| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Endocrinology and Metabolism |
1 University of Geneva, School of Pharmacy, Geneva, Switzerland. 2 Centre Interuniversitaire de Recherche et d’Enseignement, "Pharmapeptides", Campus Universitaire, Archamps, France.
aAddress correspondence to this author at: University of Geneva, School of Pharmacy, Quai Ernest-Ansermet 30, CH-1211 Geneva, Switzerland. Fax 33-450-95-28-32; e-mail Begonia.Delgado{at}pharm.unige.ch.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Methods: The simultaneous extraction of glucose and sodium by reverse iontophoresis was performed on human volunteers over 5 h, and blood glucose was measured in the conventional manner at each collection interval. These data were used for each volunteer to calculate an extraction constant (K), which equals the ratio of the extracted fluxes (JGlucose/JNa+) normalized by the corresponding ratio of the concentrations in the blood ([Glucose]/[Na+]). The values of K were compared between and within volunteers.
Results: The iontophoretically extracted glucose flux reflected the glucose concentration profiles in the blood, and sodium extraction remained essentially constant, consistent with the fact that its systemic concentration does not vary significantly. A constant value of K was established for two thirds of the study population. However, the efficiency of glucose extraction varied seasonally, whereas the reverse iontophoresis of Na+ did not; i.e., variation in K became apparent.
Conclusions: Use of the sodium ion as an internal standard could refine the determination of glycemia by reverse iontophoresis without requiring calibration with a blood sample.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
One such approach is iontophoresis, which uses a small electric current to drive charged and highly polar compounds across the skin at rates very much greater than their passive permeabilities. Two major transport mechanisms are involved: electromigration and electroosmosis. Electromigration is the movement of small ions across the skin under the direct influence of an electric field. Electron fluxes are transformed into ionic fluxes by the electrode reactions, and ionic transport proceeds through the skin to maintain electroneutrality. The total charge transported depends on the strength of the electric field and the duration of application. Iontophoresis sets in motion several ions across the skin, and all of them compete to carry a fraction of the current. The contribution of each ion to charge transport is called the transport number, the sum of which equals 1. According to Faraday’s law, the flux of each ion in the iontophoretic circuit is given by:
 | (1) |
Electroosmosis is the principal transport mechanism of uncharged molecules and of high-molecular-weight cations. The skin is negatively charged at physiologic pH and acts, therefore, as a permselective membrane to cations. This preferential passage of counterions induces an electroosmotic solvent flow that may carry neutral molecules in the anode-to-cathode direction. The volume flow, JV [volume/(time · area)] is predicted (3) to be proportional to the potential gradient (–d
/dx) established by the electric field:
 | (2) |
 | (3) |
In clinical chemistry, iontophoresis has already been established as a tool used in the diagnosis of cystic fibrosis (in this setting, pilocarpine is administered to test the secretional function of the sweat glands) (4). The symmetry of iontophoresis renders it useful not only for the delivery of drugs, but also for the extraction of endogenous substances of clinical interest and, in particular, glucose (5).
This latter application has been investigated in depth and has led to the development of the GlucoWatch® Biographer (Cygnus Inc.) (6)(7)(8)(9)(10). The wrist-worn device monitors glucose continuously for up to 13 h, recording six glucose readings per hour. However, the device needs to be calibrated against blood glucose from the fingertip to correlate the extracted glucose amounts with subdermal concentrations. This essential step has been perceived as a disadvantage despite the fact that the GlucoWatch provides tremendously more information to the individual with diabetes than (the typical) one or two fingersticks per day. A completely noninvasive calibration approach would therefore be beneficial and would open the way to other applications of the reverse iontophoresis technology.
