Low Etanercept Concentrations in Children With Obesity and Juvenile Idiopathic Arthritis

Abstract

OBJECTIVE

To evaluate the impact of obesity on etanercept (ETN) drug exposure in children with juvenile idiopathic arthritis (JIA).

METHODS

We conducted a pilot, cross-sectional, observational study in a real-world cohort of children with JIA receiving ETN as standard of care from a single center. We analyzed the relationship between body size and ETN plasma concentrations, adjusting for dosage. Body size was analyzed as a continuous measure using weight and body mass index (BMI) percentiles and categorically using BMI percentile classifications according to the CDC guidelines.

RESULTS

We enrolled a total of 29 children. Each child provided one plasma sample for ETN concentration measurement, and all participants were receiving subcutaneous ETN dosed weekly. We observed that the ETN concentration normalized for dose decreased significantly as a function of weight (p = 0.004) and BMI percentile (p = 0.04). Similarly, we observed a progressive decline in mean and median dose-normalized concentrations across higher body size categories. Because of reaching maximum ETN dosage (50 mg), 7 of 8 children (87.5%) with obesity received a weight-based dosage < 0.8 mg/kg/dose.

CONCLUSIONS

We found that higher body weight and BMI percentile are significantly and negatively associated with ETN drug serum concentration, accounting for differences in dosing. Our data suggest that children who are obese may be routinely under-dosed using current dosing strategies. Inadequate dosing may increase the risk for therapeutic failure and long-term morbidity in a developing child. As a result, characterizing adequate drug exposure in children of all sizes is an important step toward precision dosing.

Introduction

Etanercept (ETN) is FDA approved for the treatment of juvenile idiopathic arthritis (JIA) and was the first tumor necrosis factor–alpha (TNF-α) inhibitor studied in a randomized trial in children with arthritis.¹ Pediatric dosing is weight-based; when dosed once/week, children ≥ 63 kg receive a fixed dosage of 50 mg subcutaneous weekly, and children < 63 kg receive 0.8 mg/kg/dose. Despite widespread use in JIA, treatment failure with ETN occurs in up to 25% or more of children.¹ Robust data in adults suggest that obesity can exacerbate therapeutic failure with TNF-α inhibitors; for example, adult patients with obesity are 80% more likely to fail therapy.²

One potential mechanism by which obesity could worsen outcomes with TNF-α inhibitors is by altering drug pharmacokinetics.³ Alterations in monoclonal antibody pharmacokinetics seen in patients with obesity include increased proteolytic clearance, reduced absorption, changes in the volume of distribution, and other physiologic changes.³ These changes are also likely relevant for ETN because higher body weight is associated with increased ETN clearance.^3,4 In turn, high clearance can result in subtherapeutic drug serum concentrations independent from a ‘dilutional' effect of body size, thereby leading to therapeutic failure.

Despite the potential for obesity to result in sub-therapeutic ETN serum concentrations, no studies have investigated the impact of obesity on ETN exposure in children with JIA. Because obesity complicates care in up to 1 in 5 children with JIA,⁵ and because suboptimal treatment of JIA can result in morbidity and disability in the long term,⁶ optimizing drug exposure in this population is imperative. Therefore, we conducted a pilot, cross-sectional, observational study in a real-world cohort of children with arthritis with the aim of characterizing ETN drug concentrations across body size.

Materials and Methods

We enrolled children with JIA who were receiving ETN as standard of care from a single institution (Children's Mercy Hospital) between October 6, 2016, and December 31, 2018. All patients were receiving once-weekly subcutaneous dosing of ETN. We collected a single plasma sample for ETN concentration at the participant's regularly scheduled clinic visit. Random samples were obtained from participants who were on unchanged doses of ETN for at least 3 consecutive weeks and were assumed to have serum concentrations that were at steady state.

