Imaging in Transcatheter Aortic Valve Replacement (TAVR): role of the radiologist

The aortic valve complex consists of the aortic valve annulus, commissures, sinuses of Valsalva (SOV), coronary ostia (CO) and sinotubular junction (STJ). Three aortic valve cusps are connected proximally to the wall of the left ventricular outflow tract (LVOT) [10] by three anchor points at the nadir (hinge point) of each aortic cusp. Those nadir points, when virtually connected, form an oval-shaped, three-pronged coronet—the “virtual basal ring” (Fig. 1). The orientation of this virtual basal ring, or aortic annulus, is double oblique and does not correspond to conventional axial, coronal or sagittal planes of MDCT. This ring is the major target for transcatheter aortic valve prosthesis sizing.

Distally, the three semilunar leaflets are attached to the wall of the aortic sinus at the surgical annulus that corresponds to the sinotubular junction: the junction between the SOV and ascending aorta. Multiple manipulations of the raw imaging data are required to create an image that would exactly correspond to the aortic annulus (virtual basal ring). This virtual basal ring is often not orthogonal to the LVOT, and insertion of the right coronary cusp can often be inferior to the left and non-coronary cusps [11].

There are currently two transcatheter heart valves (THVs) available for use in the US: the Edwards SAPIEN valve (Fig. 2) and the Medtronic CoreValve prosthesis (CoreValve) (Fig. 3). Both devices were granted Conformité Européenne (CE) approval in Europe in 2007, and the Edwards SAPIEN valve was approved for commercial use in the US in 2011 following the landmark PARTNER trial [12, 13]. The CoreValve device is currently available in research trials only. The main characteristics of both devices are summarised in Tables 1 and 2, with some specific features outlined below. Both devices are predominately delivered through a transfemoral approach involving either percutaneous or surgical cutdown for arterial access (Fig. 4).

The Edwards SAPIEN is a ballon-expandable stent. The earlier generation valve is available in 23- and 26-mm valve sizes and is delivered via a 22- or 24-French delivery system, respectively. A newer version, the SAPIEN XT, is available in 23-, 26- and 29-mm sizes using a smaller transfemoral delivery sheath system (18 and 19 French, which requires a minimal iliofemoral luminal diameter of 6 mm). Due to its short stent frame length, the Edwards SAPIEN valve can be deployed in the aortic annulus without any significant limitations in terms of sinus of Valsalva size, sinotubular junction or ascending aortic diameter. The balloon-expandable frame assumes its shape with a single inflation, and the aortic annulus typically conforms to the shape of the device with a more circular geometry (Fig. 5).

Available in 23-, 26-, 29- and 31-mm sizes, the Medtronic CoreValve device (Table 1) is a self-expanding prosthetic valve requiring an 18-French delivery catheter (a minimal iliofemoral luminal diameter of 6 mm). The longer CoreValve device frame requires sufficient ‘outflow’ room to be deployed but the device is available in a wider range of sizes allowing patients with particularly larger annular diameters to be treated. The radial force of the self-expanding frame holds the device in position at the level of the annulus. This same radial force is thought to increase the need for a permanent pacemaker because of compression on conduction tissues, an issue more prevalent with the CoreValve device.

Depending on the specific device, there are several potential additional vascular access routes for implantation [14]:

TAVR is performed with cardiac surgery back-up and typically general anaesthesia is used. Imaging during the procedure is crucial for accurate device positioning, assessing prosthetic valve function and efficient detection of procedural complications. Positioning of the prosthetic valve is aided by pre-determination of the annular plane orientation using pre-procedure MDCT and rotational angiography with fluoroscopy during the procedure (Fig. 8). Transesophageal echocardiography (TEE) plays an important role in device positioning, final placement and determiningTransoesophageal degree of aortic regurgitation following deployment (Fig. 9). Main procedure risks and complications include vascular access complications, aortic valve regurgitation, transvalvular and paravalvular, stroke, conduction system abnormalities, prosthesis embolisation, mitral valve disruption, haemorrhage, peripheral vascular complications and death [19, 20].

Currently TAVR is reserved for patients with symptomatic severe aortic stenosis in whom pre-existing comorbidities increase the risk of sternotomy and standard valve replacement, and who are considered high or very high surgical risk (Table 3). Aortic annular size appropriate for any of the existing THV devices is an important inclusion criterion. Multiple objective estimates of surgical mortality are applied in the selection process of patients referred for TAVR [21‐23]. A logistic EuroSCORE (European System of Cardiac Operative Risk Evaluation) of 20 points or a Society of Thoracic Surgeons Score of 10 points is often used as a risk threshold to aid in decision making [23]. As of 2013, there had been more than 40,000 transcatheter valvular replacements performed worldwide and approximately 7,000 cases performed in the US.

Annular size not corresponding to any of the available THV devices would be a major exclusion criterion, followed by the lack of an appropriate vascular access route; for the CoreValve device, this criterion would be sinus of Valsalva dimensions too small or too large to accommodate the upper frame when positioned (Table 3).

