Establishment of a Indomethacin-Saccharin Raman Based PLS Model for the Quantification of Cocrystal Content

Cocrystal Content N. A. Heron, Y. Tian, D. S. Jones, G. P. Andrews Queen's University of Belfast Purpose The solubility of a drug within the gastrointestinal tract has a direct impact upon its oral bioavailability. Numerous strategies have successfully enhanced drug solubility; however, these approaches are unreliable as long term solutions due to physical instability of the drug form. Crystal engineering via cocrystallization allows for the optimization of physicochemical properties of the drug; in particular solubility and improved physical stability [1]. Cocrystals incorporate pharmaceutically acceptable neutral guest molecules (coformers) into the crystal lattice of the API, most commonly via hydrogen bonding [2]. This work uses indomethacin (IND) and saccharin (SAC) to form cocrystal that has improved solubility to the parent drug. Several characterization techniques have been used to identify IND-SAC cocrystal; produced from both solvent and non solvent methods. The focus of this work is to establish a Raman based PLS model to quantify the content of IND-SAC cocrystal in the presence of their original components [3]. This model will further be used to quantify cocrystal content engineered via solvent free, hot melt extrusion. Methods Solvent Evaporation: An equimolar (1:1) mixture of IND (1g) and SAC (0.519g) was added to 14-15ml of ethyl acetate in a 25ml conical flask. The solution was allowed to evaporate slowly over 48 hours in a controlled fume hood (temperature 22°C, air flow 0.54m/s) to produce cocrystals. The product was characterised by differential scanning calorimetry (DSC), powder x-ray diffraction (PXRD), attenuated total reflectance infrared (ATR-IR) spectroscopy and Raman spectroscopy. Results obtained indicated high pureness of the cocrystal and as a result was used as a standard INDSAC cocrystal. IND-SAC cocrystal quantification: 10 calibration samples with a ternary mixture design were used providing six levels of variation. Therefore, each sample contained different amounts of IND, SAC and cocrystal. Samples were weighed (total of 500mg) and mixed in an oscillatory ball mill mixer (Retsch, MM200, Germany) with six small stainless steel balls, to ensure maximum mixing of the materials. Raman Spectroscopy: Spectra were collected on an Avalon Instruments Raman Station R3 Raman Spectrometer (Avalon Instruments Ltd., Belfast, Northern Ireland, UK). Analysis was conducted over a range from 200 to 3200cm-1. PLS model: constructed using SIMCA 14 analysis software. Results Model 1 (untreated, raw data) for cocrystal calibration used 6 components in the analysis and had 99.6% (R2X=0.996) explained variance for X variables and 92.3% (R2Y=0.923) explained variance for Y variables. R2 is the percent of variation of the training set explained by the model, i.e. how well the model fits the data. A large R2 close to 1 is an indication of a good model. Q2 is the percent of variation of the training set predicted by the model according to cross validation. In model 1 a Q2 of 88.6% (0.886) was found, potentially as a result of noise, or outliers within the data. Model 2 was pre treated with an MSC (multiplicative signal correction) filter to give each “corrected” spectrum the same offset and amplitude and as a result used 2 components in the analysis. The improved model has yielded 97% (R2X=0.97) and 94% (R2X=0.94). Q2 was also increased to 93.4% (0.934) improving the predictability of the model. The root mean square error of cross validation (RMSECV) for cocrystal from model 1 is 10% variation and from model 2 is 8% variation. This confirms that model 2 has better predictability as less variation is seen within the data compared to the untreated data in model 1. Conclusion A successful Raman based calibration PLS model has been produced that has relative errors around 8% prediction of cocrystal variability. Development of the model is continuing to further improve the prediction. Validation of this model is also to follow as well as quantification of cocrystal yield as a result of hot melt extrusion.