Fructosamine is a glycemic biomarker which may be useful for indication and control of diabetes respectively.
The objective of the study was to evaluate fructosamine as an indicator of hyperglycaemia and glucose control in subjects with diabetes.
Design, Setting & Patients
From the AMORIS cohort, subjects with serum glucose, fructosamine and HbA1c from the same examination were studied cross-sectionally and longitudinally (n = 10,987; 5,590 overnight-fasting). The guidelines of the American Diabetes Association were followed for classification of prediabetes and diabetes. Separate analyses were performed in patients with a newly detected or a known diagnosis of type 1 or type 2 diabetes respectively.
All three biomarkers were strongly correlated. With regard to the association between fructosamine and HbA1c Pearson linear correlation coefficients in the range of 0.67–0.75 were observed in fasting and non-fasting subjects with type 1 or type 2 diabetes. Analyses of glucose control in fasting patients with type 2 diabetes having all three biomarkers measured at three separate occasions within on average 290 days of the index examination showed similar trends over time for glucose, fructosamine and HbA1c. Discrimination of subjects with and without diabetes across the range of fructosamine levels was good (area under curve (AUC) 0.91–0.95) and a fructosamine level of 2.5 mmol/L classified subjects to diabetes with a sensitivity of 61% and a specificity of 97%.
Citation: Malmström H, Walldius G, Grill V, Jungner I, Gudbjörnsdottir S, Hammar N (2014) Fructosamine Is a Useful Indicator of Hyperglycaemia and Glucose Control in Clinical and Epidemiological Studies – Cross-Sectional and Longitudinal Experience from the AMORIS Cohort. PLoS ONE 9(10): e111463. https://doi.org/10.1371/journal.pone.0111463
Editor: Marta L. Hribal, University of Catanzaro Magna Graecia, Italy
Received: June 24, 2014; Accepted: October 2, 2014; Published: October 29, 2014
Copyright: © 2014 Malmström 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: The authors confirm that, for approved reasons, some access restrictions apply to the data underlying the findings. This study is based on national registers in Sweden: the National Board of Health and Welfare, the National Cause-of-Death Register, and the National Patient Register. From Statistics Sweden: the LISA database, Surveys of living conditions (ULF). The Swede-Heart Register, the National Diabetic Register (NDR) and several research registers; WOLF, 60 years old men and women, COSM, SMC at Karolinska institutet (KI), Stockholm, Sweden. These datasets contain sensitive information. The database used in the present paper is located in a security server at the institution (IMM, KI) with restricted access. This dataset is available upon request to the corresponding author, given that the interested party can obtain approval from the data owners, the National Board of Health and Welfare in Sweden (http://www.socialstyrelsen.se/english) and Statistics Sweden (http://www.scb.se/en_/) as well as from the owners of the research registers at KI, Stockholm, Sweden.
Funding: The study was funded by the Gunnar and Ingmar Jungner Foundation for Laboratory Medicine, Stockholm, Sweden. The funder had no role in study design and analysis, decision to publish, or preparation of the manuscript but was responsible for the laboratory data Collection in 1985-1996 via the CALAB laboratory.
