Andrew Hooker, Mats Karlsson, Sebastian Ueckert

Pharmacometric models are based on (patho-) physiological and pharmacological knowledge. The complexity and heterogeneity of biological data makes the knowledge about and development of statistical data analysis methods a central part of this scientific field. There are many benefits to using pharmacometric models in the analysis of data from clinical trials, for example the ability to handle sparse data and to integrate different types of observations into one model; however, these models are complex and intrinsically non-linear which presents technical challenges in model building and estimation.

One main challenge is to reduce the time it takes to develop these models. With complex, non-linear models and data from a clinical trial that can have thousands of data points from hundreds of patients with multiple response variables, computer runtimes become non-ignorable. Generally, run-times can be divided into short (minutes), intermediate (hours to days) and long (days to months). The number of runs in a complete analysis tends to range between 30 and many hundred. We are investigating the implementation and automation of important modelling tasks through the use of new algorithms developed in our research group. Additionally, we are developing new methods of model building and new algorithm development that can shorten run times and the number of steps needed in the model building process.

Other areas of active research include the influence on parameter estimates of single observations and rational and statistically correct algorithms for adding explanatory variables, .i.e. covariates, to the models.

Andrew Hooker, Mats Karlsson, Joakim Nyberg

There are two principle ways in which models can be used to evaluate and optimize clinical and pre-clinical experiments. The first is by simulation of a set of proposed designs from a model (or set of models), followed by evaluation of those resulting data sets using metrics of interest. The simulations, repeated many times with different random seeds, provide information about the expected results of various different designs (for example, measures of the precision and bias of parameter estimates, or the power to detect a drug effect of a specific size). With this methodology we have investigated, for example, differences in randomization schemes for dose-finding trials where it was found that dose-randomized trials are more powerful in characterizing the underlying relation compared to concentration-randomized trials. In most instances, this increase in power can be achieved with a similar or lower number of observed side effects.

The second way of evaluating and optimizing trial designs is through the use of optimal experimental design methodologies. These methods often rely on calculations of an Information Matrix (e.g. the Fisher Information Matrix), which characterizes the information content of any possible design. Each design evaluation is much quicker than clinical trial simulation, thus one can investigate the landscape of possible designs (within constraints) potentially available for an experiment, and even optimize a design based on this information. We have developed methods and software (PopED) that utilize these methods with both local and global design criteria (e.g. E-family optimal designs, which take into account the underlying uncertainty in a pharmacometric model description of a biological system. Additionally, while optimal design is often focused on optimization of sampling times in an experiment, the methodology can be applied to other aspects of trial designs, such as the dose administered or the length of run-in, treatment and wash-out phases of an experiment, these aspects are investigated in our research. Further, we have investigated the extension of optimal design methodology to optimize a study for other interesting quantities such as power, as opposed to the traditional optimization based on model parameter estimation uncertainty.

The two methods of evaluating and optimizing trial designs can be combined to evaluate and explore adaptive optimal designs. In these types of trial designs, interim analyses are used to update models used to describe the system being investigated and then to use this information to re-optimize the next cohort of patients coming into a study. With combined simulation/optimization one can explore the adaptation and optimization rules one will use in an adaptive trial. We are currently developing such a tool (MBAOD), and are investigating the use of such designs in, for example, pedatric bridging studies and time-to-event type studies.

Andrew Hooker, Mats Karlsson, Maria Kjellsson, Elodie Plan,
Sebastian Ueckert

For many diseases, the primary outcome is of discrete nature: stage category, symptom severity, number of events, or occurrence of events. In pharmacometrics we distinguish (non-)ordered categorical data, count data, and (repeated) time-to-event data. Models handling this type of data are based on probabilities and, even if they have been around for ~20 years in pharmacometrics, are still not widely used and subject to important innovations. In this project, we aim to study and develop new methodologies for discrete data, in order to better describe disease progression, characterize exposure-response with a higher power, as well as simulate clinical trials in a more realistic manner.

Our recent focus has been on score-based clinical outcomes, for which we introduced the item response theory approach as well as bounded integer models. While the former allows to describe disease assessments with unique details, the latter is a more parsimonious modeling approach with excellent statistical properties. In our work, we continue to investigate the statistical properties of these complementing techniques and develop tools and techniques to simplify their application. Another current area of research is the application of hidden Markov models in NONMEM, which we use, for example, for the detection of anti-drug antibodies.

