(641d) Towards Optimized Pharmaceutical Freeze-Drying Processes Via Controlled Nucleation: Mechanistic Modeling of Vacuum-Induced Surface Freezing | AIChE

(641d) Towards Optimized Pharmaceutical Freeze-Drying Processes Via Controlled Nucleation: Mechanistic Modeling of Vacuum-Induced Surface Freezing

Authors 

Keywords: Freezing, Drying, Freeze-Drying, Lyophilization, Stochastic Nucleation, Mechanistic Modeling, Pharmaceutical Manufacturing, Quality by Design, Process Design & Development,

Motivation and Introduction

Freeze-drying is the method of choice to stabilize sensitive biopharmaceutical drug products, since it removes water at low temperatures. At commercial scale, tens of thousands of vials containing the pharmaceutical formulation are densely packed on temperature-controlled shelves and freeze-dried together. [1-2] Regulatory requirements impose that the products in all vials have to meet quality specifications, necessitating a detailed process understanding and tight control.

At the same time, the properties of dried products vary among vials on a shelf, rendering process design and control challenging. [1-3] This batch heterogeneity is expressed in a plethora of product attributes, whereby the mean pore size and the residual moisture in the dried products are of special interest. A variability in these properties directly poses a risk to the quality of the final product and thus necessitates the choice of more conservative process parameters, connected to longer cycle times. [1-2]

Literature indicates that the stochasticity of ice nucleation is a major cause for batch heterogeneity. The times and temperatures, at which nucleation occurs, inherently vary among vials; since the nucleation temperature is understood as key predictor of the mean pore size of the product, this variability affects the final properties of the dried product, including residual moisture. [1-3]

To overcome this stochasticity, a number of technical approaches have been developed over the past decades, commonly referred to as controlled nucleation techniques. [1-2] While many such techniques have been proposed, only three of them have emerged so far in laboratory scale freeze-drying devices, currently moving towards implementation at commercial scale. [3]

Among these techniques, vacuum-induced surface freezing (VISF) is the only one that does not require major technical modifications or retrofitting of the existing freeze-drying equipment and is thus of special interest for commercial applications. The technique is based on applying vacuum during the freezing stage, in order to promote nucleation via evaporative cooling. [3] While successful implementations of VISF were reported in literature, no systematic and mechanistic framework exists to date that could predict the process performance and thus guide the choice of operating conditions. And indeed, there are many parameters that were shown to affect the process outcome: They include the duration during which vacuum is applied and the corresponding pressure; the temperature of the shelf at the time of the vacuum; and finally, the design of holding steps at constant temperature both before and after the vacuum phase.

Given the plethora of process parameters, the choice of parameter values is primarily empirical, rendering process design and optimization challenging. This work aims to support lyophilization professionals in designing and understanding VISF by providing a mechanistic model of the process.

Methods

We extended a mechanistic freezing model recently developed in our group, [4-5] in order to study vacuum-induced surface freezing in detail. The model explicitly considers the stochastic nature of nucleation, and predicts the thermal gradients that form within the vial upon application of vacuum due to evaporation. Importantly, the effect of these gradients on the nucleation rate is taken into account. To promote the use and further development of the model in the wider community, we provide open source access to it in the form of a python package (available at https://pypi.org/project/ethz-snow/ ).

We applied the model to carry out sensitivity studies to deepen the understanding how various process parameters affect the process performance. Additionally, we compared the model predictions with experimental data from campaigns in laboratory freeze-driers.

Results

Model instantiations revealed that the extent of evaporative cooling induced by the application of vacuum increases for longer vacuum duration and lower vacuum pressure. The vacuum pressure was found to govern the final minimum temperature that could be achieved. In order to promote nucleation efficiently, low temperatures have to be reached fast, i.e. before a substantive amount of water has evaporated. For this, the application of lower pressures was found to be beneficial.

In addition to process conditions, the nucleation behavior of a formulation is governed by its nucleation kinetics. We therefore studied the effect of the nucleation kinetics on the performance of VISF; the findings indicate that the process becomes less efficient for slower nucleation kinetics, where lower temperatures have to be reached for nucleation to occur.

Finally, we studied the effect of holding steps on the VISF outcome. In line with the current practice in experimental case studies, we find that holding steps both before and after the application of vacuum are beneficial. A pre-nucleation holding step ensures thermal equilibration; a post-nucleation holding step ensures that solidification takes place in a slow and controlled manner. To this end, the model identifies suitable temperatures and durations for both holds.

Discussion and Implications

The finding that the performance of VISF depends on the nucleation kinetics of the specific formulation, is of great relevance. It implies that the choice of process parameters requires knowledge on the nucleation behavior of the formulation; different formulations may require different process conditions to achieve optimal process performance.

For a given formulation, nucleation occurs in average at lower temperatures the higher the concentration is. Thus, VISF eventually reaches performance limitations for highly concentrated systems, where the extent of evaporative cooling is not sufficient anymore to induce nucleation. Furthermore, the nucleation kinetics is governed by the number of available nucleation sites, which depend on the production environment: Under GMP conditions, reported nucleation temperatures typically are 5-10 K lower than those obtained for the same formulation under standard laboratory conditions. This implies a potential scale-up issue, where VISF parameters successfully tested in experiments at lab-scale may be unsuitable for commercial-scale processes. The model may be used to quantify this difference and suggest how parameter values have to be adjusted depending on the environment.

As these findings indicate, the herein proposed model provides a valuable framework to study VISF in a mechanistic fashion. It may be used to understand potential limitations of the process, and to identify suitable process conditions tailored towards the specific formulation of interest.

Acknowledgement

The authors thank the Janssen Pharmaceutical Companies of Johnson & Johnson for the support in the course of the project and the financial funding. We especially thank Juan Carlos Araque and Ryan Wall.

References

[1] JC Kasper and W Friess: The freezing step in lyophilization: Physicochemical fundamentals, freezing methods and consequences on process performance and quality attributes of biopharmaceuticals, Eur. J. Pharm. Biopharm. (2011), 78, 248-263.

[2] LT Deck, DR Ochsenbein and M Mazzotti: Stochastic Shelf-Scale Modeling Framework for the Freezing Stage in Freeze-Drying Processes, Int. J. Pharm. (2022), 613, 121276.

[3] R Pisano, A Arsiccio, K Nakagawa and AA Barresi: Tuning, measurement and prediction of the impact of freezing on product morphology: A step toward improved design of freeze-drying cycles, Drying Technology (2019), 37:5, 579-599.

[4] LT Deck, DR Ochsenbein and M Mazzotti: SNOW – Stochastic Nucleation of Water, (2021), GitHub Repository, https://github.com/SPLIfA/snow/

[5] GM Maggioni and M Mazzotti: Modelling the stochastic behaviour of primary nucleation. Faraday Discussions 2015, 179, 359-382.