Spectroscopy of reperfused tissue after stroke reveals heightened metabolism in patients with good clinical outcomes

Abstract

The aim of acute stroke treatment is to reperfuse the penumbra. However, not all posttreatment reperfusion is associated with a good outcome. Recent arterial spin labeling (ASL) studies suggest that patients with hyperperfusion after treatment have a better clinical recovery. This study aimed to determine whether there was a distinctive magnetic resonance spectroscopy (MRS) metabolite profile in hyperperfused tissue after stroke reperfusion therapy. We studied 77 ischemic stroke patients 24 hours after treatment using MRS (single voxel spectroscopy, point resolved spectroscopy, echo time 30 ms), ASL, and diffusion-weighted imaging (DWI). Magnetic resonance spectroscopy voxels were placed in cortical tissue that was penumbral on baseline perfusion imaging but had reperfused at 24 hours (and did not progress to infarction). Additionally, 20 healthy age matched controls underwent MRS. In all, 24 patients had hyperperfusion; 36 had reperfused penumbra without hyperperfusion, and 17 were excluded due to no reperfusion. Hyperperfusion was significantly related to better 3-month clinical outcome compared with patients without hyperperfusion (P=0.007). Patients with hyperperfusion showed increased glutamate (P<0.001), increased N-Acetylaspartate (NAA) (P=0.038), and increased lactate (P<0.002) in reperfused tissue compared with contralateral tissue and healthy controls. Hyperperfused tissue has a characteristic metabolite signature, suggesting that it is more metabolically active and perhaps more capable of later neuroplasticity.

Introduction

The main aim of acute ischemic stroke therapies such as alteplase is to achieve early reperfusion of acutely ischemic tissue, thereby salvaging tissue at risk and minimizing the extent of infarction. Successful reperfusion may also be an important precursor to enable interventions aimed at enhancing recovery in the subacute stage.¹ Successful reperfusion can be identified by evidence of restored regional blood flow on angiographic or perfusion imaging, in the absence of infarction. However, penumbral tissue that appears to be salvaged from infarction on brain imaging may still have sustained injury at the cellular level. Selective neuronal necrosis² and regionally disturbed metabolism such as raised lactate³ and decreased N-acetylaspartate concentrations are well described and likely reflect neuronal impairment⁴ or oxidative stress after reperfusion therapy. Such metabolic disturbances may alter recovery potential from the initial ischemic event. How these metabolic and tissue processes are related to patient outcome after human ischemic stroke is poorly understood, and may provide insight into neural repair mechanisms.

It has been recognized for over two decades that peri-lesional hyperperfusion in the early postinfarct phase is associated with favorable clinical outcome.⁵ The mechanisms underlying this longstanding observation, however, remain unclear. Using noncontrast perfusion imaging, arterial spin labeling (ASL), repeated assessments of perfusion in the peri-infarct zone can be acquired.⁶ The aim of this study was to investigate potential metabolic processes that may underpin peri-infarct hyperperfusion detected using ASL. We hypothesized that hyperperfusion may be a reflection of enhanced postischemic metabolism and that this may potentially lead to, or be a marker of enhanced recovery.

Materials and methods

We studied acute consecutive middle cerebral artery territory ischemic stroke patients treated with intravenous recombinant tissue plasminogen activator within 4.5 hours of stroke onset between January 2012 and April 2013. The standard acute ischemic stroke protocol at the John Hunter Hospital for all patients included a pretreatment acute multimodal computed tomography (CT) and follow-up magnetic resonance (MR) imaging (MRI) at 24 hours. Intravenous recombinant tissue plasminogen activator was administered in accordance with local guidelines (HNE health NSW thrombolysis guidelines), as well as additional advanced CT imaging. The National Institutes of Health Stroke Scale (NIHSS) was used to assess clinical stroke severity at baseline and 24 hours, and the modified Rankin score (mRS) was recorded at 3 months as a measure of functional outcome. Ethics approval for this study was granted by the Hunter New England Human Research Ethics Committee, and informed consent was obtained from all participants.

