Improving Dose Selection In Children through Better Estimates of Renal Function
CDER scientists are investigating how to improve pediatric dose selection by predicting drug clearance from renal function equations.
Challenges in Pediatric Dose Selection
How to select the right dose for pediatric patients is a fundamental challenge in pediatric drug development. In children, rapid changes in body size and organ function bring large inter- and intraindividual variabilities in the time course that drugs move through the body to elicit their pharmacodynamic effects. FDA is frequently consulted by sponsors about the acceptability of using renal function equations to predict exposures for renally eliminated drugs in the pediatric patient population; however, such predictions warrant careful evaluation. As revealed by the data in new drug applications, the commonly used renal function equations for children can markedly overestimate glomerular filtration rates and result in unrealistic predictions regarding pediatric dosing.
Renal Function Equations in Pediatric Dose Selection
For drugs eliminated mainly via the kidney, estimation of renal function will be used to determine the dose. Various equations have been developed to determine renal function and to interpret drug elimination profiles in pediatric patients. In drug development, an estimator of renal maturation can be incorporated into pharmacokinetic models to predict drug clearance and inform appropriate dose selection in pediatric patients. In clinical settings, individual dose adjustment can be guided through the use of estimated glomerular filtration rate (eGFR) equations as descriptors of renal function. The most commonly used eGFR equations are the “bedside” version of the Schwartz equation for children less than 12 years of age, and the Cockcroft-Gault equation for patients 12 years of age and older, providing creatinine clearance as a proxy for glomerular filtration rate. (See the FDA draft guidance for industry Pediatric Clinical Pharmacology Studies). Calculations based on eGFR equations may use one or more renal function biomarkers; demographic parameters such as body weight, height, age, and sex may also be incorporated.
To get an overview of the eGFR equations currently recommended in product labeling for pediatric patients, CDER investigators searched through FDA-approved drug labels and found that a given eGFR value can be used in multiple ways: 1) as a continuous variable to directly calculate the dose; 2) as a categorical variable to adjust the dose; or 3) as an indication of whether the drug should be avoided for a subgroup of pediatric patients. Among those drugs where an eGFR equation was specified in the label, a version of the Schwartz equation was given most often (Figure 1).
Figure 1. Use of eGFR equations to adjust dose in pediatric patients.
See text for details. (*Schwartz equation; #Cockcroft-Gault equation)
Renal Function Equations and Drug Clearance Predictability
To evaluate the potential of renal function equations to guide pediatric dose selection, CDER researchers sought to identify the major factors affecting the predictive accuracy of the Schwartz equation. For those drugs that are predominantly eliminated via the kidney, elimination from the circulatory system is expected to approximate renal clearance, thereby offering an opportunity to assess the accuracy of eGFR calculations. Following this logic, researchers analyzed clinical pharmacokinetic data of four renally eliminated drugs in pediatric patients and compared their observed values against eGFR calculations. For example, for gadobutrol, an intravenously administered imaging-contrast agent, 99% of elimination from the body occurs via glomerular filtration, thereby allowing one to determine its rate of clearance directly from clinically observed pharmacokinetic data. Figure 2 compared the clinically observed value against eGFR values calculated from a variety of equation versions. For children less than 12 years old, five out of seven equation versions, including the original and bedside Schwartz equations, in fact overpredicted the observed clearance rate for gadobutrol.
In addition to gadobutrol, analyses of three other drugs were performed using the bedside Schwartz equation. The three additional drugs—namely, gadoterate, amikacin, and vancomycin—were selected because their elimination ensues primarily (more than 90%) by glomerular filtration, with minimal tubular secretion and reabsorption. For about 90% of clinically observed subjects, the ratio of the eGFR value to the pharmcacokinetically determined value of clearance (eGFR/CL) was greater than 1, and an additional 10% had a ratio higher than 2. The overestimation of eGFR was even greater when the original (rather than bedside) Schwartz equation version was used. The results thus support the notion that eGFR calculations, regardless of drug type, overestimate drug clearance.
Figure 2. Forest plot showing the ratio of eGFR and observed drug clearance (CL) values for gadobutrol. The results are presented as mean ratios (black circles) for the given age groups with the 95% confidence interval indicated. The age groups shown on the vertical axis correspond to those pertaining to the original Schwartz equation.
eGFR Values as Overestimates of Drug Clearance Rates
The researchers noticed that eGFR values for some subjects exceeded the upper bound of the normal range (“supraphysiological eGFR”), with observed serum creatinine values less than 0.2 mg/dL. In fact, nearly all subjects with serum creatinine less than 0.2 mg/dL manifested supraphysiological eGFR values. To explain this correlation, the researchers hypothesized that the limit of reliable creatinine detection for the assay may occur around concentrations of 0.2 mg/dL. Thus, measurement errors may well contribute to the high rate of supraphysiological eGFR values calculated for subjects with serum creatinine less than 0.2 mg/dL. Even when subjects with supraphysiological eGFR values were excluded from analyses, however, the overall trend of mean eGFR/CL ratios exceeding unity persisted.
The investigators further reasoned that alterations to coefficient (k) values in the Schwartz equation might improve eGFR-based predictions of drug clearance. To address this possibility, they combined data from the four test drugs and established an integrated population pharmacokinetic model. Values of eGFR recalculated with the combined (integrated population) model-derived k values, which were all lower than the coefficient (0.413) used for the bedside Schwartz equation, agreed more closely to observed drug clearance rates.
Renal function models and equations are valuable tools for predicting drug pharmacokinetics, clearance, and dose adjustments in pediatric patients. Although the clearance of renally eliminated drugs might be overestimated when using creatinine-based renal function equations—especially for pediatric patients with low levels of serum creatinine—well-reasoned adjustments to methods of calculating eGFR values may improve the accuracy of predicted parameters. Equations based on cystatin C levels or a combination of multiple renal biomarkers might be productive in this regard. The importance of reliably selecting the proper drug dose in pediatric populations is an important element of the CDER mission, and FDA is working with industry and clinicians to determine the best way to apply modeling information to the care of children.
How does the research improve pediatric drug development and patient care?
The findings of this study will inform efforts to improve pediatric dose selection strategies for drug developers as well as pediatricians who must pay particular attention to renal function. The research project has also had a significant impact on pediatric drug regulatory review by facilitating consensus on best practices for assessing pediatric renal function and dose selection in clinical trials and product labeling.
 Zhang, Y., Sherwin, C.M., Gonzalez, D., Zhang, Q., Khurana, M., Fisher, J., Burckart, G.J., Wang, Y., Yao, L.P., Ganley, C.J. and Wang, J. Creatinine‐Based Renal Function Assessment in Pediatric Drug Development: An Analysis Using Clinical Data for Renally Eliminated Drugs. Clin. Pharmacol. Ther. 2020 doi:10.1002/cpt.1991
 Wang J, Kumar SS, Sherwin CM, Ward R, Baer G, Burckart GJ, Wang Y, Yao LP. Renal Clearance in Newborns and Infants: Predictive Performance of Population-Based Modeling for Drug Development. Clin Pharmacol Ther. 2019 Jun;105(6):1462-1470. doi: 10.1002/cpt.1332