Aorto-cardiac fistula etiology, presentation, and management: A systematic review

Abstract

Background:

Aorto-cardiac fistulae are a rare but increasingly reported entity, and data are scarce.

Method:

The authors performed a systematic review of ACFs to characterize the underlying etiology, clinical presentation, and compare outcomes of treatment strategies.

Results:

3,733 publications were identified in the search. Of those, 292 studies including 300 patients were included. Etiology of ACFs was 38% iatrogenic, 25% infectious, 14% traumatic, and 15% due to other causes. Most patients (74%) presented with heart failure. Common locations were aortic-right atrium (37%), and aortic-pulmonary artery (25%). The majority of patients (71%) were treated surgically, while 13% were treated percutaneously, and 16% were treated conservatively. Patients who were managed conservatively had a higher mortality than those treated with invasive closure (53% vs. 12% vs. 3%, p = <0.00001).

Conclusions:

This systematic review sheds light on this highly morbid condition. Once recognized, fistula closure appears to be superior to conservative management.

Introduction

Aorto-cardiac fistulae also called aorto-cavitary fistulae (ACFs) are rare abnormal connections that form between the aorta and chambers of the heart.^1–3 The incidence and prevalence of this entity is unknown, as ACFs have not been studied extensively, and they are often found on post-mortem examinations.^1,4 ACFs can be congenital or acquired. The main focus of this study was on acquired fistulae. Acquired etiologies can injure the aortic wall and adjacent cardiac chambers, these can occur traumatically or atraumatically.¹ Traumatic causes include direct blunt or penetrating trauma to the chest, deceleration trauma, and aortic dissection. Iatrogenic trauma can result from cardiac or aortic surgery, trans-septal puncture, left heart catheterization, or the presence of a catheter or foreign body in the right atrium.^1,3–7 Non-traumatic causes of ACF include infective endocarditis, inflammatory disorders such as Behçet’s disease, aortic dissection, or rupture of a sinus of Valsalva aneurysm.^1,3–7

Clinically, ACF patients can range from asymptomatic to symptomatic with signs and symptoms of heart failure and cardiogenic shock.¹ The presentation varies depending on the size of the fistula, and resultant pressure and shunt flow.² This flow of blood across the shunt produces a murmur typically heard as a continuous murmur on auscultation.^1,8,9 Patients can develop symptoms such as fever, pedal edema, dyspnea, fatigue, and chest pain.¹ Ultimately, without diagnosis, appropriate closure of the ACF, and treatment of the underlying cause, it can often lead to death.^1,10

The diagnosis of ACF requires high clinical suspicion, and the use of imaging modalities such as echocardiography, which allows the best visualization.¹ Transesophageal echocardiography (TEE) has shown to provide better visualization than transthoracic echocardiography (TTE), as it has a higher sensitivity and specificity for diagnosing ACFs.^{1–3,8,10–13} Additional diagnostic modalities such as computed tomography (CT) can be useful to characterise the underlying cause, such as in the case of aortic dissection.¹

There is no consensus on the management of ACFs and management strategies vary. Strategies may include medical management of symptoms or infection, or attempts at closure either surgically or percutaneously. The first open surgical repair and closure of an ACF was reported by Temple et al. in 1966.¹⁴ The first transcatheter closure of ACFs was reported 20 years later by Hayward et al. in 1988, where he described using a detachable balloon device.¹⁵ Recent successful attempts at percutaneous fistula closure have also been made utilising occluder devices or coil embolization.^4,16

Data on the etiology, presentation, and outcomes of various management strategies for acquired ACFs are scarce; however, if left untreated the mortality has been shown to be high.^1–3,5,8,13

Recent reviews have focussed on select forms of ACFs, in terms of fistula anatomy or etiology.^1,3,17,18 However, our systematic review sought to encompass this entity in all reported etiologies, and discuss the presentation, and common findings seen among patients. The gathered data provides insight on treatment options, and the outcomes are quantified using 300 patient cases.

