Rodrigues C (1), Hansen L (2), Skartved NJ (2), Tessier A (1)
(1) Clinical Pharmacometrics - Quantitative Pharmacology, Servier, Suresnes, France, (2) Symphogen, Ballerup, Danemark
Introduction/Objectives: Therapeutic monoclonal antibodies (mAbs) have been a dominating class of new approved drugs during the last two decades. To determine the first-in-human starting dose, human pharmacokinetics (PK) can be extrapolated from pre-clinical species, but other methods may be applied. The aim of this work is to analyze the extrapolation methods of 5 internal mAb assets and the associated prediction error when comparing with clinical observations.
Methods: The extrapolation methods to predict the human PK of 5 different mAbs were analyzed. Prediction error for each PK parameter was calculated as:
Prediction error = (θpred – θobs)/θobs*100
Where θpred is the parameter predicted by the extrapolation method and θobs is the parameter estimated by popPK modelling on observed human data.
Results: Human PK for mAb1 was considered to be similar to its main competitor as the clearance (CL) of both compounds were comparable in monkeys and mice. The final extrapolated PK model for mAb1 was a 2-compartment model with a linear elimination combined with a saturable elimination pathway of the Michaëlis-Menten type, which PK parameters were fixed to its competitor values. For mAb2 and mAb3, PK predictions in human were derived from PK in cynomolgus monkeys. Since the PK of both mAbs were linear and dose-proportional in in anti-drug antibody (ADA)-negative monkeys, the PK in humans were also assumed linear and dose-proportional. CL was predicted via one-species allometric scaling using an exponent of 0.85 as described by [1]. Typical values for central volume of distribution (V1), peripheral volume of distribution (V2), and relevant distribution clearance (Q) reported for other mAbs – reviewed in [2] – were used in the 2-compartment model. The non-linear human PK of mAb4 was also predicted from monkey but using different allometric exponents. An exponent of 0.75 was used for CL, Q and Vmax (maximum rate of elimination via saturable mechanism of elimination) and an exponent of 1 was used for V1 and V2, as described by [3]. The Km (concentration for 50% of Vmax) was assumed to be the same in monkeys and humans. As mAb5 is not cross-reactive with mouse or cynomolgus monkey and target expression is expected to be low in human, only non-target mediated and linear PK could be addressed using hFcRn Tg32 mice data. Human PK was predicted via one-species allometric scaling using exponents of 0.90 for CL, 0.67 for Q, 0.97 for V1, and 0.93 for V2 [4]. Clinical data were already obtained for mAb1, mAb2, mAb3 and mAb4 (preliminary data) and ongoing for mAb5. The structural models were all correct except for mAb3, for which target mediated drug disposition (TMDD) was observed at the lowest doses. The parameter with less prediction error was V1 (-3 to -20%). CL was well predicted (-12 to 14% prediction error) except for mAb3 (83%) probably because the non-linear part of the elimination was not taken into account in the extrapolated model. Vmax had high prediction errors ranging from -100 to 616 %, as well as Km (-100 to 527 %).
Conclusions: The results highlight the difficulty to predict the non-linear PK in humans using data from preclinical species, or from data reported for competitor antibodies. Overall, predicted typical profiles in the linear phase of the extrapolated models are similar to the typical profiles obtained with the popPK model built with observed clinical data. In summary, all the methods used to predict human PK seem to describe accurately the linear PK observed at high doses, while it remains challenging to extrapolate non-linear PK.
References:
[1] Wang, J, Iyer, S, Fielder, P J, Davis, JD, et al (2016) Biopharm Drug Dispos.; 37(2): 51-65.
[2] Dirks, N., & Meibohm, B. (2010). Clin.Pharmacokinet., 49(10), 633-659
[3] Dong et al (2011). Clin Pharmacokinet, 50(2): 131-142
[4] Betts et al (2018). mAbs, 10:4, 636-650
Reference: PAGE 30 (2022) Abstr 10030 [www.page-meeting.org/?abstract=10030]
Poster: Methodology - Other topics