Clinical characteristics in Japanese children with nonobese type 2 diabetes

1. Urine glucose screening program at schools in the Tokyo Metropolitan Area

We have annually conducted a urine glucose screening towards primary school children, aged 7–12 years, and junior high school children, aged 13–15 years, residing in the Tokyo Metropolitan Area [5,6]. During the period from 1975 to 2015, a total of 11,652,205 students, including 7,955,857 primary school students and 3,696,348 junior high school students, participated in the urine glucose screening program. Finally, a total of 301 students, including 64 primary school students and 237 junior high school students, were diagnosed as having type 2 diabetes. The incidence of type 2 diabetes (per 100,000 students/yr) during the entire study period was 2.58 overall with 0.80 in primary school students and 6.41 in junior high school students [9].

Type of diabetes was retrospectively determined to follow up the clinical course for at least 1 year after diagnosis. Subjects diagnosed with type 2 diabetes usually sustained considerable insulin secretion abilities and did not require insulin treatment, and continued adequate glycemic control under dietary management, physical exercise or by using some oral hypoglycemic drugs within at least 3 years after diagnosis. Then, some patients subsequently progressed to insulin treatment, use of basal insulin or premixed insulin, to achieve optimal glycemic control. Besides, all the patients had no evidence of autoimmunity to pancreatic β-cells, without showing positive results in any β-cell associated autoantibodies [5-7].

We previously reported that a small number of school children with slowly progressive form of type 1 diabetes were also detected by the screening program [5,10,11]. In Japan, there is clinical heterogeneity of youth-onset type 1 diabetes, including abrupt-onset type 1 diabetes, and slowly progressive type 1 diabetes. The latter type exhibits a milder onset of the disease without showing ketoacidosis at the time of diagnosis. This novel type is generally encountered in adults and has been referred to as latent autoimmune diabetes in adults [12,13]. Patients with slowly progressive type 1 diabetes require insulin treatment at least within 1.5 to 3 years from the diagnosis, and most of them have positive β-cell associated autoantibodies [5,10,11]. On the other hand, patients diagnosed with type 2 diabetes have no evidence of β-cell associated autoimmunity.

Monogenic diabetes, such as mitochondrial diabetes and maturity-onset diabetes in the young (MODY), can be detected by the screening program [9]. They are tested for mitochondrial DNA mutations and MODY related genes, and are eventually identified as monogenic diabetes.

2. Frequency and clinical features including age, sex, and family history in nonobese type 2 diabetes

Among the 301 students diagnosed with type 2 diabetes by the screening program during 1975–2015, 39 (13.0%) were identified as nonobese type 2 diabetes, with the BMI less than the 90th percentile for sex- and age-matched Japanese children [9]. Age at the diagnosis was 12.2±2.0 years, which was not different from that in obese-patients (12.8±3.0 years). On the other hand, of note, the percentage of females was significantly greater among the non-obese patients than the obese-patients (84.6% vs. 49.2%, P<0.005). This finding suggests that femaleness might play a role in developing subtype of type 2 diabetes. Besides, a positive family history in the first- and second-degree relatives was in 61.5% of the nonobese patients, which was not different from that in obese-patients (65.6%). Interestingly, 75.0% of the relatives of the nonobese patients had a similar diabetic type; i.e., nonobese type 2 diabetes. Therefore, there might be a strong hereditary component, likely a multigenetic determinant, in the development of nonobese type 2 diabetes.

3. Birth weight in nonobese type 2 diabetes

We did not analyze the data regarding birth weight in patients diagnosed with type 2 diabetes by the screening program during 2009–2015. Besides, the data from 1975 to 2008 was analyzed, and we found a significant high frequency of small for gestational age (SGA) in non-obese patients [14]. The mean birth weight of the nonobese patients was not different from that of obese-patients (2,865±480 g vs. 3,002±250 g). On the other hand, frequencies of SGA and large for gestational age in nonobese patients were 20.8% and 8.3%, respectively, and those in obese patients were 8.5% and 10.3%, respectively. The percentage of SGA of the nonobese patients was much greater than that recorded in the Japanese general population (6.3%, P<0.0001) (Fig. 2). Sugihara et al. [15] also reported that the frequencies of both low and high birth weights were higher in patients with type 2 diabetes than in Japanese control subjects. These findings suggest possibility of an association between intrauterine growth failure and the risk of type 2 diabetes, particularly in nonobese type 2 diabetes.

Fig. 2.

Birth weight in children with nonobese type 2 diabetes. SD, standard deviation.

