Perspectives in Diabetes
`Development of the Implantable Glucose Sensor
`What Are the Prospects and Why Is It Taking So Long?
`
`David A. Gough and Jon C. Armour
`
`Downloaded from http://diabetesjournals.org/diabetes/article-pdf/44/9/1005/361605/44-9-1005.pdf by guest on 14 July 2024
`
`The development of an implantable glucose sensor
`
`for use in diabetes was first suggested in the 1960s
`(1-3). The sensor would provide an alternative to
`the present discrete methods of glucose determi-
`nation that are based on intermittent blood sampling. Con-
`tinuous glucose sensing would be particularly important in
`the detection and management of hypoglycemia. It would
`also allow early detection of hyperglycemia and provide a
`basis for insulin administration at more appropriate dosages
`and timing or for automatic insulin delivery from a pump. An
`implantable glucose sensor could also be used in parallel
`with other existing or potential forms of insulin replacement,
`such as transplantation or hybrid islet devices.
`Since the 1960s, a modest research effort has been devoted
`to implantable glucose sensor development. There has been
`significant progress, a considerable number of publications
`have appeared, and several candidate implantable glucose
`sensors have been partially developed. There is no shortage
`of investigators who feel they have promising approaches.
`By many standards, implantable glucose sensor research is a
`relatively mature and well-worked research field. Results of
`the Diabetes Control and Complications Trial (4) also sug-
`gest a clear need for the sensor. Most people with diabetes
`remain enthusiastic about the possibility of eventually hav-
`ing such a device, although they are unsure about its present
`status. With all of this, one must ask, Why has an implant-
`able glucose sensor not yet appeared in the clinic?
`The reasons are subtle. In our view, there exist questions
`and controversies about several key issues related to sensor
`design and validation in certain applications. Resolution of
`these issues requires implantation studies in animals and
`humans as part of a limited program of focused basic
`research before extensive industrial development of the
`sensor and clinical trials can be effectively carried out. It is
`our experience that this research has not been adequately
`funded by public sources because of the perception that the
`technology is sufficiently advanced that industry should
`assume the responsibility. Industry, however, does not view
`
`support of the remaining research as its mission and either
`has not embraced sensor development or, in certain cases,
`has attempted to pursue development without resolution of
`key scientific issues. This has resulted in ineffective and
`unsuccessful efforts. There is a need to bridge this important
`information gap so that effective development can proceed.
`We give our perspectives here with the hope that a
`consensus can develop that leads to a more direct route for
`clinical introduction of the sensor.
`
`SENSOR CONFIGURATIONS AND POTENTIAL
`APPLICATIONS
`When ultimately implemented, the implantable glucose sen-
`sor may take several configurations.
`A short-term intravenous sensor similar to a catheter
`would be used in hospitalized patients for up to 72 h. This
`application may be useful in diagnosis and glycemic stabili-
`zation, management of ketoacidosis, and monitoring of
`surgery and recovery, labor in mothers with glucose insta-
`bilities, intensive care of certain neonates, and other similar
`situations.
`A short-term subcutaneous sensor would be inserted into
`subcutaneous tissues of nonhospitalized patients for periods
`of several days. The sensor would be connected percutane-
`ously to an external data storage and display device in the
`form of a wristwatch or belt-mounted beeper. This type of
`sensor will be attractive if it is inexpensive, does not require
`frequent recalibration, and can be considered disposable.
`Industry has devoted attention to this application at the
`expense of other implant applications because of its per-
`ceived commercial potential.
`A long-term sensor would be implanted either intrave-
`nously or in tissues for periods of up to 1 year. This sensor
`could be coupled to an implanted telemetry system that
`transmits information to an external receiver, making percu-
`taneous components unnecessary. This configuration will
`have to be easily inserted and retrieved with minimal sur-
`gery. Recalibration may be acceptable if simple and infre-
`quent (say, monthly). In each of these configurations, the
`sensor could be used either as a monitor or as part of an
`automatic feedback-controlled insulin delivery system.
`
`From Department of Bioengineering, University of California, San Diego, La Jolla,
`California.
`Address correspondence and reprint requests to Dr. David A. Gough, Depart-
`ment of Bioengineering, University of California, San Diego, La Jolla, CA 92093-
`0412.
`Received for publication 19 September 1994 and accepted in revised form 11
`May 1995.