Stroke Imaging

Acute CT imaging was acquired on a Toshiba Aquilion (Toshiba, Tochigi, Japan) One cone beam CT. Whole-brain noncontrast CT was performed in one rotation (detector width 16 cm). Next, a four-dimensional time-resolved whole-brain CTA (computed tomography angiography) and whole-brain computed tomography perfusion (CTP) were acquired simultaneously. For the CTA-CTP, 40 mL of contrast agent (Ultravist 370; Bayer HealthCare, Berlin, Germany) was injected at 6 mL/s chased by 30 mL of saline (56.0 mGy; DLP 896.3 mGy cm).⁷ Perfusion data were processed using a single value deconvolution algorithm with delay and dispersion correction.⁸

Magentic resonance imaging was performed on a 3-T MRI 24 hours after stroke onset (Siemens Magnetom Verio, Erlangen, Germany). The stroke MRI protocol included an axial isotropic diffusion-weighted echoplanar spin-echo sequence (DWI), time of flight MR angiography, bolus-tracking perfusion-weighted imaging (PWI),⁹ and four, single voxel spectroscopy scans using point resolved spectroscopy (repetition time 2,000 ms, echo time 30 ms, 2 cm × 2 cm × 2 cm=8 mL) in two locations with and without water suppression. Arterial spin labeling was performed using a pulsed sequence with a 32 channel receive only head coil (ASL -Q2TIPS: repetition time 2,500 ms, inversion time₁ 500 ms; inversion time_1s 1,500 ms; inversion time₂ 1,700 ms; FOV 240 × 240 mm, matrix 64 × 64). This acquired nine slices at 8 mm thickness with 28 repetitions for a scan time of 4:02 minutes. Perfusion-weighted imaging was performed after ASL and Magnetic resonance spectroscopy (MRS) acquisitions.

Control Participants

Twenty age-matched healthy controls were recruited from a community research volunteer register. Healthy controls underwent the same MRI protocol as the stroke patients. Control patient spectroscopy voxels were placed in the dominant middle cerebral artery territory gray matter.

Image Processing

Perfusion and diffusion imaging analysis was performed using MIStar (Apollo Medical Imaging Technology, Melbourne, VIC, Australia). For perfusion imaging, MIStar automatically performs motion correction and selects an arterial input function from the anterior cerebral artery and a venous output function from the sagittal sinus. These were checked manually for accuracy and repositioned if necessary. Chronic infarcts/gliosis and cerebrospinal fluid regions were automatically detected by a Hounsfield unit threshold and not included in the analysis. Computed tomography perfusion imaging thresholds were based on previous studies, which identified the volume and location of the acute penumbra (delay time >2 seconds) and infarct core (delay time >2 seconds and cerebral blood flow (CBF) <40%).^10,11

Major reperfusion was defined as a reduction in the acute CTP mean transit time (MTT) lesion (MTT >145% threshold) to 24 hours PWI-MTT (MTT >145% threshold)¹² perfusion lesion volume of >80%.¹³ Patients without evidence of major reperfusion on 24 hours post recombinant tissue plasminogen activator MR perfusion imaging were excluded from further analysis. Crossover of CTP-MTT and PWI-MTT modalities has been shown to have excellent intertechnique agreement.¹⁴

Hyperperfusion on ASL was defined as a CBF of >130% (reference to contralateral hemisphere) in the previously hypoperfused region as previously described.⁸

Spectroscopy Processing

Spectroscopy voxels were placed in tissue that was identified as hypoperfused on acute CTP (delay time>2 seconds) but which was reperfused on 24 hours ASL (>80% perfusion lesion reduction at 24 hours). A second MRS voxel was placed contralateral to the first voxel (Figure 1). Voxels were placed in the peri-infarct gray matter that was not hyperintense on 24 hours DWI. Thus, infarcted tissue was not included in the voxels.

Figure 1.

