Journal of Medical Imaging and Radiation Oncology 58 (2014) 408–414 bs_bs_banner
RADIOLO GY —O R I G I N A L A RT I C L E
The location of origin of spontaneous extracranial internal carotid artery dissection is adjacent to the skull base Jonathon Downer,1 Mahendra Nadarajah,2 Eliza Briggs,1 Peter Wrigley1 and William McAuliffe1,3 1 Neurological Intervention and Imaging Service of Western Australia, Sir Charles Gairdner Hospital, and 3Neurological Intervention and Imaging Service of Western Australia, Royal Perth Hospital, Perth, Western Australia, Australia, and 2National Neuroscience Institute, Singapore
J Downer MA, MRCP, FRCR; M Nadarajah B.Sci, FRCR; E Briggs BSci, MBBS; P Wrigley MBBS; W McAuliffe FRANZCR. Correspondence Dr William McAuliffe, Neurological Intervention and Imaging Service of Western Australia, Sir Charles Gairdner Hospital, Verdun St, Nedlands, Perth, WA 6009, Australia. Email: [email protected] Conflict of interest: None. Submitted 7 February 2013; accepted 16 January 2014. doi:10.1111/1754-9485.12170
Abstract Introduction: The traditional view is that spontaneous extracranial internal carotid artery (ICA) dissection (CAD) extends cranially from an intimal tear located just beyond the carotid bulb. This paper demonstrates that CAD originates in and primarily involves a more distal segment of the artery. Methods: A retrospective study of 54 dissected ICAs in 50 consecutive patients with spontaneous or traumatic CAD was undertaken. The site of the dissection, presence of ICA redundancy, rate of acute or delayed ischaemic stroke and vessel remodelling were determined. Results: Of the 51 dissections that occurred spontaneously or after indirect trauma, 25/51 (49.0%) were solely in the distal third of the artery, and 49/51 (96.1%) involved the distal two-thirds. Only 2/51 (3.9%) originated in the proximal third. ICA redundancy was seen in 27/36 (75%) of patients with spontaneous CAD, compared with only 1/11 (9.1%) of those with CAD due to indirect trauma (P = 0.0002). Acute stroke occurred in 10/12 (83.3%) of patients with ICA occlusion secondary to CAD and in 14/38 (36.8%) with non-occlusive CAD (P = 0.0074). Where follow-up was available, only 2/32 (6.3%) patients had a stroke after diagnosis, and 19/33 (57.6%) ICAs recanalised or remodelled. Conclusion: CAD occurring spontaneously or due to indirect trauma most frequently involves the distal extracranial ICA. Spontaneous CAD is associated with vessel redundancy, and the risk of acute stroke is greatest with occlusive CAD. The prognosis is good with treatment, with a low rate of recurrent stroke and a high rate of vessel remodelling. Key words: adult neuroimaging; angiography; magnetic resonance imaging; neurointerventional radiology; neuroradiology; vascular imaging.
Introduction Cervical artery dissection is an important cause of ischaemic stroke, accounting for almost 20% of cases in young adults.1 Cervical artery dissection may occur spontaneously or following trauma and is defined pathologically by the presence of a haematoma in the vessel wall.2 The traditional view has been that spontaneous dissection of the extracranial internal carotid artery (ICA) extends cranially from an intimal tear located just beyond the carotid bulb, approximately 2 cm from the vessel origin.3,4 This is primarily based on historical 408
pathology- and catheter angiography-based studies involving small numbers of patients.4,5 Our experience is that extracranial ICA dissection (CAD) originates in and is frequently confined to the distal cervical segment of the vessel, 1–2 cm below the skull base. The primary objective of this paper is to test that hypothesis. We also set out to assess the rate of acute ischaemic stroke at presentation and after diagnosis, to determine whether the degree of vessel narrowing is an important factor in determining the risk of stroke, to determine the incidence of cervical ICA loops or kinks in CAD and to find the rate and degree of vessel remodelling over time. © 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
Methods Study design A retrospective study of consecutive patients diagnosed with CAD on imaging was performed over a 5-year period from 2004 to 2009 at the three major teaching hospitals of the Perth region, with a catchment population of more than 2.4 million people.
Patient cohort Patients were identified by searching the state-wide Radiology Information System (RIS). Report text and clinical history fields were trawled using the search term ‘dissection’, generating a list of 331 patients. Review of the imaging reports of these cases identified 93 patients who were diagnosed with CAD. Diagnostic imaging was then reviewed with reference to specific inclusion criteria to arrive at a final cohort of 50 patients.
Inclusion criteria Patients with an imaging diagnosis of CAD were included, defined by demonstration of intramural haematoma, intimal dissection flap, vessel wall irregularity or pseudoaneurysm on CT/CTA and/or MRI/MRA datasets. Those patients with iatrogenic CAD or aortic dissection extending to the internal carotid artery were excluded. The final cohort therefore included cases of both spontaneous and traumatic CAD.
Data collection Clinical data was collected from the RIS record or the medical notes where this was incomplete. The mechanism of injury was categorised as spontaneous where there was no history of preceding trauma. When patients presented after traumatic injury, this was defined as indirect trauma (e.g. high-speed motor vehicle accident) or direct trauma (e.g. hanging). Imaging review was performed by an experienced consultant neuroradiologist and fellow. The longitudinal extent and location of the intramural haematoma or false lumen were determined on cross-sectional imaging at presentation. The site of maximum mural haematoma thickness was assessed in axial images. For the purpose of this analysis, the extracranial ICA was divided longitudinally into thirds over its length. ICA redundancy was assessed using the categorisation described by Barbour,6 who defined vessel loops, kinks and coils. Loops are C- or S-shaped deformities with two turns in the vessel with angles less than or equal to 90 degrees; kinks produce a sharp bend with an angle of less than or equal to 90 degrees; and coils are 360-degree turns in the vessel. A qualitative assessment of the anatomical outcome of the vessel was made on final follow-up imaging. © 2014 The Royal Australian and New Zealand College of Radiologists
Follow-up imaging Follow-up imaging was not available for 16 patients, of whom 7 were lost to follow up, 3 died, 3 returned home for follow up in another state or country, and 3 lived in remote rural areas.