In the past, we have analyzed sleep stages in patients with insomnia using Markov models and modeled pain scores rated on a Likert scale by neuropathic patients by including features for under-dispersion and serial correlation to count models. We also used daily numbers of seizures in the investigation of over-dispersion or Markov patterns in count data, and introduced repeated time-to-categorical event model to simultaneous characterize the drug effect on severity and time to acid reflux events.

Methodology-wise, we have compared parametric time-to-event models to semi-parametric Cox proportional hazard models, developed an approach to simulate large scale unbiased repeated time-to-event data, and studied methods such as dynamic inter-occasion variability or stochastic differential equations to handle within-subject variability in count models.

We also explored the performance of estimation methods available for discrete models and studied how the Laplace approximation behaves in situations with non-even distributions of ordered categories as well as for different Poisson-type models. In another study, we compared the accuracy of parameter estimation with SAEM and importance sampling to the one of Laplace in repeated time-to-event models where the frequency of individuals with events was low. We have also conducted a study investigating all methods available in NONMEM 7 for all types of discrete models.

Andrew Hooker, Mats Karlsson, Rikard Nordgren, Sebastian Ueckert

One integral part of all of our research activities is the implementation of the methods developed in freely available software to facilitate a wider and consistent use of the new algorithms.

The main software developed by the group is PsN https://uupharmacometrics.github.io/PsN/

Xpose http://xpose.sf.net

PopED http://poped.sf.net

PopED lite (http://www.bluetree.me/PopED_lite.html)

NCAPPC (https://cran.r-project.org/package=ncappc)

MBAOD (https://github.com/andrewhooker/MBAOD)

Andrew Hooker, Lena Friberg, Siv Jönsson, Mats Karlsson, Elodie Plan

Diseases:

Multiple Sclerosis is both a complex and chronic neurological disease of the CNS. The natural course of MS is slow and difficult to monitor clinically. The overall aim of this project is to establish the first population, data driven, MS disease progression model in order to construct a mathematical modelling platform where the interplay between the majority of relevant aspects of the disease, such as time course of disability progression, relapse rate dynamics, time course of the imaging data, time course of lymphocytes and population characteristics are incorporated. Model building involves sequential development of (i) separate models for all components of interest (disease progression, relapse rate dynamics, MRI dynamics and lymphocytes including CD4+, CD8+) and (ii) the covariate model, which explore the underlying patient factors influencing within- and between-subject variability in treatment response. The final anticipated model shall enable simultaneous characterization of the interplay between relapse rate dynamics, total CD4+ and CD8+ lymphocytes and their ratio, and MRI readout dynamics in the evolution of disease and, most importantly, link the time-course of MRI and clinical outcome (relapse rate and disability) and lymphocyte data.

In rheumatoid arthritis a range of variables of the disease are summarized into a clinical endpoint for evaluation of drug response - the dichotomous ACR20 score. Integrated longitudinal transition models with dropout are useful for understanding the outcome of different dosing schedules and by expanding such models to also include the more stringent ACR70 criteria more information can be preserved. To increase the information on the concentration-effect relationship in the available data, a longitudinal transition model describing the probability of ACR20, ACR50 and ACR70 responses have been developed.

Treatment:

Biological medicinal products are an important contributor in the treatment of many diseases, e.g. multiple sclerosis, rheumatoid arthritis, cancer and psoriasis. Characterization of biologics benefit from pharmacometric modelling, since they exhibit complex disposition characteristics, quite different to the processes and pathways utilized for small molecules, e.g. monoclonal antibodies exhibit target mediated drug disposition (TMDD) and disease-related changes in protein turnover

The study design of biological products needs to consider the special disposition features. We explore study design options for studies in different stages of drug development, optimal design methodology is applied to TMDD models.

A complicating factor for biologics is the occurrence of antidrug antibodies (ADA), which may affect the pharmacokinetic and efficacy features. The identification of ADAs is usually confounded by the presence of the drug itself and therefore the result from an analysis is that ADA is present, absent or unknown (missing information). Thus, there is a need to develop adequate methods to incorporate the ADA information in pharmacometric models. Currently, we are using mixed hidden Markov models (MHMM) to model the underlying unobservable ADA states.

