Anal. Chem. 2002, 74, 5942-5947
In Vivo Biocompatibility and Analytical Performance of Intravascular Amperometric Oxygen Sensors Prepared with Improved Nitric Oxide-Releasing Silicone Rubber Coating Megan C. Frost,† Steven M. Rudich,‡ Huiping Zhang,† Martı´n A. Maraschio,‡ and Mark E. Meyerhoff*,†
Departments of Chemistry and General Surgery, The University of Michigan, Ann Arbor, Michigan 48109
The in vivo biocompatibility and analytical performance of amperometric oxygen-sensing catheters prepared with a new type of nitric oxide (NO)-releasing silicone rubber polymer (DACA/N2O2 SR) is reported. The NO-release silicone rubber coating contains diazeniumdiolated secondary amine sites covalently anchored to a dimethylsiloxane matrix. Narrow diameter (0.9 mm, o.d.) silicone rubber tubing coated with this polymer can be employed to construct functional oxygen-sensing catheters that release NO continuously at levels >1 × 10-10 mol/cm2min for more than 20 h. In vivo evaluation of such sensors within the carotid and femoral arteries of swine over a 16-h time period demonstrates that sensors prepared with the new NO-release coating exhibit no significant platelet adhesion or thrombus formation, but control sensors (non-NO release) implanted within the same animals do show a high propensity for cell adhesion and bulk clot formation. Furthermore, the in vivo analytical data provided by sensors fabricated with NO-release coatings (N ) 9) are shown to be statistically equivalent to PO2 levels measured in vitro on discrete samples of blood. Control sensors (N ) 9) placed within the same animals yield average PO2 values that are statistically different (p < ) 0.05) (lower) from both the levels measured on discrete samples and those provided by the NO-release sensors over a 16-h in vivo monitoring period. Over the past two decades, there has been considerable effort focused on developing intravascular chemical sensors (electrochemical, optical, or both) that are capable of reliably monitoring blood gas and electrolyte levels on a continuous, real-time basis.1,2 To date, such sensors have not become widely used in clinical practice owing to their erratic analytical performance, largely as a result of blood compatibility issues.3-5 Indeed, regardless of the * To whom correspondence should be addressed. E-mail: mmeyerho@ umich.edu. † Department of Chemistry. ‡ Department of General Surgery. (1) Frost, M. C.; Meyerhoff, M. E. Curr. Opin. Chem. Biol. 2002, 6, 633-641. (2) Mahutte, C. K. Clin. Biochem. 1998, 31, 119-130. (3) Banerjee, R.; Nageswari, K.; Puniyani, R. R. J. Biomater. Appl. 1997, 12, 57-76. (4) Benmakroha, Y.; Zhang, S.; Rolfe, P. Med. Biol. Eng. Comput. 1995, 33, 811-821.
