Development of implantable glucose sensors based on electrically 'wired' glucose oxidase

Implantable amperometric glucose sensors based on electrically connecting the reaction centers of the enzyme glucose oxidase to gold electrodes by the electron conducting hydrogels formed through crosslinking the redox polymers poly- ((vinylimidazole)Os(bipyridine){dollar}sb2{dollar}Cl) {dollar}sp{lcub}+/2+{rcub}{dollar} (termed PVI-Os) and poly- ((1-vinylimidazole)Os(4,4{dollar}spprime{dollar}-dimethylbipyridine){dollar} sb2{dollar}Cl) {dollar}sp{lcub}+/2+{rcub}{dollar} (termed PVI-Os-DME) were constructed and characterized in vitro and in vivo. The 0.3-mm-diameter flexible sensors were made by depositing polymer solution droplets of the "wired" glucose oxidase sensing layer in a recessed cavity at the tip of a polyimide-insulated 0.25-mm gold wire. The sensing layer was then overcoated with various layers to (a) increase the linear range of the sensor by restricting the transport of glucose, (b) improve the selectivity of the sensor for glucose by eliminating the electrooxidation of interferants such as ascorbate, urate and acetaminophen, (c) increase the operational and storage stability of the sensor, and (d) improve the biocompatibility of the sensors.; Experiments in both Sprague-Dawley rats and humans demonstrated that (a) the sensors are sufficiently selective for glucose to allow their one-point in vivo calibration through independent analysis of one sample of withdrawn blood, (b) are stable enough for continuous subcutaneous in vivo monitoring over a 3 day period, (c) respond to glucose over the clinically relevant glucose concentration range of 2-30 mM, and (d) are comfortable to wear and cause no adverse effects. When the sensors were implanted in either rats or humans greater than {dollar}>{dollar}92% of the sensor-estimated glucose concentrations were in the clinically acceptable range. By using redundant sensors in rats and applying a statistical probability ratio test in order to exclude unsuitable calibration points, the percentage of the clinical accurate sensor estimates was increased to 99%. When the sensors were used to study the dynamic difference between subucutaneous and blood glucose concentrations following intravenous injection of insulin it was found that the difference can be transiently large. Finally, it was confirmed that the subcutaneous tissue is an acceptable site for monitoring of blood glucose levels at steady state.