One-Hour Postload Glucose Is a More Sensitive Marker of Impaired β-Cell Function Than Two-Hour Postload Glucose

The oral glucose tolerance test (OGTT) is widely used in clinical practice to screen for diabetes and prediabetes, the latter being defined as impaired fasting glucose or impaired glucose tolerance (IGT). Although OGTT is more costly and time-consuming than the testing of fasting plasma glucose (PG) levels and glycated hemoglobin A_1c (HbA_1c), not performing an OGTT would lead to underdiagnosis of both diabetes and prediabetes (1). On the other hand, it is well known that individuals with IGT have an increased risk for development of type 2 diabetes, and the efficacy of interventions for the prevention of type 2 diabetes has predominantly been demonstrated in individuals with IGT (2–4). Of note, longitudinal studies have shown that many people with IGT do not progress to type 2 diabetes, whereas a substantial number of patients with type 2 diabetes have normal glucose tolerance (NGT) at baseline (5,6).

Traditionally, a patient’s 2-h PG concentration (hereafter, 2hPG) during an OGTT is used to diagnose IGT or diabetes. Nevertheless, there is ample evidence that 1-h PG level (hereafter, 1hPG) is predictive of future development of type 2 diabetes and related health outcomes, with equal or even superior performance to 2hPG (7–11). Therefore, a panel of international experts advocated that 1hPG may potentially identify individuals at high risk of type 2 diabetes at an earlier stage than the current definition of IGT (12). Regarding the mechanisms behind the alteration of 1hPG, previous studies have focused mainly on the investigation of insulin secretion and action in the context of NGT (i.e., NGT with a high 1hPG vs. NGT with a normal 1hPG) (13–16). However, none have directly compared the pathophysiologic features between 1hPG and 2hPG. Therefore, in an attempt to provide more insights into the role of 1hPG, compared with 2hPG, in the assessment of future diabetes risk, we examined the measures of insulin sensitivity and β-cell function among participants categorized by 1hPG and 2hPG during OGTT who were matched by key clinical factors and selected from a large Chinese population with a high risk of type 2 diabetes.

Individuals who underwent a 75-g OGTT in the outpatient clinic of the Department of Endocrinology and Metabolism of Shanghai Jiao Tong University Affiliated Sixth People's Hospital between January 2011 and January 2020 were included in the study. They were generally referred to the clinic because they had a high risk of type 2 diabetes development, including family history of diabetes, prior recognition of elevated fasting PG levels and/or HbA_1c, history of gestational diabetes, obesity, hypertension, and hyperlipidemia. All participants had no history of diabetes and no history of use of hypoglycemic agents. We excluded those who met the diagnostic criteria for diabetes according to the 2010 American Diabetes Association criteria (17), were younger than 18 years old, or had missing or invalid data relevant for analysis (Supplementary Fig. 1).

This study was approved by the Ethics Committee of Shanghai Jiao Tong University Affiliated Sixth People’s Hospital and was conducted in accordance with either the 1964 Helsinki declaration and its later amendments or with comparable ethical standards. Informed consent was obtained from all participants included in the study.

Based on the results of OGTT results, participants free of diabetes were categorized into four categories according to 1hPG and 2hPG: 1) 1hPG <8.6 mmol/L and 2hPG <7.8 mmol/L (NGT/1h-normal); 2) 1hPG ≥8.6 mmol/L and 2hPG <7.8 mmol/L (NGT/1h-high); 3) 1hPG <8.6 mmol/L and 2hPG ≥7.8 mmol/L (IGT/1h-normal); and 4) 1hPG ≥8.6 mmol/L and 2hPG ≥7.8 mmol/L (IGT/1h-high).

All participants received a comprehensive physical examination, including measurement of height, body weight, and blood pressure. BMI was calculated as weight (in kilograms) divided by height (in meters) squared. Data on family history of diabetes were obtained by self-report. A 75-g OGTT was performed for each participant after an overnight fast for at least 10 h. Blood samples were then collected at 0, 30, 60, 120, and 180 min.