`
`SENSOR DESIGNS
`The most advanced glucose sensors are based on immobi-
`lized glucose oxidase coupled to electrochemical systems.
`One version is the hydrogen peroxide-based enzyme elec-
`trode sensor (5-12). This design has a membrane containing
`
`DIABETES, VOL. 44, SEPTEMBER 1995
`
`1005
`
`ABBOTT Exhibit 2043
`Dexcom v. Abbott Diabetes Care, IPR2023-01409
`
`

`

`Downloaded from http://diabetesjournals.org/diabetes/article-pdf/44/9/1005/361605/44-9-1005.pdf by guest on 14 July 2024
`
`IMPLANTABLE GLUCOSE SENSOR
`
`immobilized glucose oxidase coupled to a peroxide-sensitive
`catalytic anode. Glucose and oxygen diffuse into the mem-
`brane, where an enzymatic reaction occurs in which perox-
`ide is produced. The peroxide diffuses through an underlying
`porous membrane to the anode, where it is electrochemically
`oxidized to produce the signal current. The external compo-
`nent of the current passes from the anode through the
`membranes and tissue or blood to a nearby cathode.
`The signal of enzyme electrode sensors can be related to
`glucose concentration when the following conditions are met.
`First, there must be a means of countering the relatively low
`ratio of oxygen to glucose in the body, a problem known as
`the oxygen deficit (13). Ample oxygen must be made avail-
`able by membrane design or other means to avoid oxygen
`limitation of the enzyme reaction. Second, electrochemical
`interference caused by small endogenous molecules that
`pass through the membrane must be insignificant (14). Third,
`any change in sensitivity with time due to enzyme inactiva-
`tion must be accounted for (15). Fourth, the diffusion field in
`front of the sensor must remain undisturbed.
`These conditions can be met in vitro in simple solutions
`with the peroxide-based sensor, but there are differences of
`opinion as to whether all are achievable in vivo with this
`design. Membranes having relatively high oxygen solubility
`may be helpful to minimize the steady-state oxygen deficit
`(10,11,13), but this sensor design provides no means to
`account for the effects of local oxygen variations on the
`signal. Membranes have been proposed (16,17) that may
`partially counter the problem of electrochemical interfer-
`ence, which is substantial with this sensor design (14,18).
`However, peroxide-mediated enzyme inactivation (15) is
`inevitable with this design. Frequent recalibration may be
`necessary to account for interference and enzyme inactiva-
`tion. Commercial in vitro versions of this sensor have
`provisions for automatic rinsing and recalibration between
`samples, but there is no assurance that a stable continuous
`implantable sensor can be achieved by this approach. Given
`these inherent characteristics, this sensor design may be
`limited to short-term applications at best.
`An alternative is the oxygen-based enzyme electrode sen-
`sor (3,13), which has a membrane containing immobilized
`glucose oxidase coupled to a membrane-covered electro-
`chemical oxygen sensor. Glucose and oxygen diffuse into the
`membrane and react, resulting in a reduction of the amount
`of oxygen that would otherwise be detected by the oxygen
`sensor. The signal current is subtracted from that of a similar
`reference oxygen sensor without the enzyme, and a glucose-
`dependent difference current results.
`This approach has several important advantages (19).
`First, catalase can be co-immobilized in excess within the
`membrane to forestall peroxide-mediated glucose oxidase
`inactivation. This is obviously not possible with the perox-
`ide-based sensor. Second, a confluent nonporous hydropho-
`bic membrane is used between the enzyme layer and
`electrode surface to prevent access of polar molecules to the
`electrode surface. This vastly reduces electrochemical inter-
`ference compared with the peroxide-based sensors, which
`must use porous membranes. The hydrophobic membrane
`also completely retains the current within the sensor. Third,
`the use of the reference oxygen sensor in conjunction with
`an appropriate glucose sensor design can eliminate the
`oxygen deficit and render glucose determination transparent
`to oxygen. The nominal disadvantage of this approach is that
`
`the sensor has more components and may therefore be more
`difficult to fabricate.
