Reversal of DiffusionAbnormalities after IschemicStroke: Adding Difficultyand Complexity to theConundrum ofAcute Stroke Imaging

Diffusion-weighted magnetic resonance imaging(DWI) is increasingly used for the evaluation of acutestroke patients and provides important informationabout the location, extent and potential cause of thestroke.1 Combining DWI with susceptibility-weightedMRI appears to reliably distinguish between acutehemorrhagic and ischemic strokes.2 Beyond rapid con-firmation of the stroke diagnosis, DWI can be com-bined with perfusion MRI to potentially approximatethe ischemic penumbra and to choose patients morelikely to respond to thrombolytic therapy.3,4 The tissueabnormality observed on DWI represents ischemic re-gions with a decline in the apparent diffusion coeffi-cient of water. The precise mechanism for this appar-ent diffusion coefficient (ADC) decline remainsunresolved, but it is apparent that with ischemia theADC decline is associated with impairment of high-energy metabolism and the loss of ion homeostasis.5

Initially, it was thought that regions of ADC declineand hyperintensity on DWI represented regions of ir-reversible ischemic injury. Recent animal studies dem-onstrated that abnormal DWI regions could be largelyreversed with early reperfusion, although late, second-ary ADC declines occurred in a portion of the initialregion of ischemia that normalized.6,7 The UCLAgroup previously demonstrated in a small group ofstroke patients undergoing successful intraarterialthrombolysis that DWI abnormalities could be re-versed and that late secondary DWI abnormalities alsooccur in humans.8

The study by Kidwell and colleagues extends theprevious one and includes 18 ischemic stroke patientstreated within 6 hours of stroke onset with intraarterialthrombolysis alone (n � 12) or a combination of in-travenous and intraarterial thrombolysis (n � 6).9

Eight of 18 patients demonstrated some amount ofADC reversal when DWI was repeated several hoursafter thrombolysis. The day 7 MRI study showed thatfive of eight patients had partial or complete reappear-ance of a DWI abnormality that initially resolved andthree of eight had no secondary ADC changes. Clini-

cally, the latter group tended to have a better outcomethan those patients with secondary ADC declines.

The observations in this study and those derivedfrom animal studies evaluating the phenomenon ofADC reversal and secondary reoccurrence have severalintriguing implications. The study by Kidwell and col-leagues9 and the recent one by Fichler and colleaguesconfirm the animal studies that early ADC declines,even relatively severe declines, can be reversed withearly reperfusion.10 These observations support thecontention that use of the so-called diffusion-perfusionmismatch, that is, that region appearing to be normalon DWI, but abnormal on perfusion MRI, to definethe ischemic penumbra is indeed an oversimplifica-tion.3 If part of the diffusion abnormality is reversibleearly after stroke onset, then the ischemic penumbra,which has been operationally defined as ischemic tissuethat is potentially reversible,11 must include part of theDWI abnormality as well. In the future, the ischemicpenumbra will likely be defined on diffusion-perfusionMRI by relating time from stroke onset to absoluteADC and perfusion values and not just simply the re-gion of abnormal perfusion that is normal on DWI.Recent histological studies in rats subjected to 30 min-utes of temporary middle cerebral artery occlusion andmechanical reperfusion further complicate the issue oflesion reversal on DWI. Ringer and colleagues demon-strated partial or complete reversal of initial DWI ab-normalities 5 hours after reperfusion.7 Loss of MAP2immunostaining, a sensitive marker of irreversible neu-ronal, structural damage disclosed widespread changesin regions where ADC maps and T2 imaging werenormal. Using the same stroke model, Li and col-leagues observed abnormal neurons (shrunken neuronswith swollen cellular processes) and astrocytes (swollencytoplasm and vacuolation of the neuropil), changesconsistent with ischemic injury in the majority of cellsevaluated, 90 minutes after reperfusion when ADC val-ues had completely normalized.12 These studies suggestthat at the cellular level incomplete ischemic injury ispresent despite normal ADC and T2 values. One im-plication of these observations is that later ADC andT2 changes perhaps represent a continuum of ischemiclesion development and not a de novo process.

Another implication of early DWI reversal and thenlate ADC decline after successful thrombolysis is thatadditional therapy may be useful. The mechanismscausing the secondary ADC decline and delayed isch-emic injury have not been elucidated precisely. Poten-tial causes for the secondary injury include free radical–mediated cell damage, recruitment of inflammatorymediators, delayed mitochondrial dysfunction relatedto secondary energy failure, and the initiation of pro-grammed cell death pathways.6 It is certainly conceiv-able that more than one mechanism may contribute tothe development of secondary cellular injury after

EDITORIALS

© 2002 Wiley-Liss, Inc. 695

reperfusion. Precise clarification of the mechanisms in-volved will help to target therapeutic intervention toimpede the development of secondary injury. Thestudy by Kidwell and colleagues9 suggests that the de-velopment of secondary injury may be clinically rele-vant, as patients with this MRI-documented processtended to have a worse outcome. Future clinical trialswill need to determine if using one or more therapiestargeted to impede potential mechanisms of secondaryischemic injury along with intravenous or intraarterialthrombolysis leads to better ultimate clinical outcomesthan thrombolysis alone. The use of multiple therapiesafter acute ischemic stroke is an important concept forthe enhancement of stroke therapeutics.

