Molecular Pharmaceutics, 2019, vol 16, 10, pp. 4319-4338
Excipients are substances that are added to therapeutic products to improve stability, bioavailability, and manufacturability. Undesirable protein-protein interactions (PPI) can lead to self-association and/or high solution viscosity in concentrated protein formulations that are typically greater than 50 mg/mL. Therefore, understanding the effects of excipients on nonspecific PPI is important for more efficient formulation development. In this study, we used National Institute of Standards and Technology monoclonal antibody (NISTmAb) reference material as a model antibody protein to examine the physical stability and viscosity of concentrated formulations using a series of excipients, by varying pH, salt composition, and the presence of cosolutes including amino acids, sugars, and nonionic surfactants. Small angle X-ray scattering (SAXS) together with differential scanning calorimetry (DSC), dynamic light scattering (DLS), and viscosity measurements were used to obtain various experimental parameters to characterize excipient modulated PPI and bulk solution viscosities. In particular, a good correlation was found between SAXS and DLS/SLS results, suggesting that the use of DLS/SLS is valid for predicting the colloidal stability of NISTmAb in concentrated solutions. Moreover, further analysis of effective structure factor S(q)eff measured from SAXS enabled the dissection of net PPI into hydrodynamic forces due to excluded volume as well as any additional attractive or repulsive interactions with the presence of excipients. It was found that although no denaturation or aggregation of NISTmAb was observed and that the net PPI were repulsive, the use of ionic excipients such as pH and salts leads to increased short-range attraction, whereas the nonionic excipients including sugars, amino acids, and polysorbate surfactants lead to increased repulsive PPI with increasing protein concentration. Results obtained from viscosity measurements showed that the use of excipients can lead to increased solution viscosities at high protein concentrations. The use of S(q)eff, interaction parameter kD, and second virial coefficient B22 as predictors for solution viscosity was also evaluated by comparing the predicted results with the measured viscosities. Although B22 and S(q)eff appeared to be better predictors than kD, disagreement between the predicted and measured results suggests other factors apart from PPI contribute to the bulk rheological properties of concentrated protein solutions.