5942 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
material used to fabricate such sensors, the biological response of the host system can lead to significant problems. Typically, proteins adsorb onto the surface of the implanted device. Platelets then begin to adhere to the protein layer and become activated, acting as a scaffold to attract cells and to create a clot on the sensor surface.3,4,6 Cell adhesion and eventual thrombus formation changes the microenvironment surrounding the sensor, resulting in sensor readings that deviate from real blood levels of important physiological analytes such as PO2, PCO2, and pH.5-7 Thrombus formation can be avoided by using systemic anticoagulants. However, even a single monolayer of metabolically active cells adhering to the sensor surface can alter the local analyte level at the sensor’s surface. This can result in output values that do not reflect the true physiological state of the patient (i.e., oxygen will be consumed and carbon dioxide produced, lowering the PO2 levels, increasing PCO2, and decreasing pH).7 Similar deviations can be observed when blood flow at the implant site is reduced as a result of vasoconstriction of the artery in which the sensor is implanted.1,5,7 Nitric oxide (NO) has been shown to be a potent inhibitor of platelet adhesion and activation8 and is endogenously released by endothelial cells that line the walls of all healthy blood vessels.9 If materials used to make blood-contacting devices (e.g., sensors) are able to release NO continuously at or above the level of that produced by normal endothelial cells (i.e., 1 × 10-10 mol/cm2min),10 the adhesion and activation of platelets should be inhibited. Consequently, bulk thrombus formation can be prevented, and the analytical performance of intravascular sensors made with these materials may be greatly improved. Indeed, in preliminary studies by Schoenfisch et al.,11 it was demonstrated in an in vivo canine model that NO-releasing polymers can be used to fabricate intravascular oxygen sensors that exhibit enhanced blood compat(5) Mahutte, C. K.; Sassoon, C. S. H.; Hansmann, D. R.; Maxwell, T. P.; Miller, W. W.; Yafusso, M. J. Clin. Monit. 1990, 6, 147-157. (6) Kyrolainen, M.; Rigsby, P.; Eddy, S.; Vadgama, P. Acta Anaestheiol. Scand. 1995, 39 (Suppl. 104), 55-60. (7) Meyerhoff, M. E. Trends Anal. Chem. 1993, 12, 257-266. (8) Radomski, M. W.; Palmer, R. M. J.; Moncada, S. Br. J. Pharmacol. 1987, 92, 639-646. (9) Radomski, M. W.; Salas, E. Artherosclerosis 1995, 118 (Suppl.), 569-580. (10) Vaughn, M. W.; Kuo, L.; Liao, J. C. Am. J. Physiol. (Heart Circ. Physiol.), 1998, 274, H2163-H2176. (11) Schoenfisch, M. H.; Mowery, K. A.; Radar, M. V.; Baliga, N.; Wahr, J. A.; Meyerhoff, M. E. Anal. Chem. 2000, 72, 1119-1126. 10.1021/ac025944g CCC: $22.00
© 2002 American Chemical Society Published on Web 11/01/2002
this material originates from within the polymer, thereby eliminating the problems of diamine and nitrosamine leaching encountered in the original work using the dispersed DMHD/N2O2 NO donor. Most importantly, NO release will occur only locally at the polymer/blood interface and sensor surface (i.e., the diazeniumdiolate cannot leave the polymer). It will be shown that Clarkestyle amperometric oxygen-sensing catheters coated with a thin layer (∼100 µm) of DACA-6/N2O2 SR released NO for >20 h, with fluxes equal to or greater than the level of unstimulated endothelial cells. Further, it will be demonstrated that the biocompatibility and analytical performance of such sensors tested in vivo within the carotid and femoral arteries of swine for 16 h show significantly improved analytical performance and blood compatibility, as compared to control sensors, which do not release NO.
Figure 1. Diagram of DACA-6/N2O2 SR-coated intravascular oxygen sensors. The diazeniumdiolate group decomposes to release nitric oxide (NO) under physiological conditions, and the parent diamine remains within the polymer matrix, covalently attached to the silicone rubber (SR).