PG levels were obtained by the glucose oxidase method (Kehua Biological Engineering Co., Ltd., Shanghai, China) using the Glamour 2000 biochemical autoanalyzer. Electrochemiluminescence immunoassay was used to quantify the serum insulin levels on Cobas e 411 and e 601 analyzers (Roche Diagnostics GmbH, Mannheim, Germany) with inter- and intra-assay coefficients of variance of 2.50% and 1.70%, respectively. HbA_1c was detected by high-pressure liquid chromatography (Variant II hemoglobin analyzer; Bio-Rad, Hercules, CA) with inter- and intra-assay coefficients of variance of 0.75–3.39% and 0.55–2.58%, respectively.

Insulin sensitivity indexes were calculated from OGTT using the Matsuda (18), HOMA of insulin resistance (HOMA-IR) (19), and quantitative insulin sensitivity check index (QUICKI) (20) indexes. As surrogate measures of insulin secretion, the insulinogenic index (IGI) (8) was calculated as follows: (insulin [30 min] − insulin [fasting])/(glucose [30 min] − glucose [fasting]), and the insulin response during 0–120 min of OGTT was estimated by the incremental area under the curve (AUC) of insulin (ΔI[AUC]) divided by the incremental AUC of glucose (ΔG[AUC]) using the trapezoid rule. The HOMA of insulin secretion (HOMA-B) was also calculated. The product of insulin secretion and insulin sensitivity, termed the disposition index (DI) (8), was calculated for 0–30 min (IGI × Matsuda index; DI₃₀) and 0–120 min (ΔI[AUC]0–120 min/ΔG[AUC]0–120 min × Matsuda index; DI₁₂₀) during OGTT, respectively. The 180-min glucose was not used in the calculation of measures of insulin sensitivity and insulin secretion.

Statistical analyses were performed with SPSS, version 26.0 (SPSS Inc., Chicago, IL) and R, version 4.2.2 (RStudio Inc., Boston, MA). Variables are presented as mean ± SD, median (interquartile range [IQR]), or number (%), as appropriate. The correlations between 1hPG and 2hPG and indices of insulin sensitivity and secretion were evaluated by Spearman’s rank correlation coefficients. Linear mixed-effects models were used to investigate the relationships between 1hPG and 2hPG and indices of insulin sensitivity and secretion, with these being data log-transformed due to their skewed distributions. To compare the slopes relating the specific index of insulin sensitivity and/or secretion with the PG concentration between OGTT 1hPG and OGTT 2hPG, we calculated P values for the interaction terms index of insulin sensitivity or secretion × time point.

The data sets for this study are available upon reasonable request of the corresponding author.

A total of 3,965 people without diabetes who underwent the OGTT were eligible for this analysis (Supplementary Fig. 1). Based on the OGTT results, 2,249 people (56.7%) were classified as NGT, and 1,716 (43.3%) as IGT (Supplementary Fig. 2). In the NGT group, elevated 1hPG was observed in 1,069 individuals (47.5%), whereas the majority of participants with IGT had elevated 1hPG (n = 1,590; 92.7%). By contrast, of participants with elevated 1hPG, 59.8% were classified as IGT.

Next, the four groups stratified by 1hPG and 2hPG were 1:1:1:1 matched by age, sex, and BMI, resulting in a study population of 496 individuals (n = 124 for each group) (Table 1). The glucose and insulin concentrations of the four groups during the OGTT are shown in Fig. 1. In the NGT/1h-normal group, insulin levels peaked at 30 min; in the NGT/1h-high group, insulin levels peaked at 60 min; and in both the IGT/1h-normal and IGT/1h-high groups, insulin levels peaked at 120 min.

The baseline characteristics of the matched population are depicted in Table 1. The median (IQR) age was 55.0 (43.8, 63.0) years, and 42.3% were men. The median (IQR) BMI was 23.4 (21.5, 26.0) kg/m². As expected, age, sex, and BMI were comparable among the four groups, and no significant difference was observed in either systolic or diastolic blood pressure. Family history of diabetes was less prevalent in the NGT/1h-normal group than in other groups. The NGT/1h-normal group had the lowest value of HbA_1c (P < 0.05 for all, compared with other groups), whereas there were no significant differences among the other three groups.

