Aetiological heterogeneity of asymptomatic hyperglycaemia in children and adolescents
- First Online:
- Cite this article as:
- Feigerlová, E., Pruhová, Š., Dittertová, L. et al. Eur J Pediatr (2006) 165: 446. doi:10.1007/s00431-006-0106-3
- 114 Views
Randomly estimated fasting hyperglycaemia in an asymptomatic individual may represent the first sign of pancreatic β-cell dysfunction.
We aimed at specifying the genetic aetiology of asymptomatic hyperglycaemia in a cohort of children and adolescents.
Subjects and methods
We analysed the aetiological diagnosis in 82 non-obese paediatric subjects (38 males) aged 0.2-18.5 years (median: 13.1) who were referred for elucidation of a randomly found blood glucose level above 5.5 mmol/l. In addition to fasting glycaemia and circulating levels of insulin and C-peptide, the subjects were tested by an oral glucose tolerance test and an intravenous glucose tolerance test and screened for mutations in the genes encoding glucokinase (GCK), HNF-1α (TCF1), Kir6.2 (KCNJ11) (if aged <2 years) and HNF-4α (HNF4A) (those with a positive family history of diabetes).
Results and discussion
We identified 35 carriers of GCK mutations causing MODY2, two carriers of TCF1 mutations causing MODY3, one carrier of a HNF4A mutation causing MODY1 and one carrier of a KCNJ11 mutation causing permanent neonatal diabetes mellitus. Of the remaining patients, 11 progressed to type 1 diabetes mellitus (T1DM) and 9 had impaired glucose tolerance or diabetes mellitus of unknown origin. In 23 subjects, an impairment of blood glucose levels was not confirmed. We conclude that 39 of 82 paediatric patients (48%) with randomly found fasting hyperglycaemia suffered from single gene defect conditions, MODY2 being the most prevalent. An additional 11 patients (13%) progressed to overt T1DM. The aetiological diagnosis in asymptomatic hyperglycaemic children and adolescents is a clue to introducing an early and effective therapy or, in MODY2, to preventing any future extensive re-investigations.
KeywordsHyperglycaemiaGeneticsChildrenMODYType 1 diabetes mellitusPermanent neonatal diabetes mellitus
First-phase insulin release
Gene encoding glucokinase
Hepatocyte nuclear factor-1α
Gene encoding HNF-4α
Hepatocyte nuclear factor-4α
Hepatocyte nuclear factor-1
Impaired glucose tolerance/diabetes mellitus
insulin promotor factor
intravenous glucose tolerance test
Gene encoding Kir6.2
Inwardly rectifying K+ channel subunit
Maturity-onset diabetes of the young
Normal glucose tolerance
Oral glucose tolerance test
Permanent neonatal diabetes mellitus
Standard deviation score
Type 1 diabetes mellitus
Gene encoding HNF-1α
For decades, the diagnosis of paediatric diabetes mellitus has being considered trivial. In the majority of patients, suggestive symptoms of recent polyuria, polydipsia and weight loss, in some cases associated with ketoacidosis, clearly indicate the need for blood glucose measurement to establish the diagnosis of type 1 diabetes mellitus (T1DM).
However, an unexpected finding of elevated blood glucose may arise from a random measurement in children without typical symptoms of diabetes, while elaborating various medical conditions; in others, a positive dipstick test for glycosuria may have led to a subsequent estimation of hyperglycaemia. As these children may suffer from presymptomatic progressive pancreatic β-cell dysfunction, a rapid and effective diagnostic action is required.
Maturity-onset diabetes of the young (MODY) is a family of monogenic forms of impaired β-cell function. The clinical diagnosis of MODY is based on (1) young age at onset (before 25 years of age), (2) familial occurrence with autosomal dominant inheritance and high penetrance and (3) no need for insulin treatment for at least 2 years following diagnosis [20, 22]. So far, six distinct MODY subtypes (MODY1–MODY6) have been defined according to the underlying genetic defect.
