Despite declines in cardiovascular disease (CVD) mortality rates over the past few decades, CVD still remains the leading cause of morbidity and mortality in the United States (1). Endothelial dysfunction contributes to atherosclerosis development and progression, which may ultimately lead to plaque rupture and cardiovascular events. Although the vascular endothelium serves many important functions including maintaining vasomotor tone and barrier functions, the most readily detectable means to define endothelial pathology or dysfunction in humans is by quantifying vasomotor responses to endothelial dependent stressors. The development in recent years of imaging strategies to measure endothelial function of the coronary and peripheral vessels has provided insights into important contributors of coronary artery disease (CAD) and the vascular response to therapeutic intervention. In this review, we will briefly examine mechanisms relating endothelial function and atherosclerosis, review imaging strategies, both invasive and non-invasive, to quantify endothelial function of the coronary and peripheral circulation, and discuss recent insights from human endothelial function studies.

Dysfunction of the vascular endothelium is increasingly recognized as serving a prominent role in CVD pathology. The endothelium regulates vascular tone, smooth muscle cell proliferation, thrombosis, and leukocyte adhesion and platelet aggregation (2). Endothelial dysfunction, or alteration in normal function, often precedes the development of anatomic atherosclerotic disease progression and clinical manifestation. Examination of endothelial function can enhance risk stratification, improve early detection of disease and be used to assess the vascular response to therapeutic intervention (3).

Healthy endothelial cells respond to local and systemic factors by producing and releasing vasoactive molecules to maintain vascular tone, a balance between vasodilation and vasoconstriction (4). A defining feature of endothelium-dependent relaxation is the release of nitric oxide (NO), which diffuses to vascular smooth muscle cells and results in cGMP-mediated vasodilation (4). NO is released in response to a variety of signals, such as adenosine, serotonin, catecholamines, ischemia, and shear stress (5). Conversely, systemic inflammation and increased reactive oxygen species (ROS) tend to counter the effects of NO, and can result in chronic endothelial dysfunction (6). Cardiovascular risk factors such as hyperlipidemia, hypertension, and diabetes may result in dysregulation of endothelial nitric oxide synthase (eNOS) and ROS (7), leading to endothelial dysfunction, one of the earliest steps in the atherosclerotic disease process (8). Although dysfunctional endothelium is characterized by increased vascular inflammation, permeability and thrombosis, it is impaired vasodilation in response to stressors that increase NO that is the most readily measurable response in humans and detectable by imaging.

In the peripheral conduit vessels, endothelial function is typically evaluated in the brachial artery due to its accessibility, and measures can be performed invasively (forearm plethysmography) or non-invasively (brachial ultrasound for flow mediated dilation) by evaluating the vasomotor response to endothelial dependent stressors (9). Measuring endothelial function of the coronary arteries is more challenging but important as the clinical impact of coronary endothelial dysfunction is greater than other vascular beds. Coronary endothelial function (CEF) is typically examined through invasive measures during coronary angiography. Coronary arteries are prone to atherosclerosis and studying CEF provides new information about the heterogeneity of endothelial function and contributors to plaque formation in patients with, or at risk for coronary artery disease. However, the invasive measurement of CEF carries procedural risk and preclude studies in lower risk patients over time. Newer non-invasive measures of CEF including with magnetic resonance imaging (MRI) and positron emission tomography (PET) promise new insights into the pathophysiology of CVD in low risk and other populations not undergoing invasive angiography and can assess response to therapy. Finally, microcirculatory assessment of smaller vessels, comprised of pre-arterioles, arterioles, capillaries and venules, investigates endothelial function in vascular resistance, which mediates blood pressure and blood flow. The measure of endothelial function of the larger (conduit) or smaller (microvessels) provides important and complementary information which can help gauge CV risk and provide prognostic information for patients (10).

The measurement of human endothelial function primarily focuses on vasoreactivity testing, as this is the most clinically demonstrable function of the vascular endothelium (11). Coronary endothelial dysfunction predicts cardiovascular events and remains the most important vascular bed studied in vasoreactivity (10, 12, 13). The gold standard for coronary endothelial functional assessment is via invasive quantitative angiography to detect luminal changes in response to vasoactive stimuli, either pharmacologic or physiologic, that increase the endothelial release of NO (Figure 1) (14, 15).

