Microdialysis-Coupled Enzymatic Microreactor for in Vivo Glucose

Nov 7, 2013 - Byeong-Ui Moon†‡, Martin G. de Vries†§, Carlos A. Cordeiro§, Ben H. C. Westerink†§, and Elisabeth Verpoorte*‡. † Biomonit...
0 downloads 0 Views 330KB Size
Article pubs.acs.org/ac

Microdialysis-Coupled Enzymatic Microreactor for in Vivo Glucose Monitoring in Rats Byeong-Ui Moon,†,‡ Martin G. de Vries,†,§ Carlos A. Cordeiro,§ Ben H. C. Westerink,†,§ and Elisabeth Verpoorte*,‡ †

Biomonitoring and Sensoring, Groningen Research Institute of Pharmacy, University of Groningen, Antonius Deusinglaan 1, P.O. Box 196, 9700 AD Groningen, The Netherlands ‡ Pharmaceutical Analysis, Groningen Research Institute of Pharmacy, University of Groningen, Antonius Deusinglaan 1, P.O. Box 196, 9700 AD Groningen, The Netherlands § Brains On-Line B.V., P.O. Box 4030, 9701 EA Groningen, The Netherlands ABSTRACT: Continuous glucose monitoring (CGM) is an important aid for diabetic patients to optimize glycemic control and to prevent long-term complications. However, current CGM devices need further miniaturization and improved functional performance. We have coupled a previously described microfluidic chip with enzymatic microreactor (EMR) to a microdialysis probe and evaluated the performance of this system for monitoring subcutaneous glucose concentration in rats. Nanoliter volumes of microdialysis sample are efficiently reacted with continuously supplied glucose oxidase (GOx) solution in the EMR. The hydrogen peroxide produced is amperometrically detected at a (polypyrrole (PPy)-protected) thin-film Pt electrode. Subcutaneous glucose concentration was continuously monitored in anesthetized rats in response to intravenous injections of 20% glucose (w/v), 5 U/kg insulin, or saline as a control. In vitro evaluation showed a linear range of 2.1−20.6 mM and a sensitivity of 7.8 ± 1.0 nA/mM (n = 6). The physical lag time between microdialysis and the analytical signal was approximately 18 min. The baseline concentration of blood glucose was 10.2 ± 2.3 mM. After administering glucose to the rats, glucose levels increased by about 2 mM to 12.1 ± 2.3 mM in blood and 11.9 ± 1.5 mM in subcutaneous interstitial fluid (ISF). After insulin administration, glucose levels decreased by about 8 mM relative to baseline to 2.1 ± 0.6 mM in blood and 2.1 ± 0.9 mM in ISF. A microfluidic device with integrated chaotic mixer and EMR has been successfully combined with subcutaneous microdialysis to continuously monitor glucose in rats. This proof-of-principle demonstrates the feasibility of improved miniaturization in CGM based on microfluidics.

A

hypoglycemia and, when integrated into an artificial pancreas, to direct automatic adjustment of insulin dosing.7 In vivo microdialysis is a powerful technique for continuous monitoring, in which glucose is sampled from the body via a subcutaneous probe that consists of a semipermeable hollow fiber.8 The microdialysis probe is designed to mimic a blood capillary. The lumen of the probe is perfused with an artificial interstitial fluid (ISF), and glucose diffuses from a higher in vivo concentration across the membrane into the probe where glucose concentration is lower. The membrane surface is biocompatible and safe for the patient. The low molecular weight cutoff of the membrane does not allow large molecules, such as proteins or proteases, to get into the emerging microdialysis fluid. The detection electrodes, located in an external unit coupled to the probe, are therefore exposed to a relatively clean sample solution, which makes them less prone

utomated registration of glucose values with a continuous glucose monitor (CGM) is a relatively new approach to assist patients with diabetes mellitus to maintain blood glucose concentrations within near-normal levels.1−3 The manifestation of long-term diabetic complications such as retinopathy, nephropathy, and neuropathy can be prevented by maintaining tight glycemic control.4 Iatrogenic hypoglycemia should also be avoided, as it is the main limiting factor for good glycemic management.5 Continuous monitoring provides much better insight into the changes of glucose concentration during the day compared to intermittent glucose measurements by the finger-stick method or assessment of glycated hemoglobin. Patient compliance with the finger-stick method is usually limited by its painful and tedious nature,5 especially in young children, elderly, or patients with disabilities.6 CGMs are particularly suited to register episodes of hypoglycemia during the night, when finger-pricks are infrequent and the risk for (severe) hypoglycemia is increased. In addition, CGMs can also be used in prospective mode to warn a patient of progressing © 2013 American Chemical Society

Received: August 1, 2013 Accepted: October 14, 2013 Published: November 7, 2013 10949

