From January 2011 to June 2016, 20 patients with severe AS were enrolled in this retrospective study. All patients had undergone strict preoperative evaluations using transthoracic echocardiography (TTE), coronary angiography (CAG), computed tomographic angiography (CTA), and lung function tests to assess the cardiac structure and function, systemic vascular conditions, and other comorbidities.

Besides those who refused open surgery, high-risk patients with severe symptomatic AS who were deemed inoperable included those with: (1) severe AS with an aortic valve area of less than 0.8 cm² plus an aortic velocity no less than 4.0 m/s, or mean pressure gradient equal to or greater than 40 mm Hg with New York Heart Association (NYHA) functional class II or higher; (2) logistic EuroSCORE of 15% or higher. The exclusion criteria were: congenital bicuspid aortic valve, acute myocardial infarction, significant coronary artery disease (CAD), left ventricular ejection fraction (LVEF) <20%, aortic annulus area >650 mm² or <300 mm², severe aortic or mitral regurgitation, severe kidney dysfunction, and transient ischemic attack within 6 months [6].

The current study comprised 20 patients (13 male) who had congestive heart failure with NYHA class II–IV and a logistic EuroSCORE higher than 15%. The aortic annulus diameters ranged from 20 to 25 mm, as determined by TTE.

All operations were performed with the patients under general anesthesia; endotracheal intubation and mechanical ventilation were used during the procedure. Edwards SAPIEN XT Transcatheter Heart Valves (23 mm, 26 mm, and 29 mm) provided by Edwards Lifescience Corp. (Irvine, CA, USA) were used in all patients. Monitoring devices utilized for real-time supervision during the TAVR procedure included transesophageal echocardiography (TEE), three-channel electrocardiogram (ECG), continuous oxygen saturation, and hemodynamic assessment via radial artery catheter. A standard transfemoral retrograde approach was conducted as described by Kasel [9]. All procedures were performed via the transfemoral approach except for one patient with an aortic valve area of 0.22 cm² in whom the procedure was assisted by the transapical approach.

All data are summarized and displayed as mean±SD for continuous variables and numbers (%) for categorical variables. Valve function and heart function before and after TAVR were analyzed and compared by the Wilcoxon signed-rank test or the paired-samples t test (Table 1). All two-tailed test results with a value of p < 0.05 were considered statistically significant.

The etiology of coronary occlusion is often multifactorial and generally includes asymmetric expansion caused by bicuspid valve malformation, bulky calcification on the left or right coronary cusps, low take-off of the coronary ostia, large aortic valve leaflets, and shallow aortic sinus.

Aortic sinus dissection as a rare complication caused by TAVR has not been reported in the current literature. However, it can result in myocardial infarction and was considered to be the cause of death of the first patient in our study. The preoperative CTA of the first patient showed an annulus area of 358 mm² (Fig. 1a); a 23-mm artificial valve with a 17-ml balloon of 1‑ml underfill was chosen, which provided 10.0% oversizing. The take-off of the left coronary artery was 15.4 mm (lower than the right one), and was considered to be a safe distance. After releasing the valve stent, the left coronary ostium was seen to be occluded under digital subtraction angiography (DSA). After percutaneous transluminal coronary angioplasty (PTCA) and stent implantation, the blood flow of the left coronary artery recovered immediately (Fig. 1b). The patient was reported to have acute coronary syndrome (ACS) on the first postoperative day and died afterwards. The take-off of the left coronary ostia was considered safe enough and the CAG showed that the left coronary artery kept patency after stent implantation. As a result, we considered that coronary sinus dissection occurred and led to the acute occlusion of the coronary ostia. The risk factors remained unclear, but could be related to annulus calcification and the balloon-expansion procedure.

Researchers have shown that [15] a distance of <10 mm between the base of the coronary arteries and the hinge point of the corresponding cusp pose an increased risk for occlusion. Among the 20 patients, the average height of the coronary ostia was 13.64 mm (range, 12.27–15.42 mm) in the left side and 14.87 mm (range, 13.05–16.70 mm) in the right side using the three-dimensional structure software Aquarius Workstation (Version 3.7.0.13, TeraRecon, Foster City, CA, USA). In the second patient, the left coronary ostial height was 15.1 mm (lower than the right) and was considered to be a safe distance (Fig. 2a). Angiography showed that the patency of the cornary arteries were good before stent implantation (Fig. 2b). In other investigations [15], it was pointed out that if the coronary ostial height is shorter than the coronary cusp length, it could also increase the risk of coronary occlusion, which was thought to be the cause of death in the second patient. After releasing the valve stent, the left coronary ostium was covered by a self-leaflet (Fig. 2c). It indicated a shorter left main ostial height (15.1 mm) than the left coronary cusp leaflet length (16.1 mm), which was the main reason for coronary occlusion (Fig. 2a). To avoid such life-threatening situations, surgeons should analyze the preoperative imaging carefully, especially the coronary ostial height, the bulky calcification on the left or right coronary cusps, and the shape of the coronary sinus. For patients with CAD, CAG should be performed before TAVR to decide whether to implant coronary stents or not.