The purpose of this systematic review is to: Assess the trends in the reported cases of aorto-cardiac fistula, and to compare the outcomes of various management strategies. We hypothesized that there was (1) a growing temporal trend in the use of transcatheter intervention in the management of aorto-cardiac fistula (2) improved survival with transcatheter management of aorto-cardiac fistula compared with conservative or surgical management.

Overall this systematic review sheds new light on this highly morbid condition, and enables readers to be cognisant of it as a possible diagnosis, particularly in patients who may have undergone invasive procedures. The presented information will provide insight to healthcare professionals, and researchers when considering ACF as a diagnosis and area of research.

Methods

A systematic review of ACFs protocol was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting guidelines.¹⁹ We conducted a literature search of the PubMed database (March 19, 2018). Eligible studies were delineated using the Medical Subject Headings search terms and text word search. Three thousand seven hundred and thirty-three articles published on ACFs between January 1st, 1966 and January 1st, 2018 were identified.

We chose to begin our search in 1966 as the first open surgical repair and closure of the fistula was reported at this time.¹⁴ We used the MeSH search terms: “aorto” AND “heart” “atria” OR (“heart” AND “atria”) OR “heart atria” OR “atrial”) AND (“fistula” OR “fistulae”).

All articles were initially included regardless of quality or language of publication. Those that were not in English were translated using Microsoft Bing translator. We found that once translated using Bing translator, the case reports were easy to understand and reliable, particularly when considering the translation of quantitative data. After removing duplicates (1819/3733) and screening the title and abstract, 670 full text records were assessed for eligibility. The inclusion criteria consisted of any study conducted in any country, including case reports, and case series on aorto-cardiac fistulae. We included studies with patients of any gender older than 12 months. Studies were excluded if they were animal studies, congenital causes of ACF, or conducted on patients that were less than 12 months of age. Commentaries, review articles, or those that did not have quantitative data were also excluded. Overall 292 records were included for synthesis (Fig. 1). Of those studies, 287 were case reports, and 5 were case series. We did not identify clinical studies or randomized trials. Since articles were mainly case reports, the quality of the records was not formally assessed.

Fig. 1.

PRISMA search protocol.

Articles were reviewed by two authors, and disagreements were resolved through consensus and arbitration by the senior authors. The data was manually exported into a Microsoft Excel 2013 spreadsheet (Microsoft Corporation, Redmond, Washington). The following characteristics were extracted: publication year and location, number of patients, patient demographics, ACF etiology, clinical presentation, initial diagnosis, pulmonary blood flow to systemic blood flow ratio (Qp/Qs), anatomy, largest dimension, number of fistulae, and the reported outcomes (Table 1). Studies were categorized by treatment strategy: surgical, percutaneous, or conservative management. Conservative measures included any non-invasive management, including observation or medical management of the underlying condition and comorbidities.

Table 1.

