45 Biosensors for real-time in vivo measurements

Biosensors and Bioelectronics20(2005)2388–2403ReviewBiosensors for real-time in vivo measurementsGeorge S.Wilson∗,Raeann GiffordDepartment of Chemistry,University of Kansas,Malott Hall,Lawrence,KS66045,USAReceived14October2004;received in revised form1November2004;accepted2December2004Available online15January2005AbstractThe current status of sensors capable of continuous measurement of analytes in biological media is reviewed.This review containing173 references deals with devices whose use in single cells,tissue slices,animal models and humans has been demonstrated.In addition to sensors specific for glucose,lactate,glutamate,pyruvate,choline and acetylcholine,insights obtained from monitoring nitric oxide,Na+,K+,Ca2+,and dopamine are presented.Performance criteria for sensor performance are described as is the subject of biosensor calibration.Biocompatibility issues are dealt with in some detail as is the status of continuous blood glucose monitoring in humans.©2004Elsevier B.V.All rights reserved.Keywords:Biosensor;Continuous monitoring of glucose;BiocompatibilityContents1.Introduction (2389)2.Analyte-specific sensors (2390)2.1.Sensor performance criteria (2390)2.2.Sensitivity (2390)2.3.Selectivity (2391)2.4.Interferences (2392)2.5.Temporal/spatial resolution (2392)2.6.Simultaneous analyte detection (2393)3.Biocompatibility (2393)3.1.Initial inflammatory response events (2393)3.2.Physiological host response (2394)3.3.Biosensor degradation (2396)3.4.Biocompatibility tests (2396)3.5.Calibration (2396)4.Applications (2396)4.1.Real-time blood glucose monitoring (2396)4.2.Assessment of glucose sensor performance (2397)mercially available systems (2398)4.4.Cell culture studies (2399)∗Corresponding author.Tel.:+17858643475;fax:+17858645272.E-mail address:gwilson@(G.S.Wilson).0956-5663/$–see front matter©2004Elsevier B.V.All rights reserved.doi:10.1016/j.bios.2004.12.003G.S.Wilson,R.Gifford/Biosensors andBioelectronics 20(2005)2388–240323895.Future prospects ...............................................................................................2399References ........................................................................................................23991.IntroductionThe biosensor,born in the 1960s with the pioneering work of Clark (Clark and Lyons,1962)has only recently been increasingly employed in a variety of applications where continuous measurements in biological media are required.An electrochemical biosensor has been defined as a “self-contained integrated device,which is capable of providing specific quantitative or semi-quantitative information using a biological recognition element retained in direct spatial con-tact with an electrochemical transduction element”(Th´e venot et al.,2001).It is,of course,possible to employ single use de-vices,such as those used for self-monitoring of blood glucose but this review will concentrate on biosensors that are capable of in situ measurements with attendant good time and spatial resolution.Such devices may be required to function reliably for hours,days or even months in the biological medium without regeneration or addition of reagents.A biosensor is distinguished from a chemical sensor in that it possesses a biological recognition element,typically a protein,peptide or oligonucleotide.Not all biological recognition elements lend themselves to continuous monitoring because the reac-tions that they undergo with the target analyte areessentially Fig.1.Energy utilization in the brain:balance of glycolysis and oxidative phosphorylation via a lactate pool (Pellerin,L.,Magistretti,P.J.,2004.Reprinted with permission from Science,305:50–51).irreversible.Thus,there have been few antibody–antigen re-actions or oligonucleotide hybridization reactions applied to this type of application.The last 10years,the period of concentration of this re-view has shown increased confidence in the reliability of in vivo sensors,and this has led,in turn,to their use in monitor-ing biological processes in real time.Improved performance has made possible measurement of sub-second processes,such as various forms of exocytosis.The goal in this case may be to measure the rate of uptake or efflux of relevant species or to establish spatial distributions.Such processes are invariably coupled to changes in Na +,K +,and Ca 2+con-centrations as well as pH (Kennedy et al.,2002a ).For this reason,it may be essential to measure several analytes si-multaneously,so that temporal and spatial relationships can be established.The importance of this kind of information is illustrated by the schema for energy utilization in the brain as shown in Fig.1below.Upon neural stimulation (glutamater-gic neurons),there will be rapid changes in the extracellular concentration of several species and complete understanding of regulatory and signaling processes requires establishment of the phase