Positron Emission Tomography after Ischemic Brain Injury: Current Challenges and Future Developments

Abstract

Positron emission tomography (PET) is widely used in clinical and animal studies, along with the development of diverse tracers. The biochemical characteristics of PET tracers may help uncover the pathophysiological consequences of cardiac arrest (CA) and ischemic stroke, which include cerebral ischemia and reperfusion, depletion of oxygen and glucose, and neuroinflammation. PubMed was searched for studies of the application of PET for “cardiac arrest”, “ischemic stroke” and “targeted temperature management”. Available studies were included and classified according to the biochemical properties involved and metabolic processes of PET tracers, and were summarized. The mechanisms of ischemic brain injuries were investigated by PET with various tracers to elucidate the pathological process from the initial decrease of cerebral blood flow (CBF) to the subsequent abnormalities in energy and oxygen metabolism, to the monitoring of inflammation. In general, the trends of cerebral blood flow and oxygen metabolism after ischemic attack are not unidirectional but closely related to the time point of injury and recovery. Glucose metabolism after injury showed significant differences in different brain regions whereas global cerebral metabolic rate of glucose (CMRglc) declined. PET monitoring of neuroinflammation shows comparable efficacy to immunostaining. The technology of PET targeting in brain metabolism and the development of tracers provide new tools to track and evaluate the brain’s pathological changes after ischemic brain injury. Despite no existing evidence for an available PET-based prediction method, discoveries of new tracers are expected to provide more possibilities for the whole field.

Introduction

Cardiac arrest (CA) and ischemic stroke have been a focus of clinical and animal research because of their high rate of mortality and disability worldwide. In the U.S., it is estimated that the incidence of CA was 290 000 per year in-hospital and 350 000 per year out-of-hospital ¹. Furthermore, up to 800,000 stroke events occur in the U.S. per year and result in a financial burden as high as $3.4 billion ². CA outcomes may be worse for older adults. A study of the Danish population showed that over 10 years, the 30-day and 1-year-survival rates for out-of-hospital cardiac arrest (OHCA) patients improved for patients between 18 and 80 years of age, but not for patients older than 80, possibly as a result of lower physiological reserve and higher comorbidities in the most elderly patients ³. A Japanese study found that over five years, the neurological outcomes 1 month after OHCA for patients of all ages improved, but that this improvement was lowest in patients older than 85 ⁴.

Due to its relative lack of glucose and oxygen reserves, the brain is considered to be the most intolerant organ to hypoxic-ischemic attack. Numerous studies, including animal research, were conducted to explore the pathophysiology of ischemic brain injury, which consists of both primary and secondary insults. Rapid depletion of adenosine triphosphate (ATP), the primary dysfunction resulting from hypoxia-ischemia in the brain, ensues after CA or ischemic stroke onset ⁵. As a result, the dysfunction of cell membrane ion pumps induces the outflow of K⁺, leading to an increased extracellular K⁺ concentration. Meanwhile, a large influx of Na⁺, Cl⁻ and water enters cells, leading to cytotoxic edema ⁶. Total intracellular calcium increases by approximately 25% due to calcium influx during anoxic depolarization⁷. Elevated calcium levels activate the calpain system, followed by remodeling of cytoskeletal and membrane structures, altered signal transduction pathways, and apoptosis ⁶. Furthermore, anaerobic metabolism induced by hypoxia releases excessive lactic acid and H⁺, which exacerbates cellular dysfunction and leads to more extracellular potassium accumulation^{8, 9}.

Secondary injury is due to reperfusion injury, as there is massive production of reactive oxygen species (ROS) in the forms of a superoxide anion (O₂⁻), H₂O₂, and hydroxyl radical (OH⁻). The peroxidative action of ROS hinders activation of enzymes of glucose metabolism ¹⁰. Other free radicals, such as reactive nitrogen species (RNS) produced by reaction of nitric oxide (NO), can also disrupt cell homeostasis. These ROS/RNS cause variable oxidative damage, ranging from the mitochondria to the cell, leading to neuroinflammation ¹¹. The pathological features of neuroinflammation at the cellular level include activation of microglia and infiltration of inflammatory cells from the peripheral circulation ^12–15, which further exacerbates cell damage. On the contrary, neuroinflammation also leads to the removal of necrotic cell debris and improvement of the extracellular environment ^16–18.

Although PET is widely used in hypoxic-ischemic brain injury, the review of PET and hypoxic-ischemic brain injury, particularly cardiac arrest brain injury, is limited. There is limited literature investigating the pathogenesis of ischemia and hypoxia (perfusion-metabolism abnormality and reperfusion injury-inflammation) as mechanical clues. This paper attempts to review the two most common etiologies of cerebral ischemia and hypoxic injury, using PET technology as the core tool for investigating the pathological mechanisms of this injury in hopes to offer a new point of view.

Application of PET in Neurological Diseases

The mechanisms of pathology for both CA and ischemic stroke include cerebral ischemia and reperfusion, depletion of oxygen and glucose, and changes of neuroinflammation. To date, electrophysiological and biochemical methods to monitor these changes of the brain after ischemic injury have provided diverse exploratory directions. Additionally, applications of neuroimaging technology in ischemic brain injury have been shown to be increasingly valuable ¹⁹. Computed tomography (CT) and magnetic resonance imaging (MRI) are the traditional imaging methods for evaluating brain injury but have inherent limitations in the dynamic monitoring of brain function because of their technical focus on anatomical structure rather than brain function ²⁰.

