ABSTRACT
A simultaneous pharmacokinetic (PK)/pharmacodynamic (PD) analysis was conducted to characterize relationships between anti‐B‐cell maturation antigen (BCMA)‐targeting antibody‐drug conjugate, belantamab mafodotin, and longitudinal multiple myeloma (MM) disease markers (soluble BCMA [sBCMA], M‐protein). Data from patients with relapsed/refractory MM receiving belantamab mafodotin monotherapy in the DREAMM‐1 (NCT02064387), DREAMM‐2 (NCT03525678), DREAMM‐3 (NCT04162210), or DREAMM‐5 (NCT04126200) trials were included. The dataset contained 510 patients with 15,860 PK and PD observations. Previous population PK and M‐protein PD models were combined to create a new starting model. Overparameterization was reduced using sensitivity analyses to create a 2‐compartment base model for belantamab mafodotin with a linear clearance component and target‐mediated drug disposition assuming quasi‐equilibrium between belantamab mafodotin and sBCMA binding. Free belantamab mafodotin affected M‐protein elimination rate via a direct effect, and M‐protein in turn scaled sBCMA synthesis rate. sBCMA was assumed to transfer into the central free sBCMA compartment through a reservoir compartment, allowing capture of observed differences in timescales. Stepwise covariate modeling led to the final model, which contained eight covariate effects (baseline sBCMA, albumin, body weight, and sex on belantamab mafodotin clearance; baseline body weight and sex on belantamab mafodotin central volume of distribution; extramedullary disease at screening and baseline sBCMA on M‐protein death rate constant). Simulations of belantamab mafodotin doses/schedules showed concentrations of free sBCMA and M‐protein decreased as free and total belantamab mafodotin increased with higher doses. The PKPD model captured the clearance of belantamab mafodotin and the highly dynamic interactions between belantamab mafodotin PK, M‐protein, and sBCMA.
Keywords: modeling, NONMEM, oncology, pharmacodynamics, pharmacokinetics, population pharmacokinetics‐pharmacodynamics
Study Highlights
- What is the current knowledge on the topic?
-
○Population pharmacokinetics (PK)/pharmacodynamics (PD) of anti‐B‐cell maturation antigen (BCMA) agent belantamab mafodotin have been assessed separately for patients with relapsed/refractory multiple myeloma (RRMM).
-
○
- What question did this study address?
-
○A simultaneous PKPD analysis was performed to characterize relationships between belantamab mafodotin monotherapy PK and disease markers (soluble BCMA [sBCMA], M‐protein) in RRMM.
-
○
- What does this study add to our knowledge?
-
○The model captured the highly dynamic interactions between belantamab mafodotin PK, sBCMA, and M‐protein.
-
○The base model was a 2‐compartment model with a linear clearance component and target‐mediated drug disposition assuming quasi‐equilibrium between belantamab mafodotin and sBCMA binding.
-
○Free belantamab mafodotin affected M‐protein elimination rate; in turn, M‐protein affected sBCMA synthesis rate.
-
○sBCMA was assumed to transfer into the central free sBCMA compartment through a reservoir compartment, allowing capture of different timescales observed in the data.
-
○
- How might this change drug discovery, development, and/or therapeutics?
-
○The model improves understanding of drivers of belantamab mafodotin monotherapy PK and efficacy.
-
○Model refinements/updates using data from combination regimens will aid in dose recommendations for future studies.
-
○
1. Introduction
Monoclonal (M)‐protein is a well‐established biomarker secreted by plasma cells in MM [1, 2]. Changes in serum and urine M‐protein measurements are used to inform clinical response assessments per International Myeloma Working Group (IMWG) consensus criteria, making it an important indicator of efficacy in MM clinical trials [1, 3]. B‐cell maturation antigen (BCMA) is an important prognostic marker for MM expressed on mature B cells and tumor cells [4], and plays a vital role in disease progression [5]. Once shed from the tumor cell surface, BCMA is cleaved into soluble BCMA (sBCMA), and circulating levels of sBCMA have been shown to correlate with tumor burden, treatment response, and serum M‐protein levels in patients with MM [5, 6, 7, 8].
Belantamab mafodotin is an antibody‐drug conjugate (ADC) that targets BCMA [4]. It is composed of a humanized anti‐BCMA IgG1 monoclonal antibody conjugated to the microtubule inhibitor monomethyl auristatin F via a protease‐resistant maleimidocaproyl linker (cysteine maleimidocaproyl monomethyl auristatin F) and leads to tumor cell destruction through cell‐cycle arrest and apoptosis, and antibody‐dependent cellular cytotoxicity and phagocytosis [4, 9]. It has been evaluated for the treatment of relapsed/refractory multiple myeloma (RRMM) in multiple clinical studies [10, 11, 12, 13, 14, 15, 16]. Belantamab mafodotin monotherapy was evaluated for RRMM in the first‐in‐human DREAMM‐1 (NCT02064387) trial, the pivotal Phase 2 DREAMM‐2 (NCT03525678) trial which investigated two doses of belantamab mafodotin [11, 17, 18], the pivotal Phase 3 DREAMM‐3 (NCT04162210) trial which investigated belantamab mafodotin versus standard of care [10], and the Phase 1/2 DREAMM‐5 (NCT04126200) platform trial which evaluates a belantamab mafodotin monotherapy control arm against belantamab mafodotin combined with various novel agents [12, 19, 20, 21, 22]. Belantamab mafodotin‐based regimens showed robust clinical efficacy in the Phase 3 DREAMM‐7 (NCT04246047; belantamab mafodotin in combination with bortezomib and dexamethasone versus daratumumab, bortezomib and dexamethasone) and DREAMM‐8 (NCT04484623; belantamab mafodotin in combination with pomalidomide and dexamethasone versus pomalidomide, bortezomib and dexamethasone) trials, leading to their approval in several countries including Europe, Japan, Canada, and Switzerland [14, 15, 16, 23, 24, 25, 26, 27].
Separate population pharmacokinetic (PopPK) and pharmacodynamic (PD) models have been developed for belantamab mafodotin [28, 29, 30, 31]. The PopPK analysis of belantamab mafodotin monotherapy in the Phase 1 DREAMM‐1 study and Phase 2 DREAMM‐2 study described belantamab mafodotin PK using a linear, two‐compartment model with empirically described time‐varying clearance; in this model, higher baseline sBCMA levels were associated with reduced belantamab mafodotin exposure [29]. This finding was further substantiated in a subsequent PopPK model of belantamab mafodotin monotherapy and combination therapy studies, which demonstrated that belantamab mafodotin PK is affected by baseline disease characteristics (sBCMA, albumin, serum IgG) and body weight [31]. Finally, the effect of belantamab mafodotin on M‐protein was investigated using data from DREAMM‐1 and DREAMM‐2 where a concentration‐driven tumor growth inhibition model described the time course of serum M‐protein concentration, with an effect compartment included to better describe the killing effect of belantamab mafodotin on M‐protein; M‐protein concentrations gradually decreased during treatment, and lower M‐protein levels were associated with slower tumor growth [30].