relationships between,for example,oxygen and glucose uptake,glutamate release and the interplay of cations2390G.S.Wilson,R.Gifford/Biosensors andBioelectronics20(2005)2388–2403that control potential gradients at membranes as well as nu-merous other species.For this reason,we will give some attention to changes in the concentration of species that play key roles in regulation of cellular events,even if the rele-vant sensor does not meet the requirements of a biosensor. The scheme illustrates the challenges to sensor development in providing a comprehensive picture of many simultaneous processes proceeding in parallel.The scope of this review will be further restricted to the monitoring of single cells,cell cultures,tissue slices,and a variety of mammalian models:rats,dogs,rabbits,and hu-mans.It is also possible to sample the biological medium in question using microdialysis.This affords a greater range of analytical approaches to analysis,including separations and derivatization not possible with a single sensor or even a sensor array.What is sacrificed is optimal temporal and spatial resolution and in some cases,tissue damage created by the relatively large probes can affect measurements.Non-invasive spectroscopic techniques such as FTIR,light polar-ization,and NMR have also been employed,but as they are not biosensors,they will not be covered in detail here.Read-ers are referred to a recent review on this subject(Cot´e et al.,2003).When a sensor is brought in contact with biolog-ical tissues,sensor performance can deteriorate.The exact causes of this deterioration are not clear,but are a mix of pas-sive adsorption of biomolecules on the sensor/probe surface and active processes coupled to tissue response.This subject will be discussed in more detail.2.Analyte-specific sensorsPerformance criteria for in vivo biosensors are not only dependent on the specific analyte,but also on the intended application for the biosensor.Because of its importance to the treatment of diabetes,glucose biosensors have been the most extensively studied,although lactate,oxygen,reactive oxy-gen(ROS),and nitrogen species(RNS)have also been inves-tigated(Lisdat and Scheller,2000;Brovkovych et al.,1999). With increased emphasis on neurobiology lactate,glutamate, and pyruvate have been measured in the brains of mammals, single cells,cell cultures,and tissue slices as well as the influx or efflux of Ca2+,Na+,and K+ions(Kahlert and Reiser,2004; Kennedy et al.,2002a;Buck et al.,1995;Smith et al.,1999). Other important analytes involved in neurotransmission,in-cluding acetylcholine and choline(Cui et al.,2001;Mitchell, 2004),ascorbate(primarily as an interferent)(Kulagina et al.,1999),NAD(H)(Liu et al.,1999),and dopamine (Avshalumov et al.,2003)have been monitored.In conjunc-tion,with these analytes ROS and nitric oxide(NO)have also been monitored(Scheller et al.,1999;Manning et al.,1998).2.1.Sensor performance criteriaThe challenge for in vivo biosensor development is pro-viding adequate performance to distinguish among these cell signaling entities and neurotransmitters in a manner that leads to enhanced understanding of biological function.That re-quires sensor specificity with appropriate spatial and tem-poral resolution within acceptable sensitivity and limits of detection(LOD)for each analyte.It is necessary to achieve optimum balance among thefigures of merit for a specific ap-plication.For example,the addition of a permselective mem-brane may improve selectivity but simultaneously degrade response time.In addition,the biosensor must be reasonably stable,which is longer than a few hours,with days or weeks preferable.The minimum useful stability is defined by the duration of the experiment,which for in vivo applications may be hours or days in a hostile environment.The most demanding application and the greatest focus for biosensor development over the last few years has been for neurobiol-ogy research,which is the source for the majority of the new biosensor applications reviewed here.2.2.SensitivityBiosensor sensitivity,LOD,and linear range are a func-tion of the physical design and the molecular recognition element(e.g.biomolecule activity).Biosensor development for measurement of glucose and lactate is relatively simple because of reasonably high endogenous brain concentrations (millimolar range)and enzymes with adequate specific activ-ity and stability levels(Parkin et al.,2003;Yao et al.,2004; Yang et al.,2001;Gramsbergen et al.,2003).Glutamate and pyruvate are present at endogenous brain concentrations in the low micromolar range and in addition,the corresponding oxidase enzymes have low activity and stability(Parkin et al.,2003;Yao et al.,2004;Yang et al.,2001;Gramsbergen et al.,2003).The challenge increases with pyruvate where the enzymatic reaction is dependent on multiple substrates and co-factors(Gajovic et al.,2000).Some progress has been made on improving stability by