Two tertiary hospitals (Seoul St. Mary's Hospital and Samsung Medical Center) in Korea participated in this study. CPAF was defined as a communication between one or more coronary arteries and the pulmonary artery on CCTA. We retrospectively reviewed the CCTA reports generated by cardiovascular radiologists from March 2009 to June 2016 (Fig. 1). During this period, a total of 29189 patients underwent CCTA in two institutions (11956 and 17233 each). Among them, CPAF was diagnosed in 188 patients on CCTA (86 and 102 each). We excluded 57 patients who underwent less than 2 years of follow-up (FU). The FU period was defined as the period from the date of diagnosis to the date of the last medical record. Finally, 131 patients were included in the study. Data including age, sex, presence of underlying diseases, presence of coronary artery disease (CAD), awareness of CPAF before CCTA, smoking history, clinical history, prior cardiac testing, and reasons for undergoing CCTA were collected by a thorough review of the patients' electronic medical records. Underlying diseases included diabetes, hypertension, and hyperlipidemia. CAD was defined as a luminal narrowing of ≥ 50% on CCTA or invasive coronary angiographic images. Prior cardiac testing included myocardial perfusion single-photon emission computed tomography (SPECT), treadmill test, stress echocardiography, and invasive coronary angiography.

The CCTA findings of 43 and 17 patients included in the current study have been reported in two previous studies, respectively (824). In one publication, we reported the prevalence and types of CAFs detected with CCTA (8), while the hemodynamic characteristics of CPAF were evaluated using thallium-201 SPECT in the other publication (24). For the current study, we further used these data to evaluate the natural clinical history of CPAF detected on CCTA and proposed potential treatment strategies.

All CCTA scans were performed with dual-source CT scanners at the two institutions. The detailed scanning parameters are presented in Table 1. All images from both institutions were transferred to an external workstation equipped with image processing software (Advantage Workstation 4.3, GE Healthcare, Milwaukee, WI, USA; Syngo Multimodality Workplace, version 2008, Siemens Healthineers, Erlangen, Germany; Aquarius 3D Workstation, TeraRecon, San Mateo, CA, USA).

The following morphologic features of CPAFs were analyzed with consensus by radiologists with cardiac imaging experience of 14 years and 1 year: the vessel of origin, size, and presence of an aneurysmal sac. Fistula size was determined by the largest diameter of the originating vessel and the aneurysmal sac was defined as a dilation that was 1.5 times larger than the adjacent vessels.

The type of treatments administered after the diagnosis of CPAF and the occurrence of major late complications during the FU period were evaluated. Patients were divided into two groups according to treatment methods: surgical ligation and observation with optimal medical treatment (OMT). The occurrence of major late complications was considered as a major adverse cardiac event (MACE), which included death, myocardial infarction, or the need for revascularization or percutaneous intervention. To avoid the interference of confounding factors such as CAD or hypertension, we carefully interpreted the occurrence of MACE in relation to the clinical situation. Furthermore, for patients who had undergone FU CCTA imaging after the diagnosis of CPAF, morphological changes in CPAF on the FU CCTA were assessed.

Continuous variables are presented as medians (25th–75th percentile ranges) and categorical variables are expressed as frequency (percentage). We analyzed the differences in the demographic and morphologic features of CPAF between the two treatment groups by using the Mann–Whitney U test for analysis of continuous data and the Fisher's exact test or chi-squared test for categorical data. The association between surgical treatment and various factors such as age, sex, underlying disease, smoking history, symptoms, and morphologic features of CPAF was estimated using a logistic regression analysis with backward elimination. All statistical analyses were performed with the free statistical software R (version 3.5.3, The R Project for Statistical Computing, Vienna, Austria), and a p value < 0.05 was considered statistically significant.

Table 2 shows the demographic characteristics of the entire study population and a comparison between the two treatment groups. The median age of all patients was 63.0 years (range, 57.0–72.0 years), and the mean FU period was 5.72 years (range, 4.08–6.96 years). There were 65 men with a median age of 62.0 years (56.0–72.0 years) and 66 women with a median age of 65.0 years (58.3–72.0 years). The most common underlying disease was hypertension (n = 60, 45.8%), followed by hyperlipidemia (n = 40, 30.5%) and diabetes mellitus (n = 21, 16.0%). There were 28 patients (21.4%) with CAD. Four patients (3.1%) were already aware of the CPAF before CCTA. The most common prior cardiac testing procedure was invasive coronary angiography (n = 27), followed by the treadmill test (n = 26), myocardial perfusion SPECT (n = 18), and stress echocardiography (n = 7). The most common reasons for performing CCTA were chest pain (n = 57), asymptomatic screening (n = 27), preoperative evaluation (n = 17), palpitation/arrhythmia (n = 14), dyspnea (n = 12), and syncope (n = 5). There was a significant difference in the rate of awareness of the CPAF before CCTA between the surgical ligation and OMT groups (p = 0.030).

In the 131 CPAF cases detected, 76 fistulas (58.0%) originated from both coronary arteries, 48 from the left coronary artery, and only 7 from the right coronary artery (Table 3, Fig. 2). Approximately half the fistulas (n = 54, 41.2%) were less than 2 mm in size, while those in 57 cases were 2 to 10 mm in size. Only 20 fistulas were more than 10 mm in size. In 41 patients (31.3%), the CPAF was associated with aneurysmal dilatation.

Of the 131 patients, 8 (6.1%) underwent surgical fistula ligation. Although the decision regarding the treatment method was made on the basis of the subjective judgements or discretion of the cardiologist and cardiovascular surgeon, our analysis also showed that the size of the fistula was significantly different between the two treatment groups (p = 0.013) (Table 2). The origin site and presence of an aneurysm were not different in the two groups. Logistic regression analysis demonstrated a significant association between the fistula size and surgical treatment (odds ratio = 1.14, p = 0.021), but statistical analysis was limited due to incomplete data (only a small number of patients underwent surgical treatment). Symptoms disappeared after surgery in all eight patients. One patient developed palpitation as a new symptom postoperatively, which resolved after radiofrequency catheter ablation of the left atrium.

The remaining 123 patients were managed with OMT. Drugs such as beta-blockers, calcium channel blockers, anti-hyperlipidemic agents, and antiplatelet agents were used for OMT. Only one patient in the observation with OMT group reported a MACE. This patient was a 71-year-old woman incidentally diagnosed with a small CPAF of less than 2 mm on FU imaging after coronary artery bypass grafting (CABG) surgery; she needed a revascularization intervention after one year of FU. However, none of the patients treated with fistula ligation surgery experienced MACE during the FU period.

Of the 131 patients, 29 (22.1%) underwent FU CCTA imaging one or more times after the CPAF diagnosis. The median interval between diagnosis and imaging FU in these patients was 3.81 years (range, 13 days to 7.18 years). None of the 22 patients in the OMT group showed a definite change in the morphological features of CPAF on FU CCTA imaging. FU imaging examinations were performed in these patients because of other reasons such as regular FU after CABG surgery, Takayasu's arteritis, CAD, and heart valve disease, rather than CPAF.

In the surgical ligation group, seven patients (87.5%) underwent post-operative CCTA imaging, and only one patient demonstrated recurrence of a small CPAF (less than 2 mm in size), but the patient was asymptomatic. No residual fistula or recurrence was found in the remaining six patients in the surgical ligation group.