Cerebral Oximetry

Cerebral tissue oxygen saturation (SctO₂), also known as cerebral oximetry (rSO₂), estimates regional tissue oxygenation by transcutaneous illumination with ‘near-infra-red’ light in the area of the frontal cerebral cortex. Adhesive sensors applied to the forehead of the patient emit ‘near-infra-red’ light and detect the reflected waves via photodetectors, a process called ‘near-infra-red’ spectroscopy (NIRS). The waves are reflected from different tissue depths, including the cranial bone and any underlying cerebral tissue. The collected signal can be processed in different ways.

NIRS emits light of a certain frequency and amplitude, and the relative changes in light intensity of the reflected waves are related to changes in relative concentrations of haemoglobin through the modified Beer–Lambert equation. One of the factors needed to derive the absolute changes in haemoglobin concentration (THC = cerebral total haemoglobin concentration, the sum of oxy- and deoxy-haemoglobin, expressed in µmol) is the length of the path the reflected photon has travelled back. Continuous wave (CW) NIRS emits waves of constant frequency and amplitude. Because these do not allow the path length to be determined, CW NIRS can only be used as a trend monitor. Frequency-domain (FD) NIRS uses an amplitude modulated sinusoid signal [21]. The extra information contained in the reflected signals (both amplitude and phase changes) can distinguish light (photon) absorption from light scattering, making it possible to derive absolute changes in THC. Meng et al used this more quantitative approach [11].

The blood contained in brain tissue consists, on average, of 25% arterial and 75% capillary and venous blood. Normal values for SctSO₂ range between 60% and 80%. Most commercial devices consider a trend towards 20% below baseline a clinically relevant change. Many factors affect the NIRS: cardiac output, ABP, F_IO₂, temperature, PaCO₂ (both hypo- and hypercapnia), and extracranial blood [22].

Cerebral tissue oxygen saturation reflects the balance between O₂ supply and demand, while CBF only represents supply. While the reliability of SctO₂ as a surrogate measure of CBF is a matter of discussion, NIRS has been validated in numerous studies as a means to evaluate cerebral autoregulation [23, 24]. If autoregulation fails to maintain CBF, the brain will compensate for this by increasing O₂ extraction. NIRS thus becomes an indirect and non-invasive monitor of the vulnerability of cerebral perfusion. A decrease in NIRS can be attributed to any of the following, either alone or combined: (1) increased cerebral oxygen metabolic rate; (2) decreased oxygen delivery to the brain; and (3) decreased arterial and/or increased venous blood contribution(s).

A new concept derived from SctO₂ is the cerebral oxygen index (COx). The COx is obtained by calculating the correlation coefficient between 150 paired samples of 2-second recordings of mean arterial pressure (MAP) and 300-second epochs of SctO₂. A COx that approaches zero indicates that ABP is within the autoregulatory range. A COx approaching one indicates a pressure passive cerebral circulation, the situation where a decrease in cerebral perfusion pressure (CPP) causes a decrease in CBF. The COx threshold to discriminate between intact and impaired CA is arbitrary and varies between 0.25 and 0.50 [15].

Transcranial Doppler

Transcranial Doppler ultrasonography (TCD) provides relatively high-fidelity recordings of CBF velocity [25] and is one of the most commonly used methods to measure CBF in clinical studies [26]. TCD remains an indirect measure of CBF because blood velocity measurements accurately reflect volumetric blood flow only if the cross-sectional area of the insonated vessel remains constant. For this reason in some patients it might not reflect the actual CBF.

The rate of regulation (RoR) is an index calculated from TCD changes in response to a rapid, short-term decrease in ABP after, for example, sudden deflation of leg cuffs (dynamic) or after changing MAP itself by, for example, a phenylephrine bolus (static). The reaction of CBF velocity to different levels of perfusion pressure suggests that pressure-induced changes in cerebrovascular resistance (CVR) has two components, a ‘rapid response component’ (sensitive to pressure pulsations) followed by a ‘slow response component’ (sensitive to changes in MAP) [27].