ibility and improved in vivo analytical performance. The oxygen sensors fabricated in this earlier work were coated with Dow Corning RTV-3140 silicone rubber that was doped with a diazeniumdiolate NO donor, (Z)-1-[N-methyl-N-[6-(N-methylammoniohexyl)amino]]-diazen-1-ium-1,2-diolate (DMHD/N2O2), homogeneously dispersed within the polymer. This NO donor decomposes under physiological conditions to produce the parent diamine compound and two molecules of NO. These results were promising; however, it was subsequently found that a significant fraction of the DMHD/N2O2 readily leaches from the polymer matrix and decomposes in blood to release NO.12 Furthermore, once formed, the parent diamine compound can back-react with oxidized intermediates of the NO (e.g., N2O3) to form nitrosamines, which are known carcinogens.13 Additionally, because a significant fraction of the NO donor leaches from the polymer prior to releasing NO, the donor can diffuse away from the polymer/blood interface and be washed distally from the sensor, thereby reducing its effectiveness at preventing platelet adhesion and activation on the sensor surface. Herein, we report the in vivo performance and biocompatibility of intravascular oxygen sensors coated with a new type of NOreleasing silicone rubber polymer recently reported by Zhang et al.14 The NO donor in this material utilizes the same diazeniumdiolate group employed in the initial studies;11 however, the parent diamine is anchored to the cross-links of a silicone rubber polymer matrix (DACA-6/N2O2 SR; see Figure 1). All NO released from (12) Mowery, K. A.; Schoenfisch, M. H.; Saavedra, J. E.; Keefer, L. K.; Meyerhoff, M. E. Biomaterials 2000, 21, 9-21. (13) Loeppky, R. N. In Nitrosamines and Related N-Nitroso Compounds: Chemistry and Biochemistry; Loeppky, R.N, Michejda, C. J., Eds. ACS Symposium Series 553; American Chemical Society: Washington, DC, 1994; pp 1-18. (14) Zhang, H.; Annich, G. M.; Miskulin, J.; Osterholzer, K.; Merz, S. I.; Bartlett, R. H.; Meyerhoff, M. E. Biomaterials 2002, 23, 1485-1494.
EXPERIMENTAL SECTION Diamine Cross-Linked Silicone Rubber (DACA-6 SR). Poly(dimethylsiloxane) (PDMS, viscosity 2000 cSt), and dibutyltin dilaurate were obtained from Sigma-Aldrich (Milwaukee, WI). N-(6-aminohexyl)-3-aminopropyltrimethoxysilane (DACA-6)- and hexamethyldisilazane-treated fumed silica were obtained from Gelsdt (Tullytown, PA). Toluene was obtained from Fisher (Pittsburgh, PA). Nitric oxide (NO) and argon (Ar) were obtained from Cryogenic Gases (Detroit, MI). A detailed description of the synthesis and characterization of DACA-6 SR is presented elsewhere.14 Briefly, PDMS (1.6 g), DACA-6 (0.3 g), and dibutyltin dilaurate (3.2 mg) were dissolved in 8 mL of toluene and allowed to partially cross-link for 24 h at room temperature. Fumed silica (0.3 g) was suspended in 4 mL of toluene. Four mL of the PDMS solution and 3 mL of the fumed silica solution were combined and dip-coated onto silicone rubber oxygen sensor sleeves (see below). The DACA-6 SR-coated sensor sleeves were then loaded with NO to form the NO-donating diazeniumdiolate groups (see Fabrication of NO-Releasing Oxygen Sensors, below). Fabrication of NO-Releasing Oxygen Sensors. Silastic medical grade tubing (0.55 mm i.d. × 0.94 mm o.d.) was obtained as a gift from Medtronic, Inc. (Minneapolis, MN), and silicone rubber (RTV-3140) was purchased from Dow Corning Corp. (Midland, MI). Quick-setting two-part epoxy from Super Glue Corp. (Rancho Cucamonga, CA) was obtained from Meijer, Inc. (Ann Arbor, MI). Methocel 90 HG was obtained from Fluka (Milwaukee, WI). Potassium chloride and tetrahydrofuran (THF) were obtained from Fisher (Pittsburgh, PA). Hexamethyldisilazane and phosphate-buffered saline (PBS) were obtained from SigmaAldrich (Milwaukee, WI). Teflon-coated Pt/Ir and Ag wires were products of Medwire Corp. (Mt. Vernon, NY). Ultra 4-way stopcocks were obtained from Medex (Hillard, OH). The NO-releasing Clarke-style amperometric oxygen-sensing catheters were fabricated as previously described,11 the only difference being the nature of the outer NO-release coating on the surface of the silicone rubber tubing. The silicone rubber sensor sleeves were made by cutting ∼26-mm-long pieces of silastic tubing and filling an ∼2 mm length of one end with RTV3140. The sensor sleeves were allowed to dry overnight and were then dip-coated with 8 coats of DACA-6 SR at 10-min intervals. A topcoat of conventional silicone rubber was then applied to the sensor sleeves by dip-coating the sleeves with a solution containing 0.6 g of Dow Corning RTV-3140 dissolved in 4 mL of THF. The Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5943
