Abstract
Background: Aneurysmal subarachnoid hemorrhage (SAH) and surgical clipping of intracranial aneurysms are associated with substantial morbidity and mortality.
Objective: To compare cognitive outcome and structural damage in patients with aneurysmal SAH treated with surgical clipping or endovascular coiling.
Methods: Forty case-matched pairs of patients with aneurysmal SAH treated by surgical clipping or endovascular coiling were prospectively assessed by use of a battery of cognitive tests. Twenty-three case-matched pairs underwent MRI 1 year after the procedure. Matching was based on grade of SAH on admission, location of aneurysm, age, and premorbid IQ.
Results: Both groups were impaired in all cognitive domains when compared with age-matched healthy control subjects. Comparison of cognitive outcome between the two groups indicated an overall trend toward a poorer cognitive outcome in the surgical group, which achieved significance in four tests. MRI showed focal encephalomalacia exclusively in the surgical group. This group also had a significantly higher incidence of single or multiple small infarcts within the vascular territory of the aneurysm, but both groups had similar incidence of large infarcts and global ischemic damage.
Conclusion: Endovascular treatment may cause less structural brain damage than surgery and have a more favorable cognitive outcome. However, cognitive outcome appears to be dictated primarily by the complications of SAH.
For subarachnoid hemorrhage (SAH) patients that reach a neurosurgical unit, the mortality can be as high as 26%.1 Neurosurgical clipping of the aneurysm has, until recently, been the preferred treatment option to prevent further bleeding. Most patients undergoing craniotomy and surgical clipping of the aneurysm are left with significant cognitive deficits.2 Neuropsychological assessment has shown that up to 62% of patients with Glasgow Outcome Scale scores of 1 are left with cognitive impairment.3 The most important factors implicated include grade on admission4 (according to the World Federation of Neurosurgical Surgeons grading scale5), aneurysm location,6 age,7 and operative and perioperative factors. Previous attempts to assess the contribution of surgery toward cognitive outcome were hindered by the lack of an appropriate control group. Endovascular embolization8 using detachable coils is now an alternative to surgical clipping and thus provides a unique opportunity to attempt to assess the relative contribution of surgery toward cognitive outcome as well as the presence of any advantages of one treatment over the other.
The assessment of structural brain damage resulting from SAH and subsequent surgery has never been adequately explored. This is partly because of concerns about the safety of MRI in patients with intracranial aneurysm clips. The new generation of nonferromagnetic clips has made possible the safe use of MRI in investigating structural damage that occurs after surgical clipping and a comparison with endovascular coiling.
Methods.
Patient selection.
All patients (n = 109) who underwent surgery after aneurysmal SAH during a 2-year period (January 1995 to December 1996) at the neurosurgical unit at the Royal Hallamshire Hospital, Sheffield, United Kingdom, were recruited in a prospective study investigating cognitive outcome and structural brain damage. Standard neurosurgical technique was employed. To reduce the risk of ischemia, a variety of techniques were used: mild hypothermia, ventriculostomy for brain relaxation, minimization of retractor usage, cisternal irrigation of clots, and papaverine installation into the cisterns. All 109 patients were considered for case matching to 40 patients treated during the same period at the Radcliffe Infirmary, Oxford, United Kingdom, by endovascular coil embolization. All of these 40 patients were eligible for either surgery or coiling. The final decision about method of treatment was made by the patient in the first 26 cases but was a result of randomization for the last 14 patients (these 14 patients were the first to be recruited for the International Subarachnoid Hemorrhage Trial, a multicenter trial looking at outcome after randomization to surgery or endovascular treatment). Forty of the 109 surgically treated patients were selected on the basis of best match to the 40 patients treated with endovascular embolization. Case matching was based on the following criteria: grade on admission to the neurosurgical unit according to the World Federation of Neurologic Surgeons (WFNS) grading scale,5 age and location of aneurysm. In addition, patients were graded by the extent of the SAH evident on CT scans.9 This grading is said to correlate with the incidence and severity of vasospasm. Case matching was done blindly by one of the authors (M.H.), based on the above three parameters and without knowledge of patient identity. Because the primary aim of the study was to look at cognitive outcome, patients who died before the 1-year assessment were excluded. There was, however, no significant difference in the mortality rates between the two groups.
