Pediatric warfarin dosage is currently based primarily upon clinical judgment and INR monitoring. While adult warfarin dosage has potentially benefitted from the use of pharmacogenetics based on information from CYP2C9 and VKORC1 in some centers, this information has not been applied to pediatric warfarin dosage previously. This study involved leveraging prior information in the form of adult warfarin data, application of knowledge of warfarin PK and PD, developmental pharmacology, and preliminary pediatric warfarin pharmacogenetic information. Pharmacometric methods were used to bridge an adult model and dosage regimen to develop a pediatric warfarin model and propose a dosage regimen. The two primary contributions of the current study are the proposal of a science-based, reproducible pediatric warfarin dosage regimen and a tool that can be used by clinicians and researchers to arrive at a pediatric warfarin dosage regimen based on a target INR.
Based on the developed pediatric warfarin PK/ PD model, the CYP2C9 and VKORC1 genotypes should have a significant effect on warfarin clearance and INR response and on the required doses to achieve a target INR. In the simulations, using a fixed dose and assuming that each patient belongs to the same genotype category (CYP2C9*1*1 and VKORC1 rs9923231 GA) resulted in adverse INR outcomes, particularly for the most variant genotypes (CYP2C9 *2*3 and VKORC1 rs9923231 AA). Between the CYP2C9 (PK) and VKORC1 (PD) genetic effects, the polymorphisms altering PK had the most significant impact on dosage. There are two explanations for this observation. First, the dual CYP2C9 *2*2, *2*3, and *3*3 variant genotypes had a large magnitude of effect (−70% to −85% on clearance) relative to that of the VKORC1 homozygous rs9923231 AA variant genotype (−35% on potency). Second, the warfarin dose was titrated by monitoring the INR and PD response not the drug concentrations. Hence, adjusting starting doses based on genotypes relevant for PK may be more critical. Patients with the *2*2,*2*3, and*3*3 genotypes have a prolonged warfarin half-life (3–6 times longer than that of *1*1 patients), and the starting dose must account for this effect.
While supratherapeutic INRs remained well below 10% initially, they tended to increase over time, especially in patients with variant genetic polymorphisms. Despite generally conservative dosage, the lack of genotyping for CYP2C9
results in a suboptimal starting dose and/or lack of a rational warfarin dose titration scheme. In the CHLA study, a target INR range of 2.5 to 3.5 was used for patients with valve replacements, consistent with previous reports in the literature. However, the choice of a target INR range of 1.5 to 2.5 or 2.0 to 2.5 for patients undergoing the Fontan procedure or for whom Kawasaki disease was diagnosed is institution-specific and different from previous reports.21–23
For the proposed pediatric warfarin dosage based on pharmacometric modeling, the commonly accepted and used target INR range of 2.0 to 3.0 was selected to maintain consistency with the literature and generalize the outcomes to most settings. However, the modeling and simulation tool developed herein could be used to derive rational pediatric warfarin dosage for optimizing clinical outcomes for any alternative target INR range.
There are limitations to the current research. The most significant limitation may be attributed to the paucity of available clinical data on warfarin use and pharmacogenetics in pediatric patients. A multicenter pediatric warfarin pharmacogenetics trial sponsored by the FDA is now underway. The current pilot study was restricted to a small sample of pediatric subjects (n=26), which was used to validate our model. Additionally, the study population was relatively homogenous and did not represent a wide variety of ethnic groups. Next, Fontan patients were included, and this population may have different warfarin dosage requirements than other populations. The influence of interacting drugs was also not accounted for in the pediatric study population. The model was based on adult data and physiological principles rather than pediatric data, and the proposed dosage was based on simulations. Another limitation of this study is the absence of prospective validation of the dosage regimen. However, despite these limitations, the current study makes efficient use of available information and provides an important first step towards improving pediatric warfarin dosage.
This research is based on certain assumptions. The first assumption is that the concentration– response relationship for warfarin is similar between pediatric and adult patients. Some researchers have suggested intrinsic developmental differences in the coagulation systems,24
precluding extrapolation of dose–response for antithrombotic therapy from adults to the youngest subset of the pediatric population (<2 years old). The second assumption is that CYP2C9
polymorphisms reduce warfarin clearance to the same extent in pediatric patients as they do in adult patients. However, data regarding how ontogeny affects warfarin PK and PD is not available to formally challenge these assumptions. This is particularly true for VKORC1
where the patterns of developmental expression are not yet understood.
The goal of this work is to eventually establish an improved standard of care for pediatric patients who require warfarin therapy. A genetics-based warfarin dose nomogram that functions more efficiently than the conventional standard of care would represent a major advance in pediatric pharmacotherapy. Such a nomogram could be made widely available to all clinicians and may enhance the safety and effectiveness of warfarin therapy in pediatric patients.