Positron emission tomography (PET) was first used in 1975 by Phelps and Ter-Pogossian to explore dynamic brain function. They injected H₂¹⁵O and ¹³N into normal dogs to assess uptake of water and ammonia into soft tissue ^{21, 22}. The application of PET in neurology increased in the following decades with the development of various tracers, which are a group of special compounds labeled by radionuclides. The half-life and production path of radionuclides are the primary factors for their application (Table 1). Different tracers, due to their biochemical properties, target different tissues and participate in a diverse distribution of metabolic processes ²³. The process characteristics of metabolism are calculated by mathematical algorithms. With the development of new tracers, PET has been adopted for clinical applications in neurology because of its metabolism-based principles. Currently, in addition to CA and ischemic stroke, PET is widely used in dementia (early diagnosis, differential diagnosis), brain tumor (grading, staging, tumor localization, mass lesion diagnosis, tumor recurrence versus treatment effect, therapy efficacy evaluation, malignant degeneration diagnosis, prognostication), epilepsy (episodic neurologic syndrome diagnosis, localization of seizure focus), parkinsonism (early diagnosis, differential diagnosis), cerebrovascular disease (cellular viability, cellular ischemia), and traumatic brain injury (injury identification) ²⁴. Utilizing a variety of PET tracers, different brain functions can be assessed, and different brain states can be longitudinally detected (Table 2).

Table 1.

Half-life, requirement, and application of PET radionuclides

Radionuclide	Half-life	Requirement	Major application
18F	110 min	· End products · Used directly	Clinical setting 25
11C	20 min	· On-site cyclotron	Animal research 26
13N	10 min	· On-site cyclotron	Animal research 27
15O	2 min	· On-site cyclotron	Animal research 28
68Ga	68 min	· 68Ga generators · Local installation · Simple operation	Clinical setting 29
82Rb	1.3 min	· On-site cyclotron	Animal research 30
89Zr	3.27 days	On-site cyclotron	Animal research 31

F: fluorine, C: carbon, N: nitrogen, O: oxygen, Ga: gallium, Rb: rubidium, Zr: Zirconium

Table 2.

PET application and target using variety tracers in neurological disorders

Application	Target	Tracer
Hypoxia-ischemia injury	Cerebral blood flow	H215O 32
	Oxygen metabolism	15O2/ FMISO 33, 34
	Erythropoietin receptors	89Zr-DFO-EPO 35
	Glucose metabolism	18F-FDG 36
	CXCR4-expressing	68Ga pentixafor 37
Movement disorders	Dopamine storage	18F-dopa 38
	Dopamine D1 receptors	11C-SCH23390 39
	Dopamine D2 receptors	11C-Raclopride 40
Neuropharmacology	Central benzodiazepine binding	11C-Flumazenil 41
	Opioid binding	11C-Diprenorphine 42
Metabolism of CNS	Cellular amino acid uptake	11C-Methionine 43
Inflammation of CNS	Activated microglia	11C(R)-PK11195 44
	TSPO	11C-PBR28 45
		11C-DPA713 46
		18F-DPA714 47
Tumors of CNS	Glutamine synthetase	13N Ammonia 48
	Capabilities of blood brain barrier permeability	82Rb chloride 30

FMISO:¹⁸F-fluoro-misonidazole, ¹⁸F-FDG:¹⁸F-2-deoxyglucose, ¹⁸F-dopa:¹⁸F-6-Fluoro-dopa, CNS: central nervous system

As the most common hypoxic-ischemic brain injury in the clinic, ischemic stroke has been intensively studied with PET in recent years. Because of a cascade of metabolic and molecular alterations, there is cellular dysfunction and morphological damage after an ischemic stroke. PET makes it possible to detect these pathophysiological changes and permits the regional assessment of metabolic variables and the location mapping of molecular markers. To follow these pathophysiological changes, PET is noninvasively applied to longitudinally quantify the regional cerebral blood flow (rCBF), regional cerebral blood volume (rCBV), regional cerebral metabolic rate of oxygen (rCMRO2), oxygen extraction fraction (OEF), and the regional cerebral metabolic rate of glucose (rCMRglc) ⁴⁹.

Compared to the regional hypoxia-ischemia of the brain caused by stroke, global hypoxia-ischemia of the brain induced by CA is more severe and often lethal. Both primary and secondary injuries after the hypoxic-ischemic injury can be investigated by PET to monitor the metabolic and inflammatory mechanisms. In terms of primary injury, PET was applied to investigate the halt of cerebral perfusion as well as the subsequent interruption of oxygen and glucose metabolism ^{50, 51}. Neuroinflammation has been extensively studied with PET as a secondary insult of brain injury ^{46, 52–56}. In this paper, we reviewed research on PET in the field of CA and ischemic stroke. The limitations and prospects for future research are also discussed.

Evaluation of Glucose Metabolism after Ischemic Brain Injury with PET

The brain is one of the organs that consumes the most glucose per unit weight (25% of total glucose utilization compared to 2% of the bodyweight) ⁵⁷. Normally, the extraction fraction of glucose in the brain is approximately 10% ^{58, 59}. PET research using an ¹⁸F-FDG tracer for brain glucose metabolism has lasted more than 40 years ⁶⁰. The study of glucose metabolism in ischemic brain injury is the most used PET application (Table 3) because ¹⁸F is a positron-emitting radioisotope with a half-life of 109.8 min ⁶¹, which is an appropriate time for PET. To identify the glucose uptake of the brain, standardized uptake values [SUV: (mean radioactivity measured in the volume of interest /injected radioactivity) * animal weight] and cerebral metabolic rate of glucose (CMRglc), which was expanded from SVU, was introduced ^62–64. Due to the severity of patient conditions and the complexity of PET procedures, research using these parameters has generally come from animal subjects rather than humans. Representative PET images using ¹⁸F-FDG after cardiac arrest from Jia lab ⁶⁵ are presented in Figure 1.

Table 3.

