Calibration of a subcutaneous amperometric glucose sensor. Part 1. Effect of measurement uncertainties on the determination of sensor sensitivity and background current

The calibration of a continuous glucose monitoring system, i.e. the transformation of the signal I(t) generated by the glucose sensor at time (t) into an estimation of glucose concentration G(t), represents a key issue. The two-point calibration procedure consists of the determination of a sensor sensitivity S and of a background current I(o) by plotting two values of the sensor signal versus the concomitant blood glucose concentrations. The estimation of G(t) is subsequently given by G(t) = (I(t)-I(o))/S. A glucose sensor was implanted in the subcutaneous tissue of nine type 1 diabetic patients during 3 (n = 2) and 7 days (n = 7). For each individual trial, S and I(o) were determined by taking into account the values of two sets of sensor output and blood glucose concentration distant by at least 1 h, the procedure being repeated for each consecutive set of values. S and I(o) were found to be negatively correlated, the value of I(o) being sometimes negative. Theoretical analysis demonstrates that this phenomenon can be explained by the effect of measurement uncertainties on the determination of capillary glucose concentration and of sensor output.     

Rise in background current over time in a subcutaneous glucose sensor in the rabbit: relevance to calibration and accuracy

In order to calibrate a continuous glucose monitor, accurate determination of the background current (I0) is necessary, in part because I0 could change over time. We compared two methods of I0 measurement: (1), extrapolation of sensor output data (as a function of glucose level) to the intercept at zero glucose and (2) direct measurement of the output of a blank anode with no enzyme coat. We implanted telemetric sensors subcutaneously in rabbits and measured their outputs during tri-level glucose clamps once per week for 5 weeks. The two methods yielded similar results. I0 rose substantially over time and this increase reached significance during week 3 by the direct method but not until week 5 by the extrapolation method. Using the direct method, I0 rose from 3.41 (0.60-8.48 nanoamperes (nA), median and range) during week 1 to 13.42 (9.1-14.3) during week 5. Using the extrapolation method, I0 rose from 0.57 (0-16.7) during week 1 to 15.3 (12.2-21.6) during week 5. We conclude that I0 can rise over time. If this rise went undetected and was assumed to be stable, a one-point calibration procedure would overestimate glycemia in the hypoglycemic range, i.e. fail to appreciate the severity of hypoglycemia. It is recommended that during validation of a chronic glucose sensor, I0 be measured sequentially over time.     

In-vivo behaviour of hypodermically implanted microfabricated glucose sensors

The in-vivo behaviour of microfabricated GOD (glucose oxidase)/H2O2 glucose sensor implanted subcutaneously in normal anaesthetized rats has been studied. The sensor consists of a planar, three-electrode microcell, an enzyme membrane (glucose oxidase and bovine serum albumin cross-linked with glutaraldehyde) and an outer diffusion limiting polyurethane membrane. The sensor behaviour during hyperglycaemic (13.8 mM and 11.2 mM), euglycaemic (7.8 mM) and hypoglycaemic (3.5 mM) plateau levels was determined. The values of the in-vivo sensitivity (0.64 +/- 0.05 nA/mM) and background current (1.25 +/- 0.4 nA) were determined using a two-point calibration method and then used to calculate apparent subcutaneous glucose concentrations. The results show the presence of a good correlation between all the plasma glucose levels (G) and the apparent subcutaneous tissue concentrations (G'), with G' = 0.997.G - 0.066, r = 0.9782.     

Acute effects of valsartan on insulin sensitivity in obese, non-hypertensive subjects with and without type 2 diabetes.

