The Assessment of Endothelial Function – From Research into Clinical Practice

Abstract

The discovery of the endothelium as a crucial organ for the regulation of the vasculature to physiological needs and the recognition of endothelial dysfunction as a key pathological condition - which is associated with most if not all cardiovascular risk factors - led to a tremendous boost of endothelial research in the past 3 decades. Despite the possibility to measure endothelial function in the individual and its widespread use in research, its use as a clinical tool in daily medicine is not established yet. We review the most common methods to assess vascular function in humans and discuss their advantages and disadvantages. Furthermore we give an overview about clinical settings were endothelial function measurements may be valuable in individual patients. Specifically, we provide information why endothelial function is not only a risk marker for cardiovascular risk but may also provides prognostic information beyond commonly used risk scores in primary prevention, and in patients with already established coronary disease.

We conclude, that non-invasive endothelial function measurements provide valuable additional information, however, to ascertain its use for daily clinical practice, future research should determine whether endothelial function can be used to guide treatment in the individual and if this translates into better outcomes.

Introduction

The discovery of nitric oxide (NO) as a crucial endothelium-derived molecule for vascular relaxation and the recognition of the endothelium as more than a passive interface between blood and the vessel wall, led to substantial progress in the field of vascular research.¹ Endothelial dysfunction is a pathological condition, mainly characterized by an imbalance between substances with vasodilating, anti-mitogenic and anti-thrombogenic properties (endothelium-derived relaxing factors)² and substances with vasoconstricting, pro-thrombotic and proliferative characteristics (endothelium-derived contracting factors).³ Among the most important vasodilator molecule, in muscular arteries in particular, is NO, which also inhibits other key events in the development of atherosclerosis such as platelet adhesion and aggregation, leukocyte adhesion and migration as well as smooth muscle cell proliferation. Particularly in the microcirculation, prostacyclin and endothelial-derived hyperpolarization factors (EDHF; an umbrella term for substances and signals hyperpolarizing vascular myocytes by opening voltage channels⁴) also play an important role.

Generally, loss of NO bioavailability indicates a broadly dysfunctional phenotype across many properties of the endothelium. Thus, the assessment of its vasodilator properties due to NO and other molecules may provide information on the integrity and function of the endothelium. Interestingly most, if not all, cardiovascular risk factors are associated with endothelial dysfunction,⁵ and risk factor modification leads to improvement in vascular function. Endothelial dysfunction has been detected in the coronary epicardial and resistance vasculature as well as in peripheral arteries, so that endothelial dysfunction can be regarded as a systemic condition.⁶ Importantly, the process of atherosclerosis begins early in life, and endothelial dysfunction contributes to atherogenesis and precedes the development of morphological vascular changes.⁷

Over the past 25 years many methodological approaches have been developed to measure the (patho-)physiological function of the endothelium in humans.⁸ Although the ability of measuring endothelial function has boosted clinical research in this field, its use as a clinical tool in daily practice is not established, nor has any method been recommended in clinical guidelines for planning primary or secondary prevention of vascular disease.

It is the aim of this review to give a short overview of the most commonly used methods to measure endothelial function in humans, non-invasive techniques in particular (Table 1); to summarize the clinical implications of endothelial dysfunction in the population as well as in individual patients. The possible future role of endothelial function measurement for individualized medicine is also considered.

Table 1.

Advantages and disadvantages of the most commonly used techniques to assess endothelial function

Technique	Vascular bed	Advantages	Disadvantages	Stimulus (examples)
Coronary epicardial vasoreactivity (QCA)	Epicardial macrovascular Conduit arteries	Assessment directly in the coronary vascular bed Gold standard	Invasive Expensive Time intensive Limited to those undergoing coronary angiography Challenging for serial measurements	Ach Exercise Pacing CPT
Coronary microvascular function – Doppler wires	Coronary microvascular Resistance arteries	Assessment directly in the coronary microvasculature	Invasive Expensive Time intensive Limited to those undergoing coronary angiography Challenging for serial measurements	Ach Adenosine Papaverine
Flow- Mediated- Dilation (FMD)	Brachial Artery Conduit artery	Easy access Correlation with invasive epicardial vascular function Many outcome studies Inexpensive Possibility to assess other important parameters (flow, baseline arterial diameters, FMC)	Challenging to perform well Disparate protocols for performance and standardizations Need for standardization	Reactive Hyperemia
Venous occlusion plethysmography	Forearm vasculature Microvasculature	Easy access Vasoactive substances infused to generate a dose-response relationship Contralateral arm as a control	Invasive (cannulation of the brachial artery) Time consuming	Ach and other vasoactive substances
EndoPAT	Finger Microvasculature	Easy to access and perform Automated Low inter- and intraobserver variability Correlation with invasive microvascular vascular function	Expense of disposable finger probes PAT signal influenced by variable non endothelial factors	Reactive hyperemia

