Cerebral Blood Flow of the Neonatal Brain after Hypoxic-Ischemic Injury

Abstract

Objective

Hypoxic-ischemic encephalopathy (HIE) in infants can have long-term adverse neurodevelopmental effects and markedly reduce quality of life. Both the initial hypoperfusion and the subsequent rapid reperfusion can cause deleterious effects in brain tissue. Cerebral blood flow (CBF) assessment in newborns with HIE can help detect abnormalities in brain perfusion to guide therapy and prognosticate patient outcomes.

Study Design

The review will provide an overview of the pathophysiological implications of CBF derangements in neonatal HIE, current and emerging techniques for CBF quantification, and the potential to utilize CBF as a physiologic target in managing neonates with acute HIE.

Conclusion

The alterations of CBF in infants during hypoxia-ischemia have been studied by using different neuroimaging techniques, including nitrous oxide and xenon clearance, transcranial Doppler ultrasonography, contrast-enhanced ultrasound, arterial spin labeling MRI, 18F-FDG positron emission tomography, near-infrared spectroscopy (NIRS), functional NIRS, and diffuse correlation spectroscopy. Consensus is lacking regarding the clinical significance of CBF estimations detected by these different modalities. Heterogeneity in the imaging modality used, regional versus global estimations of CBF, time for the scan, and variables impacting brain perfusion and cohort clinical characteristics should be considered when translating the findings described in the literature to routine practice and implementation of therapeutic interventions.

Brain damage from hypoxia-ischemia in the perinatal period occurs in approximately 1 to 8 out of every 1,000 births in developed countries and is associated with death and long-standing disabilities worldwide.^1-4 Therapeutic hypothermia (TH) commenced within 6 hours of birth is now the standard of care for infants with moderate-to-severe encephalopathy from HIE.^5-8 While this intervention has been shown to reduce overall mortality and disability in infants (up to 22 months), rates of moderate and severe disability were similar regardless of receiving TH when compared with control patients at older ages.⁹ Sufficient delivery of oxygen and other energy substrates, as well as the removal of waste products, are required to preserve aerobic metabolism within the brain.^10,11 Cerebral blood flow (CBF) is a determining factor in brain perfusion status and outcomes have been attributed to variations in brain perfusion during the acute and reperfusion phases after injury.¹²

CBF is quantified in terms of volume per unit mass per unit time in a specific region within the brain (mL/100 g per min). Systemic and intracranial pressures, as well as blood viscosity, are important determinants of CBF. In the healthy brain, CBF is greater in cortical areas than in white matter.¹⁰ Alterations of CBF in infants after hypoxia-ischemia have been studied by using multiple different neuroimaging techniques.¹³ Consensus is lacking regarding the clinical significance of CBF estimations in infants with hypoxic brain injury in terms of management or prognosis. In this review, we discuss the challenges of estimating brain perfusion in infants who are healthy and those who have sustained a hypoxic-ischemic insult, including the benefits and limitations of available imaging modalities, and the impact of CBF on disease management and prognostication.

Quantifying Cerebral Blood Flow in Healthy Infants

CBF in healthy infants is approximately one-third of CBF in healthy adults. Values in infants can range between 10 and 20mL/100 g/min.¹⁴ This is in part related to the lower level of neuronal activity of the infant’s brain compared with older children and adults. CBF has regional specificity and is higher in the frontal lobe compared with other brain regions in neonates.¹⁵ The infant’s brain undergoes exponential growth of lobe volume, synaptogenesis, and myelination, associated with a progressive increase in CBF within the first days of postnatal life. CBF continues to increase with age until it peaks around 3 to 8 years of age at which time it decreases steadily through adolescence, plateauing in adulthood.^15-20 This evolution of flow has been demonstrated with Doppler blood flow velocity measurement, with CBF velocity increasing over the first 24 hours of life,²¹ although interobserver variability could limit these measurements.²¹

Determination of normal CBF in healthy infants represents a scientific and logistical challenge. Some CBF measurements from “healthy infants” have been performed in hospitalized infants with conditions such as preterm birth, transient neonatal tachypnea, feeding problems, hypoglycemia, or maternal gestational diabetes mellitus.^22-24 Variability in equipment used to collect the values are additional obstacles.

Evaluation of the minimum CBF required to maintain normal brain function or limit long-term neurologic sequelae poses further challenges. While adult normal neurological function may be preserved even above CBF values of 19mL/100 g/min in some cases, irreversible damage is commonly seen under 15mL/100 g/min after a period of time, with lower CBF values causing infarcts in shorter amounts of time. In infants, normal neurological examinations have been reported with CBF values below 10mL/100 g/min.²⁵ Cortical electrical activity has been shown in preterm infants during the first 48 hours of life at CBF values as low as 4.5 mL/ 100 g/min.²⁶ It has been hypothesized that external conditions, including some medications, can impair CBF or modulate CBF-related brain injury. For example, decreased metabolic rate induced by general anesthesia may also lead to neurotoxicity and decreased CBF from systemic changes in blood pressure and chemical homeostasis, as seen in animal and human studies, but this has not been completely elucidated in healthy children.^27-30

Brain Perfusion Estimations in Neonatal Hypoxic-Ischemic Encephalopathy

A newborn’s developing brain is more vulnerable to abrupt variations of CBF, which makes the newborn brain more susceptible to vascular ischemic insults.²⁷ As proposed by Bell et al, the neurovascular unit represents a synergic structure comprised of blood vessels and the nervous tissue that maintains CBF at optimal levels, and disruption of the neurovascular unit can lead to brain perfusion abnormalities and blood–brain barrier (BBB) dysfunction that can cause hypoxic-ischemic injury (HII),^31,32 potentially resulting in hypoxic-ischemic encephalopathy (HIE). Acute hypoxia-related findings that might precede HIE include basement membrane disruption, reduced BBB integrity, altered tight junction protein expression, and astrocyte hypertrophy/hyperplasia.³¹ The most important cause of HIE in newborns is perinatal asphyxia from intrauterine or postnatal events. Intrauterine insults include impaired placental perfusion and gaseous exchange due to maternal hypertensive/prothrombotic states, massive hemorrhage, cord prolapse, tight nuchal cord, or fetal intrinsic factors. Postnatal condition includes congenital heart disease or pulmonary failure.³³

