In Vitro Sugar Interference Testing With Amperometric Glucose Oxidase Sensors

Abstract

Background:

Electrochemical enzymatic glucose sensors are intended to measure blood or interstitial fluid glucose concentrations. One class of these glucose sensors are continuous glucose monitors (CGMs), indicated for tracking and trending of glucose concentrations in interstitial fluid and as an adjunct to blood glucose testing. Currently approved CGMs employ a glucose oxidase (GOx) electrochemical detection scheme. Potential interfering agents can impact the accuracy of results obtained by glucose sensors, including CGMs.

Methods:

Seven sugars, seven sugar alcohols, and three artificial sweeteners were in vitro screened for interference with amperometric glucose oxidase (GOx) sensors at concentrations greater than physiologic concentrations. Galactose was investigated further at physiologically relevant concentrations using a custom amperometric system. Furthermore, glucose and galactose calibration experiments were conducted to facilitate multiple enzyme kinetic analysis approaches (Michaelis-Menten and Hill equation) to understand the potential source and mechanism of interference from galactose.

Results:

Under in vitro testing, except for galactose, xylose and mannose, all screened compounds exhibited interference bias, expressed in mean absolute relative difference (MARD), of ⩽ 20% even at concentrations significantly higher than normal physiologic concentrations. Galactose exhibited, CGM-dependent, MARD of 47-72% and was subjected to further testing. The highest recorded mean relative difference (MRD) was 6.9 ± 1.3% when testing physiologically relevant galactose concentrations (0.1-10 mg/dL). Enzyme kinetic analysis provided calculations of maximum reaction rates ( $i_{m a x}$ ), apparent Michaelis constants ( $K_{m}^{a p p}$ ), and Hill equation h parameters for glucose and galactose substrates for the enzymes in the CGMs.

Conclusion:

Under the conditions of in vitro screening, 14 of the 17 compounds did not exhibit measuarable interference. Galactose exhibited the highest interference during screening, but did not substantially interfere with CGMs under the conditions of in vitro testing at physiologically relevant concentrations. Enzyme kinetic analysis conducted with galactose supported the notion that (1) the reactivity of GOx enzyme toward nonglucose sugars and (2) the presence of enzymatic impurities (such as galactose oxidase) are two potential sources for sugar interference with GOx glucose sensors, and thus, should be considered during device development.

Keywords

continuous glucose monitors enzyme kinetics glucose sensors glucose oxidase interferences in vitro testing

Continuous glucose monitors (CGMs) are implantable medical devices which have advanced diabetes management by allowing for better glycemic control with more frequent data point collection.^1,2 Commercial CGMs are indicated for the measuring of glucose concentrations in interstitial fluid (ISF) under the skin.³ CGMs consist of three main parts: an implantable glucose sensor, a wearable transmitter and a wireless receiver/monitor. The glucose sensors in many commercially available CGM systems are comprised of electrochemical electrode systems, which contain the immobilized glucose oxidase (GOx) enzyme. GOx converts glucose to gluconic acid and hydrogen peroxide. Hydrogen peroxide in turn is oxidized by the electrode system to generate current.^4-6 An algorithm in the wireless receiver and/or transmitter converts the current to a glucose concentration value that is displayed on the monitor.^7,8 Figure 1 illustrates the basic enzymatic electrochemical glucose sensing scheme of a typical commercially available CGM system, including the chemical reactions involved in producing a signal.

Figure 1.

Example of enzymatic electrochemical glucose sensing element in the CGM, and the chemical reactions involved.

Electrochemical sensors based on GOx may be susceptible to two types of interferences: electrochemical and enzymatic. Electrochemical interference can be caused by the electrochemical oxidation of electroactive substances such as drugs, drug metabolites, or endogenous molecules on the working electrode of the CGM, resulting in changes in electrode current independent of glucose concentration changes.^9-12 Enzymatic interferences may also result in changes in sensor signal, but occur due to either reaction of nonglucose sugars with GOx enzyme directly or with other enzyme impurities within the GOx sensing layer. Enzymatic interferences are the focus of this study, rather than electrochemical interferences.

Although the GOx enzyme is produced by a variety of different organisms, in CGM systems, the enzyme most commonly originates from the fungus Aspergillus niger.¹³ Even though the A. niger GOx enzyme is highly selective for glucose, it exhibits limited reactivity toward other sugars.^13-16 In addition, GOx extracted from A. niger may contain other enzyme impurities.^17-19

Glucose analogues and substitutes are commonly introduced to the human body, primarily orally through diet (eg, food, supplements, drugs, artificial sweeteners), but also by non-oral pathways like intravenous administration of drugs and transdermal lotions. People with diabetes rely on glucose free substitutes like sucralose, mannitol, aspartame, and stevia, among others, to help with glycemic control. Enzymatic interferences of nonglucose sugars with blood glucose meters that use the glucose dehydrogenase enzyme have been reported, in some cases leading to serious or fatal outcomes.^20-22 Given this history, it is important to understand the susceptibility of GOx-based sensors to similar interferences. The goal of this study is to evaluate the effect of nonglucose sugar interference with GOx-based sensors. Commercial CGMs were used to conduct this study, and comparative performance assessments were not the aim of the study. For this reason, the sensors used are not identified in this paper.

Materials and Methods

Interference Screening Experiments Using Commercial Continuous Glucose Monitoring Systems

Three commercially available types of CGM sensors were used to conduct this study. Sensors were inserted in a custom-made flow cell, allowed to equilibrate, and calibrated per the manufacturers’ instructions. The flow cell was filled with 1× Hank’s Balanced Salt Solution (10× HBSS, Gibco®, 14065, ThermoFisher Scientific, Waltham, MA) with 100 mg/dL glucose concentration and 350 mg/L bicarbonate, at pH 7.4 and 37°C; Solutions contained 0.1 g/L of sodium azide to prevent -microbial growth. A high-performance liquid chromatography (HPLC) pump (Waters Corp, Milford, MA) was used to run the solution through the cell at a flow rate of 1 mL/min. All chemicals were acquired from Sigma.

Glucose concentration readings from the CGM receiver in the presence of sugar analyte were compared to baseline readings without sugar analyte to determine interference. A total of seven nonglucose sugars, seven sugar alcohols, and three artificial sweeteners were tested (Sigma Aldrich, at least 98% purity). Given the absence of accepted clinical values for analyte concentrations in ISF, the concentrations of analytes during screening studies were chosen based on a maximum blood concentrations reported for these analytes. According to the two-compartment model that describes the relationship between plasma and interstitial glucose levels, at steady state conditions, interstitial glucose concentration generally correlates with the blood glucose, but with an associated lag time.²³ Table 1 lists nonglucose sugars tested in the initial interference screening. Each compound was screened twice.

Table 1.

Results of Nonglucose Sugar, Sugar Alcohol and Artificial Sweetener Interference Screening Testing.

