IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 38, NO. 8. AUGUST 1991
New Noninvasive Transcutaneous Approach to Blood Glucose Monitoring: Successful Glucose Monitoring on Human 75 g OGTT with Novel Sampling Chamber Shinsuke Kayashima, Tsunenori Arai, Member, IEEE, Makoto Kikuchi, Fellow, IEEE, Noriharu Sato, Naokazu Nagata, Osamu Takatani, Narushi Ito, Jun Kimura, Toshihide Kuriyama, and Akio Kaneyoshi Abstract-A novel noninvasive and quasi-continuous method of transcutaneous blood glucose monitoring for use with the human 75 g oral glucose tolerance test (OGTT) has been developed. The effused fluid was obtained by applying suction on the skin surface and labeled suction effusion fluid (SEF). The system consists of two main parts: a suction apparatus and the glucose sensor systep. The suction apparatus applies vacuum to the patient's skin at 400 mmHg absolute pressure to collect the SEF. The miniature ion sensitive field effect transistor (ISFET) based glucose sensor can measure glucose in small SEF quantities. The monitoring system is based on the association between the glucose concentration in the SEF and in the serum. During the 75 g OGTT, the glucose change in the SEF was measured every 10 min. Although a response delay of up to 20 min was observed in the SEF glucose change, it was possible to perform the 75 g OGTT by this noninvasive monitoring method.
INTRODUCTION N DIABETIC CARE, both noninvasive and long-term continuous blood glucose mointoring is desirable. However, this kind of monitoring procedure has not been accomplished clinically. For other blood constituents, noninvasive monitoring or continuous monitoring without blood sampling would also be an ideal goal. Extensive efforts have been concentrated on this subject. Several procedures are applicable for restricted materials, including measuring oxygen and carbon dioxide tension in blood , transcutaneous bilirubinometry [2 11, and hemoglobin oxygen saturation [ 171. Currently, these procedures play important roles in basic and clinical medicine. If both noninvasive and continuous monitoring without blood sampling could be achieved, the procedure would greatly contribute to many clinical procedures and patient care. For blood glucose monitoring, the first successful procedure for noninvasive and quasi-continuous monitoring
Manuscript received January 12. 1990; revised November 26, 1990. S . Kayashima, N. Sato, N. Nagata. and 0 . Takatani are with the Third Department of Intemal Medicine. National Defense Medical College. Saitama 359, Japan. T. Arai and M. Kikuchi are with the Department of Medical Engineering. National Defense Medical College, Saitama 3.59. Japan. N. Ito, J. Kimura. and T. Kuriyama. and A . Kaneyoshi are with Central Research Laboratories, NEC Corporation, Kanagawa, Japan. IEEE Log Number 9101 187.
during the 75 g OGTT on a human subject is reported. This procedure has the potential to contribute to treating diabetics, regarding both monitoring and control of the blood glucose concentration. Although some reports have been published using slightly invasive methods with needle glucose sensors, which were inserted into subcutaneous tissues [ 151, [ 161, noninvasive blood glucose monitoring procedures without bleeding have not been reported. We have employed suction effusion fluid (SEF) as the sample. It is a small volume sample and is difficult to handle. We originally developed suction apparatus and applied an ion sensitive field effect transistor (ISFET) glucose sensor which consists of small sensor chip that can be realized in a small size while giving accurate glucose concentration measurements , . In this paper we operated the suction apparatus by hand and measured glucose concentration by the ISFET glucose sensor which can be integrated into the suction apparatus. MATERIALS A N D METHODS A . Suction Apparatus The thin, 20 g polyvinyl chloride suction apparatus, illustrated in Fig. l , can be banded to a human upper arm so as not to impede movement of the arm [ l ] . The skin suction area is 7.1 cm2. The surface which directly contacts human skin is smooth to prevent irritating the skin. The comeous layer of the epidermis should be stripped before applying suction because it acts as a water barrier [ l l ] . The suction apparatus was easily set on a human upper arm by banding tape after stripping the corneous layer from the epidermis by adhesive tape. Suction was applied to the skin surface by this apparatus at 400 mmHg absolute pressure. Collected SEF in the suction apparatus reservoir was sampled by a 27 G syringe needle penetrating the silicone seal.
