| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Zentralkrankenhaus Bremen-Nord, 28755 Bremen, Germany
2 Zentralkrankenhaus Sankt-Juergen-Strasse, 28205 Bremen, Germany
aauthor for correspondence: fax 49-421-4973334, e-mail info{at}zkh-bremen-mitte.de
Glucose concentrations are usually measured in whole blood or plasma. Plasma values are influenced by the concentration of proteins, especially those with large volumes, such as lipoproteins. Blood values additionally depend on the total volume of the various blood cells, which is usually expressed as the hematocrit (1)(2).
The interconversion of glucose values for venous and capillary blood is further complicated by the arteriovenous difference. In the fasting state, the glucose concentrations in arterial, capillary, and (forearm) venous blood are supposed to be almost indistinguishable. In contrast, arterial blood glucose values may differ by 20% or as much as 70% in the postprandial state (3)(4). The mean arteriovenous differences are largest in lean nondiabetic individuals, smallest in diabetic individuals, and larger in deep veins than in superficial vessels (1)(5). Other factors can influence the differences in glucose concentrations among the various samples (6)(7)(8)(9). Thus, the conversion of concentration values from one system (or sample type) to another is subject to unpredictable errors.
Several authors have already rejected the practice of converting glucose concentrations and have recommended that plasma be used for all glucose determinations (2)(10)(11). In a recent editorial, glucose measurement in whole blood was considered anachronistic (12), but only whole blood is used by home monitoring and near-patient monitoring devices. Many laboratories measure the glucose concentration in whole blood, especially in capillary whole blood, for therapeutic monitoring and for diagnosing hypo-, normo-, and hyperglycemia. However, the applicability of whole blood for determining glucose intolerance is still a matter of debate. Many practitioners tend to use capillary blood (CB) for diagnostic purposes (13)(14). The decision limits usually applied for whole blood are those recommended by WHO (15)(16)(17) and the American Diabetes Association (18), which are based on epidemiologic studies with venous plasma (VP). In practice, either measured values or decision limits are converted from one sample system to another. The present study was undertaken to reinvestigate the comparability of glucose determinations in venous blood (VB), VP, and CB.
The study group consisted of 147 individuals from outpatient departments (internal medicine and dermatology) who were able to walk to the laboratory for blood collection (age range, 25–76 years). Using values recommended by WHO (15)(16) for the classification of plasma glucose concentrations, we separated the individuals into three groups according to whether they displayed a "healthy" (n = 74), impaired (n = 36), or diabetic glucose tolerance (n = 37). Oral glucose tolerance tests (GTTs) were performed according to WHO recommendations (15)(16). Participants ingested 75 g of glucose as Dextro O.G-T. (Roche Diagnostics).
VB samples were drawn into 2.7-mL monovettes containing lithium heparinate (cat. no. 05.1553; Sarstedt AG). Capillary and VB samples were collected within 2 min of each other by separate medical staff. Plasma was prepared within 10 min of blood sampling.
Glucose concentrations were determined in 500 µL of hemolyzing reagent plus 10 µL of blood or plasma (collected in heparinized end-to-end glass capillaries; cat. no. 19.414; Sarstedt AG) with an EBIO plus 6668 analyzer (Eppendorf AG) using glucose oxidase-containing electrodes (19) within 2 h of sampling. Glucose concentrations in hemolyzed samples were stable for at least 24 h. The results obtained with the glucose analyzer were referred to a glucose solution (11.11 mmol/L; certified primary reference material from NIST). All procedures were subjected to internal and external quality assurance programs. Control materials (Validate A and N) were purchased from Organon Teknika. The mean value (± SD) obtained for Validate A (lot no. 6B403; assigned value, 11.54 mmol/L) over 24 days was 11.55 ± 0.39 mmol/L glucose, and that for Validate N (lot no. 6B401; assigned value, 5.21 mmol/L) was 5.35 ± 0.20 mmol/L (obtained over 28 days).