Sugar	Tested concentration (mg/dL)	Interference? (yes/no)
Fructose	100²⁴	No
Galactose	200, 70²⁵	Yes
Inulin	50²⁶	No
Lactose	500²⁷	No
Maltose	600²⁸	No
Mannose	200²⁹	Yes
Xylose	200, 70²⁷	Yes
Sugar alcohol	Tested concentration (mg/dL)	Interference? (yes/no)
Erythritol	60³⁰	No
Glycerol	300³¹	No
Inositol	50³²	No
Maltitol	0.2³³	No
Mannitol	1000³⁴	No
Sorbitol	300³⁵	No
Xylitol	100³⁶	No
Artificial sweetener	Tested concentration (mg/dL)	Interference? (yes/no)
Aspartame	72 (60 pack)	No
Saccharin	50, 150	No
Sucralose	160	No

Interference screening data were analyzed by calculating the mean absolute relative difference (MARD) between the mean baseline glucose concentration CGM reading without interferent (M₀) and the mean glucose reading value with interferent present (M_I) as:³⁷

M A R D = \frac{| M_{I} - M_{0} |}{M_{0}} * 100 %

(1)

In this study, interference was defined as any change in the detected glucose concentration of more than 20% from a baseline glucose value.

Interference Testing of Galactose at Physiological Levels - via Amperometric Measurements

To test galactose at physiologic concentrations, direct acquisition of the electrical current from a CGM’s sensor was conducted while bypassing the CGM’s algorithm using the device’s built-in circuitry and a benchtop potentiostat. CGM sensors were inserted into the flow cell through a silicone gasket with a drop of silicone sealant applied at the insertion site to prevent leaks. The flow cell was filled with phosphate buffered saline (PBS) solution and the sensors were conditioned in this solution for at least three hours. Custom rapid prototyped sensor connections were fabricated to enable an interface between the sensors and a CHI1005B Electrochemical Analyzer potentiostat (CH Instruments, Inc, Bee Cave, TX). A syringe pump (Harvard Apparatus PhD Ultra) was used for fluid exchanges. Figure 2 shows the schematic of interference testing setup.

Figure 2.

Schematics of the interference testing setup.

Interference tests were run as follows: Glucose-only solution at one of three baseline levels (70, 100, or 250 mg/dL) was infused into the flow cell and the sensors were stabilized at an applied voltage of 0.6 V for at least 40 minutes, with a data point recorded every 0.1 seconds, and a current/voltage sensitivity of 1×10⁻⁸. Some sensors required a 90-minute soak in the glucose solution prior to stabilization, while the other sensors could be stabilized immediately. Baseline current measurements were acquired. After this procedure, the flow cell was flushed with fresh glucose-only solution and again stabilized for 40 minutes. Measurements were conducted at room temperature.

After second stabilization, a glucose + galactose solution was pumped into the flow cell using a syringe containing either 2 mg/dL or 25 mg/dL galactose, for a duration sufficient to achieve galactose concentration targets of 0.1 and 1.0 mg/dL, and 5 and 10 mg/dL, respectively. Each infusion step was conducted with an infusion at 1 mL/min flowrate, 5 minutes of additional stirring, and 5 minutes of signal acquisition. Stirring in the flow cell was conducted during all steps of the process. All fluids run through the flow cell contained 0.1 g/L sodium azide to prevent microbial growth.

Analysis of Amperometric Data

The interference due to galactose was determined by calculating the percentage mean relative difference (MRD) of the current values for glucose solutions with and without interferent. The MRD due to galactose was determined as:

M R D_{i} = \frac{i_{i} - i_{0}}{i_{0}} * 100 %

(2)

where i₀ and i_i are the mean initial current following stabilization and the mean current with interferent at each galactose infusion step, respectively. The i₀ and i_i values were obtained by calculating the means of the current collected in the last minute of each infusion step.

We observed decay in the measured current at each infusion step during glucose baseline measurements. Such current decay during infusion step would artificially reduce i_i value and thus affect the interference value. To correct for this current decay, we chose to calculate interference as a difference in means of MRD of current values between baseline (glucose-only) and galactose-spiked solution.

The MRD for the glucose-only infusion experiments were also calculated by a modified version of equation 2, where i_i is now a decayed current value, $i_{d}$ :

M R D_{o} = \frac{i_{d} - i_{0}}{i_{0}} * 100 %

(3)

Difference in means of MRD values with 95% confidence between baseline (glucose-only) and galactose-spiked solution were obtained with the following formulas:

M R D (D i f f e r e n c e o f M e a n s) = M R D_{i} - M R D_{0}

(4)

S_{M R D_{i} - M R D_{0}} = \sqrt{\frac{S_{0}^{2}}{n_{0}} + \frac{S_{i}^{2}}{n_{i}}}

(5)

C I L o w e r L i m i t = M R D_{i} - M R D_{0} - (t_{C L}) (S_{M_{0} - M_{i}})

(6.1)

C I U p p e r L i m i t = M R D_{i} - M R D_{0} + (t_{C L}) (S_{M_{0} - M_{i}})

(6.2)

In these formulas, $S_{M R D_{i} - M R D_{0}}$ is the standard error of the difference in MRD means between the galactose ( $M R D_{i}$ ) and control ( $M R D_{0}$ ) groups. The $S$ term represents the group standard deviation, where $s_{0} a n d s_{i}$ are the standard deviations of the baseline glucose and galactose-spiked solution currents, respectively, $n_{0} a n d n_{i}$ are the number of glucose and galactose-spiked samples. CI is the Confidence Interval, $M R D$ is the group mean, and $t_{C L}$ is the t-table value for the 95% confidence level.

Galactose Purity

Glucose impurity in the galactose sample was tested using a glucose/hexokinase assay kit (GAHK-20, Sigma Aldrich) with detection limit of 0.05 mg/dL of glucose. Testing was done as per instructions of the assay. Spectrophotometry at 340 nm (Genesys 20, ThermoScientific, Waltham, Massachusetts) was used to quantify glucose content.

Mechanism of Galactose Interference

The mechanism of galactose interference was assessed by performing interference testing and glucose and galactose calibration experiments.

Interference Testing

Interference testing of galactose at 70 mg/dL was evaluated at various concentrations of glucose (70, 100, and 250 mg/dL) using the experimental setup described in section - Interference Testing of Galactose at Physiological Levels via Amperometric Measurements above.- Data were analyzed in terms of MRD as described in section - Analysis of Amperometric Data, above-. In addition, the change in raw current ( $i$ ∆i) produced by galactose with each sensor type was determined as:

Δ i = i_{i} - i_{d}

(7)

Where, $i_{i}$ and $i_{d}$ represent the measured currents from glucose solutions in the presence or absence of galactose, respectively (as defined in section - Analysis of Amperometric Data, above).

Calibration Experiments and Enzyme Kinetics Analysis

Glucose calibration experiments were conducted in the presence or absence of galactose, using the setup described in section - Interference Testing of Galactose at Physiological Levels Via Amperometric Measurements, above as follows: Sensors were inserted into the flow cell and allowed to equilibrate in a buffer solution without glucose. After baseline current was reached, an appropriate volume of glucose solution at 400 mg/dL was added incrementally to reach desired concentrations up to a final concentration 400 mg/dL. The current was recorded at each glucose addition step. Galactose calibration was completed in the same manner.

