The most important predictor of AF progression is possibly AF itself[7]. At an early stage, AF determines an atrial electrophysiological, mechanical and structural atrial remodeling by shortening, mismatching and lengthening the atrial effective refractory periods (increase of dispersion) and by the depression of intra-atrial conduction and the loss of contractile function. The electrical, mechanical and structural remodeling determines the perpetuation of AF and the progression from paroxysmal to persistent and permanent forms. The longer one waits to initiate a rhythm treatment strategy, the more difficult it is to regain sinus rhythm. Dittrich et al[8] showed that patients who converted to sinus rhythm within 3 mo of onset of AF were more likely to maintain sinus rhythm at 6 mo than patients who converted more than 12 mo after onset of AF (67% vs 27%). By shortening the atrial refractory period, reducing conduction velocity and provoking contractile and structural remodeling, AF provokes AF[9].

Almost any valvular lesion that leads to significant stenosis or regurgitation is associated with the development of AF. In patients with degenerative mitral regurgitation in sinus rhythm at diagnosis, the incidence of AF occurring under conservative management is high and similar whether the cause of mitral regurgitation is flail leaflet or simple mitral valve prolapse. After onset of AF, an increased cardiac mortality and morbidity are both observed under conservative management[10]. Rheumatic heart disease is now uncommon in developed countries. It is, however, associated with high prevalence of AF. The highest frequency of AF in rheumatic heart disease occurs in those with mitral stenosis, mitral regurgitation, and tricuspid regurgitation in combination. AF, while occurring in 29% of patients with isolated mitral stenosis and in 16% with isolated mitral regurgitation, is an infrequent finding (1%) in patients with aortic valvular disease[11]. In addition, John et al[12] compared patients with mitral stenosis with 24 control patients. Patients with mitral stenosis showed, not only left atrial enlargement, but also a significantly reduced biatrial voltage (left atrium 1.8 + 0.6 mV vs 3.6 + 0.6 mV, right atrium 1.9 + 0.6 mV vs 3.3 + 0.5 mV), reduced conduction velocity, and prolonged effective refractory periods. These abnormalities may clearly play a role in the increased propensity to AF in patients with mitral stenosis.

The association between hypertension and AF is well established. A history of hypertension increases 1.42-fold the risk of developing AF[13]. Although the increase in risk is relatively modest (relative risk, 1.2-1.5), the high prevalence of hypertension in the general population renders it the most significant population-attributable risk factor for AF beyond age and sex. It is observed that hypertension is responsible for 14% of all cases of AF[14]. Although overt systolic hypertension is strongly related with the progression of AF, recent studies demonstrated that systolic blood pressure in the prehypertensive range (130-139 mmHg) and widened pulse pressure are also associated with increased incidence of AF[15,16]. Mean arterial pressure does not seem to be related with AF.

AF occurs transiently in 6%-10% of patients with acute myocardial infarction, presumably due to atrial ischemia or atrial stretching secondrary to HF[17]. These patients have a worse prognosis that is mostly due to comorbidities such as older age and HF. The Coronary Artery Surgical study which included 18000 patients showed that the incidence of AF is much lower (0.6%) in patients with chronic stable coronary artery disease (CAD)[16]. These patients probably had chronic AF; the prevalence of paroxysmal AF may be higher. AF was an independent predictor of increased mortality in patients with stable CAD[18]. Coronary artery disease can promote AF via multiple mechanisms. Myocardial infarction often causes substantial left ventricular dysfunction and HF predisposing to AF. Acute atrial ischemia/injury promotes AF by causing important atrial conduction disturbances, likely related to impaired cell-to-cell coupling[19]. Healed atrial infarctions and persistent ischemia enhances AF by causing Ca²⁺ - handling abnormalities, resulting in delayed afterdepolarizations and triggered activity resulting in ectopic firing, along with structural remodeling and reentry[20]. Chronic atrial coronary artery occlusion in conjunction with autonomic activity promotes ectopic firing and AF.

Aging is accompanied by atrial structural remodeling associated with substantial conduction abnormalities[21]. Gaborit et al[22] showed that men have greater expression of important repolarizing ion channel subunits, which could enhance atrial repolarization, shorten atrial refractoriness, and favor reentry. Moreover, men have greater left atrial dimensions that could promote AF maintenance[23].

