Thijs Zweers1, Eline van Maanen1, Sinziana Cristea1, Stephen Duffull1, Anne Chain2, Carolina de Miranda Silva2, Cindy Zhang2, Azher Hussain2, Yezhe Cheng3
1Certara, 2Merck & Co., Inc., 3Sichuan Kelun-Biotech Biopharmaceutical Co., Ltd.
Introduction: Antibody-Drug Conjugates (ADCs) are a promising class of targeted cancer therapies that combine the specificity of monoclonal antibodies (mAbs) with the potency of cytotoxic payloads. Sacituzumab tirumotecan (sac-TMT) is a TROP2-targeting ADC with a cytotoxic payload derived from the topoisomerase 1 inhibitor class. TROP2, a cell surface glycoprotein highly expressed in epithelial tumors, plays a critical role in promoting proliferation, invasion, and metastasis, making it a promising therapeutic target. TROP2-targeting ADCs selectively deliver cytotoxic effects to these tumors, demonstrating promising clinical activity [1,2]. The intricate pharmacokinetics (PK) of ADCs driven by the interplay of mAb, linker, and payload presents significant challenges in PK modeling. To address this, a semi-mechanistic simultaneous modeling approach was employed. The structural model represents the different drug-to-antibody ratio (DAR) species enabling an accurate description of ADC kinetics and comprehensive understanding of the PK for the 3 analytes: ADC, total mAb, and payload. The model was further utilized to guide the selection of tolerable dosing regimens, ensuring both efficacy and safety in therapeutic outcomes. Objectives: This analysis aims to characterize the PK of the ADC sac-TMT using a simultaneous population PK (popPK) model, leveraging total ADC, total mAb and payload data from 2 phase I/II clinical trials (NCT04152499, NCT05631262). It explores the impact of DAR distribution on PK and supports dose optimization to reduce adverse events. Methods: Data was collected in 663 subjects after multiple cycles of treatment with doses between 2 and 6 mg/kg (DCOs: 15-Jan-2024, 25-Feb-2024). A semi-mechanistic model was developed to characterize the PK of sac-TMT, which was simultaneously fitted to concentration data of total ADC (sum of all DAR species), total mAb (naked mAb + total ADC) and payload, by nonlinear mixed-effects modeling with NONMEM (v7.5.1). The modeling approach accounted for the DAR distribution in the dose. Model development included a covariate evaluation for the ADC, mAb and payload PK. Goodness-of-fit (GoF) criteria and prediction-corrected visual predictive checks (pcVPCs) were used to assess model performance across analytes. Results: The simultaneous modeling approach characterized the PK across analytes. The full model structure incorporated each DAR species ranging from the naked antibody (DAR0) to the ADC with the highest DAR (DAR9) by utilizing 10 interconnected identical model components each with a 2-compartment structure and linked to the payload, also described by a 2-compartment structure. The total ADC and total mAb data are key for distinguishing between the different types of payload release: through degradation of ADC and due to DAR species conversion. The incorporation of DAR distribution-based release of payload and measured payload concentrations enable the characterization of sequential transitions between DAR species, with each conversion releasing a single payload molecule through deconjugation. Inter-individual variability was estimated on clearance (CL) and volume of distribution (Vd) for ADC and payload. Statistically significant covariates (p <0.01) were identified on ADC and naked mAb: CL and Vd increased with bodyweight (estimated exponents 0.323 and 0.412, respectively) and CL increased with glomerular filtration rate (estimated exponent 0.367). Additionally, covariates were evaluated on payload but were not statistically significant. All parameters were estimated with good precision (RSE < 25%). GoF and pcVPCs showed that the model adequately describes the data and captures the median trend in the data and associated variability. However, the model is constrained by its assumptions and available data. It presumes that the payload release rate is independent of DAR. The DAR distribution in the dose product was assumed to be consistent across all dose levels. Additionally, the total ADC assay did not quantify DAR species in the PK data. Conclusion: This semi-mechanistic popPK model provides a comprehensive characterization of sac-TMT with multiple DAR species and payload release mechanisms. The model’s ability to characterize DAR transitions and degradation pathways enhances our understanding of sac-TMT PK and supports dose optimization to minimize adverse events in clinical development. Future studies could explore the application of this model to other ADCs.
[1] Crescioli, S. et al. (2024) mAbs 17(1). [2] Xu , B. et al. (2024) JCO 42, 104-104(2024)
Reference: PAGE 33 (2025) Abstr 11384 [www.page-meeting.org/?abstract=11384]
Poster: Drug/Disease Modelling - Oncology