Cover
Title Page
List of Figures
List of Tables
List of Contributors
Advances in Pharmaceutical Technology: Series Preface
Preface
1 Introduction to Quality by Design (QbD)
1. 1.1 Introduction
2. 1.2 Background
3. 1.3 Science‐ and Risk‐Based Approaches
4. 1.4 ICH Q8–Q12
5. 1.5 QbD Terminology
6. 1.6 QbD Framework
7. 1.7 QbD Application and Benefits
8. 1.8 Regulatory Aspects
9. 1.9 Summary
10. 1.10 References
2 Quality Risk Management (QRM)
1. 2.1 Introduction
2. 2.2 Overview of ICH Q9
3. 2.3 Risk Management Tools
4. 2.4 Practical Examples of Use for QbD
5. 2.5 Concluding Remarks
6. 2.6 References
3 Quality Systems and Knowledge Management
1. 3.1 Introduction to Pharmaceutical Quality System
2. 3.2 The Regulatory Framework
3. 3.3 The Documentation Challenge
4. 3.4 From Data to Knowledge: An Example
5. 3.5 Data Integrity
6. 3.6 Quality Systems and Knowledge Management: Common Factors for Success
7. 3.7 Summary
8. 3.8 References
4 Quality by Design (QbD) and the Development and Manufacture of Drug Substance
1. 4.1 Introduction
2. 4.2 ICH Q11 and Drug Substance Quality
3. 4.3 Linear and Convergent Synthetic Chemistry Routes
4. 4.4 Registered Starting Materials (RSMs)
5. 4.5 Definition of an Appropriate Manufacturing Process
6. 4.6 In‐Line Process Analytical Technology and Crystallization Processes
7. 4.7 Applying the QbD Process
8. 4.8 Design of Experiments (DoE)
9. 4.9 Critical Process Parameters (CPPs)
10. 4.10 Design Space
11. 4.11 Control Strategy
12. 4.12 References
5 The Role of Excipients in Quality by Design (QbD)
1. 5.1 Introduction
2. 5.2 Quality of Design (QbD)
3. 5.3 Design of Experiments (DoE)
4. 5.4 Excipient Complexity
5. 5.5 Composition
6. 5.6 Drivers of Functionality or Performance
7. 5.7 Limited Utility of Pharmacopoeial Attributes
8. 5.8 Other Unspecified Attributes
9. 5.9 Variability
10. 5.10 Criticalities or Latent Conditions in the Finished Product
11. 5.11 Direct or Indirect Impact of Excipient Variability
12. 5.12 Control Strategy
13. 5.13 Communication with Suppliers
14. 5.14 Build in Compensatory Flexibility
15. 5.15 Risk Assessment
16. 5.16 Contingencies
17. 5.17 References
6 Development and Manufacture of Drug Product
1. 6.1 Introduction
2. 6.2 Applying QbD to Pharmaceutical Drug Product Development
3. 6.3 Product Design Intent and the Target Product Profile (TPP)
4. 6.4 The Quality Target Product Profile (QTPP)
5. 6.5 Identifying the Critical Quality Attributes (CQAs)
6. 6.6 Product Design and Identifying the Critical Material Attributes (CMAs)
7. 6.7 Process Design and Identifying the Critical Process Parameters (CPPs)
8. 6.8 Product and Process Optimisation
9. 6.9 Design Space
10. 6.10 Control Strategy
11. 6.11 Continuous Improvement
12. 6.12 Acknowledgements
13. 6.13 References
7 Design of Experiments
1. 7.1 Introduction
2. 7.2 Experimental Design in Action
3. 7.3 The Curse of Variation
4. 7.4 Fitting a Model
5. 7.5 Parameter Estimates
6. 7.6 Analysis of Variance
7. 7.7 ‘To Boldly Go’ – An Introduction to Managing Resource Constraints using DoE
8. 7.8 The Motivation for DoE
9. 7.9 Classical Designs
10. 7.10 Practical Workshop Design
11. 7.11 How Does This Work? The Underpinning of Statistical Models for Variation
12. 7.12 DoE and Cycles of Learning
13. 7.13 Sequential Classical Designs and Definitive Screening Designs
14. 7.14 Building a Simulation
15. 7.15 Conclusion
16. 7.16 Acknowledgements
17. 7.17 References
8 Multivariate Data Analysis (MVDA)
1. 8.1 Introduction
2. 8.2 Principal Component Analysis (PCA)
3. 8.3 PCA Case Study: Raw Material Characterization usingParticle Size Distribution Curves
4. 8.4 Partial Least Squares Projections to Latent Structures (PLS)
5. 8.5 PLS Case Study: A Process Optimization Model
6. 8.6 Orthogonal PLS (OPLS Multivariate Software)
7. 8.7 Orthogonal PLS (OPLS Multivariate Software) Case Study – Batch Evolution Modeling of a Chemical Batch Reaction
8. 8.8 Discussion
9. 8.9 References
9 Process Analytical Technology (PAT)
1. 9.1 Introduction
2. 9.2 How PAT Enables Quality by Design (QbD)
3. 9.3 The PAT Toolbox
4. 9.4 Process Sensors and Process Analysers
5. 9.5 Analyser Selection
6. 9.6 Regulatory Requirements Related to PAT Applications
7. 9.7 PAT Used in Development
8. 9.8 PAT Used in Manufacturing
9. 9.9 PAT and Real Time Release Testing (RTRT)
10. 9.10 PAT Implementation
11. 9.11 Data Management
12. 9.12 In‐Line Process Monitoring with UV‐Vis Spectroscopy: Case Study Example
13. 9.13 References
10 Analytical Method Design, Development, and Lifecycle Management
1. 10.1 Introduction
2. 10.2 Comparison of the Traditional Approach and the Enhanced QbD Approach
3. 10.3 Details of the Enhanced QbD Approach
4. 10.4 Defining Method Requirements
5. 10.5 Designing and Developing the Method
6. 10.6 Understanding the Impact of Method Parameters on Performance
7. 10.7 Defining the Method Control Strategy and Validating the Method
8. 10.8 Monitoring Routine Method Performance for ContinualImprovement
9. 10.9 Summary
10. 10.10 Example Case Studies
11. 10.11 References
11 Manufacturing and Process Controls
1. 11.1 Introduction to Manufacturing and Facilities
2. 11.2 Validation of Facilities and Equipment
3. 11.3 Drug Product Process Validation: A Lifecycle Approach
4. 11.4 The Impact of QbD on Process Equipment Design and Pharmaceutical Manufacturing Processes
5. 11.5 Introduction to Process Control in Pharmaceutical Manufacturing
6. 11.6 Advanced Process Controls (APC) and Control Strategy
7. 11.7 The Establishment of Continuous Manufacture
8. 11.8 The Tablet Press as Part of a Continuous Tableting Line
9. 11.9 Real‐Time Release Testing and Continuous Quality Verification
10. 11.10 Acknowledgments
11. 11.11 References
12 Regulatory Guidance
1. 12.1 Introduction
2. 12.2 The Common Technical Document (CTD) Format
3. 12.3 Essential Reading
4. 12.4 What Is Not Written, or Hidden, in the Guidance Documents?
5. 12.5 Post‐Approval Change
6. 12.6 Summary
7. 12.7 References
Index
End User License Agreement