Diabetes mellitus is an independent determinant of AF prevalence but predicted incidence only among women[24]. Over a mean follow-up of 7.2 years, diabetic patients without AF at baseline developed AF at an age/sex-adjusted rate of 9.1/1000 person-years, compared with 6.6/1000 person years among non-diabetic patients. Diabetes mellitus was associated with 26% increased risk of AF among women, but diabetes was not a statistically significant factor among men. Diabetes mellitus elicits AF via both structural remodeling, mediated by advanced glycosylation end products[25] and autonomic nervous system remodeling[26].

AF and HF often occur together and each may predispose to the other. There is continuing controversy as to whether HF is merely a common coexisting condition among patients with AF or whether it is a true causative factor. Among patients with HF, the prevalence of AF is variable, depending in part upon the severity of HF. The association is not limited to systolic left ventricular dysfunction but also AF is combined with diastolic dysfunction of the left ventricle[27]. Isolated diastolic dysfunction is associated with an increased AF incidence, possibly reflecting shared risk factors such as advancing age and hypertension. Although the association between AF and HF is well established, the causative relationship between the two has not been fully elucidated. Probably, AF can cause reductions in cardiac output (because of shorter diastolic filling time, loss of atrial contractile function, and elevated filling pressures) and tachycardia-induced cardiomyopathy[28]. HF results in structural and electrical remodeling changes that predispose to AF.

AF has been reported in 10%-28% of patients with hypertrophic cardiomyopathy (HCM)[29]. AF is the most common arrhythmia in patients with HCM. Olivotto et al[30] evaluated 480 patients with HCM with a mean follow-up of 9.1 years and found the prevalence of AF to be 22%. More recently, a study in Japan examined 261 patients with HCM and found that 74 (28%) patients had documented paroxysmal AF or permanent AF[31]. The high prevalence of AF in HCM is related to atrial dilation and remodeling in the setting of diastolic dysfunction, mitral regurgitation, and atrial fibrosis[30,31]. AF is an important prognostic indicator in patients with HCM, because these patients are typically at a higher New York Heart Association functional class and have a poorer outcome. This subgroup of patients with HCM is at an increased risk of cardiovascular morbidity and mortality in the form of thromboembolic events, HF, and sudden death[32]. In a systematic review, Kumar et al[33] reported that in HCM brain natriuretic peptide, left atrial size (left atrial volume measured with cardiac magnetic resonance), higher left atrial mean extent of late gadolinium enhancement in cardiac magnetic resonance, left ventricular myocardial fibrosis determined by delayed contrast enhancement, sleep apnea, longer P-wave duration, genetic factors, and ischemia are associated with AF progression.

AF is common in patients with dilated cardiomyopathy (DCM). Epidemiologic studies have shown that 30%-40% of patients with left ventricular dysfunction and systolic HF from any cause will develop AF during the course of their disease, and AF has been associated with increased morbidity and mortality[34-36]. In experimental subjects, the increased incident of AF is associated with atrial structural abnormalities, with increased atrial fibrosis associated with slowing conduction of velocity and conduction heterogeneity[37]. In humans, Sanders et al[38] also showed that AF in patients with left ventricular dysfunction is associated with widespread areas of low voltage and electrical silence consistent with scar, and with regional atrial conduction slowing with prolongation of the P-wave duration, in addition to altered sinus node function. Pulmonary veins are responsible for arrhythmia initiation[39]. Atrial electrical and structural remodeling outside the pulmonary veins is the substrate of maintenance of persistent AF. Rotors, or high-frequency sources within the atrium, have been recently proposed as mechanisms for both initiation and maintenance of persistent AF[40].

In 2007, the European Society of Cardiology working group on myocardial and pericardial diseases redefined cardiomyopathies including peripartum cardiomyopathy (PPCM), which it is defined as a form of DCM that presents with signs of cardiac failure during the last month of pregnancy or within 5 mo of delivery[41]. Limited data regarding the association of PPCM and AF exist in the literature. Biteker et al[42] studied 42 women with PPCM. Five of them (11.9%) had AF and AF had no apparent effect on survival or recovery of left ventricular function. Kane et al[43] examined 33 women with PPCM and 1 (3%) of them had AF. Finally, Isezuo and Abubakar[44] showed that 2 out of 65 women (3.1%) developed AF strengthening the observation that PPCM is associated with AF.

