Children are not small adults: Specific findings in statin exposure and response in a growing population

With enhanced recognition that subclinical coronary artery disease (CAD) in childhood predicts CAD-associated morbidity and mortality in later life, an increasingly large number of children in the US qualify for interventions aimed at reducing hypercholesterolemia, a significant risk factor for CAD. As with adults, 3-Hydroxy-3-methyl-glutaryl Coenzyme A (HMG-CoA) reductase inhibitors (statins) represent the most common pharmacologic agents for hypercholesterolemia treatment when behavioral lifestyle modifications alone do not succeed. Given the known challenges with adherence to dietary and behavioral modifications as interventions for hypercholesterolemia, an increase of statin use in children and adolescents is inevitable. With the application of recent pediatric lipid screening and management guidelines, it is estimated that over 600,000 12-year olds to 21-year olds would now be eligible for statin therapy in the U.S. Statins are FDA approved for children ≥8 years (pravastatin) and ≥10 years (atorvastatin, fluvastatin, lovastatin, rosuvastatin, simvastatin) with familial hypercholesterolemia. Despite their success in low-density lipoprotein cholesterol (LDL-C) reduction on a population level, considerable inter-individual variability in LDL-C response to statins is observed amongst children. Genetic sources of variability contributing to risk of treatment failure and adverse events (e.g. HMGCR, SLCO1B1 genotypes) have been confirmed in adult populations. However, these same pharmacogenomic variables are not uniformly applicable to the growing child, illustrating yet again that children are not simply small adults.

Pharmacogenomics has become integrated into drug optimization programs aimed at maximizing drug efficacy while simultaneously delivering the lowest possible risk for toxicity. Pediatricians are uniquely primed and enthusiastic, given their clinical familiarity with genomic medicine, to incorporate pharmacogenomics to enhance clinical practice. However, extrapolation of the genotype-phenotype relationship observed in adults to inform the pharmacologic treatment guidelines in a growing child may not be appropriate. Ontogeny is just as important as genetic variation, as evidenced by failure of extrapolation from adults to children. This may be due to changing patterns of gene expression or in protein function over the course of development. Additionally, pediatric providers must consider the unique factors in children, such as physiological changes that occur during development (e.g. protein binding affinity, gastric transit time, drug target maturity) and underlying disease state (e.g. congenital heart disease, obesity), that can influence drug disposition and response.

Genotype-stratified pharmacokinetic studies performed in pediatric populations (e.g. atomoxetine, proton pump inhibitors, statins) have consistently demonstrated highly variable ranges in systemic exposures and in selected scenarios, were more variable compared to adults. The impact of genetic and non-genetic factors on drug exposure in children are often different than those in adults, meriting caution when using adult data to inform dosing in the developing child. Herein, we review 3 challenges occurring to implementing precision based therapy, including pharmacogenomics, for growing children, utilizing statins as an example.

Challenge #1: Pharmacogenomic variants have pediatric-specific magnitudes of effect

The most investigated source of variability in statin disposition relates to underlying genetic differences in the protein transporters (e.g. SLCO1B1) responsible for movement of statins across cell membranes. In adults, the SLCO1B1 c.521T>C genotype is associated with reduced hepatocellular statin uptake at the site of action, potentially leading to altered cholesterol biosynthesis inhibition, and increased systemic exposure, resulting in undesirable statin-association muscle symptoms.^1,2 The effect of SLCO1B1 c.521T>C variation on simvastatin acid (SVA) systemic exposure in adults was recently confirmed in a pediatric cohort (mean 14 yrs, 8–20 yrs) with each copy of the variant c.521C allele contributing to a 2.5-fold increase in SVA systemic exposure.³ However, the genotype effect on SVA systemic exposure in children being nearly 2-fold greater than that observed in adults was not anticipated (Table 1).^1,3 However, this magnitude of effect was not observed in the same study cohort dosed pravastatin.(Table 1).⁴ For both aforementioned statins, the effects of the c.521CC genotype was greater in children compared to adults (Table 1).^1–4 Collectively, one could conclude that SLCO1B1 c.521T>C genotype influences the systemic exposure of statins in children and thus appears to be more influential in the growing child compared to adults.