MRI.
In all surgical patients, MRI-compatible clips were used. At 1 year after surgery, MRI scans were obtained by use of a 1.5 Tesla imager (Siemens Magnetom SP, Munich/Erlangen, Germany). The standard imaging protocol consisted of axial and coronal dual-echo fast spin echo PD and T2-weighted images as well as a T1-weighted 3D volume MPRAGE acquisition. MRI scans were available in all of the surgical group. Only 23 patients from the endovascular group underwent MRI a year after the procedure. These were performed on a 1.5 Tesla imager (Siemens Magnetom Vision). The MRI protocol consisted of axial and coronal dual echo fast spin echo PD and T2-weighted images as well as T1-weighted sagittal images. The MRI results are thus based on 23 case-matched pairs only. The MRI scans from both groups were analyzed by the same neuroradiologist (C.R.). Blinding was not possible because of characteristic MRI appearances produced by the two procedures. Surgical damage was defined as focal encephalomalacia related to the surgical approach, and areas of infarction were defined by their size, number (small single, small multiple and large major vessel), and anatomic location. These showed high signal on T2-weighted images and low signal on T1-weighted images. Small vessel white matter ischemic changes were defined as foci of high signal on T2-weighted images with no signal change on T1-weighted images.10,11⇓ These ischemic changes were scored subjectively and semi-quantitatively (none, few but normal for age, and excessive for age). Further detailed anatomic data regarding sites of damage were also obtained.
Neuropsychological assessments.
The neuropsychological assessments were performed 1 year after the procedure and consisted of tests on which performance has previously been shown to be impaired after SAH and surgery. These tests covered the following cognitive domains: premorbid IQ was estimated by using the National Adult Reading Test (NART).12 Current IQ was assessed using the following subtests of the Wechsler Adult Intelligence Scale–Revised (WAIS-R):13 vocabulary, digit span, similarities, arithmetic, picture arrangement, and block design. Affect was assessed by using the Beck Depression Inventory.14 Memory was examined by using specially constructed and standardized word and face recognition tests, immediate and 30-minute recall of a short story, immediate and 30-minute drawing of a complex figure, and the Corsi block span test from the Cambridge Automated Neuropsychological Test Battery (CANTAB).15,16⇓ Executive function was assessed by using letter fluency, semantic fluency, digit ordering, number of attempts to the extradimensional stage, and number of stages completed in the intradimensional/extradimensional shift (ID/ED shift) test (CANTAB), strategy count and between errors of the spatial working memory test (CANTAB), and the number of moves above minimum from the Tower of London test (CANTAB). Language was assessed by using one of the equivalent forms of the Boston Confrontation Naming test.17 For visual perception, a specially constructed and standardized test of form discrimination was used. The subject was asked to determine whether a pair of two-dimensional forms were the same or different. The mean latencies per correct and incorrect response were calculated. The Trails test18 was used as a test of attention. Normative data for the neuropsychological tests were collected from 31 volunteers responding to an advertisement. They were selected to match the patient groups on premorbid IQ, age, and sex. They underwent the same neuropsychological battery as the patient groups. The neuropsychological assessments of both patients and controls were performed by seven assessors at two sites (Sheffield and Oxford). Standardization was achieved by all assessors being trained by the same person (M.H.) and by attending as an observer in at least two patient assessment sessions.
Statistical analysis.
Analysis of variance (ANOVA) with post hoc testing using Fisher’s Protected Least Significant Difference (PLSD) was used. Comparisons were made between the endovascular and the surgical groups as well as between each group and the control group. The Mann-Whitney test was used for those cognitive tests in which the data were nonparametric (marked with * on table 1). The MRI data were analyzed by using the χ2 test.