Competing interests: Niklas Hammar is an adjunct professor of epidemiology at the Institute of Environmental Medicine, Karolinska Institutet (KI) and also employee of AstraZeneca Research & Development, Mölndal, Sweden. The research of the current study has no relation to the drug development or other activities performed by AstraZeneca and AstraZeneca had no influence on the initiation, conduct or interpretation of the study. The CALAB laboratory was overtaken by another company in 1997 but the laboratory database containing data from 1985 to 1996 was still owned and managed by Ingmar Jungner. Jungner donated the database to the Institute of Environmental Medicine in 2012. Ingmar Jungner is retired since many years and is not active in any business activities. He still is a member of the AMORIS Steering Committee. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
Fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) are established indicators of glucose status and its control. Definitions from different health organizations use FPG or HbA1c cut-offs of ≥7 mmol/L and ≥6.5% (48 mmol/mol) respectively either as single indicators or combined to diagnose diabetes –. HbA1c reflects the glucose concentration during the entire lifespan of the red blood cells but to the largest extent the 6–8 weeks preceding the time of measurement . Certain factors, e.g. red blood cell disorders, could potentially bias the measurement of HbA1c and therefore alternative markers could be useful . One such marker is fructosamine which relates to average levels of glucose during the preceding 1 to 3 weeks , . Fructosamine may give an earlier indication of poorly controlled glucose compared to HbA1c. It is a simple, robust and inexpensive biomarker that could potentially be a useful tool in large epidemiological and clinical studies either as a standalone indicator of hyperglycaemia or in combination with glucose and HbA1c . Importantly, fructosamine may be reliably measured irrespective of fasting or non-fasting. However, fructosamine is rarely used in clinical practice and has not been extensively evaluated as an indicator of diabetes and its micro-and macro vascular complications in large population based studies. Various cut-offs for fructosamine as well as correlations to glucose and HbA1c have been published based on rather small cohorts of patients –. Recently, fructosamine has been showed to be an independent biomarker to predict incident diabetes and its microvascular complications .
The aim of the present study was to explore the relationship between serum glucose, serum fructosamine and HbA1c cross-sectionally and over time in a large population with a wide variation in glucose and HbA1c values. An additional aim was to investigate the sensitivity and specificity of various fructosamine cut-offs in identification of subjects with prediabetes and diabetes using established diagnostic criteria based on HbA1c and FPG.
Materials and Methods
The study was based on the AMORIS cohort that has been described in detail previously , . This cohort was recruited during 1985–1996 mainly from the greater Stockholm area. The cohort subjects were healthy and referred for clinical laboratory testing as part of a health check-up or were outpatients referred for laboratory testing. No samples were from inpatients. The AMORIS population consists of 812,073 subjects of all ages with varying information on a number of biomedical factors. Subjects were frequently followed with repeated examinations. For a total of 10,987 individuals, 5,199 males and 5,788 females, there was complete information on serum glucose, serum fructosamine and HbA1c from the same examination. Of these individuals, 5,590 were known to be overnight-fasting at the time of the index examination which was defined as the first time glucose, fructosamine and HbA1c were measured at the same occasion. For a repeated measurement analysis we identified type 2 diabetes patients (n = 254) with assessments of all biomarkers at the index examination as well as within 0–6 months and 6–12 months. In the case of multiple assessments within one of these time windows, we used the assessment farthest from the index examination. Subjects were grouped on whether they had increased or reduced glucose levels at the third examination.
All laboratory examinations were performed at the CALAB laboratory (CALAB laboratories, Stockholm, Sweden), a laboratory accredited by SWEDAC (Swedish Board for Technical Accreditation). Serum glucose was measured enzymatically with a glucose oxidase/peroxidase method based on an enzymatic colorimetric technique according to the GOD-PAP method with automated multichannel analyzers (AutoChemist-PRISMA (PRISMA) and Technicon DAX 96). Total imprecision was CV<3%. Values of serum glucose are virtually identical to plasma glucose only differing by about 1% . In the AMORIS study serum glucose values were standardized versus international standards of plasma glucose.
Fructosamine was measured by the Nitroblue Teterazolium (NBT) colorimetric technique devised by Johnson  based on the reducing ability of fructosamine in an alkaline solution and performed with the same automated analyzers as for serum glucose. Total imprecision was CV<5%.
Two methods were used to determine HbA1c. From 1992 and until the end of 1994 the method was based on affinity binding and quantitative determination of glycated hemoglobin in anticoagulated whole blood ,  and performed with the Abbot Vision System. The results were ´adjusted to be HbA1c and calibrated to DCCT values. Total CV was 4.4% at level 4.8 mmol/L and 2.8% at level 8.9 mmol/L. Reagents and three levels of calibrators from Abbott Diagnostics (Lake Forest, Illinois, USA) were used. The second method used since 1995 was based on a two-channel method, UNIMATE , automated on the analyzer Hitachi 917, for the immunological/colorimetric determination of % HbA1c. Here HbA1c was measured by latex enhanced turbidimetry  and total Hb colorimetrically . Total CV was <4.5%.
