Towards a platform quantitative systems pharmacology (QSP) model for preclinical to clinical translation of antibody drug conjugates (ADCs)

Abstract

A next generation multiscale quantitative systems pharmacology (QSP) model for antibody drug conjugates (ADCs) is presented, for preclinical to clinical translation of ADC efficacy. Two HER2 ADCs (trastuzumab-DM1 and trastuzumab-DXd) were used for model development, calibration, and validation. The model integrates drug specific experimental data including in vitro cellular disposition data, pharmacokinetic (PK) and tumor growth inhibition (TGI) data for T-DM1 and T-DXd, as well as system specific data such as properties of HER2, tumor growth rates, and volumes. The model incorporates mechanistic detail at the intracellular level, to account for different mechanisms of ADC processing and payload release. It describes the disposition of the ADC, antibody, and payload inside and outside of the tumor, including binding to off-tumor, on-target sinks. The resulting multiscale PK model predicts plasma and tumor concentrations of ADC and payload. Tumor payload concentrations predicted by the model were linked to a TGI model and used to describe responses following ADC administration to xenograft mice. The model was translated to humans and virtual clinical trial simulations were performed that successfully predicted progression free survival response for T-DM1 and T-DXd for the treatment of HER2+ metastatic breast cancer, including differential efficacy based upon HER2 expression status. In conclusion, the presented model is a step toward a platform QSP model and strategy for ADCs, integrating multiple types of data and knowledge to predict ADC efficacy. The model has potential application to facilitate ADC design, lead candidate selection, and clinical dosing schedule optimization.

Keywords: Antibody drug conjugate; Enhertu; HER2; Kadcyla; Oncology; Quantitative systems pharmacology.

Fig. 1

Schematic representation of the ADC QSP model. The full QSP model consists of several submodels connected in a modular fashion. Submodels include a Tumor cell model, b Soluble target model, c Healthy cell model, d Tumor disposition model and e Tumor growth inhibition model. The full human model is shown. The in vitro cellular model (shown in Figure S1A) and the mouse model (shown in Figure S3) are simplifications of the human model. The model describes the disposition and PK of ADC, antibody and released payload in central, peripheral and tumor compartments. ADC is dosed in the central compartment, where it can deconjugate to release payload and naked antibody or distribute to peripheral and tumor compartments. The ADC and the Ab can bind to receptors expressed on healthy cells in the central and peripheral compartment, receptors expressed on tumor cells in the tumor compartment, and soluble receptors in central, peripheral, and tumor compartments. The ADC, antibody and payload are eliminated in central and peripheral compartments. The model incorporates a mechanistic characterization of tumor uptake, implemented using a Krogh cylinder model. In the tumor the ADC can deconjugate, ADC and Ab can bind to receptors on the surface of the tumor cell and be internalized into the endo/ lysosomal compartment. Here, intracellular bound ADC can either be recycled back to the cell surface, or the payload can be released either by linker cleavage (cleavable linkers) or via ADC degradation (non-cleavable linkers). Payload can escape into the cytosol and bind to its target or exit the cell. Free payload can also re-enter the cell. Tumor payload concentrations predicted by the model were linked to a model of TGI in xenograft mice. The state variables used in the model are described in Table S1, the model parameters are described in Table S2, and the reactions and resulting ordinary differential equations are shown in Table S3

Fig. 2

Workflow of ADC QSP model development and translational strategy, connecting in vitro, in vivo, and clinical data, models, and predictions. At each step, the model integrates the input data and knowledge of the biological mechanism to perform simulations and predictions which informs the next step in the strategy

Fig. 3

Observed (symbols) and in vitro cellular model calibrated (lines) T-DM1 in vitro disposition data. BT-474EEI, SK-BR-3 and MCF-7-neo/HER2 cells were incubated with T-[³H]DM1 for 2 h on ice (gray shaded area), washed, and intracellular and extracellular concentrations of DM1 catabolites were measured [16]. The in vitro cellular model was used to simulate these concentrations and compared to observed data

Fig. 4

In vivo mouse model calibration for T-DM1 and T-DXd. A Observed (symbols) and in vivo PK model calibrated (lines) DM1 disposition data. Following iv administration of T-[H]³-DM1 (300 µg/kg DM1 based dose) to BT-474EEI tumor bearing xenograft mice, levels of DM1 were measured in plasma and tumor [16]. Observed (symbols) and in vivo model calibrated (lines) tumor growth inhibition following IV administration of A T-DM1 to BT474-EEI, B T-DM1 to N87 and C T-DXd to N87 mouse xenograft studies. Model simulations of predicted payload concentrations in plasma and tumor (total and unconjugated), and model predicted TGI versus observed data are shown for T-DM1 and T-DXd [, , –29]

Fig. 5

Observed (symbols) and model predicted (lines) clinical PK of T-DM1 and T-DXd. Simulations of ADC, total antibody and released payload following IV administration of A T-DM1 to HER2+ MBC patients at 3.6 mg/kg, compared with clinical data from cycle 1 [31], and B T-DXd to HER2+ patients with breast, gastric or gastro-oesophageal carcinomas at 6.4 mg/kg, compared to data from cycles 1–3 [32]. Simulations of ADC concentrations in the same patient populations following IV administration of C T-DM1 from 0.3 to 4.8 mg/kg and D T-DXd from 0.8 to 8.0 mg/kg Q3W, with comparison to clinical data

Fig. 6

Model predictions of T-DM1 and T-DXd plasma PK, tumor payload concentrations and efficacy (tumor growth inhibition) using nominal patient parameters, following IV administration at 0.3–6.4 mg/kg. T-DM1 and T-DXd have very similar ADC plasma PK, but T-DXd delivers higher concentrations of payload to the tumor, one of the factors that could result in greater predicted efficacy compared to T-DM1 at the same doses

Fig. 7

Comparison of clinical trial simulations of progression free survival (PFS) with observed clinical trial data for T-DM1 and T-DXd. A T-DM1 PFS simulations (solid lines) and observed data (dotted lines) from a phase 2 clinical trial [39] following administration of T-DM1 to HER2+ metastatic breast cancer (MBC) patients at 3.6 mg/kg Q3W for 14 months. B T-DXd PFS simulations (solid lines) and observed data (dotted lines) from a phase 2 [40] and a phase 3 clinical trial [41] following administration of T-DXd at 5.4 mg/kg Q3W to HER2+ MBC patients for 29 months. For both T-DM1 and T-DXd, clinical simulations and data were stratified based upon HER2 high (1e6/cell) and HER2 low (2e4) expression levels in MBC patients