B. Glucose Concentration Comparisons Between SEF and Serum on Human Subjects in Static Condition The SEF was collected from five healthy males in 3 h using the suction apparatus. The subjects sat on chairs (it
0018-9294/91/0800-07.52$01.00 @ 1991 IEEE
TRANSCUTANEOUS APPROACH TO BLOOD GLUCOSE: I MONITORING -I
GClcose sensor system
steel mesh Reservor part
Fig. 1 . Suction apparatus. This shows a qide view of the suction apparatus. Its lower side directly contacts the human skin. After its reservoir part is decompressed to 400 mmHg absolute pressure by the decompressing pump, SEF emerges into it. Every I O min a syringe with a 27G nccdle penetrates the silicone sealing portion vertically (the upper part of the figure) to collect the SEF. The suction apparatus is 40 mm in diameter. 12.8 mm thick, and 7. I cm' in skin sucking area. Its weight is 20 g. It is made of vinyl chloride in the body. silicone in the necdlc penetrating part of the cap, and stainless steel mesh to maintain a sucking space between the suction apparatus and the skin and to make sure the suction is applied to the skin.
was not necessary for them to lie down) and were allowed no food intake or exercise. Just before and after collecting the SEF, blood was sampled and centrifuged to sera (1000 g, 10 min). The sera were stored at 4"C, and glucose concentration was measured within 24 h. The collected SEF volumes were measured every 10 min during the 3 h suction, in order to observe any changes in the SEF effusion rate. The glucose concentrations in the integrated SEF and the sera were measured by the glucose dehydroxylase method. The room temperature and humidity during the experiment was between 20 and 27"C, and between 50 and 75 % , respectively. C. Glucose Monitoring System and Transcutaneous Glucose Monitoring During 75 g OCTT The experimental system and the transcutaneous 75 g OGTT procedure is illustrated in Fig. 2. The experimental system consists of the suction apparatus and the glucose sensor system. The suction apparatus was banded onto the human upper arm as described above. The glucose sensor system was composed of a miniature ISFET glucose sensor, a pen recorder, an amplifier, a mechanical vibrator, and a sampling plate. A miniature ISFET was adopted as a glucose sensor. It was able to measure glucose concentration in volumes as small as 5 pL. It was developed for this experiment because the collected SEF volume was too small to use for measuring glucose concentration continually by conventional methods. Characteristics of this sensor have been reported in detail previously . The ISFET glucose sensor gives stable output and has a response time of less than 2 min (71, . The ISFET glucose sensor consists of a glucose oxidase field effect transistor (CHEMFET) and reference field effect transistor (REFET), both embedded on a 0.3 X 1.6 X 8 mm sapphire plate. Its output was stable within 1.5 % coefficient of variation measuring standard solutions 30 times. For this experiment, the sensor is thought to show reliable glucose output without sensor membrane degradation. Just
Fig. 2. Glucose monitoring system. This 5hows the glucose monitoring system. It consists of the suction apparatus (on the left) and the glucose sensor system (on the right). SEF sampling is done by hand using a syringe with a 27G needle (refer to Fig. I ) . The 5 pL SEF sample is dropped on the hydrophobic sampling plate and measured by the ISFET glucose sensor. The procedure involving dropping thc SEF sample and recording the data is implemented automatically within two minutes for one sample.
before and after this glucose monitoring, the output from the ISFET glucose sensor was measured, using standard glucose solutions (100, 200, 300, 400 mg/dL glucose concentrations) with 4 g / d L bovine serum albumin. * Also, the response curve for the ISFET output was recorded while measuring 100 mg/dL glucose concentration serum. The 5 pL SEF sample was dropped on the hydrophobic sampling plate in the glucose sensor system, and the SEF was diluted by buffer solution (20 mM pH 7.5 HEPES, ionic strength = 0.15) to one-fifth concentration. This was necessary due to the upper limit of the glucose concentration measurement range for the ISFET glucose sensor. The glucose concentration in the SEF was measured automatically after output from the sensor became stable. Then the ISFET glucose sensor was washed in preparation for the next measurement. The ISFET glucose sensor takes about 2 min to complete measurement of one sample. A 75 g OGTT was carried out on a healthy male with impaired glucose tolerance, using the glucose monitoring system described above. The SEF and blood were sampled every 10 min for about 3 h, before and after oral 75 g glucose loading. The glucose concentrations in the SEF were measured by this glucose monitoring system. The glucose concentrations in the sera were measured by a Beckman Glucose Analyzer 2.
