Validation of a multipoint near-infrared spectroscopy method for in-line moisture content analysis during freeze-drying

• Published in: Journal of pharmaceutical and biomedical analysis 2014-07, Vol.95, p.229-237
• Main Authors: Kauppinen, Ari, Toiviainen, Maunu, Lehtonen, Marko, Järvinen, Kristiina, Paaso, Janne, Juuti, Mikko, Ketolainen, Jarkko
• Format: Article
• Language: English
• Subjects: Accuracy profile; Calibration; Freeze Drying; Least-Squares Analysis; Moisture; Multipoint near-infrared spectroscopy; Spectroscopy, Near-Infrared - methods; Validation; Water - analysis
• ISSN: 0731-7085; 1873-264X
• DOI: 10.1016/j.jpba.2014.03.008

•In-line moisture content analysis method for freeze-drying process was validated.•Method was based on multipoint NIR spectroscopy and quantitative PLS modeling.•Validation was performed using accuracy profile concept.•Method resulted in valid moisture predictions at the latter stages of the process.•Method can be utilized in pharmaceutical freeze-drying research and development. This study assessed the validity of a multipoint near-infrared (NIR) spectroscopy method for in-line moisture content analysis during a freeze-drying process. It is known that the moisture content affects the stability of a freeze-dried product and hence it is a major critical quality attribute. Therefore assessment of the validity of an analytical method for moisture content determination is vital to ensure the quality of the final product. An aqueous sucrose solution was used as the model formulation of the study. The NIR spectra were calibrated to the moisture content using partial least squares (PLS) regression with coulometric Karl Fischer (KF) titration as the reference method. Different spectral preprocessing methods were compared for the PLS models. A calibration model transfer protocol was established to enable the use of the method in the multipoint mode. The accuracy profile was used as a decision tool to determine the validity of the method. The final PLS model, in which NIR spectra were preprocessed with standard normal variate transformation (SNV), resulted in low root mean square error of prediction value of 0.04%-m/v, i.e. evidence of sufficient overall accuracy of the model. The validation results revealed that the accuracy of the model was acceptable within the moisture content range 0.16–0.70%-m/v that is specific for the latter stages of the freeze-drying process. In addition, the results demonstrated the method's reliable in-process performance and robustness. Thus, the multipoint NIR spectroscopy method was proved capable of providing in-line evaluation of moisture content and it is readily available for use in laboratory scale freeze-drying research and development.