the breadth and expertise of the group.In the context of pediatric pharmacogenetics, interest was growing not only in characterizing genotypephenotype relationships in pediatric populations, but also in in vitro studies characterizing the developmental trajectories of clinically relevant drug biotransformation pathways. For example, in 1989 Drs. Evans and Relling followed up their introduction to pharmacogenetics with the first CYP2D6 in vivo phenotyping study in children, using dextromethorphan (DM) as the phenotyping probe in children aged 3 to 21 years .15 In 1990, in vivo CYP2D6 activity was compared between 13 children with autoimmune hepatitis and 31 unaffected children by a French group who also used DM as the phenotyping probe,16 and a year later Evans and Relling demonstrated CYP2D6 genotype-phenotype concordance in 116 pediatric participants with a median age of approximately 10 years.17 Around the same time, in vitro studies were revealing that CYP2D6 expression and activity were relatively low in fetal liver and increased after birth.18,19 Thus, existing knowledge was that CYP2D6 was poorly expressed in fetal liver and genotype-phenotype concordance was present later in childhood, setting the stage for our efforts to characterize the developmental trajectory of CYP2D6 in vivo. Our laboratory at the time was just down the hall from the operating theaters, and it was not uncommon for one of the young orthopedic surgeons to drop by for a chat. On one occasion he was particularly disturbed by an interaction he had had with the mother of one of his patients. She was upset by the fact that the codeine her child was prescribed at discharge was not anywhere near as effective as the morphine the child had been prescribed as an inpatient. Dr. Gaedigk and I explained that codeine had limited analgesic activity itself and had to be metabolized to morphine by CYP2D6 to be effective. Furthermore, CYP2D6 activity was limited at birth and perhaps his patient had not yet acquired the capacity to convert codeine to morphine. He then asked us how long it took for an infant to acquire functional CYP2D6 activity – a question to which we had no answer.This interaction ultimately led to the design of a longitudinal phenotyping study in healthy term infants that was coordinated with well-baby visits at two weeks, one month, two months, four months, six months and 12 months of age. Participating mothers were provided with a small dose of DM (0.3 mg/kg) that they administered after the last evening feed, and all diapers were collected overnight. Urine was expressed from the diapers, and DM and its metabolites (dextrorphan (DX), 3-methoxymorphinan and 3-hydroxymorphinan) were measured. CYP2D6 activity was then assessed as the ratio of DM/DX in the urine samples and compared to CYP2D6 genotype determined from genomic DNA. As an aside, the urine collection process turned out to be somewhat more challenging than we initially anticipated. Commercially available diapers were designed to keep babies dry using a bead matrix from which it was virtually impossible to recover urine. During this formative phase of study development we were assisted by John van den Anker, MD, PhD (Yaffe Award, 2019), who was visiting our group from Erasmus University in Rotterdam on a mini-sabbatical. For his efforts, Prof. van den Anker was renamed “Dr. van den Diaper”, for which I am sure he is extremely proud. The problem was eventually solved by identifying a source for wood pulp fiber-based diapers from which DM and DX could be successfully extracted from recovered urine. Ultimately, five other sites in the PPRU network contributed participants to this study.The results of this study revealed that CYP2D6 phenotype, as assessed by urinary DM/DX ratios, was concordant with phenotype by two weeks of age, the first of the study visits.20 Rapid acquisition of CYP2D6 activity in the first days of life has also been observed in preterm infants using tramadol O-demethylation to assess CYP2D6 activity, and the results of a series of studies published between 2005 and 2008 reveal that while genetic variation in CYP2D6 is an important factor contributing to inter-individual variability in tramadol clearance in critically ill newborn infants, it is not the only factor.21 In fact, genotyping may be of limited value for guiding dosing of CYP2D6 substrates in a neonatal intensive care setting due to the impact of non-genetic factors, such as maturation of renal function. However, this situation also suggests that a genotype-stratified study design may allow for a comparative analysis of the contribution of genotype by assessing the magnitude of effect on clearance or systemic exposure between genotypic extremes (e.g., poor metabolizers vs extensive or ultra-rapid metabolizers) and by identifying additional factors that contribute to inter-individual variability within a genotype group.An increasing number of studies investigating the role of genetic variation as a factor contributing to variability in drug disposition and response have been published over the past several years, and several excellent reviews on the subject have been written by investigators at several institutions around the world. Examples of disease areas (drugs) that have been the subject of investigation include attention-deficit/hyperactivity disorder (atomoxetine), autism, asthma, bone marrow transplant, congenital heart disease (warfarin), cystic fibrosis, epilepsy, fungal infection (voriconazole) HIV infection (efavirenz), inflammatory bowel disease (azathioprine), juvenile idiopathic arthritis (methotrexate), Kawasaki disease, oncology (6-MP, cisplatin, daunorubicin, among others), neonatology, pain (morphine), and solid organ transplant (mycophenolic acid, tacrolimus). Unfortunately, review of published CPIC guidelines reveals that pediatric-specific guidance within these guidelines is generally quite limited. For example, the CPIC guideline for CYP2D6 and CYP2C19 genotypes and dosing of selective serotonin reuptake inhibitors (SSRIs) contains the statement: “Data describing the relationship between CYP2D6 or CYP2C19 genotype and SSRI systemic exposure or steady state concentrations in pediatric patients are scarce. Because CYP2D6 activity is fully mature by early childhood, it may be appropriate to extrapolate these recommendations to adolescents or possibly younger children with close monitoring.” 22Similar language can be found in other CPIC guidelines, such as for tacrolimus and statins; in the absence of pharmacokinetic data assessing the magnitude of the effect of drug metabolizing enzyme or transporter genotype on drug clearance in pediatric age groups, extrapolation from adult experience becomes the norm, one exception being atomoxetine, a medication used in the management of children and adolescents with ADHD.23 Nevertheless, this situation provides an opportunity for the design and conduct of studies to address this knowledge deficit in the future.Over my career I have benefited to a tremendous extent from ongoing interactions with many individuals, not the least of which been the trainees and early career investigators I have had the privilege to mentor. Through my own and their investigations we have learned several lessons and gained many insights that inform new directions for pediatric pharmacogenetics research in the future. Some thoughts for consideration by the next generation of pediatric clinical pharmacologists are included in the following 4 areas.Genotype-stratified pharmacokinetic studies. Design of genotype-stratified pharmacokinetic studies to efficiently establish the expected extremes within a population, the magnitude of pharmacogenetic effect on the dose → exposure relationship, and secondary analyses to explore additional factors contributing to within-genotype variability. As an example, SLCO1B1-genotype-stratified pharmacokinetic studies of simvastatin,24 pravastatin,25 and rosuvastatin26 in dyslipidemic children and adolescents have revealed considerable inter-individual variability in dose-corrected systemic exposure following fixed doses of the statins, and that within-genotype variability (~10-fold) is greater than the difference between SLCO1B1 genotypes (~2.3-fold). Furthermore, non-genetic factors contributing to the within genotype variability are unique to each statin, indicating that SLCO1B1 genotype alone is not sufficient to reduce inter-individual variability in the dose-exposure relationship and inclusion of additional factors will be necessary to develop model-informed strategies for individualized dosing of statins in children and adolescents.Opportunistic sampling strategies. Adopt rich, intensive opportunistic sampling strategies whenever possible. Collection of plasma samples obtained for clinical purposes before they are discarded has become a popular strategy for developing population pharmacokinetics models, particularly in premature newborns. Although quite labor-intensive, concurrent near-quantitative collection of urine over an extended period of time and pooling in 12-hour bins, for example, may lead to important insights into the ontogeny of drug disposition pathways in vivo, as we discovered for preterm newborns treated with indomethacin.27 The original study design called for collection of urine over 24 hours after the first dose of indomethacin, and analysis revealed less that 1% recovery of the administered dose within this time period. As a result, urine collection was extended for up to seven days after the last dose. Analysis of the urine for indomethacin, its acylglucuronide metabolite and O-desmethyl indomethacin revealed three distinct patterns of metabolite recovery, and more importantly that changes in the relative contribution of competing drug disposition pathways occurring in vivo during the first few days of life will not be captured by limiting sample collection to the first 24 hours after drug administration. Furthermore, inter-individual variation in the developmental trajectory of drug biotransformation pathways appears to be present, and accurate characterization is an essential first step to identifying factors contributing to the observed variability and the clinical consequences.Prospective validation and refinement of pharmacokinetic models for clinical implementation. There are many published papers describing the development of pharmacokinetic models characterizing the disposition of drugs in pediatric patients, but few are ever prospectively validated to ascertain how well they actually perform in a clinical setting. I would argue that an amount of effort equivalent to the model development process (and probably considerably more) should be expended for prospective validation. I acknowledge that this situation creates a challenge from the perspective of resources (financial and personnel) required to recruit the independent, prospective cohort, but an ultimate goal of improving the safety and efficacy of medications used in pediatrics through model-informed precision dosing necessitates that the performance of these models be validated and refined in a prospective manner.Ontogeny and genetic variation impact on drug action and response. One area that has received relatively little attention, at least relative to pathways involved in drug disposition, is characterization of the impact of ontogeny and genetic variation on the targets of drug action and ultimately response to medications in pediatric patients. However, when genetic variation in drug clearance results in a wide range of exposures for a given dose, the variability in exposure obscures the contribution of variation in drug target to the observed variability in drug response and is difficult or impossible to detect, as is illustrated in Figure 2d in reference 28. In this context dose escalation study designs are not helpful or realistic given the large number of patients that would need to be recruited to overcome the limitations imposed by the excessive variability in exposure. An alternative approach is use of exposure-controlled/escalation studies to investigate the exposure → response relationship for a given medication and to establish therapeutically relevant exposure ranges. The concept of controlling systemic exposure is not new, having been proposed by Dr. Carl Peck and colleagues more than 20 years ago.29 However, individualization of dose to achieve a target systemic exposure requires robust, prospectively validated models as described above, further supporting point 3 above as a critical need for the future.The origin of pediatric pharmacogenetics is tightly linked to the development of pediatric clinical pharmacology as a discipline, and Dr. Sumner Yaffe was a driving force for both. Although there have been a few cases of demonstrable impact of pediatric pharmacogenetics (e.g., TPMT and NUDT15 in ALL and regulatory changes in the codeine label), there are few examples of widespread, data-informed integration into clinical practice. Genotype-stratified pharmacokinetic studies have the potential to efficiently capture the magnitude of pharmacogenetic effect and to help identify additional sources of variability. Future investigations should utilize all sources of new data, especially the value of opportunistic sampling for pharmacokinetic and pharmacodynamic studies, including identification of biomarkers predictive of drug disposition and response. Finally, generating more pharmacokinetic models is not sufficient; prospective validation of model performance in clinical settings is essential to advance the field.In concluding remarks at the first international symposium on pharmacogenetics30 Professor Kalow indicated that clinical pharmacologists were most likely to drive the field of pharmacogenetics forward, (“the more clinical pharmacologists there are, the more rapidly is pharmacogenetics likely to advance”) and I believe his statement can be extended to the more pediatric clinical pharmacologists there are, the more likely is pediatric pharmacogenetics to advance. Certainly, pediatric pharmacogenetics has been closely intertwined with pediatric clinical pharmacology since its early days, and I am extremely honored to have been a part of this process.September 12, 2021