Article Text
Abstract
Background and purpose Limited data exist to guide patient selection for preventive treatment of unruptured cerebral aneurysms. Cerebral aneurysms have been associated with circle of Willis anomalies but whether this association is also related to aneurysm rupture is not known. The occurrence of cerebral aneurysm rupture when a circle of Willis anomaly was present or absent was compared.
Methods Patients admitted over a 2 year period with a diagnosis of a cerebral aneurysm and an anterior communicating artery (ACoA) or posterior communicating artery (PCoA) aneurysm were included in the analysis. Brain vascular imaging was reviewed for aneurysm size, morphology and presence of circle of Willis anomaly. Relevant medical history and demographics were obtained from the medical records.
Results Of the 113 patients with ACoA or PCoA aneurysms, 85 (75.2%) cases were ruptured. There were 49 (43.4%) PCoA aneurysms and 64 (56.6%) ACoA aneurysms. Mean aneurysm size was 5.65 mm (SD 3.31). A circle of Willis anomaly was identified in 46 (40.7%) of all patients. Circle of Willis anomalies were present in 38 (46.9%) ruptured aneurysm cases and eight (29.6%) unruptured aneurysm cases. Multivariate analysis revealed a higher risk of aneurysm rupture when a circle of Willis anomaly was present (p=0.0245, OR 3.72 (CI 1.18 to 11.66)).
Conclusions This series shows that circle of Willis anomalies are more commonly found in ruptured as opposed to unruptured cerebral aneurysms of the anterior and posterior communicating arteries. The presence of a circle of Willis anomaly may be an important characteristic for selecting patients for preventive aneurysm treatment.
Statistics from Altmetric.com
Introduction
Preventive treatment of aneurysms via either clipping or coiling, as shown in the International Study of Unruptured Intracranial Aneurysms (ISUIA), was associated with an overall morbidity and mortality of approximately 10% at 1 year.1 Aneurysmal subarachnoid hemorrhage however, is a leading cause of stroke disability and death in young patients, with an estimated initial mortality rate of 50% and up to 50% morbidity in survivors.2 Rupture is not a common event though, with 51 (3%) of the 1692 patients with cerebral aneurysms developing a confirmed aneurysmal rupture during follow-up in the ISUIA study.1 Therefore, identifying risk factors for aneurysm rupture is fundamental in selecting patients who may truly benefit from preventive treatment.
Current methods to risk stratify patients with unruptured cerebral aneurysms are based on a limited understanding of its natural history. Published risk factors for aneurysm rupture have been largely limited to demographic factors and rudimentary anatomic factors, including female gender, a history of smoking and/or hypertension, family history, as well as aneurysm dome diameter and location in the posterior circulation.1 3 The paucity of detailed imaging based, quantifiable risk factors has led to considerable controversy about appropriate patient selection for preventive treatment. One example is the discrepancy in the literature on the risk of hemorrhage of aneurysms that are 7 mm or less in diameter. ISUIA concluded that aneurysms less than 7 mm without a history of subarachnoid hemorrhage located in the anterior circulation have zero risk of subarachnoid hemorrhage at 5 years.1 In contrast, several series of exclusively ruptured aneurysms demonstrate an overwhelming majority to have diameters of less than 10 mm.4
With increased availability and use of non-invasive brain vascular imaging, such as CT angiography (CTA) and MR angiography (MRA), detailed vascular anatomic information is now available that, in addition to maximal diameter, may provide additional anatomic information to adjudicate the risk of aneurysm rupture.
Rahman et al used a morphometric accounting for both maximum aneurysm dome diameter and the parent vessel diameter in correlating risk of cerebral aneurysm rupture.5 Other studies have similarly looked at more anatomic detail to indirectly determine direction of flow into the aneurysm based on vascular configurations in ruptured and unruptured groups.6 What these studies all suggest is that easily measurable and reproducible cerebrovascular anatomic information, perhaps serving as a surrogate for physiologic information, can be of practical value to the practicing neurovascular physician in better understanding cerebral aneurysm pathophysiology.
