Contribution of Convexal Subarachnoid Hemorrhage to Disease Progression in Cerebral Amyloid Angiopathy
Background and Purpose—Cerebral amyloid angiopathy–related cortical superficial siderosis (cSS) seems to indicate an increased risk of subsequent intracerebral hemorrhage (ICH). We wanted to identify the mechanisms and sequence of hemorrhagic events which are responsible for this association.
Methods—During a 9-year-period, we identified patients with spontaneous convexal subarachnoid hemorrhage (cSAH) and performed a careful longitudinal analysis of clinical and neuroimaging data. A close imaging–histopathologic correlation was performed in one patient.
Results—Of 38 cSAH patients (mean age, 77±11 years), 29 (76%) had imaging features of cerebral amyloid angiopathy on baseline magnetic resonance imaging. Twenty-six (68%) had cSS. Sixteen subjects underwent postcontrast magnetic resonance imaging. Extravasation of gadolinium at the site of the acute cSAH was seen on all postcontrast scans. After a mean of 24±22 (range 1–78) months of follow-up, 15 (39%) had experienced recurrent cSAHs and 14 (37%) had suffered lobar ICHs. Of 22 new ICHs, 17 occurred at sites of previous cSAHs or cSS. Repeated neuroimaging showed expansion of cSAH into the brain parenchyma and evolution of a lobar ICH in 4 patients. Propagation of cSS was observed in 21 (55%) patients, with 14 of those having experienced recurrent cSAHs. In the autopsy case, leakage of meningeal vessels affected by cerebral amyloid angiopathy was noted.
Conclusions—In cerebral amyloid angiopathy, leakage of meningeal vessels seems to be a major cause for recurrent intrasulcal bleedings, which lead to the propagation of cSS and indicate sites with increased vulnerability for future ICH. Intracerebral bleedings may also develop directly from or in extension of a cSAH.
Intracerebral hemorrhage (ICH) is the most devastating form of stroke and a major public health problem.1 Despite all preventive efforts, the incidence of ICH among elderly people rises dramatically.2 An increasing rate of bleeding-prone cerebral small vessel disorders in the aging population might account for this phenomenon. Among those, cerebral amyloid angiopathy (CAA)—an up to now untreatable disorder characterized by deposition of amyloid in cerebral vessels—seems to be most relevant. CAA was previously considered to be a curiosity, but—as people grow older—is emerging as a rather common cerebral small vessel disorders and major cause of spontaneous ICH.3 The risk of ICH in CAA patients has been related to various factors, such as the extent of associated leukoaraiosis,4 the number of lobar microbleeds5 (MB), or the number of previous clinical episodes of hemorrhage.6 Cortical superficial siderosis (cSS) has recently been suggested as another indicator of increased risk for future ICH in this population.7,8 cSS is characterized by the deposition of blood breakdown products in the superficial layers of the cerebral cortex and meninges and seems to be a frequent and characteristic imaging feature of bleeding-prone CAA.9 In a longitudinal observation of 51 patients with cSS, ≈50% experienced intracranial bleedings during a period of 35 months.7 A European multicenter study of 118 CAA patients confirmed that the presence of cSS on magnetic resonance imaging (MRI) significantly increases the risk of future symptomatic lobar ICH and suggested cSS as a useful and independent prognostic marker of intracerebral bleeding risk in CAA.8 The mechanisms for this close association, however, are not entirely clear.
Nontraumatic convexal subarachnoid hemorrhage (cSAH) has also been associated with CAA,10–12 and it has been speculated that recurrent intrasulcal bleedings potentially cause cSS.11 This is supported by experimental data, which showed that cSS originates from recurrent bleeding into the subarachnoid space13 because blood products from subarachnoid hemorrhage (SAH) penetrate the pia mater and are deposited in superficial cortical layers. Furthermore, isolated CAA-related cSAH also seems to indicate poor outcome.12 However, clinical observations of cSAH or associated lobar ICH are scarce.10,14
Based on the assumption that CAA-associated cSAH, cSS, and lobar ICH are intimately related, we therefore wanted to explore the mechanisms and sequence of events in this association by performing a careful longitudinal analysis of clinical and imaging data of a series of consecutive patients with spontaneous isolated cSAH. This was supported by the possibility of a close imaging–histopathologic correlation in one patient.