METHODS TO ASSESS VASCULAR FUNCTION

The first demonstration of endothelial dysfunction in atherosclerotic coronary arteries using intracoronary infusion of acetylcholine and quantitative coronary angiography dates back to 1986 by Ganz, Selwyn and colleagues.⁹ Their seminal studies heralded an important shift in paradigm in understanding of human atherosclerosis, which had previously been regarded as a purely structural disease. Their research drew attention to the functional manifestations of atherosclerosis, such as exaggerated vasoconstriction, as a consequence of poorly functioning endothelium. Later, less invasive techniques were developed mainly using the forearm circulation as a surrogate for coronary arteries.^{6, 10, 11} All approaches have their advantages and disadvantages and most importantly different vascular beds are examined (Figure 1). The basic principle, however, is similar: healthy arteries such as the coronary or brachial arteries dilate in response to reactive hyperemia (flow-mediated vasodilatation) or after pharmacological stimuli including intra-arterial infusion of endothelium-dependent vasodilators such as acetylcholine (Ach), bradykinin or serotonin, via release of NO and/or other endothelium-derived vasoactive substances.² In disease states, such endothelium-dependent dilatation is reduced or absent. However, whichever technique is applied, vascular responses are not only determined by the functional status of the vasculature at the place of measurement, but also by the structural condition of the resistance arteries in the microvasculature. Furthermore, to differentiate endothelium-dependent from endothelium-independent responses, exogenous NO donators (e.g. glycerol-trinitrate) or direct non NO donators such as adenosine can be applied. Impaired endothelial-independent function is associated with structural vascular alterations and alterations in smooth muscles cells rather than changes in the endothelium.

Figure 1.

Figure 1 depicts the principles of the most commonly used methods for the assessment of endothelial function.

Coronary epicardial and microvascular function

To assess coronary endothelial function, a “functional” test is performed to measure epicardial as well as resistance vessel endothelial function. Although these methods are limited by the invasive nature, their advantage is to measure endothelial function directly in this clinically important vascular bed.

Epicardial endothelial function

To image vasomotor responses of epicardial coronary arteries, quantitative coronary angiography (QCA) or intravascular ultrasound (IVUS) is used and changes in vessel diameters and cross-sectional areas in response to endothelium-dependent interventions are documented. After acetylcholine infusion, vessels and segments with an intact endothelium vasodilate, whereas vessels and segments with dysfunctional or disrupted endothelium, will respond with vasoconstriction due to a direct activation of muscarinic receptors on vascular smooth muscle cells.⁹ Similar induced functional changes in vascular reactivity have been demonstrated with salbutamol¹² and other substances (Table 2) and with more physiological interventions, for example increased coronary blood-flow. However, dose titration is more difficult. Physical measures of endothelium-dependent responses include exercise,^{13, 14} or pacing induced tachycardia as a surrogate for exercise,^{15, 16} and induce an increase in coronary blood flow and thus shear stress on the coronary circulation, which leads to flow mediated endothelium-dependent vasomotion of the epicardial vessels. Similar responses can be seen in response to mental stress.¹⁷ The observation of endothelium-dependent flow mediated dilatation in the coronary epicardial vessels and its impairment in atherosclerosis^{18, 19} provided the rationale to study similar responses in the peripheral vasculature later (see below). Another “physiologic” test to assess epicardial vasoreactivity is the use of the cold pressor test (CPT), where subjects put their hand into ice water. The activation of the sympathetic nervous system leads to release of NO and endothelium-derived hyperpolarizing factors (EDHFs) via stimulation of endothelial α₂-adrenergic receptors and consequently vasodilation in healthy.²⁰ However, in dysfunctional endothelium, α₁-adrenergic mediated constriction of smooth muscle cells will dominate,²¹ closely mirroring the responses to Ach. ^{21, 22}

TABLE 2.

Pharmacological triggers for the assessment of coronary vascular function

	Epicardial vessels	Microcirculation
Endothelium dependent vascular function	Acetylcholine Salbutamol Serotonin Substance P Calcitonin gene-related peptide	Acetylcholine Salbutamol Bradykinin
Endothelium independent vascular function	Nitroglycerin Nitroprusside Papaverine	Adenosine Dipyridamole Nitroprusside Papaverine

Coronary microvascular function

Changes in coronary (or myocardial) blood flow (CBF) can be used as a surrogate parameter for microvascular function.²³ Coronary flow reserve (CFR) is the ratio of maximal CBF during maximal coronary hyperemia with provocative stimuli (such as adenosine infusion, pacing or exercise), divided by the resting CBF. This maximal blood flow response (CFR) is both endothelium and non-endothelium dependent and a CFR below 2.0 is considered abnormal.²⁴ To measure endothelium-dependent microvascular function, the percent increase in CBF in response to endothelium-dependent vasodilators (commonly Ach) infused at increasing concentrations is analyzed.

Other methods to estimate microvascular function have been introduced, as for example the measurement of the number of cineangiographic frames that it takes to fill a distal vessel with proximal injection of contrast. The corrected TIMI (Thrombolysis in Myocardial Infarction) frame count (CTFC) provides a semiquantitative assessment of epicardial coronary blood flow.²⁵

Taking the main disadvantage into account – the invasive nature of the above mentioned tests – non-invasive functional tests to assess the coronary microvasculature have been developed, among them positron emission tomography,²⁶ myocardial perfusion imaging,²⁷ blood oxygen level-dependent (BOLD) MRI²⁸ and echocardiography,²⁹ however, in detail discussion would go beyond the scope of this review.