The additional MODY subtypes (MODY1 and 3–6) result from defective β-cell transcriptional regulation (Fig. 1). The affected individuals usually manifest in late puberty or early adulthood and suffer from progressively impaired insulin secretion and impaired glucose regulation and a high risk of late diabetes-associated complications.
Neonatal diabetes mellitus (either transient or permanent) is characterised by hyperglycaemia revealed within the first months of life requiring insulin treatment [7, 12]. Transient neonatal diabetes mellitus resolves within a median of 3 months . On the contrary, patients with permanent neonatal diabetes mellitus (PND) remain insulin dependent . A defect of KCNJ11 encoding the Kir6.2 subunit of the β-cell ATP-sensitive K+ channel has recently been established as a cause for PND in a substantial proportion of affected children (Fig. 1) .
To make the spectrum of diabetic conditions among children and adolescents even more complex, type 2 diabetes mellitus is recently being reported among severely overweight young people from countries with an epidemic of obesity .
Here, we studied mutations and the phenotypic expression in the genes TCF1, GCK, HNF4A and KCNJ11 in an unselected cohort of 82 children and adolescents, consecutively referred for investigation of asymptomatic fasting hyperglycaemia.
Subjects and methods
Between January 1998 and December 2004, a total of 82 children and adolescents (38 males, 44 females; aged 2 months–18.5 years, median: 13.1 years) were referred by general paediatricians, paediatric endocrinologists or paediatric departments of local hospitals to the Department of Paediatrics of the 3rd Medical Faculty in Prague for elucidation of asymptomatic fasting hyperglycaemia. All patients were of Caucasian origin. Their body mass index (BMI) ranged between 13.7 and 25.0 kg/m2 (median: 18.9). None of the subjects was severely overweight (BMI >97th percentile of the age- and gender-matched reference population) .
Non-symptomatic elevated fasting blood glucose was originally estimated either within the elaboration of an acute condition (tonsillitis, gastroenteritis/vomiting, bronchitis, influenza, pyelonephritis, otitis media, fatigue, abdominal pain, head injury, vertigo, dyspnoea, collapse or tachycardia) or following a positive glycosuria testing at a routine preventive examination or at urine examination for various other reasons. None of the patients suffered from polyuria/polydipsia/weight loss or was ketotic at initial evaluation.
The age of patients at the first recognition of hyperglycaemia was 2 weeks–16.1 years (median: 12.0 years) and fasting plasma glucose level at the referring physician’s office ranged from 5.6 mmol/l (101 mg/dl) to 20.9 mmol/l (376 mg/dl) (median: 7.1 mmol/l; 127 mg/dl). Those with fasting plasma glucose <5.6 mmol/l (101 mg/dl) at initial examination were excluded from the study cohort. At the first examination following referral, levels of fasting plasma glucose ranged from 4.0 mmol/l (71 mg/dl) to 22.6 mmol/l (407 mg/dl) (median: 5.9 mmol/l; 106 mg/dl).
The baseline evaluation included family history, fasting glycaemia, fasting serum insulin and C-peptide, and glycosylated haemoglobin (HbA1C). Family history was considered positive if at least one parent and/or one sibling had diabetes mellitus.
All individuals were tested with an oral glucose tolerance test (OGTT) which was evaluated according to the American Diabetes Association (ADA) criteria . Children with results within normal range (fasting plasma glucose <5.6 mmol/l and 2-h postload plasma glucose <7.8 mmol/l; <140 mg/dl) were considered to have normal glucose tolerance (NGT). Intravenous glucose tolerance test (IVGTT) to assess pancreatic β-cell function by estimating the first-phase serum insulin release (FPIR) was performed in all children above 2 years of age. All subjects were investigated for mutations in the genes encoding GCK and HNF-1α. The KCNJ11 gene encoding the Kir6.2 subunit of the β-cell ATP-sensitive K+ channel was analysed in all children younger than 2 years.
