Atraumatic convexity subarachnoid haemorrhage describes spontaneous bleeding into the convexities of the brain sulci without parenchymal involvement. Its many causes include reversible cerebral vasoconstriction syndrome, cerebral sinus venous thrombosis, posterior reversible encephalopathy syndrome and (in older people) cerebral amyloid angiopathy. We describe the clinical and radiological features of non-traumatic convexity subarachnoid haemorrhage with its various presentations, causes, treatments and prognoses, and use clinical vignettes to highlight important clinical points and pitfalls.
- cerebrovascular disease
- clinical neurology
- subarachnoid haemorrhage
Data availability statement
No data are available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
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.
Atraumatic non-aneurysmal convexity subarachnoid haemorrhage (cSAH) represents approximately 6% of SAH and is distinct from aneurysmal-related SAH.1 Both are due to spontaneous haemorrhage into the subarachnoid space between the pia and arachnoid membranes. cSAH is confined to the cortical surfaces, without extending into the Sylvian or hemispheric fissure, basal cisterns, parenchyma or ventricles. By contrast, aneurysmal SAH tends to occur around the Circle of Willis, with subarachnoid bleeding and intracerebral haemorrhage (ICH) concentrated around the site of the aneurysm.2
The causes of cSAH in younger people include reversible cerebral vasoconstriction syndrome (RCVS), posterior reversible encephalopathy syndrome (PRES), cerebral sinus venous thrombosis and others.1 3 In people aged over 60 years, cerebral amyloid angiopathy (CAA) is a common cause.1 4 Although people with CAA-related cSAH generally recover well from the episode, they remain at increased risk of lobar haemorrhage (by 13% per year (95% CI 10% to 17%)), recurrent cSAH (by 11% per year (95% CI 8% to 15%)) and ischaemic stroke (by 5% per year (95% CI 3% to 8%)).5 6 Furthermore, cSAH may also rarely accompany inflammatory CAA, a condition causing significant neurological disability if not recognised and treated early.7 People with Alzheimer’s disease receiving aducanumab, a monoclonal antibody targeting amyloid-β (Aβ), also have imaging abnormalities resembling amyloid angiitis.8
CAA probably results from an imbalance between the production and clearance of Aβ, a breakdown product of Aβ precursor protein. Excessive Aβ accumulates in the small leptomeningeal and cortical blood vessels, which become brittle and fragile, allowing blood leakage into the subarachnoid space.9 The triggers for inflammatory CAA remain unknown, but there is probably an inappropriate perivascular immune response, with predominantly granulomatous inflammatory reactivity to vascular Aβ deposition within the meningeal and cortical vessels.10 11
cSAH presents with a range of symptoms, depending on its cause and the patient’s age. Common symptoms include headache, impaired level of consciousness, seizures, confusion, transient focal neurological episodes, persistent focal neurological deficits, visual abnormalities, nausea and vomiting (table 1).
Reversible cerebral vasoconstriction syndrome
cSAH presenting with thunderclap headache—typically in young people and especially women—is usually caused by RCVS. Occasionally there are associated focal neurological symptoms, but these may not be immediately apparent. It is important to look for potential triggers, including recent use of vasoconstrictive agents, including herbal medicines, illicit drugs and several prescribed medications. Other relevant medical history includes migraine, which may be associated with RCVS.
Posterior reversible encephalopathy syndrome
While RCVS and PRES are considered distinct entities, they overlap in terms of triggers and clinical features. In suspected PRES, clinicians should identify specific triggers, including elevated blood pressure, pre-eclampsia/eclampsia, cytotoxic or immunosuppressive agents, renal failure and autoimmune disorders. Common presenting features of PRES include headache, cognitive impairment or altered conscious state, visual symptoms and seizures.12
Cerebral venous sinus thrombosis
Patients with cerebral venous sinus thrombosis may have worsening subacute headaches, often with focal neurological symptoms. It is important to identify thrombotic risk factors including pregnancy, dehydration and other thrombotic tendencies, with or without a family history.
