REVIEW ARTICLE

Imaging of Experimental Stroke Models

Marc Fisher & Bernt Tore Bråtane

Received: 24 August 2011 /Revised: 4 October 2011 /Accepted: 6 October 2011 /Published online: 15 November 2011# Springer Science+Business Media, LLC 2011

Abstract The ischemic penumbra is the target of acutestroke therapy. It can be approximated on diffusion/perfusion MRI as the ischemic region with abnormalperfusion and normal diffusion imaging. Using arterial spinlabeling perfusion MRI and diffusion MRI, our group hasstudied the evolution of the diffusion/perfusion mismatch inrat stroke models. Additionally, we have evaluated theeffects of high-flow oxygen on the natural history ofpenumbral evolution, demonstrating that high-flow oxygen“freezes” the evolution of the mismatch and allows for laterbeneficial use of intravenous tissue plasminogen activator(tPA). Two neuroprotective drugs, Granulocyte colonystimulating factor and a PSD95 inhibitor both impededthe evolution of the mismatch into infarcted tissue in vivoand by histological analysis. Employing a novel techniqueof clot imaging, our group was able to demonstrate that thecombination of tPA plus Annexin-2 was superior to tPAalone in dissolving an embolus and also in reducing theextent of hypoperfused brain tissue of perfusion imaging.The use of these advanced MRI techniques in animalexperiments will help to advance clinical imaging of theischemic penumbra and hopefully contribute to the exten-sion of the therapeutic time window in stroke patients.

Keywords Mismatch . Diffusion . Perfusion . ADC . CBF.

MRI

Imaging of acute ischemic stroke is a key component in theevaluation of this disorder and provides valuable informa-tion for the development and implementation of therapies.The two imaging modalities used to evaluate focal brainischemia, computerized tomography, and magnetic reso-nance imaging reflect tissue changes that occur in relation-ship to this dynamic and complex process. As aconsequence of the interruption of oxygen and nutrient-rich blood flow to the brain tissue injury ensues [1]. Thelevel of cerebral blood flow (CBF) decline in the region ofischemia is related to the rapidity of the tissue becomingirreversibly injured, i.e., infarction. The region with adrastic reduction in CBF with levels less than 12 ml/100 g/min becomes infarcted rapidly and is called theischemic core. The region with a more modest CBF declinein the range of 12–25 ml/100 gm/min evolves more slowlyand is known as the ischemic penumbra. Other definitionsof the ischemic penumbra have been proposed such as theischemic region with preserved ion homeostasis withelectrical failure and the region with failure of high-energy metabolism failure. The rate of evolution of theischemic penumbra from potentially salvageable intoirreversibly injured varies across species from rodents toprimates and is affected by metabolic factors such asglucose levels, plasma oxygenation, and temperature [2]. Athird region of reduced CBF, termed benign oligemia,represents a region with reduced CBF, but not to a degreethat will lead to infarction even with prolonged exposure.

With both CT and T2 or fluid attenuated inversionrecovery (FLAIR) MRI sequences abnormalities can beimaged that are related to an increase in total tissue watercontent that occurs early after the development of irrevers-ible tissue injury. This increase in tissue water contentoccurs secondary to the development of vasogenic edemaand can take many hours to manifest. Although subtle

