ABSTRACT
Background
Glycated albumin (GA), 1,5‐anhydroglucitol (1,5‐AG), and fructosamine have attracted considerable interest as markers of hyperglycemia. This study aimed to evaluate the optimal cutoff values for GA, 1,5‐AG, and fructosamine and to determine their respective diagnostic efficacies in relation to hyperglycemia.
Methods
We enrolled 6012 individuals who had undergone fasting blood glucose (FBG) and Hemoglobin A1c (HbA1c) tests along with at least one alternative glycemic marker. Receiver operating characteristic (ROC) curves and the upper or lower limit of the reference range (97.5 or 2.5 percentiles) were used to ascertain the optimal cutoff values. Follow‐up data from healthy individuals were used to identify patients who developed diabetes mellitus (DM).
Results
The ROC cutoff values for GA, 1,5‐AG, and fructosamine were 13.9%, 13.3 μg/mL, and 278 μmol/L, respectively, with corresponding area under the curve (AUC) values of 0.860, 0.879, and 0.834. The upper limits of the reference intervals for GA and fructosamine were 15.1% and 279 μmol/L, respectively, and the lower limit for 1,5‐AG was 5.3 μg/mL. Among the GA cutoff values, the ROC cutoff had the highest sensitivity. Analyzing the follow‐up data showed that lowering the GA cutoff from 16.0% to 13.9% identified an additional 40 people with DM progression.
Conclusions
Lowering the GA cutoff values significantly increased the sensitivity of DM diagnosis and enhanced its potential as a screening marker by identifying more individuals with diabetes progression. Conversely, modifications to the cutoff values for 1,5‐AG and fructosamine did not confer any discernible diagnostic or predictive advantages.
Keywords: 1,5‐anhydroglucitol; cutoff; diabetes; fructosamine; glycated albumin; hyperglycemia
We assessed the optimal cut‐off values for glycated albumin (GA), 1,5‐anhydroglucitol (1,5‐AG), and fructosamine using receiver operating characteristic (ROC) curve and the upper or lower limit of the reference range (97.5 or 2.5 percentiles). Lowering the GA cutoff values yielded a marked enhancement in sensitivity and detected more individuals with diabetes progression. In contrast, adjusting the cutoff values for 1,5‐AG and fructosamine offered no noticeable diagnostic benefits.
1. Introduction
Diabetes mellitus (DM) is a chronic disease that is the leading cause of end‐stage renal disease, nontraumatic amputations, and new blindness in adults aged 20–74 years [1]. Early detection of hyperglycemia in clinical and healthcare programs could prevent the progression from prediabetes to diabetes and reduce the risk of complications in DM patients [2, 3, 4].
DM is diagnosed based on hemoglobin A1c (HbA1c) or blood glucose criteria, either via the fasting blood glucose (FBG) or oral glucose tolerance test (OGTT); however, these tests have limitations [5]. The disadvantage of the OGTT is that it requires a longer stay at a medical facility for the procedure, as it requires a glycemic load and multiple blood tests. Meanwhile, FBS tends to overdiagnose diabetes, and the results have imperfect concordance with OGTT results [6]. In contrast, HbA1c has several advantages over FBG and OGTT, including convenience in testing, preanalytical stability, and fewer day‐to‐day perturbations, as HbA1c reflects the mean glucose levels over the preceding 3 months [7]. However, these advantages may be offset by lower sensitivity, underdiagnosis of diabetes, analytical costs, and interference from red blood cell turnover, including anemia, hemolysis, or hemoglobinopathies. Additionally, the 3‐month period reflected in HbA1c is an inappropriately long time for the diagnosis of DM and monitoring of treatment response [8, 9].
Glycated albumin (GA), 1,5‐anhydroglucitol (1,5‐AG), and fructosamine are of increasing interest in research and clinical practice as alternatives to or additional markers of FBG and HbA1c [10]. These biomarkers reflect a shorter integration period for the hyperglycemic state, from 1 week to 1 month, as opposed to 2–3 months for HbA1c [11, 12, 13]. GA levels were measured as the ratio of serum GA to total albumin. As albumin has a shorter half‐life and is more sensitive to glycation than hemoglobin, GA reflects 2–3 weeks of hyperglycemia [14, 15]. The 1‐deoxy form of glucose, 1,5‐AG, is ubiquitous in various food sources [16, 17]. This marker is reabsorbed in the renal tubules and competes with very high glucose levels for reabsorption. High circulating glucose levels over a few days induce lower serum 1,5‐AG levels and glycosuria [18, 19]. When glucose binds to any protein, including albumin, it forms a ketoamine called fructosamine through glycation [20]. Fructosamine reflects blood glucose levels over the previous 2–4 weeks because glycated proteins have a more rapid turnover than HbA1c, which depends on erythroid turnover. Therefore, GA and fructosamine are not affected by erythrocyte or hemoglobin disorders and are more susceptible to glycemic variability [21, 22].
Despite the prospective convenience and cost‐effectiveness of these alternative markers, routine clinical implementation is limited. One of the main reasons for this is the need for more consensuses on clinically useful cutoffs for the diagnosis of DM, owing to the lack of large‐scale studies evaluating the reference levels of these markers reflecting hyperglycemia and estimating their diagnostic performance. Therefore, this study aimed to establish cutoff values for these three alternative markers using health examination data and to estimate the diagnostic performance of each cutoff value for current hyperglycemia and early detection of DM progression.
