(S-112) Sparse Sampling Strategies for Monoclonal Antibody Pharmacokinetics: Evaluating Estimation Methods for Reliable Clearance Assessment

Kyeongmin Kim – The Ohio State University; Jin Gyu Kim – The Ohio State University; Bryan Remaily – The Ohio State University; Mitch Phelps – The Ohio State University

Objectives: The clearance (CL) of monoclonal antibody (mAb) immune checkpoint inhibitors (ICIs) is associated with clinical outcomes. Though ICI exposure is not associated with response across dose levels, higher baseline CL links to short duration of response, and CL decline over time correlates with better outcomes [1]. Therefore, CL is a useful biomarker for outcomes from ICI and other mAb therapies [2]. However, accurate measurement and estimation of mAb baseline and time-varying CL requires longitudinal blood sampling, which is costly and impractical for standard-of-care treatment. This study evaluated the accuracy of baseline and time-varying CL determination using sparse sampling with varying pharmacokinetic (PK) parameter estimation methods, sampling designs, and data quantity.

Methods: PK profile simulations, dataset generation, and statistical analyses were performed using R, and PK estimation was conducted in NONMEM 7.5 using a published time-dependent PK model (Li et al. 2017) [3]. A virtual cohort of 100 adult NSCLC patients was simulated using parameter distributions from the original population model. Patients received 3 to 15 doses of pembrolizumab 200 mg intravenously every 3 weeks. Simulated true PK parameters served as references for evaluating estimation performance of Maximum-A-Posterior (MAP) Bayesian (M1), maximum likelihood estimation (MLE) with $PRIOR record (M2), and MLE without $PRIOR (M3). Performance was assessed across rich and sparse sampling designs, varying weighting and bias of priors on THETA and OMEGA, and different numbers of dosing intervals. Mean absolute error (MAE) was used to assess performance.

Results: M1 was sensitive to biased priors even with rich sampling. M3 produced unstable estimates under sparse sampling schemes and was excluded from further analysis. M2 showed robust performance across all sampling schemes. Under sparse designs, M2 maintained MAE < 15.4% for all PK parameters, even with biased priors, and was therefore further evaluated. Uninformative OMEGA priors outperformed informative OMEGA priors when these priors were biased. However, non-informative THETA priors led to poor estimates of Q and V2 (MAE >30%), which was corrected by selectively increasing prior weights on these vulnerable parameters. Estimation accuracy for all parameters improved with sampling across increasing number of dosing intervals (longer sample collection period). Cycle-specific evaluation of baseline and time-varying CL showed that baseline CL can be reliably estimated even with short sampling periods. However, accurate estimation of time-varying CL required sampling in later cycles to achieve low MAE. Reducing RUV from 25% to 15% consistently improved estimation of all parameters.

Conclusions: Our findings support the use of MLE with $PRIOR (M2) as a robust and flexible method for estimating individual PK parameters in sparse sampling designs. M2 outperforms MAP-Bayesian estimation when prior information is limited or biased and can be optimized by adjusting prior weights based on the sampling design and model structure. These results offer practical guidance for determining ICI and other mAb CL as a biomarker for outcomes when dense sampling is often infeasible.

Citations: [1] Turner DC, Kondic AG, Anderson KM, et al. Pembrolizumab Exposure–Response Assessments Challenged by Association of Cancer Cachexia and Catabolic Clearance. Clin Cancer Res. 2018;24(23):5841-5849. doi:10.1158/1078-0432.CCR-18-0415
[2] Guo Y, Remaily BC, Thomas J, et al. Antibody Drug Clearance: An Underexplored Marker of Outcomes with Checkpoint Inhibitors. Clinical Cancer Research. 2024;30(5):942-958. doi:10.1158/1078-0432.CCR-23-1683
[3] Li H, Yu J, Liu C, et al. Time dependent pharmacokinetics of pembrolizumab in patients with solid tumor and its correlation with best overall response. J Pharmacokinet Pharmacodyn. 2017;44(5):403-414. doi:10.1007/s10928-017-9528-y

Keywords: Immune Checkpoint Inhibitors, Clearance Estimation, Sparse Sampling