Small animal PET with spontaneous inhalation of 15O-labelled oxygen gases: Longitudinal assessment of cerebral oxygen metabolism in a rat model of neonatal hypoxic-ischaemic encephalopathy

Perinatal hypoxic-ischaemic encephalopathy (HIE) is the leading cause of irreversible brain damage resulting in serious neurological dysfunction among neonates. We evaluated the feasibility of positron emission tomography (PET) methodology with 15O-labelled gases without intravenous or tracheal cannulation for assessing temporal changes in cerebral blood flow ( CBF ) and cerebral metabolic rate for oxygen ( CMRO2 ) in a neonatal HIE rat model. Sequential PET scans with spontaneous inhalation of 15O-gases mixed with isoflurane were performed over 14 days after the hypoxic-ischaemic insult in HIE pups and age-matched controls. CBF and CMRO2 in the injured hemispheres of HIE pups remarkably decreased 2 days after the insult, gradually recovering over 14 days in line with their increase found in healthy controls according to their natural maturation process. The magnitude of hemispheric tissue loss histologically measured after the last PET scan was significantly correlated with the decreases in CBF and CMRO2. This fully non-invasive imaging strategy may be useful for monitoring damage progression in neonatal HIE and for evaluating potential therapeutic outcomes.


Introduction
Perinatal asphyxia resulting in hypoxic-ischaemic encephalopathy (HIE) is a major cause of premature mortality worldwide, and 25% of survivors suffer from long-term neurological and neurodevelopmental impairments such as cerebral palsy. 1 The incident occurs in approximately 1.5 per 1000 live births in developed countries. 2The pathophysiology starts with primary energy failure during the hypoxic-ischaemic (HI) event, involving oxidative metabolism failure, cytotoxic oedema and excitotoxin accumulation.The subsequent secondary energy failure results in glutaminergic excitotoxicity, apoptosis and neuroinflammation. 35][6] The predominantly employed Rice-Vannucci neonatal rodent HIE model 7,8 has provided extensive advances toward developing promising therapies that could enhance the neuroprotective effect of hypothermia.
Given the large variation in brain damage severity in this animal model, 9 in vivo follow-up observations on an individual basis are vital for consistent progress monitoring.This interindividual variability is also the case clinically in patients and thus reliable prognostic indicators are highly desired, particularly in the acute phase.In neonates, cerebral oxidative metabolism exhibits a distinct elevation in the early maturational process. 10,11Oxygen supply derangement to the brain at this stage accordingly causes devastating outcomes including reperfusion-related injury after the initial decrease in cerebral blood flow (CBF) post-HI incident. 12,13The subsequent progressive brain damage or healing process may be detected by a direct physiological parameter for brain tissue viability, i.e. the cerebral metabolic rate for oxygen (CMRO 2 ).Thus, longitudinal monitoring of CBF and CMRO 2 in neonatal patients with their growth may aid in robustly assessing the disease course and potential treatment outcomes in association with normal developmental changes in brain energy metabolism.Moreover, this experimental approach should be minimally invasive, particularly for vulnerable neonatal rodents.
Studies have utilized the blood oxygenation leveldependent (BOLD) technique 14,15 to quantitatively assess blood oxygenation level in the superior sagittal sinus region in humans using magnetic resonance imaging, from which the oxygen extraction fraction (OEF) can be accurately quantified for the whole brain.A combination of BOLD with blood velocity assessment using the phase-contrast technique in the carotid artery region provides CBF and thus CMRO 2 for the whole cerebral region. 16,17However, regional assessment or parametric imaging of OEF and CMRO 2 remains uncertain because of unknown signal behaviour in the diseased brain due to a varied vasculature abnormality that may violate the essential assumptions required in the theoretical model of Yablonskiy. 18ositron emission tomography (PET) with 15 O-labelled gases has been the gold standard tool for decades for quantitative measurements of CBF, CMRO 2 and OEF.Despite the proven clinical value of 15 O-PET, its application to small animals requires exceptionally intensive procedures like tracheotomy and repeated arterial blood sampling 19,20 or the use of 15 O-oxygen-labelled haemoglobin. 21,22The complexity of the technical procedure and kinetic modelling approaches that require correction for recirculating 15   O-H 2 O are additional burdens.Recently, Temma et al. demonstrated a 15 O-PET method that enables spontaneous inhalation of 15 O-labelled gases without needing tracheotomy in mice. 23The technique enabled quantitative evaluations of CBF, CMRO 2 and OEF over a one-month time course in a mouse model of chronic cerebral hypoperfusion with bilateral common carotid artery stenosis.Currently, no other methodology allows a non-invasive, repetitive and fully quantitative assessment of cerebral metabolic variables together with tissue perfusion in the 3D domain in experimental settings.
This study depicted temporal changes in CBF, CMRO 2 and OEF after HI insult in rat neonates.We first assembled a novel integrated PET imaging system that enables sequential assessment of CBF, CMRO 2 and OEF in rat neonates with spontaneous inhalation of 15 O-labelled oxygen ( 15 O-O 2 ), carbon dioxide ( 15 O-CO 2 ) and carbon monoxide ( 15 O-CO) gases.We also examined the ageing variation in those parametric values in healthy controls at term-equivalent ages to define a comparable time course of normal postnatal brain development across HIE pups.