Albumin was colorimetrically measured based on its binding with bromocresol green. The CV was ≤2.0% at level 35 and 46 g/L and ≤2.5% at level 26 g/L. The concentration of albumin could potentially influence the concentration of fructosamine . Hence, when possible we also calculated albumin corrected fructosamine using the following formula: (fructosamine (mmol/L)/albumin (g/L) x100) . Glomerular filtration rate (eGFR) was estimated by the Epidemiology Collaboration (CKD-EPI) formula . Classification of chronic kidney disease was defined as an eGFR less than 60 mL/min per 1.73 m2.
Classification of glucose tolerance
We used the plasma glucose levels and HbA1c values as defined by the American Diabetes Association (ADA) guidelines ,  for the diagnosis of type 2 diabetes. Subjects were classified into five groups: normal glucose tolerance, prediabetes, newly diagnosed type 2 diabetes (NewT2D), previously diagnosed type 2 diabetes (DiagnosedT2D) and type 1 diabetes (T1D). This classification was based on glucose level at the index examination or by information of type 1 or type 2 diabetes prior to index examination based on other sources. Following the guidelines of ADA, we classified fasting serum glucose (FSG) below 5.6 mmol/L or HbA1c below 5.7% (39 mmol/mol) as normal glucose tolerance. FSG of 5.6–6.9 mmol/L or HbA1c between 5.7% and 6.4% (39–46 mmol/mol) was classified as prediabetes. FSG above 6.9 mmol/L or HbA1c of 6.5% (48 mmol/mol) and above was classified as type 2 diabetes (NewT2D). In addition, if a subject had a random glucose of 11.1 mmol/L or above this was considered as NewT2D. Subjects who had information about hospitalization with diabetes as discharge diagnosis prior to the index examination were classified according to the discharge diagnosis (DiagnosedT2D). Subjects registered in the Swedish National Diabetes Register (NDR) were classified as type 1 (T1D) or type 2 diabetes (DiagnosedT2D) according to the diagnostic criteria set out in the register. As regards NDR, if no specific type was defined the clinical diagnosis by the referring physician was used. When registered as unspecified diabetes in the NDR or detected by the glucose criterion in AMORIS we classified subjects less than 30 years of age as T1D. Subjects diagnosed by the Swedish medical birth register (MFR) with gestational diabetes within one year prior or one year after the index examination were excluded from this study.
Socioeconomic factors were obtained from several databases maintained by the Statistics Sweden. Those databases were the1990 national education database, the 1990 national census for occupation and the multi-generation register to obtain country of birth. Information on medical history including cardiovascular events, overall anemia, nephrosis and cancer prior to the index examination were obtained from inpatient registers and the cancer register maintained by the Swedish National Board of Health and Welfare. BMI was collected from the AMORIS laboratory database when assessed at any examination or from the NDR, national survey of living conditions (ULF, Statistics Sweden), the medical birth register (MFR, National Board of Health and Welfare), the Swedish Twin Registry or quality of care registers included in SWEDEHEART (Uppsala Clinical Research Center, UCR). The BMI closest to the date of the index examination was used for adjustment of the associations. Data on BMI were available in about 40% of subjects. Serum albumin, creatinine, lipoproteins and apolipoproteins were obtained from the index examination. Triglycerides, total cholesterol, albumin and creatinine were available for about 90% and high and- low density lipoprotein cholesterol (HDL-C, LDL-C) as well as apolipoproteins A-1 and B (ApoA-1, ApoB) for about 50% of the subjects.
This study is part of a project entitled “Metabolic abnormalities and inflammation in relation to chronic disease primarily cardiovascular diseases, kidney disease, cancer, dementia, rheumatoid arthritis and psychiatric illness – epidemiological studies based on the AMORIS population”.
This project has been approved by the Regional Ethical committee at Karolinska Institutet, Stockholm, Sweden (Registration number 2010/1∶7). All the study data was de-identified and the Ethical committee approved the project without collecting any additional informed consent.