In addition patients with a positive family history of diabetes mellitus in first-degree relatives but with negative search for mutations in GCK, TCF1 and KCNJ11 genes were screened for mutations in the gene encoding HNF-4α.
Low-dose insulin therapy and home blood glucose monitoring were initiated in patients who had abnormal glucose tolerance (a pathological result during an OGTT) in association with decreased FPIR (lower than 1st percentile). Although that is still not a standard therapeutic option, we personally believe that it may prevent a rapid progression to overt T1DM.
Clinical follow up
All study participants were followed prospectively. The median follow-up time was 3.9 years (range: 0.7–6.9 years). Venous blood was sampled for HbA1C and profiles of blood glucose levels over 24 h (three to five measurements before main meals and at bedtime) were performed every 6 months.
In those on insulin therapy, the treatment was aimed at maintaining near-normoglycaemia. The insulin requirements were recorded at regular outpatient visits every 3 months.
Informed written consent was obtained from all subjects and/or their parents before entering the study protocol. The study was approved by the Ethical Committee of the 3rd Faculty of Medicine, Charles University of Prague.
Both OGTT and IVGTT  were performed according to standard protocols. The subjects were on regular diet with unrestricted carbohydrate intake at least 3 days preceding the test. Excessive physical activity was not allowed 1 day before testing. The test was not provided in cases of acute illness and/or administration of drugs with potential effect on blood glucose levels (including inhaled corticosteroids). After an overnight fast for 10–12 h, testing was initiated between 8 and 9 a.m.
For IVGTT, two contralateral antebrachial veins were cannulated. Sampling was performed from one cannula to measure basal plasma glucose and serum insulin and C-peptide. Immediately thereafter, 0.5 g glucose per kg of body weight (maximum: 35 g) as a 40% aqueous solution was infused into the second cannula within 3 min±15 s. Serum insulin and C-peptide levels at time points 1 and 3 min were used for calculation of the first-phase insulin response (FPIR). The results were evaluated according to published standards [9, 18].
For OGTT, 1.75 g glucose per kg of body weight (maximum: 75 g) was given. Blood samples to measure plasma glucose and serum insulin and C-peptide were obtained at time points 0, 60 and 120 min. The results were evaluated according to ADA criteria .
Routine laboratory assays
Plasma glucose concentration was measured by the enzymatic hexokinase method using the automatic analyser Konelab 60 (Thermo Clinical Labsystem Oy, Espoo, Finland).
C-peptide and insulin
C-peptide and insulin in serum were analysed by a chemiluminescent immunometric technique using the commercial sets Immulite 2000 C-peptide and Immulite 2000 Insulin (Diagnostic Products Corporation, Los Angeles, CA, USA).
Estimation of HbA1C was performed with the DS5 Analyser (Drew Scientific Ltd., Barrow in Furness, Cumbria, UK) using cation exchange chromatography in conjunction with gradient elution. The assigned values of HbA1C were calibrated to the International Federation of Clinical Chemistry (IFCC) system (normal levels: 2.0–4.5%).
Preparation of genomic DNA
Genomic DNA was isolated from leukocytes in blood samples anticoagulated with ethylenediaminetetraacetate (EDTA).
Analyses of genes encoding HNF-1α, GCK and HNF-4α
Denatured high-performance liquid chromatography (dHPLC) and direct sequencing were used for analysis of all exons, the intron-exon boundaries and the promoter regions of the TCF1 and GCK genes . Analysis of the HNF4A gene and of its P1 promoter was performed by direct sequencing using ABI PRISM Dye Primer Cycle Sequencing Kit with AmpliTaq DNA polymerase FS.
