Assessment of mefenamic acid polymorphs in commercial tablets using chemometric coupled to MIR and NIR...

Mefenamic Acid (MFA) is a widely-used non-steroidal anti-inflammatory drug. MFA presents four possible crystal forms; Form I and Form II being the only two pure crystals that have been isolated and fully characterized. Both Form I and Form II were prepared following the literature and completely characterized by middle (MIR) and near (NIR) infrared spectroscopy, digital optical microscopy, differential scanning calorimetry, melting point and dissolution properties. In order to develop quantitative models to assess Form I in formulated products, two sets of samples, training (n=10) and validation (n=8) sets, were prepared by mixing both polymorphs and the matrix of excipient (simulating commercial tablets). The particle size of the samples was homogenized by sieving and samples were mechanically mixed. A batch of commercial tablets was gently disaggregated, sieved and mechanically mixed for further analysis. For each sample, full MIR and NIR spectra were acquired and used as input of partial least squares (PLS) algorithm separately. Method optimization and internal validation were performed by leave one out procedure. Full spectra and 5 PLS-factors were used for MIR; while, 5 PLS-factors and mean center spectra of full spectra were the optimal conditions for NIR. Accuracy and precision were assessed by evaluation of the actual vs. predicted curve of validation set; and by calculating validation set recoveries and deviations (104.3±8.2% and 100.4±1.0% for MIR and NIR respectively). Only NIR-PLS yielded acceptable results and low deviations during commercial samples evaluation (102.8±0.1%). The same behavior was observed when spiked tablets were analyzed (103.5±0.5%). Additionally, for the calibration set ten dissolution profiles (average of 6 curves each), were obtained under optimized test conditions (900 ml of buffer phosphate pH 9 with surfactant, apparatus II USP, 100rpm, detection at 342nm). A multiple linear regression (MLR) was carried out using dissolution profiles and Form I content. The developed MLR model could correlate dissolution profiles and polymorphic richness. This approach, coupled to previously developed NIR-PLS, may act as a valid tool to estimate dissolution profiles from solid forms.