All indices of insulin sensitivity, including HOMA-IR, QUICKI, and Matsuda, differed significantly among the four groups (P < 0.05 for all). However, there were no significant differences in any of the three indices of insulin sensitivity when comparing the IGT/1h-normal group with the NGT/1h-high group (P < 0.05 for all). Regarding the measures of insulin secretion, IGI and ΔI[AUC]_0–120/ΔG[AUC]_0–120, but not HOMA-B, differed significantly among the four groups (P < 0.05 for all). Specifically, compared with the IGT/1h-normal group, the NGT/1h-high group had significantly lower measurements of IGI (0.6 [0.4, 1.1] vs. 1.1 [0.7, 1.8]; P < 0.05 after Bonferroni correction). The ΔI[AUC]_0–120/ΔG[AUC]_0–120 values in the NGT/1h-high group were numerically lower than in the IGT/1h-normal group (19.7 [12.7, 29.8] vs. 22.0 [14.3, 29.9]; P > 0.05 after Bonferroni correction). To account for the influence of insulin resistance on insulin secretion, we compared the indices of DI (i.e., DI₃₀ and DI₁₂₀) among the four groups. The NGT/1h-normal and the IGT/1h-high groups had the highest and lowest DI values, respectively (Table 1). Of note, both DI₃₀ (2.5 [1.9, 3.5] vs. 4.3 [3.2, 6.0]; P < 0.05 after Bonferroni correction) and DI₁₂₀ (70.3 [56.4, 88.3] vs. 82.5 [65.7, 102.7]; P < 0.05 after Bonferroni correction) were significantly reduced in the NGT/1h-high group compared with the IGT/1h-normal group.

The 1hPG was significantly correlated with all indices of insulin sensitivity and secretion examined (Table 2). Generally, a stronger correlation was noted for measures of insulin secretion than for insulin sensitivity. A similar pattern of correlation was found for 2hPG, although the strength of correlations with indices of insulin secretion seemed to be weaker compared with 1hPG. Figure 2 illustrates that the slopes of fitted regression lines relating glucose levels to log-transformed HOMA-IR and QUICKI were similar between 1hPG and 2hPG. However, the slopes of fitted regression lines were significantly steeper (P < 0.05 for interaction for all) for 1hPG compared with 2hPG in relation to log-transformed indices of Matsuda, HOMA-B, insulinogenic index, ΔI[AUC]_0–120/ΔG[AUC]_0–120, DI₃₀, and DI₁₂₀, with the most apparent differences observed in DI₃₀.

We next separated the study sample according to age (<40 vs. ≥40 years), sex (male vs. female), BMI (<24 vs. ≥24 kg/m²), and family history of diabetes (yes vs. no). The slopes of fitted regression lines relating glucose levels to log-transformed DI₃₀ and DI₁₂₀ were significantly steeper for 1hPG than 2hPG in each subgroup (P < 0.05 for interaction; Supplementary Figs. 3–6), consistent with the main findings.

Among 496 participants without diabetes stratified according to 1- and 2-h OGTTs, who were matched by age, sex, and BMI and selected from a large study population at high risk of type 2 diabetes, our study demonstrated that those with elevated 1hPG but normal 2hPG were characterized by significantly lower β-cell function, especially early-phase insulin secretion, compared with their counterparts with normal 1hPG but elevated 2hPG levels. Importantly, compared with 2hPG, 1hPG was more sensitive to the deteriorations in β-cell function, suggesting that it possibly could be used to identify individuals at high risk of type 2 diabetes at an earlier stage.

During the past two decades, there has been ample evidence from longitudinal studies across multiple ethnic groups that the 1-h OGTT glucose level is closely related to future risk of type 2 diabetes (7, 14, 21–23). Moreover, in some studies, the 1-h glucose level has been demonstrated to be superior to the 2-h glucose level in predicting the development of type 2 diabetes. For instance, in 1,551 participants without diabetes in the San Antonio Heart Study, Abdul-Ghani et al. (24) found that 1hPG had a greater discriminatory power than 2hPG for incident type 2 diabetes after 7–8 years of follow-up. Additionally, numerous prospective studies indicated that 1hPG was significantly linked to microvascular (10, 11) and macrovascular complications (11), and death (9, 11). Given the performance of 1hPG in predicting type 2 diabetes and related outcomes, and the practicability, a panel of international experts advocated that 1hPG, instead of 2hPG, should be determined in routine OGTT (12).

