Creatinine-Based Renal Function Assessment in Pediatric Drug Development: An Analysis Using Clinical Data for Renally Eliminated Drugs

Abstract

The estimated glomerular filtration rate (eGFR) equations based on serum creatinine (SCR) have been used for pediatric dose adjustment in drug labeling. This study evaluated the performance of those equations in estimating individual clearance of drugs that are predominantly eliminated by glomerular filtration, using clinical data from the renally-eliminated drugs gadobutrol, gadoterate, amikacin, and vancomycin. The eGFR was compared with the observed drug clearance (CL) in 352 pediatric patients from birth to 12 years of age. Multiple eGFR equations overestimated the drug clearance on average, including the original and bedside Schwartz equations, which showed an average eGFR/CL ratio between 1 and 3. Further analysis with bedside Schwartz equation showed a higher eGFR/CL ratio in the subjects with a lower SCR or CL. Supraphysiological eGFR as high as 380 mL/min/1.73m² was obtained using bedside Schwartz equation for some of the subjects, most of whom are children less than two years of age with SCR < 0.2 mg/dL. Excluding the subjects with supraphysiological eGFR from the analysis did not change the overall trend of overestimation. In conclusion, Schwartz equations led to an overestimation of drug clearance for the drugs evaluated. When greater precision is required in predicting eGFR for pediatric patients, such as in drug dosing, revised k constants for the Schwartz equation or new methods of GFR estimation may be necessary.

Introduction

Dose selection is challenging for pediatric patients due to the dynamic changes in drug pharmacokinetics and pharmacodynamics related to physiological growth, maturation, and organ function. In pediatric drug development, an initial dose needs to be proposed before starting a clinical trial. Drug clearance (CL) is a critical parameter for selecting an age-appropriate dose and is often predicted by developmental changes in renal and hepatic function.

Various equations have been developed to obtain the estimated glomerular filtration rate (eGFR) as a measure of renal function based on serum creatinine (SCR), cystatin C, and other biomarkers. Those equations are potential predictors of renal drug clearance, which could inform dose selection of drugs that are predominantly renally eliminated. Based on our survey of U.S Food and Drug Administration (FDA)-approved drug labeling,¹ the Schwartz equations have been used for guiding dose adjustment for pediatric patients with renal impairment (Table 1). These equations have been applied either as a continuous variable to calculate dosage directly, or as a categorical variable to adjust the dose for pediatric patients (Table 1). Additionally, the predictive performance of these formulas on CL in pediatrics is unknown, especially in the first 2 years of life when renal function may not be fully mature.

Table 1.

Use of the eGFR equations to adjust dose in pediatric patients based on product labeling

Drug	Pediatric Use in Drug Labels
Digoxin*	Maintenance dose is calculated using CrCl
Valganciclovir*	Dose (mg) = 7 × BSA × CrCl, with the CrCl capped at 150 mL/min/1.73 m2, and a maximum dose of 900 mg
Ceftazidime/ Avibactam*	Adjust dose if CrCl ≤ 50 mL/min/1.73 m2
Cefdinir*	Reduce dose if CrCl < 30 mL/min/1.73 m2
Peramivir#	Reduce dose if CrCl < 50 mL/min
Lacosamide*	Reduce dose if CrCl < 30 mL/min/1.73 m2
Deferasirox*	Contraindicated if eGFR < 40 mL/min/1.73 m2. Reduce dose if eGFR is 40 – 60 mL/min/1.73 m2
Teduglutide	Reduce dose if eGFR < 60 mL/min/1.73 m2
Ceftaroline*	Information insufficient for pediatric dosing if CrCL < 50 mL/min/1.73 m2
Lisinopril, Benazepril, Losartan	Not recommended in patients < 6 years or with GFR < 30 mL/min/1.73 m2
Candesartan, Enalapril	Not recommended in pediatric patients with GFR < 30 mL/min/1.73 m2

The terms eGFR and CrCl are used interchangeably to be consistent with the language in drug labels, and they both represent the values calculated from creatinine-based renal function equations.

eGFR: estimated glomerular filtration rate; CrCl: creatinine clearance; BSA: body surface area.

^*Schwartz equation or

^#Cockcroft-Gault equation is recommended for pediatric patients in labeling.

