In-line UV-Visible process analytical technology for optimisation of continuous manufacture of piroxicam products using hot melt extrusion - a quality by design approach

Continuous manufacture (CM) and process analytical technology (PAT) have been proposed as innovative solutions for the pharmaceutical industry to decrease costs and prevent shortages of medicines in the market. This concept implies that raw materials are continuously fed and processed in a sequence of unit operations while being monitored by non-destructive analytical techniques. Furthermore, this strategy provides flexibility as products can be diverted in response to process deviations. The work carried out in this thesis concerns the use of Hot-Melt Extrusion (HME) as a continuous process system with in-line UV-Vis spectroscopy to develop and manufacture piroxicam (PRX) amorphous dispersions (ASD) with enhanced solubility. Solubility parameters of the drug and polymers were calculated using group contribution methods to select a carrier miscible with PRX. The Just-Breitkreutz total solubility parameter value for PRX is (δ = 21 MPa1/2) and for Vinyl pyrrolidone – vinyl acetate copolymer (PVPVA) it is (δ = 19.71 MPa1/2). The small difference (Δδ < 7 MPa1/2) between PRX and PVPVA solubility parameters (Δδ = 1.30 MPa1/2) indicated those would be miscible, therefore PVPVA was selected as the ASD carrier to be evaluated experimentally. The Flory-Huggins thermodynamic model and differential scanning calorimetry (DSC) melt depression data were applied to build a phase diagram of PRX and PVPVA mixtures. The phase diagram estimated that 8-24% PRX processed with PVPVA around 140 °C would form a metastable amorphous mixture. The phase diagram results were validated using extrusion experiments. Samples containing 15-20 % of PRX with PVPVA yielded amorphous mixtures whereas samples with concentration ≥ 25 % PRX produced unstable dispersions that resulted in phase separation. Design of experiments (DoE), multivariate analysis (MVA) and Process Analytical Technology (PAT) were used to investigate the impact of die temperature, screw speed, solid feed rate and PRX % on the ASD extrudates. Statistical models built from in-line UV-Vis responses (L*, b*, a*) revealed interactions between PRX% and die temperature, feed rate and screw speed. The design space (DS) of the process included PRX < 20 % w/w, temperature 140 °C, solid feed rate 7-10 g/min and screw speed 200-300 rpm. The model indicated that screw speed and feed rate should be adjusted according to PRX % to achieve residence time and thermal and mechanical energy that would prevent failure modes such as oversaturation and bubbles (Abs@680 nm > 0.1) and colour changes (L* < 90, a* > -9). In-line UV-Vis spectroscopy provides a powerful tool to monitor quality of extrusion products in real-time. A quantitative method based in-line UV-Vis absorbance and partial-least square model to predict the PRX % in PVPVA during HME was proposed using analytical quality by design principles (AQbD) and accuracy profile approach. The accuracy profile showed that In-line UV-Vis could predict PRX % within the trueness and precision acceptance limits (±5 %) for all PRX levels analysed (10.58-18.46 %). Additionally, the quick response and simplicity of in-line UV-Vis responses enabled a machine learning Nelder-Mead (NM) algorithm to converge to screw speed (7.81 g/min) and feed rate (354.44 rpm) settings that matched L* (93.07) and b* (83.00) targets. The optimised PRX dispersions with 16 % PRX were formulated as immediate-release tablets. The analysis of the Hausner ratio (1.11), angle of response (37-40 °) and compressibility index showed that the ASDs have excellent flow, thus are good candidates for direct compression (DC) process. The tabletability plot demonstrated that ASDs achieved considerable low tensile strength (TS) at high force (TS ~ 1.5 at 160 MPa). A mixture design experiment determined optimal combinations of fillers (Avicel® PH 102, Pearlitol® SD 200) and the disintegrant (Ac-Di-Sol®), whereas a classical design evaluated compaction pressure, speed and lubrication effects on tablet critical quality attributes (CQAs). The mixture design determined a minimum Avicel® PH 102 threshold of 27 % to achieve TS > 1.7MPa at 120 MPa, while the process experiment excluded lubricant % and compression speed effects in TS and SF. Formulations with 30-50 % Avicel® PH 102, 15-30 % Pearlitol® SD 200, 3-5 % Ac-Di-Sol® manufactured between 120-180 MPa resulted in tensile strength > 1.7 MPa, solid fraction ≈ 0.85, ejection pressure < 3 MPa, friability < 0.5 %, disintegration time < 300 s and 80 % PRX released within 30 min of the in-vitro dissolution test, confirming the formulation design space and compression pressure predicted effects. This work demonstrated the role of in-line UV-Vis spectroscopy in optimisation and monitoring the extrusion of amorphous PRX. The PAT detected the PRX saturation limit, previously predicted by Flory-Huggins thermodynamic modelling. The L* and b* were successfully implemented as targets of the NM algorithm to optimise screw speed and feed rate. Finally, a new tablet dosage form was developed based on the optimised extruded amorphous PRX and PVPVA.