Clinical Radiology 69 (2014) 350e356
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Value of delayed MRI in angiogram-negative subarachnoid haemorrhage J. Woodfield a, *, N. Rane b, S. Cudlip a, J.V. Byrne b a b
Department of Neurosurgery, John Radcliffe Hospital, Oxford, UK Department of Neuroradiology, John Radcliffe Hospital, Oxford, UK
article in formation Article history: Received 7 October 2013 Accepted 5 November 2013
AIM: To assess the efficacy of delayed magnetic resonance imaging (MRI) in identifying a structural cause for angiogram-negative subarachnoid haemorrhage. MATERIALS AND METHODS: All patients presenting with spontaneous subarachnoid haemorrhage who had negative computed tomography (CT) angiography and catheter angiography between 2006 and 2012 were reviewed. RESULTS: During the 6 year period, 1023 angiograms were performed for a new presentation of subarachnoid haemorrhage. Of these, 242 (23.7%) did not show a cause for the haemorrhage. A second catheter angiogram was performed in 48 patients, and aneurysms were identified in two patients. Of the remaining 240 patients, 131 underwent a subsequent MRI brain. One hundred and five (80.2%) MRI examinations were performed 4 or more weeks after angiography. In two patients, cavernomas were identified as the likely bleeding source. In both patients, the pattern of subarachnoid haemorrhage surrounding a small intraparenchymal haemorrhage on the initial CT suggested the diagnosis. Thirty-nine patients underwent MRI of the cervical spine, none of which identified a cause for the haemorrhage. None of the patients re-presented to our centre during the 6 year study period. CONCLUSION: Delayed MRI following angiogram-negative subarachnoid haemorrhage has a low (1.5%) yield and is not routinely necessary. MRI may be useful to characterize the diagnosis in patients with clinical or radiological features of an underlying abnormality such as a cavernoma. Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Initial digital subtraction angiography (DSA) is negative in 5e28% of patients after spontaneous subarachnoid haemorrhage (SAH).1,2 In 10e15% of patients the distribution of SAH is perimesencephalic, and these patients usually follow a more benign course.3 However, there are other nonaneurysmal aetiologies of SAH such as cerebral amyloid angiopathy, arterial dissections, pituitary apoplexy, cerebral infarction, neoplasia, coagulopathy, and spinal vascular * Guarantor and correspondent: J. Woodfield, Neurosurgery Department, Ninewells Hospital, Dundee, DD1 9SY, UK. Tel.: +44 0 1382660111. E-mail address: julie.woodfi[email protected] (J. Woodfield).
anomalies.1,2 It is important to identify such abnormalities to risk stratify such patients and prevent re-haemorrhage.4,5 In patients where the computed tomography (CT) at presentation clearly shows perimesencephalic pattern SAH, it has been suggested that a single negative CT angiogram,6,7 or the combination of a negative CT angiogram and negative DSA1 is sufficient neuroradiological investigation to exclude a structural vascular cause. However, in patients with non-perimesencephalic angiogramnegative SAH (AN-SAH), the optimal imaging strategy remains unclear. Magnetic resonance imaging (MRI) has been advocated as a potentially useful imaging tool in patients with AN-SAH. One report of MRI within 72 h of SAH identified 10
0009-9260/$ e see front matter Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2013.11.002
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
out of 71 possible cases in which MRI identified a hitherto unknown structural cause for SAH.8 Another study of 49 patients with AN-SAH undergoing MRI within 10 days of the ictus identified no likely cause for SAH.9 The most recent and largest retrospective review of MRI within 72 h of SAH in 179 patients with normal catheter angiograms failed to identify any cases where MRI added useful diagnostic information.10,11 Only one study investigated delayed MRI performed up to 3 months after the ictus, and this identified one incidental capillary telangiectasia from 18 patients with perimesencephalic AN-SAH and questions the value of MRI in identifying the cause of AN-SAH.12 At John Radcliffe Hospital, patients with AN-SAH were often undergoing delayed MRI at 8 weeks following the ictus to exclude angiographically occult vascular abnormalities that may be more clearly identified following resolution of the acute haemorrhage. However, the utility of this strategy was unknown. The aim of the present study was to assess the diagnostic yield of delayed MRI in identifying a cause for AN-SAH.
Materials and methods All cerebral DSA examinations carried out at John Radcliffe Hospital between July 2006 and March 2012 were identified from the departmental records. Radiologists’ reports of all procedures were screened to identify those performed as the first procedure for acute spontaneous SAH. Patients undergoing DSA who presented with acute intracerebral haematomas, subdural haematomas, parenchymal contusions without SAH, or a history of trauma were excluded. Follow-up angiograms for patients previously investigated or treated for SAH were also excluded. Reports and images were reviewed to identify those in which no cause for SAH was diagnosed. These patients were classified as having AN-SAH and all imaging associated with this event was reviewed. The majority of these patients had also undergone CT angiography (CTA), which was negative in all cases. All patients presenting to John Radcliffe Hospital with SAH and a negative CTA during the study period routinely also underwent DSA. Studies performed in patients at surrounding hospitals were reviewed where possible. Follow-up imaging was assessed to identify patients who underwent delayed MRI prior to July 2012. DSA was performed within a dedicated neuroangiography suite by a neuroradiologist using a standard protocol with selective hand injections within the right and left internal carotid, external carotid and vertebral arteries. MRI was performed using either a 1.5 T system (Phillips Achieva/GE) or 3 T system (GE) using a standard protocol. This included the following sequences: sagittal T1, axial T2, T2*, diffusion-weighted imaging (DWI), and coronal fluidattenuated inversion recovery (FLAIR). All MRI brain examinations included the craniocervical junction. Spinal MRI was performed using a standard protocol with sagittal T1 and T2 imaging and axial T2 imaging. Data were analysed using Microsoft Excel and statistical analysis was performed with GraphPad InStat. The X2 test
351
Figure 1 Flow diagram of angiography suite procedures over the 6 year study period.
and Fisher’s exact test were used to assess associations between groups with a significance level set at p < 0.05.