Baseline characteristics of the study cohort

Characteristics	Overall	Percutaneous	Surgical	Conservative	P-value
Total Number of Articles	292	39	204	49
Number of patients	300 (100)	39 (13)	212 (71)	49 (16)
Age (mean ± SD)	52±13.4	59.4 ± 4.9	49±11.3	58.8 ± 26.8	0.00
Male	208 (69)	27 (69)	149 (70)	32 (65)	0.793
Etiology:
Iatrogenic	113 (38)	31 (79)	64 (30)	18 (37)	<0.00001
Infectious	76 (25)	2 (5)	53 (25)	21 (43)	0.0003
Trauma	42 (14)	1 (3)	40 (19)	1 (2)	0.0008
Ruptured SOV	23 (8)	2 (5)	19 (9)	2 (4)	0.417
Other	46 (15)	3 (8)	36 (17)	7 (14)	0.326
Presentation:
Incidental Finding	28 (9)	6 (15)	16 (7)	6 (12)	0.226
Heart Failure	221 (74)	30 (77)	155 (73)	35 (72)	0.838
Chest Pain	32 (11)	0 (0)	27 (13)	5 (10)	0.030
Other Symptoms	19 (6)	3 (8)	14 (7)	3 (6)	0.956
Hemodynamic Instability	78 (26)	5 (13)	56 (26)	17 (35)	0.065
Initial Diagnosis:
Murmur	186 (62)	15 (38)	144 (68)	27 (55)	0.0013
Imaging	114 (38)	24 (62)	68 (32)	22 (45)	0.0013
QP/QS (mean ± SD)	2.06 ± 0.96	1.65 ± 0.32	2.28 ± 1.16	1.6 ± 0.54	0.00
Number of ACF:
1	278 (93)	35 (90)	196 (92)	47 (96)	0.531
2	18 (6)	3 (8)	13 (6)	2 (4)	0.769
>2	4 (1)	1 (2)	3 (2)	0 (0)	0.521
Total Number of ACF	326	44	231	51
Anatomy of the ACF:
AO-RA	119 (37)	14 (32)	91 (39)	14 (27)	0.217
AO-LA	60 (18)	5 (11)	43 (19)	12 (24)	0.309
AO-RV	58 (18)	10 (23)	36 (15)	12 (24)	0.266
AO-LV	6 (2)	1 (2)	4 (2)	1 (1)	0.968
AO-PA	83 (25)	14 (32)	57 (25)	12 (24)	0.573
Largest ACF Dimension	30 mm	18 mm	30 mm	10 mm
Reported Outcomes:
In-Hospital Data	286 (95)	39 (100)	204 (96)	43 (88)	0.020
In-Hospital Data Only	158 (53)	10 (26)	122 (57)	26 (53)	0.001
Follow up Data	128 (43)	29 (74)	82 (39)	17 (35)	0.00008
Length of follow up (months) (median ±IQR)	8 ± 9	5.5 ± 9	12 ± 9	10.5 ± 23	0.012
Publication Year:
<2000	73 (25)	1 (2)	65 (32)	7 (14)	0.00009
2000–2010	104 (36)	10 (26)	69 (34)	25 (51)	0.029
2011–2017	115 (39)	28 (72)	70 (34)	17 (35)	0.00005
Publication Location:
US	96 (33)	14 (36)	64 (31)	18 (37)	0.704
Non-US	196 (67)	25 (64)	140 (69)	31 (63)

Values are n (%) or mean ± SD.

SOV, sinus of Valsalva; ACF, aorto-cardiac fistulae; AO-RA; aorto-right atrium; AO-LA, aorto-left atrium; AO-RV, aorto-right ventricle; AO-LV, aorto-left ventricle; AO-PA, aorto-pulmonary artery; IQR, interquartile range; US, United States.

Data synthesis and analysis

IBM-SPSS statistics package 20 (IBM Corporation, Armonk, New York) software, Chi-Square Test Calculator (Social Science Statistics, 2018),²⁰ and Freeman-Halton extension of Fisher’s exact test (Statistics Calculators, 2018)²¹ were used for statistical analysis. Normally distributed continuous variables were reported as a mean. The standard deviation, and the median and interquartile range (IQR) were determined for continuous variables that were not normally distributed. Categorical variables were presented as frequencies and percentages, and chi-square or Fisher exact test and p-values were used to evaluate correlations; p values were considered statistically significant at <0.05.

Results

A total of 3733 publications were identified and screened from the database search. We retrieved 670 full-text records to evaluate, of which 292 satisfied the inclusion criteria for synthesis. Of those 292 studies, which were mainly case reports, 300 patients were presented (Fig. 1). The mean age was 52 ± 13.4, and 69% of patients were male. The majority of patients (71%) were treated surgically, while 13% were treated percutaneously, and 16% were treated with conservative measures.

The majority of articles in our systematic review were published between 2011 and 2017 (39%), 36% were published between the years 2000 and 2010 (104/292), and 25% of the articles were published in the year 2000 or earlier (73/292). Notably, 72% of the articles on percutaneous management of ACFs were published between 2011 and 2017 (28/39). Many of the studies were completed outside of the United States as opposed to in the United States (67% vs 33%). Six out of seven continents were represented. Studies were completed in Europe (41%), North America (35%), Asia (19.5%), South America (2%), Africa (1.4%), and Australia (1.4%).

