South Africa 2002 PAGANZA


20 November 2002 – 22 November 2002


Wednesday 20 – Friday 22 November 2002

Department of Pharmacology, University of Cape Town, South Africa

Last Updated  04 October 2004

Please contact Justin Wilkins <[email protected]> for details of local arrangements.


PAGANZA and PAWS will be presented by a panel of South African and International Faculty



Duration – 2 days.

The course will consist of lectures and hands-on exercises at a computer. During the hands-on exercises, tutors will be available for consultation and assistance.

The Beginners Course and the Intermediate Course will be conducted as parallel sessions. Attendance at the Intermediate Course requires experience with the conduct of at least one population approach analysis.


Day 1 – 20 November 2002

Beginners and Intermediate

08h00 – 11h00
Lecture Session
Introduction to Modelling NH
Review of Basic PK and PD Models LMcF
Introduction to the Population Approach DM
Fixed Effects
Random Effects
Covariate Models


11h00 – 13h30 JW
Poster session & lunch




13h30 – 16h30 NH/LMcF
Hands On Session (NONMEM)

Population Approach NH
Covariate Models NH

Theophylline Pharmacodynamic Model NH


Review of Theophylline PD Results



13h30 – 16h30 DM/RG/CP
Lecture Session
Introduction to Logistic Regression (Odd Type Data) Analysis DM
Initial Estimates
Verbatim Code
Control Stream

Day 2 – 21 November 2002


08h00 – 11h00 CP/RG/DM
Lecture Session
Compartmental Pharmacokinetics CP
The PREDPP Library
Data File Format for Pharmacokinetics
Examining NONMEM Output
Common Errors and Error Messages


08h00 – 09h00 NH/LMcF
Lecture Session
Introduction to the Use of Differential Equations NH
Why use Differential Equations?
NM-TRAN control stream

Data Set CMT Item
Use of Bioavailability Fraction to Initialize Compartments

The Use of TIME and T


09h00 – 11h00 NH/LMcF
Hands On
Ethanol Pharmacokinetic Model NH


Ethanol Effect on CYP2E1 Turnover


11h00 – 13h30 JB
Question and Answer session & Lunch


13h30 – 16h30  CP/RG/DM
Hands On Session (NONMEM):
Theophylline Pharmacokinetic Model RG




Structural, Covariate and Residual Error Model Exploration

Review of PK Results


13h30 – 16h30 NH/LMcF
Lecture Session
Review of Hands-On Session
Hypothesis Testing and Model Building LMcF
Population Model Confidence Intervals (bootstrap) NH
Randomization Test NH
Day 3 – 22 November 2002





Attendance at the Population Approach Workshops is not a pre-requisite for attendance. The focus of the presentations will be towards a clinical audience with only necessary technical details. The objective is to show potential for use of the methodology.


Each presentation will be 40 minutes with 20 minutes question time.


08:45 Introduction to PAGANZA


09:00 Osteoporosis

Selection of Bisphosphonate Dosage Regimens in Osteoporosis by Modelling of Bone Resorption Biomarker Response – Colin Pillai


The ultimate clinical response in osteoporosis clinical trials is fracture efficacy – however, measurement of this endpoint requires long clinical trials of 3-5 years duration. This emphasises the importance of careful dose and regimen selection for the Phase III trials and argues in favour of identifying and using appropriate biomarkers to provide early evidence that may be used to inform dose and regimen selection. During this presentation, PKPD modelling and clinical trial simulation techniques will be used to obtain an understanding of the dose-regimen-response surface of biomarker response in osteoporosis.


10:00 Oncology

Logistic Regression and Mixed Effects Modelling as Complementary Tools in Clinical Drug Development: capecitabine as an example – Ronald Gieschke


In the development of XELODA (capecitabine), a novel orally administered tumor selective cytotoxic (5-FU) agent, logistic regression was used in Phase I to select capecitabine metabolites for subsequent PK model based evaluations. Using data from a bioequivalence study, amultiresponse population pharmacokinetic model was developed for jointly describing the time course of plasma concentrations of 5′-DFUR, 5-FU and FBAL. After addition of sparsely sampled data from Phase III colorectal cancer patients, the model detected e.g. a statistically significant relationship between alkaline phosphatase and 5-FU clearance. Individual PK parameter values were subsequently used in an exposure-response (safety, efficacy) analysis.


11:00 Coffee/Tea


11:30 Using PKPD to Evaluate and Optimize Dose for Hematopoietic Factors -Diane Mould


Hematopoietic growth factors stimulate differentiation and maturation of pluripotent stem cells, resulting in increased cell counts of a specific lineage.  These agents are used to treat patients with leukocytopenia, thrombocytopenia and anemia resulting from disease or therapy.  New approaches in formulation and manufacturing processes have promoted the development of novel chemical modifications for proteins, such as pegylation.  These chemical modifications alter pharmacokinetics (PK) and pharmacodynamics (PD), resulting in a greater flexibility in dose regimens. Consequently, utilizing PK and PD to optimize dose regimen or drug formulations has become an important tool.