BACKGROUND: Two studies were designed to determine whether a single dose (80 mg) of the angiotensin II receptor blocker (ARB), valsartan, alters insulin sensitivity in obese, non-hypertensive subjects with and without Type 2 diabetes. METHODS: Insulin sensitivity (S(I)), glucose effectiveness (S(G)), and acute insulin response (AIR(0-10 min)) were measured by means of a 3-hour insulin-modified frequently sampled intravenous glucose tolerance test (FSIVGTT) before and after a single dose of valsartan. Study 1: obese, normotensive non-diabetic male subjects (n = 12), mean (SD) age 37.2 +/- 11.2 years, BMI 32.8 +/- 6.8 kg/m (2); Study 2: obese, normotensive Type 2 diabetic patients (n = 12), mean age 55.7 +/- 6.9 years, BMI 35.0 +/- 6.8 kg/m (2)/l. Both studies were randomised, double-blind, placebo-controlled, single-dose crossover group studies involving subjects in two study days, two weeks apart. After fasting samples were taken, a 300 mg/kg iv glucose bolus was injected at 0 min, and 0.05 U/kg iv insulin was given 20 min later. Blood samples for analysis of glucose and insulin were taken throughout the 3-hour study period. RESULTS: Study 1 (non-diabetic subjects) S(I) 2.81 vs. 2.63 x 10 (-4) min (-1) per microU/ml (p = 0.54), S(G) 0.020 vs. 0.020 min (-1) (p = 0.90), AIR(0-10) min 3305 vs. 3450 microU/min/ml (p = 0.71); Study 2 (patients with type 2 diabetes) S(I) 0.59 vs. 0.85 x 10 (-4) min (-1) per microU/ml (p = 0.15), S(G) 0.013 vs. 0.014 min (-1) (p = 0.71), AIR(0-10) min 65 vs. 119 microU/min/ml (p = 0.14), placebo vs. valsartan, respectively. CONCLUSION: In obese, non-hypertensive non-diabetic and Type 2 diabetic subjects a single dose of valsartan does not alter insulin sensitivity.     

Calibration in dogs of a subcutaneous miniaturized glucose sensor using a glucose meter for blood glucose determination.

The feasibility of calibrating a glucose sensor by using a wearable glucose meter for blood glucose determination and moderate variations of blood glucose concentration was assessed. Six miniaturized glucose sensors were implanted in the subcutaneous tissue of conscious dogs, and the parameters used for the in vivo calibration of the sensor (sensitivity coefficient and extrapolated current in the absence of glucose) were determined from values of blood glucose and sensor response obtained during glucose infusion. (1) Venous plasma glucose level and venous total blood glucose level were measured simultaneously on the same sample, using a Beckman analyser and a Glucometer II, respectively. The regression between plasma glucose (x) and whole blood glucose (y) was y = 1.12x-0.08 mM (n = 114 values, r = 0.96, p = 0.0001). The error grid analysis indicated that the use of a Glucometer II for blood glucose determination was appropriate in dogs. (2) The in vivo sensitivity coefficients were 0.57 +/- 0.11 nA mM-1 when determined from plasma glucose, and 0.51 +/- 0.07 nA mM-1 when determined from whole blood glucose (t = 1.53, p = 0.18, n.s.). The background currents were 0.88 +/- 0.57 nA when determined from plasma glucose, and 0.63 +/- 0.77 nA when determined from whole blood glucose (t = 0.82, p = 0.45, n.s.). (3) The regression equation of the estimation of the subcutaneous glucose level obtained from the two methods was y = 1.04x + 0.56 mM (n = 171 values, r = 0.98, p = 0.0001).(ABSTRACT TRUNCATED AT 250 WORDS)     

Optimizing insulin sensitivity assessment using the minimal model in chronic heart failure.

BACKGROUND: Minimal model analysis of the intravenous glucose tolerance test (IVGTT) has been used successfully to demonstrate that patients with chronic heart failure (CHF) are insulin-resistant. Continuing experience in minimal model methodology has raised questions about how best to assign basal glucose concentrations during such analyses. METHODS AND RESULTS: IVGTT data from randomly selected patients with CHF (n = 15) and controls (n = 15) were analysed using the minimal model, with the basal glucose concentration (G (b)) assigned the value of fasting plasma glucose concentration (G (fast)), or the value of plasma glucose concentration 180 minutes after the start of the IVGTT (G (180)). Insulin sensitivity (S (I)) was significantly higher with G (b) = G (fast), than with G (b) = G (180) (controls: 5.60 +/- 0.78 vs. 3.36 +/- 0.25/min/muU/ml x 10 (4), p = 0.0017; patients 4.19 +/- 0.54 vs. 2.36 +/- 0.15/min/microU/ml x 10 (4), p = 0.0004). At G (b) = G (fast), CHF patients showed a non-significant 25 % reduction in S (I) in comparison to controls (p = 0.15). In contrast, at G (b) = G (180), CHF patients showed a significant 30 % reduction of S (I) in comparison to controls (p = 0.0018). S (I) estimates derived at G (b) = G (fast) exhibited twice the variability of those estimated using G (b) = G (180) (coefficients of variation of S (I) in patients with CHF were 50.0 % and 24.8 %, respectively). CONCLUSION: In studies of patients with CHF, greater precision and discriminatory power of insulin sensitivity estimates is obtained when the basal glucose concentration is taken as the plasma glucose concentration 180 minutes after the start of the IVGTT.     

