Pros and Cons of Current Diagnostic Tools for Risk-Based Screening of Prediabetes and Type 2 Diabetes in Children and Adolescents with Overweight or Obesity

The worldwide prevalence of overweight and obesity dramatically increased in the last decades, from 4% in 1975 to over 18% in 2016 among children and adolescents [1]. High body mass index in childhood has been associated with an increased risk of type 2 diabetes (T2D) [2]. T2D in youth has specific characteristics compared to adults, including an accelerated deterioration in insulin secretion and a rapid development of diabetes complications [3]. Thus, the early diagnosis of T2D or prediabetes by risk-based screening is crucial to detect asymptomatic children and adolescents and promptly manage the condition. The aim of this study is to review available diagnostic tests for risk-based screening and diagnosis of prediabetes and T2D in children and adolescents with overweight or obesity and analyze their pros and cons for clinical practice.

The validity of a screening test is based on its sensitivity, specificity, predictive value, and negative results. The predictive value is correlated with the prevalence of the condition, and in low-prevalence settings, even excellent tests have poor predictive positive value [4]. In adults, it has been demonstrated that the screening for prediabetes and T2D, and thus the early identification and treatment of individuals, is effective and cost effective for reducing the rates of diabetes [5]. However, the prevalence of T2D in children is low, compared to adults and the current evidence is insufficient to assess the balance of benefits and harms of screening for T2D in asymptomatic children and adolescents [6]. Nevertheless, the screening for T2D in the high-risk pediatric population, i.e., children and adolescents with overweight and obesity, was introduced in 2000 in response to an increased report of obesity and T2D phenotype in tertiary care and high-risk populations [7]. Thus, it is important to identify high-risk children that could benefit from screening programs and in Table 1 are reported the criteria according to the American Diabetes Association (ADA) Guidelines [8]. The decision to screen should also be guided by clinical judgment, as some youth with obesity may have earlier onset of puberty and therefore need earlier screening, but also depending on the prevalence of diabetes in different parts of the world [9]. Noteworthy is that risk-based screening for prediabetes or T2D in children and adolescents allows early diagnosis of alterations in glucose metabolism that enable early and targeted preventive and therapeutic interventions that could prevent or delay progression to T2D, so fast in early ages [10, 11]. Moreover, the early diagnosis and intensive treatment of T2D during youth is important to maintain good glycemic control and prevent T2D complications [12]. Potential harms of screening include overdiagnosis and overtreatment. While the use of medication can include possible side effects (e.g., hypoglycemia and gastrointestinal problems), lifestyle interventions alone, as a nonpharmacological approach, are essentially free of serious side effects [6].

The ADA recommends fasting plasma glucose (FPG), 2-h plasma glucose (PG) during a 75-g oral glucose tolerance test (OGTT), and glycated hemoglobin (HbA1c) to test for prediabetes and T2D in children and adolescents at high risk (Table 2) [3, 8]. Laboratory diagnostic thresholds used in pediatrics are derived from epidemiological studies in adults. Indeed, the specific suitability of those criteria remains to be further investigated in pediatric patients and the optimal screening tests for diabetes in high-risk children and adolescents are debated.

Noteworthy is that the three recommended tests do not identify the same population, in terms of type of glucose intolerance [13‒15]. The estimated prevalence of diabetes or prediabetes is therefore different, depending on the test used, and the use of one of these methods alone would fail to identify all individuals at risk of developing diabetes [15]. Pros and cons of current recommended screening tests are reported in the following paragraphs and Table 3.

Oral glucose tolerance test is largely employed to assess glucose tolerance in youth [16]. After an overnight fast, the standard glucose load of 1.75 g/kg body weight, up to a maximum dose of 75 g, is given orally. Blood samples are collected at baseline fasting and in the following 2 h from glucose ingestion, to measure PG concentrations. Concurrently, insulin, C-peptide, and other hormones could be assayed [16].