All analyses were performed using SAS software (SAS Institute Inc., Cary, NC, USA). Pearson’s correlation coefficients and partial correlations were calculated using the SAS correlation procedure. A change in the point estimate of 10% or more was used as criteria for inclusion of a covariate in the multivariable models for confounding adjustment purposes. Sensitivity and specificity to classify diabetes were calculated using binary tables at various levels of fructosamine and with ADA criteria as the reference standard. Receiving operator characteristic analyses (ROC) were performed and area under the curve (AUC) with 95% Wald confidence limits estimated according to Mann-Whitney with PROC LOGISTIC. Series of the arithmetic mean values for glucose and HbA1c per percentile of fructosamine were graphically displayed with corresponding 1 SD range using PROC MEANS and PROC SGPLOT. For the repeated measurement analyses, arithmetic mean values by biomarker and time window were calculated and graphically presented.
Characteristics of subjects
Subject characteristics are presented in Table 1. There were 5,714 subjects (51% of all) who were categorized as having normal glucose tolerance, 1,877 with prediabetes (17%) and 3,116 with type 2 diabetes (29%). Of the subjects classified as type 2 diabetes, 1,256 were diagnosed at the index examination i.e. NewT2D and 1,860 had a known diagnosis of type 2 diabetes. A total of 280 patients were diagnosed with type 1 diabetes. Fasting subjects in the NewT2D group (n = 759) were detected by a glucose value of 7 mmol/l or above in 94% of the subjects. Simultaneously, 51% of those subjects also had HbA1c above 6.5%. The gender distribution was balanced, 47% males and 53% females, although females were relatively fewer among subjects with type 2 diabetes.
The mean age in the study population was 57 years with a range from 2 to 97 years. The subjects with normal glucose tolerance were somewhat younger than subjects with T2D. Those with DiagnosedT2D had more often a history of cardiovascular disease compared to the group with normal glucose tolerance. A history of cancer was similar across the categories. The NewT2D group had similar clinical characteristics compared to the DiagnosedT2D group except for a lower frequency of history of cardiovascular diseases. Mean values of triglycerides, total cholesterol, ApoB and the ApoB/ApoA-I ratio increased and mean values of HDL-C and ApoA-I successively decreased from subjects with normal glucose levels to those with pre-diabetes and newly diagnosed type 2 diabetes.
Diabetes-induced increases of markers of glycaemia
In the fasting state, fructosamine increased from an average of 2.03 mmol/L in subjects with normal glucose tolerance to 2.11 mmol/L in the prediabetes group and to 2.62 mmol/L in the NewT2D group. Similarly to fructosamine, HbA1c increased from an average of 4.76% (29 mmol/mol) to 5.14% (33 mmol/mol) in the prediabetes group and to 6.93% (52 mmol/mol) in the NewT2D group. Neither for fructosamine nor HbA1c did DiagnosedT2D markedly diverge from NewT2D. Concentrations were markedly higher for the T1D group compared to the groups with type 2 diabetes (2.93 mmol/L for fructosamine and 7.62% for HbA1c).
Associations between various glucose markers
Across glucose levels, strong associations were found for fructosamine with glucose and HbA1c respectively, r = 0.75 and r = 0.78 respectively in the entire cohort (Figure 1). Fructosamine correlated with similar strength to HbA1c, r = 0.79 and glucose, r = 0.80 in the fasting state. The linear correlations for fructosamine and HbA1c was identical, r = 0.75, in the NewT2D and DiagnosedT2D group respectively (Figure 2). Corresponding linear correlations for non-fasting subjects were r = 0.69 and r = 0.72. Subjects with T1D showed a similar association between fructosamine and HbA1c as the subjects with DiagnosedT2D or NewT2D. The linear correlation was r = 0.67 in fasting subjects with T1D. Much lower correlations were found in the prediabetes group, r = 0.11 and r = 0.17, in the fasting and non-fasting subjects respectively and no clear association in subjects with normal glucose tolerance. Adjusting for gender, age, albumin, BMI, total cholesterol and triglycerides, affected associations only marginally. Exclusion of subjects with previously diagnosed anemia, chronic kidney insufficiency or diagnosed nephrosis respectively did not alter the associations.
