The Formulation and Process Development of a Novel Multiparticulate Extended Release Pharmaceutical Delivery Platform for a Highly Water Soluble Compound

This research compared a traditional drug development approach with an enhanced "Quality by Design" (QbD) approach to foster greater process and formulation understanding. Propranolol HCl extended release capsules and Metoprolol Succinate extended release tablets served as targets for development. The formulation and process parameters utilized well-established techniques, such as wet granulation, extrusion, spheronization, and fluid bed processing with commercially available aqueous or organic polymeric systems. Propranolol HCl extended release capsules were a benchmark for current generic pharmaceutical process development to identify basic parameters for a suitable product. Metoprolol Succinate extended release tablets utilized the ICH Q8 annex guidelines approach to identify target profiles, and then; define, test, and link Critical Quality Attributes (CQAs). Preliminary data supported factor and level selections for the 2 full factorial and Box-Behnken experimental designs for elements of a conceptual design space. Physical and chemical characterization of commercially available competitors established product target profiles. Particle size distribution, sphericity, moisture content, and dissolution profiles were studied as CQA's. Preliminary studies for the immediate release beads identified the water quantity during granulation, kneading time during granulation, and the duration of spheronization as significant factors for particle size generation, and sphericity. The traditional approach determined an organic system was necessary for Propranolol HCl to yield a stable extended release product. The aqueous sustained release coating studies for Metoprolol Succinate found the polymer coating level, humidified curing condition, and curing duration were important factors . Altering the excipient blend formulation during initial tabletting trials minimized segregation and bead damage. The full factorial design for the development of immediate release beads identified spheronization time as a statistically significant factor in determining the standard deviation and relative standard deviation for sphericity. The Box-Behnken design for sustained release beads found the polymer coating levels to be a statistically significant main effect for the dissolution profile. Significant surface damage was apparent throughout the full factorial tabletting design, with a "best case" approach yielding an improved dissolution profile. Traditional approaches incorporated into the QbD approach facilitate variable and level selection throughout the development process. Data generated via these statistical methods supports process understanding and future decision-making. ACKNOWLEDGEMENTS I would like to express my gratitude to the following people for their unwavering support throughout this process: Dr. Needham, Dr. Palinaswamy, the Staff at Cherokee Pharma Inc., my friends and family.


Introduction
This chapter serves to orient the reader with the challenges associated with drug development and the use of organic solvents in the pharmaceutical industry.
The objectives and overview of the approach to the research proposed are described.
Additionally, the potential benefits of this work for the pharmaceutical community and the general public are presented.

Statement of the Problem
Tablet manufacturing has been described as a "paradox"; formulating and manufacturing a mass of raw materials into a usable product is complicated, and then upon administration, the tablet must release the active ingredients in the desired manner requiring additional design considerations (Swarbick 2007). To improve formulation and process understanding during development, aspects of the Quality by Design (QbD) approach described by the International Conference on Harmonization (ICH) to guide experimentation can be used to enhance innovative efforts in the marketplace. This approach is different from the traditional method of drug development by supporting a risk-based approach to the identification and evaluation of critical parameters that negatively impact product characteristics.
Experimental design to evaluate formulation and processing parameters with benchmarks gained through reverse engineering (product characterization of marketed competitors) serves as an improved model for generic drug development. 1 Utilizing reverse engineering techniques to understand how the products are formulated and manufactured is not a novel concept. Generic pharmaceutical manufacturers use reverse engineering to understand the performance of innovator products in order to model their formulations to mimic the marketed product closely enough to gain FDA approval. While this method of deconstruction to copy the original product may be effective, it does not improve the product design or manufacturing practices. Instead of attempting to duplicate ineffective, expensive, or other disadvantageous methods, improvements can be made to the drug product, delivery system and manufacturing process to provide a well understood generic product. This conceptual generic product development strategy will serve as the intellectual basis of this work and will provide generic manufacturers with an understanding of the advantages of implementing these methodologies. As a result of improved product development, consumers of generic products will ultimately reap the rewards of a more competitive market with theoretically lower costs and decreased time to market entry.

Objectives of the Study
The primary aim of the study is the comparison of formulation and process development for two extended release water-soluble pharmaceutical compounds, propranolol HCl and metoprolol succinate. Propranolol HCl development will largely be empirical, from raw ingredients to a final sustained release capsule using a traditional approach. General information gained through the traditional development process of propranolol HCl will be used to support decisions for a QbD 2 developmental approach of metoprolol succinate extended release tablets. While the development of a single drug product is dependent on its chemical and physical properties, the traditional approach can be understood and improved to yield a stronger process understanding. Process understanding can be gained through the use of multivariate statistical approaches to support a design space. Utilizing a systemic approach facilitates understanding of the material attributes and process parameters that are linked to a drug product's critical quality attributes (CQA's). Figure 1 is an adapted overview of the ICH Q8 Annex strategy for Pharmaceutical Development, from concept through commercialization. The Q8 strategy provides a general framework of drug development, which will be applied to metoprolol succinate throughout this research. Throughout the development and product life cycle, changes in formulation and manufacturing practices offer opportunities to gain greater knowledge of product characteristics and performance under various conditions. Inclusion and analysis of relevant experimental and experiential information can be used to create, support, and expand the control and design spaces. While absolute operational and process understanding is impossible, it is important to recognize what is known and operate within those parameters. This project will use the sequential approach presented by the ICH Q8 annex to guide the enhanced product development approach to support elements of a conceptual design space. Scale up, development of a control strategy, and product life cycle management are important aspects for commercialization, but are beyond the scope of this work. Due to equipment and financial limitations, development will avoid costly high tech machinery and utilize a practically based QbD approach of statistical methodology and process understanding to support elements of the design space.
A secondary objective is the evaluation of aqueous systems in place of organic solvents where feasible, as an environmentally friendlier alternative. It can be hypothesized that both brand and select generic products utilize organic solvents in selected aspects of their manufacturing. Solvent usage is not limited to sustained release coatings, and may be used during the production of immediate release beads.
Therefore, both the immediate release bead preparations and the sustained release coatings provide areas where aqueous systems may be explored.  The formulation and process development aspects of the project are divided into four phases based on common processing steps where intermediate testing occurs. Figure 2. serves as an overview of the study approach for drug product development, with processing steps presented in squares, and potential variables as rounded squares. Phase I deformulates the competitor's products and uses physical, chemical, and published literature to support the development of product profile targets. Phase II evaluates the formulation and process development of an immediate release bead, from raw ingredients to final dried beads. Phase III is the development of a sustained release bead, which spray coats and cures the polymer system onto the immediate release beads. Phase IV topcoats the sustained release beads, before blending and compressing them into multiparticulate sustained release tablets.

Study Approach
Identification and evaluation of intermediate metrics will facilitate a comprehensive system understanding in addition to the final product characterization. Evaluation throughout the process will bolster process understanding and a systemic approach to support formulation and production changes. Each stage of development has an impact on the final product; therefore it is essential to understand processing ranges and yields for each stage.
Drug development does not occur in a vacuum, therefore it must be fluid and flexible in order to adapt to changes. For example, the development of spherical immediate release beads in a controlled particle size distribution is an important intermediate goal. The traditional approach would strive to optimize the batch yield and proceed to the next phase. Optimization of a poorly understood process is not a wise allocation of time and resources. In contrast, the goal for the immediate release 7 beads should be to evaluate the output over a reasonable processing range to understand the impacts of variability and processing conditions. Therefore, when the second phase of development begins, sustained release coating and curing, and the data suggests an adjusted particle size distribution, the necessary changes are made based on process understanding.
The ICH Q8 approach for drug development is divided into four stages for

Introduction
Generic pharmaceuticals are a vital element of healthcare and often deliver quality medications at a fraction of the price of the brand product. This chapter aims to familiarize the reader with the current regulatory, legal landscape and competitive climate of the generic pharmaceutical market.

Regulatory Background of Generic Pharmaceuticals
The "Drug Price Competition and Patent Term Restoration Act of 1984'', commonly referred to as the Hatch-Waxman Act, legalized the approval of Abbreviated New Drug Application (ANDA) submissions for generic equivalents to currently marketed products (Mathias, Dole et al. 1984 In addition to demonstrating bioequivalency to the innovator product, new generic drug products must be adequately labeled, and manufactured in compliance with good manufacturing practices (cGMP's) for the FDA to approve the ANDA (Holovac 2004) Generic pharmaceutical manufacturers can make minor formulation modifications, such as limited excipient (inactive ingredients) substitutions, and changes to the manufacturing method, such as processing equipment and procedure, as long as the final product is found to be bioequivalent to the innovator.
Manufacturers must use caution about radical formulation changes to a referencelisted drug when submitting an ANDA. The FDA states, "Any product variations because of differences in excipients (e.g. absorption enhancers or hydrophobic agents) or other changes in formulation that may significantly affect absorption of the active drug ingredient or active moiety should be submitted in separate applications (FDA 1998)." Meeting regulatory requirements and a manufacturer's desire to improve a product's process and/or formulation must be balanced. Advances in manufacturing technology, equipment, and materials should be explored in order to gain a competitive advantage in the marketplace. While the exclusivity associated with being first to the generic market is highly desirable, an efficient formulation and process will reap long-term profits and allow sustainability throughout the product lifecycle.  (Joneckis 2006) to encourage innovation in the pharmaceutical industry.

PAT Background
The FDA defines PAT as: A scientific, risk-based framework intended to support innovation and efficiency in pharmaceutical development, manufacturing, and quality assurance. The framework is founded on process understanding to facilitate innovation and risk-based regulatory decisions by industry and the Agency. The framework has two components: (I) a set of scientific principles and tools supporting innovation and (2) a strategy for regulatory implementation that will accommodate innovation (FDA 2004a).
The definition for PAT presented does not define the overall strategy of the initiative, but serves as an introduction to the regulatory practice. A practical definition of PAT is given as "systems for continuous analysis and control of manufacturing processes based on real-time measurements, or rapid measurements during processing, of quality and performance attributes of raw and in-process 14 materials and processes to assure acceptable end product quality at the completion of the process" (Hussain 2002).
These revised guidance documents have created a new era for the pharmaceutical industry. The documents shift away from conventional thinking, an example is a current guidance document for the submission of products using on line process controls in place of end product sterility testing for terminally moist heat sterilized products. Parametric release is defined as "a sterility assurance release program where demonstrated control of the process enables a firm to use defined critical process controls in lieu of sterility testing . .. (FDA 2008c)." Traditionally, these sterile products were subject to end product testing, which sampled a small amount of material and was limited to identifying only the most serious of contaminants due to scientific limitations. Through greater process understanding, the decision was made to validate and control the process parameters to monitor the product bioburden. The process understanding, approach supports an environment of continuous improvements.

Design Space-Background
The new focus is to understand the product, the manufacturing process, and operations. This approach has been described as the "design space", defined by the FDA as: . .. the multidimensional combination and interaction of input variables and process parameters that have been demonstrated to provide assurance of quality. Working within the design space is not considered as a change. Movement out of the design space is considered to be a change and would normally initiate a regulatory postapproval change process.
Design space is proposed by the applicant and is subject to regulatory assessment and approval (FDA 2006c).
Within the design space, the control space has been described as a: "Multidimensional space that encompasses process operating parameters and component quality measurements that assure process or product quality. It is a subset of the design space" (Desai 2006). Exploration and understanding of these areas will lead to the identification of critical parameters, as well as metrics and methods to capture their impact on the process, enabling quality management through a risk based approach. A new control strategy aims to minimize risks associated with failures when critical and non-critical process parameters fall outside the control space but remain within the design space.

Quality by Design (QbD) Background
The FDA has recognized, similarly to outside industries, that quality must be built into the design of the product, and that it cannot be achieved through testing or inspection alone. In order to promote the idea of incorporating quality into product development, the ICH adopted "Quality by Design (QbD)." QbD is defined as, "A systematic approach to development that begins with predefined objectives and emphasizes product and process understanding and process control, based on sound science and quality risk management (ICH 2007)".
By the standards of the 21st century the development and manufacturing of pharmaceutical products is generally considered to be inefficient when compared to other industries (FDA 2004a). The FDA has introduced cGMP's for the 21st century (FDA 2004b) to facilitate the improvement of pharmaceutical manufacturing through the use of a risk-based approach. Figure 3 depicts a conceptual representation of the hierarchy of manufacturing control strategy. Data generated through experimental designs provide the framework for creating and supporting the manufacturing practices.
Changes to the control space, within the design space, supported by adequate data as defined by the manufacturer would conceivably not require supplemental FDA approval.
,.  I  I  I  I  I  I  I  I  I   I minimize the risks associated with failures when the critical and non-critical process parameters fall outside the control space but remain within the design space. The ICH presents the concept of Critical Quality Attributes (CQA), defined as "A physical, chemical, biological, or microbiological property or characteristic that should be within an appropriate limit, range, or distribution to ensure the desired product quality (ICH 2007)." Ideally all CQAs should be evaluated early in development, but the process of continuous improvement necessitates vigilance throughout the development and life cycle of the product as guided by the ICH QlO documentation. The first step is to identify all of the factors, which may impact the product throughout the process, and then to identify those factors with the theorized greatest impacts to evaluate experimentally. Changes in formulation and manufacturing practices offer opportunities to gain greater knowledge of product characteristics and performance under different conditions. Inclusion and analysis of relevant experimental and experiential information can be used to create, support, and expand the control and design spaces. Figure 4. is an example of an Ishikawa diagram, which can be used as a tool to identify key areas of interest in a tableting process.