There are numerous publications on the biology and PKPD of hematopoietic factors.  Although the models describing response to drug differ to some extent, the majority utilize some form of an indirect effect model. Growth factors have in common several properties that must be considered in PK and PD modeling.  They stimulate up-regulation of cell surface receptors, resulting in increased drug clearance.  As a consequence of binding and uptake by cells, many of these agents also exhibit nonlinear PK.  Hematopoietic factors are low molecular weight proteins, which undergo rapid clearance, requiring frequent administration. The PD of growth factors is often linked to the desired clinical outcome, making PK/PD modeling valuable.  Their PD involves hemeostatic mechanisms and the time required for stem cell differentiation and maturation.  It is necessary to consider receptor availability and the effects of endogenous growth factors when assessing the PD these agents.  Following chemotherapy, white cells are diminished, causing endogenous growth factor concentrations to increase.  Administration of exogenous growth factors immediately following chemotherapy is of little benefit, however, since the number of stem cells available to respond is also diminished.  Delaying administration of the exogenous growth factors too much results in less activity since endogenous growth factor levels have increased. Because these agents are administered parenterally, chemically modifications, such as pegylation of the parent compound have been developed.  Pegylation reduces the non receptor mediated clearance of these agents, permitting less frequent dosing.  However, chemical modification can also alter the binding affinity, resulting in an attenuated response. Development of chemically modified growth factors requires balancing  PK improvements with demonstrated PD response.  Clinical trial simulation is useful for guiding drug development strategies for these agents.

PKPD is useful to develop optimal treatment regimens and select appropriate chemical modifications for hematopoietic factors.


12:30 Lunch


14:00 Cardiovascular

Clinical versus statistical significance in covariate selection in population modelling: Dofetilide as an example.– LynnMcFadyen


With large phase 3 data sets it is possible to find a large number of covariates that test significant even if  p<0.001 is used in covariate modelling. Dofetilide is an antiarrhythmic agent where dosing is adjusted according to creatinine clearance and QTc interval. Its disposition is affected mainly by creatinine clearance. However in the phase 3 analysis a number of additional covariates where shown to have statistically significant effects on pharmacokinetic parameters . The clinical significance of these parameters were considered in selecting relevant covariate effects for drug labelling.


15:00 Parkinson’s Disease

Parkinson’s Disease: Progression of Disease and Drug Action – Nick Holford


Studies of patients with Parkinson’s disease over several years have shown how the severity of the disease changes with time. The population approach has been used to describe these changes in disease as well as characterising the systematic and random variability in pharmacokinetic and pharmacodynamic parameters. Levodopa appears to slow the progression of Parkinson’s disease while the maximum symptomatic effect of levodopa increases slowly over the first year of treatment.


16:00 Panel Discussion


16:30 Meeting Ends





Colin Pillai trained as a pharmacist at the University of Durban-Westville, South Africa. His post-graduate qualifications have involved PK and PKPD modelling. His therapeutic area interests have included asthma, multidrug resistant tuberculosis and HIV infection. In SA, he has held academic positions at UDW and WITS and been a researcher and consultant to the MRC’s National TB Research Programme. During the past 3 years he has worked in Switzerland at F Hoffmann-La Roche and at Novartis as a Clinical Pharmacologist involved with PKPD modelling and Clinical Trial Simulation in clinical drug development.


Ronald Gieschke received his mathematical (MS) and medical (MD) training at the University of Goettingen, Germany. Subsequently, in 1986, he joined F. Hoffmann-La Roche, Basel, Switzerland, as a clinical pharmacologist supporting drug development in the Central Nervous System area. Since 1994 he has been primarily involved in pharmacometrics topics, such as pharmacokinetic/pharmacodynamic mathematical modelling and clinical trial simulation. His special research interests are in the area of oncology, virology, osteoporosis and diabetes.


Diane Mould received her PhD in Pharmaceutics and Pharmaceutical Chemistry from the Ohio State University College of Pharmacy, USA in 1989.  She spent 12 years as a pharmacokineticist in the pharmaceutical industry, where she specialized in population PK and PD modeling.  During that time, she conducted population PK/PD analyses of hematopoietic agents, anti-cancer and anti-viral agents.  She also worked on models for sedative/hypnotic and cardiovascular agents.  Diane has been involved in trial simulation and study design in drug development.  She has published 14 peer-reviewed articles, 4 book chapters, and has made numerous national and international presentations of advanced modeling and simulation work.


Lynn McFadyen received her Pharmacy degree from the University of Potchefstroom, South Africa. She became interested in clinical pharmacology at the University of Cape Town while studying for her masters degree. From 1982 to 2000 she taught pharmacology at theUniversity of Durban-Westville, where Raymond Miller had started a population modelling group in the mid 1980s. Her research interests are pharmacokinetics, pharmacodynamics and reproductive health. In July 2001 she joined the Modelling and Simulation Group (Clinical Sciences) of Pfizer, Sandwich, UK

Nick Holford received his medical training at the University of Manchester, United Kingdom, and subsequently became a clinical pharmacologist at the University of California San Francisco, USA. Since 1983 he has been involved in teaching and research at the University of Auckland, New Zealand. His research interests are in pharmacokinetics, pharmacodynamics and disease progression with a particular focus on Alzheimer’s and Parkinson’s Disease.