List of Tables

Chapter 02
1. Table 2.1 Commonly used risk assessment (RA) tools (adapted from [9]).
2. Table 2.2 Example of risk assessment tool: check sheet (basic tool).
3. Table 2.3 Example of risk assessment tool: risk ranking (basic tool).
4. Table 2.4 Example of risk assessment tool: failure mode effect analysis (FMEA) (advanced tool). Note: red = dark grey; amber = light grey and green = medium grey
5. Table 2.5 Example of risk assessment tool: failure mode effect and criticality analysis (FMECA) (advanced tool).
6. Table 2.6 Example of risk assessment tool: hazards analysis and critical control points (HACCP) (advanced tool).
7. Table 2.7 Risk matrix for medicinal product.
8. Table 2.8 Risk score definitions.
9. Table 2.9 Detectability score.
10. Table 2.10 FMECA for medicinal product.
Chapter 04
1. Table 4.1 Impurity reduction for dirithromycin by re‐crystallization from various solvents [43].
2. Table 4.2 Potential critical process parameter.
3. Table 4.3 Scoring table.
4. Table 4.4 Scoring table for a crystallization process.
Chapter 06
1. Table 6.1 TPP for a solid oral dosage form.
2. Table 6.2 TPP for an injection dosage form.
3. Table 6.3 TPP for an inhalation dosage form.
4. Table 6.4 An example of a Pugh matrix for formulation technology design option evaluation. Note: red = dark grey; green = light grey and yellow = medium grey
5. Table 6.5 A typical QTPP for an oral solid dosage form product.
6. Table 6.6 A typical QTPP for a biopharmaceutical large molecule parenteral product.
7. Table 6.7 Typical example of a summary risk chart linking CQAs and prior knowledge to the unit operations for an oral solid dosage form.
8. Table 6.8 Example of ‘impact’ definition and scale from the CMC‐Vaccines Working Group, 2012 [19].
9. Table 6.9 Example of ‘uncertainty’ definition and scale (CMC‐Vaccines Working Group, 2012).
10. Table 6.10 The QTPP dictates the product design and development work.
11. Table 6.11 Criticality of parameter types (from Mitchell, 2013 [23]).
12. Table 6.12 Structured approach to build in vivo understanding for dissolution CQA.
13. Table 6.13 Developing appropriate dissolution CQA tests.
14. Table 6.14 Example of a control strategy for an oral solid dosage form.
Chapter 07
1. Table 7.1 Typical Microsoft Excel data:table (left) and reformatted for analysis (right).
2. Table 7.2 Data subsets (Study A and Study B).
3. Table 7.3 An alternative data collection plan.
4. Table 7.4 The relationship between factors and experiments.
5. Table 7.5 The tableting results data table.
6. Table 7.6 The first experiment in the sequential approach – a 10 run fractional factorial Resolution III design.
7. Table 7.7 Second experiment in the sequential approach – a 26 run central composite design.
8. Table 7.8 The experiment in the definitive screening approach – a 13 run definitive screening design (DSD).
Chapter 09
1. Table 9.1 Elements of a PAT system.
2. Table 9.2 Typical examples of attribute measurements by process analysers within common unit operations.
3. Table 9.3 Typical attribute measurements by process sensors.
4. Table 9.4 Typical attribute measurements by process analysers.
Chapter 10
1. Table 10.1 The role of analytical testing in pharmaceutical development across the entire control strategy.
2. Table 10.2 Summary analytical control strategy and risk assessment for a drug substance manufacturing process. Note: red = dark grey; green = light grey and amber = medium grey
3. Table 10.3 Generic design parameters for chromatographic methods.
4. Table 10.4 Generic design parameters for chromatographic methods.
5. Table 10.5 Elements of an established analytical control strategy.
6. Table 10.6 System suitability criteria.
Chapter 11
1. Table 11.1 A typical ranking system for severity, probability, and detectability.
2. Table 11.2 Example of a process impact assessment for the manufacture of a tablet product.Attribute: Tablets will not be contaminated or compromised in any way by the manufacturing activity.
3. Table 11.3 Key differences between the ISPE Baseline® Guide and ASTM E2500.
4. Table 11.4 Summary comparison of the EU and US process validation guidance.
5. Table 11.5 Typical measurements of common unit operations.
6. Table 11.6 Control strategy options for a simple IR tablet.
7. Table 11.7 Comparing batch and continuous processes.
8. Table 11.8 Unit operations currently used in oral solid dose manufacture.
Chapter 12
1. Table 12.1 Summary of where to present the control strategy in the CTD.

