Chemometric Evaluation of Near Infrared, Fourier Transform Infrared, and Raman Spectroscopic Models for the Prediction of Nimodipine Polymorphs

Abstract

The objective of this study was to assess the performance of the chemometric model to predict the proportion of the recrystallized polymorphs of nimodipine from the cosolvent formulations. Ranging from 100% to 0% (w/w) of polymorph I, the two polymorphs mixtures were prepared and characterized spectroscopically using Fourier transformed infrared spectroscopy (FTIR), near-infrared spectroscopy (NIR), and Raman spectroscopy. Instrumental responses were treated to construct multivariate calibration model using principal component regression (PCR) and partial least square regression approaches. Treated data showed better model fitting than without treatment, which demonstrated higher correlation coefficient (R(2) ) and lower root mean square of standard error (RMSE) and standard error (SE). Multiple scattering correction and standard normal variate exhibited higher R(2) and lower RMSE and SE values than second derivative. Goodness of fit for FTIR and NIR (R(2) ∼ 0.99) data was better than Raman (R(2) ∼ 0.95). Furthermore, the models were applied on the recrystallized polymorphs obtained by storing nimodipine-cosolvent formulations at selected stability conditions. The relative composition of the polymorphs differed with storage conditions. NIR-chemical imaging on recrystallized sample of nimodipine at 15°C qualitatively corroborated the model-based prediction of the two polymorphs. Therefore, these studies strongly suggest the importance of the potential utility of the chemometric model in predicting nimodipine polymorphs.