Results A total of 3088 cerebral angiography or embolization procedures were performed between July 2006 and March 2012. Of these, 1023 were the initial investigation or treatment of acute spontaneous SAH. In 242 (23.7%) patients, the initial angiogram did not identify a vascular cause for SAH (Fig 1).
Characteristics of the study population One hundred and fifty (62%) patients with AN-SAH were male. Patients ranged from 18e87 years old with a median age of 51 years. The initial diagnosis of SAH was made by CT in 153 (63.2%) patients and lumbar puncture in 89 (36.8%) patients. Patients presented to hospital up to 20 days following the ictus. The pattern of SAH on CT was perimesencephalic in 90 patients. The Fisher grade of SAH is shown in Fig 2. The median interval for DSA was 2 days (range 0e21 days) after SAH.
Figure 2 Fisher grading of SAH.
352
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
Figure 3 Investigation of AN-SAH.
Repeat DSA Of the 242 patients with AN-SAH, repeat DSA was performed in 48 (19.8%) patients during the same admission (Fig 3). Repeat DSA occurred 1e17 days after the original DSA (median 8 days). Repeat DSA was more likely to occur in patients diagnosed using CT rather than LP (p < 0.0001)
and in patients with a higher Fisher grade at CT (p < 0.0001). Patients with perimesencephalic SAH at CT were less likely to undergo repeat DSA than those with a non-perimesencephalic pattern at CT (p ¼ 0.02). Patients undergoing repeat DSA did not differ from those not undergoing repeat DSA in terms of age, sex, or year of presentation.
Figure 4 Comparison of patients undergoing MRI head with those not undergoing MRI head.
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
353
identified. These two patients were treated endovascularly and excluded from the analysis of follow-up imaging.
MRI Of the 240 patients with AN-SAH, 131 (54.6%) underwent MRI brain (Fig 3). These were performed 0e52 weeks following the initial DSA (median 8 weeks). A comparison of patients who did and did not undergo MRI brain is displayed in Fig 4. Twenty-six (19.8%) MRI examinations were undertaken during the initial admission and within 3 weeks of the original DSA. Nineteen (73.1%) of these were carried out within 1 week of the ictus. One hundred and five (80.2%) were performed as outpatients and were delayed 4 or more weeks after the original DSA. Early MRI was more likely to occur in patients with a higher Fisher grade (p ¼ 0.006). There was no difference between those undergoing early or late MRI in terms of age, sex, or year of presentation. Underlying vascular abnormalities responsible for SAH were identified in two (1.5%) of the 131 patients undergoing MRI head. Both of these patients had associated intraparenchymal haemorrhage along with SAH, and in both cases a cavernoma was identified. Patient one was a 37-year-old man who presented with sudden onset of severe headache. CT head showed intraventricular haemorrhage in the third and fourth ventricles with a small focus of adjacent intraparenchymal haemorrhage involving the quadrigeminal plate (Fig 5). He underwent DSA on day 1 and again on day 7. Neither angiogram showed a vascular cause for the haemorrhage. MRI after 5 weeks identified a midbrain cavernoma (Fig 5). Follow-up MRI 2 years later showed no change in the lesion. The second patient was a 37-year-old woman who also presented with a sudden severe headache. Initial CT showed SAH at the left cerebello-medullary angle with haemorrhage in the fourth ventricle and a small left middle cerebellar peduncle intraparenchymal bleed (Fig 6). DSA on day 2 was negative. MRI on day 5 showed a cavernoma within the left flocculus of the cerebellum (Fig 6). She underwent repeat DSA on day 12, which was also negative. At 8 months after presentation, no other follow-up imaging had been performed. MRI of the cervical spine (C-spine) was undertaken in 39 (16.3%) of the patients with AN-SAH. No C-spine examinations showed any vascular abnormality that could account for the SAH.
Vasospasm and rebleeding Figure 5 Patient 1. (a) CT at presentation showing SAH, intraventricular and quadrigeminal plate haemorrhage. (b) T2* MRI 5 weeks later with cavernoma at this site.
