Preface

Nonlinear mixed effects modeling is an analytical approach capable of the efficient use of data to support informed decision making. In drug development and in clinical practice, the culmination of the work of the pharmacometrician is making informed decisions regarding the treatment of patients, either during the development of new treatments or in the clinical care of specific patients. The application of nonlinear mixed effects models to sparse clinical data was a concept brilliantly conceived and implemented via the NONMEM system by Dr. Lewis B. Sheiner and Dr. Stuart L. Beal, supported over the years by Alison J. Boeckmann. The impetus for development of the analytical approach and supporting software was the need to study drugs in the population in which they were to be used, and realization of the direct correlation between the patients who are most vulnerable to pharmacokinetic differences and those who are most difficult to study (Sheiner and Benet 1985). Of course, there have been many additional contributors to the development of this field over the past few decades; their contributions have created an environment where pharmacometric analysis approaches now flourish and contribute significantly to informed decision making.

The realization of the value of the field of pharmacometrics has caused significant demand for new pharmacometricians (Gobburu 2010). In its Pharmacometrics 2020 Strategic Goals statement, the FDA set a goal to train 20 pharmacometricians by the year 2020 (FDA 2013). In Pharmacometrics 2020, Dr. Gobburu suggests a goal that an additional 500 should be trained in that same time through the support of academia and the pharmaceutical industry (Gobburu 2010). This book is intended to contribute to that effort. It is written at an introductory level to bring clarity to the basic concepts of nonlinear mixed effects modeling for those new to the field. The book is focused on using the NONMEM® software for pharmacometric analysis. This is not to say that other software applications are not useful, but NONMEM was selected in part, because the authors are most familiar with this tool, but also because it is among the most flexible and powerful tools in use today, and because in our estimation, it remains the most often used software for nonlinear mixed effects modeling. There are also clear signs of its continued development and improvement by Dr. Robert J. Bauer and ICON Development Solutions, as evidenced by the release of new versions adding important new estimation algorithms, modernization of the code to improve efficiency, and enhancements to the program output. This book is also written with a particular focus toward drug development applications, but we hope that it will be useful to clinical pharmacists as well, to increase the use of such tools in defining clinical practice and in supporting patient care.

Several researchers have made significant efforts, contributing additional tools to make the application of pharmacometrics more efficient and more attainable to a greater number of people. Tools are now available to assist with the validation of one’s NONMEM installation, the production of graphics to assess a dataset and modeling results, the automated implementation of extended processing for applications such as visual predictive checks and bootstrapping, and cataloging and comparing the history of models run during a project. A few of these excellent and generous efforts include the work of Dr. Mats Karlsson and colleagues at Uppsala University through the development and support of the Xpose and Perl Speaks NONMEM (PsN) software applications and the work of Dr. Marc Gastonguay and colleagues at Metrum Institute with the development and support of the nmqual package for validation of NONMEM installation and for their nmfuns set of tools. Wings for NONMEM by Dr. Nick Holford and PLT Tools by Dr. Dennis Fisher also prove to be useful for many. Most of these products are available free of cost but may have licensing fees to access support or additional features. The list of pharmacometric tools available has certainly been enhanced by the work of Dr. Tom Ludden, Dr. Robert Bauer, and colleagues at ICON Development Solutions, through their continued development of NONMEM and their add-on package PDx-POP™. This is by no means an exhaustive list of available tools as many others have made important contributions as well; those listed were simply the programs most familiar to the authors at the time of this writing.

The learning curve for the new pharmacometric analyst is quite substantial. In addition to the theory of the data analytic and statistical methodology, the understanding of experimental design, data collection, quality control, and many other critical components, there are at least three basic types of software with which the analyst must be familiar. These include: (i) a software tool for building the analysis dataset, (ii) software for the pharmacometric analysis itself, and (iii) pre- and postprocessing software for the preparation of graphics which are critical to understanding the data, presenting the results of modeling, and communicating to others the implications of the analysis. It is the latter, effective communication of modeling results and their implications to support critical decision making milestones that makes the significant effort and cost of population modeling worthwhile.

Once the basics of nonlinear mixed effects modeling are understood, one must continue learning new technical skills and gaining additional experience in effectively using the results of models to make informed decisions. And equally important, one must learn to communicate those decisions clearly and authoritatively to others. After mastering this introductory text, continued learning regarding theoretical and technical skills can be enhanced by the study of other texts that deal with the topics of pharmacometrics in more depth, but are perhaps more challenging as an introduction to the field than the present text (Ette and Williams 2007; Bonate 2011). It is our hope that the reader will find the present text a useful introduction to the field, one that may be utilized by graduate students, pharmaceutical scientists, statisticians, pharmacists, engineers, and others interested in gaining a sound, practical foundation in the concepts and techniques relevant to pharmacokinetic-pharmacodynamic modeling applications with NONMEM.

Supplemental content is provided online (http://booksupport.wiley.com) to support use of this text. The content varies across chapters, but includes some datasets, programming code, and questions for application in teaching or further contemplation.

The text includes the topics the authors feel are the most critical building blocks for a comprehensive understanding of population modeling and those that should be considered in most population modeling efforts. Several topics were therefore left behind in favor of increased attention to those included. Those topics not included herein should not be viewed as unimportant, but may require further understanding or background, or the benefit of experience from a project or two. Since we anticipate the potential audience for this text to be quite diverse, we do not assume a particular background or training as a foundation for the concepts herein. Instead, we believe that pharmacometrics involves the unique merging of many disciplines, any one of which might be the initial study of focus during one’s academic training. As such, concepts that rely on advanced pharmacokinetic or statistical training (for example) are explained in detail at the level at which understanding is required for the application described. And as they do in their teaching, the authors have attempted to impart as much practical wisdom and consideration of the subject matter as possible, so that the material is tangible for even the most naive student.

The authors’ considerable experience in applying modeling and simulation approaches to solve real-world drug development problems comes from more than 20 years of consulting for large and small pharmaceutical and biotechnology companies. The authors’ understanding of the researcher or student new to pharmacometrics comes from their personal experience in addition to more than 10 years of teaching introductory courses and workshops in population modeling. This book is written in response to the continual requests and inquiries of such students interested in further or continued study after attending an introductory population modeling workshop or course. For those who have already attended such a course, we hope that this book will fill in any gaps in understanding that may remain as a result of the necessary pace of such a workshop.

Acknowledgments

The authors are indebted to numerous mentors and colleagues, most either formerly or currently at Cognigen Corporation, with whom scientific discourse over the years has improved our understanding of the field in general and many of the specific topics herein. First and foremost, we wish to thank Ted Grasela for his tremendous influence on our thought processes and whose vision for the ever-increasing importance of quantitative approaches in the future pharmaceutical industry is a natural one to which we subscribe. In addition, we wish to thank several colleagues, including Luann Phillips, David Jaworowicz, and Brenda Cirincione, for sharing generously their wisdom and perspectives over the years. In particular, the authors wish to thank Rebecca Blevins, Amir Youssef, and Lineau Vilson, Jr. for their technical assistance with tables and careful review of the text. Most...