Placement of magentic resonance (MR) spectroscopy (MRS) voxels using acute computed tomography (CT) perfusion (CTP) imaging as a reference. Identification of hyperperfused tissue required acute CTP imaging (Top row, first map). Next, 24 hours magnetic resonance imaging (MRI) using arterial spin labeling (ASL-CBF, bottom row, first image) and diffusion-weighted imaging (DWI, bottom row second imaging) was performed to measure reperfusion and penumbral salvage. MRS voxels were placed in the reperfused region on ASL that did not show infarct on DWI (bottom row second imaging, yellow box) to measure metabolic concentrations. CBF, cerebral blood flow.

Raw time-domain 1H MRS data from 4.0 to 1.0 p.p.m. in in the spectral dimension were analyzed using LCModel ¹⁵ with the unsuppressed water scan as a concentration reference. As a quality-assurance measure, LCModel produces a Cramer–Rao lower bound of the fit to the peak of interest. If this value was greater than 15%, then the fit was deemed unreliable and was excluded from the analysis. Metabolite concentrations in millimoles (mmol) were recorded using LCModel for total creatine (tCR), Guanidinoacetate (GUA), Glutamate (GLU), glutamate+glutamine (Glx), N-Acetylaspartate (NAA), lactate (Lac), glycerophosphocholine (GCP), Total choline (tCH), and myo-inositol (Ins).

Statistical Analysis

Patients in whom major reperfusion was detected were dichotomized based on whether they had hyperperfusion in the region of previously hypoperfused tissue. Statistical comparison was performed between the patient's ischemic hemisphere against the patient's contralateral hemisphere and with control values. Spectroscopy concentrations were analyzed using linear regression for comparison with ASL and clinical data, and two-way analysis of variance to compare patients with controls. Statistical significance was defined as a P value of <0.001.

Results

During the study period, a total of 109 acute stroke patients treated with intravenous recombinant tissue plasminogen activator were recruited. Thirty-two patients were excluded from further analysis due to the absence of major reperfusion on 24 hours ASL imaging (24) or motion affected MRS (8). Of the 77 eligible patients, the age range was 39 to 91 (mean 61), and 32 were female. The median acute NIHSS was 13 (range 5 to 22), the median 24-hour NIHSS was 7 (range 0 to 18), and the median 3-month mRS was 2 (range 0 to 5). The median stroke onset to treatment time was 171 minutes (range 32 to 270 minutes) with a median door to needle time of 50 minutes. The mean baseline infarct core and penumbra volume were 35 mL (range 10  to 73 mL) and 146 mL (range 10  to 214 mL) respectively, and the mean 24-hour DWI infarct core volume was 48 mL (range 25  to 160 mL). The 20 control participant's ages were from 55 to 67 (mean 61), half were female.

At 24 hours, 24 patients had hyperperfusion; 36 had reperfused penumbra without hyperperfusion, and 17 were excluded due to no reperfusion on 24-hour imaging (Figure 2). Patients with hyperperfusion had a much higher rate of excellent outcome (mRS 0 to 1, OR 4.1, P=0.002). Peri-infarct hyperperfusion was independently associated with better clinical recovery in a multivariate regression that included baseline infarct core volume and acute NIHSS. Mean 3-month mRS in the hyperperfused patients was 1.05 versus 2.61 (P=0.007, corrected for age and acute NIHSS P=0.016, Table 1). Patients with hyperperfusion were also younger (mean 68 versus 74 years, P=0.015). There were no significant differences between the groups in terms of baseline or 24-hour infarct core and penumbral volume, or baseline and 24-hour NIHSS (Table 1). Time to follow-up MRI was not associated with detection of hyperperfusion using binary logistic regression (P=0.591).

Figure 2.