The most common etiology of ACFs was iatrogenic in 38%, infectious in 25%, traumatic in 14%, ruptured sinus of Valsalva in 8%, and other in 15%. The number of ACFs was 1 in most patients (93%), but multiple in 7%. Thus, the total number of ACFs in 300 patients was 326. Most common locations were from the aorta to the right atrium (37%), from the aorta to the pulmonary artery (25%), from the aorta to the left atrium (18%), from the aorta to the right ventricle (18%), and from the aorta to the left ventricle (2%) (Fig. 2). The largest diameter ACF was managed surgically, as opposed to percutaneously, or conservatively (30 mm vs 18 mm vs 10 mm).

Fig. 2.

Anatomy of the aorto-cardiac fistulae. RA, right atrium; LA, left atrium; AO, aorta; RV, right ventricle; LV, left ventricle; PA, pulmonary artery; AO-RA, aorto-right atrium fistula; AO-LA, aorto-left atrium fistula; AO-RV, aorto-right ventricle fistula; AO-LV, aorto-left ventricle fistula; AO-PA, aorto-pulmonary artery fistula.

Many of the patients with an ACF presented with heart failure (74%). Patients also presented with hemodynamic instability (26%), chest pain (11%), other symptoms (6%), or as an incidental finding (9%). Their initial diagnosis was often investigated due to the presence of a murmur on examination (62%); however, in 62% of patients who were treated percutaneously, the ACF was initially diagnosed via imaging (24/39). The average Qp/Qs ratio was 2.06±0.96.

In-hospital outcome data was not present in 5% of studies. Of those studies with reported outcomes, the majority of cases that were invasively managed were successfully closed (97% percutaneous and 91% surgical). Patients who were managed conservatively had a higher mortality than those treated with surgery or percutaneous closure (53% vs. 12% vs. 3%, p = <0.00001). Thus, the in-hospital survival for percutaneous, surgical, and conservatively managed patients was 97%, 88%, and 46% respectively (p = <0.00001). Patients were followed for an overall median length of 8 months ± 9 IQR, and the majority of patients who survived the hospitalization were alive at short-term follow-up (100% percutaneous, 98% surgical, and 65% conservative, p = <0.00009). The patients who were managed invasively showed greater symptomatic improvement at follow-up than patients who were conservatively managed (100% vs. 95% vs 65% respectively p = 0.0004) (Table 2).

Table 2.

Outcomes stratified by management strategy

Clinical Characteristics	Percutaneous	Surgical	Conservative	P-value
Successful ACF Closure	38 (97)	193 (91)	NA	0.175
Reported Outcomes:
In-Hospital Data (mortality) Only	1/39 (3)	24/204 (12)	23/43 (53)	<0.00001
Follow up Data (mortality)	0/29 (0)	2/82 (2)	6/17 (35)	0.00009
Length of follow up (months) (median ±IQR)	5.5 ± 9	12±9	10.5 ± 23	0.012
In-Hospital Survival	38/39 (97)	180/204 (88)	20/43 (46)	<0.00001
Survival at Maximum follow up	29/29 (100)	80/82 (98)	11/17 (65)	0.00009
Symptomatic Improvement at follow up	29/29 (100)	78/82 (95)	11/17 (65)	0.0004

Values are n (%) or mean ± SD.

ACF, aorto-cardiac fistula; IQR, interquartile range.

Discussion

The main findings of this systematic review are that: 1) ACFs were commonly iatrogenic or infectious in greater than 64% of the cases; 2) ACF patients presented with heart failure, and/or hemodynamic instability; 3) Common locations are aortic-right atrium, and aortic-pulmonary artery in greater than 61% of the patients in this study; 4) Fistula closure with transcatheter or surgical means appears to be superior to conservative management.

