Rolien Bosch 1, Sjaam Jainandunsing 2, Eric J.G. Sijbrands 3, Nelleke Snelder 1
1 LAP&P Consultants B.V. (Leiden, The Netherlands), 2 Department of Internal Medicine, Anna Hospital ( Geldrop, The Netherlands), 3 Department of Internal Medicine, Erasmus MC, University Medical Centre ( Rotterdam, The Netherlands)
Objectives
Type 2 diabetes mellitus (T2DM) is a major global health challenge, marked by insulin resistance and progressive dysfunction of pancreatic beta cells. While insulin resistance is a key feature, the inability of beta cells to adequately compensate plays a central role in the onset and progression of the disease. Understanding β-cell function, particularly insulin biosynthesis and secretion, is therefore of great value. Changes in the glucose concentration are the most important signal for the β-cell response. In addition, gastrointestinal hormones Glucagon-Like Peptide-1 (GLP-1) and Glucose-dependent Insulinotropic Peptide (GIP) play an essential role in in vivo insulin secretion and biosynthesis during oral food intake by stimulating the glucose-dependent insulin secretion. Assessing the individual β-cell function and insulin sensitivity are key to optimise diabetes treatment, including tailoring lifestyle interventions and pharmacological strategies such as incretin-based therapies versus exogenous insulin. Currently, individual estimates of these metabolic indices are commonly obtained using the oral minimal model (OMM) method, which combines glucose, C-peptide and insulin minimal models[1]. Although it is recognised as a reference method in metabolic research, its clinical application is limited by its multistep nature and the substantial technical expertise required. We introduce a model-based approach that simultaneously quantifies and visualises population- and individual-level β-cell function and insulin sensitivity using our previously published glucose-glucagon-GLP-1-GIP-insulin (4GI) [2], [3], [4] quantitative systems pharmacology (QSP) model.
Methods
To quantify the insulin release from β-cells prior to hepatic extraction, the 4GI model was extended with C-peptide dynamics, resulting in the 4GIC model. C-peptide was linked to insulin via a descriptive liver compartment and insulin delivery rate was described via distribution of insulin from the liver to the central compartment. C-peptide and insulin release from the pancreas were separated in a first and second phase secretion. Furthermore, the model was updated to describe 4GIC biomarker dynamics during Oral Glucose Tolerance Test (OGTT) provocation. For these model updates additional literature data were added to the existing 4GI model dataset. The updated 4GIC model was applied to OGTT data from 44 Caucasian participants (twenty-seven controls, ten persons with pre-diabetes, seven patients with type 2 diabetes). All system parameters remained fixed, except for the β-cell function and insulin sensitivity related parameters. More precisely, the insulin secretion rate (ISR), first-phase insulin secretion (FPI), insulin delivery rate (IDR), glucose sensitivity (GS) and insulin sensitivity (IS) were estimated, including inter-individual variability. These population and individual parameters were normalised and compared to a typical control subject.
Results
Individual OGTT profiles of glucose, insulin, C-peptide, GLP-1, glucagon and GIP were adequately described by the model. The individual predicted versus observed area-under-the-curve values of these biomarkers demonstrated good agreement, with root-mean-square percentage error of 2.5, 11.9, 5.0, 16.1, 11.5 and 15.8%, respectively. The 4GIC model enabled simultaneous estimation of β-cell function and insulin sensitivity at both the population and individual levels, with significant correlations with HOMA scores and related parameters estimated using the OMM, including insulin sensitivity and glucose-dependent static and dynamic β-cell responsivity (Φs, Φd) [5] (p < 0.001) [6], [7]. Population and individual IS and β-cell function-related parameters were normalised to a typical control subject, and subsequently visualised and compared using radar plots. Distinct profiles were observed across healthy, pre-diabetic, and diabetic states, supporting mechanistic characterisation of disease progression. A typical pre-diabetic subject has a 32.5 and 46.0 % lower FPI and IS compared to a typical healthy subject, respectively. A typical diabetic patient has a 55.1, 57.1 and 24.3 % lower FPI, IS and GS, and a 59.4 and 117.6 % higher ISR and IDR compared to a typical healthy subject, respectively. Individual radar plots provide a quantitative overview of a subject’s β-cell function and insulin sensitivity status.
Conclusions
The 4GIC model facilitated detailed, personalised quantification of β-cell function and insulin sensitivity and supports the development of a patient-specific digital twin for glucose regulation. Building on prior validation in predicting responses to GLP-1R/GCGR/GIPR agonists and lifestyle changes, future applications could refine diabetes care by integrating individual metabolic profiles into tailored treatment decision algorithms.
References:
[1] Cobelli C et al. Diabetes (2014) 63(4):1203–13.
[2] Bosch R et al., CPT:PSP (2022) 11(3):302–317, doi: 10.1002/psp4.12752
[3] Bosch R et al. Br J Pharmacol (2024) 181(12):1874-1885 doi: 10.1111/bph.16336
[4] Bosch R., et al. CPT:PSP (2025) 14(9): 1515-1525 doi: 10.1002/psp4.70074
[5] Cobelli C Am. J. Physiol. - Endocrinol. Metab. (2007) 293(1): E1–E15, doi: 10.1152/ajpendo.00421.2006.
[6] Jainandunsing S et al. Acta Diabetol., (2015) 52(1):11–19, doi: 10.1007/s00592-014-0588-9.
[7] Geragotou T et al. Journal of Diabetes Research (2016) 6, doi: 10.1155/2016/9286303.
Reference: PAGE 34 (2026) Abstr 12108 [www.page-meeting.org/?abstract=12108]
Poster: Clinical Applications