Transient Ischemic Attacks Before Ischemic Stroke: Preconditioning the Human Brain?
A Multicenter Magnetic Resonance Imaging Study
Background and Purpose— We investigated whether transient ischemic attacks (TIAs) before stroke can induce tolerance by raising the threshold of tissue vulnerability in the human brain.
Methods— Sixty-five patients with first-ever ischemic territorial stroke received diffusion- and perfusion-weighted MRI within 12 hours of symptom onset. Epidemiological and clinical data, lesion volumes in T2, apparent diffusion coefficient (ADC) maps and perfusion maps, and cerebral blood flow and cerebral blood volume values were compared between patients with and without a prodromal TIA.
Results— Despite similar size and severity of the perfusion deficit, initial diffusion lesions tended to be smaller and final infarct volumes were significantly reduced (final T2: 9.1 [interquartile range, 19.7] versus 36.5 [91.2] mL; P=0.014) in patients with a history of TIA (n=16). This was associated with milder clinical deficits.
Conclusions— The beneficial effect of TIAs on lesion size in ADC and T2 suggests the existence of endogenous neuroprotection in the human brain.
Two previous clinical studies have suggested that transient ischemic attacks (TIAs) before ischemic stroke in the same vascular territory are associated with milder initial clinical symptoms and more favorable outcome in stroke (“protective TIA”).1,2 Among possible explanations for this phenomenon is the concept that a prodromal TIA may be the clinical correlate of experimental preconditioning and thus provoke adaptive cellular responses to ischemia. These have been elicited in tissue cultures and animal models in an early and a delayed time window of protection. Ischemic preconditioning in the brain follows the delayed pattern. It develops with a latency period of at least 1 day and is sustained for days to weeks (for review, see Kirino3). As an alternative to this concept of ischemic protection on a cellular level, this phenomenon may be due to a better vascular supply induced by TIA or a beneficial collateral circulation indicated by TIA, so that less tissue is destroyed by subsequent stroke. However, in a rat model of focal preconditioning, development of tolerance did not involve alterations in regional cerebral blood flow (CBF) before or during the ischemic challenge, resulting in infarction.4 To date, evidence regarding the existence of preconditioning in the human brain is still lacking. The aforementioned studies that showed a potentially beneficial effect of prodromal TIAs were based on clinical parameters only. One could argue that these findings may be explained simply by prodromal TIAs being an indicator of smaller areas of ischemia or less severe ischemia. If this were the case, the initial perfusion deficit in patients with prodromal TIAs should differ from that in patients without TIAs. The constellation of a similar perfusion lesion but smaller infarct size in those patients, however, would be consistent with an “endogenous preconditioning” hypothesis. To address this issue, we assessed perfusion and tissue damage in patients with first-ever ischemic stroke with and without prodromal TIA using perfusion-weighted (PWI) and diffusion-weighted (DWI) MRI.5,6 The extent of the initial perfusion and diffusion deficit and final infarct volume were compared. Additionally, CBF and cerebral blood volume (CBV) within the area of restricted perfusion were analyzed qualitatively in both groups.
