Jones J A, Last I R, MacDonald B F, Prebble K A
Analytical Development Laboratories, Burroughs Wellcome Co., Greenville, NC 27835-1887.
J Pharm Biomed Anal. 1993 Nov-Dec;11(11-12):1227-31. doi: 10.1016/0731-7085(93)80108-d.
Near-infrared reflectance (NIR) methods for non-invasive and non-destructive measurement of moisture in a lyophilized product were developed independently at two sites. NIR spectra were collected with the same model instrument at each site. The spectra were scanned through the bases of unopened glass vials using a horizontal instrument accessory. The primary reference data for moisture content were generated using Karl-Fischer titration on the same individual vials. The NIR calibration equations were developed with second derivative spectral data using regression programs within the NIR software. These calibration equations were validated using independent test data from additional vials of product from the original site, and cross tested against similar data from the other site to check their robustness. This cross comparison demonstrated that a calibration equation from one site could satisfactorily be used for predicting moisture contents in product manufactured at the other site. Expected variations arising from differences in vial material, manufacturing process, analysts and instruments appeared to be satisfactorily accommodated. Finally, two test data sets were obtained by scanning a set of samples on two spectrometers at one site. The results from these data sets were comparable using the calibration equation developed solely on one of the two instruments. The results from all calibration and test sets are presented and discussed, and an assessment of the method transferability between instruments and sites is given.