Coronary angiography for epicardial arterial dimension measurement is often performed with intracoronary infusion of acetylcholine. Acetylcholine is an endothelial-dependent vasodilator that is suitable for intracoronary infusion and is the most commonly used drug for the purposes of invasive vasomotor testing. Healthy endothelium should result in coronary arterial vasodilation and increased blood flow (by >50%) in response to low dose acetylcholine, while dysfunctional endothelium may lead to diminished blood flow response and even to paradoxical vasoconstriction. At higher doses, acetylcholine can result in constriction of small arteries via direct effect on smooth muscle cells, and may be used to evaluate microvascular function (16). Less commonly, other agents used in endothelial-dependent vasomotor testing have included bradykinin, papaverine and Substance P (17). In addition, adenosine has partial endothelial dependent effects (18). These vasoactive agents act on coronary microvasculature through vasodilation and increased flow, resulting in NO release and proximal coronary artery vasodilation, or flow-mediated dilation (FMD) (19, 20), permitting the study of epicardial endothelial function.

Coronary microvascular function can be studied invasively by measuring coronary blood flow changes and thereby coronary flow reserve using a Doppler wire. Generally, this is accomplished by placement of a Doppler-tipped guide wire into the coronary artery of interest, whereby continuous blood flow velocity is measured both at baseline and during intracoronary infusion of vasoactive substances (acetylcholine, adenosine, or papaverine) through the guiding catheter (17, 21, 22).

Further, invasive CEF assessment can also be performed by cold pressor testing (CPT) or exercise testing, both endothelial-dependent stressors (23). Exercise stress testing can be performed while supine using a bicycle ergometer with concurrent hemodynamic monitoring (24). Healthy coronary arteries dilate in response to these stressors, while paradoxical vasoconstriction occurs in diseased coronary arteries, suggesting underlying endothelial dysfunction.

Endothelial dysfunction diagnosed by invasive methods has been reported in several cardiometabolic disease states and is associated with future atherosclerosis and other adverse outcomes (13, 25–27). These techniques have also been used in the assessment of endothelial dysfunction reversal with treatment therapies (28). The advantages of catheter-based methods of coronary endothelial assessment include the precision and accuracy of results obtained using this gold standard of testing, particularly in comparison to techniques that rely on surrogate measures of coronary arterial function (17). With this approach, however, come the limitations of an invasive procedure with intra-arterial injection of vasoactive medications that can have systemic adverse effects, along with exposure to radiation and contrast. Given these risks, repeat evaluation is often not performed. Invasive techniques are therefore largely limited to patients undergoing coronary angiography for clinical reasons. Additionally, in patients with CAD, vessel area measurements may be limited in coronary segments with atherosclerosis.

Nuclear imaging methods can be used to evaluate myocardial blood flow and response to endothelial-dependent stressors. PET can be used to estimate coronary flow reserve and myocardial regional perfusion using intravenously injected tracers (¹⁵Oxygen-labeled water, ¹³Nitrogen-ammonia, and ⁸²Rubidium), and studies have revealed abnormalities in endothelial function prior to visible atherosclerosis on angiography (40, 41). These techniques have been successfully combined with CPT to assess CEF. CPT protocols typically involve immersion of the subject’s hand or foot into an ice bath at 2°C for at least 1 min prior to radioactive tracer injection and PET scan (42). CPT functions to increase myocardial oxygen demand via sympathetic activation, which should cause vasodilation and an endothelial-dependent increase in coronary blood flow in healthy subjects (43). Using these principles, cardiac PET during CPT has been shown to reflect epicardial vasomotor dysfunction in subjects at high risk for CAD (44). Abnormalities in myocardial blood flow on PET, regardless of concurrent CAD, appear to confer an increased relative risk of death and heart failure (42, 45). It is important to recognize that myocardial blood flow is affected by epicardial coronary vasomotor tone and microvascular function, making it challenging by PET imaging alone to determine whether changes in flow are related to conduit or resistance vessels (41).

The addition of computed tomography (CT) to PET can further enhance the sensitivity for atherosclerosis detection (46). A hybrid PET/CT approach has the ability to quantify changes in coronary cross-sectional area in response to stress, global, and relative myocardial perfusion, left ventricular functional performance, and coronary calcium score. This non-invasive tool for assessing coronary vascular health may represent a clinically relevant evaluation that can be performed in early disease or to predict downstream risk, however, its use has been primarily research-related (47).

