If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To assess the incidence of post-operative non-ischaemic cerebral complications as a pivotal outcome parameter with respect to size of cerebral infarction, timing of surgery, and peri-operative management in patients with symptomatic carotid stenosis who underwent carotid endarterectomy (CEA).
Retrospective analysis of prospectively collected single centre CEA registry data. Consecutive patients with symptomatic carotid stenosis were subjected to standard patch endarterectomy. Brain infarct size was measured from the axial slice of pre-operative computed tomography/magnetic resonance imaging demonstrating the largest infarct dimension and was categorised as large (> 4 cm2), small (≤ 4 cm2), or absent. CEA was performed early (within 14 days) or delayed (15 – 180 days) after the ischaemic event. Peri-operative antiplatelet regimen (none, single, dual) and mean arterial blood pressure during surgery and at post-operative stroke unit monitoring were registered. Non-ischaemic post-operative cerebral complications were recorded comprising haemorrhagic stroke and encephalopathy, i.e., prolonged unconsciousness, delirium, epileptic seizure, or headache.
646 symptomatic patients were enrolled of whom 340 (52.6%) underwent early CEA; 367 patients (56.8%) demonstrated brain infarction corresponding to stenosis induced symptoms which was small in 266 (41.2%) and large in 101 (15.6%). Post-operative non-ischaemic cerebral complications occurred in 12 patients (1.9%; 10 encephalopathies, two haemorrhagic strokes) and were independently associated with large infarcts (adjusted odds ratio [OR] 6.839; 95% confidence interval [CI] 1.699 – 27.534) and median intra-operative mean arterial blood pressure in the upper quartile, i.e., above 120 mmHg (adjusted OR 13.318; 95% CI 2.749 – 64.519). Timing of CEA after the ischaemic event, pre-operative antiplatelet regimen, and post-operative blood pressure were not associated with non-ischaemic cerebral complications.
Infarct size and unintended high peri-operative blood pressure may increase the risk of non-ischaemic complications at CEA independently of whether performed early or delayed.
For patients undergoing carotid endarterectomy (CEA) for symptomatic carotid stenosis the size of the corresponding infarct is an independent predictor of non-ischaemic complications such as encephalopathy and haemorrhagic stroke. These hyperperfusion induced complications are related neither to timing of CEA (early vs. delayed) nor intensity of antiplatelet premedication but to an intra-operative blood pressure level of stage 2 hypertension during most of the surgical time, i.e., a median intra-operative mean arterial blood pressure of 120 mmHg or above.
In patients with a transient ischaemic attack (TIA) or a minor stroke due to severe carotid stenosis nowadays endarterectomy is recommended within two weeks of symptom onset
The former recommendation for late surgery was intended to prevent cerebral haemorrhage by not exposing patients with acute ischaemia induced blood brain barrier disruption to operative re-perfusion and peri-operative antiplatelet or anticoagulant agents.
A large infarct is often associated with severe functional deficit, questioning the benefit of a prophylactic intervention and whether severely affected patients should undergo early carotid intervention before potential successful rehabilitation. Thus, management of carotid endarterectomy (CEA) in patients with a substantial cerebral infarction has gained increasing significance in terms of balancing risk and benefit.
In the present study, the hypothesis that infarct size, timing of CEA, and peri-operative management in patients with symptomatic carotid artery stenosis are associated with post-operative encephalopathy and haemorrhagic stroke was investigated. It was postulated that hyperperfusion, triggered by inflammatory reaction of small vessels in necrotic and peri-infarct tissue, could lead to headache, epileptic seizure, or even local bleeding.
Retrospective analyses were performed on prospectively collected data of the Magdeburg University Vascular and Stroke Centre Registry.