Plant Factors
Precoll'l>ressing  Tablet   Table I is an overview of the differences that can be found between the "minimal" (traditional), and "enhanced"(QbD) developmental approaches. The ICH recognizes in the Q8 Annex that manufacturers will most likely utilize tools from both of these approaches, with their processes' described between the two extremes.

Raw Materials
The emphasis is to incorporate these techniques from the initial stages throughout the entire product life cycle. A hybrid approach between the two extremes, would help to introduce new techniques to well established systems to encourage improvement.
For many small to medium manufacturers, an incremental approach may be the only economically feasible option. Knowledge gained from utilizing the QbD approaches on a small scale or through a partial implementation can serve to guide improvement for future expansions and products. The FDA has given the pharmaceutical industry an opportunity to enter the 21st century of manufacturing by expanding the control space of their operations while still staying within safe operating conditions in the design space. PAT, the design space, and QbD has excited many in the pharmaceutical industry, who see the potential to continuously improve processes as they occur in other manufacturing industries.
This drastic change in the regulatory mindset has raised many questions regarding feasibility and practicality. The FDA has promoted the adoption of these techniques on a general level. For example, PAT has become an industry "buzzword" with much of the knowledge only attainable through consultants, and/or is guarded closely within the industry. Current seminars and workshops address very specific ideas of using novel techniques for limited areas, such as specific assays used during processing to verify quality (Tyler 2006). Some consultants may propose vague "buzzword" laden approaches to improvements without scientific background and appear to be more oriented to selling their services. Widespread adoption of these new approaches has been slow due to a lack of technical knowledge and trepidation over the interpretation of these guidance documents by regulators.
Moving toward an "enhanced" developmental approach is still in its early stages with skepticism and uncertainty of how the FDA will regulate this new area.
While the long-term benefits of manufacturing improvements are clear, it may be difficult to make the argument for a sizable investment in innovation on a product currently being developed for fear of clinical failure and the uncertainties of FDA product approvals. Further, if a product will be approved, companies want to avoid delaying or jeopardizing the approval process due to changes in manufacturing approaches. Additionally, there are concerns over technological limitations, which prevent online process monitoring and understanding for complex products such as protein drugs (Glaser 2006).

Organic Solvents
The advantages of aqueous systems over organic solvents has been widely accepted in the pharmaceutical industry for over the past 20 years (McGinity 1997).
Organic solvents require specialized equipment and recovery systems for production.
Due to the volatility of the solvents, explosion proof equipment and processing areas are necessary. Systems for solvent scrubbing and recovery are necessary to minimize the environmental impacts and contain waste materials (Olsen 1989).
Organic waste produced in the manufacturing of pharmaceuticals including: liquids, volatile gasses, and solid materials, all contribute to make the pharmaceutical industry one of the leading producers of organic waste. It is estimated that "the pharmaceutical industry has the highest waste generation and the highest amount of organic waste used per mass of product produced for any commercial sector (Slater and

Class 2 Solvents: Solvents to be Limited
Non-genotoxic animal carcinogens or possible causative agents of other irreversible toxicity such as neurotoxicity or teratogenicity. Solvents suspected of other significant but reversible toxicities. Examples: Methylene Chloride, Methanol, Xylene.

Class 3 Solvents: Solvents with Low Toxic Potential
Solvents with low toxic potential to man; no health-based exposure limit is needed. Class 3 solvents have PDEs of 50 mg or more per day. Examples: Acetone, Isopropyl Alcohol, Ethanol (ICH 2005).
Specialized equipment, waste disposal, analytical testing for residuals, and many other considerations raise the cost of organic processing (Olsen 1989). Ideally, aqueous systems can be substituted for organic solvents during development.
Unfortunately due to pharmaceutical feasibility, organic solvents are still used where aqueous systems are inadequate. In cases where organic solvents are necessary, the risk classification and environmental impacts should be minimized.

Legal Background
Traditionally, manufacturers in many industries would develop novel products and kept their methodologies secret. These "trade secrets" became targets of commercial espionage and could be stolen by competitors. In order to protect innovation, patents become an important and useful barrier that manufacturers use to block or slow their competitors. A general "utility" patent offers 20 years of protection from the date of filing, and allows multiple claims on a single patent (USPTO 2006 2) Patenting the processing methods or synthesis of the API and/or drug product.
3) Patenting the formulation of the drug product (API and excipient blend  14,2008). In this fast paced and competitive market, strong formulation and process development is key to a manufacturer's long-term growth and survival.

Chapter Review
This chapter gives a brief overview of the current legal and regulatory climate of the generic pharmaceutical industry, and addresses challenges associated with developing a generic product. Opportunities for advancement in to identify different approaches in drug development.

Reverse Engineering Background
Theoretically, the simplest method for one manufacturer to create an identical drug product as their competitor is to steal the formulation and a copy of the master Reverse engineering is defined as "the process of extracting the knowledge or design blue-prints from anything man-made (Eilman 2005)." Classically general manufacturers used reverse engineering techniques in order to gain insight into the methods of their competitors. Semiconductors, software source code, and protected digital media have received protection from reverse engineering techniques because of the efficiency of reverse engineers (Samuelson and Scotchrner 2002). Currently the EEA and TRIPS laws do not specifically sanction or condemn the practice of reverse engineering, which leaves room for enforcement interpretation.

Deformulation Background
The term "deformulation" is often used interchangeably with reverse engineering, but must be considered as a tool within the broader definition of reverse engineering. Business sectors related to the pharmaceutical industry such as the polymer and paint industries, utilize deformulation techniques. A general definition of deformulation is "A comparative analysis of unknown materials, utilizing product specific methodology to separate and identify each unknown component in the formula" (Chen, Tseng et al. 2001 ). In the paint industry this methodology may be applied for several different reasons, such as: a loss of documentation during formulation, to investigate a competitors product to ensure that there is no patent infringement, investigation of whether competitors marketing claims are supported by their formulation, and the identification of advantages held by the competition (Bruck and Willard 2006). While the polymer industry is complex and requires extensive methodology to understand the molecular weights, copolymerization, and other necessary attributes for optimal performance (Nuwaysir, Wilkins et al. 1990), deformulation can provide critical information.

Pharmaceutical Deformulation
In the pharmaceutical industry, the term deformulation is often used when referring to understanding the genetic structures of existing bacteria and viruses in order to reengineer the function to meet their objectives. Two examples of current published research are: the reverse engineering of bacteria to create highly efficient antibiotic producing organisms (Lum, Huang et al. 2004), while vaccine discovery has been expedited and improved by applying these strategies to genomics (Zagursky 2003  While it is not required, it is customary for the manufacturer of oral dosage forms to list inactive ingredients used in the final product. The inactive ingredients listed are useful for the formulation scientist, but information about quantities, molecular weights or grades of materials, and methods must be determined in order to fully understand the target product. This aspect is most important when a branded product is no longer available and there are only generic equivalents on the market. The process of product characterization is product specific, but general techniques utilized for the dosage form are useful. For solid oral dosage forms such as tablets and capsules, an important characteristic is the dissolution profile of the product. While this information may be available through research journals or other references, it is important to understand that manufacturers do experience variability within their approved specification ranges. This can be especially true for controlled release products, with acceptable but significant batch-to-batch dissolution variations. Therefore, looking at multiple lots and/or multiple manufacturers is an important step in understanding fluctuations which may occur (Kanfer, Walker et al. 2004). Additional areas for consideration are the formulation ' s structural characteristics, chemical and physical parameters.

SEM
SEM is a quantitative method for identifying physical and structural parameters of controlled released oral dosage forms. Images obtained using SEM are of high resolution and have a three dimensional appearance attributed to the instrument's large depth field, which allows a large amount of the sample to be in focus at one time (Crowder, Hickey et al. 2003). Multiparticulate systems can be analyzed to gain a stronger understanding of the manufacturing techniques utilized.
By separating the internal components of active and inert ingredients of the multiparticulate system, the formulation scientist can use SEM to focus on the active system in the formulation. Images of the surface morphology can be useful to understand the film integrity and the impact of damage during processing. Cross sectional imaging can give valuable information on polymer coating thickness and core structure characteristics (Metha and Jones 1985).

Particle Size Analysis
Sieving is a useful method for determining the particle size distribution of the discharged product. Sieves are usually cylindrical open containers with predetermined and calibrated mesh sizes stacked upon each other with increasing aperture size, with the largest openings on top to catch agglomerated particles, and the smallest screen on the bottom to remove the fines (Crowder, Hickey et al. 2003).

4. 3 Dissolution
Dissolution testing has been used extensively for drug development and the four approved apparatuses are described in detail in the USP chapter <711> (USP 2007a formula of C 1 6 H 21 N0 2 ·HC1, and the chemical structure found in Figure 5 Table 5. is an overview of the metoprolol succinate formulation patent developed by (Sriwongjanya, Yuk et al. 2008), which is the property of Watson Pharmaceuticals via the acquisition of AND RX Pharmaceuticals. This formulation utilizes a water/acetone solvent system for the sustained release polymer coating and follows a similar process as the innovator. Curil!_g Colloidal Silicon Dioxide Anti-Caking 1% Table 6. is the patented formulation by Ravishankar, Patil et. al. 2006, with ut a two-polymer system approach for their dosage form. The Eudragit® NE30D polymer acts to quench the ionic charges of the salt cores and create a suitable surface for drug layering. While the Eudragit® RS30D acts to slow the release of the Metoprolol Succinate from the drug loaded cores. The authors targeted a "chronopharmaceutical" profile, which attempts to control the release of drug based on the circadian rhythm to optimize the therapeutic dosing regimen. The in vitro and in vivo release of the drug was designed to slowly release the drug initially and then 40 more rapidly release the drug after the 7-hour period. A traditional zero order release profile would require optimization of both polymeric layers to adequately match the brand product and sustain the release over a 20 hour period. Additionally, the research focused on the sustained release pellets, which were not tabletted into a final dosage form. Compression of the coated pellets cannot be overlooked, as there can be a significant impact on the release profile.  Table 7. is Dias 2007 formulation, which utilized a multi step formulation on the inactive microcrystalline (Celphere®) cores. Opadry is a film coating system composed of polymer, plasticizer, and pigment to form an adhesive film with high tensile strength (Colorcon 2008a). The seal coat acts as a barrier between the Metoprolol drug layer and the sustained release layer. The Surelease polymeric suspension system acts to sustain the drug's release over the dosing interval. The topcoat is necessary to prevent "blocking," a static buildup that prevents material flow at the end of processing of the Surelease polymer (Colorcon 2006 therefore it is difficult to hypothesize the performance of this delivery system.
Toprol XL (Metoprolol Succinate) extended release 1 OOmg, and 200 mg tablets (Astra-Zeneca), potassium phosphate monobasic, sodium phosphate monobasic, phosphoric acid, sodium hydroxide, and analytical grade deionized water prepared in house. The water system consists of Millipore® pre filters for municipal water to feed a Waters® Elix 5 reverse osmosis system, and is stored in a 60-L Waters® Reservoir equipped with a UV Automatic Sanitation Module® (ASM), which feeds the Waters® A-10 Gradient reverse osmosis system equipped with a terminal 0.22 µm filter, where the analytical water is dispensed.

SEM Sample Preparation Methods
The contents of sample capsules were emptied while tablet samples were placed in approximately 1 OOml of purified water, USP and manually agitated until the drug and water-soluble components were in solution. The remaining insoluble material was screened and the active beads were separated from the other excipients.
Samples were air dried, cross-sectioned with a razor blade, and analyzed by scanning electron microscopy (SEM) on a JSM-5900 JEOL (Japan). Magnification and background settings were adjusted based on the sample type. Whole pellet and cross sectional samples were imaged and measured using the instrument's software. One set of Inderal LA 160mg beads were studied after completion of the USP dissolution procedure.

3.J0.2 Physical Characterization and Particle Size Evaluations
Propranolol HCl capsules were weighed, emptied, and the beads were sieved through screens #16-30 mesh. U.S. Standard Test Sieves (Newark, Clifton, NJ), which meet ASTM E-11 (Formerly the American Society for Testing and Materials) specification were used to screen the material. The quantity retained on each screen was weighed and calculated as a percent of the total weight.
Toprol XL Tablets were weighed and measured with a Mitutoyo® CD-8"CSX digital caliper. The tablets were gently broken apart and screened through sieves #25-60 mesh. U.S Standard Test Sieves (Newark, Clifton, NJ), which meet ASTM E-11 Specification were used to screen the material. The quantity retained on each screen was weighed and calculated as a percent of the total weight.

