A semi-mechanistic population pharmacokinetic model for trastuzumab emtansine (T-DM1) antibody-drug conjugate and total antibody in patients with metastatic breast cancer (mBC)
Franziska Schaedeli Stark (1), Vaishali L. Chudasama (2), Manish Gupta (3), Jay Tibbitts (3), Nicolas Frey (1), Donald E. Mager (2)
(1) Translational Research Sciences, F. Hoffman-La Roche Ltd, Basel, Switzerland, (2) Department of Pharmaceutical Sciences, University at Buffalo, SUNY, Buffalo, NY, (3) Pharmacokinetics and Pharmacodynamics, Genentech Inc., South San Francisco, CA
Objectives: Antibody-drug conjugates (ADC) are a class of targeted drugs with antibodies bearing covalently bound cytotoxic agents designed to target antigen-specific cells to enhance efficacy and reduce the toxicity associated with the cytotoxic agent alone. T-DM1, an ADC conjugating the antimicrotubule agent DM1 with trastuzumab (Tmab), exhibits faster elimination than total trastuzumab (TT, i.e. T-DM1 + unconjugated Tmab). The objective of this study is to characterize the pharmacokinetics (PK) of T-DM1 by integrating prior preclinical knowledge with clinical data.
Methods: T-DM1 and TT serum concentrations from a phase I dose escalation study (TDM3569g) in 52 patients , and from a phase II study (TDM4258g) in 111 patients  with mBC were available for model building. T-DM1 drug-antibody-ratio (DAR), i.e. the number of DM1 molecules per Tmab, ranges from 1 to 8. A semi-mechanistic PK model developed for monkey data, linking 8 measured DAR species with a chain of transit compartments , was used as a starting point. T-DM1 elimination kinetics is characterized by the net effect of transitions from higher to lower DAR species with loss of DM1 containing moieties, and proteolytic antibody elimination. Stepwise reduction of the number of compartments was performed to account for the reduced information (T-DM1 is the sum of DAR>0) in the clinical data set. The model was developed in S-ADAPT using Monte Carlo parametric expectation maximization.
Results: The transition from T-DM1 to Tmab was described with 5 distinct DAR species (DAR4 to DAR0), and each was characterised by a 2-compartment model with the same central volume (VC: 3.21 L; inter- individual variability: 17.5%), peripheral volume (VP: 1.70 L; 62.3%), inter-compartmental clearance (Q: 0.576 L/day; 63.1%), and Michaelis-Menten elimination (Km: 11.1 mg/L; 129% and Vmax/Km 0.751 L/day; 75.8 %). A first-order transition rate constant was used to link the different DAR species (Ktd: 0.35 day-1; 15.5%) and the fraction of DAR0 in the administered T-DM1 dose was estimated (fr0: 0.045; 63.6%). Both T-DM1 and TT PK data were adequately described by the model.
Conclusions: A semi-mechanistic PK model was able to successfully link the different DAR species of T-DM1 and to capture the extended terminal half-life of TT relative to T-DM1. This modeling framework may be useful to further investigate the release kinetics of DM1, and to characterize the PK of other ADCs.
 Krop IE et al. J Clin Oncol. 28:2698-2704 (2010)
 Burris HA et al. J Clin Oncol. E-publication ahead of print. Dec 20, 2010.
 Leipold D et al. AACR Meeting. 100, 2914 (2009)