1.

Introduction

The concentration of glucose in the capillaries of the skin is a very important parameter in the field of medical diagnostics. Glucose plays a major role within the complex nutritive supply system of the tissue. In healthy persons, the concentration of glucose is stable within a narrow range (dependent on age and fasting: 70 to $110\mathrm{mg}∕\mathrm{dl}$ ). In patients with diabetes, the regulation system is disordered. The disease causes the metabolism to malfunction, and without therapeutic support, the probability of survival is low. Diabetic therapy requires that blood glucose concentration be measured several times a day. More than 10% of the costs in a clinical chemistry laboratory result from tests used to determine blood glucose concentration, thereby being the most frequently measured parameter. The standardized laboratory measurement methods include several work steps, and are therefore time- and cost-intensive methods.

A minimally invasive, fast, and reliable method for the determination of blood glucose is still of high economic relevance, because such a method has not yet been found. The development of a method for the quantitative determination of glucose in blood by means of infrared spectroscopy has been the subject of recurrent research and publication. Such research has involved investigations not only on dried blood,^{1, 2, 3} but also on serum and whole blood⁴ utilizing both transmission spectroscopy and ATR spectroscopy.^{5, 6} Cylindrical $\mathrm{ZnSe}$ crystals have been employed for these experiments. However, this approach has various practical shortcomings, such as the adsorption of proteins on the surface of the crystal.

In other methods, blood samples have been applied drop-wise on polyethylene cards and dried before measurement to overcome the interference of water absorption.⁷ One early report about the state of the art⁸ and a recently published review⁹ have discussed the use of analytical instrumentation for glycaemic control. For an overview of various other laboratory assays, see Heise and Abel.¹⁰ Multivariate methods¹¹ and Raman spectroscopy have also been used to determine glucose concentration in vivo in blood.¹² A recently published article used a system based on liquid-core optical fiber Raman and absorption spectroscopy, followed by partial least square regression to determine chemical concentrations in blood serum or urine.¹³ Glucose in water has proved to be a challenge to analysts in the past, because water has strong absorption bands in the infrared wavelength range of interest. The new FTIR spectrometers allow the detection of glucose quantitatively within a range of 1300 to $1000{\mathrm{cm}}^{-1}$ . Whole blood shows bands of other substances within the range of interest such as proteins, urea, cholesterol, and triglycerides, in addition to glucose absorption. Moreover, there is an additional loss of light caused by erythrocyte scattering.

Vonach examined the direct determination of glucose in whole blood, without any sample preparation, using FTIR spectroscopy.¹⁴ The measurements were carried out in a flow cell. The whole spectral region from 1182 to $964{\mathrm{cm}}^{-1}$ had to be analyzed for determination of glucose concentration. A clear correlation could be obtained between the measured data and clinically determined blood glucose concentration.

In the investigations described in this work, FTIR measurements were used to investigate the influence of temperature on glucose in water, the influence of concentration on glucose in water and in whole blood, and the influence of the surrounding media (water versus blood) on glucose spectra. The results were compared to glucose determination carried out by the application of a two-wavelength ${\mathrm{CO}}_2$ laser.

2.

Materials and Methods

2.3.

Two-Wavelength ${\mathrm{CO}}_2$ Laser

A two-wavelength ${\mathrm{CO}}_2$ laser was developed and manufactured at the Institute of Applied Physics in Nishny Novgorod, Russia. The two laser wavelengths were selected at ${\lambda }_1=1080{\mathrm{cm}}^{-1}$ and ${\lambda }_2=950{\mathrm{cm}}^{-1}$ , because ${\lambda }_1$ is located within the glucose absorption band and ${\lambda }_2$ is used for background corrections.

The two laser beams with ${\lambda }_1$ and ${\lambda }_2$ were superposed in the optical cell containing the blood sample (Fig. 1 ). The transmitted light was detected separately for each wavelength behind the sample. Part of both laser beams was detected on a reference detector. The ratio of the laser power detected at ${\lambda }_1$ and ${\lambda }_2$ representing the glucose absorption was determined and transferred to data analysis.

Fig. 1

Scheme of the two-wavelength ${\mathrm{CO}}_2$ laser setup.

The optical cell consisted of two ${\mathrm{CaF}}_2$ plates with a spacer of $50\mathrm{\mu }\mathrm{m}$ . The cell was filled with the blood sample using capillary action. The blood volume required for the measurements was about $1\mathrm{\mu }\mathrm{l}$ .