2. Materials and Methods
2.1. Study Participants
Data were obtained from the Kangbuk Samsung Health Study, a cohort study comprising Koreans who underwent comprehensive examinations at the Kangbuk Samsung Hospital Total Healthcare Center in Seoul, Republic of Korea, from August 2013 to September 2014 [23, 24]. Of the 6274 individuals who were tested for FBG, HbA1c, and at least one alternative glycemic marker (GA, 1,5‐AG, or fructosamine), 262 who had already been diagnosed with DM were excluded to mitigate potential bias from DM treatment and lifestyle changes. The remaining individuals (n = 6012) were enrolled in this study.
This study was approved by the Institutional Review Board of the Kangbuk Samsung Hospital (Approval No. 2016‐12‐015). The need for informed consent was waived because of the use of de‐identified retrospective data routinely collected during the health screening process.
2.2. Laboratory Examination
All tests were conducted at the laboratory department of Kangbuk Samsung Hospital Total Healthcare Center in Seoul. Blood samples were obtained after an 8‐h fast from the antecubital vein. Serum GA, 1,5‐AG, and fructosamine levels were measured using an automatic chemistry analyzer (Modular P800; Roche Diagnostics, Tokyo, Japan). Lucica GA‐L reagent (Asahi Kasei Pharma Co., Tokyo, Japan), Determiner L 1,5‐AG reagent (Kyowa Medex, Tokyo, Japan), and fructosamine reagent (Roche Diagnostics GmbH, Mannheim, Germany) were used for each test based on the enzymatic method for GA and the colorimetric method for 1,5‐AG and fructosamine. The upper or lower limit of reference interval provided by the manufacturer for these tests is GA 16.0%, 1,5‐AG >14.0 μg/mL, and fructosamine 285 μmol/L. These values are commonly used in clinical settings and are referred to as conventional reference values in our study [25, 26, 27, 28, 29, 30]. FBG levels were measured using the hexokinase method with a Cobas 8000 c702 (Roche Diagnostics), and HbA1c levels were measured using a turbidimetric inhibition immunoassay with an Integra 800 (Roche Diagnostics, Rotkreuz, Switzerland). The Laboratory Medicine Department of Kangbuk Samsung Hospital in Seoul, Korea is accredited by the Korean Society of Laboratory Medicine (KSLM) and the Korean Association of Quality Assurance for Clinical Laboratories (KAQACL) and participates in the College of American Pathologists (CAP) Survey Proficiency Testing.
2.3. Data Analysis
Demographic data and medical history were collected using standardized and self‐administered questionnaires (Table 1). The study participants examined for each alternative marker numbered 4848 for GA, 2309 for 1,5‐AG, and 2539 for fructosamine and were used to calculate cutoff values using the receiver operating characteristic (ROC) curve analysis (Figure 1). We evaluated the upper limit of the reference intervals of GA and fructosamine and the lower limit of 1,5‐AG in healthy individuals after excluding those with hyperglycemia. Individuals with hyperglycemia were identified using FBG ≥126 mg/dL or HbA1c ≥6.5% as the diagnostic criteria [5]. After dividing the study group by sex, mean comparisons and ROC analyses were conducted for the three markers.
TABLE 1.
Demographic characteristics and initial examination results of the study participants (n = 6012).
| Variables | N | Mean ± SD or percentage | Range |
|---|---|---|---|
| Questionnaire | |||
| Sex—female | 2317/6012 | 38.5% | |
| Age, years | 6012 | 42.1 ± 9.0 | 15–77 |
| Diagnosed HTN | 687/5894 | 11.7% | |
| Diagnosed dyslipidemia | 1171/5876 | 19.9% | |
| Alcohol intake | |||
| None | 1568/5544 | 28.3% | |
| ≤3 times/week | 3653/5544 | 65.9% | |
| ≥4 times/week | 323/5544 | 5.8% | |
| Smoking | |||
| Non | 2843/5554 | 51.2% | |
| Current | 1195/5554 | 21.5% | |
| Ex | 1516/5554 | 27.3% | |
| Examination | |||
| Glucose (mg/dL) | 6012 | 94.9 ± 12.2 | 58–344 |
| HbA1c (%) | 6012 | 5.6 ± 0.4 | 4.5–12.3 |
| Hyperglycemia | |||
| FBG ≥126 mg/dL or HbA1c ≥6.5% | 133/6012 | 2.2% | |
| Glycated albumin (%) | 4848 | 12.1 ± 1.8 | 7.0–38.6 |
| 1,5‐Anhydroglucitol (μg/mL) | 2309 | 17.6 ± 7.3 | 0.5–58.1 |
| Fructosamine (μmol/L) | 2539 | 244.1 ± 21.1 | 177–604 |
| Total cholesterol (mg/dL) | 6011 | 200.3 ± 34.7 | 89–450 |
| Triglyceride (mg/dL) | 6011 | 118.3 ± 79.4 | 25–1080 |
| HDL cholesterol (mg/dL) | 5998 | 59.0 ± 15.6 | 13–146 |
| LDL cholesterol (mg/dL) | 5994 | 122.0 ± 31.7 | 30–347 |
| Low GFR | |||
| CKD‐EPI < 60 mL/min/1.73 m2 | 496/6002 | 8.3% | |
| Overweight | |||
| BMI≥25 | 2015/5958 | 33.8% | |
| High BP | |||
| SBP≥140 or DBP≥90 mmHg | 386/5996 | 6.4% | |
FIGURE 1.
