The oral glucose tolerance test (OGTT) is regarded as the "gold standard" for the diagnosis of diabetes mellitus (DM) and impaired glucose regulation (IGR) [1]. However, the OGTT is poorly reproducible, time-consuming, and unsuitable for large-scale screening [2]. Clinical doctors have begun to recognize that point-of-care blood glucose testing cannot accurately monitor glycemic fluctuation, or "drift of glucose," and its clinical significance. Glycated hemoglobin (HbA_1c) is an indicator that ref lects the average plasma glucose (PG) levels over the recent 2 to 3 months. The HbA_1c test is relatively stable and has less variability and fewer interferences with environmental factors than point-of-care glucose testing [3,4].

Recently, the application of HbA1c in the diagnosis of DM was fully endorsed by The American Diabetes Association (ADA) in an updated and promulgated 2010 version of "The Clinical Practice Recommendations for the Diagnosis and Treatment of Diabetes" [5]. The recommendation was made that HbA_1c ≥ 6.5% can be used as a diagnostic criteria for DM. In this study, we examined the HbA_1c and OGTT results of 497 Chinese subjects with impaired fasting glucose (IFG). We evaluated the validity of using HbA_1c as a screening and diagnostic test for diabetes in Chinese patients with IFG.

Of the 497 subjects, according to the OGTT results, 155 (31.2%) subjects had DM, 230 (46.3%) had IGR, and 112 (22.5%) had NGT. Table 1 summarizes the clinical parameters of these subjects according to their glycemic status. In particular, HbA_1c values in subjects with NGT, IGR, and DM were 5.84 ± 0.77%, 6.06 ± 0.77%, and 6.88 ± 0.99%, respectively (p value < 0.05 comparing the three groups). Age, BMI, WC, systolic BP, fasting and 2 hours PG, TG, and HDL-C were significantly different between any two groups of NGT, IGR, and DM.

In total, 213 subjects with DM were diagnosed by either HbA_1c ≥ 6.5% and/or the OGTT. Among these 213 subjects, 154 were diagnosed by HbA_1c and 155 were diagnosed by the OGTT; 96 fulfilled both criteria (the "overlapping" group), accounting for 45.1% of all the newly diagnosed patients who had DM. Table 2 summarizes the clinical characteristics of the DM group diagnosed by the different criteria with or without excluding the "overlapping" group. In those diagnosed by HbA_1c, the HbA_1c level was higher, and the fasting and 2-hour PG values were lower than those diagnosed by the OGTT. After excluding the "overlapping" group, the DM subjects, regardless of the diagnostic test used, had significantly lower fasting and 2-hour PG and HbA_1c levels, compared with those included in the "overlapping" group. No significant difference was noted for other clinical indexes such as BP and lipid profile (Table 2).

With OGTT as the reference standard for diagnosing DM, the optimal cutoff point of HbA_1c was 6.3%, with sensitivity and specificity at 79.6% and 82.2%, respectively. The corresponding area under the curve (AUC) of the ROC curve was 0.87 (95% confidence interval [CI], 0.83 to 0.92). Using HbA_1c ≥ 6.5% as a diagnostic tool, the sensitivity and specificity were 62.7% and 93.5%, respectively (Fig. 1).

DM is an important clinical and social health problem. Alarmingly, many undiagnosed subjects remain, who may account for up to 50% of the diabetic population [7]. The onset time of type 2 diabetes has been estimated to be 9 to 12 years before diagnosis. Currently, the OGTT is still the "gold standard" for diagnosing DM and IGR. However, the glucose values obtained by the OGTT are poorly reproducible, and many problems related to the standardization of glucose testing remain [2,8,9]. In addition, the subjects need to receive strict preparation before the OGTT. Due to the above reasons, an increasing demand for adopting HbA_1c to diagnose DM has arisen. Recently, the ADA revised their "Clinical Practice Recommendations for the Diagnosis and Treatment of Diabetes" such that HbA_1c was recommended for diagnosing diabetes [1]. The International Diabetes Federation, the European Association for the Study of Diabetes, and the World Health Organization have also incorporated HbA_1c as an alternative method for diagnosing DM [10].

