Abstract
Keywords
Introduction
The near infrared (NIR) part of the electromagnetic spectrum corresponds to light wavelengths starting from approximately 700 nm up to 1400 nm. The upper wavelength limit can also be found as 3000 or 5000 nm depending on the categorization scheme and it should be noted that for wavelengths higher than 3000 nm the physical principle of absorption changes and the spectrum is then called “thermal infrared”.
Water has a very low light absorption in the NIR wavelength area between 700 and 1000 nm [3, 11] and the same is true for the human tissue which consists mainly of water (≈ 70%). Consequently, in this area of the NIR spectrum there is a “window” where the light can travel some centimeters deep inside the human tissue (deep tissue spectrophotometry).
The biological properties, extracted from absorption NIR spectra, refer to the whole tissue including fat, muscle, bone, connective and nervous tissue, large blood vessels and microcirculation. Major influence to these spectra have specific chromophores such as oxygenated and deoxygenated hemoglobin (HbO2 and dHb respectively) in flowing blood. Therefore, NIR Spectrophotometry (NIRS) is not tissue or vessel specific as other techniques such as Doppler ultrasound applied to large vessels or video microscopy applied to microvessels [17]. Gayda et al. [7] have recently shown that microvascular function, assessed with NIRS, was impaired in patients with metabolic syndrome and coronary heart disease. They also demonstrated that microvascular function, assessed with NIRS, was more strongly related to cardiovascular risk factors compared to macrovascular function [7].
Intermittent claudication (IC) is the most common symptom of peripheral arterial disease (PAD) [10] and a classical, although subjective, clinical classification scheme of IC was proposed by Fontaine [6]. PAD is associated to an increased risk of ulceration in the affected lower limbs [10] and recent findings suggest that PAD impairs downstream cutaneous microvascular vasodilation which is probably explained, at least to some extent, by altered nitric oxide (NO) signal [10]. In addition, peripheral arterial disease and poorly compressible arteries are associated to altered hemorheological factors [4, 27]. Stage 2 PAD patients have “lower than average” blood fluidity and in particular, lower blood cell filterability, higher blood viscosity and higher red cell aggregation [4].
Using 6 NIR laser light wavelengths between 770 and 910 nm and making assumptions in the in vivo application of the Lambert-Beer law to the human calf muscle, Cheatle et al. [1] found significantly (p < 0.03) reduced muscle oxygen consumption (VO2), at rest, in peripheral arterial disease (PAD) patients in comparison to physiological controls. However, they did not classify the PAD patients and there was a broad overlap between the groups [29].
After the above result, many researchers have concentrated their efforts on improving PAD diagnosis using treadmill exercise tests [14, 23]. On surplus, Kooijman et al. [16] found no significant difference in VO2 at rest, between controls and patients clinically classified at Fontaine stage 2. This result pushed again the research efforts on exercise tests [13, 21] and Komiyama et al. [15] classified claudicate patients into 3 distinct types using NIRS patterns during and after treadmill exercise. Manfredini et al. [21] proposed a dynamic (exercise) NIRS-based test, for quantifying muscle metabolic response, using metabolic parameters expressed as area-under-curve. They found statistically significant differences between normal and PAD patients and different metabolic parameter patterns in relation to the severity of the disease.
With the advent of spatially resolved spectroscopy (SRS) the quantification of the tissue oxygenation index (TOI) with commercial instruments became possible [26] with TOI defined as the ratio (in percent) of the oxygenated hemoglobin concentration ([HbO2]) over the sum of the oxygenated and deoxygenated hemoglobin concentration ([HbO2] + [dHb]). However, the reported normal TOI level ranged between 47% [13] and 65% [28] and no difference in tissue oxygenation was found, at rest, between claudicates and normals [13, 28].
So, after more than 30 years from the introduction of the NIRS to noninvasive tissue monitoring, NIRS measurements have to be combined with a treadmill exercise test in order to assist in the diagnosis of PAD. The ultimate goal of discriminating PAD patients objectively without any exercise has not yet been achieved.
The second derivative of the absorption NIR spectrum has the advantage of removing both baseline offset and linear slope [2, 24] and was used in a special technique for the TOI quantification in the neonatal brain [2].
In this work, a second derivative NIR spectrophotometric technique was used in order to see firstly, if it can discriminate peripheral arterial disease (PAD) patients from control subjects before exercise, secondly, if the discrimination efficiency improves during or after a standard treadmill exercise test and thirdly, if two closely related groups of PAD patients (Fontaine 2a and 2b) can be distinguished, before, during or after the exercise test.
