Sarah Lobet (1), Gilles Paintaud (3,4), Thierry Lecomte (1, 2), David Ternant (3,4)
(1) Tours University, Inserm UMR 1069, Nutrition Croissance et Cancer (N2C), Tours, France ; (2) Department of Gastroenterology and Digestive Oncology, CHRU de Tours, Tours, France; (3) CHRU de Tours, Plateforme Recherche, Centre Pilote de suivi Biologique des traitements par Anticorps (CePiBAc), Tours, France; (4) Tours University, EA4245 Transplantation, Immunologie, Inflammation, Tours, France.
Introduction: Monoclonal antibodies (mAbs) exposure-response relationship in oncology is complex and associated with an important inter-individual variability [1–3]. Several confounding factors [4,5], including baseline disease status and time-varying phenomena, such as target engagement and disease improvement (i.e. tumor growth and cancer-related cachexia), may bias the description of this relationship and should be considered. This work aimed to investigate the variability of cetuximab (CTX) and bevacizumab (BV) exposure-response relationships in metastatic colorectal cancer patients, while mitigating potential sources of bias.
Methods: The relationship between mAbs concentrations, target kinetics and time-to-progression (TTP) was described using joint pharmacokinetic-TTP models in two clinical studies. Three potential relationships between pharmacokinetics and TTP were considered: (i) no relationship between mAb concentration or target kinetics and hazard to progression («dummy»-link model), (ii) a relationship between increasing mAb concentration and decreasing baseline hazard function (C-link model) or (iii) a relationship between decreasing unbound target level and decreasing baseline hazard function (R-link model). Parameters were estimated using non-linear mixed-effects modeling and the Monolix software [6], following a sequential approach [7].
Several strategies were applied to mitigate the “prognostic-driven” and “response-driven” bias on the exposure-response relationship [4,5]. Firstly, the joint kinetics of these mAbs and their target were described using target-mediated drug disposition (TMDD) modeling to mitigate the potential target engagement bias. Then, the addition of a time-varying clearance term to the TMDD model’s first-order clearance term and mutual influences of clearance (CL) and TTP were investigated to mitigate disease activity bias.
Results: Concentration-time data for CTX and BV were well described using quasi-steady-state approximation. Complexes eliminations (kint = 0.95 day-1 for CTX and 0.52 day-1 for BV) is faster than endogenous eliminations (ke = 0.17 day-1 for CTX and 0.04 day-1 for BV) for both drugs. Baseline target levels (R0) were 43 nM for CTX and 8.4 nM for BV. An increase in R0 for CTX was not significantly associated with baseline tumor markers (i.e. carcinoembryonic antigen CEA). Conversely, it was associated with an increase of these markers (i.e. CEA, circulating VEGF and presence of extrahepatic metastases) for BV.
The addition of a time-varying clearance component improved the performance of the TMDD QSS model for CTX (Δ AIC = -18.26), but not for BV (Δ AIC = +6.02). Estimated time-varying clearance parameters for CTX were baseline CL (CL0= 0.38 L/day-1), time-variant component of CL (CL1= 0.058 L/day-1) and first-order rate of CL1 decreasing over time (kdes = 0.049 day-1).
Unbound target levels (R) significantly influenced progression hazard functions for both drugs. Indeed, the joint TMDD-TTP model with R-link led to better performance than the « dummy »-link (ΔAIC = − 11.93 for CTX and ΔAIC = -6.37 for BV). Models assuming an influence of mAb concentrations on hazard led to inaccurate TTP parameter estimates (RSE > 60%). Higher R0 led to lower unbound mAb concentrations, higher R levels and shorter risk to progression.
A part of the relationship between target and TTP was TTP-driven, with CL and TTP inversely correlated. Shorter covariate TTP values were associated with higher CL values (p=0.016 for CTX and p=0.054 for BV) in the absence of any baseline tumor biomarker. However, the addition of the influence of the baseline tumor biomarkers on R0 compensated the influence of TTP on CL for BV (p=0.45). In addition, higher CL values were significantly associated with a higher risk of progression for both drugs (p=0.0000067 for CTX and p=0.047 for BV)[8,9].
Conclusion: This work revealed the complexity of CTX and BV exposure-response relationships and highlighted the need for bias mitigation strategies. Late response (progression) to mAbs treatment was linked to treatment exposure, target engagement, and changes in clinical condition. Patients with high target expression (high R0) and a clinical condition associated with high proteolytic activity (high CL) might benefit from optimized dosing.
References:
[1] Paci A, Desnoyer A, Delahousse J, Blondel L, Maritaz C, Chaput N, et al. Pharmacokinetic/pharmacodynamic relationship of therapeutic monoclonal antibodies used in oncology: Part 1, monoclonal antibodies, antibody-drug conjugates and bispecific T-cell engagers. European Journal of Cancer. 1 mars 2020;128:107-18.
[2] Bensalem A, Ternant D. Pharmacokinetic Variability of Therapeutic Antibodies in Humans: A Comprehensive Review of Population Pharmacokinetic Modeling Publications. Clin Pharmacokinet. juill 2020;59(7):857-74. [3] Wang Y, Booth B, Rahman A, Kim G, Huang SM, Zineh I. Toward greater insights on pharmacokinetics and exposure-response relationships for therapeutic biologics in oncology drug development. Clin Pharmacol Ther. mai 2017;101(5):582-4. [4] Kawakatsu S, Bruno R, Kågedal M, Li C, Girish S, Joshi A, et al. Confounding factors in exposure–response analyses and mitigation strategies for monoclonal antibodies in oncology. British Journal of Clinical Pharmacology. 2021;87(6):2493-501. [5] Dai HI, Vugmeyster Y, Mangal N. Characterizing Exposure-Response Relationship for Therapeutic Monoclonal Antibodies in Immuno-Oncology and Beyond: Challenges, Perspectives, and Prospects. Clin Pharmacol Ther. déc 2020;108(6):1156-70. [6] Monolix version 2020R1. Antony, France: Lixoft SAS, 2020. http://lixoft.com/products/monolix/ [7] Kerioui M, Bertrand J, Bruno R, Mercier F, Guedj J, Desmée S. Modelling the association between biomarkers and clinical outcome: An introduction to nonlinear joint models. British Journal of Clinical Pharmacology. 2022;88(4):1452-63. [8] Lobet S, Paintaud G, Azzopardi N, Passot C, Caulet M, Chautard R, et al. Relationship Between Cetuximab Target-Mediated Pharmacokinetics and Progression-Free Survival in Metastatic Colorectal Cancer Patients. Clin Pharmacokinet. sept 2023;62(9):1263-74. [9] Lobet S, Caulet M, Paintaud G, Azzopardi N, Desvignes C, Chautard R, et al. Confounding mitigation for the exposure-response relationship of bevacizumab in colorectal cancer patients. British Journal of Clinical Pharmacology [Internet]. [cité 14 déc 2023];n/a(n/a). Disponible sur: https://onlinelibrary.wiley.com/doi/abs/10.1111/bcp.15983
Reference: PAGE 32 (2024) Abstr 10778 [www.page-meeting.org/?abstract=10778]
Poster: Drug/Disease Modelling - Oncology