One of the options to analyse data from haemodynamic or brain waveforms is ICM+® [28]. ICM+ collects data coming from the analogue output of the arterial blood pressure and TCD monitors. Using a method called time-wave integration, average values for MAP, intracranial pressure (ICP) and mean flow velocity (FVm) can be estimated. Mean Index (Mx) is a parameter extracted from those waveforms [29]. Mx reflects the moving linear correlation coefficient between values of CPP (see definition below) and FVm. Its strength is that it allows spontaneous fluctuations in MAP to be used to characterize autoregulation, eliminating the need for drugs or manoeuvres such as compression of the carotid artery or bilateral thigh cuffs to induce changes in MAP. Furthermore, it allows the efficiency of autoregulation to be quantified, with a higher Mx indicating a less efficient autoregulation.

While Mx and systolic (Sx) indices were originally calculated using CPP [29], Piechnik et al [1] found that MAP could be used as well. This modification does not invalidate the sensitivity of Mx and Sx to CO₂ induced changes of cerebral vasodilatory capacity. Although their reliability is not as good as RoR obtained with Aaslid’s leg cuff tests [25], Mx and Sx have the advantage of being parameters that can be continuously measured and are considered particularly useful when continuous monitoring of cerebrovascular reactivity is needed.

The availability of new, non-invasive technologies that measure cerebral oxygen flow via image or signal analysis provides the clinician with reliable tools to assess the state of the brain almost in real time.

Cerebral Perfusion Pressure

CPP is defined as the difference between MAP and ICP (or central venous pressure if it is higher than ICP). CPP represents the pressure gradient for CBF and thus is one of the key factors that determine oxygen and metabolite delivery. CPP must be kept within a very narrow range to avoid inadequate blood inflow to the brain.

Because the transmural pressure in distal cerebral veins is very close to zero, the ICP becomes the principal determinant of the postcapillary venous outflow pressure in a manner that is similar to a Starling resistor [30]. It has been suggested that the sum of ICP and the active tension produced by vascular smooth muscle contraction at the arteriolar level can be considered the pressure at which flow ceases. The arterial pressure at which cerebral flow ceases is called the zero flow pressure (ZFP). It is determined by arterial tone and represents the effective downstream pressure [31]. ZFP can be indirectly calculated from the instantaneous relationship between the blood flow velocity (FV) of the middle cerebral artery (MCA) and MAP during a cardiac cycle. The gradient between MAP and ZFP determines CPP. Because critical closing pressure can be as high as 30 mmHg in the supine position [32], the common practice of only considering MAP without taking effective downstream pressure into account can lead to marked overestimation of true CPP.

Maintaining an adequate CPP is fundamental when managing patients under anaesthesia or in critical condition. Abrupt changes in CPP may induce resistance changes in cerebral autoregulatory arterioles, i.e. arteriolar vasoconstriction when pressure increases and arteriolar vasodilation when pressure decreases. The resultant change in arterial to venous blood volume ratio can explain the observed change in SctO₂ [8, 16].

Cerebral Flow Autoregulation

A low MAP can cause inadequate cerebral perfusion, a situation where O₂ supply fails to meet O₂ demand. However, whether or not this actually occurs in a given patient will depend on the autoregulatory response range (Fig. 12.1 [33]).

Fig. 12.1 Cerebral autoregulation plot (reproduced with permission from [33]).

Cerebral autoregulation is visualized as a correlation plot between cerebral blood flow (CBF) and cerebral perfusion pressure (CPP). CBF remains stable between the lower limit (LL) and the upper limit (UL) (portion B, plateau). CBF is pressure passive at the CPP range below the lower limit (portion A) and above the upper limit (portion C). This illustration uses a CPP of 60 mmHg as the lower limit, a CPP of 150 mmHg as the upper limit, and a CBF of 50 mL/min per 100 g as the plateau. However, these regularly quoted numbers are not fixed; rather, they vary inter-individually and intra-individually depending on a variety of factors. Therefore, we take a note of SD to emphasize that these parameters have a wide range of distribution. The cerebrovascular reactivity is also illustrated. SD, standard deviation.

