Work

Pharmacometric Modeling of Disease Progression and Biomarker Dynamics in Type 2 Diabetes

Year: 2026
Type: dissertation
Abstract: In antidiabetic drug development for type 2 diabetes (T2D), short-term changes in glucose-related biomarkers need to be translated into longer-term treatment response and clinical outcomes. This translation is complicated by the delayed HbA1c response, the progressive nature of T2D, and the variability between patients in both biomarker trajectories and complication risks. Pharmacometric models can characterize this variability and quantify relationships between treatment, biomarkers, and clinical outcomes, thereby supporting prediction and extrapolation. This thesis focused on the development and application of pharmacometric models to address methodological challenges across antidiabetic drug development and the clinical course of T2D, using simulated, clinical trial, and registry data. Simulation-based evaluation showed that, under appropriate model assumptions, trials using HbA1c as an endpoint may be shortened while still supporting extrapolation of the long-term response. A model based on continuous glucose monitoring data was developed to describe glucose profiles in the absence of recorded meal times and meal sizes. By integrating a meal-input model, glucose-insulin dynamics, and insulin glargine pharmacokinetics, the model described population-level glucose profiles, providing a tool for prospective simulations. To quantify T2D disease progression in newly diagnosed adult patients, a multi-state model with competing risks was developed using registry data. The model described progression from uncomplicated T2D to macrovascular complications, diabetic kidney disease, the combined state, and death, quantifying how complications accelerate mortality. To complement this event-based analysis, routinely collected biomarkers were integrated into composite measures of cardiometabolic dysfunction using item response theory. Two underlying dimensions were identified, blood pressure-related and metabolic. These latent composite measures, together with individual biomarkers, were then evaluated as covariates using alternative representations within the multi-state framework. In this integrated framework, time-varying biomarker representations were generally more informative than baseline values for transition-specific hazards. However, several of these associations showed counterintuitive directionality, highlighting how treatment-by-indication confounding and informative missingness can complicate the interpretation of covariate-outcome relationships in routine clinical data. In conclusion, this thesis demonstrates how pharmacometric modeling can support antidiabetic drug development and characterize T2D progression by informing trial duration, improving interpretation of glucose profiles, characterizing the risks of complications and death, and evaluating how the representations of longitudinal biomarkers influence associations with long-term clinical outcomes.
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W7204512780
Cites:
Cited by:
Related to: 10