Subjects and Methods
Selection of Patients
We retrospectively searched the computed tomography (CT) reports at an academic primary and tertiary care hospital for all patients identified with an SAH from October 2004 to March 2014. From all patients with SAH, we extracted subjects with nontraumatic, nonaneurysmal SAH for a review of the CT scans (Figure 1). According to own previous work, we defined cSAH by evidence for acute blood in ≥1 adjacent cortical sulci at the convexity of the brain. Patients with concomitant evidence of blood in the basal cisterns, interhemispheric or Sylvian fissures, with intracerebral bleedings that might have ruptured into the subarachnoid space, or with damage to the brain parenchyma adjacent to the cSAH were excluded.12 We also excluded patients with cSAH related to recanalization procedures or subjects who had not undergone MRI at baseline. When our search method identified multiple cSAHs in a single patient, the first cSAH was considered the baseline event. The study was approved by the hospital institutional review board and ethics committee.
We used the medical and nursing documentation and communication network of Styria (MEDOCS) to collect clinical and neuroimaging data for a careful longitudinal analysis. MEDOCS is a hospital information system implemented in all 21 public hospitals in the district of Styria15 with 1 210 971 registered inhabitants in the year 2013 and 450 000 people living in the direct catchment area of the University Hospital of Graz. MEDOCS provides access to all laboratory, clinical, and neuroimaging data acquired in the 21 public hospitals. We collected all available information on clinical symptomatology that prompted the admission of the cSAH patients to hospitals or the initiation of neuroimaging and extracted from the system all CT and MRI scans of the brain for a subsequent systematic review. Moreover, we invited all identified patients for a clinical and MRI follow-up examination.
MRI of the brain was performed at 1.5 T and included axial fluid–attenuated inversion recovery, diffusion-weighted imaging, and gradient echo T2*-weighted sequences (slice thickness 5 mm) for detection of past bleedings.16,17
A single expert (C. Enzinger), who was blinded to clinical data, evaluated the baseline MRI scans in a standardized manner, assessed the exact location and extent of the cSAH, and recorded the presence of old intraparenchymal hemorrhages (lesion with presumed hemosiderin deposition of >5 mm in diameter), MBs (areas of T2* signal loss ≤5 mm),16,17 and cSS which was defined by linear signal loss along the cerebral cortex on gradient echo T2*-weighted sequences.9 The presence and number of focal diffusion-weighted imaging abnormalities, old infarcts/lacunes, and other morphological abnormalities were also recorded. White matter hyperintensities were rated as absent (score 0), punctate (score 1), early confluent (score 2), or confluent (score 3).17,18 Images were also assessed for the presence of focal or diffuse brain swelling and evidence for contrast enhancement. Possible or probable CAA was defined according to the modified Boston criteria.9
For longitudinal analysis, we retrieved all brain imaging data from MEDOCS and reviewed them for recurrent acute cSAH, new ICH, or ischemic infarction. The blinded expert also performed a side-by-side comparison of the baseline and subsequent scans regarding the overlap in location of initial abnormalities and subsequent bleeding events. Repeat MR scans were assessed for evidence of propagation of cSS, which was defined by the occurrence of new areas with signal loss at the surface of the cerebral cortex on the T2*-weighted sequence.
We reviewed medical records for clinical symptoms at presentation, presence of cerebrovascular risk factors, and demographics. Hypertension, diabetes mellitus, and dyslipidemia were coded as previously described.12
The clinical information collected regarding neurological symptomatology was reviewed by an experienced stroke neurologist (M. Beitzke). An acute stroke syndrome was diagnosed if a patient had typical symptoms that lasted longer than 24 hours. Transient neurological attacks (TNA) were defined as attacks of sudden neurological symptoms that completely resolved within 24 hours. TNAs were further dichotomized into (1) TNA with transient focal symptoms, including motor symptoms like weakness or limb shaking, sensory symptoms, or aphasia and (2) TNA with transient nonfocal symptoms, including sudden onset of confusion, dizziness, or unwell feelings.19 Clinical follow-up was performed in 2 phases (2004–2010 and 2010–2014). The stroke neurologist screened all eligible participants in person and interviewed the patient or next of kin or their guardians for occurrence of focal or nonfocal TNA by asking for transient neurological symptoms and carefully registered symptoms and attack characteristics. Additional clinical information was obtained by contacting family physicians directly or by telephone.
