CT-Visible Convexity Subarachnoid Hemorrhage is Associated With Cortical Superficial Siderosis and Predicts Recurrent ICH

Abstract

Objective

To investigate whether acute convexity subarachnoid hemorrhage (cSAH) detected on CT in lobar intracerebral hemorrhage (ICH) related to cerebral amyloid angiopathy (CAA) is associated with recurrent ICH.

Methods

We analyzed data from a prospective cohort of consecutive acute lobar ICH survivors fulfilling the Boston criteria for possible or probable CAA who had both brain CT and MRI at index ICH. Presence of cSAH was assessed on CT blinded to MRI data. Cortical superficial siderosis (cSS), cerebral microbleeds, and white matter hyperintensities were evaluated on MRI. Cox proportional hazard models were used to assess the association between cSAH and the risk of recurrent symptomatic ICH during follow-up.

Results

A total of 244 ICH survivors (76.4 ± 8.7 years; 54.5% female) were included. cSAH was observed on baseline CT in 99 patients (40.5%). Presence of cSAH was independently associated with cSS, hematoma volume, and preexisting dementia. During a median follow-up of 2.66 years, 49 patients (20.0%) had recurrent symptomatic ICH. Presence of cSAH was associated with recurrent ICH (hazard ratio 2.64; 95% confidence interval 1.46–4.79; p = 0.001), after adjusting for age, antiplatelet use, warfarin use, and history of previous ICH.

Conclusion

cSAH was detected on CT in 40.5% of patients with acute lobar ICH related to CAA and heralds an increased risk of recurrent ICH. This CT marker may be widely used to stratify the ICH risk in patients with CAA.

Classification of Evidence

This study provides Class II evidence that cSAH accurately predicts recurrent stroke in patients with CAA.

Cerebral amyloid angiopathy (CAA) is a small vessel disease that is commonly associated with lobar intracerebral hemorrhage (ICH) in the elderly.¹ It is characterized by progressive deposition of β-amyloid peptide in the small cortical and leptomeningeal vessels and contributes to cognitive impairment.^2,3 The clinical spectrum of CAA varies widely from transient focal neurologic episodes to dementia.^4–6 Symptomatic lobar ICH is the most devastating neurologic manifestation of CAA and is associated with high morbidity and mortality. Despite recent advances in risk factor control, the risk of recurrent ICH events remains high in patients with CAA-related ICH (CAA-ICH).^7–9

Convexity subarachnoid hemorrhage (cSAH) is a particular form of nontraumatic intracranial hemorrhage that occurs within one or a few cortical sulci at the convexity of the brain.¹⁰ cSAH may be related to a wide spectrum of neurologic disorders ranging from cerebral vein thrombosis to reversible cerebral vasoconstriction syndrome.^3,10,11 It is increasingly recognized as an imaging marker associated with CAA and has distinct pathophysiologic traits from aneurysmal or traumatic subarachnoid hemorrhage.^12,13

Recently, cortical superficial siderosis (cSS) has emerged as an important imaging marker for CAA according to the revised Boston criteria.^8,14,15 cSAH may be involved in cSS physiopathology and promote development of future cSS. However, they are different imaging markers of hemorrhage and our understanding of the pathophysiologic features of these 2 imaging markers remains limited.^1,7 In addition, the association between cSAH and other imaging markers of small vessel disease remains unclear. cSS is a useful MRI-based imaging marker that is associated with high risk of recurrent ICH.⁷ Because cSS is an imaging marker visible only on MRI, developing a novel prognostic marker that can be assessed in the acute setting with noncontrast CT would be desirable, particularly when MRI is not available, not tolerated, or contraindicated.

In this study, we aimed to assess the association between cSAH and small vessel disease markers and to investigate whether cSAH on CT is associated with increased risk of recurrent bleeding in acute lobar ICH with CAA.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the institutional review board of Massachusetts General Hospital. Informed consent was obtained from all participants or their legal representatives.

Patient Selection

We analyzed data from a prospective observational cohort study of consecutive survivors of lobar ICH admitted to the Massachusetts General Hospital between February 1997 and November 2012. Patients were included in the study if they fulfilled the following inclusion criteria: (1) acute symptomatic lobar ICH; (2) diagnosis of definite, probable, or possible CAA based on the modified Boston criteria¹⁵; (3) available brain noncontrast CT and MRI including T2*-weighted gradient-recalled echo (T2*GRE) or susceptibility-weighted imaging (SWI) sequences performed within 10 days after symptom onset; (4) still alive at discharge; and (5) availability of follow-up data. Patients were excluded from the study if they had craniotomy before MRI or baseline CT scan. Patients with traumatic ICH were also excluded.

