Next Article in Journal
Prevalence of Diabetes Mellitus and Cardiovascular Risk Factors in El Trébol, Argentina: A Population-Based Cross-Sectional Study
Next Article in Special Issue
Diabetes-Related β-Cell Dysfunction Across COVID-19 and Metabolic Syndrome Is More Closely Associated with Chronic Oxidative Stress than with Transient Hypoxia
Previous Article in Journal
Infrared Thermography in Diabetic Foot Assessment: Review
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Diabetology
Volume 7
Issue 3
10.3390/diabetology7030048
diabetology-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Beta-Cell Function Assessment by In-Silico Modeling Using Three Samples from an Oral Glucose Tolerance Test During Pregnancy Possibly Complicated by Gestational Diabetes

by
Christian Göbl
1,*,
Agnese Piersanti
2,3,
Florian Heinzl
1,
Tina Linder
1,
Micaela Morettini
3 and
Andrea Tura
2,*
1
Department of Obstetrics and Gynaecology, Medical University of Vienna, Währinger Gürtel 18–20, 1090 Vienna, Austria
2
CNR Institute of Neuroscience, Corso Stati Uniti 4, 35127 Padova, Italy
3
Department of Information Engineering, Università Politecnica delle Marche, Via Brecce Bianche 12, 60131 Ancona, Italy
*
Authors to whom correspondence should be addressed.
Diabetology 2026, 7(3), 48; https://doi.org/10.3390/diabetology7030048
Submission received: 10 December 2025 / Revised: 12 January 2026 / Accepted: 26 February 2026 / Published: 3 March 2026
(This article belongs to the Special Issue Beta-Cell Failure and Death: A Cornerstone in Diabetes Pathogenesis)
Browse Figures
Versions Notes

Abstract

Background/Objectives: In pregnancy, beta-cell function is of interest since not only insulin resistance but also beta-cell dysfunction is common, especially when gestational diabetes mellitus (GDM) occurs. Typically, model-based beta-cell function is assessed with (at least) five-sample oral glucose tolerance test (OGTT). The aim of this study was to investigate whether the clinically common three-sample OGTT is sufficient for model-based beta-cell function assessment in pregnancy. Methods: We studied a group of pregnant women undergoing a 2 h five-sample OGTT with glucose, insulin, and C-peptide measurement at early and/or mid-pregnancy, for a total of 152 OGTTs. The five-sample OGTT was used for model-based beta-cell function assessment, yielding three beta-cell function parameters, i.e., glucose sensitivity (GSENS), potentiation factor ratio (PFR), and rate sensitivity (RSENS). GSENS, PFR, and RSENS assessment was repeated with the three-sample OGTT (at 0, 60, 120 min) and related values were compared to those from the five-sample OGTT (reference). Results: We found that, for GSENS, regression and Bland–Altman analyses showed satisfactory results (conditional and marginal R2 values: 0.56 and 0.75, p < 0.0001, and limits of agreement containing 94.2% of samples). Moreover, five-sample and three-sample OGTT GSENS versions were fully consistent in patient subgroup analyses. Results for PFR were less satisfactory but acceptable, whereas those for RSENS were not reliable. Conclusions: The three-sample OGTT is acceptable for model-based beta-cell function assessment in pregnancy, although not for all parameters. Our methodology may be used to explore the effect of time sample reduction in other in-silico models.

Graphical Abstract

1. Introduction

In-silico mathematical models have been developed over the years for the assessment of pancreatic beta-cell function in single individuals from an oral glucose tolerance test (OGTT), typically requiring 4–5 samples of glucose, insulin, and C-peptide [1,2,3,4]. During pregnancy, a 2 h OGTT is recommended for the possible diagnosis of gestational diabetes mellitus (GDM), based on glycemia at fasting, as well as at 1 h and 2 h following the OGTT [5]. This means that in the specific condition of pregnancy, an OGTT with three samples is available in the clinical routine. This is different from other clinical conditions (such as being at risk for type 2 diabetes, T2D), where only the two-sample OGTT, with fasting and 2 h post-load glucose, is usually performed [6].
While the application of in-silico modeling for beta-cell function assessment appears unfeasible with two samples, the question arises as to whether an OGTT with three samples (as performed in pregnancy) may be sufficient for a model-based beta-cell function assessment. The interest in beta-cell function is due to the reason that beta-cell function impairment is often present in GDM (with consequences on maternal, neonatal, and perinatal outcomes) [7,8,9,10,11,12,13,14,15,16,17,18,19,20], and even postpartum in former GDM (this being relevant for later T2D) [21,22,23,24,25,26,27,28,29,30,31,32,33]; thus, insulin resistance (a physiological trait of pregnancy) is not the only defect observed in GDM. On the other hand, the model-based approach has proven advantageous in studying beta-cell function in women with GDM or former GDM because it is able to identify even modest beta-cell function impairment compared to simple data-driven indices (also called “empirical” indices), such as the insulinogenic index [34].
In light of the indicated superiority of the model-based approach compared to data-driven indices, and considering that a three-sample OGTT is routinely available during pregnancy, this study aims to investigate whether such an OGTT may be sufficient for a reliable model-based assessment of beta-cell function in pregnancy. For this purpose, we will compare the model-based findings obtained with three samples to those obtained with a five-sample OGTT as the reference. Specifically, we will focus on the performance of one of the mentioned beta-cell function models [4], which has already been used in previous studies in the context of pregnancy and GDM [34,35,36]. If we succeed in our aim, this will allow application of the model approach to several datasets, either already collected or to be collected in the future. Additionally, another aim of the study is to provide a methodological workflow to explore the effect of sample reduction in other in-silico models that are applicable to dynamic glucose test data.