Postmortem analyses were performed on one patient within 5 days from CT scanning that had shown an acute cSAH in one hemisphere and a lobar ICH without rupture into the subarachnoid space in the other. At autopsy, the brain was coronally dissected, guided by the CT images, to target the intrasulcal bleeding and the adjacent cerebral cortex. Tissue from that area was immediately fixed in 10% neutral buffered formalin and embedded in paraffin.
10-μm-thick sections of formal-fixed paraffin-embedded brain tissue were deparaffinized and treated with heat-induced epitope retrieval at 98°C for 40 minutes in water bath, cooled down to room temperature for 20 minutes, and treated with the blocking solution S2023 from DAKO. The primary Amyloid P antibody (polyclonal rabbit antihuman P-Component, Cat. No. A0302; DAKO, Vienna, Austria) was stained in a dilution of 1:300 for 30 minutes at room temperature. The antibody reactions were detected with DAKO K5001 Detection System using AEC Substrate as chromogen. H&E staining was done on deparaffinized tissue with a progressive Meyer’s hematoxylin and counterstaining with Eosin Y.
Relevant demographic, clinical, and radiological data were tabulated. Quantitative data are expressed as mean±SD. The Mann–Whitney U test was used to compare age between patient groups. The Fisher exact test and Pearson Chi-Square test were used to compare the clinical and imaging characteristics between patients with and without a new ICH or a recurrent cSAH. The level of significance was set at P<0.05. Incidence rates were calculated based on person-years of observation.20 The Statistical Package for the Social Sciences (version 20.0; SPSS Inc., Chicago, IL) was used for data analysis.
We identified 1178 patients diagnosed with SAH. Two-hundred-forty-nine (21%) had nontraumatic, nonaneurysmal SAH and 45 (3.8%) fulfilled the criteria of cSAH. Comprehensive clinical and neuroimaging work-up, including baseline MRI, and follow-up data were available in 38 cSAH patients (Figure 1). Demographic, clinical, and MRI characteristics of the patients at baseline cSAH are detailed in Table. Baseline clinical and imaging data of 18 of these patients have been reported previously.12
Neuroimaging Characteristics at Baseline
Figure 2 shows representative neuroimaging findings from a 68-year-old man.
Of the 38 cSAH patients, 29 (76%) had imaging features of CAA on the baseline MRI (Figure 1). Ten (26%) had possible and 19 (50%) had probable CAA. Imaging features of CAA consisted of past parenchymal bleedings in 18 (47%) and cSS in 26 (68%) patients. Seven of the 18 patients with old parenchymal bleedings had old ICHs and 16 had cerebral MBs. In the latter, a total of 486 MBs was found with 431 (89%) of the MBs located cortico-subcortically. A single patient (under chronic hemodialysis) had 360 MBs. Of the 26 patients with cSS, 10 (38%) had cSS in the absence of any evidence of past intracerebral bleedings (MBs or old ICH).
Post-contrast MRI was available in 16 patients and showed linear extravasation of gadolinium at the location of the cSAH in all of them. Eight (50%) of these patients had more widespread leptomeningeal enhancement, which was not restricted to the site of the cSAH but occurred also in adjacent cerebral sulci (Figure 2).
Nineteen cSAHs were located in the right hemisphere, 14 in the left hemisphere, and 5 were bilateral. Convexal SAH involved 1 sulcus (n=19), 2 to 3 sulci (n=11), or >3 sulci (n=8). Focal swelling of the cortex in the immediate vicinity of the intrasulcal bleedings was noted in 34 patients (89%) on CT and confirmed on MRI in all of those. Five (13%) cSAH patients had concurrent acute lobar ICH, which was located in the contralateral hemisphere (n=4) or in the ipsilateral hemisphere but remote from the cSAH (n=1), and 2 patients had developed a lobar ICH at the site of the cSAH when the baseline MRI was performed. Three (8%) patients had territorial ischemic infarcts and 8 (21%) had concurrent small cortico-subcortical lesions on diffusion-weighted imaging, which were also remote from the cSAH.
