Clinical Evaluation of Near-Infrared Spectroscopy for Testing Cerebrovascular Reactivity in Patients With Carotid Artery Disease
Background and Purpose Near-infrared spectroscopy (NIRS) derives information about the concentrations of oxyhemoglobin (HbO2) and deoxyhemoglobin (Hb) from measurements of light attenuation caused by these chromophores. The aim of this study was to assess NIRS as a tool for testing CO2 reactivity in patients with carotid artery disease.
Methods One hundred patients with symptomatic carotid occlusive disease were examined (age range, 44 to 83 years). The severity of stenosis ranged from 30% to 100% (median, 80%) on the ipsilateral side and 0% to 100% (median, 30%) on the contralateral side. Monitored parameters included transcranial Doppler flow velocity, changes in concentration of HbO2 and Hb, cutaneous laser-Doppler blood flow, end-tidal CO2, arterial blood pressure, and arterial oxygen saturation. Hypercapnia was induced with the use of a 5% CO2/air mixture for inhalation. To estimate the contribution of skin flow to NIRS during reactivity testing, the superficial temporal artery was compressed, and the NIRS changes in response to the fall in laser-Doppler blood flow were recorded. Finally, reproducibility of reactivity testing was assessed in 10 patients who were subjected to repeated examinations over 3 days.
Results Flow velocity– and HbO2-derived reactivity values were related to the severity of the stenosis (P=.0001 and P=.017, respectively). The correlation between the two reactivity modalities was significant (r=.49, P<.000001). The median estimated contribution of skin flow to NIRS changes was 15.8%. Another variable affecting HbO2 signal changes during the CO2 challenge was arterial blood pressure (P=.025). Reproducibility of HbO2 reactivity was similar to flow velocity reactivity (14.3% and 18.6% variation, respectively).
Conclusions NIRS shows potential as an alternative technique for testing CO2 reactivity in patients with carotid disease provided that conditions are carefully controlled. Marked changes in arterial blood pressure may render the NIRS reactivity indices unreliable, and the contribution from extracranial tissue must be taken into account when significant.
Near-infrared spectroscopy has explored several clinical applications. Since the technique can measure changes in concentration of HbO2 and Hb with high temporal resolution,1 it is suitable for trend measurements of changes in cerebral oxygenation and cerebral blood volume.2 3 4 In addition, by manipulating blood gas tensions it is possible to quantify absolute concentration of tHb5 and calculate cerebral blood flow.6 A recent application for NIRS is in the assessment of cerebrovascular reactivity.7 8 9
A persisting concern with NIRS is the extent to which light is attenuated by the extracranial tissues. Although initial impressions indicated that the contribution of skin is negligible,1 recent investigations indicated that this is not the case.10 11 12 13 Attempts to reduce cutaneous factors include methods that subtract the signals recorded from two receiving optodes placed at different distances from the transmitting optode (spatial resolution).14 The extracranial contributions from each optode theoretically cancel out. However, these instruments have been disappointing.11 15 A second approach is to increase the intracranial contribution by adopting a large interoptode separation (≥5 cm)10 and to record other modalities that monitor changes in skin blood flow, allowing an estimation of extracranial influences.9 16
Our earlier experience suggests that extracranial contamination does not appear to be significant during a CO2 challenge in normal adult volunteers, since the recorded changes in cutaneous blood flow are small.9 The NIRS probes used in this study were supported with a pressure bandage for light exclusion, and therefore cutaneous blood flow may have been reduced by relative pressure ischemia. Such probe application is uncomfortable for the patient and may cause skin necrosis when the probes are kept for a few hours in the same position (such as during long-term monitoring of comatose patients).2 More recent commercial equipment provides smaller probes covered in a light-shielding rubber mold. These probes are applied without pressure, thereby revealing any cutaneous influence. In the presence of carotid artery disease, in which extracranial to intracranial collaterals develop, the cutaneous component of the attenuation of NIRS signals may become important. Thus, the technique may not be as reliable for CO2 reactivity testing in these patients.
The aim of this project was to validate NIRS as an alternative tool for testing CO2 reactivity in patients with carotid artery disease. The following questions have been explored: (1) Are NIRS-derived reactivity indices related to severity of stenosis? (2) Do NIRS and TCD reactivity indices show a correlation in patients with carotid artery disease? (3) What is the influence of extracranial circulation on NIRS results, and can this be corrected? (4) What is the day-to-day reproducibility of the CO2 reactivity indices using NIRS and TCD?