Statistical analysis Statistical analysis was performed using GraphPad QuickCalcs (GraphPad Software, San Diego, CA, USA). P values were calculated using a two-tailed Fisher’s exact test.
Results Demographics Fifty patients with 54 dissected ICAs were included: 46 patients with unilateral CAD and 4 patients with bilateral CAD. The mean age of the patients was 44.8 years with a range of 19 to 68 years. Thirty-one of 50 (62%) were male, and 19 of 50 (38%) were female. MRI/MR angiography was performed in 25/50 patients (50%) and CT angiography in 41/50 (82%) as part of the imaging work-up at presentation.
Mechanism of arterial injury Thirty-six of 50 patients (72%) had spontaneous CAD, 11/50 (22%) had CAD due to indirect trauma, and 3/50 (6%) had CAD due to direct traumatic injury.
Presenting symptoms Twenty-four of 50 patients(48%) presented with an ischaemic stroke and one patient (2%) with amaurosis fugax. Four of 50 (8%) had Horner’s syndrome, and one patient (2%) had hypoglossal nerve palsy due to local effects of the dissection.
Stroke risk at presentation and vessel patency Ten of 12 patients (83.3%) presenting with CAD resulting in ICA occlusion had suffered an acute ischaemic stroke at diagnosis (Fig. 1), compared with 14/38 patients (36.8%) with non-occlusive CAD. This difference was statistically significant (P = 0.0074). Five of 18 patients with CAD (27.8%) with a less than 50% stenosis presented with acute ischaemic stroke, compared with 9/20 (45%) among those with a greater than 50% stenosis. This difference was not statistically significant (P = 0.3276).
Origin of dissection defined by location of maximum mural haematoma thickness Spontaneous CAD Thirty-six of 38 vessels (94.7%) had maximum mural haematoma thickness in the distal third of the ICA 409
J Downer et al.
(a)
(b)
(c)
(d)
Fig. 1. Forty-five-year-old female with no history of trauma who presented with left-sided weakness and right neck pain. (a) Axial CT angiogram. Thickened mural haematoma, maximal 1 cm below the entry of the right internal carotid artery (ICA) into the skull base, is clearly demonstrated (two white arrows), with occlusive dissection originating at this site. (b) Sagittal CT angiogram. The typical truncated rat’s tail appearance (black arrow) of the stump of the ICA can erroneously lead to the belief that the ICA dissection commences proximally at this site. Axial data (not shown) demonstrated that the right ICA wall was of normal thickness in the proximal and midcervical region, with only vessel luminal collapse. (c) Axial postcontrast CT at presentation, showing the established right middle cerebral artery infarct. (d) Time-of-flight MR angiogram reconstruction 18 months later, showing remodelling and recanalisation of the vessel.
(Fig. 2), 1/38 (2.6%) in the middle third and 1/38 (2.6%) equally over the distal two-thirds. In no case was the mural haematoma thickest in the proximal third.
36/38 (94.7%), dissection involved the distal two-thirds, and in 2/38 (5.3%) it extended distally from the proximal third.
CAD due to indirect trauma
CAD due to indirect trauma
Eleven of 13 vessels (84.6%) had maximum mural thickness in the distal third (Fig. 3) and 2/13 (15.4%) in the middle third. Again, in none of these cases was the mural haematoma thickest in the proximal third.
Eight of 13 vessels (61.5%) had dissection confined to the distal third of the ICA, whereas in 1/13 (7.7%) it involved only the middle third, and in 13/13 (100%) it involved the distal two-thirds.
CAD due to direct traumatic injury One of three vessels (33.3%) had maximum mural thickness in the distal third of the ICA, 1/3 (33.3%) in the proximal third and 1/3 (33.3%) in the common carotid artery.
CAD due to direct traumatic injury One of three (33.3%) vessels had dissection confined to the distal third of the ICA, whereas in 2/3 (66.6%) it involved the proximal third and the distal common carotid artery.
Vessel redundancy Extent of arterial segment involved Spontaneous CAD Seventeen of 38 vessels (44.7%) had dissection confined to the distal third of the ICA (Figs 1–4), whereas in 410
Twenty-seven of 36 patients (75%) with spontaneous CAD demonstrated vessel redundancy (Figs 3,4), compared with 1/11 (9.1%) patients with CAD due to indirect trauma. This difference was statistically significant (P = 0.0002). © 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
Fig. 2. Forty-year-old male who presented with right neck pain with no focal neurology. (a) Axial T2 time-of-flight MRA data set image and (b) axial T1-weighted image just below the level of the skull base. (c) Axial CTA and (d) 2D time-of-flight MRA data set images at the level of C4. Mural wall haematoma with marked thickening of the internal carotid artery (ICA) at the level of the skull base is demonstrated on MRA and T1-weighted imaging (large white arrows) (a,b). The collapsed unthickened ICA is shown at C4 (small white arrows) by both CTA and MRA (c,d) with a normal-appearing left ICA (black arrow). The dissection is confined to the upper third of the ICA despite occlusion of the entire cervical ICA.
(a)
(b)
(c)
(d)
Treatment
Follow-up interval
Eighteen of 50 patients (36%) received anticoagulant therapy with intravenous heparin or subcutaneous lowmolecular-weight heparin followed by warfarin; 28/50 patients (56%) received antiplatelet agents (predominantly aspirin), and 2/50 (4%) had endovascular stenting (Fig. 4). Two of 50 patients (4%) were not treated with anticoagulants or antiplatelet agents due to severe concurrent traumatic brain injury.
The mean total imaging follow-up interval for the cohort was 11.9 months, with a range of 0 to 47 months.
