Watershed Infarcts in Transient Ischemic Attack/Minor Stroke With ≥50% Carotid Stenosis
Hemodynamic or Embolic?
Background and Purpose— Watershed ischemia is a significant cause of stroke in severe carotid disease, but its pathophysiology is unsettled. Although hemodynamic compromise has long been regarded as the main mechanism—particularly with deep watershed infarction—there is some contradictory evidence from clinical and pathological studies for a role of microembolism, thought to result from plaque inflammation. However, no study so far has directly addressed these conflicting scenarios.
Methods— In 16 consecutive patients with recent transient ischemic attack/minor stroke and ipsilateral 50% to 99% carotid stenosis, we prospectively obtained (1) plaque inflammation mapping with 18F fluorodeoxyglucose positron emission tomography; (2) brain MRI and perfusion MR; and (3) transcranial Doppler detection of microembolic signals (MES). Patients were excluded if on dual antiplatelets or with a potential cardiac source of emboli or contralateral MES.
Results— We found the expected significant relationship between (1) degree of stenosis and severity of distal hemodynamic impairment in the watershed areas; and (2) degree of in vivo plaque inflammation and rate of MES/hr. Deep watershed infarcts were present in 8 patients and MES in 8 (3 with both). There was no systematic association between the presence of deep watershed infarcts and either hemodynamic impairment or MES, but deep watershed infarcts were present only when either hemodynamic impairment or MES was present (P=0.01).
Conclusion— This pilot study supports the idea that in symptomatic carotid disease, deep watershed infarcts result either from hemodynamic impairment secondary to severe lumen stenosis or from microembolism secondary to plaque inflammation. There was no direct evidence that both mechanisms act in synergy.
Cerebral events distal to atherosclerotic carotid disease are thought to be mostly embolic1 and to originate from the unstable plaque as supported by a parallel time course of microembolic signals (MES) detected on transcranial Doppler (TCD).2,3 Plaque inflammation, a key component of this process,4 can now be assessed in vivo using 18F fluorodeoxyglucose positron emission tomography.5
Events distal to severe internal carotid artery (ICA) disease may also be due to hemodynamic impairment (HDI)6–8 and this is particularly relevant to the pathophysiology of watershed (WS) infarction, a significant cause of stroke in severe ICA disease.9 Historically, HDI was regarded as the sole determinant of cortical WS (CWS) infarction owing to the latter’s prevalence after severe systemic hypotension, and indeed HDI has been repeatedly demonstrated in association with CWS infarcts (reviewed by Momjian-Mayor and Baron8). However, some pathological reports also support a role for microemboli in CWS infarcts,8,10 and synergism between these 2 mechanisms has been proposed11 although not directly evidenced so far.
The second type of WS infarct found in carotid disease affects the white matter and is referred to as deep watershed (DWS).12 There is strong evidence that DWS infarcts are more strongly related to HDI than are CWS infarcts,8 but 2 perfusion studies13,14 found HDI not to be systematically associated with DWS infarcts and suggested microemboli may play a role here too, a notion supported by 1 pathological study.15
The present pilot study is the first to directly address these issues by assessing both the cerebral hemodynamics using perfusion-weighted MR (pMR) and the occurrence of MES by TCD in recent patients with transient ischemic attack (TIA)/minor stroke with >50% stenosis of the relevant carotid bifurcation, in whom the presence and distribution of WS infarcts was determined on structural MRI. In addition, carotid plaque inflammation as a potential source of MES was mapped in vivo with 18F fluorodeoxyglucose positron emission tomography (FDG PET). Based on current understanding and literature reviews,8 we hypothesized that DWS infarcts would be related to the presence of HDI, whereas MES would be associated with the presence of plaque inflammation but not of DWS infarcts. Conversely, CWS infarcts would not be exclusively associated with either mechanism.