The objective of this study was to evaluate the performance of creatinine-based renal function equations for estimating individual CL of renally eliminated drugs in pediatric patients. Clinical pharmacokinetic data of four drugs in pediatric patients were pooled to compare the observed CL with individual eGFR obtained using creatinine-based equations. Gadobutrol and gadoterate are imaging agents that are >99% eliminated by glomerular filtration, and thus are ideal model drugs for comparison of eGFR with drug clearance. Amikacin and vancomycin are antibiotics predominantly eliminated by glomerular filtration (>90%).

Methods

Clinical Data

Gadobutrol, gadoterate, amikacin, and vancomycin were selected as model drugs since they are predominantly (>90%) eliminated by renal glomerular filtration with minimal tubular secretion/reabsorption. All four drugs were administered intravenously as previously described.^{2, 3} The PK data of gadobutrol, gadoterate, amikacin and vancomycin were measured in the time frames of 2007 – 2013, 2015 – 2015, 2001 – 2016, and 2006 – 2016, respectively. Data of gadobutrol⁴ and gadoterate⁵ were collected from their clinical pharmacology reviews, in which SCR was analyzed using IDMS-traceable Jaffe and enzymatic methods for pediatric patients without renal impairment. For amikacin and vancomycin, the demographics, lab values, concomitant medications, and clinical PK data were extracted using SQL from the enterprise data warehouse of the Intermountain Healthcare system,⁶ and were subsequently checked manually for correctness and completeness. Data were excluded from the analysis if any of the following information was missing: (a) date and times of drug administration or drug concentrations; (b) relevant covariates including age, body weight, height, and SCR. SCR was quantified at IH Laboratories and ARUP Laboratories using IDMS Traceable Vitros CREA Slides and the Vitros 5.1 FS Chemistry System analyzer (Ortho Clinical Diagnostics, Inc, Rochester, New York) across all hospitals. The patient characteristics of each data set are summarized in Table 2. Both preterm and term infants are included in the study, with the lowest gestational age at 23 weeks and 22 weeks for amikacin and vancomycin, respectively.

Table 2.

Drug elimination pathway and patient characteristics (N = 352)^a

CL, drug clearance; eGFR, estimated glomerular filtration rate; PK, pharmacokinetic; SCR, serum creatinine.

^aIn the group of 0 to < 1 year of age, all the data of gadobutrol and gadoterate are collected from full-term infants (gestational age ≥37 weeks). The subjects for amikacin and vancomycin each include 9 preterm infants (gestational age < 37 weeks). Values are shown as mean ± SD.

^beGFR was calculated using the bedside Schwartz equation.

Calculation and Comparison of eGFR and Drug CL

The eGFR was calculated using the equations as listed in Supplementary Table 1. Body surface area (BSA) was calculated using the Du Bois & Du Bois formula: BSA = 0.007184 × Height^0.725 × (Body Weight)^0.425. The 5% – 95% range of age-appropriate normal GFR was calculated as mean ± 1.96 × SD based on published data.⁷ The individual drug CL values were determined with the observed PK data of renally-eliminated drugs using a standard population PK analysis approach.⁶ For the equations that calculate eGFR in the unit of mL/min/1.73m², the obtained eGFR values were divided by 1.73 and multiplied by individual BSA, to be consistent with the unit of CL which is mL/min. The eGFR/CL ratio was subsequently calculated.

Data analysis was performed using NONMEM® version 7.3 (ICON Development Solutions, Ellicott City, MD, USA) and R version 3.5.1 (https://cran.r-project.org/).

Nonlinear Mixed Effects Modeling

Based on observed individual PK data for each age group, non-linear mixed effects modeling was used to estimate the k values following the same covariate model as the Schwartz equation. The analysis was conducted using NONMEM® version 7.3 (Icon Development Solutions, Ellicott City, Maryland, USA) and R version 3.5.1 (https://cran.r-project.org/). Data of the four drugs were combined to establish an integrated two-compartment model with first-order elimination kinetics. The CL was modeled with the following equation:

$$\text{CL}=k\times \text{Height}/\text{SCR}\times \text{BSA}/1.73$$

Where k values were estimated for each subgroup of patients as listed in Supplementary Figure 4. The drug-specific parameters were estimated individually for each drug. Interindividual variability was described by an exponential error model. Residual variability was described by a proportional error model (gadobutrol and gadoterate), or a combined proportional plus additive error model (amikacin and vancomycin). The first-order conditional estimation method with interaction was used for parameter estimation. Standard diagnostic plots were used to assess the final population model (Supplementary Figure 1), including the observed values vs. individual predictions, observed values vs. population predictions, conditional weighted residuals vs. population predictions, and conditional weighted residuals vs. time. The parameter estimates are shown in Supplementary Table 2.