However, only few have investigated the pathophysiologic features of elevated 1hPG, especially in comparison with 2hPG. In 1,205 healthy volunteers from the Relationship between Insulin Sensitivity and Cardiovascular Risk (RISC) study, participants with NGT and 1hPG >8.95 mmol/L had lower insulin sensitivity and higher insulin secretion compared with those with 1hPG ≤8.95 mmol/L. Meanwhile, participants with NGT and 1hPG >8.95 mmol/L had lower β-cell glucose sensitivity, β-cell rate sensitivity, and potentiation factor (14). Reduced insulin sensitivity in participants with NGT but elevated 1hPG (≥8.6 mmol/L), compared with individuals with normal 1hPG, was confirmed in another study comprising 305 offspring without diabetes of individuals with type 2 diabetes (15). Although the OGTT-derived estimate of insulin secretion did not differ between the two groups, the acute insulin response during the intravenous glucose tolerance test and the related DI were significantly reduced in the former group. In a subsequent study, conducted with 926 participants at high risk of type 2 diabetes, participants with NGT and a high 1hPG (>8.6 mmol/L) were observed to be more insulin resistant and have worse β-cell function and β-cell sensitivity than those with a normal 1hPG (16). It is noteworthy that although all these studies compared the pathophysiological characteristics between NGT/1h-high and IGT, a large proportion of participants with IGT also had elevated 1hPG levels [e.g., 90% in IGT in the study by Bianchi et al. (16)]. Therefore, it is not surprising that almost no significant differences in measurements of insulin sensitivity or β-cell function were observed in the aforementioned studies (14–16). On the other hand, the direct comparison of NGT with elevated 1hPG to IGT with normal 1hPG may provide more insight into the metabolic abnormalities characterizing 1- and 2hPG. Unfortunately, this comparison was not available in previous studies, probably due to the limited number of participants with IGT but normal 1hPG, which represents a minority group in individuals with IGT (12).

To our knowledge, this is the first study to examine the pathophysiological distinctions between isolated defects in 1hPG and 2hPG. The relatively large sample size of individuals with IGT and normal 1hPG (n = 124) selected from a large population at high risk of type 2 diabetes, and the matching of age, sex, and BMI allow a more confident and straightforward interpretation of our findings. Although insulin sensitivity was comparable between NGT/1h-high and IGT/1h-normal, we found that the insulinogenic index was significantly lower in the former than in the latter. Further adjustment for insulin resistance showed that individuals with NGT with a 1hPG of ≥8.6 mmol/L had significantly lower DIs (0–30 min and 0–120 min) than those with IGT with 1hPG of <8.6 mmol/L. These findings suggest that the main pathophysiologic difference between 1hPG and 2hPG resides in β-cell function, especially early-phase insulin secretion. Consistent with our results, in the Botnia study, 8.5% of participants (n = 15 of 176) with NGT and a 1hPG ≥8.6 mmol/L progressed to type 2 diabetes after 7–8 years of follow-up. In contrast, none of the participants with IGT and a 1hPG <8.6 mmol/L (n = 23) developed type 2 diabetes (25). In addition, researchers conducting a prospective, population-based cohort study of 4,867 Swedish men observed that, compared with people with NGT and a 1hPG <8.6 mmol/L, those with NGT and a 1hPG ≥8.6 mmol/L at baseline had increased risk of type 2 diabetes (hazard ratio [HR] 2.93; 95% CI 2.48–3.46) and mortality (HR 1.29; 95% CI 1.19–1.39) after 39 years of follow-up, whereas the risk was not significantly heightened in people with IGT and a 1hPG <8.6 mmol/L (for type 2 diabetes: HR 1.17 [95% CI 0.43–3.15]; for mortality: HR 0.8 [95% CI 0.49–1.31]) (11). Another interesting finding of our study is that the fitted regression line demonstrating the relationship between PG levels and DI was significantly steeper for 1hPG than 2hPG, implying that 1hPG is more sensitive to the decline in β-cell function. Because progressive β-cell failure is the principal factor responsible for the development of overt hyperglycemia (26), results of our study further support the notion that 1hPG is an earlier marker of dysglycemia than 2hPG (27). Therefore, the adoption of 1hPG during OGTT may lead to earlier identification of individuals at high risk of type 2 diabetes and facilitate more timely intervention. Regarding potential predisposing factors, we observed that older age, but not other traditional variables such as sex, BMI, and family history of diabetes, was significantly related to higher odds of isolated IGT compared with isolated elevation in 1hPG (Supplementary Table 1), suggesting that tailored screening strategies may be warranted and are an interesting issue for further investigation.