2. Materials and Methods

2.1. Patients and Measures

We studied 152 OGTTs from 96 pregnant women who were recruited at the pregnancy outpatient department at the Medical University of Vienna. Women after bariatric surgery, with preconceptional (type 1 or type 2) diabetes or other endocrine disorders, HIV, hepatic infection, or malignant tumors were excluded. At early pregnancy (median gestational age of 15.9 weeks), all women underwent assessment of maternal biometry and blood tests at fasting. Women at high risk for GDM received a 2 h 75 g OGTT to assess dynamic parameters of glucose metabolism. After an overnight fast, women ingested a solution containing 75 g of glucose. Venous blood samples were taken at fasting and for 120 min (at 30, 60, 90, and 120 min) for blood glucose, insulin, and C-peptide measurements. At a median gestational age of 26.0 weeks, maternal biometry and blood tests at fasting were repeated, and a 2 h 75 g OGTT was conducted again. Thus, we collected data from 64 OGTTs at early pregnancy and 88 at mid-pregnancy (152 OGTTs in total). Of note, 56 women received both early and mid-pregnancy OGTTs. GDM was diagnosed according to international recommendations [6].
All laboratory parameters were measured according to the standard methods provided by the Clinical Institute for Laboratory Medicine (https://labormedizin.meduniwien.ac.at/, accessed on 9 December 2025). The data analyzed in this study were assessed within clinical investigations, described in previous studies [37,38], performed at the Medical University of Vienna, approved by the local Ethics Committee (approval number: 1637/2014, 1179/2018), and performed in accordance with the Declaration of Helsinki (all patients provided written informed consent).

2.2. Beta-Cell Function Assessment

The in-silico model for beta-cell function assessment that we used in the present analysis was originally described in a study by Mari et al. [4], which reports details such as the expression of the model equations. In our model approach, plasma glucose and C-peptide levels measured during the OGTT are required, whereas plasma insulin level may be useful but not strictly necessary. In such a model, insulin secretion is represented as the sum of two main components, i.e., Sg(t) and Sd(t), where t is the time during the OGTT. The first component describes the dependence of insulin secretion on absolute plasma glucose levels (GLU), and is characterized by a nonlinear dose–response function, f(GLU). The mean value of the dose–response slope is named glucose sensitivity (GSENS) and represents the sensitivity to glucose of the beta cell. The dose response is modulated by a time-varying potentiation factor, P(t); thus, Sg(t) = P(t)·f(GLU). The ratio of the potentiation factor at the end of the OGTT to that at the beginning of the test is named PFR (potentiation factor ratio, which is precisely the ratio of the mean value of the potentiation factor in the 100–120 min interval to that in the 0–20 min interval). The second insulin secretion component, Sd(t), describes the dynamic dependence of insulin secretion on the rate of change in glucose levels, and it is indicated as the derivative component. Sd(t) is proportional to the glucose time derivative (when the glucose derivative is positive, otherwise Sd(t) is null), and the proportionality constant is named rate sensitivity (RSENS). Thus, the model provides three main parameters of beta-cell function: GSENS, PFR, and RSENS. The model was first run using data from the complete (five-sample) OGTT, and hence, the derived GSENS, PFR, and RSENS parameter values were assumed as the reference values for this study. Subsequently, the model was run using data from an OGTT with restricted sampling, which included only the time samples prescribed for the GDM diagnostic OGTT, i.e., 0, 60, and 120 min (three-sample OGTT).

2.3. Data Preparation and Statistical Analysis

Data preparation and statistical analyses were performed in RStudio (version 2023.12.1 Build 402). Different statistical tests were performed to compare beta-cell function as assessed by the five-sample OGTT (the reference) and the three-sample OGTT. For more robust analysis, we first eliminated some unreliable values. For the glucose sensitivity, GSENS, it is known from previous studies that values higher than 300 pmol·min−1·m−2·mM−1 are unreliable [39]. Thus, we excluded cases where the reference GSENS (i.e., obtained with the five-sample OGTT) was higher than the indicated value. Similar criteria were applied to the potentiation factor ratio, PFR, and to the rate sensitivity (to be excluded if higher than 3 (dimensionless) and 10,000·pmol·m−2·mM−1, respectively) [39]. Independent from the above a priori criteria, we also visually analyzed the variable value distribution for both the five-sample and the three-sample OGTT, and excluded those cases that, according to our experience, could be assumed to be outliers in relation to the specific variable distribution of the present study. We then assessed the distribution of the variables using a graphical test of normal distribution and the Shapiro–Wilk test. A square root transformation was performed in the case of skewed distribution. Following the above-described data preparation, we visually analyzed the variable distribution again and excluded cases presenting as outliers. Of note, since we performed these operations independently for each variable of interest, including the elimination of unreliable/unphysiological values or outliers, we ended up with a slightly different number of values for the studied variables. This approach was advisable in order to include the highest number of values possible for each variable in the following analyses. Moreover, it was acceptable since each variable in this study was analyzed separately from the other variables.
After the preliminary operations described above, a comparison between the five-sample and the three-sample OGTT parameters was performed using linear regression, taking into account the repeated measures condition (some patients received both early and mid-pregnancy OGTT). This was accomplished using a linear mixed-effects model (random intercept model, using the lme function in the nlme package, available in the RStudio tool). Subsequently, we performed a Bland–Altman analysis, again considering the presence of repeated measures (with the loa_lme function, available in the SimplyAgree package).
Furthermore, to possibly get improved results, following the calculation of the variables’ mean value in both the five-sample and the three-sample OGTT, we performed individual value adjustments to get the same variable mean over the five-sample and the three-sample OGTT versions. Therefore, we repeated the indicated analyses (linear regression and Bland–Altman).
We also tested the ability of each pair of variables (that is, the five-sample and the three-sample OGTT versions) to provide similar results in terms of the difference (or lack of difference) between patients divided into subgroups. Specifically, patients were divided according to glucose tolerance (GDM or non-GDM) and maternal body mass index (BMI) (overweight/obese with a maternal BMI equal to or above 27 kg/m2 or non-overweight/obese with a maternal BMI less than 27 kg/m2), defined at either the early or mid-pregnancy OGTT. For this purpose, we performed the repeated measures ANOVA test.
Data are presented as mean ± standard deviation, SD. Two-tailed p-values less than 0.05 were considered statistically significant. All results in this study were interpreted in an exploratory manner, aiming to generate new hypotheses.
In the Supplementary Materials, Figure S1 summarizes the workflow of the described data preparation and analysis. The workflow is presented in a generic format, for possible application to other similar problems.

