Purpose: To develop resource-sparing in silico approaches that aim to reduce experimental effort and material required by developability assessments (DA) of monoclonal antibody (mAb) drug candidates.
Methods: A battery of standardized biophysical experiments was performed on high concentration formulations of 16 drug product development stage mAbs using a platform buffer. Full-length molecular models of these mAbs were also generated via molecular modeling. These models were used to computationally estimate molecular descriptors of these 16 mAbs. Pairwise and multi-parameter correlations among experimentally measured biophysical attributes and calculated molecular descriptors were obtained via statistical analyses.
Results: Diffusion interaction parameter (kD) showed statistically significant pairwise correlations (p-values <0.005) with thermal stability, viscosity, isoelectric point, and apparent solubility of the antibodies in our dataset. kD also showed statistically significant pairwise correlations (p-values <0.005) with several computationally calculated molecular descriptors (pI, net charge, charge on the Fv region, and zeta potential.) These pairwise correlations were further refined by multivariate analyses. These analyses yielded several useful equations for prediction of kD from antibody sequences, structural models, and experimentally measured biophysical attributes.
Conclusions: Diffusion interaction parameter (kD) was found to be a key biophysical property for the mAbs in our dataset. It connects conformational heterogeneity of an antibody with its colloidal and rheological behaviors. The equations derived in this work shall enable rapid, resource-sparing, and cost-effective DAs of biologic drug candidates.
Keywords: biopharmaceutical informatics; diffusion interaction parameter; high concentration; molecular modeling; monoclonal antibody.