This study evaluates the ability of the Conductor-like Screening Model for Real Solvents (COSMO-RS) to predict the solubility of naproxen (NAP) in three in-house synthesized poly(2-oxazoline)s (POx)-poly(2-ethyl-2-oxazoline) (PEtOx), poly(2-n-butyl-2-oxazoline) (PBuOx), and poly(2-phenyl-2-oxazoline) (PPhOx)-and to assess the effect of their side-chain structures. Solid-liquid equilibrium data were obtained by differential scanning calorimetry using the step-wise dissolution method. For each NAP-POx system, two COSMO-RS-based SLE curves were generated: one combining the drug with a CH3-capped COSKF trimer, and one combining the drug with a more realistic CH3/OH-terminated pentamer. Pentamer-based calculations provided the most reliable thermodynamic description and correctly reproduced the experimental compatibility ranking (PEtOx > PBuOx > PPhOx) when PBuOx was treated as amorphous, consistent with its crystallinity loss in NAP-rich mixtures. Absolute average deviations ranged from 4.3 to 11.9 wt% for pentamers, compared with 8.3 to 19.5 wt% for COSKF trimers, reflecting the improved accuracy of more realistic polymer models. Sigma-profile analysis offered qualitative insight into polarity and potential interactions, complementing the quantitative predictions. Overall, COSMO-RS effectively differentiates closely related POx excipients in early formulation screening.
Zemánková et al. (Sat,) studied this question.