Article Text

Download PDFPDF

Endothelial dysfunction and pathophysiological correlates in atrial fibrillation
  1. M Guazzi1,
  2. R Arena2
  1. 1
    Cardiopulmonary Unit, University of Milan, Italy
  2. 2
    Virginia Commonwealth University, Richmond, Virginia, USA
  1. Professor Marco Guazzi, Cardiopulmonary Unit, Cardiology Division, University of Milan, San Paolo Hospital, Via A di Rudinì, 8, 20142 Milan, Italy; marco.guazzi{at}


Endothelial dysfunction (ED) increases oxidative stress and proinflammatory agents, and impairs nitric oxide (NO)-dependent vasorelaxation. Atrial fibrillation (AF) is a risk factor for ED as documented by (1) impaired acetylcholine-mediated blood flow increase; (2) reduced plasma nitrite/nitrate levels; (3) additive impairment of flow-mediated dilatation by comorbidities causing ED; and (4) efficacy of cardioversion. Several possible mechanisms sustain the AF-ED association: (1) An impaired rheology. Endothelial NO release is tightly regulated by laminar shear stress and AF induces a turbulent flow which may impair arterial vessel distension and responsiveness. Specifically, Ca2+ elicits NO synthase (eNOS) activation, and shear stress application to endothelial cells increases intracellular Ca2+ primarily in response to regular pulsatile flow at a rate higher than that observed in the presence of oscillatory pulsatile flow. (2) The atrium activity on arterial vessels. The left atrium produces NO and may serve as an endocrine organ releasing nitroso compounds. A disorganised atrial contraction markedly reduces eNOS expression. (3) AF induces atrial inflammation and elevation of C reactive protein and cytokines, exerting a proinflammatory activity on endothelial cells. (4) Systemic factors such as the renin-angiotensin system (RAS) may be prominent. In fact, RAS and inflammation reciprocally “cross-talk”. Angiotensin II increases atrial cell death and RAS contributes to myocardial and vascular oxidative stress in AF. RAS inhibition prevents AF. Important clinical correlates of ED in AF patients are muscle underperfusion, premature lactic acidosis and ergoreflex oversignalling during physical activity. This review focuses on the evidence of an association of AF with ED, the possible underlying mechanisms and the pathophysiological correlates.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Endothelial cells importantly regulate oxidative stress, vascular permeability to plasma constituents, platelet aggregation, thrombosis, vascular tone and blood pressure by controlling the release of several vasoactive substances including nitric oxide (NO).1 A number of risk factors for cardiovascular disease, such as age,2 diabetes,3 hypertension,4 obesity,5 hypercholesterolaemia6 and smoking7 may cause the endothelium to be subjected to a maladaptive state, currently called endothelial dysfunction (ED), which is characterised by abnormal regulation of the vessel tone, heightened oxidative stress, increased proinflammatory and prothrombotic states, overexpression of adhesion molecules and, as a main characteristic, impaired endothelium-dependent vasorelaxation.8

Several studies indicate that ED is a systemic vascular process which may herald atherogenesis, promote plaque development, be a mechanistic link between risk factors and atherosclerosis, and modulate its progression and clinical course.8 Recent evidence also suggests that ED is significantly involved in exercise intolerance, hyperventilation and breathlessness in patients with chronic heart failure (CHF). The impaired physiological endothelium-mediated vasodilation,9 in fact, withdraws a mechanism that is basically implicated in the increase in arterial conductance, upregulation of muscle perfusion and modulation of exercise neurogenic vasoconstriction.1012

Laminar shear stress exerted by the viscous drag of flowing blood is an important physiological stimulus in the regulation of NO release from the endothelium, which, in turn, has a key role in inhibition of inflammation, cellular proliferation, thrombosis and in the release of vasoactive substances for the vascular tone and organ perfusion regulation. Which strongly suggests that an irregular pulsatile blood flow, like that caused by atrial fibrillation (AF), may impair the endothelial responsiveness to vascular shear stress. Although this is evidently in favour of AF as a risk factor for ED, it was only in 1997 that the association of AF with a maladaptive state of the endothelium was proposed. Minamino and co-workers13 first showed that plasma levels of nitrite/nitrate and platelet cyclic guanosine monophosphate (cGMP) were lower in patients with persistent AF compared with age-matched and sex-matched controls in sinus rhythm.

In this article we review the evidence of an association of AF with ED, the possible underlying mechanisms and the pathophysiological correlates.


