This chapter describes a number of methods for measurement of blood flow, nearly all tried in newborns because they had worked in adults. Most often they have been applied to the brain, with good-quality flow imaging achieved. None of the methods has really been established as a reference, but the “consensus” is that the normal global cerebral blood flow in healthy term newborn infants during the postnatal period is approximately 20 mL/100 g/min and, in preterm infants, somewhat lower (i.e., less than half of the values in adults). To be used for acute clinical care in newborns, a method must be precise to within ±10% and able to detect normal autoregulation and normal reactivity to partial pressure of carbon dioxide and must furthermore be feasible and safe in the intensive care unit. It must also be possible to relate the values to an ischemic threshold and to continuously monitor the changes. Unfortunately, no method so far has lived up to these requirements. Therefore monitoring of tissue oxygenation has been explored as a surrogate.
Keywords133Xenon clearance, arterial spin labeling, blood flow, diffuse correlation spectroscopy, Doppler, infant, MR-phase contrast, near-infrared spectroscopy, positron emission tomography, single photon emission tomography
Almost all methods for measuring cerebral blood flow (CBF) have been applied to newborn infants.
The normal global CBF in healthy newborn infants is likely to be only approximately 20 mL/100 g/min.
Flow imaging with magnetic resonance imaging (and positron emission tomography) yields expected patterns with high flow to the brainstem, midbrain, cerebellum, and central gray matter and intermediate flows to cortex.
Monitoring of cerebral oxygenation may serve as a surrogate for CBF, but its clinical value is still undetermined.
This chapter describes the methods available to assess organ blood flow in the neonate and discusses their strengths, weaknesses, and sources of errors. Most detail is given on Doppler ultrasound and diffuse optical methods with near-infrared (NIR) light because these are the most practical methods. The focus is on the measurement of blood flow to the brain (cerebral blood flow [CBF]) because this is where almost all the experience is. A brief section at the end addresses the experience with assessment of blood flow to other organs.
Organ blood flow is usually expressed in mL/min because this is the most descriptive unit when an organ is supplied from a single artery and/or drained via a single vein. Indeed, in animal experimentation the simplest method to measure blood flow to an organ is to drain the venous outflow into a calibrated container. The description of blood flow in mL/min has been used to judge the flow to an organ expressed as a fraction of cardiac output. Whether this fraction changes during development or under pathologic conditions, such as hypoxia, arterial hypotension, and/or low cardiac output states, is an important and clinically relevant question in the field of developmental physiology and pathophysiology. Finally, to allow comparisons among groups of infants of different gestational age and thus body weight, it is useful to normalize blood flow for body weight and express it as mL/min/kg.
However, organ blood flow can also be expressed in mL/100 g tissue/min. This measure may refer to an organ as a whole or to a specific region or compartment in a given organ, depending on the method of measurement. The simplest methods to assess organ blood flow use the Fick principle for inert tracers: “Flow equals the rate of change in tissue concentration of tracer divided by the arteriovenous concentration difference of the tracer.” This measure is especially useful to assess the flow in relation to metabolism and organ function.
It is important to emphasize that blood flow is a complex and dynamic variable. Aside from physiologic fluctuations in organ blood flow governed by the changes in functional activity and thus the metabolic demand of a given organ, blood flow may significantly change within seconds under pathologic conditions, such as with abrupt changes in blood pressure or the onset of hypoxia. In addition, blood flow may vary from one part of an organ to the other, as during functional activation, or, during stress, the distribution may change markedly.
For several decades, authors studying the management of circulation during provision of intensive care to the neonate have been pointing out the need to consider blood flow, rather than only arterial blood pressure. Therefore thoughtful neonatal practice would include the use of indirect measures of blood flow, such as skin color, peripheral–core temperature difference, capillary refill time, urine output, and lactic acidosis. Unfortunately, these indirect signs of tissue perfusion either lack sensitivity and specificity (i.e., peripheral–core temperature difference, capillary refill time) or do not represent the changes in the hemodynamic status in a timely manner (i.e., urine output and lactic acidosis).
As for the methods available for the more direct assessment of organ blood flow in neonates, very few units routinely use these tools. Why is this so?