In two cases, an aneurysmal cause was identified on repeat DSA that was not identified on the original DSA. In one case, a dissecting right ICA aneurysm was identified, and in another, an enlarging right MCA aneurysm was
Of the 240 patients who underwent AN-SAH, 17 (7.1%) had radiological evidence of vasospasm. This was less likely to occur in patients diagnosed via lumbar puncture rather than CT (p ¼ 0.004) and more likely to occur the higher the Fisher grade (p ¼ 0.0021). Thirty-five (14.6%) patients with AN-SAH had evidence of hydrocephalus on one or more imaging techniques. Hydrocephalus was also much less likely to occur in patients diagnosed via lumbar puncture rather than CT (p < 0.0001) and much more likely to occur
354
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
Figure 6 Patient 2. (a) CT at presentation with left posterior fossa SAH. (b) T1, (c) T2, and (d) T2* MRI showing a lesion typical for a cavernoma at this site.
with a higher Fisher grade (p < 0.0001). There was no difference in the likelihood of vasospasm or hydrocephalus on imaging between patients with perimesencephalic SAH and non-perimesencephalic SAH. No patient with AN-SAH undergoing initial angiography between July 2006 and March 2012 re-presented to the centre with SAH or any other haemorrhagic event up to July 2012, and all were still alive at July 2012.
Perimesencephalic SAH Of the 90 patients with perimesencephalic pattern SAH at CT, 20 underwent repeat DSA, none of which showed a vascular cause for the SAH. Fifty-seven patients with perimesencephalic AN-SAH underwent MRI head and 17 underwent MRI C-spine. None of the MRI examinations identified any likely cause for the SAH. Ten patients with perimesencephalic AN-SAH had findings of radiological
vasospasm, and 22 showed increased ventricular size on imaging.
Lumbar-puncture-positive SAH Of the 89 patients with AN-SAH diagnosed via lumbar puncture, three underwent repeat DSA, 40 underwent MRI head, and 15 underwent MRI C-spine. None of these investigations revealed a cause for the SAH. One patient had evidence of radiological vasospasm and none had hydrocephalus on any imaging.
Discussion Of 240 patients with AN-SAH, 131 (54.6%) underwent MRI brain, and the majority (105, 80.2%) were delayed four or more weeks after the initial DSA. Two of the 131 patients
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
(1.5%) had angiographically occult lesions responsible for the initial haemorrhage identified on MRI brain. In both cases, a focus of intraparenchymal haemorrhage suggestive of the presence of a cavernoma was present on the initial CT. MRI may be useful in confirming the diagnosis of cavernoma where this is suspected from the presentation CT, but is unlikely to identify an occult cause of AN-SAH in patients where the initial imaging or clinical presentation is not suggestive of an underlying angiographically occult vascular lesion. In the 97 patients with perimesencephalic SAH or SAH diagnosed by lumbar puncture, MRI did not provide any additional useful diagnostic information. These results are similar to those of Maslehaty et al.10,11 who reported 179 cases of AN-SAH in which early MRI within 48 h failed to identify any cause for SAH. Topcuoglu et al.9 also failed to identify any cause for AN-SAH on 49 MRI examinations performed within 48 h of the ictus. Wijdicks et al.12 did identify a capillary telangiectasia at MRI in a patient with AN-SAH. However, it is likely that this lesion was incidental and not responsible for the SAH.12 Rogg et al.8 found multiple potential causes of AN-SAH with early MRI; however, many of the causes identified could have been identified clinically or on initial imaging with CT or angiography. The results of the present study add to the growing body of evidence against performing MRI routinely in AN-SAH. Cavernomas occur in 0.4e0.6% of the population, and approximately 36% of these will present with haemorrhage.13 Cavernomas usually result in intraparenchymal haemorrhage, although isolated SAH has also been reported.14 As very few cases of AN-SAH will be attributable to cavernomas, the series would need to be large in order to identify them. It is possible that none of the previous series included sufficient numbers of patients to identify a case of AN-SAH caused by a cavernoma. It is also possible that strict exclusion criteria in other studies excluded patients with any intraparenchymal haemorrhage on CT at presentation. This study population is representative of patients presenting to a UK neurosurgical department for investigation and treatment of SAH. Patients were included in this study if they had a history strongly suggestive of SAH, a negative CT and the cerebrospinal fluid (CSF) spectroscopy showed an oxyhaemoglobin peak sufficient to impair the detection of bilirubin. Patients were also included if they had a suddenonset severe headache and the CT showed SAH along with intraparenchymal or subdural haemorrhage. However, patients without any subarachnoid component to the haemorrhage were excluded. There was no exclusion for patients presenting late. Patients in this study presented up to 20 days post ictus, and DSA was performed up to 21 days post ictus. Reported rates of LP diagnosed SAH vary from 9e65% depending on the timing of CT post ictus.1 The relatively high proportion (36.8%) of patients diagnosed by lumbar puncture and the inclusion of all patients being investigated for SAH may account for the 23.7% rate of AN-SAH. This study is limited by its retrospective analysis. The nature of follow-up imaging was determined by the requesting neurosurgeon and the neuroradiologist undertaking the request. Some patients did not attend their follow-up
355
imaging, and in some instances, imaging was delayed several times due to patient cancellation and re-booking. It is possible some patients underwent delayed MRI at other surrounding centres. Only 39 of the 240 patients underwent MRI of the whole C-spine. This number is similar to the largest reported series of patients undergoing MRI C-spine after AN-SAH, in which one spinal arteriovenous malformation was found in 41 patients.8 This patient had reported neck and back pain and had experienced four similar events in the preceding 8 years.8 In the present 39 patients without any symptoms localizing the source of the haemorrhage to the spine, none of the MRI C-spine examinations identified any occult cause for AN-SAH. The decision to undertake MRI C-spine or whole spine following AN-SAH would need to be made on an individual patient basis, bearing in mind the lack of evidence for or against this strategy. The most widely studied subgroup of patients with ANSAH is those with a perimesencephalic distribution of haemorrhage on CT. In patients with the strictly defined perimesencephalic pattern of SAH on CT3,15 and a negative DSA, vascular causes for the haemorrhage have not been identified.1,16 Patients with perimesencephalic SAH rarely require intervention for hydrocephalus or vasospasm,17 and outcomes after up to 6 years of follow-up are favourable.18 Of the present patients with a perimesencephalic pattern of SAH, 24.4% had hydrocephalus and 11.1% had radiological vasospasm. Repeat DSA and delayed MRI did not identify any anatomical or vascular lesion responsible for the haemorrhage. All of these findings are consistent with the current position that perimesencephalic SAH is a more benign condition that does not require follow-up imaging. Due to the rarity of angiographically occult vascular abnormalities causing AN-SAH, large numbers of patients are required to determine the most appropriate imaging strategy for their detection. The present study was a large, retrospective study of imaging follow-up of patients with AN-SAH. Performing routine delayed MRI in patients with AN-SAH has a low diagnostic yield and is probably unnecessary. For patients in whom a parenchymal lesion is suspected on initial imaging, MRI brain can provide the diagnosis. In patients with signs or symptoms suggestive of spinal vascular anomalies, MRI of the whole spine may provide useful diagnostic information.