Identifiability and online estimation of diagnostic parameters with in the glucose insulin homeostasis

Today, diagnostic decisions about pre-diabetes or diabetes are made using static threshold rules for the measured plasma glucose. In order to develop an alternative diagnostic approach, dynamic models as the Minimal Model may be deployed. We present a novel method to analyze the identifiability of model parameters based on the interpretation of the empirical observability Gramian. This allows a unifying view of both, the observability of the system's states (with dynamics) and the identifiability of the system's parameters (without dynamics). We give an iterative algorithm, in order to find an optimized set of states and parameters to be estimated. For this set, estimation results using an Unscented Kalman Filter (UKF) are presented. Two parameters are of special interest for diagnostic purposes: the glucose effectiveness S(G) characterizes the ability of plasma glucose clearance, and the insulin sensitivity S(I) quantifies the impact from the plasma insulin to the interstitial insulin subsystem. Applying the identifiability analysis to the trajectories of the insulin glucose system during an intravenous glucose tolerance test (IVGTT) shows the following result: (1) if only plasma glucose G(t) is measured, plasma insulin I(t) and S(G) can be estimated, but not S(I). (2) If plasma insulin I(t) is captured additionally, identifiability is improved significantly such that up to four model parameters can be estimated including S(I). (3) The situation of the first case can be improved, if a controlled external dosage of insulin is applied. Then, parameters of the insulin subsystem can be identified approximately from measurement of plasma glucose G(t) only.     

Continuous glucose monitoring and control in a rotating wall perfused bioreactor

A glucose control system consisting of a single in-line glucose sensor, concentrated glucose solution, and computer hardware and software were developed. The system was applied to continuously control glucose concentrations of a perfusion medium in a rotating wall perfused vessel (RWPV) bioreactor culturing BHK-21 cells. The custom-made glucose sensor was based on a hydrogen peroxide electrode. The sensor continuously and accurately measured the glucose concentration of GTSF-2 medium in the RWPV bioreactor during cell culture. Three sets of two-point calibrations were applied to the glucose sensor during the 55-day cell culture. The system first controlled the glucose concentration in perfusing medium between 4.2 and 5.6 mM for 36 days and then at different glucose levels for 19 days. A stock solution with a high glucose concentration (266 mM) was used as the glucose injection solution. The standard error of prediction (SEP) for glucose measurement by the sensor, compared to measurement by the Beckman glucose analyzer, was +/-0.4 mM for 55 days.     

In vivo calibration of the subcutaneous amperometric glucose sensors using a non-enzyme electrode

A new two-point calibration method for the subcutaneous amperometric continuous glucose sensor is reported. The proposed method is based on direct measurement of the background current (I(o)) using a non-enzyme electrode. For in vivo test, three electrodes were implanted in rabbits. Two of the three were identical needle-type enzyme electrodes with perfluorinated polymer outer layers (Pt/enzyme layer/Kel-F/PTFE/Kel-F/Nafion) that were placed in subcutaneous tissue and in a vessel (ear artery), respectively. And one non-enzyme electrode with exactly the same membrane composition as those of other two was in the subcutaneous layer to measure the background current. Implantation in the subcutaneous layer generated many crevices on the protecting layers of the electrodes. The signals from enzyme electrodes were effectively corrected by the measured background current from the non-enzyme electrode. In addition, a telemetric monitoring system was developed and evaluated for in vivo continuous glucose monitoring in order to alleviate the problems of motion artifact.     