Results

PK data were collected for four renally-eliminated drugs in pediatric patients from birth to 12 years of age. Gadobutrol, gadoterate, amikacin, and vancomycin were chosen as candidate drugs because they are predominantly eliminated by glomerular filtration.⁶ Baseline SCR, eGFR, observed CL, and other demographic variables stratified by age groups for each drug are summarized in Table 2.

The creatinine-based eGFR equations are listed in Supplementary Table 1. With eGFR obtained using those creatinine-based equations, Figure 1 shows the ratio of eGFR to observed CL for gadobutrol, a drug that is exclusively renally eliminated through glomerular filtration (>99%). For children 0–12 years of age, five out of the seven equations showed a mean ratio of eGFR/CL greater than 1.0 regardless of age subgroup, including the Cockcroft-Gault equation, Leger equation, Counahan-Barratt equation, and the original and bedside Schwartz equations (Figure 1). Flanders Metadata equation and Lund-Malmo equation led to eGFR/CL mean ratios lower than 1 for infants less than one-year-old, and greater than 1 for children at 1–12 years.

Figure 1. Forest plot showing the ratio between the equation-derived estimated glomerular filtration rate (eGFR) and the observed drug clearance (CL) of gadobutrol.

The results are presented as mean ratios (black dots) in each age group with the 95% confidence interval (horizontal lines). The y-axis represents the age groups as in the original Schwartz equation.

The individual eGFR values obtained using the bedside Schwartz equation were compared to the reference range of normal GFR.⁷ Of the total subjects, 10 % (34 out of 352) had a “supraphysiological eGFR”, which is defined in the context as an estimated GFR that exceeded the 95^th percentile of the normal GFR range,⁷ with a maximum value of 380 mL/min/1.73m² (Table 3). Notably, all these subjects whose eGFR exceeded the 95^th percentile of the normal range had a low SCR concentration (Figure 2A), most of which were lower than 0.2 mg/dL (Supplementary Figure 2). Given that the drugs are predominantly (>90%) eliminated by glomerular filtration, it would be reasonable to assume that CL and eGFR are proportional. However, although the eGFR values were dramatically higher than the population mean (183.8 vs. 108.0 mL/min/1.73m²; Supplementary Figure 3C), their CL was close to the population average (absolute CL: 32.4 vs. 28.8 mL/min; BSA-adjusted CL: 97.9 vs. 77.2 mL/min/1.73m²; Figure 2B). The discrepancy between eGFR and CL, which was associated with low SCR values, suggests that the high eGFR values may not be reliable for individuals exceeding the 95^th percentile of the normal range.

Table 3.

Accuracy of bedside Schwartz equation and the original Schwartz equation in estimating drug CL

		Bedside Schwartz equation		Original Schwartz equation
Total subjects included in analysis		n=352	n=318*	n=352	n=209*
Subjects with eGFR/CL ratio > 1		316 (90%)	282 (89%)	334 (95%)	191 (91%)
Subjects with eGFR/CL ratio > 2		36 (10%)	27 (8%)	142 (40%)	47 (22%)
eGFR/CL ratio (mean ± SD)	gadobutrol	1.57 ± 0.74	1.54 ± 0.78	2.15 ± 1.01	1.95 ± 1.27
	gadoterate	1.29 ± 0.42	1.23 ± 0.39	1.62 ± 0.59	1.43 ± 0.50
	amikacin	1.65 ± 1.00	1.62 ± 0.98	2.12 ± 1.11	2.02 ± 1.09
	vancomycin	1.50 ± 0.57	1.41 ± 0.38	1.95 ± 0.88	1.61 ± 0.52
Subjects with supraphysiological eGFR		34 (10%)	0	143 (41%)	0
Maximum eGFR (mL/min/1.73m2)		380	176	506	185

CL, drug clearance; eGFR, estimated glomerular filtration rate.

^*The subjects whose eGFR values are higher than 95th percentile of age-appropriate glomerular filtration rate (mean ± 1.96 × SD) were excluded from this analysis.⁷

Figure 2. Ratio of the estimated glomerular filtration rate (eGFR) to the observed drug clearance (CL) versus serum creatinine (SCR, A), CL (B), age (C), and height (D).