Materials and methods
Patients admitted to the Magdeburg University Vascular and Stroke Centre with an ischaemic cerebrovascular or retinal event underwent standardised admission and work up. All diagnostic and therapeutic procedures were performed within standard clinical practice and prospectively collected data were entered into the institution’s registry in accordance with federal and national data protection laws. Therefore, the responsible ethics committee confirmed the study and waived the need for patient consent for data analyses. For the present retrospective study, consecutive patients admitted with a symptomatic extracranial carotid stenosis, i.e., symptoms within the last 180 days assigned to the hemisphere or the ophthalmic artery ipsilateral to the stenosis, which underwent subsequent secondary preventive CEA at the institution between January 1997 and December 2017 were selected. TIA and stroke were classified according to clinical World Health Organisation definition independently from imaging findings.
The standardised pre-operative diagnostic procedure included documentation of medical history (including antithrombotic and statin therapy), neurological examination, laboratory examination, electrocardiogram, echocardiography (if clinically indicated), and cerebral imaging (computed tomography [CT] or magnetic resonance imaging [MRI]). Vascular pathology was assessed and quantified regularly in all patients by Doppler and colour coded duplex sonography. Since 2011 patients admitted with an acute stroke underwent admission CT angiography (CTA). In case of ambiguous sonographic findings without prior CTA, additional digital subtraction angiography (until 2005) or supporting magnetic resonance angiography (since 2002) were performed to diagnose stenosis. The degree of carotid stenosis was quantified sonographically according to published criteria
and expressed as distal diameter reduction (North American Symptomatic Carotid Endarterectomy Trial [NASCET] method) in steps of 10%. In general, CEA was considered as indicated for a stenosis of ≥ 60%. Patients with a sufficient transcranial bone window underwent transcranial Doppler carbon dioxide testing to assess cerebrovascular reserve capacity.
The size of the corresponding infarct was measured in the axial CT or MRI slice demonstrating the largest dimension of the infarct. Within the MRI, diffusion weighted sequences were used for measurements for acute ischaemia and fluid attenuated inversion recovery sequences for ischaemic events which occurred some time ago and no longer demonstrated diffusion weighted hyperintensity. The shape of the (largest) infarct was categorised as rectangular, triangular, circular/oval, or parallelogram shaped and its size was subsequently calculated according to the appropriate formula (Fig. 1). Infarct size was dichotomised as large, i.e., exceeding 4 cm
, or small, i.e., at maximum of 4 cm2, because infarcts exceeding a diameter of 2 – 2.5 cm diameter – approximately corresponding to a size of 4 cm2 – are often associated with major stroke and were excluded in former studies.
To investigate the whole range of infarct size, patients without infarct, e.g., in case of ocular symptoms, were also included and infarct size was assessed as 0 cm2.
For multiple infarcts, areas were not summated, but the largest infarct size was used because the distance of the infarct core from normal perfused tissue was considered as a key parameter for blood brain barrier breakage and subsequent complications. However, a potential effect of multiple and confluent infarcts on outcome was investigated by recording and including both, as co-variables, into multivariable analyses.
Anaesthetic and operative procedures
Until 2001, CEA was performed as standard patch endarterectomy under general balanced anaesthesia. Since 2002 regional anaesthesia has been performed as combined deep and superficial cervical plexus blockade. Patients under general anaesthesia were neuromonitored by transcranial Doppler sonography for selective shunt placement during carotid cross clamping; shunt placement was obligatory in those without a sonographic temporal bone window. During regional anaesthesia patients were monitored clinically by means of repetitive fist closure contralateral to the CEA side. For neurological worsening while clamping, a selective shunt was placed. Heparin was applied intravenously before carotid cross clamping followed by intravenous protamine sulphate at the end of surgery usually - at a dose to compensate for half of the administered heparin. Intra-operative completion angiography was carried out sporadically until 2003 and has been implemented as routine quality control since 2004. Peak, lowest, and median mean arterial blood pressure (MABP) during surgery and during 24 hours post-operative monitoring were assessed by close cardiovascular monitoring for each patient.
Post-operative cerebral complications and outcome
Non-ischaemic cerebral complications, i.e., post-operative encephalopathy and haemorrhagic stroke, were the primary outcomes of this study. Post-operative new onset ischaemic stroke served as a secondary outcome variable.