4. Laboratory data at diagnosis in nonobese type 2 diabetes

The data from 1974 to 2008 also showed comparison of laboratory data at the time of diagnosis between nonobese patients and obese-patients as indicated Table 1 and Fig. 3 [14]. There was no significant difference in the mean levels of fasting plasma glucose and glycosylated hemoglobin, expressed as the National Standardization Program value, between the two groups. However, nonobese patients had significantly lower mean values of fasting plasma immunoreactive insulin (IRI) and homeostasis model assessment of insulin resistance than obese patients. Moreover, they also showed significantly decreased mean value of homeostasis model assessment of β-cell function than obese patients, and impaired first phase insulin secretion response, expressed as insulinogenic index (calculated by ΔIRI (30 minutes on OGTT)/ΔPG (30 minutes on OGTT), of less than 0.4. However, we also found that nonobese patients showed significantly higher insulin resistance as compared with age- and pubertal stage-matched Japanese people without diabetes in other studies (Fig. 4) [16,17]. Accordingly, nonobese patients show lower insulin secretion abilities, but have milder, but evident, insulin resistance even from the time of diagnosis.

Table 1.

Laboratory data of nonobese and obese patients with type 2 diabetes at the time of diagnosis

Fig. 3.

Laboratory data at diagnosis in children with nonobese type 2 diabetes. IRI, immunoreactive insulin; HOMA-IR, homeostasis model assessment of insulin resistance; PG, plasma glucose.

Fig. 4.

Comparison of fasting IRI levels and HOMA-IR at diagnosis among children with type 2 diabetes of different % overweight. IRI, immunoreactive insulin; HOMA-IR, homeostasis model assessment of insulin resistance. % overweight: ^aNonobesity: n=17, <20%; Mild obesity: n=37^b, n=30^c; ^dSevere obesity:n=26.

5. Accumulation of visceral adiposity-possible cause in nonobese type 2 diabetes

Yoon et al. [18] reported that even though the average BMI of Asians was lower than that in White Caucasian descents, the prevalence of type 2 diabetes was higher in Asians, possibly due to more accumulation of visceral adiposity. Some previous studies also discussed that Asians might have a greater amount of visceral adiposity than Caucasians [19,20]. Then, we examined the ratio of a visceral fat area to a subcutaneous fat area in some patients with obese and nonobese type 2 diabetes. It is notable that some nonobese patients have a large amount of visceral fat areas, even though their body weights and subcutaneous fat areas reveal normal for their age and sex, and show higher ratio of a visceral fat area to a subcutaneous fat area (Fig. 5). Therefore, it is necessary to evaluate the amounts of visceral adiposity, other than subcutaneous adiposity, which could more precisely reflect insulin resistance in patients with nonobese type 2 diabetes.

Fig. 5.

Distribution of subcutaneous fat (SF) and visceral fat (VF) in patient with nonobese type 2 diabetes. WC, waist circumference; BMI, body mass index; SF, subcutaneous fat; VF, visceral fat.

6. Treatment option

Dietary management and physical exercise are principal in treatment of type 2 diabetes. Most patients with obese-type 2 diabetes can improve glycemic control by reducing their body weight [8,21,22]. However, some patients fail to improve glycemic control even though therapeutic modification of the diet and exercise habits. Particularly, patients with nonobese type 2 diabetes tend to progress earlier to pharmacological treatment using oral hypoglycemic agents (OHA) or insulin. We found over 95% of patients with nonobese type 2 diabetes eventually progressed to pharmacological treatment: i.e., 45.8% treated with OHA and 50.0 % with insulin [14]. The mean interval to start of pharmacological treatment was 3.1±2.3 years (2.2±2.1 years for OHD and 4.8±2.9 years for insulin). Fig. 6 shows various kinds of pharmacological treatment used in patients with obese and nonobese type 2 diabetes. Metformin should be a first-line drug, particularly in obese patients. Sulfonyl urea (SU), glimepiride, is used in mostly nonobese patients. Alpha-glucosidase inhibitors, dipeptidyl peptidase-4 inhibitors and glucagon-like peptide-1 receptor agonists come to be used in both types of diabetes. New drugs, SGLT2 inhibitors hereafter may be used mainly among obese patients [22]. However, only metformin and glimepiride are currently permitted for official use in pediatric patients. Therefore, it is necessary to get informed consent from the patients and their family members and be careful for occurrence of the adverse events when use of new medications other than metformin and glimepiride.

Fig. 6.

Pharmacological treatment in patients with nonobese and obese type 2 diabetes. SU, sulfonyl urea; α-GI, α-glucosidase inhibitors; DPP-4, dipeptidyl peptidase-4 inhibitors; GLP-1, inhibitors and glucagon-like peptide-1 receptor agonists.