Patients and Methods
Consecutive patients were prospectively recruited from Addenbrooke’s Hospital TIA clinic with the following clinical inclusion criteria: age >40 years; symptomatic within <3 months with amaurosis fugax, hemispheric TIA, or minor stroke clinically localized to the carotid territory; and 50% to 99% stenosis (North American Symptomatic Carotid Endarterectomy Tria [NASCET] criteria) of the clinically appropriate ICA on duplex ultrasonography. Exclusion criteria were: history of major stroke; anticoagulants or dual antiplatelets (a single antiplatelet agent was permitted because it does not significantly reduce the incidence of MES16); inadequate TCD temporal bone window; presence of MES bilaterally (see subsequently); and concomitant potential cardiac source of embolism clinically or on routine investigations, including patent foramen ovale on echocardiography or TCD contrast study.
Detection of MES was done by simultaneous insonation of both middle cerebral arteries (Doppler box/multi-Doppler X4 with 2-Mz probe; DWL) at 2 depths for 1 hour on the same day as PET and, whenever feasible, repeated on separate days to account for MES variability. Identification of MES followed international consensus recommendations17 (unidirectional signals <0.3 seconds and intensity ≥7 dB over background) and was performed both on- and offline. Patients with bilateral MES were excluded for suspected noncarotid source. MES were recorded as present or absent and their rate expressed as MES/hr.
18F Fluorodeoxyglucose Positron Emission Tomography
The methodology for FDG plaque imaging and associated neck and plaque MRI is described in detail elsewhere.18 Briefly, we used a GE Advance scanner (GE Medical Systems, Milwaukee, Wis) with dynamic scanning in 3-dimensional mode for 75 minutes after injection of 185 MBq of 18F-FDG. A transmission scan was acquired before the FDG scan for attenuation correction. Neck MRI (1.5 T Signa Excite; GE Medial Systems) was obtained to coregister PET to individual anatomy and high-resolution MRI of the carotid plaque to quantify FDG plaque uptake.18 We used standardized uptake value (SUV) as a measure of plaque uptake.18,19 All patients harbored contralateral plaques on high-resolution MRI, as expected, markedly smaller than the index carotid (P=0.001). Regions of interest (ROIs) were defined on the high-resolution MRI and transferred onto the coregistered mean 55 to 75 minutes PET image to measure mean SUV from both carotid plaques. Partial volume effects were corrected as described elsewhere.18 The index/contralateral partial volume effect-corrected SUV ratio was calculated and used in all analyses.
Structural Brain MRI
This was done within 2 days of PET and TCD on the same scanner described previously using fast spin echo T2-weighted (TE 100; TR 5800), fluid-attenuated inversion recovery (TE 121; TR 2181), and diffusion-weighted imaging (DWI; TE 66; TR 4925; 3 directions; b=1000; slice thickness 5 mm). The MR protocol purposely included DWI to maximize the detection and yield of watershed infarcts/ischemic lesions in the population studied12 and hence optimally address our hypothesis.
To determine infarct presence and topography, the MRIs were anonymized and analyzed for the presence of supratentorial infarctions by 2 observers (R.R.M., J.-C.B.) on 2 separate occasions. Infarcts were classified as8: (1) territorial: ≥2 subdivisions of the middle cerebral artery; (2) CWS: cortical border zone between middle cerebral artery and anterior cerebral artery or middle cerebral artery and posterior cerebral artery; (3) other cortical: nonwatershed nonterritorial cortical infarct; (4) DWS infarct: rosary-like, confluent, striated, or solitary located in the supraventricular or paraventricular areas (corona radiata or centrum semiovale), excluding immediately subcortical lesions; and (5) other deep: large striatocapsular lesions (>15 mm), single perforator (<15 mm). There was excellent intraobserver agreement (κ=0.9 for the any lesion, κ=0.82 for CWS lesion, and κ=0.75 for DWS lesions; P<0.01) and interobserver agreement (κ=1, κ=0.71, and κ=0.8, respectively; all P<0.01). Final adjudication was by consensus.
pMR was acquired during the same session using spin echo echoplanar imaging (TE 65; TR 1500; field of view 220 mm; thickness 6 mm, interslice gap 1 mm; 14 slices). Gd-DTPA (0.1 mmol/kg) was injected as a bolus using a power injector and a total of 42 volumes acquired. The data were processed in nordicICE (NordicNeuroLab, Bergen, Norway), which involves automated estimation of the arterial input function and singular value decomposition with block-circulant deconvolution,20 which controls for tracer delay. Maps of relative cerebral blood volume, relative cerebral blood flow, time to peak, and mean transit time (MTT) were generated.
