In the diagnosis of diabetes mellitus, hemoglobin A1c (HbA1c) is sometimes measured to determine the need of an oral glucose tolerance test (OGTT). However, HbA1c does not accurately reflect glycemic status in certain conditions. This study was performed to test the possibility that measurement of serum glycated albumin (GA) better assesses the need for OGTT. From 2006 to 2012, 1559 subjects not known to have diabetes or to use anti-diabetic medications were enrolled. Serum GA was measured, and a 75-g OGTT was then performed to diagnose diabetes. Serum GA correlated significantly to age (r = 0.27, p<0.001), serum albumin (r = –0.1179, age-adjusted p = 0.001), body mass index (r = -0.24, age-adjusted p<0.001), waist circumference (r = -0.16, age-adjusted p<0.001), and plasma GA (r = 0.999, p<0.001), but was unaffected by diet (p = 0.8). Using serum GA at 15% for diagnosis of diabetes, the sensitivity, specificity, and area under the receiver-operating characteristic curve were 74%, 85%, and 0.86, respectively. Applying a fasting plasma glucose (FPG) value of < 100 mg/dL to exclude diabetes and of ≥ 126 mg/dL to diagnose diabetes, 14.4% of the study population require an OGTT (OGTT%) with a sensitivity of 78.8% and a specificity of 100%. When serum GA value of 14% and 17% were used to exclude and diagnose diabetes, respectively, the sensitivity improved to 83.3%, with a slightly decrease in specificity (98.2%), but a significant increase in OGTT% (35%). Using combined FPG and serum GA cutoff values (FPG < 100 mg/dL plus serum GA < 15% to exclude diabetes and FPG ≥ 126 mg/dL or serum GA ≥ 17% to diagnose diabetes), the OGTT% was reduced to 22.5% and the sensitivity increased to 85.6% with no change in specificity (98.2%). In the diagnosis of diabetes, serum GA measurements can be used to determine the need of an OGTT.
Citation: Wu W-C, Ma W-Y, Wei J-N, Yu T-Y, Lin M-S, Shih S-R, et al. (2016) Serum Glycated Albumin to Guide the Diagnosis of Diabetes Mellitus. PLoS ONE 11(1): e0146780. doi:10.1371/journal.pone.0146780
Editor: Juergen Eckel, GDC, GERMANY
Received: May 26, 2015; Accepted: December 22, 2015; Published: January 14, 2016
Copyright: © 2016 Wu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data cannot be made publicly available due to ethical restrictions. The use of the data for further research needs to be approved by the Institutional Research Ethical Committee of National Taiwan University Hospital. Requests to access the data may be submitted to firstname.lastname@example.org.
Funding: This work was funded by NTUH 102-M2262 from National Taiwan University Hospital to HY Li. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
In accordance with the recommendations of the American Diabetes Association (ADA), International Diabetes Federation, and World Health Organization [1–4], diagnosis of diabetes mellitus should be based on either an oral glucose tolerance test (OGTT) or hemoglobin A1c (HbA1c) findings. However, the OGTT is time-consuming and requires two blood samplings. In the diagnosis of diabetes, therefore, HbA1c values are often used to access the need for an OGTT [5–9]. For example, subjects with fasting plasma glucose 100–125 mg/dl or HbA1c 6.1–6.4% were recommended to receive OGTT, according to the clinical practice guidelines by Diabetes Association of the R.O.C. in 2015 , which is modified based on our previous report . Nonetheless, measurement of HbA1c may not accurately reflect blood glucose concentrations in certain conditions. For example, lower HbA1c values are found in conditions associated with a decrease erythrocyte lifespan, such as recent transfusion and increased erythropoiesis secondary to hemolysis or blood loss [10–12]. In patients with chronic kidney disease (CKD), erythrocyte lifespan is decreased by anemia, recent transfusions, and other alterations which result in reduced HbA1c values. In these patients, carbamylated hemoglobin is recognized to interfere with HbA1c measurements . HbA1c values are reported to be higher in conditions, such as asplenia, wherein erythrocyte lifespan increases . Iron deficiency anemia is also associated with higher HbA1c value due to the enhancement of hemoglobin glycation by increased plasma malondialdehyde [13, 14]. Additionally, genetic variants of hemoglobin, such as HbS trait and HbC trait, and elevated fetal hemoglobin (HbF) concentration can affect the accuracy of HbA1c measurements. Based on the prevalence of anemia and chronic kidney disease, up to one fourth of the population is estimated to have a condition wherein HbA1c values do not accurately reflect blood glucose concentrations [15–17]. Therefore, an alternative procedure to determine the need for an OGTT in the diagnosis of diabetes is needed.
Glycated albumin (GA), a ketoamine formed by non-enzymatic glycation of serum albumin, reflects short-term (2–3 weeks) mean glycemic levels [18, 19]. GA values are not affected by changes in erythrocyte lifespan , and measurement of GA is not influenced by anemia or other conditions which invalidate HbA1c measurements in the diagnosis of diabetes . Furthermore, GA is reported to serve as a better indicator of glycemic control than HbA1c in diabetic patients on dialysis [21–23]. Findings from the DCCT/EDIC and other studies [24–28] revealed the association of increased GA values with the presence of diabetic retinopathy, nephropathy, and cardiovascular complications; these findings also support the use of GA measurements in the diagnosis of diabetes. The present study was undertaken to ascertain whether measurement of GA can be employed to determine the need for an OGTT in the diagnosis of diabetes.
