CHEMPHYSCHEM
REVIEWS
DOI: 10.1002/cphc.201300172
Continuous Glucose Monitoring: 40 Years, What We’ve
Learned and What’s Next
Raeann Gifford*[a]
In memory of Steven W. Gifford
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2013, 14, 2032 – 2044
2032
CHEMPHYSCHEM
REVIEWS
After 40 years of research and development, today continuous
glucose monitoring (CGM) is demonstrating the benefit it provides for millions with diabetes. To provide in vivo accuracy,
new permselective membranes and mediated systems have
been developed to prevent enzyme saturation and to minimize interference signals. Early in vivo implanted sensor research clearly showed that the foreign body response was
a more difficult issue to overcome. Understanding the biological interface and circumventing the inflammatory response
continue to drive development of a CGM sensor with accuracy
and reliability performance suitable in a closed-loop artificial
pancreas. Along with biocompatible polymer development,
other complimentary algorithm and data analysis techniques
have improved the performance of commercial systems signifi-
1. Introduction
The importance of glucose monitoring is well established for
people with diabetes. The incidence of diabetes is now estimated at 366 million people world wide (2011) and is expected
to reach 552 million by 2030.[2] Half of this population remains
undiagnosed, and this leads to an ever-growing financial
burden on already-stressed health care systems. The U.S.A.
alone spent 471 billion dollars on diabetes-related health care.
The need to improve quality of life for millions and the need
to reduce financial burden continue to drive glucose sensor research, and current projects strive to develop methods that are
discrete, painless, and accurate to help monitor and control
glucose levels and ultimately to reduce complications.
The 1993 landmark Diabetes Control and Complications Trial
(DCCT) was key to demonstrating that tight glucose control
would reduce the complications of diabetes.[3] The DCCT and
the United Kingdom Prospective Diabetes Study (UKPDS) further confirmed this outcome with type 2 diabetes patients.[4] It
was clear that a system that could aid in the maintenance of
tight glucose control would be beneficial. However, the studies
also showed an increase in hypoglycemic events while adhering to the tight glucose control regimen. Since those first studies, the benefits of better glycemic control and detection of
hypoglycemia have been confirmed in multiple studies by
using currently available commercial continuous glucose monitoring (CGM) systems.[5] In addition, subsequent studies with
CGM systems have demonstrated the value of CGM for applications in pediatrics, the intensive care unit, pregnancy, behavioral studies, and as an outcome measurement in clinical studies.[6]
As of this writing, there are three commercially available accurate and user-friendly CGM systems that provide continuous
glucose level information and act as an indicator for impending hypoglycemia. After more than 10 years since the first
[a] Dr. R. Gifford
Life Science
Acreo Swedish ICT AB
Box 787 SE-601 17 Norrkçping (Sweden)
E-mail: [email protected]
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
cantly. For example, the mean average relative difference of
Dexcom’s CGM system improved from 26 to 14 % and its uselife was extended from 3 to 7 d. Significant gains in usability,
including size, flexibility, insertion, calibration, and data interface, have been incorporated into new generations of commercial CGM systems. Besides Medtronic, Dexcom, and Abbott,
other major players are also investing in CGM. Becton Dickinson is conducting clinical trials with an optical galactose glucose binding system. Development of fully implanted sensor
systems fulfills the desire for a discreet, reliable CGM system.
Research continues to find innovative ways to help make living
with diabetes easier and more normal, and new segments are
being pursued (intensive care unit, surgery, behavior modification) in which CGM is being utilized.
CGM device was approved for patient use, many studies have
demonstrated reduced glycosylated hemoglobin (HbA1c) and
a reduction in hypoglycemic events by about 50 %.[7] Other
studies have concluded that the use of CGM in long-term projections is cost effective at the $100 000/QALY (quality-of-life
year) threshold.[5b] The ultimate goal in CGM has been to improve quality of life (QoL) and well-being for people with diabetes, and studies have shown improved well-being and
device satisfaction for CGM users, particularly in reducing fear
of hypoglycemia. However, ranking of QoL is similar between
CGM and traditional methods, which can be attributed to increased hassle of use (e.g. false alarms, insertion). Pediatric patients and their parents indicate more significant evidence that
CGM improves QoL and well-being with higher satisfaction.[7, 8]
It has taken more than 40 years from conception until these
positive benefits of CGM have been clearly demonstrated.
2. History
Clark and Lyons first introduced the concept of an electrochemical glucose sensor on the basis of Scheme 1 (GOx = glucose oxidase, FAD = flavin adenine dinucleotide).[8] Because
oxygen is ubiquitous in tissue, as well as in the atmosphere, it
acts as a very convenient mediator to regenerate the enzyme.
In the original concept, monitoring the consumption of
oxygen by reducing oxygen at a Pt electrode ( 0.6 V) produced a current that was proportional to the glucose concentration. For this mechanism, attributing changes in the oxygen
concentration only to reaction 2 (Scheme 1) meant that more
complex dual electrode systems to monitor ambient oxygen
might be required. One advantage of electrochemical sensors
is the relatively simple technology required for read-out elec-
Scheme 1. Glucose is specifically recognized with glucose oxidase. This reaction scheme produces the electroactive hydrogen peroxide species, and regeneration of glucose oxidase is mediated by oxygen.
ChemPhysChem 2013, 14, 2032 – 2044
2033
CHEMPHYSCHEM
REVIEWS
tronics and miniaturization. The concept of electrochemical
biosensors based on Clark’s original work took two paths: glucose strips for self-monitoring blood glucose (SMBG) and CGM.
Shichiri et al. took the first step toward CGM by implanting
a glucose sensor in dogs for 7 d.[9] However, the Shichiri team
utilized the oxidation of hydrogen peroxide, as shown in
Scheme 2. They were also the first to create a “needle-type”
Scheme 2. Oxidation of hydrogen peroxide at the Pt electrode (0.6 V)
sensor small enough (2 cm long, 0.4–1 mm in diameter) for
in vivo subcutaneous implant in humans.[10] Shichiri et al.
added a telemetry unit, measuring 4 6 2 cm and weighing
50 g, to create the first essential components for a wearable
CGM system.[11] Again in 1983, Shichiri et al. led the way
toward a wearable artificial pancreas when he coupled the
sensor with the an insulin pump that measured 12 15 6 cm
and weighed 400 g. Shichiri et al. was not the first to develop
an artificial pancreas. In 1977, Miles Laboratories produced the
Biostator, a relatively large bedside unit. The Biostator incorporates an in-line venous cannula to measure glucose and then
calculates the correct insulin and dextrose infusion rate. The Biostator is still used for conducting diabetes-related clinical research, often for controlling glucose levels in clamp studies for
CGM evaluation.
The Shichiri team demonstrated viability of several key components for a successful CGM system. They used hydrogen
peroxide oxidation rather than oxygen depletion, which simplified the system for practical use. The sensor was miniaturized
to facilitate subcutaneous implantation. Subcutaneous implantation was fundamentally essential because a portable, wearable system implanted in the blood proved technologically impractical due to risk of blood clotting. Vadgama et al. developed a system requiring continuous perfusion to circumvent
the clotting issue.[12] Coupling the system with telemetry
Raeann Gifford started her work with
CGM at iSense (acquired by Bayer Diabetes Care) in Portland, OR. Raeann received her Ph.D. in Bioanalytical
Chemistry from the University of
Kansas (KU) under the direction of
George S. Wilson. Her focus was CGM
sensor biocompatibility studies. Collaborating between Tokushima University
and TOYO Precision Parts Mfg. Co. Ltd.
(Japan), she worked to industrialize
a CGM sensor. She then moved to industry to lead an early-stage technology innovation group at Bayer
Diabetes Care in Tarrytown, NY. Raeann now directs the Life Science business development for Acreo Swedish ICT, a Swedish Research institute, and is a visiting scientist at Linkçping University,
Sweden.
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
added another key component that is now incorporated into
all the CGM systems that are currently commercially available.
Companies with approved CGM systems, and many with systems in development, have programs coupling CGM to insulin
delivery. Therefore, the early work by Shichiri’s team for a CGM
system combined with insulin delivery was central to the
future development of an artificial pancreas.
To improve treatment compliance and to create a more
“normal” life for people with diabetes, a fully implanted sensor
coupled to insulin delivery (artificial pancreas) is an essential
tool. The history of these efforts started with Updike’s group
who successfully obtained data for three months from a fully
implanted glucose sensor in dogs.[13] With these monumental
successes in the 1970s and 1980s, it seemed a commercial
CGM or artificial pancreas would be available very soon. Unfortunately, as research continued the fundamental technological issues became more evident.
3. Sensor Performance Issues
3.1. Electrochemical Systems
To function accurately for extended in vivo use (one week to
several months) a CGM sensor must meet fundamental performance characteristics: sensitivity, specificity, selectivity, linearity
within a biologically relevant range (3–20 mm), and lifetime.
