SEX AND VASCULAR BIOMECHANICS: A HYPOTHESIS FOR THE MECHANISM UNDERLYING DIFFERENCES IN THE PREVALENCE OF ABDOMINAL AORTIC ANEURYSMS IN MEN AND WOMEN

Abstract

The prevalence of abdominal aortic aneurysms differs greatly between men and women across the spectrum of ages. The reason for this discrepancy is not clear and likely involves several factors including the impact of sex hormones. We hypothesize that the unique spatial localization of abdominal aortic aneurysms is dictated in part by local hemodynamic forces on the vascular wall. Specifically, we propose that oscillatory shear stress is a specific signal to the endothelium that initiates the events ultimately leading to abdominal aortic aneurysm formation. We are proposing that sex-dependent differences in oscillatory shear stress in the infra-renal aorta may explain the observed differences between men and women. Initial observations suggest that, indeed, men and women have different degrees of oscillatory blood flow in the infra-renal abdominal aorta. The challenge is to extend these observations to show a causal relationship between oscillatory flow and aneurysm formation.

INTRODUCTION

Abdominal aortic aneurysms (AAAs) are a common occurrence in elderly men with a prevalence of up to 2.5% in men older than 70 years (1). There is a striking age-dependent increase in AAA prevalence with some estimates suggesting that more than 15% of men older than 75 years have an AAA (2). However, in women, the prevalence of AAA is markedly lower across the spectrum of ages (1,2). AAAs are relatively uncommon in women and, indeed, routine screening of women for AAA is not recommended.

The physiologic basis for this sex-dependent difference in the occurrence of AAA is unknown. A major emphasis of much basic research examining sex-dependent differences in atherosclerosis has focused on the obvious differences in the humoral environments as a likely mechanism for the dissimilarities in atherosclerosis between men and women. One study has suggested that similar mechanisms may be at play in animal models of AAA. However, an early study of AAA using an angiotensin II infusion model actually demonstrated a high incidence of AAA formation in female mice. Thus, it is not clear that these animal data translate to humans.

Although it is likely that more generalized humoral mechanisms play at least some role in sex-dependent differences in AAA formation, the unique spatial localization of AAA raises the possibility that other, local factors may be important. One potential reason for highly localized effects in the arterial tree is differences in the local hemodynamic environment. Blood flowing across the surface of the blood vessel imparts a shear stress to the endothelium. It is well known that localized variations in wall shear stress correlate with similarly localized atherosclerotic lesions. We are proposing that spatial differences in hemodynamics can explain in part the sex-dependent differences in AAA prevalence. We suggest that there are critical differences between men and women in the fluid mechanics of the arterial tree due to the unique anatomic features of men and women and that the resultant differences in localized shear stress are responsible for the sex-dependent discrepancy in AAA formation.

Observations Supporting a Relationship Between Oscillatory Flow and Aneurysm Formation

One of the key concepts that we have put forward is that a critical mechanism driving the localized formation of AAA is the presence of oscillatory flow in the distal abdominal aorta. This hypothesis is supported by several key lines of reasoning. First is the well-known clinical phenomenon of post-stenotic dilatation. This is often observed in the setting of severe aortic stenosis where there is aneurysmal dilatation of the ascending aorta. Similarly, with significant aortic coarctation, it is almost uniformly observed that the descending thoracic aorta distal to the coarctation becomes dilated and ectatic. In both settings, there is markedly disturbed flow in the aorta distal to the stenosis that is characterized in part by flow oscillations and eddies. Pathologically, the areas of the aorta distal to the stenosis have some similarities to abdominal aortic aneurysms in that there is loss of the media and accumulation of extracellular matrix and intense expansion of the adventitia. Similar changes in arterial morphology occur on many vascular beds downstream from a discrete stenosis which presumably generates a region of disturbed flow. Thus, there is precedence in other human disease states where areas of disturbed flow are associated with aneurysmal arterial dilatation.