Lifespan of subcutaneous glucose sensors and their performances during dynamic glycaemia changes in rats

Performances of a glucose sensor have been investigated during dynamic variations of plasma glucose levels. Subcutaneous glucose concentrations measured by the sensors were calculated by a one-point calibration, performed in basal conditions. A first group of sensors were chronically implanted in the subcutaneous tissue of normal rats. The animals were submitted to glucagon and insulin injection, in order to induce rapid modifications of their glycaemia. This test was repeated at different days after implantation in order to investigate the lifespan and the performance of the sensors. All the sensors were working 1 or 2 days after implantation, and 70% adequately responded to glycaemia variations at day 3 or 4. The quality of the sensors' performance remained constant as a function of the time. With a second group of sensors, we demonstrated that an efficient sterilization procedure did not alter the sensors' characteristics. At the day of implantation, the sterilized sensors' performance, during dynamic variations of plasma glucose levels, was closely similar to that of the non-sterilized sensors. The animals bearing the sterilized devices were rendered diabetic by steptozotocin (STZ) injection. Once the rats had developed a severe hyperglycaemia (1–3 days after STZ), they were injected with intravenous insulin. The subcutaneously implanted glucose sensors correctly followed the decline in plasma glucose levels. We therefore conclude that our sensor could represent a useful tool for short-term continuous blood monitoring.     

Accuracy and repeatability of a new method for measuring facet loads in the lumbar spine

We assessed the repeatability and accuracy of a relatively new, resistance-based sensor (Tekscan 6900) for measuring lumbar spine facet loads, pressures, and contact areas in cadaver specimens. Repeatability of measurements in the natural facet joint was determined for five trials of four specimens loaded in pure moment (+/- 7.5 N m) flexibility tests in axial rotation and flexion-extension. Accuracy of load measurements in four joints was assessed by applying known compressive loads of 25, 50, and 100 N to the natural facet joint in a materials testing machine and comparing the known applied load to the measured load. Measurements of load were obtained using two different calibration approaches: linear and two-point calibrations. Repeatability for force, pressure, and area (average of standard deviation as a percentage of the mean for all trials over all specimens) was 4-6% for axial rotation and 7-10% for extension. Peak resultant force in axial rotation was 30% smaller when calculated using the linear calibration method. The Tekscan sensor overestimated the applied force by 18 +/- 9% (mean+/-standard deviation), 35 +/- 7% and 50 +/- 9% for compressive loads of 100, 50, and 25 N, respectively. The two-point method overestimated the loads by 35 +/- 16%, 45 +/- 7%, and 56 +/- 10% for the same three loads. Our results show that the Tekscan sensor is repeatable. However, the sensor measurement range is not optimal for the small loads transmitted by the facets and measurement accuracy is highly dependent on calibration protocol.     

Improvement of glucose tolerance in type 2 diabetic patients: traditional vs. modern insulin regimens (results from the Austrian Biaspart Study).

OBJECTIVE: Major advantages of modern insulin regimens containing premixed insulin analogues in comparison to traditional insulin regimens have not been evaluated yet. The aim of the present study was to investigate whether meal-related (breakfast, lunch, dinner) application of biphasic insulin aspart 30 (BIAsp 30) provides better glycaemic control than administration of biphasic human insulin 30 (BHI 30) twice per day. RESEARCH DESIGN AND METHODS: In a multi-centre, randomized, open-label parallel trial, a total of 177 patients with type 2 diabetes mellitus were exposed to the two different insulin regimens described above over a study period of 24 weeks. HbA1c and glycemic exposure parameters were measured at predefined intervals. RESULTS: The mean difference between treatment groups in HbA1c after 24 weeks of treatment was 0.08% (p = 0.6419). Analysing the 7-point blood-glucose (BG) profiles, significant differences in BG levels were observed after lunch (156 vs. 176 mg/dl, p = 0.0289), before dinner (142 vs. 166 mg/dl p = 0.006) and after dinner (154 vs. 182 mg/dl p = 0.002) in favour of BIAsp 30 insulin. Prandial BG increment was lower in the BIAsp 30 group at breakfast (p = 0.057) and lunch (p < 0.0005). No difference was found regarding safety parameters in the two treatment groups. CONCLUSIONS: This study demonstrates that meal-related BIAsp 30-insulin maintains postprandial BG control more effectively than traditional BHI 30 insulin twice per day in type 2 diabetic patients.     