For glucose sensors in biological tissue, impediments to meet
these characteristics are 1) interfering substances 2) oxygen dependence, and 3) biocompatible function. Fundamental research innovations to overcome these issues were plentiful, in
addition to algorithm development to compensate for inadequate performance. Ultimately it is a combination of these two
techniques that imparts the performance criteria of today’s
commercial CGM systems.
The accuracy of current CGM systems is commonly reported
as the mean average relative difference (MARD) relative to a reference glucose measurement (Yellow Springs laboratory glucose is often used), among several other factors. The accepted
average for regulatory approval is about 20 % over all biologically relevant glucose ranges. A correlation analysis, the Clarke
error grid analysis (EGA) is included in most studies. The chart
is divided into therapeutically relevant zones. Points in zones A
and B are considered therapeutically acceptable, zone C could
lead to unnecessary treatment, and zones D and E could lead
to potentially dangerous therapeutic decisions. The EGA was
originally designed to characterize SMBG individual points relative to laboratory methods. Because CGM reports continuous
blood glucose rather than discreet point values fixed in time,
the continuous glucose-EGA, which adds rate error grid analysis (REGA), was developed. The REGA includes therapeutic
error indications for rate and direction of change information
that is unique to CGM data. Kovatchev and Clarke et al. have
developed extensive CGM evaluation methods that account
for the rich data set that CGM systems produce.[14] Some of
these methods are included in the recommendations for evaluating and standardizing CGM clinical evaluations (guidance
ChemPhysChem 2013, 14, 2032 – 2044
2034
CHEMPHYSCHEM
REVIEWS
document) issued by the Clinical Laboratory and Standards Institute (CLSI).[15]
Specificity and selectivity are major figures of merit required
to meet accuracy requirements. The selected biological recognition element imparts the specificity, for which glucose oxidase has historically been the main actor. Selectivity, or eliminating reactivity to inferring substances, has been significantly
more challenging. For systems that use H2O2 oxidation as the
transduction mechanism, a relatively high polarization voltage
will also oxidize interfering species commonly found in vivo,
such as ascorbic acid, acetaminophen (paracetamol), and
urate. The design developed by Wilson et al., a two-electrode
needle-type sensor, overcame this issue by applying a multilayer cellulose acetate and Nafion permselective membrane.[16] Selectivity of permselective membranes is primarily achieved by
size and charge exclusion. Several other researchers have developed polymeric permselective membranes such as Vadgama
et al. with a sulfonated polyether–ether sulfone–polyether sulfone (SPEES-PES) membrane.[17] Others have applied electrodepositon of polypyrrole, phenylenediamine, and phenol to exclude interfering species, although the final product was often
not stable enough for reliable in vivo use.[18] Commercial CGM
developers incorporate proprietary membranes to create selective glucose sensors.
The permselective membrane concept is also used to reduce
sensitivity to changes in oxygen concentration. This is particularly important considering reaction (2) (Scheme 1), in which
oxygen limitation would cause enzyme saturation and nonlinearity at the upper end of the relevant range (3–20 mm). The
situation is compounded because the normal in vivo oxygen
level is 220 mm and glucose is 5 mm, which creates a natural
disparity for the stoichiometric 1:1 reaction. Limiting glucose
diffusion to create an environment with an excess amount of
oxygen has been realized by combinations of hydrophilic and
hydrophobic membranes of polyurethane, polytetrafluoroethylene, polycarbonate, and a variety of other polymers.[19, 20] A significant body of intellectual property (IP) has been created by
commercial manufacturers and researchers in this effort.[4, 21]
The second method to overcome selectivity issues is to
design a sensing system that does not require the high polarization voltage (0.6 V) to oxidize H2O2 and that does not use
oxygen as the enzyme regeneration mediator. This has been
achieved with a general mediated reaction shown in Scheme 3
(Medox = mediator in the oxidized form, Medred = mediator in
the reduced form).
www.chemphyschem.org
Borrowing technology that has been subsequently employed in the glucose strip industry, Updike et al. developed
a ferrocene derivative mediated system for implanted sensors.[22, 23] Producing a system for implantable sensors that is
safe and efficacious proved difficult due to the tendency for
the mediator to leach out of the polymer system. Heller’s
group overcame this by synthesizing a redox hydrogel of poly[(vinylpyridine)](bipyridine) with an osmium center that was
co-deposited with GOx.[22] A commercially utilized version of
this “wired” chemistry is incorporated into glucose monitoring
systems sold by Abbott.[23] Many of these technologies used to
overcome oxygen dependency and selectivity issues are employed in commercial CGM systems and on-going CGM development programs.
Although Updike’s group had success with his mediated
system in pigs, the impact of the inflammatory response and
its affect on sensitivity and long-term function was not yet understood.[24] A significant appreciation for the impact of the
biological response to in vivo sensor function was gained from
the earliest implanted glucose sensor research, both short and
long term. One of the first realizations was that sensors implanted in interstitial tissue would require calibration to equate
blood glucose concentration with the sensor response because
of the initial change in sensitivity from in vitro performance.
In vivo sensors do not produce equivalent in vitro and in vivo
response; thus, a finger-stick SMBG test is required to calibrate
the CGM sensor.[25] The biological response creates a continuously changing environment that is exhibited by changes in
background response and sensitivity; these changes are most
significant in the first 24 to 48 h, but they can be compensated
for with frequent calibration.
The calibration accuracy can be affected by a lag of 5 to
30 min between the blood glucose and interstitial tissue. The
lag was studied both in early research by Wilson et al. and in
clinical studies with Medtronic sensors by Boyne et al. as well
as many other studies too numerous to cite.[24] The lag varied
depending on the rate and direction glucose was changing;
thus, a poorly timed calibration could create significant error.
Multiple clinical studies have indicated that the lag is attributed to both sensor diffusion characteristics and physiological
factors.
The “biocompatibility” of implanted glucose sensors, after
years of research, remains the most challenging aspect to provide long-term accurate and reliable implanted glucose sensors. “Biocompatibility” in this case was aptly defined by Reichert and Sharkawy as minimal perturbation of the in vivo environment, and likewise, the in vivo environment does not adversely affect the sensor performance.[26]
3.2. Biological Response
Scheme 3. Mediated electron-transfer mechanism to minimize oxygen dependence and to reduce oxidation of interfering species. The regeneration
of the mediator to the oxidized form at the electrode, preferably at relatively
low potentials, produces a current that is proportional to the glucose concentration.
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
The in vivo response to implanted sensors continues to be one
of the greatest challenges to developing a CGM system with
appropriately accurate performance for use in a closed-loop artificial pancreas. Figure 1 illustrates the complex inflammatory
response when a sensor is inserted subcutaneously. The infiltration of proteins, the release of cytokines and reactive
ChemPhysChem 2013, 14, 2032 – 2044
2035
CHEMPHYSCHEM
REVIEWS
www.chemphyschem.org
and polyvinyl alcohol that form a shield of surface water,
which is thought to prevent protein adhesion. Other biomimetic compositions have also been formulated to deter protein
attachment, including collagen, chitosan, alginate, hyaluronan,
and phospholipids.[30] Work by Kobayashi et al. has shown the
relationship between nanofunctionalized surface characteristics
with poly(2-hydroxyethyl methacrylate) (PHEMA) polymer
brushes and the size exclusion of protein adsorption on in vivo
surfaces.[31] The packing density and orientation of the fibers
has a significant impact on the degree of effective protein exclusion, which is shown in Figure 2. Protein adhesion is essen-
Figure 1. Depiction of the complex inflammatory response cascade when
a foreign body (biosensor) is implanted in subcutaneous tissue.
oxygen species (ROS), and subsequent influx of inflammatory
cells all contribute to the instability of implanted glucose sensors. This includes an initial decrease in sensitivity, sensor detection of ROS, decomposition of sensor components, and consumption of oxygen and glucose by the inflammatory cells.[24]
The biological effort to heal the wound and eliminate the foreign body in the acute phase creates a very unstable and reactive environment. The unpredictable changes in the in vivo
milieu can create an unstable background response for electrochemical sensors.
In a foreign body response, the acute phase continues for
about one to two weeks before a fibrous avascular collagen
capsule is formed around the long-term implant. The capsule
inhibits diffusion of glucose and oxygen to the implant, which
is not representative of “normal” tissue. There is significant historical research to understand and circumvent the foreign
body reaction. With the success of the first commercial glucose
sensors, companies are focusing on the biocompatibility issue,
and researchers work toward greater understanding and solutions.