There are three main reasons why assessment of organ blood flow has not become routine in clinical practice. First, none of the many methods used for research has achieved broader application because none satisfies the requirements in terms of ease of use, precision, accuracy, noninvasiveness, and cost. In reality, therefore no method is truly available to clinicians who want to upgrade their clinical practice. Methods using “standard” equipment (e.g., ultrasound) require much skill and expensive equipment, whereas the ones, which in principle are “push-button” methods, require special instruments. Second, no method has truly sufficient enough precision. From this standpoint, for research purposes a method of measurement is appropriate if it is unbiased, even if it lacks high precision. Accordingly, although the findings in an individual infant may be imprecise (i.e., uncertain), it is still possible to achieve meaningful and statistically significant results by analysis of the findings from groups of infants. However, for clinical use, it is absolutely necessary that a single measurement is sufficiently precise. Third, research on organ blood flow in neonates has focused on physiology and pathophysiology and done little in the way of defining the clinical benefit of having measures of blood flow in ill infants. This means that there is minimal incentive for the clinician to overcome the difficulties presented earlier.
Doppler ultrasound to assess changes in CBF was first used in neonates in 1979. Because clinical and research interest focused on CBF during this time, several methods, including Doppler ultrasound, were introduced to assess blood flow to the brain of the neonate. The use of Doppler ultrasound for functional echocardiography in the neonate is described in Chapter 10 in detail.
According to the Doppler principle, the frequency shift of the reflected sound (the “echo”) is proportional to the velocity of the reflector. Because erythrocytes in blood reflect ultrasound, blood flow velocity can be measured based on simple physics. The equation states that the frequency shift equals the flow velocity multiplied by the emitted frequency divided by the speed of sound in the tissue. However, there are several factors that need to be taken into consideration and corrected for with the use of the Doppler principle. First, the apparent velocity has to be corrected for the angle between the blood vessel and the ultrasound beam. Second, it should be kept in mind that multiple frequencies are detected when performing an ultrasound study of a vessel, because the flow velocity decreases from the center of the bloodstream toward the vessel wall. In addition, even the vessel wall itself contributes to the signal. Finally, the velocity is pulsating in nature because it is faster in systole than in diastole ( Fig. 16.1 ).
The instruments of the 1970s and early 1980s were continuous wave (with no resolution of depth, with no image, and a crude mean frequency shift estimator). Finding an arterial signal was done blindly, using general anatomic knowledge and an audible signal with the frequency shift in the 50 to 500 Hz range while searching for the loudest pulsating signal with the highest pitch. This left the angle and the true spatial average undetermined, and therefore the scale of measurement is uncertain. Indices of pulsatility (resistance index = [(peak systolic flow velocity − end diastolic flow velocity)/end diastolic flow velocity] and pulsatility index = [(peak systolic flow velocity − end diastolic flow velocity)/mean flow velocity]) are used since these indices are independent of the angle of insonation.
Indices of Pulsatility
Indices of pulsatility reflect downstream resistance to flow. The pulsatility in the umbilical artery has achieved great clinical importance in fetal monitoring. However, in newborn infants the resistance index in the anterior cerebral artery was only weakly associated with CBF as measured by 133 Xe clearance. In addition, more sophisticated modeling reveals that arterial blood pressure pulsatility and arterial wall compliance are just as important determinants of the indices of pulsatility as is the downstream resistance. In summary, Doppler data on resistance indices may be biased. However, in a seminal clinical study the pulsatility index was shown to carry independent prognostic ability in term infants with neonatal hypoxic-ischemic encephalopathy. With increased computational capacity, it has become possible to make full two-dimensional Doppler images at full spatial resolution and a sufficient time resolution to track the cardiac cycle—called fast Doppler. This allows measurement of indices of resistance in all vessels in the field and thereby will allow the study of functional activation and localized pathology. Central hemodynamics will not bias local differences in pulsatility, but differences in local artery wall compliance may still be mistaken for local vasodilation or vasoconstriction.
Blood Flow Velocity
Since the 1980s, duplex scanning combining imaging and Doppler, range-gating limiting flow detection to a small sample volume, and frequency analysis allowing proper estimation of maximum and mean frequency shift has been possible and all contributed to more reliable measurement of blood flow velocity. However, if arterial diameter is not measured, it is not straightforward to compare one infant with another, one organ with another, and even one state with another in the same infant because arterial diameter varies dynamically in the immature individual.
Absolute organ blood flow in mL/min equals flow velocity (cm/s) multiplied by arterial cross-sectional area (cm 2 ). To measure left and right ventricular output, the diameters of the ascending aorta and the pulmonary trunk, respectively, need to be precisely determined. The diameter of these two major vessels is 6 to 10 mm, and an error of 0.5 mm on the first generation of duplex scanners translated to a reproducibility of 10% to 15%. Recently, for measurement of superior vena cava (SVC) flow at the vessel’s entry into the right atrium with a diameter of 3 to 6 mm, a reproducibility of SVC flow of 14% was reported using a 7-MHz transducer. With the advent of color-coded imaging and the use of higher ultrasound frequencies, volumetric measurement of distributary arteries has also become possible in the newborn. For instance, measurement of blood flow in the right common carotid artery with a diameter of 2 to 3 mm was reported to have a reproducibility of 10% to 15% using a 15-MHz transducer, and that in both internal carotid and both vertebral arteries with diameters of 1 to 2 mm was found to be 7% for the sum of the blood flows in the four arteries using a 10-MHz transducer and the mean value 14 mL/100 g/min for infants born at 32 to 33 weeks’ gestation and 19 mL/100 g/min for infants born at term.