Cerebral amyloid angiopathy–related convexity SAH
cSAH in CAA typically presents in elderly people with recurrent transient focal neurological episodes (also known as amyloid spells), manifesting as brief motor, somatosensory, visual or language disturbances.4 13 14 These may easily be misdiagnosed as recurrent transient ischaemic attacks and treated with antithrombotic medications, with potentially fatal complications. The episodes probably result from cortical spreading depolarisation with its stereotypical migratory spread; their duration of several minutes rather than seconds also helps their clinical distinction from seizures.13–15 When in doubt, early electroencephalogram (EEG) may help. Although inflammatory CAA rarely presents with cSAH, this condition is important to consider early in elderly people with subacute onset of confusion and imaging features of CAA, especially if there are also confluent regions of vasogenic oedema. Delayed diagnosis may lead to significant neurological disability. Depending on the location, patients may have other focal clinical cortical features including dysphasia and neglect. Often, however, there are no focal motor deficits. Early EEG may also help in managing inflammatory CAA, as epileptic seizures are common.10
While it is important to recognise that cSAH is distinct from aneurysmal SAH—so avoiding unnecessary digital subtraction angiography—it can be challenging to identify its cause. After careful clinical history and examination, several investigations may help with the diagnosis (table 1 and figure 1).
CT scan of head
The first-line investigation for all patients with suspected cSAH is a non-contrast CT scan of the head (figure 2A). This needs scrutinising carefully, as it is easy to overlook small amounts of subarachnoid blood. We find the coronal view especially helpful in identifying subtle bleeds. CT scanning has high sensitivity (98%) for detecting SAH within 6 hours of onset, though only about 90% after 6 hours.16 cSAH probably has a lower sensitivity, in view of the smaller blood volume.
In RCVS, a CT cerebral angiogram may identify segmental narrowing of multiple intracranial arteries, suggesting vasospasm, usually in multiple vascular territories. A CT cerebral venogram is appropriate if suspecting cerebral venous sinus thrombosis.
MRI/A/V with SWI
In the acute phase of cSAH, an MR scan of the brain may show an increase in T2/FLAIR and T1 signals in the corresponding sulcus (figure 2B,C). However, very early after the presentation, the scan may be normal if there is only a small leptomeningeal protein leak.17 Susceptibility-weighted imaging (SWI) and gradient recalled echo (GRE) are MRI T2*sequences that detect paramagnetic material as a hypointense signal. This susceptibility effect may also not be apparent in the early phase of cSAH (figure 2D). As time passes and blood products degenerate leaving haemosiderin deposits, the SWI and GRE phase becomes hypointense, giving the appearance of superficial siderosis. By this time, the T2/fluid-attenuated inversion recovery (FLAIR) and T1 hyperintense signal have typically resolved (figure 2F,G). Interestingly, these dynamic transient MRI sulcal T2/T1 and SWI abnormalities also occur in people with Alzheimer’s disease treated with monoclonal anti-amyloid therapy: this is termed amyloid-related imaging abnormalities with haemosiderin deposits (ARIA H). Similar transient sulcal changes without superficial siderosis may also develop without haemorrhage, probably from proteinaceous fluid leaking into the leptomeningeal space (ARIA E).18
MR brain imaging and MR cerebral angiography can provide important clues to the underlying cause of cSAH. The Boston criteria are traditionally used to diagnose CAA but these require a biopsy or autopsy to confirm definite CAA. The recently revised Boston criteria (version 2.0) include updated clinical and MRI criteria, increasing the sensitivity of diagnosis.19 For example, probable CAA can now be diagnosed in those aged over 50 years presenting with spontaneous ICH. Also, transient focal neurological episodes or cognitive impairment can be diagnosed in those with MRI evidence of either two lobar haemorrhagic lesions (including either ICH, cerebral microbleeds, or superficial siderosis) or one lobar haemorrhagic lesion and another feature such as large perivascular spaces in the centrum semiovale region or multiple (>10) white matter hyperintensity lesions. Cerebral microbleeds are small collections of blood appearing as hypointense signal visible with T2* imaging. Although punctate DWI lesions may be associated with cSAH, they are not included in version 2.0 of the Boston criteria9 19 (figure 3G).
