Modelling of Anti-Drug Antibodies Directed Against a Monoclonal Antibody
R. Niebecker (1, 2), C. Kloft (1)
(1) Department of Clinical Pharmacy and Biochemistry, Freie Universitaet Berlin, Berlin, (2) and Graduate Research Training program PharMetrX, all Germany
Objectives: Monoclonal antibodies (mAbs) are an important class of biopharmaceuticals. Due to their nature as foreign proteins, they can induce the formation of anti-drug antibodies (ADAs) in patients, which may affect pharmacokinetics (PK), pharmacodynamics and/or safety of the therapeutic mAb. Concerning the PK changes, it is generally assumed that the formation of immune complexes results in a more rapid clearance of the therapeutic mAb [1]. Population PK analyses of mAbs accounting for an immune response are sparse; usually this influential factor was incorporated as a binary variable on clearance in the covariate submodel [2]. Modelling of the immune response could facilitate more complex approaches. Consequently, the objective of the current analysis was to develop and compare different structural models for an immune response surrogate.
Methods: Data from 51 patients with solid tumours enrolled in two phase I/II clinical trials of a mAb were available for analysis. Weekly surface plasmon resonance (SPR)-based measurements (considered as ADA surrogate, given the constraints of immunogenicity analytics [1]) were performed to detect immunogenicity. The assay signal was assumed to comprise background (time-independent) and immune response-dependent signal. For the latter part linear, exponential as well as sigmoidal descriptions of the time course were compared. Modelling was performed using the PRED routine in NONMEM 7.2.
Results: According to SPR measurements, 13 patients developed an immune response. For these patients, the analysis revealed a significantly increased background signal compared to the patients not experiencing an immune response. Based on the observed time of seroconversion (i.e. ADA formation), it was possible to determine a shift factor characterising the start time of the immune response. The resulting signal increase was adequately described by linear or exponential functions. Since the available data did not provide information on the shape of the immune response-dependent signal beyond the phase of linear increase, models featuring additional complexity were not supported.
Conclusions: The development of a model describing the SPR assay signal-time profile as a surrogate of the immune response was possible. Next, the model predictions can be incorporated in population models for the therapeutic mAb to better characterise the alteration of PK caused by formation of ADAs.
References:
[1] R. Niebecker, C. Kloft. Curr Drug Saf 5: 275–86 (2010).
[2] N.L. Dirks, B. Meibohm. Clin Pharmacokinet 49: 633–59 (2010).