Non-invasive assessment of impaired myocardial blood flow, which contributes to ischemia in patients with CAD and cardiomyopathy, can be performed using stress perfusion cardiovascular magnetic resonance (CMR), which may be especially helpful for serial examinations evaluating treatment success (62). Stress perfusion CMR, distinct from coronary vasoreactivity approaches mentioned above, typically uses vasodilator stress (i.e., adenosine) to detect macrovascular (i.e., coronary stenosis) and microvascular differences in myocardial blood flow in response to stress. Recently, studies have employed fully quantitative stress myocardial perfusion techniques in patients with no obstructive CAD and detected reduced myocardial perfusion reserve, not explained by cardiac hypertrophy or fibrosis (63). The ability of CMR to study ventricular function/structure and fibrosis make it well-suited to be used in combination with stress perfusion techniques, especially in patients with left ventricular hypertrophy.

Stress perfusion CMR techniques have also been employed to evaluate patients at risk for microvascular dysfunction. Clinical guidelines have recently added microvascular dysfunction to epicardial stenosis and epicardial coronary spasm as one of the mechanisms of myocardial ischemia in patients with CAD (64). One study used stress CMR techniques and showed that myocardial perfusion reserve index was impaired in women with no obstructive CAD on coronary angiography, reflecting microvascular dysfunction compared to reference controls (65). A randomized trial in this setting showed that medical therapy with ranolazine improved angina and reduced ischemic burden in woman with myocardial ischemia detected by stress CMR in the absence of obstructive CAD, suggesting a possible use of MRI for therapeutic assessment (66). Additionally, in patients with infiltrative heart disease such as amyloidosis, it has been demonstrated that impaired myocardial perfusion is related to abnormalities in myocardial structure and function not only at stress, but also at rest (67). Taken together, studies support the use of stress perfusion CMR to investigate myocardial perfusion reserve, which reflects microvascular dysfunction in the absence of CAD. In addition, stress CMR has already demonstrated high prognostic value and cost-effectiveness compared to invasive strategies (68, 69). While classically, limitations of this technique included the need for highly specialized equipment and providers, recent technical developments now allow quantitative and fully automated assessment of myocardial ischemia using stress CMR, which may enable the broad use of this modality outside of specialized centers (70).

The homogeneity of myocardial perfusion can be readily assessed by its uptake of iodine contrast medium and its associated X-ray attenuation. George et al. demonstrated that myocardial perfusion can be quantified using CT and that reversible perfusion defects can be identified after vasodilator challenge analogous to nuclear imaging techniques (71). CT scanning is performed using injection of an iodinated contrast agent with prospective electrocardiographic gating. Microvascular function may be assessed by determining MBF at rest and after vasodilator challenge with abnormal flow reserve typically defined as a ratio of <2.0 (60). In the absence of obstructive CAD and local myocardial perfusion defects, reduced MFR can be attributed to microvascular dysfunction. Figure 4 shows an imaging example of a patient with a severe myocardial perfusion defect in the lateral and posterolateral walls post infarct.

Advantages of CT include faster image acquisition than with nuclear techniques and markedly superior spatial resolution. Directly compared to nuclear myocardial perfusion imaging using exercise or vasodilator challenge, CT myocardial perfusion yields at least equivalent accuracy for identifying patients with CAD (72). Another major advantage of cardiac CT is the assessment of both coronary arterial anatomy and myocardium. Using contemporary technology, rest-vasodilator CT for coronary angiography and myocardial perfusion imaging can be performed with radiation doses lower than standard nuclear perfusion imaging using SPECT, though requiring two contrast applications of approximately 60 ml each (72). Determining MBF and coronary flow reserve by CT myocardial perfusion imaging is possible using dynamic imaging, i.e., continued imaging over several cardiac cycles (73).

Application of dynamic CT imaging had been hindered by high associated radiation exposure to the patient but new protocols have been developed using lower tube settings which have reduced radiation to levels similar to that by conventional rest-vasodilator myocardial perfusion protocols (73). Comparison to PET revealed high accuracy of dynamic CT for detecting abnormal MBF using a mean radiation dose of 8.4 mSv (74). Further radiation dose reductions are feasible using intermittent instead of continuous scanning, thus overcoming one of the major limitations of dynamic CT perfusion imaging and opening the possibility of comprehensive coronary arterial and myocardial assessment.