Subjects and Methods
Eligibility and Inclusion Criteria
This multicenter study incorporated data from 4 university hospitals (Düsseldorf, Hamburg, Heidelberg, and Berlin) within the German Competence Network Stroke study group MRI in Acute Stroke (B5 Project; Principal Investigator, Arno Villringer). The study was approved by the local ethics committees. We retrospectively reviewed the prospectively collected stroke databases of the 4 participating stroke centers between 1997 and 2001. Inclusion criteria were acute ischemic stroke with MRI and clinical (National Institutes of Health Stroke Scale and modified Rankin Scale) assessment within 12 hours of symptom onset as well as on follow-up (earliest time point included for final assessment was day 3) and completion of a validated interrogation sheet regarding history of TIA.2 Exclusion criteria were intracerebral hemorrhage, older lesions on initial T2-weighted images, lack of informed consent, and insufficient clinical or radiological material. Ninety-three patients fulfilled all inclusion criteria for this study, among them 28 patients with lacunar infarctions. Because of presumably different underlying pathogenesis,7 we analyzed these patients separately and here report only on patients with territorial infarctions. Clinical and epidemiological data were taken from patients’ charts. Stroke etiology was defined according to Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria.8
MRI Acquisition and Analysis
MRI was performed on 1.5-T clinical scanners at each study center. Typical sequence parameters of the joint protocol are described elsewhere6,9,10 and in the Appendix. PWI and DWI data were postprocessed with routines developed by Peter Brunecker and Christian Kerskens (Berlin) using IDL 5.0 (Research Systems Inc). Maps of the apparent diffusion coefficient (ADC), mean transit time (MTT), CBF, and CBV were generated from DWI and PWI data, respectively. For CBF and CBV, an arterial input function was calculated following the approach suggested by Ostergaard et al.11–13 Lesion volumetry was performed by 2 independent observers blinded to clinical data using National Institutes of Health Image software. Contralateral mirror regions of interest were manually generated as control, and a threshold was set for the definition of the ischemic lesion on the basis of the mean signal intensity of the control region of interest. For analysis of CBF and CBV, masks from the perfusion deficit on MTT maps generated with a planimetric method13 were transferred to the corresponding and mirrored CBF and CBV images with the use of the software package MRIcro,14 thus yielding the matching area of the contralateral hemisphere.15 Recanalization was assessed from day 1 MRI on the basis of MR angiography and PWI studies according to the modified Thrombolysis in Myocardial Infarction (TIMI) criteria.16 Vessel occlusion types were determined from MR angiography as follows: 0=no vessel occlusion, 1=internal cerebral artery occlusion at origin, 2=carotid-T occlusion, 3=proximal middle cerebral artery (MCA) trunk occlusion, 4=distal MCA trunk occlusion, and 5=M2 occlusion.16
Statistical analyses were performed as nonparametric because of small (and different) samples sizes and nonnormal distributions. Results are expressed as medians and interquartile ranges (IQRs). Differences between groups were investigated with the Mann-Whitney test (continuous and categorical data) or Fisher’s exact test (frequencies). Differences in infarct volumes and CBV and CBF values between the 3 groups (no TIA, TIA ≤4 weeks, TIA >4 weeks) were analyzed with the Kruskal-Wallis exact test and Mann-Whitney test in a closed testing procedure, securing the multiple level α (in this test procedure the probability of a false-positive decision is not larger than α, regarding all group comparisons). Recanalization and occlusion types were tested with Fisher’s exact test. Probability values are always given as 2-sided and exact. The multiple regression analysis was accomplished with feature selection in a stepdown procedure and without feature selection. Significance was assessed at the P<0.05 level. Statistical analyses were performed with the use of SPSS for Windows (release 11.0.1, SPSS Inc, 1998–2001) and StatXact5 (CYTEL Software Corp, 2001).
Of the patients with acute ischemic stroke admitted to the participating stroke centers during 1997–2001, 139 had MRI assessment within 12 hours of symptom onset according to the joint multiparametric protocol and had clinical as well as MR data available. Among the 65 patients with territorial strokes who met all inclusion criteria, we identified 16 prodromal TIAs. We categorized TIAs on the basis of the time interval between TIA and stroke (≤/>4 weeks) because clinically observed protection had been shown to wane within this time period.1 Four remote TIAs (in a different vascular territory than subsequent stroke, which, according to experimental studies,17 are expected to provide ischemic protection as well) were included in the analysis, 2 of these ≤4 weeks and 2 >4 weeks from stroke. Baseline characteristics of our stroke patient population, presumed stroke etiology, sex, risk factors, thrombolysis, and prestroke medication with acetylsalicylic acid or phenprocoumon (Table 1) resembled previously published cohorts,18 with the exception of a higher prevalence of smokers in the TIA group. Thrombolytic treatment was administered in 40.8% (n=20) of patients without TIA and in 31.3% (n=5) of patients with prodromal TIA at a time interval of 157.5 (IQR, 80.0) minutes and 152.5 (IQR, 83.8) minutes after symptom onset, respectively (P=0.57). Time of first MRI assessment was 4.5 (IQR, 4.5) hours in the no-TIA group and 4.9 (IQR, 4.5) hours in the TIA group (P=0.54). Stroke symptoms and functional disability were significantly less severe in patients with prodromal TIA at the acute stage (<12 hours) as well as on discharge (Table 1).
For lesion volumetry, we calculated ADC and MTT maps from DWI and PWI data, respectively. We further analyzed the area of restricted perfusion delineated on the MTT maps for changes in CBF and CBV after determination of the arterial input function11,12 (Table 2 and Figure).