Perinatal asphyxia is characterized by an initial decrease in CBF followed by relative hyperperfusion that can cause injuries mediated by the release of inflammatory molecules.¹⁵ Rapid reperfusion causes deleterious effects in brain tissue also known as reperfusion injury.³⁴ Persistently high CBF in the subacute phase after neonatal HIE is associated with poor clinical outcomes. High CBF can be seen both globally and in specific brain regions and at various times after injury. This association between hyperemia and unfavorable clinical outcome is also observed in children and adults.^35-38

General clinical findings in perinatal asphyxia include short- and/or long-term manifestations, such as seizures and spastic cerebral palsy, respectively, although other manifestations are possible depending on the location and extent of the ischemic insults.⁴ Classical magnetic resonance imaging (MRI) patterns of HIE in infants have been described extensively in the literature.^33,39,40 Early changes in blood flow velocity can be detected before the development of observable hypoxic-ischemic changes on computed tomography (CT) and MRI.⁴¹ In light of this, timely detection of abnormal brain perfusion, obtained through CBF measurement, could direct treatment to improve outcomes in HIE. Table shows a comparison of studies of brain perfusion in HIE population (►Table 1).

Table 1.

Reported cerebral blood flow values in healthy/relatively healthy infants (without major cerebral lesions or congenital heart diseases), infants with hypoxic-ischemic encephalopathy, and infants with multiple pathologies including or not hypoxic-ischemic injury (mixed group)

Study	Year	Number of subjects	Type of population	Mean GA in weeks±SD (range)	Mean time (DOL) of scan ±SD (range)	Technique	Brain area evaluated	Mean CBF (mL/100 g/min) value±SD (range)a
Edwards et al104	1988	9	Mixed	(26–44)	1–10	NIRS	Global	18 (7–33)
Altman et al25	1988	16 preterms 14 terms	Mixed	Preterms (26–36)	Preterms (2–39) Terms (3–39)	PET	Global	Preterms (4.9–23) Terms (9–73)
Frewen et al43	1991	9	HIE	38.3±2.6	(1–3) (every 24 h)	No solubility	Global	UO: 83.2 ±44.9 FO: 25.6 ±8.2
Chiron et al104	1992	7	Healthy/relatively healthy		(2–45)	SPECT	Global	50 ±3.4
Meek et al16	1998	11	Healthy/relatively healthy	26 (24–31)	(1–3)	NIRS	Global	(6.3–15.2)
Kehrer et al21	2002	67	Healthy/relatively healthy	36.3 (24.2–41.6)	9 (1–83)	TcD	Global	20
Kusaka et al23	2005	17	Mixed	32.9±4.3	18±26	NIRS	Parieto-temporal region	15.3±4.2
Massaro et al136	2013	36	Healthy/relatively healthy (18) HIE (18)	Healthy: 39.5 ±1.5 HIE: 38.8±1.5	7–10	ASL MRI	Global	Healthy: 19±2 HIE: 24 ±5
Iwata et al22	2014	32	Healthy/relatively healthy	38.3±2.6	21±17	NIRS	Global	12.9±1.5
Wintermark et al116	2014	7	HIE (3 moderate, 4 severe)	≥36	(1–2)	ASL MRI	Frontal	Moderate: 37.42 ±1.11 Severe: 24.07±2.49
De Vis et al76	2014	9	Mixed	38 (34–41)	5 (2–7)	ASL MRI	Global	12 ±4
Boudes78	2014	19	HIE	≥36	(1–3)	ASL MRI (PASL and pCASL)	Global	PASL: 27.48±8.21 pCASL: 24.65±8.82
De Vis et al70	2015	28	HIE	39.3 (36.4–42.0)	(1–2)	ASL MRI	Basal ganglia, thalami and frontal cortex	UO: 63 (28–108) FO: 28 (12–51)
Proisy et al71	2019	28	HIE	39.7 (36.7–41.9)	4	ASL MRI	Global	23.8 (14.6–48.3)

Abbreviations: ASL, arterial spin labeling; CBF, cerebral blood flow; FO, favorable outcome; GA, gestational age; MRI, magnetic resonance imaging; NIRS, near infrared spectroscopy; NO, nitrous oxide; PASL, single inversion-time pulsed ASL; PC, phase contrast; pCASL, pseudo-continuous ASL; PET, positron emission tomography; SPECT, single photon emission computed tomography; TcD, transcranial doppler; TGM, total gray matter; UO, unfavorable outcome; WM, white matter.

^aCBF measured as mL/100 g/min.

Current and Emerging Techniques for Measuring CBF in Neonates

CBF Estimations Using Nitrous Oxide, Xenon Clearance and SPECT

The first CBF measurements were made by using nitrous oxide in the 1940s. This method was highly invasive and was not practical in a clinical setting.²⁷ Further, CBF measurements were performed by using the isotope 133-Xenon (¹³³Xe)⁴² in the 1960s. ¹³³Xe has an advantage over nitrous oxide in that it does not interfere with brain metabolism²⁷: this noble gas crosses the BBB and is highly diffusible with fast clearance. ¹³³Xe was initially administered directly onto the carotid artery, providing CBF information about one cerebral hemisphere. Subsequently, inhalation and intravenous administration allowed global estimation of CBF. This technique involved exposure to ionizing radiation due to inhalation or intravenous injection of the radionuclide.¹⁴