Longitudinal Analysis of Neuroimaging Data
The 38 subjects were followed over a mean period of 24±22 (range 1–78) months (75.5 person-years) during which they underwent a total of 123 CT and 49 MRI scans of the brain. New intracranial bleedings, that is, a new ICH or recurrent cSAH occurred in 19 (50%) of 38 patients. Sixteen (84%) of them had imaging features of CAA on the baseline MRI, and all 16 patients had cSS (Table I in the online-only Data Supplement). Twenty-five patients underwent repeated T2* MRI for the assessment of cSS propagation.
During the observational period, 14 (37%) patients experienced 22 symptomatic intracerebral bleedings. The incidence rate for a new ICH was 19×100 years−1. Imaging features of CAA on baseline MRI had been present in 11 (78%) of those 14 new ICH (ICH+) patients. All subsequent intracerebral bleedings were lobar, that is, occurred in a cortico-subcortical location. Seventeen new ICHs were located at sites of previous cSAHs or cSS. In 8 subjects, new ICHs occurred at the site of a previous cSAH (Figure I in the online-only Data Supplement). In 4 patients, serial neuroimaging documented expansion of a primary cSAH into the adjacent brain parenchyma as the cause of intracerebral bleeding (Figure 3). Ten patients had preexistent cSS at the sites of new lobar ICHs. Two patients suffered large fatal ICHs at sites where cSAH, cSS, and MBs had been documented on repeated previous scans. Five new ICHs had no relation to previous cSAH, cSS, or MBs.
There were no clinical factors that predicted subsequent intracerebral bleedings, except arterial hypertension (new ICH+ versus new ICH−; 10 of 14 versus 9 of 24, P=0.044). Neuroimaging findings of more advanced cerebral small vessel disorders were associated with new ICH (Table II in the online-only Data Supplement). There was no significant difference of age between new ICH+ and new ICH− patients.
Neuroimaging provided evidence of 34 recurrent acute cSAHs in 15 (39%) of the 38 patients. The incidence rate for recurrent acute cSAH thus was 20×100 years−1. Imaging features of CAA on the baseline MRI had been present in 14 (93%) of the 15 patients who suffered a recurrent cSAH. Eight (53%) had several recurrent intrasulcal bleedings. New cSAHs occurred together with large lobar ICHs in 6 patients. Neuroimaging features of advanced cerebral small vessel disorders were associated with recurrent cSAHs (Table III in the online-only Data Supplement). There was no association of cerebrovascular risk factors and recurrent cSAH and no significant difference of age between recurrent cSAH+ and recurrent cSAH− patients.
Propagation of cSS
Propagation of cSS was noted in 21 of 25 (84%) patients who underwent repeated T2* MRI (incidence rate for cSS propagation: 28×100 years−1). Six patients had propagation of cSS at sites of baseline intrasulcal contrast enhancement where acute bleeding was not seen on baseline MRI. Fourteen (66%) of 21 patients with cSS propagation had recurrent cSAHs on repeated imaging. Propagation of cSS was associated with both recurrent cSAH and new ICH (recurrent cSAH+ versus recurrent cSAH−, 14 of 14 versus 7 of 11, P=0.014 and new ICH+ versus new ICH−, 13 of 13 versus 8 of 12, P=0.023).
Subsequent Ischemic Strokes
Three patients suffered territorial infarcts and 7 patients had small cortico-subcortical diffusion-weighted imaging lesions (incidence rate for ischemic events: 13×100 years−1).
The most frequent clinical presentation of cSAH were TNAs (26 [68%] of the 38 patients; Table I in the online-only Data Supplement). Repeated stereotyped transient symptoms were recorded in 18 (47%) patients. TNA with transient focal symptoms occurred in 20 and TNA with transient nonfocal symptoms in 6 patients.
cSAH was a clinically silent (incidental) imaging finding in 13 subjects. Of those, 3 had territorial ischemic infarcts and 10 had acute lobar ICH. Recurrent transient neurological symptoms occurred in 14 (66%) of the 21 patients in whom propagation of cSS was documented on repeated MRI. Seven patients had propagation of cSS in the absence of clinical symptoms. Focal TNAs were associated with both new ICH and recurrent cSAH (new ICH+ versus new ICH−; 12 of 14 versus 8 of 24, P=0.002 and recurrent cSAH+ versus recurrent cSAH−; 11 of 15 versus 9 of 23, P=0.039).
