Anatomical Classification and Posttreatment Remodeling Characteristics to Guide Management and Follow-Up of Neonates and Infants With Coronary Artery Fistula: A Multicenter Study From the Coronary Artery Fistula Registry

CAF in neonates and infants was identified in 48 patients from a total of 115 patients from 20 centers participating in the Congenital Cardiovascular Interventional Study Consortium CAF registry. Demographics, clinical features, diagnostic studies, and treatment are listed in Table 1. At baseline evaluation, an audible murmur was present in all patients. HF symptoms were reported in 28 of 48 (58%) patients. Nonspecific ST changes on ECG were noted in 6 and ventricular dysfunction by echocardiography in 3 patients. Based on angiographic evaluation, the CAFs were classified as proximal in 30 (62%) patients and distal in 18 (Figure 1). The CAF caliber was large in 39 patients (81%). Most CAFs arose from the left coronary system (65%) and drained predominantly to right sided heart structures (88%). The median pulmonary to systemic flow (Qp:Qs) ratio for the cohort was 1.73. In a total of 42 of 48 (87.5%) patients, 24 had TCC and SC was performed in 18. Indication for closure was HF in 28 of 42 (66%) patients, and the remaining 14 were asymptomatic. These 14 patients underwent closure for hemodynamic evidence of Qp:Qs ≥1.5, chamber dilation, and for concerns of fistula size progression with age. Six of 48 (12.5%) patients were placed on medical observation. Of these, 1 died due to HF and ischemic symptoms before intervention, treatment was deferred for later in childhood in 2 patients, and the remaining 3 had severely restrictive large CAF with the plan to continue observation and follow-up. Procedural success was 92% for TCC and 94% for SC. There were no procedure-related deaths, strokes, or infections.

Follow-up was available in 34 of the 48 patients (70%) with a median follow-up duration of 2.9 (range, 0.1–18) years. All surviving patients were asymptomatic on clinical evaluation with normal EKG and echocardiogram (Table 2). Coronary angiograms to assess posttreatment remodeling characteristics (Figure 2) were available in 20 patients (Table 3).

I. Optimal remodeling was noted in 10 patients. Among them, 7 with proximal CAF demonstrated fistula segment occlusion with thrombus without encroaching the adjacent epicardial arteries. The remaining 3 were distal CAF showing thrombus extension from point of closure to the nearest coronary branch, and the conduit/native coronary artery demonstrated regression in caliber toward normal (Figure 3).

II. Suboptimal remodeling was observed in 7 patients and this included the following:

A. Symptomatic coronary thrombosis occurred in 2 patients, and both had distal CAF. An infant with a large right coronary artery (RCA) fistula (Figure 4B through 4D) had TCC using a vascular plug and was placed on aspirin only. One week later, the infant became extremely irritable with elevated troponins, and an echocardiogram demonstrated moderate right ventricular dysfunction. The infant was treated medically with anticoagulation and a β-blocker. A follow-up coronary angiogram was not performed. The second patient had a moderate sized RCA fistula and underwent TCC using coils. An immediate postclosure angiogram demonstrated sluggish flow with thrombus extending proximally along the conduit RCA occluding nearby coronary branches (Figure 4E through 4H). The troponins were elevated with qualitatively normal right ventricular function. The patient was treated with intravenous heparin and transitioned to warfarin and aspirin.

B. Asymptomatic coronary thrombosis and revascularization with collaterals and vascular channels was observed in 3 patients. One patient was a 7-day-old neonate with a large proximal left coronary artery fistula who underwent an emergent SC at the drainage site. A surveillance angiogram at 4 years of age demonstrated asymptomatic thrombus extension proximally in the fistula segment with occlusion of left circumflex artery and retrograde filling of the left circumflex artery with collaterals (Figure 5B through 5D). The remaining 2 patients had large distal RCA fistula with distal closure. Follow-up angiograms showed complete thrombus occlusion of the entire conduit RCA in one and partial occlusion of the distal RCA with thrombus in the other. Both angiograms demonstrated revascularization of the conduit RCA and its branches with multiple thread-like vascular channels (Figure 5F through 5H).

C. Partial thrombosis was noted in 2 patients. Partial thrombus extension with a residual cul-de-sac in the proximal fistula segment was observed following SC distally in one patient with a proximal CAF. Vessel irregularity and area of stenosis due to partial thrombus in the conduit left circumflex artery was noted in 1 neonate with a distal CAF following TCC (Figure 6).

