Correlation between protein-polydimethylsiloxane visible particle formation and computationally derived antibody...

Proteinaceous visible particles (pVPs) form through protein adsorption and aggregation at interfaces. Their formation risk varies widely among biopharmaceutical proteins. We developed a computational chemistry approach to quantitatively compare and predict the risk of protein-polydimethylsiloxane visible particles (psVPs)-a representative pVP type in poloxamer 188 (PX188)-based monoclonal antibody (mAb) formulations-across multiple mAbs by quantifying computationally derived surface charge and hydrophobic patches. First, we established an accelerated assessment system by lowering PX188 concentration, enabling rapid quantification of psVP formation risk for eight mAbs in siliconized polymer pre-filled syringes (PFS). Next, we quantified surface charge and hydrophobic patches on each mAb's molecular surface and assessed their association with psVP risk. Charge patches contributed more to psVP formation than hydrophobic patches in this system. A composite feature combining the total area of charge patches and the combined area of the top five hydrophobic patches showed the highest correlation with psVP formation among the evaluated features. Finally, we quantified the effect of reducing headspace in PFS as a mitigation strategy. These findings could be utilized for sequence and structure‑based prediction of psVP risk, supporting the design of lower‑risk mAbs and early decision‑making on mitigation strategies-such as headspace reduction-during biopharmaceutical development.