The complications encountered in the earliest devices have
driven many researchers to examine the interface between
sensor function and the biological response. Gerritson et al.
and Wilson et al. showed that the initial source of sensitivity
loss was protein infiltration into and onto the membranes of
the sensor.[24, 27] Wisneiwski et al. characterized the “biofouling”
process and worked toward developing methods to circumvent initial protein adhesion.[28] Anderson’s group contributed
significantly to the understanding of the foreign body response mechanism to implants. His group provided key understanding for the initial protein interaction with early inflammatory cells through the cascade of events leading to macrophage fusion into giant cells to encapsulate an implant.[29]
The fundamental understanding of the biological response
led to specific technologies to circumvent the adverse effects,
some of the first being to prevent protein adhesion. The concept is to mimic natural tissue or cellular structures and thus
to prevent protein adhesion. Morais et al. has reviewed many
of the hydrophilic hydrogel coatings of polyethylene glycol
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 2. Illustration of the size-exclusion effect for a) concentrated brush
and b) semiconcentrated brush nanostructures for in vivo protein exclusion:
d is the average distance between the nearest-neighbor graft points, D is
the average distance between the neighbor graft chains, and Rg is the
radius of gyration. Reprinted with permission from ref. [31]
tial in the inflammatory cascade for subsequent cellular adhesion, and as demonstrated by Kobayashi et al., the concentrated brush architecture effectively deters cellular adhesion. The
concentrated polymer brushes approach appears to be a very
effective method to improve biocompatibility. However, for
practical applications, fabricating the nanotextured surfaces on
CGM sensors in production manufacturing must be accomplished.
A more global solution to the foreign body reaction is to
mediate the inflammatory response. Several researchers have
contributed to this effort by release of microencapsulated antiinflammatory corticosteroids such as Dexamethasome. This has
shown some success for electrochemical and optical (“smart
tattoos”) applications.[28, 32] Safety concerns have been raised regarding the release of drugs or other substances that induce
angiogenesis, because of the possible correlation with cancer
mechanisms. The release of a naturally occurring biological
species, such as nitric oxide (NO), may have a simpler path to
regulatory success. Nitric oxide is a short-lived signaling molecule that has been demonstrated to mediate the inflammatory
response from glucose sensors. Figure 3 illustrates that a glucose sensor with short-term NO release effectively mediates
the inflammatory response.[33] In the work of Meyerhoff et al.
and Schoenfisch et al., NO appears to provide an in vivo environment that is similar to unperturbed tissue and induces angiogenesis.[33, 34] For optimum effectiveness on biological implants, Schoenfisch et al. developed a controlled release
system, which may be applied to mass production.
ChemPhysChem 2013, 14, 2032 – 2044
2036
CHEMPHYSCHEM
REVIEWS
Figure 3. Histological hematoxylin and eosin (H&E) stained tissue slice showing effect of short-term NO release from a subcutaneously implanted glucose sensor. a) Significant tissue necrosis and inflammatory cell influx (dark
purple); b) nearly normal tissue surrounding implant with NO release.
Inducing angiogenesis and disrupting the inflammatory response by blocking cytokine release are additional mechanisms
that researchers have employed to create more normal tissue
surrounding the sensor implant. Updike et al. developed coatings and topographical properties to promote angiogenesis,
whereas others used gene expression methods to induce the
generation of VEGF (vascular endothelial growth factor) to increase tissue vascularization.[32, 35] Inhibiting cytokine release
with surface moieties such as TGF-b (transforming growth
factor b) by using antisense genes or antibodies has shown
some success.[36]
A variety of the technologies described have been adapted
to achieve the best performance for the three currently approved sensors as well as for the next generation of CGM sensors being developed. Knowledge gained from following patient use and multiple clinical studies with available sensor systems (both approved and in development) has contributed to
the continual improvement in the performance of CGM. Performance reliability and accuracy are very important, but users
also highly value convenience and comfort. From a patient
perspective, it is often assumed that an approved device will
provide safe and effective performance, whereas the usability
can have a considerable impact on which device encourages
patient compliance. The size (sensor, body-worn transmitter,
and receiver), comfort, intrusiveness (alarms), and ease-of-use
(insertion, calibration, and data interpretation) of the CGM
system have occupied researchers since the first systems were
approved.
4. Commercial Continuous Glucose Monitoring
Systems
4.1. Evolution
Before describing what is currently available and what is next
on the horizon, a brief history of the evolution of the current
CGM system is appropriate. The first CGM system approved (in
2001), the GlucoWatch biographer by Cygnus, was considered
“minimally invasive”, as it did not puncture the skin.[37] However, the reverse iontophoresis sampling method created significant skin irritation. The system was also designed to shut
down if excess sweating was detected, which often occurs
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
with hypoglycemia. These two factors led to a commercially
nonviable product, and thus this model was withdrawn from
the market.
MiniMed (subsequently purchased by Medtronic) produced
the first approved system in 1999, which was a Holter-style
monitor for clinical use only. In 2005, they followed with
a home-use system, the Guardian. John Mastrototaro, who
picked up technology from an Eli Lilly product called the
“Direct 30/30”, brought the Guardian sensor technology to
MiniMed. This system used Updike’s H2O2 oxidation technique
to develop a reusable glucose finger stick system that was
commercially unsuccessful.[38] Dexcom, originally Markwell
Medical, was a company founded on the basis of Updike’s
technology for a fully implanted glucose sensor. Dexcom suspended work on the fully implanted system and surprised the
competition with a transcutaneous implanted sensor. The Dexcom STS (short-term sensor) was approved for home use in
2006. Adam Heller started Therasense with his “wired”
osmium-mediated technology and developed a glucose strip
system while working on a CGM system. Abbott purchased
Therasense, taking on both the strip and CGM technology.
After a long and arduous regulatory path, Abbott’s Navigator
was approved in 2008. The difficulty Abbott may have encountered was an effort to commercialize a nonadjunctive CGM
system, for which a therapeutic decision could be made on
the basis of the readings from the CGM. As of today, all the approved systems are adjunctive, and the Food and Drug Administration (FDA) requires labeling advising the patient to take
a blood glucose measurement with an approved system
before adjusting their therapy. The blood glucose finger stick is
also required to calibrate the systems periodically, currently
about once every 12 h, except for the Abbott Navigator, which
requires calibration less than once per day after initial calibration. A CGM “replacement” for SMBG is a common goal among
CGM companies
4.2. Initial Performance
The first generation of the Medtronic and Dexcom approved
systems reported an average MARD of 19.7 and 26 %, respectively, over the full range (40–400 mg dL 1), whereas Abbott’s
first-generation device reported an average MARD of about
12.8 %. All three of these companies have developed improvements in sensors and systems since their first offering. A comparison of the commercial CGM systems available at the time
of the clinical study, Abbott Navigator, Dexcom Seven Plus,
and Medtronic Guardian, was compiled from a closed loop
study published by Damiano et al. in 2013. It was reported
that the Navigator (5 d wear) had superior performance with
an average MARD of 11.8 3.8 % relative to the performances
of the Medtronic Guardian (3 d wear) and the Dexcom Seven
Plus (now 7 d wear) with an average MARD of 20.2 6.8 and
16.5 6.7 % respectively.[39] Figure 4 shows a representative
sampling of the performance of the three systems during the
study and it clearly illustrates this performance difference.
However, since that study was conducted, all three companies
ChemPhysChem 2013, 14, 2032 – 2044
2037
CHEMPHYSCHEM
REVIEWS
www.chemphyschem.org
Figure 4. a) Comparative data for a representative result from 1 of 12, 48 h closed-loop experiments. b) The 48 h
MARD for each sensor in the 12 experiments. Mean and standard deviations are superimposed on data for each
device. Reprinted with permission from ref. [39]
Figure 5. Medtronic Enlite sensor that can be coupled to the Paradigm Veo
and the redesigned sensor insertion device. Reprinted with permission from
Medtronic.
ment are also keenly focused
on creating CGM systems with
enhanced usability features.
Features that not only improve
accuracy but that also address
insertion, calibration convenience, baseline drift, reduce size,
power consumption, and data
communication.
The latest offer from the commercially marketed home-use
systems illustrates improvements in usability and safety.
The Medtronic Enlite sensor
system, available in Europe
since 2011 and expected for
FDA approval in early 2013, saw
a complete overhaul.[41] The
electrode layout was reconfigured to minimize potential
changes in oxygen tension
most likely due to inflammatory
response effects. Medtronic also
decreased the sensor size,
changed the implant angle (perpendicular to the skin surface),
extended wear time to 6 d, and
created a user-friendly inserter
device. All these changes signifi-
Figure 6. Dexcom G4 Platinum showing the receiver, transmitter, and insertion system with attached sensor. Reprinted with permission from ref. [43]
have both new sensors and systems.[7] The new systems are
shown in Figures 5 (Medtronic), 6 (Dexcom), and 7 (Abbott).
4.3. Current CGM Systems
The Medtronic Enlite and Dexcom G4 demonstrate significant
improvement over the versions reported in the Damiano team
study. Performance data for the Navigator 2 was not available.
Improved performance is most likely a result of sensor and
software algorithm redesign to overcome nonselectivity,
oxygen dependence, and in vivo instability.[40] The companies
with approved systems and those with systems in develop 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 7. Abbott Navigator 2 showing transmitter and receiver. Reprinted
with permission from Abbott.