optimizing the immobiliza-tion conditions(Chen et al.,1998;Heller and Heller,1998).The required sensitivity for a particular analyte is de-termined by the concentration levels found in the mea-surement environment of interest.For enzymatic biosensor measurements of glutamate,the difference between the ex-tracellular milieu of10␮M and1–10mM for intracellu-lar measurement is significant(Kahlert and Reiser,2004). The discrepancy between intra-and extra-cellular concentra-tion is not confined to analytes measured with amperomet-ric biosensors,but also to other important species such as oxygen with intra-and extra-cellular concentrations given, respectively(0.032mM/0.24mM),Na+(20mM/140mM), and K+(130mM/5mM)(Clark et al.,1998;Kahlert and Reiser,2004).These differences create concentration gra-dients across the cell membrane,which may be changing rapidly in the second to millisecond time domain.More of-ten,it is the concentration changes,such as a50–500%vari-ation in lactate,pyruvate or glucose concentrations seen dur-ing neurological stimulation experiments that are of interest (Yang et al.,2001;Yao et al.,2004).G.S.Wilson,R.Gifford/Biosensors andBioelectronics20(2005)2388–24032391Biosensor sensitivity can be enhanced by changing the surface of the electrode.For oxidation of peroxide,which is most rapid on Pt,carbonfibers can be platinized(Clark et al.,1998).Deposition of platinum black on carbon or Pt electrodes also increases the active surface area,further en-hancing sensitivity.However,this strategy also increases the electrode sensitivity to electroactive interferences and makes the sensor vulnerable to decreased response resulting from adsorption of surface-active species.Specialfilms that serve to concentrate analytes by adsorption,such as Nafion TM for dopamine,will increase sensitivity;sometimes at the cost of diminished temporal resolution(Venton et al.,2002).A review of dynamic measurements in microenvironments in-dicates that reduction of the background is one key to en-hancing the sensitivity and reducing the LOD for in vivo measurements(Wightman et al.,1999).This can be accom-plished by reducing the complexity of the environment by us-ing tissue slices or cultured cells rather than the virtually un-controlled in vivo environment and/or eliminating interfering species(Avshalumov et al.,2003).Using technologies with inherently high signal-to-noise ratios such as electro-optical chemiluminescent sensors with appropriate molecular recog-nition elements are promising methods to enhance sensitivity (Szunerits and Walt,2003).In addition,some electronicfil-tering of background noise can be effective;however,the use offiltering that could obscure important temporal events has been pointed out(Wightman et al.,1999).Extending the lin-ear range by creating diffusion-limiting barriers with permse-lective polymer membranes is a widely employed approach. The sensor response is no longer controlled by the kinet-ics of the enzyme reaction,but by mass transfer.This has the advantage of minimizing temperature effects(∼2.5%/◦C increase in the rate of mass transfer versus∼10%/◦C for en-zyme catalysis).The disadvantage is a decrease in sensitivity and response time and under some conditions,complicated sensor response(Jablecki and Gough,2000).2.3.SelectivitySelectivity for biosensors is gained by employing ana-lyte specific molecular recognition elements,primarily en-zymes.To produce reagentless sensors,oxidases are used most frequently for implantable applications,because the co-substrate,oxygen,is relatively abundant in biological ap-plications.Some analytes require more specialized enzymes such as the FIA system developed for␥-aminobutyric acid (GABA)that used gabase(a dual enzyme system)to convert GABA to succinic semialdehyde(SSA)and glutamate.The glutamate can then be monitored with a glutamate oxidase biosensor system(Niwa et al.,1998).However,gabase re-quires the co-substrate␣-ketoglutarate that must be supplied as an external reagent,precluding this system from use as a reagentless biosensor.Although there are a large number of dehydrogenases that could serve as biological recognition el-ements,they typically require co-factors,which restricts their use to microdialysis-based applications(Liu et al.,1999).Other methods are used to render specificity,most no-tably the work by Wightman’s group in using fast scan cyclic voltammetry with microelectrodes to produce a unique voltammogramfingerprint for dopamine(Bath et al.,2000; Wightman et al.,1999).Measures that should be employed to guarantee specificity,primarily with regard to interferent species have been described(Phillips and Wightman,2003). This analysis applies to intrinsically electroactive endoge-nous species,but it is also important to systematically evalu-ate sensors by checking potential interferences while,at the same time,measuring the analyte of interest(Hu and Wilson, 1997a,b;Wilson and