A second line of evidence supporting a possible causal relationship between oscillatory flow and aneurysm formation is the previously defined pro-inflammatory effects of oscillatory blood flow on the endothelium. Inflammation has been implicated as a causal event in AAA formation (3,4). Relating disturbed flow patterns to local areas of inflammation is an important component for our overall hypothesis. By the nature of cardiac contraction, cardiac output is pulsatile resulting in non-steady flow patterns throughout much of the cardiovascular system. Additional complexity is imparted to blood flow patterns as a result of the geometry of the arterial vasculature. Branching, curvature, downstream resistance, and changes in the cross sectional area of the vasculature all alter the flow patterns in complex, but predictable ways. The most fundamental work in support of the relationship between oscillatory blood flow and inflammation comes from studies of cultured cells exposed to carefully proscribed flow patterns. Initial studies performed using cultured cells under steady flow conditions demonstrated that low wall shear stress induced expression of a variety of inflammatory proteins implicated in atherosclerosis, whereas higher shear stresses were associated with decreased expression of these same genes. These findings led to the concept of high shear stress being considered “atheroprotective” and, conversely, lower shear stress defining a state that is “atherosensitive.” Subsequent studies specifically examined the additional complexity of oscillatory flow and determined that oscillatory flow induces an even higher level of inflammatory gene expression. Our research group extended these observations to the in vivo setting to examine inflammatory protein expression in areas of oscillatory flow (5). Computational fluid dynamic models of blood flow in the aorta of the mouse were developed from micro–computer tomographic images and cardiac ultrasound (Figure 1). The average wall shear stress was mapped to the aorta which revealed that the inner curvature of the aortic arch (a site known to be predisposed to atherosclerosis) had extremely low mean wall shear stress, whereas the outer curvature of the aortic arch (a relatively atheroprotected region) had a much higher mean wall shear stress. Interestingly, when the direction and magnitude of the wall shear stress was calculated throughout the cardiac cycle, the inner curvature of the aorta exhibited a high degree of flow oscillation and the outer curvature showed essentially unidirectional shear stress. The polar plots in Figure 1 show the vector for the shear forces at different times in the cardiac cycle documenting reversal of shear stress in the inner curvature of the aorta. These areas of differing flow profiles also exhibit disparate degrees of inflammation. Vascular cell adhesion molecule–1 (VCAM-1) is a prototypical endothelial inflammatory gene known to be involved in atherosclerosis. As can be seen in Figure 2, endothelial expression of VCAM-1 was highest at sites of disturbed flow on the inner curvature of the aortic arch (Fig 2B). In areas of unidirectional flow, expression was markedly lower (Fig 2A). These data demonstrate an association but not causality. To demonstrate causality, we developed an animal model of pseudo-coarctation of the aorta by placing a non-obstructive nitinol band around the abdominal aorta (6). This mild constriction induced a discrete area of disturbed flow distal to the site of banding without a significant pressure drop (Figure 3). This acute induction of disturbed flow resulted in a robust and localized upregulation of VCAM-1 expression in the region of disturbed flow (Figure 3). Taken together, these data demonstrate that areas of disturbed flow characterized in part by flow reversal exhibit increased expression of pro-inflammatory proteins in the endothelium.

Fig. 1.

Mapping of flow patterns in the mouse aortic arch. To determine the spatial localization of flow patterns in the mouse aorta, computational fluid dynamic models were constructed using micro computed tomography (top left image) and Doppler ultrasound (bottom left image). The image of the aortic arch shows color encoded mean wall shear stress. Note that the inner curvature is an area of low mean wall shear stress whereas the outer curvature of the ascending aorta experiences relatively higher mean wall shear stress. The polar plots depict the shear stress at multiple points throughout the cardiac cycle. Note that in the outer curvature of the ascending aorta, an area that is relatively protected from atherosclerosis, the shear is unidirectional. In contrast, shear stress in the inner curvature of the aortic arch varies dramatically throughout the cardiac cycle with both positive and negative vectors. This is an area that is much more susceptible to atherosclerosis. This type of modeling demonstrates that there are spatially distinct areas of flow reversal in the aortic arch. Figure adapted from Suo et al (5).