Both established and newer cardiovascular risk factors can adversely affect endothelial function, including obesity, diabetes, smoking, and inflammation/oxidative stress (7, 8). To this end, the measurement of endothelial vasoreactivity serves as an index of the sum total effects of environmental and genetic factors on the vasculature. Furthermore endothelial dysfunction is a marker for subclinical disease, an independent predictor of adverse cardiovascular events, and a potential target for medical interventions (78, 82). One study using PET showed that cigarette smokers have reduced MFR, with improvement seen with smoking cessation (85). Similarly, initiation of antihypertensives can result in improved endothelial function in patients with hypertension (86). Obese patients were reported to demonstrate impaired MBF with improvement after bariatric surgery (59). In addition, MFR is reduced in patients with diabetes, with some suggestion that endothelial function (measured by FMD) may improve with dapagliflozin (55, 87).

Recently, coronary microvascular dysfunction has been implicated in multiple disease processes including microvascular angina, a common encountered disorder which can lead to ischemia or myocardial infarction, even in the absence of obstructive coronary artery disease (60). Microvascular angina due to ischemia with non-obstructive coronary arteries (INOCA) can be challenging to diagnose, with a heterogenous approach to patients and many knowledge gaps with regards to treatment. The CorMicA trial showed that guiding therapy by invasive provocative coronary testing in patients with INOCA identified to have microvascular dysfunction may be of clinical benefit (88). Recent methods using stress perfusion CMR are being employed in the CorCMR study to evaluate whether a non-invasive approach to assess coronary microvascular dysfunction in INOCA patients improves cardiovascular risk and anginal symptoms (89). The results of study may have important clinical implications in this patient population, where there is less evidence in terms of diagnostic testing and treatment. Furthermore, using these approaches to quantify coronary microvascular function may provide clinically meaningful information beyond what is possible using standard anatomic and ischemia assessment with the ultimate goal of improving patient outcomes. It is important to note that many of the techniques described in this review (PET, specialized CMR, and CT perfusion) are not yet widely available clinically, however, may play an important role in the evaluation of patients with INOCA and to test early therapies to justify larger clinical trials with hard end points.

Both coronary and systemic endothelial function measures have been used as endpoints in clinical intervention trials after the techniques were shown to be reproducible in the short and intermediate term (90). Studies using endothelial function as an endpoint enable the assessment of the vascular impact of emerging treatment strategies and can guide novel drug development, such as approaches to target oxidative stress or inflammation. Studies targeting the xanthine oxidase (XO) system, a significant source of vascular oxidative stress, or systemic inflammation using colchicine have used CEF testing as a surrogate imaging endpoint over time in randomized placebo-controlled clinical trials (90, 91). Recently, impaired CEF in people with HIV and dyslipidemia improved with short term treatment with the PCSK9 inhibitor, evolocumab, indicating that the MRI-CEF technique can detect rapid improvements in CEF in response to treatment (92). Therefore, this approach enables future studies focused on repeated CEF measures in healthy and lower risk populations over time.

Endothelial dysfunction is now a well-established gauge of cardiovascular risk and predicts future adverse events. Recently, endothelial dysfunction has been implicated as a contributor to a variety of cardiovascular diseases including INOCA, stress cardiomyopathy, preeclampsia and heart failure with preserved ejection fraction among others (26, 93). We have summarized multiple methods that are available for probing coronary and peripheral endothelial, each with specific strengths and weaknesses, and different values for defining pathology (Table 2). Currently available endothelial testing methods are helpful for mechanistic understanding of disease and for risk stratification and prognostication. It is important to recognize, though, that using a pharmacological stressor for imaging to assess endothelial function will often detect function in response to a combination of endothelial and non-endothelial dependent mechanisms, and depending on the stressor and imaging modality, this should be considered. Increasingly endothelial function testing is being explored for clinical management and evaluation of therapeutic response, although there are currently no guidelines recommending use of endothelial function in routine patient management. Nonetheless, evidence continues to grow in the role of the vascular endothelium in disease pathophysiology and ongoing large-scale studies are essential for the evaluation of therapies targeting endothelial function.

AM, EG, MC, AA-Z, SK, and AH drafted the manuscript. TL and AL edited the manuscript. All authors contributed to the article and approved the submitted version.

AA-Z has received research support from Canon Medical Systems. AL serves as a consultant for Itamar Medical. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.