Peripheral techniques to assess endothelial function

The aforementioned techniques to measure both coronary epicardial vascular function and the assessment of the coronary microcirculation are very well suited for patients requiring a coronary angiogram for clinical indications. However, to assess vascular function and health in the asymptomatic patient, performing an invasive functional coronary angiogram is usually not appropriate. Therefore non- or less invasive surrogate techniques to assess macrovascular as well as microvascular endothelial function have been developed. Although these do not measure vascular function in the coronary circulation directly, they have been shown to correlate reasonably with its more invasive counterparts.^30–32 Whereas all these techniques assess the generalized function of the vasculature, one has to keep in mind that certain phenomena cannot be explained by systemic endothelial dysfunction; it is likely that local factors (e.g. flow patterns) and local vascular dysregulation observed at branch points related to disturbed shear stresses also contributes to disease.^33–37

Plethysmography of the Forearm Circulation

Although still limited by its semi-invasive nature (arterial puncture), with this technique changes in forearm blood flow are measured by venous plethysmography in both arms before and after infusion of vasoactive substances into a cannulated brachial artery.¹⁰ The main advantage is that vasoactive molecules, hormones or drugs (for instance Ach or nitroglycerin) can be infused, thus respectively quantifying endothelium-dependent and endothelium-independent vasodilation, in a dose-dependent manner. The dosages required have limited systemic effects, allowing the contralateral limb to serve as an internal control. The results are expressed as the ratio of the changes in flow measured in both arms and are reproducible.³⁸ The response to acetylcholine is significantly reduced by intraarterial infusion of L-NMMA (but not by acetylsalicylic acid),³⁹ demonstrating a key role for NO. However, it has to be taken into account that, especially in patients with multiple risk factors, EDHF also play an essential role for resting microvascular tone,⁴⁰ as well as for agonist-stimulated vasodilation.^{41, 42} The technique is well suited to measure differences in blood flow to various stimuli or inhibitors in a single patient, however, due to different initial arterial pressure, forearm blood flow, different sizes of the forearm and other factors, comparisons between groups or serial studies in the same patient are of limited value.⁴³ Although pharmacologically induced vasodilation with this technique gives interesting insights into microvascular pathophysiology, the response not necessarily mimics microvascular vasodilation to transient ischemia or exercise.

Flow-mediated vasodilation of brachial artery

Due to its non-invasive approach, flow-mediated vasodilatation of the arm arteries (FMD) has become the most widely used technique to measure endothelial function. The technique measures the ability of the arteries to respond with endothelial NO release during reactive hyperemia (flow-mediated) after a 5-minute occlusion of the brachial artery with a blood pressure cuff. Celermajer, Deanfield and their colleagues were the first to measure this response in vivo by measuring the respective diameter changes of the brachial or radial artery by ultrasound¹¹ a response later demonstrated to be mainly NO dependent,^{39, 44, 45} although other vasodilator pathways may contribute as well.⁴⁶ Importantly, peripheral endothelial function as assessed by FMD correlates with coronary artery endothelial function.^{30, 32} However, although the principle of this technique seems to be simple, its application is technically challenging and requires extensive training and standardization (Table 3).^47–49 Study preparation, image acquisition and site selection, sphygmomanometer probe position, cuff occlusion time, the accurate use of edge-detection software as well as the correct characterization of the FMD response are crucial, as recently outlined in detail in guidelines by Charakida et al.⁴⁹, Harris et al.⁵⁰ and Thijssen et al.⁵¹ These publications are useful as they draw attention to the need to standardize the different protocols and indeed, if efforts to standardize the technology are followed, reproducibility of FMD can be considerably improved.⁵²

TABLE 3.

Technical considerations in FMD measurements

Subject preparation fasting state (>6h) no smoking or any tobacco consumption at least 6h before study no exercise or food/beverages that contain alcohol, caffeine or are rich in polyphenols (cocoa, tea, fruit juices for >12h no vitamins for at least 72h withheld vasoactive medications on the morning of the study if possible and carefully note use and timing of any drugs no exercise >12h prior to test quiet, temperature-controlled room in females, repetitive studies should be made at the same time of menstrual cycle (ideally assess on days 1–7 of the menstrual cycle) rest for at least 10 minutes prior to measurements supine position arm resting comfortable with cradle support with the imaged artery at the heart level test should be performed at the same time of the day (especially if multiple tests are performed)

Sphygmomanometer probe position and cuff occlusion time placement of the cuff 1–2 cm distal to the elbow crease other sites are discouraged because proximal cuff postioning affects the magnitude of the peak vasodilatory response occlusion time: 5 minutes (shorter inflation attenuates FMD response) cuff inflation to at least 50 mmHg above systolic pressure

Site selection brachial artery with a minimum diameter (usually >2 mm). Small arteries are difficult to measure and changes in absolute diameter correspond to big relative changes if repetitive measurements are planned, site has to been replicated -anatomic landmarks should be used.

Image acquisition longitudinal images obtained by high-resolution ultrasound (7.5–12 MHz) a clear interface between the near and far arterial wall should be achieved diameter measurements are obtained in end-diastole or averaged over the heart cycle stereotactic adjustable prop-holding is essential to endure image quality recording of the baseline diameter for at least 1 min simultaneous acquisition of pulse-wave Doppler velocity signals for quantification of shear stress (stimulus), if feasible. Insonation angle should be <60°

Measurement automated edge-detection should be used reported as maximal percentage change from baseline diameter (most reproducible) report also baseline diameter, absolute change characterization of the hyperemic stimulus (ideally the flow-velocity time integral)

Sources of Information:^49–51

Recently, some new aspects of this technique have emerged and are under investigation. While FMD measures conduit artery vascular function, the stimulus for FMD itself (reactive hyperemia flow and the induced shear stress on the endothelium) might be an important measure of peripheral microvascular function because reactive hyperemia is highly dependent on maximal forearm resistance.^{53, 54} Notably, both hyperemia induced shear stress and velocity changes (measured by calculation of the velocity-time integral, adjusted for heart rate) showed even stronger correlations with the presence of cardiovascular risk factors than FMD⁵⁵ and also predict cardiovascular outcomes.^{56, 57}