3. Results

3.1. Glucose Sensitivity

The focus of the results is on GSENS since it represents the main model parameters. Some basic information on PFR and RSENS are then briefly reported.
Regarding GSENS, out of the 152 OGTTs, 12 cases were excluded for the indicated a priori criterion (five-sample OGTT GSENS higher than 300 pmol·min−1·m−2·mM−1, as explained above). Afterward, following visual inspection of the distribution of variable values, two OGTTs were excluded for outlier detection in three-sample OGTT GSENS (values above 1000 pmol·min−1·m−2·mM−1). Then, since the distribution of both the five-sample and three-sample OGTT GSENS values was skewed, we used a square root transformation to achieve a normal distribution. After this transformation, we continued to perform a visual inspection of the variables’ distribution and excluded one outlier in the three-sample OGTT GSENS (lower than 5 pmol·min−1·m−2·mM−1 after the square root transformation). Therefore, we ended up with 137 values to be compared between the five-sample and the three-sample OGTT GSENS. The mean value of the square root transformed five-sample OGTT GSENS was 11.30 ± 2.35 pmol·min−1·m−2·mM−1. For the three-sample OGTT GSENS, the mean value was 13.74 ± 4.12 pmol·min−1·m−2·mM−1. Of note, none of the excluded OGTTs represented GDM cases.
Linear regression analysis showed a good association between five-sample and three-sample OGTT GSENS values, with marginal and conditional R2 values equal to 0.560 and 0.752, respectively (p < 0.0001). On the other hand, the three-sample OGTT GSENS tended to be higher than the five-sample OGTT GSENS. The Bland–Altman analysis emphasized the offset in the difference (i.e., bias) between the five-sample OGTT and the three-sample OGTT GSENS (2.452 pmol·min−1·m−2·mM−1), although the limits of agreement contained 93.4% of the data (i.e., only 9 out of 137 samples were outside the limits of agreement).
Based on these findings, which indicated that the five-sample and the three-sample OGTT GSENS values are strongly related but that the three-sample OGTT GSENS overestimates the five-sample OGTT GSENS, we adjusted the former by multiplying by a factor equal to the mean of the five-sample OGTT GSENS divided by the mean of the three-sample OGTT GSENS (this factor was equal to 0.8224). This yielded the same mean value for both variables. Following this adjustment, when repeating the regression analysis, the marginal and conditional R2 value and the p-value remained the same as those reported previously. However, the regression plot indicated that the three-sample OGTT GSENS values were more similar to those of the five-sample case (the regression plot after the adjustment is reported in Figure 1).
This result is even clearer from the Bland–Altman analysis, which shows the improvement in terms of bias (i.e., no more offset between the two variables) and has limits of agreement containing 94.2% of the data (i.e., only 8 out of 137 samples were outside the limits, as visualized in Figure 2), even though a tendency for overestimation from the three-sample OGTT GSENS remained. On the other hand, when calculating the bootstrap-based 95% CIs, we found satisfactory results, as 95% CIs were virtually the same for the five-sample and three-sample cases (10.90–11.69 and 10.78–11.91 pmol·min−1·m−2·mM−1, respectively).
Thereafter, we tested the behavior of the five-sample and the three-sample OGTT GSENS in terms of revealing differences (or lack of differences) between patient subgroups. Regarding stratification according to glucose tolerance in our cohort, we found a prevalence of GDM equal to ~14.5%, that is, 14 cases. When testing the possible difference in GSENS between GDM and non-GDM subgroups, both the five-sample and the three-sample OGTT GSENS consistently revealed significant impairment in the former, with extremely similar p-values (p = 0.0044 and p = 0.0048, respectively). When testing possible differences between overweight/obese or non-overweight/obese subgroups (n = 62 and n = 75, respectively), no significant differences were observed, again yielding consistent results with both the five-sample and three-sample OGTT GSENS (p = 0.49 and p = 0.12, respectively).

3.2. Potentiation Factor Ratio and Rate Sensitivity

For the potentiation factor ratio, PFR, out of the 152 OGTTs, six were excluded for the indicated a priori criterion (five-sample OGTT PFR higher than 3). Following visual inspection of variable distribution, five further OGTTs were excluded as outliers in the three-sample OGTT PFR (values above 2). Then, as was done for GSENS, since both the five-sample and three-sample OGTT PFR values distribution was skewed, a square root transformation was performed. Following further visual inspection of the data, two other outliers in three-sample OGTT PFR were excluded (lower than 0.5). This resulted in 139 values to be compared between the five-sample and the three-sample OGTT PFR. Linear regression and Bland–Altman analysis showed again the need for adjustment. Therefore, we adjusted the three-sample values by again multiplying by a factor equal to the mean of the five-sample OGTT PFR divided by that of the three-sample OGTT PFR (factor equal to 1.3548). Following the adjustment, the regression plot indicated three-sample OGTT PFR values similar to those of the five-sample case (Figure 3), and the Bland–Altman analysis showed the expected improvements in terms of offset (Figure 4).
With regard to the rate sensitivity, it is worth noting that RSENS values equal to zero (or close to zero) are possible because in some OGTTs, the early insulin secretion component is negligible. This typically happens in a relatively small number of cases; however, this was observed in several cases with the three-sample OGTT, likely due to the missing 30 min OGTT sample. Specifically, with the five-sample OGTT, RSENS was almost zero (precisely, lower than 1 pmol·m−2·mM−1) in 18 out of 152 cases, whereas with the three-sample OGTT, this was observed in 93 cases, leading to the conclusion that the three-sample OGTT cannot be used for reliable assessment of RSENS.

3.3. Practical Steps: How to Proceed When Analyzing a New Dataset

Based on the performed analyses and related results, in Figure 5 we summarize the workflow to be performed by investigators analyzing their own three-sample OGTT pregnancy datasets with the beta-cell function modeling approach examined in the present study. The workflow refers to the beta-cell glucose sensitivity, GSENS. However, the workflow can be similarly applied to the potentiation factor ratio, PFR, provided that the recalibration factor is 1.3548, and unreliable values are those above 3.
It is, however, worth noting that the recalibration factors may be different in different study settings. Ideally, one investigator may have his/her own initial dataset with five-sample OGTTs. In this case, the investigator may consider performing the steps reported in the present study to get his/her own (i.e., internal/center-specific) recalibration factors.