Subjects and Methods
One hundred patients with symptomatic unilateral carotid artery occlusive disease were examined (73 men, 27 women; age range, 44 to 83 years; mean age, 67 years). The severity of stenosis ranged from 30% to complete occlusion (median, 80%) on the ipsilateral side and 0% to total occlusion (median, 30%) on the contralateral side, as demonstrated by carotid angiography.
The monitoring configuration was similar to that used on volunteers.9 Briefly, in a quiet environment, patients lay in the supine position and were asked to breathe through a face mask. The mask was connected by a system of valves and tubes to an elastic bag filled with a mixture of 5% CO2 in air. A TCD probe (PC DOP 842; 2-MHz probe) was applied to the temporal window for insonation of the middle cerebral artery and fixed in position with an elastic band. Near-infrared probes (NIRO 500, Hamamatsu Photonics) were positioned on the side of TCD, with the receiving probe 2 cm from the midline and 2 cm above the supraorbital ridge. The transmitting optode was placed 6 cm away toward the hairline. The probes were secured with a lightproof holder and a bandage. With the exception of the first 41 patients, a cutaneous laser-Doppler probe (LDF, MBF3D, Moor Instruments) was placed in the immediate vicinity of the NIRS probes. Light emitted by the laser diode was adjusted to 635 nm to avoid interference with the near-infrared spectrometer. ABP was measured continuously from the index finger with a noninvasive pressure monitor (Finapress, Ohmeda 2300). EtCO2 and arterial oxygen saturation were also monitored (Multinex 4200, Datascope).
Signals of ABP, EtCO2, FV, LDF, HbO2, and Hb were sampled (50 Hz) and captured with specific software (ICM17 and CO2 reactivity test analyzer, P.S.). This software calculated the time trends of the signals and enabled automatic calculation of reactivity indices. The signals were calibrated in appropriate units (FV, centimeters per second; Hb and HbO2, microlar; ABP, millimeters of mercury; and EtCO2, kilopascals) with the exception of LDF, which was expressed in arbitrary units. For calibration of the near-infrared spectrometer readings, a path length factor of 5.93 was adopted.18 Changes in tHb were calculated as a sum of changes in HbO2 and Hb signals. The difference between HbO2 and Hb (Hbdiff) was also calculated to improve the signal-to-noise ratio.
The protocol consisted of a 5-minute period of baseline recording, followed by 5 minutes during which the patient breathed increased CO2. Hypercapnia was terminated when at least 1 minute of a stable recording in both TCD and NIRS had been acquired. In the last 10 patients entered into the study, the CO2 challenge was repeated after a 5-minute rest period for assessment of the reproducibility of the test. The patients from this small group were reexamined within the next 3 days. The same protocol was performed for both sides. To assess the influence of skin flow changes on NIRS readings, two superficial temporal artery compressions were performed at the end of the reactivity test, and the variation in NIRS parameters during the fall in cutaneous blood flow was captured.
Cerebrovascular reactivity indices were defined as follows:\mathit|<|FV_|<|reactivity|>||>||<|=|>|\frac|<||<|\Delta|>|\mathit|<|FV_|<|CO_|<|2|>||>||>||>||<||<|\Delta|>|\mathit|<|CO_|<|2|>||>||>||<|\cdot|>|\frac|<|1|>||<|\mathit|<|FV_|<|baseline|>||>||>|\mathit|<|HbO_|<|2|<|\,|>|reactivity|>||>||<|=|>|\frac|<||<|\Delta|>|\mathit|<|HbO_|<|2(CO_|<|2|>|)|>||>||>||<||<|\Delta|>|\mathit|<|CO_|<|2|>||>||>|\mathit|<|tHb_|<|reactivity|>||>||<|=|>|\frac|<||<|\Delta|>|\mathit|<|tHb_|<|(CO_|<|2|>|)|>||>||>||<||<|\Delta|>|\mathit|<|CO_|<|2|>||>||>|\mathit|<|Hb|>|diffreactivity|<|=|>|\frac|<||<|\Delta|>|\mathit|<|Hb|>|_|<|diff_|<|(CO_|<|2|>|)|>||>||>||<||<|\Delta|>|\mathit|<|CO_|<|2|>||>||>|where Δx(CO2) denotes increase in the x parameter value due to increase in CO2, and ΔCO2 denotes the increase in EtCO2 during hypercapnia.