The pMR maps were visually analyzed for markers of HDI in the carotid territory of the symptomatic relative to contralateral hemisphere, namely a delay in MTT or time to peak or increase in cerebral blood volume and/or decrease in cerebral blood flow. Based on these criteria, a judgment was made as to whether HDI was absent, mild (limited abnormalities involving parts of the ICA territory), or significant (extensive and/or severe abnormalities). The analysis was carried out by the same 2 investigators and repeated on a separate session with scans presented in random order with very good intra- and interobserver agreement (κ 0.85 and 0.78, respectively; P<0.05).
To substantiate the visual analysis as well as to focus on the WS areas more specifically, the MTT maps were also quantitatively analyzed by placing ROIs in 3 axial planes (spanning the body of the lateral ventricle) in the symptomatic hemisphere over the anterior CWS and posterior CWS (16-mm diameter circular ROIs) and DWS areas (20×5-mm radius ellipsoid ROIs) based on classic descriptions of the vascular anatomy.21 The ROIs were mirrored on the contralateral hemisphere and a symptomatic/contralateral hemisphere MTT ratio was calculated for each patient for the CWS and DWS areas.
Correlations were assessed using Spearman rho (Kendall τ in case of ties). Nonparametric tests were also used for comparing continuous variables (Wilcoxon signed ranks, Mann-Whitney U test, and Kruskal-Wallis). Proportions were compared using corrected χ2. All values are presented as median and interquartile range unless otherwise specified, and 2-tailed P<0.05 was considered significant.
Clinical details are summarized in Table 1. Ten patients presented with a TIA, of which 6 were hemispheric and 4 retinal. One had both a hemispheric TIA and amaurosis fugax. Six presented with a minor stroke. No patients reported precipitating events such as postural episodes or systemic hypotension; 9 of 16 patients were current or exsmokers but none had chronic respiratory insufficiency.
Lumen stenosis in the index and contralateral carotid arteries was 69.8% (50% to 87.5%) and 0% (0% to 15%), respectively (P<0.001).
On brain MRI (Table 1), 10 of 16 patients had infarcts, all located in the WS except 1 of these patients also had a territorial parietal infarct. Five had infarcts in the DWS only, 2 had infarcts in the CWS only, and 3 had infarcts in both. Of the 8 patients who had DWS infarcts, 4 had a rosary-like pattern in the centrum semiovale.
pMR was marred by head movement in 1 patient (Patient 9). On visual analysis, HDI was present in 7 of 15 patients (HDI+) and absent in 8 (HDI−; Table 1). The MTT ratio was significantly higher in the DWS areas in patients with significant visually assessed pMR abnormalities (n=4) than in those with mild (n=3) or no abnormality (1.18 [1.0 to 1.3] versus 0.87 [0.68 to 0.9] versus 0.92 [0.81 to 0.98], respectively; P=0.05). Figure 1 illustrates 3 possible associations between HDI and DWS infarcts in 4 patients.
MES assessment was of inadequate quality in 1 patient (Patient 7). MES were present in 7 patients (MES+) and absent in 8 (MES−; Table 1).
Lumen Stenosis Versus HDI and MES
The degree of ICA stenosis differed significantly between patients with different grades of HDI, being highest in those with significant HDI than in those with mild or absent HDI (91% [87.5% to 95%] versus 50% [50% to 75%] versus 57.5% [50% to 76%], respectively; P=0.01). There was also a positive correlation between percent stenosis in the index carotid and the MTT ratio in the DWS (ρ=0.51; P=0.05) but not in the CWS. The degree of stenosis was not significantly different in MES+ and MES− patients (55% [50% to 90%] versus 81.3% [52.5 to 87.5%], respectively; P>0.05). There was minimal or no HDI in the 3 patients who presented with amaurosis fugax only (Table 1).
