- Email: [email protected]
Usefulness of Fructosamine for Monitoring Outpatients with Diabetes
Usefulness of Fructosamine for Monitoring Outpatients with Diabetes GARTH E. AUSTIN, MD, PHD; ROBERT WHEATON, MS; JANET RUBIN, MD; RICHARD E. MULLINS, PHD
ABSTRACT: Background: Measurements of total glycohemoglobin (glycoHb) or hemoglobin Ale are routinely used to evaluate intermediate-to-Iong term glycemic control in patients with diabetes. However, despite the recent availability of more rapid methods for glycohemoglobin determination, it remains difficult in many institutions to obtain same-day glycoHb determinations in time to assist physicians with management of outpatients with diabetes. Hence, we investigated whether fructosamine, which reflects very recent (2 to 3 weeks) glycemic control and which can be assayed more rapidly in our laboratory, could serve as a useful adjunct to glycoHb for management of these patients. Methods: Diabetes control in outpatients managed using fructosamine, fasting serum glucose, and glycoHb (concentrations from the prior visit) was compared with that for outpatients monitored using fasting serum glucose and prior glycoHb alone. The relative usefulness of fructosamine, current and prior glycoHb, and "fasting" serum glucose for evaluation and management of outpatients with diabetes was compared. In addition, the acceptance of
I
t is now generally accepted that optimal glycemic control in patients with diabetes is essential to minimize the long-term complications associated with this disease. 1- 3 The quality of glucose control in patients with diabetes is routinely determined by measuring the extent of glycosylation of hemoglobin or serum proteins. 4 ,5 Such measurements provide an estimation of time-averaged serum glucose concentrations over a period of 2 to 3 months before the assay.6,7 Glycohemoglobin (glycoHb) determinations (hemoglobin A1c or total glycohemoglobin) are the most commonly used procedures used to estimate
From the Departments of Pathology and Laboratory Medicine and Medicine (MSN, JR), Veterans Affairs Medical Center, Decatur, Georgia, and Emory University School of Medicine (GEA, RW, MSN, JR, REM), Atlanta, Georgia. Submitted December 4, 1998; accepted in revised form February 16, 1999. Correspondence: Garth E. Austin, M.D., PhD., Laboratory Service (113), Atlanta VA Medical Center, 1670 Clairmont Road, NE, Decatur, GA 30033 (E-mail: [email protected]). (GEA)
316
MARK S. NANES, MD;
fructosamine by physicians was evaluated by a questionnaire. Results: Same-day fructosamine concentrations correlated better with current glycoHb than did either prior glycoHb or current fasting serum glucose concentrations. However, the avai labi I ity of same-day fructosamine results did not objectively improve diabetes control compared with that obtained using only fasting serum glucose values and prior glycoHb concentrations. Nonetheless, most examining physicians stated that same-day fructosamine concentrations helped them significantly with diabetes management, primarily because many patients do not adequately monitor home blood glucose levels. Conclusions: Fructosamine may be a useful adjunctive test for management of outpatients with diabetes in situations where it is not practical to obtain same-day glycoHb concentrations, for patients less compl iant with home glucose monitoring, or where recent changes in insulin dose or clinical presentation might not be reflected in the glycoHb levels. KEY INDEXING TERMS: Fructosamine; Diabetes mellitus; Glycohemoglobin [Am J
Med Sci 1999;318(5):316-23.]
protein glycosylation. However, in the case of outpatients with diabetes, same-day glycohemoglobin concentrations often cannot be provided to the physician quickly enough to assist with patient evaluation in the clinic. In addition, glycohemoglobin values may be misleading when the interval between patient visits is 1 month or less, when recent changes in insulin dose have been made, or in the case of anemic patients with high rates of red blood cell turnover. The fructosamine test, a rapid, inexpensive assay that provides a measure of glycated serum proteins, is a potential substitute for or adjunct to glycoHb determinations.B,g This test measures ketoamines, which are formed by the reaction of glucose moieties with specific amino acids of the serum proteins. The fructosamine assay is more sensitive to changes in glucose concentration than are glycoHb assays because of the short (average, 17 days) half-life of serum proteins. The first generation offructosamine assays suffered from problems of lack of specificity, November 1999 Volume 318 Number 5
Austin et 01
lack of standardization among laboratories, difficulty in calibrating the assay, and susceptibility to interference by hyperlipidemia. 10 - 12 However, second-generation assays have largely overcome these problems. These newer assays seem to be highly specific and free from interference by urates and triglycerides. 13 Large numbers of fructosamine tests can be performed rapidly using anyone of a variety of automated instruments, permitting results to be available for same-day clinic visits. At the time this study was instituted, we were unable to provide same-day glycoHb values at our institution in time for use in managing outpatients with diabetes and had to rely on values obtained at the time of the previous clinic visit (prior glycoHb). We hypothesized that same-day serum fructosamine concentrations would (1) correlate better with sameday glycoHb concentrations than would prior glycoHb concentrations, (2) provide a satisfactory substitute for same-day glycoHb when it could not readily obtained, and (3) provide information on recent glycemic control that would be useful to physicians managing these patients. To test this hypothesis we compared the relative usefulness of fructosamine, glycoHb, fasting serum glucose, and home glucose monitoring in long-term evaluation and management of outpatients with diabetes in a VA outpatient clinic setting. The study design was to compare diabetes control for patients who were managed using only fasting serum glucose values and prior glycoHb values with that for patients managed using same-day fructosamine values in addition to glucose values and prior glycoHb values. All patients were encouraged to monitor home blood glucose and to bring records of their monitoring to clinic visits.
Methods All patients being treated in the outpatient diabetes clinic of the Atlanta VA Medical Center over an 1S-month period were invited to join the study. The subjects who agreed to participate by informed consent were classified as patients with type 2 diabetes 14 (non-insulin-dependent diabetes according to pre-1997 American Diabetes Association criteria13 ) or type 1 diabetes (insulin-dependent diabetes). The patients were divided into control and study rroups by social security number (SSN) (even SSN, study; odd SSN, control). The participating physicians were staff endocrinologists at this medical center and residents working under their direction. On the morning of each clinic visit, blood samples were drawn and serum fioucto!!/lmine, serum glucose, and total glycoHb concentrations were determined. At the time of the visit, physicians treating study group patients received results of serum glucose and fructosamine tests performed that morning, and total glycoHb results from the date of the previous (3 to 6 months) clinic visit (prior glycoHb). Physicians treating control group patients received only current serum glucose and prior glycoHb values. Patients were observed serially (for all clinic visits) over an 1S-month period. Physicians provided feedback on each patient by filling out patient data sheets and answering a questionnaire. The patient data sheet is shown in Figure 1; the questionnaire is described below. Fructosamine tests were performed on a Cobas-Mira chemistry THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
Patient Data Sheet Date:
Pm~-tN~am--e:===~____________
Visit Number: _ _ _ _ ___ Soc Sec No: _ _ _ _ _ ___
Type of Diabetes: IDDM (Type I) _____ NIDDM (Type 2) _____ Other: Examining Physician: A.
B.
C.
D.
E.
F.
Current Therapy: Diet: _____________________________________ Oral _________________________________ Insulin: __________________________________ ag~ts:
Laboratory Data: Fasting serum glucose: ___....,-_-:Prior glycoHb: Date performed: _________ Fructosamine: -:-_ _ _ _ __ Average home glucose: Less than 8.3 mmollL 8.3-13.9 mmollL 13.9-19.4 mmollL Greaterthan 19.4mmollL Clinical Information: _________________________________
Clinical Assessm~t: Good Control:
Moderate Control:
Poor Control:
Therapeutic Decision: No change in therapy: _____ Change therapy to:
Figure 1. Patient data sheet filled out by examining physician at the time of each visit. analyzer using ROTAG reagents (Roche Diagnostic Systems, Inc., Nutley, NJ). Total glycoHb was assayed using Glyco-Gel columns (Pierce, Rockford, IL). Serum glucose was analyzed by ASTRA and CX3 analyzers (Beckman Instruments, Inc., Palo Alto, CA). Home glucose monitoring was performed by the patients during the intervals between clinic visits using Accuchek II reflectance meters (Boehringer-Mannheim, Inc., Indianapolis, IN) and Chemstrip-BG reagent strips (Boehringer-Mannheim, Inc.). Physicians had access to the home glucose monitoring results. However, those data are not reported in this study.