ChemPhysChem 2013, 14, 2032 – 2044
2038
CHEMPHYSCHEM
REVIEWS
cantly improve usability and accuracy. Clinical studies with the
Enlite sensor by using the Paradigm Veo calibration routine reported an average MARD (over entire range) of 13.86 %. Not
only does the Enlite sensor substantially improve average performance over earlier versions, but by using Veo calibration
routines, the accuracy in the critical hypoglycemic region is improved over that of the Guardian by 3.5 %. At the 2013 Advanced Technologies and Therapeutics for Diabetes (ATTD)
meeting, Medtronic reported an improved Enlite sensor in
which they removed the cannula surrounding the sensor, developed a new polymer and Pt electrode materials, and reconfigured the connector design.[42] The Enlite Improved showed
a MARD of 12 % compared to 17 % for the original Enlite.
The Dexcom G4 system, FDA approved in October 2012,
offers a completely redesigned smaller sleek receiver and
a transceiver with a similar design but slightly larger. The
sensor also incorporates improvements, most notably a slimmer, more flexible wire that reportedly decreases risk of breakage, although there is still a warning on the label. The system
is reported to attain calibration stability after 1 h. According to
the user’s guide, the overall performance of the system for inhome use (40–400 mg dL 1) as compared to a YSI laboratory
glucose analyzer was 14.1 % MARD.[43] Dexcom also reported
the MARD as compared to the SMBG readings, which was
12.6 %. These reflect significant improvements over the first-approved generations. However, Dexcom still advises that inaccuracy may occur with the use of acetaminophen; therefore, it
seems they have not significantly changed the interference
screening technology. Dexcom is also working on G5, which
provides wireless communication to a smartphone; the G6,
a complete new design with true interference blocking; and
GlucoClear-2 for hospital use.[44] Early trials of the GlucoClear-2
showed a MARD of 5.2 %.[45]
Abbott introduced the Navigator 2 in a few European countries (e.g. Germany). Like Dexcom, Abbott completely redesigned the receiver, reduced the size of the transmitter, and
eliminated the replaceable battery.[35] A body-worn transmitter
unit without a sealed battery, labeled as functional in wet conditions, may have been a problem. The sensor size is slightly
reduced, with the same 1, 2, 10, 24, and 72 h calibration
scheme and use-life of 5 d as the Navigator 1.5. The clinicaltrials.gov website indicates if a study for FDA approval has
been completed, but performance data is not reported. The
Navigator 1.5 is still available in Europe, but distribution of the
original Navigator in the U.S.A. was discontinued in August
2011 due to supply issues.
Medtronic, Dexcom, and Abbott are not the only companies
to have commercial CGM systems. The other systems are not
available for home use. Marco Mascini developed a microdialysis system, the GlucoDay, which is now marketed by A. Menarini for physician and research use only.[46] Via Medical, Inc., also
sells the Glucoscout to measure blood glucose from an arterial
or venous catheter in a hospital or research setting.[47] Glycemic control in a hospital setting, particularly critical care, has
been recognized as a vital factor in improving outcome.[6a, 48]
Therefore, several companies are investing in the development
of hospital monitoring systems including Dexcom (Gluco 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
Clear 2) and Medtronic (CE-approved Sentrino). Companies
without commercial CGM systems are also developing hospital
systems as described in the following section.
5. Next Generation
In the realm of CGM, enumerable technologies have been developed that have been reviewed in plenary talks (e.g. Jay
Skyler)[49] and recent papers.[42, 50] After more than 40 years, the
driving force behind the ongoing research and development
for superior CGM sensors is the desire to provide diabetes
management tools that are more convenient, accurate, smaller,
and less painful and obtrusive. These features will increase
user satisfaction, which results in better compliance and, therefore, reduced complications and higher QoL.[6b, 8a] Another compelling reason for continued research is to achieve a true
closed-loop artificial pancreas. The Juvenile Diabetes Research
Foundation (JDRF) founded the Artificial Pancreas Project in
2005, which provides guidance and funding toward developing a closed-loop system to deliver insulin automatically on
the basis of blood glucose concentration. A great deal of research has contributed to reaching that goal, which requires
an accurate and reliable glucose sensor.[40b, 51] A few of the
more significant efforts to develop in vivo glucose sensors by
other companies and researchers with some likelihood of success are presented.
5.1. Electrochemical
At the 2013 ATTD conference, Medtronic revealed multiple
new technical development efforts to improve sensor performance. They showed an integrated infusion cannula and
sensor by using a split needle requiring only one insertion for
co-located insulin infusion and CGM. An early study showed
little impact on the performance of the sensor during bolus injections with a reported MARD of 12.1 %. The most innovative
new sensor development from Medtronic is the double-sided
redundant individually addressed sensor system, which compensates for tissue heterogeneity. Medtronic applies an electrochemical impedance spectroscopy (EIS) system that periodically interrogates each sensor to ascertain function within a factory-set threshold limit, and thereby, erroneous signals from
CGM analysis can be eliminated.[42] Medtronic proposes that
this may lead to replacement technology that is no longer adjunctive and possibly providing factory calibration.
In addition to companies with approved systems, other
groups are pursuing electrochemical CGM systems. Bayer acquired iSense, Inc., and is continuing to develop a system
based on the Wilson et al. needle-type technology.[52] Bayer
Diabetes Care reported the clinical results of their new CGM
system at the 2011 Diabetes Clinical Technology Meeting.[53] Although they have several patents, there is no additional information about future regulatory submission or commercialization plans.[54] Glysens continues to develop a fully implanted
system based on the work of Gough et al.[55] The system uses
the differential measurement of oxygen reduction with a nonlinear calibration scheme. Roche Diagnostics, at the 2013
ChemPhysChem 2013, 14, 2032 – 2044
2039
CHEMPHYSCHEM
REVIEWS
ATTD, reported preliminary clinical results from a novel needletype CGM sensor. There were three key design components
that allowed Roche to achieve a MARD of 8.6 % compared to
SMBG and 8.1 % as compared to a laboratory method.[56] The
first design component is the use of multiple WEs (working
electrodes) (redundancy) to reduce effects of tissue heterogeneity. Roche also reported the development of a biocompatible
membrane and the use of a catalyst for electrochemical transduction. Of particular interest is that besides for the calibration
of CGM data no compensation for lag was used. This supports
the fact that as sensor performance improves due to greater
understanding of sensor design characteristics relative to biological response, the majority of lag for short-term implanted
sensors (less than 7 d) can be attributed to sensor characteristics and software smoothing anomalies. Review of the clinicaltrials.gov website does not indicate they are conducting clinical trials yet, and no information was available about the technology.
5.2. Optical Detection
Optical detection methods can overcome some of the challenges of electrochemical transduction, but they also introduce
new issues. Some of the earliest optical glucose detection
technologies with potentially viable mechanisms were conducted by Lackowitz et al., Cote et al., and Ballerstadt et al.,
which evolved into some of the commercial development projects described below.[57] Many of the optical technologies that
are reviewed by McNichols et al. are targeted for noninvasive
detection (polarimetry, near-IR, and Raman), which is not the
focus of this review. However, modified versions of the fluorescent techniques described can be more effectively adapted to
implanted systems.[58]
The biological recognition utilized in the optical technologies reviewed here is based on equilibrium rather than analyte
consumption. The equilibrium glucose binding mechanism
does not consume analyte that may deplete the analyte concentration in the tissue immediately surrounding the sensor.
The equilibrium condition is established quickly, runs continuously, and does not require re-establishment of a diffusion gradient for the signal transducer to make a measurement. This
offers an “instant on” condition, for which the reader (transmitter) can be disconnected and reconnected without a relatively
lengthy stabilization period, as required by electrochemical systems. Optical or spectroscopic methods often have inherently
higher sensitivity and less in vivo interference than electrochemical mechanisms. However, the system design is critical to
avoid detection of tissue autofluorescence. The development
of glucose-specific optical binding systems in biologically relevant dissociation constant (KD) ranges, in addition to the availability of miniaturized robust optical detection and electronic
systems offer a realistic path for optical CGM commercialization. The benefits compared to electrochemical include greater
accuracy, “instant on”, and minimal interference.
Optical detection technology research and development is
quite active, and several companies are demonstrating good
clinical function. Becton Dickinson (BD) has demonstrated
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
in vivo functionality of an implanted sensor on the basis of
a mutated galactose glucose binding protein (GGBP). The mutated GGBP is modified with derivatives of Red Nile fluorescent
dye to create glucose-specific binding in an appropriate range
(KD = 7.4 mm). The conformation change upon binding produces a fluorescence resonance energy transfer (FRET) between
the fluorophores, which thus conveys optical transduction,
shown in Figure 8.[59]
Figure 8. The structures of GGBP in the open (2FW0.pdb) and closed
(2FVY.pdb) state. The 34 kDa protein has two large domains connected by
a hinge region. Upon glucose binding, a large conformational change (318)
occurs, which brings the two domains together and entraps glucose. Reprinted with permission from ref. [60]. Copyright (2011) American Chemical
Society.
In 2011, BD published a study in which the sensor was implanted in subcutaneous tissue and intradermal dermal tissue
to produce a calibration signal 15 min after implant. By using
a predetermined time constant to adjust for time lag, t, the
median average percent error (MAPE) was about 10 % or less
across all glucose ranges.[61] At the 2013 ATTD conference, BD
reported a 24 h pilot clinical study by using a prototype optical
GGBP device. The median absolute relative difference
(MedARD) of 12.4 % overall was consistent in all glucose regions. The relatively short 15 min warm up was again reported.