4. Discussion

In this study we have analyzed the performance of a known in-silico model for beta-cell function assessment [4] with data derived from the three-sample OGTT in pregnancy. To our knowledge, the analyzed beta-cell function model is currently the model with a more extensive application in the field of pregnancy, as mirrored by several studies [34,35,36].
For the purpose of the study, the results obtained from the model [4] by the three-sample OGTT data were compared to those obtained by a five-sample OGTT, which is traditionally used for a beta-cell function modeling analysis. Of note, the methodological procedure used in this study, composed of several statistical tests applied in both the entire cohort and in derived subgroups, may be useful for future studies to assess the performance of other OGTT-based beta-cell function in-silico models. Examples of such beta-cell function models were previously mentioned [1,2,3]. More generically, our methodological procedure, as summarized in Figure S1, may be applied in other studies for the validation of a parameter or index derived by a new approach against the corresponding parameter/index derived by the reference method. It should also be noted that each clinical center that makes use of the model analyzed in the present study [4] should ideally calculate its own recalibration factors using an internal dataset of five-sample OGTTs.
In addition, it is worth noting that we focused on model performance with the three-sample OGTT in pregnant women, since in this category of patients, the clinical routine already recommends a three-sample OGTT to diagnose GDM [5]. However, the methodological procedure of this study may also be used in future studies on different patient populations. Indeed, the one-hour OGTT sample (i.e., glucose levels obtained at 60 min after oral ingestion) will likely become more relevant as a diagnostic parameter for type 2 diabetes. As a matter of fact, the International Diabetes Federation (IDF) is already recommending the use of a one-hour OGTT for the diagnosis of prediabetes and diabetes [40,41,42]. Thus, the three-sample OGTT may become clinically common not only in pregnant women, but also in several other patient categories at risk for diabetes.
It should be emphasized that in our methodological procedure we have been extremely conservative. Specifically, we excluded from the analysis OGTT cases where the modeling analyses provided results that may have been unphysiological or not reliable for computational reasons. Indeed, this introduces some lack of power in the statistical steps, but it definitely makes the whole methodology more robust, and in our experience, this aspect should be prioritized in a context such as that addressed in this study (specifically, to ensure the reliable prediction of the five-sample GSENS, i.e., the reference, through the three-sample GSENS values). Nonetheless, we explored what would happen in the eventuality that all OGTT cases were retained in the analysis. Interestingly, although the results tend to deteriorate as expected, they were still found to be relatively good, both in terms of the regression analysis and the Bland–Altman analysis, especially regarding the limited number of samples that fell out of the limits of agreement (see Supplementary Materials, Figure S2 and Figure S3, respectively). These findings further demonstrate the appropriateness and reliability of our methodological procedure.
One may wonder why the three-sample OGTT GSENS tended to overestimate the five-sample reference, whereas the PFR showed the opposite behavior. This is likely due to the reason that, in conditions where the assessment of the beta-cell function parameters may be problematic (as with three OGTT samples only), less robust parameters (PFR and RSENS, derived by the potentiation factor and by the early derivative component, respectively) tend to show a smaller effect, assuming small values. Notably, the potentiation factor and the early derivative component represent the potentiation of the “core” insulin secretion (as described by the dose–response function) and the insulin secretion increase in the early OGTT phase, respectively. Both contributions to the total insulin secretion are, however, typically modest as compared to the contribution of the dose–response. Thus, when the information provided to the model is limited (three samples only), the accurate estimation of the potentiation factor and the early derivative component may be problematic. It is therefore possible that the estimation procedure for the model parameters converges to a “cautionary” solution, i.e., a solution that tends to minimize the size of the potentiation factor and derivative component estimates. Thus, the tendency of GSENS to assume bigger values than the “reference” values may be a compensation, and hence a consequence, of the smaller values assumed by the other parameters.
One possible source of inaccuracy in the model approach should be mentioned. Indeed, the model [4], which we have analyzed in the present study (with both five-sample and three-sample OGTT data), relies on some assumptions regarding C-peptide kinetics. Specifically, the model exploits the C-peptide kinetics calculation based on the popular method by Van Cauter et al. [43], and it remains unclear whether that method is completely suitable for pregnant women. In fact, we are not aware of studies similar to that of Van Cauter et al. [43] focusing on pregnant women. That being said, the Van Cauter method is widely used and accepted. Further, the possible effect of some related inaccuracies in the C-peptide kinetics calculation would likely be modest in our modeling approach [4], especially with regard to the model parameter of main interest, i.e., GSENS. A limitation of the study regarding the analysis of possible GSENS differences between subgroups is the small number of GDM women in the analysis. However, despite the small size of the GDM subgroup, GSENS was found, as expected, to be remarkably lower in this group. Additionally, we calculated extremely similar p-values between the five-sample and the three-sample OGTT GSENS analyses. This suggests that the limitation of the GDM subgroup size did not jeopardize the reliability of the subgroup analyses. Furthermore, in the analysis based on overweight/obese and non-overweight/obese subgroups, the number of women in each subgroup was well balanced, and again, we found consistency between the five-sample and the three-sample OGTT GSENS analyses.
Of note, the study of the main glucometabolic parameters in overweight/obese people that includes beta-cell function parameters may be of particular relevance since recent meta-analyses showed an increased risk of developing GDM in this category of pregnant women [44,45]. Similarly, a large meta-analysis also indicated that age is a factor clearly associated with the risk of GDM development [46]. In these contexts, the analysis of the OGTT data through in-silico beta-cell function modeling may be of particular relevance, considering the superiority of the model-based parameters in disclosing even modest beta-cell function defects that may remain undetected by the simple data-driven indices, as previously shown [34].

