Results

EFFECTS OF SURGERY

Preoperative versus postoperative performance in patients with unruptured aneurysms

As a group, patients who underwent repair of unruptured aneurysms showed poorer postoperative performance, compared with preoperative performance, on the following tests: (1) word fluency (the COWT; t=−4.62; df=11, p<0.003); immediate verbal recall (RAVLT; t=−2.76; df=11; p<0.02); delayed verbal recall (RAVLT, delayed; t=−3.87; df=11; p<0.005); and “frontal lobe” function (the Stroop test;t=2.36; df=10; p<0.05 (the fewer degrees of freedom for this test reflect the fact that one person could not perform the Stroop test due to premorbid colour blindness)). These differences are shown in fig 1. The probability of identifying a significant (p<0.05) decline between preoperative and postoperative performance on more than three out of 16 tests by chance alone is <0.01. Of note, however, is that only 33% of patients showed a deterioration in any or all of these scores from the preoperative to the postoperative scores. Only one patient deteriorated on all three tests. There was no difference between preoperative and postoperative performance on tests of attention (digits), confrontation naming (the Boston naming test), non-verbal memory (recognition memory test for faces, Rey-Osterreith complex figure test), visuoconstructional skills (the block design subtest of the WAIS-R) or psychomotor speed (grooved pegboard, symbol-digit subtest of the WAIS-R). However, when the data from the same subset of patients who were studied both preoperatively and postoperatively were analysed in the more traditional method for such studies, patients who underwent clipping of unruptured aneurysms performed at least 2 SD below the published mean for normal controls, postoperatively, on tests of psychomotor speed (symbol digit subtest of the WAIS-R; t=13.87; df=11; p <0.01), visuo-constructional skills (the Block Design subtest of the WAIS-R; df=11;t=−5.89; p<0.0001), as well as “frontal lobe” function (the Stroop test; t=−3.04, df=10; p<0.02). The probability of this finding (more than two of the 16 tests being significantly below normal) by chance alone is 0.04—that is, there was a marginally significant impairment on the set of tests compared with normal controls. Moreover, as there was not a significant decrease in scores from their preoperative performance on the WAIS-R subtests, it is unlikely that the significantly “impaired” scores were a result of the aneurysm surgery or associated factors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Change from preoperative to postoperative performance (in Z scores) in patients with unruptured aneurysms.

Postoperative performance in patients with unruptured aneurysms compared with published norms

For the entire group of patients with unruptured aneurysms (including those who were not tested preoperatively), mean performance was significantly below published norms on: word fluency (the COWT; t=−3.08, df=19, p<0.01); delayed verbal recall (RAVLT; t=−2.23; df=19; p<0.05); “frontal lobe” function (the Stroop test; t=−3.07; df=17; p<0.01); psychomotor speed (WAIS digit-symbol; t=−4.47, df=18; p<0.0001); and block design (t=−6.11; df=17; p<0.0001). The probability of this result (more than four of the 16 tests being significantly lower than normal) by chance alone is <0.001. Mean performance of this group was significantly above the published norms on theWechsler memory scale-revised verbal summary score (t=2.81; df=16; p<0.03; table 2). However, the finding of at least one of 16 tests being significantly higher than normal by chance alone is 0.56; in other words, this finding is likely due to chance.

EFFECTS OF SAH PLUS SURGERY

Postoperative performance in patients with ruptured aneurysms compared with published norms

