The independent relationship between hemoglobin A1c and homeostasis model assessment of insulin resistance in non-diabetic subjects

Introduction: Determining the factors affecting hemoglobin A1c (HbA1c) levels may help better interpretation of HbA1c values. In this study, we investigated if insulin resistance is a significant parameter contributing to the variability of HbA1c values. Methods: We retrospectively analyzed serum fasting glucose, fasting insulin, 2  h glucose and HbA1c records of 18–85  years aged outpatients who underwent a 75 g oral glucose tolerance test (OGTT) in our hospital during the period January 2010–May 2014. Homeostasis model assessment of insulin resistance (HOMA-IR) ≥ 2.5  was defined as insulin resistant. Results: Insulin resistant subjects with normal glucose tolerance had significantly higher HbA1c and fasting glucose levels compared to insulin sensitive subjects with normal glucose tolerance (p = 0.002, p < 0.001, respectively). Similarly, insulin resistant subjects with pre-diabetes had significantly higher HbA1c and 2-h glucose levels compared to insulin sensitive subjects with pre-diabetes (p = 0.016, p = 0.013, respectively). Regression analysis showed that HOMA-IR(log) is associated with HbA1c levels independent of fasting and 2h glucose concentrations (p < 0.001). Age was the variable with highest standardized β coefficient in regression model. Conclusion: Our data showed that HOMA-IR is associated with glycated hemoglobin values independent of glycemic status and the effect of age on HbA1c values should not be ignored in non-diabetic subjects.


Introduction
Insulin resistance (IR) is defined as the impaired ability of target tissues of fat, liver, and muscle to show various metabolic effects of insulin, including glucose uptake [1]. Insulin resistance plays an important patho-physiological role in the development of diabetes, dyslipidemia, hypertension, and cardiovascular disease [2,3]. Prospective studies have shown that it is a powerful predictor of the likelihood of an individual developing diabetes or cardiovascular disease [4]. Accurate measurement of IR requires complex techniques that are expensive and timeconsuming. A number of surrogate indices of IR had been developed. The homeostatic model of assessment-insulin resistance (HOMA-IR), which uses fasting insulin and glucose levels to calculate IR, is the most widely used [5]. However, HOMA-IR results are reasonably correlated with those of clamping studies (the "gold standard").
Glycated hemoglobin (HbA 1c ) is well recognized and widely used as a measure of glycemic control reflecting the mean blood glucose level over the preceding weeks to months. Until recently, hemoglobin A 1c has only been used to monitor people already diagnosed with diabetes, serving as the gold-standard measure of glycemic levels over a 3-month period. More recently, as with measures of glucose, HbA 1c levels have been used to describe a continuum of risk for the development of diabetes and associated conditions. Because it can be measured regardless of food intake, A 1c is simpler than fasting plasma glucose (FPG) or oral glucose tolerance tests [6].
Recently, the International Expert Committee recommended that HbA 1c be added to the diagnosis of diabetes mellitus (DM); the 2010 American Diabetes Association (ADA) clinical practice recommendation defining HbA 1c levels of over 6.5% as DM, and an HbA 1c between 5.7 and 6.4% as pre-diabetes [7,8]. It was shown that pre-diabetes is associated with insulin resistance which is a risk factor for development of type 2 diabetes. How well a HbA 1c level in the pre-diabetic range (5.7%-6.4%) predicts insulin resistance is not clear. Publications associated with the relationship between HbA 1c and insulin resistance are limited. Therefore, the aim of the current study was to clarify the relationship between insulin resistance and HbA 1c .