The classic teaching of cerebral autoregulation is that CBF is maintained at a constant level across a wide range (plateau) of MAPs, with a lower limit of autoregulation (LLA) of 50 mmHg and an upper limit (ULA) of 150 mmHg [34]. However, studies in awake healthy subjects indicate that the LLA usually is around 70 mmHg (higher than the 50 mmHg still reported in textbooks) and with a huge range (43–110 mmHg). It is therefore impossible to know the LLA in every patient at any given time [35]. In fact, ‘preserved cerebral autoregulation’ does not mean the position of the CBF–CPP plot remains unchanged; it only means that the link between the two remains intact.The LLA also depends on the cause of hypotension: in baboons, it is 65% of baseline MAP during experimental haemorrhagic hypotension but only 35% to 40% during drug-induced hypotension [36]. Meng et al [37] propose a framework that integrates the various CBF-regulating processes at the level of the cerebral arteries/arterioles. Depending on the mechanism, different segments of those vessels might be affected, e.g. sympathetic stimulation constricts large arteries, while an increase in MAP constricts the arterioles (Fig. 12.2 [37]).

Fig. 12.2 Integrated regulation of brain perfusion. (Reproduced with permission from [37].)

The conceptual framework of the integrated regulation of brain perfusion. The cerebrovascular resistance determined by the calibre of the cerebral resistance vessels is regulated by various physiological processes: (1) cardiac output (CO) likely via sympathetic nervous activity (SNA) and renin–angiotensin–aldosterone (RAA) system, depending on the chronicity of the change in CO, (2) arterial blood pressure (ABP) and cerebral perfusion pressure (CPP) via cerebral autoregulation, (3) cerebral metabolic activity via neurovascular coupling, and (4) arterial blood carbon dioxide (CO₂) and oxygen (O₂) via cerebrovascular reactivity. The SNA regulates cerebral blood flow and may play a prominent role during acute hypertension and hypercapnia as a protective mechanism preventing cerebral overperfusion (dashed line). These various regulatory mechanisms, together with other CBF-regulatory mechanisms that are not specified here such as anaesthetic effects, integrate at the level of the cerebral resistance vessels and, therefore, jointly regulate brain perfusion. The plateau of the autoregulation curve shifts downward when the CO is reduced and upward when augmented. The position of the plateau is determined by the calibre (R) of the cerebral resistance vessels at high (R_high), normal (R_norm), and low (R_low) CO. The scale of CO on the right side is smaller than that of CBF on the left side to reflect the lesser extent of change in CBF induced by an alteration of CO.

Studies under various experimental settings have all demonstrated pressure–flow curves resembling Lassen’s autoregulatory curve with the classic shape of a plateau and shoulders. However, most of these studies are inconsistent in that some report the plateau to be relatively wide (40 mmHg) [38] while others find it to be demonstrably narrow (i.e. 10–15 mmHg) [39]. Moerman et al [16] found substantial inter-subject variation in the response to pharmacology-induced pressure changes. This implies that the relation between MAP and SctO₂ may follow different patterns and that intact cerebral autoregulation may manifest itself not only as an autoregulatory plateau but also as a paradoxical response. Jones et al [40] demonstrated that the classic pattern is the result of averaging individual cerebral autoregulation curves, a finding confirmed by Moerman [16].

In patients with supratentorial tumours, preoperative cerebral autoregulation and autoregulation during the first 24 hours postoperatively was impaired in the 20% of patients (seven patients) that had a large tumour (average volume 100 cm³) and a midline shift of more than 5 mm, but not in those with smaller tumours (average volume 40 cm³) [3].

