David C. Klonoff,
Department of Medicine
University of California at San Francisco San Francisco, California
James Braig, Bernhard Sterling Ph.D., Charles Kramer Ph.D., Daniel Goldberger and Rick Trebino Ph.D.
OptiScan Biomedical Corporation
Optimal management of diabetes requires stabbing the finger with a needle up to four times daily to obtain a drop of blood for glucose measurement. Such invasive testing could be replaced by noninvasive, nearly noninvasive, or minimally invasive technologies. These new technologies will reduce or eliminate the skin trauma, pain, and blood waste associated with current glucose monitoring technology
Noninvasive blood glucose monitoring involves either radiation or fluid extraction1. The most promising radiation technologies are: 1) mid-infrared radiation (Mid-IR) spectroscopy; 2) near-infrared radiation (Near-IR) spectroscopy; 3) radio wave impedance; and 4) optical rotation of polarized light. All of these technologies except Mid-IR spectroscopy involve application of an external energy source to the body. The interaction between applied radiation (or thermal emissions in the case of Mid-IR spectroscopy) and glucose in the blood is measured and converted into a blood glucose concentration.
Fluid extraction from skin, also known as reverse iontophoresis, utilizes an electrical current applied to the skin. The current pulls out salt, which carries water, which in turn carries glucose. The glucose concentration of this extracted fluid is measured and is proportionate to that of blood.
Nearly noninvasive blood glucose monitoring involves transcutaneous harvesting and measurement of interstitial fluid which is the fluid surrounding every cell of the body. The most promising technologies for disrupting the skin barrier to obtain interstitial fluid include: 1) dissolution with chemicals; 2) microporation with a laser; 3) penetration with a thin needle; and 4) suction with a pump.
Minimally invasive blood glucose monitoring involves insertion of an indwelling glucose monitor under the skin to measure the interstitial fluid glucose concentration. Such a monitor typically utilizes a sensor containing the enzyme glucose oxidase. This enzyme oxidizes glucose and employs oxygen as an electron acceptor. The reaction liberates H2O2 which is detected by measuring its oxidation current on platinum electrodes2.
Interstitial fluid glucose concentrations were recently shown to be equal to simultaneously measured fixed or fluctuating blood glucose concentrations3. Such studies helped validate radiation technologies for noninvasive blood glucose monitoring. These technologies might measure glucose in blood as well as interstitial fluid. If these two compartments rapidly equilibrate, then they can be measured with the same method.
Mid-IR spectroscopy is an attractive technology for blood glucose monitoring because it is noninvasive, specific, rapid, and potentially inexpensive. The remainder of this article discusses that methodology.
In the mid-IR, glucose absorbs strongly with minimal interference from other absorbing species, but sources are not ideal and water absorbs very strongly. The OptiScan noninvasive glucose meter takes advantage of the realization that the human body is an excellent blackbody emitter of mid-IR light at precisely the right spectral region, so the necessary radiation is therefore already present inside the body near enough to the surface (within 10s of microns) that its emission spectrum may be used to measure glucose.
Unfortunately, the nature of blackbody radiation is such that, if the glucose to be measured is at the same temperature as the tissue that emitted the blackbody radiation, then the glucose will be effectively transparent, and no glucose spectrum will be observed in the emitted light due to the balance of absorption, spontaneous emission, and stimulated emission at equilibrium. Fortunately, this problem can be solved by introducing a temperature gradiente.g., cooling the surfacewhere the cooler glucose will absorb more than it will emit. This yields a reasonable glucose absorption spectrum superimposed on the usual smooth blackbody emission spectrum.
An approximate analytical mathematical model yields the predicted emission spectrum and signal-to-background ratio (S/B) that will be observed by OptiScans glucose monitor. It begins with the simple equation of radiation transfer, which simply says that the change in intensity is that due to spontaneous emission plus that due to stimulated emission minus that due to absorption4.
The model assumes that the tissue surface is then cooled to DT below the interior tissue temperature, T. Call tg the optical depth due to glucose over the temperature-gradient region and t the total optical depth due to everything else over this region (mainly water). The signal-to-background ratio, that is, the ratio of glucose signal to other light that will emerge from the chilled tissue can be shown to be:
where w is the frequency of observation, " is Plancks constant over 2p, and k is Boltzmans constant.
The relevant numbers
in the above equation for glucose in water are:
"w/k T~3, DT/T=0.05 (corresponding to DT~ 15° C and T ~ 300° K), tg/t ~ 0.01 (corresponding to a water concentration of ~ 1 g/cm3, a glucose concentration of ~ 100 mg/dl ~ .001 g/cm3, and, at a wavelength of 9.7 mm, absorption cross-sections of ~ 2 ´ 10-18 cm2 and 2 ´ 10-20 cm2 for glucose and water, respectively), and t ~ 10 (corresponding to a water absorption 1/e depth of 15 mm and a gradient region ~ 150 mm long).
As a result, S/B ~ 10-4, which is not unreasonable in view of the capabilities of available spectrometers and detectors. Also, tissue is, of course, not entirely water, and skin has even less water. These factors actually act to improve the S/B over that calculated because absorption in human subjects is actually less. On the other hand, humans have complex variations that reduce instrument reliability.
Figure 1 shows a composite graph of clinical results taken from four volunteer Type 1 diabetic subjects measured on two functionally identical non-invasive prototype instruments. The data was collected during modified glucose tolerance tests. Two tests were conducted in October, 1997 at the companys clinical laboratory and two other tests were conducted in January, 1998 in an ongoing clinical study at UCSF. The overall standard error is 24.7 mg/dL and the R2 is 0.94. The data was post-processed using an individual linear calibration curve for each subject. The reference instrument was a YSI laboratory glucose analyzer for the two UCSF subjects and an Accu-Chek home blood glucose monitor for the two company subjects.
In conclusion, we believe our data supports the hypothesis that mid-infrared radiation spectroscopy is a viable method for non-invasively measuring human blood glucose concentrations.
1 Klonoff, DC, Noninvasive blood glucose monitoring. Diabetes Care, 1997:20:433-437.
2 Zhang, Z., Liu, H., and Deng, J., A glucose biosensor based on immobilization of glucose oxidase in electropolymerized o-aminophenol film on platinized glassy carbon electrode, Analytical Chemistry 1996; 68:1632-1638.
3 Bantle, JP and Thomas, W., Glucose measurement in patients with diabetes mellitus with dermal interstitial fluid. Journal of Laboratory and Clinical Medicine 1997; 130:436-441.
4 Gaussorgues, G. Infrared Thermography (Chapman and Hall, London, 1994) p. 53
Back to In This Issue . . .