References 1. Schwartz TH, Solomon RA. Perimesencephalic nonaneurysmal subarachnoid hemorrhage: review of the literature. Neurosurgery 1996;39:433e40. 2. Rinkel GJ, van Gijn J, Wijdicks EF. Subarachnoid hemorrhage without detectable aneurysm. A review of the causes. Stroke 1993;24:1403e9. 3. van Gijn J, van Dongen KJ, Vermeulen M, et al. Perimesencephalic hemorrhage: a nonaneurysmal and benign form of subarachnoid hemorrhage. Neurology 1985;35:493e7. 4. Ruigrok YM, Rinkel GJ, Van Gijn J. CT patterns and long-term outcome in patients with an aneurysmal type of subarachnoid hemorrhage and repeatedly negative angiograms. Cerebrovasc Dis 2002;14:221e7. 5. Boswell S, Thorell W, Gogela S, et al. Angiogram-negative subarachnoid hemorrhage: outcomes data and review of the literature. J Stroke Cerebrovasc Dis 2013;22:750e7.
356
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
6. Cruz JP, Sarma D, Noel de Tilly L. Perimesencephalic subarachnoid hemorrhage: when to stop imaging? Emerg Radiol 2011;18:197e202. 7. Ruigrok YM, Rinkel GJ, Buskens E, et al. Perimesencephalic hemorrhage and CT angiography: a decision analysis. Stroke 2000;31:2976e83. 8. Rogg JM, Smeaton S, Doberstein C, et al. Assessment of the value of MR imaging for examining patients with angiographically negative subarachnoid hemorrhage. AJR Am J Roentgenol 1999;172:201e6. 9. Topcuoglu MA, Ogilvy CS, Carter BS, et al. Subarachnoid hemorrhage without evident cause on initial angiography studies: diagnostic yield of subsequent angiography and other neuroimaging tests. J Neurosurg 2003;98:1235e40. 10. Maslehaty H, Petridis AK, Barth H, et al. Diagnostic value of magnetic resonance imaging in perimesencephalic and nonperimesencephalic subarachnoid hemorrhage of unknown origin. J Neurosurg 2011;114:1003e7. 11. Maslehaty H, Petridis AK, Barth H, et al. Does magnetic resonance imaging produce further benefit for detecting a bleeding source in subarachnoid hemorrhage of unknown origin? Acta Neurochir Suppl 2011;112:107e9.
12. Wijdicks EF, Schievink WI, Miller GM. MR imaging in pretruncal nonaneurysmal subarachnoid hemorrhage: is it worthwhile? Stroke 1998;29:2514e6. 13. Gross BA, Lin N, Du R, et al. The natural history of intracranial cavernous malformations. Neurosurg Focus 2011;30:E24. 14. Yaghi S, Oomman S, Keyrouz SG. Non-aneurysmal perimesencephalic subarachnoid hemorrhage caused by a cavernous angioma. Neurocrit Care 2011;14:84e5. 15. Rinkel GJ, Wijdicks EF, Vermeulen M, et al. Nonaneurysmal perimesencephalic subarachnoid hemorrhage: CT and MR patterns that differ from aneurysmal rupture. AJNR Am J Neuroradiol 1991;12:829e34. 16. Rinkel GJ, Wijdicks EF, Hasan D, et al. Outcome in patients with subarachnoid haemorrhage and negative angiography according to pattern of haemorrhage on computed tomography. Lancet 1991;338:964e8. 17. Rinkel GJ, Wijdicks EF, Vermeulen M, et al. The clinical course of perimesencephalic nonaneurysmal subarachnoid hemorrhage. Ann Neurol 1991;29:463e8. 18. Brilstra EH, Hop JW, Rinkel GJ. Quality of life after perimesencephalic haemorrhage. J Neurol Neurosurg Psychiatry 1997;63:382e4.
