A subcutaneous glucose sensor with improved longevity, dynamic range, and stability of calibration

To evaluate the lifetime, response time, linearity, glucose range, and calibration stability of two different types of continuous glucose sensor implants in a dog model.

Glucose sensors based on the enzyme electrode principle that are coupled to a radio transmitter were evaluated on the bench top, sterilized, and then implanted subcutaneously in nondiabetic mongrel dogs. A multichannel radio receiver and PC data processor were used to record the sensor glucose data. Initial early reliable sensor responsivity was recognized by a vigorous hyperglycemic excursion after an intramuscular injection of glucagon. Periodically the dogs were made temporarily diabetic by blocking pancreatic insulin secretion by subcutaneous injection of a synthetic somatostatin (octreotide). By using exogenous insulin injection followed by intravenous glucose infusion, glucose levels were manipulated through the entire clinical range of interest: 2.2-38.9 mmol/l (40-700 mg/dl). Every 5-10 min, reference blood glucose samples were obtained and run in our hospital clinical laboratory. The glucose sensor data was evaluated by linear least squares optimization and by the error grid method.

Beginning as early as postimplant day 7, the in vivo performances of sensors were evaluated by using glucose infusion studies performed every 1-4 weeks. Bench-top and in vivo 90% response-time sensors were in the range of 4-7 min during sensor lifetime. Best-performing sensors from both types are summarized as follows. The earlier-stage technology was less linear with a dynamic range of no more than 22 mmol/l glucose, had a best-case recalibration interval of 18 days, and had a maximum lifetime of 94 days. The improved later-stage technology sensors, which were constructed with the addition of bioprotective and angiogenic membranes, were linear over the full extended range of clinical interest (2.2-38.9 mmol/l [40-700 mg/dl glucose]), had a best-case recalibration interval of 20 days, and had a maximum lifetime of >160 days.

Stable clinically useful sensor performance was demonstrated as early as 7 days after implantation and for a sensor lifetime of 3-5 months. This type of subcutaneous glucose sensor appears to be promising as a continuous and painless long-term method for monitoring blood glucose. Specifically sensors with top-layer materials that stimulate angiogenesis at the sensor/tissue interface may have better dynamic measurement range, longer lifetimes, and better calibration stability than our previously reported sensors.