Enrollment of study participants and statistical analysis.
Of the individuals identified as healthy at baseline, 4520 were revisited for follow‐up until August 2017. Data collected during these revisits, including FBG level, HbA1c level, and self‐reported DM history, were used to evaluate the performance of alternative markers for the early detection of DM progression. Individuals were considered to have progressed to DM if, at the follow‐up test, they had FBG ≥126 mg/dL, HbA1c ≥6.5%, or a changed response in diagnosed DM history.
2.4. Statistical Analysis
We established optimal cutoff values for each of the three alternative markers using ROC curve analysis and calculated the 97.5 and 2.5 percentile points for these markers within the healthy population to estimate their respective upper and lower limits of the reference intervals. The ROC curves were derived against current hyperglycemia according to FBG and HbA1c levels, and the cutoff values were determined using Youden's J statistic. Established cutoff values and conventional reference values (GA, 16.0%; 1,5‐AG, >14.0 μg/mL; and fructosamine, 285 μmol/L) were used to estimate the diagnostic performance for current hyperglycemia and follow‐up DM progression. We compared sensitivities and specificities using the McNemar test and area under the curve (AUC) using the DeLong test. Pearson's chi‐square test was used to validate the statistical association between the two groups, and Student's t‐test was used to compare the means of the two independent groups. A p‐value of <0.05 was considered statistically significant for all tests.
IBM SPSS version 24.0 (IBM Corp., Armonk, NY, USA), R (Version 4.4.0, R Foundation for Statistical Computing, Vienna, Austria), and Rex (Version 3.5.0, RexSoft Inc., Seoul, South Korea) were used to perform calculations and statistical analyses.
3. Results
The mean age of study participants was 42.1 ± 9.0 (range: 15–77) years old, and the participants included 38.5% females (Table 1). Self‐reported hypertension and dyslipidemia were observed in 11.7% and 19.9% of participants, respectively. All the study participants tested FBG and HbA1c with mean levels of 94.9 ± 12.2 (range, 58–344) mg/dL and 5.6 ± 0.4% (range, 4.5%–12.3%), respectively. Of the traditional DM markers, 2.2% of total participants showed FBG ≥126 mg/dL or HbA1c ≥6.5% and were thus classified as having hyperglycemia. All the individuals in the study were tested for at least one of the alternative hyperglycemic markers, and the mean levels were 12.1 ± 1.8% (range, 7.0%–38.6%) for GA, 17.6 ± 7.3 (range, 0.5–58.1) μg/mL for 1,5‐AG, and 244.1 ± 21.1 (range, 177–604) μmol/L for fructosamine.
The optimal cutoff values for the three alternative markers were calculated by plotting the ROC curves for hyperglycemic states determined by FBG and HbA1c levels tested simultaneously (Figure 2). The cutoff values of GA, 1,5‐AG, and fructosamine were found to be 13.9%, 13.3 μg/mL, and 278 μmol/L, with estimated AUC of 0.860 (95% confidence interval [CI], 0.762–0.938), 0.879 (95% CI, 0.834–0.882) and 0.834 (95% CI, 0.834–0.983), respectively. No significant differences in the AUCs were observed in the DeLong test (p‐values = 0.326–0.707). Furthermore, the upper and lower limits of the reference intervals were calculated using data from healthy individuals without hyperglycemia. These limits were established at the 97.5% percentile for GA (15.1%) and fructosamine (279 μmol/L), and at the 2.5% percentile for 1,5‐AG (5.3 μg/mL). We presented the percent distribution of the three alternative markers within the normal and hyperglycemic groups by introducing the calculated cutoffs and conventional reference values (Figure 3). The conventional reference value and ROC cutoff of 1,5‐AG separated the hyperglycemic population more clearly than the other biomarkers.
FIGURE 2.
ROC analysis of the three alternative markers against the hyperglycemic state determined by FBG and HbA1c levels. The cutoffs were determined using Youden's J statistic. The cutoff values of (a) glycated albumin, (b) 1,5‐anhydroglucitol, and (c) fructosamine were found to be 13.9%, 13.3 μg/mL and 278 μmol/L, with estimated AUCs of 0.860 (95% confidence interval [CI], 0.762–0.938), 0.879 (95% CI, 0.834–0.882) and 0.834 (95% CI, 0.834–0.983), respectively.
FIGURE 3.
Percent distribution of (a) glycated albumin, (b) 1,5‐anhydroglucitol, and (c) fructosamine within normal and hyperglycemic groups with calculated cutoffs and the upper or lower limit of the conventional reference interval. The lower limit of the conventional reference interval and ROC cutoff of 1,5‐anhydroglucitol more clearly separated the hyperglycemic group from normal populations than other markers.