Dissolution-Propranolol HCl
A Hanson® SR-8 dissolution apparatus was used in accordance with the USP test method 1 for propranolol hydrochloride extended release capsules. Key settings: apparatus 1, basket speed 100 rpm, 2 phase media, 900 mL of dissolution media. 0. lN hydrochloric acid buffer with a pH of 1.2 for the first 1.5 hours and a sodium phosphate monobasic pH 6.8 buffer from 1.5 hours to 24 hours. The bath temperature was maintained at 37 ± 0.5°C throughout the dissolution. lOmL samples were taken at the sampling time points, with the acceptance criteria found in Table 9. Concentration was primarily determined following the USP method using a Hewlett Packard® 8453 UV spectrophotometer at 320 nm absorbance. The general method was confirmed by high performance liquid chromatography following the USP method. Not more than 30% 4 Between 35% and 60% 8 Between 55% and 80% 14 Between 70% and 95% 24 Between 81 % and 110%

A Hanson SR-8 Dissolution apparatus was used for the USP test method for
Metoprolol Succinate extended-release tablets. The key settings are: Apparatus 2, Paddle speed 50 rpm, pH 6.8 phosphate buffer, 500mL dissolution media volume.
The sample volume was 5mL with sampling time points and acceptance criteria in Table 10, additional and alternative time points may be sampled where appropriate.
Concentration was primarily determined using high performance liquid chromatography. This calculation method was used to determine the concentration of the drugs throughout the dissolution period to account for concentration lost due to sampling.

.4 Concentration Determined by High Performance Liquid Chromatography
Propranolol HCl and Metoprolol Succinate were assayed for concentration on a Waters® 2695 separation module equipped with a column heater, an auto sampler, 47 and degasser. The system was equipped with a Waters® 2487 dual/.., absorbance detector for UV Nis detection. All methods had to meet system suitability criteria of less than a 2.0 % relative standard deviation (RSD) of five injections of the reference solution. Additionally, an in process standard was injected after every six injections of sample solution and with a reference standard which must be less than 3. 0 % RSD. The concentration was determined using equation 4.

Equation 4. Formula for Calculating Percent Content of Drug by HPLC Assay
Where: AT Area of the drug peak from the sample solution.
AC Area of the drug peak from the reference solution.
WR Weight (mg) of reference substance taken.
WT Weight of the sample in mg.
MW Average mass of the capsule or tablet content in mg.

PF
Purity factor % Purity of working standard on an as needed basis.

LC
Label claim for content of drug per unit.
I 00 Percent conversion factor

I I. 5 Propranolol Hydrochloride Assay Method
The Propranolol Hydrochloride USP monograph assay method was used, an overview of the method is given here. The system was equipped with a Symmetry shield®, RP18, 150 mm x 4.6 mm 5µm and the UV detector was set to 220 nm. An isocratic method was employed, with a flow rate of 2.0 ml/min, a sample injection volume of 1 O µL, an average retention time of 2-4 minutes, and a sample run time of 6 minutes. A Phosphate Buffer with a pH of 3.0, and a mobile phase of blended buffer at and acetonitrile (75:25 respectively).

J.11 . 6 Metoprolol Succinate Assay Method
The Metoprolol Succinate USP monograph assay method was used, an overview of the method is given here. The system was equipped with a Nova-Pack®, Cl8, 150 mm x 3.9 mm 4µ column, and the UV detector was set to 280 nm. An isocratic method was employed, with a flow rate of l .5ml/min, a sample injection volume of 25 µL, an average retention time of 2-4 minutes, and a sample run time of 7 minutes. A phosphate buffer with a pH of 3.0, and a mobile phase of blended buffer and acetonitrile (75:25 respectively) were used in the method.

SEM and Deformulation of Competitors
The deformulation of the competitors provided formulation and manufacturing strategies to guide developmental targets for metoprolol succinate and propranolol HCl. This section is divided by product and manufacturer to examine the SEM results and the inactive ingredients. Manufacturing techniques and formulation approaches are supported by patents, literature, and the experimental SEM images. For cases where concrete information was unavailable, formulation function and utilization are based on traditional methodology and a hypothetical "best guess." Due to a lack of sample availability, the Mylan formulation of Propranolol HCl was not investigated, fortunately the formulation is similar to the Actavis product, which can be used to understand the general development approach.

J 2. 2 Results-Wyeth Inderal LA Extended Release Capsules
Based on early patents and the current list of inactive ingredients in Table 11, the brand product consists of a simplistic formulation utilizing organic solvents.
Drug loaded cores are prepared via wet granulation with microcrystalline cellulose using alcohol, and/or methylene chloride, or an alcohol/water mix as the granulating agent (Guley, DeMeals et al. 1981) The sustained release coating is comprised of an ethylcellulose and Hypromellose blend in a methylene chloride/methanol solvent mix in a range from 1.5-5% (Guley and Farina 1992).

Figure 10 Propranolol HCl ER Caps 160mg-Cross Section after Dissolution
The porous core after dissolution, seen in Figure 10, indicates that the production method utilized a wet granulation with extrusion/spheronization approach to create pellets of the desired size. The material remaining after dissolution are water insoluble excipients, such as microcrystalline cellulose, ethylcellulose, etc .. ..

3 Results-Par Propranolol HCl Extended Release Capsules
The Par formulation is similar to lnderal LA, based on the inactive ingredient list in Table 12.and the SEM figures. The Par formulation appears to have followed a similar formulation approach to the brand product.  Figure 11 is the whole bead image for the Par formulation of propranolol HCl at a magnification of 1 OOx. The bead appears to have a smooth defect free surface morphology, and a bead sphericity near 1.0.

Figure 11 Par Propranolol 160mg
Figure 12 is alOOx magnification of the cross-sectioned bead, with the drug loaded core and the sustained release polymer coating visible. Figure 13 is a 2500x magnification of the edge of a cross sectioned bead, which shows a thicker coating of polymer than the brand product. More polymers are likely necessary due to the smaller particle size distribution yielding a greater surface area.

4 Results-Actavis Propranolol HCl Extended Release Capsules
The Actavis formulation in Table 13 is markedly different than the extrusion/spheronization techniques used by the other manufacturers, using a drug layering approach on sugar spheres. The solubility of Propranolol HCl in water would likely result in an extremely long processing time, and may employ an aqueous polymer coating. The Actavis formulation utilizes a drug layering approach, which may utilize an organic, aqueous or organic/aqueous solvent system. Research has effectively used ethanol/water (60:40 respectively) to load the required drug amount (Dashevsky and Mohamad 2006). Aqueous systems alone have been used to layer the drug (Percel, Vishnupad et al. 2002). Either approach would conceivably be a time consuming process due to the drug's solubility and the quantity necessary to achieve the target potency (Jones 1989).

Metoprolol Succinate
Metoprolol succinate brand and generic marketed products were deformulated using SEM imaging, physical and chemical techniques. The list of ingredients is presented with their theorized function to describe the formulation approach. SEM imaging was performed on the whole bead to identify surface morphology and sphericity. Cross-sectioned imaging was used to identify drug and/or polymer coating thicknesses and manufacturing strategies, such as drug layering vs. extrusion/spheronization. Table 14 is an overview of the formulation and theorized function for Toprol XL. Based on a formulation and process patent (Jonsson 1990 (Abrahamsson, Alpsten et al. 1996). The small inert cores have a large surface area and utilize a combination of the following organic solvents:

J I Astra-Zeneca (Brand)
isopropylic alcohol, ethanol 95%, and/or methylene chloride in order to avoid agglomeration during drug layering and sustained release coating to ensure a uniform distribution of the material.

62
The beads appear to be spherical, based on the SEM image in Figure 17, with a sphericity of the beads near 1.0. The surface is relatively smooth and without visual cracking or defect. Figure 17 Astra-Zeneca 25mg-Whole Pellet Figure 18 is a cross-sectioned bead at 1 OOx magnification; the bead is comprised of the silicon dioxide core, a drug layer, and a sustained release layer. Figure 19 is a 250x magnification focused on the sustained release layer of the bead.
Sizing was calculated by the system software and gives an approximate range between 32-45µm.  This formulation appears to use the greatest amount of different inactive ingredients to create a viable dosage form. The impact and cost of each excipient should be well understood when creating a generic product. While the individual excipient may not be particularly expensive to purchase, it is important to consider the costs associated with maintaining cGMP material for production and the extensive handling and documentation required for use. Additionally, all of the inactive ingredients must be examined to ensure that there are no compatibility issues or long-term stability implications. Based on the processing methodology and magnitude of excipients, this formulation is theoretically the most expensive and identifying specific excipients with their functions is challenging. Figure 20 is a whole pellet and displays a dumbbell like shape, which may be attributed to the immediate release bead processing during extrusion/ spheronization.

2 Ethex
The bead is not as spherical as the brand product, and visual observations indicated a significantly larger particle size distribution.  Figure 21 is a 75x magnification of a cross-sectioned bead, which does not contain an inert core or separate drug layering. Figure 22 is a 1 OOx magnification of the sustained release polymer level with approximately 9-11 µm thickness. Based on the ingredient list and the SEM images, the Ethex® formulation is the only currently marketed product theorized to be manufactured using a wet granulation process.  Table 16 contains the Sandoz formulation for metoprolol succinate, with the theorized function for each ingredient. This formulation is unique in the approach to sustained release coating by employing Methacrylic acid copolymer alone. While all three grades of Methacrylic acid copolymer are used for enteric coating, each has a specific pH solubility, which allows for targeted colonic delivery.    ?ft.  Propranolol HCl in Figure 27 shows a significantly larger amount of polymer applied, which may indicate that an aqueous polymer was employed for the intermediately distributed beads.  19.40 Mix ofTo_IJ_ Coat and comj)ressed exciQients Total 1030.9 This small sampling of branded tablets yields an estimate of the target end process particle size for the SR beads prior to tabletting. Excipient granules and beads could be differentiated and estimated with the naked eye. Table 18 shows that qualitatively the bulk of the active beads are retained on sieve sizes #35 and #45.

Metoprolol Succinate Particle Size Distribution Results
The excipient granules are spread out over the range of the sieves, with a significant 73 amount of fine excipients comprised of crushed excipient granules during tabletting and fine excipients (i.e. lubricant).
74    Table 19 shows that all of the marketed products have similar fill weights, and provide a reasonable fill weight to target.

~
All of the products used size 1 capsules to encapsulate the extended release beads. The Mylan brand was unavailable for evaluation. The Metoprolol Succinate 1 OOmg tablets are slightly more than half of the weight of the 200mg tablets, which supports the dose proportionally of the RLD.

2 Results-Metoprolol Succinate Extended Release Tablets-Dimensions/Fill Weight
The high dose tablets were selected for initial development, 25mg and 50mg tablets were not evaluated. Table 20 provides information about target weights and tablet dimensions for tooling selection. target in anticipation of potential formulation or processing difficulties. The sieving studies identified a controlled particle size distribution for both Propranolol HCl and metoprolol succinate. The Propranolol HCl competitors each had a tailored particle size distribution but varied in their tailored approach, with the brand product having the largest mean particle sizes. The Metoprolol Succinate beads were significantly smaller than the Propranolol HCl beads, and also showed a tailored approach to particle size distribution. The dissolution profiles of both Propranolol Hcl and Metoprolol Succinate brand products were within the USP acceptance criteria, and served as the targets for drug release for the target products.

Introduction
This chapter gives an overview of current techniques used to formulate and process multi particulate sustained release in order to support the selection of Critical Quality Attributes (CQA's). Wet granulation, extrusion/spheronization and fluid bed drying are the primary processing techniques explored for immediate release pellets, which were utilized in the development of the experimental dosage forms in order to provide a viable platform for the water-soluble drugs. The second area of interest is in polymeric membrane formulations and processing conditions required to create a viable controlled release. Finally, a basic overview oftop coating and tabletting are discussed to support the selection of the CQA's.

Stress-Strain Relationship
Throughout the manufacturing process ingredients are subject to stress and strain, which can impact product properties. Tensile strength testing is used to calculate stress ( cr ), where the force applied to the material is divided by the crosssection of the sample and strain (i::) where; the resulting length of the sample is divided by the initial length   (Hoag, Vivek et al. 2008). Beyond the "yield" (elastic limit), the material is experiencing irreversible plastic deformation, until it reaches a final "rupture" point where fracture occurs. Fracture is a disruption of the continuous connective nature throughout the system, which can result in a weakened unit

J Granulation
Common reasons for granulating pharmaceutical material described by (Parikh 2005) include: Increased drug distribution uniformity, densification of material, enhanced flow rates and rate uniformity, easier metering or volumetric dispensing, dust reduction, and improved product appearance.
Extensive research has been conducted on difficulties associated with the granulation process, examples include studies focused on: high shear granulation (Devay, Mayer et al. 2006), extrusion/spheronization (Chatchawalsaisin, Podczeck et al. 2005) (Sriamomsak, Nunthanid et al. 2007), and fluid bed processing (Lipsanen, Antikainen et al. 2007) (Rajniak, Mancinelli et al. 2007). Common general equipment used in wet granulation processes are: high shear mixers, low shear mixers and fluid bed granulators. While each method has its distinct advantages and disadvantages, high shear granulation will be explored in this research.