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry



to biofouling and more accurate than implantable needle-type biosensors.9 However, a microdialysis-based CGM can be rather bulky because of the required perfusate reservoir, delivery pump, and tubing. Promising results for the use of various CGMs in patients with diabetes have been reported.10−13 While most reported CGMs are based on invasive approaches for glucose analysis, noninvasive monitoring has also been reported.13 A prototype of one particular CGM based on microdialysis, the Subcutaneous Continuous Glucose Monitoring System, has been shown to produce reliable results in humans over a period of several days.14 This system is unusual in that the analytical signal is based on the reaction of microdialysis sample with continuously supplied glucose oxidase (GOx) solution rather than GOx immobilized on a detection electrode. However, besides being bulky, the physical lag time of the system is long (over 30 min). Therefore, miniaturization of this system is highly desirable, also to reduce the consumption rate of chemical reagents, shorten reaction times, and increase costeffectiveness. This might be accomplished with chip-based microfluidic technology. Chips with micrometer-sized fluidic channels enable efficient mixing of GOx solution and microdialysis sample in nanoliter volumes compatible with the nL-to-low-μL flow rates characteristic of microdialysis.15,16 Thanks to soft lithography, the replication of microchannel networks in elastomeric materials, the field of microfluidics combined with microdialysis has made significant advances in the past decade.17−19 On-chip mixing of GOx into nL plugs of microdialysate has been demonstrated by Wang et al. in a segmented flow device coupled to microdialysis, for example.19 Complete CGM systems are made possible on a lab-on-a-chip by integrating control modules (e.g., valves, pumps, fluidic channels, and reservoirs) and monitoring modules (e.g., sensors to measure flow rate or glucose concentration).20,21 An integrated on-chip microdialysis system with in-line sensing electrodes has been successfully demonstrated.22 In our laboratory, we recently developed an efficient enzymatic microreactor (EMR) with a chaotic mixing channel, which we combined with an on-chip electrochemical detector.23 This detector comprised a thin-film Pt microelectrode and a Ag/AgCl wire reference electrode. A major advantage of this device is that fresh GOx solution is continuously delivered into the microreactor. This system enables the more effective conversion of glucose into hydrogen peroxide and is less prone to signal drift compared to CGM systems utilizing subcutaneously implanted immobilizedenzyme electrodes for glucose detection.23,24 In the present study, we evaluated the performance of our microdialysis-coupled EMR to test its feasibility as a CGM. First, we optimized the in vitro probe recovery by selecting the optimal length of the membrane. Next, we tested in vitro the linear range, sensitivity, and physical lag time of our CGM and the specificity of the polypyrrole (PPy)-protected electrodes against faradaic interference from ascorbic acid, uric acid, and acetaminophen. Finally, we investigated the performance in vivo by monitoring subcutaneous glucose in anesthetized, healthy rats. Interstitial glucose monitored by the subcutaneously implanted CGM was experimentally modulated by intravenous administration of glucose or insulin and compared to concurrent blood glucose by frequent sampling from the jugular vein and analysis using a commercially available colorimetric method. Signal stability, the amplitudes of the changes, and response lag times were evaluated.

Article

EXPERIMENTAL SECTION

Chemicals and Reagents. All chemicals were analyticalreagent grade. Ascorbic acid was obtained from Merck (Germany). Uric acid and acetaminophen were purchased from Sigma-Aldrich (Germany). Pyrrole was also purchased from Sigma-Aldrich and used to prepare a 200 mM solution in PBS solution. The PBS was prepared in-house by mixing Na2HPO4 and NaH2PO4 in water to final concentrations of 0.04 and 0.01 M, respectively. NaCl was added to this buffer to a final concentration of 0.10 M. Artificial ISF, also prepared inhouse, was used as the microdialysis infusion solution and had a composition of 147.0 mM NaCl, 3.0 mM KCl, 1.2 mM MgCl2, and 1.2 mM CaCl2. Saline solution (0.9 %, sterile) was purchased from Baxter and stored at 2 °C. Insulin (recombinant human insulin, Humulin, 100 IU/mL) was purchased from Eli Lilly (USA) and stored at 2 °C. D-Glucose and GOx were supplied by Merck (Germany) and used to prepare 1 M and 5000 U/mL stock solutions in water, respectively. D-Glucose stock was stored in a refrigerator at 2 °C, while GOx stock was stored at −20 °C. All solutions were prepared with 18 MΩ·cm ultrapure water purified in an Arium 611 system (Sartorius Stedim Biotech, Germany). Fabrication of the Microfluidic Chip. The microfluidic chip with microchannels in polydimethylsiloxane (PDMS) was replicated from a structured silicon wafer as a master. Briefly, a 4 in., 525 μm-thick, p-type, polished (100) silicon wafer (SiMat, Germany) was employed as a substrate. After wafer cleaning, it was treated with hexamethyldisilazane (HMDS) (Sigma-Aldrich, Germany) to improve adhesion of the photoresist. The structure on the silicon master was processed in a thick photoresist layer with two steps of standard photolithography. A thick layer of positive Photoresist, AZ 4562 (Microchemicals GmbH, Germany), was coated onto the silicon wafer using a spin coater and then rehydrated at ambient temperature for 3 h. The coated wafer was exposed to ultraviolet (UV) light (365 nm, 10 mW/cm2) using a photomask made on a transparency sheet (resolution 3810 dpi; Pro-Art B.V., The Netherlands) to structure the ridges which are the negative of the microchannels. The exposed photoresist was then removed by dipping in a developer solution and rinsed in deionized water. Subsequently, the substrate was again exposed to UV light for 10 s using a photomask to structure the grooves on top of the microchannel-ridges and developed. For microchannel replication in PDMS, a mixture of PDMS resin and curing agent (Sylgard 184, Dow Corning, USA) was prepared and cast onto the silicon master. The mixture was allowed to cure at 50 °C for 4 h. Once hardened, the PDMS slab was peeled off from the silicon master. Microfabricated thin-film platinum (Pt) electrodes were obtained from microLIQUID (Arrasate, Mondragón, Spain). These electrodes were formed on a 4 in. Pyrex glass wafer using a standard photolithographic and lift-off process. Groups of three electrodes were prepared on one wafer (each electrode had an active area of 100 μm × 1500 μm). The glass wafer was then diced into individual glass chips, each with several electrodes. The PDMS slab was cut into individual devices for bonding to the glass chip with electrodes. After UV−ozone treatment, the PDMS slab and glass chip with electrodes were immediately aligned under a microscope and brought into contact with each other to form an irreversible bond. The entire bonding procedure was done in a cleanroom. 10950