Until recently little data existed on the etiology, presentation, and outcomes for acquired aorto-cardiac fistulae (ACFs). This analysis sought to clarify the characteristics of this increasingly reported entity and its management. Our analysis systematically reviewed data from 292 studies. Among the 300 patients in our review the majority presented with an iatrogenic ACF (38%). The etiology of ACFs can be due to primary causes such as congenital malformations, or secondary causes such as paravalvular abscesses, ruptured sinus of Valsalva aneurysms, aortic dissections, trauma, endocarditis, or iatrogenic causes such as occurring after intravascular procedures or valve replacements.^1,3–7 Our study focussed on secondary ACFs. Infectious ACFs occurred in 25% of patients, and was the most common etiology in patients undergoing conservative management (Table 1). As expected, less infectious ACF cases were managed invasively. Five percent of infectious ACFs patients were managed percutaneously and 25% surgically. Due to the increased risk for device related infection, infectious ACF patients would be less likely to be treated percutaneously. Traumatic ACFs (14%), ruptured sinus of Valsalva (8%), or other causes (15%) occurred less frequently in all management groups (Fig. 3).

Fig. 3.

Etiology of the aorto-cardiac fistulae. SOV, sinus of Valsalva.

ACFs can be categorized by the fistula location from the aorta to the cardiac chamber. Fistula formation from the aorta to the right atrium has been commonly reported.^4,8 Studies have also suggested that fistulae form with equal prevalence between the aorta and any of the cardiac chambers.^10,22 Our review found that ACFs developed most commonly from the aorta to the right atrium (37%), and secondarily to the pulmonary artery in 25% of cases. Fistulae from the aorta to the left atrium, right ventricle, or left ventricle occurred less frequently (18%, 18%, and 2% respectively).

Patients that present with ACFs can be asymptomatic and thus be identified incidentally. Or ACF patients often present with mild or severe symptoms depending on the size of the fistula, and the resulting pressure and shunt flow.² The gradient of shunt flow can be determined by the Qp/Qs ratio, which is usually determined via oximetry during cardiac catheterization.²³ The mean Qp/Qs ratio that was found was 2.06 ± 0.96, indicating a left-to-right shunt. Although the Qp/Qs ratio has been established in the literature to assist in the decision for shunt closure, in defects such as atrial septal defects (ASD), there has been no consensus with regards to ACFs management due to its rarity.^24–26 Often ACF patients present with volume overload, and signs of heart failure due to the shunting of blood from the aorta to the right or left atrium, and subsequently the right or left ventricle.^1,3,4,9 Large shunts have been seen to lead to an increased severity of heart failure, and greater volume overload.¹⁰ Thus the Qp/Qs ratio may have value in predicting patient outcomes; patients with a large Qp/Qs ratio are likely to experience greater volume overload and severe heart failure. It may prove to be a valuable quantitative entity to assist in ACF treatment decision making, similar to its usage in ASD treatment decision making.^24–26

We found that the majority of patients in this review, 74% presented with heart failure symptoms or hemodynamic instability (26%), as compared to symptoms of chest pain (11%), or other symptoms (6%). Only 9% presented with ACFs as an incidental finding (Fig. 4). The diagnosis of ACFs via imaging occurred less frequently (38%) (Table 1). A common sign of ACFs at presentation is a continuous murmur on auscultation, due to the gradient of flow from the aorta to the cardiac chamber.^1,8,9 We found that a murmur was present in many of the cases at initial diagnosis (62%).

Fig. 4.

Presentation of ACF patients within each management strategy.