Wet Granulation
Wet granulation uses liquid to wet the seed particles to create controlled conditions of agglomeration. Liquid binding forces are responsible for the particle size generation, while solid bridging is the key factor for granule strength (Crowder, Hickey et al. 2003). Figure 30. gives an overview of the rate processes of agitative agglomeration that includes: wetting, growth, consolidation, and breakage. The  (Ansel, Allen et al. 1999). The sized particles can then be further processed with sustained release coating or blended with excipients for tab letting.
Water-soluble drugs are well suited to wet granulation methods which can use water as the granulating liquid to provide adhesion of the particles to one another. Water acts to raise the contact angle of water soluble materials at the solid liquid interface to enhance the distribution of the drug throughout the mass (Cantor, Augsburger et al. 2008). Challenges associated with this approach are found during 85 the establishment of a compatible uniform formulation due to possible drug migration and robust processing parameters to create a pharmaceutically viable product (Allen, Popovich et al. 2005).  Figure 31 is a schematic of a typical vertical high shear granulator for wet granulation. Dry material is introduced to the granulator through the load port and can be mixed with the main impeller to achieve a uniform dry blend. Dust created 86 from dry materials is captured in the product filter to prevent external contamination.

High Shear Wet Granulation
Granulating liquids, such as water or organic solvents, are introduced through the spray port during main impeller and/or chopper mixing. When an acceptable granule is formed, the wet mass exits the system through the discharge port.
Granule growth in a high shear granulator is largely dominated by coalescence or layering. Coalescence is the agglomeration of materials based on collision and binding of granules to one another. Binders facilitates fine particle adherence to larger particles, to achieve layering (Gokhale, Sun et al. 2005).
Adequate water is essential for success for both of these methods of granule growth.
Two types of water are inherent in the system: internal water, which is captured within the particles during agglomeration, and "free" surface water remaining from the addition of granulating liquids. Both forms of water are necessary to create bonding strength and plasticity to allow coalescence and layering (Ghebre-Sellassie 1989).
High shear granulation has been shown to create denser particles with lower porosity to slow disintegration times compared to alternative methods of single step extrusion granulation (Keleb, Vermeire et al. 2004b) and fluid bed granulation (Gao, Jain et al. 2002) approaches. This method's unique physical properties are associated with the high shear granulation's adhesion of material with water before repeated cutting and compaction to yield a denser and harder material (Gao, Jain et al. 2002).
Research has shown the importance of shear on a granule's growth and its final properties. (Oulahna, Cordier et al. 2003) found that higher impeller speeds during processing result in lower granule porosity and friability, with a narrower particle size distribution overall. This research also found that granule size is a critical factor for a granule's properties regardless of the impeller speed and increasing shear alone does not result in more homogenous granules.

2. 4 Extrusion
Extrusion can be defined as "a method of applying pressure to a mass until it flows through an orifice or defined opening. It is a technique that determines two dimensions of an agglomeration of particles (Hicks and Freese 1989)." The two dimensions of the particles defined by extrusion are: 1) the cross sectional diameter which is a function of the screen size the material passes through.
2) The length of the extrudate, which is dependent on the formulation and processing parameters.
There are a number of different types of extruders, but all achieve the same objective of converting a wet mass into cylindrical particles. A wet mass can be created using high or low shear granulation, and is forced through a screen with holes of uniform diameter to create spaghetti like rods. The extrudate hangs down and breaks under its own weight into similar lengths. The critical formulation parameter during extrusion is the material's plasticity, which must break but avoid adherence to other particles during spheronization (Mehta, Singh Rekhi et al. 2005).
Water in the formulation, added during granulation, aids the extrusion process by increasing the plasticity of the material, and provides lubrication to the die during processing (Tomer and Newton 1999 al. 2007), use of twin screw extruders (Keleb, Vermeire et al. 2004a), and hot melt extrusion (Andrews, Jones et al. 2008). Equipment modifications and operational parameters have been studied to evaluate non-formulation based variables. The study of water distribution and loss during extrusion is important for aqueous wet granulation formulations. Research has demonstrated water migration during slower speeds of extrusion resulting in wetter extrudates early in the process (Tomer and Newton 1999). The authors associated this phenomenon with slow extrusion speeds allowing water greater time to travel through void spaces to the die.
Additionally, the loss of surface water has been associated with evaporation due to a rise in temperature of the extruder and die during processing ).

2. 5 Spheronization
The spheronizer was patented over 40 years ago to rapidly create small (<2.0 mm diameter) uniform spherical granules using crossing grooves to cut and rub the material (Nakahara 1966). Figure 32 is comprised of A) Motor to drive the system. 89 B) A cross-hatched friction plate, which spins to shape the material. C) A sidewall to contain the product. D) A material loading port. E) A material discharge port.

------+--+t-Motor
Figure 32 An Adapted Schematic of an Early Spheronizer (Nakahara 1966) Spheronization occurs after extrusion, where the cylindrical particles are broken into short lengths by contact with the rotating frictional plate, and collisions at the particle/particle level and the particle/wall interface to create spherical shapes with nearly uniform diameters. The centrifugal force generated by the rotating plate throws the material to outside of the plate where it climbs up the sidewall before 90 gravity results in a tumbling ("rope like") motion back to the friction plate, where it repeats the same cycle (Nakahara 1966). The cross hatch angle, pattern, and groove distance (space between edges) can all be adjusted to improve the efficiency of spheronization for different sized beads (Hicks and Freese 1989). (Reitz and Kleinebudde 2008) evaluated jacketed vessels for temperature control during processing to alter product viscosity, plasticity and sticking more basic spheronizer research focuses on spheronization speed and duration.
Spheronization speeds are generally evaluated to yield an optimal rope like movement of the material (Dukie-Ott, Remon et al. 2007) and adequately densify the material . Spheronization cycle times directly impact the sphericity of a particle by increasing the number of collisions the material undergoes during a cycle, and serves as an important area for research (Pinto, Lameiro et al. 2001). In general, the longer the cycle is run the greater chance the material will be round, but there is also an increased chance for unintended particle size growth.

2. 6 Fluid Bed Drying
~ i .' . . · · · · · · : : : : The air passes through a lower screen (distribution plate); screen mesh sizes are changed for different products to accommodate varying loads and to alter the flow pattern and restriction of airflow to lift the product. Heated air will lift and dry the wet mass; the drying rate is described by Equation 5: dw dt Equation 5. Drying Rate  Where aw is the mass transfer rate (drying rate), h is the heat transfer dt coefficient, A is the surface area, His the latent heat of evaporation, and !1T is the temperature difference between the air and the material surface. Heat transfers to the material during drying to supply latent heat to evaporate the liquid, while simultaneously mass transfers as the internal liquid/vapor diffuses and evaporates from the surface .
Physical interactions with other particles, chamber walls, and the air distribution plate will aid in the breakage of agglomerated particles that may have formed during earlier stages of processing. A critical balance must be maintained during fluidization for drying; over fluidization will break the particles and create excess dust resulting in concentration or excipient loss, while under fluidization can leave particles wet, agglomerated and unsuitable for further processing (Olsen 1989).
The air then passes through the expansion chamber, where cooling occurs, and to the filter housings.
Filters may consist of bags, cartridges, or a drying screen similar in design to the air distribution plate on the bottom of the chamber. Filter pore size must be selected carefully to balance adequate airflow to sustain fluidization, and prevent the escape of API or excipients. Generally, a smaller pore size is chosen when drying become clogged during processing, thereby limiting the airflow. The filters are shaken or "blown back" using compressed air to reintroduce the drug or excipients to the immediate release pellets to prevent concentration loss during processing. While every effort is made to keep the material off the filters and on the product, filters may gradually become clogged and decrease the fluidization level. It is critical to monitor the filter pressure and product pressure for machinery equipped with these gauges. Adjustments to the fan speed during processing may be required to maintain an optimal fluidization height and must be evaluated to understand and control the process parameters (Parikh 1991).
An "air blast" or a "bed blast" can be utilized to lift the product off the air distribution plate during the process if necessary. This technique closes the valves above the filters on the top of the chamber to allow vacuum pressure to build just above the product chamber. The valves are rapidly opened and a large amount of vacuum pressure pulls the material off of the air distribution plate. The abrupt movement can lift and break up wet materials that may have formed large agglomerates. Standard operating parameters designed to avoid over agitation of the product and may not sufficiently lift the wet materials early in the drying stages when it is heaviest. After the fluid bed is discharged, it is important to separate process generated agglomerates and fine particles (also referred to as "fines") from the usable material (Olsen 1989).
Sieving is performed to remove agglomerated particles, fines, and undesirable particle sizes that may be generated during processing. This controls the particle size of the immediate release pellets that will be subsequently coated for sustained release.

Fluid Bed Spray Coating
Product Container  When spraying polymeric membranes onto drug loaded beads, it is important to understand the impact of the starting material. Figure 3 5 is a useful estimator of particle surface area based on diameter, in order to estimate the required polymer quantity. As the particle size increases, the surface area and the amount of polymer required for 1 mg/cm 2 decreases. This is important when considering the target formulation 's particle size distribution to understand the impacts on polymer coating

Curing
After completion of the aqueous sustained release coating process, the film applied may not completely coat the immediate release bead. This is due to the incomplete coalescence of the polymer particles into a homogenous film, resulting in variable drug release from the beads (Bodmeier, Guo et al. 1997). Curing is a processing step that may use heat and/or humidity to facilitate the rapid coalescence of the polymer into a uniform coating to protect the beads to create a uniform drug release profile. The extreme conditions of curing, such as: high temperature, high humidity, and long durations of curing have been studied. The results are highly dependent on the polymer coating material used and cannot be generalized among 97 polymer systems (Siepmann, Muschert et al. 2008). Curing polymers has also been found to reduce the brittleness when compared to uncured beads, which is advantageous during tabletting to minimize bead crushing (Abbaspour, Sadeghi et al. 2007). For Aquacoat ECD aqueous dispersion, the manufacturer recommends a curing cycle of two hours at 60°C (FMC-BioPolymer 2006). Curing approximately 10-12°C above the Tg allows relaxation of polymer chains and an alteration of the film wetting properties reduces instability during storage for beads coated with an Aquacoat ECD and DBS (Wheatley and Steuemagel 1997).

Bead Top Coating
A protective excipient blend is of primary importance to minimize bead crushing during tabletting, top coating aqueous ethylcellulose beads may offer an additional level of protection during processing (Dias 2007). Mannitol, polyethylene oxide, polyethylene glycol and microcrystalline cellulose have also been explored as protective excipients to provide cushioning, especially for high potency beads in poorly uniform blends (Torrado and Augsburger 2008 (Hoag, Vivek et al. 2008) Direct compaction may be a viable approach to a limited number of products as an effective tabletting method. These products do not require adjustments to their flow properties, density, and/or particle size distribution, to yield appropriate physical parameters for an acceptable dissolution profile (Parikh 2005 Varying granule properties such as: particle size, sphericity, and porosity can all influence the granular deformation during tabletting. Plastic deformation is primarily seen during tabletting, but porous and irregular granules may experience fragmentation and breakage (attrition) resulting in rapid drug release (Hoag, Vivek et al. 2008).
One common tabletting technique matches the active and excipient particle sizes in order to gain better homogeneity to minimize segregation due to the flow properties of the powders during the blending steps. If homogeneity and adequate flow is not achieved, the dissolution properties, tablet weight, hardness, and manufacturability may all be adversely effected (Crowder, Hickey et al. 2003).

. 7.1 Fillers/Bulking Agents
There are a number of fillers and bulking agents used to create a core structure for the active ingredient. Examples include: lactose, sugars, dicalcium phosphate, starch (pregelatinized), and microcrystalline cellulose ( MCC is ubiquitously found throughout pharmaceutical manufacturing in a multitude of functions, available in a range of mean particle sizes and grades. It is primarily used for oral capsules/tablets as a binder and diluent in wet granulation and direct compression (Weller 2003). MCC primarily undergoes plastic deformation during compression, in contrast to crystalline lactose and sucrose which may be more prone to experience fracturing (Hoag, Vivek et al. 2008). MCC's addition to an excipient blend improve the plastic characteristics during compression and lubricant efficiency to protect sustained release coatings on the beads during compression (Torrado and Augsburger 2008).
Smaller mean particle sized grades ofMCC (e.g. Avicel PH-101 -50µm (FMC-BioPolymer 2008) are utilized in wet granulation, due to their beneficial rheological properties as a wet mass during extrusion and spheronization (Faure, York et al. 2001). Avicel PH-102 (-100 µm) (FMC-BioPolymer 2008) may be used during tabletting to provide excipient protection for smaller coated beads, but potential formulation dependent segregation issues must be understood (Torrado and Augsburger 2008). Larger sized grades of MCC (e.g. Avicel PH 200 -200 µm) have been found to yield lower tablet weight variations, while maintaining compactibility similar to smaller sized materials (Doelker, Massuelle et al. 1995). In addition to MCC ' s enhanced compactibility, wetting and drying occur at a rapid and even pace to prevent variable distribution of soluble ingredients in the granule (Cantor, Augsburger et al. 2008).