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry

Article

Electrode Coating. In order to prevent protein adsorption and reaction with interfering substances, the thin-film Pt electrodes were coated by electrodeposition of PPy, which was subsequently made nonconductive by overoxidation, as described below.25−27 A solution of 200 mM PPy was prepared in 100 mM PBS (pH 7.4, composition as above) solution and used for electrode coating. The PPy solution was first bubbled for 20 min with nitrogen (N2) gas in order to lower the oxygen concentration in solution. The electrodes were then coated by introducing the PPy solution into the microfluidic channel and applying a constant potential of 0.85 V for 5 min. Subsequently, to make the layer of PPy nonconductive, it was overoxidized by maintaining a potential of 0.85 V for 6 h in PBS solution. The chip was then cleaned with ultrapure water and ready for use. Optimization of in Vitro Microdialysis Recovery. In the first experiment, we optimized the flow rate of the perfusion fluid through the microdialysis probe and the length of the semipermeable membrane in order to obtain nearly 100 % probe recovery. Concentric I-shaped microdialysis probes (Brainlink B.V., The Netherlands) with a polyacrylonitrile membrane (45−50 kDa cutoff, 250 μm ID, 340 μm OD) of 1 or 3 cm in length were perfused with artificial ISF at flow rates varying from 0.3 to 20.0 μL/min. The probes were submerged in a stirred, 10 mM solution of glucose in ultrapure water maintained at 37 °C. Microdialysate samples were collected at 30 min intervals, mixed with quinoneimine-based glucose reagent (CMA, Sweden), and colorimetrically analyzed for glucose at a wavelength of 546 nm on a microplate reader (Molecular Devices, U.S.A.). Relative recovery was expressed as percent of beaker glucose content. Evaluation of CGM Performance in Vitro. In the second experiment, we coupled the microdialysis probe to the EMR (see Figure 1) and evaluated the performance of this CGM in

on the electrodes was sufficient to limit faradaic interference from electroactive species common in blood and ISF.11 We therefore compared the analytical current obtained with 5.6 mM glucose to that obtained with this glucose concentration together with 150 μM ascorbic acid, 500 μM uric acid, or 150 μM acetaminophen. The electrodes of the EMR were either bare or covered with a protective layer of overoxidized PPy. The interference was expressed as percent of the glucose content. The effectiveness of the protecting PPy layer was evaluated by paired t tests for each of the analytes separately. P < 0.05 was considered significant. Evaluation of CGM Performance in Vivo. In the final experiment, we evaluated the performance of our CGM in vivo by using it to monitor ISF glucose concentrations in subcutaneous adipose tissue of anesthetized, nondiabetic rats. This study was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Groningen. Male Wistar rats (350−410 g, Harlan, The Netherlands) were singly housed in Plexiglas cages (30 × 30 × 40 cm) under light controlled (12:12 h light/dark cycle, lights on at 07:00 h) and temperature controlled (21 ± 1 °C) conditions. The rats had free access to fresh water and standard rodent chow and were acclimated for one week before surgery and the start of the experiment. On the day of the experiment, the rats were anesthetized with isoflurane gas (5% isoflurane in oxygen for induction, followed by 2% isoflurane for maintenance). A silicone catheter (0.50 mm ID, 0.90 mm OD, 10 cm length) fitted with a 21gauge stainless steel connector (0.80 mm OD, 40 mm length) was placed into the right jugular vein and exteriorized from the neck. The catheter was kept patent by infusing 30 μL of heparin (20 μL/mL in saline) every 15 min. A second incision was made in the dorsal part of the thorax, and a microdialysis probe (1.0 cm membrane length) was placed in a deep subcutaneous tissue layer. Both wounds were closed with suture. Monitoring of ISF glucose in the rats started immediately after surgery. The animals were kept under anesthesia; body temperature was maintained at 37 °C, and dehydration was prevented by hourly subcutaneous injections of saline at a site distant from the microdialysis probe. The probe was connected to the EMR by silica tubing (0.005 in. ID, 0.02 in. OD, 10 cm length) and perfused with artificial ISF at a flow rate of 0.5 μL/min. GOx was continuously delivered to the EMR via silica tubing (30 cm length). The microdialysis probe and EMR were allowed to stabilize for 90 min, followed by recording of a stable baseline for 1 h. Glucose concentrations were experimentally modulated by consecutive intravenous injections of saline (1 mL, control), 20% (w/v) glucose, and 5 U/kg insulin (Humulin, Lilly, USA) at 1 h intervals. Venous blood (70 μL) was sampled from the jugular vein catheter at 15 min intervals, and loss of fluid volume was replaced by hourly saline injections. Blood glucose was colorimetrically quantitated with an Accu-chek analyzer (Roche Diagnostics, Switzerland). After completion of the experiment, the rats were immediately sacrificed with 20% Euthasol. Acquired current values generated by the EMR were averaged over 1 min intervals and converted to glucose concentration by multiplying with a conversion factor, k, with k being the blood glucose concentration at baseline divided by the EMR current at baseline. The responses of glucose in blood and ISF to the experimental intervention were expressed as changes from their respective baseline. Differences in response

