Ethnicity modifies the relation between fasting plasma glucose and HbA1c in Indians, Malays and Chinese.

AIMS: To study whether HbA(1c) , and its relationship with fasting plasma glucose, was significantly different among Chinese, Malays and Indians in Singapore. METHODS: A sample of 3895 individuals without known diabetes underwent detailed interview and health examination, including anthropometric and biochemical evaluation, between 2004 and 2007. Pearson's correlation, analysis of variance and multiple linear regression analyses were used to examine the influence of ethnicity on HbA(1c) . RESULTS: As fasting plasma glucose increased, HbA(1c) increased more in Malays and Indians compared with Chinese after adjustment for age, gender, waist circumference, serum cholesterol, serum triglyceride and homeostasis model assessment of insulin resistance (P-interaction < 0.001). This translates to an HbA(1c) difference of 1.1 mmol/mol (0.1%, Indians vs. Chinese), and 0.9 mmol/mol (0.08%, Malays vs. Chinese) at fasting plasma glucose 5.6 mmol/l (the American Diabetes Association criterion for impaired fasting glycaemia); and 2.1 mmol/mol (0.19%, Indians vs. Chinese) and 2.6 mmol/mol (0.24%, Malays vs. Chinese) at fasting plasma glucose 7.0 mmol/l, the diagnostic criterion for diabetes mellitus. CONCLUSIONS: Using HbA(1c) in place of fasting plasma glucose will reclassify different proportions of the population in different ethnic groups. This may have implications in interpretation of HbA(1c) results across ethnic groups and the use of HbA(1c) for diagnosing diabetes mellitus.

Development of ipsilateral adrenocortical carcinoma sixteen years after resection of an adrenal tumour causing Cushing's syndrome.

INTRODUCTION: At times, it may be difficult to differentiate early stage, low-grade adrenocortical carcinoma from benign adrenal adenoma. CLINICAL PICTURE: A 53-year-old lady underwent right adrenalectomy for a 4-cm adrenocortical tumour causing Cushing's syndrome. Histology revealed an adrenocortical adenoma. Sixteen years later, she presented with a 14-cm adrenal tumour, again on the right side. TREATMENT: She underwent surgical removal of the tumour. Histology confirmed adrenocortical carcinoma. OUTCOME: She died of metastatic disease 17 months later. CONCLUSIONS: This case highlights the importance of long-term, systematic follow-up of patients treated for benign adrenal adenomas, especially if the tumour size exceeds 4 cm.

HLA microsatellite associations with insulin-dependent diabetes mellitus in Singaporean Chinese.

Singaporean Chinese with insulin-dependent diabetes mellitus (IDDM) have previously been shown to be associated with the DRB1*0301 haplotype and the joint occurrence of DRB1*0301/*0901 and DRB1*0301/*04. The present study extended previous HLA associations by investigating the HLA region using four microsatellites (TNFa, D6S273, TAP1, DQCARII). Seventy-five IDDM patients and 80 healthy controls were studied. TNFa*3 (RR = 2.26), TNFa*12 (RR = 3.30), TAP1*9 (RR = 2.55) showed increased frequencies while TNFa*11 (RR = 0.29), TAP1*4 (RR = 0.50) showed decreased frequencies in patients compared to controls. Linkage analysis suggested that the positive associations of TNFa*3 and TAP1*9 were secondary to that of DRB1*0301. However, TNFa*12 appeared to provide additional risks to IDDM besides the DRB1*0301 haplotype, whereas TNFa*11 and TAP1*4 conferred an independent protective effect against IDDM. Our findings reinforce the notion that susceptibility to and protection against IDDM may include TNF region. In the present study, TNFa*12 seemed to be the primary association in the DRB1*0405 haplotype and may play an independent role in the pathogenesis of IDDM through TNF-alpha function.

Influence of gender and age at onset on the HLA associations in Chinese with insulin-dependent diabetes mellitus.

IDDM in Singaporean Chinese was associated with HLA B58, DRB1*0301, DQB1*0201, and joint occurrences of DRB1*0301/*0901 and DRB1*0301/*04. Of the DR4s the frequencies of DRB1*0401, *0404, and *0405 were higher and *0406 was lower in patients compared to controls. DRB1*0301/*0901 was observed mainly in female patients and the frequency showed an inverse relationship with age at onset, whereas DRB1*0301/*04 was observed mainly in male patients and also showed an inverse relationship with age at onset. DRB1*1202 showed an increasing frequency with age at onset. IDDM patients had a higher frequency of homozygous NAsp57 DQ beta chains and a lower frequency of homozygous Asp57 DQ beta chains compared to controls, especially in younger onset patients.

