Hemodynamics in Intracranial Aneurysm Formation

Abstract

Intracranial aneurysms (IAs) are dilated lesions of the arterial wall caused by a dysfunction of the hemodynamic stress, leading to subarachnoid hemorrhage, which can be devastating. In initiating IAs, two different hemodynamic factors, high wall shear stress (WSS) and mechanical stretch, simultaneously stimulate vascular endothelial cells and adventitial fibroblasts, to recruit macrophages into the vessel wall and cause chronic inflammation. Interestingly, whereas IA initiation is triggered by high WSS, its growth and rupture are suggested to be induced by low WSS. This tentative chapter describes the pathophysiology of IAs, focusing on hemodynamic features. Subsequently, recent advancements in diagnostic and therapeutic approaches to IA growth and rupture including computational fluid dynamics and artificial intelligence are discussed.

1. Introduction

Intracranial aneurysms (IAs) are defined as balloon-like dilations of the arterial wall caused by a dysfunction of hemodynamic stress and inflammatory processes, leading to the development of subarachnoid hemorrhages (SAHs). Following SAH, approximately 30% of cases succumb to the event, whereas approximately 50% suffer severe disabilities [1, 2]. Therefore, understanding the pathophysiology of IAs is crucial to overcoming this disease.

Recent evidence has shown that chronic inflammation is caused by hemodynamic stress, resulting in degenerative changes in the vascular wall. In particular, high wall shear stress (WSS) and mechanical stretch on the vessel wall are significant factors in IA development [3, 4, 5]. Conversely, factors such as low WSS and turbulence within the IA wall have been reported to contribute to its growth and rupture [6, 7]. However, the pathophysiology of IAs is yet to be elucidated because of the different key factors of hemodynamic stress and induced inflammatory cells during their growth, development, and rupture. Moreover, the analysis of human surgical specimens alone is insufficient to understand the causal relationship of IA formation and its pathological features due to the diversity of patient backgrounds, differences in disease stage, and inability to monitor pathological changes. Thus, animal models of IAs are now employed to conduct studies to elucidate IA pathogenesis.

This chapter outlines the mechanisms underlying hemodynamic stress-induced inflammation, its histopathological manifestations, the intricate relationship between these factors in the context of IA development, and future perspectives for the diagnosis and treatment of IAs.

2. Blood flow analysis in IAs

It is intuitive that high hemodynamic factors contribute to IAs; however, the precise mechanisms by which they disrupt vascular homeostasis and cause the eventual rupture remain unclear. To approach this question, it is first necessary to understand what forces blood flow exerts on the vessel wall and what pathological changes it causes in the vessel wall.

In IAs, the distal neck is defined as the “inflow zone” from the direction of blood flow, whereas the proximal neck is the “outflow zone” (Figure 1). The apex and its surrounding area are referred to as the “dome,” and the rest of the body is called the “body.” If the dome or body is accompanied by a protrusion, this lesion is called a “daughter sac,” which is a risk factor for rupture [8].

Figure 1.

A schematic image of the intracranial aneurysm formation and definition of its terminology.

2.1 History of blood flow analysis using models

In 1972, Roach et al. [9] fabricated IA models made of glass with diverse lumen diameters, bifurcation angles, and IA diameters prepared by skilled glass workers. They visualized blood flow by applying a steady or pulsatile flow using Evans blue dye to examine flow patterns.

In 1991, Nakatani et al. [10] harvested IA specimens from a rat model (Hashimoto model) [11] and visualized the streamline by using latex particles that were added to the water for continuous tracking and imaging with high-speed cameras ex vivo. They reported high WSS in the distal neck of the aneurysm and significant pressure gradients in the proximal neck. The manipulations required for the use of the Hashimoto model are listed below [11].

2.1.1 Increased blood flow stress in localized vascular bifurcations

Unilateral common carotid artery ligation increases blood flow to the contralateral intracranial vascular bifurcation.

2.1.2 Hypertension induction

Renal hypertension is induced by ligation of the (1) unilateral anterior and posterior renal arteries or (2) bilateral posterior renal arteries. Systemic hypertension is further exacerbated by a high-salt diet.

2.1.3 Fragile vascular walls

In addition to the above manipulations, a lysyl oxidase inhibitor (3-aminopropionitrile: BAPN), which inhibits intercellular cross-linking enzymes, is mixed with the food to weaken the vascular walls.

