CAFs are infrequent abnormalities of coronary artery termination. In our angiographic series, the prevalence of CAFs was 0.2%.

Although in earlier studies CAFs were most often described as involving the right coronary artery, more recent studies have found CAFs to arise most frequently from the LAD.2–4 The most commonly reported drainage site is the pulmonary artery.2,4,5 As in those studies, most CAFs in our series originated from the LAD and the most common pattern was between the LAD and the pulmonary artery.

Drainage of CAFs into the left heart chambers has been reported as a very rare finding.2 However, in our series the left ventricle was the second most common drainage site, either in the form of one or more distinct small channels or as multiple microfistulas. Canga et al.4 also reported the left ventricle as the second most common drainage site in their series.

While CAFs are most commonly congenital, they can be acquired, as in the setting of infective endocarditis or chest trauma. Although rarely reported, acquired CAFs can also be iatrogenic, occurring after procedures such as percutaneous coronary intervention, CABG, valve replacement, permanent pacemaker implantation and endomyocardial biopsy.1 Chiu et al.6 described an incidence of 0.44% of acquired CAFs after open heart surgery for congenital heart disease.

In our study, 10.9% of patients (n=6) had a history of open cardiac surgery. In one of those patients, who had undergone CABG and aortic/mitral valve replacement surgery, the fistula was considered to be clearly iatrogenic, since it was not present in the preoperative ICA. In the four patients who had undergone previous open cardiac surgery for correction of congenital heart disease, the fistula could also have been iatrogenic, but we cannot state this with certainty as these patients had not undergone a previous ICA for comparison. In the patient with a heart transplant, as all the endomyocardial biopsies were performed in the right ventricle and the fistula was from the distal LAD to the left ventricle, we could not find a direct cause-effect relationship.

Congenital CAFs occur most frequently as an isolated anomaly, but they can be associated with another congenital heart disease in 10-45% of cases.1 Concomitant congenital cardiovascular disorders were present in 12.7% of the patients in our series, the most frequent being atrial septal defect.

The observed high prevalence of cardiovascular risk factors among this population is probably explained by the fact that CAFs were detected incidentally in a large number of cases during a coronary angiography performed due to suspected atherosclerotic coronary artery disease (acute coronary syndromes or assessment of stable angina).

The clinical presentation of CAFs is mainly determined by the shunt volume, site, duration and the presence of underlying cardiac disease. Most CAFs are clinically silent.7 However, volume overload of both ventricles or solely of the left chambers (depending on the drainage site) can result in heart failure, atrial or ventricular arrhythmias and pulmonary hypertension. Fistulas can also give rise to the coronary steal phenomenon, which causes ischemia by diverting blood from the high-resistance myocardial capillary bed into the low-resistance fistula and by preventing coronary flow reserve from increasing during exercise.7 Dilatation of the fistula and affected coronary artery can occur with time and give rise to complications such as local compression or rupture.

Clinical presentations include congestive heart failure, chest pain and palpitations, while a soft continuous murmur (crescendo-decrescendo in both systole and diastole) is the most commonly reported physical examination finding in CAF patients.5,8 Presentation can also be related to complications such as myocardial infarction, infective endocarditis and rupture or thrombosis of a fistula or an associated aneurysm.5,9 Premature coronary atherosclerosis has also been reported.7

Regarding the 15 patients in our series in whom CAFs were considered to be the main cause of the patient's clinical status, chest pain was the most common symptom (40%), while heart murmur was the most frequent sign (53.3%). As for the other patients, in whom fistulas were considered to be an incidental finding (n=40), symptoms are difficult to discern due to the presence of concomitant disease which might explain them, particularly obstructive coronary disease or valvular heart disease.

