Image Findings of Acute to Subacute Craniocervical Arterial Dissection on Magnetic Resonance Vessel Wall Imaging: A Systematic Review and Proportion Meta-Analysis

Background and Purpose: This systematic review and meta-analysis aimed to evaluate the pooled proportion of image findings of acute to subacute craniocervical arterial dissection (AD) direct signs on magnetic resonance vessel wall imaging (MR-VWI) and to identify factors responsible for the heterogeneity across the included studies. Methods: A systematic literature search in the Ovid-MEDLINE and EMBASE databases was performed for studies published on the relevant topic before April 14, 2020. Pooled sensitivity and specificity values and their 95% confidence intervals (CIs) were calculated using bivariate random-effects modeling. Meta-regression analyses were also performed to determine factors influencing heterogeneity. Results: Eleven articles with data for 209 patients with acute to subacute craniocervical AD who underwent MR-VWI were included in this systematic review and meta-analysis. The most common findings on MR-VWI were wall hematoma (84%; 95% CI, 71%−92%), abnormal enhancement (72%; 95% CI, 49%−88%), aneurysmal dilatation (71%, 95% CI, 53%−84%), and intimal flap or double lumen signs (49%; 95% CI, 29%−71%). Among the potential covariates of heterogeneity, the presence of contrast-enhanced T1-weighted imaging (CE-T1WI) within the MR-VWI sequence combination significantly affected the pooled proportion of the intimal flap or double lumen signs. Conclusion: Wall hematoma and intimal flap or double lumen signs were the most common and least common direct sign image findings, respectively, on MR-VWI in patients with acute to subacute craniocervical AD. Furthermore, the absence of CE-T1WI in MR-VWI protocol was the cause of heterogeneity for the detection of the intimal flap or double lumen signs. This data may help improve MR-VWI interpretation and enhance the understanding of the radiologic diagnosis of craniocervical AD.


INTRODUCTION
Conventionally, digital subtraction angiography (DSA) had been the method of choice for diagnosing craniocervical arterial dissection (AD) using findings such as double lumen and pearl and string signs (1,2). Due to the invasiveness of DSA, magnetic resonance angiography (MRA) emerged as a non-invasive alternative to the luminal evaluation of craniocervical AD (3). Conventional MRA can provide information about geometric vascular changes, hemodynamic alterations, and collateral blood vessel development, but these are indirect signs, and there are inevitable limitations for the depiction and evaluation of arterial wall pathology (1,2). Owing to its non-invasiveness and superior performance compared to luminal angiography, high-resolution magnetic resonance vessel wall imaging (MR-VWI) is now widely used for the diagnosis of AD (2,(4)(5)(6).
However, despite improving the diagnosis of AD considerably, MR-VWI remains a subject of active ongoing research in terms of aspects such as sequence combinations, spatial resolution, and acquisition techniques (2,(4)(5)(6). Correspondingly, studies reporting image findings of direct signs of vascular wall pathology have focused on positive findings on MR-VWI in patients with acute or subacute craniocervical AD (4,(7)(8)(9)(10)(11)(12)(13)(14)(15). With this in mind, the recognition of direct signs of MR-VWI would enhance a comprehensive understanding of the radiologic diagnosis of craniocervical AD. Furthermore, the factors responsible for the heterogeneity in techniques, protocols, and image findings of direct signs across studies need to be identified and crossvalidated. In this context, this systematic review and metaanalysis aimed to evaluate the pooled proportion of image findings of direct signs of craniocervical AD on MR-VWI and to identify factors responsible for heterogeneity across the included studies. To the best of our knowledge, this is the first systematic review and meta-analysis that pools image findings on MR-VWI in patients with AD.

MATERIALS AND METHODS
This study was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (16).

Literature Search
A search of the MEDLINE and EMBASE databases was performed to find original literature reporting image findings of direct signs on MR-VWI for patients with acute to subacute craniocervical AD. The following search terms were used: [(Magnetic resonance vessel wall imaging) OR (MR * vessel wall imaging) OR (vessel wall imaging) OR (vessel wall MR * ) OR (high-resolution MR * ) OR (high-resolution magnetic resonance)] AND [(dissect * )]. No initial search date was set, and the literature search was updated until April 14, 2020. The search Abbreviations: AD, arterial dissection; BB, black blood; CE-T1WI, contrast enhanced T1-weighted imaging; MRA, magnetic resonance angiography; MR-VWI, magnetic resonance vessel wall imaging. was limited to publications in English. The bibliographies of relevant articles were searched to identify other relevant articles.