Assessment of Severity of Stenosis
For measurement of severity of stenosis, the diameter of the most narrow segment of the internal carotid artery was taken against the diameter of the widest part of the distal internal carotid artery recorded on biplanar digital subtraction carotid angiography. The severity of stenosis was graded into categories: 1, mild (stenosis <50%); 2, moderate (stenosis of 50% to 70%); and 3, severe (stenosis >70%). The last category was additionally divided into two groups: those without and with significant (≥50%) stenosis on the contralateral side. This grading system gave a uniform distribution of cases into the specified categories and allowed inclusion of information about contralateral stenosis in patients with severe cerebrovascular pathology, in which case the influence of the contralateral side is expected to be significant. Very severe stenosis causing narrowing of the distal part of the internal carotid artery and underestimation of the severity ratio were placed into the severe groups.
Calculation of the Influence of Cutaneous Changes
The skin flow contribution was assessed by brief (10 seconds) compressions of the ipsilateral superficial temporal artery, which usually caused a small change in the NIRS signals. The results were used to estimate the cutaneous component of the NIRS changes during CO2 reactivity tests and used to calculate “corrected” NIRS reactivity values. Details of the method and calculations are given in the “Appendix.”
The data were tested for normality with the use of the Shapiro-Wilk W statistic. Most variables failed to show normality, and therefore nonparametric tests were used: Wilcoxon matched pairs test, Kruskal-Wallis ANOVA, Friedman repeated measures ANOVA, and Spearman rank correlation analysis. For assessment of reproducibility, the standard deviation was calculated.
Five patients were excluded from the study because of inability to define a suitable ultrasound window. No NIRS-related exclusions occurred.
NIRS Reactivity Indices Versus Severity of Carotid Stenosis
All NIRS reactivity values were significantly lower on the ipsilateral side than on the contralateral side (Table 1⇓), and there was significant association with the severity of stenosis (Fig 1⇓). Reactivities were most severely depleted when a severe ipsilateral stenosis was combined with significant contralateral disease (Table 2⇓). The maximum resolution was obtained with the Hbdiff parameter (P=.02). The difference between severe ipsilateral stenosis, with and without significant contralateral disease, was not statistically significant.
Reactivity With TCD Versus Reactivity With NIRS
FV reactivity indices showed a relationship with the side (Table 1⇑) and severity of stenosis (Table 2⇑) that was similar to that with NIRS, but the associations were statistically more significant. As with NIRS, the degree of stenosis on the contralateral side did not exert a statistically significant influence on the ipsilateral reactivity (Fig 1⇑). NIRS-derived reactivity indices demonstrated significant correlation with FV reactivity (Table 3⇓). The highest correlation was obtained with Hbdiff (r=.49, P<.000001, Fig 2⇓).
Estimated Extracranial Contribution to NIRS Signals
The sensitivity of HbO2 to changes in LDF was 1.85 μmol/L per 100% change in LDF (range, 0 to 4.9 μmol/L per 100%) on the ipsilateral side and 1.56 μmol/L per 100% change (range, 0 to 5.4 μmol/L per 100%) on the contralateral side. The difference between sides was not significant. Change in LDF during hypercapnia was on average 9.1% (range, −64% to 86%) on the ipsilateral side and 8.8% (range, −30% to 108%) on the contralateral side. The difference was again not significant. The estimated contribution of skin to HbO2 reactivity varied from −10% to 105% (median value, 15.8%) on the ipsilateral side and from −69% to 101% (median value, 6.4%) on the contralateral side. The difference was significant at a level of P=.04. Increases in LDF were predominantly associated with increases in ABP (84% of cases on the ipsilateral side and 89% of cases on the contralateral side). However, the direct correlation between relative ABP and LDF changes was not statistically significant.
NIRS Reactivity Indices Corrected for Cutaneous Changes
Results of analysis performed with corrected values of HbO2, tHb, and Hbdiff reactivity indices are given in Table 3⇑. The correlation coefficient between FV- and NIRS-derived indices improved when corrected values were used.