Post-operative neurological complications were diagnosed by a neurologist who routinely monitored patients during hospitalisation. Only patients with new onset clinical symptoms underwent routine CT/MRI post-operatively.
Encephalopathy was diagnosed in patients with prolonged post-operative disturbed consciousness after general anaesthesia impeding extubation (somnolence/sopor/coma), convulsion, headache ipsilateral to the CEA side, and/or delirium. The diagnosis of delirium was based on clinical findings, i.e., acute onset and fluctuating course of disturbed consciousness, cognitive disturbance, psychomotor agitation, and/or sleep wake phase disorder. Patients with encephalopathy symptoms underwent CT/MRI scanning and transcranial colour coded duplex sonography (TCCD) to differentiate between an ischaemic, haemorrhagic, or other aetiology. Stroke was diagnosed in case of new onset or worsened focal neurological deficit with compatible finding in subsequent CT/MRI.
A focal ischaemic event was considered a surgical complication if occurring within 30 days after CEA. Post-hospitalisation data were recorded at the six week post-operative routine outpatient follow up examination. Encephalopathy was considered as associated with surgery if it occurred post-operatively during hospitalisation.
Statistical analyses were performed with the SPSS26 software package. Baseline characteristics and surgical procedures were presented as frequencies, mean/standard deviation, and median/interquartile range (IQR) as appropriate. Backward multivariable logistic regression analysis was used to investigate the influence and independence of the variables of interest, i.e., infarct size, timing of CEA, and peri-operative antithrombotic and blood pressure status, on post-operative cerebral complications. Investigated variables of interest were forced into regression analysis within the first block. The year of surgery was forced into the first block to allow for compensation of time trends within the observation period in addition to co-variables known to have changed within the observation period. Independence of the variables of interest from confounding ones as listed in Table 1, Table 2 – including all peri-procedural and surgical variables having changed within the observation period (Table 3) – was estimated by backward stepwise removal (p > .10) and entry selection (p < .050) of confounding variables at the second block based on their likelihood ratios. Estimated relative risks and 95% confidence intervals (CI) of variables of interest and of significant co-variables on the primary and secondary outcome event were calculated. For continuous variables of interest, additional calculations were performed for clinically relevant dichotomisations; i.e., infarct size (large vs. small or absent), timing of CEA (early vs. delayed), and blood pressure (upper quartile vs. lower three quartiles). Significance was set at a p value < .050.
Table 1Baseline characteristics of 646 consecutive patients with symptomatic carotid stenosis undergoing carotid endarterectomy (CEA) studied for relevance of infarct size, timing of surgery and peri-operative management for non-ischaemic cerebral complications
Demographic and peri-operative characteristics
Patients (n = 646)
Age – y
67.3 ± 10.5
Ischaemic heart disease
Type of ischaemic event
Ischaemic deficit at CEA
Minor, modified Rankin scale 1 to 2
Major, modified Rankin scale 3 to 5
Cerebral infarction on CT/MR images
Maximal axial infarct size ≤4 cm2
Maximal axial infarct size >4 cm2
Time from ischemic event to CEA – d
Early CEA, 0 to 14 days after ischemic event
Degree of carotid stenosis
Moderate, <70% NASCET
Severe, ≥70% NASCET
Contralateral carotid occlusion/severe stenosis
Unknown, missing temporal bone window
Antiplatelet agents at CEA
Statins since ≥2 weeks before CEA
Data are presented as n (%), mean ± stand deviation, or median (interquartile range). NASCET = North American Symptomatic Carotid Endarterectomy Trial; TIA = transient ischaemic attack.
Table 3Longitudinal course of peri-operative and intra-operative variables during the 21 years observation period in 646 consecutive patients with symptomatic carotid stenosis undergoing carotid endarterectomy (CEA)
1997–2003 (n = 210)
2004–2010 (n = 191)
2011–2017 (n = 245)
Early (vs. delayed) CEA
Time from ischaemic event to CEA – d
MRI (vs. CT) prior to CEA
Time from ischaemic event to CT – d
Time from ischaemic event to MRI – d
Regional (vs. general) anaesthesia
No antiplatelet agent at CEA
Single antiplatelet agent at CEA
Dual antiplatelet agents at CEA
Statins since ≥2 weeks before CEA
Data are presented as n (%) or median (interquartile range). CEA = carotid endarterectomy.