Evaluations of glucose metabolism after ischemic brain injury with PET

Types	Publication	Clinical or preclinical	Main parameters and results	New findings
Animal	Liang, 201866	9 SD rats of stroke (MCAO)	18F-FDG uptake increased in part of contralateral brain	· Abnormal metabolic connectivity in acute ischemic stroke · Metabolism decreased in ipsilateral hemisphere and increased in contralateral cerebellum with voxel-wise analysis
Animal	Li, 2018 67	6 rats exposed to cerebral ischemia	Lower SUV in CIMT group compared to control group	· Evaluation of glucose metabolism based on CIMT · Assessment of glucose metabolism in different brain regions at baseline, before, and after treatment.
Animal	Joya, 2018 68	6 SD rats of stroke (MCAO)	18F-FDG uptake showed the lowest value at day 1 and gradually rose	· Combined with MRI detection · Assessment of CCD after cerebral ischemia as a predictive factor of neurological outcome
Animal	Wu, 2017 69	54 SD rats of stroke (MCAO)	Higher SUV was exhibited in EA group in CPu, MCTX and SCTX	· Detection of glucose metabolism after EA treatment · Phosphorylation of AMPK increased
Animal	Han, 2015 70	SD rats of stroke (MCAO)	18F-FDG uptake decreased and remained low for at least 14 days after ischemic stroke	· Evaluation of glucose metabolism based on HUK treatment · HUK increased angiogenesis in stroke rats
Animal	Hwang, 201571	10 SD rats of stroke (MCAO)	Cortical 18F-FDG uptake increased after liposomal delivery of angiogenic peptides compared with that of control and liposomes group	· Glucose metabolism after therapy of angiogenic peptides incorporated into liposomes ·Angiogenic peptide delivery facilitated glucose utilization
Animal	Arnberg, 2015 72	25 SD rats of stroke (M2CAO)	Increase of 18F-FDG uptake in the peri-infarct region	· Stroke model of M2CAO · Combined with MRI detection
Animal	Jiang, 2014 73	36 SD rats of stroke (MCAO)	18F-FDG uptake increased after post-stroke exercise	· Exercise therapy post-stroke · Exercise reverses hypometabolism caused by ischemia back to baseline from 7d to 2d on the ipsilateral brain
Animal	Wang, 2013 74	24 rats of stroke (MCAO)	18F-FDG uptake ratio (lesion/normal) of cerebral ischemic area increased in the ESCs-treated group and iPSCs-treated group at 1 week after transplantation	· Treatment of iPSCs or ESCs · Treatment groups showed significantly higher 18F-FDG accumulations in the ipsilateral cerebral infarction
Animal	Walberer M, 2012 75	10 Wistar rats of stroke (MCAO)	K1 and net influx rate constant Ki related to 18F-FDG uptake could predict tissue fate	· Unidirectional transport parameter K1 of FDG correlated well with rCBF at 60 minutes after the stroke · The infarct volume determined by MRI was well predicted by FDG-PET
Clinical observation	Stender, 2015 76	41 patients in the states of VS/ UWS/ MCS	Median global cortical CMRglc decreased in patients in the states of VS/ UWS/ MCS	· Averaged 42% of normal in VS/UWS, 55% of normal in MCS · ROC curve showed in MCS/ VS/ UWS patients can be differentiated with 82% accuracy based on cortical metabolism
Clinical observation	Soddu, 2015 77	11 VS/UWS and 4 locked-in patients	SUV (averaged over gray matter) was higher in healthy controls (5.5 ± 1.3) compared to VS/UWS patients (1.9 ± 1.3)	· Conjunction analysis with PET and fMRI show decreases in frontoparietal and medial regions in VS patients
Animal	Kim, 2019 78	18 SD rats of 5min CA	SUVdelta (SUV changes after CA) of forebrain regions were significantly decreased in good outcome group (p < 0.05)	· SUVdelta was firstly used to predict the outcome · Morris water maze test was applied to evaluate functional outcome
Animal	Putzu, 2018 79	10 rats of 8min CA	CA altered 18F-FDG uptake in the range of −36% ~ +4% in different brain structure	· Focus on metabolism from cortex to brainstem, and deep structure · Autoradiography was applied
Animal	Li, 2015 80	6 beagle dogs of 6min CA	CMRglc decreased in whole brain, frontal cortex, cerebellum, and thalamus at 4 time-point	· Comparison of glucose metabolism at 4 time points (baseline, 4h, 24h, 48h) · Assessment of hexokinase I (HXK I) and HXK II and glycolysis-related genes (GLUT3, HXK-I, GPI, PGK1, ENO2, and PKM2)
Animal	Zhang, 2015 81	44 miniature inbred pigs of 8min CA	Maximum value of SUV decreased in both VFCA and ACA groups compared to sham group at 24 h post-ROSC (parietal lobe, frontal lobe, brain stem and cerebellum)	· Grouping strategy based on different causes of cardiac arrest · Neurological deficit scores (NDS) and NSE/S100β were quantified
Clinical observation	Schaafsma, 2003 82	10 patients of CA	Reduction of 18F-FDG uptake was most marked for cortical gray matter (54% of normal) and least for cerebral white matter (70% of normal)	· Significant differences in PET data were not found between survivors and non-survivors
Clinical follow-up research	Rudolf, 1999 83	24 patients Post-CA comatose	Cortical CMRglc in patients with PVS were significantly reduced when compared to patients in AVS	· Comparison of CMRglc in PVS and AVS patients · No significant correlation was found between the degree of evoked potential or EEG alterations and the reduction of glucose metabolism in VS patients
Clinical observation	Tommasino, 1995 84	15 unconsciousness patients of CA, TBI, Brain Ischemia	rCMRglc was significantly reduced to 3.70 +/− 61 in coma, to 3.45 +/− in VS, and to 2.33 +/− 0.34 mg/100 g/min in PVS patients	· Early clinical research · Comparison of glucose metabolism in patients of different coma levels

SUV: standardized uptake value, CIMT: constraint-induced movement therapy, EA: electro-acupuncture, CPu: caudate putamen, MCTX: motor cortex, SCTX: somatosensory cortex, AMPK: AMP-activated protein kinase, HUK: human urinary kallidinogenase, iPSCs: induced pluripotent stem cells, ESCs: embryonic stem cells, (r)CMRglc: (region) cerebral metabolic rate of glucose, VS: vegetative state, PVS: persistent vegetative state, AVS: acute vegetative state, UWS: unresponsive wakefulness syndrome, NSE: neuron specific enolase, TBI: trauma brain injury, MCAO: middle cerebral artery occlusion, M2CAO: occlusion of the M2 branch of the MCA, CCD: crossed cerebellar diaschisis, MCS: minimally conscious state, VFCA: ventricular fibrillation cardiac arrest, ACA: asphyxial cardiac arrest

Figure 1.