Interstitial glucose concentration and glycemia: implications for continuous subcutaneous glucose monitoring

The changes in plasma glucose concentration and in interstitial glucose concentration, determined with a miniaturized subcutaneous glucose sensor, were investigated in anesthetized nondiabetic rats. Interstitial glucose was estimated through two different calibration procedures. First, after a glucose load, the magnitude of the increase in interstitial glucose, estimated through a one-point calibration procedure, was 70% of that in plasma glucose. We propose that this is due to the effect of endogenous insulin on peripheral glucose uptake. Second, during the spontaneous secondary decrease in plasma glucose after the glucose load, interstitial glucose decreased faster than plasma glucose, which may also be due to the effect of insulin on peripheral glucose uptake. Third, during insulin-induced hypoglycemia, the decrease in interstitial glucose was less marked than that of plasma glucose, suggesting that hypoglycemia suppressed transfer of glucose into the interstitial tissue; subsequently, interstitial glucose remained lower than plasma glucose during its return to basal value, suggesting that the stimulatory effect of insulin on peripheral glucose uptake was protracted. If these observations obtained in rats are relevant to human physiology, such discrepancies between plasma and interstitial glucose concentration may have major implications for the use of a subcutaneous glucose sensor in continuous blood glucose monitoring in diabetic patients.     

Simultaneous direct determination of amiloride and triamterene in urine using isopotential fluorometry

A method for the simultaneous fluorometric determination of two diuretics in urine is proposed. The combination of matrix isopotential synchronous fluorescence (MISF) and first derivative techniques provides good analytical results. MISF spectra are obtained by calculating the isopotential trajectory in the three-dimensional fluorescence spectrum for a urine solution. In the spectral contour, the trajectory is taken to be the portion of the line that passes by the fluorescence maxima of both diuretics (lambda(ex) = 365 and lambda(em) = 413 nm for amiloride and lambda(ex) = 365 and lambda(em) = 437 nm for triamterene). Because contour lines connect points of identical intensity and the trajectory is part of a contour line, it is called "isopotential." Analyses was carried out in a 1/1 (v/v) ethanol/water mixture, using an apparent pH of 6.3 provided by 0.01 M sodium/citrate citric acid buffer. Urine samples are diluted 50 times and provide linear calibration plots at amiloride and triamterene concentrations up to 320 and 100 ng mL(-1), respectively. The goodness of the analytical signal was checked by using variance analysis. Signals recorded throughout the calibration range were subjected to three calibrations per each analyte, both in the absence and in the presence of variable amounts of the other analyte. Differences between individual calibrations and slopes were compared with those within individual calibrations. Based on the results, triamterene and amiloride can be accurately quantified in the presence of each other. The limit of detection calculated according to Clayton who uses error propagation throughout the calibration curve and a noncentralized security factor was 16.8 and 2.4 ng mL(-1) for amiloride and triamterene, respectively.     

Circannual Rhythm of Insulin Release and Hypoglycemic Effect of Tolbutamide Stimulus in Healthy Man

Four healthy non obese young volunteers were observed for a 24-hr period, every other month, over the course of one year. Tolbutamide was injected i.v. each day of the experiment every four hours. Tolbutamide-induced insulin secretion (T.I.I.S.) was evaluated by planimetrically measuring insulin areas above basal levels. Tolbutamide-induced hypoglycemic effect was evaluated by measuring the blood glucose difference between the Sth and 25th minute after the drug injection (δG 5′-25′). The macroscopic evaluation of T.I.I.S. and δG 5′-2S′(mean chronograms) permitted the detection of the existence of a circannual variation of both variables. In particular the maximum level of the blood glucose drop (δG 5-25) was registered in February.

Subsequently the quantification of the rhythm of T.I.I.S. was obtained by fitting a sine curve, according to the Cosinor method. The highest insulin release was confirmed in winter.

As previously documented, the existence of a statistically significant circadian rhythm of T.I.I.S. was confirmed in the morning, i.e. the same period of the day in which insulin-induced hypoglycemia occurs.     