Analysis of the KCNJ11 gene
The published primers and previously described protocol were used to carry out the polymerase chain reactions (PCR) in addition to fragment 6 for which the annealing temperature used was 70°C . PCR were performed using AmpliTaq Gold and a Gene-Amp PCR system 9700 thermocycler (Perkin Elmer, Foster City, CA, USA). After PCR, the products were purified using an ExoSAP-IT treatment (USB Corporation, Cleveland, OH, USA), and all of them were sequenced in both directions using the BigDye Terminator v3.1Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). The sequencing was performed on an ABI PRISM 3100-Genetic Analyser (Applied Biosystems, Foster City, CA, USA).
Definition of diagnostic subgroups
Positive screening for mutations in the HNF4A, GCK, TCF1 and KCNJ11 genes was considered diagnostic for MODY1, MODY2, MODY3 and PND, respectively.Subjects with negative search for mutations in genes mentioned above, but with persistent increased fasting glycaemia and abnormally high plasma glucose response to an OGTT were considered diabetic. Of these, individuals with FPIR below the 1st percentile who required a daily insulin dose ≥0.5 IU/kg to maintain near-normoglycaemia within the follow-up period were considered to manifest an early (pre-manifest) phase of T1DM. Those with normal FPIR (and no insulin treatment) or those with low FPIR but daily insulin requirements <0.5 IU/kg were assigned as impaired glucose tolerance/diabetes mellitus (IGT/DM) of undetermined origin. β-Cell-specific autoantibodies were not included into the diagnostic work-up. Individuals negative for mutations in the analysed genes who had normal fasting plasma glucose and normal OGTT and physiological FPIR at re-evaluation following referral were considered as NGT.
The values are expressed as mean±SEM if not given otherwise. The data on diagnostic subgroups were statistically evaluated if the entire subgroup included ≥9 individuals. The group differences in the clinical and metabolic variables were assessed by the one-way analysis of variance (ANOVA) and by Student’s t-test where appropriate.
Monogenic forms of hyperglycaemia
Genetic findings: 39 heterozygous mutations identified within the cohort of 82 unrelated children and adolescents with randomly found hyperglycaemia
HNF4A (1 subject)
GCK (35 subjects)
Arg36Trp (1 subject)
Glu40Lys (8 subjects)
Gly44Asp (2 subjects)
Phe150Leu (1 subject)
Glu157Lys (1 subject)
Ala188Thr (1 subject)
Cys220Stop (4 subjects)
Val226Met (2 subjects)
Met251Val (1 subject)
Glu268Stop (1 subject)
Gly294Asp (1 subject)
Leu315His (2 subjects)
Phe316Val (1 subject)
Gly318Arg (3 subjects)
Ser383Leu (1 subject)
Phe419Leu (1 subject)
Cys434Tyr (2 subjects)
Ile436Asn (1 subject)
Ala454Glu (1 subject)
TCF1 (2 subjects)
KCNJ11 (1 subject)
One patient carried a heterozygous mutation in the KCNJ11 gene, an arginine-to-histidine substitution at position 201 (R201H), causing PND. He was originally investigated for prolonged cough at age 2.5 months. In spite of a blood glucose level of 22.6 mmol/l (407 mg/dl) at referral, he was free of diabetic symptoms. We have not previously reported this patient; however, the R201H mutation is known to be the most prevalent PND-causing variant within the entire KCNJ11 gene .
Clinical and metabolic characteristics of diagnostic subgroups of children with randomly found asymptomatic hyperglycaemia
Age at first detection of hyperglycaemia (years)
Fasting p-glucose (mmol/l)
Fasting s-insulin (mIU/l)
Fasting s-C-peptide (pmol/l)
OGTT: 1-h postload p-glucose (mmol/l)
OGTT: 2-h postload p-glucose (mmol/l)
Number of affected parents (2/1/0)
Type 1 diabetes mellitus (pre-manifest phase)
Eleven children who tested negative for mutations in the screened genes (13% of the study group) had fasting hyperglycaemia (9.1±1.2 mmol/l; 164±22 mg/dl), abnormally high plasma glucose response to an oral glucose load and substantially decreased FPIR (13.7±3.0 mIU/l) (Table 2). None of them was ketotic at diagnosis or within the follow-up period of 3.8±0.4 years. The reason for the initial plasma glucose estimation was an intercurrent infection in five, fatigue in three and head injury in one subject. In two subjects, glycosuria was detected in a random urine sample at a preventive examination.