5. Conclusions

We have shown that in pregnancy, possibly complicated by GDM, a model-based beta-cell function assessment from OGTT with three samples only (0, 60, 120 min) is feasible, at least for the assessment of the most relevant parameter, the beta-cell glucose sensitivity. Additionally, it is worth noting that our methodology may be applied to other in-silico models for analyzing sample reduction effects.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/diabetology7030048/s1, Figure S1: Workflow of the data preparation and analysis. The workflow is presented in generic format, for possible application to other problems where a given parameter or index calculated with a new methodology to be validated needs comparison to the same parameter or index calculated with the reference method; Figure S2: Linear regression plot between 5-samples and 3-samples OGTT square root transformed GSENS values after adjustment for the different mean, in the case of all OGTTs retained in the analysis. Regression lines and unity lines are reported (solid and dashed lines, respectively). Marginal and conditional R2 are 0.470 and 0.564, respectively (p < 0.0001); Figure S3: Bland-Altman plot for 5-samples and 3-samples OGTT square root transformed GSENS values after adjustment for the different mean, in the case of all OGTTs retained in the analysis. After adjustment, offset was negligible, whereas upper and lower limits of agreement were 7.375 and −7.417 pmol∙min−1∙m−2∙mM−1, respectively. Solid lines are offset and limits of agreement.

Author Contributions

Conceptualization, C.G. and A.T.; methodology, C.G., A.P., F.H., M.M. and A.T.; data curation, C.G. and T.L.; writing—original draft preparation, C.G. and A.T.; writing—review and editing, A.P., F.H., T.L. and M.M. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Medical University of Vienna with protocol code 1637/2014 on 28 October 2024 and 1179/2018 on 4 April 2018.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The datasets presented in this article are not readily available because the data are part of an ongoing study. Requests to access the datasets should be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
BMIBody mass index
GDMGestational diabetes mellitus
GSENSGlucose sensitivity
OGTTOral glucose tolerance test
PFRPotentiation factor ratio
RSENSRate sensitivity
SDStandard deviation
T2DType 2 diabetes