This multifactorial approach enables IA development in a stress-dependent manner without direct lesion manipulation. The model exhibits a nearly 100% reproducibility rate of IA formation at the anterior cerebral artery-olfactory artery (ACA-OA) bifurcation, accompanied by the recreation of distinct pathological features including loss of endothelial cell (EC) lining, disruption of internal elastic lamina (IEL), and degeneration of smooth muscle cells (SMC). Thus, it is considered to be an ideal model for understanding the pathogenesis of the IAs. However, given the inherently low rupture rate in this model, additional models are needed to further understand the pathophysiology.

2.2 Blood flow analysis using computational fluid dynamics analysis

Computational fluid dynamics (CFD) has become the mainstream method for hemodynamics analysis [12]. In 2004, Shojima et al. conducted CFD analysis using three-dimensional computed tomography angiography of patients with IAs, which revealed higher WSS values at the outflow zone of the aneurysm and lower values at the dome. This result was consistent with that of Nakatani et al. [10]. With its emergence in providing accurate findings, CFD analysis has been widely used in IA studies, investigating hemodynamic parameters such as WSS and related parameters (e.g., WSS gradient [WSSG] and WSS divergence [WSSD]), oscillatory shear index (OSI), gradient oscillatory number [13], and pressure loss coefficient (PLc) [14]. The formulas of the representative hemodynamic parameters listed above are shown in Figure 2. However, to facilitate the understanding of IA pathophysiology, we broadly discuss hemodynamic stresses by classifying them into shear stress and mechanical stretch.

Figure 2.

The formulas of representative hemodynamic parameters. GON: gradient oscillatory number, OSI: oscillatory shear index, PLc: pressure loss coefficient, WSS: wall shear stress, WSSD: wall shear stress divergence, WSSG: wall shear stress gradient.

3. Pathophysiology of IAs

3.1 Initiation of IAs

Studies on the developmental process of IAs are limited due to difficulties in predicting or observing the precise moment of pathogenesis in both clinical and animal models. For instance, cerebrovascular imaging prior to IA occurrence is rarely obtained unless prompted by specific medical concerns (e.g., medical check-up and head injury examination).

To compensate for this problem, animal aneurysm models capable of CFD analysis have been established. One study by Meng et al. [15] reported the creation of a new bifurcation in the neck region by dissecting and anastomosing the bilateral common carotid arteries and the proximal end of the right common carotid artery to the flexed left common carotid artery. In this canine model, histological examination revealed anaplastic remodeling of the vascular wall, including EC and SMC loss, IEL disruption, and hyperplastic remodeling characterized by intimal hyperplasia and thickening of the SMC layer. Although aneurysm formation was not observed in this model, the reported findings were consistent with early IA development, showing associations with high WSS and WSSG.

In a clinical study, Kulcsár et al. [3] reported the relationship between high WSS and spatial WSSG values and IA formation using CFD analysis of three patients. Similarly, Fujimura et al. [5] performed CFD analysis of 10 bifurcations and observed IA formation during 4.8 years of the mean follow-up period and 34 bifurcations in 10 patients without IA development for 8.9 years. In the IA group, PLc was significantly higher, which is presumably due to impingement of the vascular wall, and WSSG was approximately twice as elevated as the group without IA formation, although the difference was not significant.

For in vivo studies, a live imaging technique of the intracranial vascular wall using two-photon microscopy in the Hashimoto model was reported [16], allowing the observation of vessel wall motion prior to IA development and revealing the extension motion at the site of IA formation [4]. Additionally, histopathological analysis of ACA-OA bifurcation specimens in the Hashimoto model revealed that transient outward bulging was observed from days 0 to 2, followed by shrinkage and subsequent enlargement from day 5. During this period, immunohistochemistry showed expression of nuclear factor (NF)-κB in the EC, as well as C–C motif chemokine ligand (CCL) 2, a macrophage migration factor, in the adventitial fibroblasts. Subsequently, on day 5, SMC loss, IEL disruptions, and CD68-positive macrophage infiltration were all observed. The simultaneous activation of adventitial fibroblasts and ECs in the early stages of IA development suggests the significance of two hemodynamic stresses: (1) shear stress as a signal to the ECs and (2) mechanical stretch as a signal to the adventitial fibroblasts [4]. In other words, it was suggested that IA formation is regulated by two distinct hemodynamic factors.

