Hyeseon Jeon1, Jong Hyuk Byun2,3*, Jae Kyoung Kim4,5*, Soyoung Lee1*, Jung-woo CHAE1,6*, Hwi-yeol Yun1,6*
1 College of Pharmacy, Chungnam National University, Daejeon, South Korea 2 Department of Mathematics and Institute of Mathematical Science, Pusan National University, Busan 46241, Republic of Korea 3 Institute for Future Earth, Pusan National University, Busan 46241, Republic of Korea 4 Biomedical Mathematics Group, Pioneer Research Center for Mathematical and Computational Sciences, Institute for Basic Science, Daejeon 34126, Republic of Korea 5 Department of Mathematical Sciences, KAIST, Daejeon 34141, Republic of Korea 6Department of Bio-AI convergence, Chungnam National University, DaeJeon, Republic of Korea *Those of authors contributed equally as correspondence.
Objectives: Target-Mediated Drug Disposition (TMDD) is a phenomenon that occurs when a potent and specific drug binds to the target site with high affinity, leading to changes in its pharmacokinetic characteristics [1]. Approximation methods have arisen from the increasing number of parameters in TMDD models, making it more feasible to develop simplified models for data fitting [2]. While previous studies have explored the validity conditions of approximation methods, they have primarily focused on conditions where drug concentration exceeds the target. This is typical since TMDD is prevalent in high-affinity-low-capacity targets. In low doses, a significant portion of the drug binds to the target, making drug concentration in the systemic circulation low. In high doses, a higher portion of the drug is in systemic concentration due to target saturation [3]. However, some cases where the concentration is lower than the target, such as micro-dosing studies or drugs exhibiting significant non-specific protein binding, also have shown TMDD [4]. This study is significant in addressing such scenarios.
Methods:
In the previous study by Byun et al.(to be published), the validity criterion of TMDD approximation methods, including Michaelis-Menten model (referred to as mTMDD), Quasi steady state (QSS) model (referred to as qTMDD), and first-order approximation of the total QSS (referred to as pTMDD), a new method suggested by the author, was presented. Byun et al. showed that when C_0+k_m≫R_tot, all three methods were accurate, whereas when C_0+k_m≪R_tot, mTMDD became inaccurate, while qTMDD and pTMDD remained accurate.
In this study, we selected two anti-drug conjugates (ADC) clinical trial studies: hil-1Ra-hyfc and anakinra(used as active comparator of hil-1Ra-hyfc), satisfying the former condition, and hil-7-hyfc, satisfying the latter. Detailed information on each study is provided by Lee et al.[5] and Ngo et al.[6]. All three studies were a semi-mechanistic model with TMDD and FcRn-mediated recycling. The models underwent parameter optimization using the first-order conditional estimation with interaction (FOCE-I) method in NONMEM 7.5 and PsN 5.3.1 software. Subsequently, the performance of the models was assessed based on model diagnostic criteria and diagnostic plots.
Results:
In our clinical trial studies, in case of C_0+k_m≫R_tot, mTMDD, qTMDD, and pTMDD were all accurate, whereas in case of C_0+k_m≪R_tot, mTMDD was inaccurate while qTMDD and pTMDD remained accurate. This assessment was based on Objective Function Value (OFV), pharmacokinetic parameters, and Visual Predictive Check (VPC) analysis. For hil-1-Ra-hyfc corresponding to the former case, OFV was similar across all models, with values of 1556.403, 1564.453, 1556.586, and 1561.906 for the original, mTMDD, qTMDD, and tTMDD, respectively. Similarly, for anakinra, which also corresponds to the former case, the OFV values were also comparable, with 359.309, 358.45, 359.69, and 361.173 for the original, mTMDD, qTMDD, and tTMDD, respectively. Conversely, for hil-7-hyfc corresponding to C_0+k_m≪R_tot, the OFV values were 1552.9, 1611.916, 1552.9, and 1553.139 for the original, mTMDD, qTMDD, and tTMDD, respectively, indicating the lowest predictive capability in the mTMDD model. Additionally, in the mTMDD model of hil-7-hyfc, the pharmacokinetic parameters such as Q, VD, and VI exhibited significant disparities compared to other models. Moreover, Visual Predictive Check (VPC) plots also showed that the mTMDD model could not predict appropriately, whereas the qTMDD and pTMDD models could appropriately describe the observation data. Last but not least, the elapsed time of hil-1-Ra-hyfc was 5345.33, 403.25, 1113.53, and 555.99 seconds for the original, mTMDD, qTMDD, and tTMDD, respectively. For Anakinra, the elapsed time was 122.40, 2.07, 2.83, and 1.70 seconds. For hil-7-hyfc, the elapsed time was 136.09, 44.35, 66.26, and 39.11 seconds for the original, mTMDD, qTMDD, and tTMDD, respectively.
Conclusions:
This study supports the validity criteria proposed by Byun et al., confirming that when C_0+k_m≫R_tot, mTMDD, qTMDD, and pTMDD were all accurate in clinical cases. In contrast, when C_0+k_m≪R_tot, mTMDD became inaccurate, while qTMDD and pTMDD remained accurate in clinical cases. Hence, in cases where the concentration of the target markedly exceeds that of the drug, the pTMDD approximation is a viable option, delivering both precision and computational efficiency.
References:
[1] Text for reference [1] Levy G. (1994). Pharmacologic target-mediated drug disposition. Clinical pharmacology and therapeutics, 56(3), 248–252. https://doi.org/10.1038/clpt.1994.134
[2] Dua P, Hawkins E, van der Graaf PH. A Tutorial on Target-Mediated Drug Disposition (TMDD) Models. CPT Pharmacometrics Syst Pharmacol. 2015 Jun;4(6):324-37. doi: 10.1002/psp4.41. Epub 2015 Jun 15. PMID: 26225261; PMCID: PMC4505827.
[3] An G. (2020). Concept of Pharmacologic Target-Mediated Drug Disposition in Large-Molecule and Small-Molecule Compounds. Journal of clinical pharmacology, 60(2), 149–163. https://doi.org/10.1002/jcph.1545
[4] Smith DA, van Waterschoot RAB, Parrott NJ, Olivares-Morales A, Lavé T, Rowland M. Importance of target-mediated drug disposition for small molecules. Drug Discovery Today. 2018;23(12):2023-30.
[5] Lee, S. W., Choi, D., Heo, M., Shin, E. C., Park, S. H., Kim, S. J., Oh, Y. K., Lee, B. H., Yang, S. H., Sung, Y. C., & Lee, H. (2020). hIL-7-hyFc, A Long-Acting IL-7, Increased Absolute Lymphocyte Count in Healthy Subjects. Clinical and translational science, 13(6), 1161–1169. https://doi.org/10.1111/cts.12800
[6] Ngo, L., Oh, J., Kim, A., Back, H. M., Kang, W. H., Chae, J. W., Yun, H. Y., & Lee, H. (2020). Development of a Pharmacokinetic Model Describing Neonatal Fc Receptor-Mediated Recycling of HL2351, a Novel Hybrid Fc-Fused Interleukin-1 Receptor Antagonist, to Optimize Dosage Regimen. CPT: pharmacometrics & systems pharmacology, 9(10), 584–595. https://doi.org/10.1002/psp4.12555
Reference: PAGE 32 (2024) Abstr 10931 [www.page-meeting.org/?abstract=10931]
Poster: Methodology - New Modelling Approaches