Fig. 2.

Expression of vascular cell adhesion molecule 1 (VCAM-1) in the aortic arch. Shown in the center is an enface image of a mouse aorta with expression of VCAM-1 identified using quantum dots conjugated to an anti VCAM-1 antibody. The green color of the tissue represents auto fluorescence. The yellow color is indicative of higher levels of VCAM expression. The two insets are confocal microscopy images demonstrating localized VCAM-1 expression (red). Blue staining indicates the location of nuclei. In the region of low oscillatory shear stress this is very low expression of VCAM-1 (A). Note that in both the enface image and in the high-resolution images, the VCAM-1 expression is high in the endothelium of the inner curvature of the aortic arch (B). This is the site of higher oscillatory shear as shown in Fig 1. The other areas of higher VCAM-1 expression in the enface image aorta are also areas that are exposed to oscillatory shear stress (e.g., bifurcations). Figure adapted from Suo et al (5).

Fig. 3.

Demonstration of induction of inflammatory protein expression by disturbed flow in vivo. A pseudocoarctation was induced in the abdominal aorta of a mouse using a nitinol clip. Shown is a map of the predicted flow patterns on the left demonstrating an area of induced oscillatory flow distal to the clip. On the right are confocal microscopy images demonstrating induction of expression of vascular cell adhesion molecule 1 in the region of flow oscillation. Figure adapted from Willett et al (6).

The final data set that supports an association between oscillatory flow and AAA formation comes from a comparison of the sites of AAA formation and disturbed flow in mice and man. In man, the most common site of AAA formation is in the infra-renal aorta. In man, this is also the precise location of flow reversal. Mice do not normally develop AAAs. However, when atherosclerotic-prone mice are treated with angiotensin II, they develop AAAs that have many features in common with human aneurysms (7,8). However, there is one very distinct difference. These AAAs occur above the renal arteries. Interestingly, the flow patterns in the abdominal aortas of mice also differ from what is seen in man in that although there is a discrete area of flow reversal, it occurs above the renal arteries (9). Thus, whereas the location of AAA differs between mice and man, the spatial concordance of AAA formation and disturbed oscillatory flow is intact.

It is highly unlikely that oscillatory flow is the sole cause of AAA formation. It is far more likely that oscillatory flow is a “necessary but not sufficient” factor that works in concert with other risk factors such as sex, environmental exposures, and genetic variations to initiate AAA formation. However, we have developed three lines of data to support a causal relationship between oscillatory flow and AAA formation in the abdominal aorta: 1) oscillatory flow that occurs as a result of common pathophysiological conditions causes aneurysmal dilatation of arteries in man; 2) in both cell culture models and animal models of oscillatory flow, there is an upregulation of inflammatory proteins, reactive oxygen species, and matrix modifying enzymes, all of which have been mechanistically linked to AAA formation; and 3) the site of AAA in different species co-localizes with regions of oscillatory flow in the abdominal aorta.

METHODS

Study Population

We recruited healthy (cardiovascular disease–free) volunteers aged between 30 years and 42 years. Volunteers were screened for magnetic resonance imaging (MRI) safety per normal clinical protocols (i.e., no medical implants, no foreign objects in the eye, no claustrophobia, etc.). Enrollment of these study subjects was approved under the Emory Institutional Review Board.

Phase-contrast MRI

Phase-contrast MRI was performed on a Siemens 3T scanner (Malvern, PA, USA) using standard techniques. Parameters for the electrocardiogram – gated phase-contrast images are as follows: TR of 39.85s, TE of 3.39s, flip angle of 30°, field of view of 255 mm² × 340 mm², resolution of 1.3281 mm³ x 1.3281 mm³ x 5 mm³, Venc of 150 cm/s, yielding 30 time step images across the cardiac cycle.