Diagnostic Criteria. Diagnostic criteria based on 2-h post-load glucose concentration (2hPG) (140–199 mg/dL) have been considered the most reliable for the diagnosis of impaired glucose tolerance (IGT) [15]. In fact, the OGTT is the gold standard for the diagnosis of this high-risk condition for the development of T2D and cardiovascular disease [8, 17]. Weiss et al. [18] longitudinally studied 117 children and adolescents with obesity showing that among youths classified as IGT at baseline, 30% remained IGT, 24% developed T2D, and 45.5% reverted to NGT in about 2-year follow-up. Lower rates of progression were reported for European children with IGT, with 2% of children with IGT developing T2D, 16% remaining IGT, and 75% converting to NGT in 3–5 years of follow-up [19]. The IGT is also associated with cardiovascular risk factors in children, including high blood pressure, atherogenic dyslipidemia, low insulin sensitivity, and hepatic steatosis [20, 21]. Detecting patients with IGT is therefore fundamental, especially if FPG levels are within normal ranges. Indeed, evidence in adults showed that 2hPG detects individuals with diabetes earlier in the progression of the disease, when FPG might be not increased [22]. Thus, an early diagnosis of this condition could be beneficial for identifying individuals who could benefit from early intensive lifestyle modification.

Noteworthy is that youths with obesity and NGT with 2hPG values between 120 and 139 mg/dL were found to have the disposition index obtained from hyperinsulinemic-euglycemic clamp (cDI) not different from youth with obesity and IGT [23]. The cDI was the best predictor of their 2hPG measured after 2 years. Therefore, youths with obesity and 2hPG values within 120–139 mg/dL have already an impaired β-cell function relative to insulin sensitivity, with increased future risk of IGT and T2D.

Insulin Sensitivity Assessment. The euglycemic-hyperinsulinemic clamp is the “gold standard” for measuring insulin sensitivity. However, it is costly, time demanding, troublesome for participants, especially in pediatric age, and requires a research setting [24]. Although the OGTT cannot directly measure insulin sensitivity, it allows to derive surrogate measures of both fasting and glucose-stimulated insulin sensitivity, such as inverse fasting insulin, homeostatic model assessment, quantitative insulin sensitivity check index (i.e., a permutation of homeostatic model assessment, QUICKI), and the combined PG and insulin excursion during OGTT (whole-body insulin sensitivity index, WBISI, or Matsuda-Defronzo index) [16]. Interestingly, regardless of the glucose tolerance status, fasting-derived indices of insulin sensitivity have shown higher correlations with clamp insulin sensitivity than OGTT-derived indices in youth with obesity [25]. Nevertheless, it is not appropriate to use any surrogate measure of insulin sensitivity since there is a great overlap of these measures across quartiles of clamp-measured insulin sensitivity [25].

β-Cell Function Assessment. The relationship between β-cell function and insulin sensitivity underlies a hyperbolic function, mathematically represented by the product of insulin sensitivity and β-cell function, i.e., DI [9]. The DI is therefore expression of the degree of glucose tolerance in any individual and, when low, an early marker of β-cell function impairment [9, 26]. Some parameters obtained from OGTT can be used to derive the oral DI, which has shown to predict the development of diabetes over 10 years among adults without diabetes, independently of age, sex, body mass index, family history of diabetes, FPG, and 2hPG [26].

Gut-Pancreas Axis Assessment. A specific key advantage of the OGTT is that, when compared with an intravenous glucose tolerance test, it allows to assess the physiological potentiating contribution of the gut-pancreas axis in the measure of insulin responses after the ingestion of glucose (i.e., the incretin effect) [16, 27]. Interestingly, in a study of 255 youths with obesity, the incretin effect in youth with IGT or T2D was reduced, respectively, of 32% and 38% compared with NGT, with no evidence of decreased incretin concentrations [27].

Disadvantages. The most critical disadvantages of OGTT are its poor reproducibility and the lack of validation in youth [28, 29]. Libman et al. [29] examined 60 youths with obesity who underwent two repeated OGTTs, showing that the percent positive agreement between the two OGTTs was low for both impaired fasting glycemia (IFG) (22%) and IGT (27%). Moreover, youths with discordant OGTT results were more insulin resistant and had a lower relative insulin secretion, with a higher risk of T2D [29].