`Recent innovations in this sensor design include a two-
`dimensional cylindrical configuration in which oxygen enters
`the enzyme region from the end and side, while glucose
`enters from only the end, allowing adequate oxygen avail-
`ability even at substantial concentration mismatches (20),
`and development of a three-electrode potentiostatic oxygen
`sensor (21), which is much more stable than the well-known
`two-electrode Clark oxygen sensor. In addition to new design
`features, there is now a better understanding of factors that
`affect the stability of the immobilized glucose oxidase: The
`immobilized enzyme is remarkably stable under certain
`conditions but can decay rapidly in the presence of hydrogen
`peroxide (15). Sensor designs that incorporate an enzyme
`reserve, include co-immobilized catalase, and promote rela-
`tively low catalytic generation of peroxide are advantageous
`(22).
`Another sensor design is the mediator-based enzyme
`electrode sensor (23-25), in which electron exchange mole-
`cules included in the enzyme region take the place of oxygen
`to shuttle electrons between the enzyme and electrode
`surface. This feature is intended to avoid the oxygen limita-
`tion. There is a reduction in sensitivity to oxygen with this
`design. However, the problem of electrochemical interfer-
`ence remains, and there is a potential for mediator washout.
`A successful sensor for in vitro glucose determination has
`been developed by this approach, but is difficult to adapt to
`in vivo applications.
`Continuous microdialysis is also being considered as a
`basis for glucose sensing (26,27). The concept involves the
`use of a small hollow fiber inserted under the skin through
`which a buffer is circulated and returned to an external
`glucose analyzer. Extracellular glucose is detectable in this
`manner, but there are often significant response lags due to
`glucose equilibration. A possibility exists for development of
`this type of device for short-term applications, but conve-
`nient long-term applications are unlikely because of the
`obtrusiveness of the apparatus.
`Membrane-covered catalytic electrodes have been studied
`extensively for glucose sensing (28,29), but problems with
`selectivity to glucose and electrochemical interference re-
`main substantial.
`Noninvasive optical glucose sensor concepts have re-
`ceived considerable attention in recent years. The premise is
`that light in the near-infrared or other region of the spectrum
`that has some sensitivity to glucose is beamed on a relatively
`transparent region of tissue such as the finger web (30). The
`transmitted or reflected light signal is processed by mathe-
`matical filtering techniques to maximize any aspects of the
`signal that may show some correlation with blood glucose
`samples obtained at the time of the observation. The over-
`whelming problem with this approach is the lack of adequate
`selectivity for glucose. There is substantial chemical inter-
`ference from many biological molecules, as well as physical
`interference from tissue structures, temporal and spatial
`variations in perfusion, and several optical effects. In addi-
`tion, there is a fundamental reservation about calibration of
`the device: The retrospective correlations between signal
`and blood glucose concentration described above change
`with time and are not useful for real-time monitoring. These
`issues must be resolved by systematic mechanistic studies
`before further development by industry can be justified. The
`
`1006
`
`DIABETES, VOL. 44, SEPTEMBER 1995
`
`ABBOTT Exhibit 2043
`Dexcom v. Abbott Diabetes Care, IPR2023-01409
`
`

`

`Downloaded from http://diabetesjournals.org/diabetes/article-pdf/44/9/1005/361605/44-9-1005.pdf by guest on 14 July 2024
`
`best hope for this approach may be to establish a separate
`qualitative relationship for each user that detects large
`glucose changes. However, even this degree of response will
`be very difficult to validate at best and may lead to a sensor
`that is of minimal clinical value. Despite substantial indus-
`trial investment, media attention, and considerable enthusi-
`asm on the part of proponents, prospects for implementation
`of a noninvasive glucose sensor in the near future are not
`encouraging.
`A variety of other sensor concepts have been explored,
`including glucose oxidase coupled to thermal sensors (31),
`osmotically active gels (32), and fiber-optic sensors with
`glucose-sensitive ligands (33,34). Some of these approaches
`have been the basis of devices that can respond to glucose in
`buffer under ideal conditions, but there should be little
`expectation that these devices can function as implantable
`sensors.
`Other types of sensors are sensitive to certain physio-
`logical effects of glucose. Simple bare electrodes or other
`sensors responsive to local microvascular perfusion, temper-
`ature, or potassium fluxes can often register a change in
`signal immediately after a glucose injection because of the
`effects of glucose on a variety of physiological phenomena.
`This has led to the hope that monitoring these phenomena
`might reliably indicate glucose concentration. In reality, the
`signals are only weakly related to glucose and are not
`specific. Moreover, this approach requires independent prior
`knowledge of glucose challenges and is not effective for
`detection of spontaneous glucose changes. In our view, it is
`unlikely that reliable glucose sensing can be achieved by this
`route.