Statistics Within each group of patients, means, standard deviations, and standard errors of the mean were calculated separately for all analytes and for every clinic visit. Means for each analyte for the study and control groups at each visit and between different visits were compared by unpaired t-tests. Correlations between the total glycoHb, fructosamine, prior glycoHb, current glucose, and home glucose concentrations for every patient visit were determined by linear regression analysis. The statistical significance of differences between correlation coefficients was determined using a 2-tailed Fisher Z transformation for correlation coefficients. The answers to questions on the questionnaire for control and study groups and between visits were compared using J? tests.
Results A total of 214 patients were enrolled in the protocol, 109 control subjects and 105 study group patients. One hundred and fifty-two patients returned 317
Fructosamine Monitoring of Diabetics
30,--------------------------------, III
20 (5
E
g
:g
8::l
a
10
o+---~----~--~~--~--------_i
o
10
20
30
Current GlycoHb (%) Figure 2. Comparison of fasting serum glucose concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: glucose (mmollL)= 3.77 + 0.69 (% Current glycoHb). R = 0.494.
for follow-up visits over the ensuing 18 months; these patients constituted the control and study groups, for purposes of analysis. The patients in the control group consisted of 78% with type 2 diabetes and 22% with type 1 diabetes, whereas the study group consisted of 73% with type 2 diabetes and 27% with type 1 diabetes. For every clinic visit, current "fasting" serum glucose, fructosamine, and total glycoHb values were available for our analysis. In addition, the total glycoHb concentration determined at the previous visit (prior glycoHb) was available to us. We first examined these data from all clinic visits to determine whether current fructosamine, prior glycoHb, or current "fasting" serum glucose correlated best with current glycoHb concentrations. Specific comparisons of current fasting serum glucose, prior glycoHb, and fructosamine with current glycoHb for all patient visits (including both control and study groups) are shown in Figures 2, 3, and 4. Figure 2 shows that current glucose values correlated least well with current glycoHb concentrations (R = 0.494). This correlation coefficient was significantly lower (p < 0.05) than that between either prior glycoHb (Z = 3.67) or fructosamine (Z = 5.83) and current glycoHb. The relatively poor correlation between serum glucose and current glycoHb was because (although the 7 AM blood specimen drawn to assay these parameters was intended to be a fasting specimen), interviews with the patients indicated that many of them were, in fact, not fasting. Because this is likely to be the case in many other diabetes clinics as well as our own, our data suggest that "fasting" or random glucose values obtained on the morning of diabetes clinic visits may be of question318
able value for assessing anything other than the immediate status of the patient's glucose control. As expected, prior glycoHb correlated fairly well with current glycoHb concentrations (R = 0.633) (Figure 3). However, there were many outliers. These discrepancies probably resulted from a combination of the true changes in glucose control between visits and the imprecision inherent in this and most other glycoHb assays. (The day-to-day coefficients of variation of the glycoHb assay used in this study ranged from 5 to 8% depending upon the percentage of glycoHb. This compares with day-today coefficients of variation of 3 to 5% for the fructosamine assay, data not shown.) The data illustrated in Figure 4 reveal a correlation coefficient of 0.70 between fructosamine and current glycoHb, with fewer outliers observed. This correlation coefficient was significantly higher (P < 0.05, Z = 2.16) than that between prior glycoHb and current glycoHb. Because current fructosamine concentrations correlated slightly better with current glycoHb values than did prior glycoHb values, we asked whether diabetes control was better for patients in the study groups (where physicians were provided with fructosamine concentrations in addition to prior glycoHb and glucose values) than for patients in the control groups (where fructosamine data were not available to the physicians). Table 1 shows the laboratory data for the study and control groups among the patients with type 2 diabetes, by visit, along with the mean number of days elapsed from the initial visit. All measures indicate a relatively stable population with diabetes with gradual improvement in indicators of glycemic control over the course of the study. 30
20
...
o .;:
10
Q..
O+-----r----,----~----_r~--~--__,
o
10
20
30
Current %GlycoHb Figure 3. Comparison of prior glycoHb concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: prior glycoHb(%) = 3.04 + 0.74 (% current glycoHb). R = 0.633. November 1999 Volume 318 Number 5
Austin at 01
600
...
500
2:
400
"0 E Q)
c
'E
It!
300
III
0
t5 :::J
...
LL
200 100 0 0
10
20
30
Current %GlycoHb Figure 4. Comparison of current fructosamine concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: fructosamine (JLmollL) = 97.36 + 16.65 (% Current glycoHb). R = 0.700.
Although the control group seems to show slightly lower fructosamine and glycoHb values than the study group, the differences between the 2 groups were not statistically significant (at a P-value of 0.05), except for the glycoHb values from the fourth visit. It is evident from Figure 5 that for patients with type 2 diabetes, in almost all cases, changes in current glycohemoglobin and fructosamine paralleled each other from one visit to the next. (Note that the "current glycohemoglobin" concentrations were not available to the physicians until the next visit.) To rule out the possibility that the patients who came back for 4 or more visits were a select group different from the entire population, we looked specifically at the same parameters for study and con-
trol patients with type 2 diabetes who returned for at least 4 visits during the observation period. The data for these smaller groups (not presented) show trends similar to those observed for the entire set of patients with type 2 diabetes, failing to support an argument of selection bias as the cause of the observed results. Table 2 and Figure 6 show the data for the entire group of patients with type 1 diabetes. Both study and control patients showed gradual improvement in the indicators of glycemic control over the course of the investigation. None of these measures of diabetes control in the study group patients differed significantly from those of the control group. As with the patients with type 2 diabetes, the changes in current glycoHb and fructosamine from one visit to the next paralleled each other in almost every case. The data for the subset of patients with type 1 diabetes observed for 4 or more visits (not presented) show trends similar to those seen in the overall population of patients with type 1 diabetes, again failing to support an argument of selection bias. These results showed that fructosamine correlated slightly better with current glycoHb than did prior glycoHb. However, according to the objective criteria shown in Tables 1 and 2, providing physicians with current fructosamine concentrations (in addition to fasting serum glucose and prior glycoHb concentrations obtained at the previous clinic visit) did not significantly improve glucose control in the study group patients compared with the control patients, for whom fructosamine concentrations were not available. Despite this, it was important to determine whether the physicians caring for these patients felt that fructosamine concentrations did, indeed, provide information that assisted them in managing their patients or strengthened their confidence in the clinical decisions that they made. These subjec-
Table 1. Laboratory Test Results for Patients with Type 2 Diabetes Clinic Visit Number (days after initial visit ± SEM) 1 (-)
Analyte Study group Glucose (mmollL) Prior glycoHb (%) Fructosamine (JLmollL) Current glycoHb (%)
11.2 10.9 273 10.7
± ± ± ±
0.6 (53) 0.6 9.4 0.5
1 (-) Control group Glucose (mmollL) Prior glycoHb (%) Fructosamine (JLmollL) Current glycoHb (%)
10.4 10.9 269 10.1
± ± ± ±
0.5 (62) 0.6 9 0.4
2 (163 ± 13)
10.9 10.7 256 9.7
± ± ± ±
0.6 (52) 0.5 9.0 0.4
2 (153 ± 10)
10.0 10.1 249 9.4
± ± ± ±
0.4 (62) 0.4 8 0.3
3 (302 ± 23)
11.2 10.2 258 10.2
± ± ± ±
0.8 (25) 0.5 14.6 0.4
3 (292 ± 20)
10.0 9.5 251 9.6
± ± ± ±
0.5 (32) 0.4 10 0.5
4 (358 ± 34)
9.8 10.1 275 10.3
± ± ± ±
1.2 (13) 0.7 16.4 0.6
4 (366 ± 40)
9.7 9.2 248 8.5
± ± ± ±
0.4 (10) 0.6 14 0.5
5 (408 ± 71)
10.9 10.0 268 10.0
± ± ± ±
0.9 (4) 0.5 19.3 0.9
5 (403 ± 40)
10.4 9.0 235 8.9
± ± ± ±
1.9 (4) 0.8 3 1.3
Test results are presented as mean ± SEM. The numbers in parentheses are the number of patients in the groups. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
319
Fructosomine Monitoring of Diabetics
----
4.0
N
~
Q,
N
Current%GHb
0
11 3.5
><
...I
10
'0
E
..=,
><
U
...I
>C;
3.0
... ~
9
CD
.
CD
·S
as
..
::I ()
2.5
I/)
8
0
11 .t:I J: 0 10
=== 0
E
3.0
..=,
.
.. . C
~
E as
::I ()
2.5
.. .
