Site-to-site variation of glucose in interstitial fluid samples and correlation to venous plasma glucose.
A method for sampling interstitial fluid (ISF) from which glucose can be measured has been developed in our laboratory. Studies have suggested that glucose concentrations in ISF samples collected via this method closely reflect ambient glycemia (2) and that there is not a clinical difference between glucose in the interstitial and venous compartments in subjects whose glucose concentrations are rapidly changing (3). These studies concluded that a correlation exists between ISF glucose and venous plasma glucose. Both of these studies collected ISF from the forearm using a 30-gauge needle, which penetrated into the dermal layer to extract ISF. Because we are collecting a discrete ISF sample, one could hypothesize that the glucose concentration may be dependent on the sampling site within the forearm.
To determine whether there is any variability in glucose concentrations between ISF samples collected from different areas of the forearm, a study was conducted with diabetic subjects from whom consecutive ISF samples were collected over a 30-min period. The study was reviewed and approved by the Western Institutional Review Board [R], and informed consent was obtained from all subjects before enrollment. A total of 95 subjects were enrolled in the study. Fingertip capillary blood samples were collected just before and just after the ISF samples (at t = 0 and t = 30 min); a venous blood sample was collected at t = 5 min. The first 50 subjects were removed from the data set because of a sample handling error that led to uncontrolled evaporative losses from the ISF samples. Venous samples were not obtained from 2 of the remaining 45 subjects, leaving 43 subjects for whom data are reported here. A mean of 26 ISF samples were collected from each subject (minimum, 15; maximum, 40). ISF samples of ~1.5 [micro]L were collected into 2-[micro]L heparin-containing capillary tubes (Drummund Scientific). The capillary tubes containing ISF were stored at -70[degrees]C. Duplicate glucose measurements were obtained from each of the fingertip capillary blood samples (Glucometer Elite [R]; Bayer Corporation Diagnostics Division). Venous plasma and ISF glucose concentrations were measured by a modified enzymatic hexokinase method. Modifications made to this single-reagent, manual procedure (kit 16-50; Sigma Diagnostics) were necessary to accommodate the small ISF sample size, the range of volumes collected, and the spectral sensitivity requirements at low glucose concentrations. The procedure described by Sigma requires a 10.0-[micro]L sample volume and a 1.00-mL reagent volume. The procedure was modified to accommodate sample volumes of 0.50-2.00 [micro]L and a reagent volume of 75.0 [micro]L. To account for the variable sample volumes, each ISF sample was weighed using an MT 5 microbalance (Mettler-Toledo), and a gravimetric correction was applied. Absorbances were measured with a Spectronic Genesys 5 Spectrophotometer (Milton Roy). Venous plasma samples were measured in quadruplicate on 1.00-[micro]L samples, and the mean values were used for numerical analysis.
Because the subjects were not required to be in a fasting state, the capillary blood samples collected at the beginning and end of each test served as an indicator of the amount and direction of change in blood glucose over the course of the test. To account for the trending in blood glucose, the SD in ISF glucose for each subject was calculated as the paired SD, which is the square root of the mean of paired variances. Paired variances were calculated by pairing the determined glucose values for consecutive samples. For example, samples 1 and 2 were the first pair and samples 3 and 4 were the second pair. The paired SD calculated in this way is equivalent to the RMSE of a one-way ANOVA on pairs of samples. The paired CV was calculated as the paired SD divided by the mean of ISF glucose values for a subject, taken as a percentage. The range in ISF glucose paired CVs for the 43 subjects was 2.7% to 16%, with a mean of 6.2% and a median of 5.0%. These values compare favorably with the 5.0% and 5.1% paired CVs obtained respectively for the 5.0 and 16.1 mmol/L controls, which were run in duplicate at least twice daily for the modified hexokinase method.
A comparison between glucose values in the venous, capillary, and ISF compartments was made by evaluation of Clarke Error Grid statistics, correlation coefficients, and the standard deviation of residuals ([S.sub.y/x]). The Clarke Error Grid divides a reference vs predicted glucose correlation plot into five zones (A-E) of descending clinical accuracy of glucose estimation (4). Zone A represents values that are clinically accurate, zone B values are clinically benign, and values in the C,D, and E zones are clinically inaccurate and might lead to a clinically inappropriate action by the patient. The correlation coefficient and [S.sub.y/x], give an indication of the linear relationships and overall error of the respective regressions. These comparisons are summarized in Table 1. The venous vs ISF and venous vs capillary comparisons do not adequately take into account the trending in blood glucose because only a single venous sample was collected. The capillary vs ISF comparison better accounts for this trending because both capillary and ISF glucose values are averaged over the 30-min duration of the test. This most likely accounts for the better Clarke Error Grid statistics for capillary vs ISF. The similarities between the paired CVs for the daily controls and the consecutively collected ISF samples support a conclusion that there is minimal variability in glucose concentration in ISF samples collected from the forearm when trending in blood glucose is taken into account. In addition, a comparison of the absolute glucose values shows that there is very good agreement between glucose concentrations in the interstitial, capillary, and venous compartments.
(1.) The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes in the development and progression of long term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977-86.
(2.) Service FJ, O'Brien PC, Wise SD, Ness S, LeBlanc SM. Dermal interstitial glucose as an indicator of ambient glycemia. Diabetes Care 1997;20: 1426-9.
(3.) Bantle JP, Thomas W. Glucose measurement in patients with diabetes mellitus with dermal interstitial fluid. J Lab Clin Med 1997;130:436-41.
(4.) Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care 1987;10:622-8.
Philip Stout, * Kristen Pokela, Debra Mullins-Hirte, Thomas Hoegh, Michael Hilgers, Ann Thorp, Michael Collison, and Tatyana Glushko
(Integ, Inc., 2800 Patton Rd., St. Paul, MN 55113; * author for correspondence: fax 651-639-9042, email email@example.com)
Table 1. Summary of fluid comparisons. Mean venous Mean capillary Mean venous Statistic (a) plasma HK vs blood GE vs mean plasma HK vs mean ISF HK ISF HK capillary blood GE A zone, % 90.7 97.7 88.4 B zone, % 7.0 2.3 9.3 C zone, % 0 0 0 D zone, % 2.3 0 2.3 E zone, % 0 0 0 r 0.973 0.965 0.970 [S.sub.y|x] 1.00 1.17 1.05 mmol/L (a) Zones A-E refer to the zones of the Clarke Error Grid.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Abstracts of Oak Ridge Posters|
|Author:||Stout, Philip; Pokela, Kristen; Mullins-Hirte, Debra; Hoegh, Thomas; Hilgers, Michael; Thorp, Ann; C|
|Date:||Sep 1, 1999|
|Previous Article:||Analytical characterization of electrochemical biosensor test strips for measurement of glucose in low-volume interstitial fluid samples.|
|Next Article:||Immobilization of monolayers of Fc-binding receptors on planar solid supports.|