The majority of the patients who were managed percutaneously were treated between 2011 and 2017 (72%). The number of ACF cases that were treated percutaneously increased 10 fold during the years 2001 to 2010 (Fig. 5). The number of cases of ACF that were managed percutaneously continued to rise and nearly triple in the following 7 years; this met our hypothesis. During this time the trend for surgically managed cases of ACFs also increased, however from 2001 to 2010, and 2011 to 2018, the surgical trend began to plateau (Fig. 5). Thus, percutaneous management of ACFs is a relatively recent and developing management approach, compared to surgical management, and it is a common option to manage cases of ACFs during the 21st century. The recent development of this discipline may also explain the outcome differences between management groups. Percutaneous management outcomes from earlier studies may have been affected by a lack of technical prose, and/or less device options available for transcatheter ACFs closure, compared to recent years. The first percutaneous case of ACF closure was described using a detachable balloon device.¹⁵ Recently other devices such as an occluder, or coil embolization have also been described.^4,16 The number of reported cases of ACFs doubled during the period of 1991 to 2000, and 2001 to 2010, this likely correlates to an increased awareness and diagnosis of the condition (Fig. 5). During this time-frame the use of transesophageal echocardiography also increased, which has a higher sensitivity and specificity than transthoracic echocardiography for the diagnosis of ACFs.^{1–3,8,10–13}

Fig. 5.

Number of ACF cases reported within each management strategy.

There is a consensus that ACFs require closure, particularly if they are associated with symptoms, as they result in a high level of morbidity and mortality.^1–3,5,8,13 Our results were somewhat in agreement to this statement as we found that a significant mortality occurred in patients managed conservatively, both in-hospital (53%), and at follow-up (35%). However, definitive conclusions cannot be deduced as the data provided all-cause mortality and death could not be attributed to the ACFs or a cardiovascular cause. Morbidity and mortality risk calculations prior to interventions were not delineated between studies in the review. Consideration must be given to the possibility that patients were highly morbid and as a result, high risk in the conservatively managed group, and possibly unsuitable for invasive closure. Patients who were conservatively managed had worse symptomatic improvement at follow-up than invasively managed patients. When comparing percutaneous and surgical ACFs closure, both procedures were successful in closure, and percutaneously closed ACF patients had a slightly higher in-hospital survival than patients who underwent surgical closure (97% vs 88%). However, the two invasive management strategies demonstrated comparably low mortality at maximum follow-up. These results met our hypothesis. Surgical and percutaneous management of ACFs appeared as superior to conservative management of ACF patients.

Study limitations

This study represents, to our knowledge, the largest systematic review of ACFs. Although a comprehensive search was conducted there is an inherent selection bias that occurs in systematic reviews. Many of the articles that satisfied our inclusion criteria on ACFs between the dates of our study were retrospective and thus they too carry a bias. Some studies were missing in-hospital or follow-up data on patients, and thus we cannot conclusively interpret data regarding patient survival at maximum follow-up. Patient populations also differed between each management group, in terms of ACFs etiology, and other co-morbidities. Both the surgical and conservative groups treated more patients with infectious ACFs than the percutaneously managed patients. It cannot be underestimated that the etiology of ACFs had a role in the severity of disease at presentation, and may have contributed to a higher patient mortality. Also, there were multiple possible confounding variables, which may have led to increased mortality among patients in different studies. Differences in survival between the two invasive management groups cannot be conclusively compared as there was no uniform stratification of procedural risk among studies, and between surgical and percutaneous interventions.

Conclusion

Aorto-cardiac fistula is a rare but increasingly reported entity. This systematic review sheds new light on this highly morbid condition; describing the etiology, presentation, and outcomes of various management strategies, which enables readers to be cognisant of it as a differential diagnosis. When considering ACFs, it is beneficial to remember that in greater than 64% of the cases in this study, ACFs were found to be caused by iatrogenic or infectious processes. Patients will commonly present with heart failure, and/or hemodynamic instability. Thus ACF could be a possible differential diagnosis in patients with infective endocarditis or prior invasive cardiac intervention, presenting with new symptoms of heart failure, and/or hemodynamic instability. When investigating ACFs, they are best seen on TEE which has a high sensitivity and specificity. Common locations where ACFs may be found are from the aorta to the right atrium, and from the aorta to the pulmonary artery. These locations were seen in more than 61% of the patients in this review. Once recognized, fistula closure with transcatheter or surgical means appears to be superior to conservative management.

Abbreviations:

ACF

aorto-cardiac fistula

ACFs

aorto-cardiac fistulae