In patients with acute coronary syndromes and concomitant coronary fistulas, the relationship with the fistula is also difficult to ascertain. Responsible mechanisms may be thromboembolic complications due to turbulent flow in the fistulous vessel and/or hypoperfusion of the related main coronary artery.10 In one patient in our series, non-ST-elevation myocardial infarction was considered to be directly attributed to a large left main-pulmonary artery fistula. In the other cases of acute coronary syndrome, the contribution of the fistulas was less certain.

Cardiac catheterization with coronary angiography remains the gold standard for detecting CAFs,1 also providing hemodynamic and shunt calculation data. In this series, coronary angiography was the standard for diagnosis in all patients, and multidetector computed tomography (MDCT) was performed in cases in which additional information on the fistula anatomy was needed. Cardiac magnetic resonance (CMR) and MDCT can provide further details when the CAF anatomy is difficult to discern by ICA.11 In particular, MDCT provides detailed information on the anatomy of the coronary arteries and fistulas, the patency of the shunt and the relationship with adjacent structures, and also enables assessment of the venous system. CAF prevalence has been reported to be higher in MDCT-based series than in ICA series.12,13

Transthoracic echocardiography (TTE) can detect CAFs but is often unable to delineate the fistula pathway, which could be improved with the use of transesophageal echocardiography.14,15 Even so, TTE remains vital for assessment of chamber size, systolic function and segmental contractility, as for investigation of associated cardiac anomalies.

Exercise electrocardiography and myocardial perfusion imaging or CMR can be valuable tools for the assessment of myocardial ischemia. However, it has been reported that stress testing with vasodilators could lead to negative results for ischemia due to the poor vasodilatory capacity of the fistula compared to the native coronaries, favoring flow in the original feeding vessel.7 Positron emission tomography scanning can be of great value as a non-invasive diagnostic technique to assess flow ratios of the different coronary arteries and to quantify myocardial perfusion reserve,10 but it was not applied in our patients.

The optimal management of CAFs remains controversial. Treatment options include TCC, surgical ligation and pharmacological treatment with follow-up over time. Recently, it has also been demonstrated that radiofrequency ablation may be applied for the closure of CAFs.16,17 The decision depends on the patient's age and symptoms as well as the anatomic and functional characteristics of the fistula and the risk of future complications. While it is clear that hemodynamically significant fistulas do require intervention, there is no agreement on the ideal management of small and/or asymptomatic fistulas. Regarding children, elective closure of any clinically apparent CAF after 3-5 years of age should be performed, even if the patient is asymptomatic.18 Algorithms for assessment and treatment of patients with CAFs have recently been suggested by Karazisi et al. and Buccheri et al.1,7

In our series, watchful waiting was the main approach. TCC was performed in eight patients and cardiac surgery was performed specifically for fistula closure in one patient. Six other patients had their fistulas closed during cardiac surgery for another cause. In the pediatric population undergoing intervention, percutaneous closure was conducted in all cases.

When intervention is indicated, catheter-based interventional techniques have become the preferred option in the current era, if technically feasible.9,19,20 On the other hand, surgical ligation is indicated in distal fistulas, large, high-flow fistulas, multiple complex communications, tortuous arteries, prominent aneurysms, a wide drainage site, presence of large vascular branches that can be accidentally embolized, or if the patient has associated cardiac lesions that need surgical correction.1 Both surgical and percutaneous techniques have been associated with good outcomes and a low risk of complications and seem to be equally effective.2,20,21,22 Nevertheless, periprocedural mortality and complications such as myocardial infarction have been reported.21 Among the 15 patients who underwent intervention in our series, there was one case of myocardial infarction and pericarditis immediately after TCC. There were no periprocedural deaths or procedure-related complications during follow-up. One patient needed a percutaneous reintervention due to persistent residual flow in the fistula.

Concerning pharmacological treatment, this generally includes antianginal therapy addressing the demand and supply mismatch induced by the coronary steal phenomenon and antiplatelet or anticoagulant therapy in selected cases. Although not endorsed by current guidelines, most authors recommend prophylaxis against bacterial endocarditis regardless of the type of fistula.1,7,8,18,21