After stent implantation, the coronary artery blood supply and the artificial valve stent function should be examined. If there is coronary occlusion, PTCA should be performed immediately. Furthermore, a hybrid operation room should be readily available in the case of coronary artery bypass grafting (CABG).

Larger registries [16] reported that periprocedural aortic annular rupture was less than 1% of the total procedures and was more common with balloon-expandable valves. Aortic annulus rupture usually occurs because of poor patient selection, lack of experience, measurement error, aggressive dilatation, valve oversizing (>20% was shown to be associated with an 8.4-fold higher risk of contained and uncontained ruptures with the balloon-expandable system [17]), bulky calcifications in the valve deployment zone (Fig. 3; [17, 18]), severe asymmetric subvalvular hypertrophy, and global left ventricular hypertrophy in the elderly, especially in female patients with relatively weak myocardium.

The third patient who died in our series was a 72-year-old woman with NYHA class IV. Asymmetric hypertrophy of interventricular septum and decreased left ventricular compliance were detected with echocardiography. An aortic annulus area of 500 mm² with no obvious calcification was found by CTA (Fig. 4). A 26-mm valve with a 22-mm balloon was needed (6.2% oversizing). During the balloon dilatation procedure, there were temporary system errors in the pacemaker, which was observed under DSA and resulted in a slight sliding of the balloon. After balloon displacement, cardiac tamponade occurred. Pericardiocentesis was carried out and the symptoms were gradually controlled. After entering the ICU, however, the patient’s heart rate and blood pressure dropped again within 4 h and she died. Advanced age, female gender, as well as reduced cardiac structure and function were thought to be critical risk factors. The balloon displacement might be the main reason for the aortic annulus rupture.

Extensive calcification of the valve deployment zone is associated with lower device success and increases the risk of rupture during the process of balloon expansion [19]. The fourth patient in our series had an aortic annulus area of 568 mm² and extensive calcification was found in the valve deployment zone via CTA. A 29-mm valve and a 23-mm balloon with 1‑ml underfill were selected, which had an oversizing of 13.3%. After balloon expansion, hydropericardium was seen under TEE observation. Although pericardiocentesis and exploratory thoracotomy were performed immediately, it still failed to save the patient’s life; the patient died of cardiac tamponade caused by aortic annulus rupture. Whether patients with extensive calcification of the valve deployment zone are suitable for a balloon-expandable valve system and how to select the rate of oversizing are issues that still need further exploration.

Selecting suitable patients and the correct size of the valve system via preoperative imaging assessment is important to reduce the rates of rupture. The rate of oversizing is recommended to be 5–20% when the Sapien XT system is used [17]. In cases of extreme calcification, an average rate of 7.5% is recommended [20] to balance the oversizing that may result in rupture and the undersizing that may result in increased paravalvular regurgitation. Balloon underfilling during valve deployment may also lower the risk of rupture. In addition, it is equally important to ensure the effectiveness and stability of the pacemaker before and during the procedure.

Acute hemodynamic collapse immediately follows the rupture, which can be observed with TEE or DSA. In order to maintain hemodynamic stability, a hybrid operating room is usually required to perform cardiopulmonary bypass and sternotomy with conventional cardiac procedures. Alternatives include isolated pericardial drainage and a conservative strategy if the rupture is contained and the patient is not in an immediate life-threatening situation. Recently, researchers described endovascular repair with the use of tissue glue injection via a microcatheter to seal an annular perforation defect [20]; this requires further validation and long-term follow-up before entering the guidelines.

Patient selection and assessment were highly subjective. Family members held different attitudes toward exploratory thoracotomy and autopsy, and some of them rejected the study proposal. Thus, the causes of death in some patients could only be judged by the clinical experiences. The study was a single-center, non-randomly controlled study and had a limited number of patients. It was not adequately powered for clinical or procedural endpoints with the 30-day follow-up period, and therefore further postoperative studies are required.