The calculated cutoffs were used to estimate the performance of the three alternative markers in detecting hyperglycemia (Table 2). Among the GA cutoffs, the conventional reference value showed the highest accuracy and a positive likelihood ratio. However, it had only 42% sensitivity, and the conventional reference value was considerably higher than the ROC cutoff, which showed the highest sensitivity. In contrast, the ROC cutoff for 1,5‐AG had a similar value and performance as the conventional reference value. However, the lower limit of the reference interval had a significantly lower level than the other two cutoffs, along with the lowest sensitivity. The three fructosamine cutoffs showed similar values and performance. Comparing the ROC cutoffs of the three alternative markers, fructosamine had the highest accuracy and 1,5‐AG had the highest sensitivity.
TABLE 2.
Diagnostic performances of three alternative markers to detect hyperglycemia, as determined by FBG and HbA1c.
| Alternative markers | Cutoffs | Abnormal population | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%) | LR+ | LR− |
|---|---|---|---|---|---|---|---|---|---|
| Glycated albumin | Conventional (16.0%) | 82/4848 (1.7%) | 42.0 (32.7–51.7) | 99.3 (99.0–99.5) | 57.3 (45.9–68.2) | 98.6 (98.3–99.0) | 97.9 (97.5–98.3) | 56.8 (38.2–84.3) | 0.6 (0.5–0.7) |
| ROC (13.9%) | 626/4848 (12.9%) | 72.3*** (63.1–80.4) | 88.5*** (87.6–89.4) | 12.9 (10.4–15.8) | 99.3 (99.0–99.5) | 88.1 (87.2–89.0) | 6.3 (5.5–7.2) | 0.3 (0.2–0.4) | |
| Upper limit of R.I. (15.1%) | 187 / 4848 (3.9%) | 51.8** (42.2–61.3) | 97.3*** (96.8–97.7) | 31.0 (24.5–38.2) | 98.8 (98.5–99.1) | 96.2 (95.7–96.8) | 19.0 (14.9–24.3) | 0.5 (0.4–0.6) | |
| 1,5‐Anhydroglucitol | Conventional (>14.0 μg/mL) | 782/2309 (33.9%) | 87.8 (75.2–95.4) | 67.3 (65.3–69.2) | 5.5 (4.0–7.3) | 99.6 (99.2–99.9) | 67.7 (65.8–69.6) | 2.7 (2.4–3.0) | 0.2 (0.1–0.4) |
| ROC (>13.3 μg/mL) | 695/2309 (30.1%) | 87.8 (75.2–95.4) | 71.2*** (69.2–73.0) | 6.2 (4.5–8.2) | 99.6 (99.2–99.9) | 71.5 (69.6–73.3) | 3.0 (2.7–3.4) | 0.2 (0.1–0.4) | |
| Lower limit of R.I. (>5.3 μg/mL) | 77/2309 (3.3%) | 40.8*** (27.0–55.8) | 97.5*** (96.7–98.1) | 26.0 (16.6–37.2) | 98.7 (98.1–99.1) | 96.3 (95.4–97.0) | 16.2 (10.6–24.7) | 0.6 (0.5–0.8) | |
| Fructosamine | Conventional (285 μmol/L) | 61/2539 (2.4%) | 43.6 (30.3–57.7) | 98.5 (98.0–99.0) | 39.3 (27.1–52.7) | 98.8 (98.2–99.2) | 97.3 (96.6–97.9) | 29.3 (18.9–45.4) | 0.6 (0.5–0.7) |
| ROC (278 μmol/L) | 103/2539 (4.1%) | 50.9 (37.1–64.7) | 97.0*** (96.2–97.6) | 27.2 (18.9–36.8) | 98.9 (98.4–99.3) | 96.0 (95.1–96.7) | 16.9 (12.0–23.7) | 0.5 (0.4–0.7) | |
| Upper limit of R.I. (279 μmol/L) | 90/2539 (3.5%) | 47.3 (33.7–61.2) | 97.4*** (96.7–98.0) | 28.9 (19.8–39.4) | 98.8 (98.3–99.2) | 96.3 (95.5–97.0) | 18.3 (12.7–26.5) | 0.5 (0.4–0.7) |
Note: Statistically significant differences in sensitivity and specificity compared with those of the manufacturer's cutoffs are marked **p < 0.01 and ***p < 0.001.
Abbreviations: Conventional, upper limit of conventional reference interval; LR+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; R.I., reference interval; ROC, receiver operating characteristic.
Using follow‐up data to evaluate the potential of alternative markers for the early detection of DM progression (Table 3), we observed that lowering the GA cutoff from 16.0% to 13.9% could contribute to the detection of an additional 40 individuals with DM progression. This adjustment increased the sensitivity and AUC from 23.2% to 49.0% and from 0.614 to 0.691, respectively. In contrast, adopting the calculated cutoffs for 1,5‐AG had no advantage in terms of sensitivity and AUC compared with the conventional reference value. Lastly, the two calculated cutoffs of fructosamine also had no significant advantage in terms of sensitivity, specificity, or AUCs compared with the conventional reference value.
TABLE 3.