Non-restricted immunoglobulin-G subclass islet cell antibodies in Chinese.

Islet cell antibodies (ICAs) in Chinese (23 IDDM, 13 NIDDM and 6 non-diabetic) were characterized for immunoglobulin isotypes and light chain specificity. All ICAs were IgG-type and none were IgM- or IgA-type (median titre: 20 JDF units; range 10-160). Light chain specificity showed that 25/36 (69.4%) of the diabetic patients had lambda and kappa chains. Half of the non-diabetic subjects had both lambda and kappa chains. The rest had only lambda chains. Isotyping for ICA-IgG subclass combination with IUIS/WHO reference monoclonal antibodies in the diabetic patients gave the following: IgG1 alone-9 (25%), IgG1+2+3-8 (22.2%), IgG1+2-11 (30.6%), IgG1+3-6 (16.7%), IgG2+3-2 (5.6%). No ICA-IgG4 was detected. The frequency of the subclasses would be: IgG1-94.4%, IgG2-58.3% and IgG3-44.4%. The distribution of ICA-IgG subclasses was not affected by diabetes type (IDDM or NIDDM) or duration of disease. Of the 6 non-diabetic subjects only one had a single ICA-IgG subclass (IgG1). Serum levels of IgG subclasses in a subgroup of the patients (n = 16) were not significantly different from normal individuals. Biochemical modification of pancreatic tissue prior to ICA testing showed that acetylneuraminic acid residues, lipid and protein components were associated with binding of ICAs. The co-existence of other autoantibodies was also tested in these 42 ICA-positive sera. Twelve individuals (1 non-diabetic) had thyroid autoantibodies. Antibodies to thyrotrophin receptor, gastric parietal cell and rheumatoid factor were not detected.(ABSTRACT TRUNCATED AT 250 WORDS)

Tyrosine phosphatase-like protein (IA-2) and glutamic acid decarboxylase (GAD65) autoantibodies: a study of Chinese patients with diabetes mellitus.

Type 1 diabetes in most Asian populations may not have a salient autoimmune basis when assessed with single determinations of the major markers, islet cell antibodies (ICAs) and glutamic acid decarboxylase antibodies (GAD65ab). With the inclusion of antibodies to tyrosine phosphatase-like protein IA-2 (IA-2ab) as an additional major marker, we re-examined autoimmune diabetes in a group of Chinese patients. We studied 272 subjects at various stages of disease with blood samples procured for biochemical analysis. ICAs were measured by immunofluorescence, GAD65ab and IA-2ab by radioimmunoassay. Sixty-seven patients fulfilled clinical diagnosis of type 1 diabetes and the remaining 205 patients were type 2. Prevalence of single autoantibody type in recent-onset type 1 diabetes ( < 1 year duration; n = 47) showed 10.6% with ICAs, 44.7% GAD65ab and 36.2% IA-2ab. GAD65ab account for more than two-thirds of the markers found in type 1 diabetes. Combined analysis further showed that 51.1% had at least one antibody type, 31.9% with two or more antibodies and 8.5% with all three antibodies. Islet autoimmunity presence in childhood-onset type 1 diabetes improved with the addition of IA-2ab, though less impact was seen in the adult-onset. Similarly, combined analysis for type 2 patients with recent diabetes showed a modest increase to 13% with islet autoimmunity compared to 8% when assessed by GAD65ab alone. Combining IA-2ab and GAD65ab assays results detected slightly more immune-mediated diabetes, compared to using a single GAD65ab determination. Non-autoimmune causes need to be considered in the pathogenesis of type 1 diabetes in Chinese, particularly in adults.

Corneal arcus and cardiovascular risk factors in Asians in Singapore.