Circle of Willis variations, or a configuration that is not completely symmetric, is found in up to 40% of the population,7 and often includes a dominant anterior cerebral artery with a hypoplastic or absent contralateral anterior cerebral artery, or a dominant posterior communicating artery with a hypoplastic or absent P1 segment of the ipsilateral posterior cerebral artery. Such anatomic variations have been shown using quantitative in vivo imaging to be associated with physiologic changes in blood flow as well. Tanaka et al demonstrated in 125 subjects, using two-dimensional cine phase contrast MRI, that flow rates in both the carotid and basilar arteries were significantly different with the presence of a circle of Willis variation.8
Therefore, in an effort to improve patient selection for preventive treatment of unruptured cerebral aneurysms, we studied a familiar, easily measured, anatomic feature using unique criteria to see if there may be pathophysiologic implications in aneurysm rupture. We hypothesized that circle of Willis anomalies were more commonly seen in ruptured versus unruptured cerebral aneurysms of the anterior and posterior communicating arteries.
Methods
Patient selection
We identified all patients with an International Statistical Classification of Diseases and Related Health Problems (ICD-9) diagnosis code for subarachnoid hemorrhage (430) or non-ruptured cerebral aneurysm (437.3) admitted to a single urban tertiary care medical center over a 2 year period. Only patients with anterior communicating artery (ACoA) or posterior communicating artery (PCoA) aneurysms were included in the analysis. This study was approved by the Institutional Review Board of Rush University Medical Center.
Data extraction and imaging measurements
Demographic and clinical information was obtained from patient charts. All patients had prior cerebrovascular imaging with digital subtraction angiography (DSA), CTA or MRA. Cerebral angiograms for all patients were independently reviewed on a picture archiving and communications system using a submillimeter measurement tool for aneurysm location morphology and parent vessel diameter. Cases with evidence of vasospasm, poor image quality CTA or incomplete catheter angiograms were excluded. Vessel diameter was measured perpendicular to the direction of the vessel and aneurysm size was measured at the greatest diameter. Two authors (MAL and MC) reviewed all images and had access to all clinical data. Cases were divided between the two reviewing authors. After initial independent measurements were made by one of the two reviewing authors, images were masked for identification and adjudicated by the other reviewing author.
Definitions
Circle of Willis anomalies of the ACoA group were defined as hypoplasia (diameter ≤50% of the dominant A1) or absence of A1 segment contralateral to the anterior cerebral artery supplying the aneurysm (figure 1). A circle of Willis anomaly in the PCoA group was defined as presence of a fetal origin posterior cerebral artery (PCoA diameter greater than the P1 segment of the ipsilateral posterior cerebral artery) (figure 2). Aneurysm size was defined as maximal length, rounded to the nearest one-tenth of a millimeter. The shape of the aneurysm was defined as spherical when the maximal width was similar to its length. All other shapes were considered non-spherical.
Statistical analysis
For univariate analysis, the χ2 test and Student t test were used. Variables studied included circle of Willis anomaly, gender, aneurysm size, shape and age. Multivariate logistic regression analysis was performed with independent variables, including circle of Willis anomaly, gender, ACoA aneurysm location or PCoA aneurysm location, aneurysm size, shape, age, history of smoking and history of hypertension. The association between circle of Willis anomalies and ruptured aneurysms was measured. Statistical analysis was performed using SAS V.9.1 analytical software (SAS Institute Inc).
Results
During 2008 and 2009, 132 patients had cerebral aneurysms of the ACoA or PCoA treated at our institution. Of these, 19 were excluded for vasospasm, mislabeled date, insufficient angiographic data or mislabeled aneurysm site (ie, anterior choroidal artery versus PCoA). The remaining 113 patients were included in the analysis. Patient characteristics are summarized in table 1.
Mean age was 57.5 years, 83 (73.5%) patients were female and 85 (75.2%) cases were ruptured aneurysms. Nearly all patients had either DSA or CTA imaging (112/113, 99.1%) and one patient had only MRA imaging. Circle of Willis anomalies were present in 38 (46.9%) ruptured aneurysm cases and eight (29.6%) unruptured aneurysm cases. There were 49 (43.4%) PCoA aneurysms and 64 (56.6%) ACoA aneurysms. Aneurysm shape was non-spherical in 74 (65.5%) cases. Mean overall aneurysm size was 5.65 mm (SD 3.31) and mean dominant vessel diameter within anomaly groups was 2.31 mm (SD 0.39) and 1.96 mm (SD 0.52) for the A1 artery and PCoA, respectively. A circle of Willis anomaly was identified in 46 (40.7%) patients. A circle of Willis anomaly was present in 29 (45.3%) patients with an ACoA aneurysm and in 17 (34.7%) patients with a PCoA aneurysm. There was no significant difference between the two groups on univariate analysis (table 2). Multivariate regression analysis revealed a higher risk of aneurysm rupture when a circle of Willis anomaly was present (p=0.0245, OR=3.72 (CI 1.18 to 11.66)).