Notably, there was no significant lag time (< 1 min), which indicates that previous lag issues may be related more to sensor
design than to physiological phenomena.[62] A search of the
clinicaltrials.gov website reveals that BD has completed at least
one clinical trial for their optical GGBP system and is recruiting
to study a second-generation device; this is indicative of their
active effort to commercialize the product. El-Laboudi et al. reviewed a significant body of work targeting microneedles for
CGM, a technology in which BD owns significant IP.[63] The reported sensor uses fiber optic technology; however, BD may
incorporate microneedles in a CGM system with insulin delivery, an area in which BD Technology has reported some success.[64]
In vivo fiber optic sensors have been developed with boronic acid derivatives coupled with fluorescence derivatives that
are quenched in the presence of glucose. Glumetrics uses a version of the technology developed by Gamsey et al.[65] The
system is intended for vascular implant, and Glumetrics supports many patents applicable for a CGM sensor.[66] Recent
in vivo studies have shown that excellent MARD ( 10 %) is
possible in a vascular-implanted sensor. However, this low
MARD was only demonstrated in 60 % of the sensors.[67] The inChemPhysChem 2013, 14, 2032 – 2044
2040
CHEMPHYSCHEM
REVIEWS
tended hospital application by using a vascular implant is particularly difficult with the added complexity of a thrombotic response. Therefore, it may be projected that Glumetrics may develop a subcutaneous implanted version of this sensor.
Several other optical technologies based on variations of
binding proteins have shown promising results. Ballerstadt
et al. employs Concanavalin A, a lectin, with a fluorescent-labeled dextran in a competitive assay.[68] An extension of that
system in a short-term (4 h) pilot study by BioTex, Inc., by
using retrospective calibration gave a MARD of 13 15 %.[69]
Spectroscopic systems that use intensity to measure the
change in glucose concentration can suffer from biological
bleaching phenomena. A method to overcome this drawback
was developed by Lacowitz et al. by using time-resolved fluorescence detection.[57a, 70] A phase shift in the fluorescence lifetime is measured when the glucose binding equilibrium shifts
with concentration. John Pickup, who has also developed boronic acid glucose detection systems, has utilized time-resolved
fluorescence on a fiber optic sensor.[59] For their system, Pickup
et al. created a site-mutated GGBP that changes the glucose KD
to 11 mm, which is a more relevant biological range than the
native protein.[71, 72]
www.chemphyschem.org
Figure 9. Senseonics glucose sensor system. Top) The wireless miniature implantable glucose sensor, which is fully implanted just under the skin, and
the reader platform, which is inductively coupled directly over the implant
site. Bottom) Schematic of the sensor function. Reprinted with permission
from ref. [77]
5.3. Fully Implanted
The ultimate objective for commercial CGM systems is to make
them reliable, in addition to discreet and worry free. A fully implantable CGM system could certainly be discreet, and with adequate accuracy and reliability, it could work in an artificial
pancreas. The goal is a sensor with a lifetime of 6–12 months
that slides inside a small incision that is made in a doctor’s
office. Several of the earliest CGM attempts targeted long-term
implants, including unpublished commercial efforts by Medtronic (Minimed) in the vascular bed and Dexcom (Markwell
Medical) with a fully implanted device. Adverse effects from
the biological response significantly impeded progress toward
commercialization. Biocompatibility is key for a long-term implant to be successful, and the power source must also be
long lived, as changing batteries periodically would be another
invasive procedure. One company that has achieved excellent
results in this realm is Senseonics (formerly Sensors for Science
and Medicine). The device is powered through coupling with
an external induction coil, which therefore eliminates the battery. They use a variation of the anthracene-derivatized diboronic acid chemistry developed by James et al.[73] The company
has demonstrated 29 d of continuous in vivo data from a site
in the upper arm with a MARD of 12.4 0.7 %; the device is
shown in Figure 9.[74] To function long term, a fully implanted
system must surmount a hostile chronic inflammatory in vivo
environment, in which the components are continuously subjected to species designed to breakdown the foreign body.
Senseonics utilized a unique method to circumvent break
down of the boronic acid recognition element coupled to the
fluorescent transduction molecule. The researchers sputter
coated a sacrificial Pt layer to decompose the ROS (H2O2) produced during the inflammatory response, which resulted in
minimal signal loss over 29 d.[75]
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Precisense developed a proprietary mutated glucose binding
lectin in a competitive glucose binding concept, similar to the
Concanavalin A/Dextran system BioTex, Inc., has demonstrated.
The fluorescence lifetime is measured from a FRET signal rather
than intensity.[76] The original fully implanted sensor was designed to biodegrade over time and thus did not require an
explant procedure, which may be difficult to get regulatory approval. Medtronic acquired Precisense in 2009 and is now collaborating with JDRF on an orthogonal redundant system incorporating both the electrochemical redundant system reported at the 2013 ATTD meeting and optical technology,
quite likely by using the Precisense IP.[42]
Positive ID Corporation developed a fully implanted CGM
sensor called the GlucoChip by using radio-frequency identification wireless technology. Receptors LLC used their patented
platform to create an optimized dendrite–boronic acid glucose
binding system.[78] Smart tattoos implanted just beneath the
skin are novel but are farther from commercialization. Competitive binding FRET and near-infrared (NIR) tattoos have been
demonstrated.[79, 80] These would require an optical reader
through the skin, but anomalies can occur due to variations in
skin characteristics between patients, in addition to potential
scattering effects. Another significant concern for fully implanted CGM sensors is compensating for lag. Although short-term
implanted sensors are reporting significant decrease in lag because of design improvements, long-term implants will be affected by capsule formation and the lack of capillary formation
near the sensor. Novak et al. modeled the effect of the biological response components relative to sensor lag and showed
how capsule thickness and diffusion from vessels affects lag
time.[81]
ChemPhysChem 2013, 14, 2032 – 2044
2041
CHEMPHYSCHEM
REVIEWS
To avoid the effects of subcutaneous inflammatory response,
implants in the conjunctiva in the eye have been reported
with some success.[82] EyeSense technology uses ConA/Dextran
FRET technology and has recently demonstrated clinical success with a 16 week implant in the eye conjunctiva. The current version is not CGM, but it does satisfy the desire to eliminate blood glucose finger sticks. Another popular mechanism
for “noninvasive” CGM is the development of contact lenses
that measure glucose in tears or possibly intraocular lenses.
Polymeric optical concepts and electrochemical mechanisms
with miniaturized electronics have been produced.[83] Significant controversy regarding the correlation of tears to glucose
concentration for daily use in real-life conditions creates skepticism for this approach.[84]
5.4. Next-Generation Optical Challenges
It seems like the optical technologies outlined could be the
answer to the quest for an accurate, discreet, and worry-free
CGM system. However, optical methods still present significant
challenges. Many of the fluorophores adopted can be toxic
and thus require careful design and testing to insure safety. In
addition, some of the recognition elements are toxic, such as
Concanavalin A, and both Concanavalin A and boronic acid derivatives can be unstable. The fluorophore signal may have
issues with stability due to photobleaching, pH, background
autofluorescence, or degradation due to biological oxidative
species. A primary analytical problem is analyte binding in
a biologically relevant range. To insure this, all of the publications for optical sensors describe an effort to design a recognition element with binding affinity in an appropriate range. Although much success has been achieved, these proprietary
modifications add an element of complexity to the manufacturing process. Even though advances in optics and electronics
will now make appropriate miniaturization and cost targets
possible, implementing those designs is challenging. Electronics for electrochemical detection is relatively simple compared
to the robust alignment and signal analysis required for
a home-use optical CGM system. These challenges and the associated benefits, in addition to conquering the foreign body
response, is why keen focus on optical CGM development continues. However, for the optical systems to take a dominant
role, the issues described must be overcome and configured in
a small unobtrusive package.
The optical and electrochemical techniques described only
make a small dent in all the research involving glucose detection. In some cases, glucose is used as a model system to
prove the concept, and other times, glucose detection is the
end goal. A few of the other early-stage technologies being investigated for glucose detection include spectroscopy, pressure, quartz crystal microbalance (QCM), field-effect transistors
(FET), nanomaterials, holographic sensors, and other more esoteric methods.
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
6. Summary
Since the initial work of Clark and Lyons, the field of CGM has
advanced significantly. There are now three approved systems
available, from Medtronic, Dexcom, and Abbott. The research
continues to improve the systems both in analytical accuracy
and usability. The level of quality from the CGM systems has
provided a significant body of clinical work understanding the
impact of glucose variation and patient health. These studies
have demonstrated that the wait for a working system has
provided benefits in lower HbA1c values and hypoglycemic
awareness. The systems have not reached their peak performance, and therefore, researchers continue to overcome negative biological response effects and to design systems based
on new transduction methods. Optical equilibrium binding systems using FRET and time-resolved fluorescence for transcutaneous implants and full implants seem to be close to showing
performance levels worthy of regulatory approval. Many novel
technologies have been applied to improve the stability of
in vivo performance, but biocompatibility continues to be
a major challenge. Physical and chemical technology has been
applied to improve the systems, but these technologies have
also been combined with algorithms that compensate for potential raw signal issues and have been applied with success.