I/)
8
0
u
U
>C; ~
9
CD C
C
C
Current%GHb
3.5
ID ::E: 0
12 Fructosamine
~
I
CD
0
....
----
4.0
12 Fructosamine
U
::I
::I
u..
u.. Ole
Two
A
Three
Visit
7
2.0
7
2.0 Five
Four
Ole
Two
B
Number
Three
Visit
Four
Five
Number
Figure 5. Comparison of mean values for fructosamine and current glycoHb at each clinic visit for patients with type 2 diabetes. (A) Study group; (B) control group.
group patient visits as they were during study group patient visits. In 74 of 185 responses (40%) from control office visits, the physician made a change in therapy. By comparison, a change in therapy was made in 83 of 227 (37%) responses from study office visits (P > 0.05). Questions 2 and 3 applied only to the study group patients. The physicians felt that the same-day fructosamine value was consistent with other laboratory data and with the clinical status of the patient in 184 of 244 (75%) of the office visits (83% of res ponses). The fructosamine concentration was not felt to be consistent with other information in 37 of 244 (15%) of the visits (17% of responses); this question was not answered in 23 (9%) of the visits. Hence, the fructosamine value was thought to accurately reflect the clinical status of the patients in the vast majority of cases. The physicians who treated study group clinic
tive impressions of the physicians regarding the usefulness of same-day fructosamine results for management of their patients, were quantified by assessing their responses to the questionnaire, which the clinicians were asked to complete after every patient visit. Three questions were asked: 1. Was a change in therapy made during the
visit? 2. Was the fructosamine result consistent with other laboratory data and the clinical status of the patient? 3. Was patient management affected by availability of the fructosamine value? These questionnaires were completed for 241 control group office visits and 244 study group office visits. The results are shown in Table 3. The answers to question 1 revealed that physicians were as likely to make a change in therapy during control
Table 2. Laboratory Test Results for Patients with Insulin-Dependent Diabetes Clinic Visit Number (days after initial visit ± SEM) 1 (-)
Analyte Study group Glucose (mmollL) Prior glycoHb (%) Fructosamine (MmollL) Current glycoHb (%)
12.8 10.8 292 10.6
± ± ± ±
2 (147 ± 22)
1.2 (20) 0.6 14 0.5 1 (-)
Control group Glucose (mmollL) Prior glycoHb (%) Fructosamine (MmollL) Current glycoHb (%)
12.9 10.5 294 10.7
± ± ± ±
1.6 (17) 1.0 17 0.8
10.7 10.6 280 10.3
± ± ± ±
0.8 (20) 0.5 14 0.5
9.6 10.3 268 9.7
2 (206 ± 36)
9.7 10.7 270 10.5
4 (359 ± 48)
3 (279 ± 26)
± ± ± ±
1.1 (17) 0.8 19 0.8
± ± ± ±
1.2 (14) 0.6 16 0.6
11.1 8.4 260 9.5
± ± ± ±
2.1 (7) 0.5 28 0.7
3 (255 ± 28)
11.7 9.8 279 9.3
± ± ± ±
1.7 (6) 1.7 28 1.2
5 (368 ± 24)
11.0 8.1 268 8.6
± ± ± ±
1.0 (4) 0.3 28 0.9
4 (425 ± 33)
9.1 9.0 260 9.2
± ± ± ±
2.6 (5) 1.4 27 1.4
Test results are presented as mean ± SEM. The numbers in parentheses are the number of patients in the group.
320
November 1999 Volume 318 Number 5
Austin et al
4.0 4.0
-----
N' iii 0
...
--0--
I
CD
.... 0
3.5
.a ::E: 0
~
10 3.0
9
CD
c
E III
.. III
Cl
E
~
....
0
#.
~
.
CD
C CD
8
..
12 Fructosamine Current % GHb
11
.a
::E: 0
....
:::
CJ
0
><
U
:s
2.5
-0-
3.5
11
Ie
E
N
Fructosamine Current % GHb
......."0
---
-.
12
.5 E ca
..•
10 3.0
u
>C;
.. ..
"#. 9
C CD
~
2.5
0 8
0
..
u
u
:s
~
II..
2.0
7 Q1e
Two
Three
Four
Five
u..
2.0
7 Ole
Two
Three
Four
B Visit Number Visit Number A Figure 6. Comparison of mean values for fructosamine and current glycoHb at each clinic visit for patients with type 1 diabetes. (A) Study group; (B) control group. patients thought that fructosamine data influenced their management of these patients in 92 of 244 total visits (38%) (42% of all responses) and did not influence it in 52% of the visits (58% of responses). No answer was given for the remaining 10% of visits. In the 83 study group visits in which a change in therapy was reported, the fructosamine value was thought to influence the decision in 46 of the visits (55%), whereas fructosamine did not influence the decision in 35 of the cases (42%). In the remaining 2 cases, no answer was provided. Of the 144 visits in which the physician indicated that there was no change in therapy, 46 of these decisions (32%) were influenced by the availability of fructosamine, whereas 89 (62%) were not influenced by the fructosamine result. In 9 cases, this question was not answered. Discussion
The results of this study demonstrate that fructosamine concentrations can be used to effectively monitor glucose control in patients with insulindependent or non-insulin-dependent diabetes. Fructosamine concentrations generally correlated well with current glycoHb values and were consistent with the clinical status of the patients in the majority of cases. Importantly, fructosamine concentrations correlated significantly better with current glycoHb concentrations than did prior glycoHb values, suggesting that current fructosamine concentrations provide a better way to monitor outpatients with diabetes than do prior glycoHb concentrations. On the other hand, current serum glucose concentrations showed poor correlation with current glycoHb concentrations and were of value only in the evaluation of the immediate status of the patients. The responses to the questionnaires indicated THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
that a substantial percentage of physicians found that the availability of fructosamine concentrations (in addition to prior glycoHb concentrations) provided them with useful information that helped them make treatment decisions for their patients. Nonetheless, patients in both the study and control groups showed similar glycohemoglobin concentrations over time. Hence, it seems that although sameday fructosamine concentrations were useful and gave the clinicians increased confidence in the clinical decisions they made, fructosamine concentrations were not absolutely necessary for managing most of the patients as long as current glucose concentrations, prior glycoHb concentrations, home glucose values, and other clinical information were available. One probable reason that long-term glucose control was similar in control and study group patients is the stable nature of the population with diabetes under study, consisting primarily of patients with type 2 diabetes. Another reason may be that this study was carried out in the pre-Diabetes Control and Complication Trial era, such that the perceived need to make frequent treatment interventions to achieve tight control of diabetes was undoubtedly less than at present. In fact, a change in therapy was made in only a third of the visits. Perhaps the most important reason that availability of fructosamine levels did not significantly improve glucose control is because physicians rely on multiple types of clinical and biochemical information in clinical decision making; in this study, fructosamine values constituted only a small part of the total information available. Because our study showed that same-day fructosamine concentrations had subjective value, the participating clinicians requested that in lieu· of 321
Fructosamine Monitoring of Diabetics
Table 3. Responses to Questionnaire Question 1: Was a change in therapy made? Number of Patient Visits (%) Response
Total Population
Control Group
Study Group
Yes No No response Total
157 (32%) 255 (53%) 73 (15%) 485
74 (31%) 111 (46%) 56 (23%) 241
83 (34%) 144 (59%) 17 (7%) 244
Question 2: For study group patients, was fructosamine consistent with other laboratory data? Response
Number of Patient Visits (%)
Yes No No response
184 (75%) 37 (15%) 23 (9%)
Question 3: Was management of study group patients influenced by fructosamine result? Response
Number of Patient Visits (%)
Yes No No response
92 (38%) 126 (52%) 26 (10%)
Question 4: What was the impact of fructosamine levels on decisions to change therapy? Change of Therapy Impact of Fructosamine on Decision
Yes (%)
No (%)
Yes No No response
46 (55%) 35 (42%) 2 (2%)
46 (32%) 89 (62%) 9 (6%)
same-day glycoHb, our hospital laboratory make available same-day fructosamine testing for all their subsequent outpatient diabetes clinic visits. For 6 months, we continued to make available total glycoHb concentrations on a next-visit availability basis, in addition to the same-day fructosamine concentrations, and about 20% of the patients received both tests during this interim period. By the end of this 6 month period, the clinicians were satisfied that separate total glycoHb determinations were unnecessary, and further in-house glycoHb testing was stopped. More recently, with the availability of more rapid methods to perform hemoglobin Alc, we have begun provision of same-day hemoglobin Alc testing. This provides an added degree of assurance with regard to assessment of the intermediate and long-term glucose control in these patients. However, because our method for Alc determination requires a longer time for completion than that for fructosamine, we continue to experience problems in providing hemoglobin Alc values on all patients in time for their clinic visits. The most appropriate role for fructosamine in 322