Dynamic images of 18F-FDG metabolism monitored by PET. (A) PET image of similar coronal brain slices for a representative rat 60 minutes after 7 minutes cardiac arrest in a rat. Mean coronal images take from an average of 10-60 minutes post-injection. Images are displayed on the same scale. (B) Image for last time frame (55-60min). S1HL: somatosensory cortex of the hindlimb. VPL: ventral posterolateral nucleus. (C) Image for all time frames (0-60min).

Some studies pay attention to the relationship between CMRgl and time after ROSC. After excluding the effects of surgery and anesthesia (sham group), CA led to a reduction of absolute ¹⁸F-FDG uptake from −32 to −36% in all cortical regions in ten rats. Unlike cortical manifestations, there was no statistically significant reduction in the striatum (−18%), hippocampus (−17%), and thalamus (−16%). No changes were observed in the posterior brain regions, such as the cerebellum or brainstem (−5 to +4%) ⁷⁹. One study included 6 dog CA models reporting that glucose metabolism decreased significantly after ROSC at 4, 24 and 48h compared to the baseline. The low value persisted and recovered slowly--less than 50% of the baseline at 48 hours after ROSC. Moreover, there was a coinciding transformation in the expression of hexokinase I and II, enzymes related to the phosphorylation of cellular glucose ⁸⁰. In addition to studying the time course of glucose metabolism after CA, PET has also focused on mapping glucose metabolism in different regions of the brain. The parietal lobe, frontal lobe, brain stem, and cerebellum post-CA all showed low metabolism of glucose, which supported the conclusion that brain injury after CA was diffuse ⁸¹. A clinical study investigating 8 patients with hypoxic-ischemic encephalopathy drew the map of ¹⁸F-FDG uptake and revealed that no distribution border existed between the gray and white matter of the brain. Without correlation with the outcome, the quantitative evaluation of ¹⁸F-FDG uptake showed a prominent decrease in glucose consumption compared to uptake in the control group (cortical down to 54% of normal and subcortical down to 70% of normal) ⁸². Previous studies compared the glucose metabolism rate detected by PET between patients with persistent vegetative state (PVS) and those who died post-CA through an 18-month follow-up, and the results indicated that there was a small and insignificant difference between the two groups ⁸³. In another study, patients with altered consciousness post-CA were divided into three subgroups of the comatose group (n = 5), vegetative state group (VS, n=6), and persistent vegetative state group (PVS, n=4). Ten normal age-matched subjects were grouped as control. The results showed that rCMRglu measured by PET was markedly reduced in all three groups compared to control, and the reduction of PVS group (50–72%) was significantly lower than that of the coma (36–54%) and VS (34–54%) groups in all the structures, although rCMRglu did not correlate with neurological outcome ⁸⁴.

When referring to PET and stroke, the areas of interest extended widely from basic glucose metabolism changes to tissue fate, neurological recovery, and even differential performance of PET after physical therapy ⁶⁹. The ¹⁸F-FDG value was proposed as the most promising tracer to discriminate the recoverable tissue (with medium value) from recruited tissue (with a lower value) and normal tissue (higher value) after mild ischemia at early times in both permanent and transient ischemic animal models ⁸⁵. Longer-term animal research evaluated the glucose metabolism with ¹⁸F-FDG in regions of the brain and the whole brain, and at 1, 3, 7, 14, 21, and 28 days after occlusion of the middle cerebral artery (MCAO). At day 1 after ischemia, the uptake of ¹⁸F-FDG fell to the lowest level and then gradually increased from day 3 to days 7–14 ⁶⁸. However, when rat models preserved the collateral flow, inconsistent conclusions were drawn. PET image showed an increased ¹⁸F-FDG uptake within ischemic regions succumbing to infarction and in the peri-infarct region ⁷². To understand the potential relationship between glucose metabolism and brain tissue fate by PET, a rate parameter of K₁ derived from rCBF and the permeability surface product of FDG, which represents the transport from blood to the brain, was introduced ^{86, 87}. It was reported that rCBF strongly correlated with the K₁ of FDG in a rat MCAO model. Meanwhile, tissue fate could be predicted based on K₁ ⁷⁵. In a photochemically induced thrombosis (PIT) model, at 1 and 3 days after PIT, there was a lower level of ¹⁸F-FDG uptake in the infarct regions and higher ¹⁸F-FDG uptake in the area surrounding the infarct lesion at day 7, contradicting expectations. Further immunohistochemical studies showed that there was a high peak of ¹⁸F-FDG uptake on day 7, which may be due to neuroinflammation, including microglial cell activation. Therefore, to interpret F-FDG solely as an estimation of brain glucose metabolism may result in an overly optimistic estimation of brain damage ⁸⁸. More complex experimental results show that the metabolic variations in the bilateral cerebral hemisphere after stroke may be variable ⁶⁶. MCAO rats showed low metabolism in some areas of the impaired hemisphere, such as the striatum, somatosensory cortex, auditory cortex, thalamus, bilateral amygdala, insular cortex, hippocampus, and olfactory cortex, while hypermetabolism manifested in the contralateral hemisphere. For example, there was hypermetabolism in the somatosensory cortex, auditory cortex, thalamus, posterior parietal cortex, hippocampus, and visual cortex ⁶⁶. PET was also used to explore the mechanism of treatment after stroke. A study aiming to investigate the mechanism of human urinary kallidinogenase in MCAO rats found that treatment promoted angiogenesis in rats, and this facilitation was associated with an increase of ¹⁸F-FDG uptake detected by PET ⁷⁰.