7.2 Granulating Agent
A variety of granulating agents can be employed depending on the physicochemical properties of the dry mass of excipients and active ingredient(s).
Water, ethanol, acacia, alginate, pectin, HPMC, sodium carboxy methylcellulose (CMC), polyvinylpyrrolridone (PVP), citric acid and calcium chloride solution (in water) were studied as granulating agents for the wet granulation process (Sriamornsak, Nunthanid et al. 2007). This research found that the higher viscosity agents resulted in dumbbell formation, while low viscosity watery agents with calcium chloride, which reduced the swelling potential of the excipient blend and yielded desirable spheres Water has become the primary granulating agent of choice, with organic or hydro alcoholic solvent blends employed when hydrolysis of the active ingredient or other concerns are apparent (Cantor, Augsburger et al. 2008). The amount of water used during granulation will dictate the properties of the intermediate materials throughout process. If the mass is over wetted during granulation agglomerates may form during the spheronization process and yield an undesirable particle size distribution and shape . Conversely, ifthe mass is not sufficiently wet to lubricate the extrusion die, excessive die pressure and heat due to frication can result in a failed process (Tomer and Newton 1999).

7. 3 Binding Agents
Binders provide cohesion for bonding of solid particles to promote size enlargement to produce granules and improve the blend flow during processing properties. There are a number of binders which are commonly used and can be divided into three categories, see Table 21 .

Polymers
Polymers can be made from a variety of natural or synthetic materials, and are used throughout the pharmaceutical industry to conquer challenging formulations. Cellulose is a natural polymer found in the fibrous tissues of cotton and wood, which pass unchanged through the human digestive tract (Kim 2004

8.1 Methylcellulose
Hydroxypropylmethyl cellulose (HPMC) is soluble in water, isopropyl alcohol (IP A), and hydro alcoholic mixtures depending on the grade of material (Harwood 2002). The polymer is used in many areas of the pharmaceutical industry, including film coating and sustained release matrix tablets. The HPMC acts to form viscous gels to control the diffusion of water and drug release and when it is combined with a water insoluble polymer, such as ethylcellulose, it facilitates the creation of a "non-continuous film" (Dow 2008), or commonly referred to as a pore former. HPMC can be used as an effective pore former when combined with ethylcellulose for both organic and aqueous systems. Upon hydration HPMC acts to open channels within the polymeric matrix and facilitates drug release (Bodmeier, Guo et al. 1997). Research has indicated that the use of HPMC can result in flocculation and unstable coating systems when used with aqueous dispersions of ethylcellulose (Ong 2006b) The flocculation was overcome when the dispersion was adequately mixed, which is consistent with standard processing parameters (Ong 2006a (Gunder, Lippold et al. 1995). Overall the challenges of coating with HPMC in aqueous systems are considered greater than in organic systems, due to stricter processing controls to prevent undesirable results (Nagai, Ohara et al. 1997).

8. 2 Ethyl cellulose
Ethylcellulose is a hydrophobic polymer, used to coat granules and/or tablets to create a sustained release profile (Dahl 2002). Ethylcellulose is brittle due to inter chain hydrogen bonding and bulky glucose subunits and requires plasticization (Bodmeier and Paeratakul 1994). The hydrophobicity of the polymer necessitates that it be dissolved in an organic solvent or be dispersed in an aqueous system.
While aqueous or organic systems may be employed for ethylcellulose, the film formation mechanism is different. Organic solvents dissolve the ethylcellulose and other components to form a film when the organic solvent evaporates leaving the individual polymer molecules in contact (Osterwald 1984 ).
In contrast, aqueous systems undergo the following film forming process: 1) Individual polymer spheres containing hundreds of polymer chains dispersed in water coat the surface. 2) As water evaporates, the interfacial tension between the remaining water and polymer spheres increases and results in an ordered arrangement of polymer spheres. 3) Capillarity resulting from the increased interfacial tension provides a driving force to overcome the repelling forces and cause deformation to fuse and coalesce the particles together (Wheatley and Steuemagel 1997).
In addition to avoiding the negatives of organic solvents, highlighted in chapter 1; aqueous systems offer several additional advantages: the lower viscosity dispersion allows a greater amount of polymer to be applied per unit of volume, and lower water vapor transmission rates due to the coalescence of small latex spheres (Wheatley and Steuemagel 1997).
Aquacoat® ECD and Surelease® are commercially available ethylcellulose aqueous dispersions. The Surelease® system is ready for use after the addition agitation of water to the desired solids concentration, and a 15 minute mixing time (Colorcon 2006). A disadvantage of the Surelease® system is the need for additional barrier coating, adding processing time and material expenses. In addition to ethylcellulose, the Aquacoat® ECD system contains, cetyl alcohol, and sodium lauryl sulfate which act as an emulsifier and stabilizer, respectively. The Aquacoat® ECD system requires the addition of a plasticizer, and a 30 minute mixing time before it is ready for application (FMC-BioPolymer 2006).

Plasticizers
Organic and aqueous ethylcellulose dispersions require plasticizers due to their brittle nature to prevent cracking, improve flexibility, reduce the polymer's glass transition (Tg) temperature to promote uniform film (Bodmeier, Guo et al. 1997). Plasticizers provide flexibility to the polymer by increasing the free space between the polymer chains and decrease rigid polymer-polymer binding (Aulton, Abdul-Razzak et al. 1981). Additionally, plasticizers facilitate water uptake into the film to improve the coating's permeability to drugs (Lippold, Gunder et al. 1999). Aquacoat (Tarvainen, Sutinen et al. 2003). DBS is initially emulsified in the water phase of the system, prior to incorporation into the polymer phase within 30 minutes of mixing (Bodmeier, Guo et al. 1997

Other Inactive Ingredients for Tabletting
Other inactive ingredients that are used in the formulation of tablets are lubricants and disintegrants. Lubricants act to overcome the increased friction generated during compression between the tablet and die walls (Armstrong 2008).
Examples of lubricants used in tabletting include: Stearic acid and its salts (calcium, magnesium, zinc), hydrogenated vegetable oil, waxes, and mineral oils (Kottke and Rudnic 2002).
Tablet disintegrants are defined as, "Any solid, pharmaceutically acceptable material included in the formulation that acts to cause the tablet matrix to break up when the tablet comes into contact with aqueous media (Moreton 2008

Particle Size Distribution
The particle size distribution generated at the completion of the granulation process is identified as a potential CQA in the ICH Q8 Annex. The importance of the particle size distribution on down stream processes is widely recognized and must be monitored (ICH 2007). Products with tailored particle size distributions have narrower specifications for acceptance criteria, with poor process control resulting in highly variable and poor yields (Dukie-Ott, Remon et al. 2007). The yield can have a significant impact on the profitability and sustainability of a product throughout its life cycle. For sustained release products, understanding how the formulation and process effect the immediate release bead particle size distribution is critical. Variable particle size distribution will yield varying surface areas for coating and can drastically affect the dissolution profile of the system. Figure  In Dashevsky's research, both pellets were coated to 20% w/w gain with a polymer for modified release. It is clear from Figure 36. that the smaller sized pellets release the drug faster due to an increased surface area and consequently a thinner polymer coating than the larger particles. While smaller pellets in general undergo less mechanical stress during compression and improved distribution into the void space, pellet rupturing appears similar for both sizes for this research (Dashevsky, Kolter et al. 2004)

9. 2 Sphericity
The sphericity of a pellet indicates the roundness of material, by direct comparison of perpendicular lengths of the material or by how well the material rolls . The closer the measured ratio is to 1.0, the rounder the material. Rounder sustained release coated pellets have been associated with a more uniform drug release, than irregularly shaped material (Chopra, Alderbom et al. 2002). This research concluded that rounder pellets had more uniform polymeric coating, while dumbbell shaped pellets experienced disproportionate coating at the body and edges resulting in variable drug release. Mathematically, rounder pellets will increase the surface area for hydration and drug release.
The pellet sphericity can also impact down stream processes. Spherical pellets will move through the manufacturing process easier and create a more uniform dosage form. Conversely, pellets with an aspect ratio of> 1.2 have been found to cause variability during filling operations for capsules (Chopra, Podczeck et al. 2002

9. 4 Dissolution Profile
A simplistic definition of the dissolution rate is "the amount of active Ideally the formulation provides a Level A correlation, but often the anticipated bioavailability cannot be adequately projected through dissolution alone prior to performing the in vivo studies. While it is an imperfect method, it is widely considered as the most sensitive predictor of in vivo performance (Banakar 1991 ).
Chapter 5 Linking Materials and Processes to CQA's-Risk Assessment

Introduction
This chapter outlines the general approach for each phase of drug development to initially link the materials and processes to CQA's. The Risk Assessment approach used Ishikawa diagrams to help identify potential factors for study. The materials and methodology used for this section and throughout the rest of the study are given. The results of the traditional/exploratory batches are given to guide the design of experiments.

0.1 Phase JI: Formulation and Process Development of an Immediate Release Drug Pellet
Phase II-The De-.elopment of an Immediate Release Pellet initially be evaluated for Metoprolol Succinate using exploratory batches to determine feasible operating parameters. The factors will be narrowed down and the operating levels will be fixed for a factorial experiment to support the QbD approach. Traditionally, non-aqueous and organic solvent systems were utilized to enhance the solubilization of polymers. A clear polymeric solution will form uniform coatings on substrates with reproducible film characteristics to create an extended release product. Solvent processing can be seen in product patents (Dhalinder 1993), but are generally not described in depth in literature, because it is viewed as an intermediate processing step; a classic example can be seen in the formulation of Metoprolol Succinate (Ragnarsson, Sandberg et al. 1987) (Sandberg, Rangarsson et al. 1988).
Due to environmental hazards, pollution, and costs associated with handling, processing, and disposal of solvents, some manufacturers may prefer aqueous polymeric coating systems to solve their needs for polymer application. However, more environmentally friendly and theoretically cost effective over the long term, aqueous polymer coatings can be difficult to process due to their delicate nature.
The coating process is highly dependent on processing conditions, with inconsistent results often due to a lack of process control or inappropriate processing parameters.
Additionally, aqueous polymer dispersions may experience variability or inappropriate physicochemical properties dependent on the polymer(s) selection (Porter and Bruno 1990). Figure 41 is an Ishikawa diagram for the development of sustained release beads, which describes potential variables for exploration. The immediate release pellets developed in Phase III will be coated with aqueous polymers first to evaluate the feasibility of the system against benchmarks set by the marketed product. For products, which cannot reasonably use aqueous systems, an organic solvent system will be employed. For Propranolol HCl, Ultra Violet (UV) Spectroscopy was used for concentration determination, and the method was verified by high performance liquid chromatographic (HPLC). Metoprolol Succinate did not use UV spectroscopy due to interference and concentration was determined via HPLC. Scanning electron microscopy (SEM) or other visual imaging methods will be used to evaluate the surface morphology and coating patterns on the pellets. Pellets will be sieved to determine the particle size distribution. Additional physical characterization will be performed as needed.

.2.Excipient Granule (Formulation #1) for Metoprolol Succinate SR Tabs
Microcrystalline Cellulose NF (PH102), Crospovidone NF K29/32, Pregelatinized Starch (1500) NF, Lactose Monohydrate NF, and Purified Water, USP were granulated. All dry ingredients were first screened through a #25 sieve to remove large particles, prior to processing. Low shear wet granulation was performed on the Jaygo 10-L mixer, at a low impeller speed without the chopper blade. The granules were dried in a Freas® Scientific 625 oven, until an LOD of less than 5% was achieved. The granules were milled in a Fitzpatrick® Comminutor, with a screen size of 60 and knives facing forward.

Propranolol HC! IR Bead Processing
The approach to development was to evaluate a selected factor while holding the other parameters constant to determine their effects, and to use visual observations to make adjustments. Early development utilized 400g batch sizes, while the final three lots RB005054, RB005055 , and RB005056 were 800g batches to evaluate the process parameters effects of doubling the batch size. Propranolol HCl followed the general processing procedure described above. During wet granulation, impeller blade speed and chopper speeds were fixed throughout water addition and kneading. A l .Omm stainless steel screen was used during extrusion to yield larger extrudates. The experimental process parameters can be seen for Propranolol immediate release bead production in Table 22. An example of the fluid bed drying cycle and processing parameters for the spheronized moist pellets can be seen in Figure 44. The moisture content specification was set to not more than 2.5%, based on the initial moisture content of the active ingredient and the excipient to ensure an adequately dry product.