managing outpatients with diabetes remains somewhat controversial. In the recent Diabetes Control and Complications Trial, 1 hemoglobin Alc was used to monitor the patients, and hemoglobin Alc (or total glycoHb) remains the criterion standard method for monitoring long-term diabetes control, in the view of most people in the field. Two fairly recent studies l6,l7 found relatively poor correlations between hemoglobin Alc and fructosamine in patients with diabetes. Likewise, in specific groups, such as dialysis patients, in whom hypoalbuminemia is common, fructosamine must be normalized for albumin concentration or it may give misleading information concerning diabetes control,18 Furthermore, blood samples with elevated concentrations of highly glycosylated proteins such as IgA may show spuriously high fructosamine concentrations. On the other hand, most other studies have suggested that fructosamine may be an acceptable alternative to hemoglobin Alc for monitoring diabetes control, particularly in specific population groups. Lin et al l9 found a high correlation between fructosamine and both hemoglobin Alc and fasting serum glucose and noted that fructosamine offered advantages of convenience, speed, and cost over hemoglobin Alc. Meneilly and Tessier20 reviewed glucose monitoring in elderly patients with diabetes and concluded that fructosamine may be superior to hemoglobin Alc for monitoring long-term glucose control in such patients. Tahara and Shima2l found that changes in both fructosamine and hemoglobin Alc decreased in a manner consistent with theory, after normalization of serum glucose, which suggests that either test might be used to monitor changes in diabetes control. Gunter et al 22 and Hughes et a12 3 found fructosamine satisfactory for monitoring glucose control in pregnancy, provided that changes in total protein concentration are taken into account, whereas Uncu et a12 4 found fructosamine and hemoglobin Alc to be valuable as screening tests for gestational diabetes. Tvorogova et al 25 found fructosamine to correlate better with glucose concentrations than did glycosylated hemoglobin. Defallah et al 26 found that fructosamine concentrations are not affected by sickle cell anemia or G6PD deficiency and provide a useful tool for monitoring patients with diabetes in regions with high prevalence of hemoglobinopathies and G6PD deficiency. Likewise, Islam et a12 7 found fructosamine to be a useful alternative to hemoglobin Alc, particularly in developing countries, where its low cost and convenience are important advantages. Our data suggest that fructosamine may be particularly suitable, at least in an adjunctive role, for monitoring outpatients with diabetes. Because fructosamine provides information about quite recent (2 to 3 weeks) diabetes control, whereas hemoglobin Alc levels provide estimates of average glucose control over a considerably longer (2 to 3 months) period November 1999 Volume 318 Number 5
Austin et al
of time, the information provided by these 2 tests should generally be viewed as additive rather than redundant. Imperfect correlations between fructosamine and Alc levels found in some studies probably reflect this fact, together with the high degree of imprecision inherent until recently in most hemoglobin Alc assays. Although factors such as hypoalbuminemia or high levels of IgA may, in theory, affect fructosamine levels (as mentioned above), we have found this not to be a problem for the vast majority of patients, particularly for observing clinic outpatients over time. In those few instances in which fructosamine results seem not to correlate with other information, the baseline fructosamine assay can be compared against hemoglobin Alc or total glycoHb concentrations. Even with the recent availability of new, rapid methods for hemoglobin Alc determination, fructosamine concentrations often can still be obtained more rapidly, conveniently, and cheaply than glycoHb concentrations. Furthermore, when outpatient visits are spaced only a few weeks apart, fructosamine may provide a much better indication of changes in glycemic control in response to therapy than glycated hemoglobin because of the differences in their half-lives. For these reasons, we believe that fructosamine continues to provide a useful adjunct (and in selective cases, alternative) to glycoHb concentrations for monitoring outpatients with diabetes, particularly in an outpatient setting.
7. 8. 9.
10. 11. 12. 13. 14. 15. 16. 17. 18.
19.
Acknowledgments
We gratefully acknowledge the assistance of other members of the Endocrinology Division of the Medical Service of the Atlanta VAMC in carrying out this study.
21.
References
22.
1. The Diabetes Control, and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977-86. 2. Reichard P, Nilsson B-Y, Rosenqvist U. The effect of long-term intensified insulin treatment on the development of microvascular complications of diabetes mellitus. N Engl J Med 1993;329:304-9. 3. Wang PH, Lau J, Chalmers TC. Meta-analysis of effects of intensive blood-glucose control on late complications of type 1 diabetes. Lancet 1993;341:1306-9. 4. Mayer TK, Freedman ZR. Protein glycosylation in diabetes mellitus: a review of laboratory measurements and of their clinical utility. Clin Chim Acta 1983;127:147-84. 5. Benjamin RJ, Sacks DB. Glycated protein update: Implications of recent studies, including the diabetes control and complications trial. Clin Chem 1994;40:683-7. 6. Nathan DM, Singer DE, Hurxthal K, et al. The clinical
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
20.
23. 24.
25.
26.
27.
information value of the glycosylated hemoglobin assay. N Engl J Med 1984;310:341-6. Austin GE, Mullins RH, Morin LG. Nonenzymic glycosylation of individual plasma proteins in normoglycemic and hyperglycemic patients. Clin Chem 1987;33:220-4. Armbruster DA. Fructosamine: structure, analysis, and clinical usefulness {Review}. Clin Chem 1987;33:2153-63. Gebhart S, Wheaton R, Mullins RE, et al. A comparison of home glucose monitoring with determination of HbA1c, total glycoHb, fructosamine, and random serum glucose in diabetic patients. Arch Intern Med 1991;151:1133-7. Schleicher ED, Mayer R, Wagner EM, et al. Is serum fructosamine assay specific for determination of glycated serum protein? Clin Chem 1988;34:320-3. Phillipou G, Seaborn CJ, Phillips PJ. Re-evaluation of the fructosamine reaction. Clin Chem 1988;34:1561-4. Schleicher ED, Vogt BW. Standardization of serum fructosamine assays. Clin Chem 1990;36:136-9. Kallner A. Influence of triglycerides and urate on methods for determination offructosamine. Clin Chim Acta 1992;207: 99-106. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997;20: 118397. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039-57. Shield JP, Poysner K, Hunt L, et al. Fructosamine and glycated haemoglobin in the assessment of long term glycaemic control in diabetes. Arch Dis Child 1994;71:443-5. Furuseth K, Bruusgaard D, Rutle 0, et al. Fructosamine cannot replace HbA1c in the management of type 2 diabetes (NIDDM). Scand J Prim Health Care 1994;12:219-24. Ichikawa H, Nagake Y, Takahashi M, et al. What is the best index of diabetic control in patients with diabetes mellitus on hemodialysis? Nippon Jinzo Gakkai Shi 1996;38: 305-8. Lin MJ, Hoke C, Ettinger B, et al. Technical performance evaluation of BMlHitachi 747-200 serum fructosamine assay. Clin Chem 1996;42:244-8. Meneilly GS, Tessier D. Diabetes in the elderly. Diabet Med 1995;12:949-60. Tahara Y, Shima K. Kinetics of HbA1c, glycated albumin, and fructosamine and analysis of their weight functions against preceding plasma glucose concentration. Diabetes Care 1995;18:440-7. Gunter HH, Ritter C, Reinhardt W, et al. Influence of non-diabetic pregnancy on fructosamine and HbA1 concentration. Z Geburtshilfe Neonatol 1995;199:148-55. Hughes PF, Agarwal M, Newman P, et al. An evaluation of fructosamine estimation in screening for gestational diabetes mellitus. Diabet Med 1995;12:708-712. Uncu G, Ozan H, Cengiz C. The comparison of 50 grams glucose challenge test, HbA1, and fructosamine concentrations in diagnosis of gestational diabetes mellitus. Clin Exp Obstet Gynecol 1995;22:230-4. Tvorogova MG, Zaitseva DTM, Aronskaia EE, et al. Comparison offructosmine, glycosylated hemoglobin and glucose in patients with type II diabetes. Klin Lab Diagn 1995; 1:10-2. Defallah AA, Eskandarani H, Rehaimi A, et al. Fructosamine in HbS and G6PD-deficient Saudi Arabs in the Eastern Province of Saudi Arabia. Br J Biomed Sci 1994;51: 332-5. Islam N, Akhter J, Kayani N, et a1. Fructosamine: an alternative assessment of past glycaemic control in developing countries. J Pak Med Assoc 1993;43:238-40.
