Target-mediated drug disposition model of rituximab in patients with diffuse large B-cell lymphoma
Armando Tratenšek (1), Jorge Nuno Resende Major (1, 2), Jurij Aguiar Zdovc (1), Samo Rožman (3), Iztok Grabnar (1), Srdjan Novakovic (4), Barbara Jezeršek Novakovic (5)
(1) University of Ljubljana, Faculty of Pharmacy, Ljubljana, Slovenia, (2) University of Lisbon, Faculty of Pharmacy, Lisbon, Portugal, (3) Institute of Oncology, Pharmacy Department, Ljubljana, Slovenia, (4) Institute of Oncology, Department of Molecular Diagnostics, Ljubljana, Slovenia, (5) Institute of Oncology, Department of Malignant Lymphomas, Ljubljana, Slovenia
Objectives: Rituximab is a chimeric anti-CD20 IgG1 monoclonal antibody which specifically binds to the CD20 antigen expressed on the surface of both normal and neoplastic B lymphocytes and is approved in the therapy of various lymphoid malignancies including diffuse large B-cell lymphoma (DLBCL)[1]–[3]. The standard dose of rituximab is 375 mg/m2 usually given in 2- or 3-week regimens (R-CHOP regimen)[1]–[3]. Previous studies with rituximab suggest that pharmacokinetics is nonlinear[2], [3] and influenced by antigen mass (amount of CD20 available for rituximab binding)[1], [3]. Target mediated drug disposition models (TMDD) usually describe nonlinear PK of therapeutic antibodies and characterize the joint kinetics of the antibody, its target antigen and the immune complex[3], [4]. As such, Rožman et al. developed a two compartment model comprising linear nonspecific clearance and time-varying specific clearance, corresponding to target-mediated drug disposition of rituximab in patients with DLBCL[2]. Similarly, Ternant et al. described nonlinear PK of rituximab in non-Hodgkin lymphomas with a two compartment model where target-mediated elimination was described as irreversible binding between rituximab and its target[3].
The objectives of this study were to evaluate rituximab pharmacokinetics in newly diagnosed patients with DLBCL to reveal and characterize nonlinear kinetics. We aimed to fit various nonlinear models, including time-dependent elimination rate approximated by exponential decrease in clearance (Model 1), concentration-dependent (Michaelis-Menten type) elimination (Model 2), target-mediated elimination with irreversible binding of rituximab to latent target and zero-order antigen input (Model 3), irreversible binding of rituximab to latent target with no description of target turnover (Model 4) as described by Ternant et al.[3] and quasi steady-state (QSS) approximation of the full TMDD model as described by Gibiansky et al. (Model 5) [4], [5].
Methods: We used the data from our previous study[2], where 29 newly diagnosed patients with DLBCL (median age 62 years, median body weight 74 kg) who received eight cycles of R-CHOP regimen every 3 weeks were included in the analysis. In total, 18 blood samples per patient were collected for the PK analysis (cycles 1-7: peak and trough serum concentrations of rituximab; cycle 8: peak, 1, 3 and 6 months after rituximab infusion). During follow-up (median: 52.9 months, range 9.7-66.3 months) 6 patients (21%) experienced disease progression. Population PK model was developed using NONMEM software. One- and two-compartment structural models were evaluated followed by inclusion of nonlinear elimination (Models 1-5). Comparison between models was based on the objective function value (OFV) and Akaike’s information criterion (AIC). In the final step we associated the mechanism of nonlinear pharmacokinetics with disease progression through covariate modelling.
Results: The data were best described by a two-compartment model (OFV 3997.953, AIC 4009.953). Introduction of IIV on CL and V1 improved both the OFV and AIC (3547.05 and 3563.05, respectively). Furthermore, nonlinear Models 1-5 with IIV on CL and V1 improved the model’s performance as follows: Model 1 (OFV 3536.21, AIC 3556.21), Model 2 (OFV 3531.364, AIC 3551.364), Model 3 (OFV 3535.765, AIC 3557.765), Model 4 (OFV 3535.768, AIC 3555.768) and Model 5 (OFV 3522.99, AIC 3546.99). Based on OFV, AIC and goodness-of-fit plots PK were most adequately described by Model 5. The final model was a two-compartment TMDD model with QSS approximation. The estimated clearance was 0.1766 L/day, volume of central compartment 4.15 L, volume of peripheral compartment 6.96 L, distribution clearance 1.279 L/day, target synthesis rate constant (ksyn) 4.656 nmol/(L×day), target degradation rate constant 5736 day-1, steady-state constant 3.81x10-4 nmol/L and internalization rate constant 1.682x10-2 day-1. IIV was introduced on CL, V1 and ksyn. A combination of additive and proportional error model was used for residual variability. Disease progression was associated with ksyn (p=0.026), in patients with disease progression ksyn was 63.7% higher compared to patients with no disease progression during the follow-up.
Conclusions: We demonstrated that PK of rituximab in DLBCL patients is nonlinear. Nonlinearity was described by a semi-mechanistic TMDD model applying quasi steady-state approximation. Ksyn was related with disease progression and could serve as a prognostic marker of rituximab response.
References:
[1] M. Tout et al., “Rituximab exposure is influenced by baseline metabolic tumor volume and predicts outcome of DLBCL patients: a Lymphoma Study Association report,” Blood, vol. 129, no. 19, pp. 2616–2623, May 2017, doi: 10.1182/blood-2016-10-744292.
[2] S. Rozman, I. Grabnar, S. Novakovic, A. Mrhar, and B. Jezeršek Novakovic, “Population pharmacokinetics of rituximab in patients with diffuse large B-cell lymphoma and association with clinical outcome,” Br J Clin Pharmacol, vol. 83, no. 8, pp. 1782–1790, Aug. 2017, doi: 10.1111/bcp.13271.
[3] D. Ternant et al., “Nonlinear pharmacokinetics of rituximab in non-Hodgkin lymphomas: A pilot study,” Br J Clin Pharmacol, vol. 85, no. 9, pp. 2002–2010, Sep. 2019, doi: 10.1111/bcp.13991.
[4] L. Gibiansky and E. Gibiansky, “Target-mediated drug disposition model: approximations, identifiability of model parameters and applications to the population pharmacokinetic-pharmacodynamic modeling of biologics,” Expert Opin Drug Metab Toxicol, vol. 5, no. 7, pp. 803–812, Jul. 2009, doi: 10.1517/17425250902992901.
[5] L. Gibiansky, E. Gibiansky, T. Kakkar, and P. Ma, “Approximations of the target-mediated drug disposition model and identifiability of model parameters,” J Pharmacokinet Pharmacodyn, vol. 35, no. 5, pp. 573–591, Oct. 2008, doi: 10.1007/s10928-008-9102-8.