It is important to note that the findings of our study do not justify the replacing of 2hPG with 1hPG. Specifically, most participants (92.7%) with IGT in our study population also had elevated 1hPG. Compared with individuals with isolated elevations in either 1hPG or 2hPG, those with combined elevation in 1hPG and 2hPG exhibited significantly greater impairments in estimates of insulin sensitivity and DI. Consistent with this observation, the Malmö Preventive Project revealed that participants with IGT/1h-high had the highest incidence rate of type 2 diabetes among the four groups stratified by 1hPG and 2hPG after both 12 and 39 years of follow-up (11). Additionally, most interventional programs aimed at preventing type 2 diabetes have included participants with IGT (2–4), demonstrating the possibility of slowing or halting the progression to type 2 diabetes. Taken together, these findings suggest that 2hPG still plays an important role in identifying those with substantially increased risk of type 2 diabetes, a risk that can be effectively reduced with timely intervention.

Several limitations of our study should be noted. First, due to the cross-sectional nature of our study, it remains unclear whether the disturbances in insulin sensitivity and β-cell function associated with elevated 1hPG are responsible for the future deterioration in glucose homeostasis. Additionally, because participants with IGT/1h-normal constitute a minority in our study population, further prospective studies are warranted to characterize the natural history of this subset, particularly in comparison with the NGT/1h-high group. Second, the OGTT was only performed once in this study. Although this represents a routine in clinical practice, the chance of misclassification could not be fully ruled out. Third, insulin concentrations, rather than C-peptide levels, were used for the estimation of insulin secretion. Although insulin and C-peptide are produced in equimolar amounts, C-peptide has negligible hepatic clearance and a more constant clearance rate in peripheral circulation (28). Therefore, the measurement of C-peptide allows for a more direct and accurate estimation of insulin secretion. Consequently, the insulin-based indices used in our study should be regarded as surrogate measures of insulin secretion and interpreted with caution, although they have been used widely in epidemiologic studies. Moreover, the use of insulin concentrations for both the indices of insulin sensitivity and secretion could theoretically introduce autocorrelation. However, the use of an index of insulin secretion divided by insulin resistance, or DI, may mitigate this concern. Finally, only Chinese participants from a tertiary medical center were included in the analysis. Given that the relative contributions of insulin resistance and β-cell dysfunction to the development of type 2 diabetes may vary across multiple ethnic groups, our findings may not be generalized to other populations.

In conclusion, individuals with NGT but elevated 1hPG were characterized by similar insulin sensitivity but more impaired β-cell function, especially early-phase insulin secretion, compared with those with IGT but normal 1hPG. Moreover, 1hPG seemed to be more sensitive to the deterioration in β-cell function than was 2-h glucose, suggesting that the use of 1hPG could aid the early identification of individuals at high risk of type 2 diabetes.

Acknowledgments. The authors thank Robert A. Vigersky (Medtronic Diabetes, Northridge, CA) for dedicating time and expertise to the revision of the manuscript. The authors also thank all research staff, students, and patients who participated in this work.

Funding. This work was funded by the Program of Shanghai Academic Research Leader (22XD1402300), the Shanghai Oriental Talent Program (Youth Project), the Shanghai Key Discipline of Public Health Grants Award (GWVI-11.1-22), the National Key Clinical Specialty (Z155080000004), and the Shanghai Research Center for Endocrine and Metabolic Diseases (2022ZZ01002).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.Z. conceived and designed the study. J.L. and J.N. contributed to data collection, data analysis, and writing the manuscript. H.S., X.H., W.L., W.Z., and Y.W. contributed to data collection and analysis. W.L. and W.Z. contributed to conducting of the study and to data collection. X.M., Y.B., and J.Z. contributed to interpretation of data and revision of the manuscript. J.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.