Cognitive data
Results.
Patient characteristics.
Because of the matching process, there were no significant differences in age, grade on admission (there was one Grade 2 patient in the endovascular group matched to a Grade 1 patient in the surgical group) and location of aneurysm (one patient with left internal carotid aneurysm at the bifurcation from the endovascular group was matched to a patient with left posterior communicating aneurysm from the surgical group). The female-to-male ratio was 27 to 13 in the surgical group and 29 to 11 in the endovascular group. Table 2 also contains the distribution of the grading of SAH based on the CT scan on admission.9 No significant differences were noted between the two groups; data from four patients in the endovascular group were not available. There was a significant difference in the time to operation from the onset of the hemorrhage between the two groups. The mean time for the surgical group was 275 ± 248 hours, but for the coil group this was 113 ± 116 hours. Outcome as assessed with Glasgow Outcome Score (GOS) at 1 year did not differ apart from one more patient with GOS of three in the surgical group for one more patient with GOS of two in the endovascular group.
Patient characteristics
Cognitive data.
There was no difference in premorbid IQ as estimated by the NART between the surgical and endovascular groups (see table 1). The IQ for the control group was estimated by using the vocabulary subtests of WAIS-R. The difference between the estimated IQ of the control group and the estimated premorbid IQ of the two other groups (their IQ before the SAH) is not significant. Table 1 contains the mean cognitive scores and standard deviations for each test per group. p values are given for comparisons between the surgical and endovascular groups as well as comparisons between each group with controls. Both treatment groups were significantly impaired when compared with the control group in all cognitive domains. The surgical group scored significantly worse than controls in 20 of 25 tests as compared with 17 of 25 of the endovascular group. On direct comparison, the surgical group scored significantly worse than the endovascular group in four of the tests: Semantic fluency test (p = 0.05), vocabulary subtest of the WAIS-R (p = 0.02), complex figure recall test (p = 0.04) and the extradimensional stage of the ID/ED shift test from the CANTAB (p = 0.01).
Structural damage.
MRI results for the two treatment groups are based on 23 case-matched patients (table 3). Focal encephalomalacia was seen in 19 patients from the surgical group, but in none of the endovascular group (p < 0.001). This damage varied in size from large areas adjacent to the sylvian fissure to small areas of resected gyrus rectus seen exclusively in patients operated on for anterior communicating artery aneurysms. Areas of infarction were present in 20 patients (87%) from the surgical group, compared with 13 (57%) of the endovascular group (p < 0.05). All infarcts occurred within the vascular territories of the vessel(s) from which the aneurysm arose. Eight patients from the surgical group and nine patients from the endovascular group had large vessel infarcts. In the remaining patients, the infarcts were small, single, or multiple and occurred in isolation or in combination with large infarcts. Small vessel ischemic changes thought to be excessive for the age of the patient were present in 11 of the clipped and 14 of the coiled groups, but the difference was not significant.
MRI results based on 23 case-matched pairs
Discussion.