Three hundred and thirty-three of 646 symptomatic CEA patients (51.6%) underwent pre-operative brain CT, and 313 patients (48.4%) brain MRI. Three hundred and sixty-seven patients (56.8%) demonstrated brain infarction corresponding to stenosis induced symptoms with a median of maximum infarct size of 1.8 cm2 (IQR 0.6, 4.4). If dichotomised into small and large infarct size, median maximum infarct size in those 266 patients with a small infarct was 1.1 cm2 (IQR 0.5, 2.3), and it was 7.1 cm2 (IQR 5.2, 10.7) in those 101 with a large infarct. Thirty-six of 101 patients (35.6%) with a large infarct, 159 of 266 (59.8%) with a small infarct, and 118 of 279 (42.3%) without an infarct underwent MRI as the pre-operative imaging technique. Multiple infarcts were found in 202 patients with brain infarction; they were more frequently detected in patients with small infarcts (167/266, 62.8%) than in those with large infarcts (35/101, 34.7%) and were confluent in 67 (33.2%).
Three hundred and forty patients (52.6%) underwent early CEA; in 306 patients (47.4%) CEA was delayed. Median time from index ischaemia to CEA in those operated early was 5.3 days (IQR 2.7, 8.7), and 43.8 days (IQR 23.5, 80.1) in those with delayed operation.
One hundred and seventy-nine CEA patients had a median intra-operative MABP in the upper quartile (median 125 mmHg), 73 of those (40.8%) despite intra-operative application of one or more blood pressure lowering agents. In those 467 with median intra-operative MABP below the 75th percentile, 131 patients (28.1%) had received blood pressure lowering medication during surgery. Blood pressure elevating agents were administered in 33 of 179 (18.4%) patients with intra-operative MABP in the upper quartile, and 179 of 467 patients (38.3%) with intra-operative MABP below the 75th percentile. During the 24 hours post-operative monitoring in the intensive care or stroke unit, 154 patients had a median MABP in the upper quartile (median, 103 mmHg). Further baseline characteristics and operative data are presented in Table 1, Table 2.
As a consequence of the long observation period, surgical routines and peri-operative procedures changed over time. The longitudinal course of those variables is presented in Table 3.
Post-operative cerebral complications
Thirty-eight (5.9%) patients suffered from adverse post-operative cerebral symptoms, which were non-ischaemic in 12 (1.9%) and ischaemic in 26 (4.0%). Non-ischaemic complications were focal in two patients (caused by intracerebral haemorrhage) and encephalopathic in 10. Encephalopathy appeared as delirium in six, epileptic seizure in three, and headache in one patient. They were caused by cortical subarachnoid bleeding in one, cerebral hyperperfusion in six, and unknown in three patients. The last three had no TCCD bone window to verify aetiology, the others had confirmed hyperperfusion by TCCD. Eight encephalopathic patients recovered by discharge and two were discharged with minor clinical signs of delirium. Both patients with intracerebral haemorrhage suffered from persisting functional deficit (modified Rankin Scale 3) after three months.
All 26 ischaemic complications (4.0%) were caused by de novo brain infarction, resulting in a persisting functional deficit in 24 and death in two patients. An additional four patients (0.6%) died post-operatively between 14 and 21 days of surgery, two from myocardial infarction, one from Listeria meningitis under immunosuppression, and one from massive upper gastrointestinal bleeding.
