Abstract
Many islet transplant recipients have medical conditions that could interfere with the accuracy of HbA1c measurements (e.g., anemia/dapsone use). Fructosamine is less prone to have clinical interferences and reflects glucose control in a shorter period of time than HbA1c. This study aimed to validate fructosamine use in islet transplant subjects and to evaluate its effectiveness as a predictor for islet graft dysfunction. Thirty-three islet transplant recipients who had concomitant fructosamine and HbA1c data available were retrospectively analyzed. HbA1c, fructosamine, mean capillary blood glucose, and islet graft function (fasting C-peptide/glucose ratio) were assessed. There was a significant and positive association between fructosamine and HbA1c (p < 0.0001). Both variables were also positively associated with mean overall and fasting capillary glucose. Neither fructosamine nor HbA1c was shown by ROC analysis to significantly discriminate between periods with and without subsequent graft dysfunction. HbA1c >6% was predictive of this outcome 1 month in advance (OR 2.95, p = 0.003). However, although significantly associated with graft dysfunction, use of this cutoff as a predictor of dysfunction has poor sensitivity (50%) and specificity (77.6%). Fructosamine above the normal range (>270 μmol/L Quest Diagnostics) was also predictive of ensuing dysfunction (OR 2.47, p = 0.03); however, it had similarly poor sensitivity (62%) and specificity (64%). Fructosamine can be used as an alternative to HbA1c for glycemic assessment in islet transplant recipients in situations with HbA1c assay interference. Neither HbA1c nor fructosamine are good predictors of islet graft dysfunction.
Introduction
Previous studies have supported the benefit of intensive glycemic control for the prevention of diabetes-related complications (4,27). Therefore, glucose monitoring is an essential feature of diabetes management. HbA1c is a result of a posttranslational glycation of the N-terminal of the beta chain of hemoglobin A, reflecting glycemic control in the previous 90 days, predominantly the previous 30 days (11). It is a standard method for glucose assessment and it has been shown to correlate with mean fingerstick glucose readings and to predict diabetes complications (18,27).
Fructosamine is a product of nonenzymatic glycation of albumin, reflecting the glycemic control in the prior 2–3 weeks (17). The clinical use of fructosamine has been validated (2). However, the within-subject variability and lack of conclusive evidence relating its levels with diabetes complications limit the clinical use of this assay (11,15). In a study comparing predicted HbA1c and fructosamine in diabetic subjects and normal controls, there was a fair correlation between the two measures in diabetic patients but not in normal controls (19).
Islet transplantation currently under evaluation as a treatment for patients with unstable type 1 diabetes and hypoglycemia unawareness, which has been shown to result in more stable glycemic control, less hypoglycemia, improved hypoglycemia awareness, and improvement of quality of life (8,21,23). As with all diabetic therapies, A1c remains an important outcome measure. However, many recipients can have conditions associated with unreliable HbA1c measurements; for example, dapsone use and erythropoietin administration can result in falsely low HbA1c values and iron deficiency anemia can result in falsely high values (7,14,16,26).
Whether fructosamine could be a tool for assessing glycemic control in islet transplant subjects, which commonly have near normal blood glucose, needs to be established (10). Secondly, it has been shown that islet graft dysfunction occurs in virtually all islet transplant recipients over time (8,23). The diagnosis of graft dysfunction is based on capillary glucose and HbA1c levels (12). Both have their limitations. Reporting of fingerstick glucose values can diminish over time, and HbA1c is slow to change. As fructosamine is capable to reflect the glucose control of a shorter period of time, it may represent a better way to identify islet graft dysfunction in comparison with HbA1c.
This article evaluates fructosamine as an alternative outcome measure in this population and assesses HbA1c and/or fructosamine as predictors of graft dysfunction.
Material and Methods
Subjects
We retrospectively analyzed the data of 33 islet transplant recipients (15 males and 18 females) who had concomitant fructosamine and HbA1c measurements during 2000–2008.
Transplant procedure and immunosuppressive therapy were performed as previously described (1,3,8,9,22,24). The mean age at the time of first islet infusion was 43 ± 7.5 (25–60) years old and the mean follow-up posttransplantation was 1,199 ± 651 (51–2,100) days. Subjects underwent islet transplantation alone (n = 25), islet after kidney transplantation (n = 7), and islet transplantation with bone marrow infusion (n = 1). Twenty-six of the 33 subjects (79%) achieved insulin independence. This was defined as detectable C-peptide levels and stable glycemic control (fasting capillary glucose <140 mg/dl and 2-h postprandial capillary glucose levels <180 mg/dl) for at least 2 consecutive weeks (8).
At the time of data analysis, 23 subjects had demonstrated graft dysfunction and had required reintroduction of insulin. Graft dysfunction was defined as positive C-peptide levels and fasting and/or 2-h postprandial capillary glucose levels >140 and >180 mg/dl, respectively, on three or more occasions over a week, or two consecutive monthly HbA1c values >6.5% (11).