Evaluation of Cerebral Perfusion with PET

The principal regulators of cerebral blood flow (CBF) include the partial pressure of arterial carbon dioxide (PaCO2), cerebral metabolism, and mean arterial pressure (MAP). As a major pathophysiological change after cerebral ischemia, the decrease of CBF has been the focus of many studies. Cerebral perfusion is significantly sensitive to changes in PaCO2. Techniques commonly used to evaluate cerebral perfusion and their advantages/ disadvantages in clinical context are displayed in Table 4. PET demonstrates its unique advantage based on continuous and multifactorial detection. However, the application of PET is limited compared to other techniques due to its cost and technical requirements.

Table 4.

Techniques commonly used to evaluate cerebral perfusion in clinical research

Techniques	Main method	Advantages	Disadvantages
PET	Detection of 15O-H2O distribution with CBF 89	• Dynamic • Multiple metabolic factors	• Unclear anatomical structure • Radioactive
MRI	• Dynamic susceptibility contrast of contrast agent 90 • Arterial spin labeling 91	• Clear anatomical structure • Multiple perfusion parameter (MTT, TTP, CBV, CBF)	• Non dynamic • Avoiding special metals
CT	Gamma-variate fits applied to the vascular time-density curves after injection of contrast agent 92	• Multiple perfusion parameter (MTT, TTP, CBV, CBF) • Wide availability	• Contrast-noise ratio • Radioactive • Risk of allergy and acute kidney injury
TCD	The Doppler effect of pulsed wave Doppler to image vessels at various depths 93	• Real-time detection of CBF • Repeat assessment/follow-up • Non-invasive, radiation-free	• Quantitative difficulties
LSCI	The random granular effect produced by a laser when illuminates a diffuse surface 94	• Excellent spatial and temporal resolution on blood flow • Dynamic and in real time	• Invasive and limited clinical application • Stationary scatteres affect the measure

MTT: mean transit time, TTP: time to peak, CBV: cerebral blood volume, TCD: transcranial doppler, LSCI: laser speckle contrast imaging

As opposed to other technologies such as LSCI ⁹⁵, detection of CBF by PET depends on special tracers, and the most commonly used tracer is ¹⁵O labeled H₂O, which can freely pass through the blood-brain barrier (BBB). CA usually causes acute, global cerebral ischemia after cardiopulmonary resuscitation (CPR). When return of spontaneous circulation (ROSC) is achieved, the stability of hemodynamics does not directly lead to the complete recovery of cerebral perfusion. It is generally accepted that the timeline of CBF consists of three phases: early hyperemia (0-20min after ROSC), hypoperfusion phase (20min-12h after ROSC) and restoration of normal blood flow (12-72h after ROSC) ⁹⁶.

In a CA model with nine young pigs, a more detailed survey by PET (¹⁵O-H₂O) uncovered that, compared to CBF at baseline of (32+/−5)· ml·g⁻¹·min⁻¹, there was a hyper-perfusion period around 10 min after ROSC (most evident in the diencephalon) followed by a 30-minute period of hypoperfusion until a gradual recovery to baseline values ⁹⁷. Clinical PET observation found that global CBF in CA patients continued to be lower than that in the normal population within a week from CPR to neurological recovery ⁹⁸. This trend was present in all patients of awake or persistent coma at the endpoint. In addition to global CBF, the variation of local CBF was relatively significant. For example, in the putamen, CBF was higher in the group with better neurological prognosis ⁹⁸. In addition to the temporal characteristics of CBF, the spatial characteristics of CBF were also assessed in a clinical study ⁸². A single qualitative analysis showed that, in general, CBF images of patients had clear boundaries between cortex and white matter after CA. However, when the area of cortex or white matter was investigated separately, the distribution of trace was uniform, which indicated that CBF in the different areas had close relationships with the number and density of neurons ⁸².

Compared to CBF and CA, PET research focusing on CBF and stroke has been more thorough. There have been studies not only on perfusion, but also on thresholds for judging ischemic penumbra and cerebral infarction ⁹⁹. The penumbra was distinguished as an area with a reduction of CBF, raised OEF, and stable CMRO2 in PET studies. The absolute value of CBF in the ischemic penumbra was estimated to be between 12ml·100g⁻¹·min⁻¹ and 22ml·100g⁻¹·min^{−1 100, 101}. When we considered the CBF value in the normal hemisphere as a baseline, there was a 40% decrease in CBF value in the insulted hemisphere detected by PET, indicating tissue infarction ¹⁰². As a stroke subtype, silent brain infarction (SBI) is considered to be a risk factor for severe stroke, although its mild symptoms make it difficult to detect clinically. A PET study provided a new breakthrough point: it found that patients with SBI showed a decreased mean CBF value in the bilateral cortex and deep grey matter ¹⁰³. PET was also used as a monitoring tool of CBF while stroke therapy was conducted. For example, an animal study using ¹⁵O-H₂O PET showed that pre-ischemic exercise significantly improved CBF during reperfusion after MCAO ¹⁰⁴. A clinical observational study involving 40 ischemic stroke patients and 40 healthy volunteers showed that the brain temperature difference (affected hemisphere temperature minus contralateral hemisphere temperature) was correlated with cerebral blood volume (CBV) (r = 0.570) and oxygen extraction fraction (OEF) (affected hemisphere /contralateral hemisphere) (r = 0.641) in the deep white matter ¹⁰⁵. Another double-blinded clinical study including 15 patients (11 men and 4 women with ischemic stroke) evaluated CBF with PET. It reported that CBF improved after 14 days of vinpocetine (a neuroprotective agent) treatment ¹⁰⁶.