Interestingly, some recent multi-center studies demonstrated that simple baseline brachial artery diameter readings correlate with clinical outcomes, nearly as well as the FMD itself.^{58, 59} One explanation for this intriguing finding might be that the larger the diameter of an artery, the smaller the relative percent changes (FMD) and, additionally, shear stress generated seems to be lower in larger vessels.^{60, 61} As shear stress is the stimulus for FMD, it seems reasonable to correct for the impact of shear stress on FMD. However, several methodological as well as physiological factors influence shear stress and ratio normalizations of the dilatory response to shear rate have provided conflicting results.⁵¹

Recently, Gori et al.⁶² raised the issue of endothelial function in the resting (not hyperemic) state. While FMD provides crucial information about the ability of the endothelium to respond to a specific stimulus (reactive hyperemia), it is not a measure of the resting production of vasoactive substances. In this respect vasoconstriction of the brachial artery after inflation of a wrist cuff to suprasystolic was first reported years ago,⁶³ and is mainly mediated through vasoconstrictor substances such as endothelin.⁶⁴ This concept has garnered new attention and the term “low-flow-mediated constriction (L-FMC) was introduced.^{62, 65, 66} In principle, L-FMC detects the change in brachial artery diameter in response to a decrease in blood flow and shear stress, respectively, after occlusion of the artery by a distally placed cuff. In a recent study an association between traditional risk factors and impaired L-FMC as well as a relationship with the severity of coronary artery disease has been demonstrated.⁶⁷

Taken together, the information obtained by functional vascular ultrasound is manifold and future protocols should not only measure percent changes in arterial diameter in response to hyperemia but should consider also the above mentioned measurements.

Finger Plethysmography

Recently measuring endothelial function using peripheral arterial tonometry (PAT) has gained increasing attention and a proprietary device has been developed to measure observer independent pulsatile arterial volume changes by finger plethysmography (EndoPAT, Itamar Medical).^{68, 69} With PAT beat-to-beat plethysmographic recordings of the finger arterial pulse wave amplitude with pneumatic probes are captured. With the device a counterpressure of 70 mmHg on the digit is applied to avoid distal venous distention thus inhibiting venous pooling and venoarteriolar reflex responses.⁶⁸ In principle, an increase in arterial blood volume in the finger tip causes an increase in pulsatile arterial column changes thus increasing the measured signal. Similar to the assessment of endothelial function with the FMD technique, a pressure cuff is placed on the arm and after obtaining baseline blood volume changes, the blood pressure cuff is inflated above systolic pressure and deflated after 5 minutes to induce reactive hyperemia on one arm. A main advantage of the system is that the contralateral arm serves as its internal control that can be used to correct for any systemic drift in vascular tone during the test and an index between the two arms is calculated to adjust for any such drift. This index is validated marker for endothelial function; however, augmentation of the pulse amplitude after reactive hyperemia is a complex response to ischemia. It reflects changes in flow, as well as in digital microvessel dilatation and is only partly dependent on nitric oxide.⁷⁰ Validation studies demonstrated that impairment in peripheral finger endothelial function measured with EndoPAT is correlated with coronary microvascular function in patients with early atherosclerosis³¹ and predicts cardiovascular events.⁷¹ Two large cross-sectional studies (in over 1900 patients in the Framingham cohort^{72, 73} and over 5000 individuals in the Gutenberg Heart Study⁷⁴) digital vascular dysfunction was associated with traditional and metabolic cardiovascular risk factors but not or only modestly with FMD, thus likely measuring different aspects of vascular biology (see below).

CLINICAL IMPLICATIONS OF ENDOTHELIAL DYSFUNCTION IN POPULATIONS AND IN THE INDIVIDUAL

Is endothelial function a marker for cardiovascular risk?

In the coronary arteries, impairment of endothelial function occurs early in the course of atherosclerosis, in relation to systemic risk factors⁷⁵ and abnormal hemodynamic shear stresses.³⁷ The more systemic cardiovascular risk factors present, the worse epicardial vascular function.⁷⁵ There is extensive literature documenting that endothelial dysfunction is associated with almost every condition predisposing to atherosclerosis and cardiovascular disease.⁷⁶ There are many studies correlating endothelial dysfunction (conduit artery and microvasculature likewise) with cardiovascular risk. For example, endothelial dysfunction has been observed in patients with arterial hypertension,^77–79 in normotensive subjects with a family history of hypertension,⁸⁰ in smokers,^{81, 82} as well as passive smokers,^{83, 84} in dyslipidemia,^{85, 86} in ageing,⁷⁸ diabetes mellitus,^87–91 in obesity,⁹¹ in hyperhomocysteinaemia,^{92, 93} in humans with low intracellular magnesium levels⁹⁴ and in patients with inflammatory or infectious diseases.^95–97 Importantly, the effects of cardiovascular risk on the endothelium can be seen in children as early as 8 years of age.^{98, 99} Thus, endothelial dysfunction may represent the effect of these risk factors on vascular health.