323
ABSTRACT: Background: Measurements of total glycohemoglobin (glycoHb) or hemoglobin Ale are routinely used to evaluate intermediate-to-Iong term glycemic control in patients with diabetes. However, despite the recent availability of more rapid methods for glycohemoglobin determination, it remains difficult in many institutions to obtain same-day glycoHb determinations in time to assist physicians with management of outpatients with diabetes. Hence, we investigated whether fructosamine, which reflects very recent (2 to 3 weeks) glycemic control and which can be assayed more rapidly in our laboratory, could serve as a useful adjunct to glycoHb for management of these patients. Methods: Diabetes control in outpatients managed using fructosamine, fasting serum glucose, and glycoHb (concentrations from the prior visit) was compared with that for outpatients monitored using fasting serum glucose and prior glycoHb alone. The relative usefulness of fructosamine, current and prior glycoHb, and "fasting" serum glucose for evaluation and management of outpatients with diabetes was compared. In addition, the acceptance of
I
t is now generally accepted that optimal glycemic control in patients with diabetes is essential to minimize the long-term complications associated with this disease. 1- 3 The quality of glucose control in patients with diabetes is routinely determined by measuring the extent of glycosylation of hemoglobin or serum proteins. 4 ,5 Such measurements provide an estimation of time-averaged serum glucose concentrations over a period of 2 to 3 months before the assay.6,7 Glycohemoglobin (glycoHb) determinations (hemoglobin A1c or total glycohemoglobin) are the most commonly used procedures used to estimate
From the Departments of Pathology and Laboratory Medicine and Medicine (MSN, JR), Veterans Affairs Medical Center, Decatur, Georgia, and Emory University School of Medicine (GEA, RW, MSN, JR, REM), Atlanta, Georgia. Submitted December 4, 1998; accepted in revised form February 16, 1999. Correspondence: Garth E. Austin, M.D., PhD., Laboratory Service (113), Atlanta VA Medical Center, 1670 Clairmont Road, NE, Decatur, GA 30033 (E-mail: [email protected]). (GEA)
316
MARK S. NANES, MD;
fructosamine by physicians was evaluated by a questionnaire. Results: Same-day fructosamine concentrations correlated better with current glycoHb than did either prior glycoHb or current fasting serum glucose concentrations. However, the avai labi I ity of same-day fructosamine results did not objectively improve diabetes control compared with that obtained using only fasting serum glucose values and prior glycoHb concentrations. Nonetheless, most examining physicians stated that same-day fructosamine concentrations helped them significantly with diabetes management, primarily because many patients do not adequately monitor home blood glucose levels. Conclusions: Fructosamine may be a useful adjunctive test for management of outpatients with diabetes in situations where it is not practical to obtain same-day glycoHb concentrations, for patients less compl iant with home glucose monitoring, or where recent changes in insulin dose or clinical presentation might not be reflected in the glycoHb levels. KEY INDEXING TERMS: Fructosamine; Diabetes mellitus; Glycohemoglobin [Am J
Med Sci 1999;318(5):316-23.]
protein glycosylation. However, in the case of outpatients with diabetes, same-day glycohemoglobin concentrations often cannot be provided to the physician quickly enough to assist with patient evaluation in the clinic. In addition, glycohemoglobin values may be misleading when the interval between patient visits is 1 month or less, when recent changes in insulin dose have been made, or in the case of anemic patients with high rates of red blood cell turnover. The fructosamine test, a rapid, inexpensive assay that provides a measure of glycated serum proteins, is a potential substitute for or adjunct to glycoHb determinations.B,g This test measures ketoamines, which are formed by the reaction of glucose moieties with specific amino acids of the serum proteins. The fructosamine assay is more sensitive to changes in glucose concentration than are glycoHb assays because of the short (average, 17 days) half-life of serum proteins. The first generation offructosamine assays suffered from problems of lack of specificity, November 1999 Volume 318 Number 5
Austin et 01
lack of standardization among laboratories, difficulty in calibrating the assay, and susceptibility to interference by hyperlipidemia. 10 - 12 However, second-generation assays have largely overcome these problems. These newer assays seem to be highly specific and free from interference by urates and triglycerides. 13 Large numbers of fructosamine tests can be performed rapidly using anyone of a variety of automated instruments, permitting results to be available for same-day clinic visits. At the time this study was instituted, we were unable to provide same-day glycoHb values at our institution in time for use in managing outpatients with diabetes and had to rely on values obtained at the time of the previous clinic visit (prior glycoHb). We hypothesized that same-day serum fructosamine concentrations would (1) correlate better with sameday glycoHb concentrations than would prior glycoHb concentrations, (2) provide a satisfactory substitute for same-day glycoHb when it could not readily obtained, and (3) provide information on recent glycemic control that would be useful to physicians managing these patients. To test this hypothesis we compared the relative usefulness of fructosamine, glycoHb, fasting serum glucose, and home glucose monitoring in long-term evaluation and management of outpatients with diabetes in a VA outpatient clinic setting. The study design was to compare diabetes control for patients who were managed using only fasting serum glucose values and prior glycoHb values with that for patients managed using same-day fructosamine values in addition to glucose values and prior glycoHb values. All patients were encouraged to monitor home blood glucose and to bring records of their monitoring to clinic visits.
Methods All patients being treated in the outpatient diabetes clinic of the Atlanta VA Medical Center over an 1S-month period were invited to join the study. The subjects who agreed to participate by informed consent were classified as patients with type 2 diabetes 14 (non-insulin-dependent diabetes according to pre-1997 American Diabetes Association criteria13 ) or type 1 diabetes (insulin-dependent diabetes). The patients were divided into control and study rroups by social security number (SSN) (even SSN, study; odd SSN, control). The participating physicians were staff endocrinologists at this medical center and residents working under their direction. On the morning of each clinic visit, blood samples were drawn and serum fioucto!!/lmine, serum glucose, and total glycoHb concentrations were determined. At the time of the visit, physicians treating study group patients received results of serum glucose and fructosamine tests performed that morning, and total glycoHb results from the date of the previous (3 to 6 months) clinic visit (prior glycoHb). Physicians treating control group patients received only current serum glucose and prior glycoHb values. Patients were observed serially (for all clinic visits) over an 1S-month period. Physicians provided feedback on each patient by filling out patient data sheets and answering a questionnaire. The patient data sheet is shown in Figure 1; the questionnaire is described below. Fructosamine tests were performed on a Cobas-Mira chemistry THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
Patient Data Sheet Date:
Pm~-tN~am--e:===~____________
Visit Number: _ _ _ _ ___ Soc Sec No: _ _ _ _ _ ___
Type of Diabetes: IDDM (Type I) _____ NIDDM (Type 2) _____ Other: Examining Physician: A.
B.
C.
D.
E.
F.
Current Therapy: Diet: _____________________________________ Oral _________________________________ Insulin: __________________________________ ag~ts:
Laboratory Data: Fasting serum glucose: ___....,-_-:Prior glycoHb: Date performed: _________ Fructosamine: -:-_ _ _ _ __ Average home glucose: Less than 8.3 mmollL 8.3-13.9 mmollL 13.9-19.4 mmollL Greaterthan 19.4mmollL Clinical Information: _________________________________
Clinical Assessm~t: Good Control:
Moderate Control:
Poor Control:
Therapeutic Decision: No change in therapy: _____ Change therapy to:
Figure 1. Patient data sheet filled out by examining physician at the time of each visit. analyzer using ROTAG reagents (Roche Diagnostic Systems, Inc., Nutley, NJ). Total glycoHb was assayed using Glyco-Gel columns (Pierce, Rockford, IL). Serum glucose was analyzed by ASTRA and CX3 analyzers (Beckman Instruments, Inc., Palo Alto, CA). Home glucose monitoring was performed by the patients during the intervals between clinic visits using Accuchek II reflectance meters (Boehringer-Mannheim, Inc., Indianapolis, IN) and Chemstrip-BG reagent strips (Boehringer-Mannheim, Inc.). Physicians had access to the home glucose monitoring results. However, those data are not reported in this study.
Statistics Within each group of patients, means, standard deviations, and standard errors of the mean were calculated separately for all analytes and for every clinic visit. Means for each analyte for the study and control groups at each visit and between different visits were compared by unpaired t-tests. Correlations between the total glycoHb, fructosamine, prior glycoHb, current glucose, and home glucose concentrations for every patient visit were determined by linear regression analysis. The statistical significance of differences between correlation coefficients was determined using a 2-tailed Fisher Z transformation for correlation coefficients. The answers to questions on the questionnaire for control and study groups and between visits were compared using J? tests.