Study design
An integrated 15 O-PET system was developed enabling quantitative CBF, CMRO 2 and OEF assessment in small animals by spontaneous inhalation of 15

Animal subjects and ethical statement
For the first experiment that determined the k w , thirteen Sprague-Dawley rats consisting of six rats (five males and one female) at 40.3 AE 0.8 days old (d/o) with a body weight of 155.0 AE 30.1 g and seven (all males) at 25.7 AE 1.0 d/o with 65.6 AE 1.0 g were studied.
For the second experiment, nine male and nine female Sprague-Dawley rat pups at 9 d/o, whose brain developmental stage is considered equivalent to that of human infants at birth, 24,25 from two litters were randomly divided into HIE, non-operated and sham groups, as summarized in Table 1.
All animals were group housed under standard conditions (12- Integrated 15 O-PET imaging system A comprehensive PET system was assembled to sequentially assess CBF, CMRO 2 and OEF in small animals as an extension of previous work. 23The entire system consisted of an 15 O-oxygen-dedicated small cyclotron (C3D, IBA solutions, Louvain, Belgium), radiosynthesis and purification unit, gas chromatography for quality control, inhalation controller, dedicated PET scanners for each experiment and a radioactive gas evacuation system for effective scavenging of radioactive gases from the animal holder and radioactivity decay.
Radioactive gas production, purification/qualification and inhalation control. 15O was produced using the cyclotron by a 14 N(d,n) 15 O nuclear reaction.The target gas was 1.0% O 2 in N 2 and 1.0% CO 2 in N 2 , producing 15 O-O 2 and 15 O-CO 2 gases in the target, respectively.Each gas was transported to the hot cell, where 15 O-O 2 and 15 O-CO 2 gases were purified.The 15 O-CO gas was synthesized from the target 15 O-O 2 gas by a charcoal column heated at 950 C. The radiochemical and chemical purity was assessed each time before inhalation by gas chromatography (Model GC-2014, Shimazu, Kyoto, Japan) with Molecular Sieve and HayeSep Q columns (Figure 1(a)).
The quality-assured radioactive gases were then mixed with pure oxygen and air so that animals inhaled 20% oxygen during the entire period when animals were in the holder, even while radioactive gases were supplied or when the radioactivity supply rate was changed.The typical flow rate was 300 ml/min and the radioactivity supply rate was 250, 150 and 300 MBq/ min, corresponding to 15 O-O 2 , 15 O-CO 2 and 15 O-CO, respectively.Using multiple radioactivity detectors and mass flow controllers reliably controlled these parameters.The inhalation gas was mixed with evaporated isoflurane gas at a given concentration (typically 2-2.5%) before it was transferred to the animal holder.Dedicated 15 O-PET system animal holder.Two sets of animal holder assemblies were constructed to adapt to the HRRT PET scanner (CTI PET Systems, Knoxville, TN, USA) for the first experiment and the b-Cube PET scanner (Molecubes, Ghent, Belgium) for the second experiment.Both holders were sealed to prevent radioactive gas leakage.A corn-shaped facemask made of a fabric sheet (100 lm thickness) gently covered the snout to which the radioactive gases were supplied at a flow rate of 300 ml/min.The air inside the animal holder was evacuated at 7-10 l/min from the front through small holes surrounding the supply tube inside the animal holder.Fresh air was then passively carried to the animal holder from the tail to the front (Figure 1(b)).
Radioactive gas evacuation system.The scavenged radioactive gas from the animal holder was carried to 28 decay ducts (each 9.5 m long and 150 mm diameter, total volume of 4649 l) implemented in the facility building, allowing sufficient radioactivity decay before releasing it from the facility (Figure 1(c)).

Animal preparation and experimental procedures
First experiment (assessment of metabolized 15 O-H 2 O).Each animal assigned for the first experiment was anaesthetized with isoflurane (1.5-3%), and the left common carotid artery was cannulated with a polyethylene tube (SP-31, outer diameter: 0.8 mm, inner diameter: 0.5 mm, Natsume Seisakusho Co., Ltd, Tokyo, Japan).The cannula was secured by tight ligation around the cannulated vessel with 5-0 silk sutures (Ethicon, NJ, USA) and flushed with heparinized 0.9% NaCl to prevent blood clotting.
The animal was then moved to the animal holder and placed in the HRRT PET scanner.After ensuring the stabilisation of the animal under isoflurane anaesthesia, a 6 min attenuation correction scan was conducted.Next, a 6 min dynamic PET scan with a 5 s frame duration was started 15 s before initiation of 15 O-O 2 inhalation.Arterial blood samples (0.1 ml each) were manually obtained 4-5 times at 30 s intervals after 30 s of 15 O-O 2 inhalation from the catheter placed in the carotid artery.The blood samples were immediately divided into two samples.One sample was quickly centrifuged and the plasma radioactivity concentration was measured using a BeWell-Q3 well counter (Molecular Imaging Labo, Suita City, Japan), and the other sample was used to determine the whole blood radioactivity concentration.
Second experiment (sequential 15 O-PET imaging).HIE was induced in 9 d/o rats as described previously. 7,8Briefly, under isoflurane anaesthesia (1.5-3%), the left common carotid artery was exposed, ligated (6-0 silk sutures: Ethicon, NJ, USA) and then completely cut with electrocautery.After resting with their dam for 60 min, pups were exposed to 120 min of hypoxia in a chamber containing 8% oxygen and 92% nitrogen (P360 ProOx, BioSpherix, NY, USA) at 36-37 C rectal temperature.Pups were then returned to their dam again for recovery for over 60 min and kept in standard housing condition after sequential imaging studies.Sham pups underwent a similar surgical procedure without carotid artery ligation and hypoxic insult.Non-operated pups experienced no interventions.
PET and computed tomography (CT) scans were sequentially performed on eleven HIE, four nonoperated and three sham pups using b-Cube and X-Cube small animal scanners (Molecubes).During all scans, the animal was placed on a heating pad to maintain its body temperature, and respiration rate was continuously monitored using a sensor pad placed underneath the animal's chest.
After ensuring the stabilisation of the animal under isoflurane anaesthesia, a series of list-mode PET scans were acquired (Figure 1(d)).Two 14 min scans were started at the time of initiating continuous inhalation of both 15 O-O 2 and 15 O-CO 2 over 8 min.A separate 10 min scan was acquired from the time of initiating continuous inhalation of 15 O-CO over 3 min on one HIE pup on days 1, 7 and 14 before the 15 O-O 2 and 15 O-CO 2 scans to assess cerebral blood flow (CBV).Thereafter, the holder containing the animal was moved to the CT scanner for attenuation correction.Ten-to-fifteen-minute breaks separated PET scans to change cyclotron settings and allow radioactivity decay in the animal's body.Since the b-Cube PET scanner has a limited data transfer rate of 8 Mcps in the electric circuit, the radioactivity supply rate was adjusted not to exceed this limitation during each inhalation.After PET/CT scans, the animal was kept on a heating pad to recover from anaesthesia and returned to its regular housing.
These sequential scans were repeated four times, first within 6 h (day 0) and then at 24 h (day 1), 48 h (day 2) and 7 days (day 7) post-insult for HIE pups in scan group 1, and on days 1, 2, 7 and 14 post-insult for HIE pups in scan group 2. Age-matched non-operated pups were sequentially scanned four times at ages equivalent to those of HIE pups on days 0, 2, 7 and 14.Sham pups were scanned only on day 7 after the surgical intervention.The PET imaging time course with postoperative days and ages in the three subject groups is presented in Table 1.Image data analysis excluded animals that indicated poor radioactivity inhalation dose in the body during scanning or inconstant body position between 15 O-O 2 and 15 O-CO 2 scans.