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Value of delayed MRI in angiogram-negative subarachnoid haemorrhage J. Woodfield a, *, N. Rane b, S. Cudlip a, J.V. Byrne b a b
Department of Neurosurgery, John Radcliffe Hospital, Oxford, UK Department of Neuroradiology, John Radcliffe Hospital, Oxford, UK
article in formation Article history: Received 7 October 2013 Accepted 5 November 2013
AIM: To assess the efficacy of delayed magnetic resonance imaging (MRI) in identifying a structural cause for angiogram-negative subarachnoid haemorrhage. MATERIALS AND METHODS: All patients presenting with spontaneous subarachnoid haemorrhage who had negative computed tomography (CT) angiography and catheter angiography between 2006 and 2012 were reviewed. RESULTS: During the 6 year period, 1023 angiograms were performed for a new presentation of subarachnoid haemorrhage. Of these, 242 (23.7%) did not show a cause for the haemorrhage. A second catheter angiogram was performed in 48 patients, and aneurysms were identified in two patients. Of the remaining 240 patients, 131 underwent a subsequent MRI brain. One hundred and five (80.2%) MRI examinations were performed 4 or more weeks after angiography. In two patients, cavernomas were identified as the likely bleeding source. In both patients, the pattern of subarachnoid haemorrhage surrounding a small intraparenchymal haemorrhage on the initial CT suggested the diagnosis. Thirty-nine patients underwent MRI of the cervical spine, none of which identified a cause for the haemorrhage. None of the patients re-presented to our centre during the 6 year study period. CONCLUSION: Delayed MRI following angiogram-negative subarachnoid haemorrhage has a low (1.5%) yield and is not routinely necessary. MRI may be useful to characterize the diagnosis in patients with clinical or radiological features of an underlying abnormality such as a cavernoma. Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Initial digital subtraction angiography (DSA) is negative in 5e28% of patients after spontaneous subarachnoid haemorrhage (SAH).1,2 In 10e15% of patients the distribution of SAH is perimesencephalic, and these patients usually follow a more benign course.3 However, there are other nonaneurysmal aetiologies of SAH such as cerebral amyloid angiopathy, arterial dissections, pituitary apoplexy, cerebral infarction, neoplasia, coagulopathy, and spinal vascular * Guarantor and correspondent: J. Woodfield, Neurosurgery Department, Ninewells Hospital, Dundee, DD1 9SY, UK. Tel.: +44 0 1382660111. E-mail address: julie.woodfi[email protected] (J. Woodfield).
anomalies.1,2 It is important to identify such abnormalities to risk stratify such patients and prevent re-haemorrhage.4,5 In patients where the computed tomography (CT) at presentation clearly shows perimesencephalic pattern SAH, it has been suggested that a single negative CT angiogram,6,7 or the combination of a negative CT angiogram and negative DSA1 is sufficient neuroradiological investigation to exclude a structural vascular cause. However, in patients with non-perimesencephalic angiogramnegative SAH (AN-SAH), the optimal imaging strategy remains unclear. Magnetic resonance imaging (MRI) has been advocated as a potentially useful imaging tool in patients with AN-SAH. One report of MRI within 72 h of SAH identified 10
0009-9260/$ e see front matter Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2013.11.002
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
out of 71 possible cases in which MRI identified a hitherto unknown structural cause for SAH.8 Another study of 49 patients with AN-SAH undergoing MRI within 10 days of the ictus identified no likely cause for SAH.9 The most recent and largest retrospective review of MRI within 72 h of SAH in 179 patients with normal catheter angiograms failed to identify any cases where MRI added useful diagnostic information.10,11 Only one study investigated delayed MRI performed up to 3 months after the ictus, and this identified one incidental capillary telangiectasia from 18 patients with perimesencephalic AN-SAH and questions the value of MRI in identifying the cause of AN-SAH.12 At John Radcliffe Hospital, patients with AN-SAH were often undergoing delayed MRI at 8 weeks following the ictus to exclude angiographically occult vascular abnormalities that may be more clearly identified following resolution of the acute haemorrhage. However, the utility of this strategy was unknown. The aim of the present study was to assess the diagnostic yield of delayed MRI in identifying a cause for AN-SAH.
Materials and methods All cerebral DSA examinations carried out at John Radcliffe Hospital between July 2006 and March 2012 were identified from the departmental records. Radiologists’ reports of all procedures were screened to identify those performed as the first procedure for acute spontaneous SAH. Patients undergoing DSA who presented with acute intracerebral haematomas, subdural haematomas, parenchymal contusions without SAH, or a history of trauma were excluded. Follow-up angiograms for patients previously investigated or treated for SAH were also excluded. Reports and images were reviewed to identify those in which no cause for SAH was diagnosed. These patients were classified as having AN-SAH and all imaging associated with this event was reviewed. The majority of these patients had also undergone CT angiography (CTA), which was negative in all cases. All patients presenting to John Radcliffe Hospital with SAH and a negative CTA during the study period routinely also underwent DSA. Studies performed in patients at surrounding hospitals were reviewed where possible. Follow-up imaging was assessed to identify patients who underwent delayed MRI prior to July 2012. DSA was performed within a dedicated neuroangiography suite by a neuroradiologist using a standard protocol with selective hand injections within the right and left internal carotid, external carotid and vertebral arteries. MRI was performed using either a 1.5 T system (Phillips Achieva/GE) or 3 T system (GE) using a standard protocol. This included the following sequences: sagittal T1, axial T2, T2*, diffusion-weighted imaging (DWI), and coronal fluidattenuated inversion recovery (FLAIR). All MRI brain examinations included the craniocervical junction. Spinal MRI was performed using a standard protocol with sagittal T1 and T2 imaging and axial T2 imaging. Data were analysed using Microsoft Excel and statistical analysis was performed with GraphPad InStat. The X2 test
351
Figure 1 Flow diagram of angiography suite procedures over the 6 year study period.
and Fisher’s exact test were used to assess associations between groups with a significance level set at p < 0.05.