Siv Jönsson, Mats Karlsson, Elisabet Nielsen

Large unexplained inter-patient variability in pharmacokinetics (PK) and pharmacodynamics (PD) may be a cause for suboptimal treatment in the individual patient. We aim to facilitate the use of and advance the methodologies for model-based treatment individualization using therapeutic drug monitoring (TDM) data. Ongoing projects relate to dose individualization of antibiotics in critically ill patients and factor VIII replacement therapy in hemophilia A patients.

Intensive care units (ICU) has a high prevalence of infections and a vulnerable and heterogeneous patient population. ICU patients exhibit extreme PK variability resulting in wide variations in drug exposures. Furthermore, the causative organisms in the ICU show a large variability in drug susceptibility, accordingly the drug exposure to target also needs to be personalised. We aim to develop and evaluate novel methods for dose individualization of antibiotics based on PK and PD modelling with the ultimate goal to improve patient outcome and minimize the emergence of resistance in the ICU.

Hemophilia are genetic disorders impairing blood coagulation due to deficiency of coagulation factor VIII (hemophilia A) and IX (hemophilia B), whereof hemophilia A is the more common although rare: in Sweden 14 of 100 000 boys and men have hemophilia, (80 % hemophilia A). In prophylactic treatment with factor VIII and IX, pharmacokinetic (PK) tailored dose individualization is promoted, and is indicated to reduce the total doses administered. However, extensive unexplained variability is present in the bleeding risk and there is a need to consider the bleeding phenotype when individualising the dose. We aim to develop model-based dose individualisation strategies/methods in haemophilia accounting for the patient’s PK and bleeding characteristics.

Lena Friberg, Mats Karlsson, Elisabet Nielsen

Antibiotics are considered one of the greatest discoveries of modern medicine. Today, treatment failures due to multidrug-resistant bacteria are becoming more frequently observed, and both a use and misuse of antibiotics accelerate this phenomenon.

Our research aim to advance the understanding of the pharmacokinetic/pharmacodynamic (PKPD) relationships for antibiotics of value for a more streamlined drug development process and an improved therapeutic use of clinically available antibiotics.

Typically, the PKPD characterization of antibiotics is done based on pre-clinical data and high performing translational methods are thus central in the assessment of an appropriate antibiotic drug use. Mechanism-based models describing time-kill curves from in vitro experiments form the basis of our modelling. The developed models have shown to be applicable across drugs and bacterial strains, for both static and dynamic concentration experiments, for different sizes of start inocula, and for predicting selection of resistance in competition experiments. Based on the developed models, optimal experimental design techniques are applied to find experimental protocols that increase the efficiency of both pre-clinical and clinical studies.

The use of a mechanism-based PKPD modelling approach in dose selection has been suggested for increased robustness and extrapolation potential, especially for special patient populations. To further increase the translatability of pre-clinical results, our current research aim to incorporate the activation and effect of the innate immune response in the model predictions. We also focus on describing the efficacy of antibiotic combination therapies, where the use of mechanism-based modelling that describes the combined effect on the bacterial killing while taking the time-aspect of PK as well as PD into account, is highly advantageous and may facilitate the translation of in vitro information to in vivo.

We develop PKPD models describing the time-courses for a range of variables of interest for cytotoxic, target and immunotherapies, and characterize relationships between them; biomarkers, drug-induced toxicity, tumor size measurements (diameters, volumes), tumor and immune response activity, patient reported outcomes (PRO) as well as overall survival. The models are aimed at being of value to support development of new and existing drug therapies, including being a tool for individualized dose-adaptations. By integrating information of different variables into a modelling framework the variables’ relations and predictive value can be tested, and a better overview of both desired and adverse effects from a changed dosing regimen can be obtained. For example, a modelling framework, including biomarkers, side-effects, tumor response and survival, has been developed for sunitinib in gastrointestinal stromal tumors that can be used to explored consequences on survival and different adverse effect from a changed dose. We also explore what metrics of tumor size, constant and time-varying, are predictive of the hazard of death, and if changes in individual lesions, and their location, are predictive of patient outcomes and better than the standard measure sum of longest diameters (SLD). The models can also be used to explore different concepts of study design in oncology, for example to identify a strategy to define an optimal combination of drugs.

