Techniques to preserve product quality in pharmaceutical freeze drying

Therefore, the operating conditions of the freeze drying process – namely the temperature of the heating shelf (or that of the heating fluid) and the pressure in the drying chamber – have to be carefully selected for a given formulation. In most cases, their values are set on the basis of a trial-and-error approach, selecting the same cycle used for similar (if available) or different products as a starting point, and testing product quality at the end of the run. This method is time-consuming and therefore expensive, and it does not guarantee the optimal cycle at the end of the development stage.

In this framework, in 2004 the U.S. Food and Drug Administration issued the Guidance for Industry PAT, strongly encouraging adoption of new technical solutions to build product quality in the manufacturing run, instead of testing product characteristics at the end of the process. A Process Analytical Technology (PAT) is a mechanism to design, analyse and control pharmaceutical manufacturing processes through the measurement of Critical Process Parameters (CPP) which affect Critical Quality Attributes (CQA). In a pharmaceutical freeze drying process, PAT should be able to measure CPP like the temperature of the product and the residual amount of ice, with the goal to check if the constraint on product temperature is fulfilled and to identify the ending point of the drying stage, and to modify the operating conditions with the goal to minimise drying duration, beside avoiding product overheating.

Nowadays, various techniques are available to freeze-drying practitioners to avoid testing product quality at the end of the freeze-drying cycle, both at laboratory scale and in industrial units. This way, product quality can be built-in or it can be by design, and a suitable (and optimal) cycle can be obtained rapidly: this is of outmost importance as a very high number of pharmaceuticals are currently being freeze-dried, and this figure is only going to rise in the coming years.

A suitable, or optimal, cycle can be obtained both in-line and off-line. In the first case it is required to use a monitoring tool to get information about the state of the product, and some previous knowledge (rule of thumb) or a mathematical model of the process to identify suitable (or optimal) operating conditions. In the second case it is necessary to carry out a preliminary experimental investigation, in most cases at laboratory scale, with the goal to get process knowledge (the same monitoring tools used to design in-line the process can be used to this purpose), and then to determine the design space of the process, i.e. the values of the operating conditions that fulfil some operation constraints (i.e., in this case, maintaining product temperature below the limit value).

The Pressure Rise Test is a method frequently used to get some insight into product state and dynamics, in particular in laboratory scale and pilot-scale freeze-dryers used for cycle development. It consists of closing the valve in the duct connecting the drying chamber to the condenser for a short time interval (usually ranging from 5 to 30 seconds), and measuring the increase of the value of chamber pressure. A mathematical model is then used to calculate the curve of pressure rise that appears to be a function of many parameters, and in particular of:

• The temperature of the product
• The residual amount of ice
• The heat transfer coefficient (K_v) used to express the dependence of the heat flux to the product (J_q) on the difference between the temperature of the heating fluid (T_fluid) and that of the product at the bottom of the container (T_B):

J_q = K_v (T_fluid – T_B)

• The resistance of the dried cake (R_p) used to express the dependence of the mass flux from the product (J_w) on the difference between the partial pressure of water vapour at the interface of sublimation (p_w,i) and in the drying chamber (p_w,c):

Looking for the best fit between measured and calculated values of chamber pressure, it is therefore possible to get information about the state of the product (temperature and residual amount of ice) and to get the values of some model parameters (K_v and R_p). The test can be repeated during the drying stage with the goal to track product dynamics.

The Smart Freeze-Dryer is an expert system that uses the Pressure Rise Test, and the Manometric Temperature Measurement (MTM) algorithm, as well as some rules of thumb to identify suitable values of the heating fluid temperature and of the chamber pressure¹.

