Etiologic spectrum and clinical characteristics of pediatric diabetes among 276 children and adolescents with diabetes in a single academic center

The prevalence of monogenic diabetes is estimated to be 1–5% of patients with diabetes mellitus (DM). The overlapping clinical features of various forms of diabetes make differential diagnosis challenging. Therefore, this study investigated the etiologic distribution and clinical characteristics of pediatric diabetes, including monogenic diabetes, in a single tertiary center over the last 20 years. Methods This study included 276 consecutive patients with DM diagnosed before 18 years of age from January 2000 to December 2019. Clinical features, biochemical ndings, β-cell autoantibodies, and molecular characteristics were reviewed retrospectively. This study showed that genetically conrmed monogenic diabetes accounts for 5.1% of patients referred to pediatric endocrinology clinics. The frequency of mutations in the major genes of MODY is low among pediatric patients in Korea. Healthcare providers should be consider that diabetic patients with a family history or extra-pancreatic features without β-cell autoantibodies might have monogenic diabetes.

Monogenic diabetes is a single gene disorder caused by mutations in one of more than 20 genes that control either the secretion or action of insulin [5]. The prevalence of monogenic diabetes is estimated to be 1-6% among children and adolescents with DM in Europe [6][7][8][9][10][11].
It is important to make an accurate etiologic diagnosis of DM considering that it can affect the therapeutic decisions, the prognosis of chronic complications, and genetic counseling [12]. However, the overlapping clinical features of various forms of DM make it di cult to perform differential etiologic diagnosis. Information on the distribution of the type of DM and the prevalence of monogenic diabetes has not been established in the Korean pediatric population. A few studies of MODY in Korea demonstrated that mutations in the major MODY gene were rare including some cases with GCK mutations [13][14][15][16]. Therefore, this study was performed to investigate the etiologic distribution and clinical characteristics of pediatric diabetes including monogenic diabetes in a single tertiary center over the last 20 years. Clinical and endocrine characteristics such as the patients' age at diagnosis, gender, C-peptide level, presence of βcell autoantibodies [glutamic acid decarboxylase (GAD) antibody and insulin autoantibody], and presence of diabetic ketoacidosis (DKA) at diagnosis were retrospectively reviewed.
Type 1 DM was diagnosed in the presence of low C-peptide level, β-cell autoantibody positivity, and the absence of acanthosis nigricans or any extra-pancreatic features suggesting monogenic diabetes. The diagnosis of type 2 DM was based on clinical ndings such as a family history of type 2 DM, obesity, and signs of insulin resistance (acanthosis nigricans). The patients were categorized as having monogenic diabetes when they showed clinical features such as a family history of DM, absence of β-cell autoantibodies, normal or high C-peptide levels, a lowdose insulin requirement, and signs of extra-pancreatic features [12].
Four patients with monogenic diabetes previously reported by our group [17][18][19][20] were included to delineate the clinical and molecular spectrum of monogenic diabetes in our cohort. This study was approved by the Institutional Review Board of Asan Medical Center, Seoul, Korea (IRB No. 2020 − 0667). The blood samples for DNA analysis were collected after obtaining written informed consent from the patients or their parents. In pediatric patients aged 7-17 years, the informed consent was obtained from both a parent and the patients.

Molecular analysis for patients with monogenic diabetes
Molecular analysis was performed by Sanger sequencing for patients with a clinical diagnosis of monogenic diabetes according to the patients' phenotype. Sanger sequencing of the major genes (HNF1A, HNF4A, and GCK genes) involved in maturity-onset diabetes of the young (MODY) was performed for cases with clinically suspected MODY: 1) a family history of diabetes in one parent and rst-degree relatives of that affected parent; 2) negative autoantibodies; 3) lack of the characteristics of type 2 DM (obesity and acanthosis nigricans); and 4) good metabolic control with diet, sulfonylurea, or low-dose insulin [12]. In a case with defects of the urogenital tract, HNF1B was analyzed. DNA sequencing of INSR was performed for a patient with severe insulin resistance.
For patients with MODY, whole exome sequencing was performed when there were no rare sequence variants in the major MODY genes. SureSelect Human All Exon V5 (Agilent Technologies, Santa Clara, CA, USA) was used for library preparation. Sequencing was performed using the NextSeq500 platform (Illumina Inc., San Diego, CA, USA), generating 2 × 150 bp paired-end reads. The sequence reads were aligned to the human reference genome (hg19) using the Burrow-Wheeler Alignment program (BWA version 0.7.12). SAMtools 0.1.19 and Genome Analysis Toolkits (GATK version 3.5) were used for single nucleotide polymorphism variant calling from aligned sequence reads. GATK version 3.5, FreeBayes 0.9.2.1, and Scalpel 0.5.3 were used for insertion-deletion variant calling. After removing duplicates with PICARD (version 1.96, https://broadinstitute.github.io/picard/), annotation was performed using Variant Effect Predictor [21] and dbNSFP [22]. The sequence variants of known genes previously associated with MODY were scanned [23]. All sequence variants were classi ed as pathogenic, likely pathogenic, variant of uncertain signi cance (VUS), likely benign, or benign, in accordance with the guidelines of the American College of Medical Genetics Laboratory Practice Committee Working Group [24].