Results A total of 3088 cerebral angiography or embolization procedures were performed between July 2006 and March 2012. Of these, 1023 were the initial investigation or treatment of acute spontaneous SAH. In 242 (23.7%) patients, the initial angiogram did not identify a vascular cause for SAH (Fig 1).
Characteristics of the study population One hundred and fifty (62%) patients with AN-SAH were male. Patients ranged from 18e87 years old with a median age of 51 years. The initial diagnosis of SAH was made by CT in 153 (63.2%) patients and lumbar puncture in 89 (36.8%) patients. Patients presented to hospital up to 20 days following the ictus. The pattern of SAH on CT was perimesencephalic in 90 patients. The Fisher grade of SAH is shown in Fig 2. The median interval for DSA was 2 days (range 0e21 days) after SAH.
Figure 2 Fisher grading of SAH.
352
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
Figure 3 Investigation of AN-SAH.
Repeat DSA Of the 242 patients with AN-SAH, repeat DSA was performed in 48 (19.8%) patients during the same admission (Fig 3). Repeat DSA occurred 1e17 days after the original DSA (median 8 days). Repeat DSA was more likely to occur in patients diagnosed using CT rather than LP (p < 0.0001)
and in patients with a higher Fisher grade at CT (p < 0.0001). Patients with perimesencephalic SAH at CT were less likely to undergo repeat DSA than those with a non-perimesencephalic pattern at CT (p ¼ 0.02). Patients undergoing repeat DSA did not differ from those not undergoing repeat DSA in terms of age, sex, or year of presentation.
Figure 4 Comparison of patients undergoing MRI head with those not undergoing MRI head.
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
353
identified. These two patients were treated endovascularly and excluded from the analysis of follow-up imaging.
MRI Of the 240 patients with AN-SAH, 131 (54.6%) underwent MRI brain (Fig 3). These were performed 0e52 weeks following the initial DSA (median 8 weeks). A comparison of patients who did and did not undergo MRI brain is displayed in Fig 4. Twenty-six (19.8%) MRI examinations were undertaken during the initial admission and within 3 weeks of the original DSA. Nineteen (73.1%) of these were carried out within 1 week of the ictus. One hundred and five (80.2%) were performed as outpatients and were delayed 4 or more weeks after the original DSA. Early MRI was more likely to occur in patients with a higher Fisher grade (p ¼ 0.006). There was no difference between those undergoing early or late MRI in terms of age, sex, or year of presentation. Underlying vascular abnormalities responsible for SAH were identified in two (1.5%) of the 131 patients undergoing MRI head. Both of these patients had associated intraparenchymal haemorrhage along with SAH, and in both cases a cavernoma was identified. Patient one was a 37-year-old man who presented with sudden onset of severe headache. CT head showed intraventricular haemorrhage in the third and fourth ventricles with a small focus of adjacent intraparenchymal haemorrhage involving the quadrigeminal plate (Fig 5). He underwent DSA on day 1 and again on day 7. Neither angiogram showed a vascular cause for the haemorrhage. MRI after 5 weeks identified a midbrain cavernoma (Fig 5). Follow-up MRI 2 years later showed no change in the lesion. The second patient was a 37-year-old woman who also presented with a sudden severe headache. Initial CT showed SAH at the left cerebello-medullary angle with haemorrhage in the fourth ventricle and a small left middle cerebellar peduncle intraparenchymal bleed (Fig 6). DSA on day 2 was negative. MRI on day 5 showed a cavernoma within the left flocculus of the cerebellum (Fig 6). She underwent repeat DSA on day 12, which was also negative. At 8 months after presentation, no other follow-up imaging had been performed. MRI of the cervical spine (C-spine) was undertaken in 39 (16.3%) of the patients with AN-SAH. No C-spine examinations showed any vascular abnormality that could account for the SAH.
Vasospasm and rebleeding Figure 5 Patient 1. (a) CT at presentation showing SAH, intraventricular and quadrigeminal plate haemorrhage. (b) T2* MRI 5 weeks later with cavernoma at this site.