The appearance of hyperperfusion after acute ischemic stroke treatment was independent of the acute clinical and imaging criteria, however, hyperperfusion predicted good outcome (modified Rankin score (mRS) 0 to 2). Columns are divided by post processed imaging maps, the first being an acute computed tomography perfusion (CTP) looking at the penumbra (green) and infarct core (red) volumes and acute CTP cerebral blood flow (CTP-CBF). The last three columns are from 24-hour magnetic resonance imaging (MRI) including arterial spin labeling (ASL-CBF), diffusion-weighted imaging (DWI) and 24-hour bolus tracking perfusion-weighted imaging cerebral blood flow (PWI-CBF). The second and fourth patients show hyperperfusion on follow-up ASL imaging (white arrows) while the first, third, and fifth patients do not show hyperperfusion on 24-hour ASL-CBF. Note that hyperperfusion observed on ASL-CBF is not apparent on the 24-hour PWI-CBF maps obtained concurrently. Also, the bottom patient did not show reperfusion and had persistent hypoperfusion on ASL-CBF, and was not included in the magnetic resonance spectroscopy (MRS) studies. Units are in mmol/L.

Table 1. Clinical and imaging characteristics of patients with and without hyperperfusion at 24 hours.

	Hyperperfusion (95% CI) n=24	No hyperperfusion (95% CI) n=36	Sig (P)
Acute core	36 mL (7–59 )	48 mL (15–67)	0.343
Acute penumbra	81 mL (42–129)	72 mL (39–102)	0.756
24-hour core volume	37 mL (19–66)	49 mL (18–83)	0.548
Acute NIHSS	15 (11–19)	12 (9–16)	0.163
24-hour NIHSS	7 (1–11)	8 (5–11)	0.449
Age	68 (60–75)	74 (70–85)	0.015
3-month mRS	1 (0–1.7)	2.5 (1.6–3.3)	0.007

CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale, mRS, modified Rakin Scale. A comparison between the clinical and imaging characteristics of reperfused patients with and without hyperperfusion at 24 hours and 3 months.

Spectroscopy of Reperfused Tissue

Hyperperfused versus nonhyperfused patients

Magnetic resonance spectroscopy voxels in the hyperperfused peri-infarct region showed elevated GLU, NAA, tCR, Lac, and GLX compared with peri-infarct voxels in patients without hyperperfusion. Other metabolites in the peri-infarct region were not different between hyperperfused and nonhyperperfused patients (Table 2). In the contralateral hemisphere, there were no differences between hyperperfused and nonhyperperfused patients with the exception of NAA, which was lower in patients without hyperperfusion (Table 2, Figure 3).

Table 2. Comparison between hyperperfused and nonhyperperfused metabolite concentrations.

		Reperfused penumbra		Contralateral		Control
		Hyperperfused patients	Hypoperfused patients	Hyperperfused patients	Hypoperfused patients
GLU	Mean	7.38	5.4	6.29	6.29	6.8
	95% CI	6.77–7.99	4.32–6.6	5.6–6.9	5.2–7.3	6.5–7.1
GPC	Mean	1.62	1.19	1.55	1.85	1.4
	95% CI	1.4–1.7	1–1.5	1.38–1.7	1.2–1.9	1.1–1.8
INS	Mean	4.36	4.49	4.95	5	5
	95% CI	3.35–5.3	3.67–5.2	4.2–5.6	3.9–5.8	4.7–5.6
Lac	Mean	2.1	1.3	0.88	1.13	0.58
	95% CI	0.9–3.3	0.2–2.3	0–1.8	0.07–2.2	0.28–0.9
NAA	Mean	6.97	5.4	6.8	5.22	7.5
	95% CI	6–7.8	4.5–6.4	6–7.6	4.3–6.1	7.1–7.8
GUA	Mean	2.05	2	1.97	2	1.3
	95% CI	1.4–2.7	0.75–3.2	1.22–2.7	0.9–3.29	1.2–1.5
tCH	Mean	2.37	2.23	2.42	2.69	1.6
	95% CI	0–3.9	0.84–4.5	0.45–4.3	0.19–5.2	1.1–1.8
tCr	Mean	6.2	4.57	5.75	5.57	5.6
	95% CI	5.48–6.92	3.94–5.43	5–6.41	4.65–6.41	5.3–5.9
GLX	Mean	8.88	7	6.3	6.49	6.4
	95% CI	7.83–9.85	5.76–8.24	5.5–7	5.2–7.7	5.4–7.5

