Antibody-Drug Conjugates (ADCs) are an exciting class of targeted anti-cancer therapeutics, combining the selectivity and specificity of biologics (monoclonal antibodies) with the potent cytotoxic activity of small molecule payloads. While proven to yield clinical benefit in different cancer types (5 ADCs have been approved by the FDA), many molecules fail in late stage clinical testing. The fine balance of anti-tumor activity vs. toxicity ultimately originates from the ADC ?design space?: the choices of target, backbone (usually monoclonal antibodies (mAb)), linker chemistry, cytotoxic payload, and drug-to-antibody ratio (DAR) make for a vast number of possible combinations that cannot be fully explored experimentally. ADCs are thus currently designed empirically, often based on variations of existing ADCs, supported by very limited and highly-imperfect pre-clinical assays, and clinical dosing schedules selected from sparse human toxicity data. Mechanism-based computational models that could synthesize the different preclinical mechanistic data to predict human efficacy and toxicity, and anticipate the therapeutic index (TI) of novel ADCs in silico would be highly valuable to guide both molecule design during early development, and clinical decisions. Specifically, if target selection and candidate screening could be performed computationally, better ADCs would be taken into clinical testing. Similarly, if the effect of alternate dosing schedules and patient populations could be evaluated pre-emptively, molecules that enter clinical testing would have a higher chance of success, trials would be accelerated, and clinical benefit would be improved. We propose developing a Quantitative Systems Pharmacology (QSP)-based platform ADC model that could do so - the ADC Workbench. By integrating the disparate body of data and biological knowledge available for successful ADCs into one platform model, the ADC Workbench will enable systematic candidate evaluation based on simulated clinical activity and toxicity (i.e., the TI). Leads with a poor chance of success will be weeded out early, and those with better prospects taken forward. The ADC Workbench will allow dosing schedules to be evaluated in large numbers of diverse virtual patient populations, providing a rational approach to clinical trial designs that maximize TI. The platform will be constructed in a modular way so that innovative new ADC molecules (e.g. with novel mAB backbones, linkers or payloads) can be incorporated as data becomes available. The ADC Workbench tool will be preloaded with several parameter sets for approved ADC molecules and their individual components (mAB, linker, payload), to allow for rapid in silico prototyping and benchmarking of potential new candidates. Continuous improvements to the built-in parameter database will be made as more data of clinical success and failure becomes available. Combining the model- and parameter database with the powerful high performance computing (HPC) analysis tools of Applied BioMath?s cloud based simulation engine will allow for routine and timely contribution to the ADC drug discovery process. 1 of 1
Antibody-Drug Conjugates (ADCs) are a promising class of anti-cancer therapeutics, with proven clinical benefit in a range of cancer types. However, the design of such molecules is highly complex, the pre-clinical experimental models imperfect, and unexpected toxicities often tank programs in clinical testing. The ADC Workbench is a Quantitative Systems Pharmacology (QSP)-based platform computational model that will integrate multiple data types and established knowledge to predict how the molecular properties of such agents relate to therapeutic index. This will be used in the de novo design of new molecules, lead candidate selection, and optimization of dosing schedules for ADCs in clinical development.