`
`IMPLANT STUDIES
`Short-term subcutaneous implants. The short-term sub-
`cutaneous sensor is the most difficult application. The per-
`oxide-based sensor has been fabricated in the form of a
`needle and implanted in dogs (35,36), rats (8,12), and pigs
`(24). The sensors typically remained in place for several days
`and signals were recorded in response to intravenous glu-
`cose challenges. The sensitivity and baseline of each sensor's
`output were independently adjusted after each experiment.
`Anesthesia was used during testing in most cases. The most
`common experimental protocol was to determine how
`closely statistical averages of signals at discrete times from
`various sensors correlated with statistical averages of blood
`glucose concentration. Responses of sensors that did not
`meet expectations were often ignored. Because of the em-
`phasis on statistical correlation, there was no attempt to
`interpret the signals of individual sensors, although an inter-
`pretation of this kind is necessary for actual application of
`the sensor. Little understanding of the factors that affect
`glucose transport to implanted sensors was revealed. These
`studies show that subcutaneous sensors can sometimes
`produce a response to blood glucose challenges under ideal
`conditions, but it remains unclear whether the response can
`be useful for glucose monitoring.
`There has also been an attempt to use the peroxide-based
`sensor as a long-term implant in dog subcutaneous tissue
`(11). The sensor and a telemetry unit (37) were implanted for
`periods of up to 90 days. The sensor signal decayed contin-
`uously over the implant period, and the sensor required
`recalibration before each recording session. No determina-
`
`D.A. GOUGH AND J.C. ARMOUR
`
`tion was made of the maximal period the sensor could
`operate accurately without recalibration, although this was
`apparently only a few days at most. Thus, even though the
`implant remained in place for up to 90 days, it could not be
`considered a reliable, functioning sensor for that period. It
`would not be feasible for such a sensor to accurately indicate
`spontaneous glucose fluctuations or be used for purposes of
`continuous control. The authors reported that the range of
`sensitivity to glucose decreased significantly over the im-
`plant period because of an increasing oxygen limitation, even
`after recalibration. This may have been caused by enzyme
`inactivation, progressive electrochemical interference, or
`an advancing tissue reaction. The important question of
`whether the limitations were due to sensor design, tissue
`remodeling, or local physiological phenomena was not con-
`sidered. The many questions raised by this study remain to
`be addressed.
`The peroxide-based sensor in the form of a fine needle has
`been used as a short-term subcutaneous implant in humans
`(7,10,38-40). In these studies, sensors were connected by
`percutaneous leads to wearable instrumentation and there
`was no need for local anesthesia. Statistical correlations
`were reported between groups of sensor responses, and
`averaged blood glucose concentrations were determined by
`a standard method. As judged by in vitro characterization
`before implantation, sensors typically had a rapid response
`to concentration changes and were linear over clinically
`useful ranges. Most studies noted a substantial decay in
`signal over the several-day implant period, but the sensor
`responses were typically adjusted after the experiment to
`correspond to actual blood glucose values. One group (40)
`used a rapid recalibration device that adjusted readings
`during use based on glucose values determined by frequent
`blood sampling. Extensive histological studies of the sensor
`environment were not feasible in humans.
`Although most investigators interpreted these results as
`highly promising, in our view, there are a number of reser-
`vations. First, there is the problem of decay in sensitivity
`over the implant period. This has led to the use of retrospec-
`tive calibration, in which the sensor sensitivity is adjusted
`after the experiment to match independently determined
`steady-state blood glucose values. In addition, the actual
`values of the signal without recalibration are not often
`reported, making it difficult to compare actual rates of signal
`decay. The correlation between steady-state blood glucose
`and retrospectively adjusted signal conveys the impression
`that the sensor can be useful in real time, but in fact,
`monitoring in real time cannot be based on retrospective
`calibration unless there is a highly reproducible pattern of
`signal decay, which has yet to be established. Improved
`means of calibration must be developed.
