Imaging Characteristics of Growing and Ruptured Vertebrobasilar Non-Saccular and Dolichoectatic Aneurysms
Background and Purpose—Vertebrobasilar, nonsaccular, and dolichoectatic aneurysms generally have a poor natural history. We performed a study examining the natural history of vertebrobasilar, nonsaccular, and dolichoectatic aneurysms receiving serial imaging and studied imaging characteristics associated with growth and rupture.
Methods—We included all vertebrobasilar dolichoectatic, fusiform, and transitional aneurysms with serial imaging follow-up seen at our institution over a 15-year period. Two radiologists and a neurologist evaluated aneurysms for size, type, mural T1 signal, mural thrombus, daughter sac, mass effect, and tortuosity. Primary outcomes were aneurysm growth or rupture. Univariate analysis was performed with chi-squared tests for categorical variables and Student’s t test or analysis of variance for continuous variables. Multivariate logistic regression analysis was performed to identify variables independently associated with aneurysm growth or rupture.
Results—One hundred and fifty-two patients with 542 patient-years (mean 3.6±3.5 years) of imaging follow-up were included. Aneurysms were fusiform in 45 cases (29.6%), dolichoectatic in 75 cases (49.3%), and transitional in 32 cases (21.1%). Thirty-five aneurysms (23.0%) grew (growth rate=6.5%/year). Eight aneurysms (5.3%) ruptured (rupture rate=1.5%/year). Variables associated with growth and rupture on univariate analysis were size >10 mm (57.6% versus 16.0%, P<0.0001), mural T1 signal (39.7% versus 16.3%, P=0.001), daughter sac (56.3% versus 21.3%), and mural thrombus (45.5% versus 13.4%, P<0.0001). 26.7% of fusiform aneurysms, 9.3% of dolichoectatic aneurysms, and 59.4% of transitional aneurysms grew or ruptured (P<0.0001). The only variable independently associated with rupture was transitional morphology (P=0.003).
Conclusions—Vertebrobasilar, nonsaccular, and dolichoectatic aneurysms are associated with a poor natural history with high growth and rupture rates. Further research is needed to determine the best treatments for this disease.
Vertebrobasilar, nonsaccular, and dolichoectatic aneurysms (VBDA) are generally associated with a poor natural history with high rates of ischemic stroke, mass effect, and rupture.1–5 With the evolution of intraluminal flow diverters, there has been growing interest in treatment of these lesions.6,7 However, these treatments are associated with a substantial degree of morbidity and mortality when applied in the posterior circulation.7 Ideally, the only VBDAs that would undergo any interventional treatment are those with a rupture risk that exceeds procedure-related morbidity. For patients receiving conservative management, questions arise regarding the need for imaging surveillance. Imaging surveillance may be useful in cases in which the aneurysm is at a high risk of growth, thus potentially indicating an increased risk of rupture.
Because VBDAs have a substantially lower prevalence in the general population than saccular intracranial aneurysms and are likely heterogeneous in their etiologies, little is known regarding their natural history.5,8 Thus, we performed a study examining the natural history of VBDAs that underwent serial imaging at our institution and studied imaging characteristics associated with growth and rupture.
Materials and Methods
After IRB approval, we performed a search of our imaging and clinical database for all patients with clinical or radiological notes between January 1, 2002, and January 1, 2015, with the key words dolichoectasia, fusiform aneurysm, vertebrobasilar dolichoectasia, tortuous, dolichoectatic, and nonsaccular aneurysm. Two radiologists and a neurologist examined all neuroimaging performed on these patients to determine which patients had VBDA. VBDAs were defined using the definitions proposed by Flemming et al and had to meet at least one of the following imaging definitions: (1) fusiform aneurysmal dilatation: 1.5× normal diameter without a definable neck involving a portion of an arterial segment (either vertebral or basilar) with any degree of tortuosity, (2) dolichoectasia: uniform aneurysmal dilatation of any artery >1.5× normal involving either the entire basilar or vertebral or both with any degree of tortuosity, or (3) transitional: uniform aneurysmal dilatation of an artery >1.5× normal involving the vertebral, basilar artery, or both with a superimposed dilatation of a portion of the involved arterial segment.1 Patients with dissecting aneurysms were excluded because of the fact that these lesions are associated with a distinct natural history as compared with VBDA.9 Dissecting aneurysms were defined as those with ≥1 of the following: double lumen, string sign, alternating areas of stenosis and dilatation, semilunar hematoma with luminal narrowing, or arterial occlusion.