Risk factors for non-ischaemic cerebral complications
Larger pre-operative infarcts were an independent predictor of post-operative encephalopathy and haemorrhagic stroke (Table 4). If dichotomised by infarct size, patients with a large infarct had a 6.839 fold higher risk (adjusted OR; 95% CI 1.699 – 27.534; p = .007) than those without or with only a small infarct. Higher intra-operative blood pressure was associated with a higher risk of non-ischaemic cerebral complications, too (Table 4). After dichotomisation, median MABP within the upper quartile (120 – 190 mmHg), led to a 13.318 fold higher risk (adjusted OR; 95% CI 2.749 – 64.519; p = .001) than a median MABP below 120 mmHg. As high blood pressure might had been intended before final declamping and restoration of blood flow because of suspected low cerebral perfusion pressure, a subsequent analysis was performed that also included intra-operative blood pressure management. This analysis demonstrated that only patients with an intra-operative median MABP in the upper quartile despite intra-operative antihypertensive therapy showed an increased risk (adjusted OR 23.627; 95% CI 4.394– 127.049; p < .001), but not patients in the upper quartile without antihypertensive or with blood pressure elevating therapy – thus with potential need for high blood pressure (adjusted OR 1.616; 95% CI 0.140 – 18.696; p = .70). Intra-operative peak and lowest blood pressure as well as blood pressure levels/values during 24 hours post-operative monitoring were not associated with the patients’ outcome. Neither CEA timing after the ischaemic event, type of anaesthesia, peri-operative antithrombotic drugs, nor multiple infarcts nor their confluence were associated with the appearance of non-ischaemic cerebral complications.
Table 4Multivariate binary logistic regression demonstrating infarct size and median intra-operative mean arterial blood pressure (MABP) as independent predictors of post-operative encephalopathy and haemorrhagic stroke in 646 patients treated with carotid endarterectomy (CEA) for symptomatic carotid stenosis
Adjusted OR (95% CI)
Year of observation period
Time period from event to CEA per day
Infarct size per cm2
Single antiplatelet medication at CEA
Dual antiplatelet medication at CEA
≥ 5000 IU heparin not antagonised at end of CEA
Median MABP during CEA per mmHg
Age – y
MABP = mean arterial blood pressure; OR = odds ratio; CI = confidence interval; CEA = carotid endarterectomy; IU = international units.
Antiplatelet therapy up to the day before CEA reduced the risk of a post-operative ischaemic stroke in the cohort to one third (single antiplatelet; adjusted OR 0.302; 95% CI 0.112 – 0.812; p = .018) or even one sixth (dual antiplatelets; adjusted OR 0.167; 95% CI 0.042 – 0.661; p = .011) compared with pausing this therapy several days before surgery. Regional anaesthesia was the only other variable in the series independently associated with the risk of a post-ischaemic stroke, substantially lowering that risk compared with general anaesthesia (adjusted OR 0.242; 95% CI 0.073 – 0.799; p = .020).
The post-operative risk of non-ischaemic cerebral complications, i.e., encephalopathy and haemorrhagic stroke, was more than six fold higher in patients with large infarcts than in those without or with only small infarcts. Even though the existence vs. the absence of an infarct is a known risk factor of non-ischaemic complications after CEA,
a strong relationship between those symptoms and the infarct size was demonstrated. Timing of surgery or antiplatelet therapy was not associated with the occurrence of non-ischaemic cerebral events. Intra- and post-operative cerebral hyperperfusion was corroborated in seven and may be considered as a probable cause in the other patients with non-ischaemic symptoms. Thus, a hyperperfusion syndrome after CEA in this study accounted for 1.9% of complications which is comparable to previous studies.
Hyperperfusion after recanalisation, triggered by inflammatory reaction of small vessels in necrotic and peri-infarct tissue, seems to depend on infarct volume. No correlation with CEA timing could be demonstrated, suggesting that effect lasts longer than 14 days.
Hypertension – in general and peri-operatively – has been identified as a risk factor for non-ischaemic complications.
However, the results enable a more specific evaluation concerning the contribution of hypertension to that risk. A therapy resistant, stage 2 hypertensive intra-operative blood pressure level during the majority of surgery time, assessed as a median MABP of 120 mmHg or above, seems to be the key risk factor. Neither temporary intra-operative hypertensive peaks, elevated post-operative blood pressure during stroke unit monitoring, nor an intra-operative iatrogenic increased blood pressure to compensate for low cerebral perfusion, especially during carotid cross clamping, contributed additionally to that risk. An at most stage 1 hypertensive intra-operative blood pressure, spontaneously or after administration of antihypertensive agents, served as the only protective factor against these complications.