Diagnostic performances of three alternative markers of DM progression within 3 years of follow‐up visit.
| Alternative markers | Cutoffs | The proportion of individuals having DM progression within the group of | p a | Expected performances for DM progression | |||
|---|---|---|---|---|---|---|---|
| Normal alternative marker | Abnormal alternative marker | Sensitivity (%) (95%CI) | Specificity (%) (95%CI) | AUC (95%CI) | |||
| Glycated albumin (n = 3618) | Conventional (16.0%) | 3.3% (119/3643) | 69.2% (36/52) | <0.001 | 23.2 (16.8–30.7) | 99.6 (99.3–99.7) | 0.614 (0.581–0.647) |
| ROC (13.9%) | 2.4% (79/3235) | 16.5% (76/460) | <0.001 | 49.0*** (40.9–57.2) | 89.2*** (88.1–90.2) | 0.691*** (0.651–0.731) | |
| Upper limit of R.I. (15.1%) | 3.0% (108/3571) | 37.9% (47/124) | <0.001 | 30.3** (23.2–38.2) | 97.8*** (97.3–98.3) | 0.641** (0.604–677) | |
| 1,5‐Anhydroglucitol (n = 1776) | Conventional (>14.0 μg/mL) | 1.3% (16/1207) | 10.6% (64/601) | <0.001 | 80.0 (69.6–88.1) | 68.9 (66.7–71.1) | 0.745 (0.699–0.790) |
| ROC (>13.3 μg/mL) | 1.6% (20/1272) | 11.2% (60/536) | <0.001 | 75.0 (64.1–84.0) | 72.5*** (70.3–74.6) | 0.737 (0.688–0.786) | |
| Lower limit of R.I. (>5.3 μg/mL) | 3.3% (58/1751) | 38.6% (22/57) | <0.001 | 27.5*** (18.1–38.6) | 98.0*** (97.2–98.6) | 0.627*** (0.578–0.677) | |
| Fructosamine (n = 1948) | Conventional (285 μmol/L) | 3.5% (68/1948) | 47.2% (17/36) | <0.001 | 20 (12.1–30.1) | 99.0 (98.4–99.4) | 0.595 (0.552–0.638) |
| ROC (278 μmol/L) | 3.4% (65/1923) | 32.8% (20/61) | <0.001 | 23.5 (15.0–34.0) | 97.8*** (97.1–98.5) | 0.607 (0.561–652) | |
| Upper limit of R.I. (279 μmol/L) | 3.4% (66/1932) | 36.5% (19/52) | <0.001 | 22.4 (14.0–32.7) | 98.3*** (97.6–98.8) | 0.603 (0.558–0.648) | |
Note: The statistical significance of the differences in sensitivity, specificity, and AUC compared with the manufacturer's cutoff values are denoted as **p < 0.01 and ***p < 0.001.
Abbreviations: AUC, area under the curve; Conventional, upper limit of conventional reference interval; R.I., reference interval; ROC, receiver operating characteristic.
Pearson's chi‐square test was used to validate the statistical association between alternative markers and DM progression.
Additionally, the mean comparison of the three markers between the sex groups revealed statistically significant differences (p < 0.05) for all three markers (Table S1). However, the means of GA for females and males were 12.3 ± 1.7% and 12.1 ± 1.9%, respectively, and fructosamine levels were 242.0 ± 19.4 μmol/L and 244.8 ± 21.6 μmol/L, respectively, indicating no substantial differences. In contrast, 1,5‐AG levels exhibited a notable difference, with values of 15.1 ± 6.7 μg/mL in females and 18.5 ± 7.3 μg/mL in males. ROC analysis also indicated a distinct cutoff for 1,5‐AG, which was higher in females (14.8 μg/mL) than in males (12.5 μg/mL), contrasting with the mean analysis. However, the derived 1,5‐AG cutoff for the female group had a significantly low specificity of 47.1%, and although the AUC was relatively high at 0.754, its wide CI rendered it clinically unreliable.
4. Discussion
Several studies have discussed various reference ranges for GA. One study derived an 11.9%–15.8% range from 201 healthy individuals in the US [31]. In another study, a healthy reference population of 1799 individuals showed a 10.7%–15.1% GA reference range [32]. An optimal threshold of 15.2% has been reported in the Japanese population using ROC curve analysis and OGTT results [33]. In a study by Hwang et al., a relatively low cutoff of 14.3% was established using ROC analysis in 852 Korean individuals [34]. The ROC cutoff in our study (13.9%) was markedly lower than the conventional reference value of 16.0% and the upper limit of the reference interval of 15.1%. Interestingly, lowering the GA cutoff from 16.0% to 13.9% significantly increased the sensitivity from 42.0% to 72.3% but slightly decreased the specificity from 99.3% to 88.5%. Moreover, our ROC cutoff detected twice the number of individuals with DM progression within 3 years of follow‐up examination. Therefore, our findings and those of Hwang et al. suggest applying a lower cutoff compared to the prevailing GA reference values in the Korean population.
