OBJECTIVEThe long-term performance stability of minimally invasive glucose biosensors was evaluated in vitro and in vivo.METHODSCoil-type glucose biosensors were constructed using an epoxy-polyurethane membrane. Seven sensors were continuously polarized for 12 weeks in a 5 mM glucose-phosphate-buffered saline (PBS) solution, and the sensor sensitivities were tested weekly. Glucose biosensors (n = 44) were also subcutaneously implanted in rats, and the in vivo sensitivities were determined for up to 4 weeks. Histological analysis was performed on the tissue surrounding the sensors.RESULTSDuring a period of 12 weeks, the normalized sensitivity (S/S(0)) of the sensors tested in vitro first increased from 1.10 +/- 0.13 (week 1) to 2.30 +/- 0.90 at week 6 and then decreased to 1.07 +/- 0.24 at week 12 (n = 7). After 6 weeks, the sensors showed a much more significant response to acetaminophen. With continuous polarization in 5 mM glucose-PBS, the sensor functioned for at least 3 months, or about a half of the observed lifetime of sensors stored in the solution with occasional sensitivity measurements (e.g., tested twice each month). For the 15 implanted sensors that lasted for at least 28 days, the average sensitivities values were 4.4 +/- 2.0 (S(0), in vitro), 3.5 +/- 1.3 (day 7, in vivo), 3.3 +/- 1.1 (day 14), 3.6 +/- 1.4 (day 21), and 2.9 +/- 2.2 nA/mM (day 28). Histological analysis showed that the implanted sensors were covered by a 200-800-mu-thick fibrous capsule after 1 week. Blood vessels were found in the fibrous tissue from day 7 through day 34. In addition, the background current that was observed during in vivo sensor testing could be successfully eliminated by using an enzyme-free sensor.CONCLUSIONThis study confirms that coil-type glucose biosensors based on an epoxy-polyurethane membrane can perform stably in vitro for months and in vivo for weeks.