Plaque Inflammation Versus MES and HDI
There was a very significant positive correlation between the rate of MES/hr and the index/contralateral partial volume effect-corrected SUV ratio (τ=0.64, P=0.003; Figure 2). The time interval from symptom onset to PET scanning was not different in MES+ and MES− patients (56 [21 to 70] days versus 30.5 [9.5 to 80.5] days, respectively; P=0.68). There was no difference in index/contralateral partial volume effect-corrected SUV ratio between patients with significant HDI on pMR and those with mild and or absent pMR abnormalities (76 [0.53 to 1.18] versus 1.01 [0.62 to 1.14] versus 0.99 [0.86 to 1.29], respectively; P=0.5).
WS Infarction Versus HDI and Microembolism
On visual analysis, DWS infarcts were present in 5 of 7 HDI+ and 3 of 8 HDI− patients (P>0.05). Consistent with this visual analysis, there was no difference in MTT ratio between patients who did and those who did not have DWS infarcts on MRI (P>0.05 for all ROIs). Similarly, the presence of a rosary-like pattern on MRI was not associated with a different MTT ratio in either the CWS or the DWS areas (P>0.05 for both). Regarding MES, DWS infarcts were present in 4 of 8 MES− patients but also in 3 of 7 MES+ patients (P>0.05).
These results did not support our hypothesis of a systematic association between DWS infarcts with HDI but not with MES. However, they suggested that DWS infarcts can be associated with either HDI or MES. Accordingly, we then tested this double association. The results are shown in Table 2. None of the 4 patients who did not have either HDI or MES had DWS infarcts, whereas 7 of 10 patients with HDI, MES, or both had DWS infarcts (P=0.01). Due to their small incidence (Table 1), no further analysis was considered regarding CWS infarcts.
In this pilot study on recently symptomatic patients with TIA/minor stroke with >50% ICA stenosis, we found significant relationships between degree of stenosis and severity of hemodynamic impairment in the WS and between degree of PET-based plaque inflammation and density of microembolic signals on TCD, both biologically expected. However, contrary to our hypothesis, there was no systematic association between the presence of DWS infarcts and HDI nor between their absence and MES. In other words, inconsistent with our hypothesis, DWS infarcts were observed in some but not all patients with HDI and conversely were present in some patients without it. This led us to test whether DWS infarcts would be associated with the presence of either HDI or MES but not when both are absent. The data, shown in Table 2, support this post hoc hypothesis. Overall, therefore, our results suggest that, in this sort of population, WS infarcts are not exclusive to either mechanism but appear to develop secondary to microembolism from plaque inflammation, HDI from plaque stenosis, or both in association (Figure 3). Finally, due to the low occurrence of CWS infarcts, no meaningful conclusion about their pathophysiology could be reached.
Although there is ample evidence for the correlation of MES with carotid atherosclerotic disease,2,3 the precise mechanism underlying microemboli formation is not entirely clear. In 1 study, the presence of MES was associated with higher plaque macrophage inflammatory burden on carotid endarterectomy specimen.22 By showing in vivo a strong relationship between MES and PET-derived plaque inflammatory content, our present results corroborate these earlier findings.