This study was a cross-sectional random survey of the whole of Singapore, based on 2143 subjects (aged 18-69 years, response rate 60.3%). The presence of corneal arcus was determined by a doctor using the naked eye in good light. Cardiovascular risk factors were measured by standardized techniques. The prevalence rates overall of corneal arcus were: 18-29 years (males 0.5%, females 0.3%), 30-49 years (males 18.1%, females 13.3%) and 50-69 years (males 70.7%, females 55.3%). In the 30-49 age group, people with arcus had higher serum low density lipoprotein (LDL) cholesterol concentrations than people without arcus, the mean differences being, males 0.31 mmol/l (P = 0.040) and females 0.62 mmol/l (P less than 0.001) with an increased likelihood of having values greater than 5.5.mmol/l of males 1.8 (95% confidence interval (95% CI): 1.0-3.4) and females 2.6 (95% CI: 1.4-4.8). There were no significant differences for LDL-cholesterol in the 50-69 age group. Arcus was weakly associated with fasting plasma glucose in the 30-49 age group. Arcus was not associated with serum high density lipoprotein (HDL) cholesterol, serum fasting triglyceride, blood pressure and cigarette smoking. It is concluded that while corneal arcus is primarily an age-related change, its formation is accelerated by high serum LDL-cholesterol so that in people under 50 years it is a marker for the condition.

Cardiovascular diseases in Chinese, Malays, and Indians in Singapore. II. Differences in risk factor levels.

STUDY OBJECTIVE: The aim of the study was to examine cardiovascular risk factors to see how these might explain differences in cardiovascular disease mortality among Chinese, Malays, and Indians in the Republic of Singapore. DESIGN: The study was a population based cross sectional survey. Stratified systematic sampling of census districts, reticulated units, and houses was used. The proportions of Malay and Indian households were increased to improve statistical efficiency, since about 75% of the population is Chinese. SETTING: Subjects were recruited from all parts of the Republic of Singapore. SUBJECTS: 2143 subjects aged 18 to 69 years were recruited (representing 60.3% of persons approached). There were no differences in response rate between the sexes and ethnic groups. MEASUREMENTS AND MAIN RESULTS: Data on cardiovascular risk factors were collected by questionnaire. Measurements were made of blood pressure, serum cholesterol, low and high density lipoprotein cholesterol, fasting triglycerides and plasma glucose. In males the age adjusted cigarette smoking rate was higher in Malays (53.3%) than in Chinese (37.4%) or Indians (44.5%). In both sexes, Malays had higher age adjusted mean systolic blood pressure: males 124.6 mm Hg v 121.2 mm Hg (Chinese) and 121.2 mm Hg (Indians); females 122.8 mm Hg v 117.3 mm Hg (Chinese) and 118.4 mm Hg (Indians). Serum cholesterol, low density lipoprotein cholesterol and triglyceride showed no ethnic differences. Mean high density lipoprotein cholesterol in males (age adjusted) was lower in Indians (0.69 mmol/litre) than in Chinese (0.87 mmol/litre) and Malays (0.82 mmol/litre); in females the mean value of 0.95 mmol/litre in Indians was lower than in Chinese (1.05 mmol/litre) and Malays (1.03 mmol/litre). Rank prevalence of diabetes for males was Indians (highest), Malays and then Chinese; for females it was Malays, Indians, Chinese. CONCLUSIONS: The higher mortality from ischaemic heart disease found in Indians in Singapore cannot be explained by the major risk factors of cigarette smoking, blood pressure and serum cholesterol; lower high density lipoprotein cholesterol and higher rates of diabetes may be part of the explanation. The higher systolic blood pressures in Malays may explain their higher hypertensive disease mortality.

Absence of ion channels CACN1AS and SCN4A mutations in thyrotoxic hypokalemic periodic paralysis.

Muscle weakness in patients with thyrotoxicosis during hypokalemic episodes (thyrotoxic periodic paralysis [TPP]) occurs sporadically and mostly in males. It is treated by infusion or oral supplementation with potassium and with resolution of the thyrotoxicosis state. The clinical features of TPP resemble familial hypokalemic periodic paralysis (hypoKPP), which has been linked to two mutations in the gene encoding the skeletal muscle calcium channel alpha-1 subunit (CACN1AS; Arg528His and Arg1239His) and to the sodium channel alpha-subunit (SCN4A; Arg672His). We screened for the mutations (CACN1AS by polymerase chain reaction-restriction fragment length polymorphism [PCR-RFLP]; SCN4A by single-strand conformation polymorphism analysis) described in hypoKPP in 20 unrelated patients with documented episodes of TPP (mean age, 40.0 +/- 12.3 years 19 males). Forty-eight patients with hyperthyroidism resulting from Graves' disease (48.5 +/- 12.3 years; 13 males), 1 patient with idiopathic hypoKPP (a 32-year-old male) and 32 healthy subjects (41.0 +/- 19.1 years; 16 males) were included. We found none of the TPP patients carry CACN1AS and SCN4A mutations. The hyperthyroid patients and control subjects were also negative for the mutations. The patient with idiopathic hypoKPP was genotyped to have the Arg528His mutation. These results suggest that despite close similarities between TPP and hypoKPP, a likely genetic basis for TPP does not involve the same gene mutations associated with hypoKPP.