A community-based cohort study, termed the Taiwan Lifestyle Study, was conducted. From 2006 to 2012, subjects older than 18 years were invited to participate in the study [5, 29, 30]. All participates completed a questionnaire, underwent a physical examination, and received blood tests. Subjects with diabetes or who were receiving anti-diabetic drugs were excluded. Written informed consent was obtained from each participant. This study was reviewed and approved by the Institutional Review Board of the National Taiwan University Hospital.
Body height (to the nearest 0.5 cm) and body weight (to the nearest 0.1 kg) were measured by attending nurses. Body mass index (BMI) was calculated by dividing body weight (kg) by the square of body height (m). Waist circumference (WC) was measured in the horizontal plane midway between lowest rib and the iliac crest .
The oral glucose tolerance test
A standard 75-g OGTT was performed after a fast at least 8 hours. Blood samples were collected and stored at -80°C for serological assay. Normal glucose tolerance (fasting plasma glucose [FPG] < 5.56 mmol/L and 2-h plasma glucose [PG] < 7.8 mmol/L), impaired fasting glucose (IFG) (FPG = 5.56–6.9 mmol/L), impaired glucose tolerance (2-h PG = 7.8–11.0 mmol/L), and diabetes mellitus (FPG ≥ 7 mmol/L and/or 2-h PG ≥ 11.1 mmol/L) were defined according to the ADA diagnostic criteria . GA was measured using the LUCICA GA-L kit (Asahi Kasei Pharma, Tokyo, Japan) and the Beckman Coulter AU2700 Chemistry Analyzer (Beckman Coulter Systems Co., Nyon, Switzerland). GA in paired serum and plasma samples from twelve subjects were measured to determine whether the method of sample collection affected GA measurements. Plasma glucose and HbA1c were measured using automatic analyzers (Toshiba TBA 120FR, Toshiba Medical Systems Co., Tokyo, Japan and HLC-723 G7 HPLC systems, Tosoh Corporation, Tokyo, Japan, respectively). The HbA1c assay was certified by the National Glycohemoglobin Standardization Program  and standardized to the DCCT reference assay.
Findings are presented as means ± standard deviations (SDs) for continuous variables and as percentages for categorical variables. The differences between the GA < 14% and the GA ≥ 14% groups in Table 1 were determined using the Student's t test or chi-square test. The performance of serum GA in the diagnosis of diabetes by the OGTT was analyzed by the receiver operating characteristic (ROC) curve. The optimal cutoff for serum GA was derived from the ROC curve with the shortest distance to sensitivity = 1 and 1–specificity = 0 and with the Youden index (Y = sensitivity + specificity– 1). The performance of different criteria (from GA of 10.0% to 23.0%, with a 0.1% intervals) to diagnose diabetes was calculated and included the proportion of subjects receiving an OGTT (OGTT%), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), false negative rates (FNR), and false positive rates (FPR). Optimal cutoffs with minimal OGTT% were chosen among cutoffs with sensitivity > 80% and specificity > 95%. The difference in serum GA values between men and women was determined by the Student's t test. Differences between serum and plasma GA values and between fasting and postprandial GA values were determined by the paired t test. The relationships of serum GA to age, serum albumin, body mass index (BMI), waist circumference (WC), and plasma GA were analyzed by Pearson's correlation coefficients and linear regression models. The relationships of serum GA to FPG, to 2-hr PG after the OGTT, and to HbA1c were analyzed by quadratic regression models. A two-tailed p value < 0.05 was considered to be statistically significant. The statistical analyses were performed with STATA/SE 11 for Windows (StataCorp LP, College Station, TX, USA).
A total of 1559 participants were enrolled in the present study, including 624 (40%) men and 935 (60%) women. Mean ± SD for age, BMI, and WC were 50.4 ± 12.6 years, 24.3 ± 3.4 kg/m2, and 81.8 ± 10.3 cm, respectively. Serum GA distributed normally (mean 14%, SD 2.6%, 95% CI 11.4%–17.1%). Mean ± SD for FPG, 2-h PG, and HbA1c for these subjects were 5.2 ± 1.1 mmol/L, 7.0 ± 3.2 mmol/L, and 5.8 ± 0.8%, respectively. Clinical characteristics for the study group are summarized in Table 1. Subjects with serum GA measurement ≥ 14% were older and had higher FPG, 2-h PG, and HbA1c values as compared to those below 14%. Additionally, the male/female ratio was lower in the group with serum GA measurements ≥ 14%.