Vasogenic oedema in PRES is predominantly posterior parietal and occipital, although occasionally it involves the brainstem and frontal lobes.20 Typically, vasogenic oedema appears as a hyperintense signal on FLAIR images and increased diffusion on apparent diffusion coefficient images. These changes resemble a ‘finger glove’ pattern as the oedema spreads along the white matter tracts. While these changes often develop in the subcortical white matter, they can extend to involve the grey matter. Patients with inflammatory CAA can have single or multiple confluent hyperintense lesions on FLAIR images, suggesting vasogenic oedema. There is often corresponding increased leptomeningeal enhancement on post-contrast T1 images21 (figure 4E).
In RCVS, CT or MR angiography shows diffuse and segmental narrowing or “beading” of multiple vessels. The improved resolution of these scans means that digital subtraction angiography is now rarely used for RCVS (figure 2E).
Positron emission tomography (PET) imaging
The gold standard of diagnosis for CAA requires pathological sampling.19 However, brain biopsy is invasive and rarely performed, unless there is suspicion of inflammatory CAA. While Pittsburg compound B and other amyloid-specific ligands used in PET imaging show increased uptake in the cortical region in CAA,22 their use is currently limited to the research setting.
Lumbar puncture is traditionally recommended for people with “thunderclap” headaches if the CT scan of the head showed no evidence of SAH and there is no contraindication, looking for red blood cells and xanthochromia. Red blood cells may be from a traumatic tap, but xanthochromia is more specific for haemorrhage, although its sensitivity relates to the timing of lumbar puncture after headache onset. Xanthochromia typically develops as early as 2 hours after onset, with maximum sensitivity around 12 hours after symptom onset.16 Lumbar puncture is not usually needed where imaging has established cSAH unless suspecting infection or other immune-inflammatory causes. In most other instances, cerebrospinal fluid (CSF) is normal but may show non-specific changes of increased protein concentration and pleocytosis.
EEG should be considered in patients with cSAH and focal seizures or altered mental states. While the altered mental states may be from parenchymal involvement, it is important to exclude non-convulsive status epilepticus. This is particularly pertinent in patients with suspected inflammatory CAA or in PRES with a fluctuating conscious state. EEG may show generalised or focal slowing or active epileptic discharges.
Serological and other tests
Patients with cSAH often undergo tests for metabolic derangement, electrolyte abnormalities, inflammatory and hypercoagulable conditions, vasculitic screen and urinary drug screen. These are usually normal, but when abnormal may provide clues as to the cause and prognosis. This is particularly important in patients with PRES, who may have deranged electrolytes including hypomagnesaemia, renal failure, elevated C reactive protein (associated with poor outcomes) and hypoalbuminaemia.
There are proposed clinical and radiological criteria incorporating T2/FLAIR asymmetrical lesions and the Boston criteria for the diagnosis of probable inflammatory CAA.23 However, in patients with T2/FLAIR oedematous lesions—where other MR scan findings are inconclusive and other inflammatory or immune causes (including primary angiitis of the central nervous system) and malignancy cannot be ruled out with certainty—brain biopsy should be considered early. Immunosuppressive therapy should be started without delay, particularly if suspecting primary angiitis of the central nervous system (a particularly aggressive vasculitis that can be fatal if untreated). Corticosteroids probably have little effect on the diagnostic yield of the biopsy in the acute setting. The area of biopsy is often more helpful and should be targeted to the region of MRI T2/FLAIR oedematous abnormality, preferably in the cortex and leptomeninges.24 25
The management of cSAH includes treating complications, removing potential triggers, minimising the risk of haemorrhage extension and recurrence, and addressing underlying causes.
Patients with associated seizures need timely antiseizure medication, especially in PRES or inflammatory CAA, which can mask non-convulsive status epilepticus. Anyone with cSAH requires close monitoring and blood pressure control but particularly so in PRES. Selective use of antihypertensive medication depends on the cause of cSAH. In PRES, intravenous labetalol, magnesium and hydralazine are often used to manage hypertension. In RCVS, calcium blockers such as nimodipine are often used for their vasodilating as well as antihypertensive properties. It is important to avoid identified and other known triggers.