Recurrent stroke after presentation Recurrent stroke occurred in 2/32 (6.3%) patients. The first had unilateral spontaneous CAD causing a greater than 50% stenosis and experienced a recurrent stroke 6 days after initial imaging diagnosis. The vessel had progressed to occlusion despite anticoagulation with heparin followed by warfarin. The second patient had spontaneous unilateral CAD treated with aspirin and dipyridamole. He presented with a recurrent stroke 18 months later; a coexistent atheroma, rather than a persistent dissecting pseudoaneurysm, may have been responsible. © 2014 The Royal Australian and New Zealand College of Radiologists
Vessel outcome on final follow-up imaging After exclusion of the two patients who were stented, accessible imaging follow-up was available for 33 ICAs in 32 patients. On final follow-up, 19/33 vessels (57.6%) had recanalised (Fig. 1) or remodelled, 9/33 (27.3%) were stable, and 4/33 (12.1%) had shown progression.
Discussion Our data show that the location of origin of spontaneous extracranial ICA dissection is typically adjacent to the skull base. The traditional view that CAD involves the vessel just distal to the bulb7 arose in part from the fact that the diagnosis was traditionally made using angiography, a luminal imaging technique. A ‘rat-tail sign’ or ‘flame-shaped’ ICA occlusion was previously regarded as pathognomonic of CAD.8 However, as CAD is primarily a condition of the vessel wall with secondary luminal changes, imaging the vessel wall itself allows us to more 411
J Downer et al.
(a)
(b)
(c)
(d)
accurately assess the extent of the vessel involved.9 We recognise that the ‘flame sign’ is not specific for CAD.8 In our experience, any cause of distal ICA occlusion, for example acute thromboembolic occlusion of the carotid T, can produce this appearance on catheter angiography. Although not widely available at the time of this series, MRI vessel wall imaging techniques have emerged that offer the potential for improved detection of intimal flaps, wall thickening and inflammation.10 There is a long-standing and unresolved debate as to whether CAD is initiated by an intimal tear or occurs due to direct intramural bleeding from ruptured vasa vasorum.2 Identifying an intimal tear on imaging or even pathology is notoriously difficult, and there is experimental evidence that spontaneous CAD is caused by primary intramural haemorrhage.11,12 In one pathological description, the intimal tear was identified in the distal segment of the vessel just below the skull base, the same segment in which the intramural haematoma was thickest.5 On the basis of these findings it seems reasonable to assume that the segment of maximal mural haematoma thickness is a surrogate marker for the segment of origin of the dissection. Our data demonstrate that this almost always lies in the distal third of the vessel, challenging the accepted view that the dissection extends cranially from an intimal tear in the proximal third of the vessel.3 Interestingly, a recent paper has shown that wall shear stress in patients with CAD is maximal not only just distal to the bulb but 412
Fig. 3. Forty-five-year-old male who presented with spontaneous right neck pain and Horner’s syndrome without focal cerebral infarction. (a) Axial, (b) oblique sagittal and (c) coronal reconstruction CT angiograms at the time of diagnosis. (d) Coronal CTA reformat 4 months later, showing mural haematoma with maximum thickness just below the entry of the internal carotid artery (ICA) into the skull base (a,c) (small black arrows), with extension of the haematoma into the foramen lacerum clearly demonstrated (large black arrow) (b). The coronal image (c) demonstrates the mural haematoma, but also bilateral cervical loops with a kink on the right. The CTA from 4 months later (d) demonstrates resolution of the mural haematoma and slightly less acuity in the angulation of the right ICA kink at the skull base.
also just below the skull base.4 The mechanism we propose is in large part that the ICA tears just below where it becomes relatively fixed in its passage through the bony skull base. Our results are contrary to a recent published series from New York.13 In this study, 6/11 cases of spontaneous ICA dissection (55%) involved the segment of the artery from the origin to the fifth cervical vertebra (C5), 4/11 (36%) the segment from C3 to C4 and 1/11 (9%) the segment from C2 to the skull base. However, our series includes a greater number of patients (36 out of 50 patients had spontaneous CAD) and followed a more rigid methodology. Using the same criteria as Barbour et al. in 1994,6 we found that spontaneous CAD is associated with vessel redundancy. In 2010, this finding had been challenged by Dittrich et al.,14 who demonstrated that there was no difference in mean ICA length in patients with spontaneous CAD when compared with matched controls. In light of this, our data would suggest that kinks, loops and coils, regardless of overall artery length, are associated with spontaneous CAD. Whether these changes are a mechanical risk factor for artery dissection remains uncertain. In our series, unlike in previously published series, acute ischaemic stroke was more common at diagnosis in patients with occlusive CAD.15,16 However, in those studies, diffusion-weighted imaging (DWI) was per© 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
(a)
(c)
(e)
(b)
(d)
(f)
Fig. 4. Fifty-six-year-old male patient who presented with left sided amaurosis fugax and right-sided neck pain. (a,c) Sagittal (a) and coronal (c) CT angiograms of the right internal carotid artery (ICA) at the time of presentation with lateral digital subtraction angiogram (DSA) of the right ICA (b) at presentation. (d,e) Lateral DSA of the left ICA before (d) and immediately after (e) stenting of dissection and pseudoaneurysm. (f) Anteroposterior DSA showing the left ICA at 6-month follow-up. The patient demonstrates a Barbour-type coil loop on the right and a Barbour C loop on the left, with bilateral pseudo-aneurysms (black arrows) with associated stenosis just at the infra-skull-base high cervical ICA. The patient underwent left-sided ICA stenting with a coronary balloon expandable stent across the stenosis and the psuedoaneurysm (e and f) with resolution and remodelling of abnormality, with successful right-sided stenting (not shown) performed at a later date.