Inclusion Criteria
Studies satisfying the following criteria were included: (1) involved patients with acute to subacute craniocervical AD; (2) used MR-VWI as the index test; and (4) contained sufficient information regarding the proportion of image findings of direct signs in acute to subacute craniocervical AD on MR-VWI.

Exclusion Criteria
The following articles or article subset types were excluded: (1) case reports or series including fewer than five patients; (2) letters, editorials, conference abstracts, systematic reviews or meta-analyses, consensus statements, guidelines, and review articles; (4) articles not focusing on the current topic; (5) articles with, or with suspicion of, overlapping populations; and (6) articles containing insufficient information on the proportion of image findings of direct signs of acute to subacute craniocervical AD on MR-VWI.
Two radiologists, S.J.C and B.S.C, with 7 and 19 years of experience in neuroimaging, respectively, independently performed the literature search and selection.

Data Extraction
The following data were extracted using standardized forms according to the PRISMA guidelines (16): (1) characteristics and demographic data of the included studies: author names, year of publication, institution, country of origin, period of patient recruitment, study design (prospective vs. retrospective), patient number, male-to-female ratio, mean age of enrolled patients (adult or not), number of patients with acute to subacute craniocervical AD, reference standard, clinical diagnostic criteria for AD, mean interval days from symptom to MR-VWI and their ranges, AD location (intracranial vs. extracranial), and mechanism of the AD (spontaneous vs. traumatic); (2) patient symptoms and underlying risk factors at initial presentation; (4) MR-VWI characteristics: MRI machine and vendor, magnetic field strength, MRI dimension (2D or 3D), coil type, under-sampling techniques, use of the black blood (BB) technique during image acquisition, MR-VWI sequence combination, repetition time/echo time, field of view, slice thickness, acquisition matrix of T1-weighted imaging (T1WI) as a representative sequence, analytical methods to evaluate MR-VWI (subjective or objective), and the interobserver agreement for diagnosis; and (5) the proportion of image findings of direct signs on MR-VWI, luminal stenosis on angiography, and presence of infarction.
AD was defined as acute stage if it was detected within 3 days of symptom onset, subacute stage 4-60 days, and chronic stage after 60 days (17).

Quality Assessment
The methodological quality of the included studies was evaluated using tailored questionnaires and the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) criteria (18). Two

Data Synthesis and Analyses
The primary aim was to describe the demographic data and perform pooled proportion analysis of image findings of direct signs of acute to subacute craniocervical AD on MR-VWI. The secondary aim was to identify factors responsible for the heterogeneity across the included studies by employing meta-regression analysis. Pooled proportions were calculated using an inverse-variance weighting model (19)(20)(21). A randomeffects meta-analysis of proportions was utilized to calculate the overall proportions. Study heterogeneity was evaluated using the Higgins inconsistency index (I 2 ), with substantial heterogeneity indicated by an I 2 value >50% (22). In addition, meta-regression analysis was performed for the pooled proportion of image findings to determine the factor for the heterogeneity. All statistical analyses were conducted by one author (S.J.C, with 3 years of experience in conducting systematic reviews and metaanalysis) using the "meta" package of R (version 3.6.3; http:// www.r-project.org/).

Characteristics of the Included Studies
The total number of patients with AD across all studies was 327 (range, 6 to 118 patients). The number of dissected arteries across all studies was 372 (range, 6 to 145) ( Table 1). The mean patient age across studies ranged from 40.4 to 55 years. Except for one patient in one study, all studies included adults (10). Eight of the included articles were retrospective in design (4,7,8,11,(13)(14)(15)29), whereas three studies were prospective (9,10,12). The reference standard for diagnosing AD was a combination of clinical, radiological, and luminal angiography diagnoses in three studies (12,14,29) and of clinical, radiological, luminal angiography, and follow-up MRI in eight studies (4, 7-11, 13, 15). Two of these studies specifically used the spontaneous AD criteria for clinical diagnosis (8,15), and two used the Strategies Against Stroke Study for Young Adults in Japan criteria (11,14). The mean intervals from symptom to MR-VWI were shorter than 10 days in six of the included studies (4,9,(12)(13)(14)(15) and equal to or longer than 10 days in four (7,8,11,29); the relevant information was not specified in a remaining article (10). AD  was located exclusively in posterior circulation in six studies (4,8,9,(13)(14)(15) and predominantly in posterior circulation (in more than half the cases) in two studies (11,29); the locations were variable in the other studies (7,10,12). Seven studies only evaluated intracranial arteries (4,8,9,11,(13)(14)(15) and three studies evaluated both intracranial and extracranial arteries (7,10,29); the relevant information was not specified in a remaining article (12).