List of Illustrations

Chapter 01
1. Figure 1.1 A framework for QbD.
Chapter 02
1. Figure 2.1 Medicinal product development flowchart.
2. Figure 2.2 Overview of a typical quality risk management process (ICH Q9).
3. Figure 2.3 Example of risk assessment tool: flowchart (basic tool).
4. Figure 2.4 Example of a cause and effect diagram (Ishikawa fishbone diagram).
5. Figure 2.5 Overall QRA process flow for the medicinal product.
6. Figure 2.6 Risk score approach (high level) based on current operating space.
7. Figure 2.7 Risk score approach (detailed level) based on current operating space.
8. Figure 2.8 Tacit benefits of QRM.
Chapter 03
1. Figure 3.1 Through the ages.
2. Figure 3.2 ICH QbD timeline.
3. Figure 3.3 From data to knowledge.
Chapter 04
1. Figure 4.1 Example of a convergent synthesis [19].
2. Figure 4.2 Temperature–solubility curve for AZD3342 in 1‐propanol:water and IMS:water.
3. Figure 4.3 Solubility–temperature and metastable zone limit curves.
4. Figure 4.4 Effect of cooling rate on the metastable zone width of sibenadet HCl.
5. Figure 4.5 Crystal16 data for AZD3342 polymorphs A and G.
6. Figure 4.6 Cooling crystallization scenarios: (a) natural cooling, (b) linear cooling, (c) controlled cooling.
7. Figure 4.7 Effect of seed loading on particle size.
8. Figure 4.8 UV Data for an unseeded linear cooling crystallization of AZD3342.
9. Figure 4.9 FBRM Lasentec data and Photomicrograph for an unseeded, linear cool crystallization of AZD3342.
10. Figure 4.10 UV Data for the dissolution and crystallization of AZD3342 using seeding and a cubic cooling profile (controlled cooling).
11. Figure 4.11 FBRM Lasentec data and optical microscopy of AZD3342 crystals using seeding and a cubic cooling profile (controlled cooling).
12. Figure 4.12 Typical Ishikawa diagram for an anti‐solvent crystallization process.
Chapter 05
1. Figure 5.1 Examples of “known and unknowns” to users and suppliers.
Chapter 06
1. Figure 6.1 Process of determining how a quality attribute is deemed critical.
2. Figure 6.2 Examples of where CQAs may be impacted in the manufacturing process for an oral solid dosage form.
3. Figure 6.3 PDA decision tree for designating parameter criticality.
4. Figure 6.4 Fishbone diagram recognising material attributes (MAs) and process parameters (PPs) contributing to the quality attribute, ‘solubility’, of a solid dispersion API/polymer product.
5. Figure 6.5 Use of risk assessment to determine what DoE studies to use.
6. Figure 6.6 Example of a process parameter criticality assessment decision tree.
7. Figure 6.7 The manufacturing process flow for ACE tablets.
8. Figure 6.8 Critical parameters affecting blend uniformity for ACE tablets.
9. Figure 6.9 Schematic representation of the relationship between knowledge, design, and control space.
10. Figure 6.10 Example of terminal sterilisation design space.
11. Figure 6.11 Comparison of traditional controls with advanced controls for the real‐time release of an oral solid dosage form.
Chapter 07
1. Figure 7.1 Constructing a variability gauge chart.
2. Figure 7.2 Impact of formulation of disintegration time with outlier.
3. Figure 7.3 Impact of formulation on disintegration with the corrected data point.
4. Figure 7.4 Prediction profilers showing the impact of formulation on disintegration time.
5. Figure 7.5 Statistical analysis of the disintegration time data.
6. Figure 7.6 Comparison of the statistical analysis of Study A and Study B.
7. Figure 7.7 Identifying sources of variation.
8. Figure 7.8 The GlaxoSmithKline reaction simulator.
9. Figure 7.9 ‘Understanding and Optimizing Chemical Processes’ 2015 Workshop results.
10. Figure 7.10 Costs, risks and benefits of classical designs.
11. Figure 7.11 Data visualization.
12. Figure 7.12 Analysis of the half fraction.
13. Figure 7.13 Prediction profiles for the four different designs (shaded boxes indicate statistically significant effects).
14. Figure 7.14 Interaction effects.
15. Figure 7.15 Explaining the experimental procedure to the ‘novice’ workshop delegate.
16. Figure 7.16 Comparison of ‘expert’ and ‘novice’ findings for the half fraction design.
17. Figure 7.17 Theoretical model of a specific real‐world situation.
18. Figure 7.18 Cycles of learning iterate between the real world and our model of it.
19. Figure 7.19 Resource requirements for central composite design.
20. Figure 7.20 Simulation scenario schematic.
21. Figure 7.21 Analysis of the 10 run fractional factorial experiment.
22. Figure 7.22 Prediction profiler for the 10 run fractional factorial.
23. Figure 7.23 Analysis of the central composite design.
24. Figure 7.24 Analysis of the 13 run DSD experiment.
25. Figure 7.25 Comparison of the central composite and definitive screening design (DSD).
Chapter 08
1. Figure 8.1 Notation used in principal components analysis (PCA).
2. Figure 8.2 PCA derives a model that fits the data as well as possible in the least squares sense. Alternatively, PCA may be understood as maximizing the variance of the projection coordinates.
3. Figure 8.3 The raw data curves of the 45 batches of the training set.
4. Figure 8.4 Scatter plot of the two principal components. Each point represents one batch of raw material. The plot is color‐coded according to supplier.
5. Figure 8.5 Loading line plots of the two principal components.
6. Figure 8.6 Scores and DModX plots for the particle size distribution data set. Top left: Predicted scores for the test set batches. Top right: Scores for the training set batches. Bottom left: Predicted DModX for the test set batches. Bottom right: DModX for the training set batches.
7. Figure 8.7 Line plot of the power spectral density (PSD) curves of the six early (red color) and three late (black color) batches of supplier L3.
8. Figure 8.8 Spanning batches can be selected using DoE. Such spanning batches can be subjected to a thorough investigation in order to ensure production robustness in all part of the PCA score space.
9. Figure 8.9 (Left) Scatter plot of HS_1 against HS_2, (right) PLS t₁/t₂ score plot (note the resemblance to Figure 8.8).
