The relationship between shear stress and flow-mediated dilatation: implications for the assessment of endothelial function

Abstract

Introduction

The vascular endothelium is a single layer of cells lining all of the blood vessels in the body, and over the past three decades it has emerged as a key player in vascular growth, vasoregulation and vasoprotection. During this period of discovery it was found that the endothelium is essential for vasodilatation in response to increases in blood flow-associated shear stress (Smiesko et al. 1985; Pohl et al. 1986; Rubanyi et al. 1986). This phenomenon has since been termed endothelium-dependent flow-mediated vasodilatation (FMD) and now forms an important branch of endothelial research.

Several vasodilators are released by the endothelium in response to shear stress, including nitric oxide (NO), prostaglandins (PGI₂) and endothelium-derived hyperpolarizing factor (Joannides et al. 1995; Okahara et al. 1998; Busse et al. 2002). NO is of particular interest to researchers as it is an antiatherogenic molecule, and a reduction in its bioavailability may play a role in the pathogenesis of vascular disease (Cooke & Dzau, 1997). The goal of many human studies is to create a shear stress stimulus that produces an NO-dependent response in order to use the FMD measurements as an assay of NO bioavailability (a combination of NO production by the endothelium and destruction by reactive oxygen species). A small FMD response is interpreted as indicating a low NO bioavailability and possibly an associated increased risk of vascular disease or cardiac events.

In humans FMD is typically assessed in the large peripheral conduit arteries (brachial, radial and femoral) and is often taken to represent the response in the more clinically relevant coronary circulation (Anderson et al. 1995; Takase et al. 1998). These peripheral conduit artery FMD studies create a shear stress stimulus by causing dilatation (lowering the vascular resistance) in the vascular bed that is supplied by the conduit artery. Originally created by Celermajer et al. (1992) the ‘reactive hyperaemia test’ is the most popular technique. In its most basic form the test employs a temporary limb occlusion in order to create an ischaemia-induced reactive hyperaemia and a corresponding increase in shear stress in the conduit artery feeding the ischaemic territory. Under specific conditions this stimulus has been shown to elicit a primarily NO dependent FMD response (Joannides et al. 1995). Different occlusion durations, cuff positions, degrees of ischaemic dilatation (with the addition ischaemic forearm exercise) or areas of the circulation examined (upper versus lower limb) result in distinct shear stress profiles created upon occlusion release (Joannides et al. 1995; Leeson et al. 1997; Mullen et al. 2001; Betik et al. 2004). In addition to variations of the reactive hyperaemia method, some studies increase shear stress in the feeding conduit artery via hand warming, or infusing the vasodilator acetylcholine into the forearm circulation (Mullen et al. 2001; Joannides et al. 2002). These latter techniques result in a shear stress profile that increases gradually and reaches a steady plateau. This is in sharp contrast to the large transient profile created with reactive hyperaemia.

Can it be safely assumed that the FMD response to all of these various stimulus profiles similarly reflects NO bioavailability? The latest evidence indicates that it cannot, as mechanisms of FMD in response to certain shear stress stimulus profiles have been shown to be primarily independent of NO (Mullen et al. 2001). This suggests that there are multiple mechanisms involved in FMD and that the mechanisms involved in any given response are highly sensitive to the nature of the stimulus imposed (stimulus response specificity). Hence, if the goal is to use FMD as an assay of NO bioavailability it is critical to create a stimulus which is known to evoke a primarily NO-dependent response. Unfortunately, guidelines in the current literature provide an inadequate description of and rationale for these critical conditions that are required to evoke a reliably NO-dependent FMD response (Corretti et al. 2002). Importantly, while this mechanistic isolation of NO is desirable in some circumstances, the other mechanisms of FMD should not be ignored from either a clinical or a basic research perspective.

The purpose of this review is to (1) explain the concept of stimulus response specificity and how it dictates the critical importance of the stimulus creation technique in determining the mechanisms of the response, (2) provide an update to the current guidelines for FMD assessment, specifically focusing on stimulus quantification and describing the stimulus creation technique essential to producing primarily NO-dependent FMD responses, and (3) summarize the issues that surround the clinical utility of measuring both NO- and non-NO-mediated FMD.