All calculations were performed with mean values from duplicates. The relationship between concentration ratio (target quantity) and time-specific metabolic and disease state was modeled by a linear mixed-effects model (20) using the mixed procedure of the SAS 6.12 package.
The mean VP/VB ratio from all determinations during the tolerance tests was 1.148 (Table 1
), increasing slightly but statistically not significantly (P = 0.37) from the healthy to the diabetic group (Table 1
). This increase appeared to be constant throughout the GTT. The mean VP/CB ratio was 1.048. The VP/CB ratio was related to the nutritional state, being 1.084 in the fasted and 0.972 in the postprandial state in healthy nondiabetic individuals (P <0.001). In contrast, the VP/CB ratio remained almost constant after a glucose load in diabetic individuals (P = 0.92). The VP/CB ratio was higher in diabetic than in nondiabetic individuals (P <0.001). According to these results, the WHO recommendation for the 2-h postload CB cutoff should be reduced from 11.1 to 10.0 mmol/L. The clinical consequence would be the detection of more diabetic individuals. The CB/VB ratio (Table 1
) increased during the GTT only in normotolerant individuals (P <0.001) and was lower in the diabetic than in the nondiabetic group (P <0.001). Individual ratios varied considerably. The ranges were 0.74–1.66 for the VP/VB ratio and 0.52–1.63 for the VP/CB ratio (Table 1
).
|
Glucose concentrations are usually converted from one sample system to another by use of fixed factors. These factors are either derived from the water content of the different compartments (water distribution theory) or from the glucose concentrations determined with analytical procedures (ratio of mean values or equation of regression lines).
The considerable variation in the conversion factors was demonstrated by the range of conversion factors (Table 1
). A large variation in the percentage difference between blood and plasma glucose concentrations has already been reported by others (2)(21)(22)(23). Mean VP/VB ratios reported in the literature (2)(10)(21)(22)(23)(24)(25) also vary from 1.04 (24) to 1.183 (2). The mean ratio in this study was 1.148. The even greater variation of the VP/CB ratio is probably attributable to the arteriovenous difference. The interindividual VP/CB ratio was related to the nutritional state, confirming an earlier report by Larsson-Cohn (26). The mean ratio from the entire GTT was 1.09-fold higher in diabetics than in healthy individuals (Table 1
). A higher VP/CP ratio has also been observed in gestational diabetes (27) and was explained by a decrease in the arteriovenous difference. Lind et al. (21) found that the differences between these two sample systems in the fasting state were too trivial to be worth correcting for healthy and pregnant women. However, in diabetics, the VP/CB ratio cannot be neglected.
In conclusion, conversion of glucose concentrations determined in different sample systems by use of factors is an oversimplification and probably leads to unpredictable rates of discordant disease classifications. These problems are becoming more relevant with the widespread use of point-of-care testing instruments, including blood gas analyzers with integrated glucose sensors that measure glucose in the plasma water fraction. The only solution for this dilemma is to use only one sample system. The experimental data clearly indicate that the use of plasma should be preferred to diagnose glucose intolerance, including diabetes. The logistic disadvantages are the centrifugation step and the prevention of glycolysis. In the present study, in vitro glycolysis could be neglected. In cases in which samples may need 2–4 h until processing can be started in the laboratory, unpredictable glycolysis will occur, even in the presence of fluoride (28)(29). Chan et al. (30) showed that delays in processing blood specimens in hospital practice may lead to misclassification in up to 7% of GTTs. Stahl et al. (31) proposed storage on ice for not more than 1 h until centrifugation. However, this recommendation may not be acceptable for many hospitals. The use of capillary hemolysate together with a reduced decision limit thus may be a second choice for the detection of diabetes.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  K. A Hatonen, M. E Simila, J. R Virtamo, J. G Eriksson, M.-L. Hannila, H. K Sinkko, J. E Sundvall, H. M Mykkanen, and L. M Valsta Methodologic considerations in the measurement of glycemic index: glycemic response to rye bread, oatmeal porridge, and mashed potato. Am. J. Clinical Nutrition, November 1, 2006; 84(5): 1055 - 1061. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
