Purpose We retrospectively evaluated the effect of brain atrophy on the outcome of patients after a large cerebral infarct.
Methods Between June 2003 and Oct 2008, 134 of 2975 patients with stroke were diagnosed as having a large cerebral infarct. The mean age of the patients was 70 (21–95) y. The mean infarct volume was 223.6±95.2 cm3 (46.0–491.0). The inter-caudate distance (ICD) was calculated as an indicator of brain atrophy by measuring the hemi-ICD of the intact side and then multiplying by two to account for brain swelling at the infarct site. The mean ICD was 18.0±4.8 mm (9.6–37.6).
Results Forty-nine (36.6%) patients experienced a malignant clinical outcome (MCO) during management in the hospital. Thirty-one (23.1%) patients had a favourable functional outcome (FO) (modified Rankin scale (mRS) ≤3) and 49 (36.6%) had an acceptable functional outcome (AO) (mRS≤4) at 6 months after stroke onset. In the multivariate analysis, brain atrophy (ICD≥20 mm) had a significant and independent protective effect on MCO (p=0.003; OR=0.137; 95% CI 0.037 to 0.503). With respect to FO, the age and infarct volume reached statistical significance (p<0.001, OR=0.844, 95% CI 0.781 to 0.913; p=0.006, OR=0.987, 95% CI 0.977 to 0.996, respectively). Brain atrophy (ICD≥20 mm) was negatively associated only with AO (p=0.022; OR=0.164; 95% CI 0.035 to 0.767).
Conclusions Brain atrophy may have an association with clinical outcome after a large stroke by a trend of saving patients from an MCO but also by interfering with their functional recovery.
- Brain atrophy
- cerebral infarction
- fatal outcome
- functional outcome
- image analysis
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Brain atrophy
- cerebral infarction
- fatal outcome
- functional outcome
- image analysis
Even after a pooled analysis of three recent randomised controlled trials strongly favoured offering early decompressive surgery to young patients with middle cerebral artery (MCA) infarction,1 debates continue over several issues such as the optimal timing of hemicraniectomy and the benefits of surgery in patients 60 y of age or older. In particular, the outcome after decompressive surgery for elderly patients with a large cerebral infarction has been disappointing, with 80% of patients ≥50 y of age dead or severely disabled.2 Potential reasons for this include an effect of age on the ability of the brain to compensate for a cerebral infarction2 and/or that older patients experience more brain atrophy causing them to be more vulnerable to cognitive dysfunctions such as neglect after right hemisphere infarction.3
Paradoxically, brain atrophy may protect older patients from a fatal outcome after a large stroke by providing an intracranial space to compensate for the increase in volume.4 5 This space caused by cerebral atrophy may have a role in preventing a regional increase in intracranial pressure (ICP) and further infarction, cerebral herniation and compression of the nearby brain structures, and, thus, may save patients with a large stroke from death. Conversely, a lack of cerebral atrophy in younger patients may not allow them to tolerate massive oedema as well as older patients, which consideration has been used to justify early decompressive surgery.6 7
Since the effect of brain atrophy on outcome after a large cerebral infarction has not been well investigated,2 we performed this retrospective study.
Materials and methods
Between June 2003 and October 2008, 2975 consecutive patients with stroke were managed in our hospital. Among them, 178 (5.98%) had a large cerebral infarction defined as an infarct involving one or more MCA division territories and/or the anterior cerebral artery or posterior cerebral artery territory. A total of 134 patients were included in this study after excluding patients who obtained a successful recanalisation of the occluded vessel after thrombolysis (n=18), who experienced intracerebral haemorrhage or subarachnoid haemorrhage after angioplasty or thrombolysis (n=11), who had another vascular disease such as moyamoya disease (n=2), who had reduced life expectancy due to a severe medical comorbidity such as cancer or a previous stroke (n=9) and whose relatives refused any treatments (n=4).
All patient data were based on information contained in hospital charts and radiological studies and were collected in accordance with the case record form approved by the institutional review board. Clinical data such as age, neurological status using the Glasgow Coma scale (GCS) and/or the National Institute of Health Stroke Scale (NIHSS), radiological characteristics and treatment modality were collected. Any data that were missing from the medical records due to follow-up loss were obtained through a telephone interview with the patient or with his or her relatives.