Fructosamine and HbA1c were measured monthly after the first infusion and were included in analysis if they were not more than 3 days apart. Measurements were excluded from the analysis at time points where any the following conditions that could potentially interfere with HbA1c or fructosamine were present: hemoglobinopathies, hemolytic anemia, iron deficiency anemia (defined by serum ferritin <15 ng/ml) (13), acute blood loss (defined by history or decrement of at least 1 g/dl of hemoglobin in comparison with previous measurements) (25), blood transfusion in the previous 3 months, chronic renal failure, splenectomy, dapsone/erythropoietin administration, and nephrotic syndrome or serum albumin <3 g/dl. Fructosamine and HbA1c measurement related to graft dysfunction in insulin-independent subjects, chemical and metabolic variables: fasting plasma glucose, fasting C-peptide, fasting C-peptide/glucose ratio (CPGR), hemoglobin, hemoglobin change from the previous measurement, serum ferritin, and serum albumin were recorded.
All subjects performed blood glucose self-monitoring at least four times per day. Mean fasting, 2-h postprandial, and overall capillary glucose (including both fasting and postprandial) were calculated over 1 and 3 months before fructosamine/HbA1c measurements because fructosamine and HbA1c reflect the glycemic control of the previous 2–3 weeks and 90 days, respectively (11,17).
Three different assessments were performed. First, HbA1c and fructosamine were compared with mean overall capillary glucose (all values reported). Subanalysis was then performed to see if values were more closely related to fasting or postprandial capillary glucose values.
Second, HbA1c and fructosamine were directly compared to see if there was an association between the two. For this comparison only paired values were evaluated.
Third, HbA1c and fructosamine were analyzed to see if they could indicate graft dysfunction. For this values were compared to a state of insulin independence or graft dysfunction (defined above). A cutoff value of 270 was used for fructosamine because this is the upper limit of normal as per the analyzing lab (Quest Diagnostics).
The study protocol was approved by University of Miami institutional review board (IRB) and each subject gave written informed consent.
Laboratory Analysis
Serum glucose concentrations (mg/dl) were determined by hexokinase method. HbA1c levels were measured by HPLC automated analyzer (Variant II Hemoglobin Testing System®, BioRad, Richmond CA; normal range 4.27–6.07%; <2% inter- and 1.7% intra-assay coefficient of variation). Fructosamine levels were assessed by colorimetry method (Roche fructosamine assay®, Roche, Switzerland; normal range 190–270 μmol/L; 3.7% inter- and 3.7% intra-assay CV).
C-peptide was measured by double antibody radioimmunoassay (Diagnostics Products Corp., Los Angeles, CA; 0.1–5.0 ng/ml detection limit; <10% inter- and intra-assay CV, 20% cross-reactivity with proinsulin).
Statistical Analysis
Data from this study are clustered: each patient had measures taken over multiple time points for each variable. Analyses for clustered data are used for all comparisons where appropriate.
The relationship between fructosamine and HbA1c and mean capillary glucose over 1 and 3 months, and fructosamine/HbA1c and graft function expressed by CPGR were performed with linear mixed model regression. Fructosamine and HbA1c during the insulin independence period were analyzed to investigate their usefulness in predicting islet graft dysfunction by ROC curve analysis for clustered data, presented by Obuchowski (20). Specific cutoff values were assessed for their association with islet graft dysfunction by logistic regression with generalized estimating equations to appropriately estimate test variances for the clustered nature of the data. Differences were considered statistically significant at a value of p < 0.05. The analysis was performed by SAS version 9.1.3 (Cary, NC).
Results
HbA1c was measured in a total of 910 samples. After excluding the samples with possible assay interference at different time points (at least 1 g/dl reduction of hemoglobin from the previous measurement in 32 cases, erythropoietin therapy in 11, iron deficiency anemia in 8, dapsone administration in 4, and blood transfusion in 1), there were 640 HbA1c results available for analysis. Fructosamine was measured in 760 samples. Three subjects had serum albumin <3 g/dl at least in one time point, resulting in 755 samples analyzed. Mean interval between each fructosamine and HbA1c measurements during the insulin-independent period were 39 ± 34 and 36 ± 31 days, respectively (p = 0.34).
Mean fructosamine and HbA1c of all collected samples were 277 ± 40 (163–587) μmol/L and 6.1 ± 0.67% (4.5–10.5%), respectively. Mean fasting and postprandial capillary glucose were 113 ± 15 (80–237) and 124 ± 22 (51–241) mg/dl, respectively.
Both HbA1c and fructosamine had a positive association with overall and fasting mean capillary glucose values over 3 and 1 month testing intervals, respectively (Table 1). HbA1c also demonstrated a positive association with 1 month, and fructosamine with 3 months, of overall and fasting mean capillary glucose values.
Association Between HbA1c, Fructosamine, and Mean Capillary Glucose
A positive and significant association between fructosamine and HbA1c was also observed. Each 1 unit increase in HbA1c is associated with an estimated 24.46 unit increase in fructosamine (p < 0.0001) (Fig. 1).