LyoDriver is another system that can be used to get in-line the cycle using the result obtained through the Pressure Rise Test and the Dynamic Parameters Estimation (DPE+) algorithm²: differently from the Smart Freeze-Dryer algorithm, in this case a model of the process is used to perform a real in-line optimisation of the process, looking for the values of the operating conditions that allows minimising the duration of the drying stage³. An example of the results obtained using LyoDriver to optimise in-line a freeze-drying process is shown in Figure 1 for a sucrose solution (the limit temperature is -32°C). It is possible to see that at the beginning of the run the temperature of the heating fluid is set to a high value (graph A of Figure 1): product temperature is in fact very low (graph C of Figure 1, as it starts rising from the value reached during the freezing stage. Then, as drying goes on, the heating fluid temperature is decreased, as a consequence of the results obtained through the Pressure Rise Test and product temperature remains below the limit value. Drying time is roughly 20 hours as it can be detected from the ratio between Pirani and Baratron pressure measurements (graph B of Figure 1).

While LyoDriver optimises the heating fluid temperature (some guidelines are available to set the value of the chamber pressure at the beginning of the run, and this value is maintained constant throughout the run), a Model Predictive Control algorithm has been proposed to simultaneously optimise both operating conditions⁴. Also in this case, the Pressure Rise Test is used as a sensing device, and the estimations of model parameters are used by the algorithm.

It has to be remarked that the Pressure Rise Test can be used only in case a fast-closing valve is installed in the duct, and that the increase of chamber pressure is responsible for a temperature rise in the product: therefore, the limit temperature used by the control temperature has to be few degrees lower than the collapse (or eutectic) temperature to account for the temperature rise during the test (LyoDriver, as well as the Model Predictive Controller, are able to estimate this temperature rise, and to set the safety margin accordingly).

As an alternative to the Pressure Rise Test, it is possible to monitor the state of the product, and to get that information that is required to optimise the cycle, using the Smart Soft Sensor⁵. The hardware of this sensor is very simple, as it consists of a simple thermocouple inserted in a vial of the batch. The software consists of a model of the process, used to simulate product dynamics and to calculate product temperature: the difference between measured and calculated values of product temperature is used to ‘correct’ model calculations, thus getting the desired information (that are the same provided by the Pressure Rise Test). Once the state of the product and model parameters K_v and R_p are known, the design space of the process can be calculated (using the algorithm proposed by⁶) and, finally, it is possible to set the value of the heating fluid temperature (chamber pressure has to be set at the beginning of the test) accordingly⁷. It has to be remarked that in order to use the Smart Soft Sensor a rough estimation of model parameters in required and, then, these estimations are corrected by the algorithm on the basis of the error on the calculated temperature of the product. Therefore, at the beginning of the drying stage, the accuracy of the estimated design space can be impaired by the poor accuracy of model parameters but, as product temperature is quite low, the limit temperature is never trespassed even using a high value of shelf temperature (e.g. 0°C). In any case, also when using LyoDriver, a limit value is set for the shelf temperature. Figure 2 shows the results obtained when using the Smart Soft Sensor to get in-line the optimal cycle for the same case study shown in Figure 1. The temperature of the heating fluid (graph A of Figure 2) appears to be higher with respect to that calculated by LyoDriver, thus resulting in a lower drying time (about 15 hours, as detected through the pressure ratio shown in graph B of Figure 2), and product temperature (graph C of Figure 2) remains below the limit value. This is due to the fact that when using the Smart Soft Sensor, no safety margin is required for product temperature. In fact, when calculating the design space using the estimations of model parameters obtained by the Smart Soft Sensor and by LyoDriver, assuming the same limit temperature (e.g. -33°C), almost the same results are obtained, as shown in Figure 3, but when taking into account temperature rise occurring during the Pressure Rise Test (and, therefore, decreasing the target temperature), then lower values of the heating fluid have to be used, thus resulting in a higher drying time. In this case, the safety margin calculated by LyoDriver is about 1.8°C and, thus, the limit temperature of the heating fluid is decreased, as shown in Figure 3 (solid line).

