Development and Validation of an In-Line API Quantification Method Using AQbD Principles Based on UV-Vis Spectroscopy to Monitor and Optimise Continuous Hot Melt Extrusion Process

Abstract

A key principle of developing a new medicine is that quality should be built in, with a thorough understanding of the product and the manufacturing process supported by appropriate process controls. Quality by design principles that have been established for the development of drug products/substances can equally be applied to the development of analytical procedures. This paper presents the development and validation of a quantitative method to predict the concentration of piroxicam in Kollidon^® VA 64 during hot melt extrusion using analytical quality by design principles. An analytical target profile was established for the piroxicam content and a novel in-line analytical procedure was developed using predictive models based on UV-Vis absorbance spectra collected during hot melt extrusion. Risks that impact the ability of the analytical procedure to measure piroxicam consistently were assessed using failure mode and effect analysis. The critical analytical attributes measured were colour (L* lightness, b* yellow to blue colour parameters-in-process critical quality attributes) that are linked to the ability to measure the API content and transmittance. The method validation was based on the accuracy profile strategy and ICH Q2(R1) validation criteria. The accuracy profile obtained with two validation sets showed that the 95% β-expectation tolerance limits for all piroxicam concentration levels analysed were within the combined trueness and precision acceptance limits set at ±5%. The method robustness was tested by evaluating the effects of screw speed (150-250 rpm) and feed rate (5-9 g/min) on piroxicam content around 15% w/w. In-line UV-Vis spectroscopy was shown to be a robust and practical PAT tool for monitoring the piroxicam content, a critical quality attribute in a pharmaceutical HME process.

Keywords: AQbD; HME; PAT; QbD; RTRT; analytical procedure validation; analytical quality by design; analytical target profile development; hot melt extrusion; in-line UV-Vis spectroscopy; process analytical technology; quality by design; real time release testing.

Figure 1

Schematic of the hot melt extrusion process.

Figure 2

Spectral tristimulus values.

Figure 3

Risk priority number (RPN) values of the risk areas before and after control strategy implementation as described in Table 2 for the method.

Figure 4

Methodology to build predictive model using UV-Vis spectra. WL= wavelength, PCA = principal component analysis, R_cv² = coefficient of determination, RMSECV = root mean square error, CV = cross-validation, LV= latent variables.

Figure 5

b* (blue to yellow colour parameter) vs. experiment time of the calibration data set.

Figure 6

Absorbance spectra of the calibration data set, (a) full spectrum and (b) 446 to 540 nm range.

Figure 7

Scores of principal components (PC) 1 vs. PC 2 of the calibration data set.

Figure 8

Scores of PC1 vs. experiment time with transition data.

Figure 9

Score of PC1 vs. experiment time only in steady state.

Figure 10

Loadings from a PCA of the calibration data set using (a) the entire wavelength range and (b) the selected range.

Figure 11

(a) R_cv² vs. number of components, (b) RMSECV vs. number of components and (c) stacked PLS regression vector for different number of latent variables vs. wavelength.

Figure 12

Linearity of the validation data set.

Figure 13

Accuracy profile with the dashed black lines indicating ±5% limits, the blue lines with circles are the β-expectation tolerance limits, green points are the relative bias for every measurement and the red line with circles is the method mean relative bias.

Figure 14

Contour profiler showing the influence of the interaction of screw speed and feed rate on the predicted PRX amount using the PLS model. The red line indicates the predicted PRX content of 14.45% w/w. The black cross marks the optimised extruder parameters. The black dashed lines indicate the upper and lower limits defined respectively 14.75 and 14.15% w/w of PRX.