Data processing
In the first experiment, PET images were reconstructed to generate dynamic images with 12 Â 5 s, 8 Â 15 s and 6 Â 30 s frames for 6 min using the 3D OSEM method with depth-of-interaction compensation. 27Whole blood time-activity curves were obtained from the left ventricular (LV) region in the reconstructed images.
Well-counter counts for the plasma and whole blood samples were normalized by acquisition time duration(s) and weights for each sample, followed by correction for radioactivity decay to the PET initiation time.They were further corrected for the crosscalibration factor determined using a 5 cm diameter and 10 cm long cylindrical phantom so that the well counterbased activity concentration for whole blood was equivalent to that of the LV values in the PET image, as we previously reported. 28Plasma counting rates were corrected for the plasma/whole blood radioactivity concentration ratio for 15 O-H 2 O. 29 The arterial blood concentration of 15 O-O 2 [Bq/ml] was then obtained by subtracting the 15 O-H 2 O concentrations from the whole blood radioactivity concentrations for each sample.
In the second experiment, PET images were reconstructed to generate dynamic images with a 1 min frame duration for the entire period of 14 min using 3D OSEM with depth-of-interaction compensation for 15 O-O 2 and 15 O-CO 2 and 10 min for 15 O-CO scans.Radioactivity decay correction was set uncorrected in the second experiment to avoid degradation of the pixel count precision in decay-corrected images for short-lived 15 O radioisotopes, which occurs specifically in the b-Cube scanner.

Data analysis
Determination of the production rate of metabolized 15 O-H 2 O in arterial blood.For the first experiment, a volume of interest (VOI) was selected on the LV cavity region in the dynamic PET images obtained after 30 s continuous inhalation of 15 O-O 2 using Carimas 2.1 (Turku PET Centre, Turku, Finland).The total blood arterial input function (AIF), A t ðtÞ, was determined by interpolating the LV time-activity curve (TAC), as previously validated. 28We evaluated the agreement between this and the well counter-derived whole blood radioactivity concentration curves in each animal experiment.
The following model formulation was employed to estimate the metabolized 15 O-H 2 O curve, A w ðtÞ, from the total blood AIF, A t ðtÞ: 25 A w ðtÞ ¼ k w Á A t ðtÞ e Àk w Át (1a) and then the 15  OEF and CBF (f) were both determined by simultaneous NLLSF of paired tissue time activity curves (tTACs) obtained with 15 O-O 2 and 15 O-CO 2 inhalation to the following equations. 29or the 15 O-O 2 inhalation PET data: and for the 15 O-CO 2 inhalation PET data: where the fractional volume of the arterial blood for 15 O-O 2 (V o ½ml=ml) and that for 15 O-H 2 O (V w ½ml=ml) were determined from CBV ½ml=ml assuming a fractional venous blood volume (F vein ) of 0.835 30 as: and p represents the tissue-to-blood partition coefficient of water (0.91 g/ml).The first-pass extraction fraction for 15 O-water is limited in high flow regions, such as the rat brain, because of the high blood flow in small animals.Thus, this study defines CBF as a flux of 15 O-water from the blood to the tissue space, as defined for CMRO 2 assessment using 15 O-oxygen.
The AIF for 15 O-O 2 (A o ðtÞ) and metabolized 15 O-H 2 O (A w ðtÞ) were estimated following equations (1a) and (1b), using the production rate of k w determined in experiment 1.The AIF for the 15 O-CO 2 inhalation experiment (A w t ð Þ) was determined from the LV TAC as The CMRO 2 values were calculated as the product of CBF (f), OEF and the oxygen content in arterial blood, [O 2 ] a , as: 30 where [O 2 ] a is calculated as a product of the oxygen volume confined per unit gram of haemoglobin (1.39 [ml/g]), the haemoglobin concentration in the blood [g/ml] and the fractional saturation of oxygen in arterial blood (e.g.98%).
Time course of parametric values.Time courses of absolute CBF (ml/min/g), CMRO 2 (ml/min/g) and OEF in ipsi-and contralateral brain VOIs were evaluated in the HIE, non-operated and sham groups over 14 days after the day of HI insult (day 0) or postnatal day 9. Age, group and hemispheric region dependency in CBF, CMRO 2 and OEF were evaluated at each time point and between animal groups.
The area (mm 2 ) of each hemisphere was measured on H&E-stained sections using CaseViewer 2.4 (3DHISTECH Ltd., Budapest, Hungary).The hemispheric volume of each brain was calculated by summing the hemispheric area of each brain slice and multiplying the sum of the section thickness.The mean ipsi-/contralateral hemispheric-volume ratios were compared between HIE and non-operated groups.
Additionally, the ipsi-/contralateral hemispheric-volume ratio in each HIE animal was compared with the ipsi-/contralateral ratios of CBF and CMRO 2 values obtained on day 14.

