Low Wall Shear Stress Is Independently Associated With the Rupture Status of Middle Cerebral Artery Aneurysms
Background and Purpose—We determined which hemodynamic parameter independently characterizes the rupture status of middle cerebral artery (MCA) aneurysms using computational fluid dynamics analysis.
Methods—In 106 patient-specific geometries of MCA aneurysms (43 ruptured, 63 unruptured), morphological and hemodynamic parameters were compared between the ruptured and unruptured groups. Multivariate logistic regression analysis was performed to determine parameters that independently characterized the rupture status of MCA aneurysms.
Results—Univariate analyses showed that the aspect ratio, wall shear stress (WSS), normalized WSS, oscillatory shear index, WSS gradient, and aneurysm-formation index were significant parameters. The size of the aneurysmal dome and the gradient oscillatory number were not significantly different between the 2 groups. With multivariate analyses, only lower WSS was significantly associated with the rupture status of MCA aneurysms.
Conclusions—WSS may be the most reliable parameter characterizing the rupture status of MCA aneurysms.
Recent studies using computational fluid dynamics techniques have demonstrated that hemodynamic features may be key parameters for understanding cerebral aneurysm rupture.1,2 However, previous computational fluid dynamics studies simultaneously evaluated only a limited number of hemodynamic parameters in a relatively small number of patient-specific geometries, which included different locations of the aneurysm; this may be one reason why some findings are conflicting.
The current study focused on only MCA aneurysms. We used computational fluid dynamics to simultaneously analyze different hemodynamic parameters and determine which parameter independently characterizes the rupture status of MCA aneurysms.
One hundred and six saccular MCA aneurysms (43, ruptured; 63, unruptured) diagnosed with 3-dimensional rotational angiography were analyzed. The patient-specific geometries were reconstructed from 3-dimensional rotational angiography images. Computational fluid dynamics simulations were performed under pulsatile flow conditions driven from typical blood-flow waveforms of the common carotid artery in normal humans,3 and the following hemodynamic parameters were calculated: wall shear stress (WSS), normalized WSS (NWSS),2 oscillatory shear index (OSI),4 WSS gradient (WSSG),5 gradient oscillatory number,6 and aneurysm-formation indicator (AFI).7 Morphological (aneurysm size and aspect ratio) and hemodynamic parameters were compared between ruptured and unruptured aneurysms (see expanded methods in the online only data supplement).
As not all morphological and hemodynamic parameters were normally distributed, they were analyzed with the Wilcoxon rank-sum test (Table 1).
The median of the size of the aneurysmal dome and the aspect ratio were 5.36 mm and 1.43, respectively, for ruptured aneurysms, and 5.30 mm and 1.33, respectively, for unruptured aneurysms. There was no significant difference in the size of the aneurysmal dome (P=0.122), whereas the aspect ratio was significantly different (P=0.0298).
Distributions of WSS, OSI, WSSG, gradient oscillatory number, and AFI for ruptured and unruptured aneurysms are shown in Figure. Ruptured aneurysms had significantly lower WSS, lower NWSS, higher OSI, lower WSSG, and lower AFI than unruptured aneurysms (7.19 Pa versus 9.55 Pa, P=0.00010; 0.490 versus 0.618, P=0.0129; 0.0165 versus 0.0125, P=0.00891; 7.15 Pa/mm versus 10.40 Pa/mm, P=0.00020; and 0.971 versus 0.978, P=0.00631, respectively). Gradient oscillatory number was not significantly different between the 2 groups (P=0.180).
Intercorrelations Among Parameters
Intercorrelations between significant parameters with univariate analyses were examined using Spearman’s rank correlation test. WSS was significantly correlated with NWSS (correlation coefficient: R=0.8022, P<0.0001, coefficient of determination: R2=0.644) and WSSG (R=0.8743, P<0.0001, R2=0.764). Other parameters were not significantly correlated (R<0.6).
Because WSS had the lowest P value in the Wilcoxon rank-sum test among WSS, NWSS, and WSSG, which were significantly correlated, only WSS was used as a candidate variable. Thus, the aspect ratio, WSS, OSI, and AFI, which were significant with univariate analyses, were considered independent variables (Table 2). With multivariate logistic regression analyses, only WSS was significantly associated with the rupture status (OR: 0.929; 95% CI: 0.850–0.992; P=0.0227).