The association found between HDI (especially in the DWS areas) and lumen stenosis was also expected. Cerebral HDI is largely determined by the degree of carotid stenosis.23,24 Conversely, lumen stenosis is known not to be directly associated with plaque inflammation, with plaques silently growing to cause lumen narrowing until inflammation sets in and causes them to become unstable and potentially symptomatic.4
The incidence of WS infarcts in this study (10 of 16 patients) may seem high, but their detection was optimized by adding DWI to T2 and fluid-attenuated inversion recovery sequences (see Table 1). In addition, the population studied was selected a priori to represent patients with recent TIA or ministroke (ie, not leaving any deficit beyond 7 days) and a clinically appropriate ≥50% stenosis, carefully excluding patients with completed stroke or a potential proximal source of emboli. There is no previous study on the incidence of WS infarcts in a similar population using both standard and DWI sequences to compare with, but 1 previous similar study found WS infarcts, including DWS infarcts, in one third of patients with acute stroke with >50% carotid disease.12
Regarding DWS infarcts first, the most salient finding from this investigation is that, contrary to our hypothesis, their prevalence was not different between the HDI+ and HDI− groups nor between the MES+ and MES− groups. The association of HDI to DWS infarction in carotid disease has been found in many studies using PET, single photon emission CT, pMR, or TCD (reviewed in Momjian-Mayor and Baron8). Yet, 3 studies found inconsistent results13,14,25 and proposed that emboli may play a significant role in DWS infarcts. However, because none of these studies included detection of MES in their design, any role ascribed to embolic mechanisms was by inference only. Our results derived from directly assessing both MES and HDI in the same patients are the first to provide direct evidence that either HDI or microembolism may cause DWS infarction. Thus, probably because of their small size,26 some microemboli may enter cortical “distal field” arterioles and from there medullary arteries, in turn causing centrum semiovale infarcts in the DWS territory.
Although HDI and MES may work separately to cause DWS infarcts, it is also plausible that they act synergetically. Thus, decreased perfusion may promote brain damage from otherwise innocuous microemboli entering vulnerable areas with an exhausted vascular reserve; reciprocally, by blocking end arteries, small emboli could further impair perfusion. Furthermore, reduced perfusion may facilitate local thrombosis and/or impede “clearance” of microembolic material.11 In our study, however, DWS infarcts were present in only 1 of 3 HDI+/MES+ patients, and conversely the coexistence of both mechanisms was not necessary for their occurrence. Our preliminary data do not therefore provide strong support to this hypothesis.
Evidence in favor of microembolism playing an important part is much stronger for CWS infarcts, both from hemodynamic studies8 and from pathological and experimental studies.10,15,27,28 Although their incidence (n=5) in the present study precluded formal statistical analysis, it is worth noting that MES were present in 3 of these patients—without HDI in 2—further strengthening the growing notion that CWS infarcts often result from microemboli alone.
This study has limitations. Due to the complexity of the protocol, we were able to recruit only a small sample so the results require replication before the conclusions are considered definitive. Because the investigations were not conducted immediately close in time to the symptoms, it is possible that either the MES would have subsequently cleared or that the HDI had subsided, although the latter tends to remain stable over short timespans.24 The possibility that the HDI was missed because it affected highly circumscribed WS areas that went on to infarct is unlikely, because previous evidence indicates that when present, HDI in carotid disease is quite extensive, spreading well beyond any WS infarcts.6,14,23,24
The present study provides direct support to the idea that in symptomatic ICA disease, DWS infarcts may result from HDI through severe lumen stenosis, but also from microemboli alone through plaque inflammation. These results, although preliminary, imply that the finding of DWS infarcts on structural imaging in patients with severe carotid stenosis may be compatible with an underlying microembolic mechanism from plaque inflammation, which has management implications.
We acknowledge the assistance provided by Diana J. Day, Jennifer Mitchell, and Tulasi Marrapu.
Sources of Funding
This study was funded by a program grant from the Medical Research Council to J.-C.B. (G0001219) and a British Heart Foundation (BHF) Grant (FG/03/013). D.I.-G. was supported by grants from the BHF. This work was also supported by the Comprehensive Biomedical Research Centre grant to Cambridge University Hospitals National Health Service Foundation Trust and Cambridge University (Neurosciences Theme and Cardiovascular Theme). R.R.M. is supported by a grant from the Cambridge Overseas Trust, UK.
D.I.-G. and P.S.J. contributed equally to this work.
- Received January 29, 2010.
- Revision received March 10, 2010.
- Accepted March 17, 2010.
Goertler M, Blaser T, Krueger S, Hofmann K, Baeumer M, Wallesch CW. Cessation of embolic signals after antithrombotic prevention is related to reduced risk of recurrent arterioembolic transient ischaemic attack and stroke. J Neurol Neurosurg Psychiatry. 2002; 72: 338–342.