One of the earliest studies on CBF in infants with HIE estimated CBF by using nitrous oxide clearance. In this study, CBF measurements were taken from a catheter in the jugular bulb in eight term infants. Global CBF was measured on days 1, 2, and 3 of life after HIE.⁴³ Unfavorable outcome after 1 year included neonatal demise, spastic quadriplegia, and severe global developmental delay among others; some patients were followed up to 2 years with similar findings. Greater global CBF during the first 3 days after HIE was observed in infants with unfavorable outcomes when compared with infants with favorable neurological outcome. CBF differences were independent of intracranial pressure, median arterial pressure, or presence of structural abnormalities.⁴³

Xenon-clearance measurements have been implemented in addition to CT imaging,⁴⁴ increasing the total dose of radiation exposure, but enhancing anatomic localization for accurate quantitative determination of regional blood flow. Single photo emission CT (SPECT) generates tomographic images of the tridimensional distribution of a radiotracer and can depict cerebral hemodynamics, including CBF values. Apart from ¹³³Xe, other retention tracers (technetium Tc99m hexamethylpropyleneamine oxime [⁹⁹mTc-HMPAO] or iodine 123 N-isopropyl-p-iodoamphetamine [¹²³I-IMP]) can also be used.⁴⁵ Notwithstanding, clinical utility of these techniques nowadays has been diminished due to the increasing use of noninvasive modalities.

Transcranial Doppler Ultrasonography: Macrovascular Flow Assessment

Transcranial Doppler (TCD) assesses changes in flow velocity of large caliber arteries, such as the middle cerebral artery (MCA), anterior cerebral artery (ACA) (►Fig. 1), internal carotid artery (ICA), or posterior cerebral artery (PCA), as well as venous structures. This technique has proven to be a useful tool in noninvasive monitoring of macrovascular flow. The U.S. probe emits pulsed Doppler ultrasonic beams, then received, and processed to detect vascular flow. In neonates, the open anterior fontanelle, as well as other acoustic windows including the posterior and mastoid fontanelles can be used for Doppler interrogation. Insonation depth will vary depending on which vascular segment is evaluated.⁴⁶

Fig. 1.

Transcranial Doppler ultrasonography. (A) Doppler spectral waveform of the anterior cerebral artery reveals absence of diastolic flow suggestive of high resistance in the vessel as seen in HIE. (B) Different patient with HIE who demonstrates preserved diastolic flow with resistive index within the normative range. HIE, hypoxic-ischemic encephalopathy.

Peak systolic velocity (PSV), mean flow velocity (MFV), and end-diastolic velocity (EDV) in neonates are different among arteries and are recorded based on the angle of insonation, which is defined as the angle formed between the evaluated vessel and the long axis of the probe. Peak velocity, for instance, is determined when the probe is placed in parallel (0 degrees angle of insonation) with the arterial lumen, though in practice the angle of insonation is usually between 30 and 60 degrees due to practical challenges of achieving the 0 degrees angle of insonation. On the other hand, the thalamostriate arteries from the MCA are visualized on angled sagittal views.⁴⁷ Fast Fourier transform (FFM) converts flow velocities to spectral images.⁴⁶ Nonetheless, the technique does not provide information about CBF at the microcirculatory level within the brain parenchyma.⁴⁸

Other indirect parameters can be calculated to help determine CBF status. The resistive index (RI) (normal>0.55) is obtained from the difference between the PSV and the EDV divided by the PSV. The pulsatility index (PI) (normal = 0.6–1.1) is calculated from the difference between PSV and EDV divided by the time-averaged velocity.^48-50 Measuring PI, for instance, allows minimizing the effect of vessel angulation and is well correlated with acute changes in perfusion pressures within the brain. Indices in general provide a more comparable value to be used when evaluating patients since normal velocities in neonates tend to be highly variable among studies.⁴⁷

Cerebral RI or PI is not significantly impaired in the context of mild hypoxemia. Contrarily, patients who have suffered from moderate-to-severe HIE may have extremes of RI (RI<0.50 or RI>0.90) and PI (PI<0.6 or >1.2),^48,50,51 potentially in the context of impaired cerebrovascular autoregulation.⁴⁷ Liu et al performed TCD in 40 term neonates with HIE within 24 hours after birth. Two patients with reversal in perfusion direction during diastolic phase (with end-diastolic flow velocity<0 and RI>1.0) had brain death despite timely management.⁵¹ Stark and Seibert followed 15 term neonates with history of perinatal asphyxia over 5 years. All subjects were evaluated with TCD and demonstrated high diastolic flow with a RI <0.6 on the first day of life. Two of them died and ten had severe neurodevelopmental delay when evaluated at 32 months of age. Interestingly, only half of the patients had abnormal sonographic imaging at the initial evaluation.⁵² Archer et al prospectively evaluated a larger cohort of patients with intrapartum asphyxia with TCD of the ACAs and compared with controls. TCD predicted poor neurological outcomes in cases with 86% of accuracy, 100% sensitivity, and 81% specificity.⁵³ Gray et al evaluated adverse outcomes (cerebral palsy, developmental delay, or death) in neonates with HIE, finding high specificity when abnormal RI or MFV (Vm) were obtained.⁵⁴ Other authors have investigated CBF in low-birth-weight preterms with HII who were at risk of developing periventricular leukomalacia and/or germinal matrix hemorrhage.⁵⁵ These alterations can occur in HIE as a result of loss of autoregulation, vasospasm, microangiopathy, and/or intracranial hypertension.

Portability and noninvasiveness are other advantages for using TCD in a vulnerable population such as neonates, especially in the pediatric intensive care unit. However, TCD is an operator-dependent tool necessitating education and training to improve reproducibility of results. Additionally, absolute TCD values should be interpreted in the context of multiple internal and external factors that can contribute to increased or decreased CBF (patient’s age, body temperature, cardiac output, hematocrit values, carbon dioxide partial pressure, and drugs that alter luminal arterial ratio).Given the above, this tool can help evaluate patients over time for serial monitoring rather than comparing measurements with standardized values in healthy individuals, which would be the ideal scenario.