Statistics
Results are presented as mean AE SD.The student's paired or unpaired t-test was applied to compare two variables as appropriate.Pearson's analysis was used to test for correlation with GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA).P-values <0.05 were considered statistically significant.Image data analysis was blinded to avoid any subjectivity.

Results
Five of the thirteen rats that were assigned to the first experiment were excluded from the data analysis because of unsuccessful arterial blood sampling or extremely low 15 O-O 2 inhalation dose during the PET scan.Otherwise, all well counter-based whole-blood radioactivity concentrations agreed with the PET image-derived LV TAC in all cases and were further validated by an earlier study using the same experimental setup. 28Fitting the predicted metabolized 15 O-water TAC to the manually separated 15 O-water TAC resulted in average k w values of 0.33 AE 0.06, 0.33 AE 0.07 and 0.33 AE 0.07 min À1 , corresponding to younger (n ¼ 4, 25.5 AE 1.0 d/o, 63.0 AE 0.8 g), older (n ¼ 4, 40.0 AE 0.0 d/o, 166.5 AE 20.6 g) and all animals (n ¼ 8, 32.8 AE 56.9 d/o, 114.8 AE 56.9 g), respectively.No significant difference was found in k w values between the two age groups.From these results, we defined a k w value of 0.33 min À1 in the calculation of CBF, CMRO 2 and OEF for all animals in the second experiment.
The whole-body distribution of 15 O-gases in a rat scanned on days 1, 7 and 14 post-HI insult is shown in Figure 2 as sagittal PET/CT images accumulated for a period of 0-10 min, 0-5 min and 0-8 min after initiation of 15 O-CO, 15 O-O 2 and 15 O-CO 2 gas inhalation, respectively.Radioactivity was visible on the surface of the snout inside the facemask (arrow a), in the gas supply tube (arrow b) and on the inner surface of the animal holder (arrow c).These radioactivity spots were negligibly small and did not significantly impact image data analysis.Radioactivity in the brain relative to the heart was lower with 15 O-O 2 than that with 15 O-CO 2 .Radioactivity in the lung region was greater in 15 O-O 2 images than 15 O-CO 2 images.
Figure 3 shows 15 O-gas accumulation in a HIE brain scanned on day 14 (Figure 3(a) to (c)) and LV TAC and tTACs (Figure 3(d) to (f)) for the selected VOIs defined in the 15 O-CO image (Figure 3(a)).The decreased radioactivity in the ipsilateral hemisphere is clearly visible in both 15 O-O 2 and 15 O-CO 2 PET images (white arrows), while 15 O-CO images showed no interhemispheric difference at any time point after HI insult.LV radioactivity concentrations were nearly equal between the 15 O-O 2 and 15 O-CO 2 inhalation scans although the supply rate of 15 O-CO 2 radioactivity was lower (200 MBq/ml) than that of 15 O-O 2 (500 MBq/ml).The peak values of TACs in both the ipsi-and contralateral brain regions were lower during 15 O-O 2 inhalation compared with 15 O-CO 2 , as seen in Figure 3(b) and (c), respectively.
A representative result from the NLLSF analysis for fitting CBF and OEF simultaneously to two tTACs of 15 O-O 2 and 15 O-CO 2 is shown in Figure 4. Three AIFs were determined during 15 O-O 2 inhalation PET imaging from total blood (A t t ð Þ) and metabolized 15 O-H 2 O (A w ðtÞÞ and 15 O-O 2 (A o ðtÞ), as estimated using equations (1a) and (1b), respectively (Figure 4(a)).A t t ð Þ continued to increase during the 15 O-O 2 inhalation period of 8 min.However, A o ðtÞ increased only until 3 min, plateauing at approximately 60% of the peak of A t t ð Þ, while A w ðtÞ increased until the end of 15 O-O 2 inhalation.Figure 4(b) demonstrates that the total tTAC (C t t ð Þ) as given in equation (2a) reproduced well the measured tTAC.The additional three tTACs shown in this figure correspond to each of three terms shown in equation (2a), i.e. the response for 15 . This figure also shows that while the total tTAC (C t (t)) continued to increase during the 15 O-O 2 inhalation period of 8 min, C o t ð Þ plateaued at approximately 3 min until the end of 15 O-O 2 inhalation.
In contrast, C w t ð Þ exceeded the level of that for C o t ð Þ around 4 min after 15 O-O 2 initiation.Figure 4(c) shows the AIF for 15 O-CO 2 inhalation, in which the total blood activity is equal to the AIF for 15 O-H 2 O (A t (t) 5 A w (t)), whereas Figure 4(d) presents the fit results for the 15 O-CO 2 inhalation scan given the formulation in equation (2b).The contribution of vascular radioactivity is negligibly small because of the small vascular component and large CBF values, as estimated in equation (3b).Averaged CBVs obtained from the 15 O-CO scan were 0.049 AE 0.023 and 0.053 AE 0.026 ml/ ml corresponding to the ipsi-and contralateral regions, respectively.Because of the small value, V w was neglected and the calculated V o was given as determined.
The temporal profile of mean quantitative hemispheric CBF, CMRO 2 and OEF values (Figure 5(a) to (c), respectively), and mean ipsi-/contralateral ratios of CBF, CMRO 2 and OEF (Figure 5(d) to (f), respectively), as a function of days post-HI insult or equivalent days after birth (d/o) were determined in HIE, nonoperated and sham subject groups (Figure 5).Nonoperated pups showed significant increases from 9 to 23 d/o in both CBF (p < 0.01) and CMRO 2 (p < 0.05) by approximately 2.5-fold and 2-fold increases, respectively (Figure 5(a) and (b)).CBF in the ipsilateral brains of HIE pups was significantly reduced at the lowest level on day 2 (p < 0.0001, compared to day 1) and then increased on day 7 (p < 0.01) and day 14 (p < 0.01) compared to day 2. CMRO 2 values in the ipsilateral hemispheres of HIE animals also reached the lowest levels on day 2 (p < 0.01, compared to day 1) and significantly increased on day 14 from day 2 (p < 0.05).Pearson's test showed a correlation between age and CBF (p ¼ 0.0005) or CMRO 2 (p ¼ 0.0003) in non-operated pups.The ipsilateral hemispheres of HIE pups indicated a linear relationship between CBF (p ¼ 0.0007) or CMRO 2 (p ¼ 0.0125) and postoperative days from day 2 to day 14.OEF showed no time dependent or interhemispheric differences in any animal group except for a significant increase on day 2 in the ipsilateral hemisphere compared to contralateral hemisphere (p < 0.0001) in HIE pups (Figure 5(c)).The ratios of CBF and CMRO 2 in HIE pups reached the lowest on day 2 and CMRO 2 ratios showed no temporal changes thereafter (Figure 5(d) and (e)).The ratio of OEF in HIE animals was significantly increased on day 2 (p < 0.01, compared to day 1) and then decreased on day 7 (p < 0.05) (Figure 5