How does the nature of the stimulus profile affect the mechanisms of the FMD response?

Stimulus response specificity refers to the concept that the nature of the stimulus (e.g. rate of onset, magnitude and duration, etc.) is essential in determining the nature of the response. Importantly, the nature of the shear stress may affect the response not only quantitatively, but also mechanistically. This critical point implies that (1) there are multiple mechanisms involved in FMD, and (2) that the mechanisms involved vary according to the imposed shear stress profile. Thus, FMD, often thought of as synonymous with NO bioavailability, may in fact reflect the function of different vasodilatory pathways depending on the shear stress experienced.

The concept that shear stress can stimulate the endothelial release of several vasoactive substances is not a new one. The presence of a multimechanism response has been demonstrated in microvessels and cultured cells, which may release NO (Koller et al. 1995), prostaglandins (PGI₂) (Koller et al. 1993), EDHF (Busse et al. 2002), endothelin (ET-1) (Kuchan & Frangos, 1993) and acetylcholine (ACh) (Martin et al. 1996) in response to shear stress elevations. Although these observations come from isolated or in vitro preparations, they discourage one from making the assumptions (1) that all conduit vessel responses are mediated by a single mechanism, and (2) that that mechanism is NO.

Most clearly demonstrated in humans is that the duration of the stimulus is an important determinant of the mechanisms of the response. In agreement with the findings of Joannides et al. (1995) and Doshi et al. (2001), Mullen et al. (2001) found that the radial artery FMD in response to a brief shear stress stimulus created with 5 min of wrist occlusion was abolished with the infusion of the nitric oxide synthase (NOS) inhibitor N^G-monomethyl-l-arginine (l-NMMA), indicating NO dependence. However, when they performed a 15 min wrist occlusion, which resulted in the same peak, but a more prolonged hyperaemia, neither l-NMMA infusion nor inhibition of cyclooxygenase had an effect on the FMD response. Further, when they created a very prolonged, steady increase in shear stress with hand warming or distal ACh infusion the resultant FMD was similarly unaffected by NOS and cyclooxygenase inhibition. These results indicate that the FMD response to prolonged increases in shear stress is mediated by neither NO nor PGI₂. Mullen et al. (2001) also found that while individuals with hypercholesterolaemia demonstrated an impaired response to the brief elevation in shear stress, their response to longer duration elevations in shear stress was intact. This indicates that FMD in response to short and long duration shear stress stimuli are mediated by different mechanisms and that these mechanisms may be affected differently by disease.

In support of the existence of distinct response mechanisms for different stimulus durations, in the study by Mullen et al. (2001) the brief stimulus created with reactive hyperaemia (5 min ischaemia) and the prolonged stimulus created with hand warming yielded distinct peak stimulus versus peak response relationships. Hand warming resulted in a modest peak stimulus in comparison to reactive hyperaemia (velocity time integral of 0.11 versus 0.2 m), but elicited a significantly larger FMD response (9.7 versus 5.3% change in diameter). This agrees with the observations of Joannides et al. (2002) in response to hand warming and reactive hyperaemia, respectively.

Although the precise interaction between FMD mechanisms with a prolonged shear stress stimulus remains unclear, the above observations suggest that the mechanisms primarily responsible for the observed FMD response may change over time when the shear stress stimulus is prolonged. Although there are several possibilities, one potential model of FMD that explains the observations of Mullen et al. (2001) is one where there is a sequential recruitment of mechanisms. Specifically, when a shear stress stimulus is initiated, one response mechanism may be recruited immediately, and this is most likely nitric oxide. If the shear stress remains elevated, however, another mechanism(s) may eventually take over as the one(s) primarily responsible for the FMD. If the shear stress stimulus continues, still other vasodilatory mechanisms may also become involved (Fig. 1). Further research is required to test this hypothesis. If there is in fact a sequential recruitment of mechanisms as this model suggests, it becomes essential to isolate the time course of the involvement of each mechanism to better allow isolation and examination of the desired vasodilatory pathways.