Hu,2000).Electrochemical biosensors, in spite of the use of analyte specific molecular recognition el-ements,are susceptible to interference from endogenous elec-troactive compounds such as uric acid,dopamine,and ascor-bate.Ascorbate is the most troublesome due to its compara-tively high concentration and broad oxidation potential range. Ascorbate levels also do not remain constant during neuronal stimulation.Brain ascorbate concentration has been reported from100to600␮M,and in serum at780␮M(Gajovic et al., 2000;Georganopoulou et al.,2000;Kulagina et al.,1999; Venton et al.,2002).These concentrations are compared to a lower concentration analyte like glutamate with brain basal extracellular levels from1to29␮M,which would require ex-clusion factors as high as800:1(Yang et al.,2001;Kulagina et al.,1999).Acid anions such as pyruvate,lactate,and glu-tamate are challenging because traditional anionic screen-ing methods such as Nafion TM will also exclude or retard the analyte of interest(Brown and Lowry,2003;Yang et al., 2001;Gajovic et al.,2000;Schram et al.,2002).For pyru-vate,the best solution reported was to use size exclusion by inserting the biosensor in dialysis tubing,which significantly limits the temporal resolution(Gajovic et al.,2000).In ad-dition to Nafion,several other methods have been employed to produce a combination anionic size exclusion barrier to interfering species,including electropolymerization of pyr-role and o-phenylendiamine(Gajovic et al.,1999;Fabre et al.,1997;McAteer and O’Neill,1996;Lowry et al.,1998a; Friedemann et al.,1996),cellulose acetate,and polyester sul-fonic acid for applications measuring nitric oxide,glucose, pyruvate,and glutamate(Clark et al.,1998).Experience in our laboratory has shown that electropolymerizedfilms show excellent permselectivity for short periods of time,but tend to fail rapidly when operated at37◦C.We have been able to mit-igate this problem by electrodeposition of enzyme followed by electropolymerization.In this way,we are able to deposit a thin,compact layer of enzyme,which leads to high sensi-tivity,broad dynamic range,and rapid response(Chen et al., 2002c;Matsumoto et al.,2002).Alumina sol–gels have also proven to be an effective method for enzyme immobilization as well as providing permselectivity and stability over ex-tended periods of time and a response time of less than10s (Chen et al.,2002b).The incorporation of ascorbate oxidase, which produces water rather than peroxide,has also been em-ployed(Mao et al.,2002;Phillips and Wightman,2003;Hu et al.,1994).However,the co-substrate is oxygen,which could2392G.S.Wilson,R.Gifford/Biosensors andBioelectronics20(2005)2388–2403potentially deplete the oxygen available as a co-substrate for the analyte specific oxidase that may be employed.We have also recently discovered that NO,evolved as part of the acute inflammatory response to a sensor implant,may contribute to the observed signal as it can be oxidized at0.6V(Gifford et al.,2005).2.4.InterferencesIt has long been recognized that using oxygen as the me-diator for oxidase-based sensors carries a price.The sensor response is oxygen dependent and,if peroxide is measured, a relatively high potential(0.6V versus AgCl/Ag reference) will be required.This increases significantly the number of endogenous species that can contribute to the observed sig-nal.Mediators have been developed that lower the applied potential into a window around0.0V,where few species are electroactive.These often use horseradish peroxidase (HRP)coupled to an osmium complex or polypyrrole to wire the enzymatic electron transfer directly to the electrode (Georganopoulou et al.,2000;Gajovic et al.,2000;Mao et al.,2002;Kulagina et al.,1999).Such strategies avoid the direct oxidation of electroactive interferences,but they do not avoid homogeneous chemical reactions leading to perox-ide destruction and a low signal.If a mediator is used,there is always the possibility of oxidation of reduced enzyme by oxygen and this will have a parasitic effect on the substrate signal.Few mediators are capable of competing with freely diffusing oxygen,which for glucose oxidase generates an enzyme turnover rate of about1000s−1.Willner’s group has devised highly efficient“wiring”schemes that can compete effectively with oxygen,thus rendering the sensor response truly independent of oxygen(Katz et al.,2002).These sys-tems are not100%effective by themselves,often requiring a combination of exclusion membranes(Nafion TM,cellulose acetate,polylysine,etc.)to provide adequate selectivity.Other than permselective membrane approaches to inter-ferences a dual electrode system has been employed where a blank electrode(minus the molecular recognition element) detects the background current(Burmeister and Gerhardt, 2001).A now classical approach to this problem was pro-posed by Gough(Gough et al.,1985)in which two sensing electrodes are employed:one measures the ambient