(f)).
A comparison of 15 O-O 2 PET/CT images and histology results from a HIE rat on day 14 and an agematched non-operated rat is shown in Figure 6  Additionally, increased microglial activation was displayed by Iba1 outside the margins of the infarct cores, where small reductions in 15 O-O 2 accumulation were observed.The mean ipsi-/contralateral hemisphericvolume ratio of HIE pups showed a 62.6% reduction compared with non-operated pups (p < 0.0001) (Figure 6(b)).The hemispheric volume ratios of HIE pups indicated a linear relationship with the ratios of hemispheric CBF (p ¼ 0.0112) and CMRO 2 (p ¼ 0.0091) values measured in each brain VOI (Figure 6(c) and (d), respectively).

Overview of the findings
The comprehensive 15 O-oxygen PET system for small animals developed in this study provided quantitative CBF, CMRO 2 and OEF values non-invasively by spontaneous inhalation of 15 O-O 2 and 15 O-CO 2 without tracheotomy or intravascular cannulations required using an intravenous 15 O-H 2 O administration approach.
The sophisticated radioactive gas supply and scavenging systems implemented for two PET scanners enabled good quality dynamic PET images.The physiological model-based estimation of metabolized 15 O-H 2 O in arterial blood was essential as it avoids labourintensive procedures and minimizes uncertainties in A w (t) and A o (t) values when assessing individual blood samples with small volumes.Simultaneous NLLSF to a pair of dynamic PET images for 15 O-O 2 and 15 O-CO 2 enabled reliable estimation of parametric values.This system allowed repetitive measurements even on consecutive days initiated immediately after the HI insult in vulnerable infant rats.
The technique detected time-dependent changes in CBF, CMRO 2 and OEF associated with postnatal age and post-HI injury state for different brain regions and subject categories.Important findings in this study are the following: (a) remarkable reductions in CBF and CMRO 2 on day 2 after HI insult; (b) continuous recoveries after day 2 in the affected hemisphere in parallel to the increases in the contralateral hemisphere and (c) similar increase rates in CBF and CMRO 2 with age-matched controls (Figure 5).The non-operated rats displayed a relatively large increase from 9 to 23 d/o, which was similar to a previous report demonstrating that whole brain CBF and CMRO 2 , measured by H 2 clearance and the arteriovenous difference in O 2 content, dramatically increased in infant rats from 10 to 20 d/o. 31 Our finding is also consistent with a clinical PET study that indicated 3-4 times lower oxygen consumption in term infants than adults as a reflection of brain maturation. 32This difference is considered attributed to the escalating cerebral energy demand compensating for the increased Naþ pump activity in neuronal cells 33,34 for structural and functional maturational processes at the ages examined in this study.
We confirmed that both LV TAC for 15 O-oxygen scan in the first experiment agreed with those obtained by the well counter, which is consistent with an earlier finding 28 despite the smaller average animal size in this study.This may be attributed to the contribution of the myocardium surrounding the LV chamber, which has a high flow value, resulting in a similar tTAC in the myocardial wall as the LV blood TAC.We also expect a similar phenomenon in the second experiment, in which a high-resolution b-Cube scanner was employed.However, this remains a hypothesis requiring further verification.
In HIE animals, CBF in the ipsilateral hemispheres showed a transient increasing tendency on day 1 postinsult , which may be consistent with a previous report, indicating that hyperperfusion occurring 24 h after HI intervention measured by the laser speckle method could be associated with the more severe tissue damage outcome. 13The subsequently marked reductions in both CBF and CMRO 2 on day 2 may be attributed to the neuronal cell death and apoptosis reportedly maximized at 24-72 h after neonatal HI insult. 8,35,36The pathophysiological background of the continuous increases in CBF and CMRO 2 in the ipsilateral VOI after day 2 may also be an important finding.Of note is that the ipsi-/contralateral ratio of CMRO 2 in HIE brains showed no significant increase after day 2. Thus, one may hypothesize that the surviving neuronal cells in the injured region keep growing while the damaged tissues with no energy consumption persisted.
Histological tissue losses measured after the last PET scan showed spatial agreement with severely decreased CBF and CMRO 2 areas (Figure 6).In contrast, CMRO 2 reduction was milder in the area where increased microglial activation was detected by Iba1 surrounding the infarct core region.Post-ischaemic inflammation reportedly modulates the exacerbation of neural damage in neonatal HIE, further suggesting a unique immune response in the immature brain. 37,38The application of repetitive imaging techniques of 15 O-gas inhalation PET along with neuroinflammation-targeting PET using 18 kDa translocator protein (TSPO) would be useful for evaluating the age-and time-dependent effects of microglia-mediated inflammatory response on delayed neuronal damage expansion.

