Analysis of residual moisture in a freeze-dried sample drug using a multivariate fitting regression model

Abstract

Listen

Translate article

The moisture content of many pharmaceutical products, like most vaccines, is an important control parameter in the product qualification. Freeze-drying is a widely applied technique for the preservation and storage of such products. Karl Fischer titration is the standard method for determination of residual moisture in freeze-dried samples; however, it is a time-consuming and destructive method that requires the handling of organic solvents. Instead, near-infrared (NIR) spectroscopy is a fast and non-invasive method that is well suited to the measurement of water due to the strong adsorption band of water in NIR. In this study, multivariate fitting regression with Gaussian function (MFRG) was applied to predict the water content of a freeze-dried sample drug using NIR spectra. Moreover, the prediction performance of the MFRG was compared to the results of the well-known PLS method. Several simulated datasets and experimental data were used to evaluate the performance of the MFRG method. In the case of experimental data, a number of mathematical pre-treatments were applied to evaluate whether they led to better models. In addition, several calibration and test sets were designed and applied to compare the quantification results of MFRG and PLS methods. It was found that the performance of the MFRG model was comparable or better compared to the PLS regression method in nearly all the examined cases, which is due to fact that MFRG models nonlinearity better than the PLS method.