The Concepts of Static and Dynamic Autoregulation

An arterial blood pressure tracing recorded over several seconds is a complex pulsatile waveform with several characteristic features that are the result of intrinsic cardiovascular properties (e.g. ventricular preload, vascular compliance, blood properties) and pulse wave reflection. But when recorded over a longer period (minutes, hours or days), the ABP exhibits different trends and oscillations. Superimposed, one can expect non-rhythmic blood pressure surges and dips that occur either spontaneously, or as a consequence of physical activities such as exercise, straining, coughing and changing body posture (see Fig. 12.3, and page 185). Thus, ABP is not a static variable, but rather represents a conglomerate of physiological changes that cannot be fully described in terms of simple numerical averages [30]. Clinical studies suggest that both short (minutes) and long-term (hours to days) episodes of high ABP variability increase the likelihood of adverse cerebrovascular outcomes.

Fig. 12.3 Finger arterial blood pressure and middle cerebral blood flow velocity over 900 s (presented on long axes) for a human subject in the seated resting position (upper left). The corresponding power spectra, which decompose the time series signals into their various constituent component frequencies (upper right). Note that the fundamental frequency (f0) and its harmonics (f1, f2, f3) correspond to pulsations coincident with the pulse, whereas progressively lower frequency components reflect the longer term oscillations and trends in the time domain. ULF, ultra low frequency; VLF, very low frequency; LF, low frequency; HF, high frequency. Transfer function coherence, phase, gain and normalized gain for 105 healthy individuals (mean age 26±7 years) in the supine resting position. AU, arbitrary units. Values are mean ± SE (bottom) [30].

Static cerebral autoregulation refers to the reflex adjustments in cerebral vascular resistance in response to steady-state alterations in ABP [30]. Static cerebral autoregulation can be evaluated by administering incremental doses of vasopressor drugs to reach a predetermined percentage increase or decrease of MAP while simultaneously monitoring the NIRS response (SctO₂ is used as a surrogate measure of the CBF response). The phenylephrine-induced increase in perfusion pressure induces vasoconstriction of the cerebral autoregulatory arterioles to prevent abrupt cerebral hyperperfusion, as occurs with increased sympathetic nerve activity. This causes a decreased arterial blood contribution to NIRS measurements resulting in a lower SctO₂. The SctO₂ increase with induced hypotension found by Moerman et al [16] could be explained by the same mechanism but in the opposite direction: to prevent cerebral hypoperfusion, cerebral arterioles dilate, increasing the arterial to venous blood volume ratio, which increases SctO_2. This phenomenon has been described as ‘hyperautoregulation’ [40].

Dynamic cerebral autoregulation refers to the vascular responses to higher frequency components of steady-state spontaneous ABP, or to dynamic changes in blood pressure such as those driven by altered body posture [30]; it is not affected by the acute change in CO [41]. Its integrity may be tested with the transient hyperaemic response (THR) of the middle cerebral artery after the release of a 10-second compression of the ipsilateral common carotid artery (monitored by TCD sonography). The transient hyperaemic response ratio (THRR) is calculated as the ratio of the mean systolic flow velocity from five heart cycles just prior to compression over the mean systolic value of two heart cycles after compression, with exception of the very first cycle. A value of THRR greater than 1.1 defines normal autoregulation. A valid THR test needs to meet some criteria [3]: (1) compression of the carotid artery leads to a sudden and maximum decrease in middle cerebral arterial blood velocity (Vmca); (2) heart rate and blood pressure remain stable during compression (within 10% of their respective pre-compression values); (3) no flow transients occur after release of compression; and (4) the power of the reflected Doppler signal does not change during the test.

Aaslid et al [25] tested cerebral autoregulation by using TCD to characterize the CBF velocity responses to abrupt release of inflated thigh occlusion cuffs. Cuff deflation induces sudden stepwise drops in ABP that remains depressed for approximately five to seven seconds before cardiac and vascular baroreflexes gradually restore it to baseline ABP over a 10–20 second period. The flow velocity (Vmca) response pattern is similar except for the fact that it recovered more rapidly.