ABP and Arterial Oxygen Saturation
Mean ABP on average rose during the period of hypercapnia (mean±SD increase, 13±14.5%; range, –26% to 96%). Relative changes in ABP did not show any correlation with EtCO2 and FV reactivity values. However, they were significantly correlated with NIRS reactivity values (Table 4⇓). Oxygen saturation of the arterial blood showed little change during the hypercapnic challenge (mean±SD, 0.35±0.95%; range, −1.5% to 3.5%).
Reproducibility of TCD and NIRS
Analysis of reproducibility of reactivity parameters showed similar variability in FV and in NIRS parameters. SDs for each parameter are given in Table 5⇓. Analysis of influence of ABP showed significant correlation between variability in ABP and HbO2 (r=.7, P<.025, Fig 3⇓) and between changes in LDF response and changes in ABP (r=.84, P<.002). There was no significant difference between variability of these parameters measured between examinations and CO2 challenges.
We have shown previously in healthy volunteers9 that NIRS has potential as a tool for assessing cerebrovascular reactivity. However, how does the technique compare with established methods when used in patients with cerebrovascular pathology? Does the technique give reliable and reproducible results? What are advantages and limitations of the new technique? The aim of this project was to address these questions.
Significance of Correlation With Severity of Stenosis
The relation between the severity of internal carotid artery stenosis and TCD-derived reactivity indices has been described by many authors.19 20 21 22 Impaired reactivity indicates significant stenosis and an insufficient collateral blood supply.23 24 Therefore, despite the general trend of decreasing reactivity with increasing severity of stenosis, the spread of reactivity values in each severity group is expected to be high. Our data showed that FV and NIRS reactivity indices follow this pattern. All measured reactivity indices decreased with higher degrees of stenosis and were lowest when a severe ipsilateral stenosis was combined with significant contralateral disease. This observation supports the notion that there is a major intracranial hemodynamic component to the changes recorded with NIRS.
Source of Discrepancies Between TCD and NIRS Reactivity Indices
All NIRS parameters were significantly correlated with FV reactivity. However, the correlation indices for HbO2 and Hbdiff were approximately .5, and that for tHb was even lower. There may be several contributions to discrepancy. The most important factors include inaccuracy of estimation of the optical path length,18 measurement of absolute instead of relative changes in concentration of chromophores, and assessment of regional as opposed to global hemodynamic changes.26 These issues have been debated in studies with volunteers.9 Two factors, which are relevant to the type of patient addressed in this article, have not received adequate attention. The first is the anatomic variability of the territories supplied by the six major arteries of the circle of Willis.27 Thus, some measurements could have been taken from the territory supplied by the anterior cerebral artery instead of the middle cerebral artery, which becomes important when hemodynamic disturbances present in carotid disease are considered. The second factor is extracranial contamination.
NIRS Changes Caused by Extracranial Flow Variation
The extracranial contribution to NIRS changes has been estimated by employing cutaneous laser flowmetry.28 Although this technique provides local estimation of microcirculation in the skin (with skin penetration of 1 to 2 mm), we consider that this represents the major component of variation in extracranial circulation. Since the probe is situated on the skin in the immediate vicinity of the NIRS probes, skin flow changes that may affect NIRS measurements are recorded. The study showed that despite a relatively large distance between optodes, there is a significant influence of extracranial circulation on NIRS recordings. With NIRS reactivity values corrected for influence of skin, a better correlation with the severity of stenosis and with FV reactivity (Table 3⇑) was achieved. Thus, a simple method of compensating for the influence of skin may improve measurements of NIRS. This technique in itself introduces additional sources of error. Accurate assessment of relative flow with laser Doppler requires knowledge of the biological zero.29 Another source of error addresses the different sampling areas for NIRS and LDF, since heterogeneity of blood circulation in these respective areas is likely. Finally, laser-Doppler flowmetry assesses only flow in the superficial part of the skin.