Thomas Dorlo, Mats Karlsson, Maria Kjellsson, Elin Svensson

Diseases as tuberculosis (TB), HIV and malaria claim 4 million lives every year, the majority in low- and middle-income countries. In the area of global health, we use pharmacometric methodology to support development of new treatments and optimize therapy for these types of infectious diseases.

Better TB treatment for children

About one million children fall ill with TB each year. Model-based analysis is extra valuable for studies in children where is it crucial to characterize the pharmacokinetics from a limited number of sample points. In the network IMPAACT we contribute to design and data analysis of studies aimed to determine optimal dosing regimens for children of different ages, and specifically for children co-infected with HIV and TB.

Through the consortium BenefitKids we investigate new, child-friendly formulations of existing TB drugs. Together with TB Alliance we participate in the pediatric development program for pretomanid, the newest TB drug.

New compounds against multi-drug resistant TB

Development of drug-resistance among bacteria is a global health emergency and multi-drug resistant (MDR) TB is one of the worst forms. We investigate new compounds against MDR-TB and have quantified pharmacokinetic drug-drug interactions for a large number of combinations (1–4). We also demonstrated that model-based analysis is a better method than traditional statistical analyses (5), a result now incorporated in FDA’s guidelines. To enable a better description of concentration-effect relationships in TB we have developed a new type of semi-mechanistic model (see figure) which can describe the bacterial load in the body over time on treatment (6,7).

Simplifying TB treatment

Current standard treatment includes 4 drugs and takes at least 6 months to complete. Together with PanACEA and the research group Pharmacokinetics and Quantitative Pharmacology we participate in several projects aiming to simplify TB treatment. We have investigated if it is possible to shorten treatment duration (8), to give the same dose to all adults instead of using a weight-based schedule (9), and developed methods for evidence-based design of fixed-dose combination tablets (10).

Deadly parasites

Parasitic infections hit children and pregnant women extra hard, two populations that are typically excluded from clinical trials. Pharmacometric models can be used to optimize treatment for these vulnerable patients. We have investigated how antimalarial drugs should be dosed best in pregnant women (11)and why children have a higher risk of relapse after the end of leishmaniasis treatment (12).

References:

1. Svensson EM, Aweeka F, Park J-G, Marzan F, Dooley KE, Karlsson MO. Model-Based Estimates of the Effects of Efavirenz on Bedaquiline Pharmacokinetics and Suggested Dose Adjustments for Patients Coinfected with HIV and Tuberculosis. Antimicrob Agents Chemother. 2013 Jun 1;57(6):2780–7.
2. Svensson EM, Dooley KE, Karlsson MO. Impact of Lopinavir-Ritonavir or Nevirapine on Bedaquiline Exposures and Potential Implications for Patients with Tuberculosis-HIV Coinfection. Antimicrob Agents Chemother. 2014 Nov;58(11):6406–12.
3. Svensson EM, Murray S, Karlsson MO, Dooley KE. Rifampicin and rifapentine significantly reduce concentrations of bedaquiline, a new anti-TB drug. J Antimicrob Chemother. 2015 Apr;70(4):1106–14.
4. Brill MJE, Svensson EM, Pandie M, Maartens G, Karlsson MO. Confirming model-predicted pharmacokinetic interactions between bedaquiline and lopinavir/ritonavir or nevirapine in patients with HIV and drug-resistant tuberculosis. Int J Antimicrob Agents. 2017 Feb;49(2):212–7.
5. Svensson EM, Acharya C, Clauson B, Dooley KE, Karlsson MO. Pharmacokinetic Interactions for Drugs with a Long Half-Life—Evidence for the Need of Model-Based Analysis. AAPS J. 2016 Jan;18(1):171–9.
6. Svensson EM, Karlsson MO. Modelling of mycobacterial load reveals bedaquiline’s exposure-response relationship in patients with drug-resistant TB. J Antimicrob Chemother. 2017 Dec 1;72(12):3398–405.
7. Tanneau L, Karlsson MO, Svensson EM. Understanding the drug exposure-response relationship of bedaquiline to predict efficacy for novel dosing regimens in the treatment of multidrug-resistant tuberculosis. Br J Clin Pharmacol. 2020 May;86(5):913–22.
8. Svensson EM, Svensson RJ, Te Brake LHM, Boeree MJ, Heinrich N, Konsten S, et al. The Potential for Treatment Shortening With Higher Rifampicin Doses: Relating Drug Exposure to Treatment Response in Patients With Pulmonary Tuberculosis. Clin Infect Dis Off. 2018 18;67(1):34–41.
9. Susanto BO, Svensson RJ, Svensson EM, Aarnoutse R, Boeree MJ, Simonsson USH. Rifampicin can be given as flat-dosing instead of weight-band dosing. Clin Infect Dis. 2019; Epub ahead of print
10. Svensson EM, Yngman G, Denti P, McIlleron H, Kjellsson MC, Karlsson MO. Evidence-Based Design of Fixed-Dose Combinations: Principles and Application to Pediatric Anti-Tuberculosis Therapy. Clin Pharmacokinet. 2018;57(5):591–9.
11. Lohy Das J, Rulisa S, de Vries PJ, Mens PF, Kaligirwa N, Agaba S, et al. Population Pharmacokinetics of Artemether, Dihydroartemisinin, and Lumefantrine in Rwandese Pregnant Women Treated for Uncomplicated Plasmodium falciparum Malaria. Antimicrob Agents Chemother. 2018;62(10).
12. Dorlo TPC, Kip AE, Younis BM, Ellis SJ, Alves F, Beijnen JH, et al. Visceral leishmaniasis relapse hazard is linked to reduced miltefosine exposure in patients from Eastern Africa: a population pharmacokinetic/pharmacodynamic study. J Antimicrob Chemother. 2017 Nov 1;72(11):3131–40.