RESULTS

Sex-Dependent Differences in Flow Patterns in the Abdominal Aorta

To explore the potential differences between men and women in flow patterns in the human aorta, we performed preliminary studies using phase-contrast MRI to ascertain the flow patterns in age-matched men and women without evidence of cardiovascular disease. This approach allowed us the opportunity to examine the aortic flow patterns at distinct anatomical locations in the abdominal aorta. Shown in Figure 4 are representative tracings derived from these studies. In the example from a healthy male, there is a distinct triphasic pattern of flow reversal in the infra-renal aorta with approximately 20% of the flow reversing during diastole. This pattern was not observed in the more proximal abdominal aorta. In contrast, we found that in women, there was significantly less flow reversal in the distal abdominal aorta (Figure 4).

Fig. 4.

Blood flow patterns in the infrarenal aorta of men and women demonstrate differing magnitudes of reverse flow. Phase contrast magnetic resonance imaging was used to measure the flow velocities in the abdominal aorta of healthy male and female volunteers. Shown are representative examples. Flow velocities are plotted throughout the cardiac cycle. Positive flow is flow towards the feet. Measurements were made in the infrarenal abdominal aorta. In males, there is a characteristic tri-phasic flow pattern with pronounced flow reversal in diastole. In females, the magnitude of the reversal is markedly lower.

Possible Reasons for Sex-Dependent Differences in Abdominal Aortic Flow Patterns

This very clear difference in flow patterns in the abdominal aorta of women versus men could be caused by several different factors. One obvious difference between men and women is the presence of the uterus in women and the associated blood flow requirements. The uterus contains a very low resistance vascular bed with a relatively high blood flow rate per gram of tissue. It is well known that in the normal female blood flow in the uterine artery does not exhibit any diastolic flow reversal (10). Normally blood flow from the extremities does have a strong reverse flow component. Thus, it is possible that the uterine vascular bed acts as a low-resistance “sink” to pull this reversed flow from the extremities into the uterine artery and thus reduce flow reversal in the abdominal aorta. To investigate this, we performed preliminary studies using phase-contrast MRIs of the internal and external iliac arteries of healthy men and women to see if there are differences in the flow patterns. Figure 5 presents representative images of flow patterns in the iliac arteries of men and women. There is diastolic flow reversal in both the internal and external iliac arteries of men; whereas in women, diastolic flow in the internal iliac artery is always antegrade (Figure 6).

Fig. 5.

Differences between men and women in blood flow reversal in the internal iliac arteries. Shown are representative flow profiles obtained using phase contrast magnetic resonance of the internal and external iliac arteries of normal men and women. Note that there is diastolic flow reversal in both the internal and external iliac arteries of the men whereas in the women, the internal iliac artery demonstrates only forward blood flow.

Fig. 6.

Schematic of hypothesized hemodynamic cause of reduced flow reversal in the abdominal aorta of women as compared to men. In both men and women, there is diastolic flow reversal in the external iliac arteries reflecting blood flow return from the extremities. Men also exhibit flow reversal in the internal iliac artery. However, in women, the relatively lower resistance of the uterine vascular bed preserves forward flow into the uterus during diastole and essentially draws the retrograde flow from the lower extremities into the internal iliac artery, thus protecting the distal abdominal aorta from experiencing oscillatory flow.

DISCUSSION

The clinical presentation, prevalence, outcomes, and perhaps even the etiologies of cardiovascular disease differ between men and women. We have focused on AAA because of the striking sex-dependent differences in this disease. In general, women are far less likely to develop AAA during their lifetime. As a result, routine screening of women for AAA is not recommended (11). Unfortunately, women are also more likely to experience an adverse outcome (12). Thus, there is a need to better understand the impact of sex on the pathogenesis of AAA. The reasons for sex-dependent differences in AAA formation are likely multifactorial and almost assuredly include differences in hormonal milieu (12–14). However, other factors are likely involved and we have proposed that unique, sex-dependent differences in blood flow patterns in the abdominal aorta may provide a relative protective effect in women for freedom from AAA formation.