All fasting states and all levels of glycemia included (All subjects).
Fructosamine, HbA1c and glucose in repeated measurements
Figure 3 shows the results for fasting subjects with type 2 diabetes (n = 254) in whom three repeated measurements of glucose, fructosamine and HbA1c were available. In the NewT2D group glucose levels had increased on average by 27% after a mean of 290 days. Simultaneously fructosamine and HbA1c increased by 4.6% and 3.3%, respectively. Corresponding results for DiagnosedT2D subjects were a 29% increase in glucose, a 6.8% increase in fructosamine and a 7.1% increase in HbA1c. In subjects with NewT2D who manifested a decrease of mean glucose level by 18% after a mean of 290 days, the mean decrease of fructosamine was 7.4% and that of HbA1c 6.6%. Corresponding results for the DiagnosedT2D group were for glucose 22%, fructosamine 10.2% and HbA1c 11.9%. Thus, there were parallel changes over time in the evolution of glucose, fructosamine and HbA1c in those who increased, and those who decreased their glucose level. Figure 4 shows individual changes in the biomarkers over time. The correlation coefficients for the changes were similar regardless of time point for follow-up.
Fasting subjects with new or diagnosed type 2 diabetes and with three simultaneous measurements of the biomarkers within one year are included.
Hyperglycemia (diabetes) based on fructosamine compared with ADA diagnostic criteria
Mean values were calculated for HbA1c and glucose respectively among the fasting subjects by percentiles of fructosamine (Figure 5). At a level of fructosamine of 2.5 mmol/L, the mean values of both glucose and HbA1c were diagnostic for diabetes by ADA criteria. When ADA criteria for classification of diabetes, i.e. FSG≥7 mmol/L or an HbA1c≥6.5% (48 mmol/mol) was used a fructosamine cut-off of 2.5 mmol/L correctly diagnosed 61% as diabetes and correctly rejected 97% of the non-diabetes individuals. Corresponding sensitivity and specificity for non-fasting subjects were 82% and 94% respectively. The area under the ROC curve (AUC) for diagnosing diabetes by fructosamine when the ADA criteria were used to define diabetes was AUC = 0.91 in the fasting and AUC = 0.95 in the non-fasting state, a finding which indicated good discrimination across levels of fructosamine. By separately modelling HbA1c and fructosamine respectively to explain diabetes defined solely based on glucose criterion, the AUCs were similar for the two biomarkers fructosamine and HbA1c, 0.91 and 0.90 respectively. The AUC obtained by defining diabetes by fructosamine when both glucose and HbA1c criteria were met was AUC = 0.96.The fructosamine cut-off 2.5 mmol/L was applied in analyses of all fasting subjects in the AMORIS cohort, n∼255,000, using solely the glucose ≥7 mmol/L as criterion. The results showed identical sensitivity and somewhat higher specificity (61% and 98%) compared with the current study population (results not shown).
Sensitivity/Specificity at various fructosamine cut-off values for diagnosis of diabetes according to ADA criteria, 2.3 mmol/L (79%, 88%), 2.5 mmol/L (61%, 97%), 2.6 mmol/L (52%, 99%). Diagnostic discrimination of diabetes by S-Fructosamine using ROC analysis: AUC = 0.91 (fasting) AUC = 0.95 (non-fasting).