Andrew Hooker, Lena Friberg, Mats Karlsson, Maria Kjellsson, Elodie Plan, Sebastian Ueckert

The term degenerative or progressive disorders commonly summarizes diseases in which the function or structure of the affected tissues or organs worsens over time. In most cases, progressive disorders are caused by degenerative processes that gradually reduce the physiologic operation of organs and tissues. These disorders share several features that complicate the evaluation of treatment effects in a clinical trial, such as increased drop out, complex clinical endpoints, challenging patient populations, and relatively small effect sizes. In this research area, we focus describing these data in models and with modelling techniques addressing these challenges in many disease areas.

Multiple Sclerosis, for example, is both a complex and chronic neurological disease of the CNS. The natural course of MS is slow and difficult to monitor clinically. For this disease, we are developing the first population, data-driven, MS disease progression model in order to construct a mathematical modelling platform where the interplay between the majority of relevant aspects of the disease, such as time course of disability progression, relapse rate dynamics, time course of the imaging data, time course of lymphocytes and population characteristics are incorporated.

Another example is rheumatoid arthritis, a progressive, autoimmune disease that causes inflammation, swelling, and pain in the joints. Here we developed integrated longitudinal transition models for the dichotomous ACR20 score to understand the outcome of different dosing schedules.

We also introduced the item response theory approach to the field, that is uniquely well suited to describe score based clinical outcomes often used in this class of diseases. The application of IRT does not only results in a more exact description of the assessment score and increased statistical power but also provides insight into the assessment properties. We have applied such models to Alzheimer’s (with ADAS-cog), Schizophrenia (with PANSS), multiple sclerosis (with EDSS) and Parkinson’s (with UPDRS). Currently, we aim to continue to extend this technique to more diseases.

Mats Karlsson, Maria Kjellsson

Around 6-10% of the world population is estimated to suffer from diabetes[1] (depending on region) and since the rule of halves framework applies to diabetes, only a minor part of these patients lives healthy lives; half of the 422 million people with type 2 Diabetes are not diagnosed; half of those diagnosed do not receive care; half of the cared for patients do not achieve treatment target; and half of the patients achieving target do not achieve desirable outcomes[2]. There are many challenges to be tackled in prevention, diagnosis, treatment availability and optimisation and protection against long-term complications, e.g. cardiovascular disease and kidney failure.

The first mathematical model in the diabetic field was published in 1961 by Bolie[3]; a differential equation system describing the glucose-insulin dynamic. Since then mathematical models have been used to understand and predict the complex aspects of diabetes, e.g. glucose homeostasis for diagnosis and insulin pumps, epidemiology of diabetes and its complications and cost-effectiveness of diabetes care.

Pharmacometric models is used to support drug development in diabetes. Pharmacometric models have been used to understand pharmacokinetics of anti-hyperglycaemic drugs and pharmacodynamics of short- to medium-term biomarkers, e.g. FPG, HbA1c, i.e. treatment optimisation. In recent years, focus of pharmacometric aided-drug development has shifted towards prevention with models of impaired glucose tolerance and obesity with associated progression, as well as protection against and treatment of complications.