oxygen level,the other the concentration of oxygen in the chamber where the oxidase reaction is taking place.The difference sig-nal is relatively independent of ambient oxygenfluctuations and electrochemical detection takes place behind a polymeric membrane that can effectively exclude virtually all electroac-tive interferences.2.5.Temporal/spatial resolutionThe importance of detecting transient events that occur in the second to millisecond time frame has been emphasized (Pellerin and Magistretti,2004).Fig.1emphasizes the sig-nificance of temporal and spatial resolution obtained in this case byfluorescent microscopy detection of NADH and gives an indication of the performance characteristics that will be required of a biosensor(Kasischke et al.,2004).Thus,other methods with temporal resolution of1–15s for multi-cell domains,and millisecond time scale for single cell events are required(Clark et al.,1998).ROS lifetimes tend to be very short and the effect of diffusion can yield concentra-tions in the nanomolar range if the detection method is too slow and the sensor is not properly placed near the source of the evolved species(Mao et al.,2002;Clark et al.,1998).Im-planted biosensors are superior tofluorescent imaging tech-niques due to their small size,micron to submicron diameters (Wightman et al.,1999).To measure cellular dynamics,the capability to detect neuronal communication between cells at distances of10–100␮m and for bundles of cells distances of100–500␮m is required(Clark et al.,1998).To measure synaptic release events,it will be necessary to take into ac-count vesicles with a size of ca.50nm,with distances be-tween synapses(site of release and uptake)in the single digit nanometer range(Clark et al.,1998;Pellerin and Magistretti, 2004).Analyte measurements within these dimensions re-quire probes at most in the low micron size range of1–10␮m, most often fabricated with carbonfibers,as demonstrated by measurements of ascorbate and glutamate(Kulagina et al., 1999).Ultramicroelectrodes have been used to measure ex-ocytotic events from pancreatic␤-cells with a probe diam-eter of0.9–1.4␮m(Paras et al.,2000).The sensor dimen-sions have a significant impact on the temporal resolution.A correlation between electrode size and response time,where 270␮m diameter probes produced92±10ms response times and1.5␮m probes produced190±14ms response times has been demonstrated(Meyerhoff et al.,1999).The effect of dif-fusion on the measurement of species released by cells is now well established.Convolution of the observed signal can re-sult from the dilution of the released species as it moves away from a cell yielding a broader signal of lower inten-sity(Wightman et al.,1999).The response can be further convoluted by the presence of the multilayer structures that characterize most biosensors.Improved performance will re-sult if the multiple layers can be made very thin.In addition, the kinetics of the associated enzymatic reaction must also be considered.The importance of simultaneous multi-analyte in vivo measurements was illustrated in the dentate gyrus of the hip-pocampus of a rat where it was possible to detect spikes of10–15s duration indicating rapid changes in lactate,glu-cose,and oxygen in response to repeated neuronal stimulation (Hu and Wilson,1997a,b).This study,using a miniature im-planted electrochemical sensor(ca.110␮m)with a response time of about5s is contrasted to microdialysis studies where the fastest temporal resolution achieved is ca.1–2min,with most around10min(Yao et al.,2004;Yang et al.,2001).Even so the observed sensor response is convoluted by its response characteristics.Slow response translates to measurements that are averages of multiple events,which in the case of mi-crodialysis,is further exacerbated by diffusion effects in theG.S.Wilson,R.Gifford/Biosensors andBioelectronics20(2005)2388–24032393sample tubing.Diffusional and temporal improvements can be achieved by placing biosensors directly in the microdial-ysis sampling probe which is demonstrated by the ability to see concentration spikes with an immunosensor as compared to the standard method(Cook and Devine,1998).The advan-tage of the improved temporal resolution was demonstrated by Astra–Zeneca using a glutamate biosensor(Smagin,2004) compared to microdialysis.An increase in glutamate was ob-served after stimulation as multiple30–60s spikes compared to an average concentration increase over about10min dura-tion as detected by microdialysis.2.6.Simultaneous analyte detectionMany of the previously cited references illustrate the de-sire of researchers to detect multiple analytes simultaneously. The utility of this approach mentioned above,was demon-strated with three different analyte sensors implanted to mon-itor glucose,lactate,and oxygen simultaneously,thus show-ing the temporal relationship among the three species and ad-dressing the issue depicted in Fig.1:lactate as a brain