AF is more prevalent in patients with chronic renal dysfunction (CRD). AF risk increases with severity of kidney dysfunction (HR of 1.3-1.6 and 1.6-3.2 with an estimated glomerular filtration rate of 30-59 and < 30 mL/min per 1.73 m², respectively, vs estimated glomerular filtration rate ≥ 60 mL/min per 1.73 m²)[45]. These two entities (AF and CRD) share common associated factors such as coronary heart disease, HF, hypertension, left ventricular hypertrophy and systemic inflammation. In addition, macroalbuminuria and microalbuminuria were significantly associated with higher AF risk.

Accumulating data have demonstrated a clear and significant association between obstructive sleep apnea (OSA) and AF[46,47]. The occurrence of AF in 400 middle-aged patients who had moderate or severe OSA (apnea-hypopnea index ≥ 25) was more than 3%. Furthermore, twelve of the study patients who underwent tracheostomy, bypassing the obstructed airway, had complete elimination of AF up to 6 mo later, something that clearly shows the straight correlation between AF and OSA[46]. In the largest registry until now, Gami et al[47] showed that OSA and AF were significantly associated. Body mass index and the decrease in nocturnal oxygen saturation were independent predictors of AF. This study, also, proves the correlation between obesity and AF. Multiple pathophysiological pathways link OSA with AF. Increased left atrial size, hypertension and diastolic dysfunction may coexist in OSA and AF[48]. AF probably occurs as a complex interaction of several hemodynamic and sympathetic consequences of OSA. These include autonomic dysregulation[49], elevated sympathetic tone, oxidative stresses, endothelial dysfunction, and left atrial stretch[50]. OSA is associated with systemic inflammation, increased levels of C-reactive protein (CRP), serum amyloid A, and interleukins[51]. These observations makes us believe that OSA and AF share common pathways, which contribute to atrial fibrosis and structural and electrical remodeling. Finally, Al Chekakie et al[52] showed that central obesity and pericardial fat is associated with AF. Pericardial adipose tissue contributes to inflammation and progression to AF. Patients with paroxysmal AF had significantly greater pericardial fat volume on average than patients in sinus rhythm (93.9 mL vs 76.1 mL) and the persistent AF patients had a significantly larger volume of pericardial fat volume on average than the paroxysmal AF patients (115.4 mL vs 93.9 mL).

AF has been reported in approximately 20% of adults with an arial septal defect[53]. AF and atrial flutter also occurs in other forms of congenital heart disease that affect the atria, including Ebstein’s anomaly[54] and patent ductus arteriosus[55], and after surgical correction of some other abnormalities, including ventricular septal defect, tetralogy of Fallot, pulmonary valve stenosis, and transposition of the great vessels.

Patients with hyperthyroidism have an increased risk of AF progression[56]. Frost et al[57] showed that among 40628 patients with clinical hyperthyroidism, 8.3% had ΑF or atrial flutter. Increased beta adrenergic tone play a crucial role for the development of AF in hyperthyroidism, often combined with rapid ventricular response. Furthermore, hyperthyroidism increases the likelihood of AF in experimental models, even in the presence of beta receptor and vagal blockade[58]. The pathophysiology remains unknown, but may be related to an increased automaticity and enhanced triggered activity of pulmonary vein cardiomyocytes[59]. The risk for development of AF is also increased in patients with subclinical hyperthyroidism[60,61]. It remains controversial whether patients with AF associated with previous treated thyrotoxicosis are at increased risk of thrombo-embolism, in the absence of other known risk factors[62].

AF is associated with a variety of other types of cardiopulmonary disease. AF is seen in 10% to 14% of patients with documented pulmonary embolism[63]. AF also occurs in chronic obstructive pulmonary disease[64], myocarditis[65] and acute pericarditis[66]. In addition, electrolytic disturbances like hypokalemia or low serum magnesium[67] initiates AF. Alcohol consumption contributes, also, to the development of AF[68]. Finally, prior surgery, especially and coronary artery bypass grafting[69] predispose to AF.