The concept addressed here is that of an internal standard (11). Because iontophoresis is nonspecific, many ions and small, uncharged species (in addition to the analyte of interest) are moved across the skin by the applied electric field. Instead of detecting uniquely the single target substance and calibrating its transdermal measurement by a blood assay, the idea is to monitor the extraction of two species simultaneously: the compound of interest, the temporal change in concentration of which is of clinical importance (i.e., glucose), and a second analyte, the physiologic concentration of which is known and essentially fixed. If the iontophoretic transport of the analyte (A) and the latter, "internal standard" (IS), are independent of one another, then their fluxes (J) out of the skin should obey the relationship:
 | (4) |
is obeyed even when the subdermal Na+ concentration is allowed to vary over its maximum physiologic range (125–145 mmol/L). The aim of the research described here is to test the internal standard hypothesis in vivo. The following questions were addressed: (a) what is the flux ratio Jglc/JNa+ in vivo, (b) can a common proportionality constant K be established for a subject population, (c) how accurate is the prediction of blood glucose when using the internal standard concept, and (d) what is the significance of inter- and intraindividual variability with respect to iontophoretic extraction? In addition, the potential utility of potassium as another cationic internal standard was examined.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
chemicals
Tris base, NaCl, KCl, D-glucose, HCl, NaOH, and methanesulfonic acid were purchased from Sigma-Aldrich Co. and were at least of analytical grade. Deionized water (resistivity >18.2 M
/cm2) was used to prepare all solutions.
iontophoresis
Two cylindrical glass cells (diameter, 1.6 cm; extraction surface, 2 cm2) were fixed with foam tape (3M; Health Care) on each volunteer’s ventral forearm with a distance of 7 cm between them. The anodal chamber was filled with 1.2 mL of 10 mmol/L Tris buffer (pH 8.5) containing 100 mmol/L NaCl; the cathodal chamber contained the same volume of 10 mmol/L Tris buffer alone. Custom-made Ag/AgCl electrodes were inserted into the solutions and fixed 3–4 mm above the skin surface to ensure that no physical contact with the skin occurred. Direct current (I = 0.6 mA; current density = 0.3 mA/cm2) was passed for a total of 5 h and was controlled by a Phoresor II Auto (Iomed), a US Food and Drug Administration-approved constant current, iontophoretic power supply. Every 15 min after initiation of the current, the entire cathodal solution was collected and replaced by 1.2 mL of fresh buffer. The samples were immediately frozen until analysis.
After 2.5 h of iontophoresis, the volunteers ingested either a meal rich in carbohydrates or 75 g of glucose dissolved in 300 mL of water (Glucosum monohydricum, Pharmacopoeia Europea; Hänserler AG), so as to provoke a significant change in blood sugar. From this point onward, glycemia was measured before each subsequent 15-min collection interval by use of a conventional blood glucose monitor (Glucotrend 2; Roche Diagnostics). The "within-run" and "day-to-day" imprecision of the glucose monitor was 
3%, and regression analysis of a comparison with an automated hexokinase reference method was: y (mmol/L) = 0.98x + 0.08 mmol/L (information from the supplier).
analyses
Analytes were assayed separately by HPLC on an Ion Chromatograph 600 system (Dionex). Glucose was quantified by anion separation with pulsed amperometric detection on a gold electrode; Na+ and K+ were quantified by cation separation with suppressed conductivity detection. Calibration was performed with at least six external calibrators in each chromatographic run, covering linear ranges (r2 
0.999) of 0–30 µmol/L for glucose, 0–5 mmol/L for Na+, and 0–1 mmol/L for K+. The limits of quantification (defined as a signal-to-noise ratio of 10) for glucose, Na+, and K+ were 0.15, 1.4, and 2.5 µmol/L, respectively. The within-run imprecision (CV) of injection was 2.1% for a mean concentration of the analytes.
data analysis and statistics
Iontophoretic fluxes were calculated from the amounts extracted in each collection interval and plotted at the mid-time point of each interval; blood glucose concentrations were at the actual time of measurement.