Table 1.

Comparison of SLCO1B1 Genotype Effect on Systemic Exposure in Children and Adults.

	AUC (ng/mL*h)	Fold change in mean AUC vs. c.521TT
Simvastatin Acid  Children (n=32)#3   521TT (n=15)   521TC (n=15)   521CC (n=2)  Adult (n=32)#1   521TT (n=15)   521TC (n=15)   521CC (n=2)	1.9 (± 1.8) 4.5 (± 2.5) 11.6 (± 7.6) 13.9 (± 5.3) 17.4 (± 8.0) 44.5 (± 7.4)	N/A 2.4 6.3 N/A 1.3 3.2
Pravastatin Acid  Children (n=32)*4   521TT (n=15)   521TC (n=15)   521CC (n=2)  Adult (n=41)*2   521TT (n=15)   521TC (n=15)   521CC (n=2)	133.1 (± 113.4) 225.4 (± 127.4) 293.8 (± 16.1) 89.5 (± 64.0) 184.7 (± 105.0) 140.1 (± 39.3)	N/A 1.7 2.2 N/A 2.1 1.6

^*Data shown represents mean (±SD) AUC and fold change in mean AUC amongst c.521TC and c.521CC genotype groups compared to those with the c.521TT reference genotype from (3) Wagner JB et al. J Clin Pharm 2018. Jun;58(6):823–833, (1) Pasanen MK et al. Pharmacogenet Genomics. 2006;16(12):873–879, (4) Wagner JB et al. Clin Pharmacol Ther 2018 Dec 14th epub ahead of print and (2) Niemi M et al. Pharmacogenetics. 2004;14:429–40. AUC-area under the curve;

^#denotes AUC_0-n

^*denotes AUC_0−∞

The additional impact of growth and development within the pediatric age group on this discordant genotype-phenotype relationship remains unknown. Prasad et al have previously described age-dependent levels of SLCO1B1, SLCO1B3, ABCB1 transporter expression through proteomics analysis, implying that the genotype-phenotype relationship could change during the growth and development in childhood.⁵ However, this study was not specifically designed to elicit whether the SLCO1B1 genotype impacts protein expression levels in the growing child to a greater extent than in adults. Our own work, however, did not reveal a correlation between SVA or PVA systemic exposure and age or Tanner developmental stage, although these studies were not powered to elicit this analysis.^3,4 Regardless, the influence of development on statin transport requires further elucidation in larger demographically and developmentally diverse pediatric cohorts.

Current Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines from 2014 stated there remained no data available regarding SLCO1B1 genotype effect on SVA response or myopathy in the pediatric population or data to suggest that c.521T>C genotype affects SVA disposition differently in the growing child compared to adults.⁶ Given the aforementioned discordant data in children with regards to SLCO1B1 genotype and simvastatin exposure, SLCO1B1 testing should be considered for pediatric providers initiating simvastatin and be cautious with simvastatin initiation in patients with ≥1 c.521C allele.

Challenge #2: Pharmacogenomic variants have pediatric-specific (and ancestry-specific) directions of effect

In adults, variation in the genes encoding the drug target, HMG-CoA reductase (HMGCR), is associated with reduced baseline total cholesterol (TC) and LDL-C, but attenuated LDL-C response to pravastatin and simvastatin.⁷ A well described sequence variation in HMGCR, rs3846662, impacts the ratio of alternatively spliced mRNA transcripts, HMGCR13(−), to canonical transcripts, HMGCR13(+), with a larger ratio of HMGCR13(−)/HMGCR13(+) associated with the aforementioned reduction in baseline LDL-C and LDL-C response to statins. Replication in pediatric liver tissue samples (n=62) revealed a trend towards increased HMGCR13(−)/HMGCR13(+) ratios in the presence of the “A” allele, an observation in concordance with adult data.⁸ Pediatric patients (n=195) with the presence of the “A” allele had lower LDL-C, but the difference was not statistically significant (Mean (95%CI): G/G: 166 (156,177); G/A: 160 (150,171); A/A: 151 (143, 160); p=0.19). Interestingly, in African-Americans (n=27) with G/A genotype TC and LDL-C levels were significantly higher compared to those with G/G genotype (Mean TC (95%CI): G/G: 223 (204,245); 283 (226,354); p=0.01: Mean LDL-C (95%CI: G/G 156 (138, 176); G/A 210 (145, 304); p=0.03), which is discordant to the relationship observed in African-American adults where carriers of “A” allele had a significantly lower baseline LDL-C (p=0.0006) and blunted response to statins (p=0.0008). The differential effect observed amongst individuals of different ethnic backgrounds highlights the importance of incorporation of additional non-genetic factors to individualizing pharmacotherapy for the developing child.