In this study, we compared cognitive outcome as well as structural damage after aneurysmal SAH and treatment by two different procedures. Since its introduction by Guglielmi et al. in 1991,8 coil embolization in selected patients has gained in popularity because of the avoidance of craniotomy. Although data on long-term safety and the risk of recanalization after endovascular coiling are currently being collected,19 the potential advantages of this procedure over surgery in terms of cognitive outcome and associated brain damage are only just beginning to be evaluated. Several studies have shown that GOS is a very crude outcome measure and that patients with so-called GOS of 1 have significant cognitive deficits.3 Detailed neuropsychological testing has proved more sensitive in teasing out more subtle but nonetheless important deficits that contribute to outcome. This study illustrates that despite the good GOS outcome (Grades 1 and 2) in 36 of 40 in the surgical group and 37 of 40 of the endovascular group, patients in both groups were significantly impaired in all cognitive domains when compared with controls. Although this study did not involve randomization of patients, close case-matching has minimized any potential differences between the two groups in prognostic factors related directly to the hemorrhage rather than the treatment. In addition, grading the extent of hemorrhage on CT provides a useful indirect measure of the likelihood of the development of vasospasm.9 There was no significant difference in the CT grading between the two groups. The only significant difference between the two groups was the time from hemorrhage to intervention, endovascular coiling being performed sooner than surgery. Because the incidence of rebleeding is highest immediately after hemorrhage, earlier intervention may be an additional advantage of endovascular coiling over surgery. A study specifically looking at cognitive outcome after early versus late surgery found no differences between the two groups.6 Similarly, within the whole of the surgical group we have studied in Sheffield (total 109 patients), we found no differences in cognitive outcome between those operated on early and those operated on late. We therefore believe that this factor alone cannot account for the difference in cognitive outcome and structural damage between the two groups. The trend toward favorable cognitive outcome for the endovascular group was present throughout the majority of tests despite achieving statistical significance in only four tests. The use of Bonferroni correction in the context of multiple comparisons may be too conservative; three of the tests—semantic fluency (a test of word retrieval), complex figure recall (a test of memory), and the extradimensional stage of ID/ED test from the CANTAB (a test of executive function)—are primarily sensitive to frontotemporal damage and are therefore related. The surgical group would be expected to score worse in these tests because they have more frontotemporal damage, as illustrated by the MRI findings: structural damage, seen in the surgical group, affected temporal and frontal lobes in 83% of patients as compared with 52% in the endovascular group. In addition, all patients in the surgical group, unlike those in the endovascular group, underwent cognitive assessments at 6 weeks and 6 months after surgery before the final 1-year assessment. Though equivalent versions of the same cognitive tests were used, some practice effect is inevitable. Such an effect may have resulted in masking wider differences in cognitive outcome between the two groups. Another reason why this trend did not achieve significance in more than four tests may be that the sample size was too small or test sensitivity was insufficient to distinguish differences between two cognitively impaired groups. It is also possible that these four tests are more sensitive than the rest for this particular study population. Almost all tests are, however, sensitive enough to show significant impairment in all cognitive domains when comparing the two groups with the control subjects.
The prevalence of large vessel infarcts was 35% in the surgical group and 39% in the endovascular group. This difference is not significant. These infarcts may represent delayed ischemic neurologic deficits related to vasospasm. The surgical group, however, had a significantly larger number of patients with small single or small multiple infarcts within the vascular territory of the aneurysm. The cause of these is unclear, but surgery appears to play a role. One possibility is that clipping the aneurysm may also result in compromise of other adjacent minute perforators or that the direct exposure and handling of the artery may result in subsequent ischemic deficits.
Both groups, however, appeared to have similar prevalence of white matter hyperintensities excessive for age. This may explain the more global cognitive impairment seen in both groups and may reflect the impact of the hemorrhage itself. An alternative but less likely explanation is that these changes may simply reflect preexisting ischemic damage in a population with a high incidence of smoking and hypertension.
The global and significant cognitive impairment in both groups suggests that outcome after aneurysmal SAH is primarily dictated by the hemorrhage possibly with some contribution from preexisting vascular disease. Our results, however, provide preliminary evidence that surgical clipping may result in more frontotemporal brain damage as illustrated by MRI and that such damage may be accompanied by additional cognitive deficits seen in the surgical group when compared with the endovascular group. Future studies with larger number of patients and particular emphasis on frontotemporal tests may further clarify potential outcome differences between the two procedures.
Acknowledgments
Supported by a Welcome Foundation Grant.
Acknowledgment
The authors thank the ISAT investigators for allowing them to use 14 of their patients for this study.
- Received June 19, 2000.
- Accepted February 24, 2001.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