Systemic levels of cytokines and GAD-specific autoantibodies isotypes in Chinese IDDM patients.

It is not clear if a Th1/Th2 imbalance in Type 1 diabetes (insulin-dependent diabetes mellitus, IDDM) would lead to a particular antigen-specific IgG subclass dominant as had been shown in the mouse model. In new-onset Type 1 diabetics, an autoantibody response to glutamate decarboxylase (GADab) is frequently observed but the GADab subclass repertoire is not well-established. We determined the systemic levels of representative Th1 and Th2 cytokines and the GADab IgG subclass distribution in 41 Chinese IDDM patients of whom 26 were recently diagnosed (< or = 1 year) and 32 had GADab, to ascertain a likely association of antigen-specific antibody isotype and the Th1/Th2 dichotomy. With high-sensitivity ELISA systems that measure sub-picogram cytokine concentrations, 26 of the 41 patients (63.4%) had at least one of the pro-inflammatory Th1 cytokines (TNF-alpha, IFN-gamma and IL-12) detected. Fewer patients (4/41) had the anti-inflammatory Th2 cytokine IL-4 detected. For IL-10, all subjects had measurable quantities but only three diabetics had levels above the upper limit for healthy subjects (n = 20). Grouped according to the profile of detectable cytokines, there were 24 Th1, 2 Th2 and 2 Th0 patterns. GAD-specific IgG1 antibody was more frequently expressed; 22 of 32 GADab[+] patients. The rank order for the GADab subclasses was IgG1 > 4 > 3 > 2; IgG2 was found in 11 GADab[+] patients. Recent-onset diabetics have a similar ranking of the GAD-specific IgG subclasses. In human Type 1 diabetes, a predominance of GAD-specific IgG1 antibody response is observed together with a dominant Th1 cytokine pattern.

Three new glucose reflectance meters: Diascan, Glucometer II, and Reflolux II.

More and more frequently, diabetics use portable blood glucose meters to assess their own glycemic control. New meters are constantly being introduced commercially. We evaluated three recently marketed meter/strip systems in terms of accuracy, precision, and the stability of the colors on the test strips. We compared the results from the meters with those from a Beckman Glucose Analyser and calculated the correlation and regression coefficients (Diascan, r = 0.93, y = 0.64x + 41.76; Glucometer II, r = 0.94, y = 0.86x + 2.32; and Reflolux II, r = 0.99, y = 0.99x + 5.15). Both the Diascan and Glucometer II meters tended to underestimate blood glucose whereas Reflolux II gave unbiased measurements throughout the clinical range. Reflolux II was also the most precise of the three. Furthermore, the Reflolux test strips kept their color over 6 days of storage at 4 degrees C, while strips from the other two systems faded. All three meters are compact, simple to calibrate, and convenient. They remain, however, relatively expensive, as do the test strips.

Correlates of diabetes markers with erythrocytic enzymes decomposing reactive oxygen species.

Superoxide dismutase (SOD), glutathione peroxidase (GPX) and catalase were assayed in the erythrocytes of a diabetic population on various treatment regimens (diet, oral therapy, and insulin), to investigate any relationships between their activities and diabetes markers (serum glucose, lipids, and fructosamine, as well as glycated haemoglobin). In the group of patients as a whole, there was significant negative correlation of SOD, but not of the other two enzymes with glycated haemoglobin and fructosamine. Specifically, there was a lower activity of the enzyme in the poorly-controlled patients. It is concluded that SOD in particular is potentially an additional marker for long-term diabetic pathophysiology.

Prevalence of ICA and GAD antibodies at initial presentation of type 1 diabetes mellitus in Singapore children.

