Evaluation of the Robustness of A Novel NIR-based Technique to Measure the Residual Moisture In Freeze-dried Products

Abstract

(Bio)pharmaceutical products freeze-dried in vials must meet stringent quality specifications: among these, the residual moisture (RM) is crucial. The most common techniques adopted for measuring the RM are destructive, e.g. Karl Fisher titration, thus few samples from each batch are tested. Being a high intra-batch variability an intrinsic feature of batch freeze-drying, a high number of samples needs to be tested to get a representative measurement. Near-Infrared (NIR) spectroscopy was extensively applied in the past as a non-invasive method to quantify the RM. In this paper, an accurate Partial Least Square (PLS) model was developed and calibrated with a single product, focusing on a small but significative wavelength range of NIR spectra (model SR), characteristic of the water and not of the product. The salient feature of this approach is that the model SR appears to provide fairly accurate estimates with the same product but at a higher concentration, with other excipients and in presence of an amino acid at high concentration, without requiring any additional calibration with KF analysis, as in previous techniques; the irrelevance of the vial shape was also shown. This approach was compared to a simpler one, based on a single-variable linear regression, and to more complex one, using a wider wavelength range or calibrating the PLS model with several products. Model SR definitely ended up as the most accurate, and it appeared to have a great potential as a robust model, suitable also for products that were not involved in the calibration step.

Keywords: Freeze-drying; Karl Fischer titration; Mathematical modeling; Multivariate analysis; Near-infrared spectroscopy; Partial least square; Residual moisture; Robust model.