Data Collection

The baseline clinical data including demographic, medical history, medication, and vascular risk factors were prospectively collected at the time of event using standardized data collection methods as previously described.¹⁶ APOE genotype was determined in a subgroup of patients who consented to donate blood samples for genetic analysis.⁷

Imaging Acquisition and Analysis

All CT scans of the indexed event were independently reviewed by 2 readers (Q.L. and M.Z.) who were blinded to the MRI results. The volumes of parenchymal hematoma were calculated with semiautomated planimetric methods (Alice, PAREXEL International Corporation, Waltham, MA; and Analyze 10.0, Mayo Clinic, Rochester, MN). The presence of acute cSAH was visually assessed.¹⁷ cSAH was defined as linear hyperintense signal in the subarachnoid space that could be adjacent or remote from the ICH. cSAH was classified as adjacent when the bleeding was strictly confined to sulci within 1 or 2 sulci from acute ICH or as remote when cSAH was observed away from >2 unaffected sulci of acute ICH. The interrater agreement (Q.L. and M.Z.) for the presence of cSAH was excellent (Cohen κ = 0.772). Discrepancies were settled by consensus reading after independent review of cases.

Brain MRI was acquired on a 1.5T scanner and included at least fluid-attenuated inversion recovery (FLAIR) and blood-sensitive sequence (T2*-weighted GRE or SWI). All MRIs were assessed by the investigators according to the Standards For Reporting Vascular Changes on Neuroimaging (STRIVE).¹⁷ Severity of cSS was classified as focal (≤3 sulci) or disseminated (≥4 sulci) as previously described.⁸ Cerebral microbleeds (CMBs) were defined as small areas of signal void on T2*-weighted MRI or SWI.^17,18 White matter hyperintensities (WMH) including periventricular and deep WMH were visually assessed on the axial FLAIR images using the Fazekas rating scale.¹⁹

Follow-Up

The clinical data including recurrent lobar ICH and death were collected during follow-up from consenting survivors and their caregivers by telephone interview after ICH, as previously described.¹⁶ All recurrent ICH events were confirmed by brain imaging studies and medical records.

Statistical Analysis

Statistical analyses were performed using SPSS (version 25.0). The baseline demographic, clinical, genetic, and imaging characteristics were compared between patients with and without cSAH, using χ² test, Fisher exact test, Student t test, or Mann-Whitney U test, as appropriate. We used multivariable logistic regression analysis to investigate factors associated with cSAH on CT. Multicollinearity was measured by variance inflation factors (VIFs) and tolerance. Predictors with VIF >5 were removed from the model. We determined the presence of cSAH as a predictor of recurrent ICH using Kaplan-Meier plots together with log-rank test. Data were censored after the first recurrent event for patients with multiple recurrent ICH or at the end of follow-up, whichever came first. Multivariable Cox proportional hazards models were used to calculate the hazard ratios (HRs) for recurrent ICH. The proportional hazards assumption of the Cox models were tested by graphical check and Schoenfeld residual tests. We have included prespecified plausible predictors of recurrent ICH as well as factors with p < 0.1 in univariable Cox regression analysis. All p values presented are 2 sided, with a p value of 0.05 or less considered statistically significant.

Data Availability

The authors have documented all relevant data used to conduct the research. Anonymized data will be shared by request from qualified investigators.

Results

Of 443 lobar ICH survivors, 100 were evaluated with CT only, 18 with MRI only, and 325 with head CT followed by brain MRI. Among them, 244 patients (mean age 76.4 ± 8.7 years; 54.5% female) were eligible and were included in the final analysis. A flow diagram of patient selection is illustrated in figure 1. Compared to patients with lobar ICH evaluated with CT only, included patients with were younger (76.4 vs 78.8 years, p = 0.017) and had smaller baseline hematoma (24.1 mL vs 32.5 mL, p = 0.001). The median time from symptom onset to baseline CT scan was 1 day (interquartile range 0–2). Our final cohort consisted of 10 (4.0%) patients with pathologically proven CAA, 135 (55.3%) with probable CAA, and 99 (40.5%) with possible CAA.

Figure 1. Flow Diagram of Patient Selection.

ICH = intracerebral hemorrhage.