OBJECTIVE: To determine the prevalence of islet cell antibodies (ICA) and antibodies to glutamic acid decarboxylase (GAD) in Asian children with diabetes mellitus (DM) at the time of diagnosis. PATIENTS AND METHODS: 41 children were studied at their initial presentation from 1993 to 2000. RESULTS: Mean age of onset (+/- SD) of DM was 7.6 (+/- 4.2) years. One or both of the two autoantibodies, ICA and anti-GAD, were present in 17 of the 41 children (41.5%) at the time of diagnosis. Comparing the group of children with autoantibodies and the group without detectable autoantibodies, there were no significant differences in body mass index (15.4 vs. 16.3 kg/m2), age of onset (7.4 vs. 7.8 years), random C-peptide levels (203 vs. 311 pmol/l), HbA1c levels (13.2 vs. 12.7%), and frequency of diabetic ketoacidosis at presentation (53.3 vs. 55%). CONCLUSION: Prevalence of antibodies at presentation of DM in Singapore (41.5%) is lower than in Caucasian populations (60-90%). Other autoantibodies yet to be identified may be contributory. Alternatively, non-immune mediated mechanisms may be responsible for a significant proportion of type 1 DM in Singapore children.

The glucose challenge test: a screening test for gestational diabetes mellitus.

The 50g glucose challenge test (GCT) was evaluated as a method to screen for gestational diabetes in 540 low-risk pregnant women to establish its sensitivity and specificity, as well as to establish a relevant threshold plasma glucose value above which a diagnostic 75g oral glucose tolerance test (OGTT) would be indicated. If a threshold of 140 mg/dl is used, the diagnostic yield would be 28.5%. At a threshold of 130 mg/dl, the diagnostic yield fell to 25.4%; the sensitivity rose to 87.7% and the specificity declined to 67.1%. There was progressive increase in diagnostic sensitivity when the GCT was performed after 24 weeks without significant decrease in specificity. In low-risk populations, a 50g GCT should be performed between 24-28 weeks gestation.

Rapid desktop method for the measurement of glycated haemoglobin HbA1c.

A new desktop monoclonal antibody-based test system (Bayer Diagnostics DCA 2000) for quantitation of glycated haemoglobin HbA1c is described and evaluated. This method involves a monoclonal antibody to HbA1c which defines its specificity. It requires 9 minutes to complete, and shows good intra- and inter-run precisions (1.3%-3.7%), at 5 HbA1c levels tested--4.1%, 5.7%, 6.1%, 9.2% and 12.4%. Results from 81 blood samples obtained from diabetic patients (3.9%-13.2% HbA1c) showed excellent correlation with a laboratory-based ion-exchange HPLC technique (y = 1.03 [Lab] + 0.103%; Pearson coefficient, r = 0.99). The test can be performed either with a capillary fingerprick or venous blood sample. Only 1 microliter of blood volume is required. Comparison of HbA1c levels of 43 paired capillary and venous samples showed excellent correlation (y = 1.00 [venous] + 0.042%; r = 0.99). The HbA1c values obtained from a cohort of 37 healthy adults, mean (+/- SD) age 33 +/- 9.01 years, gave a value of 5.5 +/- 0.42%. The calculated 95% confidence limits are 4.7%-6.3%. This quick method provides 'stat' HbA1c results, which were hitherto not possible with the laboratory-based methods.

Glycosylated haemoglobins in women with low risk for diabetes in pregnancy.

Glycosylated haemoglobin levels (HbA1) were determined in 489 normal pregnant Singaporean women, attending a diabetic screening programme using a 50g glucose challenge test. All subjects had no risk factors for diabetes mellitus nor a history of previous gestational diabetes. They were selected when the 1h 50g glucose challenge (GCT) is < 6.66 mmol/l or a 2h post-glucose level of < 7.77 mmol/l (a 75g OGTT is repeated within a week of an abnormal GCT). Another 18 subjects with normal OGTT but did not have a 50g glucose challenge done were also included in the study. Mean and normal range (2 SD) of HbA1 levels in all subjects were 4.64% and 3.66-5.62%, respectively. Mean HbA1 levels also varied with gestation and ethnic group. The HbA1 appeared to be low at early gestation with nadir at 21-24 weeks and thereafter returned to initial levels at term. In Chinese, mean HbA1 levels, random plasma glucose and 1h glucose challenge were significantly higher than those of the Malays but not the Indians. In referring to these levels, the variation within gestation and ethnic group must also be considered.