Contrast-Enhanced Ultrasound: Microvascular Flow Assessment

Contrast-enhanced ultrasound (CEUS) is clinically used to assess tissue perfusion by using an intravascular ultrasound contrast agent called microbubbles, which are made of biologically inert gases encapsulated with a phospholipid monolayer. Due to its size of 2 to 3 microns, smaller than red blood cells, they can better characterize tissue perfusion at the microvascular level by increasing the echogenicity and enhancing the visualization of blood vessels, which allows quantification of flow dynamics.⁵⁶ Microbubbles flow into regions of interest, whose intensity changes according to wash-in and wash-out phases of the time-intensity curve (TIC). Multiple standardize perfusion metrics (peak enhancement, time-to-peak, rise time, wash-in slope on perfusion, wash-in area under the curve, perfusion index, etc.) can be obtained and interpreted from the TIC. CEUS for the evaluation of the brain is still an off-label indication in the United States.⁵⁷

Brain injury can be indirectly inferred on CEUS due to the associated perfusion abnormalities. Perfusion abnormalities of HIE include focal, multifocal asymmetric, and symmetrical patterns⁵⁷ (►Fig. 2). Spatiotemporal evolution of tissue perfusion after brain injury can be qualitatively and quantitatively assessed.^57-60 Moreover, when there are focal and/or multifocal perfusion lesions, the contralateral nonaffected regions can be used as controls.

Fig. 2.

Ultrasound perfusion maps and time-intensity curves. Contrast-enhanced US images of a mid-coronal slice through the brain at peak enhancement (column labeled CEUS), corresponding US perfusion maps of the wash-in slope or perfusion (column labeled PER), US perfusion maps of peak enhancement (column labeled PE), US perfusion maps of the wash-inAUC (column-labeled Wi-AUC), time intensity curves or wash-in curves (column labeled time-intensity curve), and central gray nuclei-to-cortex perfusion ratio over time (column labeled gray nuclei to cortex). The first three rows (A–C) denote the normal group, and the subsequent rows (D–H) denote the affected group. Note that the red line in the time-intensity curve refers to microbubble intensity in the central gray nuclei, and the yellow line refers to that of the cortex. Note also that the red areas the US perfusion maps denote increased flow or hyperperfusion, and blue areas denote decreased flow or hypoperfusion. In controls, the red line lies above the yellow line, denoting a central gray nuclei-to-cortex intensity ratio of greater than 1 during wash-in and immediate postwash-in time points. In the affected group (D–H), the central gray nuclei-to-cortex intensity ratio is 1 or less. Patient D was a neonate after prolonged cardiac arrest. There was near-absent perfusion to the brain, which can be seen by the paucity of perfusion on the original CEUS image and corresponding perfusion maps created. Patient E was a neonate after ECMO and multiple cardiac arrests. Note the mildly dilated frontal horns and avid perfusion to the cortex, to a greater extent than the central gray nuclei. Patient F was a neonate who presented with seizures and was found to have multifocal injury. The imaged brain is a coronal slice through the parieto-occipital lobes. Heterogeneous paucity of microbubbles is seen due to multifocal injury. For this patient in particular, the central gray nuclei were not imaged in this slice, which led to an absence of a red line on the corresponding time-intensity curve. Patient G was a neonate after ECMO who showed avid hyperperfusion to the cortex and central gray nuclei to a similar extent, which can be seen by the similar time-intensity curves. Patient H was a neonate with diffuse white matter injury who showed global hyperperfusion in the immediate postinjury setting. Note the slightly more avid perfusion to the cortex than the central gray nuclei on the time-intensity curve. The last column labeled GNC shows graphs of the gray nuclei-to-cortex ratio of perfusion over time in the normal (A–C) and affected (D–H) groups. Patient F was excluded from the gray nuclei-to-cortex ratio analysis because of a lackof perfusion information in the central gray nuclei. Note that in the affected group, the plotted data points lie either at or below the red line, which denotes a central gray nuclei-to-cortex microbubble intensity ratio of 1. This contrasts with the normal group, in which all data points immediately after wash-in lie above the red line. This visually apparent trend can distinguish the affected from the normal group; au indicates arbitrary unit. Images and legends reprinted with permission from Hwang M, Sridharan A, Darge K, et al. Novel quantitative contrast-enhanced ultrasound detection of hypoxic ischemic injury in neonates and infants: pilot study. J Ultrasound Med 2019;38(8):2025–2038. AUC, area under the curve; CEUS, contrast-enhanced ultrasound; ECMO, extracorporeal membrane oxygenation; US, ultrasonography.

Hwang et al showed that the quantitative detection of HIE using CEUS is possible and demonstrated that the normally high central gray nuclei to cortex (GNC) perfusion ratio is altered in HIE, configuring central or peripheral patterns. Normal children would have a GNC equal to or slightly higher than 1. Values significantly greater than 1 can be seen in HIE with a central pattern. On the other hand, a GNC ratio lower than 1 may demonstrate peripheral pattern injury. This technology may be optimally suited to serve as a quantitative marker to screen infants for potential HIE.^56,57,59 In addition, the tool can offer important insights into the reperfusion response following HIE, which can affect the long-term neurologic sequelae and therefore is important to monitor serially.