The research in our group have followed a similar trend as that of the pharmaceutical industry and we have developed models describing pharmacokinetics[4], glucose homeostasis[5],[6],[7],[8],[9],[10],[11],[12] and time-course of HbA1c[13],[14],[15], investigated design of studies with short- and medium-term biomarkers[16],[17],[18],[19], and quantified disease progression in diabetes[20] and diabetes onset[21],[22]. In recent years, we have focused more on factors predictive of onset e.g. obesity[23],[24],[25], and smoking[26],[27], and complications of diabetes, e.g. cardiovascular disease and decreased kidney function[28] and overall survival[29].

In 2018, the group received a grant from the Swedish Research Council to develop a tool, CARE (Cardiovascular Absolute Risk Estimator). This tool will, based on an individual’s laboratory values and weight, visualise the absolute risk of cardiovascular disease. Prognostic predictions of how the risk change for various treatments and success in reaching treatment target, will hopefully motivate patient adherence to therapy as well as help engage the patient in treatment decisions.

References:

[1] Global statistics on diabetes – ESC. https://www.escardio.org/Sub-specialty-communities/European-Association-of-Preventive-Cardiology-(EAPC)/News/global-statistics-on-diabetes. Accessed Sept 2019

[2] Hart JT. Rule of halves: implications of increasing diagnosis and reduced dropout from future workload and prescribing costs in primary care. Br J Gen Pract. 1992; 42: 116-119.

[3] Bolie VW. Coefficients of normal blood glucose regulation. J Appl Physiol. 1961; 16: 783-788.

[4] Stage TB, Wellhagen G, Christensen MMH, Guiastrennec B, Brøsen K, Kjellsson MC. Using a semi-mechanistic model to identify the main sources of cariability of metformin pharmacokinetics. Basic Clin Pharmacol Toxicol. 2019; 124: 105-114.

[5] Choy S, Hénin E, van der Walt JS, Kjellsson MC, Karlsson MO. Identification of the primary mechanism of action of an insulin secretagogue from meal test data in healthy colunteers based on an integrated glucose-insulin model. J Pharmacokinet Pharmacodyn. 2013; 40: 1-10.

[6] Røge RM, Klim S, Kristensen NR, Ingwersen SH, Kjellsson MC. Modeling of 24-hours glucose and insulin profiles in patients with type 2 diabetes mellitus treated with biphasic insulin aspart. J Clin Pharmacol. 2014; 54: 809-817.

[7] Alskär O, Bagger JI, Røge RM, Knop FK, Karlsson MO, Vilsbøll T, Kjellsson MC. Semimechanistic modelling describing gastric emptying and glucose absorption in healthy subjects and patients with type 2 diabetes. J Clin Pharmacol. 2016; 56: 340-348.

[8] Røge RM, Klim S, Ingwersen SH, Kjellsson MC, Kristensen NR. The effect of a GLP-1 analog on glucose homeostasis in type 2 diabetes mellitus quantified by and integrated glucose insulin model. CPT Pharmacometrics Syst Pharmacol. 2015; 4: e00011.

[9] Parkinson J, Hamrén B, Kjellsson MC, Skrtic S. Application of the integrated glucose-insulin model for cross-study characterization of T2DM patients on metformin background treatement. Br J Clin Pharmacol. 2016; 82: 1613-1624.

[10] Ibrahim MMA, Largajolli A, Karlsson MO, Kjellsson MC. The integrated glucose insulin minimal model: an improved version. Eur J Pharma Sci. 2019; 134: 7-19.

[11] Røge RM, Bagger JI, Alskär O, Kristensen NR, Klim S, Holst JJ, Ingwersen SH, Karlsson MO, Knop FK, Vilsbøll T, Kjellsson MC. Mathematical modelling of glucose-dependent insulinotropic polypeptide and glucagon-like peptide-1 following ingestion of glucose. Basiv Clin Pharmacol Toxicol. 2017; 121: 290-297.

[12] Alskär O, Karlsson MO, Kjellsson MC. Model-based interspecies scaling of glucose homeostasis. CPT Pharmacometrics Syst Pharmacol. 2017; 6: 778-786.