The eGFR values were calculated using bedside Schwartz equation (k = 0.413), and the observed CL values were obtained from a population pharmacokinetic analysis. ■ = amikacin, ● = gadobutrol, ▲ = gadoterate, ▼ = vancomycin. The red symbols represent the subjects whose eGFR exceeded the 95th percentile of normal GFR.⁷ The red solid line shows the cutoff of eGFR/CL ratio at 1, and the black dashed lines represents the boundaries with eGFR/CL ratio of 0.5 and 2, respectively. The loess smooth curves for amikacin (blue), gadubutrol (green), gadoterate (orange), and vancomycin (purple) are also shown. Log scale was used for the x-axis of panel A and y-axis of all the panels.

The eGFR/CL ratio was calculated for each individual, and the result was summarized, either including or excluding the subjects with supraphysiological eGFR values (Table 3). The eGFR calculated using bedside Schwartz equation overestimated the gadobutrol CL by approximately 1.6-fold, and a similar trend was consistently observed with gadoterate, amikacin, and vancomycin, no matter whether the supraphysiological eGFR values were excluded or not (Table 3). With the bedside Schwartz equation, 316 out of 352 total subjects (90%) had an eGFR/CL ratio > 1, with 36 subjects (10%) showing a ratio > 2. The same analyses were repeated using the original Schwartz equation, where the overestimation was more pronounced for each drug compared with the result using the bedside Schwartz equation (Table 3).

A higher eGFR/CL ratio was observed in the subjects with a lower SCR or CL (Figure 2A–B). All individuals with an SCR lower than 0.2 mg/dL had an eGFR/CL ratio higher than 1.5 (Figure 2A). Furthermore, both the magnitude and variability of the ratio were greater for those individuals with a CL lower than 30 mL/min (Figure 2B). No significant trend was observed with a different drug, age, height, body weight, body surface area (BSA), or eGFR (Figure 2C–D, and Supplementary Figure 3).

Nonlinear mixed effects modeling was used to evaluate the relationship between CL and the covariates height and SCR, i.e., to calculate k values in the format of the Schwartz equation where CL= k × height/SCR, based on observed individual PK data. The derived k values in the CL equation were 0.244 for low birth weight (LBW) infants, 0.277 for non-LBW infants, and 0.296 for children 1–12 years of age (Supplementary Figure 4). These calculated k values are markedly lower than the k values in original and bedside Schwartz equations. Using those revised k values to calculate individual eGFR, the obtained eGFR/CL ratios became evenly distributed around 1 (Supplementary Figure 5 and Supplementary Table 3), indicating improved agreement between individual drug clearance and eGFR.

Discussion

Stringent estimation of drug clearance is critical for dose-selection in both drug development and clinical management, especially for drugs with a narrow therapeutic index. This study assessed the utility of SCR-based prediction equations in estimating GFR and how the resultant eGFR values compared to observed CL in the same patients, with four specific drugs known to be predominantly renally eliminated. Of note, the Leger equation and the Cockcroft-Gault equation led to eGFR/CL mean ratios higher than 3 for children less than one-year-old. Those equations calculate eGFR in the unit of mL/min without taking body surface area into account, which may be inaccurate for estimating GFR in growing children.⁸ The original Schwartz equation, bedside Schwartz equation, and Counahan-Barratt equation, which share a similar formula with different coefficients, overpredicted CL to a lesser extent. Our further analysis was focused on the bedside Schwartz equation, the most widely used eGFR equation by clinicians in pediatric patients.

According to FDA drug labels, the eGFR calculated from the Schwartz equation has been used as a covariate to determine the pediatric dose of several drugs (Table 1). This equation has been recommended for both renal function assessment and dose adjustment in the FDA Draft Guidance on General Clinical Pharmacology Considerations for Pediatric Studies for Drugs and Biological Products.⁹ However, the accuracy of estimating CL using eGFR largely remains unexplored. It is particularly challenging for pediatric patients under two years of age due to the ontogeny of drug elimination that involves glomerular filtration, transporters, and drug metabolism. Despite the known inherent limitations for interpreting the resultant eGFR values, reliance on the Schwartz equations to estimate GFR is widely accepted in clinical practice and is clinically useful when the limitations are understood. Nevertheless, methods that can improve accuracy in the estimation of GFR in children would be desirable.