Materials and methods
We retrospectively analyzed serum fasting glucose, fasting insulin, 2 h glucose and HbA 1c records of 18-85 yearsold outpatients who underwent a 75 g OGTT in Department of Biochemistry, Tepecik Teaching and Research Hospital during the period January 2010-May 2014. The samples for all parameters were collected in the same day. Patients with a serum creatinine level above reference range and patients with DM according to ADA criteria (fasting glucose > 125 mg/dL or 2 h glucose > 199 mg/dL or HbA 1c > 6.4%) were excluded. The present study included a total of 365 patients of whom 281 were women and 84 were men. The study was conducted with the approval of the Local Hospital's Ethics Committee.
Glucose tolerance status was assessed with the 75-g OGTT and serum fasting glucose defined according to the 2010 ADA criteria [8]. Our study used the HOMA-IR as the diagnostic criteria for insulin resistance and HOMA-IR ≥ 2.5 was identified as an indicator of insulin resistance. HOMA-IR was calculated using the following formula: HOMA-IR = fasting serum glucose (mg/dL) × fasting serum insulin value (μU/mL)/405 [5]. Serum glucose levels were measured by the hexokinase method using the Olympus AU 2700 analyzer (Olympus Diagnostics. GmbH, Hamburg, Germany).
Serum insulin levels were measured using direct chemiluminescence technology on Siemens Immulite 2000 XPI analyzer (Siemens Healthcare Diagnostics, Deerfield, IL, USA). HbA 1c analysis was performed by affinity chromatography HPLC on Primus Ultra2 Analyzer (Primus Corporation, Kansas City, Kansas, USA).

Statistical methods
Statistical analyses were conducted using the statistical package SPSS version 17 (SPSS Inc., Chicago, IL, USA). p-Value < 0.05 was considered as statistically significant. Data were expressed as mean ± SD or with 95% confidence intervals (CI). Independent samples t-test and χ 2 test were used for comparison of insulin resistant and insulin sensitive groups. Multivariate linear regression analysis was used for determining the association between HbA 1c and independent variables. Variables with non-normal distribution were log transformed.

Results
A total of 365 non-diabetic subjects were included in the study. Mean age of the subjects were 40 ± 11 and 281 (77%) of the subjects were female, while 84 (23%) of the subjects were male. There were 191 individuals with normal glucose tolerance and 174 individuals with pre-diabetes (impaired fasting glucose, impaired glucose tolerance, or both according to ADA criteria). Insulin resistance (HOMA-IR > 2.5) were detected in 147 (40%) of the subjects. Table 1 shows the characteristics of the study population.
The subjects were divided into two groups according to their HOMA-IR values: group 1, insulin sensitive (HOMA IR ≤ 2.5) and group 2, insulin resistant (HOMA-IR > 2.5). HbA 1c and other biochemical parameters of insulin resistant and insulin sensitive subjects were compared. Analysis showed that insulin resistant subjects with normal glucose tolerance had significantly higher HbA 1c and fasting glucose levels compared to insulin sensitive subjects with normal glucose tolerance (p = 0.002, p < 0.001, respectively). Similarly, insulin resistant subjects with pre-diabetes had significantly higher HbA 1c and 2-h glucose levels compared to insulin sensitive subjects with pre-diabetes (p = 0.016, p = 0.013, respectively) ( Table 2). To estimate the independent relationship between HbA 1c and HOMA-IR, multiple linear regression analysis was used. Age, gender, fasting glucose, 2-h glucose and HOMA-IR were entered in the regression model as independent variables. Since HOMA-IR values had skewed distribution, log transformed HOMA-IR was used in analysis. Regression analysis showed that age, fasting glucose, 2-h glucose, and log transformed HOMA-IR values are significant predictors of HbA 1c . Higher HOMA-IR values are associated with higher HbA 1c levels in non-diabetic subjects. Standardized β coefficients demonstrate the relative importance of predictor variables in regression model. In the regression model age showed the highest standardized β coefficient. In non-diabetic subjects age was the most important factor affecting HbA 1c values (Table 3). HbA 1c values were also presented in different age groups (Table 4).
Dividing the study population as normal and pre-diabetic subjects and re-performing the regression analysis did not change the results fundamentally. Age was still most effective factor in predicting HbA 1c values in both normal group and pre-diabetic group. Log HOMA was still a significant variable (data not shown).
To visualize the data a regression plot showing the relation between HbA 1c and HOMA-IR was generated. The relation between HbA 1c and HOMA-IR was similar in different sample groups (group 1: fasting glucose < 96 mg/ dL, group 2: fasting glucose > 96 mg/dL) (Figure 1).