The pressure and blood flow velocity recordings collected over prolonged periods (see above and Fig. 12.3) can be decomposed into their various oscillatory components. Transfer functions can be derived that define the relationships between blood pressure (input) and flow velocity (output) in terms of their linear statistical dependence (coherence), relative magnitudes (gain) and timing (phase) as a function of the frequency component of interest [30]. Between 0.07 and 0.20 Hz, there is a pattern of progressively increasing coherence and gain indicating that higher frequency blood pressure components are more linearly related to flow velocity than lower frequency components, and that the ratio between them increases. The transfer function characteristics within the 0.02–0.4 Hz range have been attributed to the capacity for cerebral arterioles to dilate and constrict dynamically in response to increases and decreases in the blood pressure. As yet, which of these metrics or combination of metrics one should use to assess dynamic cerebral autoregulation remains unknown.

Dynamic cerebral autoregulation can also be evaluated by using Mx, the moving linear correlation coefficient between values of cerebral perfusion pressure (CPP) and FVm [6].

Tiecks et al [42] demonstrated that in normal anaesthetized subjects measurement of dynamic cerebral autoregulation yields similar results to static testing of both the intact and pharmacologically impaired autoregulation.

Cerebrovascular Carbon Dioxide Reactivity

The interaction between the autoregulatory response and CO₂-induced dilation of cerebral vessels mainly involves the smaller arterial vessels [43]. Ito et al used positron emission tomography to demonstrate that changes in cerebral blood volume (CBV) during hypocapnia and hypercapnia are caused by changes in arterial blood volume without any changes in venous and capillary blood volume [44].

Carbon dioxide is a potent cerebral vasodilator, and a linear relationship exists between CBF and PaCO₂ within the range of 20 to 60 mmHg (2.5 and 8.0 kPa) [45]. The proposed mechanism for this effect is thought to be mediated via CO₂-related changes in extracellular pH [46]. Changes in CVR and CBF in response to changes in PaCO₂ are referred to as ‘cerebrovascular reactivity to CO₂’ or CVR-CO₂. Absolute CVR-CO₂ is defined as the change in CBF (mL/min or cm/s) per mmHg change in PaCO₂. Generally accepted values under anaesthesia are 1–2 mL/100g/min/mmHg (2.0 to 5.0 cm/s/mmHg). Relative CVR-CO₂ is defined as the percentage change relative to baseline: 2% to 4% change in mL/min/mmHg or 2.5% to 6% change in cm/s/mmHg from the baseline value. The relative reactivity value is by definition less dependent on baseline values than absolute reactivity and thus the better indicator of CVR-CO₂ for statistical analysis; it is also used as a measure of cerebrovascular integrity and cerebral autoregulation.

CVR-CO₂ can be influenced by disease states. A strong correlation has been reported between impaired CVR-CO₂ and high glycosylated haemoglobin (HBA1C) in insulin-dependent DM [47]. This impaired reactivity may be due to micro- and macro-angiopathic changes in cerebral arterioles associated with chronic hyperglycaemia. In patients with supratentorial tumours, however, CO₂ reactivity was normal preoperatively and remained so after tumour decompression [3].

Carbon dioxide is a modulator of cerebral vascular tone. Hancock et al [31] found that EtCO₂ changes produce significant changes in estimated CPP (eCPP) and in ZFP in a group of healthy volunteers. Hypocapnia decreases eCPP and increases the pressure needed to perfuse the brain, while hypercapnia increases eCPP and decreases ZFP. The association of hypocapnia and hypotension decreases eCPP and increases ZFP, which implies that a higher MAP is needed to maintain adequate brain perfusion.

A common practice in neuroanaesthesia and neurocritical care is to induce hypocapnia to reduce cerebral blood volume and intracranial pressure. Hypocapnia results in a shift of the plateau to lower cerebral blood flows, almost no change in LLA, and a poorly defined change of the ULA. Conversely, hypercapnia causes the CBF autoregulation plateau to progressively ascend as CO₂ rises, shifts the LLA to the right and the ULA to the left, thus narrowing the range over which CBF is maintained until it disappears, ultimately resulting in a pressure passive system [33] where CBF flow passively follows pressure.