The data from glucose and galactose calibration experiments were analyzed using Michaelis-Menten kinetics.³⁸ Michaelis-Menten enzyme kinetics analysis is a classic approach to examining enzyme-substrate binding behavior and is represented with the following equation:

v_{0} = \frac{V_{m a x} S}{K_{m} + S}

(8)

In this equation, $v_{0}$ is the initial reaction velocity of enzyme binding with a substrate, $V_{m a x}$ is the maximum achievable reaction rate, $K_{m}$ is the Michaelis constant that represents the substrate concentration needed to achieve half of the maximum reaction velocity, and $S$ is the substrate concentration.

While this approach was initially developed to analyze free enzymes in solution, it may be modified and applied to immobilized enzymes used in biosensors. When applied to immobilized enzymes, analysis gives the apparent rather than the true values of $V_{m a x}$ and $K_{m}$ . Amperometric data from enzyme biosensors may be analyzed using this approach, with the measured current serving as the reaction velocity.^39-41 The equation that has been utilized for these purposes is shown below:

i = \frac{i_{m a x} S}{K_{m}^{a p p} + S}

(9)

where $i$ is the measured current at a given substrate concentration, $i_{m a x}$ is the maximum achievable current measured from the enzyme reaction/transduction mechanism, the apparent Michaelis constant ( $K_{m}^{a p p}$ ) is the substrate concentration needed to achieve half of the $i_{m a x}$ value, and $S$ is the substrate concentration. The $i_{m a x}$ and $K_{m}^{a p p}$ values were the key parameters analyzed to assess the behavior of the CGMs.

Traditionally, the values of V_max ( $i_{m a x}$ ) and K_m ( $K_{m}^{a p p}$ ) have been determined using linear translations of the Michaelis-Menten equation to create inverse linear plots whose x- and y-intercepts define the Km and Vmax respectively. Lineweaver-Burk,^42,43 Eadie-Hofstee,^43,44 and Hanes-Woolf⁴⁵ are the most common approaches to perform these transformations. Direct linear plotting,⁴⁶ another linear translation approach to determine Michaelis-Menten kinetics, avoids amplification of experimental error that is inherent to the inverse plotting approaches.⁴³ Nonlinear regression approaches are also available for enzyme data analysis.^47,48

Enzyme behaviors do not always follow Michaelis-Menten kinetics and may display sigmoidal distributions rather than hyperbolic. In these cases, linear transformation of the Michaelis-Menten equation is not appropriate. The Hill equation, shown in equation 10, is a means of modeling sigmoidal data.

i = \frac{i_{m a x} S^{h}}{K_{m}^{a p p^{h}} + S^{h}}

(10)

Equation 10 takes the same form as the Michaelis-Menten equation, but with the addition of the $h$ term, which represents a measure of enzyme binding cooperativity. The enzyme binding cooperativity coefficient value indicates whether the enzyme binding affinity increases, decreases or is independent of the initial binding event. Similarly to the nonlinear regression for the hyperbolic Michaelis-Menten equation, a least squares nonlinear regression approach may be used to fit data to the Hill equation.⁴⁹ Furthermore, the Hill equation has been utilized previously to model biosensor enzyme data,⁵⁰ including amperometric GOx sensors.⁵¹

In this study, the data have been analyzed using Lineweaver-Burk and direct linear plotting as linear translation approaches. Furthermore, they have been assessed with nonlinear regression fits to both the Michaelis-Menten and Hill kinetics equations.

Results and Discussion

Interference Screening

Table 1 shows interference screening results for seven sugars, seven sugar alcohols and three artificial sweeteners with two CGMs at 100 mg/dL glucose. Interference was defined as a change in glucose readings by ⩾20% from the mean baseline value, as calculated by MARD (equation 1).

Under the conditions of the initial screening, most tested compounds did not show interference ⩾ 20% MARD with CGMs’ sensors. The artificial sweeteners were not expected to be enzymatically or electrochemically active, but they were included due to their prevalence as sugar substitutes that may be consumed by CGM users. As expected, the artificial sweeteners did not exhibit an interfering effect during screening. Three of the tested nonglucose sugars, galactose, xylose, and mannose exhibited an interfering effect with CGMs when present at higher than physiological concentrations. Galactose showed the largest interfering effect, followed by mannose and xylose, respectively. The MARD value for 200 mg/dL galactose, mannose, and xylose at 100 mg/dL baseline glucose ranged from 47 to 72%, 37 to 48%, and 19 to 42%, respectively, depending on the type of sensor.

Interference Testing via Direct Amperometric Measurements in Physiological Ranges

Given that the largest interferences at screening concentrations were observed with galactose, the galactose interference at physiologically relevant concentrations was tested via direct amperometric measurements (experiment described in Section Interference Testing of Galactose at Physiological Levels Via Amperometric Measurements above, -).

According to the Interference Testing Clinical Standard, the therapeutic concentration for galactose is reported at ~0.1 mg/dL and the recommended testing concentration for galactose interference is 15 mg/dL.⁵² An FDA guidance document for blood glucose monitoring test systems has recommended a similar interference testing range for galactose, with a therapeutic concentration listed at 0.1 mg/dL and a threshold of high toxicity at 10 mg/dL.⁵³ Following these guidelines, the in vitro interference testing was performed at galactose concentrations between 0.1 and 10 mg/dL, as described in Section Interference Testing of Galactose at Physiological Levels Via Amperometric Measurements, above.

Figure 3 shows the interference of galactose at physiological concentrations with Sensor 1 (Figure 3A) and Sensor 2 (Figure 3B) at 70, 100, and 250 mg/dL of glucose. The difference in current between the baseline glucose and galactose-spiked solution, expressed in MRD values with 95% confidence is plotted as a function of glucose and galactose concentrations. For both sensors, the galactose interference was the largest at 70 mg/dL glucose but was reduced as the baseline glucose concentration increased to 100 or 250 mg/dL.

Figure 3.

Difference in means between baseline glucose and galactose-spiked solution currents in MRD values for Sensor 1 (A) and Sensor 2 (B) at 70, 100, and 250 mg/dL glucose and physiologic levels of galactose. Error bars indicate 95% CI, n = 3-4.

The interference with Sensor 1 (Figure 3A) was less pronounced as compared to that observed with Sensor 2 (Figure 3B). For Sensor 1, at 70 mg/dL glucose, a positive interference trend was noted with increasing galactose concentrations and MRD point values were ⩽ 4.1% at 10 mg/dL galactose. At 100 mg/dL glucose and 250 mg/dL, MRD point values were less than ±1.3% and ±1.9%, respectively for all tested galactose concentrations.

For Sensor 2, increase in current above the baseline was also observed when galactose concentrations increased from 1.0 to 10 mg/dL (Figure 3B). The highest MRD of 6.9 ± 1.3% was calculated at 10 mg/dL galactose and 70 mg/dL glucose. At 100 mg/dL glucose and 10 mg/dL galactose MRD is 4.0 ± 3.7%. Finally, no significant interference occurred in 250 mg/dL glucose.

In summary, at all glucose concentrations tested under the conditions of this in vitro study, the galactose interference did not exceed the 20% deviation from a baseline value (MRD) when present within the physiologically relevant concentration range of 0.1-10 mg/dL.

Mechanism of Interference

Interference Experiments

Even though the enzymatic interferences of all compounds tested with GOx amperometric CGM sensors are largely negligible, it is important to understand and consider the source and mechanisms contributing to the interfering effects during sensor designs. Since galactose showed the highest interference among all compounds tested, additional tests with high (70 mg/dL) galactose concentration were conducted to understand the mechanism of enzymatic interference (Section -, Galactose Purity ).