formed in all patients, and this was used to define the diagnosis of ischaemic stroke rather than a clinical stroke syndrome. In our study, stroke was diagnosed clinically and on imaging, and our cohort included patients who where initially imaged using CT, which is less sensitive than DWI, particularly for detecting small strokes. Both Bonati et al.15 and Naggara et al.16 found that occlusive CAD was associated with larger infarct size, which is therefore more likely to be detected clinically and on unenhanced CT, offering a logical explanation for our findings. However, a subgroup analysis including only those patients imaged using MRI (with DWI) at presentation demonstrated that 9/18 (50%) of patients with non-occlusive CAD had acute ischaemic stroke, compared with 7/7 (100%) of those with occlusive CAD, a difference that was statistically significant (P = 0.0267). We are cautious in interpreting these results due to selection bias for those patients undergoing MRI in our study design, unlike the previous series. Our data confirm that there is no increase in the risk of stroke with increasing arterial stenosis in patients with non-occlusive CAD, consistent © 2014 The Royal Australian and New Zealand College of Radiologists
with distal thromboembolism being the most common mechanism of ischaemic stroke in this condition. As in previous published series, our data demonstrate a relatively good anatomical prognosis for the affected vessel in CAD,17–19 with nearly 60% remodelling at a mean follow-up interval of 1 year. Even in patients with occlusive CAD at diagnosis, the majority had recanalised at final follow-up (8/9, 88.9%). The key clinical question of whether ICA dissection is best treated using anticoagulants or antiplatelet agents has not yet been conclusively answered. The Cervical Artery Dissection in Stroke Study is an ongoing multicentre randomised controlled trial that we hope will provide the best evidence to guide our therapeutic decision-making in these patients.20
Conclusions In this cohort study of 50 patients, we have demonstrated that CAD most frequently originates in the distal ICA just below the skull base, just caudal to where the 413
J Downer et al.
vessel becomes fixed. Dissection is usually confined to the distal two-thirds of the cervical ICA, and spontaneous CAD is associated with vessel redundancy. The risk of acute ischaemic stroke is greatest in patients with occlusive CAD, with a low rate of acute recurrent stroke and a high rate of vessel remodelling.
References 1. Leys D, Bandu L, Henon H et al. Clinical outcome in 287 consecutive young adults (15 to 45 years) with ischemic stroke. Neurology 2002; 59: 26–33. 2. Debette S, Leys D. Cervical-artery dissections: predisposing factors, diagnosis, and outcome. Lancet Neurol 2009; 8: 668–78. 3. Schievink WI. Spontaneous dissection of the carotid and vertebral arteries. N Engl J Med 2001; 344: 898–906. 4. Callaghan FM, Luechinger R, Kurtcuoglu V, Sarikaya H, Poulikakos D, Baumgartner RW. Wall stress of the cervical carotid artery in patients with carotid dissection: a case-control study. AJP: Heart Circ Physiol 2011; 300: H1451–8. 5. Anderson RM, Schechter MM. A case of spontaneous dissecting aneurysm of the internal carotid artery. J Neurology Neurosurg Psychiatry 1959; 22: 195–201. 6. Barbour PJ, Castaldo JE, Rae-Grant AD et al. Internal carotid artery redundancy is significantly associated with dissection. Stroke 1994; 25: 1201–6. 7. Rodallec MH, Marteau V, Gerber S, Desmottes L, Zins M. Craniocervical arterial dissection: spectrum of imaging findings and differential diagnosis. Radiographics 2008; 28: 1711–28. 8. Arnold M, Bousser MG, Baumgartner RW. Prognosis and safety of anticoagulation in intracranial artery dissections in adults. Stroke 2007; 38: e140. 9. Kirsch E, Kaim A, Engelter S et al. MR angiography in internal carotid artery dissection: improvement of diagnosis by selective demonstration of the intramural haematoma. Neuroradiology 1998; 40: 704–9.
414
10. Hunter MA, Santosh C, Teasdale E, Forbes KP. High-resolution double inversion recovery black-blood imaging of cervical artery dissection using 3T MR imaging. AJNR Am J Neuroradiol 2012; 33: E133–7. 11. Volker W, Besselmann M, Dittrich R et al. Generalized arteriopathy in patients with cervical artery dissection. Neurology 2005; 64: 1508–13. 12. Volker W, Dittrich R, Grewe S et al. The outer arterial wall layers are primarily affected in spontaneous cervical artery dissection. Neurology 2011; 76: 1463–71. 13. Lleva P, Ahluwalia BS, Marks S et al. Traumatic and spontaneous carotid and vertebral artery dissection in a level 1 trauma center. J Clin Neurosci 2012; 19: 1112–14. 14. Dittrich R, Nassenstein I, Harms S et al. Arterial elongation (‘redundancy’) is not a feature of spontaneous cervical artery dissection. J Neurol 2010; 258: 250–4. 15. Bonati LH, Wetzel SG, Gandjour J, Baumgartner RW, Lyrer PA, Engelter ST. Diffusion-weighted imaging in stroke attributable to internal carotid artery dissection: the significance of vessel patency. Stroke 2008; 39: 483–5. 16. Naggara O, Morel A, Touze E et al. Stroke occurrence and patterns are not influenced by the degree of stenosis in cervical artery dissection. Stroke 2012; 43: 1150–2. 17. Nedeltchev K, Bickel S, Arnold M et al. Recanalization of spontaneous carotid artery dissection. Stroke 2009; 40: 499–504. 18. Rao AS, Makaroun MS, Marone LK, Cho JS, Rhee R, Chaer RA. Long-term outcomes of internal carotid artery dissection. YMVA 2011; 54: 370–5. 19. Redekop GJ. Extracranial carotid and vertebral artery dissection: a review. Can J Neurol Sci 2008; 35: 146–52. 20. Kennedy F, Lanfranconi S, Hicks C et al. Antiplatelets versus anticoagulation for dissection: CADISS nonrandomized arm and meta-analysis. Neurology 2012; 79: 686–9.