General Metoprolol Succinate JR Bead Processing
Metoprolol Succinate followed the general processing procedure described above. During wet granulation, impeller blade speed and chopper speeds were fixed throughout water addition and kneading. A 0.6 mm stainless steel screen was used during extrusion to yield finer extrudates. An example of the fluid bed drying parameters for the spheronized moist pellets can be seen in Figure 45. The moisture content specification was set to not more than 2.5%, based on the initial moisture content of the active ingredient and the excipient to ensure an adequately dry product.

3. 3 Exploratory Metoprolol Succinate JR Bead Processing
Initial exploratory batches were run based loosely on the processing parameters observed during the development of the Propranolol HCI IR beads. The raw and coded process parameters can be seen in Table 23

3. 5 General Spray Coating Methods
Drug loaded immediate release pellets produced in phase II were spray coated in the Mendel Fluid Bed® processor. For aqueous trials, the inlet temperature was set to 65°C, the product temperature was maintained between 35°C-45°C.
Organic trials operated at an inlet temperature of 50°C, and maintained the product temperature between 32°C-42°C. The fan speed was adjusted during processing to maintain adequate fluidization. The Wurster column height was fixed at a setting of 6, the spray nozzle had an internal bore size diameter of 0.8mm, the spray rate was between 2-5g/min with an atomizing air range between 0.2-1.2 bar and Tygon® 3350 tubing used to deliver the coating material via an external peristaltic pump. The machine was adequately purged during organic solvent processing to prevent an explosion.

3. 6 Aquacoat ECD Trials-Propranolol HCl
Immediate release beads were coated with an Aquacoat ECD aqueous dispersion. Dibutyl Sebecate, NF acted as a plasticizer, and Hydroxypropylmethyl Cellulose E6, was used as a pore former, both were added as fixed percentage of polymer weight throughout experimentation. Varying polymer concentrations were evaluated at 6 different levels from 4-15% polymer level. Screening studies will utilize subsampling to determine the effects of polymer levels within a batch.
Subsamples and final samples were cured where appropriate.

. 3. 7 Aquacoat ECD Solids Concentration Trials-Propranolol HCZ
Immediate release Propranolol HCl beads were coated to the same polymer weight gain(% w/w) using two different concentrations of polymer in the coating suspension. The low concentration coating suspension formula had 40% less solids than the high concentration coating suspension. The coating suspensions were prepared and administered following the same procedures in the coating level trials.

Organic Ethylcellulose Coating Trials
Immediate release Propranolol HCl beads were coated to two different levels of polymer weight gain, but contained the same solids content. The polymer levels were significantly lower than the aqueous polymer levels applied. Curing was performed at various conditions and durations in the VWR® 9005 Stability Chamber.

Aquacoat ECD Trials-Metoprolol Succinate
Immediate release beads were coated with an Aquacoat ECD aqueous dispersion. Di butyl Sebecate, NF acted as a plasticizer, and was added as fixed percentage of polymer weight throughout experimentation. Varying polymer concentrations were evaluated from 3-30% polymer level. Screening studies utilized sub sampling to determine the effects of polymer levels within a batch. Subsamples and final samples were cured where appropriate.

3.10 Curing Trials Overview
Uncured and cured material will be compared to understand the effects of curing. Non-humidified curing was performed in either the VWR® 1415 Vacuum oven (without pulling vacuum), or the Freas® Scientific 625 convection oven. While both humidified and dry curing trials will be performed in a VWR® 9005 Stability Chamber. The baseline curing for all batches was performed in dry heat at 60°C for 2 hours. Curing studies compared dry heat, humidification, and duration were performed to understand the effects on the sustained release beads.

Methods-Sustained Release Bead Top Coating
Cured sustained release beads were charged into the Mendel Fluid Bed (MFB-1) processor. Inlet temperature was set to 65°C, product temperature was maintained between 40-45°C, the fan frequency was adjusted during the coating process to sustain adequate fluidization, the coating suspension was sprayed at approximately 3g/min, through a nozzle with an inner bore diameter of 0.8mm, and Tygon 3350 tubing was used. After the top coating suspension was applied, the beads continued to fluidize for 5-15 minutes to allow the beads to dry. A wide range of top coating quantities was studied, from low (<3%) to high (>15%) to evaluate the cushioning and protective effects of the topcoat.

3.12 Tabletting Methods-Blending
A 1: 1 ratio of excipient blend to active top coated beads were blended, with the exception of Lubritab® which was added at 3-8% tablet weight at the final stage of blending. A Patterson Kelley® V-Blender, equipped with a 4-qt shell, and a shell speed of 25 rpm was used for blending. All excipients except the Lubritab® were blended for 4 minutes, top coated Metoprolol succinate beads were then added and blended for 4 minutes, and finally Lubritab® was added and blended for 5 minutes.
The final blend was tested for LOD to ensure an acceptable level of moisture in the tablet.

3.13 Methods-Tabletting
A Natoli® Type BB bilayer tablet press, with a 13/32" (.4062 and end of the run to ensure that hardness was maintained within± 1 Kp.

4.1 Dissolution and Concentration
USP methodology described in Chapter 3 was used.

General Statistical Methodology
Minitab and Microsoft Excel software was employed for the design and analysis of the experiments. Exploratory data was fit into appropriate general linear models to study the significance of their main effects. A Full factorial design was employed for phase II and while a box benhken design was used for phase III. The analysis of variance (ANOVA) and DOE will be used to determine main effects and interactions between independent and dependent factors. Dissolution profiles will be the primary metric by which sustained release bead and final product formulations and processing parameters will be evaluated. Each method, statistical, mathematical or graphical interpretation of dissolution profiles has drawbacks, (O'Hara, Dunne et al. 1998); therefore, appropriate methodology will be chosen on an individual basis. Table 25 gives the coded trial parameters and the percentage of material yielded on each sieve size. The coded variables and the yields from each of the trials 137 were fitted into a general linear model and the ANOVA using the Tukey's method For complete results for the general linear model using the Tukey's method for all of the sieve sizes, see Appendix 2. The Tukey's method was used to compare multiple processes to simultaneously evaluate ifthe means are equal (NIST/SEMATECH 2006b). The Tukey's method is useful for making comparisons across multiple factors and is an extension of the ANOVA method (Cobb 1998  .

5.1 Statistical Methods-Exploratory Batches
Tables 22 and 23 previously presented in the process methods section gives the specific and coded process parameters for Propranolol and Metoprolol, which were investigated during the exploratory batches to create the general linear model and represents unbalance nested retrospective design. This design was used because the trials were not a balanced or a crossed design, where each variable is evenly tested. The results for the full general linear model for the ANOVAs is presented in Appendices 2 and 4, the results section presents the reduced general linear models.
Model reduction was performed in accordance with statistically valid methods.
Variables were evaluated in the full model and removed in a step wise approach with careful consideration of the impacts on the R 2 and adjusted R 2 (Colton 2004).
The main effects plots indicate the major changes in the response value.
Evaluation of the slope is the critical parameter for the main effects plot; with a steep slope indicating a strong effect, and a near 0 slope indicating little effect

6.1 Particle Size Distribution Analysis
U.S. A. Standard Test Sieves (Newark, Clifton, NJ), which meet ASTM E-11 Specification were used to screen the material. Sieve sizes and their dimensional conversions used for both Metoprolol Succinate and Propranolol HCl are listed in Table 24. Deblinding of screens was performed as needed when sample particles size and shape resulted in screen blinding.

Loss on Drying (LOD)
An Ohaus MB200 was used to determine the LOD of the immediate release beads. All samples were at least 5.0g, and the LOD was calculated as the percentage of weight lost after 10 minutes at 105°C.

Sphericity
The immediate release pellet sphericity was determined on a Nikon TE2000-E inverted research microscope set to 4X magnification. The width and length were calculated with the NIS-Elements AR software for at least 20 beads per sample and used to calculate the aspect ratio, see Figure 46. for a screenshot of the sphericity measurements.  The sample weight taken was as close to 6.5g of whole tablets for each test, and run for 100 rotations. A maximum mean loss of 1.0% was considered acceptable.

Exploratoryffraditional Results
The traditional immediate release approach is presented in this section for Propranolol HCl and Metoprolol Succinate. The development of the final exploratory drug products is presented separately. Results from these studies will be incorporated into the enhanced QbD approach to support the drug development process.

8.1 Propranolol IR Statistical Results
The ANOV A from the general linear model for the material retained on all of the sieve sizes and the overall target(%) yield are in Appendix 4. Due to its greatest importance for yield, only sieve size 18 results are presented in this section. The only factor that can be considered to be statistically significant from this data is the amount of water added during the wet granulation of the material. The R 2 value of 88.74% and the adjusted R 2 of78.59% is a good indicator of fit for the data.
Appendix 2 presents the full general linear models for all of the sieve sizes used. o Mean Sphericity Figure 48 compares the batches with regard to the amount (adjusted to a 1 % scale) versus the mean sphericity ratio. Note that not all batches were analyzed for sphericity; due to availability and undesirable characteristics (failed batches), which may give the appearance that, all mean sphericities are relatively uniform (1.1-1.23).
Early batches did not produce spherical beads, but served to guide development to improve the shape. A "Target" column was placed on the left side, to indicate the theoretical targets for each value. Due to particle size growth anticipated in the sustained release coating phase a tailored particle size distribution of -65% retained on sieve size# 18 and-35% retained on sieve size #20, and a sphericity near 1.0 was set. The batches can be divided into separate phases throughout the development: Initial processing information, identifying critical processing parameters, confirmation/optimization of those parameters, doubled batch size to understand the impacts of batch size.

Initial Processing Information
The early batches 091907-RB005013 were essential to determining general processing parameters and their effects. Batch RB005008 had the best yield for sieve sizes #18, #20, and overall yield(%). The first consideration during development was to identify the appropriate amount of water during the high shear granulation. Kneading time and spheronization can both impact the particle sizes generated but the granulating liquid will dictate all of the downstream processes (Ghebre-Sellassie 1989) and should be identified and evaluated early on. The batch used a medium amount of water (275g) and was selected as the model for the next developmental phase.

Identifying the Critical Parameters
Lots RB005021, RB005022, and RB005023, were the pivotal batches for the selection of the water level amount added during high shear granulation. Lots RB005021 and RB005023 both used medium (275g) amounts of water, while RB005022 used a high (290g) amount. The particle size distributions were relatively similar between the two lots with a medium amount of water, but the high water batch yielded an unacceptably high level of material retained on sieve #16, and an insufficient yield on sieve #20. Some of the process parameters were varied for each lot to explore their impacts, and RB005023 yielded the most promising results and guided the remainder of the development process.

Confirmation/Optimization of Critical Parameters
In order to optimize the yield of the process, RB005048 was modeled after RB005023 , but varied a number of the process parameters. The granulation water was added at a faster rate (50g/ml to 75g/ml), the kneading time was reduced, and the spheronization speed was increased slightly ( 450 to 460 rpm). These parameters increased the amount of surface water available (Gokhale, Sun et al. 2005) and may have resulted in greater particle size grow during spheronization due to greater adhesive forces in the formulation (O'Connor and Schwartz 1989). These parameters also yielded the best sphericity ratio and supported the decision to move to a larger batch size of 800g. Note: RB005049 was a failed batch that was attempted to optimize the process parameters.

.8.Doubling the Batch Size
The process parameters were then evaluated at twice the batch size to explore the impacts of the increased load. RB005054 was first prepared according to the parameters developed in RB005048, with minor adjustments; the water was added at a slower rate (75 to 62.5 g/min) and slightly slower spheronization speed (460 to 440 rpm) in anticipation of the larger batch size. The batch yielded larger particles than had been anticipated, with unacceptable retention of sieve sizes # 14 and # 16. In response to the failed batch, the addition rate water was reduced (62.5 to 50 g/min), the spheronizer rpm was raised slightly ( 440 to 450 rpm) and the spheronization time was increased (5 to 6 min) to RB005055. This improved the yield to an acceptable level of 84. 7% within the desired target. Finally, RB005056 was run following the same parameters as the previous batch, RB005055, with the exception of an increased water addition rate back to the RB005054 level (50 to 62.5g/min) and yielded the best results of the three batches. The small changes between RB005054 and RB005055 in process parameters may be responsible for a portion of the improvement, but an uncharacteristically erroneous batch cannot be ruled out. This incremental improvement approach was successful in generating basic process parameters and demonstrated reproducibility between the final two batches.

Figure 49 Propranolol IR Bead-Round (RB005055/56 Blend)
Batches RB005055 and RB005056 were blended together and an example of a spherical bead is seen in Figure 49. The pellets are reasonably round, and the surface morphology contained smoothed bumps and appears consistent with acceptable bead formations (Mehta 1989 Table 26 is an overview of the Propranolol HCl sustained release coatings which displayed an acceptable f 2 values over 50.