From analysis of the receiver-operating characteristic curve for WSS, the area under the curve was 0.723, and the cut-off value was 7.263, with a sensitivity of 0.698 and a specificity of 0.698 (Figure I in the online-only Data Supplementhttp://stroke.ahajournals.org).
This study showed that not only the aspect ratio, but also WSS and some WSS-related hemodynamic parameters, such as NWSS, OSI, WSSG, and AFI, were significantly different between ruptured and unruptured MCA aneurysms. Moreover, the most important finding is the first demonstration that only WSS is an independent factor for characterizing the rupture status of MCA aneurysms.
There are 2 theories explaining the mechanisms of cerebral aneurysm rupture: high- and low-flow theories.8,9 Consistent with these theories, previous computational fluid dynamics studies reported different findings regarding the rupture status of cerebral aneurysms. Cebral et al1 reported that ruptured aneurysms have higher WSS compared with unruptured aneurysms in 210 aneurysms, but they provided no information about the location and size. Xiang et al2 analyzed 119 aneurysms in various locations and reported that ruptured aneurysms have a lower WSS and higher OSI. Because aneurysm location may affect computational fluid dynamic findings, this study simultaneously determined various hemodynamic parameters only in MCA aneurysms, using a statistically reasonable number of patient-specific geometries. As a result, the results of this study support the low-flow theory. In addition, this study is the first to demonstrate that lower WSSG and lower AFI were significantly correlated with the rupture status of cerebral aneurysms. Previous studies demonstrated that OSI is associated with complex flow,4 and WSSG is associated with disrupted flow and intimal hyperplasia.5 In contrast, AFI could detect stagnation zones, and low AFI was found in the location where the aneurysm formed.6 Therefore, our results suggest that intra-aneurysmal flow stagnation and complex flow may be linked to aneurysm rupture, at least in MCA aneurysms.
This article has several limitations. First, blood was modeled as a Newtonian fluid with a fixed density and viscosity, and vessel wall was considered to be rigid. Although previous computational fluid dynamic studies reported that the assumptions limitedly affected the hemodynamics,10 it would be worthwhile to examine effects of the assumptions on hemodynamic parameters, especially in patients with smoking, a well-known risk factor for aneurysm rupture. Second, this study did not take aneurysm histology, or physiological and humoral parameters into account. Third, in this study, only WSS distinguished the rupture status of MCA aneurysms in multivariate analyses, but the results cannot be extrapolated to other sites of aneurysms that may have different hemodynamics. Forth, this is a retrospective study that is potentially biased. Lastly, ruptured aneurysms can undergo some structural changes during and after the rupture, which affect the hemodynamics. Thus, the hemodynamic features in a ruptured aneurysm may differ from that of the aneurysm just before the rupture. A large-scale prospective cohort study of unruptured aneurysms should be performed to characterize the hemodynamics of unruptured aneurysms that eventually rupture and those that do not, taking account of many factors, including physiological and humoral parameters that may affect the results. This would be an important step to determine if intra-aneurysmal flow dynamics is useful to predict the individual risk of rupture of unruptured aneurysms.
In the largest population examined thus far, this study demonstrated that WSS may be the most reliable indicator for discriminating the rupture status of MCA aneurysms. To confirm whether this finding is a cause, result, or mere epiphenomenon of an aneurysm rupture requires further studies.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.112.675306/-/DC1.
- Received September 3, 2012.
- Revision received October 1, 2012.
- Accepted October 16, 2012.
- © 2013 American Heart Association, Inc.
- Cebral JR,
- Mut F,
- Weir J,
- Putman C
- Xiang J,
- Natarajan SK,
- Tremmel M,
- Ma D,
- Mocco J,
- Hopkins LN,
- et al.,
- Meng H
- Mantha A,
- Karmonik C,
- Benndorf G,
- Strother C,
- Metcalfe R
- Boussel L,
- Rayz V,
- McCulloch C,
- Martin A,
- Acevedo-Bolton G,
- Lawton M,
- et al