Results A total of 214 patients were enrolled in the protocol, 109 control subjects and 105 study group patients. One hundred and fifty-two patients returned 317
Fructosamine Monitoring of Diabetics
30,--------------------------------, III
20 (5
E
g
:g
8::l
a
10
o+---~----~--~~--~--------_i
o
10
20
30
Current GlycoHb (%) Figure 2. Comparison of fasting serum glucose concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: glucose (mmollL)= 3.77 + 0.69 (% Current glycoHb). R = 0.494.
for follow-up visits over the ensuing 18 months; these patients constituted the control and study groups, for purposes of analysis. The patients in the control group consisted of 78% with type 2 diabetes and 22% with type 1 diabetes, whereas the study group consisted of 73% with type 2 diabetes and 27% with type 1 diabetes. For every clinic visit, current "fasting" serum glucose, fructosamine, and total glycoHb values were available for our analysis. In addition, the total glycoHb concentration determined at the previous visit (prior glycoHb) was available to us. We first examined these data from all clinic visits to determine whether current fructosamine, prior glycoHb, or current "fasting" serum glucose correlated best with current glycoHb concentrations. Specific comparisons of current fasting serum glucose, prior glycoHb, and fructosamine with current glycoHb for all patient visits (including both control and study groups) are shown in Figures 2, 3, and 4. Figure 2 shows that current glucose values correlated least well with current glycoHb concentrations (R = 0.494). This correlation coefficient was significantly lower (p < 0.05) than that between either prior glycoHb (Z = 3.67) or fructosamine (Z = 5.83) and current glycoHb. The relatively poor correlation between serum glucose and current glycoHb was because (although the 7 AM blood specimen drawn to assay these parameters was intended to be a fasting specimen), interviews with the patients indicated that many of them were, in fact, not fasting. Because this is likely to be the case in many other diabetes clinics as well as our own, our data suggest that "fasting" or random glucose values obtained on the morning of diabetes clinic visits may be of question318
able value for assessing anything other than the immediate status of the patient's glucose control. As expected, prior glycoHb correlated fairly well with current glycoHb concentrations (R = 0.633) (Figure 3). However, there were many outliers. These discrepancies probably resulted from a combination of the true changes in glucose control between visits and the imprecision inherent in this and most other glycoHb assays. (The day-to-day coefficients of variation of the glycoHb assay used in this study ranged from 5 to 8% depending upon the percentage of glycoHb. This compares with day-today coefficients of variation of 3 to 5% for the fructosamine assay, data not shown.) The data illustrated in Figure 4 reveal a correlation coefficient of 0.70 between fructosamine and current glycoHb, with fewer outliers observed. This correlation coefficient was significantly higher (P < 0.05, Z = 2.16) than that between prior glycoHb and current glycoHb. Because current fructosamine concentrations correlated slightly better with current glycoHb values than did prior glycoHb values, we asked whether diabetes control was better for patients in the study groups (where physicians were provided with fructosamine concentrations in addition to prior glycoHb and glucose values) than for patients in the control groups (where fructosamine data were not available to the physicians). Table 1 shows the laboratory data for the study and control groups among the patients with type 2 diabetes, by visit, along with the mean number of days elapsed from the initial visit. All measures indicate a relatively stable population with diabetes with gradual improvement in indicators of glycemic control over the course of the study. 30
20
...
o .;:
10
Q..
O+-----r----,----~----_r~--~--__,
o
10
20
30
Current %GlycoHb Figure 3. Comparison of prior glycoHb concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: prior glycoHb(%) = 3.04 + 0.74 (% current glycoHb). R = 0.633. November 1999 Volume 318 Number 5
Austin at 01
600
...
500
2:
400
"0 E Q)
c
'E
It!
300
III
0
t5 :::J
...
LL
200 100 0 0
10
20
30
Current %GlycoHb Figure 4. Comparison of current fructosamine concentrations at each patient visit with corresponding current glycoHb values. Linear regression equation: fructosamine (JLmollL) = 97.36 + 16.65 (% Current glycoHb). R = 0.700.
Although the control group seems to show slightly lower fructosamine and glycoHb values than the study group, the differences between the 2 groups were not statistically significant (at a P-value of 0.05), except for the glycoHb values from the fourth visit. It is evident from Figure 5 that for patients with type 2 diabetes, in almost all cases, changes in current glycohemoglobin and fructosamine paralleled each other from one visit to the next. (Note that the "current glycohemoglobin" concentrations were not available to the physicians until the next visit.) To rule out the possibility that the patients who came back for 4 or more visits were a select group different from the entire population, we looked specifically at the same parameters for study and con-
trol patients with type 2 diabetes who returned for at least 4 visits during the observation period. The data for these smaller groups (not presented) show trends similar to those observed for the entire set of patients with type 2 diabetes, failing to support an argument of selection bias as the cause of the observed results. Table 2 and Figure 6 show the data for the entire group of patients with type 1 diabetes. Both study and control patients showed gradual improvement in the indicators of glycemic control over the course of the investigation. None of these measures of diabetes control in the study group patients differed significantly from those of the control group. As with the patients with type 2 diabetes, the changes in current glycoHb and fructosamine from one visit to the next paralleled each other in almost every case. The data for the subset of patients with type 1 diabetes observed for 4 or more visits (not presented) show trends similar to those seen in the overall population of patients with type 1 diabetes, again failing to support an argument of selection bias. These results showed that fructosamine correlated slightly better with current glycoHb than did prior glycoHb. However, according to the objective criteria shown in Tables 1 and 2, providing physicians with current fructosamine concentrations (in addition to fasting serum glucose and prior glycoHb concentrations obtained at the previous clinic visit) did not significantly improve glucose control in the study group patients compared with the control patients, for whom fructosamine concentrations were not available. Despite this, it was important to determine whether the physicians caring for these patients felt that fructosamine concentrations did, indeed, provide information that assisted them in managing their patients or strengthened their confidence in the clinical decisions that they made. These subjec-
Table 1. Laboratory Test Results for Patients with Type 2 Diabetes Clinic Visit Number (days after initial visit ± SEM) 1 (-)
Analyte Study group Glucose (mmollL) Prior glycoHb (%) Fructosamine (JLmollL) Current glycoHb (%)
11.2 10.9 273 10.7
± ± ± ±
0.6 (53) 0.6 9.4 0.5
1 (-) Control group Glucose (mmollL) Prior glycoHb (%) Fructosamine (JLmollL) Current glycoHb (%)
10.4 10.9 269 10.1
± ± ± ±
0.5 (62) 0.6 9 0.4
2 (163 ± 13)
10.9 10.7 256 9.7
± ± ± ±
0.6 (52) 0.5 9.0 0.4
2 (153 ± 10)
10.0 10.1 249 9.4
± ± ± ±
0.4 (62) 0.4 8 0.3
3 (302 ± 23)
11.2 10.2 258 10.2
± ± ± ±
0.8 (25) 0.5 14.6 0.4
3 (292 ± 20)
10.0 9.5 251 9.6
± ± ± ±
0.5 (32) 0.4 10 0.5
4 (358 ± 34)
9.8 10.1 275 10.3
± ± ± ±
1.2 (13) 0.7 16.4 0.6
4 (366 ± 40)
9.7 9.2 248 8.5
± ± ± ±
0.4 (10) 0.6 14 0.5
5 (408 ± 71)
10.9 10.0 268 10.0
± ± ± ±
0.9 (4) 0.5 19.3 0.9
5 (403 ± 40)
10.4 9.0 235 8.9
± ± ± ±
1.9 (4) 0.8 3 1.3
Test results are presented as mean ± SEM. The numbers in parentheses are the number of patients in the groups. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
319
Fructosomine Monitoring of Diabetics
----
4.0
N
~
Q,
N
Current%GHb
0
11 3.5
><
...I
10
'0
E
..=,
><
U
...I
>C;
3.0
... ~
9
CD
.
CD
·S
as
..
::I ()
2.5
I/)
8
0
11 .t:I J: 0 10
=== 0
E
3.0
..=,
.
.. . C
~
E as
::I ()
2.5
.. .
I/)
8
0
u
U
>C; ~
9
CD C
C
C
Current%GHb
3.5
ID ::E: 0
12 Fructosamine
~
I
CD
0
....