Two different cutoffs of 1,5‐AG, 10.0 μg/mL, and 14.0 μg/mL, have been discussed to reflect frequent exposure to hyperglycemia above the renal threshold [27, 28, 35, 36, 37]. Conversely, Selvin et al. reported the lower cutoff as 6.0 μg/mL, which was related to increased coronary heart disease, stroke, heart failure, and death [38]. In our study, both the calculated ROC cutoff (13.3 μg/mL) and the lower limit of the reference interval (5.3 μg/mL) were lower than the conventional reference value (14.0 μg/mL). These two values did not enhance the sensitivity for diagnosing current DM status or detecting more individuals with DM progression. Interestingly, 1,5‐AG clearly distinguished the hyperglycemic group from the normal group in the percentage distribution graph compared to other markers (Figure 3). This result could be due to the characteristic of 1,5‐AG that is reabsorbed competitively with glucose and is distinctively decreased in a hyperglycemic state exceeding the threshold.
For fructosamine, 200–285 μmol/L is generally known as the reference range for nonfdiabetic individuals [10]. Selvin et al. reported the community‐based reference interval as 194.8–258.0 μmol/L from 1799 healthy individuals in the US [32]. Our ROC cutoff and the upper limit of the reference interval (278 and 279 μmol/L) were slightly lower than the widely used value. Moreover, in contrast to GA, fructosamine showed no significant increase in sensitivity when the cutoff value was lowered. Furthermore, analysis based on 3 years of follow‐up data did not suggest significant clinical benefits from lowering the fructosamine cutoff point for the early detection of DM progression.
In our study, 1,5‐AG demonstrated a significant difference in mean values between the sex groups compared to the other two markers, with males exhibiting a higher mean (18.5 ± 7.3 μg/mL) than females (15.1 ± 6.7 μg/mL). This finding was also reported in previous studies [32, 39]. Conversely, the ROC cutoff for females was higher than for males, displaying a trend contrary to the mean values. Specifically, the cutoff for females, with a specificity of 47.1%, suggests that the threshold might be set excessively high, thereby increasing the risk of overdiagnosing DM. Additionally, the broad CI for the AUC, ranging from 0.647 to 0.837, highlights the considerable uncertainty in the diagnostic accuracy of this marker. The unreliable results in the female group may be attributed to the fact that the study population primarily consisted of healthy screening participants and that the subgroup of females tested for 1,5‐AG constituted only 36.8% of the male cohort, potentially failing to include an adequate number of DM cases.
This study used health examination data to establish new cutoff values for the three hyperglycemic markers. We established optimal cutoff values using ROC analysis of FBG and HbA1c levels and calculated the upper or lower limits of the reference ranges for community‐based estimation. Our study is valuable for developing clinically useful cutoff values for the three alternative markers based on a substantial data set. Moreover, the estimated performance for DM detection underlines the usefulness of these three markers as screening or diagnostic markers for DM.
However, this study has several limitations because it was a retrospective analysis using health examination data. Generally, examinees stay only for a short time during their health check‐up, so the OGTT, which requires a glucose load, is usually not performed. Instead, FBG and HbA1c are tested for DM screening. Therefore, impaired glucose tolerance was not included in the classification of DM and its progression groups. Similarly, it was impossible to include diverse races and sufficiently hyperglycemic individuals for performance estimation because the study participants were mainly healthy Koreans. We also evaluated the clinical performance of the upper and lower limits of the reference interval as equivalent to the ROC cutoff and conventional reference values. However, it should be noted that these reference ranges differ significantly from the clinically meaningful cutoff points for DM diagnosis.
We found that lowering the GA cutoff to 13.9% significantly increased the sensitivity for diagnosing DM and its potential for the early detection of DM progression, suggesting the possibility of higher efficiency for DM screening. However, changing the 1,5‐AG and fructosamine cutoff values had no advantage in screening for DM in terms of diagnostic performance. Recently, combinations of alternative and traditional hyperglycemic markers have been proposed to detect DM and prediabetes with adequate sensitivity and specificity [34, 40]. Further studies on the diverse uses of the three alternative hyperglycemic markers are warranted to develop effective early screening and diagnostic strategies for DM.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1.
Hui‐Jin Yu and Chang‐Hun Park contributed equally to this work.
Funding: The authors received no specific funding for this work.
Contributor Information
Eunju Sung, Email: eunjusung68@gmail.com.
Min‐Jung Kwon, Email: mjkkmd@gmail.com.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author under institutional review on reasonable request.