References

  1. Hovorka, R.; Chassin, L.; Luzio, S.D.; Playle, R.; Owens, D.R. Pancreatic Beta-Cell Responsiveness during Meal Tolerance Test: Model Assessment in Normal Subjects and Subjects with Newly Diagnosed Noninsulin-Dependent Diabetes Mellitus. J. Clin. Endocrinol. Metab. 1998, 83, 744–750. [Google Scholar] [CrossRef] [Scilit]
  2. Breda, E.; Cavaghan, M.K.; Toffolo, G.; Polonsky, K.S.; Cobelli, C. Oral Glucose Tolerance Test Minimal Model Indexes of Beta-Cell Function and Insulin Sensitivity. Diabetes 2001, 50, 150–158. [Google Scholar] [CrossRef] [Scilit]
  3. Cretti, A.; Lehtovirta, M.; Bonora, E.; Brunato, B.; Zenti, M.G.; Tosi, F.; Caputo, M.; Caruso, B.; Groop, L.C.; Muggeo, M.; et al. Assessment of Beta-Cell Function during the Oral Glucose Tolerance Test by a Minimal Model of Insulin Secretion. Eur. J. Clin. Investig. 2001, 31, 405–416. [Google Scholar] [CrossRef] [Scilit]
  4. Mari, A.; Tura, A.; Gastaldelli, A.; Ferrannini, E. Assessing Insulin Secretion by Modeling in Multiple-Meal Tests: Role of Potentiation. Diabetes 2002, 51, S221–S226. [Google Scholar] [CrossRef] [Scilit]
  5. Metzger, B.E.; Gabbe, S.G.; Persson, B.; Lowe, L.P.; Dyer, A.R.; Oats, J.J.; Buchanan, T.A. International Association of Diabetes and Pregnancy Study Groups Recommendations on the Diagnosis and Classification of Hyperglycemia in Pregnancy. Diabetes Care 2010, 33, 676–682. [Google Scholar] [CrossRef] [Scilit]
  6. American Diabetes Association Professional Practice Committee 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2024. Diabetes Care 2023, 47, S20–S42. [Google Scholar] [CrossRef] [Scilit]
  7. Mittendorfer, B.; Patterson, B.W.; Haire-Joshu, D.; Cahill, A.G.; Cade, W.T.; Stein, R.I.; Klein, S. Insulin Sensitivity and β-Cell Function During Early and Late Pregnancy in Women with and Without Gestational Diabetes Mellitus. Diabetes Care 2023, 46, 2147–2154. [Google Scholar] [CrossRef] [Scilit]
  8. Retnakaran, R.; Ye, C.; Kramer, C.K.; Hanley, A.J.; Connelly, P.W.; Sermer, M.; Zinman, B. Deteriorating Beta Cell Function Is the Dominant Determinant of Progression from Normal Glucose Tolerance to Prediabetes/Diabetes in Young Women Following Pregnancy. Diabetologia 2023, 66, 2154–2163. [Google Scholar] [CrossRef] [Scilit]
  9. Gao, M.; Wang, H.; Li, W.; Wang, L.; Li, N.; Qiao, Y.; Zhang, T.; Li, J.; Yu, Z.; Hu, G.; et al. Maternal Insulin Resistance and Maternal β-Cell Function during Pregnancy for Offspring Overweight before 2 Years of Age among Women with Gestational Diabetes. Pediatr. Obes. 2023, 18, e12995. [Google Scholar] [CrossRef] [Scilit]
  10. Ellerbrock, J.; Spaanderman, B.; van Drongelen, J.; Mulder, E.; Lopes van Balen, V.; Schiffer, V.; Jorissen, L.; Alers, R.-J.; Leenen, J.; Ghossein-Doha, C.; et al. Role of Beta Cell Function and Insulin Resistance in the Development of Gestational Diabetes Mellitus. Nutrients 2022, 14, 2444. [Google Scholar] [CrossRef] [Scilit]
  11. Dong, Y.; Luo, Z.-C.; Nuyt, A.M.; Audibert, F.; Wei, S.-Q.; Abenhaim, H.A.; Bujold, E.; Julien, P.; Huang, H.; Levy, E.; et al. Large-for-Gestational-Age May Be Associated with Lower Fetal Insulin Sensitivity and β-Cell Function Linked to Leptin. J. Clin. Endocrinol. Metab. 2018, 103, 3837–3844. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  12. Liu, Y.; Hou, W.; Meng, X.; Zhao, W.; Pan, J.; Tang, J.; Huang, Y.; Tao, M.; Liu, F. Heterogeneity of Insulin Resistance and Beta Cell Dysfunction in Gestational Diabetes Mellitus: A Prospective Cohort Study of Perinatal Outcomes. J. Transl. Med. 2018, 16, 289. [Google Scholar] [CrossRef] [Scilit]
  13. Ellerbrock, J.; Bohnen, J.M.H.A.; Lopes van Balen, V.A.; Mulder, E.G.; Aardenburg, R.; Spaanderman, M.E.A. Homeostatic Model Assessment of Beta Cell Function Predicting Abnormal Oral Glucose Tolerance Testing in Pregnancy: A Systematic Review and Meta-Analysis. Gynecol. Endocrinol. 2017, 33, 911–917. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  14. Wang, Y.-H.; Wu, H.-H.; Ding, H.; Li, Y.; Wang, Z.-H.; Li, F.; Zhang, J.-P. Changes of Insulin Resistance and β-Cell Function in Women with Gestational Diabetes Mellitus and Normal Pregnant Women during Mid- and Late Pregnant Period: A Case-Control Study. J. Obstet. Gynaecol. Res. 2013, 39, 647–652. [Google Scholar] [CrossRef] [Scilit]
  15. Miyakoshi, K.; Tanaka, M.; Saisho, Y.; Shimada, A.; Minegishi, K.; Kim, S.H.; Asai, S.; Itoh, H.; Yoshimura, Y. Pancreatic Beta-Cell Function and Fetal Growth in Gestational Impaired Glucose Tolerance. Acta Obstet. Gynecol. Scand. 2010, 89, 769–775. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  16. Retnakaran, R.; Qi, Y.; Connelly, P.W.; Sermer, M.; Hanley, A.J.; Zinman, B. Low Adiponectin Concentration during Pregnancy Predicts Postpartum Insulin Resistance, Beta Cell Dysfunction and Fasting Glycaemia. Diabetologia 2010, 53, 268–276. [Google Scholar] [CrossRef] [Scilit]
  17. Qvigstad, E.; Voldner, N.; Godang, K.; Henriksen, T.; Bollerslev, J. Overweight Is Associated with Impaired Beta-Cell Function during Pregnancy: A Longitudinal Study of 553 Normal Pregnancies. Eur. J. Endocrinol. 2010, 162, 67–73. [Google Scholar] [CrossRef] [Scilit]
  18. Lapolla, A.; Dalfrà, M.G.; Mello, G.; Parretti, E.; Cioni, R.; Marzari, C.; Masin, M.; Ognibene, A.; Messeri, G.; Fedele, D.; et al. Early Detection of Insulin Sensitivity and Beta-Cell Function with Simple Tests Indicates Future Derangements in Late Pregnancy. J. Clin. Endocrinol. Metab. 2008, 93, 876–880. [Google Scholar] [CrossRef] [Scilit]
  19. Lencioni, C.; Volpe, L.; Miccoli, R.; Cuccuru, I.; Chatzianagnostou, K.; Ghio, A.; Benzi, L.; Bonadonna, R.C.; Del Prato, S.; Di Cianni, G. Early Impairment of Beta-Cell Function and Insulin Sensitivity Characterizes Normotolerant Caucasian Women with Previous Gestational Diabetes. Nutr. Metab. Cardiovasc. Dis. 2006, 16, 485–493. [Google Scholar] [CrossRef] [Scilit]
  20. Catalano, P.M.; Drago, N.M.; Amini, S.B. Longitudinal Changes in Pancreatic Beta-Cell Function and Metabolic Clearance Rate of Insulin in Pregnant Women with Normal and Abnormal Glucose Tolerance. Diabetes Care 1998, 21, 403–408. [Google Scholar] [CrossRef] [Scilit]