Basic demographics (age and sex) and comorbidities that were present at any time during imaging follow-up (hypertension, coronary artery disease, diabetes mellitus, current tobacco use, peripheral artery disease, hyperlipidemia) were collected for each patient. Patient symptoms at the time of the imaging test that first identified the VBDA were categorized as asymptomatic/incidental, cerebrovascular accident, subarachnoid hemorrhage, cranial nerve compression, brain stem mass effect, and headache.
Patients received computed tomographic angiography on either 16-slice, 32-slice, or 64-slice computed tomographic scanners. Patients receiving magnetic resonance imaging (MRI)/magnetic resonance angiography had imaging performed on 1.5 T or 3 T scanners. All images were evaluated on a GE PACS system with image reconstruction performed using on an independent 3D work station (TeraRecon, San Mateo, CA). Image reconstruction was performed to evaluate the morphology of the aneurysm, as well as to assess the maximum cross-sectional diameter of the aneurysm. All images were reviewed by 2 radiologists and a neurologist. Each aneurysm was classified into 1 of the 3 categories listed above by consensus. For each aneurysm, images were evaluated, and the following data were collected: maximum aneurysm diameter as measured on cross-sectional imaging, presence of T1 signal in the aneurysm rim on MRI, suggesting the presence of subacute thrombus manifests as T1-shortening, presence of a daughter sac, thrombus formation, mass effect on the brain stem, presence of other intracranial dolichoectasias, and presence of other intracranial saccular aneurysms. Growth was determined by parallel examination of images. All aneurysms were also given a score using the Smoker criteria, which assessed basilar artery diameter, laterality, and height of bifurcation.10
Studied outcomes were imaging evidence of aneurysm growth, rupture, and new posterior circulation infarct. Last follow-up was defined as the last time of imaging. The primary outcomes were growth and rupture of the VBDAs. Aneurysm growth was defined as an increase in the maximum diameter of the aneurysm by at least 2 mm. New infarcts were defined as either new areas of restricted diffusion in the posterior circulation or, in cases of new chronic infarcts, areas of high T2 signal with a corresponding T1 hypointensity. We studied the association between baseline imaging and anatomic characteristics and aneurysm growth and rupture. Clinical outcomes, such as disability and mortality, were not included in this study because the primary outcomes were growth and rupture of these lesions.
All categorical variables were analyzed using a chi-squared test. All continuous variables were analyzed using a Student’s t test. We performed a subgroup analysis of imaging characteristics and outcomes by aneurysm type (ie, fusiform, dolichoectatic, transitional). We also performed a multivariate logistic regression analysis to determine which imaging characteristics were independently associated with aneurysm growth and rupture. Variables were included in this model if they were significantly associated with growth or rupture on univariate analysis. Length of imaging follow-up was included in the multivariate analysis. We assessed hemorrhage and growth rate based on the number of hemorrhages or aneurysm growths that had occurred during the follow-up period divided by the number of person-years of follow-up. We also performed a survival analysis using the Kaplan–Meier estimate. The event of interest in the Kaplan–Meier estimate was aneurysm growth and rupture. A cox proportional hazards model was performed to determine the association between aneurysm growth and rupture. All statistical analyses were performed using JMP 12.0 (http://www.jmp.com).
Patient Population and Baseline Imaging Characteristics
Baseline characteristics are provided in Table 1. The process of patient selection is detailed in Figure I in the online-only Data Supplement. Ten patients with dissecting aneurysms which had a characteristic pearl on strings appearance with a focal stenosis just proximal to the aneurysmal dilatation were excluded from the study. A total of 152 patients with 6505 months (542 patient-years, range 0.5–17.5 years, mean 3.6±3.5 years) of imaging follow-up underwent serial imaging with MRI/magnetic resonance angiography, computed tomography angiography, or DSA during the course of our study. Of these patients, 5 were treated because of growth or rupture of the aneurysm and were censored at the time of treatment. Mean patient age at time of diagnosis was 60.4±12.8 years. One hundred and sixteen patients (76.3%) were male. Conventional angiography was performed on 45 patients (29.6%), computed tomographic angiography on 59 patients (38.8%), and MRI/magnetic resonance angiography on 145 patients (95.4%). In 67 cases (44.1%), aneurysms were incidentally discovered during work-up of nonvascular-related diseases, such as trauma, movement disorder, tumor, and so on. In 49 cases (32.2%), aneurysms were discovered during work-up for acute ischemic stroke or transient ischemic attack. In 15 cases (9.9%), aneurysms were discovered as part of a work-up for cranial nerve compressive symptoms; in 12 cases (7.9%), aneurysms were discovered for workup of brain stem mass effect type symptoms; and in 4 cases (2.3%), aneurysms were discovered as part of workup for headache.