Results
Clinical and molecular characteristics of patients with type 1 and type 2 DM Of the 276 patients, 206 patients (74.6%) were diagnosed with type 1 DM, 49 patients (17.8%) were diagnosed with type 2 DM, and 21 patients (7.6%) were diagnosed with clinically suspected monogenic diabetes. The clinical characteristics at diagnosis of type 1 DM, type 2 DM, and monogenic diabetes are shown in Table 1.  The patient was initially misdiagnosed as having type 1 DM; however, insulin therapy was successfully switched to oral sulfonylurea therapy.
Wolfram syndrome. Two male siblings were initially misdiagnosed as having type 1 DM at the age of 4.9 and 6.1 years, respectively, until urinary incontinence and bilateral optic nerve atrophy occurred. They were compound heterozygous for a known pathogenic c.2171C > T (p.P724L) mutation [26] and a novel c.1725_1742del (p.G587_G592del) mutation in WFS1.
IPEX syndrome. The patient was initially misdiagnosed as having type 1 DM at the age of 11 months. However, he showed unusual clinical features including pure red cell aplasia and membranous glomerulopathy at the age of 39 months, and posterior reversible encephalopathy syndrome after a vaccination against in uenza A (H1N1) virus at the age of 11 years. DNA analysis of the FOXP3 gene identi ed a splice site mutation of c.201 + 1G > A, which was inherited from his mother [17].
Fanconi-Bickel syndrome. A 3-day-old female presented with hyperglycemia, glycosuria, and galactosemia. She failed to grow during the follow-up period, and her liver function deteriorated with micronodular cirrhosis and marked fatty changes on liver biopsy. Homozygous mutation of c.13A > T (p.K5*) in exon 1 was identi ed by Sanger sequencing of SLC2A2 [20].
Wolcott-Rallison syndrome. A 3-month-old Arab girl presented with neonatal DM, epiphyseal dysplasia, and liver failure and was diagnosed with Wolcott-Rallison syndrome caused by the pathogenic homozygous mutation of c.1293G > A (p.W431*) in EIF2AK3, which was previously reported to be pathogenic [27].
Cystic brosis-related diabetes. Cystic brosis-related diabetes was detected in two patients with CFTR mutations. They showed pancreatitis-related hyperglycemia at the age of 7.7 and 14.2 years, respectively, without β-cell autoantibodies.
Maternally inherited diabetes and deafness (MIDD). Three patients with m.3243A > G mutation in MTTL1 were diagnosed as having MIDD with an HbA1c level of 8.2 ± 2.8% without β-cell autoantibodies. All patients have been treated with insulin injection.