Evaluation of Cerebral Metabolism of Oxygen

The origin of interest in PET research on brain oxygen metabolism is the close relationship between oxygen metabolism and brain tissue fate. Oxygen diffuses into tissue based on the difference between the partial pressure of oxygen in the capillary and the tissue ¹⁰⁷. Once the brain is injured, obstruction of oxygen diffusion leads to the microvascular collapse, endothelial swelling, and tissue edema, which are all further exacerbated by hypoxia, causing a vicious cycle ¹⁰⁸.

The targets of most studies interested in brain oxygen metabolism with PET were oxygen extraction fraction (OEF) and cerebral metabolic rate of oxygen (CMRO2) ^{109, 110}. A CA study in pigs reported that CMRO2 in the cerebellum decreased to 68±20μM·Hg⁻¹·min⁻¹ at 60min, 63±5μM·Hg⁻¹·min⁻¹ at 180min and 63±9μM·Hg⁻¹·min⁻¹ at 300min compared to a baseline of 77±7μM·Hg⁻¹·min⁻¹. The analysis further showed that regional CMRO2 (rCMRO2) in the cortex was lower than in the subcortex (p<0.05) at 60, 180 and 300min after ROSC ¹¹¹. Clinical research of CA suggested that CMRO2 of the whole brain was lower after ROSC regardless of the neurological outcome. However, the whole brain CMRO2 substantially decreased to 55mmol·100g⁻¹·min⁻¹ (30% of normal) in the poor prognosis group after 1 week ⁹⁸. Corresponding to this, the patients from the better prognosis group had a mean value of 85mmol·100g⁻¹·min⁻¹ (50% of normal) after 1 week. The whole-brain oxygen extraction ratio (OER) showed both decreased and increased values after CPR. These results suggested that the progressive declines of CMRO2 over the first week may indicate a prolonged coma ⁹⁸.

The changes in cerebral oxygen metabolism after ischemic stroke are more complex than those of CA. An MCAO study using nine adult cats reported that CMRO2 fell to 55% and eventually to 25% compared to control. OEF increased primarily in the center of ischemia at the early stages. With time, areas of increased OEF moved from the center of the ischemic area to the periphery ¹¹². Another animal study including seven baboons that underwent sequential ¹⁵O₂-PET demonstrated that the mean infarction volume (2.4 +/− 0.6 cm³) was comparable to a tissue volume with oxygen consumption < 40%. In addition, the OEF increased significantly at 1 hour ¹¹³. A serial PET investigation in 5 baboons that underwent MCAO also validated that OEF increased during the early stage ranging from 60 to 95 minutes after MCAO ¹¹⁴. Similar changes in OEF were also reported in a PET study of MCAO with eleven pigs. In the ischemic core area, OEF was elevated within 1hour post-MCAO but decreased thereafter, and CMRO2 was reduced initially followed by a further decrease over the subsequent 2–3 h ¹¹⁵. The fate (infarct or not) of high OEF regions was indeterminate ^{116, 117}.

In recent years, a novel tracer that selectively stains hypoxic tissue, ¹⁸F-misonidazole (FMISO), was developed. Selective staining occurs because, in hypoxic tissues, FMISO’s nitro functional group will be reduced to form compounds that bind irreversibly to cell structures, whereas in normal tissues, FMISO can diffuse out after being re-oxidized ¹¹⁸. A pilot FMISO study of 7 rats with temporary, permanent, or no MCAO showed that FMISO was trapped in the affected cortex only for the permanent MCAO animals, whose cortical pan-necrosis was confirmed via histopathology. These findings support the validation of FMISO as a marker of viable hypoxic tissue ¹¹⁹. A prospective study revealed FMISO trapping was present in all patients and overlapped mainly with the penumbra but also with the core ¹²⁰. As a promising tracer, its advantages need to be further developed.

Neurovascular coupling of CBF/CMRO₂

PET determination of OEF and CMRO2 can be another technique to identify abnormal neurovascular coupling in stroke patients, which has already been revealed from ultrasound studies ¹²¹. A pattern of “luxury perfusion” determined from a combined PET/CT scan, which consisted of a fall in regional OEF combined with elevation in CBF (suggesting that CBF is higher than expected given the CMRO2 in a particular area which indicated the coupling of CBF and oxygen metabolism was abnormal) ¹²², allowed the diagnosis of circumscribed ischemic infarction for one patient who showed hypermetabolism in the territories of the anterior and middle cerebral arteries ¹²³. Another PET study measuring CBF, CMRO2, and OEF in 37 patients up to 1 month after ischemic stroke found that 15 patients exhibited patterns of luxury perfusion at some point during the month ¹²⁴.

PET can also be used to study neurovascular coupling following cardiac arrest. A PET study of young pigs undergoing CA found that OEF was 60% at baseline, increased at 60 min following ROSC, and declined to 50% thereafter, indicating a transition to an ischemic state ¹¹¹. Cortical regions showed an oligemic pattern whereas the cerebellum showed a pattern of luxury perfusion. A study of comatose patients after ROSC found that all patients initially had low CMRO₂ and CBF ⁹⁸. Those who remained in a coma showed a progressive decline in CMRO₂ and after 1 week had a lower CMRO₂ than the patients who woke up.