Meta-Regression
We used meta-regression analysis to determine the causes of heterogeneity. The covariates evaluated were mean interval from symptom to MR-VWI (<10 vs. ≥10 days), a predominance of posterior circulation in terms of AD location (more than half), magnetic resonance field strength (1.5 T vs. 3 T), performance via the BB technique during image acquisition, presence of CE-T1WI within the MR-VWI sequence combination, repetition  time (<500 vs. ≥500 ms), and slice thickness (≥0.6 vs. <0.6 mm).
Only the presence of CE-T1WI within the MR-VWI sequence combination was significantly associated with heterogeneity in the proportion of image findings of direct signs on MR-VWI, as it affected the pooled proportion of intimal flap or double lumen signs (p < 0.001). All other associations were statistically insignificant.

Quality Assessment
The bias risks according to the QUADAS-2 criteria were evaluated (Figure 3). We considered low bias risk in all studies in the index test domain as they all enrolled patients based on clinico-radiologic consensus, despite the absence of surgery considering the in-nature characteristics of diagnosis of this disease. In the patient selection domain, three studies were considered to have an unclear risk of bias due to unclear blinding or non-consecutive patient enrollment (7,13). In the reference standard domain, one study was considered to have an unclear risk of bias due to the absence of follow-up MRI confirmation of AD (12) and another due to unclear blinding to the index test (11). One study was considered to have a high risk of bias due to both unclear blinding and absence of follow-up MRI confirmation of AD (14). In the flow and timing domain, one study was considered to have an unclear risk of bias due to the time intervals from symptom to MRI being too heterogeneous (1 to 45 days) and a few suspicions of inappropriate patient exclusion (7). All studies were considered to have low applicability in the patient selection, index test, and reference standard domains.