ROC curves for both fructosamine (sensFA) and HbA1c (sensHbA1c) demonstrate a poor sensitivity for the detection of graft dysfunction.
Neither the ROC curves generated from fructosamine levels nor from HbA1c levels were able to significantly discriminate periods with subsequent graft dysfunction. However, from classification tables, we find that with a cutoff point of HbA1c >6% for this test showed a sensitivity of 50% and a specificity of 77.6% for the diagnosis of graft dysfunction (OR 2.95, 95% CI 1.41–6.14, p = 0.003). If the cutpoint was higher (HbA1c >6.5%), a lesser sensitivity and more specificity was shown (Table 2). Although the association between specific fructosamine levels and capillary blood glucose levels has not been defined, an abnormal fructosamine level >270 μmol/L (as determined by the lab range) showed a sensitivity of 62% and a specificity of 64% for the diagnosis of graft dysfunction (OR 2.47, 95% CI 1.09–5.59, p = 0.03).
The Sensitivity, Specificity, and Positive and Negative Predictive Values of HbA1c in Different Cutoff Values as a Predictor for Islet Graft Dysfunction 1 Month in Advance
Discussion
HbA1c is the gold standard for monitoring of diabetes control, while the within-subject variability and lack of conclusive evidence relating its levels with diabetes complications limit the clinical use of fructosamine. In the setting of islet transplantation, however, there are many conditions that can interfere with the accuracy of HbA1c measurements.
In this subject group, nearly all subjects (97%) met the criteria of blood loss (decrease in hemoglobin of at least 1 g/dl) at least at one time point, 33% of subjects were treated with either a single dose or a short course of erythropoietin, 24% had iron deficiency anemia, and 12% were on dapsone treatment at some time points during the follow-up. These might lead to falsely low or high HbA1c. By comparison, only three subjects (<10%) demonstrated a variable that was known to interfere with fructosamine levels. Whether this could be solved by correcting fructosamine levels for albumin is still under debate (11).
When the data were corrected for interferences, fructosamine proved to be a potentially useful tool for evaluating glycemic control in islet transplant recipients, as shown by its significant associations with HbA1c and mean overall and fasting capillary glucose. Fructosamine results, however, were more variable, lowering the validity of any single value. HbA1c is still preferred except in situations where there is obvious interference with this result (e.g., dapsone therapy) (7).
Both HbA1c and fructosamine were strongly associated with fasting capillary glucose levels as expected given the consistency of this measure in patients.
An association between fructosamine and 3-month mean capillary glucose prior to the test might not be expected. The observed significant association between these variables is likely explained by the stable glycemic control attained after islet transplantation. The significant association between HbA1c and 1-month mean capillary glucose reflects the predominant effect of the last 30 days of glycemic control on this measure.
Currently, no predictor of graft dysfunction has been identified. Capillary glucose measurements remain one of the best ways to detect dysfunction; however, they are also an indication that some graft loss has already occurred. In the future early intervention may be a first step to preserving remaining functional graft tissue. The burden of such frequent reporting of capillary glucose values in a population that has stable near normal glycemic control is a difficult one that research subjects undertake but that may not be feasible if islet transplantation were to become a more widely used treatment. The shorter time interval for fructosamine has the potential in theory to allow earlier identification of graft dysfunction.
We observed a poor sensitivity between HbA1c >6% and graft dysfunction, making it a poor predictor of graft dysfunction. At higher values the test became more specific but even less sensitive. Findings were similar with fructosamine. The greater variability of clinical results, in a population that has relatively stable glycemic control, does not allow identification of graft dysfunction with accuracy from any single value using these indicators. More commonly a trend upwards or careful evaluation of changes in capillary glucose values will demonstrate dysfunction more accurately.
In conclusion, fructosamine can be an alternative to HbA1c as a tool for glycemic assessment in islet transplant recipients who have situations that interfere with HbA1c measurement. The main limitation of fructosamine use is the variability of the test and the lack of data supporting its correlation with chronic diabetes complications (8,15).
Neither HbA1c nor fructosamine is a good predictor of islet graft dysfunction. In the future, instruments such as continuous glucose monitoring or the new glycemic marker, 1,5 anhydroglucitol (1, 5-AG), which is associated with glucose fluctuation and postprandial glucose levels (5,6), may predict graft dysfunction. The best possible measurement would be an identifier that could predict dysfunction before any change in glycemic control.
Footnotes
Acknowledgments
The authors are grateful to Suthep Wanichapol and Oludara Amole for their assistance and also the staff of the Clinical Islet Transplant Program, Islet Cell Processing Center, General Clinical Research Center for the continuous support. This study was supported by: NIH-National Center for Research Resources (U42 RR016603, M01RR16587); NIDDK (5R01-DK55347, 5R01-DK056953, R01-DK025802, 1RO1-DK25802–21; 1RO1-D59993-04); JDRF International (4-2000-946 and 4-2004-361); State of Florida; and Diabetes Research Institute Foundation.