An association between pre-operatively assessed exhausted cerebrovascular reserve and non-ischaemic complications could not be evaluated. However, regional loss of cerebrovascular reactivity even in large infarcts may not be detected by pre-operative testing at the level of the middle cerebral artery and a different kind of perfusion monitoring was not applicable. Contralateral carotid occlusion/severe stenosis was not associated with non-ischaemic symptoms in this series, even though results in the literature are discrepant.
This may be explained by a varying collateralisation in patients with severe bilateral carotid disease which usually is not investigated. Patients with a predominant collateralisation via the posterior communicating arteries may not be at risk in case of contralateral severe carotid disease.
Due to the quite different shapes of infarcts, area measurements were considered more representative of the infarct extent than a single diameter (Fig. 1). The calculation of the “true” infarct volume from consecutive CT/MRI slices was considered too complex for routine daily performance either for scans done on the institutional CT/MRI or from external images from outpatients. Although infarcts exceeding 4 cm2 were defined on an axial CT/MRI slice as large, it should be emphasised that the median size of large infarcts was 7.1 cm2 with an upper IQR of 10.7 cm2. Thus, the results may not or only with caution be transferred to infarcts involving for example most of the middle cerebral artery territory.
The findings confirm two trends on infarct imaging which might bias the results: the proportion of MRI in contrast to CT scans increased over time during the observation period and the proportion of MRI scans was higher in patients with small infarcts compared with those without or with large infarcts. Thus, sensitivity of the detection of small infarcts may has increased over time. However, the imaging technique did not predict non-ischaemic complications on multivariable regression analyses.
Further on, it was found that continuance of antiplatelet agents until immediately before surgery, substantially reduced the risk of a post-operative ischaemic stroke, which is concordant with literature findings in patients with atherothrombotic stroke not submitted to CEA.
The single centre analyses and the long observation period with changing operative and peri-operative practice/standards appears at first glance as a limitation. An attempt to compensate for this was made by including known and unknown variables which varied over time, the latter represented by the year of CEA, into multivariable regression analyses. The values of the co-variables (like type of anaesthesia or shunt placement) did not differ enough to enable investigation of their effects on outcome variables without an unacceptably high type 2 error.
The findings are in line with the results of nationwide German registry data
to provide regional anaesthesia not only to improve peri-procedural monitoring but to reduce ischaemic complications, even though a Cochrane analysis and review could not prove a better outcome for patients operated on under regional in contrast to general anaesthesia.
Predominantly thromboembolic events causing ischaemic complications may be recognised intra-operatively in clinically monitored patients with higher sensitivity compared with any technical monitoring or shunted unmonitored patients. The practice of immediate angiography and subsequent intra-operative endovascular thrombectomy for intracranial large vessel occlusion might be responsible for this result.
To conclude, not only the existence but also the size of an infarct increases the risk of non-ischaemic post-operative complications like haemorrhagic stroke and encephalopathy irrespective of the timing of CEA. An at most stage 1 hypertensive intra-operative blood pressure level, spontaneously or after intra-operative antihypertensive treatment, helps to reduce that risk. Antiplatelet agents (even dual) continuing to immediately before surgery may not increase the risk of non-ischaemic complications but withholding them pre-operatively may be associated with an elevated risk of a post-operative ischaemic stroke.
Conflict of interest and funding
We thank Jane Heisinger for excellent assistance in sonographic examinations and Annette Glaenzel for competent CEA register documentation.
Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/American Stroke Association.
To submit a comment for a journal article, please use the space above and note the following:
We will review submitted comments as soon as possible, striving for within two business days.
This forum is intended for constructive dialogue. Comments that are commercial or promotional in nature, pertain to specific medical cases, are not relevant to the article for which they have been submitted, or are otherwise inappropriate will not be posted.
We require that commenters identify themselves with names and affiliations.