© 2014 The Royal Australian and New Zealand College of Radiologists
RADIOLO GY —O R I G I N A L A RT I C L E
The location of origin of spontaneous extracranial internal carotid artery dissection is adjacent to the skull base Jonathon Downer,1 Mahendra Nadarajah,2 Eliza Briggs,1 Peter Wrigley1 and William McAuliffe1,3 1 Neurological Intervention and Imaging Service of Western Australia, Sir Charles Gairdner Hospital, and 3Neurological Intervention and Imaging Service of Western Australia, Royal Perth Hospital, Perth, Western Australia, Australia, and 2National Neuroscience Institute, Singapore
J Downer MA, MRCP, FRCR; M Nadarajah B.Sci, FRCR; E Briggs BSci, MBBS; P Wrigley MBBS; W McAuliffe FRANZCR. Correspondence Dr William McAuliffe, Neurological Intervention and Imaging Service of Western Australia, Sir Charles Gairdner Hospital, Verdun St, Nedlands, Perth, WA 6009, Australia. Email: [email protected] Conflict of interest: None. Submitted 7 February 2013; accepted 16 January 2014. doi:10.1111/1754-9485.12170
Abstract Introduction: The traditional view is that spontaneous extracranial internal carotid artery (ICA) dissection (CAD) extends cranially from an intimal tear located just beyond the carotid bulb. This paper demonstrates that CAD originates in and primarily involves a more distal segment of the artery. Methods: A retrospective study of 54 dissected ICAs in 50 consecutive patients with spontaneous or traumatic CAD was undertaken. The site of the dissection, presence of ICA redundancy, rate of acute or delayed ischaemic stroke and vessel remodelling were determined. Results: Of the 51 dissections that occurred spontaneously or after indirect trauma, 25/51 (49.0%) were solely in the distal third of the artery, and 49/51 (96.1%) involved the distal two-thirds. Only 2/51 (3.9%) originated in the proximal third. ICA redundancy was seen in 27/36 (75%) of patients with spontaneous CAD, compared with only 1/11 (9.1%) of those with CAD due to indirect trauma (P = 0.0002). Acute stroke occurred in 10/12 (83.3%) of patients with ICA occlusion secondary to CAD and in 14/38 (36.8%) with non-occlusive CAD (P = 0.0074). Where follow-up was available, only 2/32 (6.3%) patients had a stroke after diagnosis, and 19/33 (57.6%) ICAs recanalised or remodelled. Conclusion: CAD occurring spontaneously or due to indirect trauma most frequently involves the distal extracranial ICA. Spontaneous CAD is associated with vessel redundancy, and the risk of acute stroke is greatest with occlusive CAD. The prognosis is good with treatment, with a low rate of recurrent stroke and a high rate of vessel remodelling. Key words: adult neuroimaging; angiography; magnetic resonance imaging; neurointerventional radiology; neuroradiology; vascular imaging.
Introduction Cervical artery dissection is an important cause of ischaemic stroke, accounting for almost 20% of cases in young adults.1 Cervical artery dissection may occur spontaneously or following trauma and is defined pathologically by the presence of a haematoma in the vessel wall.2 The traditional view has been that spontaneous dissection of the extracranial internal carotid artery (ICA) extends cranially from an intimal tear located just beyond the carotid bulb, approximately 2 cm from the vessel origin.3,4 This is primarily based on historical 408
pathology- and catheter angiography-based studies involving small numbers of patients.4,5 Our experience is that extracranial ICA dissection (CAD) originates in and is frequently confined to the distal cervical segment of the vessel, 1–2 cm below the skull base. The primary objective of this paper is to test that hypothesis. We also set out to assess the rate of acute ischaemic stroke at presentation and after diagnosis, to determine whether the degree of vessel narrowing is an important factor in determining the risk of stroke, to determine the incidence of cervical ICA loops or kinks in CAD and to find the rate and degree of vessel remodelling over time. © 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
Methods Study design A retrospective study of consecutive patients diagnosed with CAD on imaging was performed over a 5-year period from 2004 to 2009 at the three major teaching hospitals of the Perth region, with a catchment population of more than 2.4 million people.
Patient cohort Patients were identified by searching the state-wide Radiology Information System (RIS). Report text and clinical history fields were trawled using the search term ‘dissection’, generating a list of 331 patients. Review of the imaging reports of these cases identified 93 patients who were diagnosed with CAD. Diagnostic imaging was then reviewed with reference to specific inclusion criteria to arrive at a final cohort of 50 patients.
Inclusion criteria Patients with an imaging diagnosis of CAD were included, defined by demonstration of intramural haematoma, intimal dissection flap, vessel wall irregularity or pseudoaneurysm on CT/CTA and/or MRI/MRA datasets. Those patients with iatrogenic CAD or aortic dissection extending to the internal carotid artery were excluded. The final cohort therefore included cases of both spontaneous and traumatic CAD.
Data collection Clinical data was collected from the RIS record or the medical notes where this was incomplete. The mechanism of injury was categorised as spontaneous where there was no history of preceding trauma. When patients presented after traumatic injury, this was defined as indirect trauma (e.g. high-speed motor vehicle accident) or direct trauma (e.g. hanging). Imaging review was performed by an experienced consultant neuroradiologist and fellow. The longitudinal extent and location of the intramural haematoma or false lumen were determined on cross-sectional imaging at presentation. The site of maximum mural haematoma thickness was assessed in axial images. For the purpose of this analysis, the extracranial ICA was divided longitudinally into thirds over its length. ICA redundancy was assessed using the categorisation described by Barbour,6 who defined vessel loops, kinks and coils. Loops are C- or S-shaped deformities with two turns in the vessel with angles less than or equal to 90 degrees; kinks produce a sharp bend with an angle of less than or equal to 90 degrees; and coils are 360-degree turns in the vessel. A qualitative assessment of the anatomical outcome of the vessel was made on final follow-up imaging. © 2014 The Royal Australian and New Zealand College of Radiologists
Follow-up imaging Follow-up imaging was not available for 16 patients, of whom 7 were lost to follow up, 3 died, 3 returned home for follow up in another state or country, and 3 lived in remote rural areas.