[13] Møller JB, Overgaard RV, Kjellsson MC, Kristensen NR, Klim S, Ingwersen SH, Karlsson MO. Longitudinal modeling of the relationship between mean plasma glucose and HbA1c following antidiabetic treatment. CPT Pharmacometrics Syst Pharmacol. 2013; 2: e82.

[14] Møller JB, Kristensen NR, Klim S, Karlsson MO, Ingwersen SH, Kjellsson MC. Methods for predicting diabetes phase III efficacy outcome from early data: superior performance obtained using longitudinal approach. CPT Pharmacometrics Syst Pharmacol. 2014; 3: e122.

[15] Claussen A, Møller JB, Kristensen NR, Klim S, Kjellsson MC, Ingwersen SH, Karlsson MO. Impact of demographics and disease progression on the relationship between glucose and HbA1c. Eur J Pharm Sci. 2017; 104: 417-423.

[16] Kjellsson MC, Cosson VF, Mazer NA, Frey N, Karlsson MO. A model-based approach to predict longitudinal Hba1c, using early phase glucose data from type 2 diabetes mellitus patients after anti-diabetic treatment. J Clin Pharmacol. 2013; 53: 589-600.

[17] Wellhagen GJ, Karlsson MO, Kjellsson MC. Comparison of power, prognosis, and extrapolation properties of four population pharmacodynamic models of HbA1c in type 2 diabetes. CPT Pharmacometrics Syst Pharmacol. 2018; 7: 331-341.

[18] Ibrahim MMA, Ghadzi SMS, Kjellsson MC, Karlsson MO. Study design selection in early clinical anti-hyperglycemic drug development: a simulation study of glucose toelrance tests. CPT Pharmacometrics Syst Pharmacol. 2018; 7: 432-441.

[19] Sheikh Ghadzi SM, Karlsson MO. Kjellsson MC. Implications for drug characterization in glucose tolerance tests without insulin: simulation study of power and predictions using model-based analysis. CPT Pharmacometrics Syst Pharmacol. 2017; 6: 686-694.

[20] Choy S, Kjellsson MC, Karlsson MO, de Winter W. Weight-HbA1c-insulin-glucose model for describing progression of type 2 diabetes. CPT Pharmacometrics Syst Pharmacol. 2016; 5: 11-19.

[21] Choy S, de Winter W, Karlsson MO, Kjellsson MC. Modeling the disease progression from healthy to overt diabetes in ZDSD rats. AAPS J. 2016; 18: 1203-1212.

[22] Ibrahim MMA, de Melo VD, Uusitupa M, Tuomilehto J, Lindström J, Kjellsson MC, Karlsson MO. Competing risks analysis of the Finnish diabetes prevention study. PAGE 28. 2019; abstr 9033.

[23] Leohr J, Heathman M, Kjellsson MC. Semi-Physiological model of postprandial triglyceride response in lean, obese and very obese individuals after a high-fat meal. Diabetes Obes Metab. 2018; 20: 660-666.

[24] Leohr J, Heathman M, Kjellsson MC. A semi-physiological model of postprandial triglyceride response following anti-obesity therapy. PAGE 26. 2017; abstr 7227.

[25] Leohr J, Kjellsson MC. A categorical model of sweet/fat preference taste in lean, obsess and very obese subjects. PAGE 27. 2018; abstr 8521.

[26] Germovsek E, Hansson A, Kjellsson MC, Perez Ruixo JJ, Westing Å, Soons AP, Vermeulen A, Karlsson MO. Relating nicotine plasma concentration to momentary craving across four nicotine replacement therapy formulations. Clin Pharmacol Ther. 2019; epub head of print.

[27] Germovsek E, Hansson A, Karlsson MO, Westin Å, Soons PA, Vermeulen A, Kjellsson MC. A time-to-event model relating integrated craving to risk of smoking relapse across different nicotine replacement therapy formulations. PAGE 28. 2019; abstr 9074.

[28] Wellhagen G, Hamrén B, Kjellsson MC, Åstrand M. Modeliing UACR as a clinical endpoint. PAGE 28. 2019; abstr 9152.

[29] Kunina H, Kjellsson MC. Diabetes progression modelling of competing risks of long-term complications and mortality using Swedish registry data. PAGE 28. 2019; abstr 9083.