Understanding the relationship between belantamab mafodotin exposure, sBCMA, and serum M‐protein, using simultaneous semi‐mechanistic modeling of PKPD, may guide future clinical decision‐making by enhancing our understanding of belantamab mafodotin PK and MM disease markers at baseline and over time. This paper describes a simultaneous PKPD model in MM that demonstrates the longitudinal and dynamic relationship between belantamab mafodotin, sBCMA, and serum M‐protein.
2. Methods
2.1. Studies Included in the Models
Data were included from patients who received belantamab mafodotin monotherapy in the DREAMM‐1, DREAMM‐2, DREAMM‐3, and DREAMM‐5 studies. Full details of these studies have been published previously, and a summary of each study, including PK, sBCMA, and M‐protein sampling, is described in Table S1 [10, 11, 12, 13, 21, 22].
All studies included in the analysis were conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Protocols and informed consent forms were reviewed and approved by institutional review boards/independent ethics committees according to local guidelines. Patients in each study provided written informed consent.
2.2. Modeling and Simulation Analyses
Data from patients were included if there was ≥ 1 PD measurement (serum M‐protein or sBCMA) with an associated sampling time and a quantifiable baseline value of > 0.5 g/L serum M‐protein.
Plasma and serum samples were analyzed using validated analytical methods to assess total belantamab mafodotin concentrations (free and complexed with sBCMA), free and complex sBCMA concentrations, and M‐protein concentrations. Post‐baseline M‐protein values below the limit of quantification (BLQ) were imputed to the lowest value observed in the source dataset overall. This imputation resulted in more realistic model predictions. All other BLQ observations were excluded from the analysis.
NONMEM (v7.4.3) was used for PKPD modeling and further software details and model codes are provided in the Supporting Information. Stochastic approximation expectation maximization and importance sampling estimation methods were used for model development. Dependent variable data were log‐transformed prior to model run execution.
2.2.1. Base Model Generation
Parameter estimates from the previously published belantamab mafodotin PopPK analysis [29] and the M‐protein PD model [30] were used as starting points for the current analysis. A new starting model was obtained by combining these models with several adaptations, including the log‐transformation of observations with an additive residual error model on the log scale.
The interindividual random effects on the parameters were modeled using a log‐normal distribution as given by the following expression: θki = θk x eηki; where θki denotes the parameter value for a given patient, θk denotes the typical parameter value, and ηki denotes the interindividual random effect for the given patient, assumed to have a mean of 0 and a variance of ωk 2. Standard diagnostic plots were used to assess the ability of each model to describe the observed data. The new starting PKPD model was stabilized and simplified by excluding records that resulted in an absolute CWRES > 6 (or |CWRES| > 6) and using local and global sensitivity analyses to identify parameters with redundant effects or low impact on model outcomes that could be simplified, fixed, or removed, resulting in the base model [32, 33]. Local sensitivity analyses assessed the impact of changing one parameter at a time (OAT analysis), allowing for variations in the model outcome to be attributed to a specific parameter, while global sensitivity analyses varied all parameters simultaneously using the extended Fourier amplitude sensitivity test (eFAST) to account for interactions between parameters. Local and global sensitivity analyses were also conducted for parameters in the final PKPD model, and further details are provided in the Supporting Information.
2.2.2. Identification of Covariates
Baseline covariates were tested to determine their influence on belantamab mafodotin central clearance (CLADC), belantamab mafodotin central volume of distribution (VC), belantamab mafodotin peripheral volume of distribution (VP), M‐protein death rate constant (KDEATH), and M‐protein growth rate constant (KGR) (Table S2). Continuous covariates were incorporated into the population model via a power model using the scaled value of the covariate based on the median value in the population or a standard value to ensure that covariate effects were relative to an individual in the middle of the population distribution. All categorical covariates were incorporated into the population model using a proportional structure with the most common level of the covariate or a level relevant to the analysis being used as the reference. This ensured that categorical covariate effects were relative to a reference group or category. Further details are provided in the Supporting Information. Starting from the base model, iterative stepwise forward addition and backward elimination were performed, with significant covariates defined using the likelihood ratio test (forward addition significance level [α] < 0.01; backward elimination α < 0.001). Further details are provided in the Supporting Information. Covariate effects lacking clinical relevance were excluded from the final model. Following the stepwise covariate modeling, myeloma immunoglobulin type was tested on KDEATH, and IgG at baseline was tested on CL.
2.2.3. Model Finalization
After stepwise covariate modeling, the near‐final model was defined once the following criteria were met: there were no estimates close to a boundary; relative standard errors (RSEs) of the estimates were preferably < 30% for fixed‐effect parameters and < 50% for random‐effect parameters; and there were no unacceptable trends in goodness‐of‐fit plots. A final update to the model was made to allow total belantamab mafodotin, instead of free belantamab mafodotin, to diffuse to the peripheral compartment. Additionally, the initialization of the reservoir was set to the concentration of sBCMA at baseline, and the concentration of sBCMA was set to the central compartment only. It is this final model that is presented. Visual predictive checks were used to evaluate the predictive ability of the final model.
2.2.4. Simulations
A representative virtual population with covariates that were included in the final model was created using multivariate imputation by chained equations [34]. The final PKPD model was used to predict concentration–time profiles for free sBCMA, M‐protein, and belantamab mafodotin (free, complexed to sBCMA, and total [free and complexed]) using multiple doses/schedules including 30 mg, 100 mg, 200 mg, and 300 mg every other week (Q2W) and every 4th week (Q4W). Simulations are reported for up to 28 days.
3. Results
3.1. Patients
In total, 565 patients were identified; after exclusions, including four patients whose random effects could not be estimated, the total analysis dataset included data from 510 patients. Table 1 shows baseline covariate values and demographics for patients included in the analysis. The median (range) age was 66.0 (34.0–89.0) years and the median (range) baseline body weight was 74.7 (37.0–142) kg for the overall population. The majority of patients were male (54.1%) and White (80.4%).
TABLE 1.