In the whole cohort, cSAH was observed in 99 patients (40.5%), cSS in 50 patients (20.4%), and lobar CMBs in 135 (55.3%) patients. Of 99 patients with cSAH, 70 had strictly adjacent cSAH and 29 had remote cSAH. Compared to those without cSAH, patients with cSAH had larger baseline ICH volumes (p < 0.001), and were more likely to have prior (before index ICH) symptomatic ICH (p = 0.003), preexisting dementia (p = 0.009), and cSS (p = 0.002) (table 1). In univariable analysis, presence of cSS (odds ratio [OR] 2.72, 95% confidence interval [CI] 1.44–5.14; p = 0.002), previous history of ICH (OR 3.80, 95% CI 1.50–9.62; p = 0.005), APOE ɛ2 (OR 2.78, 95% CI 1.12–6.91; p = 0.028), baseline hematoma volume (OR 1.03, 95% CI 1.02–1.05; p < 0.0001), and preexisting dementia (OR 2.36, 95% CI 1.22–4.56; p = 0.01) were associated with cSAH on CT. After adjusting for age, history of prior ICH, presence of lobar CMBs, cSS, APOE ɛ2, and preexisting dementia remained independently associated with cSAH in multivariable logistic regression analysis (table 2).

Table 1.

Baseline Demographic, Clinical, and Imaging Characteristics of Patients With vs Without Convexity Subarachnoid Hemorrhage (cSAH)

Table 2.

Univariable and Multivariable Regression Analyses of Factors Associated With Convexity Subarachnoid Hemorrhage in Cerebral Amyloid Angiopathy–Related Intracerebral Hemorrhage (ICH)

During a median follow-up of 2.66 years (interquartile range 0.89–5.20 years), 49 of 244 patients (20.0%) had recurrent ICH. Patients who had recurrent ICH were more likely to have prior history of ICH (20.4% vs 6.7%, p = 0.003), significantly higher prevalence of cSS (32.7% vs 17.4%; p = 0.018), and cSAH (63.3% vs 34.9%; p < 0.0001) as compared with those without recurrent ICH during follow-up. The age, vascular risk factors, and presence of lobar CMBs and WMH severity was similar between the 2 patient groups.

In Kaplan-Meier analysis, the presence of cSAH on baseline CT was a predictor of recurrence of ICH (p < 0.001, by the log-rank test) (figure 2). The prespecified univariable predictors of recurrent ICH were illustrated in table 3. In univariable Cox regression analysis, presence of cSAH (HR 2.95, 95% CI 1.65–5.28; p < 0.0001), presence of cSS (HR 2.63, 95% CI 1.44–4.81; p = 0.002), and history of previous ICH (HR 2.69, 95% CI 1.34–5.41; p = 0.005) were associated with recurrent symptomatic ICH. After adjusting for age, antiplatelet use, warfarin use, history of previous ICH, lobar CMBs, and cSS, presence of cSAH (HR 2.53, 95% CI 1.39–4.62; p = 0.002) remained significant in multivariable Cox regression model. Using a prespecified clinically applicable model controlling for age, antiplatelet use, warfarin use, and previous ICH, cSAH on CT (HR 2.64, 95% CI 1.46–4.79; p = 0.001) was an independent predictor of recurrent ICH (table 3).

Figure 2. Recurrent Intracerebral Hemorrhage (ICH) in Patients With and Without CT Convexity Subarachnoid Hemorrhage (cSAH).

Kaplan-Meier curve estimates time to recurrent symptomatic ICH according to the presence or absence of cSAH on baseline CT in all patients with cerebral amyloid angiopathy ICH.

Table 3.

Prespecified Univariable and Multivariable Regression Analyses of Predictors of Recurrent Intracerebral Hemorrhage (ICH) in Cerebral Amyloid Angiopathy–Related ICH

In multivariable Cox regression analysis, presence of adjacent cSAH (HR 2.2, 95% CI 1.2–4.0, p = 0.008) was an independent predictor of recurrent ICH, whereas remote cSAH was not significantly associated with recurrent ICH (HR 1.7, 95% CI 0.8–3.6, p = 0.20).

Discussion

Results from our prospective cohort of ICH survivors demonstrated that cSAH is detected by noncontrast CT in up to 40.5% of patients with acute lobar ICH related to CAA. The presence of cSAH is associated with several markers of CAA such as cSS and APOE ε2 allele but also appears as an important prognostic marker, heralding an increased risk of recurrent ICH, independently from cSS. Because CT is the most commonly used diagnostic imaging method for patients with ICH, detection and analysis of cSAH on CT may be useful for assessment of future risk of ICH recurrence in patients with CAA, especially in resource-poor areas that have limited access to MRI.