The fact that endothelial dysfunction is a systemic condition⁶ may explain why peripheral endothelial function (microvascular and macrovascular) correlates with endothelial function in the coronary arteries.^{30, 31, 32} The pathophysiology behind the functional changes in impaired endothelial function also leads to structural changes of the vessel over time. In a cross-sectional study in middle aged healthy men, there is no evident correlation between brachial FMD and the carotid intima media thickness (IMT),¹⁰⁰ however, in a similar population free of cardiovascular disease, FMD correlated with IMT progression over a 6 years follow up. Interestingly, in this study, in contrast to FMD, Framingham Risk was not correlated with IMT progression.¹⁰¹ Similar, FMD also predicted IMT progression in less healthy patients as demonstrated after 1 year in hypertensive, postmenopausal women.¹⁰²

Taken together, there is good evidence that endothelial dysfunction is significantly associated with the burden of cardiovascular risk and can be considered as a barometer of the total risk burden (the risk of the risk factors). However, transient endothelial function impairment, for example by intercurrent acute illnesses, ^{99, 103, 104}, after strenuous exercise¹⁰⁵, or with certain foods¹⁰⁶ have to be taken into account, posing a potential limitation for interpretation. It may thus be that endothelial function measurements should not rely on a single test but rather on the average of several tests.

Does endothelial function provide prognostic information beyond commonly used risk scores, in primary prevention?

As outlined above, endothelial dysfunction is an important mechanism for cardiovascular risk, therefore its association with prognosis is not surprising. In the clinical setting, however, it is relevant to ask whether endothelial dysfunction provides additional information beyond traditional risk score algorithms, defined for example by the Framingham, PROCAM or SCORE projects or if it just reflects cardiovascular risk. Already early, invasive studies demonstrated that endothelial dysfunction measured in the coronary vasculature is an important prognosticator for the incidence of further cardiovascular events even in patients without coronary artery disease.^{107, 108} However, in the setting of primary prevention, invasive measurements are not feasible and most studies using peripheral endothelial function tests addressing this issue were limited by the small sample size, the evaluation of a particular subset of patients, by the long follow up periods required or by the difficulties to correct for the influence of pre-existing cardiovascular risk factors.

Peripheral flow-mediated vasodilation is predictive of cardiovascular events beyond traditional risk factors in special subsets of patients, such as after elective vascular surgery¹⁰⁹ in postmenopausal women¹¹⁰ and in patients with chest pain.¹¹¹ Several (but not all) large-scale studies recognized the additional value of endothelial function in the primary prevention setting.

In one study of 435 and one of 268 consecutive healthy subjects without heart disease and low clinical risk, brachial artery FMD independently predicted long-term adverse cardiovascular events in addition to traditional risk factor assessment.^{112, 113} In the Cardiovascular Health Study, the relationship between endothelial function (as measured by FMD) and subsequent cardiovascular events was measured in a cohort of more then 2700 apparently healthy subjects older then 72 years. Over a 5 years follow-up period, event free survival was significantly higher in those patients with normal endothelial function, a relation which still held true after adjusting for traditional risk factors.⁵⁸ Similarly, in the Multi-Ethnic Study of Atherosclerosis, a study in White, Black, Hispanic and Chinese subjects (>3000 persons), FMD predicted future CV events, again, even after adjusting for the Framingham risk score. Furthermore FMD helped better to classify cardiovascular risk in combination with the Framingham score, compared to FMD or Framingham score alone.⁵⁹ Similar to the above mentioned study, in another large cohort (n>2000) of asymptomatic postmenopausal women, FMD provided additional prognostic information to the Framingham risk score.¹¹⁴

In these studies the adjustments for traditional risk factors weakened the correlation of endothelial function and outcomes. This is not surprising, as endothelial dysfunction is a key biological mechanism by which cardiovascular risk factors exert their propensity to atherosclerosis and adverse events. However, as seen in a large number of patients in the above mentioned studies, information on endothelial function was able to add additional information, potentially suggesting that we have not yet found all individual risk factors and that in each individual subject, risk factors alone may not be linked predictably to clinical outcome. It also has to be taken into account that predictors found in multivariate analysis may only be of restricted clinical utility and provide only limited information for individual risk assessment.

In discussing whether endothelial dysfunction is able to predict future events beyond traditional risk scores, it also has to be taken into account in which vascular bed and in which patient population the measurements were performed. In the FATE study (Firefighters And Their Endothelium), for example, which included 1574 middle aged apparently healthy men at low cardiovascular risk (Framingham 7.9%), FMD was not associated with future clinical events. One explanation for the neutral findings in terms of FMD might be that these firefighters were very physically active, with the consequence of adaptive vascular remodeling and thus lower FMD.¹¹⁵ An interesting finding in the FATE study was that hyperemic velocity (a measure of microvascular function) in the brachial artery was significantly related to events in a multivariable analysis (also containing Framingham risk score).⁵⁷ Similar to the above mentioned studies the addition of hyperemic velocity (instead of FMD) to the Framingham Risk score led to a risk-reclassification improvement. In another recent large (1016 persons) community-based cohort study, forearm microvascular endothelium function as assessed with Ach infusion was associated with CV events in elderly patients, whereas FMD was not.¹¹⁶ Again, the addition of microvascular endothelial dysfunction to the Framingham score improved risk discrimination. Similarly, microvascular endothelial dysfunction measured with EndoPAT was useful in predicting non-obstructive coronary atherosclerosis, not well predicted by the Framingham score,¹¹⁷ and in independently predicting adverse cardiac events in 270 outpatients.⁷¹ Thus microvascular and macrovascular function may give important extra information, above and beyond traditional risk factors, in certain situations. Altogether if looking at the above mentioned studies it seems that macrovascular endothelial function might be more important in patients with existing atherosclerosis but microvascular function might be more important in the younger subjects; in other words, microvascular function may be an earlier indicator of risk.