M. Fisher :B. T. BråtaneDepartment of Neurology,University of Massachusetts Medical School,Worcester, MA, USA

M. Fisher (*)UMASS/Memorial Healthcare,119 Belmont Street,Worcester, MA 01605, USAe-mail: fisherm@ummhc.org

Transl. Stroke Res. (2012) 3:16–21DOI 10.1007/s12975-011-0113-1

changes on standard CT scans and T2/FLAIR MRI can beseen within a few hours after the onset of focal brainischemia, more extensive and obvious changes may not beseen for 12–24 h [3]. This late time window for the accuratedepiction of focal brain ischemia is helpful for confirmingthe diagnosis but beyond the time window for helping withtreatment decisions because by definition the changesobserved reflect irreversible injury. A more recent MRIapproach for evaluation of focal brain ischemia was thedevelopment of diffusion-weighted and perfusion imaging.Diffusion-weighted MRI (DWI) employs an imagingsequence that quantifies the movement of water moleculeprotons in the brain tissue. Shortly after the onset of focalbrain ischemia, a shift of water from the extracellular tointracellular space occurs secondary to high-energy metab-olism failure induced by a reduction of CBF [4]. As aconsequence of high-energy metabolism failure ionichomeostasis is compromised and various ions and extra-cellular water move intracellularly, a process termedcytotoxic edema [5]. This early increase in intracellularwater is associated with less diffusivity as intracellularwater molecule protons are relatively restricted in theirmovement as compared to water molecule protons in theextracellular space. DWI takes advantage of this change inwater molecule proton movement and is designed tomeasure the apparent diffusion coefficient (ADC) of thebrain tissue, which is reduced in tissue with high-energymetabolism failure as compared to normal brain tissue [6].This reduction of the ADC occurs within minutes afterstroke onset, especially in the ischemic core. Regions ofreduced ADC are hyperintense on DWI and readilyidentifiable. A major advantage of DWI compared to T2/FLAIR MRI is the depiction of ischemic tissue destined forinfarction within minutes of onset as compared to the manyhours needed to image infarction with the latter techniques.Perfusion MRI (PWI) is the other novel MRI technique thatprovides much useful information in acute ischemic stroke.PWI can be performed by either the bolus contrast approachor the arterial spin-labeling approach. Bolus contrast PWI isobtained by the rapid injection of a paramagnetic contrastagent such as gadolinium followed by the acquisition of arapid sequence of susceptibility-weighted, gradient-echoimages [7]. The magnetic susceptibility effects of thecontrast agent as it transits the brain microvasculatureinduce signal loss in the brain tissue, and a signal washoutcurve is generated. In this washout curve, the area under thecurve is proportional to the cerebral blood volume (CBV),and the time needed to reach the maximal signal intensitychange or time to peak is related to the mean transit time(MTT) of the contrast bolus [8]. A qualitative representa-tion of CBF can be derived from this data as CBF=CBV/MTT. In an ischemic brain, the bolus of contrast reachesthis tissue in a much delayed fashion as compared to a

normal brain, so MTT is delayed, CBV is typically reduced,and the relative CBF of this ischemic tissue is lower thannormal brain tissue. With bolus contrast PWI, the locationand level of blood flow changes can be identified. Arterialspin-labeling (ASL) PWI is entirely noninvasive and isperformed by pulse labeling of blood protons in the neck[9]. As the labeled blood protons flow into the brain, theyinteract with tissue water and alter tissue magnetization.This interaction can be quantified and reflects the quantityof blood flowing into the brain and can be used to derivequantitative CBF values in the brain tissue. In general,bolus contrast PWI is used in most clinical situations, andASL PWI has been largely restricted to experimental strokeimaging in various animal stroke models because of signalto noise issues.

The utility of DWI and PWI to rapidly demonstrateregions of ischemic compromise and to depict the extentand severity of blood compromise in ischemic stroke led totheir widespread use for the diagnosis of ischemic stroke.Another important use of DWI/PWI is for potentialidentification of the ischemic penumbra. As initiallyproposed, ischemic regions that were abnormal on boluscontrast PWI in humans but did not demonstrate a DWIabnormality were proposed to approximate the ischemicpenumbra because this region of so-called DWI/PWImismatch has reduced CBF but has not yet developedenough high-energy metabolism compromise to induceDWI changes [10]. The initial proposal of the DWI/PWImismatch concept was an important step forward inpenumbral imaging because with accessible imagingtechniques that have become available in many hospitalsan approximation of the ischemic penumbra could beobtained, as compared to the inaccessibility of positronemission tomography, the gold standard technique forpenumbral identification. There are, however, some remain-ing unresolved problems with the DWI/PWI mismatchapproach to penumbral imaging. The first of these problemsis that at early time points the DWI lesion can at least inpart be reversed, especially in regions where the ADC dropis relatively modest [11]. A second and very importantproblem is that defining the PWI lesion on the boluscontrast approach is dependent on accurate thresholding ofthe abnormal PWI region [12]. Many suggestions to solvethe problem of accurate PWI thresholding were proposedand currently the most widely acknowledged approach is touse a Tmax threshold of 5.5 or 6 s as compared to the normalside to define hypoperfused ischemic regions destined tobecome infarcted [13]. This approach remains to bedefinitively validated, but current human imaging trialsare attempting to do so. A third problem with the DWI/PWIhypothesis is how to distinguish between benign oligemiaand the ischemic penumbra. Tissue in the benign oligemicrange of perfusion compromise is not destined for infarc-