As a group, patients who underwent clipping of ruptured aneurysms showed postoperative performance significantly below the mean for normal controls on nearly all tests: verbal fluency (COWT; t=−7.78; df=18; p<0.0001); delayed verbal recall (RAVLT; t=−3.38; df=20; p<0.002); non-verbal recognition memory (recognition memory test for faces; t=−2.96; df=17, p<0.01); psychomotor speed (grooved pegboard;t=−2.53; df=21; p<0.03); visuoconstructional skills (block design subtest of the WAIS; t=−4.23; df =13; p<0.001); “frontal lobe function” (the Stroop test; t=−2.80; df=20; p<0.02); attention (digit span; t=−5.03; df=16; p<0.0001); and confrontation naming (Boston naming test;t=−3.42; df=24; p<0.01). The probability of this many of the tests being significantly below normal by chance alone is essentially nil (p<0.000001). Mean scores and SD are shown in table 2. Unfortunately, however, it is not possible to compare this performance with their premorbid performance, as these patients presented with SAH. It is also remarkable that only a few patients with severely impaired postoperative performance accounted for the group differences on several tests (making the ttest an invalid measure of group differences). To illustrate, fig 2shows that on the Warrington recognition memory test for faces, nearly half (44%) of the patients scored at the 10th percentile or below, and another 44% of the patients scored at the 60th percentile or greater. More than half of the patients who scored below the 10th percentile had required shunts (intraventricular catheter or lumbar drain) for hydrocephalus, and one of the patients had an intraparenchymal haemorrhage.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Number of cases in each centile range on the recognition memory test for words in the ruptured aneurysm group. Percentile scores are based on published norms for the test.

EFFECTS OF SAH

Performance of patients with ruptured aneurysms versus unruptured aneurysms 3 months after aneurysm clipping

Performance of the patients who underwent clipping of ruptured aneurysms was significantly below that of patients who underwent clipping of unruptured aneurysms on a subset of the memory tests, but no other tests. There were significant group differences on the Wechsler memory scale-visual memory summary score (F=12.8; df=1, 23; p<0.003) and verbal memory summary score (52.0 v77.6; F=17.4; df=1, 23; p<0.001), and the Warrington recognition memory test for words (40.2v 45.1; F=4.46; df=1, 36; p<0.05), as shown in table 2. The probability of group differences on more than two of the 16 tests by chance alone is 0.04; thus, the documented differences between groups was marginally significant (p<0.05). On most tests, patients with ruptured aneurysm repair performed no worse postoperatively than did the patients after unruptured aneurysm repair. As most patients with unruptured aneurysm repair performed no worse postoperatively than preoperatively, it is likely that neither surgery nor SAH leads to diffuse deterioration in cognitive skills at 3 months, in the absence of complications such as hydrocephalus. Nevertheless, these results indicate that SAH itself may be associated with some circumscribed memory deficits.

There was also no significant difference between the unruptured and the ruptured aneurysm groups on the Zung depression inventory (F=0.44; df=1,40; NS). Altogether, 36% of patients rated themselves as not depressed, 36% as minimally to moderately depressed, 23% as moderately to markedly depressed, and 5% as severely depressed.

THE EFFECTS OF RECOVERY OVER TIME

Three month versus 12 month postoperative performance in patients with ruptured aneurysm

Regrettably, only a subset of six patients with ruptured aneurysms returned for long term follow up testing (mean 361 days postoperation). These patients, as a group, showed significant improvement only in confrontation naming (the Boston naming test;t=3.05; df=5; p<0.03), a finding that could be due to chance alone in making multiple comparisons (16 tests at two time periods). There may have been inadequate power to detect additional group improvements, particularly as most patients performed in the normal range at 3 months. That is, there was a ceiling effect for most patients. Also, it is not clear that this subset is representative of the entire ruptured aneurysm group. Nevertheless, this significantly improved test performance is consistent with the well known finding that patients with residual cognitive deficits at 3 months (who were few in this study) often show further recovery over the first year.