Figure 1. Experimental setup used for the evaluation of the in vitro performance of a microdialysis probe coupled to an EMR.

vitro by establishing linear range, sensitivity for glucose, and physical lag time of the response. The microdialysis probe with a membrane length of 1.0 cm was perfused at a flow rate of 0.5 μL/min. In the EMR (10 cm long mixing channel), dialysate was mixed with a 150 U/mL GOx solution delivered at 1.5 μL/ min. A 20 μL air bubble was entrapped in each gastight syringe to dampen the pulsation from the pumps (Harvard Apparatus Ltd., UK).28 The microdialysis probe was submerged in a stirred solution at 37 °C containing 0, 2.1, 5.6, 10.1, 15.0, or 20.6 mM of glucose in water. At each concentration level, the current was recorded for 15 min. The glucose concentration was quantitated electrochemically as described previously.23 Data was acquired every 2 s. In a separate experiment with a similar setup as above, we investigated whether the protective layer of overoxidized PPy 10951

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry

Article

amplitudes and lag times to maximum response between glucose in blood and ISF were statistically analyzed by one-way ANOVA with repeated measures, followed by the BonferroniHolm posthoc test. p < 0.05 was considered significant.



RESULTS Optimization of in Vitro Microdialysis Recovery. In the first experiment, we evaluated the influence of the flow rate of the perfusion fluid through the microdialysis probe and the length of the semipermeable membrane on recovery efficiency (see Figure 2). In vitro recovery efficiency was higher with a 3.0

Figure 3. Effect of permselective protection by coating overoxidized PPy onto the Pt microelectrodes on faradaic interference from other electroactive species. In vitro, the oxidation current obtained from 5.6 mM glucose was compared to that from the glucose combined with 150 μM ascorbic acid, 500 μM uric acid, or 150 μM acetaminophen. This experiment was performed with bare Pt electrodes and with electrodes protected by overoxidized PPy (n = 4). Interference is expressed as percent of glucose current. Data are means ± SE.

Figure 2. Effect of the flow rate of the perfusion fluid through the microdialysis probe and the length of the semipermeable membrane (1.0 or 3.0 cm) on in vitro recovery efficiency of glucose (n = 3 probes, 3 experiments per probe). Data are means ± SE.

cm-long membrane compared to 1.0 cm, and increased exponentially with decreasing flow rate. At a flow rate of 0.5 μL/min, the recovery was 95−100% with both membranes. For further experiments, the 1.0 cm-long membrane was chosen because of size-related benefits relevant to subcutaneous implantation. Evaluation of CGM Performance in Vitro. We investigated the effectiveness of overoxidized PPy as permselective layer on the electrodes of the EMR in protecting against common interfering electroactive species (see Figure 3). The EMR was a microchannel with an embedded groove array which has been shown to enhance mixing and reaction of glucose and GOx solutions.23 Compared to bare electrodes, the applied permselective protection reduced interference by 500 μM uric acid from 2.1 ± 1.1% to 0.1 ± 0.4% (n = 4; p < 0.05). Interference from 150 μM ascorbic acid (1.9 ± 0.9%) or 150 μM acetaminophen (1.5 ± 0.9%) was lower in each electrode coated with PPy compared to uncoated electrodes, although this effect was not statistically significant. The influence of these interferants on electrode signal is thus very small. In the second experiment, we evaluated the in vitro performance of the microdialysis probe coupled to the EMR. There was a linear relationship between EMR current and glucose concentration in the beaker between 2.1 and 20.6 mM (see Figure 4). The sensitivity was 7.8 ± 1.0 nM (n = 6 devices) with a correlation coefficient, R, of 0.997. An example of a recorded current−time trace is shown in the inset. The physical lag time between switching the beaker content and the change in EMR current was measured to be approximately 18 min.