  21. Vounzoulaki, E.; Khunti, K.; Abner, S.C.; Tan, B.K.; Davies, M.J.; Gillies, C.L. Progression to Type 2 Diabetes in Women with a Known History of Gestational Diabetes: Systematic Review and Meta-Analysis. BMJ 2020, 369, m1361. [Google Scholar] [CrossRef] [Scilit]
  22. Rayanagoudar, G.; Hashi, A.A.; Zamora, J.; Khan, K.S.; Hitman, G.A.; Thangaratinam, S. Quantification of the Type 2 Diabetes Risk in Women with Gestational Diabetes: A Systematic Review and Meta-Analysis of 95,750 Women. Diabetologia 2016, 59, 1403–1411. [Google Scholar] [CrossRef] [Scilit]
  23. Ratner, R.E. Prevention of Type 2 Diabetes in Women with Previous Gestational Diabetes. Diabetes Care 2007, 30, S242–S245. [Google Scholar] [CrossRef] [Scilit]
  24. Kim, C.; Newton, K.M.; Knopp, R.H. Gestational Diabetes and the Incidence of Type 2 Diabetes: A Systematic Review. Diabetes Care 2002, 25, 1862–1868. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  25. England, L.J.; Dietz, P.M.; Njoroge, T.; Callaghan, W.M.; Bruce, C.; Buus, R.M.; Williamson, D.F. Preventing Type 2 Diabetes: Public Health Implications for Women with a History of Gestational Diabetes Mellitus. Am. J. Obstet. Gynecol. 2009, 200, 365.e1–365.e8. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  26. Dennison, R.A.; Chen, E.S.; Green, M.E.; Legard, C.; Kotecha, D.; Farmer, G.; Sharp, S.J.; Ward, R.J.; Usher-Smith, J.A.; Griffin, S.J. The Absolute and Relative Risk of Type 2 Diabetes after Gestational Diabetes: A Systematic Review and Meta-Analysis of 129 Studies. Diabetes Res. Clin. Pract. 2021, 171, 108625. [Google Scholar] [CrossRef] [Scilit]
  27. Buchanan, T.A. Pancreatic B-Cell Defects in Gestational Diabetes: Implications for the Pathogenesis and Prevention of Type 2 Diabetes. J. Clin. Endocrinol. Metab. 2001, 86, 989–993. [Google Scholar] [CrossRef]
  28. Bentley-Lewis, R.; Levkoff, S.; Stuebe, A.; Seely, E.W. Gestational Diabetes Mellitus: Postpartum Opportunities for the Diagnosis and Prevention of Type 2 Diabetes Mellitus. Nat. Clin. Pract. Endocrinol. Metab. 2008, 4, 552–558. [Google Scholar] [CrossRef] [Scilit]
  29. Bellamy, L.; Casas, J.-P.; Hingorani, A.D.; Williams, D. Type 2 Diabetes Mellitus after Gestational Diabetes: A Systematic Review and Meta-Analysis. Lancet 2009, 373, 1773–1779. [Google Scholar] [CrossRef] [Scilit]
  30. Baptiste-Roberts, K.; Barone, B.B.; Gary, T.L.; Golden, S.H.; Wilson, L.M.; Bass, E.B.; Nicholson, W.K. Risk Factors for Type 2 Diabetes among Women with Gestational Diabetes: A Systematic Review. Am. J. Med. 2009, 122, 207–214.e4. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  31. Khan, S.R.; Mohan, H.; Liu, Y.; Batchuluun, B.; Gohil, H.; Al Rijjal, D.; Manialawy, Y.; Cox, B.J.; Gunderson, E.P.; Wheeler, M.B. The Discovery of Novel Predictive Biomarkers and Early-Stage Pathophysiology for the Transition from Gestational Diabetes to Type 2 Diabetes. Diabetologia 2019, 62, 687–703. [Google Scholar] [CrossRef] [Scilit]
  32. Huopio, H.; Cederberg, H.; Vangipurapu, J.; Hakkarainen, H.; Pääkkönen, M.; Kuulasmaa, T.; Heinonen, S.; Laakso, M. Association of Risk Variants for Type 2 Diabetes and Hyperglycemia with Gestational Diabetes. Eur. J. Endocrinol. 2013, 169, 291–297. [Google Scholar] [CrossRef] [Scilit]
  33. Batchuluun, B.; Al Rijjal, D.; Prentice, K.J.; Eversley, J.A.; Burdett, E.; Mohan, H.; Bhattacharjee, A.; Gunderson, E.P.; Liu, Y.; Wheeler, M.B. Elevated Medium-Chain Acylcarnitines Are Associated with Gestational Diabetes Mellitus and Early Progression to Type 2 Diabetes and Induce Pancreatic β-Cell Dysfunction. Diabetes 2018, 67, 885–897. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  34. Tura, A.; Mari, A.; Winzer, C.; Kautzky-Willer, A.; Pacini, G. Impaired Beta-Cell Function in Lean Normotolerant Former Gestational Diabetic Women. Eur. J. Clin. Investig. 2006, 36, 22–28. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  35. Stopp, T.; Feichtinger, M.; Rosicky, I.; Yerlikaya-Schatten, G.; Ott, J.; Egarter, H.C.; Schatten, C.; Eppel, W.; Husslein, P.; Mittlböck, M.; et al. Novel Indices of Glucose Homeostasis Derived from Principal Component Analysis: Application for Metabolic Assessment in Pregnancy. J. Diabetes Res. 2020, 2020, 4950584. [Google Scholar] [CrossRef] [Scilit]
  36. Tura, A.; Mari, A.; Prikoszovich, T.; Pacini, G.; Kautzky-Willer, A. Value of the Intravenous and Oral Glucose Tolerance Tests for Detecting Subtle Impairments in Insulin Sensitivity and Beta-Cell Function in Former Gestational Diabetes. Clin. Endocrinol. 2008, 69, 237–243. [Google Scholar] [CrossRef] [Scilit]
  37. Salvatori, B.; Linder, T.; Eppel, D.; Morettini, M.; Burattini, L.; Göbl, C.; Tura, A. TyGIS: Improved Triglyceride-Glucose Index for the Assessment of Insulin Sensitivity during Pregnancy. Cardiovasc. Diabetol. 2022, 21, 215. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  38. Eppel, D.; Feichtinger, M.; Lindner, T.; Kotzaeridi, G.; Rosicky, I.; Yerlikaya-Schatten, G.; Eppel, W.; Husslein, P.; Tura, A.; Göbl, C.S. Association between Maternal Triglycerides and Disturbed Glucose Metabolism in Pregnancy. Acta Diabetol. 2021, 58, 459–465. [Google Scholar] [CrossRef] [Scilit]
  39. Tura, A.; Grespan, E.; Göbl, C.S.; Koivula, R.W.; Franks, P.W.; Pearson, E.R.; Walker, M.; Forgie, I.M.; Giordano, G.N.; Pavo, I.; et al. Profiles of Glucose Metabolism in Different Prediabetes Phenotypes, Classified by Fasting Glycemia, 2-Hour OGTT, Glycated Hemoglobin, and 1-Hour OGTT: An IMI DIRECT Study. Diabetes 2021, 70, 2092–2106. [Google Scholar] [CrossRef] [Scilit]
  40. International Diabetes Federation Position Statement Recommending Use of 1-Hour Plasma Glucose Test to Diagnose Intermediate Hyperglycaemia and Type 2 Diabetes. Diabetes Res. Clin. Pract. 2024, 210, 111646. [CrossRef] [Scilit]
  41. Bergman, M.; Tuomilehto, J. International Diabetes Federation Position Statement on the 1-Hour Post-Load Plasma Glucose for the Diagnosis of Intermediate Hyperglycaemia and Type 2 Diabetes. Diabetes Res. Clin. Pract. 2024, 210, 111636. [Google Scholar] [CrossRef] [Scilit]