Congenital abnormalities and pregestational diabetes mellitus in pregnancy.

Despite falling perinatal mortality rate, congenital malformation remains the major cause of mortality in infants of mothers with established diabetes. The perinatal mortality rate in this group of infants is 5 times the overall perinatal mortality rate in this hospital. It is well established that pre-pregnancy counselling and maintenance of euglycaemia during the periconception period are the keys to prevention of congenital malformation. We are able to offer pre-pregnancy counselling to 29% of our diabetic mothers who are diagnosed to have pregestational diabetes mellitus only after the 6 weeks postnatal oral glucose tolerance test. Even in the known established diabetes mellitus, 95% of the patients were first seen after the period of organogenesis (> 8 weeks). This could explain our high congenital malformation rate of 15.7%.

The use of semi-quantitative urine test-strip (Micral Test) for microalbuminuria screening in patients with diabetes mellitus.

BACKGROUND: Microalbuminuria is an early marker of prognostic significance in diabetic renal disease. However, testing for microalbuminuria in a timed sample of urine using the double antibody radioimmunoassay (RIA) method is cumbersome and requires special laboratory facilities. Recently, a test strip for microalbuminuria, the Micral Test was available and we evaluated the performance of this test strip as a screening method for detection of microalbuminuria. METHODS: One hundred consecutive diabetic patients who were tested to be dipstick-negative (Albustix) for proteinuria were enrolled for the study. Micral Tests were performed on a paired first morning and random urine specimen from the same patient and the results compared with a timed 24-hour urine measurement of urine albumin excretion using the RIA method. RESULTS: Eighteen specimens were tested positive by the RIA method with a urinary albumin range of 32-177 mg/24 hours. With the Micral Test, the following sensitivity, specificity, positive and negative predictive values were obtained: 66.7%. 97.6%, 85.7% and 93.0% for the first morning urine specimens, and 77.8%, 91.5%, 66.7% and 94.9% for the random urine specimens. CONCLUSIONS: These results suggest that Micral Test with either the first morning or random urine specimen offers a simple, reliable, rapid and convenient method for screening of microalbuminuria in the diabetic patient.

Gestational diabetes: obstetric and neonatal outcome in 411 cases.

AIM: To study the obstetric and neonatal outcome of women with gestational diabetes mellitus. METHODS: Four hundred and eleven (411) women with gestational diabetes mellitus were studied retrospectively. The control group was 3,391 non-diabetic women delivered in the same period. RESULTS: Women with gestational diabetes mellitus had a significantly higher incidence of pregnancy-induced hypertension, and more were likely to present with malpresentation in labour. They had a higher incidence of surgical induction and an increased incidence of elective and emergency Caesarean section rate compared to controls. The neonatal outcome was excellent. The perinatal morbidity and mortality were similar to the non-diabetic controls. CONCLUSION: The excellent outcome of the women with gestational diabetes mellitus in this study is attributed to tight control of blood glucose level and close attention given to this group of patients.

The 75 and 50g OGTT response in normal pregnant women.

The 75 and 50g OGTT (Oral Glucose Tolerance Test) responses were studied in 86 normal healthy pregnant women (mean age 28.7 +/- 0.4 (SEM) years) at 28 and 32 weeks respectively. Of these were 50 Chinese, 20 Malays and 16 Indians. Mean glucose responses at fasting, 1 and 2h post glucose load were 78.3 +/- 0.7, 132.2 +/- 2.8 and 116.2 +/- 2.1 mg/dl respectively for the 75g OGTT and 78.5 +/- 0.7, 130.5 +/- 2.5 and 106.7 +/- 1.8 mg/dl respectively for the 50g OGTT. Except for the 2h responses, corresponding responses between both OGTTs were not significantly different. All races showed a similar OGTT response. Using a set criteria for diagnosis of abnormals resulted in gross inconsistency in the number of abnormals diagnosed for both OGTTs. However, the use of 95th percentile shows a closer agreement in the diagnosis of abnormal cases for both OGTTs. Also, the 2h OGTT response for the 75g OGTT is higher than that of WHO's criteria for impaired glucose tolerance. This emphasizes the need to establish our own reference range.