Open in a separate window

Figure 1

Hypothetical schematic of temporal recruitment of mechanisms for FMD

If a shear stress stimulus is prolonged the mechanisms primarily responsible for the observed vasodilatory response may change. The time courses of the mechanisms for FMD (i.e. when one begins and when it ceases to contribute) are currently unknown. It is also unknown how many mechanisms may be recruited over time in response to a sustained stimulus.

Clinical researchers may question the importance of the discovery that the gradual increase in shear stress brought about by hand warming or distal ACh infusion is not NO mediated. These methods are employed less frequently than reactive hyperaemia, which is the tool of choice in clinical FMD research. However, the work of Mullen et al. (2001) highlights the fact that even reactive hyperaemia stimulus profiles cannot be relied upon to produce an NO-dependent response. Importantly, the difference in the stimulus profile with 5 versus 15 min of reactive hyperaemia was fairly minor yet it completely reversed the NO dependence of the response. This suggests that the conditions which allow for primarily NO-dependent FMD responses may be very narrow indeed, and that extreme care must be taken to conform to those conditions if the FMD response is going to be used to infer NO bioavailability.

There are several examples in the literature that clearly demonstrate a need for an improved understanding of the stimulus-specific nature of FMD mechanisms. Similar to increasing occlusion duration, ischaemic handgrip exercise prolongs the duration of reactive hyperaemia (Wendelhag et al. 1999; Agewall et al. 2001; Agewall et al. 2002; Betik et al. 2004). The prolonged hyperaemia results in a larger peak FMD response. The addition of ischaemic handgrip exercise to reactive hyperaemia protocols has been suggested as an effective means of increasing the typically small reactive hyperaemia and small FMD response seen in older individuals (Wendelhag et al. 1999; Agewall et al. 2002). In the study by Wendelhag et al. (1999) it is not appreciated that the larger peak FMD observed may not be primarily NO dependent and thus may not have the anticipated association with vascular health. The study by Agewall et al. (2002) actually investigated the NO dependence of FMD responses post-ischaemic handgrip exercise. They found that the area under the curve (AUC) of the dilatory response was significantly reduced with l-NMMA infusion and concluded that the response was at least partially NO mediated. However, the peak dilatory response (what is normally used as the index of endothelial function) was almost unchanged with NOS inhibition (∼9.5 versus∼8%). Although the authors conclude that the reduced area under the curve marks a partial NO dependence, the results clearly suggest that there is a large non-NO-mediated component to the response. This directly illustrates the tenuous connection between NO bioavailability and FMD in response to reactive hyperaemia with ischaemic handgrip exercise.

In addition to the duration of occlusion and the addition of ischaemic handgrip exercise, the position of the occlusion cuff is also an important determinant of the NO dependence of the response. Specifically, it depends upon whether the occlusion cuff position is proximal or distal to the area of FMD measurement. Proximal occlusion has been shown variably to result in a larger or a similar stimulus to distal occlusion (Berry et al. 2000; Agewall et al. 2001; Doshi et al. 2001; Betik et al. 2004). However, regardless of the comparative stimulus magnitude, proximal occlusion usually results in a larger dilatory response compared to distal occlusion (Berry et al. 2000; Agewall et al. 2001; Doshi et al. 2001; Betik et al. 2004). This use of proximal occlusion has been advocated on the grounds that a larger response makes it easier to detect subtle differences between groups (Berry et al. 2000). However, proximal occlusion introduces some confounding factors in that the area of measurement is itself ischaemic and undergoing a dramatic pressure change with occlusion. This may result in metabolic or myogenic sources of dilatation in addition to an FMD response. It has now been established that the mechanisms of FMD are occlusion cuff position dependent. Doshi et al. (2001) demonstrated that while the FMD in the brachial artery in response to distal occlusion-induced reactive hyperaemia was abolished with NOS inhibition, the response to proximal occlusion was only partially attenuated from ∼12% down to ∼8%. The authors correctly suggest that the FMD in response to distal occlusion is a better marker of NO-mediated endothelial function.