Management of the patients
Based on the symptom onset time and the results of brain imaging, including a CT angiogram, MRI composed of perfusion and diffusion-weighted imaging (DWI) and a cerebral angiogram, we decided which treatments would be performed according to the stroke guideline.8
Upon signs of neurological deterioration, brain stem herniation and/or major midline shift in follow-up brain imaging despite maximum medical treatment, decompressive surgery was performed in patients whose relatives agreed with surgery after a detailed discussion regarding the surgery's chances of improving survival and the expected long-term neurological deficits. Invasive monitoring of the ICP was not part of the routine management of these patients.
All patients were assessed upon admission using the GCS (n=125) and/or the NIHSS (n=83). A NIHSS ≤20 for the left or ≤15 for the right hemisphere was regarded as favourable. A malignant clinical outcome during admission to hospital was defined as the occurrence of one of the following as previously described9: (1) a loss of brain stem reflexes and death within the first 7 days after symptom onset; (2) transient or persistent clinical signs of cerebral herniation (anisocoria) within the first 7 days after symptom onset plus basal intracranial mass effect on the corresponding CT scan; or (3) a decision to perform decompressive surgery because of space-occupying swelling of the infarction with imminent herniation as judged by the neurologist and neurosurgeon responsible.
Patients alive at discharge were usually followed-up at 3 and 6 months after stroke and then every 3–6 months for clinical evaluation. The functional clinical outcome was assessed with the modified Rankin scale (mRS).10 An mRS ≤4 at 6 months after stroke onset was regarded as an acceptable functional outcome and an mRS ≤3 was considered a favourable functional outcome.
Brain atrophy and radiological characteristics
We used the inter-caudate distance (ICD) in MR images as a tool for measuring brain atrophy, as in the previous study.11 MR images were performed at the median time of 4 h from the first abnormal time (range 0.5–49) and at the median time of 7 h from the last normal time (range 0.5–240). The ICD was defined as the minimum distance between the medial borders of the head of the caudate nuclei at the level of the foramen of monro in an axial T1-weighted MR slice. However, to control for swelling of the involved cerebral hemisphere, in this study the hemi-ICD was first measured and then multiplied by two to estimate the ICD. The hemi-ICD was defined as the shortest length between the medial border of the caudate nucleus on the opposite side of the infarction and the midline, which was usually the septum pellucidum (figure 1A).
The infarct volume was measured from MR DWI scans (b value=1000 s/mm2) as previously described (figures 1B).12 The index scan (the scan that was rated) was the first MR scan performed. The site of the infarction and the involvement of the MCA territory and other vascular territories were defined based on the arterial territories of the human brain.13 A single rater (JHH) blind to the clinical outcome data reviewed all of the MR images entered into the study. The MR imaging was obtained with 1.5 T scanners (Gyroscan Intera 1.5T; Philips Medical Systems, Washington, DC, USA) and included T1-weighted sequences (repetition time (TR)/echo time (TE) = 450–550 msec/11msec) and DWI (TR/TE = 3000–5000 msec/60–90 msec). The slice thickness was 5 mm and the matrix size was 512×512 pixels for the T1-weighted sequences and 256×256 pixels for the DWI.
Continuous variables were reported as means ±SD. Categorical variables were recorded using numbers and percentages. A Student t test was performed to compare the means between the two groups. A Pearson's correlation analysis was performed to identify bivariate correlations between two continuous variables. A p value of 0.05 was considered significant. A malignant clinical outcome, an acceptable functional outcome and a favourable functional outcome at 6 months after stroke onset were the dependent variables in the analysis. Univariate and multivariate binary logistic regression analyses (level of significance α=0.05) were used to test associations with the independent variables and each of the dependent variables. To reduce the chance of type II error due to the modest sample size, variables were considered for multivariate analysis only if they were associated with a dependent variable in each analysis at the p<0.10 level. All statistical analyses were performed using SPSS v12.0.