References
- 1. McPherson R. A., Henry's Clinical Diagnosis and Management, 22nd ed. (Philadelphia, PA: Elsevier Inc, 2015), 214 p. [Google Scholar]
- 2. Olafsdottir E., Andersson D. K., Dedorsson I., Svardsudd K., Jansson S. P., and Stefansson E., “Early Detection of Type 2 Diabetes Mellitus and Screening for Retinopathy Are Associated With Reduced Prevalence and Severity of Retinopathy,” Acta Ophthalmologica 94, no. 3 (2016): 232–239. [DOI] [PubMed] [Google Scholar]
- 3. Singh A., Donnino R., Weintraub H., and Schwartzbard A., “Effect of Strict Glycemic Control in Patients With Diabetes Mellitus on Frequency of Macrovascular Events,” American Journal of Cardiology 112, no. 7 (2013): 1033–1038. [DOI] [PubMed] [Google Scholar]
- 4.“Standards of Medical Care in Diabetes – 2013,” Diabetes Care 36, no. Suppl 1 (2013): S11–S66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.“(2) Classification and Diagnosis of Diabetes,” Diabetes Care 38, no. Suppl (2015): S8–S16. [DOI] [PubMed] [Google Scholar]
- 6. Davidson M. B., “Counterpoint: The Oral Glucose Tolerance Test Is Superfluous,” Diabetes Care 25, no. 10 (2002): 1883–1885. [DOI] [PubMed] [Google Scholar]
- 7.“2. Classification and Diagnosis of Diabetes,” Diabetes Care 39, no. Suppl 1 (2016): S13–S22. [DOI] [PubMed] [Google Scholar]
- 8. Cowie C. C., Rust K. F., Byrd‐Holt D. D., et al., “Prevalence of Diabetes and High Risk for Diabetes Using A1C Criteria in the U.S. Population in 1988‐2006,” Diabetes Care 33, no. 3 (2010): 562–568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Cohen R. M., Franco R. S., Khera P. K., et al., “Red Cell Life Span Heterogeneity in Hematologically Normal People Is Sufficient to Alter HbA1c,” Blood 112, no. 10 (2008): 4284–4291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Bergman M., Abdul‐Ghani M., DeFronzo R. A., et al., “Review of Methods for Detecting Glycemic Disorders,” Diabetes Research and Clinical Practice 165 (2020): 108233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Juraschek S. P., Steffes M. W., E. R. Miller, 3rd , and Selvin E., “Alternative Markers of Hyperglycemia and Risk of Diabetes,” Diabetes Care 35, no. 11 (2012): 2265–2270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Saudek C. D., Derr R. L., and Kalyani R. R., “Assessing Glycemia in Diabetes Using Self‐Monitoring Blood Glucose and Hemoglobin A1c,” Journal of the American Medical Association 295, no. 14 (2006): 1688–1697. [DOI] [PubMed] [Google Scholar]
- 13. Lee J. E., “Alternative Biomarkers for Assessing Glycemic Control in Diabetes: Ffructosamine, Glycated Albumin, and 1,5‐Anhydroglucitol,” Annals of Pediatric Endocrinology & Metabolism 20, no. 2 (2015): 74–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Neelofar K. and Ahmad J., “An Overview of in Vitro and in Vivo Glycation of Albumin: A Potential Disease Marker in Diabetes Mellitus,” Glycoconjugate Journal 34, no. 5 (2017): 575–584. [DOI] [PubMed] [Google Scholar]
- 15. Koga M., Hashimoto K., Murai J., et al., “Usefulness of Glycated Albumin as an Indicator of Glycemic Control Status in Patients With Hemolytic Anemia,” Clinica Chimica Acta: International Journal of Clinical Chemistry 412, no. 3–4 (2011): 253–257. [DOI] [PubMed] [Google Scholar]
- 16. Yamanouchi T. and Akanuma Y., “Serum 1,5‐Anhydroglucitol (1,5 AG): New Clinical Marker for Glycemic Control,” Diabetes Research and Clinical Practice 24, no. Suppl (1994): S261–S268. [DOI] [PubMed] [Google Scholar]
- 17. Yamanouchi T., Tachibana Y., Akanuma H., et al., “Origin and Disposal of 1,5‐Anhydroglucitol, a Major Polyol in the Human Body,” American Journal of Physiology 263, no. 2 Pt 1 (1992): E268–E273. [DOI] [PubMed] [Google Scholar]
- 18. Yamanouchi T., Minoda S., Yabuuchi M., et al., “Plasma 1,5‐Anhydro‐D‐Glucitol as New Clinical Marker of Glycemic Control in NIDDM Patients,” Diabetes 38, no. 6 (1989): 723–729. [DOI] [PubMed] [Google Scholar]
- 19. Stettler C., Stahl M., Allemann S., et al., “Association of 1,5‐Anhydroglucitol and 2‐h Postprandial Blood Glucose in Type 2 Diabetic Patients,” Diabetes Care 31, no. 8 (2008): 1534–1535. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Mosca A., Carenini A., Zoppi F., et al., “Plasma Protein Glycation as Measured by Fructosamine Assay,” Clinical Chemistry 33, no. 7 (1987): 1141–1146. [PubMed] [Google Scholar]
- 21. Rondeau P. and Bourdon E., “The Glycation of Albumin: Structural and Functional Impacts,” Biochimie 93, no. 4 (2011): 645–658. [DOI] [PubMed] [Google Scholar]
- 22. Hirsch I. B., “Clinical Review: Realistic Expectations and Practical Use of Continuous Glucose Monitoring for the Endocrinologist,” Journal of Clinical Endocrinology and Metabolism 94, no. 7 (2009): 2232–2238. [DOI] [PubMed] [Google Scholar]