Discussion
Our series demonstrates that, among ACoA and PCoA aneurysms, circle of Willis anomalies occur more frequently in patients with ruptured cerebral aneurysms than in those that have not ruptured. This finding suggests that the presence of a circle of Willis anomaly in association with an ACoA aneurysm or PCoA aneurysm may be a risk factor for rupture. Investigating the association between circle of Willis anomalies and aneurysm rupture that includes a control group of this size is unprecedented in the literature. A challenge in the retrospective study of aneurysm rupture is identifying an appropriate control series. We used ICD-9 diagnosis coding to identify all patients admitted to a single institution over a specific time period to identify a series of ruptured cerebral aneurysms and a control series of unruptured cerebral aneurysms. Circle of Willis anomalies have been studied comparing normal cerebral vessels and those with cerebral aneurysms.9 Circle of Willis anomalies have not yet been extensively studied comparing unruptured aneurysms and those that had ruptured. Defining associations with ruptured aneurysms has greater clinical significance than predicting the presence of aneurysms. Therefore, we designed this study to evaluate the association of these anatomic features with rupture.
The association between circle of Willis anomalies and aneurysm rupture in previous reports has been primarily descriptive or consisted of much fewer cases.6 10 De Rooij et al reviewed CT angiograms of 126 patients, looking at incompleteness, asymmetry or dominance of the circle of Willis configuration and found no association with rupture on a per site basis. However, their series included only 22 ACoA aneurysms and 20 PCoA aneurysms, and the definition of A1 segment hypoplasia differed.6
We focused our study exclusively on ACoA and PCoA aneurysms, and used specific anatomic definitions for circle of Willis anomalies for several reasons. Firstly, previous reports have described an association between asymmetries of the anterior cerebral artery with ACoA aneurysms, and PCoA variations with the presence of aneurysms.6 9 11 Secondly, these locations seem most relevant to hemodynamic changes due to the most common circle of Willis anomalies. For example, the excess flow through a fetal origin PCA may not contribute unusual hemodynamic stress on a middle cerebral artery bifurcation aneurysm. In considering the relationship between anomaly and aneurysm, we included only the ACoA and PCoA vessels that were directly associated with the aneurysm and, therefore, conceivably involved in the pathogenesis of aneurysm rupture. For example, a fetal origin PCA contralateral to a PCoA aneurysm was not considered in the analysis. Lastly, cerebral aneurysms have been consistently reported to be most commonly located at these two locations.12
Pathophysiologic basis
First described by Thomas Willis (1621–1675), an English physician, in his 1664 work Cerebri anatomi, the Arterial circle of Willis provides for vascular redundancy at the skull base that optimizes blood flow to the brain. This vessel structure enables interhemispheric flow through the ACoA and in two directions through both PCoAs in the event of diminished blood supply through one of these vessels. Flow through these collateral pathways that comprise the circle of Willis deviate from geometric optimality principles in that certain bifurcations, in particular the ACoA and PCoA, do not achieve minimal energy expenditure.13
It has been observed clinically that most cerebral aneurysms are located in areas subjected to increased hemodynamic forces—that is, at the apex of arterial bifurcations or on the curve of tortuous vessels (figure 1)—rather than along straight vessel segments.14 This may explain the remarkable predictability with regard to aneurysm location despite the wide variability in aneurysm size and morphology. The two most common aneurysm locations within the circle of Willis are at the ACoA and the PCoA.12 Bor et al demonstrated the putative role of hemodynamics in cerebral aneurysm formation in evaluating the CTA or MRA of 26 patients who developed de novo aneurysms after follow-up of normal cerebral vessels compared with 78 controls who did not develop an aneurysm.15 They found significantly higher rates of aneurysms associated with hypoplastic branches and in those bifurcations with sharp angles. This study was unique and significant in that the cerebrovascular geometries were measured prior to aneurysm development, making their findings more consistent with a true risk factor.
Therefore, on a theoretical basis, increased flow in combination with a vascular configuration that does not minimize energy expenditure at bifurcations may, via turbulent flow and increased wall shear stress, contribute to cerebral aneurysm formation and progression.