Factors potentially associated with the measurement of serum GA were examined. The relationship of serum GA to age, gender, serum albumin, BMI, and WC in subjects without diabetes is presented in S1 Fig. No difference in serum GA was observed as a function of gender (14.0 ± 2.6% for females and 13.9 ± 2.6% for males, p = 0.26). Serum GA was associated with age in these subjects (r = 0.27, p < 0.001). Ten—year interval increases in age were each associated with a 0.31% increase in serum GA. Serum GA was negatively associated with serum albumin in these subjects (r = -0.12, age-adjusted p = 0.001). For every 1 g/dl increase in serum albumin, serum GA was observed to decrease by of 0.32%. Serum GA was inversely correlated with anthropometric parameters, including BMI and WC, in subjects without diabetes (r = -0.24 with age-adjusted p < 0.001 and r = -0.16 with age-adjusted p < 0.001, respectively). Each 1 kg/m2 increase in BMI was associated with a 0.12% decrease in serum GA, and each 1 cm increase in WC was associated with a decrease in serum GA of 0.04%. The influence of sample collection method on GA measurements in non-diabetic subjects is presented in S2 Fig. Serum GA was closely correlated to plasma GA (r = 0.999, p < 0.001; serum GA = 0.98 x plasma GA + 0.29). Furthermore, serum and plasma GA values did not change in sample collection during a fast, postprandial, or after glucose challenge (mean serum GA of 17.9 ± 6.7% during a fast and mean serum GA of 18.0 ± 6.5% at 2 h after glucose challenge, p = 0.8; mean plasma GA of 15.8 ± 4.3% during a fasting and mean plasma GA of 15.8 ± 4.3% at 2 h post-prandial, p = 0.8).
Quadratic relationships between serum GA and FPG, 2-h PG, and HbA1c were observed for subjects with and without diabetes (S3 Fig). Table 2 shows the numeric relationships between serum GA and FPG, 2-h PG, and HbA1c. According to these quadratic relationship, serum GA of 14% corresponds to a HbA1c value of 5.7% whereas serum GA of 18% corresponds to a HbA1c value of 6.5%.
The area under the ROC curve for serum GA in the diagnosis of diabetes by the OGTT was 0.86 (95% CI 0.82–0.90) whereas the area under the ROC curve for HbA1c in the diagnosis of diabetes by the OGTT was 0.90 (95% CI 0.86–0.94). The optimal cutoff value for serum GA was 15%; the sensitivity, specificity, PPV, and NPV for serum GA in the diagnosis of diabetes were 74%, 85%, 32%, and 97%, respectively.
The use of two GA cutoffs with and without FPG cutoffs for diagnosis of diabetes was examined. Table 3A–3C and Fig 1A–1C describe the performance of serum GA alone and in combination with IFG criteria as well as the performance of IFG criteria alone for diagnosis of diabetes by the OGTT. Using the IFG criteria of FPG < 5.56 mmol/L to exclude and FPG ≥ 7 mmol/L to diagnose diabetes (Fig 1A), the sensitivity, specificity, and OGTT% were 78.8%, 100%, and 14.4%, respectively. As shown in Table 3A, the choice of cutoff GA value was associated with a tradeoff between sensitivity/specificity and OGTT%. For example, the sensitivity and OGTT% increased as the cutoff to exclude diabetes decreased. However, the specificity and OGTT% increased as the cutoff to diagnose diabetes increased. Based on the performance of the IFG alone criteria, cutoff values which can minimize OGTT% with sensitivity greater than 80% and specificity greater than 95% were selected. The proposed algorithms are summarized in Fig 1. As shown in Fig 1B, use serum GA < 14% to exclude and ≥ 17% to diagnose diabetes, the sensitivity, specificity and OGTT% were 83.3%, 98.2% and 35%, respectively. As seen in Fig 1C, use of FPG < 5.56mmol/L and serum GA < 15% to exclude and of FPG ≥ 7.0 mmol/L or serum GA ≥ 17% to diagnose diabetes, the sensitivity, specificity and OGTT% were 85.6%, 98.2% and 22.5%, respectively.
Proportions of population in specific diagnostic category were shown. (A) By impaired fasting glucose (IFG) criteria, that is, fasting plasma glucose (FPG) < 100 mg/dL (5.56 mmol/L) to exclude and FPG ≥ 126 mg/dL (7.0 mmol/L) to diagnose diabetes. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), false positive rate (FPR), and false negative rate (FNR) for this strategy were 78.8%, 100%, 100%, 98.1%, 0%, and 21.2%, respectively. (B) By serum glycated albumin (GA) at 14% and 17%, the sensitivity, specificity, PPV, NPV, FPR, and FNR for this strategy were 83.3%, 98.2%, 80.9%, 98.5%, 1.8%, and 16.7%, respectively. (C) By combine IFG and GA criteria, that is, FPG < 100 mg/dL (5.56 mmol/L) and GA < 15% to exclude diabetes and FPG ≥ 126 mg/dL (7.0 mmol/L) or GA ≥ 17% to diagnose diabetes. The sensitivity, specificity, PPV, NPV, FPR, and FNR for this strategy were 85.6%, 98.2%, 81.3%, 98.7%, 1.8%, and 14.4%, respectively.
Using serum GA alone as the criterion for diagnosis of diabetes (serum GA < 14% to exclude and ≥ 17% to diagnose diabetes), 22 subjects with diabetes and 26 subjects without diabetes were falsely diagnosed. The clinical features of the false negative and the false positive subgroups were therefore compared; those with diabetes but who were falsely diagnosed had higher BMI, WC, and serum albumin concentrations (Table 4).
This report presents several new findings. First, findings of this study reveal that GA can be measured accurately in plasma or serum samples, and in samples from non-fasting subjects. Second, quadratic relationships were demonstrated among serum GA, plasma glucose, and HbA1c for subjects with and without diabetes. Third, this study is the first to develop two screening strategies to diagnose diabetes according to serum GA values. Both strategies were found capable of reducing the need for an OGTT and, when compared with the performance of IFG criteria alone, performed well. These findings support the use of serum GA as an alternative to HbA1c under conditions wherein use of the latter is known to be doubtful for diagnosis of diabetes.