`Other reservations pertain to the tissue structure around
`the sensor. The technique of sensor insertion may be of
`crucial importance, and there must be a means of knowing
`whether a functional placement has been achieved. Trauma
`associated with sensor placement leads to tissue inflamma-
`tion and the wound-healing process, which may interfere
`with stabilization of the sensor signal. The insertion must not
`cause significant bleeding or lead to the formation of a
`reservoir of tissue fluid around the sensor. The resulting
`local hyperemia must stabilize rapidly after insertion and
`give rise to a stationary mass transfer field if reliable readings
`are to be expected. Given these considerations, it is unlikely
`
`DIABETES, VOL. 44, SEPTEMBER 1995
`
`1007
`
`ABBOTT Exhibit 2043
`Dexcom v. Abbott Diabetes Care, IPR2023-01409
`
`

`

`Downloaded from http://diabetesjournals.org/diabetes/article-pdf/44/9/1005/361605/44-9-1005.pdf by guest on 14 July 2024
`
`IMPLANTABLE GLUCOSE SENSOR
`
`that the sensor can be inserted and used shortly thereafter to
`confidently administer medication. There is a need for better
`understanding of the effects, time course, and reproducibility
`of trauma associated with sensor insertion.
`Data selection in these studies is also problematic. Be-
`cause it is not possible to present all results, representative
`results must be selected. However, results selected with the
`intent of demonstrating that the sensor is working often do
`not indicate the range of responses typically obtained, which
`would be useful for a deeper understanding. Furthermore,
`reports of the percentage of apparently functional sensor
`implants do not convey the troubling inability to predict if a
`specific sensor that is functional in vitro will give meaningful
`signals when implanted subcutaneously. Data selection prac-
`tices and incomplete description of results may unintention-
`ally convey the impression that favorable results are more
`common than is actually the case.
`In addition to issues of calibration, tissue stabilization, and
`data selection, it is our view that there is inadequate defini-
`tion of the relationship between the signal of a sensor
`implanted in tissues and actual blood glucose values. A
`simple, stable proportionality that applies with confidence in
`all situations probably does not exist. If a generally useful
`relationship is found, it will be more complicated than
`presently appreciated. The glucose signal is affected substan-
`tially by dynamic physiological aspects of the sensor envi-
`ronment and changes in the environment with time. For
`example, microvascular perfusion of the site and therefore
`delivery of glucose to the sensor may be strongly influenced
`by transient events such as sympathetic stimulation, thermal
`effects, and hydrostatic variations due to posture, movement,
`etc. Unfortunately, most research to date has been directed
`at finding a simple statistical correlation between the signal
`and blood glucose concentration, rather than understanding
`the role of relevant physiological phenomena and developing
`a deterministic model of glucose transport in living tissue
`that can be used in a predictive fashion. Moreover, in many
`cases blood sampling has not been sufficiently frequent to
`clearly define blood glucose fluctuations. More carefully
`conceived research is needed to establish the generality of
`signal-blood glucose relationships.
`There is likely to be no consensus on the ultimate promise
`of short-term tissue implant applications until these contro-
`versies are resolved. These questions have led to the most
`important difference of opinion—whether human trials of
`short-term subcutaneous sensors are presently justified.
`Early studies in humans have emphasized correlative, trial-
`and-error observations, rather than mechanistic research.
`Many questions can be more effectively addressed at present
`with animal studies. There is an urgent need for directed
`basic studies that lead to a better understanding of the
`sensor response in the tissue environment before the results
`of clinical trials can be interpreted. It will not be possible to
`effectively address deficiencies in response without a better
`basic understanding. Support for basic studies of this type
`does not fall in the domain of industrial research and must
`come from public sources. In all, there is much to be learned
`before the subcutaneous sensor can be operated confidently.
`Short-term intravenous sensor. Technically, the short-
`term intravenous sensor for inpatient use is much more
`straightforward. The sensor could be coupled to an intrave-
`nous fluid flush system and may have no greater propensity
`for complications than the intravenous catheters presently in
`
`use. Both the oxygen-based and the peroxide-based sensors
`have been operated intravenously in animals (8,41). No
`further basic research is needed, and this is the most direct
`route to clinical application of an implantable sensor, al-
`though the demand may be small relative to other potential
`applications.
