Application of quality by design to formulation and processing of protein liposomes

Abstract

Quality by design (QbD) principles were explored in the current study to gain a comprehensive understanding of the preparation of superoxide dismutase (SOD) containing liposome formulations prepared using freeze-and-thaw unilamellar vesicles (FAT-ULV). Risk analysis and D-optimal statistical design were performed. Of all the variables investigated, lipid concentration, cholesterol mol%, main lipid type and protein concentration were identified as critical parameters affecting SOD encapsulation efficiency, while the main lipid type was the only factor influencing liposome particle size. Using a model generated by the D-optimal design, a series of three-dimensional response spaces for SOD liposome encapsulation efficiency were established. The maximum values observed in the response surfaces indirectly confirmed the existence of a specific SOD–lipid interaction, which took place in the lipid bilayer under the following optimal conditions: (1) appropriate membrane thickness and curvature (DPPC liposomes); and (2) optimal “pocket size” generated by cholesterol content. With respect to storage stability, the prepared SOD liposomes remained stable for at least 6 months in aqueous dispersion state at 4 °C. This research highlights the level of understanding that can be accomplished through a well-designed study based on the philosophy of QbD.

Introduction

Owing to their unique biological and physico-chemical properties, liposomes are established yet still very promising drug delivery system (Xu and Burgess, 2011). In several areas, including small molecule anti-cancer and anti-fungal therapy (Lasic et al., 1992, Maurer et al., 2001), liposome formulations have been proven extremely effective. In other areas, they show great promise, such as in gene therapy (Web-source, 2011), vaccination (Gregoriadis, 1995, Zhuang et al., 2012), and protein therapeutics (Torchilin and Lukyanov, 2003). In particular, liposomal protein therapeutics have generated great interest. From a clinical point of view, the potential ability of liposomes to deliver protein/enzyme directly into the cytoplasm or lysosomes of live cells is of crucial importance for the treatment of inherited diseases caused by the abnormal functioning of some intracellular enzymes and cancer (Torchilin, 2005). However, from a manufacturing perspective, the extremely low protein encapsulation efficiency has been limiting the broad use of liposome delivery systems, especially in the predominantly used small vesicle size range (50–150 nm). In addition, poor protein stability during preparation elicits concern over the use of harsh processing conditions and/or organic solvents. Furthermore, manufacturing variability as a result of a lack of understanding of the preparation process means a much more stringent review is necessary in terms of product safety (Rathore and Winkle, 2009, Vogt, 1992). Hence, it is the objective of this study to utilize quality by design (QbD) principles to assist formulation and process design to improve the protein encapsulation efficiency and protein stability as well as understand the sources of variability in order to improve product quality.

Superoxide dismutase (SOD) was used as the model protein in this study. It is one of the most potent antioxidants known in nature. SOD catalyzes the dismutation of the superoxide radical into hydrogen peroxide and oxygen and it has been used for the treatment of oxidative stress diseases such as rheumatoid arthritis, cancer, and respiratory distress syndrome. While SOD has demonstrated great potential as an alternative to conventional therapies (Keele et al., 1971, McCord and Fridovich, 1969, Okado-Matsumoto and Fridovich, 2001, Zhang et al., 2002), its current use is limited by several key drawbacks, such as its extremely short circulation time, non-specific tissue distribution, and inability to penetrate through the cellular membrane to the intracellular targets. Accordingly, a liposomal SOD formulation is expected to provide a better therapeutic index due to carrier-facilitated intracellular transportation as well as the targeting effect.

Previously (Xu et al., 2012a), an improved freeze-and-thaw cycling technique was reported where the protein containing liposome preparation process was separated into two steps: the generation of unilamellar vesicles, and freeze–thaw cycling to encapsulate protein. Because the liposomes obtained using this approach remained as unilamellar vesicles and no significant change in particle size was observed, they are referred to as freeze-and-thaw unilamellar vesicles (FAT-ULV). Compared with traditional preparation methods, the FAT-ULV method is very effective in improving protein encapsulation efficiency (up to 50%). However, this process is relatively new. Hence it is very crucial to use the QbD approach to help understand the formulation and processing design space.

Pharmaceutical QbD emphasizes that the product quality should be built (designed) into the product rather than tested (Yu, 2008). This requires that quality-improving scientific methods be used upstream in the beginning stages of the research, development and design phases (Wu et al., 2007). QbD identifies characteristics that are critical to quality from the perspective of patients, translates them into the attributes that the drug product should possess, and establishes how the critical process parameters can be varied to consistently produce a drug product with the desired characteristics (Yu, 2008). A complete QbD study usually involves the following five stages: (1) define target product quality profile based on scientific prior knowledge and appropriate in vivo relevance, (2) design product and manufacturing processes to satisfy the pre-defined profile, (3) identify critical quality attributes, process parameters, and sources of variability (risk assessment), (4) use a design of experiment (DOE) approach to screen and obtain variable response surfaces in order to establish the product design space (the range of process and/or formulation parameters that have been demonstrated to provide assurance of quality), and (5) control manufacturing processes to produce consistent product quality over time through operation within the established design space, thus assuring that quality is built into the product (ICH Q8).

The current study focused on the first four stages of QbD implementation in a laboratory setting. Briefly, the desired product quality profiles were defined and risk assessment was conducted to identify potential high risk factors. Subsequently, a D-optimal experimental design was used to screen high-risk variables and to obtain the variable response surfaces (Bodea and Leucuta, 1997, El-Hagrasy et al., 2006). The optimal criterion for D-optimal design is that the determinant of the X′X matrix is maximized, where X is the design matrix (Atkinson et al., 2007). Compared with standard designs (e.g. factorial designs), the D-optimal design gives the most precise estimate of the factor effects; however, it requires statistical software to calculate the determinant of the X′X matrix. For this reason, JMP software (SAS Institute) was used to create the design. After obtaining the response surface, the optimal formulation and process conditions were identified. Further experimental tests were performed to test the robustness and accuracy of the generated model.