Oscillatory blood flow patterns are particularly pro-inflammatory through their impact on the endothelium (15–20). The pathophysiology of AAA formation is dependent on developing an inflammatory state in the abdominal aorta (21–23) and the location of AAAs is very site specific. Thus, we are proposing that differences in flow oscillation in the abdominal aorta may impact the initiation of AAA formation. We used phase-contrast MRI to examine blood flow patterns in the abdominal aorta and we found that men do indeed have a very pronounced diastolic flow reversal pattern in the infrarenal abdominal aorta, the most common site of AAA formation. Interestingly, women had a much less prominent diastolic reversal of flow, consistent with our hypothesis. Furthermore, we examined blood flow patterns in the iliac vascular beds and found that although both sexes had prominent diastolic blood flow reversal in the external iliac arteries, only men had a similar pattern in the internal iliac arteries. We propose that, in women, the relatively lower resistance vascular bed of the uterus functions to draw the retrograde flow from the lower extremities into the internal iliac artery and thus mitigate the impact of this flow on the distal abdominal aorta (Fig 6). As a result, women have less diastolic flow reversal in the infra-renal aorta.

It is possible that other factors could be impacting the blood flow patterns in women. In addition to more obvious potential differences including muscle mass, body size, and vessel size, differences in vascular compliance could be important contributors. Women tend to have more compliant aortas (12) and this may reduce the impact of flow reversal through a Windkessel mechanism whereby the increased compliance of the abdominal aorta functions as an elastic reservoir and in essence absorbs the retrograde flow. Other possibilities include differences in curvature of the aorta or differences in the angulation of the branching points of the iliac arteries.

It is interesting to speculate about the potential impact of normal physiologic as well as pathological changes in uterine blood flow on abdominal aorta hemodynamics. For example, what is the impact of hysterectomy? One would predict that this would result in a “male” pattern of blood flow. In post-menopausal women, uterine blood flow is also reduced. Does this result in changes in abdominal aorta blood flow? Conversely, in pregnancy, one could hypothesize that there is less flow reversal due to the effects of increased blood flow to the uterus.

Finally, what are the implications of this hemodynamic hypothesis for AAA susceptibility? Current recommendations are against the routine screening of women for AAA. Perhaps we could identify high-risk groups (e.g., early hysterectomy) through additional biomechanical studies that would warrant screening. This is particularly important given the likelihood of worse outcomes from AAA in women. We should also consider the possibility that other factors that impact peripheral vascular resistance could alter the risk of developing AAA. In patients with severe peripheral vascular disease, it has been assumed that the association with AAA is related to the underlying pathology of atherosclerosis. One could hypothesize that the increase in resistance imparted by occlusive peripheral vascular disease could negatively impact flow patterns in the abdominal aorta. Other conditions that impact peripheral resistance that are worth considering include amputation and lower extremity paralysis.

In summary, we have put forth a hemodynamic hypothesis to explain the sex-dependent differences in AAA prevalence and implicating a role for the impact of oscillatory shear stress on the endothelium of the abdominal aorta. While this remains a fairly speculative hypothesis without support from true causal data, it is interesting to consider the potential implications. If this hypothesis is proven to be true, it not only provides a novel mechanism for AAA formation, but it may also assist us in identifying high-risk groups to screen for AAA.

DISCUSSION

Baum, New York City: Given that hypothesis, one might expect that if you did an oophorectomy hysterectomy that that siphoning off or run -off might be reversed.

Taylor, Atlanta: So we have done a small group of women that have had hysterectomies and in fact they look very much like men; they have very large flow oscillations and we haven’t studied them sequentially but just in a cross-sectional study. They have flow oscillations with about 15% flow reversal.

