Diagnostic accuracy of cardiac computed tomographic angiography and transesophageal echocardiography in evaluation of patients with prosthetic paravalvular leakage

Computed Tomography (CT) scan has been well addressed to provide diagnostic information for patients with prosthetic heart valve (PHV) dysfunction. However, its role in the assessment of patients with prosthetic paravalvular leakage (PVL) has not been studied thoroughly. So, this study was conducted to assess the feasibility, reproducibility, and accuracy of CT for diagnosis of prosthetic PVL using surgical findings as the reference standard.


F I G U R E 1 Diagram for enrolled patients
Cardiac computed tomography (CT) has received attention for its role in the evaluation of prosthetic valves dysfunction. 8 Although cardiac CT had a role diagnostic value in the diagnosis and guidance of successful transcatheter closure of mitral PVL, Diagnostic performance of cardiac CT for detection of prosthetic PVL has not been studied in detail. [9][10][11][12] The purposes of our study were to assess the feasibility, reproducibility, and diagnostic accuracy of CT compared to TEE for the diagnosis of prosthetic PVL using surgical findings as the reference standard.

"Institutional review board approval and informed consent were obtained."
From March 2017 to March 2019, we prospectively enrolled all patients with clinically suspected prosthetic paravalvular dysfunction (e.g., suspected prosthetic heart valve [PHV] obstruction, endocarditis or paravalvular leak referred to the echocardiography laboratory). A total of 80 patients with suspected prosthetic valve dysfunction were enrolled. Twenty-six patients with prosthetic significant PVL were included in the study, whereas fifty-four patients were excluded as mentioned in Figure 1.

2.2
Exclusion criteria 1. Patients who refused to participate in the study and refused to sign consent.
2. The contraindications to multidetector CT angiography include pregnancy and contrast allergy.

Study population characteristics
All patients were subjected to the following work-up on initial presentation.

Complete transthoracic and transesophageal echocardiography
All echocardiographic studies were conducted with a Philips iE33 ultrasound system and X7-2t transesophageal transducer (Philips Medical Systems, Andover, MA, USA). Standard views for combined TTE and TEE were acquired by experienced non-invasive cardiologists. Cardiac chambers quantification was done according to the American Society of Echocardiography for assessment of cardiac chambers including calculation of left ventricle end-diastolic and end-systolic volume and diameter, septal and posterior wall thickness, right-sided chamber volume and dimensions, aortic and left atrium diameter. 13 Assessment of prosthetic cardiac valves was done according to the American Society of Echocardiography for assessment of prosthetic cardiac valves. 14 TEE was done for better assessment of prosthetic cardiac valves specifically looking for: • Paravalvular leak: Location, number, size of defect, severity of regurgitant jets depending on criteria for assessment prosthetic cardiac valves in American Society of Echocardiography. 14 For assessment of PVL across mitral prosthesis, we specifically focused on the presence of a regurgitant jet using continuous-wave Doppler. The severity of PVL was assessed semiquantitatively using the maximal widths of the vena contracta; mild, moderate, and large PVL were defined with widths of 1-2, 3-7, and >7 mm, respectively.
In addition, the presence of occult mitral prosthesis regurgitation was suspected when the following signs were present: flow convergence downstream of the prosthesis during systole, increased mitral peak E-wave velocity (>2 m/s), and increased mean gradient (>5-7 mm Hg) or unexplained or new worsening of pulmonary arterial hypertension For assessment of aortic prosthetic PVL: we specifically focused on the presence of regurgitant jet seen in the long axis view. The severity of PVL was assessed semiquantitatively using the maximal widths of the vena contracta; mild, moderate, and large PVL were defined with widths of 1-2, 3-6, and >6 mm, respectively. Also, we assessed the percentage of regurgitant jet width across the LVOT.

CT acquisition and image reconstruction
All the cardiac computed tomographic angiography studies were done with the Aquilion 64 machine, (Toshiba Medical Systems, Nasu, Japan) which provides 64-sections for optimal imaging. Our 64-section multidetector CT protocol was based on retrospectively ECG gated CT aortography protocols.
Raw data were reconstructed into 10 equally spaced datasets within the R-R interval of the cardiac cycle and were loaded simultaneously into the dedicated cardiac analysis software (vitreous vital image). The imaging planes were aligned parallel with and perpendicular to the valve leaflets as well as in plane with the valve in three perpendicular imaging planes.
For contrast material-enhanced imaging, we planned the acquisition from 2 cm above the carina (including the ascending aorta) to the apex to achieve complete imaging of the heart. When desired, the scanning range was reduced in the craniocaudal direction to reduce radiation exposure.
For anatomic assessment, the best systolic and diastolic reconstruction phases were selected. Images were reconstructed from 0% to 90% of the RR Interval with 10% intervals for subsequent off-line analysis.
The optimal systolic reconstruction windows were at 30% and 35% for assessment of the mitral prosthetic PVL. While the optimal diastolic reconstruction window was at 75% for assessment of aortic prosthetic PVL. For dynamic evaluation, cine images in the plane perpendicular to the valve leaflets were recorded after appropriate alignment and windowing.
○ Paravalvular regurgitation presence and location. We also measured the para-prosthetic anatomical regurgitant orifice area to determine the accuracy and reproducibility of Cardiac CT in assessment of prosthetic paravalvular regurgitation. The size of the PVL was assessed by measuring the perimeter on long-axis views and area on the short-axis view. When CT image quality was suboptimal and not assessable for PVL evaluation, PVL was considered absent for the purpose of data analysis. 16

Statistical analysis
Data were coded and entered using the statistical package SPSS

Patient population
Eighty patients with suspected PHV dysfunction were enrolled.
Twenty-six patients were eventually included in the study as shown in

Population characteristics
Our study populations had a mean age of 41 ± 12 years, with almost three-quarters of patients were males. Heart failure symptoms (dyspnea at rest, orthopnea, and paroxysmal nocturnal dyspnea) were the most common presenting symptoms, followed by symptoms suggestive of infective endocarditis (fever, embolic manifestations).