10. Figure 8.10 (Left) Summary of fit plot of the PLS model of the SOVRING subset with complete Y‐data. Five components were significant according to cross‐validation. (Right) Individual R²Y‐ and Q²Y‐values of the six responses. The most important responses are PAR, FAR, %Fe_FAR and %P_FAR.
11. Figure 8.11 (Left) PLS w*c₁/w*c₂ loading weight plot. Ton_In is the most influential factor for PAR and FAR. The model indicates that by increasing Ton_In, the amounts of PAR and FAR increase. There is a significant quadratic influence of HS_2 on the quality responses %Fe_FAR and %P_FAR. This nonlinear dependence is better interpreted in a response contour plot, or a response surface plot. (Right) PLS w*c₃/w*c₄ weight plot.
12. Figure 8.12 (Left) Response contour plot of PAR, where the influences of Ton_In and HS_2 are seen. (Right) Response contour plot of FAR.
13. Figure 8.13 (Left) Response contour plot of %P_FAR. (Right) Response contour plot of %Fe_FAR.
14. Figure 8.14 SweetSpot plot of the SOVRING example, suggesting the SweetSpot should be located in the upper and right‐hand part.
15. Figure 8.15 A design space plot of the SOVRING example. The green area corresponds to design space with a low risk, 1%, of failure.
16. Figure 8.16 A schematic representation of how the concepts knowledge space, design space, and normal operation (control space) region relate to one another.
17. Figure 8.17 The design space hypercube represents the largest regular geometrical structure that can be inserted into the irregular design space.
18. Figure 8.18 Batch control chart showing predictions for two bad batches. The BEM fitted using orthogonal PLS readily detects the deviating batches.
19. Figure 8.19 Batch control chart showing predictions for two bad batches. The BEM was fitted using classical PLS.
20. Figure 8.20 Contribution plot for batch DoE7 at 1.7 min showing which variables are contributing to the process deviation. The process upset is caused by a few of the process parameters. For example, the contribution plot indicates that the reaction temperature (highlighted by red color) is much higher than for a normal batch.
21. Figure 8.21 Line plot of the reaction temperature for the non‐NOC batch DoE7. At the time point of 1.7 min, the reaction temperature is almost 15° higher than the average temperature across the six NOC batches. The deviation from normality of the reaction temperature for DoE7 increases further into the lifetime of the batch.
Chapter 09
1. Figure 9.1 Elements of a PAT system (each element is further described in Table 9.1).
2. Figure 9.2 Examples of PAT applications used for an oral solid dosage process and benefits.
3. Figure 9.3 PAT applications used during processing in lieu of conventional end‐product QC testing.
4. Figure 9.4 Elements to consider for inclusion in a PAT application, whether it is for gaining process understanding or for process control.
5. Figure 9.5 Manufacturing process based on hot‐melt extrusion showing the main unit operations from powders to final product, tablets.
6. Figure 9.6 Preliminary risk assessment (RA) analysis (Ishikawa diagram). The factors highlighted were considered critical for the extrusion process.
7. Figure 9.7 Representation of about 5000 UV‐Vis spectra from DoEs 1–3 and verification experiments.
8. Figure 9.8 Raw spectra from a DoE set of runs.
9. Figure 9.9 UV‐Vis spectra from the first screening design (DoE1) showing a sample of the spectra collected.
Chapter 10
1. Figure 10.1 Enhanced QbD lifecycle approach to analytical methods.
2. Figure 10.2 Example of the impact of method variability on overall variability for drug product assay (LSL=95%LC and USL=105%LC).
3. Figure 10.3 Science‐ and risk‐based approach to analytical method design, development and lifecycle management.
4. Figure 10.4 Ishikawa (fishbone) diagram for a drug product chromatographic assay method.
5. Figure 10.5 Gage repeatability and reproducibility study. (Illustration based on Minitab® Help printed with permission of Minitab Inc. All such material remains the exclusive property and copyright of Minitab Inc. All rights reserved.)
6. Figure 10.6 I‐MR chart of drug product assay in manufacturing order – produced using Minitab® Statistical Software.
7. Figure 10.7 Science and risk assessment process.
8. Figure 10.8 Centred and scaled coefficients for the resolution and separation factor models.
9. Figure 10.9 Alternative statistical approach – main effect plot amount of TFA in the mobile phase.
10. Figure 10.10 Centred and scaled coefficients for the signal‐to‐noise model.
11. Figure 10.11 Dissolution data from multivariate experimental study (coloured by analyst) – individual tablet (open circles) and mean data (solid circles) – produced using Minitab® Statistical Software.
12. Figure 10.12 Main effects plot from multivariate experimental study – produced using Minitab® Statistical Software.
Chapter 11
1. Figure 11.1 The “V” model concept of validation.
2. Figure 11.2 The ASTM E2500 system lifecycle and validation approach.
3. Figure 11.3 Sequence of activities for formulation, process design, and optimization incorporating process validation activities according to the lifecycle approach.
4. Figure 11.4 Stages of process validation showing potential changes.
5. Figure 11.5 The elements of a simple, single‐loop control system for a liquid process. LT =sensor; LC = controller; LV = actuator.
6. Figure 11.6 Control strategy and design space for an IR tablet.
7. Figure 11.7 Typical batch process flow for a granulated tablet.
8. Figure 11.8 ConsiGma™ continuous tablet production line.
Chapter 12
1. Figure 12.1 Summary of the intrinsic value of applying QbD.
2. Figure 12.2 The CTD triangle [4].
3. Figure 12.3 A typical example of an Ishikawa or fishbone diagram. Key: red = potential high impact on CQA; Yellow = potential impact on CQA; green = unlikely to have significant impact on CQA. Example: roller compaction.
4. Figure 12.4 Types of post‐approval changes available in the EU relative to the adopted level of risk during the evaluation of the procedure.