The present results are based on a large population representative for the working general population of Stockholm at the time of laboratory examinations. We found close associations between fructosamine, glucose and HbA1c, overall and particularly in subjects with either type 1 or type 2 diabetes. These findings are consistent with earlier but smaller studies – and also with recently published results from the ARIC study where an overall correlation of r = 0.82 was found . Importantly, we found similar relations between fructosamine and HbA1c irrespective of fasting or non-fasting. The inter-relationship between fructosamine and HbA1c as well as association with glucose levels in those with prediabetes was less striking and was not seen in those with normal glucose values. Hence, the relationship between fructosamine and HbA1c to fasting glucose levels respectively indicates a curvilinear relationship as has also previously been pointed out , . Importantly, our results from AMORIS were obtained in a much larger population of men and women in all ages and with a much wider range of glucose (up to 40 mmol/L), fructosamine (up to 7.5 mmol/L) and HbA1c (up to 16.5%), than in studies published previously –. Subjects with prediabetes and especially manifest diabetes were characterized by features common in subjects with metabolic syndrome and manifest diabetes such as high BMI, hypertriglyceridemia, low HDL cholesterol high apoB and low apoA-I giving the typical manifestations of high atherogenic apoB/apoA-I ratios as reviewed . However, the relations between glucose, fructosamine and HbA1c remained basically unchanged when controlling for age, gender, BMI or lipid levels. In addition, albumin correction of the fructosamine may only improve precision to a small degree regarding monitoring of variability of glucose values. Hence, the cross-sectional association of fructosamine with different degrees of hyperglycaemia appears robust.
Previously, Baker et al. showed strong association between fructosamine and HbA1c in the interval r = 0.70 to r = 0.87 in patients with type 2 diabetes –. A more recent study by Cohen et al. included 153 subjects with varied types of diabetes . They observed a correlation of r = 0.78. Juraschek et al. also reported a similar linear correlation but also suggested polynomial models as the best fit . All of the above studies had rather small study populations with limited number of subjects with high glucose levels. Thus, the present AMORIS study confirmed the correlations from earlier work but in a much larger study population.
Moreover, Juraschek et al. evaluated various criteria for diabetes by presenting AUC measures. They reported an AUC of 0.83 for both fructosamine and fasting glucose if diabetes was defined by an HbA1c of 6.5% (48 mmol/L) . The diagnostic discrimination by using fructosamine in the AMORIS study, which is based on a much larger population, was higher, up to AUC = 0.95.
When subgroups of individuals in which glucose, fructosamine and HbA1c were measured simultaneously up to 6 and 12 months (on average 120 and 290 days respectively), fructosamine and HbA1c changed in parallel both in those who increased their glucose values and in those whose glucose values were considerably reduced (possibly due to better adherence to dietary and/or pharmacological therapy). Although individual discordant differences in directions were observed the major part of those was in individuals with minor changes of hyperglycaemia which likely would be more sensitive to random variation. Thus, these data indicate that fructosamine can be used to monitor glucose control over time in a similar way as HbA1c.
The results of the present study indicate that fructosamine discriminates well between subjects with and without diabetes across all levels of the biomarker with an AUC over 90%. For specific levels of fructosamine it is also possible to identify subjects with type 2 diabetes with a reasonable sensitivity and a high specificity. Our results would suggest a fructosamine somewhere in the range of 2.3–2.6 for this purpose. Recently, Selvin et al showed an adjusted hazard ratio for incident diabetes of 4.96 for those above 2.64 mmol/L compared to those below 2.41 mmol/L . These levels are in line with those of our study. Notably, in follow-up studies of the outcome in diabetes patients a non-differential misclassification of exposure (in this case type 2 diabetes) in terms of reduced sensitivity does not influence estimates of relative risks. A reduced specificity would tend to bias relative risk estimates towards the null. For estimates of prevalence rates, and in particular in a clinical setting, the effects of a reduced sensitivity and specificity could be misleading or unacceptable. In our estimates of sensitivity and specificity we used diagnostic criteria based on fasting serum glucose and HbA1c according to ADA guidelines. It cannot be ruled out that this ‘gold standard’ also misclassified some subjects. Thus, the interpretation of our findings should be made with the awareness of a certain degree of misclassification of the disease in all definitions based on the different biomarkers. The results of this study indicate that fructosamine does very well in the comparison to the established diagnostic biomarkers both in the identification of subjects with hyperglycaemia or diabetes and in the evaluation of long term glucose control in subjects with manifest diabetes. It would be of greatest importance to examine the prognostic association between fructosamine and major outcomes including cardiovascular disease and hence subsequent studies from the AMORIS cohort will have that focus.