Six patients presented with thunderclap headache, suggestive of a reversible cerebral vasoconstriction syndrome at the first occurrence of cSAH, and none of them experienced a recurrent cerebrovascular event during the observational period.
Histopathologic analyses were performed on postmortem brain samples in a 71-year-old woman who had been admitted with a large lobar ICH in the left hemisphere and an acute concurrent cSAH in the right central sulcus. On day 4, she had developed several episodes of limb shaking of the left arm. On day 5, she finally died of treatment-resistant cardiac arrhythmia.
Macropathological inspection of the coronally dissected brain revealed focal edematous swelling of the cerebral cortex in close vicinity to the right central sulcus (Figure 4).
Histopathologic evaluation of this region by H&E stains showed irregular hyalinosis of blood vessels in the arachnoidea. Perivasal and diffuse edema, indicative of distinct vessel-leakage, was pronounced adjacent to the arachnoidea. Extravasation of erythrocytes illustrative of vessel leakage was also noted in the cerebral cortex adjacent to the cSAH. Immunostaining revealed deposition of amyloid in the wall of several small leptomeningeal vessels but only in a small amount of cortical vessels (Figure 4).
This clinical study confirms the findings of previous work on hemorrhagic events in CAA and extends them in several directions. This encompasses (1) frequent recurrence of CAA-related cSAH, which was associated with a substantial risk for future symptomatic ICH, (2) widespread leakage of meningeal vessels in association with intrasulcal bleedings, which appeared to add to the propagation of cSS, and (3) direct development of lobar ICH from or in extension of cSAH.
Experimental data indicate that repeated bleeding into the subarachnoid space causes cSS.13 However, clinical observations of recurrent cSAH as the cause of cSS propagation are scarce.11 We observed repeated cSAH in 39% of our patients, and 66% of those with cSS propagation had repeatedly experienced a cSAH. At MRI follow-up, cSS often extended to sites without actual intrasulcal bleeding, but where leakage of contrast material had been present at baseline imaging. We cannot determine whether this indicates that already minimal seepage of erythrocytes suffices to cause cSS or whether small intrasulcal bleedings had developed later at these sites. It is quite obvious from our observations, however, that (recurrent) cSAHs may easily escape detection because they may be clinically silent or cause only subtle symptoms which are noncharacteristic for intracranial bleedings, including focal or nonfocal TNA. The observation that TNAs associated with cSS often occur repeatedly over limited time periods is in line with these considerations.21 Furthermore, CT scanning will also miss the detection of smaller cSAHs because they become isodense after a few days.
Cortical SS has been reported to indicate an increased risk for future symptomatic ICH.7,8 We also found that cSS on the baseline MRI was present in 84% of patients with acute cSAH who suffered subsequent bleedings. On the other hand, subsequent hemorrhage was infrequent when baseline MRI showed no preexisting cSS. The sequence of events and mechanisms underlying the association between cSS and future ICH, however, are not yet fully clear8,22 and several scenarios seem possible. cSS and ICH may occur as independent phenomena from increased fragility of cerebral small vessels. ICH itself may cause meningeal hemosiderosis by ruptures into the subarachnoid space. We here suggest a further possibility, that is, the development of ICH directly from or in extension of initially subarachnoid or cortical bleeding. In our series, we captured 4 patients with hematomas that expanded from an acute intrasulcal bleeding into the parenchyma. We assume that such evolution may have escaped previous investigations because it will only become apparent with rapid imaging of patients in the early phase of the disease, that is, before the hematoma has developed. In this context, the variable and often nonspecific clinical presentations of cSAH pose a particular problem. Our observation is also supported by some autopsy data, which suggest that bleedings in patients with sporadic type of CAA often occur first in the cerebral sulci and only subsequently expand into the brain parenchyma.23 Hematoma expansion from cSAH into the brain parenchyma may thus be an under-recognized mechanism of lobar ICH development in CAA.
