Evaluation of a subcutaneous glucose sensor out to 3 months in a dog model

To advance the feasibility of an implantable long-term glucose sensor with bioprotective sensor membranes and test protocols using a somatostatin analog (octreotide).

Implantable sensors were constructed with one of eight bioprotective membranes and screened in vitro for stable response to glucose. Sensors were implanted subcutaneously into nondiabetic mongrel dogs and monitored at 4-min intervals via radiotelemetry. When implanted sensor responses showed evidence of tracking blood glucose after glucagon challenge (8-21 days postimplant), a glucose infusion protocol was used to assess performance. Sensor data were collected every 4 s after octreotide inhibition of endogenous insulin release. Reference plasma glucose samples were taken every 4-10 min.

Preimplant in vitro testing of sensors verified linearity to 33.3 mM glucose and response times to 90% of equilibrium in 2-7 min. Ten implanted sensors tracked glucose for 20-114 days, during which 25 separate glucose infusion studies were conducted. The resulting regression data yielded a mean slope of 0.99 +/- 0.06, an intercept of 0.24 +/- 0.53 mM glucose, and a correlation coefficient 0.98 +/- 0.01. Long-term sensor stability was not judged adequate for clinical application, although two sensors tracked within +/- 15% for 33 and 42 days. In vivo oxygen delivery was shown to affect sensor performance. On explant, two of eight tested bioprotective membranes were found to be biostable and to fully protect the sensor's enzyme membrane. The foreign body capsule was adequately vascularized adjacent to the sensor up to 91 days postimplant. Sensor units eventually failed because of electronic problems (package leakage) or because of biodegradation or biofouling of test bioprotective membranes.

Further development of this type of sensor may provide diabetic patients with a better means of monitoring blood glucose.