Keeping Watch On Glucose

New monitors help fight the long-term complications of diabetes

11 min read

Imagine sticking yourself with a pin on the pad of your finger up to seven times a day. It's an unpleasant thought, but that is exactly what many diabetics should be doing. Several large clinical studies have shown that tight control of blood sugar slows the progression and development of long-term complications of diabetes, such as blindness and kidney failure. So sufferers must collect multiple blood samples to provide feedback for insulin dosing and other treatment.

Apart from the pain, though, tight control also introduces a risk of severe hypoglycemia--blood sugar so low that it can lead to coma or seizure. The risk exists because tight control seeks to reduce the patient's mean glucose levels to a more normal state, but from that lower point there is a greater probability that glucose levels could accidentally become dangerously low. To counter that threat, new glucose-metering technologies that should provide relatively painless and much safer blood sugar control are entering the market. 

Rushing to pull down the barriers to tight glucose control, several firms have developed devices that all but eliminate the need for blood samples. In June 1999, an implantable sensor providing continuous readings, made by Medtronic MiniMed Inc. (Northridge, Calif.), became the first alternative glucose monitor to gain approval by the U.S. Food and Drug Administration. And Cygnus Inc. (Redwood City, Calif.) followed in March 2001 with its own device, which monitors glucose continuously through the skin.

While a great many innovative technologies have been proposed, only a fraction are in advanced stages of development. They fall into three categories: implantable monitors, transdermal (through the skin) meters, and meters depending on spectroscopic methods. Each is a far cry from the painful methods used over the last several decades.

By offering greater convenience and less pain, the devices encourage people to test more frequently, the extra testing providing previously unobtainable information on trends in glucose levels in response to insulin dosage and other treatments. Also, monitors that provide frequent, automatic readings can be equipped with preset alarms to warn the user of high or low glucose levels. Parents of children with diabetes, for example, can be warned of overnight low glucose, rather than having to wake up each night to check their child's blood sugar. These advances will lead to better decisions about treatment and ultimately reduce the long-term medical complications of diabetes.

Metering from the inside

Implantable glucose sensors are inserted beneath the skin or inside a blood vessel. This technology has been in development for quite some time, but the first product, MiniMed's monitor, became commercially available only in the past few years [see ]. Such monitors, which rely on advances in chemical sensors and biocompatible materials, are a step toward the ultimate goal: a glucose sensor that can be connected directly with an insulin delivery system to make an artificial pancreas, the organ that controls blood glucose levels in the body.

MiniMed's device operates for up to three days--not a permanent solution, but one that yields enough information to improve a person's treatment regimen. The device records glucose levels for a healthcare professional to view on the patient's next office visit. First the user performs two to four finger-prick blood glucose measurements to calibrate the device, which then supplies readings at five-minute intervals. Clinical researchers report identifying low glucose levels at night and high glucose after a meal that were previously unobserved by periodic pinpricks.

MiniMed's product is an example of an enzymatic electrochemical sensor. In brief, an enzyme called glucose oxidase is fixed to an electrode and catalyzes the conversion of glucose into gluconic acid and hydrogen peroxide. The hydrogen peroxide then reacts at the sensing electrode, which is typically biased at 0.6 V, resulting in a measurable electric current.

Generally, implantable sensors can be categorized by the site of implantation and the method of measuring glucose. Subcutaneous sensors are inserted beneath the skin through a needle and measure glucose in interstitial fluid, the liquid between the cells. Other sensors are surgically affixed to the inside of a large vein and measure glucose in blood. Most sensors, like the MiniMed device, employ an enzymatic conversion step to turn glucose into a chemical signal that can be easily measured electrochemically or optically.

The main challenge in developing a glucose sensor for implantation beneath the skin or in a vein is to maintain the sensor's performance when it is exposed to the inside of the body over long periods of time. Almost without exception, interactions with the body cause a decrease in sensor performance [see illustration, ]. For example, the body's immune system inevitably launches an attack and tries to encapsulate the sensor in protein. The glucose-blocking barrier thus created blunts sensor sensitivity and lengthens response time.

