On the use of model-based tools to optimize in-line a pharmaceuticals freeze-drying process

This article is focused on the use of model-based tools to design and optimize in-line a pharmaceutical freeze-drying process. Two control systems have been compared, a predictive one that uses the pressure rise test to monitor the state of the system and to estimate in-line the values of model parameters, named LyoDriver in the previous literature, and a controller where a soft sensor uses the temperature measurement obtained by a thermocouple to get the same information and to calculate on-line the design space of the process. In both cases, the goal of the controller is to maintain product temperature as close as possible to a limit value, without trespassing it, throughout the primary drying stage. An extended experimental campaign has been performed, where various products, with different characteristics, have been processed, namely, aqueous solutions containing sucrose, mannitol, or polyvinylpyrrolidone. Results evidence that both systems are effective in optimizing in-line the freeze-drying process, but shorter cycles can be obtained using the soft sensor. This is due to the fact that the soft sensor is not responsible for any product overheating and, thus, product temperature can be maintained very close to the limit value, while when using the pressure rise test as monitoring tool, a safety margin has to be used, because of the temperature increase during the pressure rise test. Besides, when using the soft sensor no least-square optimization problem is solved to estimate model parameters, and this can improve the robustness of the system. The main drawback is represented by the fact that this system requires thermocouples to measure product temperature, and this can be difficult in industrial-scale freeze-dryers, used to process large batches of vials in sterile conditions, but it can be performed quite easily in lab-scale units used for process design.     

CFD modelling and scale-up of Confined Impinging Jet Reactors

Confined Impinging Jet Reactors (CIJRs) are appealing devices for precipitation of nanoparticles because of their high mixing efficiency. In fact, since precipitation processes are generally very fast, mixing plays a crucial role and it is of great importance to operate under very fast mixing conditions. In this work mixing and reaction in CIJRs are studied by means of Computational Fluid Dynamics (CFD). Mixing at the molecular level is modelled with a presumed Probability Density Function (PDF) approach: the Direct Quadrature Method of Moments coupled with the Interaction by Exchange with the Mean (DQMOM-IEM) model. The influence of operating conditions and reactor geometry on mixing is also evaluated and a scale-up criterion for CIJRs is developed, showing that scaling up by means of CFD is a practicable path, worth of further investigation.     

Developmental expression of dystrophin, dystrophin-associated glycoproteins and other membrane cytoskeletal proteins in human skeletal and heart muscle

Dystrophin, utrophin and the dystrophin-associated glycoproteins, beta-dystroglycan and adhalin, were analyzed, together with the membrane cytoskeletal proteins beta-spectrin, vinculin and talin, and adult and fetal myosin heavy chains, in 25 normal human fetuses from 8 to 24 weeks of gestation. Dystrophin was present in heart and skeletal muscle from 8 weeks although in the latter was mainly in the cytoplasm at this stage. Utrophin expression increased until around gestational weeks 19/21, but by 24 weeks immunostaining and immunoblot band intensities had reduced. Beta-dystroglycan was scarce in skeletal muscle at 8 weeks, increased with maturation and was more abundant in heart of the same age. Adhalin appeared later than beta-dystroglycan on skeletal muscle fiber surfaces, positivity became more intense as the fibers matured. In heart adhalin was detectable only in groups of cells at 12-16 weeks. From 8 weeks all fetal myotubes expressed beta-spectrin on their surfaces, while vinculin and talin positivity was mainly at the periphery of the fascicles, increasing with age. Adult slow myosin was seen in most myotubes at 10 weeks. Secondary myotubes then formed which increasingly expressed adult fast myosin, while still retaining fetal myosin. By 24 weeks most fibers expressing adult slow myosin had lost fetal myosin and were more mature in the expression of most membrane proteins. Muscle membrane organization during human fetal development is a complex process and takes place earlier in heart than skeletal muscle.     

Evaporative drying of aqueous dispersions of solid lipid nanoparticles.

Solid lipid nanoparticles (SLNs) have been proposed as alternative colloidal drug carriers. SLNs are obtained by dispersing warm oil-in-water microemulsions into cold water. The aim of this research was to investigate an evaporative drying process for aqueous dispersions of SLNs. For this purpose, a special apparatus, namely a thermostatic minidesiccator having alumina as the drying medium, was designed to carry out the evaporative drying at a controlled temperature. Besides the water removal kinetics, the mean particle size and the size distribution of SLNs were measured during the during with the aim of detecting the highest temperature at which the drying process can be carried out without significantly affecting the SLN average diameter. The SLN dispersions were evaluated with and without a hydrophilic excipient, commonly used as a cryoprotector (trehalose). The drying temperature of 10 degrees C was found to be the most suitable for obtaining SLNs as a powder, maintaining almost the same size as that of the SLNs in dispersion.     

A Model-Based Framework to Optimize Pharmaceuticals Freeze Drying

This article is focused on the design of pharmaceutical freeze-drying recipes using in-line or off-line tools. In particular, the model predictive control system is here used to optimize the in-line process, and the design space is used for off-line optimization. Because both methods uses a mathematical model of the process, the problem of estimating the model parameters, including their uncertainty or variability in the lot of vials, is addressed. Then, the strengths and the weaknesses of the various methods are discussed, with particular emphasis on their robustness and their application in industrial-scale freeze dryers. In particular, the ability of the model predictive control tool to obtain the optimal recipe in only one run and its capacity to manage the system in case of an in-line modification of the product properties are shown. For this purpose, experimental results obtained for sucrose- and mannitol-based formulations are presented.