Aneurysms were fusiform in 45 cases (29.6%), dolichoectatic in 75 cases (49.3%), and transitional in 32 cases (21.1%). 94.7% of aneurysms (144/152) had some degree of tortuosity. Mean maximum diameter at time of diagnosis was 8.7±4.7 mm. Range of aneurysm size was 5 to 30 mm, median was 7.2 mm, and the interquartile range was 6 to 9 mm. T1 signal in the aneurysm rim was present in 58 cases (38.7%), daughter sacs were present in 16 cases (10.5%), and thrombus formation was present in 56 cases (35.6%).
Outcome data are summarized in Table 2. Examples of growing and ruptured aneurysms are provided in Figures 1–3. Thirty-five aneurysms (23.0%) demonstrated aneurysm growth during the course of our study for an annual aneurysm growth rate of 6.5%. Eight aneurysms (5.3%) ruptured during follow-up for an annual rupture rate of 1.5% per year. Thirty-seven aneurysms (24.7%) either grew or ruptured over the course of this study. Twenty patients had new posterior circulation infarctions (13.2%). Kaplan–Meier estimates are provided in Figure IIA in the online-only Data Supplement.
The mean baseline size of growing aneurysms was significantly higher than that of stable aneurysms (12.4±6.4 mm versus 7.6±3.3 mm, P<0.0001). Nineteen aneurysms ≥10 mm (57.6%) grew, indicating an annual growth rate of 25.7%, whereas 16 aneurysms in the <10 mm group (13.5%) grew for an annual growth rate of 3.4% (P<0.0001). Other variables associated with aneurysm growth were presence of T1 signal in the aneurysm rim (39.7% versus 13.0%, P=0.0007), daughter sac (50.0% versus 19.8%, P=0.01), and thrombus formation (45.0% versus 11.3%, P<0.0001). 56.3% (18/32) of transitional aneurysms grew compared with 24.4% of fusiform aneurysms (11/45) and 8.0% (6/75) of dolichoectatic aneurysms (P<0.0001). Aneurysm laterality/tortuosity and height of bifurcation were not associated with aneurysm growth.
Ruptured aneurysms were larger than stable aneurysms but this difference was not statistically significant (12.3±5.7 versus 8.5±4.6, P=0.11). Five aneurysms ≥10 mm ruptured (15.2%) for an annual rupture rate of 6.8% of aneurysms compared with 3 aneurysms (2.5%) in the <10 mm group with an annual rupture rate of 0.6% (P=0.004). 10.9% of aneurysms with mural thrombus ruptured compared with 2.1% of aneurysms without thrombus (P=0.02). No other variables were associated with aneurysm rupture on univariate analysis, likely because of the small sample size of the rupture group. Of the ruptured aneurysms, 6 (75.0%) had imaging evidence of growth. Imaging growth was associated with a hazard ratio of 8.01 (95% confidence interval [CI]=1.84–54.80, P=0.005) for future rupture.
Subgroup Analysis by Aneurysm Type
The subgroup analysis by aneurysm type is presented in Table 3. There was no difference in mean age or sex between the 3 different aneurysm types. Transitional aneurysms had the largest mean diameter (13.6±6.5 mm), and dolichoectatic aneurysms had the smallest mean diameter (6.5±1.3 mm) (P<0.0001). 71.9% of transitional aneurysms had T1 signal in the aneurysm rim compared with 27.0% of dolichoectatic and 34.1% of fusiform aneurysms (P<0.0001). Daughter sacs were present in 21.9% of transitional aneurysms compared with 2.7% of dolichoectatic and 15.6% of fusiform aneurysms (P=0.005). Thrombus formation was present in 71.9% of transitional aneurysms, 22.7% of dolichoectatic aneurysms, and 33.3% of fusiform aneurysms (P<0.0001). Patients with transitional and fusiform aneurysms were more likely to have other intracranial aneurysms, as well as mass effect resulting from the aneurysm.
Aneurysm growth occurred in 56.3% of transitional aneurysms (growth rate of 15.8%/year), 24.4% of fusiform aneurysms (growth rate of 6.4%/year), and 8.0% of dolichoectatic aneurysms (growth rate of 2.4%/year; P<0.0001). Aneurysm rupture occurred in 12.5% of transitional aneurysms (rupture rate of 3.5%/year), 6.7% of fusiform aneurysms (rupture rate of 1.7%/year), and 1.3% of dolichoectatic aneurysms (rupture rate of 0.4%/year; P=0.04). There was no difference in infarct rates between groups (P=0.26). Kaplan–Meier estimates by aneurysm type are provided in Figure IIB in the online-only Data Supplement.