The hypothesis that AF and irregular blood pulsatile flow might trigger ED and this may increase cardiovascular risk and negatively impact clinical status and symptoms appears intriguing. In 2001, Takahashi and collaborators14 measured acethylcholine-induced changes in forearm blood flow (FBF) in patients with AF with or without underlying heart disease, and repeated measurements after cardioversion to sinus rhythm (CV). In patients, compared with healthy controls, they found a significantly smaller increase in FBF, and a substantial improvement in the responsiveness to acetylcholine following CV. The same investigators15 also showed that, after CV, FBF on exercise was greater than before, and that NO synthesis inhibition with NG-monomethyl -l- arginine (l-NMMA) inhibited the exercise-induced FBF increase, coming to the conclusion that, in the presence of AF, NO-dependent flow-mediated vasodilation (FMD) is low both at rest and on exercise. In line with these findings a set of observations demonstrated that when AF is associated with diseases causing ED (like diabetes mellitus and high blood pressure), baseline values of brachial artery FMD, evaluated by ultrasound imaging,9 16 are significantly lower than in patients with lone AF, and that, at variance with findings of a study including a small number of diabetic patients,17 the expected improvement in FMD after CV is smaller in cases with comorbidities.18 19

Limitation to physical activity and early occurrence of fatigue and dyspnoea sensation on exertion are typical manifestations in AF patients.20

One of the mechanisms that control ventilation on exercise is the so-called ergoreflex,21 22 which originates from group III and IV neural afferents located in the skeletal muscle. These afferents, conventionally named ergoreceptors, have two subtypes: metaboreceptors (sensitive to metabolic byproducts of muscle work) and mechanoreceptors (activated by deformational changes in limbs and joints). The ergoreceptors evoke increase in autonomic outflow and ventilation, and their contribution to ventilation can be tested with handgrip exercise and are represented as the difference of changes in ventilation between rest and recovery with and without post-handgrip circulatory occlusion of the exercising arm. Studies by our group were aimed at testing the hypothesis that a defective NO release on exercise due to AF should translate into an increase of the ergoreflex component of ventilation which is reversed to normal by CV. In agreement with these premises, in patients with lone AF we found23 that the ergoreflex component of ventilation fell from a mean value of 5.3 (SD 2.3) l/min pre-CV (p<0.01) to 1.2 (1.6) l/min post-CV and reverted to 5.9 (2.2) l/min after AF relapse, with corresponding values of brachial artery FMD of 0.30 (0.06) mm, 0.41 (0.08) mm (p<0.01 v pre-CV) and 0.33 (0.07) mm, respectively.

To provide further evidence of an endothelium-ergoreceptor interaction, we aimed at demonstrating18 that: (a) the receptor stimulation is proportional to the degree of baseline ED; (b) interventions that improve endothelium also attenuate ergoreflex activation; (c) the response is paired independently of the type of intervention set in motion. AF was judged to be an ideal pathophysiological model to work on, because ED of this arrhythmia is, as already mentioned, further emphasised by comorbidities like diabetes or hypertension,18 19 and because endothelial activity can promptly be restored by CV. Baseline FMD was increasingly more severe and ergoreflex activity more pronounced in hypertensive and diabetic patients compared with lone AF. Vitamin C, a rather weak antioxidant,24 25 improved FMD and reduced metaboreflex activity in AF alone or associated with high blood pressure and no additive amelioration was obtained when administered after CV. Neither vitamin C nor CV were significantly effective in the presence of diabetes. It appears, therefore, that AF generates oxidative injury, which is less when the arrhythmia is lone and greater when the arrhythmia is associated with hypertension. The baseline endothelial damage by diabetes is probably such as to prevent appreciable additive influences of AF and, independently on comorbid diseases, in AF a causal link exists between ED and ergoreflex.18

Summing up, then, there is solid evidence in favour of AF as an elicitor of ED, which is emphasised by some comorbidities known to interfere with endothelial function. AF is associated with ergoreflex activation, an influence that appears to be mediated by a deregulation of endothelial control of vessel tone.


The mechanisms involved in the AF-ED coupling may be synthesised as: (a) impaired blood rheology caused by the arrhythmia; (b) fibrillating atrium influencing the vessels; (c) atrial endocardium and vascular endothelium as twin targets of the same systemic process.