total thickness of the DACA-6 SR and RTV-3140 layers was ∼100 µm. The sensor sleeves were allowed to cure overnight in ambient moisture. The underlying DACA-6 SR coating was then loaded with NO to form the diazeniumdiolate functional groups (DACA-6/N2O2 SR) in a high-pressure reactor. The sensor sleeves were purged with Ar to remove oxygen and then exposed to 80 psi of NO for 24-36 h. The NO-loaded sensor sleeves were used in the final assembly of the oxygen sensors. The sensors were fixed into 4-way stopcocks with epoxy to allow their secure attachment to 14-gauge cannulae. This also provides a means to attach a slow saline drip to prevent blood from flowing back through the cannula while implanted in vivo. Sensing sleeves for control sensors were prepared in a similar manner, except that the NO-loading step of the cured DACA-6 SR/RTV-3140 coated sensor sleeves was omitted. For the final oxygen-sensing catheters, the working electrode was a Teflon-coated Pt/Ir wire (0.076 mm o.d.). Current was monitored at an applied potential of -0.70 V vs Ag/AgCl using in-house potentiostats, as well as a Diamond Electro-Tech electrochemical analyzer (Ann Arbor, MI). Output currents were recorded using a DATAQ Instruments DATAQ-700 USB data acquisition card (Akron, OH) with the accompanying WinDAQ/ Lite software. NO-Release Measurements. Nitric oxide release from the DACA-6/N2O2 SR-coated sensor sleeves was monitored via chemiluminescence with a Seivers Nitric Oxide Analyzer 280 (Boulder, CO) in PBS, pH 7.4. In Vivo Evaluation. Assessment of the in vivo biocompatibility and analytical performance of the NO-releasing oxygen-sensing catheters was performed by implanting sensors in five juvenile farm swine weighing 20 to 35 kg. The animals were anesthetized using isoflurane and mechanically ventilated with 21% oxygen. Both the stomach and urinary bladder were surgically drained. An arterial line was placed into the splenic artery and used to monitor blood pressure and to acquire samples for in vitro arterial blood gas measurements. Cut-downs were made on both groins as well as the bilateral neck for isolation of the femoral and carotid arteries, respectively. Fluids (lactated ringers) were administered via the external jugular vein at a rate of 300 mL/h. Arterial blood gas measurements were performed using a Radiometer Medical ABL-505 standard blood gas monitor (Copenhagen, Denmark), with samples drawn into heparin-coated syringes. The animal protocol was approved by the University of Michigan University Committee on Use and Care of Animals. For each swine, four sensors were implanted: two NOreleasing sensors and two control sensors. One NO-releasing sensor and one control sensor were placed in the femoral arteries, and the second pair was placed in the carotid arteries via BectonDickinson Angiocath 14-gauge, 1.16 in. cannulae (Sandy, UT). An ∼1-cm length of the sensor extended past the end of the cannula and was exposed to blood flow. The artery distal to the cannula insertion was not ligated. Slow saline drips (0.25 mL/min) containing 1 U/mL of heparin were attached to each sensor to prevent blood from flowing retrograde into the cannulae. Sensors were implanted such that the sensor tip was exposed to oncoming blood flow and the small amount of heparin present in the saline drips was washed distally from the sensor, never flowing over 5944
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