Figure 4 shows the results of interference tests conducted by holding the galactose concentration at 70 mg/dL while varying the glucose concentration at 70, 100, or 250 mg/dL. The figure shows plots of the difference in means between baseline glucose and 70 mg/dL galactose-spiked solution, in MRD values, for Sensor 1 and 2 (n = 3 or 4). For Sensor 1, none of the MRD values exceeded 20%. Measurements at 70, 100, and 250 mg/dL glucose indicated MRD of 18.1 ± 10.7%, 6.7 ± 10.6%, and 0.3 ± 4.2%, respectively. While there was a decreasing MRD trend with increasing glucose levels, the latter two values were not statistically different from baseline at α = 0.05. For Sensor 2, measurements at both 70 mg/dL and 100 mg/dL glucose indicated MRD ⩾ 20%, reaching 33.8 ± 4.2% and 22.8 ± 6.2%, respectively. In 250 mg/dL glucose, the MRD deviation point value was lower, at 8.2 ± 2.9%.

Figure 4.

Percentage difference in means of current values between baseline glucose and 70 mg/dL galactose-spiked solution in MRD values for Sensor 1 and Sensor 2. Error bars indicate 95% CI. *, **, ***P < .05, n = 3-4.

The combination of the amperometric and screening data suggests that galactose interference exhibits dose-response behavior; however, the magnitude of interference decreases as the glucose concentration increases.

Figure 5 compares the difference in means for the raw current ( $Δ i)$ ∆i) measured between baseline glucose and 70 mg/dL galactose-spiked solutions. There were noticeable differences in amount of current produced due to the same amount of galactose between the two sensor types. For Sensor 1, $Δ i$ in the presence of 70 mg/dL of galactose noticeably decreased with increasing glucose concentrations. T-tests for these point values had p < 0.05, indicating a statistically significant difference between current measurements. For Sensor 2, $Δ i$ in the presence of 70 mg/dL of galactose did not decrease much with increasing glucose concentrations. However, further statistical analysis of the data failed to indicate significant differences or equivalence of $Δ i$ between groups; the two-sided t-tests (α = .05) and 2 one-sided t-tests (TOST) (α = .05, Ɵ = ±20% ) were used to determine significant differences and equivalence in means of raw current, respectively.

Figure 5.

Difference in means for the raw current values between baseline glucose (70, 100, and 250 mg/dL) and 70 mg/dL galactose-spiked solution. Error bars indicate 95% CI. *P < .05, n = 3-4.

Overall, while the MRD decreased with increasing glucose concentration, the difference in raw current response to galactose across the glucose concentration range tested suggests that there may be a difference in the underlying interfering mechanism(s) between the sensor types. For Sensor 1, the magnitude of the raw current produced in response to galactose noticeably decreases as the glucose concentration increases. If galactose is reacting directly with GOx, this behavior would be expected, as increasing amounts of glucose would lead to a greater likelihood of glucose molecules out-competing galactose for interaction with GOx. As the concentration of glucose increases, the raw current produced from galactose interaction with GOx would decline. It would be expected that at high glucose concentration or when enzyme saturation threshold is reached, the addition of galactose would not cause any current increase. At the tested glucose concentration range the enzyme would not be expected to be saturated, as is demonstrated by changes in MRD in Figure 4.

Conversely, for Sensor 2, there is only a slight qualitative decline in the raw current as glucose concentration increases. These data suggest that interference may not only be influenced by galactose/glucose competition for GOx-binding, but may also result from reaction of galactose with galactose oxidase (GalOx) impurity.

Calibration Experiments and Enzyme Kinetics Analysis

In addition to the interference tests discussed above, calibration experiments with glucose and galactose were conducted to elucidate the enzyme kinetic behavior of the CGMs. Figure 6 shows examples of current increase as a function of glucose or galactose concentrations during the calibration experiments. As expected, data clearly demonstrates the greater reaction rate of GOx sensors to glucose than galactose. These data were utilized to estimate the kinetic parameters of the sensors.

Figure 6.

Examples of current increase as function of glucose or galactose concentrations during the calibration experiments with Sensor 1 (A), Sensor 2 (B), and Sensor 3 (C).

Figure 7 shows the effect of glucose concentration on reaction rate (measured in current), in a glucose calibration experiment for Sensor 2. Michaelis-Menten enzyme kinetics constants were obtained by fitting data with the linear and nonlinear Michaelis-Menten enzyme kinetic models. Figure 7A shows the linear Lineweaver-Burk double reciprocal plot of the data. It should be noted that this trend line provides a nonsensical negative y-intercept for the data, translating to negative $i_{m a x}$ and $K_{m}^{a p p}$ values. Similarly, the direct linear plot approach on Figure 7B indicates a median intersection in Cartesian quadrant III, again indicating negative $i_{m a x}$ and $K_{m}^{a p p}$ values. However, the nonlinear fits to the data, shown in Figures 7C and 7D, provide better fits to the data and results in more reasonable $i_{m a x}$ and $K_{m}^{a p p}$ values (shown in Table 2), with the best fit obtained by fitting the Hill equation. Figure 8 shows similar analysis of the galactose calibration data for Sensor 2. Again, the Lineweaver-Burk plot indicates a negative y-intercept, suggesting negative $i_{m a x}$ and $K_{m}^{a p p}$ values. The direct linear plotting approach gives more reasonable apparent values, and the nonlinear regressions also provide good fits to the data. Table 2 displays the apparent Michaelis-Menten enzyme kinetic constants $i_{m a x}$ and $K_{m}^{a p p}$ for all CGM tested sensors using linear and nonlinear fits.

Figure 7.

Effect of glucose concentration on reaction rate (measured as current) obtained from glucose calibration experiments with Sensor 2. Enzyme kinetics constants were obtained by fitting data to the linear and nonlinear enzyme kinetic models: (A) Lineweaver-Burk double reciprocal linear plot, (B) direct linear plot, (C) nonlinear regression, and (D) Hill equation nonlinear regression.

Table 2.

The Apparent Enzyme Kinetic Constants Determined Using Linear and Nonlinear Models.

Linear approaches	Lineweaver-Burk		Direct plot		Nonlinear approaches	Solver fit		Hill equation
Glucose
	$i_{m a x}$	$i_{m a x}$	$K_{m}^{a p p}$		$i_{m a x}$	$K_{m}^{a p p}$	$i_{m a x}$	$K_{m}^{a p p}$	h
Sensor 1 (n = 2)	421	1096	153	339	Sensor 2 (n = 4)	33.4	3253	7.7	352	2.3
					Sensor 3 (n = 2)	28.3	1907	10.7	467	1.2
Galactose
	$i_{m a x}$	$K_{m}^{a p p}$	$K_{m}^{a p p}$		$i_{m a x}$	$K_{m}^{a p p}$	$i_{m a x}$	$K_{m}^{a p p}$	h
Sensor 1 (n = 2)	11.5	89	15.0	115	Sensor 2 (n = 4)	12.3	2423	5.9	927	1.1
					Sensor 3 (n = 2)	1.3	257	2	599	0.8
Glucose: galactose kinetic parameter ratios
	$i_{m a x}$	$K_{m}^{a p p}$	$i_{m a x}$	$K_{m}^{a p p}$		$i_{m a x}$	$K_{m}^{a p p}$	$i_{m a x}$	$K_{m}^{a p p}$	h
Sensor 1	36.6	12.3	10.2	2.9	Sensor 2	2.7	1.3	1.3	0.4	2.0
					Sensor 3	21.8	7.4	5.4	0.8	1.5

Units: $i_{m a x}$ (nA), $K_{m}^{a p p}$ (mg/dL), h (unitless).

