Subcutaneous delivery, coupled with prefilled syringe and autoinjector technology, is increasingly desired as a preferred mode of administration for biological therapeutics, benefiting from convenience and compliance for patients. To deliver large doses, ultra-high concentration liquid formulations (>150 mg/mL) could be required. This presents a challenge, as only well-behaved biologics can be formulated at these extremely high concentrations. Therefore, it is vital that discovery pipeline candidates are screened, prior to development, for their feasibility in high concentration liquid formulations. To ensure that we do not encounter the drawback of high viscosity, we have developed a workflow to predict antibody viscosity at high concentrations using descriptors derived from static antibody structures that are modeled against a dataset of in-house antibodies for which viscosity is known.
Learning Objectives:
Upon completion, participants will be able to understand the workflow for generating in silico descriptors
Upon completion, participants will be able to understand the machine learning prediction of viscosity from descriptors
Upon completion, participants will be able to understand where this process fits in the drug development pipeline