This edition first published 2018
© 2018 John Wiley & Sons Ltd

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

The right of Walkiria S. Schlindwein and Mark Gibson to be identified as the authors of the editorial material in this work has been asserted in accordance with law.

Registered Offices
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial Office
111 River Street, Hoboken, NJ 07030, USA
350 Main Street, Malden, MA 02148‐5020, USA
9600 Garsington Road, Oxford, OX4 2DQ, UK
The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
Boschstr. 12, 69469 Weinheim, Germany
1 Fusionopolis Walk, #07‐01 Solaris South Tower, Singapore 138628

For details of our global editorial offices, customer services, and more information about Wiley products, visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of Warranty
In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of experimental reagents, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each chemical, piece of equipment, reagent, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging‐in‐Publication Data

Names: Schlindwein, Walkiria S., 1961– editor. | Gibson, Mark, 1957– editor.
Title: Pharmaceutical quality by design : a practical approach / edited by Dr. Walkiria S. Schlindwein, Mark Gibson.
Description: First edition. | Hoboken, NJ : John Wiley & Sons, 2018. | Series: Advances in pharmaceutical technology | Includes bibliographical references and index. |
Identifiers: LCCN 2017030338 (print) | LCCN 2017043153 (ebook) | ISBN 9781118895221 (pdf) | ISBN 9781118895214 (epub) | ISBN 9781118895207 (cloth)
Subjects: LCSH: Drugs–Design. | Drugs–Quality control.
Classification: LCC RS420 (ebook) | LCC RS420 .P47 2018 (print) | DDC 615.1/9–dc23
LC record available at https://lccn.loc.gov/2017030338

Cover design by Wiley
Cover image: (Background) © ShutterWorx/Gettyimages; (Graph) Courtesy of Walkiria S. Schlindwein and Mark Gibson

Advances in Pharmaceutical Technology: Series Preface

The series Advances in Pharmaceutical Technology covers the principles, methods and technologies that the pharmaceutical industry uses to turn a candidate molecule or new chemical entity into a final drug form and hence a new medicine. The series will explore means of optimizing the therapeutic performance of a drug molecule by designing and manufacturing the best and most innovative of new formulations. The processes associated with the testing of new drugs, the key steps involved in the clinical trials process and the most recent approaches utilized in the manufacture of new medicinal products will all be reported. The focus of the series will very much be on new and emerging technologies and the latest methods used in the drug development process.

The topics covered by the series include the following:

Formulation: The manufacture of tablets in all forms (caplets, dispersible, fast‐melting) will be described, as will capsules, suppositories, solutions, suspensions and emulsions, aerosols and sprays, injections, powders, ointments and creams, sustained release and the latest transdermal products. The developments in engineering associated with fluid, powder and solids handling, solubility enhancement, colloidal systems including the stability of emulsions and suspensions will also be reported within the series. The influence of formulation design on the bioavailability of a drug will be discussed, and the importance of formulation with respect to the development of an optimal final new medicinal product will be clearly illustrated.
Drug Delivery: The use of various excipients and their role in drug delivery will be reviewed. Among the topics to be reported and discussed will be a critical appraisal of the current range of modified‐release dosage forms currently in use and also those under development. The design and mechanism(s) of controlled release systems including macromolecular drug delivery, microparticulate controlled drug delivery, the delivery of biopharmaceuticals, delivery vehicles created for gastrointestinal tract targeted delivery, transdermal delivery and systems designed specifically for drug delivery to the lung will all be reviewed and critically appraised. Further site‐specific systems used for the delivery of drugs across the blood–brain barrier including dendrimers, hydrogels and new innovative biomaterials will be reported.
Manufacturing: The key elements of the manufacturing steps involved in the production of new medicines will be explored in this series. The importance of crystallization; batch and continuous processing, seeding; and mixing including a description of the key engineering principles relevant to the manufacture of new medicines will all be reviewed and reported. The fundamental processes of quality control including good laboratory practice, good manufacturing practice, Quality by Design, the Deming Cycle, regulatory requirements and the design of appropriate robust statistical sampling procedures for the control of raw materials will all be an integral part of this book series.
An evaluation of the current analytical methods used to determine drug stability, the quantitative identification of impurities, contaminants and adulterants in pharmaceutical materials will be described as will the production of therapeutic bio‐macromolecules, bacteria, viruses, yeasts, moulds, prions and toxins through chemical synthesis and emerging synthetic/molecular biology techniques. The importance of packaging including the compatibility of materials in contact with drug products and their barrier properties will also be explored.
Advances in Pharmaceutical Technology is intended as a comprehensive one‐stop shop for those interested in the development and manufacture of new medicines. The series will appeal to those working in the pharmaceutical and related industries, both large and small, and will also be valuable to those who are studying and learning about the drug development process and the translation of those drugs into new life‐saving and life‐enriching medicines.