Our results showed that the comparison of HbA_1c levels between any two groups of NGT, IGR, and DM had significant differences (p value < 0.01). This observation might be explained by recent evidence showing that HbA_1c could reflect different stages of glucose metabolism [11]. Compared to PG as a point-of-care test, HbA_1c reflects the patients' average glycemic status over the past 2 to 3 months and it is not influenced by current diet, drug treatment, and blood sampling time [12-14]. However, many studies have found that ethnicity might influence the level of HbA_1c [14-16]. A large international study comparing the relationship between HbA_1c and the average PG levels among African Americans and Caucasians showed that African Americans had a higher HbA_1c under the same PG level [14].

In our study, the optimal cutoff point for HbA_1c in diagnosing DM was 6.3% with a sensitivity and specificity of 79.6% and 82.2%, respectively. In addition, the AUC under the ROC curve was high at 0.87 (95% CI, 0.83 to 0.92). These values are similar to the reported optimal HbA_1c cutoff point of 6.2% used by the United Kingdom Prospective Diabetes Study (UKPDS), 6.1% by the DCCT [17], 6.1% by Hu et al. [18] from Shanghai, China, and 6.1% by Mohan et al. [19] in Asian Indians, but higher than that reported by Nakagami et al. [20] from the Japan National Diabetes Survey (5.6%). Our study had 259 subjects whose HbA_1c levels were lower than 6.3%. In other words, if only those with HbA_1c ≥ 6.3% need to receive an OGTT to confirm their glycemic status, then 52.2% (259/497) of the subjects would not require an OGTT. Currently, the ADA recommends that either HbA_1c (≥ 6.5%) or the OGTT can be used as diagnostic criteria for DM. Since HbA_1c may be affected by ethnicity, understanding the effectiveness of using HbA_1c to diagnose DM in the Chinese population is important. Based on the ROC curve, the sensitivity and specificity were 62.7% and 93.5%, respectively, when using HbA_1c ≥ 6.5% to diagnose DM in our population. Similarly, various studies reported that using HbA_1c ≥ 6.5% for diagnosis gave a high specificity (> 95%), while the sensitivity was suboptimal (40-50%) [21,22]. Nevertheless, since DM is a serious disease with significant medical implications, a more specific diagnostic approach is necessary to avoid a large number of false positives.

In this study, we enrolled subjects with fasting PG ≥ 5.6 mmol/L (those with IFG) so as to recruit more patients with IGR or DM. Therefore, the HbA_1c levels of the subjects may be slightly higher than those of the general community. Among the 497 subjects in our study, 42.9% (n = 213) were diagnosed with diabetes by HbA_1c (n = 154), the OGTT (n = 155), or both (n = 96). This particularly high prevalence of diabetes should not be interpreted as the usual rate in the general population in Jiangsu Province. We found that the PG and HbA_1c levels were not well correlated in our population, especially among the subjects with DM who were diagnosed by either one but not both criteria. The HbA_1c level was higher, while the fasting and 2-hour PG was lower in subjects with DM diagnosed by HbA_1c as compared to those diagnosed by the OGTT. This result highlights the importance of validating the recommended HbA_1c diagnostic level in different populations before formally adopting it. The International Expert Committee recommended HbA_1c ≥ 6.5% as the diagnostic cutoff for DM based on the existence of retinopathy with data from the Evaluation of Screening and Early Detection Strategies for Type 2 Diabetes and Impaired Glucose Tolerance (DETECT-2) Study [10,23]. Our data examined this approach and reported a high specificity of 93.5% with a HbA_1c cutoff level of 6.5%, suggesting this newly recommended diagnostic tool could be applicable to the Chinese population, thus validating the recommended standards of the ADA [1]. However, our sample size was relatively small and with a low proportion of women (27.4%). Further studies are needed to confirm our results.

In conclusion, with OGTT as the "gold standard" reference for diagnosing DM in Chinese in Nanjing, the optimal cutoff point of HbA_1c was 6.3% with a sensitivity and specificity of 79.6% and 82.2%, respectively. HbA_1c ≥ 6.5% has reasonably good specificity (93.5%) for diagnosing diabetes, which is in concordance with the ADA recommendations.