Discussion
This study described the clinical characteristics of different types of DM based on the experience of a single tertiary center over the last 20 years. Type 1 DM accounted for most cases (74.6%) of DM in our cohort, followed by type 2 DM (17.8%). Genetic etiologies were con rmed in 14 patients (5.1%) from 13 families. Diverse genetic etiologies underlie in pediatric monogenic diabetes, and extra-pancreatic features are an important clue to the diagnosis of monogenic diabetes.
The frequencies of type 1 DM, type 2 DM, and MODY were 85.6%, 10.8%, and 1.2%, respectively, in the SEARCH study (USA); on the other hand, these ratios were 95.5%, 1.3%, and 1.5%, respectively, in the SWEET study (Europe) [28][29][30]. The diagnosis of type 2 DM (17.8%) and genetically con rmed monogenic diabetes (5.1%) was more common in the present study compared with the previous studies. The variation in the frequencies could be explained by the availability of genetic testing and the prevalence of obesity in the region. Antibody positivity has been reported in up to 15% of patients with type 2 DM, and these autoantibody-positive patients are usually younger, less overweight/obese, and have higher hemoglobin A1c levels [31]. However, there were no signi cant differences in our study.
Monogenic diabetes comprises various phenotypes including neonatal DM, MODY, and rare syndromic diabetes with extra-pancreatic features including neurological, renal, intestinal, or skeletal abnormalities [5]. Therefore, monogenic diabetes might be initially misdiagnosed as type 1 or type 2 DM before the manifestation of extrapancreatic features, as in the case of the patients with Wolfram syndrome, DEND syndrome, and IPEX syndrome in our study. Establishing the etiology of DM is important for therapeutic strategies, the prognosis of chronic complications [12], and appropriate genetic counseling for monogenic diabetes [5]. For example, the molecular diagnosis of monogenic diabetes can lead to changes in treatment often with improved glycemic control as some patients with monogenic diabetes carrying mutations in speci c genes (e.g., HNF1A, HNF4A, KCNJ11, and ABCC8) can be treated with oral sulfonylureas instead of insulin [32].
MODY is de ned as an autosomal dominantly inherited familial form of non-autoimmune diabetes due to a primary defect in pancreatic β-cell function with an age of onset before 25 years [33]. Mutations in 14 different genes are known to be associated with MODY [https://www.omim.org/entry/606391, accessed on August 2020]. Among them, mutations in HNF4A, HNF1A, and GCK are the most common causes of MODY [34]. The prevalence of MODY has been estimated to be 1-2% of cases of DM [35]. In an Italian study, MODY represented the second most prevalent cause (5.5%) of DM after type 1 DM; however, mutations in MODY-related genes were documented in approximately 1.85% of patients [7]. In Asian countries, sequence variants in MODY genes were identi ed in 15-19% of patients with clinically diagnosed MODY in India [36,37]. Among them, HNF1A or ABCC8 mutations were the most common [37]. The frequency of mutations in the major MODY genes (HNF4A, GCK, and HNF1A) has been shown to be extremely low among Korean patients with MODY [15]. However, with the advent of nextgeneration sequencing, genetic diagnostic rate of MODY has been increased. In 28 patients with early onset diabetes in Korea, four pathogenic or likely pathogenic variants were identi ed in three patients by whole exome sequencing [38]. In targeted panel sequencing, molecular genetic diagnosis was possible in 21.1% of patients among 109 patients with clinically suspected monogenic cause of diabetes [39].
The prevalence of MODY is dependent on the active referral of patients with highly probable MODY, indicating that some patients with MODY remain underdiagnosed [40]. Childhood type 2 DM can be confused with MODY due to a family history, presenting features, and a possible confounding factor of obesity/overweight [9]. Furthermore, MODY, especially in cases involving HNF1A mutations, can be misclassi ed as type 1 DM [7].
A limitation of this study is that it is not a national multicenter study, and it included data from a single tertiary center. Therefore, the frequency of speci c types of DM may not re ect the actual frequency. As this study investigated etiologic distribution of DM in pediatric patients and included consecutive patients under 18 years of age at diagnosis who were referred to a pediatric endocrinology clinic in a single institute, the actual frequency of DM including MODY could not be estimated because of sampling bias.

Conclusions
This study showed that genetically con rmed monogenic diabetes accounts for 5.1% of patients referred to pediatric endocrinology clinics. The frequency of mutations in the major genes of MODY is low among pediatric patients in Korea. Healthcare providers should be consider that diabetic patients with a family history or extrapancreatic features without β-cell autoantibodies might have monogenic diabetes. . Sanger sequencing and whole exome sequencing were performed after obtaining written informed consent from patients or their parents. In pediatric patients aged 7-17 years, the informed consent was obtained from both a parent and the patients. When the clinical data were obtained retrospectively without molecular analysis, informed consent was waived by the IRB.

Consent for publication
Not applicable Availability of data and materials Not applicable.

Competing interests
The authors declare that they have no competing interests.

Funding
Whole exome sequencing in this study was supported by a grant (2018-481) from Asan Medical Center Children's Hospital, Seoul, Korea.
Author contributions JHK and JHC designed the study and wrote the manuscript; YL, YC, GHK, and HWY collected the data and involved in data analysis. All the authors made nal corrections and approved the manuscript.