Baseline covariate values and demographics for patients in the PKPD analysis.
| DREAMM‐1 (N = 45) | DREAMM‐2 (N = 216) | DREAMM‐3 (N = 215) | DREAMM‐5 (N = 34) | Overall (N = 510) | |
|---|---|---|---|---|---|
| Baseline age, years | |||||
| Median (range) | 60.0 (45.0, 79.0) | 66.0 (34.0, 89.0) | 68.0 (43.0, 86.0) | 67.5 (56.0, 82.0) | 66.0 (34.0, 89.0) |
| Baseline body weight, kg | |||||
| Median (range) | 77.3 (51.1, 112) | 73.3 (42.4, 130) | 74.0 (37.0, 142) | 79.1 (39.5, 109) | 74.7 (37.0, 142) |
| Baseline eGFR, mL/min/1.73 m2 | |||||
| Median (range) | 82.0 (41.0, 139) | 71.9 (13.7, 215) | 76.4 (16.1, 13,100) | 79.4 (31.4, 129) | 75.5 (13.7, 13,100) |
| Baseline albumin, g/L | |||||
| Median (range) | 37.0 (19.0, 44.0) | 39.0 (23.0, 48.0) | 38.0 (19.0, 52.3) | 39.5 (23.7, 46.0) | 38.0 (19.0, 52.3) |
| Baseline sBCMA, ng/mL | |||||
| Median (range) | 70.6 (11.0, 1000) | 88.3 (3.74, 2030) | 54.2 (2.08, 1180) | 52.2 (5.16, 769) | 72.9 (2.08, 2030) |
| Baseline M‐protein, g/L | |||||
| Median (range) | 22.0 (1.30, 51.0) | 18.0 (1.00, 92.0) | 18.0 (0.700, 102) | 18.0 (1.50, 96.0) | 18.0 (0.700, 102) |
| Baseline IgG, g/L | |||||
| Median (range) | 22.2 (0.300, 63.0) | 8.24 (0.350, 97.9) | 15.4 (0.510, 119) | 11.4 (0.350, 86.7) | 12.0 (0.300, 119) |
| Baseline B2M, nmol/L | |||||
| Median (range) | 254 (105, 1030) | 364 (102, 1970) | 313 (113, 1570) | 293 (139, 792) | 320 (102, 1970) |
| Sex, n (%) | |||||
| Male | 24 (53.3) | 118 (54.6) | 115 (53.5) | 19 (55.9) | 276 (54.1) |
| Female | 21 (46.7) | 98 (45.4) | 100 (46.5) | 15 (44.1) | 234 (45.9) |
| Race, n (%) a | |||||
| White | 37 (82.2) | 182 (84.3) | 165 (76.7) | 26 (76.5) | 410 (80.4) |
| Black or African American | 8 (17.8) | 28 (13.0) | 3 (1.4) | 2 (5.9) | 41 (8.0) |
| Asian | 0 (0) | 5 (2.3) | 47 (21.9) | 6 (17.6) | 58 (11.4) |
| Missing | 0 (0) | 1 (0.5) | 0 (0) | 0 (0) | 1 (0.2) |
| Baseline renal function category, a n (%) | |||||
| Normal | 16 (35.6) | 69 (31.9) | 62 (28.8) | 12 (35.3) | 159 (31.2) |
| Mild impairment | 19 (42.2) | 88 (40.7) | 96 (44.7) | 15 (44.1) | 218 (42.7) |
| Moderate impairment | 10 (22.2) | 53 (24.5) | 51 (23.7) | 7 (20.6) | 121 (23.7) |
| Severe impairment | 0 (0) | 5 (2.3) | 6 (2.8) | 0 (0) | 11 (2.2) |
| Myeloma immunoglobulin type, n (%) | |||||
| IgA | 9 (20.0) | 44 (20.4) | 48 (22.3) | 9 (26.5) | 110 (21.6) |
| IgG | 34 (75.6) | 168 (77.8) | 164 (76.3) | 19 (55.9) | 385 (75.5) |
| IgM | 1 (2.2) | 3 (1.4) | 0 (0) | 0 (0) | 4 (0.8) |
| IgA or IgG or IgM | 1 (2.2) | 1 (0.5) | 3 (1.4) | 6 (17.6) | 11 (2.2) |
Abbreviations: B2M, beta‐2 microglobulin; eGFR, estimated glomerular filtration rate; IgG, immunoglobulin G; PKPD, pharmacokinetic‐pharmacodynamic; sBCMA, serum B‐cell maturation antigen.
Missing values were imputed to the most common subcategory.
The analysis included 15,860 PK and PD observations (4586 total ADC, 4624 free sBCMA, 3228 M‐protein, and 3422 complex sBCMA observations), 14,829 of which were post‐first dose observations (Table S3).
3.2. PKPD Model Development
The new starting model (Figure S1) obtained after combining the previous PopPK and PD models was unstable. Local and global sensitivity analyses of the starting PKPD model identified several parameters that could be removed from the starting model to prevent overparameterization. Based on local sensitivity analyses, the following simplifications were made: (1) the direct effect of M‐protein on the elimination of belantamab mafodotin was removed, (2) the effect of M‐protein on free sBCMA synthesis was set equal to the current M‐protein divided by baseline M‐protein, (3) the effect of the resistance rate constant on M‐protein death was redundant with other parameters and removed, (4) the effect compartment was removed so that there was a direct effect of free belantamab mafodotin concentration on the M‐protein death rate. The elimination rate of the drug‐sBCMA complex (KINT) was assumed to be equal to the linear elimination rate of the free drug (KEL). A consequence of this simplification is that the total drug clearance is a linear, time‐invariant process, while the TMDD structure serves to model the dynamic partitioning between free and target‐bound drug. Based on global sensitivity analyses, variability in the rate (EMAI) and saturation (EMAS) of M‐protein effect on free belantamab mafodotin elimination were removed from the model. Models were highly sensitive to CLADC in both local and global sensitivity analyses. An example of local sensitivity analyses can be seen in Figure S2 and global sensitivity analyses are presented in Figure 1; both showed that the analytes were highly sensitive to changes in the parameters remaining in the final model.
FIGURE 1.
Heatmap of global sensitivity results for the final model. ADC, antibody‐drug conjugate (belantamab mafodotin); CLADC, ADC central clearance; KD, ADC‐sBCMA complex equilibrium dissociation constant; KDEATH, M‐protein death rate; KDEG, free sBCMA degradation rate; KDIFF, diffusion rate out of sBCMA reservoir; KGR, M‐protein growth rate; QA, ADC intercompartmental clearance; sBCMA, serum B‐cell maturation antigen; VC, ADC central volume of distribution; VP, ADC peripheral volume of distribution Sensitivity indices are color‐coded from gray (low sensitivity; 0–0.05) to red (high sensitivity: > 0.1) and represent variation in model output with respect to each parameter (rows), by first order (direct effect of each parameter), interactions (synergistic effect of two parameters) and total order (overall effect). Columns represent different timepoints post‐dose.
The resulting base model (Figure 2) had a 2‐compartment model structure with linear clearance and target‐mediated drug disposition (TMDD) of free belantamab mafodotin. TMDD assumed quasi‐equilibrium between rates of association and dissociation of the belantamab mafodotin‐sBCMA complex [35]. M‐protein was modeled with logistic growth. Free belantamab mafodotin affected M‐protein KDEATH through a direct effect, and M‐protein in turn affected the sBCMA synthesis rate. sBCMA synthesis was assumed to occur in a reservoir compartment, allowing the model to capture different timescales present in the data, and sBCMA subsequently transferred into the central free sBCMA compartment. Model equations for the base model are shown in Figure S3.