Dennis Douroumis

Alfred Fahr

Jürgen Siepmann

Martin Snowden

Vladimir Torchilin

List of Figures

Figure 1.1	A framework for QbD.
Figure 2.1	Medicinal product development flowchart.
Figure 2.2	Overview of a typical quality risk management process (ICH Q9).
Figure 2.3	Example of risk assessment tool: flowchart (basic tool).
Figure 2.4	Example of a cause and effect diagram (Ishikawa fishbone diagram).
Figure 2.5	Overall QRA process flow for the medicinal product.
Figure 2.6	Risk score approach (high level) based on current operating space.
Figure 2.7	Risk score approach (detailed level) based on current operating space.
Figure 2.8	Tacit benefits of QRM.
Figure 3.1	Through the ages.
Figure 3.2	ICH QbD timeline.
Figure 3.3	From data to knowledge.
Figure 4.1	Example of a convergent synthesis.
Figure 4.2	Temperature–solubility curve for AZD3342 in 1‐propanol:water and IMS:water.
Figure 4.3	Solubility–temperature and metastable zone limit curves.
Figure 4.4	Effect of cooling rate on the metastable zone width of sibenadet HCl.
Figure 4.5	Crystal16 data for AZD3342 polymorphs A and G.
Figure 4.6	Cooling crystallization scenarios: (a) natural cooling, (b) linear cooling, (c) controlled cooling.
Figure 4.7	Effect of seed loading on particle size.
Figure 4.8	UV Data for an unseeded linear cooling crystallization of AZD3342.
Figure 4.9	FBRM Lasentec data and Photomicrograph for an unseeded, linear cool crystallization of AZD3342.
Figure 4.10	UV Data for the dissolution and crystallization of AZD3342 using seeding and a cubic cooling profile (controlled cooling).
Figure 4.11	FBRM Lasentec data and optical microscopy of AZD3342 crystals using seeding and a cubic cooling profile (controlled cooling).
Figure 4.12	Typical Ishikawa diagram for an anti‐solvent crystallization process.
Figure 5.1	Examples of “known and unknowns” to users and suppliers.
Figure 6.1	Process of determining how a quality attribute is deemed critical. *A severity scale is used to assess relative magnitude.
Figure 6.2	Examples of where CQAs may be impacted in the manufacturing process for an oral solid dosage form.
Figure 6.3	PDA decision tree for designating parameter criticality.
Figure 6.4	Fishbone diagram recognising material attributes (MAs) and process parameters (PPs) contributing to the quality attribute, ‘solubility’, of a solid dispersion API/polymer product.
Figure 6.5	Use of risk assessment to determine what DoE studies to use.
Figure 6.6	Example of a process parameter criticality assessment decision tree.
Figure 6.7	The manufacturing process flow for ACE tablets.
Figure 6.8	Critical parameters affecting blend uniformity for ACE tablets.
Figure 6.9	Schematic representation of the relationship between knowledge, design, and control space.
Figure 6.10	Example of terminal sterilisation design space.
Figure 6.11	Comparison of traditional controls with advanced controls for the real‐time release of an oral solid dosage form.
Figure 7.1	Constructing a variability gauge chart.
Figure 7.2	Impact of formulation of disintegration time with outlier.
Figure 7.3	Impact of formulation on disintegration with the corrected data point.
Figure 7.4	Prediction profilers showing the impact of formulation on disintegration time.
Figure 7.5	Statistical analysis of the disintegration time data.
Figure 7.6	Comparison of the statistical analysis of Study A and Study B.
Figure 7.7	Identifying sources of variation.
Figure 7.8	The GlaxoSmithKline reaction simulator.
Figure 7.9	‘Understanding and Optimizing Chemical Processes’ 2015 Workshop results.
Figure 7.10	Costs, risks and benefits of classical designs.
Figure 7.11	Data visualization.
Figure 7.12	Analysis of the half fraction.
Figure 7.13	Prediction profiles for the four different designs (shaded boxes indicate statistically significant effects).
Figure 7.14	Interaction effects.
Figure 7.15	Explaining the experimental procedure to the ‘novice’ workshop delegate.
Figure 7.16	Comparison of ‘expert’ and ‘novice’ findings for the half fraction design.
Figure 7.17	Theoretical model of a specific real‐world situation.
Figure 7.18	Cycles of learning iterate between the real world and our model of it.
Figure 7.19	Resource requirements for central composite design.
Figure 7.20	Simulation scenario schematic.
Figure 7.21	Analysis of the 10 run fractional factorial experiment.
Figure 7.22	Prediction profiler for the 10 run fractional factorial.
Figure 7.23	Analysis of the central composite design.
Figure 7.24	Analysis of the 13 run DSD experiment.
Figure 7.25	Comparison of the central composite and definitive screening design (DSD).
Figure 8.1	Notation used in principal components analysis (PCA).
Figure 8.2	PCA derives a model that fits the data as well as possible in the least squares sense. Alternatively, PCA may be understood as maximizing the variance of the projection coordinates.
Figure 8.3	The raw data curves of the 45 batches of the training set.
Figure 8.4	Scatter plot of the two principal components. Each point represents one batch of raw material. The plot is color‐coded according to supplier.
Figure 8.5	Loading line plots of the two principal components.
Figure 8.6	Scores and DModX plots for the particle size distribution data set. Top left: Predicted scores for the test set batches. Top right: Scores for the training set batches. Bottom left: Predicted DModX for the test set batches. Bottom right: DModX for the training set batches.
Figure 8.7	Line plot of the power spectral density (PSD) curves of the six early (red color) and three late (black color) batches of supplier L3.
Figure 8.8	Spanning batches can be selected using DoE. Such spanning batches can be subjected to a thorough investigation in order to ensure production robustness in all part of the PCA score space.