In summary, the mechanisms of the FMD response are highly sensitive to the nature of the imposed stimulus. Specifically, changing the duration of the stimulus by (1) using an occlusion duration that is longer than 5 min (Mullen et al. 2001), or (2) adding ischaemic handgrip exercise (Agewall et al. 2002) can result in FMD responses that are not NO dependent and therefore do not reflect NO bioavailability. In addition, proximal occlusion adds the complication of potentially direct metabolic or myogenically mediated vasodilatory responses resulting in an overall dilatation that may not be entirely flow mediated, or endothelial dependent. In addition, it has been directly shown that proximal occlusion results in dilatation that is not primarily NO mediated (Doshi et al. 2001). Results from studies that prolong the duration of hyperaemia or introduce non-shear stress stimuli for dilatation must not be taken as a reflection of NO bioavailability.

Guidelines for the assessment of primarily NO-mediated FMD in human conduit arteries

The following guidelines are not intended to provide another set of comprehensive, detailed technical instructions for FMD investigations. Rather, they may be considered as an update to the current guidelines (Corretti et al. 2002) providing a more solid conceptual rationale for the best way to create and measure a stimulus that will evoke an NO-dependent FMD. This section will first discuss some critical points for accurate stimulus measurement and response interpretation and then, based on the previous discussion of stimulus response specificity, detail the stimulus creation technique that provides the most reliable assessment of NO-mediated FMD.

Can volumetric flow be used as a quantification of the stimulus for FMD?

Despite the term endothelium-dependent flow-mediated dilatation and the ubiquitous measurement of flow in human studies (Anderson & Mark, 1989; Celermajer et al. 1993, 1994; Corretti et al. 1995; Sorensen et al. 1995; Uehata et al. 1997; Esper et al. 1999; Berry et al. 2000), shear stress is the established stimulus for FMD in both conduit arteries and microvessels (Melkumyants et al. 1989; Koller et al. 1993). Shear stress is directly related to the velocity and the viscosity of the blood but inversely related to the vessel diameter. Accurate measurement of shear stress is very difficult in vivo due to the pulsatility of the flow and the necessity for velocity and viscosity measurements very close to the vessel walls (Wootton & Ku, 1999). However in conduit arteries where the blood flow is generally laminar and unidirectional, shear stress may be estimated by the equation:

(Davies & Tripathi, 1993; Gnasso et al. 2001). Shear rate (shear rate = velocity/diameter), although limited by potential viscosity changes or between subject differences, can be an adequate surrogate measure (Joannides et al. 1995; Gnasso et al. 1996; Yashiro & Ohhashi, 1997; Duffy et al. 1999; Betik et al. 2004; Pyke et al. 2004).

In contrast with the frictional or drag force that shear stress represents, flow is a measure of the volume of blood passing through the vessel over time (Flow = velocity ×πr², where r represents the radius). In the same vessel or vessels with similar diameters, increases in flow and shear stress are well correlated. However, vessels with different diameters may have the same flow but substantially different levels of shear stress and thus a different degree of stimuli for FMD (Fig. 2). This is demonstrated by Pyke et al. (2004) who observed that when reactive hyperaemia tests were performed in a group of young healthy subjects with a wide range of baseline diameters (∼3–5 mm) there was no relationship between the peak shear rate and the peak flow (Fig. 3). Pyke et al. (2004) also observed that the shear rate created by reactive hyperaemia was inversely related to the baseline diameter while peak flow was directly related to baseline diameter. Thus small vessels experienced a larger shear rate but a smaller flow than bigger vessels. The smaller vessels had a larger FMD response, which is at least partially explained by their larger shear rate (Fig. 4). The larger response in the small diameter vessels does not appear to be the result of an intrinsically greater sensitivity because when shear rate was controlled to create a uniform stimulus in all vessels, FMD was no longer a function of baseline diameter (Pyke et al. 2004). If flow had been measured as the stimulus it would have led to confusion when trying to interpret why small vessels experienced a smaller flow but had a larger dilatory response than large vessels. Many studies have in fact noted an inverse relationship between vessel size and FMD (Celermajer et al. 1992; Anderson et al. 1995; Herrington et al. 2001) but the mechanism (that smaller vessels experience a larger stimulus) has not been reported. The importance of this effect of vessel calibre is underscored by observations that baseline diameter is increased in the elderly, and tends to be smaller in women and patients with spinal cord injury (Herrington et al. 2001; Levenson et al. 2001; de Groot et al. 2004).