Influence of ABP Changes
ABP proved an important covariate during the CO2 reactivity test. Any form of compensation for its effect (such as dividing relative increase in FV by relative increase in blood pressure) is inappropriate, since vasoconstriction caused by increasing the perfusion pressure is counteracted by vasodilatation induced by an increase in CO2. Thus, the final product of these different mechanisms cannot be readily separated. In the whole sample of patients studied, ABP predominantly increased and in one case rose by 193% of its baseline value. In such individuals, the increases in perfusion pressure will significantly affect measurement of both NIRS and FV reactivity (Fig 4⇓). Since the overall correlation between FV reactivity and ABP was not significant, we conclude that a significant correlation of NIRS to changes in ABP is probably mainly due to the effect of extracranial circulation responding passively to ABP. The importance of considering this component of NIRS reactivity measurement is therefore highlighted.
Despite the theoretical and practical considerations outlined above, the variability of NIRS reactivity was found to be comparable with the variability in FV reactivity (Table 5⇑). Although changes in ABP seem to have an effect on the reproducibility of both TCD and NIRS reactivity indices, the influence of ABP on NIRS is greater by virtue of the influence on the extracranial circulation. Even in a small group of 10 patients, the variation in HbO2 appeared to be significantly associated with variation in ABP (Fig 3⇑), reflecting a poor autoregulatory capacity for the skin circulation.
Is NIRS Likely to Replace TCD for CO2 Reactivity Testing?
One of the disadvantages of TCD is that it is not always possible to obtain an FV signal of sufficient quality. The published incidence of failure to obtain an acceptable “ultrasound window” is between 5% and 20%,30 31 which is dependent on the equipment used, the power and ability to detect the outline of a poor-quality signal, and the experience of the examiner. In our study, five patients (5%) were excluded because of this problem.
A comparison between TCD- and NIRS-derived reactivity indices indicates that NIRS identifies patients with severe impairment of CO2 reactivity and those with normal reactivity, with high specificity and sensitivity (Table 6⇓). In contrast, patients with intermediate NIRS reactivity show a significant variation in FV reactivity. This may indicate a limitation of the NIRS technique. Alternatively, territorial cerebrovascular heterogeneity may account for some of these discrepancies. Furthermore, since NIRS measures parenchymal (small-vessel) reactivity, these findings may reflect differences in reactivity values of the various components of the vascular tree, which tend to coincide only at the extreme of cerebrovascular reactivity.
The use of NIRS to assess cerebrovascular reactivity in adults with carotid artery disease shows promise and may be able to supplement techniques that are presently used in clinical practice. To achieve this, attention to the examination procedure is essential, with particular attention to identifying and controlling influential variables. In particular, changes in ABP and extracranial blood flow require further consideration. The method presented to correct for skin flow changes is simple and offers an improvement over current methods. Finally, advances in NIRS development may improve quantification and spatial resolution.
Estimation of Contribution of Skin Blood Flow
Assuming that laser-Doppler flowmetry assesses skin blood flow in the region sampled by the near-infrared spectrometer, skin blood flow contributions to NIRS parameters were estimated with data acquired during superficial temporal artery compressions (Fig 5⇓). Linear regressions of Hb and HbO2 signals versus LDF changes were performed, and the following parameters were defined:\mathit|<|Sensitivity to Skin Flow Changes|>||<|=|>|\mathit|<|rHbO_|<|2|>|LDF|<|\cdot|>|LDF_|<|baseline|>||>| |<|[|>||<|\mu|>|mol/L|<|]|>|, per 100% change in LDFFor skin flow contribution,|<|\Delta|>|\mathit|<|HbO_|<|2(SkinFlow)|>||>||<|=|>|\mathit|<|rHbO2LDF|<|\cdot|>||<|\Delta|>|LDF_|<|CO_|<|>2|>||>||>| |<|[|>||<|\mu|>|mol/L|<|]|>|and\mathit|<|Skin Influence|>||<|=|>|\frac|<||<|\Delta|>|\mathit|<|HbO_|<|2(SkinFlow)|>||>||>||<|(|<|\Delta|>|\mathit|<|HbO_|<|2(CO_|<|2|>|)|>|^|<||<|^\prime|>||>||>||<|+|>||<|\Delta|>|\mathit|<|HbO_|<|2(SkinFlow)|>|)|>||>||<|\cdot|>|100%where rHbO2LDF is a slope of the HbO2/LDF regression, ΔLDFCO2 denotes change in LDF during CO2 challenge, ΔHbO2(SkinFlow) is change in HbO2 signal due to the skin flow change, and HbO2′(CO2) is change in HbO2 during hypercapnia corrected for the skin flow contribution. Analogous calculations were performed for Hb, tHb, and Hbdiff parameters. Results of these calculation were then used to correct the NIRS reactivity indices for influence of skin with the use of the following formula (analogous formulas were used for Hb, tHb, and Hbdiff parameters):\mathit|<|Corrected Reactivity of HbO_|<|2|>||>||<|=|>|\mathit|<|Reactivity of HbO_|<|2|>||<|-|>|\frac|<||<|\Delta|>|HbO_|<|2(Skin)|>||>||<||<|\Delta|>|CO_|<|2|>||>||>|where ΔCO2 is an increase in CO2 during hypercapnia.