The exam can be performed readily at the bedside and be done within minutes, allowing for the assessment of CBF in infants in a timely manner. In addition, no sedation or ionizing radiation is used. Moreover, the risk of adverse events of contrast agent used in CEUS is the lowest (0.0086%) when compared with other imaging modalities that require contrast enhancement. Further work is warranted to explore the full potential of brain CEUS as a diagnostic and prognostic modality in neonatal HIE evaluation. Safety measures such as maximum mechanical index and injection dosage should be considered.⁵⁶

Arterial Spin Labeling: Regional Cerebral Blood Flow Estimations

MRI provides high resolution of brain anatomy and is the most sensitive and specific technique for characterizing brain injury patterns in preterm neonates and children with HIE.^61-67 Diffusion-weighted imaging (DWI) and quantitative estimation of brain perfusion via perfusion-weighted imaging (PWI) provide additional information for detecting and characterizing the burden of HI injury. DWI measures the random Brownian motion of water molecules for detection of injury, and currently, the gold standard diagnostic method for diagnosis most acute brain ischemic injuries. MRI-based PWI can be performed either via gadolinium contrast injection or by nongadolinium-based methods. Dynamic susceptibility contrast PWI uses a gadolinium bolus contrast to measure cerebral blood volume and subsequently calculate relative CBF using a variety of mathematical models,^68,69 and it can be a useful adjunct tool to DWI for serial monitoring of the reperfusion phase.^61,70,71 PWI may remain abnormal for more than a week after DWI normalizes and may be more predictive of neurodevelopmental prognosis in infants with HIE.^69,72,73

In ASL imaging, however, no gadolinium injection is used. Protons in blood water flowing into the head are magnetically labeled, and the signal generated by these endogenous labels is detected in the brain, typically via subtracting MRI signals during alternating labeled and unlabeled acquisitions. These subtracted images demonstrate signal intensity proportional to the amount of perfusion present in each voxel and a quantitative measurement of cerebral blood flow can be calculated^74-76 (►Fig. 2). There are three common types of ASL: continuous, pulsed, and pseudocontinuous. Continuous ASL uses a prolonged radiofrequency pulse to label protons, which provides greater signal, but also deposits more energy in tissues and causes tissue heating. Pulsed ASL uses a short radiofrequency pulse at a single point in time. This is less technically demanding, deposits less energy, and is useful for the pediatric and neonatal populations since they have greater water content in blood and in brain parenchyma.⁷⁷ Pseudocontinuous ASL is a hybrid technique between continuous and pulsed, deploying intermittent pulses that balance signal-to-noise ratio, efficiency, and tissue energy deposition.^78,79 Finally, phase-contrast (PC) MRI is a complementary technique to ASL that may help obtaining quantitative estimation of global CBF.⁸⁰ A description of ASL techniques, requirements for absolute quantification, and possible sources of error can be found in a recent review.^77,78

ASL can be a useful tool to assess the spatiotemporal evolution of CBF in HIE (►Fig. 3). Wintermark et al used pulsed arterial spin labeling (ASL) MR imaging to assess regional CBF in 18 subjects with HIE, 11 of whom underwent TH. They demonstrated initial hypoperfusion on day of life 1, followed by hyperperfusion at days 2 to 3 in brain areas that subsequently showed injury on serial MRI scans. Brain injury was independent of TH.⁶¹ De Vis et al estimated CBF with pulsed ASL in the basal ganglia, thalami, and frontal cortex of 28 infants 2 to 7 days after HIE, most of whom underwent TH. They showed an association between higher CBF in basal ganglia and thalami and worse neurodevelopmental outcome compared with infants with favorable outcome.⁷⁰ 3D pseudocontinuos ASL (3D-pcASL) combined with postlabeling delay was used to evaluate 240 neonates with HIE and 200 healthy controls; all subjects were divided in groups according to age. CBF values were lower in HIE neonates younger than 1-day and aged 3 to 7 days when compared with controls. On the other hand, CBF was higher in the group of HIE neonates aged 1 to 3 days. Consequently, CBF values must be interpreted properly in correspondence with patient’s age.⁸¹

Fig. 3.

MRI arterial spin labeling images of infants both at 37 weeks of gestational age. (A) MRI arterial spin labeling perfusion in healthy infant. (B) Regions of high perfusion in hypoxic-ischemic injury involves the basal ganglia and left thalamus, and left parietal lobe. MRI, magnetic resonance imaging.

The advantages of ASL technique include noninvasiveness and direct quantification of regional CBF. Challenges of implementing this technique in infants include long acquisition times and sensitivity to motion artifact,^61,63-67,82 although technical advancements are underway to expedite the scan acquisition.⁸³ It is important to note that premature brains have a very slow blood velocity, which might be a limitation for ASL in this population.²⁴ Other challenges exist in neonatal ASL evaluation and reliable absolute quantitation, including differences in hematocrit and hemoglobin, which can alter the T1 of blood, transit time differences, labeling efficiency, and blood–brain partition coefficient. Sedative medications can result in underestimation of brain perfusion since many drugs alter brain perfusion.^84,85 The feed and bundle method can be used to avoid sedation in the neonatal population and decrease motion-related ASL artifacts. Moreover, portable MRI machines are in development, and the potential integration of ASL may be possible in the future.

Additional Factors Impacting Brain Perfusion: Brain Metabolism and Oxygenation

Brain Metabolism: ¹⁸F-FDG Positron Emission Tomography

Molecular imaging has emerged as a paramount subset in medical imaging, since it allows to visualize and measure metabolic processes at cellular and subcellular levels.⁸⁶ This technique help assess metabolic thresholds critical for brain function, depict in-vivo neurobiological processes, and determine neurochemical derangements. PET technology is based on the principle of radioactive signal detection of labeled positron-emitting isotopes/tracers that are intravenously injected or inhaled prior to scanning the patient.⁸⁷ Radioisotope nucleus emits a positron, which collides with an electron in the tissue, converting mass to energy in the form of photons. Then the PET camera utilizes scintillation crystals for detecting the photons. Finally, the crystals absorb the photons which are converted into electrical signals, whose data are extracted and processed using mathematic models.⁸⁸ The radiotracers ¹¹C-CO, ¹³N-NH3, and ¹⁵O-H₂O can be used to provide insight into CBF and oxygen consumption, while 18F-fluorodeoxyglucose (FDG) can assess metabolic activity by measuring glucose consumption rate.⁸⁹ Indeed, ¹⁵O-H₂O is contemplated as the reference standard for CBF assessment with the ability to freely passing through the BBB.⁹⁰ However, the other tracers such as ¹⁸F-misonidazole selectively stain hypoxic tissue and may provide further information as a marker for HIE in the future.⁹¹ It is important to note that CBF values obtained with PET can be measured as mL/cm³/min.⁸⁹