The two strategies developed in the present study for diagnosis of diabetes according to serum GA values are illustrated in Fig 1. Use of serum GA alone to screen for diabetes (Fig 1B) is convenient for patients because fasting is not required. When compared with the IFG criteria alone (Fig 1A), use of serum GA is associated with better sensitivity (83.3% vs. 78.8%) and a lower false-negative percentage (16.7 vs. 21.2); however, more subjects will require an OGTT (35% vs. 14.4%). On the other hand, if fasting is not a concern, use of FPG with serum GA as a screening strategy improves use of IFG criteria alone as a screening strategy (Fig 1C). In particular, sensitivity is increased from 78.8% to 85.6% and the false negative percentage reduced from 21.2 to 14.4 with an acceptable increase in the percentage of subjects requiring an OGTT (IFG criteria alone, 14.4; FPG plus serum GA criteria, 22.5). Two other studies using serum GA to diagnose diabetes have been reported but only one GA cutoff value was used in these studies. In the first which examined Japanese subjects , the sensitivity and specificity for GA as the diagnostic criterion were 83.3% and 83.3%, respectively. In the second which examined Chinese subjects , the sensitivity and specificity for GA as the diagnostic criterion were 73.3% and 80.1%, respectively. The present study employed two cutoffs, one to exclude and one to diagnose diabetes, such that sensitivity and specificity were improved significantly (sensitivity 83.3% and specificity 98.2%). It should be noted that the authors of the study of Chinese subjects  suggested that measurements of both FPG and serum GA (FPG at 6.1 mmol/L and GA at 17.1% as the cutoffs) had better identified those who had diabetes; however, those subjects were high-risk group for diabetes (i.e., those with FPG ≥ 6.1 mmol/L, with random plasma glucose ≥ 7.8 mmol/L, or with a positive family history of diabetes). By contrast, the present study was performed with subjects derived from the overall community.
The good performance of serum GA in the diagnosis of diabetes by the OGTT in the present study is supported by the findings of the two studies described above [34, 35], with the area under the ROC curve ranging from 0.86 to 0.91. In the present study, the optimal cutoff for serum GA was 15%, with a sensitivity of 74% and a specificity of 85%. Similarly, the optimal cutoff for plasma GA in the study of Japanese subjects was 15.5%, with a sensitivity of 83.3% and a specificity of 83.3% . In the study of Chinese subjects, a GA cutoff point of 15.7% was observed, with a sensitivity of 73.3% and a specificity of 80.3% . Taken together, these findings favor an optimal serum GA cutoff for 15–16% for Asian populations.
Findings of the present study support the convenience to patients of GA measurements. As with HbA1c measurement, fasting is not required, as was evidenced by the finding that serum GA values were similar in fasting and post-prandial samples. The correlation coefficient for serum GA and plasma GA was very high (0.9987), indicating that GA can be measured in either specimen type. By contrast, HbA1c can only be measured in whole blood samples.
In a previous study conducted by the authors of this report , strategies using HbA1c to diagnose diabetes were proposed. Using HbA1c alone (HbA1c < 5.9% [41 mmol/mol] to exclude and ≥ 7.0% [53 mmol/mol] to diagnose diabetes), the performance was found to be better than using serum GA alone (HbA1c criteria vs. serum GA criteria: sensitivity of 89.1% vs. 83%, false negative percentage of 10.9 vs. 16.7, and OGTT% of 26.5 vs. 35). However, using IFG plus HbA1c criteria (FPG < 5.56 mmol/L and HbA1c < 6.1% [43 mmol/mol] to exclude and FPG ≥ 5.56 mmol/L and HbA1c ≥ 7% [53 mmol/mol] to diagnose diabetes), the performance was similar to IFG plus GA criteria (IFG plus HbA1c criteria vs. IFG plus GA criteria: sensitivity of 85.2% vs. 85.6%, false negative percentage of 10.9 vs. 14.4, and OGTT% of 18.9 vs. 22). These findings strongly support the measurement of serum GA as a reasonable alternative to the measurement of HbA1c in the diagnosis of diabetes.
Serum GA values are influenced by several factors associated with albumin turnover independently of glycemia status, such as liver cirrhosis and thyroid dysfunction. For example, thyroid hormone is known to promote albumin catabolism. Furthermore, serum GA correlates positively and significantly with serum TSH and negatively and significantly with free T3and free T4 . Additionally, GA values are reported to be influenced by age and nutritional status independently of glycemic status in non-diabetic subjects with end-stage renal disease . In diabetic patients with CKD, nephrotic-range proteinuria decreases GA values regardless of the glycemic status whereas non-nephrotic range proteinuria has no significant influence on GA values .