FIGURE 2.
Base PKPD model. Figure adapted from previous presentation in poster format the American Conference on Pharmacometrics 2024 in Phoenix, Arizona, USA. (1) KINT was forced to equal KEL (i.e., KINT = KEL). (2) Both free ADC and complexed ADC diffuse into the peripheral compartment. ADC, antibody‐drug conjugate (belantamab mafodotin); BCMA, B‐cell maturation antigen; CLADC, ADC central clearance; K12, rate constant from ADC central to peripheral compartment; K21, rate constant from ADC peripheral to central compartment; KD, ADC‐sBCMA complex equilibrium dissociation constant; KDEG, free sBCMA degradation rate constant; KDEATH, M‐protein death rate constant; KDIFF, rate constant for diffusion out of sBCMA reservoir; KEL, drug elimination rate constant; KGR, M‐protein growth rate constant; KINT, ADC‐sBCMA complex internalization rate constant; KSYN, sBCMA synthesis rate; QA, ADC intercompartmental clearance; sBCMA, serum B‐cell maturation antigen; VC, ADC central volume of distribution; VP, total ADC peripheral volume of distribution.
Parameter estimates for the base model are shown in Table S4. The base model was well estimated with RSEs < 30% and between‐subject variability (BSV) shrinkage < 30%. Goodness‐of‐fit plots (data not shown) indicated that the model adequately fit the observed data.
Parameter and covariate effect estimates for the final model are reported in Table 2. Allowing total belantamab mafodotin, instead of free belantamab mafodotin, to diffuse to the peripheral compartment lowered the peripheral volume estimates from ~30 to ~13 L. Nine covariate effects were found to be significant during forward addition, none of which were removed during backward elimination. These were baseline sBCMA, albumin, body weight, beta‐2 microglobulin, and sex on CLADC; baseline body weight and sex on VC; and presence of extramedullary disease at screening and baseline sBCMA on M‐protein KDEATH. The effect of baseline beta‐2 microglobulin on CLADC was subsequently removed due to a small effect and large RSE, and age and race did not have any significant impacts.
TABLE 2.
Parameter and covariate effect estimates from the final model.
| Parameter description | Final model | ||||
|---|---|---|---|---|---|
| Estimate | SE | RSE (%) | Bootstrap final model median [95% CI] | Shrinkage (%) | |
| Fixed effects | |||||
| CLADC: central clearance of total ADC (L/day) | 0.7 | 0.0359 | 5.12 | 0.716 [0.624, 0.788] | |
| VC: central volume of distribution of total belantamab mafodotin (L) | 4.81 | 0.0881 | 1.83 | 4.79 [4.61, 4.99] | |
| QADC: intercompartmental clearance of total belantamab mafodotin (L/day) | 0.771 | 0.0538 | 6.98 | 0.79 [0.709, 0.906] | |
| VP: peripheral volume of distribution of total belantamab mafodotin (L) | 12.7 | 1.15 | 9.06 | 12.3 [10.6, 16.2] | |
| KDEG: free sBCMA degradation rate constant (1/day) | 0.671 | 0.065 | 9.69 | 0.71 [0.54, 0.886] | |
| KD: ADC‐sBCMA complex equilibrium dissociation constant (nM/L) | 1.11 | 0.0427 | 3.85 | 1.1 [1.02, 1.19] | |
| KDEATH: M‐protein death rate constant (1/(nM**day)) | 0.00019 | 1.96E‐05 | 10.3 | 0.000193 [0.000149, 0.000233] | |
| KGR: M‐protein growth rate constant (1/day) | 0.00319 | 0.000548 | 17.2 | 0.00321 [0.00196, 0.00476] | |
| KDIFF: rate constant for diffusion out of sBCMA reservoir (1/day) | 4.45 | 0.425 | 9.55 | 4.73 [4.01, 6.22] | |
| Covariate effects | |||||
| Baseline albumin effect on CLADC | −0.879 | 0.157 | 17.8 | −0.867 [−1.18, −0.586] | |
| Baseline sBCMA effect on CLADC | 0.258 | 0.0347 | 13.5 | 0.244 [0.189, 0.319] | |
| Effect of female sex on CLADC | −0.126 | 0.0403 | 32.1 | −0.134 [−0.203, −0.0447] | |
| Body weight effect on CLADC | 0.488 | 0.0963 | 19.7 | 0.49 [0.24, 0.679] | |
| Baseline sBCMA effect on KDEATH | 0.318 | 0.0624 | 19.6 | 0.332 [0.177, 0.484] | |
| Effect of female sex on belantamab mafodotin VC | −0.136 | 0.0264 | 19.4 | −0.137 [−0.192, −0.0783] | |
| Body weight effect on belantamab mafodotin VC | 0.506 | 0.066 | 13.1 | 0.503 [0.359, 0.621] | |
| Extramedullary disease effect on KDEATH | −0.684 | 0.087 | 12.7 | −0.697 [−0.834, −0.455] | |
| Between‐subject variability | |||||
| BSV CV%: central clearance of total belantamab mafodotin (CLADC) | 40.5 | 0.058 | 38.1 | 38.5 [33.1, 53.6] | 18.5 |
| BSV CV%: central volume of distribution of total belantamab mafodotin (VC) | 25.5 | 0.00493 | 7.82 | 25.5 [22.5, 28.7] | 17.9 |
| BSV CV%: peripheral volume of distribution of total belantamab mafodotin (VP) | 135 | 0.0586 | 5.67 | 122 [80.4, 193] | 26.4 |
| BSV CV%: death of M‐protein (KDEATH) | 134 | 0.0779 | 7.58 | 133 [108, 166] | 27.4 |
| BSV CV%: growth of M‐protein (KGR) | 314 | 0.197 | 8.27 | 301 [209, 500] | 30.1 |
| Residual error—variance (additive on log scale) | |||||
| Free sBCMA | 0.424 | 0.0377 | 8.9 | 0.427 [0.353, 0.512] | |
| Complex sBCMA | 0.609 | 0.0504 | 8.28 | 0.599 [0.499, 0.704] | |
| Total belantamab mafodotin (in central) | 0.131 | 0.0142 | 10.8 | 0.132 [0.104, 0.161] | |
| Serum M‐protein | 0.108 | 0.0123 | 11.3 | 0.115 [0.0918, 0.143] | |
Abbreviations: ADC, antibody‐drug conjugate (belantamab mafodotin); BSV, between‐subject variability; CI, confidence interval; CV, coefficient of variation; RSE, relative standard error; sBCMA, serum B‐cell maturation antigen; SE, standard error.