Transl. Stroke Res. (2012) 3:16–21 17

tion and should not be included in the tissue at riskcomponent of hypoperfused tissue. This distinction ofischemic tissue with reduced perfusion but not at risk ofinfarction from hypoperfused tissue at risk of infarction willagain require accurate and validated thresholding of theperfusion lesion.

Using both the suture occlusion and embolic rat modelsof ischemic stroke, our group has evaluated the evolution ofthe DWI/PWI mismatch over the first few hours after strokeonset [14]. The studies were all performed with ASL PWIand our initial step was to validate an ADC level thatdefined ischemic tissue destined to become infarcted, andCBF values in hypoperfused regions also predicted infarc-tion without early intervention. These thresholds werevalidated by relating lesion volumes on DWI and ASLPWI to infarct volumes at 24 h. Once these ADC and CBFthresholds were defined, we then employed them toevaluate the DWI/PWI mismatch in both the sutureocclusion and embolic stroke models. In rats, the mismatchdisappeared quite rapidly, and by 2 h after stroke onset,there was no longer a significant difference in the DWI andPWI volumes, supporting the lack of significant penumbraby this time point [15]. It was also observed that themismatch was somewhat larger in the embolic model ascompared to the suture model [16]. Additionally, speciesdifferences were observed as Wistar rats had persistence ofthe mismatch for a longer time period than did Sprague–Dawley rats [15]. Using a sophisticated data analysisalgorithm that evaluated baseline ADC and CBF values,we were able to track the fate of individual voxels withinthe ischemic region based upon the severity of ADC andCBF changes early after stroke onset and to predict theprobability of infarction [17]. Employing mechanicalreperfusion by withdrawing the suture occluder, the likelyresponse of individual voxels to this reperfusion regardingtissue salvage or progression to infarction was alsoevaluated.

Oxygen was studied in several MRI experiments andshowed a dramatic effect on the ADC evolution withoutaffecting the CBF and therefore extending the mismatch.The first study was conducted utilizing the suture middlecerebral artery occlusion (MCAO) model with and withoutreperfusion [18]. Normobaric oxygen (NBO) was distribut-ed for 3 h at a rate of 1 L/min starting at 30 min after strokeonset. The size of the ADC lesions in both NBO andplacebo-treated groups was approximately 100 mm3 at20 min after stroke onset. At 45 min (15 min after NBOinitiation), the ADC lesion volumes in the NBO groupswere reduced by about 50% and remained almost the samefor the duration of the NBO treatment, while in the controlgroup ADC lesion volumes steadily increased until match-ing the CBF lesion volume at 150 min after the start ofischemia. The CBF lesion volumes were stable at approx-

imately 250 mm3 for the duration of the experiment for thepermanent occlusion group and until reperfusion at 210 minin the reperfusion group. Only the reperfusion group hada statistically significant reduction of the infarct ascompared to placebo measured on a 24-h histology. Thesecond oxygen experiment was conducted utilizing theembolic model and tPA for reperfusion [19]. NBO was alsoinitiated at 30 minutes post-MCAO, and tPA was started intwo groups at 180 min after embolization and infused over1 h. NBO in both NBO groups was terminated at 240 minafter stroke onset. ADC and CBF lesion volumes beforetreatments at 20 min post-MCAO were larger at 250 and350 mm3, respectively in this embolic stroke (ES) modelthan in the previous suture MCAO model, but the effect ofthe NBO was similar. Both NBO groups had a non-significant reversal of the ADC lesion volume, and theeffect was present for the entire experiment. tPA reducedthe CBF lesion volume in both the NBO and placebo groupbut only by about 50%, indicating incomplete reperfusionand/or no re-flow phenomena. The ADC lesion volumeswere also reduced after infusion of tPA. There was asignificant mismatch between ADC and CBF for all timepoints in the tPA/NBO group and until the 150-min timepoint in the NBO/placebo group. In the non-NBO groups,the mismatch was only significant for the first 60 min afterES. On histology at 24 h, the NBO groups both hadsignificantly reduced infarct volumes compared to room air,and the tPA, when given at 180 min, significantly reducedthe infarct size in the NBO/tPA group as compared to theroom air tPA group, indicating that NBO extended thewindow for tPA in this ES model. Representative ADC,CBF, and TTC images for the two NBO experiments arepresented in Fig. 1.

