Is near-infrared spectroscopy living up to its promises?

Seminars in Fetal & Neonatal Medicine (2006) 11, 498e502

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / s i n y

Is near-infrared spectroscopy living up to its promises? Gorm Greisen* Department of Neonatology, 5024 Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen Ø, Denmark

KEYWORDS Neonate; Intensive care; Near-infrared spectroscopy; Brain; Oxygenation; Blood flow

Summary The first clinical application of near-infrared spectroscopy (NIRS) was made 20 years ago on the head of newborn infants under intensive care. Since then NIRS has yielded much credible and some important clinical research data. The most important results have been obtained using the cumbersome but quantitative techniques for measuring cerebral blood flow, cerebral blood volume, or venous oxygen saturation with manipulation of FiO2 or impeding venous outflow from the brain. The continuous nature of NIRS has been combined with monitoring of arterial pressure to obtain measures of cerebrovascular regulation, but this method has not been applied on a larger scale. Second-generation instruments allow a running estimate of vascular haemoglobin oxygen saturation, named the tissue oxygenation index (TOI), in absolute terms. Applied to the head, this is a surrogate measure of cerebro-venous saturation, an important variable in neuro-intensive care. The precision, however, is insufficient to be useful. In conclusion: clinical application is not in sight. ª 2006 Elsevier Ltd. All rights reserved.

Introduction

NIRS methodology

Transillumination of the head of small animals is possible using near-infrared spectroscopy (NIRS). The first clinical research use of NIRS in 1985 was in newborns,1 and quantitative spectroscopy was subsequently performed in 1986.2 In the following years many papers on NIRS in newborns were published. The purpose of this chapter is to provide a short overview and discuss the potential clinical use of NIRS for neonatal intensive care.

Geometry

* Tel.: þ45 3545 4320; fax: þ45 3545 5025. E-mail address: [email protected]

The newborn infant’s head is ideally suited for NIRS. The overlying tissues are relatively thin, which ensures that the signal is dominated by brain tissue e white as well as grey matter. NIRS recordings can be performed with the light being applied to one side of the head and received on the other side (transmission mode) in low-birth-weight infants with biparietal diameters of 6e8 cm. In this situation a large part of the brain is ‘seen’ during the measurement, and the results may be interpreted as ‘global’. Larger babies can only be investigated with the emitting and receiving fibres in an angular arrangement (reflection mode), possibly with both optodes on the same side of the head. In this

1744-165X/$ - see front matter ª 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.siny.2006.07.010

Near-infrared spectroscopy

499

situation a smaller volume of tissue between the optodes is seen. This may be chosen on purpose, also for smaller babies, to obtain ‘regional’ results. With a shorter interoptode distance, a narrower and shallower tissue volume is seen with a relatively larger fraction of extracerebral tissues. Therefore, distances of less than 4 cm are not recommended.

suggest that the arterial-to-venous ratio is about 1:2.5,6 Secondly, the signal can be seriously confounded by concomitant changes in tHb, depending on which of the arterial, capillary, or venous compartments are changed. Finally, the lack of a fixed zero-point makes it impossible to specify a threshold for intervention or even an alarm level for clinical use.

Algorithms

Cytochrome aa3 oxidase

Several different types of NIRS instruments have been used. The number of wavelengths used has varied from two to six. The specific wavelengths used, and therefore the mathematical algorithms used to separate the signals of oxyhaemoglobin (O2Hb), deoxyhaemoglobin (HHb), and the cytochrome aa3 oxidase difference signal (Cytox), have differed.3 It is therefore not a trivial matter to ensure that differences in research results are not simply due to differences in NIRS methodology, particularly in the earlier papers.

Reduction of cytochrome may be a specific indicator of inadequate cellular oxygen availability. At present, however, we do not know with any precision the relationship between tissue pO2, cytochrome oxidation state, and neuronal function. Furthermore, the measurement of cytochrome oxidase with optical techniques is by no means as easy as that of haemoglobin. Firstly, the cytochrome signal is at most one tenth of the haemoglobin signal in amplitude.7 Secondly, the determination of in vivo spectrum is done through animal experimentation, which at the first attempts was complicated by residual haemoglobin signals and by agonal swelling of cells and subcellular elements which influence scattering. Thirdly, there is no reference method.