On multivariate analysis adjusting for aneurysm type, maximum diameter, presence of T1 signal in the aneurysm rim, presence of thrombus, and presence of a daughter sac, aneurysm type was the only variable significantly associated with growth and rupture. Transitional aneurysms had a significantly higher odds of growth and rupture than dolichoectatic aneurysms (odds ratio [OR]=6.63, 95% CI=1.95–23.92, P=0.003). There was no difference in the odds of growth and rupture between transitional and fusiform aneurysms (OR=2.46, 95% CI=0.83–7.46). There was a trend toward higher growth and rupture in fusiform compared with dolichoectatic aneurysms (OR=2.69, 95% CI=0.91–8.36, P=0.07). T1 signal in the aneurysm rim (OR=1.04, 95% CI=0.34–2.96, P=0.95), presence of a daughter sac (OR=1.79, 95% CI=0.51–6.27, P=0.36), presence of thrombus (OR=2.29, 95% CI=0.73–7.22, P=0.15), and increasing size (OR=1.06, 95% CI=0.96–1.17, P=0.34) were not independently associated with growth and rupture.
Our study of 152 VBDA with serial imaging follow-up demonstrated that these lesions have a poor natural history with growth rates of 6.5% per year and rupture rates of 1.5% per year overall. Factors that were strongly associated with aneurysm progression on univariate analysis were fusiform or transitional morphology, T1 signal in the aneurysm rim, presence of a daughter sac, presence of mural thrombus, and size. However, on multivariate analysis, aneurysm type was the only variable independently associated with growth or rupture. Larger aneurysms (ie, ≥10 mm in size) had growth rates of 25.7% per year and rupture rates of 6.8% per year. VBDAs were also associated with high rates of mass effect and posterior circulation infarction. These findings suggest that VBDAs, especially those with these worrisome features, may warrant close imaging follow-up. Further research is needed in identifying therapies that could mitigate the poor natural history of these lesions.
The natural history of VBDAs has been difficult to define in large part because of wide variability in nomenclature.1,3–5,11–14 In a systematic review of 9 studies and 440 patients, Shapiro et al found that definitions of VBDAs in natural history studies include a wide spectrum of definitions, including tortuosity without focal enlargement,4 fusiform or transitional aneurysms,1 and even dissecting aneurysms.15 In our study, we excluded intracranial dissections because they are thought to have their own distinct natural history1,9,15,16; however, further issues arise regarding heterogeneity in the imaging definition of dissecting aneurysms and the fact that small subintimal dissections are often difficult or impossible to visualize on conventional imaging.17 As demonstrated by Krings et al, dissection from mural hematoma may even play a role in the pathogenesis of VBDAs because it is possible that aneurysms of the fusiform or transitional type could represent healing dissections.18 Postmortem studies of VBDAs have demonstrated that growing aneurysms are characterized by microdissections and intimal disruptions with associated mural thrombus, multiple different ages of mural hemorrhage, and neovascularity.16,19–21 This may explain why high T1 signal in the aneurysm rim is strongly associated with aneurysm progression because this signal likely represents methemoglobin resulting from intramural hemorrhage.3,17,18 In addition, aneurysms of the fusiform and transitional type were more likely to have high T1 signal in the aneurysm rim when compared with aneurysms of the dolichoectatic type (ie, no focal dilatation). Furthermore, our finding that aneurysms with intraluminal thrombus have a poor natural history could be explained by the fact that partially thrombosed aneurysms are generally larger and are likely characterized by recurrent subacute and nontransmural dissections, which result in progressive enlargement of these aneurysms.17 This finding is also substantiated by the fact that thrombus formation was more likely in the fusiform and transitional aneurysms compared with the dolichoectatic.
Because of the fact that VBDAs likely represent a wide variety of pathologies, it is important to delineate the various morphological and imaging characteristics that are associated with lesion instability. In a study of 156 dolichoectatic aneurysms (ie, uniform rather than focal dilatation), Passero et al found that these lesions had a relatively poor natural history with lesion progression (defined as enlargement or increased tortuosity) occurring in 43% of cases with 10 years of follow-up and ≈5% of cases with 5 years of follow-up.4 Our study had a mean follow-up of 4 years and found that dolichoectatic aneurysms progressed in ≈8% of cases. Similar to our study, Passero et al found high rates of infarct on long-term follow-up. In this study, aneurysm size and height of bifurcation were associated with lesion progression, whereas in our study, factors, such as tortuosity and height of bifurcation, had no association with lesion progression. In a study of 52 patients receiving serial imaging follow-up, Mangrum et al demonstrated that 48% of VBDAs enlarged on follow-up (mean follow-up of 4 years). Risk factors for enlargement included aneurysm diameter, morphology, and presence of symptomatic mass effect. Similar to our study, aneurysms with a fusiform or transitional morphology were significantly more likely to progress than those of a dolichoectatic morphology. In addition, 54% of enlarging aneurysms demonstrated T1 hyperintensity in the aneurysm rim compared with 20% of those without enlargement. However, this difference was not statistically significant because of small sample size.