Left atrial size is a well-known predictor of AF development. A left atrial size greater than 4 cm has been associated with a significantly higher AF recurrence rate[85]. The left atrial volume measured by transthoracic echocardiography is possibly superior to left atrial diameter in predicting progression to persistent AF[86]. Li et al[87] reported that the E/e’ index (E, early transmitral flow velocity; e’, early diastolic mitral annular velocity), an index of diastolic dysfunction, was the best independent predictor of AF recurrence after catheter ablation. E/e’ > 11.2 before ablation has been associated with AF recurrence. Shaikh et al[88] showed that speckle left atrial strain and stiffness index can predict the possibility of maintenance in sinus rhythm after cardioversion for AF. Changes in longitudinal left atrial strain (peak systolic longitudinal strain) after cardioversion were significantly higher among individuals who remained in sinus rhythm when compared to individuals with recurrent AF[88].

Late gadolinium enhancement-magnetic resonance imaging (LGE-MRI) allows the direct visualization of atrial arrhythmic substrate. Vergara et al[89] described a new staging system for AF based on the amount of left atrial enhancement on LGE-MRI, the Utah score (Utah I ≤ 5%, Utah II > 5%-20%, Utah III > 20%-35%, and Utah IV > 35%). On the basis of patient stage, a more tailored approach to AF management can be taken. Patients with a previous stroke had a significantly higher percentage of left atrial fibrosis compared with those without (24.4% ± 12.4% vs 16.1% ± 9.8%, P≤ 0.001). There was a significant difference in the rate of thromboembolism between patients with stage I and those with stage IV of atrial remodeling as assessed by LGE-MRI. In addition patients with CHADS2 score ≥ 2 had higher amounts of left atrial fibrosis. The DECAAF study showed that left atrial fibrosis contributes to the progression of AF. The more fibrosis there is, the more likely the arrhythmia will persist following ablation[90]. Atrial fibrosis estimated by LGE-MRI in 260 AF patients, including 65% with paroxysmal AF, was a significant predictor of recurrence. Each 1% increase in fibrosis was associated with a 6% increased risk of recurrence. Fibrosis was categorized as stage 1 (< 10% of the atrial wall), 2 (≥ 10%-< 20%), 3 (≥ 20%-< 30%), and 4 (≥ 30%). At one year, 88% of patients with stage 1 fibrosis were free of AF. For those with stage 2, 3, or 4 fibrosis, 69%, 55%, and 45% were free of recurrence at one year, respectively. At 475 d, 86%, 64%, 51%, and 35% of those with stage 1, 2, 3, and 4 fibrosis were free of AF, respectively. Electroanatomic bipolar voltage mapping has proved to have great correlation with DE-MRI. Jadidi et al[91] have demonstrated bipolar voltages of 0.63 ± 0.8 in dense DE-CMRI areas, compared with 0.86 ± 0.89 in non DE-MRI areas. Moreover, Spragg et al[92] have demonstrated that the mean atrial voltage in areas identified as scar by DE-MRI was 0.39 ± 0.61 mV, while in areas identified as normal by DE-CMRI was 1.38 ± 1.23 mV.

There is now increasing evidence that even in patients with so-called lone or idiopathic AF, the AF is an arrhythmic manifestation of a structural atrial disease which has recently been defined and described as fibrotic atrial cardiomyopathy (FACM). Different expressions can be found from mild (FACM I), moderate (FACM II) to excessive fibrosis (FACM III), and wide clinical variations from asymptomatic to multiple arrhythmic manifestations (including AF, left and/or right atrial re-entrant tachycardia, sinus, and/or atrioventricular node disease)[93]. Fibrosis results from a broad range of factors related to AF inducing pathologies such as cell stretch, neurohumoral activation, oxidative stress, and possibly even AF itself[94]. Stiles et al[95] investigated 25 patients with “lone” AF, during an electrophysiological study after at least 7 d in sinus rhythm, and found slower conduction velocity, longer effective refractory periods, and significantly lower voltages (left atrium 1.7 + 0.7 mV vs 3.3 + 0.7 mV, right atrium 1.7 + 0.4 mV vs 2.9 + 0.4 mV) compared with control patients without AF. These findings confirm a substantial chronic structural biatrial substrate since the electrical remodelling is reversible within a few days. It might be that not all patients with paroxysmal “lone” AF have an underdetected chronic substrate, but many more than assumed. The debate is whether the fibrosis is causative or merely a result of AF. Several data suggest that fibrosis is causative and that AF-induced fibrosis may be part of the vicious cycle. In animal models, reversal or prevention of fibrosis prevents AF[96]. Furthermore, AF substrate in the absence of any cellular electrophysiological abnormalities has been demonstrated in a transgenic mouse model of isolated atrial fibrosis[97].