All fixed‐effect and residual variability estimates were precisely estimated, with RSEs ≤ 21%, except for the covariate effects of sBCMA on CLADC (RSE 46.6%) and sex on CLADC (RSE 37.8%). BSV for belantamab mafodotin VP, M‐protein KDEATH, and M‐protein KGR had RSEs ≤ 10.2%. However, BSV on CLADC and belantamab mafodotin VC were not estimated precisely, as RSEs were ≥ 55.9%. Shrinkage of all BSVs was reasonable at ≤ 31.1%.
3.3. Final Model Diagnostics
Goodness‐of‐fit plots for the final model for free sBCMA, belantamab mafodotin‐sBCMA complex, total belantamab mafodotin, and M‐protein indicated that the model adequately fit the observed data (Figure S4). Plots of random effects were generally centered around zero with a symmetrical distribution for each parameter (Figure S5). The overall magnitudes of the covariate effects in the final model on total ADC for maximum and minimum concentration (Cmax and Cmin) following the first dose and at steady state are visualized in forest plots in Figure S6. Overall model performance was satisfactory as demonstrated in the prediction‐corrected visual predictive checks, which generally fit the data well (Figure 3), although free sBCMA amounts were slightly overpredicted, and correspondingly, belantamab mafodotin‐sBCMA complex amounts were slightly underpredicted.
FIGURE 3.
Prediction‐corrected visual predictive check of free sBCMA (A), belantamab mafodotin‐sBCMA complex (B), total belantamab mafodotin (C), and M‐protein (D) versus time in the final model, without observations. ADC, antibody‐drug conjugate (belantamab mafodotin); CI, confidence interval; sBCMA, serum B‐cell maturation antigen.
Across the studies, model predictions for belantamab mafodotin monotherapy treatment indicated that M‐protein levels most commonly decreased by < 25% from baseline, which was consistent with study observations (Figure 4).
FIGURE 4.
Model‐predicted and observed proportions of patients achieving different degrees of M‐protein reduction at nadir in the DREAMM‐1, DREAMM‐2, DREAMM‐3, and DREAMM‐5 studiesa. The middle bars represent the median percentage of participants taken from the percentage of participants from each of the 1000 replicates from the visual predictive checks. The bottom and top bars represent the 5th and 95th percentile for the percentage of participants taken from the percentage of participants from each of the 1000 replicates from the visual predictive checks. The dots represent the observed percentage of participants for each study and category. aCategories of M‐protein ranges are based on the plasma M‐protein portion of the diagnostic criteria set out in the 2016 International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma [1]. bCategory includes patients who only had an increase in M‐protein from baseline.
3.4. Simulation Analysis
A total of 1565 patients were included in the simulation dataset, including 1000 virtual patients simulated using multivariate imputation, which ensured the covariate distribution matched the population in the PKPD dataset, and 565 patients from the PKPD dataset.
Concentration–time profiles for the analytes, simulated using the final model, support the expected result that higher belantamab mafodotin doses, for both Q2W or Q4W dosing schedules, led to lower free sBCMA, lower serum M‐protein, and higher complex sBCMA median concentrations after the first dose (Figure 5). More frequent dosing for an equivalent dosing intensity resulted in substantially less change in belantamab mafodotin and sBCMA concentrations between dosing intervals, whereas dosing frequency had less of an impact on change in M‐protein concentrations.
FIGURE 5.
Median simulated belantamab mafodotin, sBCMA, and M‐protein values over time after the first dose of ADC, antibody‐drug conjugate (belantamab mafodotin); sBCMA, serum B‐cell maturation antigen.
4. Discussion
This analysis describes a simultaneous PKPD model to capture relationships between belantamab mafodotin PK, sBCMA, and M‐protein, enabling simulations that assessed the impact of different belantamab mafodotin dose levels on the PD analytes. The final PKPD model builds on the previous PopPK and PKPD analyses [29, 30], and offers a better understanding of the two‐way relationship between disease markers and PK over time.
Simplification of this model using local and global sensitivity analyses facilitated the development of a model that was stable and yielded parameter estimates with good precision. The model was also able to account for the elements of the individual PK and PD models [29, 30], and represented the clearance of belantamab mafodotin semi‐mechanistically. The current PKPD model effectively used TMDD‐associated clearance via internalization as well as belantamab mafodotin's impact on disease burden (direct effect on M‐protein and the subsequent impact on sBCMA levels) to describe clearance, thereby providing an improved understanding of belantamab mafodotin‐sBCMA interactions.
Baseline covariates included in the final PKPD model were broadly consistent with those included in the previous DREAMM‐1/DREAMM‐2 PopPK model [29]. This is also consistent with other BCMA‐targeted and non‐BCMA‐targeted therapies, where baseline covariates, including patient characteristics, past treatments, cytogenetic risk, and health/disease status, have been found to be significant predictors of exposure or response [36, 37, 38]. Baseline markers of disease burden (higher sBCMA, lower albumin) were associated with increased CLADC in both models. The previous M‐protein model included a modified weak Allee effect to account for tumor growth reduction at low M‐protein levels [30]. An Allee effect was evaluated in the current PKPD model, but was not included as the models became unstable. The diagnostics of the current final model were comparable to original models that included an Allee effect. In the previous model, covariates of sBCMA and extramedullary disease were associated with the rate constant for belantamab mafodotin transfer to effect compartment (KEO) [30]. An effect compartment KEO was not retained as a parameter in the current PKPD model after sensitivity analyses were performed and was replaced by the direct effect of free belantamab mafodotin on M‐protein; no covariates were associated with this parameter. However, higher baseline sBCMA led to increased M‐protein KDEATH and the presence of extramedullary disease reduced M‐protein KDEATH. This indicates that patients with higher target presence at baseline may experience a greater relative decline in tumor burden for a fixed free fraction of ADC, belantamab mafodotin, with the caveat that patients with higher sBCMA may also have faster clearance and a lower free fraction at a given dose level. Patients with extramedullary disease may experience less decline in tumor burden. Taken together, the covariate analyses retained important PK and PD associations and integration of the models, and adding data from the DREAMM‐3 and DREAMM‐5 studies revealed additional insights.
Simulations indicated that increasing belantamab mafodotin doses, regardless of a Q2W or Q4W schedule, led to higher free belantamab mafodotin concentrations, larger M‐protein reductions, and lower free sBCMA levels, suggesting the potential for greater efficacy with higher doses. They also showed that despite large fluctuations in sBCMA concentrations, the relatively long half‐life of belantamab mafodotin and relatively slow dynamics of M‐protein, a more frequent dosing schedule of Q2W compared to Q4W, may not be necessary for sustained M‐protein suppression if the belantamab mafodotin dose is 200 mg or higher. There was no ‘rebound’ observed within the 4‐week interval of time following a single 200 mg, or higher, dose. If greater M‐protein suppression than that achieved with a single 200 mg dose is desired in the first 28 days, then either a higher dose level could be used or the frequency of administration could be increased. Similarly, prior exposure‐response analyses have shown that higher Cycle 1 exposure to belantamab mafodotin increases the likelihood of deeper responses without an increase in the probability of the most clinically relevant safety events [28, 39, 40, 41]. Future exposure‐response analyses integrating data across the DREAMM studies are needed to further inform on the benefit–risk ratio of various belantamab mafodotin starting doses.