The issues of dose adjustment based on eGFR has previously been raised in the pediatric labeling of valganciclovir (Valcyte).¹⁰ According to a FDA Drug Safety Communication,¹¹ the pediatric dose of valcyte was initially determined by 7 × BSA × CrCl (unit: mg, CrCl calculated using a modified Schwartz formula). Under this dosing recommendation, “pediatric patients with low body weight, low body surface area, and below normal serum creatinine could have a high calculated Schwartz creatinine clearance, resulting in a pediatric dose that approached the adult dose of 900 mg”. Consequently, FDA has updated the pediatric dosing recommendations to set up a maximum value of 150 mL/min/1.73 m² for CrCl in the formula, with a maximum dose of 900 mg.¹¹

The value of k in the Schwartz equation has been modified throughout the years. The k values in the FDA draft guidance⁹ are as follows: k = 0.33 for preterm infants <1 year of age, k = 0.45 for full-term infants <1 year of age, k = 0.55 for female child 1 – <12 years of age, k = 0.70 for male child 1 – <12 years of age. This set of k values is commonly seen in drug labels where the Schwartz equation is applied (Table 1). With the development of more accurate SCR assay methods (e.g., switching from the Jaffe method to enzymatic assay), a modified Schwartz equation was published in 2009 with k = 0.413,¹² known as the bedside Schwartz equation. A recent study found that a lower k value (0.368) significantly reduced the overestimation of GFR by the Schwartz equation in children < 13 years of age.¹³ Our analysis is consistent with previous reports suggesting that a lower k value may be necessary in younger children.¹³

In our study, most of the subjects with supraphysiological eGFR¹⁴ were associated with eGFR/CL ratio >1.5, accompanied by a measurement of serum creatinine level (< 0.2 mg/dL) close to the limit of detection for the assay (0.1 mg/dL) (Figure 2A).¹⁵ The patients with supraphysiological eGFR did not show proportionally high values in the clearance of these drugs. The discrepancy between eGFR and CL is more evident for patients with low SCR (Figure 2A). This finding indicated that the “supraphysiological eGFR” identified most likely resulted from an estimation error due to the low SCR.¹⁶ Notably, the eGFR/CL ratio was consistently higher than 1 even if those high outliers of eGFR were excluded, which suggests that the overestimation of CL was not limited to only those subjects with a high GFR.

The methodologies in obtaining eGFR and CL values are noteworthy when interpreting their ratio. The Schwartz equation assumes a fixed linear relationship between GFR and the ratio of Height/SCR. This relationship is empirically valid with the dataset previously used to establish the equation, while the coefficient k value may differ by factors such as age groups, physiological conditions, and assay methods. The bedside Schwartz equation reasonably provides an estimation of renal function as a means of monitoring kidney disease progression, however, the extent of the deviation is concerning especially in the context of guiding dosing. In contrast, drug CL was determined by the observed PK data of renally-eliminated drugs using a standard population PK analysis approach. Body weight and age are the most significant covariates for drug clearance in the final population PK models.

There are multiple factors that affect the SCR level and the estimation of GFR with SCR-based equations.⁸ Firstly, the baseline SCR for any given individual represents a steady state where daily creatinine generation from skeletal muscle and dietary protein metabolism equals urinary creatinine excretion. Factors that alter this balance between creatinine generation and excretion can impact SCR for any given individual. Reduced muscle mass, malnutrition, or a vegetarian diet is associated with decreased creatinine production – the resultant lower SCR values in these individuals can result in relatively higher eGFR values using the Schwartz equations. In addition, an abrupt decrease of SCR elimination rate (at birth or due to acute renal impairment) is associated with a slow but steady increase in SCR, and SCR-based equations would overestimate the GFR until a new steady state is reached and SCR is measured after that. Moreover, the accuracy of SCR quantification needs to be considered.⁸ Both the Jaffe and enzymatic assays were used for the measurement of SCR for the four drugs in this study. The enzymatic assay for SCR calibrated to reference measurements by isotope dilution mass spectroscopy (IDMS) is usually recommended. However, a potential issue is that the IDMS calibration relies on a standard reference for serum creatinine that only goes down to 0.6 mg/dL, which is higher than the normal SCR level of most children less than 12 years of age (Figure 2A),¹⁷ who have SCR values well below 0.6 mg/dL. Furthermore, the bedside Schwartz equation was developed for children at 1–16 years of age with chronic kidney disease and median SCR of 1.3 mg/dL, so the optimum k value in the equation may deviate from 0.413 for the pediatric population with normal renal function or less than one year of age. Lastly, the tubular secretion of creatinine is associated with a 20%−25% overestimation of GFR by the Schwartz equation,¹⁸ and this percentage could be higher for the patients with renal impairment, since the tubular secretion may not decline in parallel with progressive glomerular dysfunction.^{13, 19} Therefore, the overestimation of drug CL by the Schwartz equation in the low CL range (Figure 2B) is partially due to the overestimation of GFR by the Schwartz equation, especially when true GFR is low.