`Long-term intravenous sensor. The oxygen-based sensor
`has been implanted intravenously in six dogs for periods of
`7-108 days (19). The glucose sensor was at the tip of a
`cylinder with a 2-mm outer diameter and a total length of
`~30 cm. The sensor and a similar oxygen reference sensor
`were advanced into the jugular vein so that the tips were
`positioned in the center of the superior vena cava several
`centimeters above the entrance of the right atrium. The
`sensor leads were connected to a telemeter and instrumen-
`tation unit (42) implanted subcutaneously. No systemic
`anticoagulation was needed. The sensitivity to glucose, de-
`termined before implantation, during use, and after ex-
`plantation, was not substantially altered by long-term
`implantation (19), and there was no need for recalibration
`during the entire period of the implantation. The experi-
`ments were not limited by immobilized enzyme lifetime,
`oxygen deficit, oxygen sensor instability, chemical interfer-
`ence, or biological incompatibility. These results demon-
`strate that this glucose sensor can function when implanted
`in the bloodstream of a dog for a period of several months
`and has the potential of operating for longer periods.
`The success of the sensor as an intravenous implant in
`animals and the apparent biocompatibility suggest the pos-
`sibility of use of this site for chronic implantation in humans.
`A catheter-like sensor that can be easily introduced into the
`vena cava, retrieved, and replaced every 3-6 months with a
`nonsurgical procedure may be clinically useful. This applica-
`tion raises a concern about patient safety: there is the
`possibility of clot formation or vascular wall damage. How-
`ever, it is well documented that similar pacemaker leads and
`chronic drug delivery catheters implanted in the superior
`vena cava in humans can be biocompatible and are relatively
`innocuous (43,44). This sensor is ready for development by
`industry, and implementation in humans may be relatively
`easily justified. Lag-free intravenous blood glucose recording
`may become the gold standard and the most useful approach
`for control of insulin pumps.
`
`OTHER ISSUES AND CONCLUSIONS
`Development of an implantable sensor is not a trivial task.
`The fact that research has been under way for decades in this
`area should be no more surprising than the fact that research
`in any area takes time. However, the need for the implantable
`glucose sensor is now clearer than ever. Although there has
`been considerable progress on certain sensor concepts,
`disparate investigative approaches and development goals
`have led to some confusion.
`Priority must now be given to consummation of the most
`advanced sensor approaches. There are no fundamental
`technical barriers to the use of oxygen-based sensors for
`short- and long-term intravenous applications, and given
`appropriate industrial commitment, these applications could
`be available within a few years. The peroxide-based sensor
`also has some promise, mainly for short-term applications.
`Both the oxygen-based and the peroxide-based sensors may
`eventually be useful in some capacity as subcutaneous
`
`1008
`
`DIABETES, VOL. 44, SEPTEMBER 1995
`
`ABBOTT Exhibit 2043
`Dexcom v. Abbott Diabetes Care, IPR2023-01409
`
`

`

`Downloaded from http://diabetesjournals.org/diabetes/article-pdf/44/9/1005/361605/44-9-1005.pdf by guest on 14 July 2024
`
`implants, but more focused research is needed on the
`physiology of sensor-tissue interactions. Certain long-term
`and intravenous sensor applications may actually be closer
`to clinical introduction and should be addressed more ag-
`gressively. Other sensor concepts are much farther behind. It
`is naive to expect that a significant new solution or break-
`through will come from a serendipitous discovery of the
`proverbial
`isolated
`inventor working with minimal re-
`sources.
`A more meaningful industrial involvement is needed. At
`present, industrial attention is focused on subcutaneous
`applications, but advances must await the resolution of
`certain questions about sensor operation in vivo. The an-
`swers will come from targeted research supported by public
`sources. In addition, more emphasis is needed by industry on
`long-term applications. Some investigators have pointed out
`that large-scale manufacturing of reproducible and reliable
`sensors may also pose technical difficulties (10). This may be
`true if the cost of manufacture is a crucial factor, as it is for
`disposable short-term sensors.
`It is now time to reach a consensus of investigators,
`funding agencies, and industrial partners so that implemen-
`tation of the implantable glucose sensor can move ahead
`decisively.
`
`ACKNOWLEDGMENTS
`This work was funded in part by the Juvenile Diabetes
`Foundation International and the Whitaker Foundation.
`We thank David Stenger.