Selected Abbreviations and Acronyms
|ABP||=||arterial blood pressure|
|Hbdiff||=||difference between HbO2 and Hb|
|LDF||=||blood flow as assessed by laser Doppler|
|TCD||=||transcranial Doppler ultrasonography|
The authors would like to thank Professor David Delpy and his group from the Department of Medical Physics, University College of London, for exchange of stimulating ideas and expert support in the field of NIRS. They would also like to express their gratitude to the Raymond and Beverly Sackler Foundation and the ORS award for support of P. Smielewski. M. Czosnyka and P. Smielewski are currently on leave from the IPE, Warsaw University of Technology.
- Received June 6, 1996.
- Revision received October 14, 1996.
- Accepted October 14, 1996.
- Copyright © 1997 by American Heart Association
Wyatt JS, Cope M, Delpy DT, Richardson CE, Edwards AD, Wray S, Reynolds EOR. Quantification of cerebral blood volume in human infants by near-infrared spectroscopy. J Appl Physiol. 1990;68:1086-1091.
Smielewski P, Kirkpatrick PJ, Minhas P, Pickard JD, Czosnyka M. Can cerebrovascular reactivity be measured with near-infrared spectroscopy? Stroke. 1995;26:2285-2292.
Germon TJ, Evans P, Barnett N, Wall P, Nelson RJ. Optode separation determines sensitivity of near infrared spectroscopy to intra- and extracranial oxygenation changes. J Cereb Blood Flow Metab. 1996;15:617.
Germon TJ, Kane NM, Manara AR, Nelson RJ. Near-infrared spectroscopy in adults: effects of extracranial ischaemia and intracranial hypoxia on estimation of cerebral oxygenation. Br J Anaesth. 1994;73:503-506.
Elwell CE, Cope M, Edwards AD, Wyatt JS, Delpy DT, Reynolds EO. Quantification of adult cerebral hemodynamics by near-infrared spectroscopy. J Appl Physiol. 1994;77:2753-2760.
Van der Zee P, Cope M, Arridge SR, Essenpreis M, Potter LA, Edwards AD, Wyatt JS, McCormick DC, Roth SC, Reynolds EOR, Delpy DT. Experimentally measured optical pathlengths for the adult head, calf and forearm and the head of the newborn infant as a function of interoptode spacing. Adv Exp Med Biol. 1992;316:143-153.
Markus HS, Harrison MJG. Estimation of cerebrovascular reactivity using transcranial Doppler, including the use of breath-holding as the vasodilatatory stimulus. Stroke. 1992;23:668-673.
Kleiser B, Widder B. Course of carotid artery occlusions with impaired cerebrovascular reactivity. Stroke. 1992;23:171-174.
Obrig H, Kleinschmidt A, Merboldt KD, Dinargl U, Frahm J, Villringer A. Monitoring of cerebral blood oxygenation during human brain activation by simultaneous high-resolution MRI and near-infrared spectroscopy. Proc Soc Magn Reson. 1994;1:67. Abstract.
Van der Zwan A, Hillen B. Review of the variability of the territories of the major cerebral arteries. Stroke. 1991;22:1078-1084.
Richards HK, Czosnyka M, Kirkpatrick PJ, Pickard JD. Estimation of laser-Doppler flux biological zero using basilar artery flow velocity in the rabbit. Am J Physiol.. 1995;268:H213-H217.
Gomez CR, Gomez SM, Hall IS. The elusive transtemporal window: a technical and demographic study. J Cardiovasc Technol. 1989;8:171-172.
Halsey JH. Effect of emitted power on wave form intensity in transcranial Doppler. Stroke. 1990;21:1573-1578.