3.

Results

3.1.

Fourier Transform Infrared Attenuated Total Reflection Measurements

The absorbance values of nine different concentrations of glucose in water in the region from 40 up to $400\mathrm{mg}∕\mathrm{dl}$ are presented in Fig. 2 . The typical glucose absorption bands are shown at 1080, 1036, 1059, and $980{\mathrm{cm}}^{-1}$ . A continuous increase in the absorbance could be detected with increasing glucose concentration. Sufficient signal could be detected even for a $40\text{‐}\mathrm{mg}∕\mathrm{dl}$ glucose solution. No dependence on glucose concentration could be observed at $950{\mathrm{cm}}^{-1}$ .

Fig. 2

Mid-infrared spectra for nine aqueous glucose concentrations between 40 and $400\mathrm{mg}∕\mathrm{dl}$ obtained by ATR-FTIR spectroscopy.

The relationship between the absorbance and the glucose concentration of the solvents is demonstrated for the three main bands at 1036, 1059, and $1080{\mathrm{cm}}^{-1}$ in Fig. 3 . All three absorption bands show a linear dependence and could be used for the determination of glucose concentration in water.

Fig. 3

Linear correlation of three glucose absorption bands obtained by ATR-FTIR spectroscopy with the glucose concentration.

The influence of temperature on glucose absorption was determined in the range from 15 to $28°\mathrm{C}$ for the glucose band. The results have shown that the glucose bands from 1160 to $980{\mathrm{cm}}^{-1}$ are not significantly influenced by temperature changes.

When analyzing a blood sample containing $92\text{‐}\mathrm{mg}∕\mathrm{dl}$ glucose, it was found that the glucose spectrum changes from water to blood. If glucose was added to a whole blood sample, its spectrum changed with time and resulted in a constant spectrum after $2\min$ . The spectra of glucose in pure water, and of glucose added to a whole blood sample, measured at different times are presented in Fig. 4 . The glucose band at $1034{\mathrm{cm}}^{-1}$ has its maximum intensity in water (curve a). In blood, this absorption decreases (curve b), and after $100\mathrm{s}$ , the highest intensity was obtained at $1080{\mathrm{cm}}^{-1}$ (curve c). Furthermore, the ATR signals changed when the red blood cells started to sediment. To investigate this in more detail, a blood sample was left to sediment and different parts of the sample were measured. Figure 5 shows a spectrum of the pure supernatant, the plasma, which gave the lowest absorbance (curve a). The sediment, consisting of the red blood cells (Rbc), resulted in a spectrum with the highest absorbance (curve d). A mixture of plasma with 5% red blood cells was measured directly after application on the crystal (curve b), and $2\min$ later (curve c). The measurement taken immediately after application showed a spectrum slightly higher than the plasma spectrum, whereas the one taken after $2\min$ came close to that of the high-concentrated red blood cells.

Fig. 4

ATR-FTIR absorbance spectra for $92\text{‐}\mathrm{mg}∕\mathrm{dl}$ glucose in water (curve a), $92\text{‐}\mathrm{mg}∕\mathrm{dl}$ glucose added to whole blood and measured immediately after application (curve b), and glucose added to whole blood and measured after $100\sec$ (curve c).

Fig. 5

ATR-FTIR absorbance spectra from one blood sample (glucose concentration $92\mathrm{mg}∕\mathrm{dl}$ ); plasma and red blood cells were separated. Spectra are shown from (a) pure plasma, (b) the same plasma with 5% red blood cells (Rbc) added and measured immediately after application onto the ATR crystal, (c) the same sample as (b) but measured $2\min$ after application onto the ATR crystal, and (d) red blood cell fraction.

3.2.

Two-Wavelength ${\mathrm{CO}}_2$ Laser Measurements

The applicability of a two-wavelength laser for the quick determination of glucose was tested in blood. First of all, measurements were performed on aqueous glucose solutions and analogical to the ATR measurements, the glucose concentration exhibited a linear dependency (not shown here). Then whole blood samples were investigated at different glucose concentrations. As a result of the FTIR measurements, one laser wavelength was tuned to $1080{\mathrm{cm}}^{-1}$ where the maximum absorption for glucose was found in whole blood. The other wavelength was tuned to the isosbestic point at $950{\mathrm{cm}}^{-1}$ where no dependence on glucose concentration could be observed. Glucose concentrations of whole blood samples were determined in the Dept. of Clinical Chemistry. The samples were then analyzed using the two-wavelength ${\mathrm{CO}}_2$ laser system. The correlation between the quotient of the signal intensity of ${\lambda }_1$ and ${\lambda }_2$ , and the glucose concentration determined in the Dept. of Clinical Chemistry, is presented in Fig. 6 . A correlation factor of $R^2=0.997$ was obtained.

