Development of an effectual drug delivery system (DDS), invariably involves rational blending of a plethora of diverse functional and non-functional polymers and excipients. Optimizing the formulation composition and the manufacturing process of such a drug delivery product to furnish the desired quality traits vis-à-vis established products is usually a marathon task.

 The traditional approach of optimizing a formulation or process essentially involves studying the influence of one variable at time (OVAT), while keeping all others as constant. Using this OVAT approach, the solution of a specific challenging property be sometimes achieved, but attainment of the true optimal composition or process is never guaranteed. This may ostensibly be ascribed to the presence of interactions, i.e. the influence of one or more variables on others. The final product though may be satisfactory, but mostly sub-optimal, as a better formulation still prevails for the studied conditions.

Optimizing drug products and pharmaceutical process using Design of experiments (DoE), on the other hand, has been reported to successfully embarked upon all the potential factors systematically, simultaneously and speedily. Of late, a holistic DoE-based philosophy of Quality by Design (QbD) has been permeating into the mindset and practice in the industrial environs. This popularity of QbD paradigm in pharma circles is largely attributable to the recent impetus provided by the regulatory agencies like the ICH, FDA and EMEA through their respective federal guidelines. Since DoE has much wider domain of application, a terser jargon, viz. “Formulation by Design (FbD)”, applicable specifically to the use of DoE in drug formulation development has been proposed by us recently.

The FbD approach has been found to be highly rewarding to yield “the best possible” formulation or process, revealing the plausible synergistic or antagonistic interaction(s) among the product or process variables, and formulation economics in terms of time, developmental effort, expertise and of course money. Plus, it has a highly specific merit of amenability to scale-up of drug delivery system and post approval changes. Owing to such numerous benefits inherent FbD approach, it has lately witnessed a spurt in the systematic development of various DDS, both oral and non-oral. Figure 1 corroborates the rising FbD trend.

FbD Terminology
An experimental design constitutes the gist of FbD exercise. It is the statistical strategy for organizing the experiments in such a manner that the required information is obtained as efficiently and precisely as possible. Runs or trials are the experiments conducted according to the selected experimental design. Factors are independent variables, which tend to influence the product/process characteristics or output of the process, usually termed as response variables. Levels are the values assigned to a factor. A response surface plot (Figure 2) is a 3-D graphical representation of a response plotted between two independent variables and one response variable.

A knowledge space, i.e., entire worth-exploring realm, therefore, has to be identified from the possible vast ocean of scientific information based upon prior knowledge. A knowledge space, thereby, encompasses all those product and process variables that may even minutely affect the overall product quality. A design space has to be demarcated as a subset construct of knowledge space ensuring optimal product quality or process performance involving “selected few” influential variables. Control space is further deduced from this design space as the experimental domain earmarked for detailed studies during studies within the refined ranges of input variables. Design space applies systematic approach on archival data to convert the knowledge space to control space. As working within the design space is not considered as a “change”, it would not initiate any post-approval change process as per the federal guidelines. Figure 3 portrays the hierarchy of knowledge, design and control space.

FbD methodology
FbD uses five key strengths viz. apt choice of experimental designs, accurate computer-aided optimization, meticulous drug product development, precise definition of design and control space, and identification of critical quality attributes (CQAs), critical formulation attributes (CFAs) and critical process parameters (CPPs). Figure 4 pictorially illustrates the concept.

The theme of FbD optimization methodology provides thought-through and thorough information on diverse FbD aspects organized in a five-step sequence represented in Figure 5.

The FbD study begins with Step I, where an endeavor is made to explicitly ascertain the drug delivery objective(s). Various CQAs or response variables, which pragmatically epitomize the objective(s), are earmarked for the purpose. All the independent product/process variables are also listed.

In Step II, the response variables which directly represent the product quality (e.g. particle size for nanoparticles, emulsification time for self-emulsifying systems) are selected. Also, selection of “prominent few” influential factors among the “possible many” input variables is conducted using experimental designs through a process, popularly termed as screening. Experimental studies are also undertaken to define the broad range of factor levels.

During Step III, a suitable experimental design is worked out to map the responses on the basis of the study objective(s), responses being explored, number and the type of factors, and factor levels, viz. high, medium or low. A design matrix i.e., layout of experimental runs in matrix form, as per experimental design, is subsequently generated to guide the drug delivery scientist. The drug delivery formulations are experimentally prepared according to the design matrix, and the chosen response variables are evaluated meticulously.

In Step IV, a suitable graphical and/or numeric model is proposed on the basis of experimental data thus generated, and its statistical significance is discerned. Response surface methodology (RSM) is employed to relate a response variable to the levels of input variables. Optimum formulation compositions are searched within the experimental domain.

Step V is the ultimate phase of the FbD exercise in industrial milieu, involving validation of response predictive ability of the proposed design model. The optimum formulation is scaled-up and set forth ultimately for the production cycle.

Experimental designs employed during FbD
Experimental designs are employed for “screening” of influential variables as well as subsequent response surface analysis. Figure 6 provides bird’s eye view of key experimental designs employed for optimization of oral DDS.

Out of all the experimental designs, FD, CCD and FFD have been most frequently employed for systematic optimization of oral DDS. Figure 7 provides a succinct account on the usage of experimental designs in the development and optimization of drug delivery formulations and processes.

Model development
A model is an expression defining the quantitative dependence of a response variable on the independent variables. Numeric model is a set of polynomials of a given order or degree. The models mostly employed to describe the response(s) are first, second and very occasionally, third order polynomials.

Validation of FbD methodology
Validation is the key to verify the veracity and validity of experimental design and mathematical models. Various types of plots can be constructed for validating FbD methodology like predicted vs. observed, residuals vs. predicted, normal probability plots, etc.

Optimum search
From the models thus selected, optimization of one response or the simultaneous optimization of multiple responses needs to be accomplished by one or more methods as represented in the flow chart (Figure 8). Graphical approaches like location of the stationary point, brute-force methodology, overlay plots and canonical methods or mathematical approaches like desirability function, objective function, sequential unconstraint minimization technique (SUMT) or Lagrangian method can be adopted.

The graphical approaches tend to earmark a region, with acceptable levels of responses, within which an optimum is located by “trading off” different responses. Besides, the methods for extrapolation outside the domain viz. steepest ascent (or descent) method, optimum path method, model-independent sequential method, sequential simplex, evolutionary operations (EVOP) or machine-based computational techniques like Artificial Neural Networks (ANNs) can also be employed for optimum search.

An appropriate computer software like Design Expert®, Design Ease®, Minitab® is usually considered essential for FbD while handling the numeric calculations involved in the postulation of various models, RSM and optimum search.

Overall FbD strategy for drug delivery development
Overall, Figure 9 depicts the various salient steps involved during an FbD strategy as a whole in the form of a flow chart.

FbD optimization of oral DDS: Literature instances
Almost all types of orally administered DDS have been reported in literature to be systematically optimized using FbD. Besides screening, products as well as processes have been systematically developed using FbD. Table 1 provides a pithy account of the same.

BBD: Box-Behnken design; FD: factorial design; CCD: central composite design; SSD: spherical symmetric design; PBD: Plakett-Burman design; FFD: fractional factorial design; DCM: dicholoromethane

To conclude, a product development scientist can derive unique benefits of FbD for the development of innovator’s as well as the generic drug products. Understanding the formulation or process variables rationally using FbD can greatly help in achieving the desired goals with phenomenal ease. As a rule, when finding the correct solution is not simple, a pharmaceutical scientist should mandatorily consider the use of FbD for drug delivery development.