Figure 4. In vitro calibration of an EMR coupled to a microdialysis probe with a membrane length of 1.0 cm, perfused at a flow rate of 0.5 μL/min (n = 6). The electrodes are coated with a permselective layer of PPy. The inset shows an example of a recorded current−time trace with sequentially increasing glucose concentrations at 15 min intervals. Each value of the current in the calibration curve is an average of data acquired every 2 s for a period of 15 min. Data are means ± SE.

Evaluation of CGM Performance in Vivo. In the final experiment, we used our CGM to monitor ISF glucose concentrations in subcutaneous adipose tissue of anesthetized, healthy rats (see Figure 5). Baseline concentrations of blood glucose were 10.2 ± 2.3 mM. At baseline, the EMR current was 22.7 ± 7.3 nA and the resulting conversion factor k was 0.47 ± 0.1 mM/nA. Systemic administration of saline had no effect on glucose concentrations in blood or ISF. In contrast, intravenous injection of 20% glucose rapidly increased the concentration of glucose in blood by about 2 mM to 12.1 ± 2.3 mM (p < 0.05) and in subcutaneous ISF by about 2 mM to 11.9 ± 1.5 mM compared to baseline. Because responses of glucose values varied between animals, the increases in glucose concentrations in ISF after glucose injection were not statistically significant. However, the expected direction of change occurred in every 10952

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry



Article

DISCUSSION

Automated registration of glucose values for the benefit of good glycemic control in patients with diabetes requires an accurate CGM with small dimensions for implantation convenience, short reaction times for a real-time response, and low consumption of chemical reagents for cost-effectiveness. Here, we evaluated the performance of a novel CGM consisting of a microdialysis probe coupled to an EMR based on lab-on-a-chip technology. We demonstrated that this CGM was able to register stable ISF glucose concentrations in subcutaneous adipose tissue of rats under baseline conditions, as well as glucose changes in response to intravenous glucose injection or systemic insulin-induced hypoglycemia. Evaluation of in vitro performance revealed that our CGM has a linear response in the glucose concentration range of 2.1− 20.6 mM, which is clinically relevant in glycemic management.6,29 The sensitivity is comparable to other biosensorbased CGMs,9 while the physical lag time is only 18 min compared to the 30 min reported for the Subcutaneous Continuous Glucose Monitoring System I.14 This lag time is mainly due to the dead volume in the tubing that connects the microdialysis probe to the EMR and can be shortened by putting all CGM components in closer proximity. The relatively good specificity for glucose observed with bare electrodes compared to other electrochemical biosensors9,12 might be explained by the high reaction efficiency of fresh, dissolved GOx compared to immobilized enzyme. Interference from other electroactive species known to be common in blood and ISF was reduced to only 2−3%, when a permselective protective layer of overoxidized PPy was coated onto the electrodes. We demonstrated that the microdialysis probe in the current experimental configuration has an in vitro glucose recovery efficiency of nearly 100%. This high recovery efficiency is necessary for microdialysate to reflect the ISF while keeping to a minimum the influence of vasomotricity or changes in diffusion kinetics in the tissue during the experiment. Since apparent recovery efficiency in vivo might be different from the one established in vitro, we chose to calibrate the EMR current against the glucose concentration in the blood at steady state. In this study, the baseline concentrations of glucose in the blood were relatively high compared to resting values normally obtained in freely moving rats (10.2 versus 7.7 mM),30 likely caused here by the influence of anesthesia on altered glucose metabolism. The experimental design of the present study does not allow us to evaluate absolute concentrations of glucose in ISF compared to blood. However, previous studies reported glucose levels in blood and ISF during steady state to be similar31,32 or to be 15−30% lower in ISF.33,34 It has been suggested that exchange of glucose between blood and ISF is almost instantaneous under normal physiological conditions,29 and a gradient is only created when a significant amount of glucose is taken up by the surrounding tissue.35 Our results showed that the amplitude of the changes in glucose concentration in blood and subcutaneous ISF was similar, both when experimentally increased after glucose administration and decreased during insulin-induced hypoglycemia. It has been reported earlier that the blood−ISF gradient is smaller in adipose tissue compared to skeletal muscle31 and tends to be larger during systemic hypoglycemia compared to other glycemic states.36−38 In the present study, glucose concentrations in blood and ISF both increased rapidly after

Figure 5. Dynamic changes in glucose concentration in the blood and subcutaneous ISF of anesthetized rats (n = 5). (a) Glucose concentrations were experimentally modulated by consecutive intravenous injections of saline (1 mL/kg, control), 20% glucose (w/v), and 5 U/kg insulin at one hour intervals. (b) Expanded view of glucose trace between 3:15 and 4:15 (h/min) to better show the physiological lag between glucose concentrations in blood and ISF. Blood glucose was sampled from the jugular vein at 15 min intervals, and interstitial glucose was continuously monitored by an EMR, with data points averaged over 1 min periods. Data are expressed as changes from baseline within each subject and presented as means ± SE. ∗ indicates a significant change from baseline (p < 0.05; ANOVA).