In two cases, an aneurysmal cause was identified on repeat DSA that was not identified on the original DSA. In one case, a dissecting right ICA aneurysm was identified, and in another, an enlarging right MCA aneurysm was
Of the 240 patients who underwent AN-SAH, 17 (7.1%) had radiological evidence of vasospasm. This was less likely to occur in patients diagnosed via lumbar puncture rather than CT (p ¼ 0.004) and more likely to occur the higher the Fisher grade (p ¼ 0.0021). Thirty-five (14.6%) patients with AN-SAH had evidence of hydrocephalus on one or more imaging techniques. Hydrocephalus was also much less likely to occur in patients diagnosed via lumbar puncture rather than CT (p < 0.0001) and much more likely to occur
354
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
Figure 6 Patient 2. (a) CT at presentation with left posterior fossa SAH. (b) T1, (c) T2, and (d) T2* MRI showing a lesion typical for a cavernoma at this site.
with a higher Fisher grade (p < 0.0001). There was no difference in the likelihood of vasospasm or hydrocephalus on imaging between patients with perimesencephalic SAH and non-perimesencephalic SAH. No patient with AN-SAH undergoing initial angiography between July 2006 and March 2012 re-presented to the centre with SAH or any other haemorrhagic event up to July 2012, and all were still alive at July 2012.
Perimesencephalic SAH Of the 90 patients with perimesencephalic pattern SAH at CT, 20 underwent repeat DSA, none of which showed a vascular cause for the SAH. Fifty-seven patients with perimesencephalic AN-SAH underwent MRI head and 17 underwent MRI C-spine. None of the MRI examinations identified any likely cause for the SAH. Ten patients with perimesencephalic AN-SAH had findings of radiological
vasospasm, and 22 showed increased ventricular size on imaging.
Lumbar-puncture-positive SAH Of the 89 patients with AN-SAH diagnosed via lumbar puncture, three underwent repeat DSA, 40 underwent MRI head, and 15 underwent MRI C-spine. None of these investigations revealed a cause for the SAH. One patient had evidence of radiological vasospasm and none had hydrocephalus on any imaging.
Discussion Of 240 patients with AN-SAH, 131 (54.6%) underwent MRI brain, and the majority (105, 80.2%) were delayed four or more weeks after the initial DSA. Two of the 131 patients
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
(1.5%) had angiographically occult lesions responsible for the initial haemorrhage identified on MRI brain. In both cases, a focus of intraparenchymal haemorrhage suggestive of the presence of a cavernoma was present on the initial CT. MRI may be useful in confirming the diagnosis of cavernoma where this is suspected from the presentation CT, but is unlikely to identify an occult cause of AN-SAH in patients where the initial imaging or clinical presentation is not suggestive of an underlying angiographically occult vascular lesion. In the 97 patients with perimesencephalic SAH or SAH diagnosed by lumbar puncture, MRI did not provide any additional useful diagnostic information. These results are similar to those of Maslehaty et al.10,11 who reported 179 cases of AN-SAH in which early MRI within 48 h failed to identify any cause for SAH. Topcuoglu et al.9 also failed to identify any cause for AN-SAH on 49 MRI examinations performed within 48 h of the ictus. Wijdicks et al.12 did identify a capillary telangiectasia at MRI in a patient with AN-SAH. However, it is likely that this lesion was incidental and not responsible for the SAH.12 Rogg et al.8 found multiple potential causes of AN-SAH with early MRI; however, many of the causes identified could have been identified clinically or on initial imaging with CT or angiography. The results of the present study add to the growing body of evidence against performing MRI routinely in AN-SAH. Cavernomas occur in 0.4e0.6% of the population, and approximately 36% of these will present with haemorrhage.13 Cavernomas usually result in intraparenchymal haemorrhage, although isolated SAH has also been reported.14 As very few cases of AN-SAH will be attributable to cavernomas, the series would need to be large in order to identify them. It is possible that none of the previous series included sufficient numbers of patients to identify a case of AN-SAH caused by a cavernoma. It is also possible that strict exclusion criteria in other studies excluded patients with any intraparenchymal haemorrhage on CT at presentation. This study population is representative of patients presenting to a UK neurosurgical department for investigation and treatment of SAH. Patients were included in this study if they had a history strongly suggestive of SAH, a negative CT and the cerebrospinal fluid (CSF) spectroscopy showed an oxyhaemoglobin peak sufficient to impair the detection of bilirubin. Patients were also included if they had a suddenonset severe headache and the CT showed SAH along with intraparenchymal or subdural haemorrhage. However, patients without any subarachnoid component to the haemorrhage were excluded. There was no exclusion for patients presenting late. Patients in this study presented up to 20 days post ictus, and DSA was performed up to 21 days post ictus. Reported rates of LP diagnosed SAH vary from 9e65% depending on the timing of CT post ictus.1 The relatively high proportion (36.8%) of patients diagnosed by lumbar puncture and the inclusion of all patients being investigated for SAH may account for the 23.7% rate of AN-SAH. This study is limited by its retrospective analysis. The nature of follow-up imaging was determined by the requesting neurosurgeon and the neuroradiologist undertaking the request. Some patients did not attend their follow-up
355
imaging, and in some instances, imaging was delayed several times due to patient cancellation and re-booking. It is possible some patients underwent delayed MRI at other surrounding centres. Only 39 of the 240 patients underwent MRI of the whole C-spine. This number is similar to the largest reported series of patients undergoing MRI C-spine after AN-SAH, in which one spinal arteriovenous malformation was found in 41 patients.8 This patient had reported neck and back pain and had experienced four similar events in the preceding 8 years.8 In the present 39 patients without any symptoms localizing the source of the haemorrhage to the spine, none of the MRI C-spine examinations identified any occult cause for AN-SAH. The decision to undertake MRI C-spine or whole spine following AN-SAH would need to be made on an individual patient basis, bearing in mind the lack of evidence for or against this strategy. The most widely studied subgroup of patients with ANSAH is those with a perimesencephalic distribution of haemorrhage on CT. In patients with the strictly defined perimesencephalic pattern of SAH on CT3,15 and a negative DSA, vascular causes for the haemorrhage have not been identified.1,16 Patients with perimesencephalic SAH rarely require intervention for hydrocephalus or vasospasm,17 and outcomes after up to 6 years of follow-up are favourable.18 Of the present patients with a perimesencephalic pattern of SAH, 24.4% had hydrocephalus and 11.1% had radiological vasospasm. Repeat DSA and delayed MRI did not identify any anatomical or vascular lesion responsible for the haemorrhage. All of these findings are consistent with the current position that perimesencephalic SAH is a more benign condition that does not require follow-up imaging. Due to the rarity of angiographically occult vascular abnormalities causing AN-SAH, large numbers of patients are required to determine the most appropriate imaging strategy for their detection. The present study was a large, retrospective study of imaging follow-up of patients with AN-SAH. Performing routine delayed MRI in patients with AN-SAH has a low diagnostic yield and is probably unnecessary. For patients in whom a parenchymal lesion is suspected on initial imaging, MRI brain can provide the diagnosis. In patients with signs or symptoms suggestive of spinal vascular anomalies, MRI of the whole spine may provide useful diagnostic information.
