Claire Couty: No financial relationships to disclose
Objectives: The development of new treatments for hematological cancers faces many challenges and choosing the optimal drug regimen requires careful considerations. Indeed, the various lymphoid malignancies present distinct biological behaviors in terms of tumor proliferation, immune response and escape, and heterogeneous responses to treatment. A QSP platform could be a game-changer by allowing to explore virtually various scenarios and provide valuable insights to clinical research. We developed a mechanistic QSP platform describing different lymphoid malignancies: classical Hodgkin lymphoma (CHL), T-cell and B-cell non-Hodgkin lymphomas (TCL, BCL). Each of these disease models can be coupled with treatment models (pharmacokinetics, mechanism of action). The resulting overall model can be used to explore trial design questions such as optimizing the administration regimen (e.g. dose, frequency). As a use case, we applied this QSP platform to an antibody drug conjugate (ADC) targeting a CDX protein differentially expressed on CHL and TCL cancer cells. We ran in silico clinical trials on virtual populations to identify best treatment regimens in terms of efficacy, and characterize best responders.
Methods: We developed a modular QSP platform based on published knowledge and data, implemented as a system of ordinary differential equations. In the ADC use case: - Model was calibrated on public in vitro and human data. and a validation step was performed on PK data. - Simulations and analyses were performed on Jinkō [1].
Results: The final QSP platform integrates: - a mechanistic submodel for each disease of CHL, TCL or BCL, which describes the proliferation of malignant cells, the tumor microenvironment and the intratumor heterogeneity. - a PBPK submodel that predicts the concentration of drugs such as antibodies in the human body. - a submodel accounting for the mechanism of action of the ADC: binding to the target and its internalization, and subsequent release and effect of the payload. The model captures the observed in vitro cytotoxicity, PK dynamics, and human clinical response to treatment. In silico trials, performed in the context of the example use case, allowed to characterize best responders, and optimal dosing regimens in CHL and TCL. It highlighted the importance of both dose and frequency, with higher frequency particularly beneficial at lower doses, for achieving optimal efficacy.
Conclusion: This work paves the way towards a QSP platform for blood cancers as a powerful tool to support decision-making in drug development. One key aspect is the ability to progressively integrate a large spectrum of data and knowledge from a diverse set of diseases and treatments. Beyond optimizing trial design and drug regimens, this platform can also provide efficacy estimates for a given lead across multiple indications. Prospective validation of the platform's predictions will be crucial for its integration into drug development pipelines.