animal. As indicated in Figure 5b, the glucose concentration in ISF did change significantly directly after glucose injection (p < 0.05). The highest level of blood glucose was observed at 15 min, whereas ISF glucose reached its peak value at 20−24 min after administration. Both parameters gradually returned to baseline within 1 h. The concentration of glucose in the blood decreased by approximately 8 mM relative to baseline in response to intravenous injection of 5 U/kg insulin and reached its lowest value of 2.1 ± 0.6 mM at 60 min after administration (p < 0.05). ISF glucose significantly decreased compared to baseline starting at 20 min after insulin administration (p < 0.05) and reached its lowest concentration of 2.1 ± 0.9 mM relative to baseline at 65−69 min after administration. Euglycemia had not been restored at either site at the end of the experiment. 10953

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry

Article

fabrication of fluidic reservoirs, pumps, and electronics are important aspects to consider if business-card sized devices are to be realized. To gain more insight in physiological and pharmacological study, it is also important to study freely moving animals instead of anesthetized ones with the microdialysis-coupled EMR. Since the chemical used for anesthesia influences animal metabolism, awakened animals provide a better representation of physiological glucose delivery.

intravenous injection of glucose and decreased gradually after insulin injection. At either intervention, changes in ISF glucose trailed the changes in the blood by 5−9 min regardless of the direction of change. This is in good agreement with earlier findings in diabetic dogs35 and also recent reports in human subjects.39,40 However, other studies in the literature show conflicting results with some revealing rapid equilibration36,37 and others demonstrating the response in the ISF to be delayed by 18−30 min.32,41,42 Recovery from insulin-induced glucoprivation has been shown to last longer in ISF compared to blood, both in healthy and diabetic subjects.32,33,36 In our study, euglycemia had not yet been restored at the end of the experiment. Interestingly, some investigators have suggested the existence of a “push−pull” phenomenon; i.e., glucose concentrations in ISF decrease earlier compared to the blood specifically during hypoglycemia but have a delayed response instead during increasing concentrations.34,43 We could not confirm the “push−pull” hypothesis in this study. Diabetes per sé may also affect the blood−ISF relationship, as hypoglycohistiosis preceded systemic hypoglycemia by approximately 15 min in diabetic rats but not in healthy ones.43 In the present study, we evaluated the feasibility of our microdialysis-coupled EMR to function as a CGM in patients with diabetes, by investigating its in vivo performance in healthy, anesthetized rats as a preclinical proof-of-principle. Translation of our findings to humans should be treated with caution, because there are considerable anatomical and physiological differences at the level of subcutaneous adipose tissue.44 Humans have substantially more subcutaneous fat than rats. The more fat cells, the more tortuous the diffusion path is for glucose. In addition, the size of adipocytes might affect the amount of available ISF. It has been shown that glucose recovery by microdialysis was negatively correlated with abdominal skin-fold thickness, probably due to heterogeneity in capillary density.45 The suggested influence of the degree of adiposity on measurable ISF glucose concentrations is particularly relevant to patients with type 2 diabetes, who more often tend to be overweight or obese than healthy people.46 In addition, glucose recovery is also affected by diabetes per sé.18 Our current in vitro experiment was performed immediately after surgical implantation of the CGM and lasted only for a few hours. As the time frame between CGM implantation and its actual use is critical for continuous in vivo monitoring, our present work should be viewed merely as a proof-of-principle to demonstrate the viability of a new microfluidic glucose sensing system based on online microdialysis sampling in vivo. The results presented here may reflect artificial rapid equilibration between blood and ISF caused by disruption of the endothelial barrier and a lower rate of diffusion through the tissue caused by the presence of tissue trauma and cellular debris, as well as relatively low levels of ISF glucose caused by local inflammation and wound healing processes.47 After CGM implantation, stabilization and normalization may be required for 12 h up to 4 days.45,48 Wientjes et al. demonstrated good long-term stability of their microdialysis-based CGM in humans,49 although CGM performance might be impaired in the long run by fibrous encapsulation and biofouling.50 For future research, further miniaturization of the microfluidic system should be implemented to achieve a true μSCGM. Here, we have successfully demonstrated an on-chip detector in a microfluidic device as a first step and shown feasibility in an in vivo application. The compact design and