For in vitro studies, stress-induced mechanotransduction mechanisms have been elucidated in ECs. The lipid bilayer, a major component of the cell membrane, has shown the ability to serve as a mechanosensor, allowing changes in its hydrophilicity, fluidity, cholesterol density, and undulation in response to hemodynamic stress. This signal is then transmitted into the intracellular components through P2X4 purinoceptors, ion channels, G-protein coupling receptors, integrin, and other cell adhesion proteins, thereby inducing calcium influx and ATP production followed by inflammation [17, 18, 19]. In fact, P2X4 receptor-deficient mice and rat models treated with paroxetine, a P2X4 receptor inhibitor, have shown significant inhibition of IA development [20]. Although shear stress and mechanical stretch have been shown to elicit opposite responses [18], CCL2 expression has been observed in stretched human aortic fibroblasts, which can potentially induce macrophages [4].

3.2 Growth of IAs

IA growth, known as one of the highest risk factors for rupture, is suggested to be caused by anaplastic changes of IA wall described above as a consequence of chronic inflammation. Because of hemodynamic stress affecting the weakened vascular walls, further IA growth would be induced.

Aoki et al. reported the lack of IA formation in the Hashimoto model using p50 knockout mice, a component of the transcription factor NF-κB [21]. Moreover, the inhibition of macrophage migration and suppression of NF-κB-associated cytokines, such as CCL2, interleukin (IL)-1β, cyclooxygenase (COX)-2, nitric oxide synthase (NOS), and matrix metalloproteinases (MMP)-9, have been shown to prevent IA formation in previous studies [22, 23, 24, 25, 26]. This highlights the crucial role of macrophages and the establishment of a positive feedback loop among prostaglandin E2 (PGE2), its receptor EP2, and NF-κB, leading to chronic inflammation and the self-amplification mechanism of macrophages [23, 26, 27]. Furthermore, Yamamoto et al. suggested activation of the sphingosine-1-phosphate receptor-1, a receptor responsible for maintaining endothelial junctions, inhibited both IA development and inflammatory response through transendothelial macrophage infiltration [28].

In a different rat model, Shimizu et al. [29] transected the left common carotid artery and anastomosed it apically to the contralateral common carotid artery, which was followed by hypertension induction and BAPN administration, as described in the Hashimoto model. Magnetic resonance angiography and CFD analysis were then performed to compare between cases with and without aneurysms. Before specimen collection, ferumoxytol, a supermagnetic iron oxide nanoparticle, was intravenously injected to visualize the distribution of phagocytic cells, such as macrophages. Results showed significantly lower WSS and higher OSI values at sites of aneurysmal enlargement, with histological evidence of macrophage aggregation suggesting the potential role of WSS and OSI in the regulation of aneurysmal enlargement [29].

3.3 Rupture of IAs

Due to the severe consequences of ruptured IAs, such as SAH, prophylactic treatment based on a thorough understanding of the pathophysiology is crucial. However, only indirect risk management strategies, including antihypertensive medication and smoking cessation, are currently available. Therefore, patients diagnosed with IAs must undergo periodic imaging follow-ups, with indications for surgical treatment (e.g., clipping and craniotomy) in high-risk cases. The risk of rupture is determined based on the following factors: increased IA size over time, presence of a daughter sac, high aspect ratio (calculated by dome-neck ratio), presence of multiple IAs, history of SAH, and presence of de novo aneurysm [30, 31]. In addition, the anatomical location of the IA is considered another risk factor, wherein lesions in the anterior and posterior communicating artery and basilar artery are considered high-risk due to their contribution to the Circle of Willis [8]. Other factors that may increase the risk of rupture include older age, female sex, oral bacteria, and race (particularly Finnish and Japanese individuals, albeit for unclear reasons) [32, 33]. To address the lack of standardized measures, scoring systems, such as the PHASES score [32] and the UCAS Japan prediction model [34], have been developed to predict the risk of rupture of IAs based on accumulated epidemiological data.