Plasma ETN concentrations were analyzed using a commercial monoclonal antibody–based enzyme-linked immunosorbent assay (Eagle BioSciences, Cat No. IG-AB102, Nashua, NH). Briefly, following the manufacturer's protocol, either plasma samples diluted 1:50 in assay buffer or reference standards were added to a 96-well plate pre-coated with an ETN-specific mouse monoclonal capture antibody (clone 5A1). Immobilized ETN was quantified using a horseradish peroxidase–conjugated anti-human immunoglobulin G monoclonal antibody along with a chromogenic substrate. Sample concentrations were interpolated using a standard curve (range: 6 to 200 ng/mL; limit of detection: <2.5 ng/mL; intra-assay CV: <10%) and multiplied by their dilution factor to determine the resulting plasma concentration.

Plasma samples were additionally assayed for the presence of anti-drug antibodies (ADA) to ETN using a commercially available semi-quantitative enzyme-linked immunosorbent assay (Eagle BioSciences, Cat No. IG-BB102). Briefly, following the manufacturer's protocol, plasma samples diluted 1:10 in assay buffer or reference standards were added to a 96-well plate pre-coated with ETN. Immobilized ADA to ETN were detected using horseradish peroxidase–conjugated ETN along with a chromogenic substrate. Anti-drug antibody positivity was defined as an optical density greater than twice that of the negative control (limit of detection: approximately 10 ng/mL; intra-assay CV: <10%). In addition, we collected clinical data, including age, sex, weight, height, and ETN dosing information.

We computed basic statistics for demographics and clinical characteristics including mean, median, range, and interquartile range. All statistical analyses were done using R (version 3.5.2, R Foundation for Statistical Computing, Vienna, Austria) and RStudio (version 1.1.463, RStudio Inc, Boston, MA). Graphical analysis and linear/locally weighted scatterplot smoothing regression fits were generated using Phoenix NLME (version 8, Certara, Princeton, NJ). Dose-normalized concentrations were determined based on dose (mg); the ETN concentration (ng/mL) was divided by the dose in milligram (equivalent to ng/mL/mg). In exploratory analyses, we also corrected plasma concentration by weight-based dosage using each participant's actual weight.

We analyzed body size as a continuous measure using total body weight and BMI percentiles and categorically based on BMI percentile classifications, according to the CDC guidelines.⁷ Using these guidelines, we classified children as obese (BMI ≥ 95%), overweight (BMI 85 to <95%), healthy weight (BMI 5 to <85%), or underweight (BMI < 5%). A BMI that exceeded the 99th percentile (n = 1 for our cohort) was imputed as 99%. We separately evaluated the difference in concentrations and dose-normalized concentrations for children with and without obesity using a one-way analysis of variance. We also evaluated the difference in dose-normalized concentrations vs weight and BMI percentiles using unadjusted linear regression. In exploratory analyses, we used Phoenix NLME (Certara) to graphically visualize the relationship between time- and dose-normalized ETN concentrations vs weight (e.g., dose-normalized concentration divided by time after last dose). Unless otherwise noted, data are presented as mean ± SD.

Results

Demographics and Clinical Characteristics. We enrolled a total of 29 children into the study, more than one-third of whom were classified as “overweight” or “obese” (Table). Concomitant medications included methotrexate (n = 18) and leflunomide (n = 2). Of those who did not receive methotrexate, 3/11 (27.3%) were obese, and of those who did receive methotrexate, 5/18 (27.8%) were obese. Both participants who received concomitant leflunomide were non-obese.

Table.

Demographics and Clinical Characteristics

Characteristics (N = 29)	Result
Female, n (%)	21 (72.4)
Age, median (range), yr	10 (2–18)
Race, n (%)
White	25 (86.2)
Other	4 (13.8)
CDC weight classification, n (%)
Obese	8 (27.6)
Overweight	3 (10.3)
Healthy weight	15 (51.7)
Underweight	3 (10.3)
BMI percentile, median (range)	68 (1–99)
Weight, median (range), kg	31.4 (10.5–168.9)

Each child provided one sample for ETN concentration and one sample for ADA concentration (n = 29 samples each). All participants received ETN dosed on a weekly basis for a duration of ETN therapy of 51 ± 64 weeks. The dose for all patients was 32.8 ± 14 mg, and the weight-based dose was 0.78 ± 0.19 mg/kg. Samples were obtained at a median (range) of 6 (2–11) days since last dosage. For the entire pediatric cohort, unadjusted ETN concentration was 2268 ± 1330.3 ng/mL, and the dose-normalized ETN concentration was 79.4 ± 54 ng/mL/mg. No participants had detectable ADA to ETN.