Our study has some important limitations. Since all the studied indications were not related to renal disease, this analysis did not assess possible changes in GFR that could be related to the drug. It is unknown whether changes in GFR were related to the patient’s underlying disease or potentially to nephrotoxic effects from treatment with the drug. In addition, our evaluation was based on the drugs that are >90% eliminated by glomerular filtration, while other elimination pathways, such as drug reabsorption, secretion, and metabolism, may play a minor role that could confound our analysis. Finally, this retrospective analysis is based on clinical data of 352 patients from birth to 12 years of age. Future studies with a larger sample size are warranted to address these limitations and to increase confidence in the obtained k values.

In conclusion, CL of renally-eliminated drugs could be overestimated when using creatinine-based renal function equations alone to adjust the dose in pediatric drug development. The overestimation was more evident in subjects with low SCR (e.g., < 0.2 mg/dL). When more precision is required in estimating GFR, such as during drug development, revised k constants for the Schwartz equation or new methods of GFR estimation may be necessary.

Supplementary Material

S1

Figure S1 Goodness-of-fit plots for the model developed to estimate new k values in the Schwartz equation. (A) Observed values (DV) vs. individual predictions (IPRED). (B) DV vs. Population predictions (PRED). (C) Conditional weighted residuals (CWRES) vs. PRED. (D) CWRES vs. TIME (hr)

S2

Figure S2 Comparison of eGFR with a normal range of GFR. eGFR values were calculated using bedside Schwartz equation (A) and the original Schwartz equation in the FDA Guidance (B), respectively. The dotted lines represent a 5–95% range of normal GFR⁷

S4

Figure S4 Comparison of the k values from Schwartz equations and the k values calculated with the data in this study. The new k values were calculated in NONMEM® based on data of the four drugs assuming the CL equals to eGFR

S3

Figure S3 Ratio of estimated glomerular filtration rate (eGFR) to observed drug clearance (CL) versus body weight (A), body surface area (BSA, B), eGFR (C); and eGFR versus CL (D). eGFR values were calculated using bedside Schwartz equation (k = 0.413). ■ = amikacin, ● = gadobutrol, ▲ = gadoterate, ▼ = vancomycin. Red symbols represent the subjects whose eGFR exceeded 95% percentile of normal GFR. The red solid line shows the cutoff of eGFR/CL ratio at 1. The black dashed lines represent the boundaries with eGFR/CL ratio of 0.5 and 2, respectively. The loess smooth curves for amikacin (blue), gadubutrol (green), gadoterate (orange), and vancomycin (purple) are also shown. Log scale was used for the x-axis of panel C and y-axis of panels A, B, and C

S5

Figure S5 Ratio of estimated glomerular filtration rate (eGFR) to observed drug clearance (CL) versus age. The eGFR values were calculated using the k values (Figure S4) derived from data of the four drugs in our study. The loess smooth curves for amikacin (blue), gadubutrol (green), gadoterate (orange), and vancomycin (purple) are also shown

S6

Table S1 Serum creatinine (SCR)-based equations for obtaining estimated glomerular filtration rate (eGFR)²⁰

S7

Table S2 Parameter estimates of the model for calculating new k values based on PK data of the four drugs

S8

Table S3 Accuracy of the new equation in estimating drug CL

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

  • For renally-eliminated drugs, the first-in-pediatric dose in clinical trials is usually selected by predicting the drug clearance using body weight and age. However, this approach does not account for alterations in clearance due to renal dysfunction.
• WHAT QUESTION DOES THIS STUDY ADDRESS?

  • The Schwartz equation has been used in drug labels for pediatric dose recommendations. This study examines the performance of the Schwartz equation and other serum creatinine-based renal function equations in estimating the clearance of drugs that are predominantly renally-eliminated.
• WHAT THIS STUDY ADDS TO OUR KNOWLEDGE?

  • The current study demonstrates that (1) both the original and bedside Schwartz equations overestimate drug clearance, and (2) this overestimation is associated with low serum creatinine levels in pediatric patients.
• HOW THIS MIGHT CHANGE CLINICAL PHARMACOLOGY AND THERAPEUTICS?

  • Dose calculation in pediatric patients using creatinine-based eGFR equations may lead to drug overdosage. A lower k constant for the Schwartz equation or new estimation methods are necessary when predicting renal drug clearance and dosing in pediatric drug development.