Hoai-Thu Thai1, Geraldine Ayral3, Jean-Baptiste Fau2, Helgi van de Velde4, Dorothee Semiond2, Christine Veyrat-Follet1
1 Translation Disease Modeling, Digital and Data Science, Sanofi, France 2 Translational Medicine and Early Development, Sanofi, Paris, France 3 Lixoft, Antony, France 4 Sanofi Genzyme, US
Objectives:
Isatuximab (Isa) is an anti-CD38 monoclonal antibody with multiple modes of action for killing tumor cells through direct tumor targeting and immune cell engagement (1). The addition of Isa to pomalidomide (P) and dexamethasone (d) was associated with a significant and clinically meaningful benefit in progression-free survival (PFS) in heavily pre-treated patients with relapsed/refractory multiple myeloma (RRMM) (ICARIA-MM, NCT02514668) (2). The aim of this work was to characterize the effects of treatment and baseline covariates on serum M-protein kinetics and the link to PFS in RRMM patients in both arm of the ICARIA-MM trial using a joint modeling framework.
Methods: Model building was conducted using 256 evaluable patients from the ICARIA-MM trial. Patients received Isa intravenously at 10 mg/kg once weekly for 4 weeks followed by every other week for 28-day cycles in combination with standard Pd or Pd alone in the control arm. Longitudinal data of serum M-protein and PFS data were first analyzed separately and then in a joint model to characterize the link between serum M-protein and PFS. Different models for the relationship between serum M-protein kinetics and PFS were considered: no link, current M-protein, current M-protein slope and current M-protein AUC. Covariate analysis were also performed in each submodel to examine the influence of baseline covariates on serum M-protein kinetics and hazard rate of PFS. The significance of baseline covariate in the submodels were then confirmed in the joint model. Parameters were estimated by maximizing the exact joint likelihood with the SAEM algorithm implemented in Monolix 2019R1. Model selection was based on log-likelihood ratio tests and BIC while VPC and Kaplan-Meier plots were used to evaluate model performance. Simulation of typical patients were also performed to evaluate the impact of significant covariates.
Results:
The best longitudinal model for serum M-protein involved a two-compartment PK submodel for Isa with parallel linear and nonlinear elimination and time-dependence of the linear clearance (3), a K-PD submodel for Pd dosing data and a tumor growth inhibition model for serum M-protein data with treatment effect driven by exposure of the three drugs. A log-logistic model best characterized the underlying baseline hazard distribution of time-to-PFS and the model using the current M-protein slope value outperformed all models in terms of BIC. The significant baseline covariates in the joint model were Ig MM type, albumin (ALB), beta-2 micro-globulin (B2MG) and presence of plasmacytoma. Patient with non-IgG type myeloma had faster tumor growth rate compared to IgG MM type patients. Patients with high baseline ALB and low B2MG levels were more likely to have higher serum M-protein at baseline. Of note, these laboratory tests are part of the International Staging System (ISS) and the revised ISS and are relevant for the assessment of prognosis. The presence of plasmacytoma likely induced lower PFS probability, consistent with the results of exposure-response analysis (4). The simulation of typical patients indicated that non-IgG MM patients have similar behavior on serum M-protein kinetics for the first 30 weeks even with higher steady state exposure to Isa but tend to have a more rapid tumor regrowth afterwards and a shorter PFS compared to IgG MM patients. The simulated median PFS were in agreement with those observed in both reference (Pd) and treatment (Isa) arms.
Conclusions: The joint model well characterized the time course of serum M-protein, the treatment effect and the observed PFS of patients in the ICARIA-MM trial. It also helped to identify the important covariates on serum M-protein kinetics and PFS. By linking the full time-course of serum M-protein to PFS, this model is expected to provide reliable predictions of clinical outcomes when performing model-based simulations of new clinical trials.
References:
[1] Moreno, L. et al. The Mechanism of Action of the Anti-CD38 Monoclonal Antibody Isatuximab in Multiple Myeloma. Clin Cancer Res 25, 3176-87 (2019).
[2] Attal, M. et al. Isatuximab plus pomalidomide and low-dose dexamethasone versus pomalidomide and low-dose dexamethasone in patients with relapsed and refractory multiple myeloma (ICARIA-MM): a randomised, multicentre, open-label, phase 3 study. Lancet 394, 2096-107 (2019).
[3] FAU J-B et al. Pharmacokinetic time-dependency and covariates modelling of Isatuximab monoclonal antibody in multiple myeloma patients: analysis from pooled phase I/II and phase III studies. Abstracts of the Annual Meeting of the Population Approach Group in Europe PAGE 28 (2019) Abstr 8956 (2019).
[4] Rachedi, F. et al. Exposure-Response Analyses and Disease Modeling for Selection and Confirmation of Optimal Dosing Regimen of Isatuximab in Combination Treatment in Patients with Multiple Myeloma Blood Blood (2019) 134 (Supplement_1): 1897, (2019).
Reference: PAGE () Abstr 9491 [www.page-meeting.org/?abstract=9491]
Poster: Drug/Disease Modelling - Oncology