Path length The length of the light path traversing the tissue must be known in order to calculate concentrations, i.e. for quantitative measurements. The path length in tissue exceeds the geometrical distance between the optodes by a factor 3e6 (this factor is named the ‘differential path length factor’, DPF). Estimation of path length is one of the basic problems in NIRS.

Measurements in infants Trend monitoring of haemoglobin signals Near-infrared spectroscopy is a perfect candidate for clinical monitoring of the tiny sick preterm neonates. It is non-invasive, gives real-time information, does not interfere with intensive care, does not affect the underlying skin, and does not remove the infants from the nursery. In principle, NIRS allows on-line trending of changes in O2Hb and HHb, and hence of tHb (the sum of [O2Hb] and [HHb]), which is proportional to changes in cerebral blood volume, which in turn can be used as a surrogate measure of cerebral blood flow. The appropriateness of this, however, has only been established for reactions to changes in arterial carbon dioxide tension.4 Furthermore, constant optode distance is crucial; if head circumference changes even by a fraction of a millimetre as a result of change in brain blood or brain water content, the trends are significantly biased. Minor changes in optodeeskin contact induce large transients in the signal and/or baseline shifts. The difference between [O2Hb] and [HHb] e called Hbdiff e when divided by a factor of 2 is called oxygen index (OI) e is an indicator of the mean oxygen saturation of the haemoglobin in all types of blood vessels in the tissue. This quantity has been shown to change appropriately in many experimental and clinical studies but has important limitations. Firstly, in terms of interpretation it is not known how much of the signal comes from blood in arteries, capillaries, and veins, respectively. Observations in piglets

Optical imaging/multi-regional monitoring By the use of two or more optodes, regional information may be compared to distinguish between general, systemic responses to stimuli e e.g. due to fluctuation in arterial blood pressure or ventilation e from regional responses due, e.g., to focal neuronal activation. Neuronal activation is associated with a rapid vasodilatation to meet the increased metabolic needs and oxygen uptake with increased oxygen supply. In fact, the vasodilatation precedes and overshoots the need, and as a result [O2Hb] increases, detectable by NIRS and [HHb] decreases, detectable by the blood-oxygenation-level-dependent (BOLD) effect using functional magnetic resonance imaging (fMRI). Whereas neuronal activation in healthy adults is always associated with focal hyperoxygenation, the vascular reaction in newborns appears to be less robust, so the oxygenation reaction may be absent or even inverse.8e11

Quantification of cerebral blood volume (CBV) The effect of a small induced change in arterial oxygen saturation (SaO2) on [HbO2] may be used to quantify CBV12 on the basis of the indicatoredilution principle. It is assumed that a small change in arterial oxygen saturation within the normal range will not significantly influence cerebral blood volume, flow or oxygen consumption. ½tHb Z 100ðD½HbO2   D½HbÞ=ð2DSaO2 Þmmol=L if SaO2 is expressed in %. The total cerebral haemoglobin is directly proportional to the cerebral blood volume: CBV Z k½tHbmL=100 g where k Z 100/(Hb  R  1.05), Hb is the blood haemoglobin content in mmol/L (tetrahaeme), R is the cerebralto-large vessel haematocrit ratio (usually taken as 0.69), and the factor 1.05 g/mL is the brain density.

500

G. Greisen

Quantification of cerebral blood flow (CBF) Measurement of blood flow by NIRS is based on Fick’s principle14 and uses a rapid change in arterial oxyhaemoglobin as an intravascular tracer. By using the change in OI observed after a small sudden change in arterial concentration of oxygen, CBF can be calculated as follows: CBF Z DOI=ðk