Impaired rheology

Altered haemodynamics has been reported as a possible mechanism,14 26 basically because the arrhythmia produces irregular stroke volume and turbulent flow, and CV restores a normal stroke volume and blood rheology. The molecular factors controlling shear stress-induced NO release are complex and involve “ultra-quick” ion channel opening, “quick” phosphorylation of endothelial NO synthase (eNOS) and “slow” eNOS gene transcription.27 The balance between Ca2+, K+ and chloride channel opening determines the net membrane potential and the subsequent change in Ca2+ that affects eNOS activation and NO output. Application of shear stress to aortic endothelial cells leads to an immediate large increase of intracellular free Ca2+. Notably, intracellular Ca2+ rise occurs primarily in response to regular pulsatile flow,27 even though laminar non-pulsatile flow can induce some Ca2+ increase28 and this may be at least in part dependent on the composition of the media.29 Whether the calcium-independent NO phosphorylation, as well as the stimulation of eNOS gene transcription for long-term NO production requires regular pulsatile blood flow, is unknown. Unexplored are also the respective roles of signalling molecules on eNOS activation, such as bradykinin, adenosine, vascular endothelial growth factor (in response to hypoxia) serotonin,30 and of prostaglandins and endothelium-derived hyperpolarising factor, which are co-released with NO during FMD.31

Fibrillating atrium influencing the vessels

NO production in the left atrium exceeds that in the right one and is two to three times greater than that in the arterial endothelium.26 It has been proposed that nitrosothiols formed by the reaction of NO with compounds like haemoglobin, glutathione and albumin may serve as circulating forms of NO, affecting vascular tone throughout the systemic circulation.26 Based on these premises and considering that the entire stroke volume passes across the left atrium, Cai and co-workers26 advanced the hypothesis that left atrial endocardium may serve as an endocrine organ that releases nitroso compounds into the circulation to modulate vascular function. Because the loss of an organised atrial contraction in AF is associated with a marked decrease of endocardial NOS expression, these authors suggested that at least some of the changes in the circulating forms of NO in patients with AF13 may be attributable to decreased production of NO in the left atrium, and that cyclic stretch of atrial endocardial cells and laminar blood flow are two prerequisites for the NO-dependent circulatory homeostasis. Bruins and collaborators32 were the first to report an association between C-reactive protein (CRP) elevation and occurrence of AF, in the context of AF following cardiac surgery. Although an increase in CRP may simply be a consequence of associated comorbid factors such as diabetes and hypertension, Chung and co-workers33 found that patients with atrial arrhythmias had levels of CRP higher than controls, suggesting that arrhythmia patients in AF within 24 hours before sampling had CRP higher than paroxysmal AF patients, and both groups showed higher CRP levels than controls. The interpretation was that AF may induce or be provoked by atrial,34 vascular or extravascular inflammation and the degree of elevation of CRP may be related to the “burden” or type of AF. Interestingly, in a recent report obtained in 52 patients with persistent AF lasting more than three months, high levels of CRP were found to be associated with an increased risk of AF recurrence after cardioversion.35

An atrial source of inflammation may increase serum proinflammatory cytokines, which stimulate hepatic production of CRP. On the other hand, circulating CRP36 and cytokines, like tumour necrosis factor alpha (TNF-α), exert proinflammatory effects on endothelial cells that potentially result in a systemic noxious endothelial effect, thus leaving unresolved the issue of which comes first: fibrillating atrium or peripheral vessels?

Systemic factors and processes

Observations made over the last few years suggest that the renin-angiotensin system (RAS) may have a significant part in the pathogenesis of AF3739 through several putative mechanisms. The strongest arguments in favour of this view come from the large number of reports showing that RAS inhibition prevents emergence and recurrence of AF. Increasing evidence would support the concept of a reciprocal “cross-talk” relation between angiotensin II and inflammation,39 with the former causing the latter and inflammation acting as a stimulus for increased angiotensin II production. Actually, human atrial tissue possesses the ability of locally expressing angiotensin-converting enzyme (ACE) for local angiotensin production. AF increases angiotensin II receptor expression, which is involved in increased atrial cell death and leucocyte infiltration.40

A potent link between AF and atrial electrophysiological and structural remodelling is oxidative stress. Formation of oxygen-derived free radical by NAD(P)H oxidases is involved in the development of a wide range of cardiovascular diseases. Dudley et al34 showed activation of several ROS generating enzymes in AF. Moreover, Kim and co-workers41 showed that this oxidase system contributes significantly to superoxide production in the fibrillating human myocardium and suggested a key role in atrial oxidative injury and remodelling through a decrease of NO bioavailability produced by scavenging NO to form peroxinitrite and uncoupling NOS activity. Vascular NAD(P)H oxidase is known to be greatly increased by angiotensin II,42 suggesting that RAS system activation may be an important underlying factor for the increased oxidase activity and ensuing myocardial and vascular oxidative stress in human AF. The favourable interference that ACE-inhibitors and angiotensin II receptor blockers exert with myocardial-endothelial superoxide production41 might be one of the mechanisms whereby these agents can prevent emergence and recurrence of AF.43

Thus, RAS may play a key part in making atrial endocardium and vascular endothelium twin targets of the same systemic process. However, whether these links are merely associations or causal, is still an open question that requires further prospective and interventional studies.


