Predicting Asian glycemic responses in food using a physiologically-based pharmacokinetic-pharmacodynamic model

• Main Author: Koh, Hor Cheng
• Other Authors: Chen Wei Ning, William
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2025
• Subjects: Medicine, Health and Life Sciences; Glycemic response; Pharmacokinetics; Insulin; Glucose; Pharmacodynamics
• Online Access: https://hdl.handle.net/10356/182771
• Institution: Nanyang Technological University

Diabetes mellitus has become an ever-growing health concern especially in Singapore with a prevalence rate of 10.5%, markedly higher than the expected global average of 8.8%. Asians are naturally predisposed to greater degree of insulin secretion impairment and have a higher risk of developing prediabetes compared to their Caucasian counterparts. Unlike full-blown diabetes, prediabetes is a reversible condition if early intervention is provided. However, current gold standards for prediabetes detection suffer from suboptimal sensitivity and accuracy, missing out on as many as one-third of prediabetes cases. Insulin, on the other hand, is a more sensitive biomarker for prediabetes detection yet is rarely clinically measured due to its relatively high costs and inconvenience. Additionally, with the advent of innovative technology such as continuous glucose monitoring devices, more healthy individuals are paying heed to their dietary intake and monitoring their blood glucose level for healthier living. This naturally led to the emergence of many novel low glycemic index food products in the market. The product development cycle for novel food products is often extremely costly and time-consuming and fraught with additional regulatory hurdles compared to conventional food products. The acquisition of in vivo glucose (and optionally insulin) data can significantly help in overcoming these regulatory requirements or for early pre-screening of several candidate food products for product prioritization before committing to a clinical study. There is consequently a demand for a research tool with the ability to predict glycemic response of novel food products after oral intake to help aid industries in product development. My thesis seeks to tackle both clinical and food aspects with the aim of developing a glucose model that can help predict glycemic response and prediabetes risks of individuals in the general population. The model is constructed using concepts from physiologically-based pharmacokinetic-pharmacodynamic (PBPK-PD) modelling and properly validated with both in vitro and in vivo human clinical data that were conducted during the project timeframe. In the first chapter of my thesis, the relevant background information and existing gaps in the clinical and food settings were discussed. The second chapter reviewed various glucose homeostasis models published by others, evaluating their strengths and weaknesses for predicting glycemic response. In the third chapter, we comprehensively detailed the model construction by breaking it down into multiple different sub-models each representing different processes of glucose homeostasis. For the fourth chapter, model parameterization and validation were performed using in vivo clinical data and comparing it with the model prediction. The fifth chapter focused on the clinical applications mentioned earlier, starting off with a fasting insulin reference range meta-analysis work followed by model deployment for predicting prediabetes risks in Asians. The food aspect of my work will be covered in the sixth chapter where we explored the application of our model for predicting glycemic response in novel low glycemic index food products. Finally, the seventh chapter will provide recommendations for potential future works to build upon and improve the model or to expand the application of the model towards other areas.