When using LyoDriver, or the Smart Soft Sensor, it is also possible to optimise off-line the cycle. In this case, the monitoring system is used to evaluate model parameters K_v and R_p and, in this framework, also other systems, e.g. the TDLAS⁸, can be used to get the same result. Then mathematical modelling is used to calculate off-line the design space of the process and, on the basis of this diagram, the optimal cycle can be determined.

As a conclusion, it should be clear that various model-based tools are nowadays available to freeze-drying practitioners to identify efficiently and effectively the most suitable operating conditions for a freeze-drying process, with the goal to preserve product quality and also to minimise the duration of the process. 

References

1. Tang, X.C., Nail, S.L., Pikal, M.J., 2005. Freeze-drying process design by manometric temperature measurement: design of a smart freeze-dryer. Pharmaceutical Research 22: 685-700
2. Fissore, D., Pisano, R., Barresi, A.A., 2011a. On the methods based on the Pressure Rise Test for monitoring a freeze-drying process. Drying Technology 29: 73-90
3. Pisano, R., Fissore, D., Velardi, S. A., Barresi, A. A., 2010. In-line optimization and control of an industrial freeze-drying process for pharmaceuticals. Journal of Pharmaceutical Sciences 99: 4691-4709
4. Pisano, R., Fissore, D., Barresi, A. A., 2011. Freeze-drying cycle optimization using model predictive control techniques. Industrial & Engineering Chemistry Research 50: 7363-7379
5. Bosca, S., Barresi, A. A., Fissore, D., 2014. Use of soft-sensors to monitor a pharmaceuticals freeze-drying process in vials. Pharmaceutical Development and Technology 19: 148-159
6. Fissore, D., Pisano, R., Barresi, A. A., 2011b. Advanced approach to build the design space for the primary drying of a pharmaceutical freeze-drying process. Journal of Pharmaceutical Sciences 100: 4922-4933
7. Bosca, S., Barresi, A. A., Fissore, D., 2013. Fast freeze-drying cycle design and optimization using a PAT based on the measurement of product temperature. European Journal of Pharmaceutics and Biopharmaceutics 85: 253-262
8. Schneid, S., Gieseler, H., Kessler, W.J., Pikal, M.J., 2009. Non-invasive product temperature determination during primary drying using Tunable Diode Laser Absorption Spectroscopy. Journal of Pharmaceutical Sciences 98: 3406-3418

Biography

Antonello A. Barresi has a PhD, obtained in 1990 from the Politecnico di Torino. Antonello is currently Full Professor at Politecnico di Torino, teaching process design and development. His most recent research focuses on process transfer, scale-up, cycle development and quality control in freeze-drying of pharmaceutical products and production of nanoparticles for pharmaceutical applications. Antonello is the author of more than 150 papers on international journals and book chapters, and more than 100 conference presentations.

Roberto Pisano graduated in Chemical Engineering cum laude in 2005 at Politecnico di Torino, and received a PhD from the same university in 2009. He was Visiting Scholar at Centre de Ressources Technologiques – Institut Technique Agro-Industriel (France) in 2008. Since July 2011, Roberto has been Assistant Professor at Politecnico di Torino, where he is responsible for teaching at the undergraduate level Process Dynamics and Control. His research interests focus on process modelling and optimisation of chemical processes with complex dynamics, as well as on the design and validation of model-based systems for process monitoring and control. In particular, Roberto developed and validated various tools for the monitoring, design and control of the freeze-drying process, as well as for the scale-up of cycles developed with laboratory equipment in large-scale apparatus. Lately, he has also started working on the application of e-beam radiation as post-treatment for freeze-dried products.

Davide Fissore is Assistant Professor at Politecnico di Torino, where he is lecturer of advanced process control, food processing and technologies and data driven modelling. His research activity is focused on process modelling and optimisation, and on the design and validation of model-based tools for process monitoring and control. The field of this research comprises the freeze-drying of pharmaceuticals and foodstuffs. Recent research projects have addressed the effects of using non-aqueous solvents in a freeze-drying process, and the freeze-drying of suspensions containing nanoparticles.