Brett Matzuka 1, Jason Chittenden 1, Anasuya Hazra 1, Lutz Harnisch 1
1 Regeneron Pharmaceuticals (Tarrytown, USA)
Introduction:
Linvoseltamab is a BCMAxCD3 bispecific antibody approved for triple-class exposed relapsed/refractory (RR) multiple myeloma (MM) after 3 therapies (EU) or 4 prior lines of therapy (USA) (Bumma et al, 2024; Lee et al, 2026). Linvoseltamab competes with endogenous IgG, which is elevated in a large subset of MM patients, for FcRn mediated salvage – resulting in increased clearance and lower exposure for these patients. As patients respond to treatment, free light chains (FLC) decrease, implying a reduction in tumor burden, which reduces the secretion of sBCMA, and IgG decreases, which can result in a large, response-induced chainge in clearance. The implication is that the PK of linvoseltemab can be heavily influenced by the PD response, which affects drug clearance and free drug concentration.
The pharmacokinetic properties of linvoseltamab were assessed, based upon data from the pivotal Phase1/2 Linker-MM1 study, in serum from 281 subjects to construct a two-compartment population pharmacokinetic model with parallel linear and nonlinear clearances. A covariate analysis found a time-dependent relationship between immunoglobulin G (IgG) and albumin on linear clearance and concentration-dependent relationship between free light chain (FLC) and nonlinear clearance. These covariates showed strong effects on exposure metrics and were correlated with patient response. The use of these covariates to describe PK, however, presents a problem in that the response to treatment must be known a priori to predict PK, yet the PK drives the response.
Objectives:
Develop a semi-mechanistic population pharmacokinetic-pharmacodynamic (PKPD) model incorporating pertinent biomarkers, soluble BCMA (sBCMA) and FLC, observed to affect response long term, and address the exposure-response-exposure relationship
Methods:
This analysis included 340 subjects and 14,632 observed serum concentrations from three studies involving RRMM (281 subjects, 7275 linvoseltamab concentrations, 2895 sBCMA concentrations, and 3004 FLC concentrations), newly diagnosed multiple myeloma (NDMM) (43 subjects, 535 linvoseltamab concentrations, 314 sBCMA concentrations, and 145 FLC concentrations), and high risk smoldering multiple myeloma (HR-SMM) (16 subjects, 211 linvoseltmab concentrations, 179 sBCMA concentrations, and 74 FLC concentrations). Serum linvoseltamab, sBCMA, and FLC concentrations were fit simultaneously utilizing nonlinear mixed effect modeling (NONMEM version 7.5) to develop a PopPKPD model. Covariates were included in the model based upon visual inspection, improvement in model fit, and scientific and clinical relevance.
Results:
Linvoseltamab was characterized by a 2-compartment model with linear and nonlinear clearance. Linvoseltamab was assumed to be in rapid equilibrium with sBCMA and membrane BCMA (mBCMA) with binding calculated by solving the quadratic derived from the dissociation constant with the amount of complex bound to sBCMA versus mBCMA proportional to the amount of each target present. mBCMA was modeled by logistic growth, mutation rate into a parallel logistic growth compartment unaffected by binding, and death term linked to bound complex. sBCMA, FLC, and IgG were all characterized by a first order growth, proportional to susceptible and resistant mBCMA, as well as a constant synthesis term. Free sBCMA was eliminated by a first order process derived from a half-life of 1 day and sBCMA complex was eliminated by the same process as linvoseltamab. Free sBCMA was also created from the mBCMA kill process and was transported into the sBCMA compartment through transit compartments related to the killing process. FLC was eliminated by a first order process derived from a half-life of 4 hours and IgG was eliminated by a first order process, proportional to the ratio of current IgG versus a typical value in a power relationship. The linear clearance of linvoseltamab and sBCMA complex is scaled by the IgG amount with a power relationship. FLC at baseline, sBCMA at baseline, multiple myeloma subtype, and multiple myeloma disease type (RRMM, NDMM, HRSMM) were covariates included in the model.
Visual Predictive check and goodness of fit plots displayed a good agreement between model predicted and observed data. Chemical progression free survival (cPFS) was calculated from FLC model predictions based upon a SLiM criterion for multiple myeloma and simulations were conducted to compare different dose and treatment regimens, comparing cPFS.
Conclusion:
Incorporating the sBCMA and FLC data allowed the PopPKPD model to encapsulate the treatment/disease dynamics. This model decoupled the feedback mechanisms that confounded the exposure-response-exposure relationship and can be utilized to simulate different treatment regimens, doses, and indications.
References:
Bumma N, Richter J, Jagannath S, Lee HC, Hoffman JE, Suvannasankha A, Zonder JA, Shah MR, Lentzsch S, Baz R, Maly JJ, Namburi S, Pianko MJ, Ye JC, Wu KL, Silbermann R, Min CK, Vekemans MC, Munder M, Byun JM, MartÃnez-Lopez J, Cassady K, DeVeaux M, Chokshi D, Boyapati A, Hazra A, Yancopoulos GD, Sirulnik LA, Rodriguez Lorenc K, Kroog GS, Houvras Y, Dhodapkar MV. Linvoseltamab for Treatment of Relapsed/Refractory Multiple Myeloma. J Clin Oncol. 2024 Aug 1;42(22):2702-2712.
Hans C. Lee, Jeffrey A. Zonder, Madhav V. Dhodapkar, Sundar Jagannath, James E. Hoffman, Attaya Suvannasankha, Mansi R. Shah, Suzanne Lentzsch, Rachid Baz, Joseph J. Maly, Swathi Namburi, Matthew J. Pianko, Jing Christine Ye, Ka Lung Wu, Rebecca Silbermann, Chang-Ki Min, Marie-Christiane Vekemans, Markus Munder, Ja Min Byun, JoaquÃn MartÃnez-Lopez, Michelle DeVeaux, Tito Roccia, Dhruti Chokshi, Megan Seraphin, Kate Knorr, Anita Boyapati, Anasuya Hazra, Karen Rodriguez Lorenc, Glenn S. Kroog, Naresh Bumma, Joshua Richter. Linvoseltamab in Patients With Relapsed/Refractory Multiple Myeloma in the LINKER-MM1 Study: Longer Follow-Up and Subgroup Analyses. Clinical Lymphoma Myeloma and Leukemia, Volume 26, Issue 2, 2026,Pages e201-e212.
Reference: PAGE 34 (2026) Abstr 12114 [www.page-meeting.org/?abstract=12114]
Poster: Drug/Disease Modelling - Oncology