In our cohort, we found that CT-based cSAH occurred in 40.6% of patients with CAA-related ICH and was associated with cSS, a key neuroimaging marker of CAA. Our findings are consistent with recent observations showing that cSAH might be an imaging marker of CAA, which can be assessed in the acute setting of ICH.^20,21 Our findings are also in line with results from the Edinburgh CT study, which showed that cSAH detected on CT was associated with pathologically proven CAA in patients with severe ICH who died and had a research autopsy.²²

cSAH, cSS, and lobar CMBs are all imaging markers associated with CAA.^1,2,23 In our study, we have explored the relationship among these 3 important imaging markers of CAA. We found that cSS, but not CMB, was associated with cSAH. Consistent with previous reports, our findings suggest that cSS and cSAH are likely 2 neuroimaging markers of a single pathophysiologic process and may represent subarachnoid hemorrhage in patients with CAA.^2,24–27 However, lobar CMB may represent a distinct disease subtype from cSS and cSAH.

Although a possible link between cSS and cSAH is proposed, the underlying mechanism is unclear. cSAH may occur as direct extension of lobar hematoma into the adjacent sulci or could be the result of leakage of meningeal vessels into the sulci.^26,27 Previous study suggested that cSAH may present with transient focal neurologic episodes when occurring in eloquent areas.²⁸ Individuals with CAA who are prone to develop cSAH would therefore likely have higher burden of cSS as many cSAH events in noneloquent areas would not come to clinical attention.

Our findings suggest that cSAH visible on CT is a strong predictor of recurrent ICH, independently from other predictors such as cSS, APOE, and previous symptomatic ICH.²⁹ In a recent study, we found that cSAH on MRI is a predictor of recurrent ICH after a CAA-ICH.³⁰ Results from this current study suggested that CT-visible cSAH was an independent prognostic factor of recurrent ICH, and extend MRI findings to patients with ICH evaluated with CT. Because MRI cannot be performed in many ICH cases because MRI is not available, not tolerated, or contraindicated, this CT-based prognostic marker could be widely used in clinical practice to assess the risk of recurrent ICH. Furthermore, we have observed a strong link between cSAH and preexisting dementia that has not been reported before. This may suggest that individuals with preexisting dementia have a higher small vessel disease burden and could be identified with cSAH. Our findings, together with previous reports, suggest that cSAH and cSS are both key hemorrhagic markers of recurrent ICH in patients with CAA. We were interested to observe that adjacent SAH was more strongly associated with recurrent ICH. A possible mechanism might be that adjacent SAH reflects a cascade of rupture of adjacent CAA-laden surface vessels due to pressure and cracking, whereas remote cSAH is less likely to be affected and might occur as a result of redistribution of blood in the subarachnoid space.

Our study has several potential limitations. First, it is a retrospective analysis of a prospectively collected single-center cohort and the sample size is relatively small. Second, we have included patients with both CT and MRI, which may lead to potential selection bias toward less severe ICH cases. Third, we only evaluated cSAH in CAA-related ICH survivors; whether our findings are generalizable to all patients with lobar ICH remains to be investigated in the future. Future studies with larger sample size are needed to validate our findings.

Our study demonstrated that cSAH is associated with cSS but not lobar CMBs. cSAH and cSS might be imaging markers that reflect increased vascular fragility in patients with CAA that makes them prone to develop large hemorrhages. Although MRI is increasingly used for imaging of ICH, CT is more widely available in almost all institutions worldwide that hospitalize patients with ICH. CT cSAH seems to be a strong imaging marker associated with recurrent ICH in patients with CAA and might be used to stratify the risk of recurrent bleeding in patients with ICH with only CT scans.

Glossary

CAA

cerebral amyloid angiopathy

CI

confidence interval

CMB

cerebral microbleed

cSAH

convexity subarachnoid hemorrhage

cSS

cortical superficial siderosis

FLAIR

fluid-attenuated inversion recovery

GRE

gradient-recalled echo

HR

hazard ratio

ICH

intracerebral hemorrhage

OR

odds ratio

SWI

susceptibility-weighted imaging

VIF

variance inflation factor

WMH

white matter hyperintensity

Appendix. Authors

Study Funding

This study was supported by the following NIH grants: R01AG047975, R01AG026484, and R01 NS070834. Q. Li was supported by an NIH StrokeNet fellowship.

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.