Figure 8.9	(Left) Scatter plot of HS_1 against HS_2, (right) PLS t1/t2 score plot (note the resemblance to Figure 8.8).
Figure 8.10	(Left) Summary of fit plot of the PLS model of the SOVRING subset with complete Y‐data. Five components were significant according to cross‐validation. (Right) Individual R2Y‐ and Q2Y‐values of the six responses. The most important responses are PAR, FAR, %Fe_FAR and %P_FAR.
Figure 8.11	(Left) PLS wc1/wc2 loading weight plot. Ton_In is the most influential factor for PAR and FAR. The model indicates that by increasing Ton_In, the amounts of PAR and FAR increase. There is a significant quadratic influence of HS_2 on the quality responses %Fe_FAR and %P_FAR. This nonlinear dependence is better interpreted in a response contour plot, or a response surface plot. (Right) PLS wc3/wc4 weight plot.
Figure 8.12	(Left) Response contour plot of PAR, where the influences of Ton_In and HS_2 are seen. (Right) Response contour plot of FAR.
Figure 8.13	(Left) Response contour plot of %P_FAR. (Right) Response contour plot of %Fe_FAR.
Figure 8.14	SweetSpot plot of the SOVRING example, suggesting the SweetSpot should be located in the upper and right‐hand part.
Figure 8.15	A design space plot of the SOVRING example. The green area corresponds to design space with a low risk, 1%, of failure.
Figure 8.16	A schematic representation of how the concepts knowledge space, design space, and normal operation (control space) region relate to one another.
Figure 8.17	The design space hypercube represents the largest regular geometrical structure that can be inserted into the irregular design space.
Figure 8.18	Batch control chart showing predictions for two bad batches. The BEM fitted using orthogonal PLS readily detects the deviating batches.
Figure 8.19	Batch control chart showing predictions for two bad batches. The BEM was fitted using classical PLS.
Figure 8.20	Contribution plot for batch DoE7 at 1.7 min showing which variables are contributing to the process deviation. The process upset is caused by a few of the process parameters. For example, the contribution plot indicates that the reaction temperature (highlighted by red color) is much higher than for a normal batch.
Figure 8.21	Line plot of the reaction temperature for the non‐NOC batch DoE7. At the time point of 1.7 min, the reaction temperature is almost 15° higher than the average temperature across the six NOC batches. The deviation from normality of the reaction temperature for DoE7 increases further into the lifetime of the batch.
Figure 9.1	Elements of a PAT system (each element is further described in Table 9.1).
Figure 9.2	Examples of PAT applications used for an oral solid dosage process and benefits.
Figure 9.3	PAT applications used during processing in lieu of conventional end‐product QC testing.
Figure 9.4	Elements to consider for inclusion in a PAT application, whether it is for gaining process understanding or for process control.
Figure 9.5	Manufacturing process based on hot‐melt extrusion showing the main unit operations from powders to final product, tablets.
Figure 9.6	Preliminary risk assessment (RA) analysis (Ishikawa diagram). The factors highlighted were considered critical for the extrusion process.
Figure 9.7	Representation of about 5000 UV‐Vis spectra from DoEs 1–3 and verification experiments.
Figure 9.8	Raw spectra from a DoE set of runs.
Figure 9.9	UV‐Vis spectra from the first screening design (DoE1) showing a sample of the spectra collected.
Figure 10.1	Enhanced QbD lifecycle approach to analytical methods.
Figure 10.2	Example of the impact of method variability on overall variability for drug product assay (LSL=95%LC and USL=105%LC).
Figure 10.3	Science‐ and risk‐based approach to analytical method design, development and lifecycle management.
Figure 10.4	Ishikawa (fishbone) diagram for a drug product chromatographic assay method.
Figure 10.5	Gage repeatability and reproducibility study.
Figure 10.6	I‐MR chart of drug product assay in manufacturing order – produced using Minitab^® Statistical Software.
Figure 10.7	Science and risk assessment process.
Figure 10.8	Centred and scaled coefficients for the resolution and separation factor models.
Figure 10.9	Alternative statistical approach – main effect plot amount of TFA in the mobile phase.
Figure 10.10	Centred and scaled coefficients for the signal‐to‐noise model.
Figure 10.11	Dissolution data from multivariate experimental study (coloured by analyst) – individual tablet (open circles) and mean data (solid circles) – produced using Minitab^® Statistical Software.
Figure 10.12	Main effects plot from multivariate experimental study – produced using Minitab^® Statistical Software.
Figure 11.1	The “V” model concept of validation.
Figure 11.2	The ASTM E2500 system lifecycle and validation approach.
Figure 11.3	Sequence of activities for formulation, process design, and optimization incorporating process validation activities according to the lifecycle approach.
Figure 11.4	Stages of process validation showing potential changes.
Figure 11.5	The elements of a simple, single‐loop control system for a liquid process. LT = sensor; LC = controller; LV = actuator.
Figure 11.6	Control strategy and design space for an IR tablet.
Figure 11.7	Typical batch process flow for a granulated tablet.
Figure 11.8	ConsiGma™ continuous tablet production line.
Figure 12.1	Summary of the intrinsic value of applying QbD.
Figure 12.2	The CTD triangle.
Figure 12.3	A typical example of an Ishikawa or fishbone diagram. Key: red = potential high impact on CQA; Yellow = potential impact on CQA; green = unlikely to have significant impact on CQA. Example: roller compaction.
Figure 12.4	Types of post‐approval changes available in the EU relative to the adopted level of risk during the evaluation of the procedure.