Methodological advantages
15 O-O 2 gas is the sole tracer that can trace the kinetics of oxygen molecules and oxygen metabolism.It was shown in both the first and second experiments that 15 O-O 2 is metabolized quickly in neonatal rats so that the metabolized 15 O-H 2 O replaces 15 O-O 2 in arterial blood shortly after starting 15 O-O 2 inhalation, as typically seen in Figure 4(a).Since the 15 O-H 2 O extraction rate to the brain in the capillary bed is over two-fold higher than that of 15 O-O 2 , the contribution of metabolized 15 O-H 2 O to the brain (the second term in equation ( 2a)) exceeds the level of that of the original Simultaneous NLLSF is considered to be a reliable approach for estimating both CMRO 2 and CBF defined in equations (2a) and (2b), respectively.Using a single rate constant, k w , that takes into account the kinetics of 15 O-H 2 O in parallel to that of 15 O-O 2 plays an important role in this study in terms of stabilizing NLLSF estimation.This approach detected small but significant increases in CMRO 2 during cognitive stimulation that increased CBF to a greater amount; 29 however, another approach failed to detect an increase in CMRO 2 where the metabolized 15 O-H 2 O was neglected. 39

Limitations
There are several limitations to this study.Firstly, the animal model used exhibits hemispheric HI injury generated by unilateral ligation of the common carotid artery with subsequent global hypoxic insult.Considering the impact of interhemispheric collateral circulation, the use of ipsi-/contralateral ratios of CBF and CMRO 2 may have been biased, especially in the acute phase.Secondly, anaesthesia is a potentially confounding factor in measuring cerebral haemodynamics since isoflurane can derange CBF autoregulation. 40urthermore, isoflurane may possess neuroprotective properties, and thus its exposure time may be correlated with decreased brain infarction in the Rice-Vannucci model. 41Another anaesthetic agent may be a suitable replacement, especially for the study of cerebral haemodynamics or ischaemic disease.It should also be mentioned that limited vital monitoring was performed because of the difficulty in placing highly sensitive equipment in the limited space of the animal holder to detect stable signals from small pups.Moreover, the physiological condition is unlikely constant during the entire PET imaging protocol because of the long scan period, and systematic errors may be attributed to transient changes in CBF and/or CMRO 2 .
A standardized CBV value determined from limited animal scans of a single rat was used to calculate OEF and CMRO 2 .The contribution of CBV to OEF and CMRO 2 was small and around 1% in the present data both in the control and HIE VOIs, partly attributed to the observed higher CBF and CMRO 2 values than those in humans.However, CBV could be increased, for example, in chronic ischaemic lesions, 42 during haemodynamically altered conditions, 43 and even in neurodegenerative diseases. 44Kinetic fitting of V o and V w with CMRO 2 and CBF as proposed by previously 45 is highly desired as an alternative not only to remove the 15 O-CO scan but also to improve the accuracy by removing the need for assuming a fixed arterial-to-venous blood volume ratio.
A k w of 0.33 AE 0.07 min À1 was obtained in the first experiment for 24 and 40 d/o rats, whereas it was approximately twice in an earlier study on 7-8-weekold rats. 45One reason for this discrepancy could be the different tracer administration procedures, i.e. earlier study used intravenous injection of 15 O-O 2 -labelled oxyhaemoglobin instead of gaseous 15 O-oxygen.Further systematic investigations are warranted.
We applied the average k w value obtained from the first experiment to all analyses for the second experiment.To estimate the effects of AE20% variation in k w , the CBF and CMRO 2 values were calculated by changing k w by AE20% using the tTAC obtained from the day 14 study, which was the worst situation because of the highest CBF and CMRO 2 values.The change in k w by þ20% caused changes in CBF by À3.7% and À7.5%, whereas a change by À20% resulted in þ0.0% and þ2.9%, corresponding to the ipsilateral and contralateral hemispheres, respectively.Similarly, changes in k w by þ20% and À20% caused changes in CMRO 2 by þ8.4% and þ3.1% and À4.1% and À1.0%, respectively, in the ipsilateral and contralateral hemispheres.The range of changes in CBF and CMRO 2 were not large compared with the range of the time courses shown in Figure 5.However, a more systematic error propagation study is planned for future work.