In people with CAA-related cSAH, there are no clear guidelines on antihypertensive medications to prevent ICH, and only observational evidence suggests a risk of haemorrhage expansion to ICH in the early phase.26 Nevertheless, it seems reasonable to control blood pressure early, along similar lines to that recommended for acute ICH (ie, below 140 mm Hg systolic).14
Antithrombotic medication, particularly clopidogrel and anticoagulants, should be avoided in acute cSAH unless there is cerebral venous sinus thrombosis. Where there is a strong indication for their use, such as atrial fibrillation, it may be difficult to decide when to restart antithrombotic agents. This requires a careful risk–benefit discussion, taking account of individual circumstances. Note that CAA probably has the highest risk of recurrent cSAH and an additional increased risk of ICH.6 However, recent meta-analyses have shown that in survivors of non-traumatic ICH, oral anticoagulation reduces thromboembolic events and all-cause mortality without significantly increasing the risk of recurrent ICH.27 28 There are several randomised controlled trials underway, such as ENRICH-AF,29 to evaluate the safety of direct oral anticoagulants (DOACs) in survivors of ICH, including SAH with atrial fibrillation. Until these results are available, it is reasonable to restart DOACs at a lower dose sometime after the acute phase in patients at high risk of cardioembolic events, especially in people with a high CHA2DS2-VASc score.
Alternatively, if the risk of bleeding is considered to be too high even for a lower dose of DOAC—such as in patients with CAA-associated cSAH with previous lobar haemorrhages and extensive superficial siderosis—a possible option is left atrial appendage occlusion, although patients may still need postprocedural antithrombotics in the short term.30
The recurrent nature of CAA-related transient focal neurological episodes is probably due to cortical spreading depolarisation. Clinicians can reassure patients that such recurrent stereotypical symptoms in themselves do not necessarily reflect the recurrence of haemorrhage. Although some patients are treated with antiseizure medications, especially those with anti-migraine properties to reduce the frequency of transient focal neurological episodes, such treatment probably does not alter the natural history, as most episodes are self-limiting. It is even less certain whether suppressing cortical spreading depolarisation alters the risk of future bleeding.
In patients with cSAH related to cerebral venous sinus thrombosis, it is important to use therapeutic anticoagulation early, even in those with intraparenchymal haemorrhage or venous ischaemia.
For patients with inflammatory CAA, clinicians should consider intravenous corticosteroids early. In refractory cases with poor response to initial treatment, case series have shown possible benefits from maintenance oral prednisolone, either alone or with other immunosuppressants such as cyclophosphamide, with a possible higher chance of resolution of inflammation and reduction in relapses.31 32 Often such patients require ongoing maintenance of oral immune suppressant therapy for some time.
It is important to advise patients regarding driving following cSAH. Any residual neurological deficit that may affect driving means the patient must notify their relevant driving licensing authority and may need to be cleared through an occupational therapy on-road driving assessment. Most jurisdictions also have clear driving guidelines regarding epileptic seizures. The more challenging task is how to advise patients who have recovered with no residual neurological deficit, as often occurs in patients with RCVS or CAA-associated transient focal neurological episodes. As an example, while the updated Assessing Fitness to Drive 2022 edition in Australia briefly mentions cSAH associated with these two conditions, the only specific recommendation is that it depends on the ‘presence of neurological impairments’.33 Thus, the treating physician should check with their own local driving authority. In the absence of definite specific recommendations, it seems reasonable to advise asymptomatic patients also to notify their relevant driving authority and not to drive until they have further clarification from the authority.