We found linear contrast enhancement at the site of intrasulcal bleeding and observed that subsequent hematoma expansion developed primarily along the involved cerebral sulci. This could suggest contrast enhancement in the subarachnoid space to indicate a local clustering of multiple small diseased meningeal vessels. Such fragile vessels in the subarachnoid space would also seem susceptible to secondary shearing injury and rupture as the hematoma expands in a local cascade. Such process based on secondary shearing of blood vessels surrounding an ICH has been demonstrated in a neuropathological study,24 and also clustering of diseased vessels in CAA has been suggested previously.25 Presence of a generally increased vulnerability of the cerebral microvasculature is also supported by the rather frequent co-occurrence of remote small cortico-subcortical ischemic lesions. However, we certainly cannot exclude that some ICH in our series have arisen independent of all these mechanisms.
The conclusions from our neuroimaging observations are supported and extended by the histopathologic findings in a 71-year-old woman. We found moderate to severe CAA in several leptomeningeal vessels and evidence of distinctive meningeal vessel leakage at the site of acute cSAH. Signs of vascular leakage were also seen in the swollen cerebral cortex adjacent to the acute cSAH, albeit immunostaining did not identify a large amount of CAA-diseased cortical vessels. We can thus only speculate on the pathomechanisms causing the focal cortical swelling in the immediate vicinity of acute cSAHs observed in the vast majority of our patients. These include vessel dysfunction related directly to the presence of β-Amyloid in cortical vessels,26,27 vessel dysfunction caused by CAA-related inflammation,28 or vascular leakage caused by clusters of cortical spreading depression.29
There are several limitations to our study which come from single center observation, retrospective subject identification, nonstandard follow-up, and the limited sample size. The results should therefore be viewed with some caution and certainly cannot be transferred to patients with CAA in general. Focussing on leptomeningeal bleeding may indeed identify a subset of subjects within the broad spectrum of imaging manifestations of CAA. We also lacked statistical power to identify predictors of CAA-related ICH in a multifactorial manner. Larger multicentre prospective studies will thus be needed to confirm our results and clarify the role of CAA-related cSAH in predicting future ICH among other types of associated morphological damage. Despite these limitations, our study provides first evidence that in CAA, leakage of locally clustered, diseased, or at least affected meningeal blood vessels in the subarachnoid space causes intrasulcal hemorrhage. Several such bleedings promote the propagation of CAA-related cSS. Moreover, our results raise awareness of cSAH as a sign of active and probably more widespread meningeal disease in CAA that is prone to major lobar hemorrhage, which may develop directly from or in extension of cSAH.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.008778/-/DC1.
- Received January 15, 2015.
- Revision received March 11, 2015.
- Accepted April 1, 2015.
- © 2015 American Heart Association, Inc.
- van Asch CJ,
- Luitse MJ,
- Rinkel GJ,
- van der Tweel I,
- Algra A,
- Klijn CJ
- Béjot Y,
- Cordonnier C,
- Durier J,
- Aboa-Eboulé C,
- Rouaud O,
- Giroud M
- Charidimou A,
- Gang Q,
- Werring DJ
- Charidimou A,
- Kakar P,
- Fox Z,
- Werring DJ
- Poon MT,
- Fonville AF,
- Al-Shahi Salman R
- Beitzke M,
- Gattringer T,
- Enzinger C,
- Wagner G,
- Niederkorn K,
- Fazekas F
- Gell G,
- Madjaric M,
- Leodolter W,
- Köle W,
- Leitner H
- Roob G,
- Lechner A,
- Schmidt R,
- Flooh E,
- Hartung HP,
- Fazekas F
- Rothman KJ,
- Greenland S
- Rothman KJ,
- Greenland S
- Charidimou A,
- Peeters A,
- Fox Z,
- Gregoire SM,
- Vandermeeren Y,
- Laloux P,
- et al
- Smith EE,
- Greenberg SM
- Dreier JP,
- Victorov IV,
- Petzold GC,
- Major S,
- Windmüller O,
- Fernández-Klett F,
- et al