It is important to consider the limitations of this analysis. While BSV in the PKPD parameters was generally lower in the final model compared to the base model, uncertainty in some parameter estimates, as measured by RSE, remained high or increased in the final model. However, bootstrap analyses demonstrated narrow 95% confidence intervals, suggesting that these covariates were estimated with high precision, supporting their role as important predictors of PK. The differences between observed and model‐predicted measurements of free and complex sBCMA (i.e., residual errors) were relatively high, suggesting the potential for sources of variability not captured in the model, such as assay issues [42]. In addition, the analysis was limited by exclusion of patients whose disease was only monitored by serum free light chain, and not serum M‐protein. Future modeling analyses could assess external validation populations that include these patients to simulate M‐protein levels; this would enable evaluation of the accuracy of the model compared with actual responses. M‐protein sampling occurred only on Day 1 of belantamab mafodotin dosing cycles, so simulated dynamics for M‐protein within a dosing interval could not be verified [30]. Future inclusion of dropout modeling will enable longer‐term simulations. While this model could describe regrowth (captured using an equation set by the combined effects of tumor growth, drug effect, BSV, and actual dosing), it did not explicitly incorporate resistance to belantamab mafodotin over time or in a subset of cells, which would be valuable for future research.
This analysis was strengthened by the incorporation of data from a large number of patients across multiple studies and the use of diagnostic sensitivity analyses throughout model development. Longitudinal incorporation of belantamab mafodotin PK into a marker of response (M‐protein) improves the understanding of belantamab mafodotin efficacy. Further, simultaneous PKPD modeling allows parameter estimates to be less biased towards PK, as sequential modeling of PK followed by PD may inherently provide a better fit for PK data than PD data [43]; sequential models are also more sensitive to rerun with new data and may be less adaptive if restructure is needed [44]. The model showed good stability, enabling shorter run times, the simultaneous fitting of a large, integrated dataset, and robust covariate evaluation within a single framework. Other studies of belantamab mafodotin, including as part of combinations regimens, may inform further updates and refinements to this model, which will aid in dose recommendations for future studies such as those assessing coadministration of belantamab (monoclonal antibody alone) plus belantamab mafodotin [45, 46]. Additionally, the methodology of simultaneous and longitudinal modeling of PK and relevant biomarkers and disease marker analytes may be useful for other drugs and disease states.
5. Conclusion
The PKPD model captured the highly dynamic interactions between belantamab mafodotin PK, M‐protein, and sBCMA, further improving the understanding of the drivers of belantamab mafodotin PK and efficacy in the monotherapy setting.
Author Contributions
J.D.C., H.S., and G.F.‐B. designed the research, performed the research, and analyzed the data. I.C.S, J.K., C.N., A.T., M.S.Y., and X.C. performed the research and analyzed the data. All authors wrote the manuscript.
Funding
This study was funded by GSK (studies BMA117159, 205678, 207495, and 208887). Drug linker technology licensed from Seagen Inc.; monoclonal antibody produced using POTELLIGENT Technology licensed from BioWa.
Conflicts of Interest
I.C.S., C.N., and M.S.Y. were employees of Certara and provided consultancy to GSK at the time of this analysis. I.C.S. is a current employee of Merck & Co. Inc., C.N. is a current employee of Takeda Pharmaceutical Company Limited, and M.S.Y. is a current employee of Exelixis Inc. J.K. and A.T. are employees and stockholders of Certara and provide consultancy to GSK. J.D.C., X.C., H.S., and G.F.‐B., are employees of and hold financial equities in GSK.
Supporting information
Table S1: Summary of clinical studies included in the PKPK analysis.
Table S2: Baseline covariates evaluated in the PKPD model.
Table S3: Summary of PK and PD observations included and excluded from the PKPD analysis.
Table S4: Parameter estimates from the base model.
Figure S1: Starting PKPD model.
Figure S2: Local sensitivity analysis of CLADC for the final model.
Figure S3: Model equations for the base model.
Figure S4: Goodness‐of‐fit plots for free sBCMA (A), belantamab mafodotin‐sBCMA complex (B), total belantamab mafodotin (C), and M‐protein (D) in the final model.
Figure S5: Random effects in the final model.
Figure S6: Forest plots, derived using bootstrapped parameters, illustrating the effect of covariates on total ADC based on the final model.
Acknowledgments
Editorial support (in the form of writing assistance, including preparation of the draft manuscript under the direction and guidance of the authors, collating and incorporating authors' comments for each draft, assembling tables and figures, grammatical editing and referencing) was provided by Alexus Rivas‐John, PharmD, and Eithne Maguire, PhD, at Fishawack Indicia Ltd., part of Avalere Health, and was funded by GSK. We are grateful to the anonymous reviewers for their insightful suggestions, which strengthened the validity of our model.