Results Propranolol HCI SR
The coating levels, concentration, curing methods, and dissolution data for all of the batches is found in Appendix 6. An fi value of 96.3 is nearly point-to-point, and indicates that the coded coating level 2 and low solids content for the coating suspension yield the best results. While higher coating levels 4 and 5 resulted in an overall slower release of the drug from the sustained release beads.  an open system, which sweeps across the heater in a single pass of air before exiting the system, this system operates at a lower relative humidity (5-8% RH) because it can effectively wick away moisture.
When moisture remains in the polymeric coating the pores created by the HPMC E6 remain hydrated and act to channel the drug through the sustained release coating (Bodmeier, Guo et al. 1997). An adequate amount of moisture is necessary to maintain polymer flexibility and prevent cracking, but saturating pores with moisture must be avoided. (Gunder, Lippold et al. 1995) explored an Aquacoat, DBS, and HPMC coating system for a water-soluble compound and determined that in an acidic medium (pH 1.2) the HPMC pores open and will irreversibly close within 1-2hrs. Subsequently, a change to the alkaline media will not reopen the pores, and diffusion proceeds slowly. Additionally, HPMC has been shown to potentially cause flocculation when combined with ethylcellulose and result in an unstable dosage form (Wong and Bodmeier 1996).
Further complicating the balance of moisture in the system is the need to pass long term accelerated stability studies. ICH guidelines dictate that a product must not have appreciable degradation (>5%) in their dissolution profile throughout the stability study. The two acceptable accelerated conditions are 30°C and 60% (RH) for 6 months, or 40°C and 75% (RH) for three months (ICH 2003). Additionally, room temperature must be maintained as a control to confirm normal storage conditions seen by the patients. Preliminary stability data of 10 days at 40c/75% RH confirm that even the vacuum cured beads show an unacceptably sharp increase in drug release. Aqueous ethylcellulose systems with HPMC based pore formers have shown increased drug release over time due to physical changes in the coating layer such as hydrolysis and changes in coalescence, as well as drug migration into the coatings (Siepmann, Siepmann et al. 2005). This instability resulted in the experimentation of an organic solvent ethylcellulose sustained coating solution.  Table 27 is an overview of all of the formulations which could be considered similar to the reference dissolution profile, where f 2 is greater than 50. The complete experimental results for the organic sustained release coating of Propranolol HCl are in Appendix 7.
The organic ethylcellulose coatings are not as markedly adversely affected by curing in the humidity chamber as the aqueous dispersions. This would indicate a more uniform coating of the sustained release layer, able to seal the particle and protect against the moisture in the chamber from saturating pores in the film or pockets of plasticizer in the film (Bodmeier, Guo et al. 1997). Batch RB005072, with a low coating level and a curing condition of 6 hours @ 50°C was selected as the final formulation for long-term stability studies, (not included in this study)S.9.2 Immediate Release Beads-Metoprolol Succinate Exploratory  To adequately assess the effects of the processing parameters, the material generated within the target particle size distribution and the mean sphericity ratio must be well understood. Figure 52. depicts both the mean sphericity and the target yield (adjusted to a 1 % Scale for easier viewing) of the metoprolol succinate IR beads. It is the goal for the two values to be as close to 1 as possible, and can be considered inversely related for the interpretation of this graph. The processing parameters found in Table 26. must be considered when evaluating the output data.
The best target yield was produced for lot RB005109, with greater than 94% of the particles within the desired target particle size distribution. Unfortunately, this batch also yielded the worst mean sphericity ratio with the formation of rods. The formation of the rods may have caused a skewed distribution to appear, specifically with large narrow particles retained on #30 instead of on #25, which was more commonly seen in the other batches.
The best sphericity ratio was found in RB005119, which had initially formed large agglomerates after the completed process and required a re-extrusion and respheronization to achieve a desirable shape. This suggests that the additional processing of extrusion and spheronization facilitated the distribution and removal of water from the system. The water level, l 80g in this batch was in an intermediate range of 140-200g as compared to the other batches but was added extremely rapidly (180 g/min) and did not undergo kneading. Excess water during extrusion can over saturate the material and make the extrudates difficult to process in the spheronizer.
Conversely, insufficient water can create a high viscous mass that is difficult to cut and can result in machine blockage (Keleb, Vermeire et al. 2004b ). These process parameters imply that the water added remained as surface water and due to the lack of kneading did not get fully incorporated into the wetted mass. The primary extrusion and spheronization contained a large amount of surface water, which resulted in the initial agglomeration of the materials. The initial processing also acted to remove a portion of the water from the damp mass. While some water was removed during extrusion via heat transfer and friction, water migration in the formulation was most likely prevalent yielding disproportionately wet extrudates similarly to research conducted by .
The wet extrudates were then spheronized and visible condensation was formed on the cover during processing due to centrifugal force driving the surface liquid off of the beads, and squeezing out the internal water to the bead surface (O'Connor and Schwartz 1989). The spheronizer's purge air supplies compressed air to the system to facilitate the lift of the wet material during processing, which subsequently dries the formulation further. After the initial process was completed the re-extrusion served to redistribute the remaining water into the formulation. The re-spheronization cycle was able to shape the extrudates into a desirable shape, because they contained sufficient water incorporated into the core material and surface water. As the extrudates undergo the spheronization cycle, adequate water must be present at the surface to allow adhesion of smaller particles to reduce the formation of fine particles, but must be balanced with internal water to allow uniform extrusion and particle size growth (Gokhale, Sun et al. 2005 The reduced model ANOV A found that the amount of water used during gr anulation and the speed of the spheronizer (rpm's) are both statistically significant. the reduced model in order to follow the proper reduction approach. Figure 53. shows that lower spheronization speed and lower water amount had greater yields within the target. Lower amount of water will tend to yield smaller particles, while slower spheronization will prevent excessive particle agglomeration. The particle size distribution within the target% yield is important but must be considered with the sphericity data, prior to making any processing decisions.
the reduced model in order to follow the proper reduction approach. Figure 53. shows that lower spheronization speed and lower water amount had greater yields within the target. Lower amount of water will tend to yield smaller particles, while slower spheronization will prevent excessive particle agglomeration. The particle size distribution within the target% yield is important but must be considered with the sphericity data, prior to making any processing decisions.    The amount of water added, the kneading time, and the spheronization time were all found to be significant factors for the mean sphericity of the beads. These findings are in agreement with (Baert, Vermeersch et al. 1993) which found that the amount of granulation water added, the spheronization speed, and the spheronization time are the most important factors that determine pellet sphericity. Figure 54. shows that higher levels of water, lower kneading times, and higher spheronization time all contribute to the improvement (closer to 1.0 is the goal) of the mean sphericity. The sphericity is improved by having more available surface water, which occurs due to the higher level of water added and the diminished kneading time, and longer spheronization time which increases the bead contact with the spheronizer walls and friction plate to improve the roundness. These conditions may negatively effect the particle size distribution and yield larger particles.  while the amount of water does not play a significant role. Higher time in the spheronizer allows the material more time to contact the surfaces for rounding and approach or reach their maximal roundness (Baert, Vermeersch et al. 1993). Lower kneading times will disperse less water through the mass and leave more available surface water during extrusion and spheronization.

Results-Metoprolol Succinate Exploratory Batches-New Approach
Based on the raw data and the statistical interpretations, the greatest challenge is generating a spherical bead, while maintaining adequately sized particles. Table 32 is an overview of the significant variables and the levels, which yield the desired results.
Tabl 32  It is apparent from the table that the water amount necessary is not in agreement. Interestingly, the water addition rate was not found to be significant for any of the responses. Research has indicated that slow addition of the granulating water coupled with kneading would incorporate the granulating liquid into the system and improve the sphericity, the negative impacts of this approach was that the particle size distribution would skew to the larger size (Devay, Mayer et al. 2006).
Conversely, rapid addition of water resulted in smaller particles due to the process of faster agglomeration but may yield poorly shaped pellets (Mehta, Singh Rekhi et al. 2005).
To achieve the correct balance of internal and surface water in the formulation, a new approach was implemented during the high shear granulation stage. An initial slow water addition phase was performed, followed by a rapid addition of water. Previous research had used this technique of slow addition (50g/min) followed by faster addition (lOOg/min) to successfully create dense beads 165 (Gao, Jain et al. 2002). This method theoretically allowed more time for initial agglomeration by forming liquid bridges, and supports particle growth during compaction and cutting to promote bead hardening and densification. The remaining processes followed the results seen in Table 32 The two-phase water addition coupled with the process parameters derived via statistical methodology successfully yielded a sufficiently round bead within the desired particle size distribution. The water amount and addition rate were doubled for the 800g batches, while the remaining process parameters remained the same.
The larger batch yielded a slightly more spherical bead but lost some of the target % yield. The improved sphericity may be attributed to the increased load in the spheronizer, which results in more inter particle interactions, and a greater mass cascading onto the friction plate at the end of the rope like cycle to improve shaping.
Research has shown that load can have both negative and positive impacts on shape and particle size distribution (V ervaet, . The water added during granulation may have been distributed more evenly when kneading due to more advantageous surface to volume ratios than smaller loads. Additionally, variability within the process is expected, and doubling the batch size may not have identical yields if linear parameters are employed. Coating Level (.) 50 .0 -High Polymer (.) ....... concentration's used is approximately half the range that were ultimately used. The development strategy must account for a level of bead crushing, which will be highly dependent on the amount of protection provided by the excipient granule blends and top coating during tabletting (Abbaspour, Sadeghi et al. 2005).

Exploratory Bead Polymer Levels
To hedge against damage to the beads during compression, Figure 57 is a coating trial, which examined a range of polymer levels two to three times of that seen in Figure 56. A significant decrease in drug release is present from the previous data present. The early points of 1 and 4 hour shows that the rate of initial hydration is similar for all of the beads. A divergence in the profiles from the lowest and highest profiles appears at 8 hours due to slower diffusion rates associated with the thicker polymer levels. Polymer levels 2 and 3 are similar and can be attributed to only a 2% coating thickness difference, which may not significantly impact the coating thickness for the standard curing cycle but may have an impact when cured under extreme conditions. Therefore, polymer levels 2 through 4 served as the polymer concentrations studied for the design of experiment factor.

7 Results-Top Coating
Top coating with Opadry II suspension was found to have no negative or positive effects on the sustained release beads, data not shown. Overall, top coating at any level provided an improvement to the sustained release profile of any of the experimental tablets. Top coating the beads up to 15% w/w was studied, and the particle size distributions were found to grow too large and negatively affect the product yield. A final top coating level less than 10% was selected and fixed based on the exploratory studies for the future DOE. Figure 58 and 59 show cross sections of the finalized sustained release coating with a top coating. The topcoat can be seen to be significantly thinner than the sustained release, as it was applied at approximately 25% of the sustained release coating level. In addition to providing added protection, the topcoat imparts an aesthetic white finish to the beads.

2008).
It is important to consider that not all recent research correlates these techniques to feasible end points. One example, utilized formulation and simulation to test the power consumption monitoring method for high shear granulation (Leuenberger, Puchkov et al. 2008). The author's found that a true granule "end point" during processing could not be derived and that a tailored individualized formulation approach was necessary. QbD is a general approach with many tools that can aid process understanding, an exploration of some of those approaches are described in this section.

Statistical Methods-DOE Batches
The 2 3 factorial design were analyzed for the estimated effects and coefficients, and an ANOV A of main effects and 2-way interactions to determine significant factors and interactions. All of the full model outputs can be seen in the Appendices. Proper model reduction techniques were followed where necessary.

Main effects and interaction plots are given for significant factors. Both main effects
and Interaction plots with intersecting lines are indicative of factor interactions for that response, but do not indicate statistical significance.

l Full Factorial Design for Metoprolol Succinate IR Bead Processing
The experimental process parameters can be seen for the full factorial design for Metoprolol Succinate immediate release bead production in Table 34. The factors and levels were chosen based on the findings from the exploratory batches.
The spheronizer was run at 600 rpm for one minute, and checked for excessive balling for all experiments. The spheronizer was then set to 775rpm, run for two minutes, and checked again. Then the spheronizer was run for the remaining time, either three minutes (low) or 6 minutes (high).