References 1. Schwartz TH, Solomon RA. Perimesencephalic nonaneurysmal subarachnoid hemorrhage: review of the literature. Neurosurgery 1996;39:433e40. 2. Rinkel GJ, van Gijn J, Wijdicks EF. Subarachnoid hemorrhage without detectable aneurysm. A review of the causes. Stroke 1993;24:1403e9. 3. van Gijn J, van Dongen KJ, Vermeulen M, et al. Perimesencephalic hemorrhage: a nonaneurysmal and benign form of subarachnoid hemorrhage. Neurology 1985;35:493e7. 4. Ruigrok YM, Rinkel GJ, Van Gijn J. CT patterns and long-term outcome in patients with an aneurysmal type of subarachnoid hemorrhage and repeatedly negative angiograms. Cerebrovasc Dis 2002;14:221e7. 5. Boswell S, Thorell W, Gogela S, et al. Angiogram-negative subarachnoid hemorrhage: outcomes data and review of the literature. J Stroke Cerebrovasc Dis 2013;22:750e7.
356
J. Woodfield et al. / Clinical Radiology 69 (2014) 350e356
6. Cruz JP, Sarma D, Noel de Tilly L. Perimesencephalic subarachnoid hemorrhage: when to stop imaging? Emerg Radiol 2011;18:197e202. 7. Ruigrok YM, Rinkel GJ, Buskens E, et al. Perimesencephalic hemorrhage and CT angiography: a decision analysis. Stroke 2000;31:2976e83. 8. Rogg JM, Smeaton S, Doberstein C, et al. Assessment of the value of MR imaging for examining patients with angiographically negative subarachnoid hemorrhage. AJR Am J Roentgenol 1999;172:201e6. 9. Topcuoglu MA, Ogilvy CS, Carter BS, et al. Subarachnoid hemorrhage without evident cause on initial angiography studies: diagnostic yield of subsequent angiography and other neuroimaging tests. J Neurosurg 2003;98:1235e40. 10. Maslehaty H, Petridis AK, Barth H, et al. Diagnostic value of magnetic resonance imaging in perimesencephalic and nonperimesencephalic subarachnoid hemorrhage of unknown origin. J Neurosurg 2011;114:1003e7. 11. Maslehaty H, Petridis AK, Barth H, et al. Does magnetic resonance imaging produce further benefit for detecting a bleeding source in subarachnoid hemorrhage of unknown origin? Acta Neurochir Suppl 2011;112:107e9.
12. Wijdicks EF, Schievink WI, Miller GM. MR imaging in pretruncal nonaneurysmal subarachnoid hemorrhage: is it worthwhile? Stroke 1998;29:2514e6. 13. Gross BA, Lin N, Du R, et al. The natural history of intracranial cavernous malformations. Neurosurg Focus 2011;30:E24. 14. Yaghi S, Oomman S, Keyrouz SG. Non-aneurysmal perimesencephalic subarachnoid hemorrhage caused by a cavernous angioma. Neurocrit Care 2011;14:84e5. 15. Rinkel GJ, Wijdicks EF, Vermeulen M, et al. Nonaneurysmal perimesencephalic subarachnoid hemorrhage: CT and MR patterns that differ from aneurysmal rupture. AJNR Am J Neuroradiol 1991;12:829e34. 16. Rinkel GJ, Wijdicks EF, Hasan D, et al. Outcome in patients with subarachnoid haemorrhage and negative angiography according to pattern of haemorrhage on computed tomography. Lancet 1991;338:964e8. 17. Rinkel GJ, Wijdicks EF, Vermeulen M, et al. The clinical course of perimesencephalic nonaneurysmal subarachnoid hemorrhage. Ann Neurol 1991;29:463e8. 18. Brilstra EH, Hop JW, Rinkel GJ. Quality of life after perimesencephalic haemorrhage. J Neurol Neurosurg Psychiatry 1997;63:382e4.