Strengths of this study include the fact that a single laboratory was used for determination of all biomarkers (CALAB laboratories, Stockholm, Sweden), that all analyses were performed on fresh blood samples, that all methods are well documented and that they are rigorously standardized. It is therefore unlikely that the relations between these three markers of glucose metabolism are biased due to poorly controlled methods or erroneous classification or due to the variation of nutritional status and potential confounders in this large population. Further, the study’s size with 5,590 subjects known to be fasting whereof 3,063 with impaired or abnormal glucose levels, to our knowledge is the largest used for this purpose. Moreover it is a rather homogenous population with 80% known to be born in Sweden and additional 5% in Finland. The large AMORIS population of 812,073 men and women is representative of the larger Stockholm population as previously documented . The sub-cohort studied in this paper is also representative for the whole AMORIS study population regarding education and country of birth but with a somewhat lower proportion of blue-collar workers.
The study population was required to have glucose, fructosamine and HbA1c measured at the same time. It is likely that the individuals included in the study visited general physicians for evaluation of a possible diabetes diagnosis or for diabetes control. This could explain the fact that the proportion of subjects with type 2 diabetes in this study was much higher than in the general population. It is also of note that the proportion of individuals who had a history of cardiovascular diseases and cancer was higher in the study population compared with the whole AMORIS cohort.
We used a randomly measured glucose value of 11.1 mmol/L or above as a diagnosis of diabetes in non-fasting subjects. Although not strictly adherent to ADA criteria (to be used only for symptomatic hyperglycaemia), the small number diagnosed solely by this criterion in this study, n = 14, most likely did not affect the results. In addition, the AMORIS population consists largely of subjects born in Sweden or in other Northern European countries. With regard to age there is a broad representation including also many subjects less than 30 years of age.
In conclusion, fructosamine is closely associated with HbA1c and glucose particularly in patients with type 1 or type 2 diabetes. Fructosamine and HbA1c tend to parallel each other over time in subjects with diabetes indicating similar reflections of glucose control. The results also suggest that fructosamine discriminates well between subjects with and without diabetes. Fructosamine may be a useful biomarker of hyperglycaemia in epidemiological and clinical studies. Future outcome studies may investigate relationships between levels of fructosamine and risks to develop myocardial infarction, stroke and diabetic complications.
Conceived and designed the experiments: HM GW NH. Analyzed the data: HM. Contributed reagents/materials/analysis tools: IJ. Contributed to the writing of the manuscript: HM GW NH VG SG.
- 1. American Diabetes Association (2012) Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 35(1): S64–S71.
- 2. American Diabetes Association (2013) Standards of Medical Care in Diabetes 2013. Diabetes Care 36(1).
- 3. Rydén L, Grant PJ, Anker SD, Berne C, Cosentino F, et al. (2013) ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD. Eur Heart J 34(39): 3035–3087.
- 4. World Health Organization (2006) Definition and diagnosis of diabetes mellitus and intermediate hyperglycemia report of a WHO/IDF consultation. ISBN 92 4 159493 4.
- 5. World Health Organization (2011) Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus Abbreviated Report of a WHO Consultation. WHO/NMH/CHP/CPM/11.1.
- 6. Kilpatrick ES (1997) Problems in the assessment of glycaemic control in diabetes mellitus. Diabetes Med 14: 819–31.
- 7. Cohen RM, Sacks DB (2012) Comparing Multiple Measures of Glycemia: How to Transition from Biomarker to Diagnostic Test?. Clinical Chemistry 58(12): 1615–17.
- 8. True MW (2009) Circulating Biomarkers of Glycemia in Diabetes Management and Implications for Personalized Medicine. J Diabetes Sci Technol. 3(4): 743–7.
- 9. Baker JR, Metcalf PA, Johnson RN, Newman D, Rietz P (1985) Use of Protein-Based Standards in Automated Colorimetric Determinations of Fructosamine in Serum. Clin Chem 31(9): 1550–54.
- 10. Baker JR, O'Connor JP, Metcalf PA, Lawson MR, Johnson RN (1983) Clinical usefulness of estimation of serum fructosamine concentration as a screening test for diabetes mellitus. British Medical Journal 287: 863–67.