Pathologically, the association between circle of Willis anomalies and aneurysms has been well documented.9 11 12 16 The definitions of circle of Willis anomalies vary widely among reports, which makes interpretation of the incidence within the general population challenging. In an anatomical study of the circle of Willis in 350 normal brains, only eight (2.3%) had a hypoplastic proximal segment anterior cerebral artery, and a fetal origin posterior cerebral artery was present in 51 (14.6%) cases.17 In an anatomical study of 1000 circles of Willis, 58% of aneurysms of the ACoA had an associated anterior cerebral artery asymmetry.12 The presence of circle of Willis anomalies in our series (40.7%) is comparable with previous reports of aneurysm bearing circles of Willis and, as would be expected, higher than series of non-aneurysm bearing circles of Willis. Circle of Willis anomalies may accentuate these aneurysm inducing hemodynamic forces with greater flow at specific bifurcations to compensate for an absent or hypoplastic vessel elsewhere in the circle of Willis. Tanaka et al demonstrated, using phase contrast MRA, that internal carotid artery flow volume was increased in the presence of an ipsilateral fetal-type posterior cerebral artery and in a contralateral absent A1 segment when compared with subjects with a normal circle of Willis configuration.8 These were the two specific sites of circle of Willis variations that were included in the present study. Experimental studies of fluid dynamic characteristics suggest that wall shear stress and pressure forces are major contributors to rupture of cerebral aneurysms.18 19 An increased inertial force and wall shear stress due to increased flow volume in circle of Willis anomalies may explain the promotion and rupture of cerebral aneurysms.
Study limitations
The present study has several limitations. A considerable challenge in studies of cerebral aneurysms is the study design, which is often limited by the difficulty of accurate long term prospective follow-up to evaluate rates of aneurysm rupture. The limitations of case control design must be considered in interpretation of the present study. We attempted to reduce sampling bias by including all patients hospitalized at a single urban, tertiary referral institution with a diagnosis consistent with either a ruptured or unruptured cerebral aneurysm during a defined time period. The ICD-9 coding system is imperfect for identifying cases given the possibility of misclassification20 but we intended to minimize error in the present study by chart review to confirm accuracy. The imaging modality was not consistent across all cases. DSA was used in 80% of cases but nearly all cases (112/113, 99.1%) had either DSA or CTA. Although these modalities differ, assessment of vascular anatomy with CTA is an acceptable alternative to DSA, and likely sufficient for qualitatively determining whether a circle of Willis anomaly was present.21 The initial cerebral angiogram was used for vessel measurement in the cases of ruptured aneurysms to minimize the possibility of vasospasm underestimating vessel diameter, and cases with vasospasm were excluded.
Definitions on what constitutes a circle of Willis anomaly vary widely within the literature, which makes comparison with prior reports challenging. The definition of a hypoplastic A1 segment of the anterior cerebral artery has been defined as a difference in diameter of both A1s of >33%,6 a diameter of the A1 segment being <50% of the contralateral21 22 or an A1 segment diameter of <1 mm.23 24 The definition of a fetal origin posterior cerebral artery or ‘fetal variant’, has been more consistent throughout the literature and is defined as a PCoA diameter being larger than the ipsilateral first segment (P1) of the posterior cerebral artery.24–27
Conclusions
The management of patients harboring unruptured intracranial aneurysms remains controversial and challenging because only a few, mostly demographic, risk factors for aneurysm rupture have been identified. This series aimed to study the use of more detailed anatomic information to assess rupture status. Our analysis suggests that ACoA and PCoA aneurysms with circle of Willis anomalies are associated with rupture. Further study of the physiologic characterization of blood flow, flow patterns and fluid dynamics within different circle of Willis configurations and their role in aneurysm rupture is warranted. Non-invasive studies of volumetric flow rates with phase contrast MRI may offer additional insight to the association between circle of Willis anomalies and aneurysm rupture. Future prospective observational natural history studies could establish circle of Willis anomalies as an important and readily available imaging finding that assists in patient selection for preventive aneurysm treatment.
Key messages
The association between cerebral aneurysms and circle of Willis anomalies has been previously shown; however, whether an association with rupture is present has not been demonstrated.
This series suggests that anterior communicating artery and posterior communicating artery aneurysms with circle of Willis anomalies are associated with rupture.
This finding, if confirmed in future studies, may be an important and readily available imaging finding that assists in patient selection for preventive aneurysm treatment.
References
Footnotes
Competing interests None.
Ethics approval This study was conducted with the approval of the Institutional Review Board of Rush University Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.