- 23. Kim C. W., Yun K. E., Jung H. S., et al., “Sleep Duration and Quality in Relation to Non‐alcoholic Fatty Liver Disease in Middle‐Aged Workers and Their Spouses,” Journal of Hepatology 59, no. 2 (2013): 351–357. [DOI] [PubMed] [Google Scholar]
- 24. Chang Y., Kim B. K., Yun K. E., et al., “Metabolically‐Healthy Obesity and Coronary Artery Calcification,” Journal of the American College of Cardiology 63, no. 24 (2014): 2679–2686. [DOI] [PubMed] [Google Scholar]
- 25. Lee S. H., Sohn J. H., Kim C., et al., “Pre‐Stroke Glycemic Variability Estimated by Glycated Albumin Predicts Hematoma Expansion and Poor Outcomes in Patients With Spontaneous Intracerebral Hemorrhage,” Scientific Reports 13, no. 1 (2023): 12848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lee S. H., Jang M. U., Kim Y., et al., “Effect of Prestroke Glycemic Variability Estimated Glycated Albumin on Stroke Severity and Infarct Volume in Diabetic Patients Presenting With Acute Ischemic Stroke,” Frontiers in Endocrinology 11 (2020): 230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Wada H., Dohi T., Miyauchi K., et al., “Impact of Serum 1,5‐Anhydro‐D‐Glucitol Level on the Prediction of Severe Coronary Artery Calcification: An Intravascular Ultrasound Study,” Cardiovascular Diabetology 18, no. 1 (2019): 69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Watanabe M., Kokubo Y., Higashiyama A., Ono Y., Miyamoto Y., and Okamura T., “Serum 1,5‐Anhydro‐D‐Glucitol Levels Predict First‐Ever Cardiovascular Disease: An 11‐Year Population‐Based Cohort Study in Japan, the Suita Study,” Atherosclerosis 216, no. 2 (2011): 477–483. [DOI] [PubMed] [Google Scholar]
- 29. Connor A. E., Visvanathan K., Boone S. D., Rifai N., Baumgartner K. B., and Baumgartner R. N., “Fructosamine and Diabetes as Predictors of Mortality Among Hispanic and Non‐Hispanic White Breast Cancer Survivors,” npj Breast Cancer 5 (2019): 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Romero S. T., Sharshiner R., Stoddard G. J., Ware Branch D., and Silver R. M., “Correlation of Serum Fructosamine and Recurrent Pregnancy Loss: Case‐Control Study,” Journal of Obstetrics and Gynaecology Research 42, no. 7 (2016): 763–768. [DOI] [PubMed] [Google Scholar]
- 31. Kohzuma T., Yamamoto T., Uematsu Y., Shihabi Z. K., and Freedman B. I., “Basic Performance of an Enzymatic Method for Glycated Albumin and Reference Range Determination,” Journal of Diabetes Science and Technology 5, no. 6 (2011): 1455–1462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Selvin E., Warren B., He X., Sacks D. B., and Saenger A. K., “Establishment of Community‐Based Reference Intervals for Fructosamine, Glycated Albumin, and 1,5‐Anhydroglucitol,” Clinical Chemistry 64, no. 5 (2018): 843–850. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Ikezaki H., Furusyo N., Ihara T., et al., “Glycated Albumin as a Diagnostic Tool for Diabetes in a General Japanese Population,” Metabolism 64, no. 6 (2015): 698–705. [DOI] [PubMed] [Google Scholar]
- 34. Hwang Y. C., Jung C. H., Ahn H. Y., et al., “Optimal Glycated Albumin Cutoff Value to Diagnose Diabetes in Korean Adults: A Retrospective Study Based on the Oral Glucose Tolerance Test,” Clinica Chimica Acta: International Journal of Clinical Chemistry 437 (2014): 1–5. [DOI] [PubMed] [Google Scholar]
- 35. Dungan K. M., Buse J. B., Largay J., et al., “1,5‐Anhydroglucitol and Postprandial Hyperglycemia as Measured by Continuous Glucose Monitoring System in Moderately Controlled Patients With Diabetes,” Diabetes Care 29, no. 6 (2006): 1214–1219. [DOI] [PubMed] [Google Scholar]
- 36. Warren B., Lee A. K., Ballantyne C. M., et al., “Associations of 1,5‐Anhydroglucitol and 2‐Hour Glucose With Major Clinical Outcomes in the Atherosclerosis Risk in Communities (ARIC) Study,” Journal of Applied Laboratory Medicine 5, no. 6 (2020): 1296–1306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Ikeda N., Hara H., and Hiroi Y., “Ability of 1,5‐Anhydro‐D‐Glucitol Values to Predict Coronary Artery Disease in a Non‐Diabetic Population,” International Heart Journal 56, no. 6 (2015): 587–591. [DOI] [PubMed] [Google Scholar]
- 38. Selvin E., Rawlings A., Lutsey P., et al., “Association of 1,5‐Anhydroglucitol With Cardiovascular Disease and Mortality,” Diabetes 65, no. 1 (2016): 201–208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Welter M., Boritza K. C., Anghebem‐Oliveira M. I., et al., “Reference Intervals for Serum 1,5‐Anhydroglucitol in Children, Adolescents, Adults, and Pregnant Women,” Clinica Chimica Acta: International Journal of Clinical Chemistry 486 (2018): 54–58. [DOI] [PubMed] [Google Scholar]
- 40. Ying L., He X., Ma X., et al., “Serum 1,5‐Anhydroglucitol When Used With Fasting Plasma Glucose Improves the Efficiency of Diabetes Screening in a Chinese Population,” Scientific Reports 7, no. 1 (2017): 11968. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author under institutional review on reasonable request.