Characteristics of the patients
In total, 66 (49.3%) of the patients were male and the mean age of the patients was 70±14 y (range, 21–95). The mean GCS score upon admission was 10.4±3.4 (range, 3–15) and the mean NIHSS score was 16.7±4.2 (range, 5–26). Altogether, 33 (24.6%) of the patients had a history of heart failure, 67 (50.8%) had atrial fibrillation and 76 (56.7%) had a history of hypertension. The most frequent area of infarction was the total MCA territory (63 patients, 47.0%) and 37 (27.6%) patients had infarctions involving one or more MCA divisions plus other vascular territories. The involvement of the basal ganglia was identified in 88 (65.7%) patients. A dominant hemisphere was involved in 53 (39.6%) patients. The mean infarct volume was 223.6±95.2 cm3 (range 46.0–491.0; median value=223.5 cm3). The mean ICD was 18.0±4.8 mm (range 9.6–37.6) and the median value was 17.2 mm. However, in four (3.0%) patients we could not measure the ICD because the contralateral frontal horn to the infarct site was already compromised due to severe midline shift and brain swelling and their mean infarct volume was 266 cm3. The clinical and radiological characteristics are summarised in table 1.
Overall clinical outcome
During admission to hospital, 49 (36.6%) patients experienced a malignant clinical outcome. Among them, only 22 (16.4%) patients underwent decompressive surgery, including 3 patients who finally died of brain swelling even after surgery and 1 patient who died of pneumonia after surgery. A total of 23 (17.2%) patients died due to malignant brain swelling after infarction during admission to hospital and they all were included in the category of a malignant clinical outcome (figure 2). Nine (6.7%) patients died of unrelated causes (including pneumonia in three, sudden cardiac arrest in three without any signs of herniation, gastrointestinal bleeding in one, intestinal perforation in one and pulmonary embolism in one) and, thus, were excluded when analysing the malignant clinical outcome and the functional outcome. At 6 months after stroke onset, 53 (39.6%) patients were still in a bed-ridden state with an mRS of 5 and a total of 49 (36.6%) patients had a functional outcome with an mRS of less than 4. The distribution of patients with different mRS is shown in detail in figure 2.
Malignant clinical outcome and prognostic factors
In the univariate analyses, the infarct volume, the involvement of other vascular territories in addition to the total MCA territory and the involvement of the basal ganglia all significantly correlated with malignant clinical outcome. However, brain atrophy (ICD≥20 mm) and the presence of atrial fibrillation correlated negatively with malignant clinical outcome.
Even in the multivariate analysis, brain atrophy (ICD≥20 mm) had a significant and independent protective effect on malignant clinical outcome (p=0.003; OR 0.137, 95% CI 0.037 to 0.503). The infarct volume, the involvement of other vascular territories in addition to the total MCA territory, the involvement of the basal ganglia and the presence of atrial fibrillation also had significant and independent relationships. The results of the statistical analyses are summarised in table 2.
Functional outcome and prognostic factors
With respect to favourable functional outcome (mRS<3), only age and infarct volume reached statistical significance showing a negative association in the multivariate analysis (p<0.001, OR 0.844, 95% CI 0.781 to 0.913; p=0.006, OR 0.987, 95% CI 0.977 to 0.996, respectively). The other variables, including brain atrophy and surgical decompression, did not show any effects on favourable functional outcome.
With respect to acceptable functional outcome (mRS<4), age and infarct volume again showed a significant negative correlation with outcome (p=0.001, OR 0.878, 95% CI 0.815 to 0.947; p=0.001, OR 0.982, 95% CI, 0.971 to 0.992, respectively) in the multivariate analysis. In addition, brain atrophy (ICD≥20 mm) and history of heart failure were significantly and negatively associated with acceptable functional outcome (p=0.022, OR 0.164, 95% CI 0.035 to 0.767; p=0.012, OR 0.119, 95% CI 0.022 to 0.629, respectively). Interestingly, hemicraniectomy, which had no effect on favourable functional outcome, had a significant and independent relationship with acceptable functional outcome (p=0.034; OR 7.271, 95% CI 1.164 to 45.40), though its statistical power seems to be low considering the wide range of the 95% CI. The results of the statistical analyses are summarised in table 2.
With age, many brain structures show volumetric shrinkage,14 15 and even neurologically asymptomatic elderly subjects experience continuing brain volume loss that appears to accelerate with age.16 Though the brain is a dynamic organ that seeks to maintain homeostatic cognitive function by engaging in continuous functional reorganisation and functional repairs (called ‘compensatory scaffolding’), the aged brain is less efficient at generating scaffolding and significant pathology may entirely limit scaffolding operations.15
Based on our results, brain atrophy actually seems to provide a chance of life to patients with a large cerebral infarct; however, it may not guarantee a good functional outcome. On the contrary, brain atrophy can interfere with a patient's ability to recover from neurological deficits caused by a large cerebral infarct.