Despite data indicating a predictive value for future cardiovascular events, even after adjusting for known risk, endothelial function measurements are not yet recommended by guidelines for prevention, neither by the European (ESC)¹¹⁸ nor the more recent American (AHA/ACC)^{119, 120} guidelines (Class III indication), and receive lower classification of recommendation than carotid IMT measurements and coronary calcium score. Reasons for this Class III indication were the lack of clear additional prognostic value of endothelial function and the poorly standardized non-invasive methodology (with the exception of EndoPAT). However, most of the above mentioned FMD studies, demonstrating a clear addition of prognostic value, are rather new and are published after the release of the guidelines. Despite the high sensitivity as a prognosticator for future events, specificity still remains a concern. However, with technical modifications and more accurate analysis software the variability of FMD measurements, and thus the specificity, can be shown to be further improved. If endothelial function is appreciated as a dynamic process (perhaps several measurements should be averaged) and if standardized protocols are followed, reproducible measurements can be achieved, especially if performed in high volume experienced centers.⁴⁹ Furthermore IMT and calcium score, although they can be performed in a more standardized manner and are less impacted on transient abnormalities than endothelial function, these measures give information about the vascular structure (more established disease) rather than function. Additionally these measurements do not change rapidly with interventions, an invaluable advantage of endothelial function measurements.

Is there any role of endothelial function for prognosis in patients with already established coronary artery disease or events?

As endothelial dysfunction plays an important role in the pathogenesis of atherothrombotic disease, it is not surprising that many studies have demonstrated a potential prognostic role of endothelial function in the coronary as well as in the peripheral circulation in secondary prevention. First evidence came from patients with non-obstructive coronary artery disease, where significantly higher incidences in cardiovascular^{121, 122} and cerebrovascular events in those with impaired coronary vascular function was found.¹²³ Similarly, peripheral endothelial dysfunction assessed with FMD¹²⁴ and venous occlusion plethysmography predicted CV events in patients with coronary artery disease¹²⁵ and in patients after acute coronary syndromes.¹²⁶ In the setting of established coronary artery disease, patients with endothelial dysfunction have higher rates of adverse cardiovascular events, compared to those with normal endothelial function,¹²⁷ and impaired FMD has been shown to be an independent predictor of in-stent stenosis after single-vessel coronary interventions.¹²⁸ In patients with advanced ischemic heart failure, endothelial function is a strong and independent predictor of 1 year mortality¹²⁹ and in patients with graft vasculopathy – atherosclerosis associated with cardiac transplantation – normal endothelial function is associated with lower progression of coronary intimal thickening¹³⁰ and epicardial endothelial dysfunction independently predicts outcome in these patients.^{131, 132}

In acute myocardial infarction, especially microvascular endothelial dysfunction has been documented to be indicative of a poorer prognosis.^133–136 For example, no-reflow on angiography strongly predicts 5-year mortality, independent of infarct size in patients with acute ST-elevation myocardial infarction.¹³⁷ Interestingly, no-reflow might be reversible in some cases, which is associated with a better prognosis.¹³⁸

Endothelial function as a contributor to disease progression?

Endothelial dysfunction in the periphery and in the coronary arteries not only is a marker for cardiovascular risk, but is also a contributor to the progression of atherosclerosis¹⁰¹ and cardiovascular events. Interestingly, the atherosclerotic epicardial segments which show the most endothelial dysfunction are those with characteristics of vulnerable atherosclerotic plaques.³⁶ These segments are characterized by the loss of NO activity and increase in endothelin-1 activity,¹³⁹ the same segments progressing more likely to obstructive coronary artery disease.¹⁴⁰

Importantly, microvascular dysfunction may contribute to the impaired regulation of myocardial perfusion by reducing the capacity to increase perfusion in response to exercise or mental stress, a circumstance which may lead to myocardial ischemia.¹⁴¹ In the context of myocardial infarction, endothelial microvascular dysfunction is an important mediator of the event, rather than just a consequence.¹⁴² This is likely via reduction of coronary blood flow by altering shear stress on the epicardial level and lowering endothelial function and aggravating thrombus formation. Diabetes, and the accumulation of risk factors in the metabolic syndrome, for example, have significant deleterious effects on myocardial perfusion and infarct size in patients with an acute infarction.^{143, 144, 145, 146} Moreover, patients with pre-procedural impairment of microvascular function are more likely to have post-procedural microvascular impairment as well as procedure-related injury and a worse outcome.¹⁴⁷ Thus, pre-existing microvascular endothelial dysfunction leads to a greater vulnerability to myocardial injury, highlighting the potentially clinically relevant role of a dysfunctional microcirculation and damage.

Does endothelial function identify responders and non responders to therapy?

Many medical or lifestyle interventions are able to improve endothelial function and to reduce cardiovascular events. For example statin treatment significantly improves peripheral and coronary vascular function,¹⁴⁸ although not all studies were able to prove such an effect within a six months treatment period (ENCORE-1¹⁴⁹ and CARATS¹⁵⁰). Off note, the impact on risk reduction despite this successful intervention is limited and is in the range of about 20–45% in clinical trials. Even with the combination of all therapies proven to lower risk in secondary prevention (or primary prevention), some patients may still develop later events and therefore are obviously not completely protected by their therapy.