Traditionally, the assessment of CBV, CBF, and CMRO₂ coupling has required separate scans with three tracers: C¹⁵O for CBV, H₂¹⁵O for CBF and ¹⁵O₂ for CMRO₂, but a newer method allowed assessment of CBV, CBF, and CMRO₂ with two tracers and two scans, and the most recent method with two tracers and one scan ^{125, 126}. Although CBV, CBF, and CMRO₂ are most reliably assessed in patients through continuous administration of tracers, it was shown that a dynamic method of administering ¹⁵O via brief inhalations or bolus injections showed day-to-day test-retest coefficients of variability for CBF, CBV, CMRO₂, and OEF of 8.8%, 13.8%, 5.3%, and 9.3%, which were comparable to the within-session test-retest variability of the continuous method ¹²⁷.

PET and Neuroinflammation

TSPO and PET-tracers

Neuroinflammation is a crucial segment of the cascade reaction after hypoxic-ischemic brain injury ^{46, 128}. The progress of PET research in neuroinflammation depends on the development and utilization of tracers. The discovery of a translocator protein (TSPO) provided a potential target, which led to the development of new PET tracers. TSPO is an 18kDa protein involved in inflammation of the nervous system ¹²⁹. Upregulation of TSPO has been found with PET scan in many known neurological diseases involving an inflammatory reaction (e.g. dementia ¹³⁰, Parkinson’s disease ¹³¹, multiple sclerosis ¹³² and stroke ¹³³). In normal brains, TSPO is expressed only at low levels in microglia ¹³⁴. TSPO increased significantly after the onset of neurological impairments, led by CA and stroke. Based on the close correlation between TSPO and neuroinflammation, TSPO-targeted PET tracers have been widely used in monitoring and evaluating inflammatory response after ischemic brain injury.

Monitoring Neuroinflammation with PET

Research using PET to assess post-CA neuroinflammation is unavailable. Almost all neuroinflammation PET research has focused on stroke according to our retrieval. Rats, after left MCAO, underwent PET examination with ¹¹C-(R)PK11195 (a TSPO marker tracer) and showed increased ¹¹C-(R)PK11195 binding (suggesting neuroinflammation) at day 7 ^{88, 135}. AMD3100, an antagonist of C-X-C Motif chemokine receptor 4 (CXCR4), can limit the inflammatory reaction after MCAO by inhibiting the stromal cell-derived factor 1 (SDF1), /CXCR4 pathway ¹³⁶.To study the neuroinflammation after AMD3100 treatment, ¹⁸F-DPA714, the specific PET tracer of TSPO, was used to perform the PET imaging on days 0, 1, 3, 7, 10, and 16 in a mouse model of stroke. PET results showed that there was a prominent reduction of [18F]DPA-714 signal density and signal area observed in the AMD3100-treated group at 3 days. These results are consistent with those by immunofluorescence ⁴⁷. Another tracer target to TSPO is ¹¹C-PBR28. In similar rat models of MCAO, investigators discovered an analogous result: PET tracers signaling inflammation were enriched in the lesion area on day 4, 7 and 14, and the trend of enrichment was confirmed by immunofluorescence manifested as reactive microgliosis post-damage ⁴⁵. In addition to the high uptake of inflammation tracers (¹¹C-DPA713) in the ipsilateral cerebral hemisphere, an increase in inflammatory tracers in the contralateral cerebral hemisphere was found in animals after distal middle cerebral artery occlusion (dMCAO) compared to sham-operated mice. This increase is not uniform for ratio approximates (ipsilateral: contralateral) 1.4~2.09:1 ⁴⁶. Compared with research in the field of stroke, the study of neuroinflammation after CA by PET is lacking. As a consequence of a critical global hypoxic-ischemic disease, neuroinflammation after CA warrants further study by PET.

PET and targeted temperature management (TTM)

TTM has been recognized as an important method to protect cell functions and improve neurological outcomes after brain injury ^{137, 138}. PET was first used to monitor the effects of hypothermia on the normal brains of animals anesthetized with 65% nitrous oxide and propofol ¹³⁹. It was confirmed for the first time that in normal monkey brains, hypothermia (35°C) reduced CBF and restrained oxygen metabolism by significantly reducing rCMRO₂ in all ROIs (regions of interest) compared to normothermic temperatures (38°C) ¹³⁹. Another study was conducted by PET in the normal brains of pigs treated by hypothermia at 32°C and maintained at that level during the experiments. The results indicated that CBF decreased to 50% of the baseline in 3 h and then remained stable, and CMRO2 decreased to 50% of the baseline over the 6 h. Oxygen extraction fraction (OEF) rose to 140% of the baseline in the first 3 h and then gradually returned to the baseline ¹⁴⁰. This trend was not only seen in systemic hypothermia but also in selective brain hypothermia, which was validated in mixed breed pigs by the multiple colored microspheres (CMS) technique ¹⁴¹. In 1998, fifteen mongrel dogs were assigned into cooling (n=5) and rewarming (n=5) groups of hypothermic cardiopulmonary bypass. Continuous infusions of ¹⁵O-H₂O and bolus injections of ¹⁸F-FDG were used to detect CBF and glucose uptake in the normal brain when core temperature altered between 20°C and 37°C. Regional CBF decreased gradually in cooling duration, and regional CBF at 20°C fell to about 25% of its baseline value at 37°C. It also showed that, in rewarming duration, regional CBF at 32°C positively correlated with the regional glucose uptake after cardiopulmonary bypass ¹⁴².