In all of the subjects, low-dose insulin therapy was initiated to prevent the development of overt clinical symptoms of diabetes and the daily insulin requirements to maintain near-normoglycaemia exceeded 0.5 IU/kg during the follow-up. Thus, these subjects were considered to suffer from T1DM that was randomly detected within the presymptomatic phase.
Impaired glucose tolerance/diabetes mellitus (IGT/DM) of undetermined origin
Nine patients (11%) who tested negative at mutation screening had an abnormal plasma glucose response to an oral glucose load (OGTT) and a borderline fasting plasma glucose level (6.1±0.5 mmol/l; 110±9 mg/dl). Their FPIR ranged from low normal to moderately decreased values (54.4±19.5 mIU/l) (Table 2). The reason for the initial examination was glucosuria at a preventive examination (4), intercurrent infections (3) or fatigue (2). According to standard procedures, insulin therapy was initiated in those with FPIR below the 1st percentile. However, the daily insulin requirements to maintain near-normoglycaemia remained ≤0.3 IU/kg.
Members of this subgroup were not obese (BMI-SDS ranging from −0.61 to+0.63), making a diagnosis of type 2 diabetes mellitus improbable. However, eight of nine reported a history of diabetes in one parent (three cases were classified as gestational diabetes and five cases as type 2 diabetes).
Normal glucose tolerance
In 23 subjects (28% of the study cohort), we did not confirm the hyperglycaemia originally reported by the referring physician. These patients had normal fasting plasma glucose, OGTT and FPIR. The screening for mutations in the selected genes was negative. Furthermore, HbA1C and profiles of blood glucose levels remained normal during follow-up.
The initial examination of plasma glucose was provided when elaborating an intercurrent infection (11), fatigue (7) or abdominal pain (5).
Summary of clinical laboratory data in diagnostic subgroups
The clinical and laboratory data on subgroups of patients with asymptomatic hyperglycaemia are summarised in Table 2. Individuals with the four most prevalent conditions (T1DM, MODY2, IGT/DM, NGT) were of similar age at first examination for hyperglycaemia, had similar age- and gender-matched body mass index (BMI) and similar fasting serum levels of insulin and C-peptide.
On the contrary, fasting plasma glucose, FPIR and HbA1C differed significantly among the subgroups, distinguishing children with pre-manifest T1DM by lower FPIR and higher fasting plasma glucose in association with increased HbA1C.
Within the cohort of 82 children and adolescents with asymptomatic hyperglycaemia, we identified 35 mutation carriers in the glucokinase gene (MODY2) and 3 patients with mutations in genes encoding transcription factors: one case of MODY1, two cases of MODY3 and one infant with a KCNJ11 mutation causing PND. Thus, in 39 patients (48% of the study cohort) the randomly found hyperglycaemia led to the disclosure of a single gene defect condition, MODY2 being the most common. Among the mutation-negative subjects, 11 (13%) developed T1DM, 9 (11%) had IGT/DM of unknown cause and 23 (28%) were glucose tolerant.
These findings have important clinical implications: 35 of 59 subjects who were confirmed to be hyperglycaemic by re-investigation suffered in fact from MODY2, a benign and non-progressive form of impaired glucose regulation. No diet or drug therapy is required in most of these patients  and an annual follow-up with HbA1C measurement would suffice. The risk of diabetes-associated complications is low . However, affected women may require insulin therapy during pregnancy to prevent foetal macrosomia . Also exact genetic diagnosis is important in order to prevent redundant periodic metabolic examinations of affected individuals.