.2 Box-Behnken Designs
The ICH Q8 annex recommends the use of response surface techniques to visualize factor effects and interactions to support the design space (ICH 2007).
Two prominent methods of response surface techniques are the central composite and Box-Behnken designs. The Box-Behnken approach has the advantage of requiring fewer total runs because it only requires three levels for each factor instead of five, while remaining a balanced incomplete block design (Wu and Hamada 2000). The surface plots and contour plots that can be generated by this method are useful tools to aid in the visualization of the design space. The Box-Behnken designs have been used in pharmaceutical research such as (Zidan, Sarnmour et al. 2007) and (Shah, Zidan et al. 2007) to characterize critical factors for a novel delivery system.  The coded variables in Table 35 represent the following: -1 is the low level, 0 is the middle point, and 1 is the high level. A standard order (stdorder) is generated in accordance with the Box Behnken design; Minitab then randomizes the order in the Run Order column. The point type (PtType) is coded as 0 to represent the center point, and a 2 represents the face of the cube. The polymer coating level incrementally rises from the low level by 2% w/w, the exact coating levels cannot be described due to formulation confidentiality. The range of curing conditions was 60°C, 60°C/50% RH, and 60°C/75% RH. The curing time range was 2 hours, 4 hours, and 8 hours. Tablet   T  6M  able 3  The standard order (Std Order) of the 2 3 factorial design order is given in the first column of  It is not uncommon to find insignificant correlations between chosen variables and physical parameters. A 2 3 factorial design examining comparing high shear granulation impeller speed and binder flow was not statistically significant for three responses: Time parameter of dissolution, shape parameter of dissolution, and mean particle size diameter (Devay, Mayer et al. 2006). Interestingly, the mean particle size diameter for the responses in the study ranged from ~513 -770 µm, which would result in retention on sieves #35 and #25 respectively. While statistically insignificant, the acceptance criteria set commercially may be based on the desired mean size retained and could imply a practically significant difference. Figures 61 and 62 are examples of dumbbells and rods, respectively, generated during the DOE. These undesirable shapes are visible with the naked eye or can be evaluated with a standard optical microscope. It is important to consider that even in acceptable batches there may be limited formation of dumbbells and/or rods, and screening may not be sufficient to eliminate. Therefore it is critical to understand the overall shape of the batch to anticipate future problems, which may arise in down stream processes.   The estimated effects in Table 38 has the water level at a P value of 0.054, but this does not meet the general acceptance criteria of P <0. 05. This implies that the mean sphericity ratio range of 1.26-1.45 does not have a great enough difference associated with the selected factors at those ranges to be significant. The DOE did find significant results for processing factors with regard to the sphericity standard deviation and the sphericity RSD %. The standard deviation is a key measurement in manufacturing and is used extensively to support Six Sigma approaches for continuous improvement. Where Six Sigma links higher processing variability with product defects, and conversely lower standard deviations are associated with high quality performance (Welch 2003). The RSD is independent of units and is useful to compare the standard deviation with the mean to give an indication of the relative precision of the data (Gardiner and Gettinby 1998c ).

Main Effects Plot (data means) for Standard Deviation
Water   Table 42 shows that spheronization time and all of the two-way interactions are significant for the sphericity RSD %. Table 43 supports the main effects and interaction significance with the ANOV A. Figure 65 indicates that higher spheronization times will yield lower RSD%, while the other factors have minimal effects. Figure 66 indicates that the water level and spheronization time, as well as served as a guide when the batch size was increased to 1.0 Kg. A lack of significant fine particle generation, as reflected in pan retention can be seen in Appendix 3, supports that there was not over agitation during the drying process. The Metoprolol Succinate IR beads were run at the same processing conditions to confirm the initial findings from the DOE. The confirmation trial yielded >93% of the beads within the desired particle size distribution. The batch size was then increased to 1.0 Kg, with the water amount and rate doubling. The yields were both >90% and the mean sphericities showed improvement over the 400g batches. The increased batch size created better sphericity due to greater interactions between the material, the walls, and the spheronizer plate . temperature of at least 48°C to achieve an acceptable LOD with a specification of not more than 2.5%. An optimal LOD range of 1.5-2.0% occurs at a temperature above 48°C, with an acceptable yield at 50 C.

2. 2 Confirmation/Batch Size Increase of DOE Results for Metoprolol IR Beads
"" O"I  Table 45 is the results of the Metoprolol Succinate sustained release Box-Behnken design. The goal of the study was not to yield a high f2 value, because these beads will be further processed and the drug release is expected to increase as a function of the damage experienced during the tab letting process. Based on the exploratory tabletting study there is reason to suspect damage may occur even with an improved excipient blend, therefore a thicker sustained release coating was studied. The f 2 value is given as the main indicator of dissolution profile to the brand product to understand the relative effects of the factors studied.    signifying a potential coalescence effect. The high-level polymer beads cured at 60°C for 2 hours were chosen due to their low drug release rates for processing into multiparticulate tablets.   (Abbaspour, Sadeghi et al. 2005). It is important to consider that in a production environment, these tablets would receive an additional top coat, (i.e. Opadry) to protect against moisture, chipping, damage, and provide taste masking (Kottke and Rudnic 2002). Operating conditions during top coating for equipment such as pan coaters can be carefully selected to meet the tablet characteristics if necessary. In this situation, the friability data helps to support the understanding of the tablets that would further be top coated, but must be considered in concert with the dissolution profile. Table 48 also describes the LOD (%)of the blend prior to compression and shows little moisture differences between the blends. The ANOV A was found to be not statistically significant, see Appendix 9 for the full model.     Figure 78 compares the theoretical "best case" scenario for the tabletting study. Based on the SEM images and the dissolution profiles from the sustained release DOE, one could make the assumption that: the MCC 1: 1 ratio and the high polymer concentration would create the best scenario for protecting the beads during compression. The tablets of the best-case scenario are compared with and without loading the sieve size #25 (the largest) beads of the blend. The best results are obtained for the formulation, which does not contain the #25 sieve sized beads compressed at 2 Kp. Interestingly, there is no significant difference between the formulations at the higher hardness, indicating that crushing of the beads is still occurring. A reduction in the particle size of the immediate release bead could have a positive effect on the final tablet product. The smaller sized beads would require more coating due to an increased surface area but would theoretically be able to fill the void spaces that are created during compression (Dashevsky, Kolter et al. 2004). (Badawy, Lee et al. 2000) found that a finer grade of active ingredient can alter the size of immediate release beads yielded and achieve a shift to a smaller particle size distribution.

Establishment of Target Product Profiles
The first step of the ICH Q8 Annex was to establish the chemical and physical target product profiles. The current marketed competitors for propranolol HCl and metoprolol succinate extended release products were deformulated to establish targets to guide process development.

CQA Identification
Identification of the CQAs for this developmental process was selected based on the ICH Q8 annex guidelines, product deformulation characterization, current literature and available equipment. Particle size distribution, particle sphericity, IR bead moisture content, and the sustained release dissolution profile were selected as the CQA's to be studied. The particle size distribution dictates both the surface area 208 available for polymer coating, and means particle size of beads during compression.
Sphericity was chosen to facilitate bead flow through down stream process, and support even polymer coating application. High moisture contents have been identified as potentially causing long-term instability and potential microbial growth.
Understanding the impacts of process and formulation changes on the dissolution profile during development is essential for guiding changes in the future.

Linking CQA's to Process Parameters
The general linear model of the process parameters and the yield determined statistically significant main effects. Particle size distribution was calculated based on individual sieve size retention and as a total yielded percentage. The amount of granulating water was found to be statistically significant for determining the particle size yield for both Propranolol HCl and metoprolol succinate. Identification of the water level is the critical process parameter because it will directly effect all downstream processing. During high shear granulation, controlling the process parameters to adjust the distribution of the water is important to facilitate extrusion and control the particle size distribution generated. Metoprolol succinate immediate release beads also found spheronization speed to be a significant factor for the determination of particle size distribution. Determination of an appropriate spheronization speed early in the process to create an optimal rope like formation and avoid irregular flow patterns of the beads must be carefully evaluated to avoid agglomeration and oversized bead generation.
Sphericity was measured as the mean aspect ratio and the standard deviation between the mean aspect ratio measurements. Kneading time and spheronization time were statistically significant for both mean sphericity and standard deviation of metoprolol succinate. Additionally, the granulation water amount was statistically significant for mean sphericity. Kneading time and granulating water can be directly attributed to the distribution and quantity of water available in the system to enhance the formulation's plasticity during processing. Longer durations of spheronization are associated with greater mean sphericity and lower sphericity variability due to increased particle interactions within the system.
The incorporation of HPMC into the aqueous ethylcellulose dispersion could not provide an adequate sustained release coating system for Propranolol HCl.
Humidified and other extreme curing conditions stressed the polymeric coating and a lack of a uniform coating ultimately resulted in an unstable product. In contrast, an organic coating solution did not exhibit the same instability when humidified and extreme curing conditions were studied. Initial stability studies not presented in this work, suggest that the organic coated beads were stable and exhibited little changes over time to the dissolution profile.
The Aqueous dispersion studied for Metoprolol Succinate sustained release coating did not contain HPMC and provided a reliable and adaptable system. Curing conditions did impact the product, but to lesser degree than the Propranolol HCl sustained release beads. The sustained release bead profile was initially matched to the brand profile to understand a baseline level needed for coating. The polymer coating levels were then raised two to three times the initial polymer level in order to compensate for potential bead crushing during tabletting. Incremental increases of polymer level yielded an anticipated decrease in the drug release profile.
Bead top-coating experiments identified a medium level of coating to add adequate plasticity and extra degree of protection, without adversely affecting the drug release profile. High levels of top coating can result in beads growing to an unacceptably large size, which may have a negative impact during compression. A practical concern for the generation of oversized beads for a tailored system is the diminished product yield, which has financial and processing ramifications.
Preliminary tabletting studies for metoprolol succinate identified the variable levels of protection provided by the different excipient granule blends. A mixed excipient blend of microcrystalline cellulose provided the greatest protection to the sustained release beads. Larger excipient particle sizes match the active beads and provide structural support, while smaller sized excipients fill void spaces created due to deformation during compaction. Tablet hardness ranges were established from 2kp to 5kp, at the low end of the acceptable range for passing tablet friability testing.

Supporting Elements of a Design Space
The metoprolol succinate immediate release beads utilized a 2 3 full factorial design to evaluate granulating water amount, kneading time, and spheronization time for the effects of these process conditions on the identified CQA's. These experiments yielded greater than 80% of the target particle size but found none of the factors or interactions statistically significant. The findings indicated that the factor levels had little effect on the final output and support an acceptable processing range.
The factors were also insignificant for the mean sphericity of the IR beads, but found spheronization time and all of the interactions significance in the relative and standard deviation of the sphericity measurements. This supported earlier work where longer spheronization times contributed to more uniformly round particles due to greater particle-particle and particle-system interactions. All of the trials met the moisture content specification of not more than 2.5%, and indicated that drying was within acceptable processing ranges. The IR bead with the lowest sphericity ratio was chosen for further development. A 2 3 full factorial experimental design was utilized to study the blend ratio of two microcrystalline cellulose particle sizes within the excipient blend, the amount of sustained release polymer coating applied to the active beads, and the hardness of the tablets. Friability studies indicated that 2kp hardness resulted in significant tablet breakage, while 5 kp hardness was acceptable. Visualization of bead damage during compression was performed with SEM, with significant damage to polymer coatings seen at 5kp hardness, and at the lower ratio of MCC 200: 102. Therefore the "best case" processing parameters for the tablets were further examined through dissolution testing. The dissolution profile was faster than the branded competitor but showed improvement over the previous experimental samples. Additionally, improvement was seen when beads retained on sieve #25 , the largest mean particle size for the active beads was removed. This improvement indicated the importance of particle size during tabletting, and would support movement in the target particle size distribution to a smaller size. Smaller sized beads fill void spaces more thoroughly and support plastic deformation during compression. The results from both the exploratory and DOE batches of immediate release beads indicate that a marked shift to a smaller particle size distribution would require a smaller extrusion screen and/or a micronized API with a smaller mean particle size distribution.

Exploratory Traditional Methods vs. QbD
The traditional methods promote local optimization, when an acceptable outcome is found ; minor adjustments are made to optimize that outcome. This approach can be cost and time effective if the formulation scientist is able to discover an acceptable result in a timely fashion. If a complex interaction occurs or a problem arises in a formulation, it can be difficult to identify and quantify the cause and solution to the problem. Utilizing an appropriate design can improve the overall understanding of the system and aid in finding a global optimum.
It is important to consider that statistical significance is not equivalent to practical significance. Prudent judgment should be employed when analyzing results of the statistical methods to appropriately interpret the results and their implications.
The P value is a useful tool for determining significance but should not be the only guide to interpretation. A blended approach to drug development will save time where elaborate study designs are not necessary and support processing understanding to solve problems and changes to the system.

Study Limitations
Due to financial and resource constraints, not all generated samples could be tested and in certain situations the theoretical best guess batch was characterized.
This limited the understanding of the design space edge of failure, as batches with negative traits were less likely to undergo full testing. Industrial pharmaceutical research often does not incorporate repetition into their statistical design due to the related costs associated. Therefore it is difficult to truly understand the product ranges and traits, especially in the event of a batch or analytical anomaly. The results found in the study cannot be generalized to cover all drugs, but may help to describe other drugs with similar physical characteristics.

Future Studies
Future work to establish point-to-point in vitro relationships and establish a level A bioequivalency rating should be pursued to support the commercialization of both of these products. Investigation of raw material physical characteristics and inter batch variability would be a useful tool to identify potential issues prior to beginning the process. Aspects oflong term stability, pharmacokinetics, and scale up offer good opportunities to examine the relationships found in this study to a real world application for FDA approval. Other drugs with similar characteristics should be studied to understand the applicability of the factors across drug product