Although CFD analysis has improved our understanding of IA development, findings on ruptured IAs remain controversial, with conflicting reports suggesting the involvement of both high and low WSS values. Takao et al. [14] performed CFD analysis of 6 and 44 cases of IA with and without rupture, respectively, both of which were located in the internal carotid-posterior communicating artery (IC-Pcom) bifurcation, and seven and 43 cases of IA in the middle cerebral artery (MCA). Their results exhibited significantly lower minimum WSS in ruptured MCA aneurysms and higher PLc values in ruptured cases of both IC-Pcom and MCA [14]. Suzuki et al. [35] analyzed areas of thinning in the aneurysmal wall in seven cases of ruptured and 16 cases of unruptured IA and correlated them with CFD measurements. Their results showed that the thinned areas coincided with the areas of maximum pressure in 13 of 16 unruptured and 5 of 7 ruptured aneurysms. Furthermore, the ruptured aneurysm group showed significantly lower minimum, time-averaged, and normalized WSS values, suggesting their potential utility in predicting rupture and thinning [35]. On the other hand, Castro et al. [36] conducted a CFD analysis of 26 cases of anterior communicating aneurysms, and suspected aneurysms with small impaction zones, higher flow rates, and elevated maximum WSS were more likely to rupture. In a detailed study using human IA specimens, Frosen et al. classified the pathological features of human IAs into type I-IV and, intriguingly, reported that type IV, characterized by the loss of vascular ECs, presence of luminal thrombosis, and infiltration of inflammatory cells (e.g., macrophages and T-cells), was detected only in ruptured cases [37]. Similarly, Liu et al. [38] performed CFD analysis and histopathological classification for 113 human unruptured IA specimens. Their results showed that type IV IAs were larger in size, and normalized, averaged, and relative resident time WSS correlated with wall remodeling patterns, inflammatory marker expression, and atherosclerotic plaque formation of IAs [38].

For in vivo studies, two main models of ruptured IAs are used in literature. The first model is the modified Hashimoto model reported by Miyamoto et al. [39], which involves the ligation of the contralateral external carotid artery and pterygopalatine artery (PPA) in addition to unilateral common carotid artery ligation. As the PPA is the largest branch of the internal carotid artery in rodents, this manipulation further increased the blood flow to the unilateral intracranial side, resulting in a higher rupture rate (approximately 50%) compared to the standard model. In their study, inflammation-related factors, such as IL-1β and MMP-9, were found to be specifically expressed in ruptured lesions [39]. Different studies using the same model further revealed that neutrophils accumulated in the ruptured lesions, additional pro-inflammatory factors (e.g., TNF-α and PGE2) were expressed, and vasa vasorum were formed within the aneurysmal wall [40, 41]. The presence of vasa vasorum, in particular, has also been observed in human cases [42].

The second model is defined as the elastase injection model reported by Nuki et al. [43]. The specific steps of its inducement are described below. Currently, IAs in mast cell-deficient transgenic mice were reported to have a rupture rate of 20% compared with 80% of wild-type mice. Similar results were also obtained with stabilizer-induced mast cell suppression, suggesting the potential role of mast cells in IA ruptures [44].

3.3.1 Hypertension induction

Systemic hypertension is induced by continuously administering Angiotensin II using a subcutaneously implanted osmotic pump. Alternatively, deoxycorticosterone and sodium chloride have been reported to induce hypertension [45, 46].

3.3.2 Elastase-induced weakening of the vessel wall

Elastase is administered via stereotactic injection into the basilar cistern. This enzyme degrades the extracellular matrix of the intracranial vascular wall, resulting in IA formation. Elastase injection models have a high incidence of IA occurrence and rupture, allowing a better understanding of IA ruptures. However, it should be noted that the location of the aneurysm is not reproducible.

4. Recent developments in translational research and future direction

The most important goal in clinical practice of IA is predicting and preventing aneurysmal ruptures. The two major areas of research that hold promise of achieving this goal in the future are as follows:

1. Improvement of rupture prediction and diagnostic technology

2. Development of drug therapy for rupture prevention

As previously mentioned, the PHASES score [32] and UCAS Japan prediction model [34] are the currently available methods to prognosticate the risk of IA rupture, especially in identifying high-risk cases that require surgical intervention. However, some cases may still rupture despite the predicted low risk. Furthermore, the absence of a medical treatment underscores the need for improved prediction and diagnostic technology. In this chapter, we discuss the future direction of the findings described so far.

4.1 Improvement of rupture prediction and diagnosis technology

4.1.1 Artificial intelligence

Machine learning (ML) and deep learning (DL) have emerged as promising advancements across different research fields. In our context, various ML algorithms have been developed mainly as diagnostic aids for IA rupture prediction [47, 48], drawing on data, such as clinical information, imaging findings, imaging-based IA measurements, and hemodynamic parameters like WSS obtained by CFD analysis. A meta-analysis by Habibi et al. [49] involving 18,670 cases across 17 studies examined the use of ML for rupture prediction. They reported a sensitivity of 0.83 (95% confidence intervals [CI], 0.77–0.88), specificity of 0.83 (95% CI, 0.75–0.88), positive diagnostic likelihood ratio (DLR) of 4.81 (95% CI, 3.29–7.02), and negative DLR of 0.20 (95% CI, 0.14–0.29). Zhu et al. [50] showed similar findings by demonstrating the superiority of their three ML algorithms (random forest, artificial neural network, and support vector machine) over the PHASES score. This advantage can be attributed to its ability to handle large numbers of variables simultaneously and model nonlinear relationships, whereas the logistic regression model and PHASES score are limited to linear relationships.