Diffuse Optical Methods Using Near-Infrared Light
NIR spectroscopy (NIRS) is also discussed in Chapter 17 , Chapter 18 . The first clinical research use of this technology was carried out in newborns. Quantitative spectroscopy (NIRS) was subsequently performed in 1986. After 40 years, more than 300 published papers on diffuse optics with NIR in newborns have been published combining a variety of methodology with a range of physiologic and pathophysiologic questions. The great advantages are that it is noninvasive, usually inherently safe, and may potentially be bedside, continuous, and quantitative.
The newborn infant’s head is ideally suited for study with NIR. The overlying tissues are relatively thin, which ensures that the signal is dominated by brain tissue, including both the white and gray matter. NIR can be performed with the light applied to one side of the head and received on the other side (transmission mode) in the low birth weight infant with biparietal diameters from 6 to 8 cm. For every centimeter of source-detector (optode) separation, the intensity of the received light is reduced by a factor of 50 to 100. In transmission mode the results may be interpreted as “global.” Larger babies can be investigated only with the emitting and receiving fibers in an angular arrangement (reflection mode), usually with both optodes on the same side of the head. In this situation a smaller volume of brain tissue between the optodes is investigated. This may be chosen on purpose, also in smaller babies, to obtain “regional” results. When the source-detector separation is less than 2 cm, the extracerebral tissues are more influential.
Algorithms and Wavelengths for Spectroscopy
The purpose of spectroscopy is to measure the concentrations of the various chromophores, light-absorbing molecules, particularly hemoglobin. The number of wavelengths used has varied from two to six. With the use of different wavelengths, the mathematical algorithms used to measure oxyhemoglobin (O 2 Hb), deoxyhemoglobin (HHb), and the cytochrome aa3 oxidase difference signal have differed. This makes direct comparison of results difficult.
The pathlength of light traversing the tissue must be known to calculate concentrations (i.e., to measure quantitatively). The pathlength in tissue exceeds the geometric distance between the optodes by a factor of 3 to 6, and this factor is named the differential pathlength factor.
Measurement of Cerebral Blood Flow Using Oxyhemoglobin as Tracer
Measurement of blood flow in absolute terms is possible by NIRS and is based on the Fick principle and uses a rapid change in arterial O 2 Hb as the intravascular tracer. By using the change in the oxygenation index (ΔOI = (Δ[O 2 Hb] − Δ[HHb])/2) observed after a small sudden change in the arterial concentration of oxygen, CBF (in mL/100 g/min) can be calculated as CBF = ΔOI/( k × ∫SaO 2 × dt), where ΔOI is measured in units of μmol/L, and k = Hgb × 1.05 × 100. Hgb is blood hemoglobin in mmol/L (tetraheme), SaO 2 is given in percent, and t is time in minutes ( Fig. 16.2 ).
During the measurement of CBF, cerebral blood volume (CBV) and oxygen extraction must be constant and the period of measurement must be less than the cerebral transit time (approximately 10 seconds). This method of CBF measurement also has significant practical limitations. For instance, in infants with severe lung disease, arterial oxygen saturation (SaO 2 ) may be fixed at a low level despite administration of oxygen, whereas in infants with normal lungs, SaO 2 is near 100% in room air. Finally, manipulation of SaO 2 may not always be safe. CBF measured this way has a reported reproducibility of 17% to 24% and has been validated against 133 Xe clearance in sick newborns. These comparisons constitute important direct external validation of NIRS in the brain of human neonates.
Indocyanine Green as an Alternative Tracer
In the past a dye, indocyanine green (ICG), given by intravenous injection, has been used in place of oxygen. Because this requires measurement of the arterial ICG concentration during the injection, an alternative, a blood flow index, can be calculated as the rise time in cerebral ICG concentration after a rapid intravenous injection. This has a satisfactory coefficient of variability of 10% and has been used to study cerebral autoregulation directly during epinephrine injection.
Calculating Cerebral Blood Flow from Cerebral Blood Volume
CBV may be estimated directly by NIRS from quantitation of the total hemoglobin concentration, and changes in CBV can be used as a surrogate measure of changes in CBF, using a conversion factor (the Grubb coefficient). The appropriateness of this has been substantiated in newborns by comparing the reactions to changes in arterial CO 2 tension and during functional activation.