FDG can estimate brain injury and brain development in neonates, confirming that the neonatal brain utilizes glucose based on phylogenetic orders.⁹² Shi et al evaluated 36 preterm and term infants who were born at different gestational ages (GA), as well as examining 24 neonates with HIE who were divided in three groups based on clinical severity. Cerebral blood glucose metabolism measured by regional standardized uptake values was overall higher in terms than in preterms, except in the temporal region. Moreover, FDG uptake was lower in HIE subjects, especially in the subcortical cerebral white matter, the thalamus, and the basal ganglia. HIE severe cases demonstrated a significantly lower FDG uptake in major brain areas when compared with mild and moderate cases⁹³ (►Fig. 4). This help elucidate that in healthy neonates, deep cortical structures, thalamus, basal ganglia, and the sensorimotor cortex demonstrate high glucose uptake in term and preterm populations, while uptake is relatively low in the cerebral frontal, parietal, and occipital cortical regions.⁹⁴ This relatively high metabolic activity of deep cortical regions increases with GA. Lower glucose uptake in the temporo-occipital regions is correlated with HIE severity and poor neurodevelopmental outcomes.^95-97 Metabolic variables, such as glucose delivery and consumption, correlate with HIE severity and clinical outcome,^95,96 and these can be estimated on FDG-PET.⁹⁷

Fig. 4.

¹⁸F-fluorodeoxyglucose positron emission tomography image of infant at 33Z weeks of gestational age (above) and 18F-FDG PET in hypoxic-ischemic injury (below): relatively low in almost all regions of the brain. Images reprinted with permission from Shi Y, Zhao J-N, Liu L, et al. Changes of positron emission tomography in newborn infants at different gestational ages, and neonatal hypoxic-ischemic encephalopathy. Pediatr Neurol 2012;46(2):116–123.

Evaluation of cerebral perfusion with PET is due to oxygen ability to diffuse into tissue based on capillaries/tissue partial pressure of oxygen differences. Takahashi et al²⁰ determined normal CBF and metabolic values in the pediatric population using PET, by measuring regional CBF (rCBF), regional cerebral metabolic rate for oxygen (rCMRO₂), and regional oxygen extraction fraction (rOEF). They found that both rCBF and rCMRO₂ were lower in neonates when compared with older children and adults. PET has been used in older children with HIE.⁹⁸ Volpe et al first reported the usefulness of PET for measuring regional CBF in neonates with intraventricular and intraparenchymal hemorrhage. They found that little or no CBF could be detected in the areas of intracerebral hemorrhages, also affecting CBF within the ipsilateral hemisphere, which indicates that surrounding tissue could be affected by ischemia after an hemorrhagic event.⁹⁹ Within the same decade, values of CBF in several HIE preterm and term neonates were measured, demonstrating abnormalities in bilateral posterior parasagittal regions of the brain (also known as watershed regions). Necrosis of these regions was confirmed at postmortem examinations.^13,25

In light of this, PET is capable of quantifying brain metabolic activity and blood flow in children with HIE, so an approach contemplating both aspects is necessary. PET has disadvantages that limit its use in young patients: it exposes patients to estimated total dose of up to 0.57 mGy (57 Mrad) of radiation, is costly, has a lower spatial resolution compared with other techniques and requires patient transport.¹⁰⁰ The need for an on-site cyclotron and the radiotracers’ short half-life may also limit its use in most clinical setting around the world. Additionally, sedative medications may alter metabolic parameters that will ultimately affect proper measurements of the CBF. Nonetheless, functional information of the tissue at the global and regional levels represents a technical advantage to be further expanded.

Brain Oxygenation: Near-Infrared Spectroscopy

NIRS is a noninvasive technique that detects changes in the light absorption of oxy- and deoxyhemoglobin, allows continuous real-time bedside monitoring, and can be used as a surrogate for CBF measurement.¹⁰¹ For instance, regional cerebral oxygen saturation (rSO₂) can show if there are changes in oxygen supply and demand by reflecting the amount of hemoglobin-bound oxygen that remains in tissues after uptake has occurred,¹⁰² indirectly indicating cerebrovascular hemodynamics status. Continuous-wave NIRS (CW-NIRS), as opposed to frequency-domain NIRS (FD-NIRS), is the more common technique since most commercial manufactures have preferred its relative instrumental simplicity and data analysis straightforwardness. The advantages of FD-NIRS include its ability to measure tissue absolute optical properties and to determine absolute concentrations of oxy- and deoxyhemoglobin in brain tissue.¹⁰³ This modality was first used in nonhealthy neonates by Edwards et al, registering mean CBF values of 18mL/100 g/min.¹⁰⁴

The role of NIRS in HIE in term neonates has been extensively studied, conferring good prognostic accuracy and safety.^105-109 Toet et al determined rSO₂ and fractional cerebral tissue oxygen extraction (FTOE) within 48 hours after the insult in a cohort of 18 asphyxiated neonates that were followed for 5 years. Those infants who died or had adverse neurological outcomes were reported to have increased rSO₂ and decreased FTOE values after 1 day of the hypoxic-ischemic event.¹⁰⁵ Furthermore, Lemmers et al obtained similar results in a cohort of 39 neonates with HIE in regards to higher rSO₂ values for predicting long-term neurological impairment¹⁰⁷ As previously mentioned, TH incorporation to HIE workup has shown decreased mortality and disability rates, which is also demonstrated by reassuring CBF surrogate values obtained during cooling.

