* Identify risk factors for type 2 diabetes mellitus (DM) in adolescents, and how they may be used to suggest the diagnosis.
* Define the factors associated with progression to overt DM in young individuals.
* Contrast the various means of distinguishing between type 1 DM and type 2 DM in adolescent patients.
Over the past 5 years, a number of reports have suggested that the incidence of type 2 diabetes mellitus among children and adolescents is increasing alarmingly. While these reports have spurred increased interest in a disorder previously thought to be rare in childhood, research into the epidemiology, demographics, etiology, and treatment of pediatric type 2 diabetes is still in its infancy. A number of studies have begun to shed light on this issue, and a degree of consensus is emerging among investigators in the field. This article will review the current state of knowledge regarding the epidemiology of type 2 diabetes mellitus (type 2) in pediatrics, available information on potential explanations for the increasing incidence, along with approaches to diagnosis and treatment.
The hallmark of diabetes mellitus is hyperglycemia resulting from metabolic defect(s) causing impaired insulin secretion, insulin action, or both. Historically, diabetes has been classified along therapeutic lines, as either insulindependent or non-insulin-dependent diabetes, or by age of onset (juvenile or adult-onset), but a more etiologic classification is now preferred. According to the new classification scheme, diabetes resulting from a deficiency of insulin secretion is termed type 1 diabetes. The most common etiology is immune destruction of the insulin-producing β-cells of the pancreas (type 1a). The prevalence of detectable β-cells autoantibodies in persons newly diagnosed with type 1a is very high and is the most reliable method of identifying the disorder. In contrast, type 2 diabetes is defined as diabetes resulting from insulin resistance, often, but not universally associated with a concomitant insulin secretory defect. Clearly defined diagnostic markers for type 2 diabetes have not been identified. Thus, the definitive diagnosis of type 2 currently requires the demonstration of absence of autoantibodies, and presence of insulin resistance through physiologic testing, although diagnosis is often made clinically. In addition to these two major forms of diabetes, less common diagnostic categories exist including type 1b (β-cells destruction without evidence of autoimmunity), MODY (an autosomal dominant disorder of early onset mild diabetes for which at least three distinct genetic mutations have been identified), and mitochondrial disorders associated with diabetes.
Historically, young children and teenagers were assumed to have what we now call type 1a diabetes, while those over 30 were considered to have type 2 diabetes. However, a number of reports of adolescents and children with new onset diabetes and clinical characteristics typical of insulin resistance rather than insulin deficiency have cast doubt on the validity of age-based diagnosis. It has subsequently become clear that the incidence of type 2 diabetes mellitus in adolescents is increasing dramatically in the United States and in other industrialized countries. Currently, most available data come from referral centers and nationwide incidence is difficult to ascertain. However, regardless of the population studied, all reports indicate that minority youth are disproportionately affected.
In Cincinnati, the proportion of newly diagnosed cases aged under 19 who were not type 1a was less than 5% until 1991, then rose to 17% by 1994 and the majority were African-American, whereas more than 90% of type 1a patients were white. The average age at presentation was approximately 13 years and there was a female preponderance not present among the type 1a patients. Among adolescent age patients (10–19 years) with new-onset diabetes, 40% had non-type 1a by 1994 . In San Diego, 67% of type 2 patients were Hispanic, compared with 20% of the general population . Similar changes in diabetes demographics have been reported in San Antonio, Denver, and Arkansas, with a disproportionately high percentage of minority adolescents diagnosed with type 2 [3,4].
Thirty years of data in the high-risk Pima Indian population have shown increasing rates of antibody negative diabetes among children. A corresponding increase in body weight, and an increase in the frequency of exposure to diabetes in utero have been reported. The Indian Health Service national database also revealed increased prevalence of diabetes for persons aged 15–24 between 1988 and 1996, while the prevalence for those <15 years old remained stable. A similar increase in diagnosis of type 2 diabetes has been seen among First Nation children in Canada .
Non-type 1a may also comprise a larger proportion of youth-onset diabetes among non-European populations worldwide than previously realized. In Japanese school children, type 2 is seven times more common than type 1, and its incidence has increased more than 30-fold over the past 20 years . Higher than expected incidence rates for type 2 diabetes in youth have also been reported in Osaka and among young Asian Indian patients in South Africa. As in adults, it is likely that diabetes is underdiagnosed in adolescents. Pediatric providers typically have limited experience with type 2 diabetes, and may not consider screening patients at high risk. In addition, adolescents often avoid medical care, especially adolescent males. One third of pediatric type 2 diabetes cases in academic centers were identified by routine screening urinalysis .