Data Availability Statement
GSK makes available anonymized individual participant data and associated documents from interventional clinical studies that evaluate medicines, upon approval of proposals submitted to https://www.gsk‐studyregister.com/en/
References
- 1. Kumar S., Paiva B., Anderson K. C., et al., “International Myeloma Working Group Consensus Criteria for Response and Minimal Residual Disease Assessment in Multiple Myeloma,” Lancet Oncology 17 (2016): e328–e346. [DOI] [PubMed] [Google Scholar]
- 2. Murray D. L., Puig N., Kristinsson S., et al., “Mass Spectrometry for the Evaluation of Monoclonal Proteins in Multiple Myeloma and Related Disorders: An International Myeloma Working Group Mass Spectrometry Committee Report,” Blood Cancer Journal 11 (2021): 24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Yan X., Xu X. S., Weisel K. C., et al., “Early M‐Protein Dynamics Predicts Progression‐Free Survival in Patients With Relapsed/Refractory Multiple Myeloma,” Clinical and Translational Science 13 (2020): 1345–1354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Montes de Oca R., Alavi A. S., Vitali N., et al., “Belantamab Mafodotin (GSK2857916) Drives Immunogenic Cell Death and Immune‐Mediated Antitumor Responses In Vivo,” Molecular Cancer Therapeutics 20 (2021): 1941–1955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Shah N., Chari A., Scott E., Mezzi K., and Usmani S. Z., “B‐Cell Maturation Antigen (BCMA) in Multiple Myeloma: Rationale for Targeting and Current Therapeutic Approaches,” Leukemia 34 (2020): 985–1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ikeda D., Aikawa S., Misono C., et al., “Soluble B‐Cell Maturation Antigen Levels for Disease Monitoring in Oligosecretory and Nonsecretory Relapsed Multiple Myeloma,” Blood 145 (2025): 526–532. [DOI] [PubMed] [Google Scholar]
- 7. Tabata R., Oura M., Ikeda D., et al., “Soluble BCMA as a Surrogate Biomarker for Tumor Burden in Multiple Myeloma: Correlations With Various Biomarkers and Treatment Responses,” Blood 144 (2024): 3313. [Google Scholar]
- 8. Ghermezi M., Li M., Vardanyan S., et al., “Serum B‐Cell Maturation Antigen: A Novel Biomarker to Predict Outcomes for Multiple Myeloma Patients,” Haematologica 102 (2017): 785–795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Tai Y. T. and Anderson K. C., “Targeting B‐Cell Maturation Antigen in Multiple Myeloma,” Immunotherapy 7 (2015): 1187–1199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Dimopoulos M. A., Hungria V. T. M., Radinoff A., et al., “Efficacy and Safety of Single‐Agent Belantamab Mafodotin Versus Pomalidomide Plus Low‐Dose Dexamethasone in Patients With Relapsed or Refractory Multiple Myeloma (DREAMM‐3): A Phase 3, Open‐Label, Randomised Study,” Lancet Haematology 10 (2023): e801–e812. [DOI] [PubMed] [Google Scholar]
- 11. Lonial S., Lee H. C., Badros A., et al., “Belantamab Mafodotin for Relapsed or Refractory Multiple Myeloma (DREAMM‐2): A Two‐Arm, Randomised, Open‐Label, Phase 2 Study,” Lancet Oncology 21 (2020): 207–221. [DOI] [PubMed] [Google Scholar]
- 12. Nooka A. K., Weisel K., van de Donk N. W., et al., “Belantamab Mafodotin in Combination With Novel Agents in Relapsed/Refractory Multiple Myeloma: DREAMM‐5 Study Design,” Future Oncology 17 (2021): 1987–2003. [DOI] [PubMed] [Google Scholar]
- 13. Trudel S., Lendvai N., Popat R., et al., “Targeting B‐Cell Maturation Antigen With GSK2857916 Antibody‐Drug Conjugate in Relapsed or Refractory Multiple Myeloma (BMA117159): A Dose Escalation and Expansion Phase 1 Trial,” Lancet Oncology 19 (2018): 1641–1653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Dimopoulos M. A., Beksac M., Pour L., et al., “Belantamab Mafodotin, Pomalidomide, and Dexamethasone in Multiple Myeloma,” New England Journal of Medicine 391 (2024): 408–421. [DOI] [PubMed] [Google Scholar]
- 15. Hungria V., Robak P., Hus M., et al., “Belantamab Mafodotin, Bortezomib, and Dexamethasone for Multiple Myeloma,” New England Journal of Medicine 391 (2024): 393–407. [DOI] [PubMed] [Google Scholar]
- 16. Hungria V., Robak P., Hus M., et al., “Belantamab Mafodotin, Bortezomib, and Dexamethasone vs Daratumumab, Bortezomib, and Dexamethasone in Relapsed/Refractory Multiple Myeloma: Overall Survival Analysis and Updated Efficacy Outcomes of the Phase 3 Dreamm‐7 Trial,” Blood 144 (2024): 772. [Google Scholar]
- 17. Trudel S., Lendvai N., Popat R., et al., “Antibody–Drug Conjugate, GSK2857916, in Relapsed/Refractory Multiple Myeloma: An Update on Safety and Efficacy From Dose Expansion Phase I Study,” Blood Cancer Journal 9 (2019): 37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Nooka A. K., Cohen A. D., Lee H. C., et al., “Single‐Agent Belantamab Mafodotin in Patients With Relapsed/Refractory Multiple Myeloma: Final Analysis of the DREAMM‐2 Trial,” Cancer 129 (2023): 3746–3760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Richardson P. G., Biswas S., Holkova B., et al., “DREAMM‐5: Platform Trial Evaluating Belantamab Mafodotin (a BCMA‐Directed Immuno‐Conjugate) in Combination With Novel Agents in Relapsed or Refractory Multiple Myeloma (RRMM),” Blood 134 (2019): 1857. [Google Scholar]
- 20. Richardson P. G., Trudel S., Callander N. S., et al., “P‐212: DREAMM‐5 Platform Trial: Belantamab Mafodotin (Belamaf; GSK2857916) in Combination With Five Different Novel Agents in Patients With Relapsed/Refractory Multiple Myeloma (RRMM),” Clinical Lymphoma, Myeloma & Leukemia 21 (2021): S156. [Google Scholar]
- 21. Lonial S., Grosicki S., Hus M., et al., “Synergistic Effects of Low‐Dose Belantamab Mafodotin in Combination With a Gamma‐Secretase Inhibitor (Nirogacestat) in Patients With Relapsed/Refractory Multiple Myeloma (RRMM): DREAMM‐5 Study,” Journal of Clinical Oncology 40 (2022): 8019. [Google Scholar]
- 22. Paul C. N. R., Hus M., Ribrag V., et al., “P913: Low‐Dose Belantamab Mafodotin (Belamaf) in Combination With Nirogacestat vs Belamaf Monotherapy in Patients With Relapsed/Refractory Multiple Myeloma (RRMM): Phase 1/2 DREAMM‐5 Platform Sub‐Study 3,” HemaSphere 7 (2023). [Google Scholar]
- 23. GSK , “Blenrep (Belantamab Mafodotin) Combinations Approved by UK MHRA in Relapsed/Refractory Multiple Myeloma,” https://www.gsk.com/en‐gb/media/press‐releases/blenrep‐belantamab‐mafodotin‐combinations‐approved‐by‐uk‐mhra‐in‐relapsedrefractory‐multiple‐myeloma/.
- 24. European Medicines Agency , “Blenrep,” https://www.ema.europa.eu/en/medicines/human/EPAR/blenrep‐0#:~:text=On%2022%20May%202025%2C%20the,relapsed%20or%20refractory%20multiple%20myeloma.
- 25. GSK , “Blenrep (Belantamab Mafodotin) Combinations Approved in Japan for Treatment of Relapsed/Refractory Multiple Myeloma,” https://www.gsk.com/en‐gb/media/press‐releases/blenrep‐belantamab‐mafodotin‐combinations‐approved‐in‐japan/.
- 26. GSK , “Blenrep (Belantamab Mafodotin) Combinations Approved in Canada for the Treatment of Relapsed/Refractory Multiple Myeloma,” https://ca.gsk.com/en‐ca/media/press‐releases/blenrep‐belantamab‐mafodotin‐combinations‐approved‐in‐canada‐for‐the‐treatment‐of‐relapsedrefractory‐multiple‐myeloma/Accessed.