Annex, Charlottesville: Terrific talk; there are data in endothelial cells that endothelial cell receptors are not uniform across the endothelium and, in fact, they partition based on the directionality of flow and when you experimentally reverse flow, you can actually alter those dynamics…I am curious if you have had any chance to look at that because it would be an interesting way to consider what the longer term implications are of that reversal of flow.

Taylor, Atlanta: So we have done a little bit of that in a mouse model and looked at an ang II receptor expression. Ang II receptor expression is turned on by oscillatory flow and we were not able to reverse that but you are able to induce it…obviously in people it is a bit harder to do but in the animal models we can show that AT1, for example, is turned on.

Michael Gershon, New York City: Very nice talk Bob….so according to your model, I guess you would infer that if you have increased resistance below the legs with peripheral vascular disease that that would increase the propensity to abdominal aortic aneurysm and if you increase resistance in the renal arteries that that would decrease the propensity to abdominal aortic aneurysm; is there any evidence to support that?

Taylor, Atlanta: So I don’t know about the renal artery part but in terms of the peripheral vasculature, we have actually studied that a lot and we have looked at the ultimate increase in peripheral resistance which is amputation… So as you probably know there is some old data from the 1940s and 1950s in post–World War II veterans with increased risk of abdominal aortic aneurysm after amputation. We thought well maybe that is due to this high resistance….we have done a model in that we put a cuff on people, nice healthy cardiology fellows, and it dramatically increases their flow oscillations and we have studied a small cohort of three recent amputees and they have very, very high flow oscillations. I think you can extend that to peripheral vascular disease and of course we always see a lot of peripheral vascular disease patients with abdominal aortic aneurysm but, is it potentially mechanistically related at a level of mechanics other than the typical things.

Mackowiak, Baltimore: Very interesting….based on your slides, it looks like you did all of your studies in a horizontal position; what happens to flow in the vertical position?

Taylor, Atlanta: We don’t have a standing MR and I think that is an interesting question and maybe even more interesting, what happens in a sitting position? We spend a lot of our time sitting….and we change the geometry very dramatically so it is something we are interested in; we are trying to arrange a collaboration for that.

Alpert, Tucson: It would be difficult to do in humans because I think the human subjects committee might be a little concerned, but what about the gravid uterus you would expect to absolutely to have no flow reversal because you would have huge vasodilation but maybe you can see it in mice?

Taylor, Atlanta: We have thought about that a lot…. The problem is the geometry is so important that this is hard to study in the animals and that is why in the animals, we typically use the coarctation model. I think it is interesting and we have thought about it....and it turns out that in the U.S. you can’t do that, it turns out that in Europe you can actually do MRI on pregnant women.

LeBlond, Billings: I am getting way out of my depths here but I am wondering since the preservation of renal blood flows seems to be part of this dynamic, what happens when you transplant a kidney into the pelvis and what happens to the blood flow to that kidney versus the blood flow elsewhere below the native renal arteries?

Taylor, Atlanta: That is another good question…..we thought about that too….we have thought about scanning some of our transplant patients before and after renal transplant to see what the dynamics are because you would start potentially in some cases with very little renal flow and then add in renal flow distally but I think it’s a really good point….and you are right…sort of the driving force here is probably, if you think about it, by design is to preserve renal blood flow.

Howell, Ann Arbor: Some years ago there was an instrument called the ballistocardiogram that was used as well to measure the movement of the body as the blood flow moved down towards the feet and then came back up towards the head….and I just wonder, I doubt if there are any ballistiocardiograms around, but there is certainly are a lot of tracings left…if that is an instrument that could provide some additional insight into your hypotheses.

Taylor, Atlanta: It might be interesting….there has been a little bit of rebirth actually in ballistic cardiography and it is potentially interesting in a way of measuring volume changes…good point…thank you.