While the diagnosis of type 2 diabetes mellitus (T2DM) in children is traditionally based on fasting plasma glucose (FPG) or 2-hour plasma glucose (PG) in an oral glucose tolerance test (OGTT), the role of intermediary glucose measurements at other time points in a standard OGTT is increasingly being investigated. Large-scale population studies in adults have consistently shown that 1-hour PG during OGTT is a sensitive biomarker for diagnosing dysglycemia and detects incident T2DM and adverse cardiovascular outcomes earlier than FPG or 2-hour PG.^1-3 On the other hand, the utility of 30-minute PG has received less attention, especially among the pediatric age group. Abnormal 30-minute PG reflects inadequate first-phase insulin response, and hyperglycemia at this period is the earliest detectable point of defect of pancreatic β-cell function.^4,5 As opposed to 2-hour PG in OGTT, the 30-minute PG in OGTT can be performed with a much shorter clinic visit and might help identify those with high risk of abnormal OGTT response. With the increasing prevalence of obesity, the use of 30-minute PG in identifying high-risk individuals may save time and resources compared to performing standard OGTT in all overweight or obese children.

Our group has previously reported the utility of combining clinical and non-fasting parameters in stratifying low- and high-risk groups among a cohort of overweight or obese pediatric individuals.⁶ The positive predictive value (PPV) from combination of glycated hemoglobin (HbA1c), serum alanine aminotransferase (ALT) level, and family history of T2DM was 61.6% in the prediction of abnormal OGTT response. The combination of these parameters was a promising measure in identifying high-risk individuals who should be offered further testing with OGTT.⁶ The current study aimed to explore the optimal cutoff of 30-minute PG in predicting abnormal OGTT response. This cutoff was compared and combined with the predictive value of that derived from combinations of non-fasting parameters.

Data of 332 pediatric individuals with OGTT from the Department of Pediatrics and Adolescent Medicine, Queen Mary Hospital, The University of Hong Kong, Hong Kong, a tertiary, university-affiliated pediatric unit, from January 2012 to December 2018, were reviewed. These individuals were referred to the clinic for evaluation of overweight and obesity, which were defined as body mass index (BMI) >+1 standard deviation (SD) and >+2SD above the age and sex reference mean on the World Health Organization growth chart, respectively. Individuals who had an underlying syndromal cause of obesity and those on medications that can potentially impact glucose response in an OGTT were excluded.

Anthropometric measurements of height, weight, and BMI (kg/m²) were obtained on the same day as OGTT. Standard OGTT was carried out after an 8-hour fast, and blood for PG was collected at three time points (0 minute, 30 minutes, and 120 minutes). Prediabetes (either impaired fasting glucose with FG ≥5.6 mmol/L or impaired glucose tolerance with 2-hour PG ≥7.8 mmol/L) and diabetes (FG ≥7.0 mmol/L or 2-hour PG ≥11.1 mmol/L) glycemic response were defined according to the American Diabetes Association. Individuals with prediabetes and diabetes were considered together as a group (abnormal OGTT group) for statistical analysis. The values of 30-minute PG were compared between the abnormal and normal OGTT groups. The optimal 30-minute PG cutoff for diagnosis of abnormal OGTT was determined using a receiver operating characteristics (ROC) curve, where sensitivity and specificity were calculated for various cutoffs. The optimal cutoff was noted at maximum sensitivity and specificity. The PPV of this cutoff was compared to that derived from a combination of non-fasting parameters reported in our previous study. This cutoff was also combined with various non-fasting parameters to evaluate the PPVs for abnormal OGTT response.

The IBM SPSS version 22.2 (IBM Corp., Armonk, NY, USA) was used for statistical analyses. Area under the curve (AUC) was calculated using the SPSS ROC curve function for continuous variables. Unpaired t-test or chi-square tests were used to compare the clinical and biochemical characteristics of two groups. All data were expressed as mean±SD. unless otherwise indicated. Degree of correlation between 30-minute PG with clinical or biochemical parameters was presented as Pearson correlation coefficient. Statistical significance was inferred at a two-tailed P-value <0.05.

All clinical investigations were conducted in accordance with the guidelines of the Declaration of Helsinki and were approved by the Institutional Review Board of the Hospital Authority Hong Kong West Cluster (No. UW19-646). In view of the retrospective nature of the study, and that all data were collected on clinical grounds, informed consent from individuals was waived.

This study comprised 332 patients, of whom 323 (97.3%) were Chinese and 180 (54.2%) were males. The nine non-Chinese patients were two Caucasians and seven South East Asians. The mean age of our cohort was 15.4±2.3 years. Mean body weight was 88.8±

17.4 kg, and mean body height was 165.2±9.7 cm, with a mean BMI z-score of 2.7±0.6. Sixty individuals (18.1%) had abnormal OGTT –47 (14.2%) with prediabetes and 13 (3.9%) with diabetes. Table 1 shows a comparison of the baseline characteristics and OGTT findings between the two groups. There was no statistically significant difference in age, sex, or BMI SD score between the two groups. The 30-minute PG was statistically higher in the abnormal OGTT group compared to the normal OGTT group (9.6 vs. 7.9, P=0.001).