On another front, the body's warm, salty environment corrodes metal electrodes and can inactivate enzymes, which leads to loss of measurement sensitivity and stability. Movement by a person wearing the device can create artifacts and noise that decrease sensitivity and specificity to glucose signals and also produce mechanical stresses that affect stability.

Many other interactions with the body's environment must also be dealt with. For instance, substances such as vitamin C and acetaminophen may react at the electrode, creating spurious signal. Such chemicals can also destroy the hydrogen peroxide before it can react at the electrode, stealing signal from the system. To minimize this effect, many implantable systems include membranes that keep these substances from the sensor.

Another problem is that when glucose levels are high, oxygen may become the limiting reactant in the electrochemical sensing scheme that MiniMed and others use. The result is signal saturation and a limited system operating range. To combat this drawback, some investigators have introduced membranes that limit the amount of glucose that reaches the sensor. Or they do away with the need for oxygen by using sensing schemes that rely on alternative reactions.

Still another defense against the body's attacks is microdialysis. In this technique, dialysis tubing, made from a material that allows only small molecules to pass across it, is implanted under the skin. The tubing has a special fluid pumped through it, into which glucose diffuses. The fluid is then collected and measured with an external sensor. This strategy prevents proteins from encasing the sensor.

The design of MiniMed's sensor addresses some destructive interactions with the body. It is built on a flexible substrate to minimize the effects of motion and to fit more comfortably in the patient. The sensor is also coated with a biocompatible polyurethane to minimize the immune system's response.

Besides the subcutaneous types, some fully implantable glucose measurement systems are being developed. These systems have the ambitious goal of providing continuous blood glucose measurements for years on end and interfacing with an implantable insulin pump. The result is closed-loop control of glucose levels--in effect, an artificial pancreas.

Medical Research Group, known as MRG (Sylmar, Calif.), is developing such a device to measure glucose in the right atrial junction--the wide entrance to the right atrium of the heart. It is at this junction, where blood flows fast through a large diameter vessel, that the device is least likely to cause blood clots or blood flow perturbations. By obtaining a signal directly from blood instead of from subcutaneous tissue, the sensor's response time is minimized, making it more suitable than the subcutaneous type would be for a closed-loop system.

The sensor has been tested in dogs and in humans for months at a time, and the closed-loop system has controlled glucose levels in dogs for a full week.

Metering from the outside

Although implantable systems represent a big advance in convenience and frequency of readings, the need to implant the device into the body is a major obstacle to widespread use. Sampling through intact skin is an alternative. Broadly speaking, this transdermal technology is analogous to a nicotine delivery skin patch, except that the molecule of interest is moving out of, rather than into, the body.

The skin is normally a formidable barrier, so transdermal technologies require some type of transport enhancer to obtain a sufficient sample for analysis. Techniques to enhance transport include using electric current (iontophoresis), acoustic energy (sonophoresis), chemicals that enhance the permeation of glucose through skin (passive diffusion), or mechanical permeation (microporation). These techniques can cause a slight irritation, but many patients may find it worth the trouble.

Even with the assistance of transport-enhancing methods, the sample of fluid obtained can be orders of magnitude lower in glucose than the concentration in the blood. It will also be extremely small in volume, on the order of microliters. Both factors mandate innovative sensors or novel fluid-handling systems, or both.

Current and chemistry

Cygnus'device extracts glucose through the skin by iontophoresis and measures the extracted sample electrochemically, using the glucose oxidase reaction [again, see ]. It is meant for use in conjunction with traditional measurements to assist in diabetes management. Worn like a watch, it provides the wearer with up to three glucose readings per hour for up to 12 hours. The user first calibrates it with a blood glucose measurement using a traditional, pinprick glucose meter. The device contains user-programmable high and low glucose alerts to warn patients of dangerous blood glucose levels.

The watch consists of an electronic controller and a disposable electrochemical sensor, which is replaced with each 12-hour measuring period The controller contains the user interface, power supply, and electronics for the sensor. The sensor snaps into the underside of the controller and adheres to the skin. Each sensor contains electrodes to apply the iontophoretic extraction current, two hydrogel disks for collecting the glucose, and electrodes to sense the glucose.

