(T1130-03-16) Rational Design of Viscosity-Reducing Formulation Compositions for High-Concentration Protein Solutions

Purpose: Design of formulation compositions for protein therapeutics requires consideration of the delivery method, drug substance properties, and manufacture processes. Protein-based drugs exhibit high specificity for their targets and long half-lives but delivery routes such as subcutaneous injection often require developing low volume drug products. To ensure delivery of the efficacious dose, developers must design compositions that reduce liabilities that could negatively impact product quality, safety, and manufacturability such as solution viscosity. At high concentrations, proteins exhibit exponential increases in viscosity, which pose injectability and manufacturing challenges. Proteins exhibit high viscosity with increasing protein concentration reflecting increased protein-protein interactions (PPIs) and crowding effects due to high volume fractions of the protein solute. Existing formulation composition approaches for decreasing protein solution viscosity rely on empirical assessments of excipients’ effects on viscosity. Previously identified viscosity reducing excipients do not always work for new molecules of interest, requiring excipient screening for every molecule. Unfortunately, this approach is inefficient and increases the time required for formulation composition determination. Here, we report our efforts to identify, understand, and connect molecular properties to protein solution behavior to design formulation compositions.

Methods: A homology model of an internal monoclonal antibody (mAb) was used to calculate molecular descriptors using MOE for a range of pH and ionic strength conditions. Viscosity reducing excipient candidates were identified from previously published literature. mAb ultrafiltration product (UFP) was produced and used to generate high concentration mAb in Histidine and Sodium Phosphate. Concentrated mAb was added to a 96-well plate and concentrated stocks of excipients were added to dilute the mAb. Sample pH was measured, and volume was removed for viscosity measurements. Viscosity was measured using an UnChained Labs Honeybun at a range of shear rates. Samples showed Newtonian behavior and reported viscosities are the average across the tested shear rates.

Results: Viscosity measurements and antibody structural modelling demonstrate the interplay of protein protein and protein-excipient interactions for effective formulation composition design. Modeling shows that mAb has a strong formulation dependent zeta potential, being positive only at low ionic strength and low pH. There was not any discernible difference in charge separation between the Fv and Fc domains as a function of pH and ionic strength. The molecular dipole moment increased with increases in pH but not ionic strength. The number of positive patches decreased with increasing pH but did not appear to vary meaningfully with ionic strength. The mAb has clear hydrophobic patches. In combination with the hydrophobic patches, zeta potential, and molecular dipole data, structural modelling suggests that neither charge nor hydrophobicity alone are the mAb’s source of high viscosity but rather a combination of the two.Viscosity measurements of mAb at high concentrations show high viscosity i.e., above 20 centipoise (cP) in Histidine and even higher in sodium phosphate. Many excipients added to both the histidine formulation and the sodium phosphate formulation decreased sample viscosity relative to the control samples without any added candidate excipients. Figure 1. The magnitude of viscosity reduction varies in a manner somewhat consistent with the excipient class. Some excipients, notably polyelectrolytes, increase sample viscosity. The candidate excipients were identified based on their chemistries to identify if the viscosity reducing effects were shared among a given kind of chemistry. Some general trends exist but there are differences in the magnitude of viscosity reduction within a given class.

Conclusion: While this excipient list was selected to prioritize diversity of chemistries, the corresponding result is a variety of changes to sample viscosity suggesting a wide number of considerations are necessary for modulating sample viscosity. For example, polymers increasing sample viscosity suggest crowding plays an important role while simple inorganic salts decreasing viscosity suggests electrostatic repulsion matters even in a crowded environment. Qualitatively comparing these viscosity data to the mAb’s properties demonstrate the link between protein properties and viscosity. These findings enable future rational excipient selection and demonstrate the utility of expanding the toolbox of available excipients outside of the standard handful currently relied upon. Future studies should quantitatively analyze the relationship between excipient properties, protein properties, and viscosity reduction. Further mechanistic data describing the origin of these excipients’ viscosity reducing effects will better enable rational excipient selection for formulation design.

Acknowledgements: The Merck Postdoctoral Fellowship program funded this work.


Figure 1. Viscosity data for excipients screened against mAb in Histidine buffer at high concentration. Bars are colored according to the chemistry of the excipient. For experimental samples, N=1. The black dashed line corresponds to the control sample average viscosity (N=4).