- 27. GSK , “GSK Announces Extension of US Food and Drug Administration Review Period for Blenrep (Belantamab Mafodotin‐Blmf) in Relapsed/Refractory Multiple Myeloma,” https://ca.gsk.com/en‐ca/media/press‐releases/blenrep‐belantamab‐mafodotin‐combinations‐approved‐in‐canada‐for‐the‐treatment‐of‐relapsedrefractory‐multiple‐myeloma/.
- 28. Ferron‐Brady G., Taylor A., Kaullen J., et al., “Population Pharmacokinetics (PopPK) and Exposure‐Response (E‐R) Analyses for Belantamab Mafodotin Monotherapy in Patients With Relapsed/Refractory Multiple Myeloma (RRMM) Enrolled in DREAMM‐2 and DREAMM‐3,” Blood 142 (2023): 6730. [Google Scholar]
- 29. Rathi C., Collins J., Struemper H., Opalinska J., Jewell R. C., and Ferron‐Brady G., “Population Pharmacokinetics of Belantamab Mafodotin, a BCMA‐Targeting Agent in Patients With Relapsed/Refractory Multiple Myeloma,” CPT: Pharmacometrics & Systems Pharmacology 10 (2021): 851–863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Collins J., van Noort M., Rathi C., et al., “Longitudinal Efficacy and Safety Modeling and Simulation Framework to Aid Dose Selection of Belantamab Mafodotin for Patients With Multiple Myeloma,” CPT: Pharmacometrics & Systems Pharmacology 12 (2023): 1411–1424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Papathanasiou T., Kaullen J., Polireddy K., et al., “Population Pharmacokinetics for Belantamab Mafodotin Monotherapy and Combination Therapies in Patients With Relapsed/Refractory Multiple Myeloma,” Clinical Pharmacokinetics 64 (2025): 925–942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Sorribes I. C., Handelman S. K., and Jain H. V., “Mitigating Temozolomide Resistance in Glioblastoma via DNA Damage‐Repair Inhibition,” Journal of the Royal Society Interface 17 (2020): 20190722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Jain H. V., Sorribes I. C., Handelman S. K., Barnaby J., and Jackson T. L., “Standing Variations Modeling Captures Inter‐Individual Heterogeneity in a Deterministic Model of Prostate Cancer Response to Combination Therapy,” Cancers 13 (2021): 1872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. van Buuren S. and Groothuis‐Oudshoorn K., “Mice: Multivariate Imputation by Chained Equations in R,” Journal of Statistical Software 45 (2011): 1–67. [Google Scholar]
- 35. Dua P., Hawkins E., and van der Graaf P. H., “A Tutorial on Target‐Mediated Drug Disposition (TMDD) Models,” CPT: Pharmacometrics & Systems Pharmacology 4 (2015): 324–337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Miao X., Wu L. S., Lin S. X. W., et al., “Population Pharmacokinetics and Exposure‐Response With Teclistamab in Patients With Relapsed/Refractory Multiple Myeloma: Results From MajesTEC‐1,” Targeted Oncology 18 (2023): 667–684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Mu W., Long X., Cai H., et al., “A Model Perspective Explanation of the Long‐Term Sustainability of a Fully Human BCMA‐Targeting CAR (CT103A) T‐Cell Immunotherapy,” Frontiers in Pharmacology 13 (2022): 803693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Srimani J. K., Diderichsen P. M., Hanley M. J., Venkatakrishnan K., Labotka R., and Gupta N., “Population Pharmacokinetic/Pharmacodynamic Joint Modeling of Ixazomib Efficacy and Safety Using Data From the Pivotal Phase III TOURMALINE‐MM1 Study in Multiple Myeloma Patients,” CPT: Pharmacometrics & Systems Pharmacology 11 (2022): 1085–1099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Carreño F., Kaisermann M. C., Egger J. L., and Ferron‐Brady G., “Population Pharmacokinetics and Exposure‐Response Analyses for Belantamab Mafodotin in Combination With Standard of Care in Transplant‐Ineligible Newly Diagnosed Multiple Myeloma From Dreamm‐9,” Blood 144 (2024): 1974. [Google Scholar]
- 40. Ferron‐Brady G., Rathi C., Collins J., et al., “Exposure‐Response Analyses for Therapeutic Dose Selection of Belantamab Mafodotin in Patients With Relapsed/Refractory Multiple Myeloma,” Clinical Pharmacology and Therapeutics 110 (2021): 1282–1292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Hanafin P., Ho Y. L., Papathanasiou T., et al., “Belantamab Mafodotin With Pomalidomide and Dexamethasone in Relapsed/Refractory Multiple Myeloma: A Comprehensive Exposure‐Response Analysis of the DREAMM‐8 Study,” Targeted Oncology 20 (2025): 833–845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Fairman D. and Tang H., “Best Practices in mAb and Soluble Target Assay Selection for Quantitative Modelling and Qualitative Interpretation,” AAPS Journal 25 (2023): 23. [DOI] [PubMed] [Google Scholar]
- 43. Karlsson M. O., “In the Cradle of Pharmacometric Methodology: Introducing Population PKPD Modeling, Simultaneous Analysis, and the Effect‐Compartment Model‐Commentary on Sheiner Et al,” Clinical Pharmacology & Therapeutics 117 (2025): 1516–1532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Zhang L., Beal S. L., and Sheiner L. B., “Simultaneous vs. Sequential Analysis for Population PK/PD Data II: Robustness of Methods,” Journal of Pharmacokinetics and Pharmacodynamics 30 (2003): 405–416. [DOI] [PubMed] [Google Scholar]
- 45. ClinicalTrials.gov , “A Study to Investigate the Safety and Efficacy of Belantamab for the Treatment of Multiple Myeloma When Used as Monotherapy and in Combination Treatments (DREAMM‐20),” https://clinicaltrials.gov/study/NCT05714839.
- 46. Quach H., Fulci G., Ratia N., et al., “P983: DREAMM‐20: A Study to Investigate the Safety and Efficacy of Belantamab for the Treatment of Multiple Myeloma When Used as Monotherapy and in Combination Treatments,” HemaSphere 7 (2023): e366180e. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Summary of clinical studies included in the PKPK analysis.
Table S2: Baseline covariates evaluated in the PKPD model.
Table S3: Summary of PK and PD observations included and excluded from the PKPD analysis.
Table S4: Parameter estimates from the base model.
Figure S1: Starting PKPD model.
Figure S2: Local sensitivity analysis of CLADC for the final model.
Figure S3: Model equations for the base model.
Figure S4: Goodness‐of‐fit plots for free sBCMA (A), belantamab mafodotin‐sBCMA complex (B), total belantamab mafodotin (C), and M‐protein (D) in the final model.
Figure S5: Random effects in the final model.
Figure S6: Forest plots, derived using bootstrapped parameters, illustrating the effect of covariates on total ADC based on the final model.
Data Availability Statement
GSK makes available anonymized individual participant data and associated documents from interventional clinical studies that evaluate medicines, upon approval of proposals submitted to https://www.gsk‐studyregister.com/en/