Brain atrophy and its measurement
To accurately evaluate the role of brain atrophy in compensating for regional increases in ICP after a stroke, brain herniation and so on, it would make sense to investigate the ratio of the total intracranial volume to that of the whole brain. However, brain volume measures are almost never available to clinicians even during the management of acute stroke patients because they require sophisticated MRI acquisition techniques combined with detailed post-processing image analyses.11 Thus, one practical tool may be the application of linear markers such as the ICD (as in this study), the width of the third ventricle and/or the width of the lateral ventricle/frontal horn.
In addition, although brain shrinkage with age is widespread, one of the structures with the greatest mean shrinkage is the caudate.14 The ICD has previously been used as a valuable marker in normal ageing and various diseases.11 17–19 Thus, the authors of this study used the ICD as an indicator of brain atrophy and, considering the linear correlation between the ICD and age that was found (correlation coefficient=0.491 for 1 y of age; p<0.001; figure 3), this appeared to well-reflect the atrophy process in ageing brains.
Brain atrophy and clinical outcome
Many studies have searched for clinical and radiological factors predictive of fatal clinical outcome and/or fatal brain oedema after a large stroke and various factors have been found.2 9 20–23 One of the most important and concrete factors is the absolute infarct volume, and this was reflected in our results as well. In fact, our results indicate that the chances of a malignant clinical outcome increase by 1% for every 1 cm3 increase in infarct volume. Similarly, the involvement of other vascular territories in addition to the total MCA infarct territory seems to have as significant an effect on malignant clinical outcome as infarct volume. One of the most notable findings of this study was that brain atrophy above a certain threshold (ICD≥20 mm) had an independent protective effect on malignant clinical outcome. Simultaneously, brain atrophy (like age) seems to be one of the most significant predictors of adverse functional outcome after a large cerebral infarct because it showed no effect on favourable functional outcome (mRS score≤3), and it exposed the patients to a substantial risk of survival with a severe disability by interfering with functional recovery even to an mRS score of 4. This may be an important finding considering that there are still debates on the benefits of early or even late hemicraniectomy for older patients with a large cerebral infarct because patients with brain atrophy comprise the majority of stroke patients and they could avoid surgery merely for a room to compensate for regional increases in ICP and to increase cortex microcirculation near the infarct site.24 Additionally hemicraniectomy, which is usually performed after neurological deterioration in our institute, seems to not guarantee a favourable functional outcome, though it had a positive role in terms of acceptable functional outcome (mRS score≤4). Therefore, hemicraniectomy should be performed cautiously for patients with brain atrophy because it may not provide additional room for compensation for a large cerebral infarct and it may not offer functional independency either.
The present study had several limitations. First, as this study is retrospective, the population may have heterogeneity in their presentation, which has to be considered in the interpretation of the results. Second, the sample size of this study is small and the patients enrolled are the elderly with a mean age of 70 y, which may be a selection bias. Third, hemicraniectomy was performed in a late phase in our institute, which may have a negative effect on a functional outcome after surgery. Finally, we used only one linear marker for brain atrophy in this study instead of measurement of a brain volume. In addition, the independent effect of atrial fibrillation on malignant clinical outcome in our study may be caused by the limitations of our study because the ICD was significantly different between the group of patients with atrial fibrillation and those without (16.8±4.1 mm and 19.0±5.4 mm, respectively; p=0.012). Such a confounding effect of atrial fibrillation on brain atrophy should have been figured out in the multivariate analysis; however, the sample size of this study seems to be too small for us to distinguish such a dense confounding variable from a significant one. A large and more sophisticated study should be necessary to address the specific cut-off values of brain atrophy, age and infarct volume for decision on aggressive management.
In conclusion, brain atrophy seems to have an association with clinical outcome after a large stroke, as do age and absolute infarct volume. However, its ambivalent association between saving patients from a malignant clinical outcome and interfering with functional recovery should be considered when deciding between early decompressive surgery and medical care only, especially for older patients who usually have brain atrophy. Thus, future studies should be directed toward the use of specific quantitative indices of brain atrophy in selection of patients for acute stroke intervention.
This study was partially supported by a grant from the Korea Health 21 R&D Project, Ministry of Health, Welfare and Family Affairs, Republic of Korea (grant code: A06-0171-B51004-06N1-00040B).
Competing interests None.
Ethics approval This study was conducted with the approval of the Seoul National University Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.