4.1.2 Qualitative diagnosis of inflammation in the walls of IAs

Given the possibility of macrophage-dependent chronic inflammation in IAs, noninvasive visualization modalities may improve rupture prediction. Two mechanisms have been postulated for these concerns.

4.1.2.1 Utilizing the phagocytic ability of macrophages to internalize contrast medium

Macrophages can readily internalize nanoparticles (1–100 nm) through phagocytosis, which provides an avenue for visualization modalities [51]. Hasan et al. applied this concept with the intravenous administration of ferumoxytol as a negative magnetic resonance imaging (MRI) contrast agent. Their study showed that macrophages phagocytosing iron oxide (Fe₃O₄) nanoparticles could be visualized with MRI, allowing a noninvasive assessment of inflammation and determination of high-risk cases [52]. However, in 2015, the U.S. Food and Drug Administration issued a safety advisory regarding the potential for fatal anaphylactic shock with ferumoxytol use, and current guidelines have prohibited the administration of ferumoxytol outside of renal anemia, the original indication of this medication.

4.1.2.2 Vessel wall imaging (VWI)

Pathological hallmarks of IAs include macrophage infiltration, vasa vasorum formation, and vascular wall remodeling (neovascularization and atherosclerotic changes). These hallmarks are caused by blood flow stress and chronic inflammation, which can be detected using existing MRI contrast agents. Edjalali et al. [53] conducted VWI of 108 IAs using contrast-enhanced MRI and categorized 31 cases as unstable (defined as ruptured, symptomatic, or enlarged); and 77 cases as stable. They reported significantly better wall contrast in unstable cases (odds ratio, 9.2). Similarly, a meta-analysis of 1768 VWI cases across 12 studies demonstrated a positive association between VWI and IA rupture (prevalence ratio, 11.47; 95% CI, 4.05–32.46), as well as growth/symptomatic presentation (prevalence ratio, 4.62; 95% CI, 2.85–7.49). Longitudinal studies also demonstrated a positive association between average weighted enhancement and growth or rupture (risk ratio, 8.00; 95% CI, 2.14–29.88) [54].

4.2 Medical treatment

As mentioned earlier, surgical treatment is currently the only available therapeutic intervention. Based on recent findings, the following two concepts can be considered for medical treatment.

4.2.1 Alteration of sensitivity to hemodynamic stress

Direct alteration of vascular geometry is challenging with medical therapy. As for altering sensitivity to hemodynamic stress, paroxetine, a previously mentioned P2X4 inhibitor, shows potential in preclinical studies and awaits further clinical evaluation [20].

4.2.2 Inhibition of pathological progression by anti-inflammatory action

Among drugs with anti-inflammatory effects, three drugs show potential as treatments for IAs: (1) aspirin, a COX inhibitor; (2) hydroxymethylglutaryl-CoA (HMG-CoA) reductase inhibitors, known as “statins” (e.g., atorvastatin) which have a pleiotropic effect of NF-κB inhibitors. Antihypertensive agents and antihyperglycemic agents are also candidates for drug therapy [55, 56, 57]. For aspirin, a prospective observational study has been conducted in China (Unique identifier: NCT02846259) and in Germany and the Netherlands (Unique identifier: NCT03063541) [58, 59, 60]. Results from the study in China showed that aspirin significantly reduced IA rupture, whereas atorvastatin did not despite inhibition of IA enlargement.

5. Conclusion

IAs are chronic inflammatory diseases that are significantly influenced by hemodynamic stress. This condition follows a severe course involving its initiation, enlargement, and rupture, resulting in a devastating SAH. Although hemodynamic parameters and inflammatory cells involved in each phase are different, improved diagnostic techniques, such as CFD analysis and VWI for visualization of inflammation, hold promise for improved prediction rates in the future. Furthermore, although surgical treatment is currently the only available treatment, the establishment of medical alternatives based on an understanding of IA pathophysiology is expected.

Acknowledgments

I would like to thank Editage (www.editage.jp) for English language editing.

Conflict of interest

The authors declare no conflict of interest.

Other declarations

This article was supported by the Takeda Science Foundation, the Jikei University collaborative research fund (Grant number 2023-608 DC), and Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant number 23 K08509).