As indicated above, serum GA reflects short-term (2–3 weeks) mean glycemic values whereas HbA1c reflects longer-term (2–3 months) mean glycemic values [18, 19]. In vivo non-enzymatic glycation of albumin is approximately nine times that of hemoglobin, with the glycation reaction occurring ten times more rapidly for albumin . Because glycation of plasma proteins occurs more readily than glycation of hemoglobin, glycated albumin should serve well in the evaluation of glycemic control for both type 1 and type 2 diabetes . Because serum GA is not affected by erythrocyte survival time , GA measurements appear appropriate for patients with anemia or hemoglobinopathies . Accordingly, serum GA has been suggested as an alternative marker to HbA1c in subjects with nephropathy , retinopathy , gestational diabetes, and iron-deficiency anemia . Although HbA1c measurements underestimate glycemic values in diabetic patients on dialysis, GA measurements accurately reflect glycemic status for these patients [21–23, 44, 45].
In the present study, serum GA was found to be positively associated with age. A similar association was observed in other studies [34, 37, 46]. The finding that serum GA is inversely correlated with both BMI and WC was surprising but is also in agreement with the findings of other studies [47–50]. Since GA is also correlated with high-sensitivity C-reactive protein, this observation may be explained by the chronic, low-grade inflammation observed in obese subjects and which is known to increase the rate of albumin catabolism and to reduce rate of albumin synthesis [51, 52]. The negative association between serum GA, BMI, and WC may also explain why subjects with diabetes who were falsely diagnosed by using serum GA alone had higher BMI and WC. The quadratic relationships of serum GA to FPG, to 2-h PG, and to HbA1c observed in the present study are in agreement with the findings of a study of subjects receiving health examinations . Subjects with diabetes were diagnosed with the disease for the first time during that study and during the present study. The observed curvilinear relationships are therefore proposed to be attributable to the capacity of serum GA to measure recently-occurring (within the previous 2–3 weeks) hyperglycemia. By contrast, in studies for which subjects with a relatively stable glycemic status were recruited, linear relationships between serum GA and HbA1c were observed [24, 34, 35, 45, 46]. For example, the DCCT study  recruited subjects with type 1 diabetes well-controlled by medical treatment and who had relatively stable glycemic values as compared with the subjects in the present study with recently-diagnosed diabetes.
Strengths of the present study include its large sample size, permitting feasible and reliable statistical analyses to be performed. The present study is also comprehensive. For example, the associations between GA and several potential confounding factors were analyzed, the conversion of serum GA to HbA1c values was possible, and different diagnostic strategies for use of serum GA in the diagnosis of diabetes by the OGTT were developed, including one with a single GA cutoff value, one with two cutoff GA values, and one using both IFG and GA criteria. However, the findings of this study, which included Han Chinese adults only, may not prove applicable to other ethnic groups. Nonetheless, OGTT is not repeated in the present study when the results are in the diabetic range, which is different from the recommendations by the ADA and is another limitation of the present study.
In the diagnosis of diabetes mellitus, serum GA measurements with or without IFG criteria are more sensitive than IFG criteria alone when assessing the need for an OGTT. The optimal cutoff value for serum GA alone in diagnosis of diabetes is 15%. Use of FPG < 100 mg/dL (5.56 mmol/L) with serum GA < 15% to exclude and FPG ≥ 126 mg/dL (7.0 mmol/L) or serum GA ≥ 17% to diagnose diabetes further increases the sensitivity of the screening strategy and reduces the proportion of subjects requiring an OGTT. Measurement of serum GA is a more practical and time-saving than performance of an OGTT and can be used to determine the need for an OGTT in the diagnosis of diabetes mellitus. Serum GA presents a useful alternative to HbA1c under conditions wherein the latter does not reflect glycemic status accurately in the diagnosis of diabetes.
S1 Fig. The relationship between serum glycated albumin (GA) and (A) age, (B) gender, (C) serum albumin, (D) body mass index, and (E) waist circumference in subjects without diabetes.
The correlation coefficient between serum GA and age was 0.27 (p < 0.001). The regression coefficient between serum GA and age per 10 years was 0.31 (p < 0.001). Means ± SDs of serum GA were 14.0 ± 2.6% for women and 13.9 ± 2.6% for men (p = 0.26). The correlation coefficient between serum GA and albumin was -0.1179 (age-adjusted p = 0.001). The age-adjusted regression coefficient between serum GA and serum albumin was -0.32, suggesting that every 1 g/dl increase in serum albumin is associated with a decrease in serum GA of 0.32%. The correlation coefficient between serum GA and BMI was -0.2391 (age-adjusted p < 0.001). The age-adjusted regression coefficient between serum GA and BMI was -0.12, indicating that every 1 kg/m2 increase in BMI is associated with a decrease in serum GA of 0.12%. The correlation coefficient between serum GA and WC was -0.1607 in subjects without diabetes (age-adjusted p < 0.001). The age-adjusted regression coefficient between serum GA and WC was -0.04, suggesting that every 1 cm increase in WC is associated with a decrease in serum GA of 0.04%.
S2 Fig. The method of sample collection on the value of glycated albumin (GA).
(A) The relationship of GA in matched serum and plasma samples (r = 0.9987, p < 0.001). Serum GA = 0.98 x plasma GA + 0.29. (B) Serum GA in fasting blood samples and in blood samples collected 2 hours after OGTT. The means ± SDs of GA were 17.9 ± 6.7% in fasting samples and 18.0 ± 6.5% in samples collected 2 hours after OGTT (p = 0.8). (C) Plasma GA in fasting blood samples and post-prandial blood samples collected 2 hours after a meal. The means ± SDs of GA were 15.8 ± 4.3% in fasting samples and 15.8 ± 4.3% in post-prandial blood samples (p = 0.8).