Every 20 minutes, the device goes through an extraction and a sensing cycle. First, 0.3 mA of current is applied through the skin. Ions, which make up the current, drag glucose with them into the hydrogel disks. During sensing periods that follow extractions, the glucose undergoes the previously described enzymatic electrochemical reaction.

The current generated by the electrochemical reaction is converted into a glucose measurement by a signal-processing algorithm, which takes a number of factors into account. The calibration blood glucose measurement plugs into the algorithm to account for variations in skin permeability.

Sweat, which could interfere with measurements because it contains glucose itself, is detected by measuring the skin's conductivity through perspiration-detecting probes located on the underside of the device. A thermistor built into the device watches for temperature fluctuations that could interfere with the sensor signal. The algorithm considers all these factors for each extraction-sensing cycle and indicates if a measurement might be compromised and should be thrown out. As it turns out, about 25 percent of the readings are faulty.

Ultrasound pretreatment

Sontra Medical Inc. (Cambridge, Mass.) is working on a sonophoretic device that uses low-frequency ultrasound to enhance transport of glucose across the skin. Low-frequency ultrasound causes the formation and collapse of pockets of dissolved gases in a gel applied to the skin. This cavitation creates microscopic jets of fluid that disrupt the outer layer of the skin, making pathways through which glucose containing interstitial fluid can exit.

The ultrasound is applied as a pretreatment, and a glucose collection device is then pressed to the skin. Research has shown skin to remain permeable for at least 12 hours, and preliminary studies indicate that the extracted glucose correlates well with conventional blood glucose measurements.

Passive diffusion

Other groups such as Technical Chemical and Products Inc. (TCPI) (Pompano Beach, Fla.) took the analogy with the nicotine patch even farther. Over a five-minute period, the TCPI system extracts glucose into a transdermal patch that contains a chemical skin permeation enhancer. The patch also includes reagents that change color on contact with glucose. An optical meter then reads the magnitude of the color change.

Access through micropores

SpectRx Inc. (Norcross, Ga.) has a system in which a specially designed handheld laser forms an array of micropores in the skin, each hole about the width of a human hair. Once the skin has been perforated, so to speak, a user dons a system whose vacuum suction harvests a sample of interstitial fluid, which is then passed over an electrochemical glucose sensor. This device potentially allows for both single-use and continuous glucose sampling.

Along a similar line, PowderChek Diagnostics (Fremont, Calif.), a spinoff of drug delivery firm PowderJect Pharmaceuticals PLC (Oxford, UK), is working on a system that uses a high-pressure blast of fine particles to make holes in the skin, followed by a vacuum to extract interstitial fluid. The skin poration step is done with a handheld device that is currently used for needle-less injection of pharmaceuticals and vaccines.

Seeing sugar

The field of noninvasive glucose monitoring began about a decade ago with the promise of devices that could enable painless and bloodless measurements by shining a beam of near-infrared light through the skin. That ideal remains a goal for many researchers today.

Researchers have tried to make measurements using a number of different device configurations and at different parts of the body, including transmission through the ear lobe, tongue, or finger web, and reflectance from forearm skin and the inner lip. Appealing as this concept may appear, however, many technical challenges still block this and other techniques that depend on looking for glucose's spectroscopic signature.

Near-infrared (NIR) methods use a wavelength slightly longer than visible light (between 1000 and 2500 nm). Because water does not strongly absorb in this spectral range, NIR radiation can penetrate further through the skin than either visible or mid-IR radiation, sampling a thicker layer of tissue.

However, because glucose also absorbs weakly in the NIR and is present at much lower concentrations than water, raw glucose signals are small and can be obscured by noise.

graph normalized absorbance

Spectroscopic monitors search near-infrared light (1500-2500 nm) transmitted through tissue, such as the ear lobe, for the absorption signature of glucose [orange]. The problem with this approach is that the absorbance of water [blue] dwarfs the glucose signal. The two spectra have been normalized so they can be displayed.