Open in a separate window

Figure 2

In vessels with different diameters the same flow may represent a very different shear stress stimulus

Open in a separate window

Figure 3

Peak flow versus peak shear rate demonstrating no relationship

Data from reactive hyperaemia tests performed in 8 healthy subjects (2 tests on 2 separate days) from the data set published in Pyke et al. (2004). These data illustrate that flow and shear rate cannot be used interchangeably as the stimulus for FMD when the subject pool has a range of baseline diameters.

Open in a separate window

Figure 4

Relationship of baseline diameter to bloodflow, shear rate and FMD

A, baseline diameter versus peak shear rate demonstrating an inverse relationship. B, baseline diameter versus peak flow demonstrating a direct relationship. C, baseline diameter versus peak FMD response demonstrating an inverse relationship. Data from reactive hyperaemia tests performed on 8 healthy subjects (2 tests on 2 separate days) from the data set published in Pyke et al. (2004).

In summary, when examining the FMD response across a range of vessel diameters (or between groups with significantly different diameters) flow does not reflect shear stress and thus should not be used to quantify the stimulus. In order to measure shear rate as the stimulus, continuous blood flow velocity and diameter measurements post-cuff release are essential. In addition, the phenomenon whereby small diameter arteries dilate more in response to reactive hyperaemia may be explained by the tendency for small diameters to experience a greater shear stress. In order to compare responses of different vessel diameters that received distinct stimulus magnitudes, response normalization is recommended (see the following section).

Is it important to normalize the response to the magnitude of the stimulus?

The magnitude of a given FMD response reflects not only the functionality of the endothelium, but also the magnitude of the stimulus imposed. The magnitude of the shear stress stimulus created with reactive hyperaemia is influenced by several factors and may differ significantly between groups or between individuals (Table 1). It is therefore essential that the magnitude of the stimulus imposed is considered when interpreting the FMD response and what it indicates about the functionality of the endothelium. FMD responses can be normalized by dividing the peak percentage change in diameter by the magnitude of the stimulus achieved with reactive hyperaemia. However, post-ischaemic hyperaemia creates a brief, constantly changing stimulus that peaks and decays considerably before the peak response is observed. The quantification that best describes the stimulus responsible for FMD has not been precisely defined. However, it has been demonstrated that both the peak and the duration of the stimulus are important in determining the peak FMD response (Joannides et al. 1997; Leeson et al. 1997). Therefore at this juncture it is recommended that authors report both the peak stimulus and the whole stimulus profile post-cuff release. If the whole stimulus is quantified then the area under the curve (AUC) of a specific time interval may also be calculated. In this way, the peak and/or the AUC can be used for normalization. Whether the peak or the AUC best represents the relevant stimulus for FMD is unknown at present. However, the AUC of the stimulus until the time of peak response measurement is arguably the best time interval to use (Fig. 5) as this is a quantification of the total stimulus that could have contributed to the development of the response. The time of peak response measurement is quite variable (Berry et al. 2000) and this necessitates continuous stimulus and diameter measurement for a minimum of 90 s post-cuff release.

Open in a separate window

Figure 5

Schematic diagram of the area under the curve (AUC) of the stimulus (shear stress or shear rate) until the time of peak diameter measurement

Continuous line, stimulus; dashed line, response.

What is the stimulus technique that most reliably creates a primarily NO-dependent FMD?

