(T1330-11-64) A Novel Material-Sparing Degradation Kinetics Approach for Feasibility Assessment of Twin-Screw Melt Granulation

Purpose: Twin-screw melt granulation (TSMG) stands as a robust continuous manufacturing platform achieving commercial success formulating high-dose metformin, sitagliptin and elagolix sodium. Despite commercial success, feasibility assessment for TSMG requires substantial material quantities hindering its wider application. TSMG relies on heat and shear at the kneading zone for granulation to occur. Most drugs are stable and can be processed through TSMG due to short residence times ( < 1 min). The high-energy environment at kneading zone may lead to in-process changes like size reduction, amorphization possibly impacting drug stability. The general fundamentals of thermodynamics and particle size reduction govern the granule growth and local heat generation during TSMG. The purpose of this study was to create a blueprint for elucidating feasibility of TSMG utilizing a mechanistic understanding of granulation process. Differential Scanning calorimetry (DSC) and ball-milling was used for simulating the heat and shear crucial for granulation. Using the two material-sparing techniques, an empirical model was developed aiming to correlate actual and predicted degradation.

Methods: Gabapentin (GABA) was chosen to build the empirical model due to its sensitivity to shear and ability to degrade immediately upon melting. GABA and Gabalactam (GABA-L), the primary degradant was quantified using HPLC with sample concentration of 5 mg/mL and GABA-L limit of quantification at 0.01%. Klucel™ (Hydroxypropylcellulose) was chosen as thermal binder and binder level was fixed at 10 % based on the preliminary screening with DSC¹. GABA-Klucel blend was milled in SPEX mixer mill for 5 mins to mimic the shear stress during TSMG. GABA crystal size (in blend) was measured using laser diffraction. Milled and unmilled GABA-Klucel blend were subjected to different temperatures for various durations using DSC and % GABA-L was quantified. Predictive equations for % GABA-L as a function of time and temperature for the two blends was calculated using empirical model fitting. Granulation trials were conducted on Leistritz ZSE-18 twin-screw corotating extruder with a 25 L/D barrel configuration equipped with 30 mm kneading block (Staggering angle 60°). The feed rate was kept constant while barrel temperature (120-150 °C) and screw speed (50-200 rpm) was varied. Kittikunakorn et al. observed a strong correlation between granule temperature and drug stability². Granule temperature was measured at discharge using a Thermodyne FLIR E54 camera. The onset of granulation and morphology changes were studied using screw pullout technique by collecting samples from the different zones in the extruder. Total residence time was calculated using a dye-tracer experiment.

Results: Based on the DSC results, GABA degradation was observed to follow Arrhenius kinetics up to 160 °C. Beyond 160 °C, degradation rate increased significantly potentially attributed to pre-melting phenomena. Despite success of DSC as a screening tool to elucidate binder level, it underpredicted degradation during TSMG due to local heat generation and in-process physicochemical changes. Blend milling was performed to mimic the shear at kneading zone. Post milling, a reduction in GABA crystal size from 50 µm to 8 µm was observed along with emergence of crystal defects. In line with the increased degradation observed with smaller particle sizes, % GABA-L increased at least 2-fold with milled blend compared to unmilled under equivalent time-temperature conditions. Employing an empirical model fitting approach using response surface design, predictive equations were developed correlating % GABA-L with time and temperature for both blends (Figure 2). Screw pullout runs confirmed the kneading zone as the primary site for granulation onset and granule growth to occur. The extruder was treated as two compartments pre-granulation and post-granulation for the model based on the results of screw pullout runs (Fig 1). Utilizing the predictive equation and estimated residence times, % degradation pre-granulation was calculated and observed to be lower than post granulation across range of conditions (Table 1). Granule temperature measured at discharge over the run was notably higher than barrel temperature highlighting the impact of local heat generation (Table 1). Thus, for post granulation section, granule temperature instead of barrel temperature was utilized for prediction. Table 1 summarized the correlation between actual vs predicted degradation with the two empirical models. Comparing the actual % GABA-L to the predicted values, a 90% accuracy was observed for the granulations runs employing the chosen screw configuration.

Conclusion: The combined model integrating the unmilled and milled blends encapsulates the thermal and shear history experienced by the material, offering a material sparing approach for feasibility assessment of TSMG. The empirical model of unmilled blend reflects the pre-granulation section while the one for milled blend reflects the post-granulation section during TSMG. Granule temperature is a superior indicator of actual temperature post kneading. Granule temperature is primarily impacted by granule growth and strength. Hence future experiments are designed to utilize a rheometer for simulating rheological changes during the granule growth process and incorporating them into our model. Considering GABA degradation deviates from Arrhenius kinetics at 160 °C, the empirical model becomes invalid beyond this threshold.

References: 1. Adwait Pradhan, Nada Kittikunakorn, Fengyuan Yang, Thomas Durig, Feng Zhang: Poster presentation on ‘How to maintain drug stability during twin-screw melt granulation?’ at AAPS 2022.
2. Kittikunakorn N, Paul S, Koleng III JJ, Liu T, Cook R, Yang F, et al. How Does the Dissimilarity of Screw Geometry Impact Twin-screw Melt Granulation? European Journal of Pharmaceutical Sciences. 2021;157:105645.

Acknowledgements: The authors would like to acknowledge Mark Anderson and Stat Ease for their help with the Design of Experiments analysis during this study.

Funding: The project was funded by Ashland Specialty Ingredients.


Figure 1. Onset of granulation and morphological changes at kneading zone confirmed through SEM of screw pullout samples for TSMG of gabapentin. Pre-granulation section includes conveying and pre-densification while the post granulation section includes granule growth and attrition.


Figure 2. % Gaba-L as a function of time and temperature for the milled physical blend (8 µm Gaba crystal size). The empirical data fitting was carried out using a Response surface design on Stat Ease. T stands for temperature while t stands for time in the degradation equation.

% GABA-L= -0.32+(0.002*T) – (3.24 * t) + (0.025 * T *t)


Table 1. Sample data correlating actual vs predicted % Gaba-L for TSMG at given barrel temperature and screw speed. The given table is a subset of the dataset containing more granulation runs and used to illustrate the potential of our empirical model.