The majority of diabetes patients must prick themselves several times a day to monitor their glucose level. They then self-administer insulin or medication to ensure that their level remains within the appropriate range. Several light-based techniques promise to become noninvasive alternatives to traditional monitors by measuring glucose-induced changes in skin. However, several factors, such as variations in salt levels and in temperature as well as skin heterogeneity, can make glucose measurement with these noninvasive techniques difficult because they lower the signal-to-noise ratio. The interferometric technique optical coherence tomography (OCT) has proved to be a very effective technique for separating the light scattering caused by glucose from background scattering, contend researchers who are developing a sensor based on the technique; they are from BioTex Inc. in Houston and from the University of Texas Medical Branch in Galveston. Interferometry distinguishes the signal from background based on the principle that two waves with opposite phases will cancel each other out, whereas two waves with the same phase will add up. However, the researchers said that even OCT has problems detecting glucose levels above background. For that reason, they are developing a skin-implantable sensor that makes OCT detection more specific to glucose. Ralph Ballerstadt of BioTex, who led the team, said that, ideally, the sensor will stay in the skin for extended periods of time, and that the patient will become accustomed to having the sensor in his skin and will not even think about it. A patient would carry on his body an OCT monitor that would tell him his current glucose level. Type I diabetes patients also might carry a pump that automatically delivers insulin when the monitor indicates that it is needed. Ballerstadt believes that an implantable sensor would be more attractive to type I patients because they must poke themselves more often than type II patients.Conceptually, the sensor fits just below the top layer of skin in the upper portion of the dermis. The sensor helps OCT detect glucose levels in the interstitial fluid that surrounds cells and tissues and that is in contact with the sensor. Glucose in the fluid diffuses through the semipermeable cellulose membranes of the sensor to two compartments, one sensitive to glucose and the other insensitive to glucose, which could contain interstitial fluid or saline. OCT compares these compartments by interferometry. A mirror coating at the bottom boosts the signal (Figure 1)Figure 1. Researchers are developing a glucose sensor based on optical coherence tomography (OCT) that is designed to fit in the dermis of the skin of diabetes patients. The sensor contains glucose-sensitive and glucose-insensitive compartments (S = signal; Δ = change; Glc = glucose; w/ = with; w/o = without). Images reprinted with permission of Analytical Chemistry.The glucose-sensitive compartment would contain concanavalin A — a large molecule that binds specifically to glucose — and Sephadex from Sigma-Aldrich of St. Louis. Sephadex consists of porous beads of cross-linked dextran, which is made of multiple glucose molecules. Because concanavalin A binds to glucose, it stays bound to open glucose chains within the beads in the absence of glucose in the interstitial fluid. The solution appears cloudy because the majority of the concanavalin A remains in the beads, Ballerstadt said. Once free glucose in the interstitial fluid enters the sensors, concanavalin A leaves the beads and preferentially binds to the free glucose outside the beads. This causes the solution to become clear. The sensor is based on OCT detection of this glucose-induced change in clarity.To verify this concept, the researchers placed concanavalin A and Sephadex in a test tube and added free glucose. Before they added free glucose, the test tube was cloudy, whereas it became clear after they added glucose. This turbidity change was visible with the naked eye. Light microscopy confirmed that the concanavalin A remained inside the particles in the absence of free glucose and left the particles in its presence (Figure 2). Figure 2. In the absence of free glucose, the solution remained cloudy. When researchers pipetted free glucose in the test tube, the solution became clear. The proposed glucose sensor is based on detection by OCT of this glucose-induced change. Light microscopy showed that clusters of concanavalin A remained inside the beads in the absence of free glucose and exited the beads once free glucose was pipetted into the solution. Next they calculated the scattering coefficients of the solutions in the test tubes from turbidity measurements made at 750 nm with a PerkinElmer benchtop spectrometer, which they used for convenience. They found that scattering increased with turbidity. Ballerstadt said that they do not yet know the physical mechanism that explains this phenomenon. They hypothesize that either concanavalin A clustering within the beads or a change in the refractive index mismatch between the beads and solution could cause the scattering change.The researchers next experimented with large and small shredded beads, again using the spectrometer. The small shredded beads yielded a three times larger scattering coefficient and a four times greater sensitivity for glucose.Finally, the scientists used an OCT system to measure at 1300 nm two prototypical sensors, a rectangular disc-shaped sensor and one constructed with hollow fibers. The disc-shaped sensor took quite a long time — 23 minutes — to reach an appropriate signal-to-noise ratio. However, the hollow-fiber sensor achieved an appropriate ratio in less than 5 min. With a 0.4-mm tissue phantom over the disc-shaped sensor, they measured glucose levels down to 10 mM. The researchers concluded that they have demonstrated the feasibility of the sensor in vitro. Their experiments are detailed in the Sept. 15 issue of Analytical Chemistry.Ballerstadt said that they acknowledge the challenges of eliminating the attenuation of light as it penetrates skin and of preventing the body from rejecting the sensor. He also said that they have performed preliminary in vivo experiments in guinea pigs and rats that show how the sensor measures glucose in skin, and that they have succeeded in making a sensor that lasts for six months in solution at 37°C.