Data are expressed as the mean (SD). Mean Na+ fluxes in each experiment were determined from 12 separate points. Electroosmotic volume flows were determined by normalizing the iontophoretic flux values of glucose by the corresponding blood concentrations; according to Eq. 3
. Statistical differences were assessed by two-tailed Student t-test and ANOVA, followed by a Newman–Keuls multiple comparison test, calculated with GraphPad Prism 3.02 software. Individual values of K were determined by linear regression (extracted flux ratio vs blood concentration ratio), the significance of which was tested by ANOVA; goodness-of-fit is given by r2. A common value of K was computed by pooled regression based on six experiments from different volunteers using analysis of covariance as described by Zar (12). The accuracy of prediction was tested by plotting the predicted glucose values against the measured blood glucose in a Clark Error Grid (13).
|
Results and Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
glucose tracking
The iontophoretic transport of glucose, an uncharged, polar molecule, occurs by electroosmosis and is directly proportional to the subdermal concentration (3). However, a "warm-up" period is necessary to establish a pseudo-steady-state electroosmotic flow and to empty the glucose reservoir from the skin. The latter is attributable to local metabolism and is not reflective of glucose concentrations in the blood. The recommended warm-up period for the GlucoWatch G2 is 2 h, a period similarly adopted in this study. Subsequently, extraction every 15 min over the next 3 h allowed blood glucose tracking in 8 of the 12 volunteers studied. In these volunteers, electroosmotic flow was >5 µL/h (Table 1
). Fig. 1
shows the glucose extraction profiles for three volunteers who ingested on separate occasions either a carbohydrate-rich meal or 75 g of a standard glucose load at 150 min. The glucose reverse iontophoretic extraction fluxes accurately followed the systemic glucose concentration. A time delay between the extraction rate and the blood concentration was apparent in those volunteers receiving the oral glucose load. Rapid absorption of glucose occurred with a higher peak value obtained (14 vs 10 mmol/L) relative to those volunteers who received a carbohydrate-rich meal. In contrast, ingestion of the carbohydrate-rich meal led to more gradual glucose absorption from the gastrointestinal tract, such that differences between the plasma kinetics and the rate of extraction were blurred.
|
|
It is acknowledged that reverse iontophoresis samples the interstitial fluid (ISF) and that real differences exist between glucose kinetics in this compartment and those in the blood; this divergence is clearly relevant to the development of continuous glucose monitoring devices and the site(s) at which glucose is determined. For example, Aussedat at el. (15) have described a delayed response between increasing glucose concentrations in the interstitium compared with those in the blood in rats. On the other hand, a faster and more pronounced decrease in glucose concentrations in the interstitium was observed when blood sugar was lowered. This physiologic lag (estimated to be somewhat less than 4 min) was similarly observed in humans using the GlucoWatch (16). The possibility that insulin increases skin blood flow has also been reported (17), and this, in turn, might influence glucose extraction by reverse iontophoresis.
The warm-up period of 2 h for iontophoresis was sufficient to establish pseudo-steady-state electroosmosis. However, in volunteer 6 (see Fig. 1
), an additional 45 min was necessary before glucose fluxes stabilized; from this point, good tracking of glycemia was observed.
Accurate glucose measurements by iontophoresis required efficient extraction. Only moderate or poor (r2 0.53) correlation with plasma glucose was found when the normalized glucose fluxes were <5 µL/h (volunteers 6–9; Table 1
); above this threshold, however, good to excellent (r2 0.80) correlations were obtained (volunteers 1–5 and 10; Table 1
).