Comparison between contralateral and ischemic hemisphere concentrations of MR spectroscopy (MRS) metabolites in patients with and without hyperperfusion. Measured MRS metabolites include: total creatine (tCR), Guanidinoacetate (GUA), Glutamate (GLU), Glu +Gln (GLX), N-Acetylaspartate (NAA), lactate (lac), glycerophosphocholine (GCP), Total choline (tCH), and myo-inositol (Ins). Units are in mmol/L.

Figure 3.

Study results for the variation of Glutamate (GLU), Lactate (Lac), N-Acetylaspartate (NAA), and Creatine (Cr) between regions of hyperperfusion, nonhyperperfusion, contralateral tissue, and controls. This shows that GLU is significantly higher in hyperperfused tissue compared with controls. Additionally, the nonhyperperfused GLU concentration is less than in ‘healthy' controls. There was also less NAA loss in hyperperfused patients compared with controls. Patients without hyperperfusion slowed lower NAA compared with hyperperfused patients but not when compared with contralateral tissue suggesting that these patients may have a lower baseline NAA concentration. Lac and Cr were also significantly higher in the hyperperfused area compared with patients without hyperperfusion. Error bars for standard deviation. Units are in mmol/L. *P<0.001 ^P<0.05, >0.001.

Comparisons with controls

Controls showed lower concentrations of GLU than peri-infarct regions in hyperperfused patients (P<0.001), and higher concentrations compared with peri-infarct regions in patients without hyperperfusion (P<0.001). Patients without hyperperfusion showed lower concentrations of NAA in the reperfused penumbra compared with controls (P<0.001).

Spectroscopy and Patient Outcome

The concentration of GLU and NAA in reperfused peri-infarct tissue 24 hours after ischemic stroke onset was significantly correlated with 3-month disability on the mRS scale (R²=0.561, P<0.001, and 0.27, P<0.001, respectively). Other metabolites such as GLX, INS, Lac, and tCR were not significantly correlated with 3-month outcome. In a backward regression model predicting continuous mRS that contained GLU, INS, NAA, Lac, and tCR, only GLU remained significant in the model (R² 0.511, P<0.001, Figure 4). Additionally, in a backward logistic regression to predict continuous mRS containing GLU, INS, NAA, Lac, tCR, 24-hour NIHSS, and 24-hour infarct core volume, only GLU and infarct core volume remained in the model (R² 0.483, model P<0.001).

Figure 4.

Peri-infarct Glutamate (GLU) predicts 3-month modified Rankin score (mRS). There was a strong relationship between the peri-infarct GLU and the resulting 3-month disability of patients after successful reperfusion. Error bars for standard deviation. Units are in mmol/L.

Discussion

We have identified distinctive metabolic signatures in hyperperfused and nonhyperperfused brain tissue after apparently successful reperfusion therapy for ischemic stroke. In particular, hyperperfused peri-infarct tissue showed raised glutamate and preserved NAA compared with controls. Peri-infarct tissue in reperfused patients without hyperperfusion showed lower NAA and glutamate than controls. These marked differences in levels of key brain metabolites suggest that postischemic hyperperfused tissue may be more metabolically ‘active', a phenomenon that could underpin the recognized relationship between hyperperfusion and better clinical outcome. However, it is not possible to say based on the available data whether the metabolic changes represent a cause or an effect of the perfusion changes. It is known that astrocytes, and the neurovascular unit more broadly, have a key role in regulating blood flow and metabolism in ischemic tissue via glutamate synthesis and lactate shuttling.¹⁶ Hyperperfusion may be a reflection of increased astrocyte activity promoting blood flow to areas of recent ischemia to promote or support increased metabolic activity, leading to enhanced or preserved long-term function. An alternate hypothesis is that the metabolic and perfusion change observed in this (and other studies) are due to an overshoot of cerebral metabolic rate of oxygen consumption, which may be interpreted as a rebound of protein and neurotransmitter synthesis, or other biochemical processes whose synthesis in the cell had been arrested by the severe hypoxia.¹⁷ This alternate hypothesis also highlights that the patients with hyperperfusion are actively recovering, whereas the patients without hyperperfusion have tissue dysfunction and damage that has not lead to frank infarction.