In human conduit arteries, an NO-dependent response has only been clearly demonstrated under the following conditions.

Clinical utility of assessing FMD

Brachial artery FMD is emerging as an independent predictor of future cardiac events (Gokce et al. 2002, 2003; Widlansky et al. 2003). However it is interesting to note that meta-analysis performed by Witte et al. (2005) found that FMD was only related to low and not medium or high cardiovascular risk. The current dogma suggests that the mechanisms of the predictive ability are as follows. A small brachial FMD response is a hallmark of systemic low NO bioavailability and an associated lack of vasoprotection in clinically relevant areas of the vasculature (coronary, carotid). The lack of vasoprotection is thought to result in the development of vascular disease (Cooke & Dzau, 1997). While the potential benefits of a non-invasive test that allows early risk identification cannot be ignored, the following section provides some caveats regarding the interpretation of correlational studies and the clinical utility of the current methodology of brachial artery FMD testing.

How variable are FMD responses?

Within any given study, reactive hyperaemia endothelial function tests can consistently demonstrate a smaller degree of dilatation in atherosclerotic/risk factor patients versus controls. Improvements in FMD in response to an intervention (e.g. exercise) have been demonstrated at the group level (Widlansky et al. 2003; Green et al. 2004); however, day to day variability in subject responses may limit the utility of using reactive hyperaemia testing to track an individual's improvement in response to an intervention (Hijmering et al. 2001). Studies using automated diameter measurement (shown to reduce the observer error source of variability; Woodman et al. 2001) have reported intersession coefficients of variation of ∼14% (Hijmering et al. 2001; Woodman et al. 2001). Further, there is a significant degree of variability between testing centres to the extent that the percentage change in diameter that characterizes a normal response in one study may represent an impaired response in another (Table 2) (Berry et al. 2000). A number of technical considerations may influence the magnitude of the peak FMD response and contribute to the interlaboratory variability. These include but are not limited to occlusion cuff position, duration of occlusion, occlusion pressure, and time of peak dilatation measurement. This at least partially unaccounted-for variability between studies makes it difficult to create standardized limits defining healthy versus pathological responses. If future studies adhere more closely to the procedure that elicits the most reliable NO-dependent FMD response (see stimulus creation technique section) some of this interlaboratory variability may be eliminated.

Do the stimulus creation conditions as stated in this review ensure that the FMD response reflects NO bioavailability in diseased populations?

As previously stated, the current emphasis is that the clinical utility of peripheral FMD testing hinges on FMD's NO dependence. It has been established that straying from very specific stimulus creation conditions can change the mechanisms of the FMD response (Mullen et al. 2001). However in certain populations the FMD may reflect the function of non-NO vasoregulatory mechanisms, specifically sympathetic activation and ET-1 activity, even when applying the shear stimulus creation procedure previously recommended. The persistent ability of FMD to identify and predict disease reinforces the notion that these mechanisms (in addition to NO bioavailability) play a role in the pathogenesis of vascular disease. This aetiological distinction is important to the development of treatments for endothelial dysfunction.

Elevated sympathetic activation is common in pathologies also associated with endothelial dysfunction (Esler et al. 2001; Palatini, 2001) and has been shown to blunt the FMD response (Hijmering et al. 2002). In groups with elevated sympathetic outflow, even when the proper technique is performed the resultant FMD response is a combined reflection of NO bioavailability and sympathetic activation. In support of this statement, treatments that alleviate heightened sympathetic activation have been shown to enhance FMD responses (Ip et al. 2004). Hence, in groups where elevated sympathetic activation is likely to be present, caution should be exercised when making the connection between FMD and NO bioavailability.