cation extraction
Reverse iontophoretic extraction fluxes of sodium are shown in Table 1
for all volunteers. These values stabilized 30 min after the initiation of iontophoresis (not shown) and remained essentially constant throughout the experiments even when significant changes in glycemia were occurring (see Fig. 1
). The calculated mean (SD) transport number of sodium in these experiments was 0.55 (0.04). This is consistent with the in vitro results (11) and demonstrates that Na+ is the principal charge carrier across the skin. Given that Na+ is present in extracellular fluids at concentrations 30–50 times higher than those of other potential charge carriers, such as K+, Ca2+, or Mg2+, this result is not surprising. At this high concentration, Na+ iontophoresis is effectively independent of its subdermal concentration over the physiologic range of 125–145 mmol/L and is unlikely to be affected by changes in the plasma concentrations of the other cations. Thus, alterations in the systemic Na+ concentration during hypo- or hyperglycemia were not anticipated to affect the usefulness of Na+ as an internal standard. The results from this study support this assumption in that reverse iontophoretic extraction of Na+ was remarkably constant within and between individuals and was not significantly altered even when large excursions in glycemia occurred (see Fig. 1
).
Potassium fluxes varied among individuals and within the same experiment (data not shown), being higher at the beginning of iontophoresis before gradually decreasing after 2 h of iontophoresis to values between 0.7 and 3.9 µmol/h; the calculated mean (SD) transport number was 0.07 (0.04), corresponding to a CV more than 10 times greater than that for sodium transport. Interestingly, the highest potassium fluxes were found in individuals with the lowest electroosmotic flux, an observation that needs to be confirmed in a larger population and for which (if real) the reason is unclear. The higher variability and lower transport number of potassium precluded its use for the calibration of glucose extraction, and no further work was pursued.
common extraction constant (K)
The ratio of subdermal concentrations of glucose and Na+ (assumed to be 133 mmol/L) were calculated and plotted against the corresponding ratio of extracted fluxes (Fig. 2
). There was no systematic time delay between the blood concentration and extraction rate profiles. Linear regression (see Eq. 4
) for each individual yielded an individual value of K, given in Table 1
. These data are from six volunteers, for whom the normalized glucose flux was 8.5 µL/h, and were used to determine a mean value of K. Analysis of covariance showed no significant differences in the individual K values. Pooled ANOVA yielded a common slope, i.e., the common mean extraction constant, namely, 0.12 (SD, 0.018).
|
prediction of blood glucose
The mean K can be used to estimate blood glucose from the iontophoretic extraction flux data by rearrangement of Eq. 4
:
 | (5) |
8.5 µL/h, were analyzed in this way, and the estimated blood glucose concentrations were then compared with glycemia measured from the fingertip. The results, plotted as a Clark Error Grid, are shown in Fig. 3
. Of a total of 181 data points, 142 (78.5%) fell in region A, and 39 (21.5%) fell in region B; that is, all results were in the clinical acceptable region A + B. It must be accepted, however, that the range of glycemia covered by these experiments was limited: 4–13 mmol/L. Additional work is necessary to determine whether the predictions remain robust at more extreme degrees of hypo- and hyperglycemia. Linear correlation through all the data points in Fig. 3
yielded a slope (SE) of 0.96 (0.05) and an intercept (SE) of 0.10 (0.36) [SD of the residuals = 1.125; SE of the residuals = 0.083 (n = 182). The population of the residuals also passed a normality test (P = 0.05)]. The correlation coefficient was 0.80. For this analysis, no attempt was made to temporally adjust the extraction profile to better overlap the corresponding blood glucose values, even when an apparent time lag could be clearly discerned (e.g., Fig. 1B
). Although such a manipulation would have improved the goodness of fit for the determination of individual K values (and hence yielded a better blood sugar prediction for certain data points), a significant change in the absolute result is unlikely. This is because increasing glucose concentrations in the blood typically precede increases in the ISF concentrations, whereas decreasing blood concentrations lag behind those in the ISF (15). The objective here, in any case, was to analyze the results from the perspective of practical glucose monitoring, for which an individual’s actual lag between blood and ISF concentration profiles is generally unknown.