Statistical analysis Statistical analysis was performed using GraphPad QuickCalcs (GraphPad Software, San Diego, CA, USA). P values were calculated using a two-tailed Fisher’s exact test.
Results Demographics Fifty patients with 54 dissected ICAs were included: 46 patients with unilateral CAD and 4 patients with bilateral CAD. The mean age of the patients was 44.8 years with a range of 19 to 68 years. Thirty-one of 50 (62%) were male, and 19 of 50 (38%) were female. MRI/MR angiography was performed in 25/50 patients (50%) and CT angiography in 41/50 (82%) as part of the imaging work-up at presentation.
Mechanism of arterial injury Thirty-six of 50 patients (72%) had spontaneous CAD, 11/50 (22%) had CAD due to indirect trauma, and 3/50 (6%) had CAD due to direct traumatic injury.
Presenting symptoms Twenty-four of 50 patients(48%) presented with an ischaemic stroke and one patient (2%) with amaurosis fugax. Four of 50 (8%) had Horner’s syndrome, and one patient (2%) had hypoglossal nerve palsy due to local effects of the dissection.
Stroke risk at presentation and vessel patency Ten of 12 patients (83.3%) presenting with CAD resulting in ICA occlusion had suffered an acute ischaemic stroke at diagnosis (Fig. 1), compared with 14/38 patients (36.8%) with non-occlusive CAD. This difference was statistically significant (P = 0.0074). Five of 18 patients with CAD (27.8%) with a less than 50% stenosis presented with acute ischaemic stroke, compared with 9/20 (45%) among those with a greater than 50% stenosis. This difference was not statistically significant (P = 0.3276).
Origin of dissection defined by location of maximum mural haematoma thickness Spontaneous CAD Thirty-six of 38 vessels (94.7%) had maximum mural haematoma thickness in the distal third of the ICA 409
J Downer et al.
(a)
(b)
(c)
(d)
Fig. 1. Forty-five-year-old female with no history of trauma who presented with left-sided weakness and right neck pain. (a) Axial CT angiogram. Thickened mural haematoma, maximal 1 cm below the entry of the right internal carotid artery (ICA) into the skull base, is clearly demonstrated (two white arrows), with occlusive dissection originating at this site. (b) Sagittal CT angiogram. The typical truncated rat’s tail appearance (black arrow) of the stump of the ICA can erroneously lead to the belief that the ICA dissection commences proximally at this site. Axial data (not shown) demonstrated that the right ICA wall was of normal thickness in the proximal and midcervical region, with only vessel luminal collapse. (c) Axial postcontrast CT at presentation, showing the established right middle cerebral artery infarct. (d) Time-of-flight MR angiogram reconstruction 18 months later, showing remodelling and recanalisation of the vessel.
(Fig. 2), 1/38 (2.6%) in the middle third and 1/38 (2.6%) equally over the distal two-thirds. In no case was the mural haematoma thickest in the proximal third.
36/38 (94.7%), dissection involved the distal two-thirds, and in 2/38 (5.3%) it extended distally from the proximal third.
CAD due to indirect trauma
CAD due to indirect trauma
Eleven of 13 vessels (84.6%) had maximum mural thickness in the distal third (Fig. 3) and 2/13 (15.4%) in the middle third. Again, in none of these cases was the mural haematoma thickest in the proximal third.
Eight of 13 vessels (61.5%) had dissection confined to the distal third of the ICA, whereas in 1/13 (7.7%) it involved only the middle third, and in 13/13 (100%) it involved the distal two-thirds.
CAD due to direct traumatic injury One of three vessels (33.3%) had maximum mural thickness in the distal third of the ICA, 1/3 (33.3%) in the proximal third and 1/3 (33.3%) in the common carotid artery.
CAD due to direct traumatic injury One of three (33.3%) vessels had dissection confined to the distal third of the ICA, whereas in 2/3 (66.6%) it involved the proximal third and the distal common carotid artery.
Vessel redundancy Extent of arterial segment involved Spontaneous CAD Seventeen of 38 vessels (44.7%) had dissection confined to the distal third of the ICA (Figs 1–4), whereas in 410
Twenty-seven of 36 patients (75%) with spontaneous CAD demonstrated vessel redundancy (Figs 3,4), compared with 1/11 (9.1%) patients with CAD due to indirect trauma. This difference was statistically significant (P = 0.0002). © 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
Fig. 2. Forty-year-old male who presented with right neck pain with no focal neurology. (a) Axial T2 time-of-flight MRA data set image and (b) axial T1-weighted image just below the level of the skull base. (c) Axial CTA and (d) 2D time-of-flight MRA data set images at the level of C4. Mural wall haematoma with marked thickening of the internal carotid artery (ICA) at the level of the skull base is demonstrated on MRA and T1-weighted imaging (large white arrows) (a,b). The collapsed unthickened ICA is shown at C4 (small white arrows) by both CTA and MRA (c,d) with a normal-appearing left ICA (black arrow). The dissection is confined to the upper third of the ICA despite occlusion of the entire cervical ICA.
(a)
(b)
(c)
(d)
Treatment
Follow-up interval
Eighteen of 50 patients (36%) received anticoagulant therapy with intravenous heparin or subcutaneous lowmolecular-weight heparin followed by warfarin; 28/50 patients (56%) received antiplatelet agents (predominantly aspirin), and 2/50 (4%) had endovascular stenting (Fig. 4). Two of 50 patients (4%) were not treated with anticoagulants or antiplatelet agents due to severe concurrent traumatic brain injury.
The mean total imaging follow-up interval for the cohort was 11.9 months, with a range of 0 to 47 months.