The desire for a closed-loop artificial pancreas continues the
drive to develop more accurate, reliable, and convenient CGM
sensors and to find innovative ways to couple those with insulin delivery.
7. Outlook
As of today, there are three companies that have CGM systems
on the market. They are competing in a relatively small
market. However, soon other key players with completely new
systems will join them. Optical detection will become a player
in the CGM sensor market that is currently dominated by electrochemical transduction. As clinical research continues and as
the value of CGM becomes more evident, the CGM sensor will
move from a niche product primarily targeted for type 1 diabetes patients using a pump and become more dominant for
multiple daily injection (MDI) type 1 and type 2, type 2 to understand glycemic variation, hospital use, intensive care unit,
pregnancy, and other nontraditional applications. As the fully
implanted sensor gains performance, implant comfort, and usability enhancements, it may become the most discreet option
for many. The goal for many in the CGM business is to see
CGM replace SMBG as the option of choice for the growing
population with diabetes.
From a research perspective, glucose will still continue to be
a model system to develop novel technologies to make analytical measurements. The drive to create spin-off companies targeting CGM system development will continue because of the
huge market. However, CGM from an academic perspective is
no longer a hot topic. There may be renewed interest in noninvasive continuous or discreet monitoring, which continues to
be the Holy Grail. The current focus is to perfect a mechanical
artificial pancreas, for which research from an academic and inChemPhysChem 2013, 14, 2032 – 2044
2042
CHEMPHYSCHEM
REVIEWS
www.chemphyschem.org
dustry perspective is quite strong. The hope is that the regulatory bodies will gain confidence and knowledge enabling
them to approve and release ever better tools for people with
diabetes faster and more efficiently.
Acknowledgements
I want to thank Jiangfeng Fei of JDRF for his review and insightful comments.
Keywords: biocompatibility
·
carbohydrates
electrochemistry · glucose · optical detection
·
[1] B. P. Kovatchev, Scientifica 2012, 2012, 14.
[2] Fifth Edition Committee, Diabetes Atlas in Diabetes Atlas, 5th ed. (Eds.:
D. W. Nigel Unwin, L. Guariguata, G. Ghyoot, D. Gan), International Diabetes Federation, Brussels, Belgium, 2012.
[3] The Diabetes Control and Complications Trial Research Group, N. Engl.
J. Med. 1993, 329, 977 – 986.
[4] B. Van Antwerp, Siloxane-Containing Biocompatible Polyurethane Membranes Suitable for In Vivo Use. 96-US15386 9813685, 9960926, 1998.
[5] a) G. Y. Gandhi, M. Kovalaske, Y. Kudva, K. Walsh, M. B. Elamin, M. Beers,
C. Coyle, M. Goalen, M. S. Murad, P. J. Erwin, J. Corpus, V. M. Montori,
M. H. Murad, J. Diabetes Sci. Technol. 2011, 5, 952 – 965; b) E. S. Huang,
M. O’Grady, A. Basu, A. Winn, P. John, J. Lee, D. Meltzer, C. Kollman, L.
Laffel, W. Tamborlane, S. Weinzimer, T. Wysocki, Diabetes Care 2010, 33,
1269 – 1274.
[6] a) C. R. Sampaio, D. R. Franco, D. J. Goldberg, J. Baptista, F. G. Eliaschewitz, Diabetes Technol. Ther. 2012, 14, 117 – 124; b) N. Mauras, R. Beck,
D. Xing, K. Ruedy, B. Buckingham, M. Tansey, N. H. White, S. A. Weinzimer, W. Tamborlane, C. Kollman, Diabetes Care 2012, 35, 204 – 210; c) T. L.
Hernandez, L. A. Barbour, Diabetes Technol. Ther. 2012, 15, 172 – 179;
d) R. W. Beck, P. Calhoun, C. Kollman, Diabetes Technol. Ther. 2012, 14,
877 – 882; e) J. Wagner, H. Tennen, H. Wolpert, Psychosom. Med. 2012,
74, 356 – 365.
[7] P. Y. Benhamou, B. Catargi, B. Delenne, B. Guerci, H. Hanaire, N. Jeandidier, R. Leroy, L. Meyer, A. Penfornis, R. P. Radermecker, E. Renard, S. Baillot-Rudoni, J. P. Riveline, P. Schaepelynck, A. Sola-Gazagnes, V. Sulmont,
N. Tubiana-Rufi, D. Durain, I. Mantovani, Diabetes Metab. 2012, 38, S67 –
S83.
[8] a) M. Tansey, L. Laffel, J. Cheng, R. Beck, J. Coffey, E. Huang, C. Kollman,
J. Lawrence, J. Lee, K. Ruedy, W. Tamborlane, T. Wysocki, D. Xing, Diabetes Med. 2011, 28, 1118 – 1122; b) J. M. Lawrence, Diabetes Care 2010,
33, 569 – 576.
[9] M. Shichiri, R. Kawamori, Y. Yamasaki, N. Hakui, H. Abe, Lancet 1982,
320, 1129 – 1131.
[10] M. Shichiri, R. Kawamori, Y. Goriya, Y. Yamasaki, N. Hakui, H. Abe, Int.
Congr. Ser. Excerpta Med. 1983, 607, 339 – 346.
[11] a) M. Shichiri, N. Asakawa, Y. Yamasaki, R. Kawamori, H. Abe, Diabetes
Care 1986, 9, 298 – 301; b) M. Shichiri, R. Kawamori, Y. Goriya, Y. Yamasaki, M. Nomura, N. Hakui, H. Abe, Diabetologia 1983, 24, 179 – 184.
[12] a) G. P. Rigby, P. W. Crump, P. Vadgama, Analyst 1996, 121, 871 – 875; b) I.
Hanning, P. Vadgama, A. K. Covington, K. G. M. M. Alberti, Anal. Lett.
1986, 19, 461 – 478.
[13] a) B. J. Gilligan, M. C. Shults, R. K. Rhodes, S. J. Updike, Diabetes Care
1994, 17, 882 – 887; b) M. C. Shults, R. K. Rhodes, S. J. Updike, B. J. Gilligan, W. N. Reining, IEEE Trans. Biomed. Eng. 1994, 41, 937 – 942; c) S. J.
Updike, M. Shults, B. Ekman, Diabetes Care 1982, 5, 207 – 212.
[14] B. Kovatchev, M. Breton, W. Clarke, Methods Enzymol. 2009, 454, 69 – 86.
[15] CLSI, 1st ed. (Ed.: M. David Klonoff, FACP), Clinical and Laboratory Standards Institute (CLSI), 2008, p. 64.
[16] a) Y. Zhang, Y. Hu, G. S. Wilson, D. Moatti-Sirat, V. Poitout, G. Reach,
Anal. Chem. 1994, 66, 1183 – 1188; b) D. Moatti-Sirat, F. Capron, V. Poitout, G. Reach, D. S. Bindra, Y. Zhang, G. S. Wilson, D. R. Thevenot, Diabetologia 1992, 35, 224 – 230.
[17] Y. Benmakroha, I. Christie, M. Desai, P. Vadgama, Analyst 1996, 121,
521 – 526.
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
[18] a) D. Centonze, A. Guerrieri, C. Malitesta, F. Palmisano, P. G. Zambonin,
Fresenius J. Anal. Chem. 1992, 342, 729 – 733; b) I. M. Christie, P. Vadgama, S. Lloyd, Anal. Chim. Acta 1993, 274, 191 – 199; c) J. P. Lowry, K.
McAteer, A. S. S. El, A. Duff, R. D. O’Neill, Anal. Chem. 1994, 66, 1754 –
1761; d) R. Vaidya, E. Wilkins, Med. Eng. Phys. 1994, 16, 416 – 421.
[19] G. Reach, G. S. Wilson, Anal. Chem. 1992, 64, 381A – 386A.
[20] E. Grunstein, P. Abel, A. Gens, K. Eich, W. T. Von, Biomed. Biochim. Acta
1989, 48, 911 – 917.
[21] a) D. J. Allen, K. W. Johnson, R. S. Nevin, Hydrophilic Polyurethane Membranes
for
Electrochemical
Glucose
Sensors.
92-308837
535898,19920928., 1993; b) W. P. Van Antwerp, Polyurethane/Polyurea
Compositions Containing Silicone for Biosensor Membranes. 96-US3987
9630431,19960325., 1996.
[22] M. V. Pishko, A. C. Michael, A. Heller, Anal. Chem. 1991, 63, 2268 – 2272.
[23] A. Heller, B. Feldman, Chem. Rev. 2008, 108, 2482 – 2505.
[24] V. Thome-Duret, M. N. Gangnerau, Y. Zhang, G. S. Wilson, G. Reach, Diabetes Metab. 1996, 22, 174 – 178.