The relationship between hyperperfusion, tissue fate, and favorable clinical outcome was proposed by Marchal,¹⁷ where position emission tomography measured within the first 18 hours of middle cerebral artery occlusion revealed a number of patients showing hyperperfusion, all of whom recovered well. Interestingly, metabolic activity as measured by cerebral metabolic rate of oxygen consumption and cerebral blood volume within the hyperperfused areas was noted to be elevated, suggesting hyperperfusion was 'nutritional' and contributing to a favorable tissue outcome. We have identified a similar trend where patients with hyperperfusion had markers of increased metabolic activity such as glutamate and lactate in hyperperfused tissue. The traditional explanation for increased cerebral blood volume is vasoparalisis, however, we suspect that the increased cerebral blood volume is a more active process where blood is directed to areas of need and is supported by the results of increased glutamate in hyperperfused tissue.

Effective reperfusion has been shown to reduce tissue acidosis in ischemic stroke using PET.¹⁸ However what has not been previously recognized is that not all patients with reperfusion have the same tissue response. We have shown that there were MRS markers of tissue dysfunction in patients without hyperperfusion despite reperfusion, suggesting metabolic dysfunction despite an apparently good radiologic outcome (lack of infarction). We propose that there is a ‘reperfusion hierarchy', which significantly predicts patient outcome. First, without reperfusion the infarct core will expand into the penumbral region resulting in worse clinical outcome. Second, if there is reperfusion into already infarcted tissue, this is futile, and may be harmful due to increased risk of hemorrhagic transformation. Next, it has been assumed that reperfusion and salvage of penumbra is the ultimate prerequisite to a good outcome. Although this is indisputably true, that there is a dichotomy, even with ‘successful' reperfusion and salvage of penumbral tissue from macroscopic infarction. Patients in our study with peri-infarct hyperperfusion had much better recovery than those without. This variation in tissue pathophysiology is important to understand for future acute and recovery stroke trials because it may explain some of the large variation in patient outcomes despite apparently ‘successful' reperfusion. It also highlights the need to pay attention to reperfusion status in future trials of stroke recovery.

Glutamate is a metabolite of particular interest in ischemic stroke. This study found major differences in glutamate levels in the reperfused ischemic penumbra after stroke. Peri-infarct glutamate concentrations in hyperperfused patients were above the normal control range, while nonhyperperfused patients had low glutamate concentrations. Glutamate has traditionally been thought to have a primary role in neuronal death due to excitotoxicity during ischemic stroke, with this process beginning within seconds of vessel occlusion. After successful reperfusion, elevated glutamate in hyperperfused tissue may reflect increased cellular activity in salvaged, previously penumbral tissue. Preservation of metabolic activity was also indicated by preserved NAA in hyperfused regions.

Recent animal studies have identified that raised lactate shortly after reperfusion may be neuroprotective, as it is a product of anaerobic glycolysis and can be shuttled from astrocytes to neurons to compensate for temporary oxygen deprivation¹⁹ and attenuates neuronal death induced by intracortical glutamate injection.²⁰ Lactate is known to increase in the brain in two stages after stroke: a first increase during ischemia due to anaerobic glycolysis and a second increase beginning 1 hour after reperfusion and continuing up to 72 hours, well within our MRS time point. The second increase in lactate is not due to anaerobic glycolysis, as it occurs after reperfusion with restored oxygen supply, and may be explained by astrocyte production of lactate reflecting increased local metabolism. We postulate that this latter effect is what has been observed in this study (for the first time in human stroke), and likely explains the association between increased lactate and better recovery, whereas earlier studies suggested raised lactate may be harmful.²¹

