Spontaneous cervical artery dissection (sCAD) is an increasingly recognized cause of ischemic stroke, particularly in younger patients [¹]. Timely diagnosis is mandatory, as instant anticoagulation or antithrombotic therapy can help to prevent more serious complications [²]. Clinical symptoms are often nonspecific. Therefore diagnosis may be facilitated by dedicated imaging techniques. Several studies have shown that cardiovascular magnetic resonance (CMR) is ideally suited to establish the diagnosis of sCAD by identifying the vessel wall hematoma (VWH) using fat suppressed T1-weighted images [^3-5]. Due to the nonspecific symptoms of the disease, diagnosis is often delayed and it is often not possible to determine the exact onset of the patients' symptoms. Thus, the exact age of the arterial dissection often remains unknown. However, this information is useful as it has been shown that half of recurrent strokes/transient ischemic attacks in sCAD patients occur within the first 2 weeks after the event [⁶] and the anti-coagulation and anti-thrombotic therapies can be life threatening.

VWH is hypothesized to results from an intimal tear, allowing the blood to enter into the vessel wall, or from a rupture of the vasa vasorum. Both may lead to lumen stenosis or occlusion [⁷]. It is well known, that MR signal intensities of brain hematomas and hematomas in carotid atherosclerotic plaques change over time, due to the stepwise degradation of hemoglobin [^8-11]. Gomori et al categorized cerebral hemorrhages into hyperacute (<24 hours), acute (1-3 days), early subacute (>3 days), late subacute (>7 days) and chronic (>14 days) ¹¹ according to its signal intensities on T1- and T2-weighted images. In 1982, Lusby at al proposed a histological classification of carotid hematomas into fresh (< 1 week), recent (1-6 weeks) and old (> 6 weeks) [¹²]. Chu et al. [¹³] showed that high-resolution CMR at 1.5T is able to identify hemorrhages into carotid atherosclerotic plaques (intraplaque hemorrhage = IPH) with good correlation to histopathology and that CMR is able to differentiate between different stages of hemorrhage according to the signal intensities on T1w and T2w images. However, a recent in vivo CMR study with 40 transient ischemic attack/stroke patients with ipsilateral <70% carotid stenosis showed that in 11 out of 12 patients with intraplaque hemorrhage at baseline, the hemorrhage signal remained unchanged over a 1-year period [¹⁴], indicating that the time course of carotid intraplaque hematomas differs from that of cerebral hematomas.

The purpose of our study was to use the above mentioned classifications for carotid plaque and cerebral parenchymal hemorrhages to systematically investigate the value of CMR in determining the age of vessel wall hematoma (VWH) in patients with spontaneous cervical artery dissection (sCAD).

Our study showed that MR signal intensities of VWH are dependent on the age of the sCAD. More specifically, the age of the VWHs which were classified into acute, early subacute, late subacute and chronic hematomas according to its signal intensities on the T1w and T2w images differed, suggesting that in vivo CMR might be useful in determining the age of a sCAD. Given the fact that the risk of recurrent TIA or stroke is highest in the first two weeks after a sCAD [⁶] this information could be useful in patients with nonspecific initial symptoms, in whom the age of the sCAD is not known. From a clinical perspective, patients with more acute hematomas may therefore require a closer clinical observation than those with chronic hematomas. Furthermore, this information could be useful in cases where it is not clear whether the arterial dissection is spontaneous or traumatic. E.g., a patient with an arterial dissection and signs of an early subacute hemorrhage as determined by CMR, who suffered a car accident four weeks ago, will most likely have had a spontaneous dissection rather than a traumatic one. Finally, this multi-sequence MR protocol could give us new insights into the pathogenesis of the disease, especially in patients with multiple dissections, where the age of the VWH may be different for each individual affected vessel [¹⁸].