Although there is no consensus regarding the ideal management strategy for these lesions, our findings along with others’ suggest that imaging follow-up to evaluate for lesion progression may be warranted in the case of high-risk lesions. Growing lesions are generally associated with high rates of rupture, ischemic stroke, and compressive symptoms, and imaging follow-up could potentially be used to identify at-risk lesions so that therapies can be put in place to mitigate the poor natural history of these lesions.1,5,8,22,23 It is important to note, however, that there are no specific medical, surgical, or endovascular therapies specific to the treatment of VBDAs. In terms of stroke prevention, some studies have suggested that anticoagulation and antiplatelet therapy are not effective in preventing stroke in VBDAs and may be associated with higher hemorrhage rates,24,25 whereas other studies have suggested that anticoagulation may be protective against mortality.26 Surgical options include parent vessel occlusion with or without bypass and clip reconstruction or wrapping. However, morbidity and mortality associated with these therapies is high.27 There has been a growing interest in the use of endovascular techniques in the management of these lesions, especially with the advent of flow diverters. However, current data suggest that flow diverter treatment of VBDAs is also associated with high rates of morbidity and mortality.5,6,28
Our study has limitations. First, patients included in our study did not undergo imaging follow-up at prespecified intervals. Thus, there is marked heterogeneity in the timing and type of imaging follow-up that patients received. This could result in some bias because patients who had larger, more sinister appearing lesions may have received more close follow-up than those with a benign appearing mildly dilated and tortuous vertebrobasilar system potentially, resulting in overestimation of aneurysm growth and rupture rates. One distinct advantage that our study has over previously published studies is the use of MRI during imaging follow-up, thus allowing us to further characterize the tissue characteristics of these lesions. However, high-resolution vessel wall imaging was not performed on any patients, thus limiting our ability to examine characteristics, such as vessel wall enhancement and the presence of tiny amounts of hemorrhage in the vessel wall. There is also a limitation in ascertainment of outcomes. Another limitation is our inclusion and exclusion criteria. We excluded dissecting aneurysms and identified these based on their morphological characteristics. However, it is possible that some of the fusiform aneurysms in this study could be healed intracranial dissections. We defined growth as an increase in diameter of ≥2 mm. We chose this cutoff because differences of ≤1 mm are likely within the margin of error as a result of differences in measurement technique and scan plane. Furthermore, ≥2 mm diameter has been used in many prior studies as a definition for growth. Our center is a large national referral center, thus imaging evaluations performed at the patients’ home hospitals that were not uploaded into our imaging database would not be included in our study.
VBDAs are associated with a poor natural history with high growth and rupture rates. Predictors of aneurysm growth included large size (ie, ≥10 mm), a transitional or fusiform morphology, presence of high T1 signal in the vessel wall, mural thrombus, and a daughter sac. Further research is needed to determine the best treatments for this disease in high-risk patients.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011671/-/DC1.
- Received September 30, 2015.
- Revision received October 21, 2015.
- Accepted October 22, 2015.
- © 2015 American Heart Association, Inc.
- Shapiro M,
- Becske T,
- Riina HA,
- Raz E,
- Zumofen D,
- Nelson PK.
- Kobayashi N,
- Murayama Y,
- Yuki I,
- Ishibashi T,
- Ebara M,
- Arakawa H,
- et al
- Smoker WR,
- Corbett JJ,
- Gentry LR,
- Keyes WD,
- Price MJ,
- McKusker S.
- Echiverri HC,
- Rubino FA,
- Gupta SR,
- Gujrati M.
- Yu YL,
- Moseley IF,
- Pullicino P,
- McDonald WI.
- Krings T,
- Lasjaunias PL,
- Geibprasert S,
- Pereira V,
- Hans FJ.
- Krings T,
- Choi IS.
- Nakatomi H,
- Segawa H,
- Kurata A,
- Shiokawa Y,
- Nagata K,
- Kamiyama H,
- et al
- Sarikaya S,
- Sarikaya B.
- Passero SG,
- Calchetti B,
- Bartalini S.
- Ubogu EE,
- Zaidat OO.