Dosing in Children With and Without Obesity. In children with obesity, the average ± SD weight-based dose was 0.64 ± 0.17 mg/kg, compared with 0.84 ± 0.17 mg/kg in children without obesity (p = 0.009); the average ± SD dose was 46.9 ± 8.8 mg and 27.5 ± 11.8 mg, respectively. Because of reaching maximum ETN dosage (50 mg), 7 of 8 (87.5%) children with obesity received a weight-based dosage of <0.8 mg/kg/dose. Only one child with obesity received a dosage of ETN of <50 mg. The average ± SD unadjusted ETN concentration was 1821 ± 1115.7 ng/mL in children with obesity compared with 2438.3 ± 1389.8 ng/mL in children without obesity (p = 0.27).

Dose-Normalized Concentration and Body Size. When analyzing weight as a continuous variable, we observed that the dose-normalized ETN concentration decreased significantly as a function of weight (p = 0.004, Figure 1), although the fit was not linear. Similarly, there was a significant negative association between dose-normalized ETN concentration and BMI percentile (p = 0.04, Figure 2), although this, too, had poor linear fit.

Figure 1.

Dose-normalized (mg) concentration vs weight. Solid black line represents a LOESS fit.

Figure 2.

Dose-normalized (mg) concentration vs body mass index (BMI) percentile. Solid black line represents a LOESS fit.

When analyzing body size categorically, we observed a progressive decline in mean and median dose-normalized concentrations across larger body size categories (Figure 3). The dose-normalized ETN concentration was significantly lower in children with obesity vs in those without: 40.1 ± 23.3 vs 94.4 ± 55 ng/mL/mg (p = 0.01).

Figure 3.

Dose-normalized (mg) concentration vs weight category. Solid lines within the boxplot represent mean, whereas the dotted line represents the median. Boxes represent the IQR, and whiskers represent 1.5× the IQR.

In exploratory analyses, we normalized ETN concentration by weight-based dosage, hypothesizing that the ETN dosage cap (50 mg) would result in an increase in normalized concentration across progressively higher body weights (e.g., dividing concentration by a progressively smaller ratio of dose to weight). Consistent with our hypothesis, we observed a general trend in increased weight-based dosage-normalized concentration. Nevertheless, because of the limited number of subjects with a weight > 63 kg (n = 9), the overall visual association was poor, thereby limiting definitive conclusions.

Next, to evaluate for potential confounding between sample timing and drug concentrations, we analyzed administration timing between groups. The average (range) number of days between ETN dose and each concentration measurement was 5.2 days (range 2–11 days). Compared with children without obesity, the number of days after ETN dose was numerically greater for children with obesity (mean 6.5 vs 4.8 days), but this difference was not statistically significant (p = 0.07). In addition, when normalizing concentration by both dose and time after dose, there remained a negative association with weight.

Discussion

We conducted a cross-sectional, observational analysis of children receiving ETN in a real-world setting to evaluate the impact of obesity on drug exposure. In our cohort, where more than one-third of children we enrolled were overweight or obese, we found that higher body weight and BMI percentile were both negatively and significantly associated with ETN drug concentrations, despite accounting for differences in dosing. Similarly, we observed a dose-exposure relationship, with larger body size categories associated with progressively lower ETN drug concentrations.

There are several possible explanations for the low ETN exposure in children with obesity. First, low exposure may be due to the use of an adult dosage cap in children. For example, the average female child at the 50% percentile for weight receiving 0.8 mg/kg of ETN would not exceed the maximum dosage (50 mg) through 20 years of age; however, the average female child at the 95% percentile for weight would achieve maximum dosage around 12.5 years of age. Because of this restricted dosing, all but one child with obesity in our study received < 0.8 mg/kg of ETN. Many years of significant growth are expected to occur in the average obese child who otherwise received the maximum adult dosage of ETN, so our data suggest that restricting dosage likely results in under-dosing for children who are overweight or obese.