The pathophysiology, imaging, and treatment ofacute ischemic stroke are all areas of great complexity.Initial hopes that DWI and perfusion MRI would pro-vide easy markers for stroke diagnosis and help withmanagement have proved to be correct in some re-spects but not in others. DWI and perfusion MRI areclearly helpful for readily identifying the location andextent of ischemic stroke. However, imaging of theischemic penumbra with these MRI modalities appearsto be a complex endeavor that will require many addi-tional investigations before more precise characteriza-tion is available. The study by Kidwell and colleagues9

provides new information about ischemic lesions de-tected by DWI and sets the stage for additional inves-tigations that will likely have important implicationsfor MRI detection of the ischemic penumbra and po-tentially for the development of novel therapies to en-hance the benefits of early thrombolysis.

References1. Neumann-Haefelin T, Moseley ME, Albers GW. New mag-

netic resonance imaging methods for cerebrovascular disease:emerging clinical applications. Ann Neurol 2000;31:559–570.

2. Von Kummer R. MRI: the new gold standard for detectingbrain hemorrhage? Stroke 2002;35:1748–1749.

3. Schlaug G, Benfield A, Baird AE, et al. The ischemicpenumbra: operationally defined by diffusion perfusion MRI.Neurology 1999;53:1528–1537.

4. Marks MP, Tong DC, Beaulieu C, et al. Evaluation of earlyreperfusion and i.v. tPA therapy using diffusion and perfusion-weighted MRI. Neurology 1999;52:1792–1798.

5. Busza AL, Allen KL, King MD, et al. Diffusion-weighted im-aging studies of cerebral ischemia in gerbils: potential relevanceto energy failure. Stroke 1992;23:1602–1612.

6. Li F, Liu KF, Silva MD, et al. Secondary decline in apparentdiffusion coefficient and neurological outcome after a short pe-riod of focal brain ischemia in rats. Ann Neurol 2000;48:236–244.

7. Ringer TM, Neumann-Haefelin T, Sobel RA, et al. Reversal ofearly diffusion-weighted magnetic resonance imaging abnormal-ities does not necessarily reflect tissue salvage in experimentalcerebral ischemia. Stroke 2001;34:2362–2369.

8. Kidwell CS, Saver JL, Mattiello J, et al: Thrombolytic reversalof acute human cerebral ischemic injury shown by diffusion/perfusion magnetic resonance imaging. Ann Neurol 2000;47:462–469.

9. Kidwell CS, Saver JL, Starkman S, et al. Late secondary isch-emic injury in patients receiving intraarterial thrombolysis. AnnNeurol 2002;52:698–703.

10. Fichler J, Foth M, Kucinski T, et al. Severe ADC decreases donot predict irreversible tissue damage in humans. Stroke 2002;33:79–86.

11. Hakim AM. The cerebral ischemic penumbra. Can J NeurolSci 1987;14:557–559.

12. Li F, Silva MD, Meng X, et al. Acute postischemic renormal-ization of the apparent diffusion coefficient of water is not as-sociated with reversal of astrocytic swelling and neuronalshrinkage in rats. AJNR 2002;23:180–188.

Marc Fisher, MD

Department of NeurologyUniversity of MassachusettsWorcester, MA

DOI: 10.1002/ana.10388

Epilepsy and ReproductiveFunction in Women

As a medical student, I admitted an elderly womanwith congestive heart failure. In the course of takingher history, I discovered that she underwent a hyster-ectomy as a young woman because she had been toldthat her epilepsy would prevent her from ever having anormal baby. Later in our conversation, she showed mepictures of sister’s children, and described how sad shewas that she had never had a child of her own. Ofcourse, no one would ever recommend hysterectomyfor this indication today. Although at increased risk,most children born to women with epilepsy are nor-mal. However, women with epilepsy may be deniedmotherhood because they have an increased incidenceof reproductive hormonal disorders, impairments insexual arousal, amenorrhea, and infertility. Unfortu-nately, these problems are largely ignored in the clinicalsetting, and there have been few investigations in thisarea.

In this issue, Morrell and colleagues1 report the find-ings of a systematic investigation to determine whetherwomen with epilepsy are more likely to have anovula-tory cycles and to assess the relative contribution ofepilepsy syndromes and antiepileptic drugs (AEDs) toovulatory dysfunction. They examined 117 women ofchildbearing age who were patients with localization-related epilepsy, patients with idiopathic (primary) gen-eralized epilepsy, or healthy controls. The women wereobserved over three menstrual cycles (or 3 months) as-

696 © 2002 Wiley-Liss, Inc.