As the main therapeutic strategy for neuroprotection after CA, hypothermia (targeted temperature management, TTM) is believed to improve neurological outcomes, even when applied for different hypothermia durations ¹⁴³. Limited by the requirements for tracer preparation, clinical investigations about the connection between PET and TTM have often only measured glucose metabolism. One study included seven comatose CA patients who received hypothermia treatment shortly after CPR. The core temperature was maintained at 34°C for 3 days followed by rewarming process at a rate of 1°C each day. PET map revealed that, compared to four patients with favorable outcome, there was a reduction in ¹⁸F-FDG uptake and low OEF in the three patients with poor outcome confirmed by the Glasgow Coma Scale (GCS) ¹⁴⁴. A clinical trial involving 63 newborns with hypoxic-ischemic encephalopathy (HIE) verified that there was a significant increase in glucose metabolism post-hypothermia therapy compared to conventional treatment. The hypothermia intervention consisted of maintaining the rectal temperature at 34–35°C for 72 hours ¹⁴⁵. A study of 31 newborns with hypoxic-ischemic encephalopathy showed that in the group of newborns who died during admission or had an abnormal neurological examination before discharge, mean urinary S100B levels were significantly higher, which was correlated with the incidence of focal or multifocal hypometabolism as determined by 18F-FDG scans ¹⁴⁶. A case study of a cardiac arrest patient who received therapeutic hypothermia revealed isolation syndrome despite a normal resting neuronal metabolism across the anterior forebrain as measured by 18-FDG PET ¹⁴⁷. Despite the lack of more in-depth research, PET and its various tracers provide a method worthy of further exploration for the study of TTM therapeutics.

Challenges and Future Directions: Predictive Value and Integration Technology

Hypoxic-ischemic brain injury results in death and disability of comatose patients after cardiac arrest. 92.6% of CA survivors were discharged with neurological impairment due to irreversible brain injury ¹⁴⁸. Therefore, imaging techniques are often used to predict neurological outcome and tissue fate. Compared with the numerous CT and MRI predictive value evaluations, PET has received less attention ¹⁴⁹. Until now, the prospective and experimental animal studies have had limitations including small sample size, lack of repeatability in the experimental process and unclear conclusions ^{82, 98}. Since patients with neurological dysfunction (especially coma) after CA often have huge treatment costs and uncertain treatment outcomes, the research on the predictive value of PET has great potential to provide mechanisms and insight for optimized intervention and significant translational value.

Previous PET studies had validated the existence of CBF thresholds ⁹⁹ below which neurological physiology is abnormal ¹⁵⁰. However, neither CBF nor OEF can predict tissue fate because they lack duration and stability ^112–114. The only reliable positive conclusion is that CMRO2 could be a promising parameter in both animal and human studies. It has been shown that reduced CBF combined with preserved CMRO2 is associated with tissue fate ¹¹⁵.

As a milestone technology, PET has been available for more than 40 years ¹⁵¹. Currently, oncology and neurology remain the main fields of PET applications ^{152, 153}. However, there are still some nonnegligible elements restricting the development of PET in clinical and animal research. First of all, there is almost no progress in the study of using PET to assess the neuroinflammation after CA. Limited progress has been made in the use of radioactive tracers to track the distribution of anti-neuroinflammatory drugs. Shiyong Li et al, used framework nucleic acid (FNA) labeled with ⁸⁹Zr, to conjugate with anti-complement component 5a aptamers (aC5a). PET was used to locate the distribution of ⁸⁹Zr-FNA, which was proven to selectively reduce C5a-mediated neurotoxicity and effectively alleviate oxidative stress in the brain ¹⁵⁴.The second issue worth discussing is the innovation of tracers. For example, FDG is the most widely used, but blood glucose, as well as some drugs, have major interference with FDG uptake ¹⁵¹. There is no broad consensus about standardized protocols for ¹⁸F-FDG PET ^{155, 156}. This suggests that we need new tracers to obtain more comprehensive and rational results. Tracers that target cell surface antigens, cell proteins, and hypoxia-ischemia have good prospects ¹⁵⁷. In fact, some efforts have been made to find suitable tracers for PET by using Designer Receptors Exclusively Activated by Designer Drugs (DREADD) technology. For example, according to the latest reports, a novel trace named ¹⁸FJHU37107 exhibits high in vivo DREADD potency. This research showed that ¹⁸FJHU37107 combined with PET allows for DREADD detection of neurons, and final completion of noninvasive neuronal projection mapping in rodent and monkey models ¹⁵⁸. Thirdly, the long amount of time required from the initial stage of injection to PET imaging acquisition could create additional risk in critical patients, such as acute stroke and CA patients. Because PET is highly sensitive to head movements, slight head movements can affect the accuracy of quantitative analysis ⁵⁷, and given their neurological injuries this patient population may not be able to comply with lengthy testing. Lastly, integrated scanners (PET/CT, PET/MRI, PET/SPECT) yield unique advantages. For example, PET/CT can map the precise anatomical structure and make it easy to analyze both function and structure. With the advent of new MRI image technologies, such as arterial spin labeling (ASL) and diffusion tensor imaging (DTI) in ischemic injury, the lower radiation dose of PET/MRI compared to PET/CT will be particularly valuable in the imaging of young patients and offers the possibility of explore new application directions ¹⁵⁹. For instance, a study performed in healthy adults (n = 14) compared two methods for detection of cerebral perfusion (H₂¹⁵O-PET and ASL-MRI), and found average global CBF by ASL-MRI and H₂¹⁵O-PET was not significantly different during hyperventilation and after acetazolamide. A prominent correlation was also found between the two methods in all states (slope = 1.01, R2 = 0.82) ¹⁶⁰. It also can be predicted that the discovery of new biochemical markers in central nervous system diseases will benefit from progress in the multi-parameter analysis and molecular probes available for PET/SPECT, such as small molecular bimodal probes ¹⁶¹, nano-sized bimodal probes ¹⁶², small molecular trimodal probes ¹⁶³, nano-sized trimodal probes ¹⁶⁴. All these indicate that integrated scanners have a promising future ^{157, 165, 166}.

CONCLUSION

The technology of PET targeting in brain metabolism and the development of tracers provide new tools to track and evaluate the brain’s pathophysiological changes after ischemic brain injury. Despite no existing evidence for an available PET-based prediction method, discoveries of new tracers are expected to provide more possibilities for the whole field.