Figure 8.

Effect of galactose concentration on reaction rate (measured as current) obtained from galactose calibration experiments with Sensor 2. Enzyme kinetics constants were obtained by fitting data to the linear and nonlinear enzyme kinetic models: (A) Lineweaver-Burk double reciprocal linear plot, (B) direct linear plot, (C) nonlinear regression, and (D) Hill equation nonlinear regression.

We observed that linear fits obtained using Lineweaver-Burk plots provides better fits for Sensor 1 data, which has a greater electrode surface area and a larger linear range for glucose concentration. The greater surface area of the electrode for Sensor 1 likely corresponds to a greater quantity of GOx involved in the reaction; however, lack of exact dimensions for any of the sensor electrodes precludes normalizing current to surface area or assessing the GOx-immobilization density. Presumably, a larger amount of enzyme results in greater raw current, and wider linear range, however this may not be the case if diffusion is more limited. The nonlinear analyses provide better fits for Sensors 2 and 3, which have smaller surface areas and smaller linear ranges. The enzyme kinetics data for Sensors 2 and 3 follow a more sigmoidal pattern than the hyperbolic pattern suggested by the Michaelis-Menten kinetics, which is justification for inclusion of the Hill equation. Inconclusive data with the Lineweaver-Burk plot or other linear transformations may also be related to limitations in applying such analyses to enzyme kinetic studies of sensors with immobilized enzymes or diffusion-limited kinetics.^39,40,53,54

The comparison of kinetic data, specifically the ratio of apparent $i_{m a x}$ value for glucose to $i_{m a x}$ for galactose suggests different underlying interference mechanisms in the tested sensors. The large difference in the max rate constant ( $i_{m a x}$ ), for glucose to galactose, observed for Sensor 1 supports the notion that interference in this sensor is likely due to limited reactivity of GOx to galactose. Whereas, the smaller difference in $i_{m a x}$ values between glucose and galactose observed in Sensors 2 and 3 suggests the presence of galactose specific enzyme impurity in these sensors, such as GalOx. These points are discussed further in subsequent sections.

Implications of Different Mechanisms of Sugar Interference for Sensor Design

We hypothesized three potential mechanisms for the sugar interference: (1) the presence of glucose impurity within nonglucose sugar sample, (2) the reactivity of GOx toward nonglucose sugars, and (3) the presence of enzymatic impurities, such as galactose oxidase (GalOx) within the GOx enzyme sensing layer. Each potential mechanism was explored to explain galactose interferences.

Glucose Contamination in Galactose. Contamination of the galactose samples with glucose has a potential to create a positive interference effect. The purity of the galactose was tested in a glucose-hexokinase assay. The assay did not provide evidence for the presence of quantifiable amount of glucose contamination in the galactose samples (ie, the amount of glucose impurity in galactose sample was below the detection range of this assay, ie, < 0.05 mg/dL). Furthermore, the lot specific certificate of analysis (Lot 011M00202V) reports 100% purity for this Lot based on HPLC analysis. The results of the glucose-hexokinase assay and manufacturing HPLC purity testing suggest that the observed interferences are not due to presence of residual glucose in galactose sample, but are due to galactose.

Reactivity of Glucose Oxidase to Galactose. Reactivity of GOx toward galactose and other nonglucose sugars has been studied. However, the reports comparing the rates for GOx reactivity toward glucose and galactose are diverse.^13,15,55 One such study reports that free GOx oxidizes glucose by 270-600 times faster than galactose.¹⁵ Another study reports that an immobilized form of GOx follows Michaelis-Menten behavior in enzyme-limited experiments and exhibits catalysis efficiency (k_cat/K_M) that is 61 times greater for glucose as compared to galactose.⁵⁵ Additional studies that report K_m values for galactose interaction with GOx indicate higher Km values (lower binding affinity) compared to glucose-GOx interactions (Table 3). While GOx has a much greater affinity for glucose than galactose, it is plausible that some reactivity of GOx toward galactose can occur within the GOx-based sensors, adding to the current measured by the CGMs. The current contribution from galactose would have proportionally greater interference impact at lower rather than higher concentrations of glucose. The interference trends noted in this study fit this concept (ie, greater galactose concentrations coincide with greater measured signal, and the interference is more pronounced at lower glucose concentrations).

Table 3.

Literature-Reported K_m Values for GOx and GalOx.

Enzyme	Substrate	Source	K_m (mg/dL)	Reference
Immobilized GOx	Glucose	Aspergillus niger	7	56
			27	56
			31	56, 57
			32	58
			38	59
			52	59
			59	57
			74	58
			90	58
			101	58
			114	60
			128	56
			142	59
			144	61
			150	58
			216	61
			236	56
			324	61
			342	61, 62
			793	63
			1297	64
Immobilized GalOx	Galactose	Dactylium dendroides	220	65
			285	66
			288	67
			524	68
			775	69
Free GOx	Galactose	Talaromyces funiculosus	1386	70
		Talaromyces funiculosus	2276	70
		Penicillium amagasakiense	17 153	71

When examining kinetic data, the big difference in the max rate constant ( $i_{m a x}$ ) for glucose and galactose, observed for Sensor 1, supports the notion that interference in this sensor type is likely due to the limited reactivity of GOx to galactose. However, if GOx reactivity was the only factor contributing to interference, the increase in current is expected to be much lower than that observed during these interference studies. We hypothesize that in diffusion limited sensing, as in a case of CGM sensors, the interference due to reactivity with GOx enzyme could be amplified, since the availability of substrate is a limiting step in the production of a signal rather than substrate-enzyme reactivity. Sensor 1 seems to be influenced more by the diffusion limitations, than by enzyme reactivity. The galactose interferences noted during screening of Sensor 1 (Section - Interference Screening Experiments Using Commercial Continuous Glucose Monitoring Systems, above were similar to that of mannose, despite mannose having 2.5-12.5× greater reaction rate with GOx compared to galactose.⁷² While reactivity of GOx to glucose, galactose and mannose are different, diffusion coefficients of glucose, galactose, and mannose in aqueous media (20°C, 0.1 MPa) are similar,⁷³ suggesting that transport of these molecules through fully hydrated sensor membrane would be similar as well.

Interferences with Sensors 2 and 3 are likely influenced by a diffusion limitation mechanism and GOx reactivity with galactose. However, the comparison of kinetic data, specifically the ratio of apparent $i_{m a x}$ value for glucose to $i_{m a x}$ for galactose suggests that more than just GOx reactivity with galactose and diffusion limitations may be influencing interference. The following section discusses a potential explanation for these observed differences.