Other relevant issues are the need of the overnight fast and the length of the test, which could be demanding particularly for younger children, and associated nausea can occur in a subset of individuals [28]. Noteworthy is the preanalytic and analytic variability, more frequent for FPG and 2hPG, which require a proper storage and centrifugation of the samples [30].

The use of FPG as diagnostic tool for prediabetes and diabetes may be preferred due to its feasibility, as it is suitable for all ages and only one blood draw is required [28], higher reproducibility than the 2Hpg, and the widespread availability of the test in laboratories [29]. Noteworthy is that FPG levels within the normoglycemic range in childhood have shown to predict prediabetes and T2D in adulthood [31]. In the Bogalusa Heart Study, an increased risk (>2-fold) for developing adult prediabetes and diabetes in children with an FPG level of 86–99 mg/dL compared with those with less than 86 mg/dL, regardless of other traditional cardiovascular risk factors, has been reported. Moreover, a significant decline of about 49% in β-cell function in children in the IGF category (≥100–<126 mg/dL), independently of obesity and race, and a decline of about 23% in the category between 90 and 100 mg/dL compared with that in the lower FPG of ≤90 mg/dL was reported [32]. Thus, the impairment in β-cell function relative to insulin sensitivity is already evident in children with FPG levels within normoglycemic range.

Disadvantages of the FPG include the need of fasting and the lack of validation in youth. Furthermore, FPG is affected by stress and illness, both of great relevance in pediatric age, and can drop as low as 5–7% per hour in a sample due to glycolysis [28].

Hemoglobin A1c was included as diagnostic test for prediabetes and diabetes by ADA Guidelines in 2010 [33]. Hemoglobin A1c is an indirect measure of glycemia. It is the result of the glycation of circulating hemoglobin and reflects average blood glucose levels over approximately 120 days [34]. Therefore, HbA1c is a useful tool to assess glucose control over time, in comparison with FPG that reflects a specific moment [34]. Importantly, the test should be performed in a laboratory using a National Glycohemoglobin Standardization Program (NGSP, www.ngsp.org) certified method and standardized or traceable to the Diabetes Control and Complications Trial (DCCT) reference assay [8]. In this way, HbA1c results from different laboratories are harmonized and comparable with those reported in the DCCT [34].

The main advantages of HbA1c are that it does not require fasting prior to testing, it has greater preanalytical stability and less daily perturbations during stress, changes in diet, or illness [8]. Moreover, HbA1c levels are not significantly influenced by the current use of drugs that alter glucose homeostasis, such as corticosteroids, antibiotics, and lipid-lowering and antihypertensive agents [35]. However, HbA1c has a lower sensitivity at the established cutoff, higher cost, limited availability in certain developing countries, and a suboptimal correlation with average glucose in certain individuals [8]. In fact, being an indirect measure of glycemia, several nonglycemic factors have been described to alter HbA1c values (Table 4).

Diagnostic Criteria. Unfortunately, neither this test has been validated in pediatric population and as screening tool in children and adolescents it has raised concern [9]. Specifically, Nowicka et al. [36] studied a large multiethnic cohort of children and adolescents with obesity and reported that HbA1c at the cutoff level of 6.5% had relatively low sensitivity and specificity for detecting T2D and that the prevalence of prediabetes and T2D would be underestimated. Buse et al. [37] showed that among sixth-grade youth HbA1c levels of 5.7–6.4% were less common than IFG (respectively, 3.2% and 16.0%). After 2-year follow-up, high-risk HbA1c levels persisted in 59.4% and progressed to HbA1c >6.5% in 0.8% while IFG persisted in 46.9% and progressed to FPG >126 mg/dL in 1.1%. Among a large cohort of children and adolescents with overweight or obesity, 2.4% of study patients have been classified as having diabetes using both HbA1c and OGTT ADA criteria [14]. Among them, 68.7% had HbA1c ≥6.5% (≥48 mmol/mol) and 46.1% had FPG ≥126 mg/dL (≥7.0 mmol/L) and 2hPG levels ≥200 mg/dL (≥11.1 mmol/L), supporting the reliability of HbA1c as a diabetes screening tool.