Discussion

There are several important results of this investigation that have not been previously reported. Firstly, although the patients with ruptured aneurysm as a group had significant impairments compared with published norms on nearly all of the neuropsychological tests administered (as reported in previous studies), the impairments were found in a minority of individual patients. That is, we found a bimodal distribution of cognitive function in patients at 3 month follow up after aneurysm clipping. For example, on the Warrington recognition test for words, a large percentage (44%) of patients scored below the 10th percentile, and an equally large percentage (44%) scored above the 60th percentile for normal controls. Although the median percentile score was 10 and the mode was <5, almost half of the patients scored above average. Furthermore, the mean percentile score for this group was 45.5%, but none of the patients scored between the 20th and 60th percentile, indicating that the mean score was virtually meaningless. Although few authors have brought attention to this aspect of results, inspection of individual data reported in several investigations indicates similar findings. Ogden et al 2 reported (their table 3) that 10 week follow up testing of patients with SAH disclosed “severe” impairments on the recognition memory test for words in 10.4% of patients and no impairment in 83.1%; only 6.5% of patients scored in the moderate range of impairment. Furthermore, in the study by Barbarottoet al 13, their table 2 shows that on the verbal fluency test with phonological cue (similar to our word fluency test, the COWT, and their most sensitive measure), 30% of patients after left MCA aneurysm repair scored below the lowest 5% of normal controls, but 40% of patients obtained an average score or higher than normal controls. Similarly, the tables in the study by Romner et al 10 indicate that 30% of patients had weighted impairment indices (based on Z scores) above 20, indicating widespread, marked deviations from normal performance, but 60% had impairment indices of <10, indicating a range in performance across tests from normal to moderate impairment. Only 10% had an impairment index in the middle range from 10–20. These results are comparable with those of Stenhouseet al,6 who reported that 23% of their patients 12 to 84 months after aneurysm repair had “pervasive global impairment”, with inability to perform (or very poor scores on) the Wisconsin card sorting test, and below average performance on the Wechsler memory scale, whereas 42% of patients showed no deficits on follow up neuropsychological testing. The other 30% of patients had only focal deficits on neuropsychological testing. Other authors have reported only group means4 7 or the percentage of patients who were impaired versus unimpaired on each test,5 which may have obscured a bimodal distribution of performance. Our data indicate that, even when the group mean performance is statistically significant from normal, only a minority of patients in the group have what is likely to be a clinically significant impairment.

Secondly, although many of the neuropsychological tests showed differences between groups (ruptured and unruptured) relative to published norms, only a few tests documented a change from preoperative testing even in the subset of patients with unruptured aneurysm with significant postoperative impairments. This result indicates that previous studies comparing mean performance in patients to published norms may overestimate the effects of SAH or aneurysm repair. Preoperative data from the patients with unruptured aneurysm represent their premorbid, baseline performance. Preoperative to postoperative decline in performance occurred on a few tests, by a subset of patients. Patients in the unruptured aneurysm group showed significant declines from preoperative performance on tests of word fluency (COWT—naming words beginning with F, A, and S), verbal recall (the Rey auditory-verbal learning test), and so called “frontal lobe functions” (the Stroop test). Notably, these are the very cognitive functions that are most often found to be impaired in follow up after aneurysm repair after SAH. This finding indicates that a large proportion of the long term cognitive sequelae reported in earlier studies — impairments in verbal fluency, immediate and delayed verbal and non-verbal recall, and “frontal lobe” function — may have resulted from effects of brain surgery, such as prolonged anaesthesia, temporary regional cessation of blood flow, brain retraction, use of antiepileptic medications, or even more general effects of bed rest and debility, rather than SAH alone. Other cognitive deficits, such as postoperative psychomotor slowing and visual memory impairments were only observed postoperatively in patients with ruptured aneurysms in our study, indicating that some cognitive deficits can be attributed to the SAH itself. In addition, performance on verbal and non-verbal recall was significantly more impaired in the ruptured aneurysm group relative to the unruptured aneurysm group, indicating that SAH itself may exacerbate or cause impairment in recall beyond the impairment attributable to stay in hospital and surgical management, at least in some patients. Thus, testing of both patients with both ruptured aneurysms (with SAH) and patients with unruptured aneurysms, who underwent comparable surgical management and perioperative care, is a unique aspect of this investigation that allowed us to separate the neuropsychological deficits resulting specifically from SAH from those resulting from general effects of neurosurgical intervention, perioperative complications, general anaesthesia, depression, or debility.