S3 Fig. The relationship between serum glycated albumin, (A) fasting plasma glucose, (B) plasma glucose 2 hours after OGTT, and (C) hemoglobin A1c.
Open circles, subjects without diabetes; close circles, subjects with diabetes. Diabetes was diagnosed by an oral glucose tolerance test.
The authors would like to thank the staff of the eighth Core Lab, Department of Medical Research, National Taiwan University Hospital for their technical and computing assistance.
Conceived and designed the experiments: HYL LMC. Performed the experiments: MSL SRS CHH YJL HYL. Analyzed the data: JNW WYM WCW. Contributed reagents/materials/analysis tools: HYL WYM JNW CHH. Wrote the paper: WCW WYM TYY HYL.
- 1. American Diabetes A. Diagnosis and classification of diabetes mellitus. Diabetes care. 2012;35 Suppl 1:S64–71. doi: 10.2337/dc12-s064 pmid:22187472; PubMed Central PMCID: PMC3632174.
- 2. International Expert C. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes care. 2009;32(7):1327–34. doi: 10.2337/dc09-9033 pmid:19502545; PubMed Central PMCID: PMC2699715.
- 3. Organization WH. Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus 2011.
- 4. Federation ID. Global Guideline for T2DM2012.
- 5. Li H-Y, Ma W-Y, Wei J-N, Lin M-S, Shih S-R, Hung CS, et al. Hemoglobin A1c for the diagnosis of diabetes: To replace or to guide oral glucose tolerance tests? Journal of Diabetes Investigation. 2012;3(3):259–65. doi: 10.1111/j.2040-1124.2011.00181.x. pmid:24843574
- 6. Hu Y, Liu W, Chen Y, Zhang M, Wang L, Zhou H, et al. Combined use of fasting plasma glucose and glycated hemoglobin A1c in the screening of diabetes and impaired glucose tolerance. Acta diabetologica. 2010;47(3):231–6. doi: 10.1007/s00592-009-0143-2 pmid:19760291.
- 7. Anand SS, Razak F, Vuksan V, Gerstein HC, Malmberg K, Yi Q, et al. Diagnostic strategies to detect glucose intolerance in a multiethnic population. Diabetes care. 2003;26(2):290–6. pmid:12547851.
- 8. Ko GT, Chan JC, Yeung VT, Chow CC, Tsang LW, Li JK, et al. Combined use of a fasting plasma glucose concentration and HbA1c or fructosamine predicts the likelihood of having diabetes in high-risk subjects. Diabetes care. 1998;21(8):1221–5. pmid:9702423.
- 9. DAROC. DAROC Clinical Practice Guidelines for Diabetes Care—2015.: Taiwan, Diabetes Association of the Republic of China.
- 10. Little RR, Roberts WL. A review of variant hemoglobins interfering with hemoglobin A1c measurement. Journal of diabetes science and technology. 2009;3(3):446–51. pmid:20144281; PubMed Central PMCID: PMC2769887.
- 11. Program NGS. Factors that Interfere with HbA1c Test Results http://www.ngsp.org/factors.asp 2014 [updated January, 2014].
- 12. Panzer S, Kronik G, Lechner K, Bettelheim P, Neumann E, Dudczak R. Glycosylated hemoglobins (GHb): an index of red cell survival. Blood. 1982;59(6):1348–50. pmid:7082831.
- 13. Sundaram RC, Selvaraj N, Vijayan G, Bobby Z, Hamide A, Rattina Dasse N. Increased plasma malondialdehyde and fructosamine in iron deficiency anemia: effect of treatment. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie. 2007;61(10):682–5. doi: 10.1016/j.biopha.2007.06.013 pmid:17698317.
- 14. Hashimoto K, Noguchi S, Morimoto Y, Hamada S, Wasada K, Imai S, et al. A1C but not serum glycated albumin is elevated in late pregnancy owing to iron deficiency. Diabetes care. 2008;31(10):1945–8. doi: 10.2337/dc08-0352 pmid:18599529; PubMed Central PMCID: PMC2551632.
- 15. Organization WH. Worldwide Prevalence of Anaemia 1993–20052008.
- 16. Kuo HW, Tsai SS, Tiao MM, Yang CY. Epidemiological features of CKD in Taiwan. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2007;49(1):46–55. doi: 10.1053/j.ajkd.2006.10.007 pmid:17185145.
- 17. Coresh J, Selvin E, Stevens LA, Manzi J, Kusek JW, Eggers P, et al. Prevalence of chronic kidney disease in the United States. JAMA: the journal of the American Medical Association. 2007;298(17):2038–47. doi: 10.1001/jama.298.17.2038 pmid:17986697.
- 18. Furusyo N, Hayashi J. Glycated albumin and diabetes mellitus. Biochimica et biophysica acta. 2013;1830(12):5509–14. doi: 10.1016/j.bbagen.2013.05.010 pmid:23673238.
- 19. Koga M, Kasayama S. Clinical impact of glycated albumin as another glycemic control marker. Endocrine journal. 2010;57(9):751–62. pmid:20724796.
- 20. Kosecki SM, Rodgers PT, Adams MB. Glycemic monitoring in diabetics with sickle cell plus beta-thalassemia hemoglobinopathy. The Annals of pharmacotherapy. 2005;39(9):1557–60. doi: 10.1345/aph.1G010 pmid:16076916.