The reliability of HbA1c in predicting the future risk for T2D in children has been recently demonstrated in a group of American Indian children and adolescents followed up for 42 years [38]. The incidence of diabetes was higher in those who had prediabetes in childhood than in those who did not have prediabetes, independently from the test used (HbA1c, FPG, 2hPG). Furthermore, the areas under the curve (AUCs) were, respectively, 0.70, 0.63, and 0.73 for HbA1c, FPG, and 2hPG, with a lack of significant difference in AUC between HbA1c and FPG or 2hPG [38].

Worth mentioning is that among youth with overweight or obesity in a large primary care setting, HbA1c values within the ADA-defined prediabetes range have shown not to convey the same risk for the development of diabetes. Youth with baseline HbA1c values ranging 6.0–6.4% had higher rates of progression to diabetes than those with HbA1c between 5.7% and 6% [39]. Thus, it has been questioned whether lower HbA1c thresholds should be used, although insufficient evidence is available [14, 36, 38].

In conclusion, despite the main outstanding issues, especially in the pediatrics, HbA1c continues to be recommended by ADA Guidelines as a diagnostic tool for prediabetes and diabetes [3]. The convenience of measuring HbA1c can increase the adherence to screening in high-risk people [40]. Since HbA1c is an indirect measure of glycemia, its limits should be considered when choosing this test [8]. Nevertheless, when FPG may be affected by acute stress or concomitant illnesses, the evaluation of HbA1c may be diriment.

In adults, the 1-h post-load glucose concentration (1hPG) during an OGTT has been proposed to identify dysglycemia earlier than currently recommended biomarkers [41]. Specifically, a 1hPG level ≥155 mg/dL (8.6 mmol/L) may detect people with reduced β-cell function before the progression to prediabetes and diabetes and it is a more accurate predictor of the progression to diabetes than HbA1c or 2hPG levels [41].

Less evidence is available regarding the role of 1hPG in predicting prediabetes and diabetes among children and adolescents. A longitudinal study showed that in Latino youth with obesity a baseline value of 1hPG ≥155 mg/dL was associated with a greater reduction in β-cell function and, among the NGT participants, with an increased risk of developing prediabetes over time [42]. In adolescent girls with obesity who underwent a multiple-sample OGTT, high 1hPG levels during 1-h OGTT showed to have moderate reproducibility and a similar predictive value for the development of prediabetes, compared to the standard OGTT [43]. Additionally, in children with obesity elevated 1hPG values have been associated with worsened insulin sensitivity indices, impaired β-cell function, and unfavorable metabolic and inflammatory profile [44].

A lower cut point has been proposed in children and adolescents by Manco et al. [45]. They have found that the threshold of 132.5 mg/dL (7.36 mmol/L) was more sensitive than the cutoff value of 155 mg/dL, in predicting IGT, with similar specificity. Adopting this cut point, Tricò et al. [46] have studied 202 NGT youths with obesity from different ethnic groups, demonstrating that increased 1hPG levels may prospectively predict the progression to prediabetes.

In summary, the 1-h OGTT seems to be a promising option for early detection of people at increased risk of prediabetes and diabetes, given its high sensitivity. It would also allow to shorten OGTT duration, with the possibility of expanding its use, especially among younger individuals.

OGTT Glucose-Response Curve. Different shapes of glucose response curve during an OGTT have been described in adults and youth [47, 48]. A gradual rise of blood glucose between 30 and 90 min until a peak followed by a decline defines a monophasic (MPh) curve. Instead, if the increase in blood glucose to a peak is followed by a fall and then by another increase it defines a biphasic (BPh) curve. When blood glucose concentration continues to rise gradually without a fall, remaining elevated at 120 min, it defines an incessant increase (IIn) [48].