Therefore it is the ultimate goal to identify those patients who will develop future events despite therapy (in order to potentially escalate and intensify current treatment). One concept could be to measure the individual impact of therapy on endothelial function as a parameter of cardiovascular disease and especially target those which no improvement in vascular function. Recently, important studies in this respect have been performed:

In a study in 251 Japanese men with newly diagnosed stable coronary artery disease and concurrent endothelial dysfunction (low FMD), FMD was repeated after 6 month of optimized individualized therapy. Those patients with persistently impaired FMD had significant higher event rates in the follow up period (26% over 31 months) compared to those with normal FMD (10%).¹⁵¹ In a similar study, endothelial function was assessed in 400 post-menopausal hypertensive women without evidence of coronary artery disease at baseline and 6 month after treating blood pressure to normotensive values. In those women whose endothelial function (FMD) has not improved (37.5%) there was a nearly 7-fold increase in CV events over the average follow up of 67 month.¹⁵²

Both studies convincingly demonstrate that patients who do not respond with improvement of endothelial function with interventions are at a considerable risk of further events. In our view, more studies in this respect need to performed to give definitive answers to one of the most intriguing aspects in using endothelial function measurements.¹⁵³

Nevertheless, one has to acknowledge in the future that the pathophysiology of vascular dysfunction is extremely complex, probably differs form different vascular beds and between micro- and macrovascular vessels. Thus in the clinical endothelial function evaluation of the future, a more integrative approach should be considered, with the inclusion of peripheral macrovascular endothelial function as well as microvascular function measurements. An ideal test to identify a vulnerable patient should reflect and follow the disease state, thus should be abnormal with disease and possibly reversible with interventions (Table 4).

TABLE 4.

Criteria for an optimal endothelial function test

reflects disease state reversible with interventions mirrors coronary endothelial function improves risk stratification reproducible operator independent non-invasive (no or low risk for the patient) ease of use inexpensive

Is improvement in endothelial function an indicator of successful treatment?

It is probably a good sign when endothelial dysfunction is (partly-) reversed with treatments. First proof for this principle came from two controlled studies in 1995 where cholesterol lowering therapy (statins),^{154, 155} improved endothelial function. Statin treatment now has convincing evidence for its beneficial effect on coronary and peripheral endothelial function,¹⁴⁸ likely due to their anti-inflammatory and antioxidant properties, as well as due to the restoration of the vascular NO bioavailability.¹⁵⁶ Since the first evidence in humans with statins, numerous interventions in a broad range of patients demonstrated a beneficial effect on endothelial function. Most pharmacological intervention studies with an effect on cardiovascular risk factors also improve endothelial function. For example, antihypertensive therapy in general,^{157, 158} such as angiotensin-converting enzyme (ACE)-inhibitors,^{159, 160} angiotensin-receptor blocker,¹⁶¹ calcium channel blockers (CCB)¹⁶² and certain β-blockers, in particular the NO-group containing molecule nebivolol,¹⁶³ might reverse endothelial dysfunction, however, ACE-inhibitors seem to be particularly important.^{158, 164, 165} CCB reduce calcium entry though L-type voltage-dependent channels of the vascular muscle cells thus dilatating coronary and other arteries. Additionally, some CCB activate endothelial NO synthase or have antioxidative properties thus increasing NO bioavailability.¹⁶⁶ The ENCORE-1 and ENCORE-2 trial showed that long acting nifedipine consistently improved coronary endothelial function in patients with stable CAD that persisted even after cessation of the drug.^{149, 167} ACE-inhibitors and angiotensin receptor blockers are also preferred medication in diabetes mellitus.¹⁶⁸ Diabetes modulating drugs like metformin¹⁶⁹ or glitazones^{170, 171} may also improve vascular function in patients with type 2 diabetes, however, the later may have negative effects on cardiovascular risk,¹⁷² thus limiting its use.

Not only pharmacological agents but also lifestyle factors and medications that increase the release or prevent the degradation of endothelial derived relaxing factors, NO in particular, and those which decrease production of endothelium-derived constricting factors such as endothelin among others, can improve endothelial function. Many interventions have either demonstrated to be beneficial for microvascular or macrovascular endothelial function by increasing NO bioavailability such as physical exercise,^{173, 174, 175} weight reduction^176–178 (including bariatric surgery),^{179, 180} or enhanced external counterpulsation^{181, 182} and dietary interventions with foods rich in polyphenols, especially fruits, tea and cocoa.^183–185 An important lifestyle modification with impact on endothelial function is smoking cessation, which clearly demonstrates a favorable effect on epicardial coronary endothelial function,⁸¹ which however was not observed in the microvasculature.¹⁸⁶ This demonstrates the differences of macrovascular and microvasculature function and its complex not yet fully understood interactions.

While therapies with a proven benefit on morbidity and mortality in cardiovascular patients concordantly improve endothelial function, it is not certain whether the opposite is always true. For example vitamin C or E as well as folic acid supplementation, which were praised for their antioxidative capacity and were associated with significant acute improvement in endothelial function,¹⁸⁷ failed to show any benefit in the long-term¹⁸⁸ or in cardiovascular disease prevention so far. On the other end, the pathophysiological information derived from studies on endothelial function is not correctly applied in clinical trials. Available evidence demonstrates that vitamin C can improve endothelial function,¹⁸⁷ but also that this effect is obtained with concentrations much higher than those reached after oral administration.¹⁸⁹ Many studies evaluating pathophysiological aspects were performed in the setting of acute intravenous or intraatrial interventions. Moreover, the response to vitamins may also be dependent on the presence of increased endogenous oxidative stress.¹⁹⁰ This kind of information should be carefully evaluated when designing intervention studies. The abundance of current trials with any kind of positive effects on the endothelium are pathophysiologically interesting, but care should be taken to extrapolate the findings to cardiovascular morbidity and mortality. Finally, it is of concern that the drug effects on endothelial function may be different according to which vascular bed is considered.