Recurrent stroke after presentation Recurrent stroke occurred in 2/32 (6.3%) patients. The first had unilateral spontaneous CAD causing a greater than 50% stenosis and experienced a recurrent stroke 6 days after initial imaging diagnosis. The vessel had progressed to occlusion despite anticoagulation with heparin followed by warfarin. The second patient had spontaneous unilateral CAD treated with aspirin and dipyridamole. He presented with a recurrent stroke 18 months later; a coexistent atheroma, rather than a persistent dissecting pseudoaneurysm, may have been responsible. © 2014 The Royal Australian and New Zealand College of Radiologists
Vessel outcome on final follow-up imaging After exclusion of the two patients who were stented, accessible imaging follow-up was available for 33 ICAs in 32 patients. On final follow-up, 19/33 vessels (57.6%) had recanalised (Fig. 1) or remodelled, 9/33 (27.3%) were stable, and 4/33 (12.1%) had shown progression.
Discussion Our data show that the location of origin of spontaneous extracranial ICA dissection is typically adjacent to the skull base. The traditional view that CAD involves the vessel just distal to the bulb7 arose in part from the fact that the diagnosis was traditionally made using angiography, a luminal imaging technique. A ‘rat-tail sign’ or ‘flame-shaped’ ICA occlusion was previously regarded as pathognomonic of CAD.8 However, as CAD is primarily a condition of the vessel wall with secondary luminal changes, imaging the vessel wall itself allows us to more 411
J Downer et al.
(a)
(b)
(c)
(d)
accurately assess the extent of the vessel involved.9 We recognise that the ‘flame sign’ is not specific for CAD.8 In our experience, any cause of distal ICA occlusion, for example acute thromboembolic occlusion of the carotid T, can produce this appearance on catheter angiography. Although not widely available at the time of this series, MRI vessel wall imaging techniques have emerged that offer the potential for improved detection of intimal flaps, wall thickening and inflammation.10 There is a long-standing and unresolved debate as to whether CAD is initiated by an intimal tear or occurs due to direct intramural bleeding from ruptured vasa vasorum.2 Identifying an intimal tear on imaging or even pathology is notoriously difficult, and there is experimental evidence that spontaneous CAD is caused by primary intramural haemorrhage.11,12 In one pathological description, the intimal tear was identified in the distal segment of the vessel just below the skull base, the same segment in which the intramural haematoma was thickest.5 On the basis of these findings it seems reasonable to assume that the segment of maximal mural haematoma thickness is a surrogate marker for the segment of origin of the dissection. Our data demonstrate that this almost always lies in the distal third of the vessel, challenging the accepted view that the dissection extends cranially from an intimal tear in the proximal third of the vessel.3 Interestingly, a recent paper has shown that wall shear stress in patients with CAD is maximal not only just distal to the bulb but 412
Fig. 3. Forty-five-year-old male who presented with spontaneous right neck pain and Horner’s syndrome without focal cerebral infarction. (a) Axial, (b) oblique sagittal and (c) coronal reconstruction CT angiograms at the time of diagnosis. (d) Coronal CTA reformat 4 months later, showing mural haematoma with maximum thickness just below the entry of the internal carotid artery (ICA) into the skull base (a,c) (small black arrows), with extension of the haematoma into the foramen lacerum clearly demonstrated (large black arrow) (b). The coronal image (c) demonstrates the mural haematoma, but also bilateral cervical loops with a kink on the right. The CTA from 4 months later (d) demonstrates resolution of the mural haematoma and slightly less acuity in the angulation of the right ICA kink at the skull base.
also just below the skull base.4 The mechanism we propose is in large part that the ICA tears just below where it becomes relatively fixed in its passage through the bony skull base. Our results are contrary to a recent published series from New York.13 In this study, 6/11 cases of spontaneous ICA dissection (55%) involved the segment of the artery from the origin to the fifth cervical vertebra (C5), 4/11 (36%) the segment from C3 to C4 and 1/11 (9%) the segment from C2 to the skull base. However, our series includes a greater number of patients (36 out of 50 patients had spontaneous CAD) and followed a more rigid methodology. Using the same criteria as Barbour et al. in 1994,6 we found that spontaneous CAD is associated with vessel redundancy. In 2010, this finding had been challenged by Dittrich et al.,14 who demonstrated that there was no difference in mean ICA length in patients with spontaneous CAD when compared with matched controls. In light of this, our data would suggest that kinks, loops and coils, regardless of overall artery length, are associated with spontaneous CAD. Whether these changes are a mechanical risk factor for artery dissection remains uncertain. In our series, unlike in previously published series, acute ischaemic stroke was more common at diagnosis in patients with occlusive CAD.15,16 However, in those studies, diffusion-weighted imaging (DWI) was per© 2014 The Royal Australian and New Zealand College of Radiologists
Internal carotid artery dissection
(a)
(c)
(e)
(b)
(d)
(f)
Fig. 4. Fifty-six-year-old male patient who presented with left sided amaurosis fugax and right-sided neck pain. (a,c) Sagittal (a) and coronal (c) CT angiograms of the right internal carotid artery (ICA) at the time of presentation with lateral digital subtraction angiogram (DSA) of the right ICA (b) at presentation. (d,e) Lateral DSA of the left ICA before (d) and immediately after (e) stenting of dissection and pseudoaneurysm. (f) Anteroposterior DSA showing the left ICA at 6-month follow-up. The patient demonstrates a Barbour-type coil loop on the right and a Barbour C loop on the left, with bilateral pseudo-aneurysms (black arrows) with associated stenosis just at the infra-skull-base high cervical ICA. The patient underwent left-sided ICA stenting with a coronary balloon expandable stent across the stenosis and the psuedoaneurysm (e and f) with resolution and remodelling of abnormality, with successful right-sided stenting (not shown) performed at a later date.