----
4.0
12 Fructosamine
U
::I
::I
u..
u.. Ole
Two
A
Three
Visit
7
2.0
7
2.0 Five
Four
Ole
Two
B
Number
Three
Visit
Four
Five
Number
Figure 5. Comparison of mean values for fructosamine and current glycoHb at each clinic visit for patients with type 2 diabetes. (A) Study group; (B) control group.
group patient visits as they were during study group patient visits. In 74 of 185 responses (40%) from control office visits, the physician made a change in therapy. By comparison, a change in therapy was made in 83 of 227 (37%) responses from study office visits (P > 0.05). Questions 2 and 3 applied only to the study group patients. The physicians felt that the same-day fructosamine value was consistent with other laboratory data and with the clinical status of the patient in 184 of 244 (75%) of the office visits (83% of res ponses). The fructosamine concentration was not felt to be consistent with other information in 37 of 244 (15%) of the visits (17% of responses); this question was not answered in 23 (9%) of the visits. Hence, the fructosamine value was thought to accurately reflect the clinical status of the patients in the vast majority of cases. The physicians who treated study group clinic
tive impressions of the physicians regarding the usefulness of same-day fructosamine results for management of their patients, were quantified by assessing their responses to the questionnaire, which the clinicians were asked to complete after every patient visit. Three questions were asked: 1. Was a change in therapy made during the
visit? 2. Was the fructosamine result consistent with other laboratory data and the clinical status of the patient? 3. Was patient management affected by availability of the fructosamine value? These questionnaires were completed for 241 control group office visits and 244 study group office visits. The results are shown in Table 3. The answers to question 1 revealed that physicians were as likely to make a change in therapy during control
Table 2. Laboratory Test Results for Patients with Insulin-Dependent Diabetes Clinic Visit Number (days after initial visit ± SEM) 1 (-)
Analyte Study group Glucose (mmollL) Prior glycoHb (%) Fructosamine (MmollL) Current glycoHb (%)
12.8 10.8 292 10.6
± ± ± ±
2 (147 ± 22)
1.2 (20) 0.6 14 0.5 1 (-)
Control group Glucose (mmollL) Prior glycoHb (%) Fructosamine (MmollL) Current glycoHb (%)
12.9 10.5 294 10.7
± ± ± ±
1.6 (17) 1.0 17 0.8
10.7 10.6 280 10.3
± ± ± ±
0.8 (20) 0.5 14 0.5
9.6 10.3 268 9.7
2 (206 ± 36)
9.7 10.7 270 10.5
4 (359 ± 48)
3 (279 ± 26)
± ± ± ±
1.1 (17) 0.8 19 0.8
± ± ± ±
1.2 (14) 0.6 16 0.6
11.1 8.4 260 9.5
± ± ± ±
2.1 (7) 0.5 28 0.7
3 (255 ± 28)
11.7 9.8 279 9.3
± ± ± ±
1.7 (6) 1.7 28 1.2
5 (368 ± 24)
11.0 8.1 268 8.6
± ± ± ±
1.0 (4) 0.3 28 0.9
4 (425 ± 33)
9.1 9.0 260 9.2
± ± ± ±
2.6 (5) 1.4 27 1.4
Test results are presented as mean ± SEM. The numbers in parentheses are the number of patients in the group.
320
November 1999 Volume 318 Number 5
Austin et al
4.0 4.0
-----
N' iii 0
...
--0--
I
CD
.... 0
3.5
.a ::E: 0
~
10 3.0
9
CD
c
E III
.. III
Cl
E
~
....
0
#.
~
.
CD
C CD
8
..
12 Fructosamine Current % GHb
11
.a
::E: 0
....
:::
CJ
0
><
U
:s
2.5
-0-
3.5
11
Ie
E
N
Fructosamine Current % GHb
......."0
---
-.
12
.5 E ca
..•
10 3.0
u
>C;
.. ..
"#. 9
C CD
~
2.5
0 8
0
..
u
u
:s
~
II..
2.0
7 Q1e
Two
Three
Four
Five
u..
2.0
7 Ole
Two
Three
Four
B Visit Number Visit Number A Figure 6. Comparison of mean values for fructosamine and current glycoHb at each clinic visit for patients with type 1 diabetes. (A) Study group; (B) control group. patients thought that fructosamine data influenced their management of these patients in 92 of 244 total visits (38%) (42% of all responses) and did not influence it in 52% of the visits (58% of responses). No answer was given for the remaining 10% of visits. In the 83 study group visits in which a change in therapy was reported, the fructosamine value was thought to influence the decision in 46 of the visits (55%), whereas fructosamine did not influence the decision in 35 of the cases (42%). In the remaining 2 cases, no answer was provided. Of the 144 visits in which the physician indicated that there was no change in therapy, 46 of these decisions (32%) were influenced by the availability of fructosamine, whereas 89 (62%) were not influenced by the fructosamine result. In 9 cases, this question was not answered. Discussion
The results of this study demonstrate that fructosamine concentrations can be used to effectively monitor glucose control in patients with insulindependent or non-insulin-dependent diabetes. Fructosamine concentrations generally correlated well with current glycoHb values and were consistent with the clinical status of the patients in the majority of cases. Importantly, fructosamine concentrations correlated significantly better with current glycoHb concentrations than did prior glycoHb values, suggesting that current fructosamine concentrations provide a better way to monitor outpatients with diabetes than do prior glycoHb concentrations. On the other hand, current serum glucose concentrations showed poor correlation with current glycoHb concentrations and were of value only in the evaluation of the immediate status of the patients. The responses to the questionnaires indicated THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
that a substantial percentage of physicians found that the availability of fructosamine concentrations (in addition to prior glycoHb concentrations) provided them with useful information that helped them make treatment decisions for their patients. Nonetheless, patients in both the study and control groups showed similar glycohemoglobin concentrations over time. Hence, it seems that although sameday fructosamine concentrations were useful and gave the clinicians increased confidence in the clinical decisions they made, fructosamine concentrations were not absolutely necessary for managing most of the patients as long as current glucose concentrations, prior glycoHb concentrations, home glucose values, and other clinical information were available. One probable reason that long-term glucose control was similar in control and study group patients is the stable nature of the population with diabetes under study, consisting primarily of patients with type 2 diabetes. Another reason may be that this study was carried out in the pre-Diabetes Control and Complication Trial era, such that the perceived need to make frequent treatment interventions to achieve tight control of diabetes was undoubtedly less than at present. In fact, a change in therapy was made in only a third of the visits. Perhaps the most important reason that availability of fructosamine levels did not significantly improve glucose control is because physicians rely on multiple types of clinical and biochemical information in clinical decision making; in this study, fructosamine values constituted only a small part of the total information available. Because our study showed that same-day fructosamine concentrations had subjective value, the participating clinicians requested that in lieu· of 321
Fructosamine Monitoring of Diabetics
Table 3. Responses to Questionnaire Question 1: Was a change in therapy made? Number of Patient Visits (%) Response
Total Population
Control Group
Study Group
Yes No No response Total
157 (32%) 255 (53%) 73 (15%) 485
74 (31%) 111 (46%) 56 (23%) 241
83 (34%) 144 (59%) 17 (7%) 244
Question 2: For study group patients, was fructosamine consistent with other laboratory data? Response
Number of Patient Visits (%)
Yes No No response
184 (75%) 37 (15%) 23 (9%)
Question 3: Was management of study group patients influenced by fructosamine result? Response
Number of Patient Visits (%)
Yes No No response
92 (38%) 126 (52%) 26 (10%)
Question 4: What was the impact of fructosamine levels on decisions to change therapy? Change of Therapy Impact of Fructosamine on Decision
Yes (%)
No (%)
Yes No No response
46 (55%) 35 (42%) 2 (2%)
46 (32%) 89 (62%) 9 (6%)
same-day glycoHb, our hospital laboratory make available same-day fructosamine testing for all their subsequent outpatient diabetes clinic visits. For 6 months, we continued to make available total glycoHb concentrations on a next-visit availability basis, in addition to the same-day fructosamine concentrations, and about 20% of the patients received both tests during this interim period. By the end of this 6 month period, the clinicians were satisfied that separate total glycoHb determinations were unnecessary, and further in-house glycoHb testing was stopped. More recently, with the availability of more rapid methods to perform hemoglobin Alc, we have begun provision of same-day hemoglobin Alc testing. This provides an added degree of assurance with regard to assessment of the intermediate and long-term glucose control in these patients. However, because our method for Alc determination requires a longer time for completion than that for fructosamine, we continue to experience problems in providing hemoglobin Alc values on all patients in time for their clinic visits. The most appropriate role for fructosamine in 322