The issue of whether AF worsens the patient’s functional status has been assessed with variable methods, including respiratory gas analysis, Borg score, peak oxygen uptake (VO2),44 treadmill exercise time,4547 quality of life46 and 6-minute walk distance.47 The investigated populations were rather inhomogeneous mainly regarding the underlying cardiac (heart failure, ischaemic heart disease) or systemic disorder (hypercholesterolaemia, high blood pressure) and current drug treatments. All in all, evidence is in favour of a better functional status and quality of life with reversion to sinus rhythm. As to the pathogenetic bases sustaining a reduced exercise performance during AF, there is no question that cardiac output may become an important limiting factor, owing to a series of mechanisms including loss of atrial contribution to left ventricular filling, atrioventricular valve regurgitation, increased ventricular rate, reduced diastolic time and irregular R-R interval. However, the role of the periphery seems to be of equivalent importance in this respect. The study of Kuvin et al45 was the first to show a possible correlation between endothelium-dependent vasomotion and exercise capacity in AF, stressing the need for studies focusing on endothelial function and exercise ability.

We dealt with this issue in two studies. In the first one,23 35 patients while having lone AF, exhibited an impaired ventilatory efficiency as documented by a steeper ventilation (VE) versus carbon dioxide output (VCO2) slope on exercise, for any matched exercise time, compared with sinus rhythm after CV. That this might depend on the loss of atrial contraction increasing the pulmonary venous pressure and the alveolar interstitial fluid amount, was denied by the documentation that the alveolar capillary gas diffusion and exercise ventilation/perfusion matching as indicated by the dead space to tidal volume ratio (VD/VT) were normal and unaffected by CV. Conversely, end-tidal partial pressure of CO2 (PETCO2) was significantly reduced throughout exercise and increased to normal with CV, making it likely that extrapulmonary factors were at work. In these patients anaerobic threshold occurred at a VO2 18% less than after CV along with an increased VCO2, suggesting the occurrence of premature work rate (WR) lactic acidosis that enhances the ventilatory component of the exercising muscle ergoreflex.

In another study48 we examined whether CV of lone AF to sinus rhythm is, as far as endothelial function and exercise performance are concerned, equally beneficial in diabetes and hypertension, endothelium disturbing diseases that are often associated with AF, as it is in lone AF. Compared with lone AF, exercise workload, VO2, aerobic efficiency (ΔVO2/ΔWR) and the ratio of brachial artery diameter changes to flow changes (ΔD/ΔF) were reduced in hypertension comorbidity and, to a greater extent, in diabetes comorbidity. VE/VCO2 slope and VD/VT were similar among groups. CV had less effect on workload, VO2, ΔVO2/ΔWR, VE/VCO2 slope, ΔD/ΔF and breathlessness (Borg scale) in hypertension than in lone AF and was not significantly effective in diabetes (fig 1). Vitamin C improved FMD in lone AF and hypertension before but not after CV and was not effective in diabetes, suggesting that the oxidative injury is least in lone AF, greater in hypertension with AF and probably greater still in diabetes with AF.27 An additional indication was that advantages of CV, with respect to the patient’s functional status, may be inversely related to the extent of the underlying oxidative injury. Interestingly enough, a 30-year follow-up study, recently reported by Jahangir et al,49 has emphasised the fundamental role of comorbidities (hypertension, diabetes, heart failure) in association with ageing, in the AF progression and cerebrovascular complications. The suggestion was that AF with comorbidities may reflect the final common pathway of a vascular inflammatory process.

Figure 1 Slope of ventilation (VE) v CO2 production (VCO2) and dyspnoea sensation (Borg scale) plotted over each minute of ramp incremental exercise before (solid circles) and after (open circles) cardioversion to sinus rhythm, in patients with lone atrial fibrillation (AF) (group 1, 18 cases), or AF with high blood pressure (group 2, 19 cases), or AF with diabetes mellitus (group 3, 19 cases). (From Guazzi et al.48)

Whether AF also impairs endothelial function in the pulmonary circulation, and CV may reduce pulmonary vascular resistance and impedance to right ventricular ejection, improve ejection fraction and increase exercise capacity correlates of these effects50 is, at present, unexplored.



  • Competing interests: None.