Methods
The experiments of this work were carried out at the Vascular Laboratory of Ninewells Hospital [18], but data manipulation and conclusions are new.
Experimental set up
The experimental set up comprised a light source with a power supply, two lightguides, a spectrophotometer and a computer (Fig. 1). The light source was a 400 Watts, 36 V, Quartz Halogen lamp (Osram Xenophot HLX 64663 EVD) supplied by constant voltage. A lens, with a focal length of 10 cm and a sphere mirror focused the light onto the tip of the emitting lightguide.
The emitting lightguide consisted of 30 quartz fibres (200 m core diameter, 250 m total diameter) from which only 18 were used to transmit light, randomly arranged within the bundle. The receiving lightguide consisted of 20 fibres and the total light receiving angle was 23°.
The macro-lightguide spectrophotometer was a photal Multi Channel Photo Detector (MCPD-1000-311C, Otsuka electronics Ltd). Light from the receiving lightguide was directed to the fixed flat focus grating and then, an image was formed flatly on the one dimensional photo diode array detector composed of 1024 elements paired for each wavelength.
After the detector, the video signal was amplified, digitised and transferred to the data processing unit (Dell 210 personal computer) via the controller driving the detector main unit. The accompanying software allowed any selected part of the spectrum to be displayed on line and performed smoothing and second derivative estimation.
Human subjects
Three groups of human subjects were studied namely, A, B and C. Group A (controls) comprised 10 normal subjects (9 males and 1 female) with mean age of 26±3 years. Group B (Fontaine’s 2a) was formed by PAD patients classified as Fontaine’s stage 2a and comprised 5 patients (1 male and 4 females) with mean age of 68±5 years. Group C (Fontaine’s 2b) was formed by PAD patients classified as Fontaine’s stage 2b and comprised 10 patients (5 males and 5 females) with mean age of 69±7 years.
Stage 2 at Fontaine’s classification [6] can be subdivided into stage 2a and stage 2b. In stage 2a the walking distance of the patient is less than 1000 m whereas in stage 2b is less than 200 m. However, this subdivision of stage 2 is mainly subjective since it is based on the opinion of the patients and there is no valid clinical method of classification.
Data were acquired before the Brussels directive of 2001 [5] demanding for the European Union member states the submission of research protocols to ethics committees. In addition, it should be noted that the measurement procedure was non-invasive, without any drug testing and oral informed consent was obtained from all human subjects participating in this study.
Measurement procedure
The detector was switched on at least 40 min before measurements so as to reach a stable temperature and constant dark reference curve. The software settings were set at 15 seconds sampling and 15 points smoothing.
The first step of the procedure was recording of the dark and white reference curves in a dark room. White reference was recorded by placing the emitting lightguide in front of the receiving lightguide.
The second step was to place the lightguide holder on the human leg. The two lightguides were positioned on the sagittal line that splits the gastrocnemius muscle (calf) in two symmetrical parts and the down lightguide was approximately 3 cm above the proximal end of the calcaneal (Achilles) tendon. The distance between the lightguides was constant at 3 cm with the aid of a special holder constructed for this purpose.
The third step was taking measurements with the same protocol for all human subjects comprising 9.75 minutes (585s) of standing up (phase 1), 1 minute of treadmill exercise (phase 2) and 1 minute of rest (standing up) after exercise (phase 3). The treadmill settings for exercising were 4 km/h and 10% grading. The distance travelled with these treadmill settings was only 66.7 m which enabled all stage 2 claudicates to undertake the test.
After recording the absorption spectrums for each human subject, normalisation was performed at 800 nm which is an in vitro isosbestic point for haemoglobin [9].
Finally, the second derivative value of each recorded spectrum was estimated at 763 nm (SDV) which is an in vitro absorption peak for the fully deoxygenated haemoglobin [9]. For easier data manipulation and presentation, here the SDV was multiplied by 104.
It has been shown in vitro that SDV is linearly proportional to haemoglobin oxygen desaturation [9]. However, as it often happens, the in vivo reality is much more complicated with many different chromophores and substances affecting the absorption spectrum. For example, myoglobin has a similar spectrum to haemoglobin, especially in the NIR part of the spectrum [3] and there are reports assigning to myoglobin a greater contribution than hemoglobin to the overall NIR signal [8]. In addition, chromophores are not homogeneously distributed in human tissues and propagating light is multiply scattered [22] in a non-isotropic way.