A number of apparent risk factors have been identified from referral center series and these have been recently reviewed . The most important risk factor appears to be severe obesity. Combined data from eight studies of 461 youth with type 2 DM show an average BMI greater than 30. About a third of patients had a BMI > 40 kg/m2 indicating morbid obesity, while 17% had BMI > 45 kg/m2. Furthermore, in a representative subgroup of adolescent type 2 patients, measurement of waist-to-hip ratio indicated that the obesity in these patients is typically central. However, despite the strength of obesity as an apparent risk factor for diabetes, the excess prevalence of obesity in Hispanic and African American adult populations does not explain the high risk of diabetes in these populations. Similarly, the magnitude of excess diabetes risk associated with obesity in adolescent populations and the extent to which obesity accounts for the excess non-type 1 diabetes in this age group, remains unknown.
Family history of type 2 DM has been strongly associated with type 2 DM in youth in multiple population groups. A family history of a first or second degree relative with type 2 DM was present in 74–100% of cases from multiple centers, compared with approximately 5% positive family history in youth with type 1 DM. Moreover, firstdegree family members of patients with type 2 diabetes have been found with previously undiagnosed diabetes based on subsequent evaluation and otherwise healthy siblings may have elevated levels of C-peptide and proinsulin. Obese offspring of diabetic parents have been found to have fasting insulin levels almost double that of comparably obese controls .
Adolescent females appear to be more susceptible to development of type 2 diabetes, with an overall female to male ratio of 1.7/1 irrespective of race. However, since these gender ratios are based on series from diabetes centers and not from population screening the possibility remains that this gender discrepancy results from a higher proportion of undiagnosed cases among males due to the known lower frequency of medical visits among adolescents males than females.
Minorities are overrepresented among cases of type 2 DM in youth. Among 578 American and Canadian youth with type 2 DM, 94% were from minority communities, while the incidence of type 1 more closely parallels the population mix of the general community . However, type 2 diabetes has been reported among nearly all ethnic groups and, in an individual case, ethnicity is not a useful predictive parameter.
A number of lifestyle characteristics have been identified among adolescent patients with type 2 diabetes. For example, these patients report a high fat and low fiber diet in spite of knowledge regarding appropriate food choices. An increase in fat intake has been associated with the increased incidence of type 2 diabetes in children in Japan. At the same time, adolescents with type 2 diabetes often report a sedentary lifestyle and do not participate in structured or routine exercise programs. In adults, low physical activity and high caloric intake are known environmental risk factors for obesity and type 2 diabetes and it is presumed that a similar relationship exists for adolescents.
Insulin Resistance and Type 2 Diabetes
Signs of insulin resistance, such as acanthosis nigricans, polycystic ovary syndrome, dyslipidemia, and hypertension are commonly present in patients diagnosed with new-onset type 2 DM. For example, among youth diagnosed with type 2 DM, acanthosis nigricans is present in 60–95% of cases. Similarly, hypertension is rare in patients with type 1 diabetes at presentation but may be present in 20–30% of patients with type 2 diabetes. Thus, it is presumed that the development of insulin resistance is an important feature of risk for progression to overt type 2 DM.
A reasonably clear picture is emerging on insulin metabolism during puberty. Adolescents became relatively insulin resistant with rises in growth hormone and other counterregulatory hormones. Thus, there is a significant decrease in insulin sensitivity through puberty, resolving to near prepubertal levels by Tanner stage 5. Girls are more insulin resistant than boys at every Tanner stage, approximately 50% of this difference being accounted for by adiposity. Ethnic differences in insulin sensitivity have also been recognized, and the lower insulin sensitivity in nonobese normal African Americans, Native Americans and Hispanics compared with white adolescents has been suggested as an explanation for higher prevalence of type 2 diabetes in U.S. minority populations. Differences in insulin resistance and secretion have been observed in prepubertal children, with black children being more insulin resistant than white children. Similarly, the Bogalusa group demonstrated significantly higher 0–60 minute insulin areas after OGTT in nondiabetic black children compared with whites, controlled for age, weight, height and Tanner stage . Thus, minority children may be more vulnerable to alterations in insulin secretion because of a greater need to compensate for insulin resistance, and more vulnerable to further decompensation as a consequence of further pubertal decreases in insulin sensitivity.
While pediatric and adolescent obesity has increased dramatically over the past decade, not every obese teen appears to have the same risk for developing diabetes. Progression to overt diabetes is a consequence of both insulin resistance and pancreatic beta cell dysfunction, with insulin resistance contributing to the development of impaired glucose tolerance, and deterioration of beta cell function leading to eventual overt decomposition. Thus, an elevated fasting insulin level and increased 2-hour glucose level, both markers of insulin resistance, and increased fasting proinsulin level, a marker for beta cell dysfunction, have been correlated with increased risk of progression to type 2 DM .