The MR signal changes observed in our study had only moderate agreement with the dynamics seen in cerebral hematomas (Cohen's kappa of 0.43; p < 0.001), suggesting that the time course of signal intensity changes varies in different body regions and organs. In contrast we found strong agreement with a previously proposed histology classification of carotid plaque hematomas [¹²] with a Cohen's kappa value of 0.74 (p < 0.001). However, in the study of carotid endarterectomy specimen of Lusby et al it is not clear how the age of the hematomas was determined and there is increasing evidence that MR signal characteristics of carotid intraplaque hematomas in atherosclerotic disease remain stable over a 1 year time period in most cases as has been suggested by Takaya et al [¹⁹] and Kwee et al [¹⁴]. These differences in the speed of hematoma decomposition might be explained by recurrent microbleeds into the atherosclerotic hematomas which are caused by leaky neovessels [²⁰].

In the present study, 83% of the investigated VWH had a homogeneous appearance that could easily be staged due to their signal characteristics. The homogeneous appearance of the VWH supports the theory of a one-time event of sCAD in most cases. Only 17% of the VWH had mixed signal intensities, which could either indicate recurrent hemorrhages or a different speed of decomposition. All of these cases had a mix of acute, early and late subacute hematomas, which were then classified as late subacute hematomas (Figure 4).

Similar to previous studies, different segments of the VA and ICA have different propensities to develop a VWH [^21-24]. More specifically, ICA dissections tend to occur significantly more often in the intracranial segment than in the cervical segment of the carotid artery (80% vs. 20%; p < 0.05). The most common location for VA dissections was the V2 (55%) and the V3 (36%) segment. Only one VA dissection started in the V1 segment and no dissection originated from the V4 segment. This is similar to findings of a study of 169 patients with VA dissection [²²] who found incidences of 20%, 34%, 35% and 11% for the V1, V2, V3 and V4 segment, respectively.

About half the patients in our study (46%) developed a stroke. No significant relation was observed between the level of stenosis in dissected vessels and the occurrence of stroke (70% vs. 65%; p = 0.7). This supports the theory of Benninger et al [²⁵] that a thrombembolic stroke-mechanism is probably more likely than hemodynamic effects in cervical artery dissection patients.

Although a CMR classification of VWH in sCAD patients seems feasible, our study has several limitations: since diagnostic workup for a patient with a fresh manifestation of cervical artery dissection takes time, we were not able to evaluate the hyperacute stage of hemorrhage and only one patient with the signal characteristics of acute hemorrhage was acquired. In general a longitudinal, rather than a cross-sectional study design would be more conclusive. Repeated scans of individual patients with cervical artery dissection within short intervals of time are necessary for a better understanding of the time course of the signals changes. Furthermore it can be difficult to determine the exact time between the occurrence of sCAD and the CMR scan. Therefore we included only patients with symptoms, such as TIA, stroke or Horner's syndrome, which could be clearly attributed to the sCAD. However, we cannot exclude that some patients might initially have had an asymptomatic sCAD and a recurrent episode of VWH a couple of days later causing the symptoms. However, the majority of our patients had homogeneous VWHs which could be clearly categorized into acute, early subacute, late subacute or chronic. Only 17% of all VWHs had mixed signal intensities, suggesting that either the speed of decomposition differed at different segments of the arteries or that these patients had a recurrence of VWH.

Magnetic resonance imaging signal intensities of VWH in sCAD vary over time and multi-sequence CMR can help to determine the age of an arterial dissection. Furthermore, findings of this study suggest that the signal intensity change of carotid hematomas differs from that of cerebral hematomas, suggesting a different time course of decomposition. VWH classification by CMR in patients with sCAD could be useful to improve our understanding of the pathophysiology of the disease, by helping to identify potential causes of the disease, e.g. traumatic versus spontaneous dissections. It could also help to improve risk stratification of recurrent TIA/stroke especially in patients with nonspecific initial symptoms, where the age of the sCAD is not exactly known.

The authors declare that they have no competing interests.

MH, TP, CCC, JG, AR, MAH, CO, KN, MFR and TS were involved in the study concept/study. MH, TP and TS were involved in data analysis/interpretation. MH, TP, CCC, JG, AR, MAH, CO, KN, MFR and TS were involved in manuscript preparation and editing. MH, TP, CCC, JG, AR, MAH, CO, KN, MFR and TS gave final approval of the submitted manuscript.