Additionally, the heterogeneous nuclear ribonucleoprotein A1 (HNRNPA1) splicing factor influences alternative HMGCR transcripts.⁷ The rs1920045 variant leads to enhanced expression of an alternative transcript, HNRNPA18(+), that promotes HMGCR13(−) formation in adults. As expected, HNRNPA18(+) is associated with attenuated response to statin in adults. However, it remains unknown if rs1920045 contributes to the considerable inter-individual variability in pediatric statin response. Analysis of the aforementioned pediatric liver tissues revealed a trend towards less HMGCR13(−) in subjects with HNRNPA1 rs19020045 “T”, which is discordant with adult data where the presence of the “T” allele leads to increased HNRNPA18(+) and subsequently more HMGCR13(−).⁸

Collectively, HMGCR and HNRNPA1 genetic variation, in conjunction with development and ethnicity, may contribute to statin response during childhood, but may be discordant compared to the adult experience reinforcing that pediatric-specific guidelines are warranted.

Challenge #3: Beyond Pharmacogenomics: Analysis of children reveals pediatric-specific novel variables

Equally remarkable to the differences elicited between children and adults in SVA systemic exposure amongst SLCO1B1 genotype groups, was a finding of “negligible” SVA systemic exposure in 25% of the pediatric cohort.³ In fact, the unanticipated observation of reduced SVA exposure and delayed peaking of SVA in children with “negligible” exposure, confirms the importance of independent pediatric pharmacokinetic studies prior to extrapolating adult data to inform pediatric precision-based treatment guidelines. It is unknown whether the observation of “negligible” exposure in the pediatric study was secondary to insufficient hydrolysis of the prodrug (simvastatin lactone) to the active moiety, SVA, or enhanced hepatic uptake (e.g. SLCO1B1 *14). If “negligible” exposure is confirmed to be secondary to insufficient enzymatic hydrolysis, it will represent an additional variable contributing to SVA exposure and drug response that must be accounted for in precision-based statin algorithms for children and adults.

For pravastatin, systemic exposure was unexpectedly higher in those children within the SLCO1B1 c.521TT genotype with BMI Z-score >+2.5 and comparable to the exposure noted in the c.521TC genotype.⁴ Obesity is highly associated with the risk of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis where it is postulated that loss of glycosylation of SLCO1B1, secondary to liver adiposity, contributes to altered cellular trafficking and membrane expression. This could theoretically lead to impaired SLCO1B1-mediated transport, regardless of SLCO1B1 genotype, and ultimately, higher peripheral tissue exposure in a developing obese child prescribed pravastatin.

Overall, these examples demonstrate that pharmacogenomics in conjunction with non-genetic factors (e.g. ethnicity, development stage, liver adiposity) must be considered when attempting to optimize pharmacotherapy for the developing child.

Conclusion

Collectively, data from statin disposition and response provide another example how children are not small adults, and illustrate the need for pediatric-specific data for pharmacogenomics associations. Multiple factors must be considered when tailoring statin treatment to patients, but in children, those variables may have differential effects compared to adults. Studies designed to control “exposure” and determine optimal “exposure” in children may mitigate the inter-individual variability in response observed in previous pediatric trials. However, one must be cognizant that mere extrapolation of adult data could inaccurately estimate the systemic exposure and response a child may have. Incorporation of ontogeny, disease state, physiological changes that are unique to children into study design will allow providers the means to individualize treatment in the growing child.