A Wiley Book Series
Series Editors:
Dennis Douroumis, University of Greenwich, UK
Alfred Fahr, Friedrich–Schiller University of Jena, Germany
Jürgen Siepmann, University of Lille, France
Martin Snowden, University of Greenwich, UK
Vladimir Torchilin, Northeastern University, USA
Titles in the Series
Hot‐Melt Extrusion: Pharmaceutical Applications
Edited by Dionysios Douroumis
Drug Delivery Strategies for Poorly Water‐Soluble Drugs
Edited by Dionysios Douroumis and Alfred Fahr
Computational Pharmaceutics: Application of Molecular Modeling in Drug Delivery
Edited by Defang Ouyang and Sean C. Smith
Pulmonary Drug Delivery: Advances and Challenges
Edited by Ali Nokhodchi and Gary P. Martin
Novel Delivery Systems for Transdermal and Intradermal Drug Delivery
Edited by Ryan Donnelly and Raj Singh
Drug Delivery Systems for Tuberculosis Prevention and Treatment
Edited by Anthony J. Hickey
Continuous Manufacturing of Pharmaceuticals
Edited by Peter Kleinebudde, Johannes Khinast, and Jukka Rantanen
Pharmaceutical Quality by Design: A Practical Approach
Edited by Walkiria S. Schlindwein and Mark Gibson
Forthcoming Titles:
In Vitro Drug Release Testing of Special Dosage Forms
Edited by Nikoletta Fotaki and Sandra Klein
Characterization of Micro‐ and Nanosystems
Edited by Leena Peltonen
Therapeutic Dressings and Wound Healing Applications
Edited by Joshua Boateng
Process Analytics for Pharmaceuticals
Edited by Thomas de Beer, Jukka Rantanen and Clare Strachan
Edited by
WALKIRIA S. SCHLINDWEIN
De Montfort University
Leicester
United Kingdom
MARK GIBSON
A M PharmaServices Ltd
United Kingdom
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
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:
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 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. |
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. |
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. |