CT and TEE versus surgery
There was excellent agreement between TEE and CT versus surgery (Gold standard) for diagnosing presence of PVL with agreement k: .95 and 1.0, respectively, as shown in Table 2.
CT had better diagnostic accuracy as compared to TEE in diagnosis of aortic prosthetic PVL with sensitivity and specificity of 100%, 100% for CT and 95%, 100% for TEE, respectively. In our study, TEE incorrectly diagnosed aortic prosthetic PVL in one patient with large aortic root abscess, on the other hand, CT excluded the presence of PVL across the aortic prosthesis (finding confirmed intraoperatively).
Tachycardia and/or acoustic shadowing of mechanical prosthesis may explain the limitation of TEE in assessment of the PVL presence in this patient.

Paravalvular leakage site and location
The localization of PVL on the surgical field was done in all patients who had undergone surgery for PVL.  Table 2.
Both CT and TEE correctly revealed the location of PVL in 24 patients, although, in one patient with aortic prosthetic PVL, TEE incorrectly predicted the location due to the acoustic shadowing caused by the mitral prosthesis in this patient.

Paravalvular leakage severity
The perimeter of prosthetic PVL measured on CT was a mean of .63 ± .15 cm, while the vena contracta as a semiquantitative measure for assessment of the prosthetic PVL measured by TEE was a mean of .62 ± .2. Figure 3 illustrated quantitative methods used in both imaging modalities to assess severity of mitral prosthetic paravalvular leak in patient presented with heart failure symptoms. Table 3 showed a strong correlation between the perimeter of prosthetic PVL measured by CT and echocardiographic severity of PVL by TEE (Spearman's Correlation Coefficient, r = .83, p = .0014).

Other abnormalities
CT had lower sensitivity (75%) for detection of vegetation as compared to TEE (88%), both being compared to intraoperative findings as shown in Table 4. On the other hand, both modalities had 100% specificity for diagnosis presence of vegetation.
Both imaging modalities diagnosed aortic root abscess in seven patients with specificity of 100%.
Occluder malfunction of prosthetic valves were evaluated by both echocardiography (TTE and TEE) and cardiac computed tomographic angiography. The intraoperative assessment of occluder motion was considered the gold standard. The specificity of both modalities for diagnosing occluder malfunction was 100% as shown in Table 4.

DISCUSSION
Our study demonstrated that cardiac CT showed comparable diagnostic accuracy to TEE and intraoperative findings for the detection and assessment of severity of prosthetic PVL. It also demonstrated the role of cardiac CT in diagnosis of peri-annular findings.

Paravalvular leakage presence
CT had better diagnostic accuracy as compared to TEE in diagnosis of aortic prosthetic PVL with sensitivity and specificity were 100%, 100% for CT and 95%, 100% for TEE, respectively. In our study, TEE incorrectly diagnosed aortic prosthetic PVL in one patient with large aortic root abscess, on the other hand, CT excluded the presence of PVL across the aortic prosthesis (finding confirmed intraoperatively).
This patient had undergone surgery for persistent infection due to a large aortic root abscess that was not responding to antibiotic therapy.
The diagnostic accuracy for PVL presence in our study was closely

Paravalvular leakage severity
Considering that the major determinant of surgery is severity of PVL,  19,20 Both CT and TEE diagnosed presence of aortic root abscess in seven patients with sensitivity of 100% and specificity of 100% for both modalities (Intraoperative findings were considered the gold standard). These results were closely correlated to Feuchtner et al. results which revealed that the diagnostic performance of CT for the detection of abscesses/pseudoaneurysms combined was sensitivity 100%, specificity 100% by CT and sensitivity 89%, specificity 100% by TEE. 15

Occluder function
Both imaging modalities had an excellent agreement for diagnosing occluder malfunction with sensitivity and specificity of 100%.

Limitations of the study
The study had a relatively small sample size. We excluded patients with ball and cage prothesis due to heavy beam artefacts interfering with CT image interpretation. 3D echo was not performed for all enrolled patients, however, there were specific limitations for 3D echo as the need for optimal image quality (limited in postoperative patients), several artifacts (dropout and stitches) that may be misinterpreted as valvular perforations and periprosthetic leaks in 3D echo. [24][25][26][27] In our study, we exclusively enrolled patients with mechanical valves because of limited availability of bio-prosthetic valves.

CONCLUSION
This study demonstrates that cardiac CT showed comparable diagnostic accuracy to TEE and intraoperative findings for the detection, localization, and assessment of severity of prosthetic PVL. Moreover, CT was shown to be useful in detection of other findings related to prosthetic cardiac valves.