formed in all patients, and this was used to define the diagnosis of ischaemic stroke rather than a clinical stroke syndrome. In our study, stroke was diagnosed clinically and on imaging, and our cohort included patients who where initially imaged using CT, which is less sensitive than DWI, particularly for detecting small strokes. Both Bonati et al.15 and Naggara et al.16 found that occlusive CAD was associated with larger infarct size, which is therefore more likely to be detected clinically and on unenhanced CT, offering a logical explanation for our findings. However, a subgroup analysis including only those patients imaged using MRI (with DWI) at presentation demonstrated that 9/18 (50%) of patients with non-occlusive CAD had acute ischaemic stroke, compared with 7/7 (100%) of those with occlusive CAD, a difference that was statistically significant (P = 0.0267). We are cautious in interpreting these results due to selection bias for those patients undergoing MRI in our study design, unlike the previous series. Our data confirm that there is no increase in the risk of stroke with increasing arterial stenosis in patients with non-occlusive CAD, consistent © 2014 The Royal Australian and New Zealand College of Radiologists
with distal thromboembolism being the most common mechanism of ischaemic stroke in this condition. As in previous published series, our data demonstrate a relatively good anatomical prognosis for the affected vessel in CAD,17–19 with nearly 60% remodelling at a mean follow-up interval of 1 year. Even in patients with occlusive CAD at diagnosis, the majority had recanalised at final follow-up (8/9, 88.9%). The key clinical question of whether ICA dissection is best treated using anticoagulants or antiplatelet agents has not yet been conclusively answered. The Cervical Artery Dissection in Stroke Study is an ongoing multicentre randomised controlled trial that we hope will provide the best evidence to guide our therapeutic decision-making in these patients.20
Conclusions In this cohort study of 50 patients, we have demonstrated that CAD most frequently originates in the distal ICA just below the skull base, just caudal to where the 413
J Downer et al.
vessel becomes fixed. Dissection is usually confined to the distal two-thirds of the cervical ICA, and spontaneous CAD is associated with vessel redundancy. The risk of acute ischaemic stroke is greatest in patients with occlusive CAD, with a low rate of acute recurrent stroke and a high rate of vessel remodelling.
References 1. Leys D, Bandu L, Henon H et al. Clinical outcome in 287 consecutive young adults (15 to 45 years) with ischemic stroke. Neurology 2002; 59: 26–33. 2. Debette S, Leys D. Cervical-artery dissections: predisposing factors, diagnosis, and outcome. Lancet Neurol 2009; 8: 668–78. 3. Schievink WI. Spontaneous dissection of the carotid and vertebral arteries. N Engl J Med 2001; 344: 898–906. 4. Callaghan FM, Luechinger R, Kurtcuoglu V, Sarikaya H, Poulikakos D, Baumgartner RW. Wall stress of the cervical carotid artery in patients with carotid dissection: a case-control study. AJP: Heart Circ Physiol 2011; 300: H1451–8. 5. Anderson RM, Schechter MM. A case of spontaneous dissecting aneurysm of the internal carotid artery. J Neurology Neurosurg Psychiatry 1959; 22: 195–201. 6. Barbour PJ, Castaldo JE, Rae-Grant AD et al. Internal carotid artery redundancy is significantly associated with dissection. Stroke 1994; 25: 1201–6. 7. Rodallec MH, Marteau V, Gerber S, Desmottes L, Zins M. Craniocervical arterial dissection: spectrum of imaging findings and differential diagnosis. Radiographics 2008; 28: 1711–28. 8. Arnold M, Bousser MG, Baumgartner RW. Prognosis and safety of anticoagulation in intracranial artery dissections in adults. Stroke 2007; 38: e140. 9. Kirsch E, Kaim A, Engelter S et al. MR angiography in internal carotid artery dissection: improvement of diagnosis by selective demonstration of the intramural haematoma. Neuroradiology 1998; 40: 704–9.
414
10. Hunter MA, Santosh C, Teasdale E, Forbes KP. High-resolution double inversion recovery black-blood imaging of cervical artery dissection using 3T MR imaging. AJNR Am J Neuroradiol 2012; 33: E133–7. 11. Volker W, Besselmann M, Dittrich R et al. Generalized arteriopathy in patients with cervical artery dissection. Neurology 2005; 64: 1508–13. 12. Volker W, Dittrich R, Grewe S et al. The outer arterial wall layers are primarily affected in spontaneous cervical artery dissection. Neurology 2011; 76: 1463–71. 13. Lleva P, Ahluwalia BS, Marks S et al. Traumatic and spontaneous carotid and vertebral artery dissection in a level 1 trauma center. J Clin Neurosci 2012; 19: 1112–14. 14. Dittrich R, Nassenstein I, Harms S et al. Arterial elongation (‘redundancy’) is not a feature of spontaneous cervical artery dissection. J Neurol 2010; 258: 250–4. 15. Bonati LH, Wetzel SG, Gandjour J, Baumgartner RW, Lyrer PA, Engelter ST. Diffusion-weighted imaging in stroke attributable to internal carotid artery dissection: the significance of vessel patency. Stroke 2008; 39: 483–5. 16. Naggara O, Morel A, Touze E et al. Stroke occurrence and patterns are not influenced by the degree of stenosis in cervical artery dissection. Stroke 2012; 43: 1150–2. 17. Nedeltchev K, Bickel S, Arnold M et al. Recanalization of spontaneous carotid artery dissection. Stroke 2009; 40: 499–504. 18. Rao AS, Makaroun MS, Marone LK, Cho JS, Rhee R, Chaer RA. Long-term outcomes of internal carotid artery dissection. YMVA 2011; 54: 370–5. 19. Redekop GJ. Extracranial carotid and vertebral artery dissection: a review. Can J Neurol Sci 2008; 35: 146–52. 20. Kennedy F, Lanfranconi S, Hicks C et al. Antiplatelets versus anticoagulation for dissection: CADISS nonrandomized arm and meta-analysis. Neurology 2012; 79: 686–9.
© 2014 The Royal Australian and New Zealand College of Radiologists