- 21. Peacock TP, Shihabi ZK, Bleyer AJ, Dolbare EL, Byers JR, Knovich MA, et al. Comparison of glycated albumin and hemoglobin A(1c) levels in diabetic subjects on hemodialysis. Kidney international. 2008;73(9):1062–8. doi: 10.1038/ki.2008.25 pmid:18288102.
- 22. Inaba M, Okuno S, Kumeda Y, Yamada S, Imanishi Y, Tabata T, et al. Glycated albumin is a better glycemic indicator than glycated hemoglobin values in hemodialysis patients with diabetes: effect of anemia and erythropoietin injection. Journal of the American Society of Nephrology: JASN. 2007;18(3):896–903. doi: 10.1681/ASN.2006070772 pmid:17267743.
- 23. Freedman BI, Shihabi ZK, Andries L, Cardona CY, Peacock TP, Byers JR, et al. Relationship between assays of glycemia in diabetic subjects with advanced chronic kidney disease. American journal of nephrology. 2010;31(5):375–9. doi: 10.1159/000287561 pmid:20299782.
- 24. Nathan DM, McGee P, Steffes MW, Lachin JM, Group DER. Relationship of glycated albumin to blood glucose and HbA1c values and to retinopathy, nephropathy, and cardiovascular outcomes in the DCCT/EDIC study. Diabetes. 2014;63(1):282–90. doi: 10.2337/db13-0782 pmid:23990364; PubMed Central PMCID: PMC3868040.
- 25. Li ZL, Liu JC, Liu SB, Li XQ, Yi DH, Zhao MG. Improvement of vascular function by acute and chronic treatment with the GPR30 agonist G1 in experimental diabetes mellitus. PloS one. 2012;7(6):e38787. doi: 10.1371/journal.pone.0038787 pmid:22679517; PubMed Central PMCID: PMC3367949.
- 26. Murea M, Moran T, Russell GB, Shihabi ZK, Byers JR, Andries L, et al. Glycated albumin, not hemoglobin A1c, predicts cardiovascular hospitalization and length of stay in diabetic patients on dialysis. American journal of nephrology. 2012;36(5):488–96. doi: 10.1159/000343920 pmid:23147746.
- 27. Song SO, Kim KJ, Lee BW, Kang ES, Cha BS, Lee HC. Serum glycated albumin predicts the progression of carotid arterial atherosclerosis. Atherosclerosis. 2012;225(2):450–5. doi: 10.1016/j.atherosclerosis.2012.09.005 pmid:23040867.
- 28. Xue Y, Sheng Y, Dai H, Cao H, Liu Z, Li Z. Risk of development of acute pancreatitis with pre-existing diabetes: a meta-analysis. European journal of gastroenterology & hepatology. 2012;24(9):1092–8. pmid:22668873.
- 29. Li HY, Wei JN, Lin MS, Smith DJ, Vainio J, Lin CH, et al. Serum vascular adhesion protein-1 is increased in acute and chronic hyperglycemia. Clinica chimica acta; international journal of clinical chemistry. 2009;404(2):149–53. doi: 10.1016/j.cca.2009.03.041 pmid:19336232.
- 30. Lin MS, Li HY, Wei JN, Lin CH, Smith DJ, Vainio J, et al. Serum vascular adhesion protein-1 is higher in subjects with early stages of chronic kidney disease. Clinical biochemistry. 2008;41(16–17):1362–7. doi: 10.1016/j.clinbiochem.2008.06.019 pmid:18644360.
- 31. Ma WY, Yang CY, Shih SR, Hsieh HJ, Hung CS, Chiu FC, et al. Measurement of Waist Circumference: midabdominal or iliac crest? Diabetes care. 2013;36(6):1660–6. doi: 10.2337/dc12-1452 pmid:23275359; PubMed Central PMCID: PMC3661855.
- 32. Expert Committee on the D, Classification of Diabetes M. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes care. 2003;26 Suppl 1:S5–20. pmid:12502614.
- 33. Little RR, Rohlfing CL, Wiedmeyer HM, Myers GL, Sacks DB, Goldstein DE, et al. The national glycohemoglobin standardization program: a five-year progress report. Clinical chemistry. 2001;47(11):1985–92. pmid:11673367.
- 34. Furusyo N, Koga T, Ai M, Otokozawa S, Kohzuma T, Ikezaki H, et al. Utility of glycated albumin for the diagnosis of diabetes mellitus in a Japanese population study: results from the Kyushu and Okinawa Population Study (KOPS). Diabetologia. 2011;54(12):3028–36. doi: 10.1007/s00125-011-2310-6 pmid:21947435.
- 35. Ma XJ, Pan JM, Bao YQ, Zhou J, Tang JL, Li Q, et al. Combined assessment of glycated albumin and fasting plasma glucose improves the detection of diabetes in Chinese subjects. Clinical and experimental pharmacology & physiology. 2010;37(10):974–9. doi: 10.1111/j.1440-1681.2010.05417.x pmid:20557319.
- 36. Koga M, Murai J, Saito H, Matsumoto S, Kasayama S. Effects of thyroid hormone on serum glycated albumin levels: study on non-diabetic subjects. Diabetes research and clinical practice. 2009;84(2):163–7. doi: 10.1016/j.diabres.2009.01.013 pmid:19243849.