  42. Bergman, M.; Manco, M.; Satman, I.; Chan, J.; Inês Schmidt, M.; Sesti, G.; Vanessa Fiorentino, T.; Abdul-Ghani, M.; Jagannathan, R.; Kumar Thyparambil Aravindakshan, P.; et al. International Diabetes Federation Position Statement on the 1-Hour Post-Load Plasma Glucose for the Diagnosis of Intermediate Hyperglycaemia and Type 2 Diabetes. Diabetes Res. Clin. Pract. 2024, 209, 111589. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  43. Cauter, E.V.; Mestrez, F.; Sturis, J.; Polonsky, K.S. Estimation of Insulin Secretion Rates from C-Peptide Levels: Comparison of Individual and Standard Kinetic Parameters for C-Peptide Clearance. Diabetes 1992, 41, 368–377. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  44. Alwash, S.M.; McIntyre, H.D.; Mamun, A. The Association of General Obesity, Central Obesity and Visceral Body Fat with the Risk of Gestational Diabetes Mellitus: Evidence from a Systematic Review and Meta-Analysis. Obes. Res. Clin. Pract. 2021, 15, 425–430. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  45. Chu, S.Y.; Callaghan, W.M.; Kim, S.Y.; Schmid, C.H.; Lau, J.; England, L.J.; Dietz, P.M. Maternal Obesity and Risk of Gestational Diabetes Mellitus. Diabetes Care 2007, 30, 2070–2076. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  46. Li, Y.; Ren, X.; He, L.; Li, J.; Zhang, S.; Chen, W. Maternal Age and the Risk of Gestational Diabetes Mellitus: A Systematic Review and Meta-Analysis of over 120 Million Participants. Diabetes Res. Clin. Pract. 2020, 162, 108044. [Google Scholar] [CrossRef] [Scilit]
Diabetology 07 00048 g001
Figure 1. Linear regression plot between five-sample and three-sample OGTT square root transformed GSENS values after adjustment for the different means. Regression lines and unity lines are reported (solid and dashed lines, respectively). Marginal and conditional R2 values are 0.560 and 0.752, respectively (p < 0.0001).
Figure 1. Linear regression plot between five-sample and three-sample OGTT square root transformed GSENS values after adjustment for the different means. Regression lines and unity lines are reported (solid and dashed lines, respectively). Marginal and conditional R2 values are 0.560 and 0.752, respectively (p < 0.0001).
Diabetology 07 00048 g001
Diabetology 07 00048 g002
Figure 2. Bland–Altman plot for five-sample and three-sample OGTT square root transformed GSENS values after adjustment for the different means. After adjustment, offset was negligible, whereas upper and lower limits of agreement were 4.432 and −4.458 pmol·min−1·m−2·mM−1, respectively. Solid lines are offset and limits of agreement.
Figure 2. Bland–Altman plot for five-sample and three-sample OGTT square root transformed GSENS values after adjustment for the different means. After adjustment, offset was negligible, whereas upper and lower limits of agreement were 4.432 and −4.458 pmol·min−1·m−2·mM−1, respectively. Solid lines are offset and limits of agreement.
Diabetology 07 00048 g002
Diabetology 07 00048 g003
Figure 3. Linear regression plot between five-sample and three-sample OGTT square root transformed PFR values after adjustment for the different means. Regression lines and unity lines are reported (solid and dashed lines, respectively). Marginal and conditional R2 values are 0.344 and 0.367, respectively (p < 0.0001).
Figure 3. Linear regression plot between five-sample and three-sample OGTT square root transformed PFR values after adjustment for the different means. Regression lines and unity lines are reported (solid and dashed lines, respectively). Marginal and conditional R2 values are 0.344 and 0.367, respectively (p < 0.0001).
Diabetology 07 00048 g003
Diabetology 07 00048 g004
Figure 4. Bland–Altman plot for five-sample and three-sample OGTT square root transformed PFR values after adjustment for the different means. After adjustment, offset was negligible, whereas upper and lower limits of agreement were 0.339 and −0.339, respectively. Solid lines are offset and limits of agreement.
Figure 4. Bland–Altman plot for five-sample and three-sample OGTT square root transformed PFR values after adjustment for the different means. After adjustment, offset was negligible, whereas upper and lower limits of agreement were 0.339 and −0.339, respectively. Solid lines are offset and limits of agreement.
Diabetology 07 00048 g004
Diabetology 07 00048 g005
Figure 5. Workflow for the calculation of beta-cell glucose sensitivity, GSENS, from three-sample OGTT pregnancy datasets.
Figure 5. Workflow for the calculation of beta-cell glucose sensitivity, GSENS, from three-sample OGTT pregnancy datasets.
Diabetology 07 00048 g005
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Göbl, C.; Piersanti, A.; Heinzl, F.; Linder, T.; Morettini, M.; Tura, A. Beta-Cell Function Assessment by In-Silico Modeling Using Three Samples from an Oral Glucose Tolerance Test During Pregnancy Possibly Complicated by Gestational Diabetes. Diabetology 2026, 7, 48. https://doi.org/10.3390/diabetology7030048

AMA Style

Göbl C, Piersanti A, Heinzl F, Linder T, Morettini M, Tura A. Beta-Cell Function Assessment by In-Silico Modeling Using Three Samples from an Oral Glucose Tolerance Test During Pregnancy Possibly Complicated by Gestational Diabetes. Diabetology. 2026; 7(3):48. https://doi.org/10.3390/diabetology7030048

Chicago/Turabian Style

Göbl, Christian, Agnese Piersanti, Florian Heinzl, Tina Linder, Micaela Morettini, and Andrea Tura. 2026. "Beta-Cell Function Assessment by In-Silico Modeling Using Three Samples from an Oral Glucose Tolerance Test During Pregnancy Possibly Complicated by Gestational Diabetes" Diabetology 7, no. 3: 48. https://doi.org/10.3390/diabetology7030048

APA Style

Göbl, C., Piersanti, A., Heinzl, F., Linder, T., Morettini, M., & Tura, A. (2026). Beta-Cell Function Assessment by In-Silico Modeling Using Three Samples from an Oral Glucose Tolerance Test During Pregnancy Possibly Complicated by Gestational Diabetes. Diabetology, 7(3), 48. https://doi.org/10.3390/diabetology7030048

Article Metrics

Back to TopTop