Diffuse Correlation Spectroscopy
NIR equipment based on lasers can emit highly coherent light waves. When reflected from stationary reflectors in tissue, the coherence tends to decay slowly over 1 to 100 μs, whereas when the light is reflected from moving particles (in tissue this means red blood cells) the coherence decays more rapidly. The autocorrelation as a function of time therefore can be used as the blood flow index . The blood flow index has been correlated to Doppler ultrasound in preterm infants and to arterial spin labeling by magnetic resonance imaging (MRI) in neonates with congenital heart disease.
Cerebral Oxygenation as a Surrogate for Blood Flow
The metabolic rate for oxygen varies little, except during seizures or due to heavy sedation/anesthesia, and arterial oxygenation is almost always known and, if needed, kept within rather narrow limits. In such circumstances, it is reasonable to see cerebral oxygenation primarily as an indicator of blood flow, or more appropriately, of oxygen delivery (i.e., the product of blood flow, arterial oxygen saturation, and blood hemoglobin concentration).
Cerebral oxygenation represents the mean oxygen saturation of the hemoglobin in all types of blood vessels in the tissue. In terms of interpretation, it is not known how much of the signal originates from blood in the arteries, capillaries, or veins. Data in piglets suggest that the arterial-to-venous ratio is approximately 1:2 but may vary and that the arterial fraction may increase during hypoxemia and hypovolemic hypotension.
Three different principles are being used to measure hemoglobin-oxygen saturation in absolute terms. With spatially resolved spectroscopy (SRS), the detection of the transmitted light at two or more different distances from the light-emitting optode allows monitoring of the absolute concentration of oxyhemoglobin (O 2 Hb) as a proportion of the total hemoglobin concentration (tHb), that is, the hemoglobin saturation. This quantity is called (regional or cerebral) tissue oxygenation, (StO 2 ), or tissue oxygenation index (TOI). The measurement depends on the tissue being optically homogeneous, which is unlikely to be the case. Nevertheless, TOI values close to cerebrovenous values have been found, and appropriate changes in TOI have been documented with changes in arterial oxygen saturation and arterial partial pressure of carbon dioxide (P co 2 ); SRS is simple and relatively cheap, and the devices are simple to handle and approved for clinical use. The other two principles are time-resolved spectroscopy, which detects the time of flight of very short light pulses, and phase shift spectroscopy, which detects the phase shift and phase modulation of a continuous frequency-modulated source of light. These are more technically demanding, less commonly used, but able to determine the absorption and scattering of light separately and therefore likely to be more reliable.
Bias of Tissue Oxygen Saturation
Tissue oxygenation cannot be compared directly with any other measurement because it represents the findings in a mixture of blood in the arteries, capillaries, and veins. Interestingly, though, StO 2 has been validated on the head of young infants with heart disease during cardiac catheterization. In this study, across an StO 2 range of 40% to 80%, the mean value was almost identical to oxygen saturation in jugular venous blood as measured by co-oximetry. This suggests a significant negative bias as, in addition to venous blood, the StO 2 also represents arterial and capillary blood. This bias is likely to differ between different types of instruments.
Precision of the Tissue Oxygen Saturation
Replacing the sensor of SRS instruments multiple times in newborns and young infants, the limits of agreement after optode replacement were found to be at least −15% to +16%. For comparison, arterial hemoglobin oxygen saturation can be measured by pulse oximetry with limits of agreement of ±6% (two times absolute root mean [squared]).
Importance of the Low Precision of the Tissue Oxygen Saturation
Cerebrovenous saturation is tightly regulated, with the normal values between 60% and 70%. During hypoxemia, or cerebral ischemia, when CBF decreases without a change in cerebral oxygen consumption, cerebrovenous saturation falls. However, a 30% decrease in the CBF will lead to only a drop from 70% to 60% in the cerebrovenous oxygen saturation—and hence less in tissue oxygenation—which is well within the limits of the error of measurement.
Covariance of Cerebral Oxygenation and Arterial Blood Pressure as an Indication of Cerebral Autoregulation
During long-term cerebral monitoring of cerebral oxygenation as a surrogate for blood flow, the covariance with arterial blood pressure may be used as a measure of cerebral autoregulation. First, only changes lasting 10 seconds or more will reflect the “static” autoregulation (i.e., the full capacity of the cerebral vasculature to buffer changes in perfusion pressure) (see Chapter 2 ). Simple time-domain analysis may be better than more sophisticated frequency domain analysis, although solving the problem of directionality of the covariance may help. However, hours of data are typically required to obtain a reliable estimate, and it is difficult to avoid the bias that high variability in the blood pressure will give a better signal-to-noise ratio.