the sensor surface. No systemic anticoagulation agent was administered during the study. Sensors were calibrated via a onepoint, in situ calibration 1 h after implantation (assuming 0 current for 0 oxygen levels and that the observed current corresponds to the blood oxygen level measured in vitro at the 1-h time point). Continuous sensor output was recorded, and periodic blood gas samples were collected and analyzed to assess sensor performance. During the first 2 h, blood samples were drawn every 15 min, and during hours 3-4, samples were drawn every 30 min. During the remainder of the experiment, samples were drawn every hour for comparison to sensor output. In total, 18 sensors were evaluated (9 NO-releasing and 9 control). Outputs from one NO-releasing sensor and one control sensor were discarded due to electrical shorts that occurred during the experiments. Upon termination of each animal preparation, each femoral and carotid artery was excised distal to the implanted sensor. The animal was then euthanized with a bolus injection of Fatal-Plus (Vortech Pharmaceutical, Dearborn, MI). The sensors were carefully dissected from the arteries without pulling them out of the vessels or through the cannula (to prevent scraping off any surface bio-layer), rinsed in PBS to dislodge loosely adsorbed blood elements, and fixed in 1% glutaraldehyde for 8 h. The sensors were then dehydrated in a series of ethanol washes, followed by a hexamethyldisilazane wash and allowed to dry overnight. Sensor surfaces were sputter-coated with gold and examined with a Hitachi S-3200N scanning electron microscope. RESULTS AND DISCUSSION NO-Release DACA-6/N2O2 SR-Coated Sensors. In previous work in this laboratory with DMHD/N2O2 dispersed in silicone rubber, it was shown that significant levels of NO release from the outer gas-permeable membrane/barrier of electrochemical sensors does not interfere with the function of these devices.11,15 Specifically, Schoenfisch, et al. showed that the response time, sensitivity, linearity, and repeatability of these catheter-type oxygen sensors were essentially identical for NO-releasing sensors and control sensors.11 The same equivalency of in vitro analytical performance between the NO-releasing DACA-6 SR-coated sensors and the corresponding control sensors was found in preliminary experiments leading to the in vivo studies reported herein (data not shown). In addition, because NO is also a potent vasodilator, a secondary effect of local NO generation may also be to help reduce vasoconstriction around the implanted sensor, thus maintaining normal blood flow over the sensor. Indeed, initial experiments11 indicated that greater blood flow was maintained in the vessels containing NO-releasing sensors, as compared to flow in vessels containing control sensors. In the present work, sensor sleeves were coated with an ∼100-µm-thick layer of DACA-6 SR that was then loaded with NO to form diazeniumdiolates bound to the polymer matrix (see Figure 1). Because the flux of NO generated from the surface of the sensor sleeve is determined by the amount of DACA-6/N2O2 SR present, the coating thickness was adjusted to produce as thick a coating as possible while still allowing the sensor to fit through a 14-gauge cannula. Figure 2 shows the typical flux of NO generated from the surface of a coated silicone rubber sensor sleeve when incubated in PBS at (15) Mowery, K. A.; Schoenfisch, M. H.; Baliga, N.; Wahr, J. A.; Meyerhoff, M. E. Electroanalysis 1999, 11, 681-686.
Figure 2. Nitric oxide (NO) surface flux from a DACA-6/N2O2 SRcoated oxygen sensor incubated in PBS at 37 °C for 20 h, as measured by chemiluminescence, is greater than the flux of endogenous NO produced by unstimulated endothelial cells (- - -).
Figure 4. Scanning electron micrographs of the surfaces of the sensors shown in Figure 3. In all cases, the NO-releasing sensors show very few adhered platelets, and those that are present are not activated. The control sensors show varying degrees of platelet adhesion and activation, ranging from mature clot formation within a fibrin network of entrapped platelets (1B), to a few adhered and activated platelets near the sensing tip (2B and 3B).
Figure 3. Representative examples of oxygen sensors explanted from carotid and femoral arteries of swine after 16-19 h of implantation with no systemic anticoagulation treatment. The portion of the sensor to the left of the dashed line extended past the cannula and was exposed to flowing blood. In all cases, sensor (A) was NOreleasing, and (B) was the control (i.e., non-NO-releasing).