III. Persistent conduit coronary artery or fistula segment dilation was attributable to residual flow and observed in 4 patients (Figure 7). One patient with a proximal CAF had fistula segment dilation with a tiny jet of residual flow following SC. The remaining 3 had distal CAF with conduit artery dilation and residual flow due to intentional transcatheter partial closure in 1 and adjacent microfistulae in 2.

Overall, suboptimal remodeling was noted in 7 of 21 (33%) patients, and 6 of these had no anticoagulation post-closure. Comparing proximal versus distal CAF, 7 of 10 (70%) versus 3 of 11 (27%) patients had optimal and 2 of 10 (20%) versus 5 of 11 (45%) had suboptimal remodeling, respectively (Table 3). There was no significant difference associated among various factors between the groups (Table 4).

The rarity of CAF—a presumption that the patient is fine after the fistula is closed—and the lack of consistent follow-up evaluations have all led to a poor understanding of the actual outcomes of CAF. Valente et al² in their series reported coronary thrombosis on follow-up in 15% of their 76 patients and emphasized older age, hypertension, diabetes, hyperlipidemia, and drainage to coronary sinus as predictors of adverse coronary events. Gowda et al^11,12 were the first to highlight older age and large distal CAF as important risk factors for thrombosis causing coronary events. It is becoming apparent that post-CAF closure, sudden cessation or severe reduction of flow velocity in a dilated vascular segment is associated with a high likelihood of thrombus formation in that segment.

A proximal CAF closed distally causes stasis in the residual fistula segment leading to thrombosis that extends proximally up to its origin from the conduit coronary artery (Figure 1).^11,12 Most proximal CAF patients in the present cohort, 7 of 10 (70%), had evidence of optimal remodeling. However, in a severely dilated fistula segment, the thrombosis may extend proximally and block the adjacent epicardial coronary artery, as demonstrated in one patient with left circumflex artery occlusion (Figure 5). There are two other cases reported with a coronary event post-closure of a proximal fistula. These appeared to have thrombus extension into the dilated aortic sinus occluding the adjacent epicardial artery and required treatment with fibrinolytics and anticoagulation.^13,14 In addition, partial thrombus extension can leave a residual cul-de-sac in the proximal fistula segment with the potential for future thrombus formation and extension as was noted in one patient with the fistula closed distally (Figure 2). We, therefore, recommend closure (TCC or SC) of the fistula near the point of origin from the conduit coronary artery, with or without closure of the distal draining point, to eliminate a residual sac or dilated segment as a nidus for thrombus formation.^11,12,15,16

In a distal CAF, closure is typically performed at the drainage site distal to the last nutritive coronary branch. Post-closure, thrombus will likely extend proximally within the conduit coronary artery at least to the nearest runoff normal coronary branch. Ideally, the remaining lumen of the conduit coronary artery should regress in size toward a normal caliber with a smooth contour (optimal remodeling; Figure 2). Only 3 of 11 patients (27%) with distal CAF showed evidence of optimal conduit coronary artery remodeling (Table 3). Acute or subacute thrombus formation and occlusion of normal coronary branches cause symptomatic or asymptomatic coronary events and revascularization with vascular channels and collaterals.^{2,9–12,15,17,18} Two neonates with a large distal CAF in this study were symptomatic with acute coronary events (Figure 5). Asymptomatic coronary thrombosis was observed in 2 patients with complete and partial conduit RCA occlusion and revascularization with vascular channels (Figure 5E through 5H). The precise mechanisms and factors that promote revascularization following coronary thrombosis are unclear. The age of the patient, size of the conduit coronary artery or fistula segment, size and number of runoff normal coronary branches, hypercoagulable state, and degree or absence of anticoagulation may play a role and need further investigation. Lastly, the conduit coronary artery may develop vessel irregularity with areas of stenosis and dilation due to organized partial thrombosis as observed in one patient (Figure 6). Similarly, Hiraishi et al¹⁹ demonstrated perfusion defects on functional studies due to regression of conduit coronary artery segment with intimal stenosis in the dilated aneurysmal segments in 2 patients.

Persistent conduit coronary artery (distal CAF) or fistula segment dilation (proximal CAF) post-closure has been described previously.^11,12,17 These findings may be due to residual flow after attempted therapeutic closure or accessory coronary fistulae in adjacent branches, as noted in 4 patients in this study (Figure 7). Natural spontaneous or intentional therapeutic partial closure (Figure 7H) may actually be beneficial in a large distal CAF, as it reduces hemodynamic burden from fistula runoff. In addition, the residual flow prevents stasis and thrombus formation in the dilated native conduit coronary artery.