CONCLUSIONS We have successfully demonstrated an in vivo application of a novel CGM for continuous monitoring of the glucose concentration in subcutaneous ISF of anesthetized rats. A microfluidic device with integrated chaotic mixer and electrochemical EMR has been successfully combined with a microdialysis probe for this purpose. Compared to existing CGMs, this new device has smaller dimensions for implantation convenience, shorter reaction times for a real-time response, and lower consumption of chemical reagents for costeffectiveness. The physical lag time is much reduced by the use of lab-on-a-chip technology, and the continuous delivery of fresh, dissolved GOx enhances long-term analytical stability and specificity for glucose over other electroactive species common in blood and ISF. Specificity of the EMR is further improved by a permselective PPy membrane coated onto the electrodes. This novel CGM has good sensitivity for glucose and exhibits a linear response in the concentration range that is clinically relevant for diabetes. Our results showed concentrations of glucose in subcutaneous ISF to change in a similar fashion to those in the blood in response to experimental modulating interventions. The amplitude of responses in ISF to intravenous injection of glucose or insulin was comparable to the blood, but the ISF response trailed the blood by 5−9 min regardless of the direction of change. We therefore conclude that the combination of a microdialysis probe coupled to an EMR is feasible as a CGM in aiding glycemic control in patients with diabetes. Future studies will be necessary to further reduce the physical lag time and to investigate the performance of this novel CGM in human subjects.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Telephone: +31 (0) 50 363 3337. Fax: +31 (0) 50 363 7582. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to acknowledge Daphne Trul and Karola Jansen-van Zelm for their excellent technical assistance with the animal experiments.



REFERENCES

(1) Klonoff, D. C. Diabetes Care 2005, 28, 1231−1239. (2) Battelino, T.; Bolinder, J. Curr. Diabetes Rev. 2008, 4, 218−222. (3) Oliver, N. S.; Toumazou, C.; Cass, A. E. G.; Johnston, D. G. Diabetic Med. 2009, 26, 197−210. (4) DCCT Group. N. Engl. J. Med. 1993, 329, 977−986. (5) Davis, S.; Alonso, M. D. J. Diabetes Complicat. 2004, 18, 60−68. (6) Jeha, G. S.; Karaviti, L. P.; Anderson, B.; Smith, E. O.; Donaldson, S.; McGirk, T. S.; Haymond, M. W. Diabetes Care 2004, 27, 2881− 2886.

10954

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955

Analytical Chemistry

Article

(7) Steil, G. M.; Panteleon, A. E.; Rebrin, K. Adv. Drug Delivery Rev. 2004, 56, 125−144. (8) Heinemann, L. Diabetes Technol. Ther. 2003, 5, 545−561. (9) Wentholt, I. M.; Vollebregt, M. A.; Hart, A. A.; Hoekstra, J. B.; DeVries, J. H. Diabetes Care 2005, 28, 2871−2876. (10) Poscia, A.; Mascini, M.; Moscone, D.; Luzzana, M.; Caramenti, G.; Cremonesi, P.; Valgimigli, F.; Bongiovanni, C.; Varalli, M. Biosens. Bioelectron. 2003, 18, 891−898. (11) Schoemaker, M.; Andreis, E.; Roper, J.; Kotulla, R.; Lodwig, V.; Obermaier, K.; Stephan, P.; Reuschling, W.; Rutschmann, M.; Schwaninger, R.; Wittmann, U.; Rinne, H.; Kontschieder, H.; Strohmeier, W. Diabetes Technol. Ther. 2003, 5, 599−608. (12) Petrou, P. S.; Moser, I.; Jobst, G. Biosens. Bioelectron. 2003, 18, 613−619. (13) Weiss, R.; Yegorchikov, Y.; Shusterman, A.; Raz, I. Diabetes Technol. Ther. 2007, 9, 68−74. (14) Kapitza, C.; Lodwig, V.; Obermaier, K.; Wientjes, K. J. C.; Hoogenberg, K.; Jungheim, K.; Heinemann, L. Diabetes Technol. Ther. 2003, 5, 609−614. (15) Dempsey, E.; Diamond, D.; Smyth, M. R.; Urban, G.; Jobst, G.; Moser, I.; Verpoorte, E. M. J.; Manz, A.; Michael Widmer, H.; Rabenstein, K.; Freaney, R. Anal. Chim. Acta 1997, 346, 341−349. (16) Freaney, R.; McShane, A.; Keaveny, T. V.; McKenna, M.; Rabenstein, K.; Scheller, F. W.; Pfeiffer, D.; Urban, G.; Moser, I.; Jobst, G.; Manz, A.; Verpoorte, E.; Widmer, M. W.; Diamond, D.; Dempsey, E.; de Viteri, F. J. S.; Smyth, M. Ann. Clin. Biochem. 1997, 34, 291− 302. (17) Salieb-Beugelaar, G. B.; Simone, G.; Arora, A.; Philippi, A.; Manz, A. Anal. Chem. 2010, 82, 4848−4864. (18) Nandi, P.; Lunte, S. M. Anal. Chim. Acta 2009, 651, 1−14. (19) Wang, M.; Roman, G. T.; Schultz, K.; Jennings, C.; Kennedy, R. T. Anal. Chem. 2008, 80, 5607−5615. (20) Gravesen, P.; Raaby Poulsen, K.; Dirac, H. J. Diabetes Sci. Technol. 2007, 1, 372−374. (21) Huang, C.-J.; Chen, Y.-H.; Wang, C.-H.; Chou, T.-C.; Lee, G.-B. Sens. Actuators, B 2007, 122, 461−468. (22) Hsieh, Y.-C.; Zahn, J. D. J. Diabetes Sci. Technol. 2007, 1, 375− 383. (23) Moon, B.-U.; Koster, S.; Wientjes, K. J. C.; Kwapiszewski, R. M.; Schoonen, A. J. M.; Westerink, B. H. C.; Verpoorte, E. Anal. Chem. 2010, 82, 6756−6763. (24) Moon, B.-U.; Vries, M. G.; Westerink, B. H. C.; Verpoorte, E. Sci. China Life Sci. 2012, 55, 515−523. (25) Ramanavicius, A.; Ramanaviciene, A.; Malinauskas, A. Electrochim. Acta 2006, 51, 6025−6037. (26) Walker, E.; Wang, J.; Hamdi, N.; Monbouquette, H. G.; Maidment, N. T. Analyst 2007, 132, 1107−1111. (27) Wahono, N.; Qin, S.; Oomen, P.; Cremers, T. I. F.; de Vries, M. G.; Westerink, B. H. C. Biosens. Bioelectron. 2012, 33, 260−266. (28) Pravda, M.; Marvin, C. A.; Sarre, S.; Michotte, Y.; Kauffmann, J.M. Anal. Chem. 1996, 68, 2447−2450. (29) Aussedat, B.; Dupire-Angel, M.; Gifford, R.; Klein, J. C.; Wilson, G. S.; Reach, G. Am. J. Physiol. Endocrinol. Metab. 2000, 278, E716− 728. (30) de Vries, M. G.; Arseneau, L. M.; Lawson, M. E.; Beverly, J. L. Diabetes 2003, 52, 2767−2773. (31) Moberg, E.; Hagströ m-Toft, E.; Amer, P.; Bolinder, J. Diabetologia 1997, 40, 1320−1326. (32) Cheyne, E. H.; Cavan, D. A.; Kerr, D. Diabetes Technol. Ther. 2002, 4, 607−613. (33) Wientjes, K. J. C.; Schoonen, A. J. M. Int. J. Artif. Organs 2001, 24, 884−889. (34) Sternberg, F.; Meyerhoff, C.; Mennel, F. J.; Mayer, H.; Bischof, F.; Pfeiffer, E. F. Diabetologia 1996, 39, 609−612. (35) Rebrin, K.; Steil, G. M.; van Antwerp, W. P.; Mastrototaro, J. J. Am. J. Physiol. Endocrinol. Metab. 1999, 277, E561−571. (36) Caplin, N. J.; O’Leary, P.; Bulsara, M.; Davis, E. A.; Jones, T. W. Diabet. Med. 2003, 20, 238−241.