Among normoglycemic offspring of diabetic parents, diabetes was 61.8 times more likely in the subgroup of subjects with the lowest insulin sensitivity and an elevated fasting insulin level was associated with a 5-fold increase in the development of type 2 DM . Conversely, obese offspring of diabetic parents had fasting insulin levels almost double that of comparably obese controls. Among Pima Indians, there was a 15.8-fold risk for development of diabetes for subjects with fasting insulin levels in the 90th percentile . Pima Indian children who developed diabetes had significantly higher fasting insulin levels at baseline compared with controls that maintained normal glucose tolerance. In a similar study of Australian youth, subjects who developed impaired glucose tolerance or diabetes at follow-up 7–12 years later had elevated fasting and 2-hour insulin levels at baseline; for subjects in the highest quartile, the cumulative incidence of abnormal glucose tolerance was 32% compared with 3% in the first quartile . Among nondiabetic Mexican Americans, subjects who developed diabetes during an 8-year observation period had higher baseline levels of total cholesterol, LDL cholesterol, triglycerides, fasting glucose and insulin, 2-hour glucose, BMI, and BP, and lower HDL cholesterol. The risk of diabetes in subjects with normal baseline glucose tolerance was 1.7% among subjects in the lowest 3 insulin quartiles versus 7% among subjects in the highest insulin quartile .
Screening At-Risk Youth
Recently, observations regarding the apparent epidemiology of type 2 diabetes in adolescents were formalized in an American Diabetes Association and American Academy of Pediatrics consensus statement that addressed the criteria for selecting pediatric patients to be screened for non-type1a diabetes  (Fig. 1). This statement recommends testing for type 2 DM in overweight youth with two or more risk factors (family history of type 2 DM, high risk race/ethnicity including American Indian, African American, Hispanic, or Asian/Pacific Islander, or signs of insulin resistance such as acanthosis nigricans, hypertension, dyslipidemia, or polycystic ovary syndrome). Screening with a fasting glucose is recommended every 2 years starting at age 10.
While this consensus statement is a major advance in the recognition of type 2 diabetes as a pediatric disorder, the guidance remains inadequate in two ways. First, the recommended screening will only identify overt diabetes, since impairment of fasting glucose is the last step in the process of diabetes development. While undiagnosed diabetes is a major problem in type 2 among adults, ultimately screening must focus on identification of youth at high risk for development of diabetes. Since patients with type 2 diabetes have a high incidence of complications at the time of diagnosis, it would be preferable in the long run to prevent progression, rather than treating diabetes after it occurs. However, the optimal approach to identifying adolescents with insulin resistance is currently unclear, since we do not know the time frame for progression from insulin resistance to abnormal glucose tolerance to overt diabetes nor the point at which screening would be most effective. Furthermore, design of optimal screening will need to consider the ease of testing and its acceptability to providers and patients.
For the average adolescent, a fasting glucose may be difficult to obtain, but a full oral glucose tolerance test, which could identify youth in the impaired glucose tolerance stage, is an even more daunting prospect. At the other end of the convenience spectrum, glycosylated hemoglobin is insensitive and would offer little advantage over measurement of fasting glucose. Given that fasting insulin levels have been shown to correlate well with insulin resistance, measurement of fasting glucose and insulin and/or C-peptide would seem to be the ideal screen to identify at-risk patients. However, standardization of insulin assays and determination of values suggesting risk remain problematic issues . The problem of screening design remains to be resolved and will require larger population studies than have so far been undertaken.
The second way in which the current ADA consensus is inadequate is its reliance on currently accepted risk factors. Unfortunately, generalizations derived from diabetes referral clinics may be premature and ultimately obscure the true nature of the disease. For example, by using the degree of overweight as a primary determinant, there is a risk of “confirming” potentially biased clinic-based observations and allowing nonobese at-risk patients to remain unidentified. For example, data from the Barbara Davis Center for Childhood Diabetes indicate that, while the distribution of BMI is clearly higher among cases with apparent type 2 diabetes than among children with type 1a, there is substantial overlap—35% of type 2 patients have BMI under 28, particularly among Hispanic patients. Similarly, a substantial minority of patients reported in all series of type 2 diabetes among adolescents are not obese. Furthermore, in all case series from which current classification schemes are derived, antibody negative patients were assumed to have early onset of type 2 diabetes without determination of the true pathophysiology present (type 2, MODY, 1b etc). Thus, we do not know which of the risk factors are specific for type 2 diabetes, as opposed to other forms of non-type 1a.