D. Statistics The student t-test and linear regression were employed for data analysis. RESULTS Comparisons between the glucose concentations in the SEF and in the serum under static conditions are summarized in Table I. The glucose concentrations in the SEF are from 19 to 33% lower than those in the serum. The measured SEF effusion rates on five subjects were 6.6 to 17 pL/h/cm2, and the average effusion rate value was 11.9 3.7 pL/h/cm2. In other words, the average SEF volume obtained with the suction apparatus during 10 min *BSA, fraction 5 . lot no. STN 2231 Wako Pure Chemical Industries.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 38. NO. 8. AUGUST 1991
TABLE 1 GLUCOSE CONCENTRATION COMPARISON BETWEEN SEF
Glucose in serum Glucose in SEF
was 14.1 4.4 pL. The effusion rates were constant throughout the 3 h experiment. The effusion fluid volume over 10 min was too small for its glucose concentration to be measured by the conventional method. However, it was possible to measure it using the ISFET glucose sensor. Fig. 3 shows the comparisons between outputs from the ISFET glucose sensor, before and after the 75 g OGTT. The ISFET glucose sensor indicates nearly linear output up to 400 mg/dL in glucose concentration and almost no degradation through the 75 g OGTT. On the response curve for the ISFET output measuring serum for 100 mg/dL in glucose concentration, the sensor output reached a plateau within 45 s and remained stable. After washing the glucose sensor, the sensor output returned to the base line within 90 s. Fig. 4 shows measured glucose concentrations, both in the SEF and the serum during 75 g OGTT on one special human subject with imparied glucose tolerance. The vertical axis represents glucose concentration (mg /dL) in the SEF and the serum. The horizontal axis indicates time lapse, before and after oral 75 g glucose loading. As shown in Fig. 4, the glucose concentrations were stable in the SEF and the serum before the glucose loading. Glucose concentrations in the SEF were lower than those in the serum by 10-20% before glucose loading. The serum and SEF data before glucose loading were 123 and 100 mg/dL at -30 min, 128 and 104 mg/dL at - 10 min, and 113 and 101 mg/dL at glucose loading. The SEF/serum was 81.2 to 89.7% (84.2 4.7%). These values correlated with a coefficient of correlation 0.84, compared with those in Table I. Then the 75 g OGTT was thought to be performed under nearly the same conditions as materials and methods B: in static condition. Although there was from 10 to 20 min delay before the glucose concentration change in the SEF became apparent following that in the serum, the glucose concentration in the SEF changed in the same manner as in the serum. The peak glucose concentrations, both in the SEF and the serum, indicated nearly the same values, approximately 220 mg/dL. The subjects suffered mild skin damage, with slight redness and pain after the suction procedure was applied for about 3 h. They subjectively preferred sampling SEF to sampling blood several times. No skin biopsy was performed on humans. On rabbits, we biopsied the skin three times and compared them: before treatment, including tape stripping and suction, just after 3 h suction and on the seventh day after the treatment. These specimens were strained with hematoxylin and eosine. Slight edema but no hemorrhage was found in the just after 3 h suction
GLUCOSE CONCENTRATION (mgldl) B e f o r e experiment
Fig. 3 . ISFET glucose sensor calibration. @-the ISFET glucose sensor output before 75 g OGTT measuring standard glucose solution. 0-the ISFET glucose sensor output after 75 g OGTT measuring standard glucose solution.
t Ckal W o s e bad
Suctm ettusm tlud
Fig. 4. 75 g OGTT on a human subject.
specimen and no prominent fibrotic changes were seen on the seventh day specimen.
DISCUSSION The SEF is considered to mainly consist of interstitial fluid in the subcutaneous tissue and the intravascular fluid filtrated through the vascular wall by suciion. Therefore, the glbcose concentration change in the SEF represented some delay compared to that in the serum. There was a 10-20 min delay of the glucose concentkation change in the 9EF, which followed that in s e n " The causes for this delay are thought to be: 1) physiological delay caused by SEF requiring some time to move from the interstitial space or intravascular space in the subcutaneous tissue up to the epidermis surface, and 2) mechanical delay, which is due to the mechanical dead volume of the suction apparatus itself. The latter kind of delay can be shortened by decreasing the dead volume for the suction apparatus. Fig. 5 indicates the SEF collection principle using the suction procedure. The result indicates that the glucose concentration in the SEF correlated with that in the serum.
TRANSCUTANEOUS APPROACH TO BLOOD GLUCOSE MONITORING
Subcutaneous t ~ s s t ~
Vessels Fig. 5 . SEF generation. This shows a view of skin structure. The upper portion represents skin surface. From the top of the bottom, individual layers represent comeous layer: , epidermis excluding comeous layer: , cutis, and subcutaneous layer. The parallel lines and loops represent small blood vessels in cutis and subcutaneous tissue. The center of the comeous layer has already been stripped and sucked. SEF, shown as r'..i.-