(37) Kulcu, E.; Tamada, J. A.; Reach, G.; Potts, R. O.; Lesho, M. J. Diabetes Care 2003, 26, 2405−2409. (38) McGarraugh, G. V.; Clarke, W. L.; Kovatchev, B. P. Diabetes Technol. Ther. 2010, 12, 365−371. (39) Kamath, A.; Mahalingam, A.; Brauker, J. Diabetes Technol. Ther. 2009, 11, 689−695. (40) Kovatchev, B. P.; Shields, D.; Breton, M. Diabetes Technol. Ther. 2009, 11, 139−143. (41) Rossetti, P.; Porcellati, F.; Fanelli, C. G.; Bolli, G. B. Diabetes Technol. Ther. 2006, 8, 326−337. (42) Schmidtke, D. W.; Freeland, A. C.; Heller, A.; Bonnecaze, R. T. Proc. Natl. Acad. Sci. 1998, 95, 294−299. (43) Thome-Duret, V.; Reach, G.; Gangnerau, M. N.; Lemonnier, F.; Klein, J. C.; Zhang, Y.; Hu, Y.; Wilson, G. S. Anal. Chem. 1996, 68, 3822−3826. (44) Wisniewski, N.; Rajamand, N.; Adamsson, U.; Lins, P. E.; Reichert, W. M.; Klitzman, B.; Ungerstedt, U. Am. J. Physiol. Endocrinol. Metab. 2002, 282, E1316−E1323. (45) Lutgers, H. L.; Hullegie, L. M.; Hoogenberg, K.; Sluiter, W. J.; Dullaart, R. P. F.; Wientjes, K. J.; Schoonen, A. J. M. Neth. J. Med. 2000, 57, 7−12. (46) Cengiz, E.; Tamborlane, W. V. Diabetes Technol. Ther. 2009, 11, S-11−S-16. (47) Stenken, J.; Church, M.; Gill, C.; Clough, G. AAPS J. 2010, 12, 73−78. (48) Wientjes, K. J. C.; Grob, U.; Hattemer, A.; Hoogenberg, K.; Jungheim, K.; Kapitza, C.; Schoonen, A. J. M. Diabetes Technol. Ther. 2003, 5, 615−620. (49) Wientjes, K. J. C.; Vonk, P.; Vonk-van Klei, Y.; Schoonen, A. J.; Kossen, N. W. Diabetes Care 1998, 21, 1481−1488. (50) Dungel, P.; Long, N.; Yu, B.; Moussy, Y.; Moussy, F. J. Biomed. Mater. Res., Part A 2008, 85A, 699−706.

10955

dx.doi.org/10.1021/ac402414m | Anal. Chem. 2013, 85, 10949−10955