This study also found a significant variation in the concentration of NAA between patients with and without hyperperfusion. The function of NAA is not clear, but in healthy brain it is a marker of mitochondrial metabolism²² and its concentration is thought to be proportional to neuronal density. Interestingly, patients without hyperperfusion showed lower NAA levels remote to the infarct compared with control participants (and also compared with patients with hyperperfusion). This may suggest that patients without hyperperfusion have a reduced metabolic capacity to respond to ischemic stroke, or that there is inhibition of metabolic activity remote to the infarct (a type of diachisis perhaps). It is unknown whether NAA levels reflect potential for neuroplasticity as part of the stroke recovery process, but the variations in the current study seen remote to the infarct suggest that NAA may be a marker of a patient's capacity to recover.^23,24 The reduced NAA in nonhyperperfused patients may be as a result of selective neuronal loss^{25, 26, 27} which is hypothesized to limit recovery and affects mood despite normal appearing MRI, however we cannot confirm this in the current study.

This study has identified major differences in regional and remote brain metabolism in patients after reperfusion that strongly correlate with ultimate clinical outcome. We suggest that these findings could have major implications for studies of rehabilitation and stroke recovery.²⁸ Metabolic imaging using ASL and MRS, as in this study, could potentially be used to serially assess response to therapy. Furthermore, if we can better understand the mechanisms of these metabolic differences, then it may even be possible to influence (and monitor) tissue metabolism after reperfusion using pharmacological, brain stimulation, and physical rehabilitation approaches. Some study uncertainties should be acknowledged around the fact that hyperperfusion was better observed on ASL than PWI-CBF. Obviously, there are major differences in how the techniques quantify CBF, and neither technique is an ideal measure of perfusion. The most common cause of an artifactual observation of hyperperfusion on ASL is due to CBF overestimation is a T1 increase in formerly ischemic tissue due in large part to edema. However, it seems unlikely that hyperperfusion observed on ASL in our study was artifactual as MRI was performed at 24 hours, before the development of significant edema, and, we only placed MRS voxels in noninfarcted tissue. Further, apparent diffusion coefficient values in these regions were in the normal range and not different to the contralateral hemisphere which argues strongly against there being significant edema. By the same token, the blood transit time to ischemic regions may be significantly increased and can also result in CBF overestimation with ASL. It is true that the ASL sequence we used (pulsed ASL) was not optimized to quantify delayed perfusion. However, because we defined reperfusion by the resolution of delayed contrast transit on PWI we are confident that patients without reperfusion (who, by definition, have much greater transit delays) were excluded. It seems more likely that the discrepancy between modalities was due to the poorer sensitivity of PWI-CBF at detecting hyperperfusion. Additionally, the use of MRI at 24 hours for infarct core delineation is validated and reliable.²⁹ Finally, we did not measures the volume of hyperperfused tissue, because we suspect that first this is a vascular reaction to ischemia and may explain why ASL is a better detector of hyperperfusion than PWI, and because of the inhomogeneity of the population regarding stroke location may mean that some regions have larger hyperperfusion lesions due to their anatomic location and not because they result in better patient outcomes.

The results of this study emphasize the variability of clinical and biochemical outcome after stroke reperfusion, with hyperperfusion of the previously ischemic region being a strong marker of better outcome after therapy. Hyperperfused tissue has a characteristic metabolite signature suggesting it is more metabolically active and perhaps more capable of later neuroplasticity, possibly due to early enhanced astrocyte activity which increases local blood flow and supports cerebral reorganization. Further studies are required to attempt to identify methods of promoting increased tissue metabolism in postreperfusion patients. This may be particularly relevant in those without hyperperfusion where treatments could aim to stimulate tissue into a more metabolically active state, with the ultimate aim of enhancing neuroplasticity and stroke recovery.

The authors declare no conflict of interest.