Future implications
The present 15 O-oxygen PET system can be further improved if the entire examination duration is shortened.A PET scanner with better counting rate performance is essential for employing a single PET scan protocol during the dual inhalation of 15 O-O 2 followed by 15 O-CO 2 with a short interval, as demonstrated previously. 46,47This protocol can reduce the total anaesthetisation time and minimize possible physiological changes during PET scans.The intrinsic vascular radioactivity correction could also be compensated without the need for an additional 15 O-CO scan.Continued research in this study would be of great value to evaluate possible pharmacological interventions and the early prognostic value of 15 O-PET for neonatal HIE with a larger sample size.Such technical innovations would also facilitate the broader use of 15 O-PET in preclinical and clinical applications, including in neonatal patients.
It is also important to establish a technique that can determine the appropriate k w value for the given subject groups without the need for blood sampling.The indirect calorimeter would allow assessments of k w through whole-body oxygen metabolism by measuring O 2 consumption and CO 2 production, as demonstrated earlier. 45More sophisticated equipment, such as a ventilated chamber for small animals, 48,49 may also be a possibility.

Conclusions
We demonstrated that the present 15 O-gas inhalation PET system quantitatively assesses temporal changes in CBF and CMRO 2 associated with acute cerebral perfusion derangement and tissue damage evolving 48 h after HI insult in rat neonates.The technique also depicts subsequent pathological progression over 14 days interrelated with escalating cerebral energy metabolism along with the maturation process.This completely non-invasive imaging strategy may be of value for developing early therapeutic interventions and their response monitoring in neonatal HIE with highly individualized in vivo follow-up.

Figure 1 .
Figure 1.A schematic of the comprehensive PET system newly assembled in this study and consisting of three parts (a-c).(a) Radioactive gas production and purification/qualification system consisting of a cyclotron, target chamber, radioactive gas production/ purification unit and inhalation controller.A charcoal column heated at 950 C (CC950) was utilized for 15 O-CO synthesis.For 15 O-O 2 purification, two columns filled with charcoal at room temperature (CC) and another with MS were used.For the purification of 15 O-CO 2 , a column with charcoal heated at 450 C (CC450) was used.Gas chromatography (GC) was implemented with two columns with HayeSep Q and Molecular sieve (MS) to assess chemical and radiochemical purity.(b) Animal holder dedicated to 15 O-oxygen inhalation PET.A corn-shaped facemask made of a thin fabric sheet is placed over the snout of the animal.Radioactive gases are carried to the facemask for spontaneous inhalation.The animal holder is well sealed and actively scavenged from the front surrounding the radioactive gas supply tube.Fresh air is passively carried from the back into the animal holder.(c) Radioactive gas evacuation system.Approximately 380 l/min of evacuation carries scavenged radioactive gases from the animal holder to the decay ducts installed in the institute basement.Approximately 24 min of transit time in the ducts allows for radioactivity decay by a factor of 4000 before it is released from the chimney to the outside and (d) the 15 O-PET/CT protocol consists of three separate PET scans each with 15 O-CO, 15 O-CO 2 and 15 O-O 2 inhalation conducted in a sequential PET/CT study in the three subject groups of HIE, non-operated and sham.CT: computed tomography; HIE: hypoxic-ischaemic encephalopathy; PET: positron emission tomography; Radio-gas: radioactive gas.
. Infarct regions measured by H&E staining in the HIE brain coincided with areas showing extremely decreased accumulation of 15 O-O 2 in PET images (Figure 6(a)) .

Figure 2 .
Figure 2. Representative sagittal whole-body PET/CT images of a HIE rat placed in the animal holder on days 1, 7 and 14 post-HI insult.Summed images over 0-10 min after 15 O-CO, 0-5 min after 15 O-O 2 and 0-8 min after 15 O-CO 2 inhalation were displayed.White arrows (a-c) indicate the effects of radioactivity accumulated in the snout (a), emitted from the radioactive gas supply tube (b) and on the inner surface of the animal holder (c).CT: computed tomography; HIE: hypoxic-ischaemic encephalopathy; PET: positron emission tomography.

Figure 3 .
Figure 3. Representative transaxial, coronal and sagittal PET/CT images of the brain of a HIE rat on day 14 post-insult.PET images are summed over 0-10 min after 15 O-CO (a), 0-5 min after 15 O-O 2 (b) and 0-8 min after 15 O-CO 2 (c) inhalation.The lesion is indicated by the white arrows in the left hemisphere in both 15 O-O 2 (b) and 15 O-CO 2 images (c).VOI locations in the LV, the lesion in the ligated-side (ipsilateral) hemisphere and its symmetrical position (contralateral) are also indicated on the 15 O-CO images (a).The measured tTACs in ipsilateral (square) and contralateral (diamond) brain regions, and LV TAC (circle) from the same animal successively administered 15 O-CO (d), 15 O-O 2 (e) and 15 O-CO 2 (f) are also presented.Note that LV TAC for 15 O-CO is presented as one-tenth value (d).CT: computed tomography; HIE: hypoxic-ischaemic encephalopathy; LV, left ventricle; PET: positron emission tomography; TAC: time-activity curve; tTAC: tissue TAC; VOI: volume of interest.

Figure 4 .
Figure 4.A typical example of AIFs (a) and the results from simultaneous fitting to tTACs (b) for15 O-O 2 , AIFs (c) and results from simultaneous fitting to15 O-CO 2 tTACs (d) obtained from a HIE rat on day 14 after the insult (same case shown in Figure3).(a) Three input functions are given for the15 O-O 2 fitting analysis, i.e. the total blood AIF (A t ðtÞ) (solid line), AIF for15 O-O 2 (A o ðtÞ) (dashed line) as calculated by equation (1b) and AIF for metabolized15 O-H 2 O (A w ðtÞ) (dash-dotted line) as calculated by equation (1a) with the production rate of15 O-H 2 O, k w , determined from the first experiment.The observed tTAC (closed circles) for the15 O-O 2 scan was well reproduced by NLLSF (solid line) that consisted of the three components shown in (c), as estimated as a response to the15 O-O 2 AIF (dashed line), a response to the AIF for15 O-CO 2 (dash-dotted line) and the contribution of the blood radioactivity (dotted line), corresponding to the first, second and third terms of equation (2a), respectively.Of note is that the AIF for15 O-O 2 reached to a plateau at approximately 3 min after the initiation of continuous inhalation of15 O-O 2 , while the contribution of metabolized15 O-H 2 O continued to increase, exceeding the contribution of15 O-O 2 and reaching approximately 60% of the total radioactivity concentration at the end of15 O-O 2 inhalation.In contrast, the AIF for15 O-CO 2 inhalation consists of only a single component of15 O-H 2 O that is equal to the whole blood TAC, i.e.C t ðtÞ.AIF: arterial input function; HIE: hypoxic-ischaemic encephalopathy; NLLSF: non-linear least square fitting; TAC: time-activity curve; tTAC: tissue TAC.