Based on the findings of this registry and literature review, we propose the following management and follow-up recommendations for CAF patients diagnosed in infancy (Figure 8). In general, infants presenting with severe HF symptoms and evidence of significant hemodynamic burden should undergo early TCC or SC. Those who are asymptomatic but meet hemodynamic criteria (Qp:Qs≥1.5, chamber dilation, etc) may undergo closure as neonates. However, closure can be deferred until later in childhood (4–6 years) as it may be technically easier and safer at a larger patient size.¹⁵ In patients with medium and large CAF who have significant restriction to flow (natural partial closure at the drainage sites) and are hemodynamically insignificant, medical observation/nonintervention may be the best option in childhood. Small CAFs have been documented to become smaller and regress over time on follow-up and may not require any intervention.^15,20,21

Importantly, this study underscores the significance of follow-up coronary artery anatomic evaluation (initially around 6 months post-closure) in assessing remodeling features. The acuity, degree, and extent of thrombus formation is primarily responsible for the spectrum of suboptimal remodeling illustrated in our experience (Figure 2). Despite limited data available, we noted the use of antiplatelet agents alone, without the use of anticoagulation, in most patients with suboptimal remodeling post-closure. Most proximal CAFs can be closed with minimal postclosure concern for coronary events, with the exception of those with a severely dilated proximal fistula segment. Conversely, moderate-to-large distal CAFs are at higher risk for symptomatic or asymptomatic coronary events. Therefore, we propose aggressive anticoagulation for 3 to 6 months and antiplatelets for 6 to 12 months in this high-risk group. Anticoagulation is continued until follow-up anatomic evaluation is performed, and the type of remodeling will dictate further management as illustrated (Figure 9). The small number of patients and limited or absent long-term data in this cohort preclude definite long-term follow-up antiplatelet and anticoagulation recommendations. Therefore, the proposed long-term antiplatelet and anticoagulation therapies based on fistula type and size are at best our opinion statements only. In those with persistent severe dilation due to residual flow on follow-up, the decision to continue anticoagulation is based on individual anatomy and propensity for thrombosis and coronary events. In patients with a proximal CAF and dilated fistula segment, long-term antiplatelet therapy and subacute bacterial endocarditis prophylaxis may suffice. However, patients with a large distal CAF with persistent severely dilated conduit artery post-device closure require long-term antiplatelet therapy, anticoagulation, and subacute bacterial endocarditis prophylaxis.

Our proposal for follow-up evaluation and further long-term management is illustrated in Figure 9. Patients with suboptimal remodeling, especially symptomatic coronary thrombosis, require acute management with aggressive anticoagulation including consideration for thrombolytics and medical management for myocardial infarction. If we are to better understand the best approach to these patients, even asymptomatic patients require continued follow-up screening with coronary functional (stress imaging) and anatomic (preferably catheterization or computed tomography angiography) evaluation (we propose every 5–10 years) post-closure. In patients with persistent coronary artery dilation due to residual flow, follow-up interval anatomic evaluation should also be performed (we propose every 3–5 years) based on the degree of residual flow, and consideration for intervention should be given if there is any hemodynamic significance or evidence of progressive CAF dilation. All CAF patients treated in childhood need to be transitioned with routine follow-up in adulthood, particularly for those patients with a coronary substrate that is not normal (suboptimal remodeling). It is likely that they will be at additional risk for associated acquired coronary artery disease. Adequate counseling regarding a healthy lifestyle, exercise, diet, treatment of risk factors, such as hyperlipidemia, smoking, hypertension, diabetes, obesity, etc, and coronary symptom–driven management in accordance with existing guidelines in adults are warranted.

This study has limitations. Our patient cohort is small despite multicenter data acquired as part of a unique registry established for this rare coronary lesion. There were incomplete data available, especially regarding anticoagulation and postclosure coronary anatomic (angiographic) evaluation, understating the true incidence of asymptomatic coronary events (suboptimal remodeling). The number of coronary events were too small to identify statistically meaningful risk factors. Nevertheless, this study highlights the importance of follow-up anatomic evaluation for a better understanding of postclosure sequelae. Further long-term studies with a large patient number including anticoagulation and follow-up anatomic data are needed to provide absolute recommendations for acute and long-term management of this fascinating lesion.