Validation of an in-line Raman spectroscopic method for continuous active pharmaceutical ingredient quantification during pharmaceutical hot-melt extrusion.

A calibration model for in-line API determination was developed based on Raman spectra collected during hot-melt extrusion. This predictive model was validated by calculating the accuracy profile based on the analysis results of validation experiments. Furthermore, based on the data of the accuracy profile, the measurement uncertainty was determined. Finally, the robustness of the model was evaluated. A Raman probe was implemented in the die of a twin-screw extruder, to monitor the drug concentration during extrusion of physical mixtures containing 15, 20, 25, 30 and 35% (w/w) metoprolol tartrate (MPT) in Eudragit(®) RS PO, an amorphous copolymer of acrylic and methacrylic acid esters with a low content of quaternary ammonium groups, which are present as salts. Several different calibration models for the prediction of the MPT content were developed, based on the use of single spectra or averaged spectra, and using partial least squares (PLS) regression or multivariate curve resolution (MCR). These predictive models were validated by extruding and monitoring mixtures containing 17.5, 22.5, 25.0, 27.5 and 32.5% (w/w) MPT. Each validated concentration was monitored on three different days, by two different operators. The β-expectation tolerance intervals were calculated for each model and for each of the validated MPT concentration levels (β was set at 95%), and acceptance limits were set at 10% (relative bias), indicating that at least 95% of future measurements should not deviate more than 10% from the true value. The only model where these acceptance limits were not exceeded was the MCR model based on averaged Raman spectra. The uncertainty measurements for this model showed that the unknown true value can be found at a maximum of ±7.00% around the measured result, with a confidence level of 95%. The robustness of this model was evaluated via an experimental design varying throughput, screw speed and barrel temperature. The robustness designs showed no significant influence of any of the process settings on the predicted concentration values. Raman spectroscopy proved to be a fast, non-destructive and reliable method for the quantification of MPT during hot-melt extrusion. From the accuracy profile of the MCR model based on averaged spectra, it was concluded that for each MPT concentration in the validated concentration range, 95 out 100 future routine measurements will be included within the acceptance limits (10%).