A rigorous population-based examination of presumed epidemiology and risk factors, combined with an evaluation of the underlying physiology of antibody negative patients, will be necessary to validate these clinical observations and provide a reliable, noncircular, approach to disease classification before we can construct reliable diagnostic and screening algorithms. Meanwhile, it is important for clinicians to consider all aspects of potential risk, including ethnicity, family history, and physical signs of insulin resistance in determining which patients should be evaluated, even in the absence of frank obesity.
Classification of Diabetes in Youth
The clinical distinction between type 1 and type 2 diabetes can be difficult. In older subjects who have typical risk factors for type 2 DM and present without ketosis, diagnosis may be relatively straightforward. Conversely, the young patient with acute onset of ketotic hyperglycemia is overwhelmingly likely to have type 1 DM. However, patients may present with a mixed picture, making the classification more problematic. For example, it is not unusual for youth with type 2 DM to present with ketonuria or even frank ketosis, while patients with type 1 may present early in the course with moderate hyperglycemia in the absence of ketones. In cases where the diagnosis of type 2 vs. type 1 is uncertain, laboratory data may help clarify the diagnosis. Figure 2 reflects a suggested diagnostic approach based on the recent ADA guidelines.
Islet cell antibodies are present at diagnosis in approximately 90% of patients with type 1 DM at the time of disease onset, but are generally absent in type 2. Thus the absence of antibodies, in the right clinical context, is suggestive of type 2 diabetes, while their presence is strongly suggestive of type 1.
In patients with type 2 diabetes, both plasma insulin and C-peptide concentrations are usually high, reflecting underlying insulin resistance. However, chronic hyperglycemia can cause transient insulin deficiency (“glucose toxicity”) and, therefore, insulin and c-peptide levels may be deceptively low at the time of presentation in the presence of acute metabolic deterioration. Conversely, “insulin reserve” can persist in type 1 diabetes for up to 2 years. Thus, though measurement of insulin and c-peptide can be helpful in distinguishing type 1 and type 2 diabetes, particularly when elevated, a significant degree of overlap exists and it is difficult from current studies to choose a level below which type 2 diabetes is excluded.
Treatment of Type 2 Diabetes in Youth
Current approaches to the treatment of type 2 DM in youth are based on adult practice, as no large-scale studies of adolescents exist. As with adults, dietary modification and exercise are critical components to treatment regimens. Research in some populations suggests that lifestyle modification alone is rarely successful with long-term treatment of type 2 DM in youth, while other authors report good glycemic control with diet and exercise therapy.
Insulin is currently the only approved pharmacologic method of treatment for diabetes in the pediatric population and insulin is often required at diagnosis to overcome glucose toxicity, or during periods of illness or stress. However, most youth with type 2 DM can be successfully weaned off of insulin after diagnosis with the addition of other therapies. Chronic insulin use in type 2 DM may be a significant burden on families and lead to hypoglycemia and worsening of obesity and should be avoided if other treatment strategies are successful.
Oral hypoglycemic agents are not FDA approved for use in the pediatric population, but have been used in many centers for treatment of youth with type 2 DM following adult dosing guidelines. Preliminary studies with metformin have not shown unusual side effects of oral hypoglycemic medications in youth with type 2 DM and additional studies with a variety of agents are currently underway.
The underlying cause of the recent epidemic of type 2 diabetes in youth is unclear, but likely reflects the confluence of multiple factors. Pediatric and adolescent obesity has increased dramatically over the past decade. During the same time period, physical activity levels have decreased, promoted by significant reductions in required physical education classes in schools, “pay-to-play” organized sports replacing school based activities, and an increasing emphasis on athletic excellence rather than broad participation. Children are also spending increased time in sedentary activities such as watching television and playing video or computer games. These trends are exacerbated by deteriorating urban neighborhoods and parental work schedules that lead to a great deal of indoor time after school. Finally, the increasing availability of inexpensive, high-caloric-density foods promotes chronic excess intake of a diet that promotes both weight gain and insulin resistance. These societal changes, when combined with apparent ethnically specific differences in metabolism, then result in the increased incidence of this first pediatric lifestyle disease.
Clearly, reversal of this trend is going to require coordination of clinical, familial, and societal interventions. However, the design of effective programs will require a great deal more information than we currently have. Specifically, additional research is needed in areas such as epidemiology and classification of diabetes in youth, with larger population based studies to better understand the true incidence and prevalence of the disease. In addition, attention is required to determine practical screening approaches for at-risk youth, including whether to screen for insulin resistance, abnormal glucose tolerance, or pre-existing diabetes. Finally, studies are needed to determine appropriate treatment interventions for patients with overt type 2, as well as whether aggressive intervention in insulin resistant youth prior to the development of overt diabetes will have long-term benefits.