The clinical diagnosis of MODY2 may be supported by a positive autosomal dominant family history of mild hyperglycaemia. If maternally transmitted, affected women may have a history of gestational diabetes mellitus in all pregnancies. If transmitted by the father, the diagnosis is not necessarily known. A simple estimation of parental fasting plasma glucose levels may be helpful. The affected grandparent may be known to have “mild type 2 diabetes mellitus” and being recommended to follow a “diabetic diet”.
On the contrary, the additional MODY subtypes tend to manifest in later childhood, adolescence or young adulthood  and gradually develop to a symptomatic stage. These forms of diabetes are characterised by progressive decrease in β-cell function and high risk of microvascular complications. Therapy with insulin or oral hypoglycaemic drugs is required to maintain near-normoglycaemia . A positive autosomal dominant family history of a clinically overt diabetes mellitus may help in establishing the clinical diagnosis of MODY.
Among infants, asymptomatic hyperglycaemia may be the first sign of PND. This condition is known to result from defects of the KCNJ11 gene in a substantial proportion of cases . Our patient carrying the R201H mutation within the KCNJ11 gene was randomly detected to be hyperglycaemic at 2.5 months of age. His daily insulin requirements did not exceed 0.4 IU/kg and the metabolic regulation was excellent within the follow-up period of 3.3 years.
Currently, studies are ongoing to test the therapeutic potential of sulphonylurea derivates instead of insulin in children affected by mutations in the KCNJ11 gene. Thus, genetic diagnosis may open new future treatment options in these children .
In 11 patients (13 %), random hyperglycaemia was apparently the first sign of T1DM. Early recognition of T1DM makes it possible to initiate insulin treatment before the clinical onset of the disease and to reduce the risk of unrecognised diabetic ketoacidosis upon manifestation of T1DM. The early stages of the disease process may be detected by decreased FPIR and by β-cell-specific autoantibodies. However, in young children the levels of autoantibodies may vary, introducing difficulties in the interpretation of the results [10, 11]. Temporary positive titres of an autoantibody against molecularly defined antigens (anti GAD65, anti-IA2 or anti-IAA) might also reflect the population variability . Therefore, we performed the study irrespectively of autoimmune markers.
The IGT/DM group in which the molecular pathogenesis was not understood included nine (11%) of the examined subjects. They were all asymptomatic, had impaired glucose tolerance or diabetes mellitus according to diagnostic criteria and their FPIR was low normal or mildly decreased, making the diagnosis of T1DM unlikely. However, eight of them had a first-degree relative affected with diabetes mellitus indicating that undetected or unknown MODY gene variants may be involved in the aetiology. Low-dose insulin therapy was in some of the subjects necessary to achieve near-normoglycaemia.
In the remaining subgroup of 23 subjects (28%) with NGT, a detailed clinical and metabolic examination did not reveal any metabolic disorder. The randomly found fasting hyperglycaemia might have been due to an isolated stress hyperglycaemia, postprandial blood sampling or errors in sample handling.
In conclusion, the underlying cause of a randomly found asymptomatic hyperglycaemia was fully elucidated in a considerable number of affected individuals in our study. For many of them, the final diagnosis bears a positive message on the benign nature of the condition; however, cases of pre-manifest type 1 diabetes mellitus and of MODY1 and MODY3 require a rapid and adequate treatment to prevent an unfavourable short-term and long-term outcome. In different populations, the results might differ largely. Cases of type 2 diabetes mellitus would probably be recognised among hyperglycaemic youngsters in countries with epidemic childhood obesity. However, the spectrum of newly established monogenic conditions is worthy of inclusion in the diagnostic work-up anyway.
The study was supported by a grant of IGA MZ CR No. NB/7420-3 and by a research project of MSM 0021620814. The authors are grateful to all the referring physicians and to the families for an excellent collaboration and to Mrs. Helena Francová for the skilful assistance with metabolic tests.