Among individuals without diabetes, those with MPh curve have shown increased insulin resistance and impaired β-cell function compared with subjects with BPh curve [47, 49, 50]. In the Treatment Options for T2D in Adolescents and Youth (TODAY) study cohort of youth with T2D, MPh shape was the most frequent (68.6%), followed by IIn (21.7%) and BPh (9.7%). People with the IIn curve had worse β-cell function, the highest glycemic failure rates, and the fastest β-cell function decline [48]. The severity of β-cell dysfunction was confirmed in the RISE study of youth and adults with IGT and recently diagnosed T2D, as shown by the lowest clamp-measured β-cell responses among youth and adults with IIn curve [51]. Noteworthy is that the typical β-cell hypersecretion in youths with IIn curve was absent, maybe due to the severe β-cell impairment.

In summary, the morphology of the OGTT curve have been associated with different β-cell function and insulin sensitivity, from the most favorable BPh shape to the least one (IIn). While useful, the shape study requires at least four glucose measurements and may increase the economic and personal burden, limiting its applicability to some clinical settings.

OGTT Time-To-Glucose Peak. The timing of the glucose peak during an OGTT has been associated with the risk of the development of prediabetes and T2D in adults and youths [52, 53]. An early glucose peak (<30 min) reduced the risk of progression from NGT to IGT, or the persistence of IGT, or the progression to T2D, in youths with obesity [52]. Moreover, late-peak subjects compared with early-peak ones had impaired β-cell function relative to insulin sensitivity, lower insulin sensitivity of glucose and free fatty acid metabolism, and blunted incretin secretion [54].

In adults, time-to-glucose peak showed a reliable reproducibility (k coefficient 0.76), while moderate and poor reproducibility were, respectively, achieved by 1hPG ≥155 mg/dL and shape of the glucose curve [55]. In conclusion, a later glucose peak (>30 min) is considered a worse metabolic marker of glucose tolerance. The reliable reproducibility of this OGTT-derived parameter may have important clinical implications.

Additional nonfasting glycemic markers have been proposed as potentially screening tools for prediabetes and diabetes. In particular, 1,5-anhydroglucitol reflects the past 48-h 2-week glycemia, while glycated albumin and fructosamine the past 2–4 weeks [56]. Chan et al. [57] studied these markers in youth with overweight or obesity (n = 117) and found excellent ROC-AUCs (0.82–0.98) in predicting diabetes, whereas low ROC-AUCs were reported for prediabetes (0.5–0.66). To note, fructosamine results are unaffected by hemoglobinopathies [58]. However, further studies are needed to assess the usefulness of these biomarkers in pediatric clinical practice.

Each of the tests proposed for the screening has advantages and disadvantages providing an overview of different aspects of glucose homeostasis [28]. To the best of our knowledge, no international practical algorithms have been approved for risk-based screening of prediabetes and T2D in children and adolescents with overweight or obesity. In this review, we propose a practical flowchart for the clinical use of screening tools in this population (Fig. 1). Importantly, when using FPG, HbA1c, and 2-h OGTT as screening tests, strengths and weaknesses of each test, available resources, and the clinical setting should be considered.

Currently recommended tests for risk-based screening of prediabetes and T2D in children and adolescents have not been validated in the pediatric population, threshold values have been derived from adults, and it is still debated what test is the most suitable in the clinical practice. Since each test has its pros and cons and contributes to explore different components of glucose metabolism, a personalized choice or the performance of more than one test could be useful depending on the clinical context and the setting. Considering these potential limitations, a flowchart has been proposed for the clinical use. Nevertheless, longitudinal studies with long follow-up and large cohorts would be needed to validate any test for the diagnosis of prediabetes and T2D in children and adolescents, although they are difficult to perform in this population.

The authors have no conflicts of interest to declare.

This research received no external funding.

Claudio Maffeis and Chiara Garonzi performed the literature search. Chiara Garonzi, Claudio Maffeis, and Alice Maguolo wrote and edited the manuscript. All authors have read and agreed to the published version of the manuscript.