Endothelin (ET-1) is a powerful vasoconstrictor released by the endothelium that stimulates smooth muscle cell proliferation and platelet aggregation and antagonizes NO, making it ‘pro-atherogenic’ (Spieker et al. 2001). Elevated plasma ET-1 has been documented in both congestive heart failure and hypertension, both of which are also associated with endothelial dysfunction (Lerman et al. 1991; Rodeheffer et al. 1992). Importantly, in experimental models of hypertension, treatment with an ET-A receptor antagonist has been shown to ameliorate endothelial dysfunction (Wenzel et al. 1998; Best et al. 1999). More recently ET-A receptor antagonism has been shown to improve FMD in humans with chronic heart failure (Berger et al. 2001). Importantly FMD's NO dependence was established in healthy subjects (Joannides et al. 1995; Doshi et al. 2001; Mullen et al. 2001), in that their FMD was virtually abolished in response to l-NMMA. The above data suggest that in diseased populations with elevated ET-1 levels, high ET-1 rather than lower NO bioavailability may be responsible for their smaller (or absent) FMD responses.

How well does FMD in the brachial artery correlate with endothelium-dependent vasodilatation in the coronary circulation?

The most clinically significant vascular disease occurs in the coronary and carotid vessels. Intermittent claudication, which manifests as pain during exertion, can occur in the lower extremity (Ouriel, 2001). The vessels of the upper extremity, however, where most endothelial function testing is performed, are typically lesion free (Anderson et al. 1995). In order for brachial or radial responses to be clinically meaningful, endo-thelial dysfunction must be, at least to a certain extent, a systemic phenomenon. Specifically, a small peripheral FMD response, thought to indicate low NO bioavailability, must signify an associated lack of vasoprotection and FMD in the clinically relevant circulation. The first study to investigate the relationship between endothelium-mediated dilatation in the brachial and coronary arteries compared receptor-mediated dilatation (ACh infusion) in the left main coronary artery, and stimulated dilatation (FMD) in the brachial artery. It found a significant but modest correlation (r²= 0.36) (Anderson et al. 1995). Poor correlation between ACh-mediated dilatation and shear stress-mediated dilatation has also been observed in the brachial artery (Eskurza et al. 2001). A second study compared stimulated dilatation (FMD) in both a major coronary artery and the brachial artery and found a much stronger correlation (r²= 0.78) (Takase et al. 1998). This study was performed on a mixed group of subjects with stenotic and normal coronary arteries and varying risk factors. Future studies should focus on demonstrating the correlation between impaired coronary and brachial FMD in well defined pathological conditions.

In summary this suggests that (1) FMD in the brachial artery may be an adequate surrogate of FMD in the main coronary arteries, and (2) ACh receptor-mediated dilatation and FMD do not provide the same information about endothelial function. The latter point is important because many studies documenting coronary endothelial dysfunction have used ACh infusion methodology (Mathier et al. 1998; McDermott et al. 2001; Halcox et al. 2002) and dilatation in response to ACh infusion and shear stress are both taken as indexes of NO-mediated endothelial function. Which index provides a better reflection of clinically relevant NO bioavailability? Arguably FMD is more physiologically relevant, since shear stress and not ACh is the major stimulus for NO release in vivo. In addition, dilatation in response to ACh infusion is incompletely blocked by l-NMMA infusion, indicating that it is only partially NO mediated (Eskurza et al. 2001). This may help to explain the lack of correlation between FMD (shown under specific reactive hyperaemia conditions to be almost totally NO dependent) and ACh-mediated dilatation and suggests that FMD provides a better index of NO bioavailablity. Further, ACh infusion studies often measure dilatation in the microcirculation whereas FMD studies investigate conduit artery responses (Eskurza et al. 2001). The poor correlation between these techniques when this is the case could be in part because they test resistance versus conduit arteries, respectively. This therefore raises questions regarding the systemic nature of NO-mediated endothelial function. Further study is required to identify why ACh and FMD mediated dilatation are not correlated and to identify the clinical implications of the non-uniform response.

Is FMD only important because it reflects the level of vasoprotection?