[25] M. S. Boyne, J. Kaplan, C. D. Saudek, Diabetes 2003, 52, 2790 – 2794.
[26] W. M. Reichert, A. A. Sharkawy in Handbook of Biomaterials Evaluation:
Scientific, Technical, and Clinical Testing of Implant Materials, 2nd ed.,
Taylor and Francis, Philadelphia, 1999, vol. 28, pp. 439 – 460.
[27] M. Gerritsen, J. A. Jansen, J. A. Lutterman, Neth. J. Med. 1999, 54, 167 –
179
[28] a) N. Wisniewski, F. Moussy, W. M. Reichert, Fresenius J. Anal. Chem.
2000, 366, 611 – 621; b) N. Wisniewski, N. Rajamand, U. Adamsson, P. E.
Lins, W. M. Reichert, B. Klitzman, U. Ungerstedt, Am. J. Physiol. Endocrinol. Metabol. 2002, 282, E1316 – E1323; c) N. Wisniewski, M. Reichert,
Colloids Surf. 2000, 18, 197 – 219.
[29] a) J. M. Anderson, K. M. Miller, Biomaterials 1984, 5, 5 – 10; b) J. M. Anderson, N. P. Ziats, A. Azeez, M. R. Brunstedt, S. Stack, T. L. Bonfield, J. Biomater. Sci. Polym. Ed. 1996, 7, 159 – 169; c) T. L. Bonfield, E. Colton,
J. M. Anderson, J. Biomed. Mater. Res. 1991, 25, 165 – 175; d) W. G. Brodbeck, G. Voskerician, N. P. Ziats, Y. Nakayama, T. Matsuda, J. M. Anderson,
J. Biomed. Mater. Res. Part A 2003, 64A, 320 – 329; e) W. G. Brodbeck,
J. M. Anderson, Curr. Opin. Hematol. 2009, 16, 53; f) S. Chen, J. A. Jones,
Y. Xu, H.-Y. Low, J. M. Anderson, K. W. Leong, Biomaterials 2010, 31,
3479 – 3491.
[30] a) R. Gifford, J. J. Kehoe, S. L. Barnes, B. A. Kornilayev, M. A. Alterman,
G. S. Wilson, Biomaterials 2006, 27, 2587 – 2598; b) J. M. Morais, F. Papadimitrakopoulos, D. J. Burgess, AAPS J. 2010, 12, 188 – 196.
[31] C. Yoshikawa, S. Hattori, T. Honda, C.-F. Huang, H. Kobayashi, Mater. Lett.
2012, 83, 140 – 143.
[32] a) T. Hickey, D. Kreutzer, D. J. Burgess, F. Moussy, J. Biomed. Mater. Res.
2002, 61, 180 – 187; b) U. Bhardwaj, R. Sura, F. Papadimitrakopoulos,
D. J. Burgess, Int. J. Pharm. 2010, 384, 78 – 86.
[33] R. Gifford, M. M. Batchelor, Y. Lee, G. Gokulrangan, M. E. Meyerhoff, G. S.
Wilson, J. Biomed. Mater. Res. Part A 2005, 75A, 755 – 766.
[34] a) J. H. Shin, S. M. Marxer, M. H. Schoenfisch, Anal. Chem. 2004, 76,
4543 – 4549; b) M. H. Schoenfisch, K. A. Mowery, M. V. Rader, N. Baliga,
J. A. Wahr, M. E. Meyerhoff, Anal. Chem. 2000, 72, 1119 – 1126; c) J. P.
Cooke, Atheroscler. Suppl. 2003, 4, 53 – 60; d) S. P. Nichols, A. Koh, N. L.
Brown, M. B. Rose, B. Sun, D. L. Slomberg, D. A. Riccio, B. Klitzman, M. H.
Schoenfisch, Biomaterials 2012, 33, 6305 – 6312.
[35] S. J. Updike, M. C. Shults, B. J. Gilligan, R. K. Rhodes, Diabetes Care 2000,
23, 208 – 214.
[36] W. K. Ward, H. M. Duman, Chem. Anal. 2010, 174, 59 – 86.
[37] a) R. O. Potts, J. A. Tamada, M. J. Tierney, Diabetes Metab. Res. Rev. 2002,
18, S49 – S53; b) M. J. Tierney, J. A. Tamada, R. O. Potts, L. Jovanovic, S.
Garg, Biosens. Bioelectron. 2001, 16, 621 – 629.
[38] S. J. Updike, M. C. Shults, C. C. Capelli, H. D. von, R. K. Rhodes, N.
Joseph-Tipton, B. Anderson, D. D. Koch, Diabetes Care 1988, 11, 801 –
807.
[39] E. R. Damiano, F. H. El-Khatib, H. Zheng, D. M. Nathan, S. J. Russell, Diabetes Care 2013, 36, 251 – 259.
[40] a) A. Facchinetti, G. Sparacino, S. Guerra, Y. M. Luijf, J. H. Devries, J. K.
Mader, M. Ellmerer, C. Benesch, L. Heinemann, D. Bruttomesso, A. Avogaro, C. Cobelli, Diabetes Care 2012, DOI: 10.2337/dc12-0736; b) B. W.
Bequette, Annu. Rev. Control 2012, 36, 255 – 266.
[41] K. T. Wolfe, A. S. Kantak, E. A. Larson, D. E. Pesantez, D. Xi, C. H. Chiu, R.
Shah, Analyte Sensor Apparatuses Having Improved Electrode Configura-
ChemPhysChem 2013, 14, 2032 – 2044
2043
CHEMPHYSCHEM
REVIEWS
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
tions and Methods for Making and Using Them, 2011, US2011/0230735
A1 and EP2548013 A1.
Presentation by R. Shah at Advanced Technologies and Treatments for
Diabetes, Kenes International, Paris, 2013.
Vol. G4 Platinum User Guide, Dexcom, San Diego, 2012.
Presentation by T. Peyser at Advanced Technologies and Therapies for
Diabetes, Kenes International, Paris, 2013.
B. M. Kozak, A. S. Brown, K. L. Close, J. Diabetes 2012, 4, 113 – 126.
A. Poscia, M. Mascini, D. Moscone, M. Luzzana, G. Caramenti, P. Cremonesi, F. Valgimigli, C. Bongiovanni, M. Varalli, Biosens. Bioelectron. 2003,
18, 891 – 898.
F. Baldini, Anal. Bioanal. Chem. 2003, 375, 732 – 743.
a) J. B. Zimmermann, M. Lehmann, S. Hofer, J. Hsing, C. Alles, J.
Werner, J. Stiller, W. Knnecke, S. Luntz, J. Motsch, BMC Anesthesiol.
2012, 12, 24; b) C. Lorencio, Y. Leal, A. Bonet, J. Bondia, C. C. Palerm, A.
Tache, J. M. Sirvent, J. Vehi, Diabetes Technol. Ther. 2012, 14, 568 – 575;
c) R. Brunner, G. Adelsmayr, H. Herkner, C. Madl, U. Holzinger, Crit. Care
2012, 16, R175.
M. Jay S. Skyler, MACPDivision of Endocrinology, Diabetes, and Metabolism and Diabetes Research Institute University of Miami Miller School
of Medicine, Division of Endocrinology, Diabetes, and Metabolism and
Diabetes Research Institute University of Miami Miller School of Medicine, 2012.
A. P. Turner, Chem Soc Rev 2013, 42, 3184 – 3196..
a) M. Hoshino, Y. Haraguchi, I. Mizushima, M. Sakai, J. Artif. Organs 2009,
12, 141 – 149; b) A. J. Kowalski, Diabetes Technol. Ther. 2009, 11, S113 –
S119; c) R. Hovorka, Nat. Rev. Endocrinol. 2011, 7, 385 – 395; d) G. M.
Steil, A. E. Panteleon, K. Rebrin, Adv. Drug Delivery Rev. 2004, 56, 125 –
144; e) C. Patte, S. Pleus, P. Galley, S. Weinert, C. Haug, G. Freckmann, J.
Diabetes Sci. Technol. 2012, 6, 902 – 909.
a) V. Thom-Duret, G. Reach, M. N. Gangnerau, F. Lemonnier, J. C. Klein,
Y. Zhang, Y. Hu, G. S. Wilson, Anal. Chem. 1996, 68, 3822 – 3826; b) V.
Thom-Duret, B. Aussedat, G. Reach, M. N. Gangnerau, F. Lemonnier,
J. C. Klein, Y. Zhang, Y. Hu, G. S. Wilson, Metab. Clin. Exp. 1998, 47, 799 –
803.
A. B. Joseph Shivers, K. Close in Diabetes Close Up (Ed.: K. L. Close), Diabetes Close Up, 2011.
a) A. J. Brenneman, J. E. Smous, Continuous Analyte Monitoring Assembly
and Method of Forming the Same. WO2008115409A1, 2008; b) R. S.
Ward, Y. Tian, W. K. Ward, E. Anderson, J. House, Biosensor Membrane
Material and Drug Delivery Apparatus. US20060183871A1, 2006; c) W. K.