Contamination of Glucose Oxidase - with Other Enzymes. A third potential contributing factor to interference may be GalOx impurities in the GOx sensing layer used in the CGMs. GOx stocks acquired from commercial sources are largely pure, but vendors do report lot-specific contamination by enzymes such as α-amylase, saccharase, maltase, glycogenase, invertase, protease, and galactose oxidase.^17-19 The possible presence of GalOx creates the potential for an increase in interfering signal with increasing galactose concentration. The oxidation of D-galactose by GalOx in the presence of O₂ results in production of D-galacto-hexodialdose and hydrogen peroxide (H₂O₂). Since this reaction produces additional H₂O₂ which could be then further oxidized at the electrode, an increase in the measured current from the sensors can be expected, which is in agreement with the data acquired during this study.

Table 3 reports the Michaelis constant $K_{m}^{a p p}$ for the reaction of GOx and GalOx from various sources with glucose and galactose. A handful of studies have investigated the interaction of galactose with GOx, though these studies were conducted with free enzymes which were not produced by A. niger. These studies have indicated notably higher $K_{m}^{a p p}$ values for free GOx binding with galactose compared to $K_{m}^{a p p}$ values for immobilized GOx (A. niger) with glucose. In general, an increase in $K_{m}^{a p p}$ is indicative of a decrease in binding affinity of a substrate and enzyme. It should also be noted that GOx immobilization and encapsulation sometimes has little impact on $K_{m}^{a p p}$ ,⁷⁴ or may actually increase $K_{m}^{a p p}$ values,^75,76 due to steric impacts or diffusional constrictions. As such, $K_{m}^{a p p}$ values of galactose-GOx binding may actually be higher if GOx were immobilized or encapsulated in an enzyme layer on a sensor. As expected, the $K_{m}^{a p p}$ value obtained from Hill equation fits for galactose interactions with Sensors 2 and 3 is higher than the $K_{m}^{a p p}$ value for glucose. However, the difference between $K_{m}^{a p p}$ for GOx interaction with galactose and glucose is not as great as would be expected based on the literature. The $K_{m}^{a p p}$ values from galactose experiments are more comparable to $K_{m}^{a p p}$ values reported for galactose interaction with GalOx, suggesting possible GalOx contamination in the GOx-sensing layer. For Sensor 1, the $K_{m}^{a p p}$ value for glucose is higher than that for galactose which is opposite to expected considering the higher affinity of GOx to glucose than to galactose.

In addition, it is interesting to note the changes in the h parameter of the sensors’ data fit by the Hill equation. The h term in the Hill equation, known as the Hill coefficient, is a measure of binding cooperativity that takes into account allosteric effects of substrate binding with the enzyme.⁷⁷ When h > 1, there is positive cooperativity in the allosteric binding effects. If h = 1, there is neither positive nor negative cooperativity and the enzyme behaves in a manner analogous to Michaelis-Menten kinetics. When h < 1, there is negative cooperativity, where binding of a substrate molecule has a deleterious impact on the binding of additional molecules. While h is not necessarily indicative of the exact number of binding sites of an enzyme, and has been criticized for its simplistic representation of multiple substrate binding to an enzyme, it may give a minimum value to the number of binding sites.⁷⁷ Hill coefficients for Sensors 2 and 3, suggest that there is a higher level of cooperativity for the enzymes when binding glucose compared to galactose. The decrease in binding cooperativity in the presence of galactose or when only galactose is present suggest that there is either a dampening of cooperativity when GOx binds galactose rather than glucose, or perhaps the presence of GalOx. GOx is a dimeric protein, with two active binding sites on the molecule. This corresponds particularly well with the h = 2.3 for sensor 2. GalOx, conversely, has single active catalytic site. The h value of 1.1 in the presence of galactose, in the context of the relatively high $i_{m a x}$ , is again suggestive of the presence of GalOx for these sensor types.

Conclusion and Future Considerations

Seven nonglucose sugars, seven sugar alcohols, and three artificial sweeteners were tested in vitro for interferences with amperometric GOx-based CGM sensors. Under the conditions of testing, interferences of <20% were observed, except with very high, nonphysiological concentrations of galactose, mannose and xylose; galactose showed the highest levels of interference.

At physiologically relevant concentrations (0.1-10 mg/dL), the galactose interference was largely negligible. A maximum deviation from baseline, MRD of ⩽6.9 ± 1.3%, was observed at 70 mg/dL glucose and 10 mg/dL galactose. The interference response was found to be proportional to galactose dose. Interference appeared to diminish as glucose concentration increased from hypoglycemic- to euglycemic- to hyperglycemic-relevant concentrations, suggesting that interferences may have proportionately greater impact at lower rather than higher glucose levels. The potential for greater interferences at the low glucose concentrations should be considered during sensor design and testing.

Enzyme kinetic analysis suggests that the observed galactose interferences are impacted by the sensor design and membrane resistance to analyte diffusion, and may be attributed to either GOx reactivity toward galactose, and/or to the presence of GalOx impurities within the GOx enzyme layer. GOx reactivity and purity should be taken into consideration when designing GOx-based sensors for glucose detection. Additional research into immobilized forms of GOx would be beneficial to provide insight into the relative reactivity of immobilized enzyme toward other sugars. Much of the available enzyme cross-reactivity literature examines the free-enzyme rather than its immobilized forms that are utilized by many enzyme-based sensor technologies. Additional studies to quantify the impact of diffusion on interferences would also be helpful.

Furthermore, these data have been acquired from in vitro testing protocols. Since in vitro data do not always correlate with in vivo activity, in vivo studies would ultimately determine the clinical significance of any interferences.

Footnotes

Acknowledgements

The authors would like to thank Randy Bidinger and Bruce Fleharty for their help building the flow-cell; Bruce Schneider and Michael Gu for valuable discussions; Alex Nguyen and Roger Narayan and for assistance with 3D printing; DBCMS management for support.

Abbreviations

A. niger, Aspergillus niger; CGMs, continuous glucose monitors; CI, confidence interval; GalOx, galactose oxidase; GOx, glucose oxidase; HBSS, Hank’s balanced salt solution; HPLC, high-performance liquid chromatography; ISF, interstitial fluid; MARD, mean absolute relative difference; MRD, mean relative difference; PBS, phosphate buffered saline.

Disclaimer

The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any agency determination or policy. The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or implied endorsement of such products by Department of Health and Human Services.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: At the time of research, DS was an ORISE Fellow at CDRH/FDA, but is currently an employee at Dexcom, Inc.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by an appointment to the Research Participation Program at the Center for Devices and Radiological Health administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the US Department of Energy and the US Food and Drug Administration. Additionally, this project was supported by funding provided by the US FDA Critical Path Initiative administered through CDRH/OSEL/DBCMS.

References

Juvenile Diabetes Research Foundation Continuous Glucose Monitoring Study Group. Continuous glucose monitoring and intensive treatment of type 1 diabetes. N Engl J Med. 2008;359(14):1464-1476.

Hoeks

Greven

De Valk

Real-time continuous glucose monitoring system for treatment of diabetes: a systematic review. Diabet Med. 2011;28(4):386-394.

Joubert

Reznik

Personal continuous glucose monitoring (CGM) in diabetes management: review of the literature and implementation for practical use. Diabetes Res Clin Pract. 2012;96(3):294-305.