|
inter- and intraindividual variability
Glucose fluxes differed significantly in volunteers 6–9 compared with the rest of the population (see Table 1
). Electroosmotic glucose transport was approximately an order of magnitude lower than that seen in the other volunteers. In contrast, essentially no differences in sodium extraction were found, with all values falling in the range of 10.7–13.3 µmol/h. This contrast is shown in Fig. 4
. As a result, the K values for volunteers 6–9 were clearly smaller (Table 1
), and the use of the previously discussed common K value would not yield accurate predictions of glycemia in these individuals.
|
It might then be argued at this point that instead of attempting to use a common K for all individuals, K be determined for each individual in a separate calibration experiment. However, such an approach also demands that the within-subject variability as a function of time is reasonable. To test this hypothesis, we performed monthly experiments on volunteer 1 over the course of a calendar year. The results for the normalized glucose flux and the iontophoretic Na+ extraction kinetics are shown in Fig. 5
. Whereas Na+ transport remained constant, significant decreases in electroosmosis were observed in the winter months. Additional data were also acquired in other volunteers (see Table 2
) and revealed no clear pattern. Volunteers 2 and 10 behaved similarly to volunteer 1, whereas two other volunteers (3 and 4) showed no seasonal effect; volunteer 6, in contrast, manifested a lower electroosmotic flow in the summer. In only two volunteers (4 and 10) were Na+ fluxes significantly different, although the disparities would not have practical importance. Taken together, therefore, it appears that even an individual calibration for each individual might not permit an accurate prediction of glycemia at all times by the internal standard approach.
|
|
Although the fundamental reason for the variability in glucose extraction cannot be unequivocally deduced from this work, it can be concluded that a more appropriate internal standard will probably be another neutral substance that is transported by the same electroosmotic mechanism. In this way, any effects that change the charge on the skin (and/or its permselectivity) will similarly alter the extraction of glucose and the internal standard. The sodium ion, on the other hand, as the major charge carrier across the skin is not very sensitive to relatively subtle differences in skin charge. As such, it is a less than ideal internal standard for glucose. On the other hand, recent work has demonstrated the very appropriate use of Na+ as an internal standard for the use of reverse iontophoresis as a noninvasive tool in the therapeutic monitoring of lithium. In this case, the normalization of lithium extraction flux with that of Na+ gave improved prediction of the serum concentration of lithium in bipolar patients in vivo (18).
comparison with in vitro data
The mean K measured in this work is somewhat higher than that measured in vitro with porcine skin (11); in that study, a mean electroosmotic flow of 5.1 µL/h together with a mean sodium flux of 8.0 µmol/h gave a constant of 0.07 (11). At least two factors may contribute to the better extraction in vivo: First, the electrode formulations were modified to maximize as far as possible the electroosmotic flow (e.g., pH 8.5 was used), and second, the presence of a functioning microcirculation provides for more facile access to the subdermal compartment. Particularly interesting, however, is that the marked variability in glucose extraction in vivo was not observed in vitro. At pH 7.4, electroosmosis varied no more than 30%, and a seasonal difference was not observed. A possible explanation may lie in the physiology of the skin appendages (e.g., the sweat glands), which have been recognized as important transport pathways in iontophoresis (19). Interindividual differences in appendageal morphology and effects of climate on function have been reported (20)(21) and may account for the greater variation observed in vivo as a function of ambient conditions. Additional work is clearly needed to better understand these observations.
In summary, this study demonstrates the development of the reverse iontophoresis approach to monitor glucose noninvasively without calibration with a blood sample. The internal standard hypothesis, using the Na+ ion as the invariant endogenous calibrator, was confirmed for a significant subset of the study population. However, Na+, being extracted by a different mechanism than glucose, did not reflect the entire range of variability in glucose extraction.
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  F. M. Musteata, M. L. Musteata, and J. Pawliszyn Fast In Vivo Microextraction: A New Tool for Clinical Analysis Clin. Chem., April 1, 2006; 52(4): 708 - 715. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, blood glucose concentration; 
, glucose flux; x, Na+ flux.
) to reverse iontophoretic sodium transport flux (in µmol/h; 
).