L-methionine as an ethylene precursor in Saccharomyces cerevisiae.

L-Methionine induced production of ethylene by Saccharomyces cerevisiae growing in lactate medium. The production induced by L-methionine was inhibited by pyruvate, and elevated by glucose. Labeled ethylene was produced when L-[U-14C]methionine, but not [U-14C]glucose, was fed to the yeast. The mutant S. cerevisiae G1332 (ade-, met-) did not produce significant amounts of ethylene unless L-methionine was added. Thus L-methionine acts as a precursor of ethylene in S. cerevisiae. The role of glucose appears to be other than as a precursor.     

Circannual rhythm of insulin release and hypoglycemic effect of tolbutamide stimulus in healthy man

Four healthy non obese young volunteers were observed for a 24-hr period, every other month, over the course of one year. Tolbutamide was injected i.v. each day of the experiment every four hours. Tolbutamide-induced insulin secretion (T.I.I.S.) was evaluated by planimetrically measuring insulin areas above basal levels. Tolbutamide-induced hypoglycemic effect was evaluated by measuring the blood glucose difference between the 5th and 25th minute after the drug injection (delta G5'-25'). The macroscopic evaluation of T.I.I.S. and delta G5'-25' (mean chronograms) permitted the detection of the existence of a circannual variation of both variables. In particular the maximum level of the blood glucose drop (delta G5'-25') was registered in February. Subsequently the quantification of the rhythm of T.I.I.S. was obtained by fitting a sine curve, according to the Cosinor method. The highest insulin release was confirmed in winter. As previously documented, the existence of a statistically significant circadian rhythm of T.I.I.S. was confirmed in the morning, i.e. the same period of the day in which insulin-induced hypoglycemia occurs.     

每日2次和3次皮下注射诺和锐30疗效临床分析

目的:比较诺和锐30每日2次和3次皮下注射的疗效和安全性。方法:为期3个月的随机、开放式试验,100例2型糖尿病(T2DM)患者随机分为诺和锐30皮下注射2次(每日早、晚餐前)、3次(早、中、晚餐前)组,观测两组患者空腹血糖(FPG),中餐前血糖,晚餐前血糖,睡前的血糖值以及糖化血红蛋白(HbA1c),低血糖事件及其他不良事件差异。结果:诺和锐30皮下注射3次组总体血糖水平低于2次组,低血糖事件和其他不良反应发生次数无显著性差异。结论:两组治疗方法均能有效地降低血糖,3次治疗组控制餐后血糖更具优势,HbA1c降低更好,未增加低血糖风险。     

A Thiazolidinedione Improves In Vivo Insulin Action on Skeletal Muscle Glycogen Synthase in Insulin-Resistant Monkeys

Thiazolidinediones (TZD) have been shown to have anti-diabetic effects including the ability to decrease fasting hyperglycemia and hyperinsulinemia, increase insulin-mediated glucose disposal rate (M) and decrease hepatic glucose production, but the mechanisms of action are not well established. To determine whether a TZD (R-102380, Sankyo Company Ltd., Tokyo, Japan) could improve insulin action on skeletal muscle glycogen synthase (GS), the rate-limiting enzyme in glycogen synthesis, 4 insulin-resistant obese monkeys were given I mg/kg/ day R-102380 p.o. for a 6-week period. Skeletal muscle GS activity and glucose 6-phosphate (G6P) content were compared between pre-dosing and dosing periods before and during the maximal insulin-stimulation of a euglycemic hyperinsulinemic clamp.Compared to pre-dosing, insulin-stimulated GS activity and G6P content were increased by this TZD: GS independent activity (p = 0.02), GS total activity (p = 0.005), GS fractional activity (p = 0.06) and G6P content (p = 0.02). The change in GS activity induced by in vivo insulin (insulin-stimulated minus basal) was also increased by this TZD: GS independent activity (p = 0.03) and GS fractional activity (p = 0.04).We conclude that the TZD R-102380 improves insulin action at the skeletal muscle in part by increasing the activity of glycogen synthase. This improvement in insulin sensitivity may be a key factor in the anti-diabetic effect of the thiazolidinedione class of agents.