`
`REFERENCES
`1. Kadish A: Automation control of blood glucose: a servo mechanism for
`glucose monitoring and control. Am J Med Electron 3:82-86, 1964
`2. Clark LC Jr, Lyons C: Electrode systems for continuous monitoring in
`cardiovascular surgery. Ann NYAcad Sci 102:29-45, 1962
`3. Updike SJ, Hicks GP: The enzyme electrode. Nature 214:986-988, 1967
`4. The Diabetes Control and Complications Trial Research Group: The
`effect of intensive treatment of diabetes on the development and pro-
`gression of long-term complications in insulin-dependent diabetes melli-
`tus. N Engl J Med 329:977-986, 1993
`5. Clark LC Jr: A family of polarographic enzyme electrodes and the
`measurement of alcohol. Biotechnol Bioeng 3:377-394, 1972
`6. Abel P, Muller A, Fischer U: Experience with an implantable glucose
`sensor as a prerequisite of an artificial beta cell. Biomed Biochim Acta
`43:577-588, 1984
`7. Shichiri M, Asakawa N, Yamasaki Y, Kawamori R, Abe H: Telemetry
`glucose monitoring device with needle-type glucose sensor: a useful tool
`for blood glucose monitoring in diabetic individuals. Diabetes Care
`9:298-301, 1986
`8. Turner RFB, Harrison DJ, Rajotte RV, Baltes HP: A biocompatible
`enzyme electrode for continuous in vivo glucose monitoring in whole
`blood. Sens Actual B 1:561-564, 1990
`9. Moatti-Sirat D, Capron F, Poitout V, Reach G, Bindra DS, Zhang Y, Wilson
`GS, Thevenot DR: Towards continuous glucose monitoring: in vivo
`evaluation of a miniaturized glucose sensor implanted for several days in
`rat subcutaneous tissue. Diabetologia 35:225-230, 1992
`10. Johnson KW, Mastrototaro JJ, Howey DC, Brunelle RL, Burden-Brady PL,
`Bryan NA, Andrew CC, Rowe HM, Allen DJ, Noffke BW, McMahan WC,
`Morff RJ, Lipson D, Nevin RS: In vivo evaluation of an electroenzymatic
`glucose sensor implanted in subcutaneous tissue. Biosens Bioelectron
`7:709-714, 1992
`11. Gilligan BJ, Shults MC, Rhodes RK, Updike SJ: Evaluation of a subcuta-
`neous glucose sensor out to three months in a dog model. Diabetes Care
`17:882-887, 1994
`12. Ward WK, Wilgus ES, Troupe JE: Rapid detection of hyperglycaemia by
`a subcutaneously-implanted glucose sensor in the rat. Biosens Bioelec-
`tron 9:423-428, 1994
`13. Gpugh DA, Leypoldt JK, Armour JC: Progress toward a potentially
`implantable enzyme-based glucose sensor. Diabetes Care 5:190-198,
`1982
`14. Kerner W, Kiwit M, Linke B, Keck FS, Zier H, Pfeiffer EF: The function of
`a hydrogen peroxide-detecting electroenzymatic glucose electrode is
`markedly impaired in human subcutaneous tissue and plasma. Biosens
`Bioelectron 8:473-482, 1993
`
`DA GOUGH AND J.C. ARMOUR
`
`15. Tse PHS, Gough DA: Time-dependent inactivation of immobilized glu-
`cose oxidase and catalase. Biotechnol Bioeng 29:705-713, 1986
`16. Zhang Y, Hu Y, Wilson GS, Moatti-Sirat D, Poitout V, Reach G: Elimina-
`tion of acetaminophen interference in an implantable glucose sensor.
`Anal Chem 66:1183-1188, 1994
`17. Moussy F, Harrison DJ, O'Brien DW, Rajotte RV: Performance of subcu-
`taneously implanted needle-type glucose sensors employing a novel
`trilayer coating. Anal Chem 65:2072-2077, 1993
`18. Moatti-Sirat D, Velho G, Reach G: Evaluating in vitro and in vivo the
`interference of ascorbate and acetaminophen on glucose detection by a
`needle-type glucose sensor. Biosens Bioelectron 7:345-352, 1992
`19. Armour JC, Lucisano JY, BD McKean, Gough DA: Application of chronic
`intravascular blood glucose sensor in dogs. Diabetes 39:1519-1526, 1990
`20. Gough DA, Lucisano JY, Tse PHS: A two-dimensional enzyme electrode
`sensor for glucose. Anal Chem 57:2351-2357, 1985
`21. Lucisano JY, Armour JC, Gough DA: In vitro stability of an oxygen
`sensor. Anal Chem 59:736-739, 1987
`22. Leypoldt JK, Gough DA: Model of a two-substrate enzyme electrode for
`glucose. Anal Chem 56:2896-2904, 1984
`23. Cass A