15 O
-O 2 during the continuous inhalation of 15 O-O 2 (Figure 4(b)), at 3-5 min after 15 O-O 2 inhalation initiation.To compensate for the contribution of metabolized 15 O-H 2 O, the additional administration of 15 O-CO 2 (equivalent to intravenous 15 O-H 2 O) was applied as many PET methods that utilized 15 O-O 2 inhalation protocol.

Figure 5 .
Figure 5.The mean temporal values of (a) CBF (ml/min/g), (b) CMRO 2 (ml/min/g) and (c) OEF in VOIs selected on ipsilateral (closed circle) and contralateral (open circle) hemispheres on days 0, 1, 2, 7 and 14 post-insult in HIE, at the corresponding age of 9, 11, 16 and 23 d/o in non-operated (ipsilateral: closed square; contralateral: open square) and on day 7 after the surgical operation in sham pups (ipsilateral: closed diamond; contralateral: open diamond).The ipsi-to contralateral ratios of CBF (d), CMRO 2 (e) and OEF (f) of the three subject groups on each imaging day (HIE: circle-dot; non-operated: square-dot; sham: diamond-dot) are also presented.Values are expressed as mean AE SD and those in a pair of each imaging time point or each hemisphere are compared by student's paired t-test.Asterisks shown above closed circles indicate significant differences between each hemisphere.Ipsilateral HIE brains presented significantly lower values than contralateral hemispheres at all time points in CBF (a) and except for day 0 in CMRO 2 (b).OEF in ipsilateral HIE brains showed a significant increase on day 2 compared to contralateral hemispheres (c).*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, NS; not significant.CBF: cerebral blood flow; CMRO 2 : cerebral metabolic rate for oxygen; d/o: day old; OEF: oxygen extraction fraction; HIE: hypoxic-ischaemic encephalopathy; VOI: volume of interest.

Figure 6 .
Figure 6.(a) Representative PET/CT images of the transaxial view and each coronal slice at the striatum (1) (3) and hippocampus level (2) (4) of the brains of a HIE pup on day 14 and an age-matched non-operated pup are shown as summed images over 0-5 min after 15 O-O 2 inhalation.Coronal images stained with H&E and Iba1 of the same animals at equivalent slicing levels are also displayed.Note that the area of the histological tissue loss in H&E images as indicated by grey arrows in the HIE animal is consistent with the defect region in 15 O-O 2 PET images as indicated by white arrows.Increased microglial activation is also visible around the infarct tissues in Iba1 images as indicated by grey arrows.(b) Ratios of the ipsilateral to contralateral hemispheric volume are presented as mean AE SD, displaying a significantly decreased mean ratio in HIE (****p < 0.0001) compared to non-operated pups (n ¼ 5 and n ¼ 4, respectively).(c,d) The ipsi-/contralateral ratio of CBF and CMRO 2 on day 14 in each HIE pup showed significant correlations with the ratio of the ipsi-/contralateral hemispheric volume in histology measured on the same day (p ¼ 0.0122 and p ¼ 0.0091, respectively).CBF: cerebral blood flow; CMRO 2 : cerebral metabolic rate for oxygen; CT: computed tomography; H&E: haematoxylin-eosin; HIE: hypoxic-ischaemic encephalopathy; PET: positron emission tomography.
26light/dark cycle, temperature 21 AE 3 C, humidity 55 AE 15%) with ad libitum access to RM3(E) rodent diet (SDS, UK) and tap water at the Central Animal Laboratory, University of Turku, Finland (UTUCAL).The study was approved by the National Project Authorization Board of Finland (ESAVI/20863/2018) adhering to the 3 R principles in accordance with Finnish National legislation (Act 497/ 2013, Decree 564/2013) and EU Directive 2010/EU/63 on the protection of animals used for scientific purposes.All reported experiments complied with the ARRIVE guidelines.26

Table 1 . Age, body weight and number of animal subjects on each PET imaging day in the second experiment. Numbers in parentheses denote values after exclusion from the image data analysis.
O-O 2 AIF, A o ðtÞ, as shown below: 15o ðtÞ ¼ A t ðtÞ À A w ðtÞ (1b)where k w [min À1 ] is the production rate of the metabolized15O-H 2 O in arterial blood after15O-O 2 inhalation.Next, we determined k w for each PET scan so that the simulated A w ðtÞ in equation (1a) reproduced the well counter-derived metabolized 15 O-H 2 O curve by means of non-linear least square fitting (NLLSF).A w ðtÞ and A o ðtÞ in the second experiment were then calculated from PET-derived A t ðtÞ with the given k w following equations (1a) and (1b).Calculation of parametric values.Parametric values of CBV, CBF, OEF and CMRO 2 were calculated for two VOIs selected in the brain. CBs were first determined from a 15 O-CO image for six examinations for both ipsi-and contralateral VOIs by normalising the average radioactivity concentrations in the cerebral VOIs by that in the LV VOI.The averaged CBV value was used in the following calculation.