Tissue oxygenation index (TOI) is defined as the ratio of oxygenated to total tissue hemoglobin and, similar to rSO2, can measure brain tissue oxygenation¹¹⁰ Ancora et al evaluated 12 term neonates with moderate-to-severe HIE who underwent TH and concomitant NIRS monitoring; the group was followed for 1 year and adverse outcomes (death, cerebral palsy, low global quotient as per the Griffiths’ scale) were recorded. A higher TOI was present in the group with poor outcomes, anticipating a worse prognosis in this population.¹⁰⁶ In summary, NIRS is a good tool for predicting short- and long-term outcomes in term neonates with HIE.

Hypoxic-ischemic insults in preterm neonates have been also reported in the literature. A multisite European randomized clinical trial “safeguarding the brains of our smallest children” (SafeBoosC) was conducted in preterm neonates less than 28 weeks of GA and showed that infants with more time within the range 55 to 58% cerebral NIRS had more favorable clinical outcomes, including a reduction in severe brain injury, and all-cause mortality.¹¹¹ Importantly, dysfunctional cerebral autoregulation may play a role in decreasing tissue oxygenation and may account for decreased oxygenation and subjects with cardiopulmonary abnormalities.^112,113 When re-evaluated at 2 years of age, NIRS monitoring was not significantly associated with long-term neurodevelopmental outcomes.¹¹⁴ However, another study showed that rSO₂ values elevated greater than 85% on the first day of life were associated with adverse neurodevelopmental outcome at 2 years,¹¹⁵ and higher levels of regional oxygen saturation measured by NIRS in HIE neonates during TH may be associated with more subcortical injury on MRI and poor neurodevelopmental outcomes.¹⁰⁹

NIRS has an important advantage over some other modalities in that it can be used continuously at the bedside, is not expensive, and noninvasive. Additionally, this technique does not require either sedation or head rigid head immobilization since the optical fibers that serve as receptors are placed and attached in the neonate’s head as an elastic hat, similar to electroencephalographic devices.⁸⁷ However, it is limited in the size and depth of the area sampled, and lacks anatomic information. The technique is also prone to motion degradation as other imaging techniques. NIRS is perhaps more useful clinically when combined with adjunct evaluations, such as EEG^106-108 or MRI.¹¹⁶ Differences in values obtained by the different devices and sensors must be taken into account; similarly, randomized clinical trials and large sample-sized observational studies are needed to better depict NIRS’s role in neurodevelopment.^117,118

Functional Near-Infrared Spectroscopy

Functional NIRS (fNIRS) can measure oxy- and deoxy-hemoglobin chromophores, following similar physical principles abovementioned and assesses the activation of specific cerebral areas secondary to certain types of tasks, showing the actual connections between focal cortical activity and cognitive and developmental outcomes. As neuronal circuits are activated, there is increased oxygen consumption and metabolic demand, which is translated into increased CBF and oxygen delivery.¹¹⁹ Thus, it takes into account the difference of the quantitative regional hemodynamic state in a specific location when cortical activation occurs versus the nonactivated state.^119-121

Motor, sensory, visual, auditory, and olfactory experiments have been conducted in neonates and infants to evaluate primary cortical areas such as the superior temporal gyrus, the postcentral gyrus, and the prefrontal, orbitofrontal, and occipitotemporal cortexes.^120,122 fNIRS has been used in neonates to study language processing. For instance, Peña et al showed that native language is lateralized early in life since the left temporal area demonstrated higher activation in response to forward speech.¹²³ On the other hand, emotional components of speech are processed in the right hemisphere.¹²⁴ fNIRS offers relatively higher spatial resolution compared with EEG. In addition, its low cost and portability make it an appropriate technique for evaluating awake healthy or ill neonates without sedation.

Few studies have been done regarding fNIRS use in HIE. Cheng et al, for instance, presented auditory-evoked cerebral oxygenation changes in this population and compared them with controls, concluding that HIE neonates have lower rCBF in the frontal lobes during auditory stimulation.¹²⁵ Further research is required to better understand the role of fNIRS in neonates with HIE both at short- and long-term follow-ups. Simultaneous use of fNIRS and EEG seems promising.¹²⁶

Diffuse Correlation Spectroscopy

DCS is a novel technique derived from NIRS, which is used for continuously monitoring and quantifying microvascular CBF using temporal intensity fluctuations of the near-infrared light from tissues. These signals are measured and postprocessed by using physical models to extract a CBF index (CBFi), which is described in cm²/s.¹²⁷

Several authors have investigated the usefulness of DCS in neonates under normal and pathologic conditions. Farzam et al evaluated the optical and physiological characteristics of 30 healthy term neonates observing a positive correlation between GA, blood volume, oxygen saturation, and cerebral hemoglobin concentration. In addition, they found that CBF was higher in females compared males, bringing to attention the role of sexual dimorphism in cerebral hemodynamics.¹²⁸ Implementation of novel devices using DCS technology could potentially be used for continuous neuromonitoring at bedside in extremely vulnerable patients.¹²⁹

The role of DCS in HIE has been studied in animal models. Buckley et al proved DCS performance for calculating (CBFi) in HIE animal models. Carotid blockage decreased bilateral hemispheric CBFi, which was shortly followed by hyperemia. Subsequent TH revealed that CBFi values diminished below the baseline as opposed to the group not undergoing TH.¹³⁰ In regard to humans, Dehaes et al showed that neonates undergoing TH had decreased CBF and CMRO indexes.¹³¹ To our knowledge, no other studies in HIE using this technique have been conducted.¹¹⁸