Additionally, many lifestyle interventions, foods and drugs have been shown to improve endothelial dysfunction in a population as a whole, but this does not give information on the individual patient, a fact which should be addressed in further studies.

Are microvascular and conduit vessel endothelial function comparable?

Although cardiovascular risk factors are associated with endothelial dysfunction in virtually every arterial bed, considering the different physiological role of conduit and resistance arteries, important differences should be considered. Whereas reduced NO release in response to stimuli plays a central role in the pathophysiology of endothelial dysfunction in the conduit arteries, NO in the microcirculation may primarily modulate tissue metabolism.¹⁹¹ Furthermore metabolic and other factors are becoming increasingly important in the regulation of microvascular function. Therefore pharmacologic tests inducing NO release might not reflect the physiologic adaption of endothelial function in the microvasculature in response to exercise or ischemia.

The aforementioned differential effect of smoking on microvasculature and epicardial vasculature as outlined above might be only one example.¹⁸⁶ Furthermore, FMD is particularly sensitive to being impaired by traditional risk factors (e.g. age, hypertension), whereas the peripheral arterial tonometry reactive hyperemia index (microvasculature) is more sensitive to metabolic risk factors, especially body mass index and diabetes,^{72, 73} (and interestingly, shows a paradoxical association with age in the Framingham cohort).⁷² Micro- and macrovascular dysfunction could also reflect different stages of vascular disease as conduit artery endothelial dysfunction may be more important in patients with existing atherosclerosis and microvascular dysfunction may be an earlier indicator of risk. The fact that micro- and macrovascular endothelial function only show a weak (if any) correlation with each other⁷³ should caution against the extrapolation of findings in one circulation level to the other.

Given that macrovascular and microvascular endothelium are susceptible to different risk factors, whenever possible both should be evaluated.

Is there a role of endothelial function in drug-development programs?

For new drugs the requirement by drug regulation authorities is to prove the principle by “primum non nocere” (first, do no harm); however to test the effect of a certain drug on morbidity and mortality requires large sample sizes. Sometimes, such as for drugs with relatively small effects on the cardiovascular system or in children, such outcome trials are not feasible at all. Clinical endothelial function evaluation is of potential value in reassessing the risk of drug-development programs, especially as it gets more and more challenging to choose novel agents for clinical use.¹⁹² With endothelial function as a mechanistic surrogate integrating various types of cardiovascular risks^{49, 192} sample size can be significantly smaller, compared to clinical endpoint trials.⁵² Furthermore, endothelial function may respond rapidly to therapies (within hours, days or weeks), long before effects on clinical outcomes are seen. Thus the impact on endothelial function may give important signals of efficacy or, more important, may warn of potential harm. Therefore endothelial function is not only a valuable measure to assess drug efficacy on surrogate endpoints but also may play an important potential role in the evaluation of drug safety as exemplified in the recently completed dal-VESSEL study,¹⁹³ notably the first multicenter study to use FMD as outcome measure. For certain studies a multimodality approach, which includes peripheral endothelial function measurements, may be of particular value.¹⁹⁴

What kind of studies do we need in the future?

As outlined above, endothelial function measurement may differentiate responders versus non-responders to therapy.¹⁵³ In secondary prevention, studies demonstrate that patients who do not respond with improvement of endothelial function with interventions are at a considerable risk for further events. These early data suggest that individual endothelial function measurements guided therapy might be feasible in these settings but larger studies in this respect are needed to answer the question whether endothelial function guided therapies help to improve outcomes.

In primary prevention, it is still an unanswered question whether endothelial function should be assessed in the apparently healthy persons at low risk from traditional risk factors. To address this issue, endothelial dysfunction, which depicts mechanisms at the core of atherosclerosis and its complications, could be chosen for future similar studies, with different medical and lifestyle interventions to be tested. Designing such a trial would require very careful consideration of which non-invasive test or combination of testes of endothelial function should be included. If such results prove positive, there would be a good rationale to implement endothelial function testing into everyday clinical practice.

Current, clinical guidelines and risk management for prevention are based on the risk factors established in the Framingham study and certain cardiovascular surrogates such as carotid IMT and coronary calcium.^{119, 120}. However, the Framingham score, as well as other scores, when applied to different populations provide inconsistent results¹⁹⁵ and adjustment based on different populations might be needed. Additionally, the Framingham Score is limited to the fact that risk factors were collected years ago, where for example no statin therapy was available and most persons smoked and/or were exposed to second-hand smoke. The effect of a changing environment might be better depicted by endothelial function assessments. Assuming that endothelial function provides an integrated functional risk assessment, the question whether endothelial function might be a better predictor for cardiovascular events than the actual scoring systems is intriguing and should be tested with larger scale studies.

When using the Framingham risk score, we are aware how to deal with patients in the high or low risk category. However many patients end up having intermediate risk, where the recommendation are less clear. As demonstrated by the studies discussed above, reclassification of patients with intermediate risk according to their endothelial function seems to be feasible and reasonable, although further studies in this area are required.

CONCLUSION

In the past decade, many studies have suggested that the non invasive assessment of endothelial function may provide important information for individual patient risk, progress and guidance of therapy (in addition to its well established role in clinical research). This is underscored by the low risk of the tests and the valuable information that can be derived from them. Thus further research should be directed to determine whether measurement of endothelial function can be used to effect treatments and change outcomes and ascertain whether detection of endothelial function will be useful in clinical arena.