McGarraugh

The chemistry of commercial continuous glucose monitors. Diabetes Technol Ther. 2009;11:S17-S24.

Wang

Electrochemical glucose biosensors. Chem Rev. 2008;108(2):814-825.

Klonoff

DC.

Continuous glucose monitoring: roadmap for 21st century diabetes therapy. Diabetes Care. 2005;28(5):1231-1239.

Leal

Garcia Gabin

Bondia

, et al. Real-time glucose estimation algorithm for continuous glucose monitoring using autoregressive models. J Diabetes Sci Technol. 2010;4(2):391-403.

Bequette

BW.

Continuous glucose monitoring: real-time algorithms for calibration, filtering, and alarms. J Diabetes Sci Technol. 2010;4(2):404-418.

Tang

Louie

Kost

GJ.

Effects of drugs on glucose measurements with handheld glucose meters and a portable glucose analyzer. Am J Clin Pathol. 2000;113(1):75-86.

10.

US Food and Drug Administration—Center for Devices and Radiological Health. Guidance for Industry and Food and Drug Administration Staff: The Content of Investigational Device Exemption (IDE) and Premarket Approval (PMA) Applications for Artificial Pancreas Device Systems. Silver Spring, MD: US Food and Drug Administration—Center for Devices and Radiological Health; 2012.

11.

Lucarelli

Ricci

Caprio

, et al. GlucoMen Day continuous glucose monitoring system: a screening for enzymatic and electrochemical interferents. J Diabetes Sci Technol. 2012;6(5):1172-1181.

12.

Basu

Veettil

Dyer

Peyser

Basu

Direct evidence of acetaminophen interference with subcutaneous glucose sensing in humans: a pilot study. Diabetes Technol Ther. 2016;18:43-47.

13.

Gibson

Massey

Swoboda

BEP

. Kinetics and mechanism of action of glucose oxidase. J Biol Chem. 1964;239(11):3927-3934.

14.

Keilin

Hartree

EF.

Properties of glucose oxidase (notatin): addendum. Sedimentation and diffusion of glucose oxidase (notatin). Biochem J. 1948;42(2):221-229.

15.

Adams

Mast

Free

AH.

Specificity of glucose oxidase. Arch Biochem Biophys. 1960;91(2):230-234.

16.

Mislovicova

Patoprsty

Vikartovska

Enzymatic oxidation and separation of various saccharides with immobilized glucose oxidase. Appl Biochem Biotechnol. 2010;162(6):1669-1677.

17.

Sigma Aldrich. Glucose Oxidase from Aspergillus Niger. Available at: https://www.sigmaaldrich.com/content/dam/sigma-aldrich/docs/Sigma/Datasheet/7/g2133dat.pdf. Accessed February 3, 2016.

18.

Lee Biosolutions, Inc. Glucose Oxidase; 2013. Available at: https://www.leebio.com/content/productSpecifications/407. Accessed February 4, 2016.

19.

MP Biomedicals, LLC.Certificate of Analysis for #195196; 2015. Available at: https://www.mpbio.com/includes/msds/02195196/MP_COAT_02195196.pdf. Accessed February 3, 2016.

20.

Frias

Lim

Ellison

Montandon

CM.

Review of adverse events associated with false glucose readings measured by GDH-PQQ-based glucose test strips in the presence of interfering sugars. Diabetes Care. 2010;33(4):728-729.

21.

Schleis

TG.

Interference of maltose, icodextrin, galactose, or xylose with some blood glucose monitoring systems. Pharmacotherapy. 2007;27(9):1313-1321.

22.

Gaines

Pierce

Bernhardt

PA.

Fatal Iatrogenic Hypoglycemia: Falsely Elevated Blood Glucose Readings with a Point-of-Care Meter Due to a Maltose-Containing Intravenous Immune Globulin Product. FDA Safety and Availability Bulletin. Silver Spring, MD: US Food and Drug Administration; 2009.

23.

Cengiz

Tamborlane

A tale of two compartments: interstitial versus blood glucose monitoring. Diabetes Technol Ther. 2009;11(S1):S11-S16.

24.

Miller

Craig

Drucker

Woodward

The metabolism of fructose in man. Yale J Biol Med. 1956;29(3):335-360.

25.

Engerman

Kern

TS.

Experimental galactosemia produces diabetic-like retinopathy. Diabetes. 1984;33(1):97-100.

26.

Sjostrand

Gudbjornsdottir

Holmang

Lonn

Strindberg

Lonnroth

Delayed transcapillary transport of insulin to muscle interstitial fluid in obese subjects. Diabetes. 2002;51(9):2742-2748.

27.

Blanco

JG.

Sugar metabolism. Lactose, galactose, and xylose. J Biol Chem. 1928;79(2):6.

28.

Yoshikawa

Oda

Metabolism of maltose during surgery in patients with diabetes-mellitus under general-anesthesia. Res Exp Med. 1976;167(2):127-138.

29.

Wood

Cahill

GF.

Mannose utilization in man. J Clin Invest. 1963;42(8):1300-1312.

30.

Bornet

FRJ

Blayo

Dauchy

Slama

. Plasma and urine kinetics of erythritol after oral ingestion by healthy humans. Regul Toxicol Pharmacol. 1996;24(2):S280-S285.

31.

Guisado

Tourtellotte

Arieff

Tomiyasu

Mishra

Schotz

MC.

Rebound phenomenon complicating cerebral dehydration with glycerol: case report. J Neurosurg. 1975;42(2):226-228.

32.

Pitkänen

The serum polyol pattern and the urinary polyol excretion in diabetic and in uremic patients. Clin Chim Acta. 1972;38(1):221-230.

33.

Secchi

Pontiroli

Cammelli

Bizzi

Cini

Pozza

Effects of oral-administration of maltitol on plasma-glucose, plasma sorbitol, and serum-insulin levels in man. Klin Wochenschr. 1986;64(6):265-269.

34.

Dorman

Sondheimer

Cadnapaphornchai

Mannitol-induced acute-renal-failure. Medicine. 1990;69(3):153-159.

35.

Nau

Dreyhaupt

Kolenda

Prange

HW.

Low blood to cerebrospinal-fluid passage of sorbitol after intravenous-infusion. Stroke. 1992;23(9):1276-1279.

36.

Spitz

Rubenste

Bersohn

Bassler

KH.

Metabolism of xylitol in healthy subjects and patients with renal disease. Metabolism. 1970;19(1):24-34.

37.

Kovatchev

Anderson

Heinemann

Clarke

Comparison of the numerical and clinical accuracy of four continuous glucose monitors. Diabetes Care. 2008;31(6):1160-1164.

38.

Michaelis

Menten

ML.

Die kinetik der invertinwirkung. Biochem Z. 1913;49:333-369.

39.

Shu

Wilson

Rotating ring-disk enzyme electrode for surface catalysis studies. Anal Chem. 1976;48(12):1679-1686.

40.

Kamin

Wilson

GS.

Rotating ring-disk enzyme electrode for biocatalysis kinetic studies and characterization of the immobilized enzyme layer. Anal Chem. 1980;52(8):1198-1205.

41.