The advantages of DCS include nonionizing radiation exposure, low-cost, portability, and short time for imaging acquisitions, especially in young children who have thinner extracerebral layers. As opposed to TCD, which mostly evaluates major cerebral vessels, DCS can be focused on regional cortical microvessels, although distinction between arterial versus venous flow is not feasible with this technique.¹²⁷

Cerebral Autoregulation in Neonates with Hypoxic-Ischemic Encephalopathy

Cerebral autoregulation (CA) is understood as the ability to maintain a relatively constant CBF regardless of the changes in blood pressure values in the brain, which is usually performed by the brain arterioles.^46,132 CBF is regulated by the partial pressure of arterial carbon dioxide (PaCO₂), the mean arterial pressure (MAP), among other intrinsic metabolic parameters. CA is directly linked to fetal development stage and is well-established in healthy neonates.⁹¹ However, it is impaired in children with HIE which can make the brain susceptible to abrupt systematic blood pressure changes.¹³³

Cerebral autoregulation can be measured with TCD by determining changes in flow velocities in static or dynamic status. Autoregulation indexes can be calculated by using intrinsic systemic and cerebral blood pressure parameters in the static (normal>0.4) and dynamic (range=0–9, being 0 absent autoregulation) tests. Given the dynamic nature of cerebral autoregulation, other ratios (moving correlation index) have been introduced to overcome static evaluation limitations.⁴⁶ Cerebrovascular dysfunction post-HIE in term neonates may result in diminished or exaggerated reperfusion response implicated in reperfusion injury and permanent brain injury.¹⁰⁵

Ideally, blood pressure measurements would be done in areas where cerebral autoregulation is most optimal, so called mean arterial blood pressure with optimized function (MAP_OPT), which can be calculated by near-infrared spectroscopy (NIRS). Lee et al¹³⁴ found that neurologic injury worsened if blood pressure decreased relative to MAP_OPT, mainly in regions such as the paracentral gyri and white matter. On the other hand, blood pressure values greater than MAP_OPT were correlated with less severe injury in the paracentral gyri.¹³⁴ Carrasco et al¹³⁵ evaluated injury on diffusion tensor imaging based on deviations from MAP_OPT in neonates with HIE. They found that higher MAPs than MAP_OPT resulted in a decrease in mean diffusivity in the basal ganglia, thalamus, posterior limb of the internal capsule, and posterior centrum semiovale, which may have indicated cytotoxic edema. In light of this, it is possible to determine how closely patient’s actual blood pressure is to their calculated optimal values, based on the integrity of their cerebral autoregulation.

Recommendations and Closing Remarks

Choosing and Combining Appropriate Imaging Techniques to Improve Clinical Care

CBF can be estimated by using several advanced neuroimaging techniques. Factors affecting the choice of modality for clinical care or research use depend on acquisition time, need for sedation, need for qualitative or quantitative results, pathophysiological questions to be answered, invasiveness, and exposure to ionizing radiation. In addition, unique features of the neonatal population, such as open fontanelle, which allows the use of US, and increased water content in blood and brain parenchyma enhancing the quality of ASL, may be advantageous for some techniques over others. Convenience and repeatability are important factors. Bedside techniques such as ultrasonography or NIRS allow for easily repeated or continuous monitoring that facilitates individualized care in the midst of multiple support devices and critical care. Evaluation of CBF surrogates, such as cerebral oxygenation along with cortical electrical activity, also serve as early predictors for short- and long-term neurodevelopmental outcomes. Variation of spatial and temporal resolution among techniques confers intrinsic advantages and disadvantages for each one of them, thus the importance of not relying in individual modalities to define treatment strategies. A desirable strategy in the future would be multimodal approaches to characterizing brain health and guiding clinical management, especially in treating patients with severe pathologies with high morbidity and mortality rates whose complex brain physiology may not depicted with one modality alone.^{105-108,115,126,136}

Cerebral oxygenation in neonates with an adverse outcome is most likely explained by low energy metabolism and oxygen utilization after severe brain injury, cerebral hyperperfusion, and impaired autoregulation of the cerebral vasculature.^137,138 These conclusions, for instance, are supported by the correlation of findings on MRI with NIRS.¹³⁶

Providing Accurate and Detailed Descriptions of Methodology and Cohort Characteristics

Close monitoring of neural function during brain development and maturation is crucial for neonates who suffered CNS injury related to perinatal asphyxia. The ideal scenario would be that all techniques were comparable among themselves. However, as seen in ►Table 1, there are several discrepancies in CBF values (or their surrogates) between the abovementioned modalities, in both healthy and ill infants, which need to be taken into account in clinical and research settings.

Correspondingly, variability in estimation techniques, gestational age at birth, and clinical condition of reported values are heterogeneous. High variability in measurements underscores the need to implement serial measurements in the same individual, not only to assess disease progression, but also to ensure reproducibility. In addition, collaborative efforts should be encouraged to learn across institutions and increase sample size for a larger database. More emphasis on technique standardization, specific brain region evaluation, time of scanning, stratification to acute or subacute phases, and hemodynamic status of subjects should be clearly stated in future studies. Additionally, serial measurements will allow study of intraindividual fluctuations observed in CBF throughout hospitalization and for the correlation of spatiotemporal CBF evolution of changes to clinical outcomes.

In conclusion, the state of the field demonstrates diagnostic and prognostic utility of assessing CBF as an important biomarker in neonatal HIE. Continued efforts will undoubtedly reveal important associations between pathophysiology and CBF that can be used to improve clinical management, drug development, and prognostication. Multimodal approach for characterizing and understanding CBF in neonatal HIE will be critical.

Key Points.

• Hypoxic ischemic injury in infants can result in adverse long-term neurologic sequelae.

• Cerebral blood flow is a useful biomarker in neonatal hypoxic-ischemic injury.

• Imaging modality, variables affecting cerebral blood flow, and patient characteristics affect cerebral blood flow assessment.