managing outpatients with diabetes remains somewhat controversial. In the recent Diabetes Control and Complications Trial, 1 hemoglobin Alc was used to monitor the patients, and hemoglobin Alc (or total glycoHb) remains the criterion standard method for monitoring long-term diabetes control, in the view of most people in the field. Two fairly recent studies l6,l7 found relatively poor correlations between hemoglobin Alc and fructosamine in patients with diabetes. Likewise, in specific groups, such as dialysis patients, in whom hypoalbuminemia is common, fructosamine must be normalized for albumin concentration or it may give misleading information concerning diabetes control,18 Furthermore, blood samples with elevated concentrations of highly glycosylated proteins such as IgA may show spuriously high fructosamine concentrations. On the other hand, most other studies have suggested that fructosamine may be an acceptable alternative to hemoglobin Alc for monitoring diabetes control, particularly in specific population groups. Lin et al l9 found a high correlation between fructosamine and both hemoglobin Alc and fasting serum glucose and noted that fructosamine offered advantages of convenience, speed, and cost over hemoglobin Alc. Meneilly and Tessier20 reviewed glucose monitoring in elderly patients with diabetes and concluded that fructosamine may be superior to hemoglobin Alc for monitoring long-term glucose control in such patients. Tahara and Shima2l found that changes in both fructosamine and hemoglobin Alc decreased in a manner consistent with theory, after normalization of serum glucose, which suggests that either test might be used to monitor changes in diabetes control. Gunter et al 22 and Hughes et a12 3 found fructosamine satisfactory for monitoring glucose control in pregnancy, provided that changes in total protein concentration are taken into account, whereas Uncu et a12 4 found fructosamine and hemoglobin Alc to be valuable as screening tests for gestational diabetes. Tvorogova et al 25 found fructosamine to correlate better with glucose concentrations than did glycosylated hemoglobin. Defallah et al 26 found that fructosamine concentrations are not affected by sickle cell anemia or G6PD deficiency and provide a useful tool for monitoring patients with diabetes in regions with high prevalence of hemoglobinopathies and G6PD deficiency. Likewise, Islam et a12 7 found fructosamine to be a useful alternative to hemoglobin Alc, particularly in developing countries, where its low cost and convenience are important advantages. Our data suggest that fructosamine may be particularly suitable, at least in an adjunctive role, for monitoring outpatients with diabetes. Because fructosamine provides information about quite recent (2 to 3 weeks) diabetes control, whereas hemoglobin Alc levels provide estimates of average glucose control over a considerably longer (2 to 3 months) period November 1999 Volume 318 Number 5
Austin et al
of time, the information provided by these 2 tests should generally be viewed as additive rather than redundant. Imperfect correlations between fructosamine and Alc levels found in some studies probably reflect this fact, together with the high degree of imprecision inherent until recently in most hemoglobin Alc assays. Although factors such as hypoalbuminemia or high levels of IgA may, in theory, affect fructosamine levels (as mentioned above), we have found this not to be a problem for the vast majority of patients, particularly for observing clinic outpatients over time. In those few instances in which fructosamine results seem not to correlate with other information, the baseline fructosamine assay can be compared against hemoglobin Alc or total glycoHb concentrations. Even with the recent availability of new, rapid methods for hemoglobin Alc determination, fructosamine concentrations often can still be obtained more rapidly, conveniently, and cheaply than glycoHb concentrations. Furthermore, when outpatient visits are spaced only a few weeks apart, fructosamine may provide a much better indication of changes in glycemic control in response to therapy than glycated hemoglobin because of the differences in their half-lives. For these reasons, we believe that fructosamine continues to provide a useful adjunct (and in selective cases, alternative) to glycoHb concentrations for monitoring outpatients with diabetes, particularly in an outpatient setting.
7. 8. 9.
10. 11. 12. 13. 14. 15. 16. 17. 18.
19.
Acknowledgments
We gratefully acknowledge the assistance of other members of the Endocrinology Division of the Medical Service of the Atlanta VAMC in carrying out this study.
21.
References
22.
1. The Diabetes Control, and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977-86. 2. Reichard P, Nilsson B-Y, Rosenqvist U. The effect of long-term intensified insulin treatment on the development of microvascular complications of diabetes mellitus. N Engl J Med 1993;329:304-9. 3. Wang PH, Lau J, Chalmers TC. Meta-analysis of effects of intensive blood-glucose control on late complications of type 1 diabetes. Lancet 1993;341:1306-9. 4. Mayer TK, Freedman ZR. Protein glycosylation in diabetes mellitus: a review of laboratory measurements and of their clinical utility. Clin Chim Acta 1983;127:147-84. 5. Benjamin RJ, Sacks DB. Glycated protein update: Implications of recent studies, including the diabetes control and complications trial. Clin Chem 1994;40:683-7. 6. Nathan DM, Singer DE, Hurxthal K, et al. The clinical
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
20.
23. 24.
25.
26.
27.
information value of the glycosylated hemoglobin assay. N Engl J Med 1984;310:341-6. Austin GE, Mullins RH, Morin LG. Nonenzymic glycosylation of individual plasma proteins in normoglycemic and hyperglycemic patients. Clin Chem 1987;33:220-4. Armbruster DA. Fructosamine: structure, analysis, and clinical usefulness {Review}. Clin Chem 1987;33:2153-63. Gebhart S, Wheaton R, Mullins RE, et al. A comparison of home glucose monitoring with determination of HbA1c, total glycoHb, fructosamine, and random serum glucose in diabetic patients. Arch Intern Med 1991;151:1133-7. Schleicher ED, Mayer R, Wagner EM, et al. Is serum fructosamine assay specific for determination of glycated serum protein? Clin Chem 1988;34:320-3. Phillipou G, Seaborn CJ, Phillips PJ. Re-evaluation of the fructosamine reaction. Clin Chem 1988;34:1561-4. Schleicher ED, Vogt BW. Standardization of serum fructosamine assays. Clin Chem 1990;36:136-9. Kallner A. Influence of triglycerides and urate on methods for determination offructosamine. Clin Chim Acta 1992;207: 99-106. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997;20: 118397. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039-57. Shield JP, Poysner K, Hunt L, et al. Fructosamine and glycated haemoglobin in the assessment of long term glycaemic control in diabetes. Arch Dis Child 1994;71:443-5. Furuseth K, Bruusgaard D, Rutle 0, et al. Fructosamine cannot replace HbA1c in the management of type 2 diabetes (NIDDM). Scand J Prim Health Care 1994;12:219-24. Ichikawa H, Nagake Y, Takahashi M, et al. What is the best index of diabetic control in patients with diabetes mellitus on hemodialysis? Nippon Jinzo Gakkai Shi 1996;38: 305-8. Lin MJ, Hoke C, Ettinger B, et al. Technical performance evaluation of BMlHitachi 747-200 serum fructosamine assay. Clin Chem 1996;42:244-8. Meneilly GS, Tessier D. Diabetes in the elderly. Diabet Med 1995;12:949-60. Tahara Y, Shima K. Kinetics of HbA1c, glycated albumin, and fructosamine and analysis of their weight functions against preceding plasma glucose concentration. Diabetes Care 1995;18:440-7. Gunter HH, Ritter C, Reinhardt W, et al. Influence of non-diabetic pregnancy on fructosamine and HbA1 concentration. Z Geburtshilfe Neonatol 1995;199:148-55. Hughes PF, Agarwal M, Newman P, et al. An evaluation of fructosamine estimation in screening for gestational diabetes mellitus. Diabet Med 1995;12:708-712. Uncu G, Ozan H, Cengiz C. The comparison of 50 grams glucose challenge test, HbA1, and fructosamine concentrations in diagnosis of gestational diabetes mellitus. Clin Exp Obstet Gynecol 1995;22:230-4. Tvorogova MG, Zaitseva DTM, Aronskaia EE, et al. Comparison offructosmine, glycosylated hemoglobin and glucose in patients with type II diabetes. Klin Lab Diagn 1995; 1:10-2. Defallah AA, Eskandarani H, Rehaimi A, et al. Fructosamine in HbS and G6PD-deficient Saudi Arabs in the Eastern Province of Saudi Arabia. Br J Biomed Sci 1994;51: 332-5. Islam N, Akhter J, Kayani N, et a1. Fructosamine: an alternative assessment of past glycaemic control in developing countries. J Pak Med Assoc 1993;43:238-40.
323