We have also successfully altered ADC evolution withtwo different neuroprotective drugs. The first wasgranulocyte-colony stimulating factor (G-CSF) where120 μg/kg was injected i.v. at 60 min after MCAO andwas then given i.p. at 4 h post-MCAO in the suture model[20]. More recently, a PSD95 Inhibitor Tat-NR2B9c wasevaluated, employing a dose of 7.5 mg/kg was injected i.v.over 5 min ending at 60 min post-MCAO [21].Representative ADC, CBF, and TTC images for the G-CSF and Tat-NR2B9c experiments are presented in Figs. 2and 3, respectively.

G-CSF had an immediate effect on the ADC lesionvolume. The ADC lesion volume increased in both thetreated and vehicle groups from 130 mm3 at 25 min to160 mm3 at 45 min. In the group receiving G-CSF, theADC lesion volume was reduced to 130 mm3 at 90 min andwas stable for the remainder of the study. The placebogroup had a normal ADC lesion volume evolution andmatched the CBF lesion volume at 3 h. The mismatch inthe G-CSF groups was significantly larger than in the

18 Transl. Stroke Res. (2012) 3:16–21

placebo group starting at 90 minutes post-MCAO. The 24-h histology also showed a significantly larger strokevolume in the placebo group as compared to the G-CSF

group. Tat-NR2B9c had a similar effect on the ADC lesionvolume evolution as observed with G-CSF. With Tat-NR2B9c, the ADC lesion volume remained the same overtime while in the vehicle group it increased substantiallyover time with a significant difference in the 24-h histologically confirmed infarct volume.

To assess the effect of novel thrombolytic treatments, wedeveloped a novel parameter in vivo clot length [22].Monitoring clot length as thrombolytic treatment is admin-istered is helpful in addition to ADC and CBF lesionvolume evaluation. In order to perform in vivo monitoring,the clot was incubated with a gadolinium-based contrastagent for 24 h and then was injected with the samemethodology as in the previous ES experiments. A T1-weighted MRI sequence was used to acquire the signalfrom the clot. Clot length and CBF lesion volumes wereacquired prior to initiation of thrombolysis, every 15 minduring the 1-h infusion and 30-min posttreatment. Thetreatment groups were tPA (10 mg/kg), recombinant

Fig. 3 Representative MRI and 24-h histology images for the Tat-NR2B9c study. A suture MCAO model was applied and bothexperimental groups are presented with one central slice of the ADCand the CBF maps from one representative animal at the last imagingtime point (3-h post-MCAO) and a corresponding slice from the 24-h histology (TTC). TTC 2-3-5-triphenyl tetrazolium chloride, pMCAOpermanent MCAO

Fig. 2 Representative MRI and 24-h histology images for the G-CSFstudy. A suture MCAOmodel was applied and both experimental groupsare presented with one central slice of the ADC and the CBF maps fromone representative animal at the last imaging time point (3-h post-MCAO)and a corresponding slice from the 24-h histology (TTC). TTC 2-3-5-triphenyl tetrazolium chloride, pMCAO permanent MCAO