37 °C over a 20-h period. For reference purposes, the estimated surface flux of unstimulated endothelial cells is shown as the dotted line.10 Provided that the level of NO generated by the sensor is equal to or greater than the level of NO generated by the endothelial cells, it is anticipated that platelet adhesion and activation should be inhibited. In Vivo Performance. Sensors were implanted in porcine carotid and femoral arteries for at least 16 h with no administration of systemic anticoagulation agent. Figure 3 shows photographs of three representative pairs of sensors after they were explanted
from porcine arteries and rinsed in PBS. In each panel, (A) is the NO-releasing sensor and (B) is the control sensor. The three sensor pairs illustrate the range of results obtained when both NO-release and control oxygen sensors were implanted for at least 16 h. The portion of the sensor to the left of the vertical dotted lines was exposed to blood. All of the NO-releasing sensors were quite clean upon explantation and showed no evidence of gross thrombus formation. In contrast, control sensor 1B was completely covered with a mature blood clot. In fact, 3 of the control sensors (N ) 9) exhibited similar clots at their distal ends. The sensing tip of 2B was not covered with a blood clot; however, a large clot did form along its length and near the tip of the cannula. Four of the control sensors possessed similar clots somewhat away from the sensing tip. Control sensor 3B was relatively clean, with no thrombus formation. Two of the control sensors fit into this category upon visual inspection. Figure 4 shows the corresponding SEM images of the surfaces of the same 3 pairs of oxygen sensors explanted after at least 16 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5945
Figure 6. Plot of the average percent deviation of oxygen level measured by all NO-release (0) and control (4) sensors (nine of each type included), as compared to the oxygen level determined by standard blood gas analysis ([). Oxygen level determined by standard blood gas analysis was taken to be the true value; hence, there is 0% deviation from these values.
Figure 5. Plots of the oxygen level determined by the intravascular NO-release (0) and control (4) sensors, as compared to the oxygen level determined by discrete blood gas analysis ([) for each pair of sensors shown in Figures 3 and 4. In all cases, the NO-releasing sensors accurately measured blood oxygen levels. The oxygen level determined by the control sensor in case 1 showed large deviation from the discrete blood gas analysis. In cases 2 and 3, the control sensors showed moderate deviation from the discrete blood gas analysis.
h of in vivo monitoring. The NO-releasing sensors consistently showed virtually no platelet adhesion or activation. The surfaces (at the sensing tip) of the control sensors exhibited a much greater degree of both platelet adhesion and activation in addition to variability in the biological response compared to the NO-releasing sensors. Control sensor 1B had a very mature fibrin network with entrapped platelets at its exposed tip. The sensing tip of control sensor 2B showed few adhered platelets, although it did contain a region of thrombus formation (SEMs not shown; see Figure 2). Control sensor 3B showed very few platelets adhered to the surface, similar to the images obtained for all of the NO release sensors. Overall, nine of nine NO-release sensors appeared as sensor 1A with respect to images observed via SEM of the sensing tips; three of nine control sensors were quite similar to sensor 1B upon SEM examination; four of nine control sensors yielded images similar to sensor 2B; and two of nine controls looked like sensor 3B. Figure 5 shows the corresponding oxygen levels determined by the same three representative sensor pairs shown in Figures 3 and 4, as compared to standard in vitro blood gas analysis, over 5946 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
the 16-h period of implantation. In all three cases, the oxygen level determined by the NO-releasing sensors tracked more closely with the level determined by benchtop blood gas analysis. The control sensors showed variable behavior in their ability to measure blood levels of oxygen accurately. In the case of sensor 1B, the surface of the sensor was completely covered with a blood clot, and the level of oxygen determined by the catheter showed significant negative deviation from the discrete blood gas determination. Sensor 2B was not completely covered with a blood clot and did not show as great a level of negative deviation from the oxygen level measured by discrete in vitro determinations. However, the large clot further up the