- 37. Okada T, Nakao T, Matsumoto H, Yamanaka T, Nagaoka Y, Tamekuni T. Influence of age and nutritional status on glycated albumin values in hemodialysis patients. Internal medicine. 2009;48(17):1495–9. pmid:19721292.
- 38. Okada T, Nakao T, Matsumoto H, Nagaoka Y, Tomaru R, Iwasawa H, et al. Influence of proteinuria on glycated albumin values in diabetic patients with chronic kidney disease. Internal medicine. 2011;50(1):23–9. pmid:21212569.
- 39. Iberg N, Fluckiger R. Nonenzymatic glycosylation of albumin in vivo. Identification of multiple glycosylated sites. The Journal of biological chemistry. 1986;261(29):13542–5. pmid:3759977.
- 40. Yoshiuchi K, Matsuhisa M, Katakami N, Nakatani Y, Sakamoto K, Matsuoka T, et al. Glycated albumin is a better indicator for glucose excursion than glycated hemoglobin in type 1 and type 2 diabetes. Endocrine journal. 2008;55(3):503–7. pmid:18445997.
- 41. Koga M, Murai J, Saito H, Aoki K, Kanehara H, Bando Y, et al. Glycated albumin levels are higher relative to HbA1c levels in people with autoimmune acute-onset type 1 diabetes mellitus than in people with type 2 diabetes mellitus at the time of diagnosis. Diabetes research and clinical practice. 2011;94(1):e12–4. doi: 10.1016/j.diabres.2011.06.022 pmid:21777990.
- 42. Okumura A, Mitamura Y, Namekata K, Nakamura K, Harada C, Harada T. Glycated albumin induces activation of activator protein-1 in retinal glial cells. Japanese journal of ophthalmology. 2007;51(3):236–7. doi: 10.1007/s10384-007-0431-8 pmid:17554491.
- 43. Hashimoto K, Osugi T, Noguchi S, Morimoto Y, Wasada K, Imai S, et al. A1C but not serum glycated albumin is elevated because of iron deficiency in late pregnancy in diabetic women. Diabetes care. 2010;33(3):509–11. doi: 10.2337/dc09-1954 pmid:20032277; PubMed Central PMCID: PMC2827498.
- 44. Zheng CM, Ma WY, Wu CC, Lu KC. Glycated albumin in diabetic patients with chronic kidney disease. Clinica chimica acta; international journal of clinical chemistry. 2012;413(19–20):1555–61. doi: 10.1016/j.cca.2012.04.025 pmid:22579765.
- 45. Sany D, Elshahawy Y, Anwar W. Glycated albumin versus glycated hemoglobin as glycemic indicator in hemodialysis patients with diabetes mellitus: variables that influence. Saudi journal of kidney diseases and transplantation: an official publication of the Saudi Center for Organ Transplantation, Saudi Arabia. 2013;24(2):260–73. pmid:23538348.
- 46. Yang C, Li H, Wang Z, Zhang W, Zhou K, Meng J, et al. Glycated albumin is a potential diagnostic tool for diabetes mellitus. Clin Med. 2012;12(6):568–71. pmid:23342412.
- 47. Koga M, Matsumoto S, Saito H, Kasayama S. Body mass index negatively influences glycated albumin, but not glycated hemoglobin, in diabetic patients. Endocrine journal. 2006;53(3):387–91. pmid:16717395.
- 48. Wang F, Ma X, Hao Y, Yang R, Ni J, Xiao Y, et al. Serum glycated albumin is inversely influenced by fat mass and visceral adipose tissue in Chinese with normal glucose tolerance. PloS one. 2012;7(11):e51098. doi: 10.1371/journal.pone.0051098 pmid:23209844; PubMed Central PMCID: PMCPMC3510195.
- 49. Poon AK, Juraschek SP, Ballantyne CM, Steffes MW, Selvin E. Comparative associations of diabetes risk factors with five measures of hyperglycemia. BMJ Open Diabetes Res Care. 2014;2(1):e000002. doi: 10.1136/bmjdrc-2013-000002 pmid:25452853; PubMed Central PMCID: PMCPMC4212576.
- 50. Huh JH, Kim KJ, Lee BW, Kim DW, Kang ES, Cha BS, et al. The relationship between BMI and glycated albumin to glycated hemoglobin (GA/A1c) ratio according to glucose tolerance status. PloS one. 2014;9(2):e89478. doi: 10.1371/journal.pone.0089478 pmid:24586809; PubMed Central PMCID: PMCPMC3938490.
- 51. Koga M, Otsuki M, Matsumoto S, Saito H, Mukai M, Kasayama S. Negative association of obesity and its related chronic inflammation with serum glycated albumin but not glycated hemoglobin levels. Clinica chimica acta; international journal of clinical chemistry. 2007;378(1–2):48–52. doi: 10.1016/j.cca.2006.10.013 pmid:17141207.
- 52. Koga M, Hirata T, Kasayama S, Ishizaka Y, Yamakado M. Body mass index negatively regulates glycated albumin through insulin secretion in patients with type 2 diabetes mellitus. Clinica chimica acta; international journal of clinical chemistry. 2015;438:19–23. doi: 10.1016/j.cca.2014.07.035 pmid:25102071.