160: Modeling Age-related Changes in Kidney Function Using Aminoglycoside Clearance as a Surrogate: Can Body Composition Help?

Statement of Purpose, Innovation or Hypothesis: One out of five adults are expected to be over the age of 65 years by 2040. Drug dose modification for aging often relies on estimates of kidney function. Current kidney function equations estimate different rates of age-related decline in kidney function but it is not clear how best to model this trajectory¹. Body size is used to estimate kidney function but we presently do not incorporate body composition, which also changes with age. In this project, we tested computed tomography (CT)-based measures of body composition to see if they better explain aminoglycoside clearance as a surrogate for kidney function. We hypothesized that the inclusion of body composition can improve the precision of age-related changes in kidney function.

Description of Methods & Materials: The study population retrospectively reviewed patients who were treated with tobramycin or gentamicin. Patients with at least three measured drug concentrations and available CT imaging were included. A population pharmacokinetic (Pop-PK) model for aminoglycoside was constructed to compute the volume (V) and clearance (CL) estimates with Monolix V2024R1. Body composition measurement was performed on CT images using Analytic Morphomics. Age + body composition + serum creatinine (Scr) composite indexes were compared to age + weight + Scr as potential covariates. Wald test stochastic approximation, Pearson's correlation test, and/or ANOVA were used to assess relationships between PK parameters and covariates.

Data & Results: The final sample included 156 unique patients, 89 (58%) of which had tobramycin administered (n=89). The median (minimum, maximum) age, height, and weight of included patients were 58 (21, 93) years, 170 (147, 203) centimeters, and 80 (43.8, 139.3) kilograms. The PK of aminoglycosides was best described using a one-compartment structure with an additive and proportional error model. The estimated PK parameters (between-subject variability %CV) were: 57.18 L (41%) for V and 5.71 L/h (55%) for CL. Univariate analyses identified age as the most significant potential covariate of CL, and V was dependent on the aminoglycoside. Total psoas muscle area, skeletal muscle area, and fascia area were potential covariates. However, the inclusion of weight, morphomic parameters, and Scr did not improve the model above that of age on CL and drug on V.

Interpretation, Conclusion or Significance: A one-compartmental Pop-PK model of aminoglycosides was developed. Weight and morphomic parameters by CT were not identified as covariates of aminoglycoside PK. A 0.70 mL/min/year (0.37 to 1.28 mL/min/year) decline in kidney function is predicted in this population between the ages of 30 and 90 years. These estimates align with the Baltimore Longitudinal Study of Aging which predicts a 0.75 mL/min/year across that age range. This similar finding suggests that aminoglycoside CL can serve as a potential surrogate for age-related kidney function decline. Larger real-world datasets should be tested to confirm these findings and evaluate alternate covariates predictive of age-related kidney function decline.

Disclosures: The authors declared no competing interests for this work.

Citations/References: Alikhani R, Pai MP. Reconsideration of the current models of estimated kidney function-based drug dose adjustment in older adults: The role of biological age. Clin Transl Sci. 2023; 16: 2095-2105. doi:10.1111/cts.13643

Additional Information/Authors: N/A