The final infarct volume was significantly smaller in stroke patients with prodromal TIA than in those with first-ever brain ischemia (final T2: 9.1 [IQR, 19.7] versus 36.5 [IQR, 91.2] mL; P=0.014), which was even more pronounced when only TIAs ≤4 weeks were analyzed (5.6 [IQR, 8.6] mL; P=0.002). Initial ADC lesion volume was smaller and closely resembled final infarct size (12.7 [IQR, 20.4] versus 20.7 [IQR, 40.9] mL), but the Kruskal-Wallis test for the 3 groups did not show statistical significance (P=0.081). However, with a maximal difference of 23.4 versus 8.1, we calculated a power of only 24% (n=16). All groups had approximately the same size of perfusion restriction. Median CBF and CBV values were similar in both groups.13
To evaluate the degree of recanalization and occlusion types, we examined MR angiography and MTT maps of the initial and first follow-up examinations (Table 3). We compared no or minimal recanalization (TIMI 0 and 1) with incomplete and complete recanalization (TIMI 2 and 3) on day 2. Of the 54 patients with sufficient data, there was no significant difference between patients without or with prodromal TIA (P=0.34). Occlusion types were grouped into proximal (categories 1, 2, and 3) and distal (4 and 5) occlusions16 and into carotid (1 and 2) versus other (3, 4, and 5) occlusion types because of the a priori knowledge18 that carotid stenosis is more frequently associated with TIA compared with other subtypes of stroke. While the proportion of proximal and distal occlusion types was similar in both groups (P=1), we found more type 1 occlusions (proximal carotid stenosis) in the prodromal TIA group (P=0.022).
To determine whether this group difference or other factors were determinants of final infarct size, we conducted a multiple linear regression analysis in a stepdown procedure as well as without feature selection, including infarct etiology, age, thrombolysis, smoking, hypertension, diabetes, prodromal TIA, recanalization, and occlusion type. Separately, we tested TIA ≤4 weeks and >4 weeks to account for time of TIA (Table 4; results from feature selection for each model are marked with an asterisk). Carotid occlusion (types 1 and 2) was a significant factor in all models related to large final infarct size. Most importantly, prodromal TIA was a significant factor with a negative relationship to final infarct size, which was also evident without feature selection (P=0.039; 95% CI, −106.1 to −2.9). None of the other factors considered were significant. TIA >4 weeks was not a significant variable, while TIA ≤4 weeks showed significance (P=0.007) only after feature selection. Although the small number of patients in these TIA subgroups may account for the latter results, on the basis of these data, we cannot identify time of TIA as an independent determining variable of final T2.
Despite promising success in animal models, neuroprotective agents for stroke have been disappointing in clinical trials. This raises interest in endogenously occurring mechanisms of ischemic protection in the human brain and their potential therapeutic adaptability.19
In this study we extended the initial clinical observations regarding reduced symptom severity in stroke after prodromal TIA.1,2 The fact that sizes of the perfusion lesion as well as CBF and CBV values were not different between patients with and without prodromal TIAs indicates that both patient groups were subjected to the same extent and severity of flow restriction and microvascular capacity disruption. Nevertheless, infarcts in patients with prodromal TIA were smaller. To our knowledge, this is the first study to suggest that protection by a prodromal TIA is not explained by changes in blood flow, as would be expected from collateral recruitment or enhanced vascularization, but rather by intrinsic neuroprotective mechanisms.
A drawback of this study is the small sample size of the TIA group, which was not overcome with the multicenter approach. Differences in recanalization rate in favor of the TIA group and the higher number of carotid stenoses make interpretation of our data less straightforward, even though these potential confounders were adjusted for in the multiple regression analysis. We believe that the trend toward a higher recanalization rate in the TIA group is unlikely to be the cause for reduced final infarct volumes in this study because at the acute assessment, when vascular occlusion was still present, initial diffusion was already smaller. The higher percentage of smokers and lower incidence of a cardioembolic infarct etiology in the TIA group may also influence outcome measures, even though it does not explain the temporary nature of protection. Still, these findings suggest interesting alternative explanations, such as differences in thrombolytic activity between patient groups, possibly as a result of less tissue destruction or different thrombus characteristics. Acute DWI lesion sizes as well as final infarct volumes were smaller in patients with a prodromal TIA ≤4 weeks from subsequent stroke compared with TIAs >4 weeks in univariate analysis, but time of TIA was not a significant variable of T2 lesion size in regression analysis without feature selection. Therefore, the differences in infarct sizes present in the 2 groups may be due to factors other than time of TIA.