R

SaO2 dtÞmL=100 g=min

where OI is measured in units of mmol/L, and k Z Hb  1.05  100, Hb is blood haemoglobin in mmol/L (tetrahaeme), SaO2 is in %, and t is time in minutes. The method of measuring CBF rests on several assumptions. First, during measurement, CBF, CBV and oxygen extraction must be constant. Second, the period of measurement must be less than the cerebral transit time (approximately 10 s). Finally, this method of CBF measurement has practical limitations: in infants with severe lung disease the SaO2 may be fixed at a low level despite administration of oxygen, whereas in infants with normal lungs SaO2 is near 100%, even in room air. This can be overcome by using airenitrogen mixtures and then switching to room air. Measurements of blood flow with NIRS have been compared to 133Xe clearance in sick newborn infants. These comparisons constitute important direct external validation of NIRS in the brains of human neonates. The agreement between the two methods is acceptable.15,16

Quantification of cerebro-venous oxygen saturation Cerebral venous haemoglobin saturation reflects the balance between O2 delivery and O2 consumption. A normal cerebral venous oxygen saturation demonstrates an intact coupling between CBF and the metabolic needs. During restricted blood flow, enhanced oxygen extraction is expected to occur and to result in a drop in cerebro-venous saturation. Cerebral venous oxygen content may be estimated by near-infrared spectrophotometry.2 When venous outflow from the brain is impeded by tilting the head down or by jugular venous occlusion [tHb] increases. Assuming that this is due exclusively to pooling of blood in venoles and veins, cerebral SvO2 can be measured using the formula: cSvO2 Z 100D½HbO2 =ðD½HbO2  þ D½HbÞ% The non-invasive method of measuring cSvO2 with partial jugular venous occlusion was validated with an invasive measurement of SvO2 from co-oximetry of jugular bulb blood obtained during cardiac catheterization and gave similar values.17

Quantification of tissue oxygenation index Three different principles are used in second-generation instruments to estimate vascular haemoglobin oxygen saturation in absolute terms without the manipulation of FiO2 or the impeding venous outflow. First, detection of transmitted light at two or more different distances from the light-emitting optodes (called spatially resolved spectroscopy) allows monitoring of the ratio of absolute [O2Hb] to [tHb] e i.e. the haemoglobin saturation, called ‘tissue oxygenation index’ (TOI)18 e in the brain, called cerebral TOI (cTOI) (Fig. 1). This measure is the weighted average of arterial, capillary and venous blood oxygenation, and hence cannot be easily validated. The measurement depends on the tissue being optically homogeneous, which is unlikely to be strictly the case. Nevertheless, values near cerebro-venous values have been found, and appropriate changes have been found with changing arterial oxygen saturation, and with arterial pCO2. The signal-to-noise ratio is not as good as that of OI, and hence for quantifying the response to rapid therapeutic interventions cTOI is less useful. Similarly, by time-resolved spectroscopy detecting the time-of-flight of very short light pulses,6 or detecting the phase shift and phase modulation of a continuous frequency-modulated source of light,19 it is possible to estimate the absolute concentrations of oxy- and deoxyhaemoglobin, and from these to calculate cTOI. Bias of cTOI The first problem is that cTOI cannot be compared directly to any other measurement because it relates to a mix of blood in veins, capillaries, and arteries (see above). cTOI has been validated in young infants with heart disease during cardiac catheterization. Across a range of cTOI of 40e80%, the mean value was almost identical to saturation

90 80

Tissue oxygenation index (%)

Values for CBV obtained by NIRS are lower than those reported in adults. One explanation could be the lower cerebral blood flow in newborn infants compared to that in adults. Changes in CBV, induced by bilateral jugular venous occlusion for 5 s, as estimated by NIRS using tHb, correlated well with strain-gauge plethysmography.13 This result remains one of the few external quantitative validations of NIRS in the neonatal brain.

70 2. measurement 60 1. measurement 50 40 30 20

Optode re-siting

10 0

Figure 1 Tissue oxygenation index applied to the head is a surrogate measure for cerebro-venous saturation. This may be monitored from second to second by commercial instruments, and the value is available in absolute terms. Typical values are 60e80%. The signal noise is usually 2e3% and hence larger than what is known from, e.g. pulse oximetry. If averaging over a minute is done, however, a very precise mean value may be obtained. The problem is that this mean value differs from place to place. When re-siting the optode, differences of 7% are common and may reach 15%.