Valton L, Larrue V, le Traon AP, Massabuau P, Geraud G. Microembolic signals and risk of early recurrence in patients with stroke or transient ischemic attack. Stroke. 1998; 29: 2125–2128.
Spagnoli LG, Bonanno E, Sangiorgi G, Mauriello A. Role of inflammation in atherosclerosis. J Nucl Med. 2007; 48: 1800–1815.
Baron JC, Bousser MG, Rey A, Guillard A, Comar D, Castaigne P. Reversal of focal ‘misery-perfusion syndrome’ by extra-intracranial arterial bypass in hemodynamic cerebral ischemia. A case study with 15O positron emission tomography. Stroke. 1981; 12: 454–459.
Klijn CJ, Kappelle LJ, Tulleken CA, van Gijn J. Symptomatic carotid artery occlusion. A reappraisal of hemodynamic factors. Stroke. 1997; 28: 2084–2093.
Momjian-Mayor I, Baron JC. The pathophysiology of watershed infarction in internal carotid artery disease: review of cerebral perfusion studies. Stroke. 2005; 36: 567–577.
Ringelstein EB, Zeumer H, Angelou D. The pathogenesis of strokes from internal carotid artery occlusion. Diagnostic and therapeutical implications. Stroke. 1983; 14: 867–875.
Szabo K, Kern R, Gass A, Hirsch J, Hennerici M. Acute stroke patterns in patients with internal carotid artery disease: a diffusion-weighted magnetic resonance imaging study. Stroke. 2001; 32: 1323–1329.
Chaves CJ, Silver B, Schlaug G, Dashe J, Caplan LR, Warach S. Diffusion- and perfusion-weighted MRI patterns in borderzone infarcts. Stroke. 2000; 31: 1090–1096.
Georgiadis D, Lindner A, Manz M, Sonntag M, Zunker P, Zerkowski HR, Borggrefe M. Intracranial microembolic signals in 500 patients with potential cardiac or carotid embolic source and in normal controls. Stroke. 1997; 28: 1203–1207.
Ringelstein EB, Droste DW, Babikian VL, Evans DH, Grosset DG, Kaps M, Markus HS, Russel D, Siebler M. Consensus on microembolus detection by TCD. Stroke. 1998; 29: 725–729.
Izquierdo-Garcia D, Davies JR, Graves MJ, Rudd JH, Gillard JH, Weissberg PL, Fryer TD, Warburton EA. Comparison of methods for magnetic resonance-guided [18-F]fluorodeoxyglucose positron emission tomography in human carotid arteries. Reproducibility, partial volume correction, and correlation between methods. Stroke. 2008.
Rudd JH, Myers KS, Bansilal S, Machac J, Pinto CA, Tong C, Rafique A, Hargeaves R, Farkouh M, Fuster V, Fayad ZA. Atherosclerosis inflammation imaging with 18F-FDG PET: carotid, iliac, and femoral uptake reproducibility, quantification methods, and recommendations. J Nucl Med. 2008; 49: 871–878.
Jander S, Sitzer M, Schumann R, Schroeter M, Siebler M, Steinmetz H, Stoll G. Inflammation in high-grade carotid stenosis: a possible role for macrophages and T cells in plaque destabilization. Stroke. 1998; 29: 1625–1630.
Yamauchi H, Fukuyama H, Nagahama Y, Oyanagi C, Okazawa H, Ueno M, Konishi J, Shio H. Long-term changes of hemodynamics and metabolism after carotid artery occlusion. Neurology. 2000; 54: 2095–2102.
Derdeyn CP, Simmons NR, Videen TO, Yundt KD, Fritsch SM, Carpenter DL, Grubb RL Jr, Powers WJ. Absence of selective deep white matter ischemia in chronic carotid disease: a positron emission tomographic study of regional oxygen extraction. AJNR Am J Neuroradiol. 2000; 21: 631–638.