Sorochinskii

Kurganov

Steady-state kinetics of cyclic conversions of substrate in amperometric bienzyme sensors. Biosens Bioelectron. 1996;11(3):225-238.

42.

Lineweaver

Burk

The determination of enzyme dissociation constants. J Am Chem Soc. 1934;56(3):658-666.

43.

Ritchie

Prvan

Current statistical methods for estimating the K_m and V_max of Michaelis-Menten kinetics. Biochem Educ. 1996;24(4):196-206.

44.

Hofstee

Non-inverted versus inverted plots in enzyme kinetics. Nature. 1959;184:1296-1298.

45.

Hanes

CS.

Studies on plant amylases: the effect of starch concentration upon the velocity of hydrolysis by the amylase of germinated barley. Biochem J. 1932;26(5):1406.

46.

Eisenthal

Cornish-Bowden

The direct linear plot. A new graphical procedure for estimating enzyme kinetic parameters. Biochem J. 1974;139(3):715-720.

47.

Kemmer

Keller

Nonlinear least-squares data fitting in Excel spreadsheets. Nat Protoc. 2010;5(2):267-281.

48.

Bezerra

Fraga

Dias

AA.

Utilization of integrated Michaelis–Menten equations for enzyme inhibition diagnosis and determination of kinetic constants using Solver supplement of Microsoft Office Excel. Comput Methods Programs Biomed. 2013;109(1):26-31.

49.

Gadagkar

Call

GB.

Computational tools for fitting the Hill equation to dose–response curves. J Pharmacol Toxicol Methods. 2015;71:68-76.

50.

Kurganov

Lobanov

Borisov

Reshetilov

Criterion for Hill equation validity for description of biosensor calibration curves. Anal Chim Acta. 2001;427(1):11-19.

51.

Retama

Lopez-Ruiz

Lopez-Cabarcos

Microstructural modifications induced by the entrapped glucose oxidase in cross-linked polyacrylamide microgels used as glucose sensors. Biomaterials. 2003;24(17):2965-2973.

52.

McEnroe

Burritt

Powers

DM.

Interference Testing in Clinical Chemistry: Approved Guideline. CLSI; 2005.

53.

Food and Drug Administration.Blood Glucose Monitoring Test Systems for Prescription Point-of-Care Use; Draft Guidance for Industry and Food and Drug Administration Staff. Silver Spring, MD: US Food and Drug Administration; 2014.

54.

Silverstein

Goodney

DE.

Enzyme-linked biosensors: Michaelis-Menten kinetics need not apply. J Chem Educ. 2010;87(9):905-907.

55.

Mislovicova

Turjan

Vikartovska

Patoprsty

Removal of D-glucose from a mixture with D-mannose using immobilized glucose oxidase. J Mol Catal B. 2009;60(1-2):45-49.

56.

Kim

J-H

Cho

Bae

T-S

Lee

Y-S.

Enzyme biosensor based on an N-doped activated carbon fiber electrode prepared by a thermal solid-state reaction. Sens Actuators B: Chem. 2014;197:20-27.

57.

Feng

Cheng

Pan

Zheng

Development of glucose biosensors based on nanostructured graphene-conducting polyaniline composite. Biosens Bioelectron. 2015;70:411-417.

58.

Jung

Berlin

New water-soluble and film-forming aminocellulose tosylates as enzyme support matrices with Cu²⁺-chelating properties. Cellulose. 2005;12(1):67-84.

59.

Zhang

Huang

Dai

Wang

Fan

Tuning the redox and enzymatic activity of glucose oxidase in layered organic films and its application in glucose biosensors. Anal Biochem. 2004;329(1):85-90.

60.

Holland

Lau

Brozik

Atanassov

Banta

Engineering of glucose oxidase for direct electron transfer via site-specific gold nanoparticle conjugation. J Am Chem Soc. 2011;133(48):19262-19265.

61.

Seehuber

Dahint

Conformation and activity of glucose oxidase on homogeneously coated and nanostructured surfaces. J Phys Chem B. 2013;117(23):6980-6989.

62.

Cao

Zhang

Qian

Liu

Catalytic activity and stability of glucose oxidase/horseradish peroxidase co-confined in macroporous silica foam. Analyst. 2012;137(24):5785-5791.

63.

Szajani

Molnár

Klámar

Kalman

Preparation, characterization, and potential application of an immobilized glucose oxidase. Appl Biochem Biotechnol. 1987;14(1):37-47.

64.

Constantinides

Vieth

Fernandes

PM.

Characterization of glucose oxidase immobilized on collagen. Mol Cell Biochem. 1973;1(1):127-133.

65.

Kan

Chen

Jing

The galactose biosensor based on microporous polyacrylonitrile. Biocatal Biotransformation. 2005;23(6):439-444.

66.

Yang

Chen

Electrochemical synthesis of polypyrrole for the immobilization of galactose oxidase. Synth Met. 1998;92(2):173-178.

67.

Çevik

Şenel

Abasıyanık

MF.

Construction of biosensor for determination of galactose with galactose oxidase immobilized on polymeric mediator contains ferrocene. Curr Appl Phys. 2010;10(5):1313-1316.

68.

Jia

N-Q

Zhang

Z-R

Zhu

J-Z

Zhang

G-X.

A galactose biosensor based on the microfabricated thin film electrode. Anal Lett. 2003;36(10):2095-2106.

69.

Sung

Bae

YH.

Glucose oxidase, lactate oxidase, and galactose oxidase enzyme electrode based on polypyrrole with polyanion/PEG/enzyme conjugate dopant. Sens Actuators B: Chem. 2006;114(1):164-169.

70.

Semashko

Mikhailova

Eremin

Extracellular glucose oxidase of Penicillium funiculosum 46.1. Appl Biochem Microbiol. 2003;39(4):368-374.

71.

Witt

Singh

Kalisz

HM.

Structural and kinetic properties of nonglycosylated recombinant Penicillium amagasakiense glucose oxidase expressed in Escherichia coli. Appl Environ Microbiol. 1998;64(4):1405-1411.

72.

Leskovac

Trivic

Wohlfahrt

Kandrac

Pericin

Glucose oxidase from Aspergillus niger: the mechanism of action with molecular oxygen, quinones, and one-electron acceptors. Int J Biochem Cell Biol. 2005;37(4):731-750.

73.

Mogi

Sugai

Fuse

Funazukuri

. Infinite dilution binary diffusion coefficients for six sugars at 0.1 MPa and temperatures from (273.2 to 353.2) K. J Chem Eng Data. 2007;52(1):40-43.

74.

Godjevargova

Konsulov

Dimov

Vasileva

Behavior of glucose oxidase immobilized on ultrafiltration membranes obtained by copolymerizing acrylonitrile and N-vinylimidazol. J Membr Sci. 2000;172(1):279-285.

75.

Blandino

Macıas

Cantero

Immobilization of glucose oxidase within calcium alginate gel capsules. Process Biochem. 2001;36(7):601-606.

76.

Yang

Wang

Tan

Immobilization of glucose oxidase on chitosan–SiO₂ gel. Enzyme Microb Technol. 2004;34(2):126-131.

77.

Weiss

JN.

The Hill equation revisited: uses and misuses. FASEB J. 1997;11(11):835-841.