Near-infrared spectroscopy in jugular venous blood as measured by co-oximetry.20 This suggests a negative bias, given that TOI represents arterial and capillary as well as venous blood. TOI has not been compared to other measures of venous saturation in newborn or preterm babies, and not even internal consistency of TOI with SvO2 measured by NIRS during obstruction of venous outflow has been reported.

Precision of cTOI In the study in young infants with heart disease cited above, whereas the mean difference was negligible, limits of agreement of individual measurements were as wide as 12% to þ11%. Since co-oximetry is very precise, this represents an estimate of the error of cTOI. Using the same commercial instrument in newborns and young infants, limits of agreement after optode replacement was 17% to þ17%.21 In preterm babies we have found that the variation associated with replacement of the optodes could fully account for this poor precision (unpublished data). Presumably the main reason for the variability are small differences in anatomy causing the assumption of scatter isotropy to fail. For comparison, arterial haemoglobin oxygen saturation can be measured by pulse oximetry with limits of agreement of 6%. The significance of the low precision of cTOI cTOI is a surrogate measure of cerebro-venous saturation. This important physiological variable is tightly regulated, normal values being 60e70%. During hypoxaemia, cerebrovenous saturation will fall in parallel, whereas during cerebral ischaemia (i.e. cerebral blood flow drops without cerebral oxygen consumption dropping also) cerebrovenous saturation will fall. Hence cerebro-venous saturation is a useful measure of the sufficiency of cerebral perfusion. The problem is that a drop in cerebral blood flow by 30% will only lead to a drop in cerebro-venous saturation from 70% to 60%, i.e. well within the limits of measurement error.

Clinical correlates of cTOI In a group of 18 severely asphyxiated term infants, those with poor outcome showed a rise in cTOI during the first day of life.22 This is in agreement with the concept of delayed energy failure and ‘luxury’ perfusion. In a group of 20 newborn infants operated for congenital transposition of the great vessels, those with preoperative cTOI <35% tended to have lower developmental scores at follow-up at 2e3 years of age.23 It should be noted that in both studies the differences were seen at the group level. Imprecision of cTOI, presumably, contributed to the fact that prediction at the individual level was not possible.

Conclusion and the future NIRS has matured as a method to obtain quantitative measures of cerebral blood volume, flow and oxygenation, and has yielded credible and sometimes important information. These methods are manual, however, and too

501 cumbersome for clinical practice or large-scale research. Second-generation instruments directly quantify a tissue oxygenation index, which is a surrogate measure of venous saturation, an important parameter for judging tissue oxygen sufficiency. The precision of these instruments, however, is insufficient for clinical use.

Practice points  Clinical NIRS monitors are commercially available.  Output is ‘tissue oxygenation index’ e a surrogate measure of venous saturation.  The precision of TOI is insufficient to be useful.

Research directions  Improving precision.  Investigate clinical utility.

References 1. Brazy JE, Lewis DV, Mitnisk MH, Jo ¨bsis FF. Noninvasive monitoring of cerebral oxygenation in preterm infants: preliminary observation. Pediatrics 1985;75:217e25. 2. Wyatt JS, Cope M, Delpy DT, Wray S, Reynolds EOR. Quantification of cerebral oxygenation and haemodynamics in sick newborn infants by near infrared spectrophotometry. Lancet 1986;2:1063e6. 3. Matcher SJ, Elwell CE, Cooper CE, Cope M, Delpy DT. Performance of several published tissue near infrared spectroscopy algorithms. Anal Biochem 1995;227:54e68. 4. Pryds O, Greisen G, Skov L, Friis-Hansen B. The effect of PaCO2 induced increase in cerebral blood volume and cerebral blood flow in mechanically ventilated, preterm infants. Comparison of near infra-red spectrophotometry and 133Xenon clearance. Pediatr Res 1990;27:445e9. 5. Brun NC, Moen A, Borch K, Saugstad OD, Greisen G. Near-infrared monitoring of cerebral tissue oxygen saturation and blood volume in newborn piglets. Am J Physiol 1997;273:H682e6. 6. Ijichi S, Kusaka T, Isobe K, Kawada K, Imai T. Quantification of cerebral hemoglobin as a function of oxygenation using nearinfrared time-resolved spectroscopy in a piglet model of hypoxia. J Biomed Opt 2005;10:024026. 7. Cope M, van der Zee P, Essenpreis M, Arridge SR, Delpy DT. Data analysis methods for near infrared spectroscopy of tissue: problems in determining the relative cytochrome aa3 concentration. Proc SPIE 1991;1431:251e62. 8. Meek JH, Firbank M, Elwell CE, Atkinson J, Braddick O, Wyatt JS. Regional hemodynamic responses to visual stimulation in awake infants. Pediatr Res 1998 Jun;43(6):840e3. 9. Born P, Leth H, Miranda MJ, Rostrup E, Stensgaard A, Peitersen B, et al. Visual activation in infants and young children studied by functional magnetic resonance imaging. Pediatr Res 1998 Oct;44(4):578e83. 10. Taga G, Asakawa K, Maki A, Konishi Y, Koizumi H. Brain imaging in awake infants by near-infrared optical topography. PNAS 2003;100:10722e7.