Ward, R. G. Sass, Multiple Use Analyte Sensing Assembly. US7225008B1,
2007; d) A. Brenneman, R. G. Sass, W. K. Ward, J. Matson, D. Slomski, R.
Bruce, L. Wood, I. Federiuk, Adhesive Overbandage. US20080269657A1,
2008; e) W. K. Ward, Method and Apparatus for Background Current Arrangements for a Biosensor. US20100198031A1, 2010; f) D. Kaplan, W. K.
Ward, Additional Calibration for Glucose Sensor. WO2011028670A2, 2011.
D. A. Gough, L. S. Kumosa, T. L. Routh, J. T. Lin, J. Y. Lucisano, Sci Transl
Med 2010, 2, 42ra53., DOI: 10.1126/scitranslmed.3001148
Presentation by R. D. M. Schoemaker at Advanced Technologies and
Treatments for Diabetes, Kenes International, Paris, 2013.
a) J. R. Lakowicz, B. Maliwal, Anal. Chim. Acta 1993, 271, 155 – 164; b) R.
Ballerstadt, J. S. Schultz, Methods Biotechnol. 1998, 7, 89 – 98; c) M. J.
McShane, D. P. O’Neal, R. J. Russell, M. V. Pishko, G. L. Cote, Proc. SPIE
2000, 3923, 78 – 87.
R. J. McNichols, G. L. Cote, J. Biomed. Opt. 2000, 5, 5 – 16.
a) K. J. Thomas, D. B. Sherman, T. J. Amiss, S. A. Andaluz, J. B. Pitner, Diabetes Technol. Ther. 2006, 8, 261 – 268; b) T. J. Amiss, D. B. Sherman,
C. M. Nycz, S. A. Andaluz, J. B. Pitner, Protein Sci. 2007, 16, 2350 – 2359;
c) L. L. E. Salins, R. A. Ware, C. M. Ensor, S. Daunert, Anal. Biochem. 2001,
294, 19 – 26.
T. J. Amiss, D. B. Sherman, C. M. Nycz, S. A. Andaluz, J. B. Pitner, Protein
Sci. 2007, 16, 2350 – 2359.
K. Judge, L. Morrow, A. G. Lastovich, D. Kurisko, S. C. Keith, J. Hartsell, B.
Roberts, E. McVey, K. Weidemaier, K. Win, M. Hompesch, Diabetes Technol. Ther. 2011, 13, 309 – 317.
Presentation by B. K. S. Anderson, D. Kurisko, S. Keith, Y. Fu, B. Roberts,
K. Riley, N. Harvey at Advanced Technologies and Therapies for Diabetes,
Kenes International, Paris, 2013.
A. El-Laboudi, N. S. Oliver, A. Cass, D. Johnston, Diabetes Technol. Ther.
2013, 15, 101 – 115.
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
[64] a) R. J. Pettis, A. J. Harvey, Ther. Delivery 2012, 3, 357 – 371; b) E. McVey,
L. Hirsch, D. Sutter, C. Kapitza, S. Dellweg, J. Clair, K. Rebrin, K. Judge,
R. J. Pettis, J. Diabetes Sci. Technol. 2012, 6, 743 – 754.
[65] a) S. Gamsey, J. T. Suri, R. A. Wessling, B. Singaram, Langmuir 2006, 22,
9067 – 9074; b) S. Gamsey, A. Miller, M. M. Olmstead, C. M. Beavers, L. C.
Hirayama, S. Pradhan, R. A. Wessling, B. Singaram, J. Am. Chem. Soc.
2007, 129, 1278 – 1286.
[66] a) R. Wessling, J. Suri, S. Gamsey, Combinations of Fluorophores of Fluorophores and Pyridinium Boronic Acid Quenchers for use in Analyte Sensors, WO Patent 2,013,012,687, 2013; b) D. R. Markle, S. Gamsey, T. A.
Peyser, Improved Optical Sensor Configuration for Ratiometric Correction
of Blood Glucose Measurement, EP Patent 2,496,139, 2012; c) J. T. Suri,
R. A. Wessling, T. A. Peyser, S. Gamsey, I. Patterson, E. Eric, Antioxidant
Protection of a Chemical Sensor, EP Patent 2,534,470, 2012; d) J. T. Suri,
Flourescent Dyes for Use in Glucose Sensing, EP Patent 2,054,476, 2011;
e) J. T. Suri, HPTS-Mono and Bis Cys-MA Polymerizable Fluorescent Dyes
for use in Analyte Sensors. 2012.
[67] M. Romey, L. Jovanovic, W. Bevier, K. Markova, P. Strasma, H. Zisser, J.
Diabetes Sci. Technol. 2012, 6, 1260 – 1266.
[68] a) R. Ballerstadt, J. S. Schultz, Anal. Chem. 2000, 72, 4185 – 4192; b) G. L.
Cote, M. V. Pishko, Abstracts of Papers, 224th ACS National Meeting,
Boston, MA, United States, August 18 – 22, 2002, ANYL-022.
[69] R. Dutt-Ballerstadt, C. Evans, A. P. Pillai, E. Orzeck, R. Drabek, A. Gowda,
R. McNichols, J. Diabetes Sci. Technol. 2012, 6, 362 – 370.
[70] J. R. Lakowicz, B. P. Maliwal, P. A. Koen, Method Using Luminescence Lifetimes and Energy Transfer for Quantification of Saccharides.
CA2087411A1, 1993.
[71] a) J. Pickup, F. Khan, Glucose Sensor. US 2012,232,251, 2012; b) F. Khan,
T. E. Saxl, J. C. Pickup, Anal. Biochem. 2010, 399, 39 – 43.
[72] J. C. Pickup, F. Hussain, N. D. Evans, O. J. Rolinski, D. J. S. Birch, Biosens.
Bioelectron. 2005, 20, 2555 – 2565.
[73] a) A. E. Colvin, M. A. Mortellaro, A. Modzelewska, Preparation of Arylboronic Acid Derivatives as Oxidation Resistant Indicator Molecules.
WO2008066921A2, 2008; b) A. E. Colvin, C. J. O’Connor, D. C. Ferraro,
Wireless Optical Sensor Assembly. WO2010033901A1, 2010; c) T. D.
James, K. R. A. S. Sandanayake, S. Shinkai, Angew. Chem. 1994, 106,
2287 – 2289.
[74] Presented by X. W. A. DeHennis, C. Mdingi, O. Tymchyshyn, T. Whitehurst, S. Rajaraman at Diabetes Technology Society Meeting, Bethesda,
Maryland, 2012.
[75] A. E. Colvin, H. Jiang, J. Biomed. Mater. Res. Part A 2013, 101, 1274 –
1282.
[76] J. S. Kristensen, K. Gregorius, C. Struve, J. M. Frederiksen, Y. Yu, Sensor
for Detection of Carbohydrate. WO2006061207A1, 2006.
[77] A. E. Colvin, H. Jiang, J Biomed Mater Res A 2013, 101, 1274 – 1282.
[78] C. T. Robert Carlson Receptors LLC, Chaska, 2010, p. 12, http://www.positiveidcorp.com/files/Implantable_Glucose_Sensor_White_Paper.pdf;
http://www.positiveidcorp.com/files/Glucose-Sensor.pdf.
[79] A. Chaudhary, M. Raina, H. Harma, P. Hanninen, M. J. McShane, R. Srivastava, Biotechnol. Bioeng. 2009, 104, 1075 – 1085.
[80] A. Chaudhary, H. Harma, P. Hanninen, M. J. McShane, R. Srivastava, Diabetes Technol. Ther. 2011, 13, 827 – 835.
[81] M. T. Novak, F. Yuan, W. M. Reichert, Anal. Bioanal. Chem. 2010, 398,
1695 – 1705.
[82] a) A. J. Mller, M. Knuth, K. S. Nikolaus, P. Herbrechtsmeier, J. Diabetes
Sci. Technol. 2012, 6, 875 – 883; b) C. Hasslacher, G. Auffarth, I. Platten, T.
Rabsilber, B. Smith, F. Kulozik, M. Knuth, K. Nikolaus, A. Mller, J. Diabetes 2012, 4, 291 – 296.
[83] a) R. Badugu, J. R. Lakowicz, C. D. Geddes, Curr. Opin. Biotechnol. 2005,
16, 100 – 107; b) W. F. March, K. Ochsner, J. Horna, Diabetes Technol. Ther.
2000, 2, 27 – 30; c) H. Yao, C. Marcheselli, A. Afanasiev, I. Lahdesmaki,
B. A. Parviz in Micro Electro Mechanical Systems (MEMS), 2012 IEEE 25th
International Conference on, 2012, pp. 769 – 772.
[84] J. T. Baca, C. R. Taormina, E. Feingold, D. N. Finegold, J. J. Grabowski,
S. A. Asher, Clin. Chem. 2007, 53, 1370 – 1372.
Received: February 18, 2013
Published online on May 6, 2013
ChemPhysChem 2013, 14, 2032 – 2044
2044