Fig. 1 displays the ROC curve at various cutoffs of 30-minute PG. The optimal cutoff for prediction of abnormal OGTT was PG ≥9.2 mmol/L (AUC, 0.77; sensitivity, 62.1%; specificity, 83.7%). The PPV for abnormal OGTT response at this cutoff was 45%. The PPVs when this 30-minute PG cutoff was combined with various non-fasting and clinical parameters and their respective 95% confidence interval (CI) are shown in Table 2. The PPV differences between combinations of parameters were statistically significant. The combination of 30-minute PG ≥9.2 mmol/L, abnormal ALT, family history of diabetes, and HbA1c ≥5.5%, as derived from our previous study, achieved the highest PPV of 70.0% (95% CI, 69.1%–70.9%) for abnormal OGTT response among our cohort. This PPV was significantly higher than the 61.6% (95% CI, 60.7%–62.4%) of the combination of abnormal ALT, family history of diabetes, and HbA1c ≥5.5% without addition of 30-minute PG. There was positive correlation of 30-minute PG with 2-hour PG (r=0.48, P<0.05) and HbA1c (r=0.22, P<0.05). No correlation was found for 30-minute PG with FPG, ALT level, or BMI.

Our study showed that 30-minute PG ≥9.2 mmol/L is the optimal cutoff for predicting abnormal OGTT response in a cohort of overweight or obese pediatric individuals. Measurement of 30-minute PG, which is more time-efficient and convenient in the outpatient setting, could serve as a valuable tool in stratifying individuals referred to the obesity clinic. While the predictive power when used alone was fair, combination with other non-fasting parameters resulted in better clinical utility.

A handful of studies have demonstrated association between increased 30-minute PG and higher risk of progression to T2DM in adults.^7-11 However, few of them were conducted in Asian populations, which exhibit different trajectories of glycemia and insulin sensitivity compared to the White population.¹² Of these, one study was performed in a Japanese adult population involving more than 2,000 participants with follow-up of seven years. This study demonstrated that 30-minute PG was positively associated with future diabetes not only in those with pre-diabetes, but also in those with normal glucose tolerance. In addition, incorporation of 30-minute PG into the model including traditional diabetic risk factors plus FPG and 2-hour PG significantly improved the prediction of diabetes.¹⁰

No similar studies on the utility of 30-minute PG in children or adolescents have been performed, and none has evaluated its ability in discriminating abnormal OGTT response in the clinical setting. As peak glucose absorption occurs 30 to 60 minutes after ingesting a mixed meal, this period represents an optimal time-point to detect inadequate insulin response; hyperglycemia at this time-point could potentially be the earliest sign of metabolic dysfunction. Kim et al.¹³ conducted a clinical study comparing adolescents with a monophasic versus biphasic glucose response curve during an OGTT, where monophasic was defined by a gradual decrease after the peak of intermediate PG level without reascending at 2 hours. Adolescents with a monophasic glucose response curve had lower in vivo insulin sensitivity and inadequate compensation in first- and second-phase insulin secretion, despite having similar fasting duration and 2-hour glucose and insulin concentrations. As impaired insulin sensitivity and b-cell function are two major pathophysiological mechanisms of young-onset T2DM, this study demonstrated the role of intermediate PG measurement as a strong metabolic marker for T2DM. In addition, individuals with high 30-minute PG in an OGTT represent those with greater glycemic variability after a meal, which has been demonstrated to promote free radical production and trigger oxidative stress to result in increased insulin resistance, impaired insulin secretion, and heightened risk of T2DM.^14,15 With regard to adverse cardiovascular outcome, adolescents with obesity and elevated 30-minute PG had higher levels of inflammatory mediators that indicate atherosclerosis.¹⁶ In addition, compared to steady state hyperglycemia, an unstable glucose level in response to a glucose load can cause greater damage to vascular endothelial cells.^9,17 Our study demonstrated superior predictive value for abnormal OGTT response when 30-minute PG was combined with non-fasting parameters compared to use of non-fasting parameters alone. We validated these parameters to identify high-risk individuals and decrease time from the standard 2-hour OGTT. However, further longitudinal studies are necessary to delineate the significance of this 30-minute PG cutoff in predicting incident T2DM and adverse cardiovascular events.

This study has some limitations. First, the generalizability of the findings may be limited as it was conducted in mainly Chinese individuals. Asians, compared to Caucasians, have been demonstrated to have reduced post-load insulin secretory capacity especially in the early phase, resulting in reduced insulinogenic index in a glucose tolerance test.¹⁸ Hence, further evaluation to assess the validity of this 30-minute PG cutoff in individuals of other ethnic groups is necessary. Second, local prevalence of disease, e.g., non-alcoholic fatty liver disease, may also affect the PPV of non-fasting parameters used in our previous study. Finally, OGTT was performed once only without measurement of PG level at other time point. This may compromise the reproducibility of the findings, and comparison cannot be made with 1-hour PG in terms of the predictive value for abnormal OGTT response.

In conclusion, a 30-minute PG ≥9.2 mmol/L was the optimal cutoff in predicting abnormal OGTT response in a cohort of overweight or obese pediatric individuals in our study. When combined with non-fasting metabolic parameters, it has robust predictive value for abnormal OGTT response. Hence, shortening the standard OGTT to 30 minutes may improve clinical usability while providing effective initial stratification for high-risk individuals. Routine measurement of 30-minute PG should be encouraged in the clinical setting to allow future longitudinal studies to evaluate the outcome.

The authors declare no conflict of interest.

We would like to thank all the patients participating in this study.

Study concept and design: JYLT; acquisition of data: all authors; analysis and interpretation of data: all authors; drafting of the manuscript: SWYP; critical revision of the manuscript: JYLT; statistical analysis: SWYP; administrative, technical, or material support: all authors; and study supervision: JYLT.