A second possible explanation for the low ETN exposure relates to pharmacokinetic (PK) changes induced by obesity. Specifically, both the volume of distribution and the proteolytic degradation of many biologic drugs (most monoclonal antibodies) increase as weight increases, potentially resulting in shorter half-lives and lower drug concentrations in patients with obesity.^2,3 These changes are also relevant for children with JIA receiving ETN, in whom PK analyses found a non-linear relationship between body surface area and increased ETN clearance.⁸ High ETN clearance, in turn, requires a higher dosage to achieve the same degree of drug exposure. Because current mg/kg dosing strategies for ETN falsely assume a linear relationship between body weight and these physiologic processes,⁹ children who are obese may be routinely under-dosed. As recommended by the FDA, the effect of weight on these physiologic parameters is often better characterized using a non-linear allometric scaling function.¹⁰

From a clinical perspective, low drug exposure could result in active arthritis and therapeutic failure. Because of the opportunistic nature of data collection in our study, and the need to control for concomitant medications, we did not have the power to examine relationships between ETN exposure and disease activity. Nevertheless, several published studies in adults show an exposure response for ETN and other biologics. For example, ETN trough concentrations were significantly correlated with change in disease activity score for patients with rheumatoid arthritis (p = 0.001), with trough concentrations of 1240 ng/mL having an 81% sensitivity and 100% specificity for a good treatment response at 6 months.¹¹ Similarly, a much larger study¹² involving rheumatoid arthritis found that ETN concentrations were lower in non-responders (median 2800 ng/mL) compared with moderate (3100 ng/mL) and good (3780 ng/mL) responders. Conversely, average ETN trough concentrations in children with JIA may range from 1740 to 1910 ng/mL.¹³

Although potential differences in sample timing and assays limit formal comparison with other published studies, we observed an average unadjusted ETN concentration in children with obesity (1821 ng/mL) that suggests suboptimal exposure. Nonetheless, to fully characterize the effects of obesity on disease activity, drug concentrations, and dosing requirements for ETN and other biologics, it is critical that future studies leverage population PK/pharmacodynamic modeling using data from children from across a wide range of body sizes.

There are some limitations to our study. First, we enrolled a relatively small sample size, which limited our statistical power. While it is possible that our observed associations occurred by chance, the consistent relationship we observed between multiple measures of body size and ETN concentrations make this unlikely. Second, we leveraged an opportunistic design in which we did not directly control the timing of sample collection. Broadly, drug concentrations are dependent on both dosage and the time elapsed between the last dose and the sample measurement. In our study, we controlled for dosage by normalizing concentration based on dose, a technique widely used in clinical pharmacology studies. While dose normalization assumes linear pharmacokinetics, this assumption is very reasonable because most published pharmacokinetic models for ETN use linear parameterization.⁸ However, the time after dose varied among subjects; it is theoretically possible that differences in the time after dose created imbalances in ETN concentrations between groups. Specifically, the average difference in time after dose of 1.7 days represents approximately 0.5 half-lives for ETN in children.⁸ Therefore, although the time after dose was greater numerically (though not statistically significant) for patients with obesity in our study, we do not believe this accounts for the nearly twofold differences in concentrations between individuals with and without obesity. Lastly, while we could not measure medication adherence in our study, we would not expect adherence to differ between obese and non-obese children.

Conclusion

We found that higher body weight and BMI percentile are negatively associated with ETN drug concentrations, suggesting that children who are obese may be routinely under-dosed using current dosing strategies. Inadequate dosing may increase the risk for therapeutic failure and long-term morbidity in a developing child. As a result, characterizing adequate drug exposure in children of all sizes is an important step toward precision dosing. Larger studies leveraging pharmacokinetic/pharmacodynamic modeling are needed to fully characterize the complex relationship between obesity, disease activity, drug concentration, and the need for optimal dosing.

ABBREVIATIONS

ADA

anti-drug antibodies

BMI

body mass index

CDC

Centers for Disease Control and Prevention

ETN

etanercept

JIA

juvenile idiopathic arthritis

TNF-α

tumor necrosis factor–alpha