In conclusion, we show that patients with prodromal TIAs display different patterns of diffusion and perfusion deficit in a subsequent stroke. Despite a similar perfusion lesion, their final infarcts are smaller. The only parameter predictive of small final infarct size in a multiple regression analysis was prodromal TIA. Although a TIA is an alarming sign for patients and clinicians indicating an underlying vascular disorder, the potential clinical relevance of our findings lies in its possible application in the clinical setting: many transducers of the preconditioning signal are known from in vitro models. Further delineation and exploration of naturally occurring ischemic tolerance may be a new perspective in future neuroprotection and acute stroke therapy.
MR Acquisition Parameters of the Joint Protocol Used in This Study
MRI was performed on 1.5-T clinical scanners at each study center. The joint protocol included a spin-echo diffusion echo-planar imaging sequence (repetition time [TR] 4700 ms; echo time [TE] 114 ms; matrix 128×128; acquisition time 24 ms) with 2 different b values (0, 1000 s/mm2) and diffusion gradients in 3 orthogonal directions; a multi-echo turbo spin-echo T2-weighted sequence (TR 2900 ms; TE 15, 75, 135 ms; matrix 256×256; acquisition time 246 seconds); a T2*-weighted echo-planar sequence for PWI measurements with 20 mL Magnevist (gadopentetate dimeglumine) followed by 20 mL saline at 4 mL/s with the use of a power injector (Spectris, Medrad) and a temporal resolution of 1 second (TR 2900 ms; TE 15, 30, 45, 60, 75 ms; matrix 128×128; acquisition time 60 seconds); and a time-of-flight angiogram (acquisition time 360 seconds).
This study was supported by the German Ministry of Education and Research (Competence Net Stroke, Berlin NeuroImaging Center) and the Deutsche Forschungsgemeinschaft (Klinische Forschergruppe).
- Received June 15, 2003.
- Revision received October 27, 2003.
- Accepted November 11, 2003.
Moncayo J, de Freitas GR, Bogousslavsky J, Altieri M, van Melle G. Do transient ischemic attacks have a neuroprotective effect? Neurology. 2000; 54: 2089–2094.
Weih M, Kallenberg K, Bergk A, Dirnagl U, Harms L, Wernecke KD, Einhäupl KM. Attenuated stroke severity after prodromal TIA: a role for ischemic tolerance in the brain? Stroke. 1999; 30: 1851–1854.
Adams HP, Bendixen BH, Kapelle LJ, Biller J, Love BB, Gordon DL, Marsh EE, and the TOAST Investigators. Classification of subtype of acute ischemic stroke: definitions to use in a multicenter clinical trial. Stroke. 1993; 24: 35–41.
Fiehler J, Foth M, Kucinski T, Knab R, von Bezold M, Weiller C, Zeumer H, Röther J. Severe ADC decreases do not predict irreversible tissue damage in humans. Stroke. 2002; 33: 79–86.
Rohl L, Ostergaard L, Simonsen CZ, Vestergaards-Poulsen P, Andersen G, Sakoh M, Le Bihan D, Gyldensted C. Viability thresholds of ischemic penumbra of hyperacute stroke defined by perfusion-weighted MRI and apparent diffusion coefficient. Stroke. 2001; 32: 1140–1146.
Rother J, Schellinger PD, Gass A, Siebler M, Villringer A, Fiebach JB, Fiehler J, Jansen O, Kucinski T, Schoder V, et al, for the Kompetenznetzwerk Schlaganfall Study Group. Effect of intravenous thrombolysis on MRI parameters and functional outcome in acute stroke <6 hours. Stroke. 2002; 33: 2438–2445.
Kharbanda RK, Mortensen UM, White P, Kristiansen SB, Schmidt MR, Hoschitzky JA, Vogel M, Sorensen K, Redington AN, MacAllister R. Transient limb ischemia induces remote ischemic preconditioning in vivo. Circulation. 2002; 106: 2881–2883.
Grau AJ, Weimar C, Buggle F, Heinrich A, Goertler M, Neumaier S, Glahn J, Brandt T, Hacke W, Diener HC. Risk factors, outcome, and treatment in subtypes of ischemic stroke: the German Stroke Data Bank. Stroke. 2001; 32: 2559–2566.