Fig. 1 Representative MRI and 24-h histology images for the two NBOstudies. a Suture MCAO model. b ES model. All the experimentalgroups are presented with one central slice of the ADC and the CBFmaps from one representative animal at the last imaging time point(4.5 h post-ES) and a corresponding slice from the 24-h histology(TTC). TTC 2-3-5-triphenyl tetrazolium chloride, p/t MCAO permanent/temporary MCAO, tPA tissue plasminogen activator

Fig. 4 Representative MR images for the recombinant rA2 study. AnES model were applied, and all three experimental groups arepresented with one central slice of the ADC and the CBF maps inaddition to a T1 clot image from one representative animal at the lastimaging time point (160 min (ADC) and 3 h (T1 and CBF) post-ES).tPA tissue plasminogen activator

Transl. Stroke Res. (2012) 3:16–21 19

Annexin-2 (rA2, 5 mg/kg)+tPA (10 mg/kg) and placebo.Treatments were initiated i.v. 90 min post-ES and wasgiven for aduration of 1 h. The placebo group had nosignificant change in clot length or CBF lesion volume forthe duration of the study. The treatment groups had asignificant reduction in clot length and a significantlyreduced CBF lesion volume as compared to the placebogroup starting at 15 min and 30 min after treatmentinitiation for the rA2+tPA and tPA, respectively. Only therA2+tPA reliably dissolved the clots entirely while tPAalone was less effective. Also, for CBF lesion volume therA2+tPA group induced significantly greater reductionsthan tPA alone at the last imaging time point. For ADClesion volumes only three imaging time points wereacquired. The ADC lesion volumes were similar in allthree groups at 90 min just before treatment initiation. Thelast ADC lesion volumes at 160 min post-ES showed anon-significant reduction of about 35% and 50% for thetPA alone and the rA2+tPA, respectively versus placebo. Aplausible reason for this non-significant trend is the time oftreatment initiation and the relatively small sample size.Utilizing the information from the clot imaging, it can beconcluded that the clots are not consistently dissolved bytPA alone, and the combination of tPA and annexin 2 wasmore potent. The clot imaging technique provides a noveltool to assess the thrombolytic efficacy of newer agents incomparison to the standard of tPA. Representative ADC,CBF, and T1 images for the rA2 experiment is presented inFig. 4.

The utility of diffusion/perfusion MRI and novelimaging methodology such as clot imaging to enhanceour understanding of focal ischemic brain injury evolu-tion, and its treatment is apparent in preclinical studies.These techniques are also being used in clinical trials andpractice to evaluate patients who are more appropriatefor treatment and the effects of these treatments [23].Their use will only increase as refinements in definingthresholds predictive of infarction occur and availabilitybecomes more widespread. The evolution of the diffusion/perfusion mismatch is substantially shorter in our ratstroke models than observed in clinical studies. Thisdifference in penumbral survival suggests that the timewindow for initiating successful therapies in patients willlikely be much longer than in rodents and reinforces thesuggestion that employing DWI/PWI MRI in clinicalstroke trials should be useful for extending the therapeutictime window. The era of penumbral imaging has begun,and a bright future is likely as the ability to understandindividual patient patterns of ischemic injury are realized.It is only with the use of penumbral imaging that thetherapeutic window for acute ischemic stroke will beexpanded and the probability of individual treatmentsuccess is enhanced.

Potential Conflict of Interest The G-CSF study was partiallyfunded by Sygnis Bioscience and the PSD95 Inhibitor study waspartially funded by the Chair fund of the Neurovascular TherapeuticsProgram from the University Health Network, Toronto.

References

1. Lo EH, Moskowitz MA, Jacobs TP. Exciting, radical, suicidal:how brain cells die after stroke. Stroke. 2005;36:189–92.

2. Martini SR, Kent TA. Hyperglycemia in acute ischemic stroke: avascular perspective. J Cereb Blood Flow Metab. 2007;27:435–51.

3. Chalela J, Kidwell C, Nentwich L. Magnetic resonance imagingand computed tomography in emergency assessment of patientswith suspected acute stroke: a prospective comparison. Lancet.2007;369:293–8.

4. Liu KF, Li F, Tatlisumak T, Garcia JH, Sotak CH, Fisher M, et al.Regional variations in the apparent diffusion coefficient and theintracellular distribution of water in rat brain during acute focalischemia. Stroke. 2001;32:1897–905.