It has been proposed that the mechanism by which SAH negatively affects memory and “frontal lobe function” is focally decreased perfusion and/or depressed metabolism of brain regions critical to these skills, secondary to focal vasospasm or the clipping itself.4 6In this case, the site of aneurysm should be crucial to the effects on subsequent cognitive functions. However, in this study, site alone was not clearly a predictor of significant impairment. Of the seven patients who showed the most marked deficits in memory and “frontal lobe function” (with percentile scores <5 on the recognition memory test; see fig 1) after ruptured aneurysm, the distribution of aneurysm sites was as follows: two right posterior communicating artery; two right MCA; one left MCA, one right ACA; and one basilar artery. There was an overrepresentation of MCA aneurysms in this subset of patients with poor cognitive outcome relative to the entire group of patients with ruptured aneurysms (43% of patients with poor outcome versus 13% of all patients, had MCA aneurysms;χ ² =3.4; df=1; p=0.06; Fisher's exact: p=0.10), but this difference did not reach significance perhaps due to the few patients with poor cognitive outcome. There was also no difference in the mean Hunt-Hess scale score on admission between those with marked cognitive impairment at follow up (mean 1.8 (SD 1.3)) versus those without such impairments (mean 2.0 (SD 1.3)). However, all of the patients with marked deficits in memory and frontal lobe function at 3 month follow up had either documented basal ganglia or frontal lobe ischaemic changes on postoperative CT (n=3) or had postoperative hydrocephalus requiring shunt (n=4). These complications were significantly more common in the subset of patients with severe cognitive impairments than in those without significant impairments (100% v20%; χ ²=11.8;df=1; p<0.001; Fisher's exact: p<0.001).

By contrast with the deleterious effects of some postoperative complications, clinical vasospasm did not clearly predict impaired performance in the ruptured aneurysm group at 3 month follow up. Most patients with ruptured aneurysms, with or without lasting cognitive deficits, had at least moderate vasospasm by transcranial Doppler (TCD) criteria (for example, velocity of 140–200 in M1 or M2 segments of the MCA, or C1 segment of the internal carotid). However, only one of the seven patients with the most severe cognitive impairments had documented clinically significant vasospasm (in this case, associated with contralateral hemiparesis). Nevertheless, the frequent occurrence of vasospasm defined by TCD in the ruptured aneurysm group, compared with the absence of vasospasm, clinically or by TCD, in the unruptured aneurysm group, may account for at least some of the difference between ruptured and unruptured aneurysm groups found on three memory tests. We will report details of the relation between the degree and duration of vasospasm by TCD and performance on various neuropsychological tests at 3 months in the ruptured aneurysm group in a separate paper.

Previous studies have reported conflicting results regarding the effects of postoperative vasospasm on cognitive function after SAH. Whereas Adams et al,32 Hori and Sazuki,33 Fortuny et al,34 and Stenhouse et al 6 have all reported that postoperative vasospasm was a significant predictor of residual cognitive impairment at variable periods of follow up, Ogden et al,2 Barbarotto et al,13 and Hütter and Gilsbach5 failed to find any association between vasospasm and cognitive function. The predictive value of postoperative lesions on imaging for lasting cognitive impairments has also been controversial. Ogdenet al 2 found that ischaemic changes on neuroimaging postoperatively were predictive of more significant cognitive impairments at 1 year; and Hårdmarkeet al 9 found that ventriculomegaly on CT was related to deficits on neuropsychological testing at 1 year. By contrast, Romner et al 10 reported no correlation between tissue loss on late postoperative MRI and cognitive deficits at long term follow up (13–32 months after surgery). Contrasting results may reflect differences in the definitions or methods for identifying post operative vasospasm and ischaemic changes. For example, early MRI may show reversible lesions due to retraction injury or vasospasm that will not be seen on later MRI.35 The recent availability of technology to measure vasospasm, blood flow, and other aspects of regional metabolism, and new MRI tecniques to distinguish cerebral hypoperfusion from infarction, may be helpful in evaluating the aetiology of neuropsychological impairments after surgery in future studies.

The few previous studies that have evaluated cognitive changes after surgery in patients with unruptured aneurysms used very limited measures of cognition. The International Study of Unruptured Intracranial Aneurysms36 reported that 5.5% of patients with unruptured aneurysms and no history of SAH, and 9.6% of patients with unruptured aneurysms but history of SAH from a different aneurysm, had cognitive impairment (defined as<24 on the mini mental state examination37 or a score of less than 27 on the telephone interview for cognitive status38) at 3 months after intervention (>80% surgical; the remainder endovascular). Similar results were found at 1 year. No preintervention data are available to determine whether or not the cognitive impairment was due to the intervention (or perioperative management), or reflected premorbid variability among the study population. It seems likely from these results that the difference between groups reflected cognitive impairments associated with previous SAH. This study has been widely and severely criticised along methodological lines, and with respect to the interpretation of the findings39 and will not be further discussed here. In a separate study, Fukunagaet al 40 administered the mini mental state examination, in addition to a letter search test and a maze test, to 30 patients before and after surgery for unruptured aneurysms. They reported that 10 patients showed a deterioration in score on the mini mental state examination from the preoperative test to the 1 month postoperative test, but failed to make note of the result (data in their table 3) that a larger number of patients (12) actually showed an increase in the mini mental state examination score at 1 month after surgery. No statistical comparisons between preoperative and postoperative scores were reported.