Fig. 6

The quotient of the signal intensities at wavelengths 1080 and $950{\mathrm{cm}}^{-1}$ , obtained using the ${\mathrm{CO}}_2$ laser correlated with the glucose concentration in the range of 40 to $400\mathrm{mg}∕\mathrm{dl}$ .

4.

Discussion

In contrast to NIR measurements, the investigations in the MIR spectral range have shown that the intensity of glucose absorption bands around $1035{\mathrm{cm}}^{-1}$ does not change with temperature between 15 and $28°\mathrm{C}$ . This indicates that the absorption can be correlated with the glucose concentration and could be used for glucose determination. The linear dependency of the ATR signal could be shown for selected wavelengths. The intensities of the absorption bands changed when water was substituted by blood plasma. The changes were due to changes in the glucose structure induced by the different surrounding media.¹⁵

The maximum glucose absorption was found in whole blood samples at $1080{\mathrm{cm}}^{-1}$ . The background was determined close to the glucose absorption band at the isosbestic point at $950{\mathrm{cm}}^{-1}$ . These two wavelengths could be used for the spectroscopic determination of glucose concentration in whole blood samples. The influence of water absorption could be neglected, if this background is subtracted from the glucose absorption.

In contrast to the ${\mathrm{CO}}_2$ laser measurements, the ATR measurements show a strong dependency on red blood cell sedimentation. A part of the glucose is located in the red blood cells. This is known from the HbA1c value, which provides information about blood glucose level over the last 4 to 6 weeks. Hemoglobin is connected to glucose, a relationship that is dependent on the concentration of glucose in blood plasma. The glucose connected to hemoglobin could render the blood glucose determination faulty. To use this method for routine measurements, a very strict time schedule would be necessary, and a dependency on the hematocrit is anticipated. This strong time dependency was not observed when using the ${\mathrm{CO}}_2$ laser. Nevertheless, a time dependency due to red blood cell sedimentation could also be expected for this method, albeit to a lesser extent. The influence on this method due to red blood cells is mainly related to the scattering behavior of red blood cells. But the scattering of the cells is similar for both of the applied wavelengths and can be corrected using a ratio of the two. As a result, no marked dependency on the hematocrit is to be expected. The difference in these signals is only correlated to the glucose absorption and not influenced by changes in laser radiation intensity.

Measurement of low glucose concentrations using the ${\mathrm{CO}}_2$ laser is faster than using the ATR, as this requires several scans before a reliable signal-to-noise ratio can be attained. The measurements can be carried out on a ms time scale by using small blood samples of less than $10\mathrm{\mu }\mathrm{l}$ . In contrast to methods in the literature that require several preparation steps, such as separation of plasma and blood cells, diluting or drying the sample, this method provides a one-step procedure by filling whole blood into the cuvette or capillary. To develop a routine device based on the two-wavelength ${\mathrm{CO}}_2$ laser method, interindividual influences should be tested using blood from different persons.

5.

Conclusion

The investigations using ATR-FTIR spectrometry show that glucose in water is linearly dependent on absorbance. The absorbance maximum at $1036{\mathrm{cm}}^{-1}$ decreases when glucose is added to whole blood. All intensities change and the absorption band at $1080{\mathrm{cm}}^{-1}$ is the most prominent absorption band in whole blood. The signal intensity at $950{\mathrm{cm}}^{-1}$ is not dependent on glucose, and can be used for background correction such as water and red blood cells. Based on these results, a fast and easy method could be established without any prior preparation of whole blood by using a two-wavelength ${\mathrm{CO}}_2$ laser, and tuning the laser to the wavelengths 1080 and $950{\mathrm{cm}}^{-1}$ . The laser method is not affected by red blood cell sedimentation, and can be used to measure the glucose concentration in the range of interest 40 to $400\mathrm{mg}∕\mathrm{dl}$ on a ms timescale.