DISCUSSION
This systematic review and meta-analysis present the pooled proportions of image findings of acute to subacute craniocervical AD on MR-VWI. The most common findings on MR-VWI were wall hematoma, abnormal enhancement, aneurysmal dilatation, and intimal flap or double lumen signs. Among the potential covariates of heterogeneity, the presence of CE-T1WI within the MR-VWI sequence combination caused heterogeneity for the pooled proportion of intimal flap or double lumen signs. The pooled proportion of image findings provides information that can aid MR-VWI interpretation and can be used to enhance the comprehensive understanding of the radiologic diagnosis of craniocervical AD.
Recently, high-resolution MR-VWI has been widely adopted and has reported superior performance to luminal angiography in the diagnosis of AD (2,(4)(5)(6). Also, several studies reported image findings of direct signs of dissected vessel wall on MR-VWI (4,(7)(8)(9)(10)(11)(12)(13)(14)(15). These findings involve wall hematoma (either in the false lumen or wall layer), outer wall aneurysmal dilatation, abnormal enhancement (either in the wall or flap), and presence of intimal flap and/or double lumen signs. Wall hematoma, the most common finding in craniocervical AD, is predominantly seen as the mural iso-to hyperintensity on T1WI (3/4). However, it can be confused as intraplaque hemorrhage in atherosclerosis, and hypointensity can also be observed on T1WI. The signal intensity on T2WI also contributes to the recognition of the stage of hematoma (2). The reason for abnormal enhancement in patients with AD is not fully understood yet. Inflammation, slow blood flow in the false lumen, and enhancement of the vasa vasorum are possible reasons for enhancement (30,31). Aneurysmal dilatation of the outer wall can be measured at the perpendicular plane of the diseased vessel, as MR-VWI allows clear discrimination between the outer wall and cerebrospinal fluid (32). Although direct visualization of the intimal flap and/or double lumen signs is the most reliable indicator, their incidence was the least in the current pooled analysis (4). Although MR-VWI has improved the diagnostic accuracy of AD, it remains a developing field in terms of sequence combinations, spatial resolution, and acquisition techniques (2,(4)(5)(6). MR-VWI protocols are inevitably heterologous, giving rise to the necessity for optimization. This meta-analysis also showed heterogeneity in image findings of direct signs of craniocervical AD across studies. In terms of sequence combinations, non-enhanced T1WI, T2WI, proton density, and enhanced T1WI sequences are generally adopted for MR-VWI (3). Han et al. (4) proposed that CE-T1WI was necessary, as they found dissection flaps in almost all cases with CE-T1WI, with substantial agreement among the readers. Correspondingly, this meta-analysis also suggests that the use of CE-T1WI within MR-VWI sequence combinations caused heterogeneity for the detection of the intimal flap or double lumen signs.
Although the effects of the other sequences could not be statistically evaluated in this meta-analysis, the combinations of non-enhanced T1WI, T2WI, proton density, and enhanced T1WI sequences are considered essential for evaluation and differential diagnosis of wall hematoma, dissection states, and vessel contours (3). The effectiveness of susceptibility-weighted imaging for detection of wall hematoma and diagnosis of craniocervical AD has also been suggested (24,25). Therefore, multi-sequential combinations need to be further evaluated in terms of the protocol.
Although significance was not gauged in this meta-regression due to the relatively small number of enrolled studies, there are several factors to be considered for protocol optimization. In two of the included studies, image acquisition slice thickness was more than 1 mm (4,12). As the craniocervical vascular wall is a thin structure, the recommended slice thickness for MR-VWI is <1 mm; therefore, thicker slices need to be avoided (3). Recently, novel techniques, including the BB (32)(33)(34)(35)(36) and under-sampling techniques (37,38) that improve the resolution and clinical accessibility of MR-VWI are actively being validated. Five of the studies included in this metaanalysis used the BB technique (4,8,10,12,13); due to suppression of arterial blood and cerebrospinal fluid signals, this technique performs better in intracranial wall evaluation than 3D turbo-spin echo alone (39)(40)(41)(42), but there are also some drawbacks, including a reduced signal-to-noise ratio of the vessel wall (32). A recent study evaluated the feasibility of the BB technique in the diagnosis of vertebrobasilar AD (5) and revealed that combining the 3D CE-T1WI and BB techniques reinforced the diagnostic performance of MR-VWI. Various under-sampling techniques, such as parallel imaging techniques, have been developed to simultaneously achieve a reduction of scan time while preserving image quality and scan range (43)(44)(45). Currently, parallel imaging techniques such as sensitivity encoding (SENSE) and generalized auto-calibrating partial parallel acquisition are limited in their ability to do this (43,45) compared to non-uniform sampling techniques such as compressed sensing (CS) or combined CS-SENSE (46). Only three of the studies included in this meta-analysis used undersampling techniques, and all of these were parallel imaging techniques (8,9,15). As the use of under-sampling techniques highly depends on the vendor, they are generally not easy to adopt. Thus, non-uniform sampling (CS or CS-SENSE) for craniocervical AD needs to be further evaluated.
This study had certain limitations. First, the number of included studies and their patient sample sizes were small; therefore, the statistical power of the proportion metaanalysis was low. Second, owing to the inherent scarcity of patients who undergo surgery for craniocervical AD, studies generally enrolled patients based on clinico-radiologic consensus. Third, as AD may present with chronologic and subsequent geometric changes after onset, we only included studies focusing on acute to subacute, and not chronic, AD. Further investigations regarding chronic AD and the corresponding chronologic changes are needed. Fourth, owing to the definition of MR-VWI is established recently, some papers could be missed by our searching process. Fifth, asymptomatic or mild symptomatic patients with AD may be lost in the original studies due to the absence of MR-VWI in the diagnostic process. Finally, we did not perform pooled analysis for objective quantitative measurements, as only a few studies provided the required data.

CONCLUSION
Our results indicate that wall hematoma and intimal flap or double lumen signs are the most and least common direct image findings, respectively, on MR-VWI in patients with craniocervical AD. Furthermore, the absence of CE-T1WI in MR-VWI protocol was the cause of heterogeneity for the detection of the intimal flap or double lumen signs.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary materials, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
SC and BC: conception and design of the study and acquisition and analysis of data. SC, BC, YB, SB, LS, and JK: drafting a significant portion of the manuscript or figures. All authors contributed to the article and approved the submitted version.