5. Heiss W. Ischemic penumbra: evidence from functional imagingin man. J Cereb Blood Flow Metab. 2000;20:1276–93.

6. Baird AS, Warach SW. Magnetic resonance imaging of acutestroke. J Cereb Blood Flow Metab. 1998;18:583–609.

7. Detre JA, Alsop DC, Vives LR, Maccotta L, Teener JW, Raps EC.Noninvasive MRI evaluation of cerebral blood flow in cerebro-vascular disease. Neurology. 1998;50:633–41.

8. Wintermark M, Sesay M, Barbier E, et al. Comparative overviewof brain perfusion imaging techniques. Stroke. 2005;36:e83–9.

9. Calamante F, Thomas DL, Pell GS, Wiersma J, Turner R.Measuring cerebral blood flow using magnetic resonance imagingtechniques. J Cereb Blood Flow Metab. 1999;19:701–35.

10. Schlaug G, Benfield A, Baird AE, et al. The ischemic penumbraoperationally defined by diffusion-perfusion MRI. Neurology.1999;53:1528–37.

11. Kidwell CS, Saver JL, Mattiello J, et al. Thrombolytic reversal ofacute human cerebral ischemic injury shown by diffusion/perfusion magnetic resonance imaging. Ann Neurol.2000;47:462–9.

12. Kane I, Carpenter T, Chappell F, et al. Comparison of 10 differentmagnetic resonance perfusion imaging processing methods inacute ischemic stroke. Stroke. 2007;38:3158–64.

13. Olivot JM, Miynash M, Thijs VN, et al. Optimal Tmax thresholdfor predicting penumbral tissue in acute stroke. Stroke.2009;40:469–75.

14. Meng X, Fisher M, Shen Q, Sotak CH, Duong TQ. Characterizingthe diffusion/perfusion mismatch in a rat stroke model of focalcerebral ischemia. Ann Neurol. 2004;55:207–12.

15. Bardutzky J, Shen Q, Ren H, Duong TQ, Fisher M. Differences inischemic lesion evolution in different rat strains using diffusionand perfusion imaging. Stroke. 2005;36:2000–5.

16. Henninger N, Sicard KM, Schmidt K, Bardutzky J, Fisher M.Comparison of ischemic lesion evolution in embolic vs.mechanical middle cerebral artery occlusion in Sprague–Dawley rats using diffusion and perfusion imaging. Stroke.2006;37:1283–7.

17. Shen Q, Ren H, Fisher M, Duong TQ. Statistical prediction of tissuefate in acute in ischemic brain injury. JCBFM. 2005;25:1336–45.

18. Henninger N, Bouley J, Nelligan J, Sicard KM, Fisher M.Normobaric hyperoxia delays perfusion/diffusion mismatch evo-lution, reduces infarct volume and differentially affects neuronalcell death pathways after suture middle cerebral artery occlusionin rats. JCBFM. 2007;27:1632–42.

20 Transl. Stroke Res. (2012) 3:16–21

19. Henninger N, Bratane BT, Bastan B, Bouley J, Fisher M.Normobaric hyperoxia and late tPA in a rat embolic stroke model.J Cereb Blood Flow Metab. 2009;29:119–29.

20. Bratane B, Bouley J, Schneider A, Bastan B, Henninger N, Fisher M.Granulocyte colony stimulating factor delays mismatch evolutionand reduces final infarct volume in permanent suture and embolic ratfocal cerebral ischemia models. Stroke. 2009;40:3102–6.

21. Bratane B, Cui H, Cook DJ, Bouley J, Tymianski M, Fisher M.Neuroprotection by freezing ischemic penumbra evolution withPSD95 inhibitor. Stroke (in press).

22. Walvick RP, Bratane B, Henninger N, et al. Evaluation of clotlysis in a rat embolic stroke model. Stroke. 2011;42:1110–5.

23. Fisher M, Bastan B. Identifying and utilizing the ischemicpenumbra. Neurology (in press).

Transl. Stroke Res. (2012) 3:16–21 21