The prognosis of cSAH very much depends on its underlying causes. RCVS generally has a favourable prognosis, with long-term outcomes predominantly determined by the presence of stroke. In most patients, headaches and vasospasm resolve within weeks and few have residual deficits from their stroke. In the absence of large stroke and uncontrolled brain oedema, RCVS mortality is less than 1%.34 Recurrence occasionally occurs but is unlikely after triggers have been removed. Although PRES is traditionally considered as reversible and benign, recent case series have found up to 19% mortality, with functional impairments in up to 44%. Factors associated with poor outcome include severe encephalopathy, malignant hypertension, elevated C reactive protein, coagulopathy, low CSF glucose, underlying neoplasia, multiple comorbidities, corpus callosum involvement, extensive cerebral oedema and large haemorrhage or infarcts.35 Patients with cerebral venous sinus thrombosis usually make a good recovery, unless there is a large infarct or haemorrhage. However, because recurrence is possible, it is important that they must continue oral anticoagulation, often for life.
People with CAA-related cSAH appear at increased risk of both ICH and ischaemic stroke.5 Antithrombotics are best avoided unless strongly indicated, as in atrial fibrillation and/or high cardioembolic risk. A recent meta-analysis showed that antithrombotics taken after either a transient focal neurological episode or after motor transient focal neurological episodes lead to a greater risk of developing intracerebral (including lobar) haemorrhage; the finding of superficial siderosis was associated with greater mortality.36
Although early recognition and treatment of inflammatory CAA can give a good outcome, such patients need to be closely monitored, as recurrences can occur. A recent cohort study found that up to 38% of patients had at least one relapse at 2 years. Following initial clinical improvement, patients should be maintained on slow tapering oral corticosteroids, at least until follow up MR imaging shows radiological resolution of inflammation. Following this, it is reasonable to follow up with MR scans at 3, 6, 12 months and then yearly intervals. If new attacks occur, patients may benefit from early repeat treatment with intravenous immunosuppressants.37
Spontaneous non-traumatic cSAH is increasingly recognised. A careful history and examination with targeted investigations is important to help early recognition of the underlying cause. The causes of cSAH vary with age. In the young, while most are due to RCVS, finding posterior dominant T2/FLAIR oedematous lesions would suggest PRES. It is also important to rule out cerebral venous sinus thrombosis using CT or MR venogram. In the elderly, it is important to consider CAA-related conditions, as patients have an increased risk of future haemorrhage. A good understanding of investigation findings, in particular advanced imaging, can lead to an accurate diagnosis and avoid potentially dangerous differential misdiagnosis.
Non-traumatic convexity subarachnoid haemorrhage (cSAH) (comprising 6% of all SAHs) has causes that vary with age: in young people most are due to reversible cerebral vasoconstriction syndrome and often present with acute headache; in those aged over 60 years, clinicians should consider cerebral amyloid angiopathy-related pathology.
Associated vasogenic oedema, especially in the posterior parietal and occipital areas, suggests posterior reversible encephalopathy syndrome.
In young people with convexity SAH who present with headache and focal neurological symptoms, CT venogram or MR venogram should be considered to rule out cerebral venous sinus thrombosis.
CAA-related convexity SAH in older people typically presents with transient neurological focal episodes (amyloid spells); there may also be associated punctate DWI lesions on MRI, easily mistaken for stroke or transient ischaemic attack, thus risking treatment (dangerously) with antithrombotic therapy.
Hostettler, I.C., et al., Risk of intracranial haemorrhage and ischaemic stroke after convexity subarachnoid haemorrhage in cerebral amyloid angiopathy: international individual patient data pooled analysis. J Neurol, 2021.
Charidimou, A., et al., The Boston criteria version 2.0 for cerebral amyloid angiopathy: a multicentre, retrospective, MRI-neuropathology diagnostic accuracy study. Lancet Neurol, 2022. 21(8): p. 714–725.
Antolini, L., et al., Spontaneous ARIA-like Events in Cerebral Amyloid Angiopathy-Related Inflammation: A Multicenter Prospective Longitudinal Cohort Study. Neurology, 2021. 97(18): p. e1809-e1822.
Data availability statement
No data are available.
Contributors Conceptualisation and design of the review: JVL, TP and HM. Collection of cases: JVL and SS. Literature review: JVL, TP and BC. Drafting of the manuscript: JVL. Critical revision of manuscript: HM, SS, BC and TP. All authors have reviewed and approved the final version of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally reviewed by David Werring, London, UK.