- 11. Baker JR, Metcalf PA, Holdaway IM, Johnson RN (1983) Serum fructosamine concentration as measure of blood glucose control in type I (insulin dependent) diabetes mellitus. British Medical Journal 290: 352–55.
- 12. Juraschek SP, Steffes MW, Selvin E (2012) Associations of alternative markers of glycemia with hemoglobin A(1c) and fasting glucose. Clin Chem 58(12): 1648–55.
- 13. Cohen RM, Holmes YR, Chenier TC, Joiner CH (2013) Discordance Between HbA1c and Fructosamine. Diabetes Care 26(1): 163–67.
- 14. Ko GTC, Chan JCN, Yeung VTF, Chow CC, Tsang LWW, et al. (1998) Combined Use of a Fasting Plasma Glucose Concentration and HbA1c or Fructosamine Predicts the Likelihood of Having Diabetes in High-Risk Subjects. Diabetes Care 21(8): 1221–5.
- 15. Selvin E, Rawlings AM, Grams M, Klein R, Sharrett AR, et al. (2014) Fructosamine and glycated albumin for risk stratification and prediction of incident diabetes and microvascular complications: a prospective cohort analysis of the Atherosclrosis Risk in Communities (ARIC) study. Lancet Diabetes & Endocrinology 2(4): 279–88.
- 16. Walldius G, Jungner I, Holme I, Aastveit AH, Kolar W, et al. (2001) High apolipoprotein B, low apolipoprotein A-I, and improvement in the prediction of fatal myocardial infarction (AMORIS study): a prospective study. Lancet 358: 2026–2033.
- 17. Walldius G, Jungner I, Kolar W, Holme I, Steiner E (1992) High cholesterol and triglyceride values in Swedish males and females: increased risk of fatal myocardial infarction. First report from the AMORIS (Apolipoprotein related MOrtality RISk) Study. Blood Pressure 1(4): 35–42.
- 18. Frank EA, Shubha MC, Cletus D’Souza JM (2012) Blood Glucose Determination: Plasma or Serum? Journal of Clinical Laboratory Analysis 26: 317–320.
- 19. Johnson RN, Metcalf PA, Baker JR (1982) Fructosamine: a new approach to the estimation of serum glycosylprotein. An index of diabetic control. Clin Chim Acta 127: 87–95.
- 20. Middle FA, Bannister A, Bellingham AJ, Dean PD (1983) Separation of glycosylated hemoglobins using immobilized phenylboronic acid. Biochem J. 209(3): 771–9.
- 21. Klenk DC, Hermansson GT, Krohn RI, Fujimoto EK, Mallia AK, et al. (1982) Determination of Glycosylated Hemoglobin by Affinity Chromatography: Comparison with Colorimetric and Ion-Exchange Methods, and Effects of Common Interferences. Clin Chem 28(10): 2088–94.
- 22. Hoffman-La Roche AG, CH-4002 Basel, Schweiz.
- 23. Flückiger R, Mortensen HB (1988) Glycated hæmoglobins. J Chromatogr 429: 279–92.
- 24. Zander R, Lang W, Wolf HU (1984) Alkaline hæmatin D-575, a new tool in the determination of hæmoglobin as an alternative to the Cyanhaemiglobin method I. Description of the method. Clin Chim Acta 136: 83–93.
- 25. Lamb E, Mainwaring-Burton R, Dawnay A (1991) Effect of protein concentration on the formation of glycated albumin and fructosamine. Clin Chem 37(12): 2138–9.
- 26. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF III, et al. (2009) A new equation to estimate glomerular filtration rate. Ann Intern Med 150(9): 604–12.
- 27. Walldius G (2012) The apoB/apoA-I Ratio is a Strong Predictor of Cardiovascular Risk. In: Lipoproteins in Health and Diseases: ISBN Frank S, Kostner G, editors. 978-953-51-0773-6: pp. 95–148 http://dx.doi.org/10.5772/47869.
- 28. Holzmann M, Jungner I, Walldius G, Ivert I, Nordqvist T, et al. (2012) Dyslipidemia is a strong predictor of myocardial infarction in subjects with chronic kidney disease. Ann Med 44(3): 262–70.