energy source(Hu and Wilson,1997a,b).Many other reports mon-itor multiple analytes including glucose,lactate,pyruvate, glutamate,NADH,ascorbate,and choline based primarily on microdialysis sampling(Yao et al.,2004;Yang et al.,2001; Revzin et al.,2002;Cui et al.,2001;Parkin et al.,2003).An-other study measuring changes in dopamine response under various metabolic inhibitors and effectors demonstrated the interrelationship between glutamate,dopamine and H2O2as a messenger(Avshalumov et al.,2003).Neurological oxygen metabolism has been correlated with NO release,while the implications of ROS to astrocyte defense mechanisms have been discussed(Brown et al.,1997;Wilson,1997).These re-ports,coupled with the desired performance aspects outlined above,serve to illustrate the need for miniaturized multian-alyte implantable sensors to measure various combinations of the analytes described.As of this writing only one mul-tianalyte sensor of>500␮m diameter(lactate and glucose) tested in vivo has been reported in the literature(Ward et al., 2004).Smaller sensor arrays have been used to correct for background current(Burmeister et al.,2003).Many on-line type arrays have been developed employing microdialysis or flow injection methodology(Boutelle et al.,1996;Dempsey et al.,1997).If the various sensing elements in the array are to provide independent results,then they must not interfere with each other.Since many of these sensors must consume analyte in order to make a measurement,two kinds of“cross-talk”must be considered,electrical and chemical.The former results primarily from capacitive coupling(Sreenivas et al., 1996),the latter from overlapping diffusion layers.This lat-ter question has been examined by Yu and Wilson(2000) and Sandison et al.(2002).The optimal center-to-center dis-tance depends on a number of factors,but distances of less than100␮m will generally prove problematic due to diffu-sion layer overlap if the same species are diffusing to adjacent electrodes.3.BiocompatibilityIf a sensor is to provide a reliable reflection of the analyte concentration in the surrounding tissue,the mutual interac-tions of the sensor and the biological medium must not in-fluence results.For all in vivo measurements,the implanted device perturbs the environment and initiates a response.This can translate into ca.50%loss of sensitivity in vivo as com-pared to in vitro values(Khan and Michael,2003;Grams-bergen et al.,2003;Cui et al.,2001;Clark et al.,1998).If the sensor does not produce the expected response it is often difficult to identify the causes of such behavior since they are numerous.Biocompatibility research on materials and sensors has made it clear that biocompatibility does not mean that an implant is inert,which was the original definition.Rather biocompatibility has been defined operationally as:minimal perturbation of the in vivo environment and likewise the in vivo environment does not adversely affect the sensor perfor-mance(Reichert and Sharkawy,1999).An excellent description of the inflammatory response to implanted devices is presented by Anderson(1993).The acute inflammatory response starts immediately after the sen-sor is implanted.During the initial acute response,fluid car-rying plasma proteins and inflammatory cells migrate to the site of the foreign body(i.e.biosensor).Proteins are adsorbed initially and then phagocytic cells(neutrophils,monocytes, and macrophages)surround the biosensor and attempt to de-stroy it.However,because a biosensor is relatively large,only ‘frustrated phagocytosis’occurs,seen as the release of re-active oxygen species[ROS(H2O2,O2−,NO,OH−)]and enzymes intended to degrade the implant.The exact timing, action,and intensity of the process are dependent on the na-ture of the foreign body,which relates to size,shape,and physical and chemical properties.The acute response lasts about3days after which a chronic inflammatory response may set in or a modified version of the healing process be-gins.Ultimately afibrotic capsule is formed,which is the hallmark of the foreign body response.It has been suggested that biocompatibility for implants be considered from a perspective linked with the events of the inflammatory response(Williams,1989).For biosensors, this translates to three considerations with increasing long-term importance:(1)influence of the initial inflammatory events,specifically adsorption of biomolecules;(2)effect the implant has on the local host response that may be coupled to fluctuation in sensor response;(3)biosensor degradation.A thorough review of the inflammatory(biofouling)events with respect to glucose sensors has been provided(Wisniewski et al.,2000).The theories contributing to biosensor perfor-mancefluctuation and potential solutions are reviewed here.3.1.Initial inflammatory response eventsThe overwhelming majority of researchers agree that bio-fouling or adsorption of biomolecules on or infiltrated into the。