ADVANCES IN PHARMACEUTICAL TECHNOLOGY

Pharmaceutical Quality by Design

A Practical Approach

Advances in Pharmaceutical Technology: Series Preface

List of Figures

List of Tables

Table 2.1	Commonly used risk assessment (RA) tools.
Table 2.2	Example of risk assessment tool: check sheet (basic tool).
Table 2.3	Example of risk assessment tool: risk ranking (basic tool).
Table 2.4	Example of risk assessment tool: failure mode effect analysis (FMEA) (advanced tool). Note: red = dark grey; amber = light grey and green = medium grey
Table 2.5	Example of risk assessment tool: failure mode effect and criticality analysis (FMECA) (advanced tool).
Table 2.6	Example of risk assessment tool: hazards analysis and critical control points (HACCP) (advanced tool).
Table 2.7	Risk matrix for medicinal product.
Table 2.8	Risk score definitions.
Table 2.9	Detectability score.
Table 2.10	FMECA for medicinal product.
Table 4.1	Impurity reduction for dirithromycin by re‐crystallization from various solvents.
Table 4.2	Potential critical process parameter.
Table 4.3	Scoring table.
Table 4.4	Scoring table for a crystallization process.
Table 6.1	TPP for a solid oral dosage form.
Table 6.2	TPP for an injection dosage form.
Table 6.3	TPP for an inhalation dosage form.
Table 6.4	An example of a Pugh matrix for formulation technology design option evaluation. Note: red = dark grey; green = light grey and yellow = medium grey
Table 6.5	A typical QTPP for an oral solid dosage form product.
Table 6.6	A typical QTPP for a biopharmaceutical large molecule parenteral product.
Table 6.7	Typical example of a summary risk chart linking CQAs and prior knowledge to the unit operations for an oral solid dosage form.
Table 6.8	Example of ‘impact’ definition and scale from the CMC‐Vaccines Working Group, 2012.
Table 6.9	Example of ‘uncertainty’ definition and scale (CMC‐Vaccines Working Group, 2012).
Table 6.10	The QTPP dictates the product design and development work.
Table 6.11	Criticality of parameter types.
Table 6.12	Structured approach to build in vivo understanding for dissolution CQA.
Table 6.13	Developing appropriate dissolution CQA tests.
Table 6.14	Example of a control strategy for an oral solid dosage form.
Table 7.1	Typical Microsoft Excel data:table (left) and reformatted for analysis (right).
Table 7.2	Data subsets (Study A and Study B).
Table 7.3	An alternative data collection plan.
Table 7.4	The relationship between factors and experiments.
Table 7.5	The tableting results data table.
Table 7.6	The first experiment in the sequential approach – a 10 run fractional factorial Resolution III design.
Table 7.7	Second experiment in the sequential approach – a 26 run central composite design.
Table 7.8	The experiment in the definitive screening approach – a 13 run definitive screening design (DSD)
Table 9.1	Elements of a PAT system.
Table 9.2	Typical examples of attribute measurements by process analysers within common unit operations.
Table 9.3	Typical attribute measurements by process sensors.
Table 9.4	Typical attribute measurements by process analysers.
Table 10.1	The role of analytical testing in pharmaceutical development across the entire control strategy.
Table 10.2	Summary analytical control strategy and risk assessment for a drug substance manufacturing process. Note: red = dark grey; green = light grey and amber = medium grey
Table 10.3	Generic design parameters for chromatographic methods.
Table 10.4	Generic design parameters for chromatographic methods.
Table 10.5	Elements of an established analytical control strategy.
Table 10.6	System suitability criteria.
Table 11.1	A typical ranking system for severity, probability, and detectability.
Table 11.2	Example of a process impact assessment for the manufacture of a tablet product.
Table 11.3	Key differences between the ISPE Baseline^® Guide and ASTM E2500.
Table 11.4	Summary comparison of the EU and US process validation guidance.
Table 11.5	Typical measurements of common unit operations.
Table 11.6	Control strategy options for a simple IR tablet.
Table 11.7	Comparing batch and continuous processes.
Table 11.8	Unit operations currently used in oral solid dose manufacture.
Table 12.1	Summary of where to present the control strategy in the CTD.