sensor shaft (see Figure 2) could have caused decreased blood flow around the sensor, accounting for the lower oxygen levels measured. The control sensor 3B was relatively clean and able to measure oxygen levels with only slight negative deviations for the duration of the experiment. It should be noted that the standard deviation of sequentially drawn blood samples (N ) 5) measured in vitro over 10-min periods at the beginning and end of the experiment was less than (1.8 mmHg from the same animal, indicating that the blood oxygen level in the animal does not rapidly change during the time needed to perform in vitro data collection. A problem that has plagued intravascular sensors to date has been the inconsistent analytical results observed after implantation. As shown for all control sensors, such a high degree of variability may relate to the differences in the biological response of a given subject at a given implant site to the sensor, with some becoming completely covered with thrombus while others remain relatively clean and unaffected by clot formation. Indeed, the data shown in Figures 5-1 and 5-3 were from sensors implanted in different sites within the same animal. Such variations in the biological response (for control sensors) could be related to the differences in the diameter of the blood vessels and the concomitant change in linear velocity of the flowing blood in a given artery. Figure 6 summarizes the overall analytical results for in vivo data collected during this study. This figure shows the average percent deviation in oxygen levels determined by all of the implanted sensors compared to the oxygen level determined periodically by conventional benchtop blood gas analysis. Nine
Table 1. Average Percent Deviation in PO2 Determined with Intravascular Sensors time av % dev. for tscorea for av % dev. for tscorea for (h) NO-release sensor NO sensor control sensor control sensor 1 5 10 15
-0.8 3.1 8.5 2.9
0.041 0.089 0.147 0.190
-2.7 -7.0 -24 -28
0.052 0.165 0.195 0.189
a t score calculated with 8 degrees of freedom at 95% confidence interval.
NO-releasing sensors and nine control sensors are compared. Table 1 lists the t-scores calculated for the average deviation of oxygen level determined by the intravascular sensors, as compared to standard blood gas analysis. The difference in deviation between the NO-releasing sensors and the standard blood gas analysis was not statistically significant at the 95% confidence level when tested at 1, 5, 10, and 15 h. The range of average deviation for the NO-releasing sensors was -3.4-11%, with the greatest average level of deviation (+11%) occurring at 12 h. The differences between the oxygen levels determined by the control sensors, as compared to standard blood gas analysis, were not statistically significant after 1 and 5 h but were, however, statistically different after 10 and 15 h at the 95% confidence level. The range of average deviation was -2.7 to -28%, with the greatest average deviation occurring after 15 h. Such behavior is anticipated because the surface of the control sensors becomes further coated with adhered cells and clots as the implantation time increases.
CONCLUSIONS Results presented herein indicate that NO release originating from within a polymer coating used to fabricate intravascular oxygen sensors greatly improves the blood compatibility and analytical performance of such devices. The use of NO donors covalently linked to the polymer eliminates the problems of leaching and blood-born nitrosamine formation while ensuring that NO is present immediately at the polymer/blood interface. This local NO release effectively inhibits platelet adhesion and activation. The nitric oxide-releasing sensors performed reliably in vivo with no statistically significant difference in oxygen levels determined by the sensors when compared to standard blood gas analysis at the 95% confidence level for 15 h. Oxygen levels determined by control sensors without NO release implanted within the same animals deviated (average deviation after 15 h was -28%) from levels determined by standard blood gas analysis in a statistically significant manner at the 95% confidence level. Further work will now focus on determining the optimal level of NO release required to ensure reliable in vivo performance of intravascular sensors for even greater extended time periods. ACKNOWLEDGMENT This work is supported by the National Institutes of Health (NIH GM56991). M.C.F. is supported by the N.I.H. Cellular Biotechnology Training Program at the University of Michigan (T32 GM08353). Received for review July 16, 2002. Accepted October 2, 2002. AC025944G
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5947