Although NO's vasoprotective effects are important, improved FMD can confer direct clinical benefit irrespective of whether NO (or another vasoprotective mechanism) is the mechanism involved. FMD is an important part of the coronary response to exercise but individuals with atherosclerotic coronary arteries may experience an attenuated FMD response or even a vasoconstriction in response to increased flow (Duffy et al. 1999; Gielen et al. 2001). This exacerbates any reductions in flow that may have already been caused by structural vessel narrowing and prevents adequate oxygen delivery in times of stress or exercise. Conversely, improvements in FMD can lead to reductions in myocardial ischaemia that are not explained by lesion regression (Gielen et al. 2001; Gielen & Hambrecht, 2001). Therefore both the vasoprotective and vasoactive functions of the endothelium have clinical relevance. This emphasizes that investigations of non-NO-mediated FMD responses to physiologically relevant shear stress stimuli are also important.

Does the mechanism of FMD depend on the nature of the stimulus in the coronary circulation?

Within the radial and brachial arteries, the mechanisms of the FMD response have been shown to be stimulus profile specific. Specifically, long duration stimuli have been demonstrated to evoke non-NO-dependent FMD responses (Mullen et al. 2001; Agewall et al. 2002). It cannot be ruled out that the coronary artery endo-thelium shares this stimulus response specificity. In the best demonstration of similar endothelial function in brachial and coronary arteries the shear stress profile in the coronary artery was created with ATP infusion (Takase et al. 1998). This created a stimulus that remained elevated for 1–2 min. It is possible that this stimulus, more prolonged than that created with reactive hyperaemia, evoked a non-NO-mediated response. The results of Shiode et al. (1996), who found that FMD in response to a long duration elevation in shear stress in the coronaries is not necessarily NO mediated, support this possibility. Future research needs to be geared towards establishing the level of stimulus response specificity that exists in the coronary circulation.

Do we know the role of non-NO FMD mechanisms in the development of vascular disease?

Endothelium-dependent vasodilatation is so firmly linked to NO that the other endothelial vasodilatory factors have been largely ignored in human studies. Several lines of evidence suggest that this neglect is ill-advised. To name a few: (1) PGI₂ and ET-1 are both released by the endothelium in response to shear stress and have anti- and pro-atherogenic properties, respectively (Niebauer & Cooke, 1996; Spieker et al. 2001). (2) EDHF is known to be released by endothelial cells in response to shear stress but little is known about its physiological importance in humans (Campbell & Gauthier, 2002). As such, it cannot be ruled out as a player in vascular pathologies. (3) While an impaired NO-mediated FMD response has been demonstrated fairly universally in many disease states (Celermajer et al. 1992, 1993; Celermajer et al. 1994; Hornig et al. 1996; Mullen et al. 2001), an intact non-NO-mediated response has only been shown in one study in one disease state (hypercholesterolaemia) (Mullen et al. 2001). These results should be confirmed and further testing should investigate non-NO-mediated FMD responses in other human diseases (e.g. atherosclerosis, hypertension, chronic heart failure and diabetes). In summary, a systematic investigation of every potential mechanism involved in determining FMD, looking for potential links to vascular disease, is required.

Conclusion

It appears that in peripheral conduit arteries the mechanisms of the FMD response are highly sensitive to the nature of the shear stress stimulus. In order to treat FMD as an assay of NO bioavailability the conditions for stimulus creation outlined in this review should be followed. In order to treat FMD as an assay of NO bioavailability the conditions for stimulus creation outlined in this review should be followed. If future studies adhere to the procedure that elicits a reliably NO-dependent FMD response, some of the interlaboratory variability that has hampered FMD studies thus far may be eliminated. Isolating the function of different mechanisms of FMD (both NO and non-NO) is important to allow a clear examination of their relative importance to vascular disease and vasoregulation. Future studies should be geared towards identifying exactly what dilatory mechanisms are at work, in what proportions and over what time courses, in response to specific shear stress stimulus profiles. This will allow experimenters and clinicians to design tests that are better able to isolate and investigate specific vasodilatory pathways associated with FMD. The improved assessment techniques and interpretive ability yielded by increasing our understanding of stimulus response specificity may help to resolve apparent conflicts in the literature and bring us closer to standardization and clinical application of endo-thelial function tests.

Acknowledgments

References