502 11. Konishi Y, Taga G, Yamada H, Hirasawa K. Functional brain imaging using fMRI and optical topography in infancy. Sleep Med 2002;3:S41e3. 12. Wyatt JS, Cope M, Delpy DT, van der Zee P, Arridge S, Edwards AD, et al. Quantitation of cerebral blood volume in newborn infants by near infrared spectroscopy. J Appl Physiol 1990;68:1086e91. 13. Wickramasinghe YABD, Livera LN, Spencer SA, Rolfe P, Thorniley MS. Plethysmographic validation of near infrared spectroscopic monitoring of cerebral blood volume. Arch Dis Child 1992;67:407e11. 14. Edwards AD, Wyatt JS, Richardson CE, Delpy DT, Cope M, Reynolds EOR. Cotside measurement of cerebral blood flow in ill newborn infants by near-infrared spectroscopy. Lancet 1988;2:770e1. 15. Skov L, Pryds O, Greisen G. Estimation cerebral blood flow in newborn infants: comparison of near infrared spectroscopy and 133Xe clearance. Pediatr Res 1991;30:570e3. 16. Bucher HU, Edwards AD, Lipp AE, Duc G. Comparison between near infrared spectroscopy and 133Xenon clearance for estimation of cerebral blood flow in critically ill preterm infants. Pediatr Res 1993;33:56e60. 17. Yoxall CW, Weindling AM, Dawani NH, Peart I. Measurement of cerebral venous oxyhaemoglobin saturation in children by near infrared spectroscopy and partial jugular venous occlusion. Pediatr Res 1995;38:319e23.

G. Greisen 18. Suzuki S, Takasaki S, Ozaki T, Kobayashi Y. A tissue oxygenation monitor using NIR spatially resolved spectroscopy. Proc SPIE 1999;3597:582e92. 19. Zhao J, Ding HS, Hou XL, Zhou CL, Chance B. In vivo determination of the optical properties of infant brain using frequency-domain near-infrared spectroscopy. J Biomed Opt 2005;10:024028. 20. Nagdyman N, Fleck T, Schubert S, Ewert P, Peters B, Lange PE, et al. Comparison between cerebral tissue oxygenation index measured by near-infrared spectroscopy and venous jugular bulb saturation in children. Intensive Care Med 2005;31: 846e50. 21. Dullenkopf A, Kolarova A, Schulz G, Frey B, Baenziger O, Weiss M. Reproducibility of cerebral oxygenation measurement in neonates and infants in the clinical setting using the NIRO 300 oximeter. Pediatr Crit Care Med 2005; 6:344e7. 22. Toet MC, Lemmers PMA, van Schelven LJ, van Bel F. Cerebral oxygenation and electrical activity after birth asphyxia: their relation to outcome. Pediatrics 2006;117:333e9. 23. Toet MC, Flinterman A, van de Laar I, Vries JW, Bennink GB, Uitervaal CS, et al. Cerebral oxygen saturation and electrical brain activity before, during, and up to 36 hours after arterial switch procedure in neonates without pre-existing brain damage: its relationship to neurodevelopmental outcome. Exp Brain Res 2005;165:343e50.