For the aforementioned reasons, the SDV was not correlated exclusively to haemoglobin oxygen desaturation but it was considered as an index arbitrarily related to the calf tissue deoxygenation status. In other words, it was assumed that SDV is an arbitrary tissue deoxygenation index without trying to mathematically relate it to a particular substance concentration or to a tissue oxygenationratio.
Statistical analysis
To study differences among the 3 human groups, seven continuous variables were used for statistical analysis. Six variables (SDV15, SDV285, SDV585, SDV600, SDV645 and SDV705) were defined as the SDV at 15, 285, 585, 600, 645 and 705 seconds respectively, after the onset of the experiment. SDV15, SDV285 and SDV585 belonged to the standing phase of the experiment P1 (phase 1), SDV600 and SDV645 belonged to the treadmill exercise phase of the experiment P2 (phase 2) and SDV705 to the resting phase of the experiment P3 (phase 3). Finally, the seventh variable SDVD was defined as the SDV Drop after 1 minute of treadmill exercise: SDVD = SDV585 –SDV645.
The majority of variable distributions were not normal and therefore differences between human groups were examined using the Mann-Whitney U test. The level of significance was set at p < 0.05. The professional edition of Microsoft Office EXCEL 2016 and the version 1.4 of the SOFA (Paton-Simpson & Associates Ltd) software were used for statistical analysis.
Results
The mean SDV is shown in Fig. 2 in black dots, rectangles and triangles for groups A, B and C respectively, throughout the measurement duration of 705 seconds. The duration of phases 1, 2 and 3 is shown with double arrows P1, P2 and P3, respectively.
For group A, the mean SDV was approximately constant for the initial standing up phase P1 (first 585 s) but then a decline was observed during the 1 minute of exercise (P2) and a recovery followed during the 1 minute of rest (P3). At the end of the exercise period (645 s) the mean SDV drop was about 15% and at the end of the recovery phase (705 s) the mean SDV almost reached the value before the onset of the treadmill exercise. For groups B and C the mean SDV followed a similar time course to group A but with some important differences that are described below.
First, the mean SDV values for groups B and C were lower than that of Group A, throughout the experiment, as shown graphically in Fig. 2. Second, the mean SDV drop after 1 minute of treadmill exercise for groups B and C was more than 3 times higher than that of group A. Third, the mean SDV for groups B and C did not recover during the rest period of 1 minute after the exercise.
The SDV ranged between –23 and –512 for all human subjects. A full report of descriptive statistics for variables SDV15, SDV285, SDV585, SDV600, SDV645, SDV705 and SDVD is given in Tables 1–6 and 7 respectively. All variables were significantly higher (p < 0.05) in group A compared to those in groups B and C (Tables 1–7). It is noted that the level of significance was ten times higher (p < 0.005) at the onset (15 seconds) of the experiment (Table 1) and during the exercise and post exercise phases P2 and P3 (Tables 4–7).
On the other hand, SDV15, SDV285, SDV585, SDV600, SDV645, SDV705 and SDVD were not significantly different in group B compared to group C.
Discussion
The principal result of this work was that healthy subjects can be distinguished from PAD claudicates, in a standing position, in just 15 seconds, without the need of any exercise. Furthermore, the level of significance (p = 0.003, Table 1) was 10 times higher than previously reported (p = 0.03, [1]). These findings could speed up and facilitated the process of PAD diagnosis in the clinical field.
However, the technique did not distinguish group B from group C. This could be attributed to the fact that Fontaine’s 2a and 2b patients are close in terms of the peripheral arterial disease severity. In addition, as it was already mentioned in Section 2.2, the subdivision of Fontaine’s stage 2 is rather subjective since it is based on the opinion of the patients.
One of the limitations of this study was the age difference between group A and patient groups B and C, but this can only be overcome by repeating the experiment with age matched groups.
The standard deviation (SD) of SDV in group A was comparable to that in group B (Tables 1–7) except for Table 6 where it was about a fifth of that in group B. However, the SDs in group C were dramatically higher (on the average 5 times) than those in group A (Tables 1–7). The reasons for this outcome are unknown.
Seifalian et al. [25] reported a significant postexercise correlation between the increase in urinary albumin to creatinine ratio (ACR) and recovery time of oxyhemoglobin in claudicants and proposed that NIRS changes reflect endothelial damage. Given the complexity of endothelial cell responses, seeming to act like cardiovascular processing sensors [19] it would not be an exaggeration to assume that some endothelium derived substances affect the NIR spectrum in claudicants.
In conclusion, a second derivative NIRS technique distinguished Fontaine’s stage 2 PAD patients from healthy subjects, within 15 seconds, without any exercise. Further experiments are required in order to uncover the full potential of this method.