In the present study, we documented deficits in verbal fluency, recall, and frontal lobe functions in a subset of patients after unruptured aneurysm repair. Decline from the preoperative period to the postoperative 3 month follow up period was largely limited to four patients, who had either large left MCA (n=2) or large ophthalmic artery aneurysms (n=2). In this group the site of aneurysm may well have been the crucial variable associated with the documented deficits, although the numbers are too small to detect a significant relation. That is, it is plausible that impairments in verbal fluency, verbal recall, and the Stroop test (also a verbal test) in the two patients with left MCA aneurysms were due to focal retraction injury involving the language cortex, and that impairments on the same tests (which also depend on frontal lobe function) in the two patients with ophthalmic artery aneurysms was due to focal retraction injury involving the frontal lobes. Notably, one of the patients with MCA had small intraparenchymal haemorrhages in the deep frontal and temporal regions on CT, consistent with localised contusion. By contrast with our findings of marked impairments only in patients with MCA or ophthalmic aneurysms, Gade41 suggested that patients who undergo repair of ACA aneurysms are particularly vulnerable to amnesia. However, no data from cognitive testing were reported in that study.

One potential application of the findings of the present study is in evaluating modifications of surgical or perioperative management of patients with intracerebral aneurysms. As repair of unruptured aneurysm now carries a very low risk of mortality (1%) or major physical morbidity (4%),4 mortality and morbidity rates are not sufficiently sensitive measures of outcome to identify possible differences in outcome due to variations in operative techniques or perioperative care. By contrast, we have documented that cognitive dysfunction after (even unruptured) aneurysm repair is more common. The frequency or severity of cognitive deficits may be associated with particular aspects of surgical management, and might be reduced by changes in protocol to provide neural protection (for example, hypothermia) or aimed at avoiding hypoperfusion, hypoxia, or damage to an eloquent brain region (for example, using temporary clipping). Thus, evaluation of cognitive function is likely to provide a sensitive measure of outcome, with which variations in operative or perioperative procedures can be evaluated in future investigations.

In summary, this investigation documented performance on neuropsychological tests by patients undergoing clipping of intracerebral aneurysms. The crucial aspects of the study that distinguish it from previous studies of this type are: (1) the inclusion of patients with unruptured aneurysms as well as those with ruptured aneurysms; and (2) the analysis of preoperative versus postoperative performance on the neuropsychological battery in patients with unruptured aneurysms. Comparison of patients with ruptured and unruptured aneurysms allowed us to differentiate deficits due to SAH (and possibly associated complications such as vasospasm) versus deficits due to more general effects of neurosurgical intervention, anaesthesia, and debility. Measuring preoperative performance permitted us to determine which differences from published norms in this group were due to operative and perioperative management, and which differences were due to premorbid variability. Although our study was small by comparison with some previous studies,42 the numbers were adequate to identify significant differences between patients who underwent repair of ruptured versus unruptured aneurysms, and significant differences between preoperative and postoperative performance. Finally, analysis of individual data provided reason to doubt the prevailing view that moderate or “marked” cognitive impairments after cerebral aneurysm repair are common. Rather, the moderate differences in group means, on a subset of tests in this study, reflected uncommon severe deficits. That is, very deviant scores by a few patients substantially lowered the group means. Unfortunately, the design of the study and the relatively few patients did not allow us to determine the cause of the marked cognitive impairments experienced by a few patients. Future studies (in progress) not only aim to recruit a greater number of subjects, but also to prospectively evaluate independent variables that might influence cognitive outcome, using serial TCD, magnetic resonance perfusion and diffusing imaging, CT perfusion scans, and other advanced imaging techniques.