What's more, the spectral features of glucose are broad and overlap those of other substances such as proteins and fat tissue, all of which are found in large concentrations in the body.

Obtaining glucose measurements with adequate precision and dynamic range when the glucose absorption may total only 0.01 percent of the signal has not been easy. In fact, it has required advanced signal-processing routines. To extract the glucose signal out of the noise, the absorbances at a number of distinct wavelengths are measured. A correlation between these signals and the actual blood glucose concentration is then modeled to calibrate spectroscopic devices.

Modeling of this type usually must be done on an individual basis, and collecting the large number of data points needed can take several months, making the method less practical than others. Changes in the amount of liquid in the tissue, fat content, skin coloration, temperature, and electrolyte levels all can alter the stability of the calibration, and recalibration may be required periodically.

A prototype NIR device calibrated to individual users has been demonstrated by Instrumentation Metrics Inc. (Chandler, Ariz.). In the calibration process, the patient intentionally undergoes large changes in blood sugar over the course of several hours and produces multiple blood samples for analysis by a laboratory glucose meter. The laboratory values are then combined with data from the spectroscopic device to develop a user-specific calibration algorithm. The company also has investigated the use of a global approach, to provide a universal calibration that will work for all subjects. So far preliminary data on several dozen diabetics correlates well with their blood glucose levels.

One company, Animas (Frazer, Pa.), is combining spectroscopic technology with an implantable system. Its device searches for glucose's spectral signature using NIR transmitted across a vein. It is then intended to radio the signal out to an external display unit.

Research groups have tried a number of ways around the technical challenges of spectroscopic glucose monitoring. Using alternative wavelengths of light is one. Mid-infrared radiation, usually considered to be highly absorbed by water, does have a wavelength region, between 6 and 12 um, where water is only weakly absorbing. Although the absorbances of other compounds overlap that of glucose in this region, researchers hope that removing the interference from water--65 percent by weight of any tissue sample--will increase the glucose signal-to-noise ratio considerably.

Other techniques rely upon the interaction between glucose and specific polarizations of light to increase measurement selectivity.

Better treatment

While the methods of monitoring glucose have multiplied, each may find its own niche, because different technologies can address different user needs. The spectroscopic techniques could suit people who do not have advanced diabetes and require only infrequent measurements. Meanwhile, frequent monitoring is beneficial to people who must use insulin to control blood sugar, and also as an educational tool in helping any diabetic person to understand the effect of diet, activities, and other factors on glucose patterns during the day and night.

While transdermal technologies may be more acceptable than implantation to many users, others who are already accustomed to multiple injections or the use of insulin pump catheters may find an implantable system best.

Traditional technologies will have their place, too. Finger pricks will continue to be a necessary part of diabetes care in the near future, as the new technologies still require traditional skin prick meter readings for calibration and for confirmation of critical glucose readings. Even such blood glucose measurements, however, are becoming less painful [see sidebar, "Improvements in Pinpricks and Fingersticks"]. On top of the greater convenience of testing that the new products will provide, frequent, automatic glucose monitoring holds the promise of improving diabetes care. This technology, ushered in by implantable and transdermal glucose sensors, can supply information on the effects of diet, insulin-dosing algorithms, and glucose-lowering medications on blood sugar levels that was previously unattainable. More information gives people with diabetes more power to fight their disease.

—Samuel K. Moore, Editor

To Probe Further

A number of glucose monitors on the market and in development are reviewed at

LEOS Newsletter, from the IEEE Lasers and Electro-Optics Society, contains a series of articles about spectroscopic methods for glucose measurement. See Vol. 12, no. 2, April 1998.

Cygnus Inc. is finalizing its manufacturing capabilities for its blood sugar monitor. For more, see

The iontophoresis system is described in "The GlucoWatch Biographer: A Frequent, Automatic and Noninvasive Glucose Monitor," by Michael J. Tierney and others (Annals of Medicine, Vol. 32, 2000, pp. 632-641).

David C. Klonoff's "Noninvasive Blood Glucose Monitoring" (Diabetes Care, Vol. 20, no. 3, March 1997, pp. 433-437) surveys advances in that field.

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions