Current perspective on pediatric pharmacogenomics

The goal of pharmacogenomics is to comprehend the effect of inter-individual genetic variations on the outcome of treatment.[1,2] By understanding the role that these genetic variants play in the therapeutic response, safer and more effective individual dosing regimens can be tailored based on patient genetic profile, all in the hope of maximizing the efficacy while minimizing the side effects of treatment.[1,2] This becomes of particular importance in the pediatric population which is more vulnerable to the toxic effects of medications and is subject to age-dependent pharmacokinetics. Since the enzymatic activity of certain drug metabolizing enzymes can vary as a function of age and due to numerous developmental and maturation processes, the ontology of those enzymes and the influence of genetic variants on their activity are essential for the accurate prediction of treatment response in children.[1–4]

One of the biggest challenges to treatment in children is the first dose. Clinical essays are far more difficult to carry in children than in adults, and thus, this dose is usually extrapolated by applying scaling approaches to data acquired from dosing essays in adults. This becomes more complicated when considering the influence of inter-individual genetic variation, the maturation of organ systems, and the preferential clearance of certain drugs due to enzymatic expression switches that occur at multiple stages of development (ex. CYP3A7 to CYP3A4 in neonates).[2–4]

Recent studies demonstrated that genetically guided personalized therapy holds the promise to deliver a safer and more effective treatment in pediatric organ transplantation.[5,6] One well-described example is the dosing of Tacrolimus (TAC), an immunosuppressant used to reduce organ rejection both in pediatric and in adult patients. Its narrow therapeutic index combined with the significant variability in its pharmacokinetics (which can reach up to 10-fold in children) makes it difficult to set a standardized dosing regimen.[6] While under-dosing of TAC can lead to organ rejection, overdosing might lead to an increased risk of infection and other serious side effects. To help explaining some of this variability, pharmacogenomics investigated the impact of genetic variants on the activity of CYP3A5 enzyme – a major metabolizer of TAC. Most notably, in order to maintain a therapeutic blood concentration, patients who are homozygous or heterozygous for the wild-type allele CYP3A5*1 (i.e. CYP3A5 expressers) have 1.5–2 folds higher dose requirements compared to CYP3A5 non-expressers. CYP3A5 non-expressers are homozygous for the variant-type allele ‘CYP3A5*3’ and have a non-functional protein product.[5,6] That being said, it is important to note that therapeutic drug monitoring remains the standard practice – as demonstrated by many clinical essays – when adjusting the dose of TAC in clinical settings. Prospective trials that prove the clinical added value of genotype-guided therapy (e.g. starting dose selection) are still needed.[6]

Similar PG findings apply to the field of pain management. Codeine, for example, is a commonly used analgesic which is converted to its active metabolite, morphine, by the hepatic enzyme CYP2D6. Genetic variations in the CYP2D6 gene include point mutations, insertions, deletions, as well as copy number variants. Some of these variations can lead to protein under-expression or a complete lack of function (like in the example of CYP2D6*4 or CYP2D6*5 alleles), whereas other rearrangements can cause an over-expression like in the case of duplicated (CYP2D6*xN) alleles.[7,8] The presence of two non-functional alleles renders the individuals poor metabolizers of CYP2D6 substrates, while having two reduced function alleles or one non-functional and one reduced activity allele make their carriers intermediate metabolizers. Moreover, extensive metabolizers are defined as the ones carrying two wild-type normal-activity alleles or one of these alleles in combination with a reduced function or a non-functional allele. Finally, those who carry more than two CYP2D6 gene copies are referred to as ultra rapid.[7,8] The need for a better understanding of the codeine pathway found its roots in pharmacogenomics after fatal incidences in clinical practice in which children who were ultra CYP2D6 metabolizers (or their mothers in the case of breast-fed children) have been exposed to an excessive dose of morphine leading to subsequent severe respiratory depression.[7–9] Consequently, the Food and Drug Administration (FDA) communicated a black box warning against codeine use to manage postoperative pain in children and it recommends to use alternate analgesics for CYP2D6 ultra-rapid and poor metabolizers.[10]

In cancer therapy, patients routinely receive multi-drug combinations. Since the consequence of treatment failure is frequently life-threatening, the power to predict how a particular patient will respond to a specific treatment protocol is priceless and lies in the core interest of personalized oncology.[1] A lot of noteworthy examples of the usefulness of pharmacogenetics in pediatric oncology can be found in studies of the treatment outcome of acute lymphoblastic leukemia (ALL). Treatment-related toxicity can be life-threatening and is a main cause of treatment interruption or cessation in childhood cancer.[11] In addition, survivors of childhood ALL are at risk of higher treatment-associated morbidity and mortality with a cumulative incidence of chronic health conditions reaching 73.4%.[12] One effective drug used in chemotherapy is 6-mercaptopurine, which is mainly metabolized by the enzyme thiopurine S-methyltransferase (TPMT). Genetic variants in the TPMT gene coding for this enzyme are associated with inactive alleles and reduced enzymatic activity. Studies have shown that *2, *3A, and *3C are the most prevalent TPMT reduced-function alleles across different ethnicities with the exception of some sub-Saharan Africa populations in which a significant part of the inactivation is accounted for by the TPMT*8 allele.[13] Indeed, clinical studies have shown that patients who are homozygous for the active wild-type allele (i.e. TPMT*1) are less likely to experience myelosuppression than the carriers of TPMT-deficient alleles in whom (i.e. with two copies of variant allele) a starting-dose reduction of 10-fold is recommended.[13,14]

Another example of the childhood ALL pharmacogenomics, currently at the discovery level, is the allergic reaction in response to therapy with asparaginase.[15,16] Hypersensitivity to asparaginase can lead to dose reduction, suboptimal drug exposure, and subsequent treatment failure.[15,16] Pharmacogenomics studies identified genetic variants affecting the expression of the asparagine synthetase (ASNS) gene and variants in the genes that regulate the immune responses (i.e. human leukocyte antigen HLA-DRB1).[15,16] It can be anticipated that the combination of the two pathways can explain a larger portion of the risk for asparaginase allergy, but further studies are needed to confirm this hypothesis. Another well-studied example in the HLA region is carbamazepine-induced hypersensitivity reactions which are linked to the HLA-B*15:02 and HLA-A*31:01 haplotypes. The first is highly predictive of the risk of Stevens–Johnson syndrome and toxic epidermal necrolysis, and it is present in a relatively higher frequency in some Asian populations. The second haplotype was recently found to influence the risk of hypersensitivity reactions in other populations; it is of notable interest in Europeans given the rarity of the *15:02 allele in this ethnicity. Importantly, a prospective genotype-guided trial found that hypersensitivity reactions can be prevented by careful screening for the HLA-B*15:02 variant in Chinese patients.[8,17]

As a result of the growing body of evidences confirming the importance of genotyping for variants in the CYP2D6, TPMT and HLA genes as well as many others, the FDA and the Clinical Pharmacogenetics Implementation Consortium (CPIC) published guidelines to help clinicians understand the dimensions of the genetic component of treatment response variability and provide recommendations on how to use this information to guide the treatment.[1,6–8,13,14]

In conclusion, pharmacogenomics showed evidence-based benefits in multiple treatment areas including cardiology, neurology, oncology, pain management, organ transplantation, and immunosuppression. While genome-wide association and candidate-gene studies continue to identify significant associations between genetic variants and pharmacotherapy,[18] the clinical usefulness of these findings is still unclear. Translating these discoveries into clinical practice is not a simple process. In order to make use of such information, any identified association has to be reproducible in a validation cohort and the effect size must be convincing.[1,18] Adequately powered clinical trials that demonstrate the added benefit of pharmacogenomics-guided therapy compared to the current standard-of-care are strongly needed.

Expert opinion

While an important part of inter-individual variation in treatment response can be attributed to factors such as age, body mass index, comorbidities, and lifestyle, pharmacogenomics have the promise to enhance the efficacy and precision of existing drug dosing algorithms by factoring into the equation the effects inferred by genetic variants. This would empower clinicians to better calculate the overall risk/benefit ratio of a given treatment which remains a big challenge in the vulnerable pediatric population.

Despite the availability of genetic tests that can screen for a wide range of important genotypes, their cost combined with the lack of accessibility, lack of prospective studies, and the time needed for results hinder the widespread use of such tests among clinicians. As next-generation sequencing is advancing as a leading revolutionary tool in genetics, the ability to sequence the entire human genome holds the promise to improve the pharmacogenomics knowledge. The accumulative results banked in international databases will allow testing a large number of possible associations between variants, genes, diseases, and treatments. Indeed, this is the goal of many projects launched across the globe, like the Precision Medicine Initiative program funded by the National Institute of Health that is aimed toward improving the clinical knowledge of the risk inferred by the different factors of variability. Such studies will have the opportunity to simultaneously test for the interaction between a wide range of environmental-, genetic-, and patient-specific factors favoring the application of personalized medicine.

Finally, in order to improve treatment outcome and patient care, clinical implementation of pharmacogenomics discoveries in both pediatric and adult populations will require consensus guidelines powered by bioinformatics tools that can help physicians to interpret the genetic data and provide them with user-friendly software that can guide their decision-making process. Such guidelines are emerging through the drug agencies and implementation consortiums which provide useful recommendations on gene-based drug dosing for well-established associations.

Declaration of Interest

M Krajinovic presently holds the operating grant from Leukemia Lymphoma Society of Canada to study pharmacogenetics of childhood acute lymphoblastic leukemia. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.