Discussion
It was clearly reported that lower HbA 1c values were associated with reduced microvascular and macrovascular complications in diabetic patients [9]. In type 2 diabetes a HbA 1c target level of < 7% is currently recommended by ADA and HbA 1c is the basis guiding diabetes therapy. Recently ADA recommended using HbA 1c also for the diabetes diagnosis and consolidated its importance in diabetes mellitus. The use of HbA 1c as a diagnostic criterion necessitates a more accurate measurement and careful interpretation. Several factors other than plasma glucose levels were reported to affect glycated hemoglobin levels. Any condition affecting turnover of red blood cells like hemolytic anemia and iron deficiency affect HbA 1c levels. Previous studies showed that genetic-ethnic factors, sex hormones and age are associated with HbA 1c values [10][11][12][13][14]. In the present study we report that HOMA-IR is a factor affecting HbA 1c values independent of fasting and 2-h post-load glucose concentrations. The relationship between glycated hemoglobin levels and glucose levels are higher in diabetic patients than in non-diabetic patients therefore it can be concluded that non-glycemic factors affecting HbA 1c levels are more important in non-diabetic patients [15].
Previous studies investigated the relationship between insulin resistance and HbA 1c . Gallwitz et al. showed that with increasing HbA 1c levels, there was a statistically significant increase in HOMA-IR in patients with type 2 DM [16]. Heianza et al. reported that in subjects without a history of diabetes a HbA 1c level of > 5.9% were significantly associated with higher HOMA-IR values [17]. In another study Borai et al. showed that the correlation between HbA 1c and insulin resistance were higher in subjects with normal glucose tolerance than in patients with pre-diabetes and diabetes [18]. None of the aforementioned studies reported corrected results independent of   HbA 1c independent of fasting glucose but the effect of postload glucose on HbA 1c was lacking [19]. One recent study indicated that HbA 1c was associated with HOMA-IR independent of 0-and 120-min glucose in pregnant women with gestational diabetes mellitus [20]. The results of our study reveal that HbA 1c is associated with HOMA-IR independent of fasting and post-load glucose status also in non-diabetic subjects. Glycation is the nonenzymatic attachment of free aldehyde groups of carbohydrates to the unprotonated free amino groups of proteins [21]. The binding of glucose molecules to potential glycation sites in hemoglobin molecule leads to formation of HbA 1c . Via condensation with glucose, hemoglobin A first forms a labile intermediate adduct, which is thereafter rearranged to the more stable ketoamine adduct (HbA 1c ) form [22]. Physiological factors like pH, inorganic phosphate, oxidative stress, deglycation, and Schiff base inhibitors can affect the rate of HbA 1c formation [23][24][25][26][27]. Another point to be considered is that glycation of hemoglobin occurs in the intracellular compartment. Previously, it was demonstrated that the erythrocyte glucose-to-plasma glucose concentration ratio may affect hemoglobin glycation and contributes to the variation in HbA 1c levels [28].
Oxidative stress is a process that was proposed to be associated with the multifactorial etiology of insulin resistance. It was shown that plasma markers of oxidative stress were correlated with the degree of insulin resistance [29,30]. Oxidative stress which is a factor co-existing with insulin resistance may also be responsible for the increased hemoglobin glycation. Oxidative stress biomarkers as lipid peroxides were reported to be associated with hemoglobin glycation [31,32]. LDL oxidation was also suggested to increase HbA 1c values [33]. Furthermore there is evidence that antioxidants can partially inhibit the formation of HbA 1c [31]. Consistent with previous studies, age was a significant factor affecting HbA 1c levels independent of glycaemia [13,34]. However the mechanisms involved in the age related HbA 1c increase remain to be established. Since our model showed that age was the most effective factor contributing to the variation of HbA 1c levels in non-diabetic subjects it should not be ignored when interpreting an HbA 1c result. The question whether age-specific diagnostic and treatment criteria would be appropriate was previously mentioned [13].
Due to its retrospective design the current study has some limitations. Firstly the information of some possible confounding variables (e.g. BMI) could not be gathered; therefore confounding factors may exist. Second, HOMA-IR is not the gold standard method for measuring insulin sensitivity. The euglycemic hyperinsulinemic clamp technique is the gold standard for quantifying insulin sensitivity; however, this technique requires insulin infusion and repeated blood sampling. HOMA-IR is a relatively simple method to determine insulin sensitivity. It can be calculated from a single blood sample and it was reported to have a linear correlation with glucose clamp technique [35].
In conclusion, our data showed that HOMA-IR is associated with glycated hemoglobin values independent of glycaemia and age is a very important factor affecting HbA 1c values in non-diabetic subjects.