Two- and three-dimensional transthoracic (TTE), transesophageal echocardiography (TEE), ECG-gated MDCT and rarely cardiac magnetic imaging (CMR) (Figs. 10 and 11) can be used for aortic valve apparatus assessment, with each imaging technique providing complementary information for optimal valve prosthesis sizing and THV selection. TTE, TEE and MDCT are currently the primary modalities for aortic root anatomy assessment.

For precise aortic valve annular sizing and root evaluation, ECG-gated CT angiograms of the ascending aorta and heart are obtained with either prospective or retrospective ECG triggering. Aortic valve and root assessment during systole has been shown to be preferable to during diastole because of the larger annular size noted in systole as well as dynamic changes [28, 29]. Thus, with both prospective and retrospective gating, the data should be acquired during systole (usually 20–50 % phase of the cardiac cycle) with no radiation during the rest of the cycle in prospective ECG triggering and very aggressive dose modulation with retrospective gating, allowing substantial dose savings. No routine administration of β-blockers is used for scanning purposes because of concerns with underlying severe aortic stenosis. A high incidence of arrhythmia in TAVR candidates precludes routine use of prospective gating on a routine basis. As a result, the estimated radiation dose may be relatively high, but acceptable given the advanced age of the vast majority of TAVR candidates and the amount of information acquired. For younger patients, with stable and low heart rates, prospective axial acquisition during the systolic phase is suggested [26]. Tube potential of 100 kV is suggested for patients weighing less than 90 kg or with a body mass index (BMI) less than 30, whereas a tube potential of 120 kV is usually indicated for patients weighing more than 90 kg (BMI >30). The lowest setting possible should be selected in keeping with acceptable image noise [30].

Image acquisition triggering can be obtained automatically or by using a bolus injection technique for determining the transit time. The region of interest (ROI) can be placed on the proximal descending aorta using a pre-selected HU threshold. Arterial enhancement in studies performed with a 64-slice scanner demonstrates more consistency with the bolus injection technique, but requires higher contrast volume (typically an additional 20 ml of contrast medium). The vast majority of medical centres imaging TAVR candidates follow one of the two two-step protocols (Table 4). Reduction of contrast volume can be achieved by using lower flow rates such as 3 ml/s or less [31]. Dual-energy techniques allow adequate enhancement with lower intravenous contrast volumes [32]. In extreme cases of renal insufficiency, direct aortic injection with extremely low volumes of contrast (,20 ml) can be used [33, 34].

Image reconstruction of the aortic valve apparatus requires a complex reconstruction technique based purely on aortic valve anatomy. No single image reconstruction protocol exists, but similarities between clinical centres and approaches exist. Currently the most widely used method for reconstruction of aortic valve annulus is based on sequential double-oblique reconstructions that ultimately provide the image corresponding to the aortic valve annulus. The post-processing approach used for assessment of the aortic root and ascending aorta has been described in detail by Achenbach et al. and Leipsic et al. (Fig. 12) [11, 26]. As with all other types of CT post-processing, the quality and precision of reconstructed images depend on the quality of the raw CT data. The presence of a high heart rate, arrhythmia and severe valve calcifications can substantially degrade image quality (Fig. 13). If these limitations present, they should be addressed in the report, and a decision should be made on the necessity of repeated imaging or acquiring the data with alternative imaging modalities.

Measurements of the aortic valve as well as qualitative assessment should be done based on properly reconstructed images (see above). Three main measurements have been suggested as being most informative regarding aortic annular sizing and THV selection (Table 5) (Fig. 14): (1) Aortic annulus diameters: long and short diameters are measured, and the mean diameter is calculated by averaging short and long diameters [35]. (2) The circumference (perimeter) of the aortic annulus can be either calculated under the assumption that the area corresponds to full circularity or measured manually (Fig. 15). (3) Aortic annulus area can be calculated based on the mean diameter or measured manually using the planimetry technique. Aortic annulus dimensions should be measured in systole, similar to echocardiography, since both the area and mean diameters are larger in systole [36]. No substantial difference in circumference was found across the cardiac cycle [26, 37].

Additional aortic valve apparatus measurements pertinent for TAVR dimensions include the aortic valve cusps, aortic valve commissures, sinuses of Valsalva, coronary ostia and sinotubular junction. The parameters important for pre-TAVR assessment are summarised in Table 6. The most important factors are discussed below.

Aortic valve calcifications (AVCs) are either diffusely or focally located on the aortic surface of thickened aortic valve cusps. If excessive, these calcifications may prevent appropriate alignment of the prosthesis, leading to gaps between the prosthetic valve and the aortic annulus and subsequent paravalvular aortic regurgitation (leak), as well as increased risk of annular rupture, device dislodgement or obstruction of coronary ostia due to opposition of calcified aortic cusp [34, 38‐42] (Fig. 16). Calcification of the aortomitral continuity may increase the risk of annular rupture with balloon dilatation (Fig. 17). AVC can be assessed in the cross-sectional view of the sinus of Valsalva (SOV) either qualitatively (mild, moderate, severe) or quantitatively using Agatston units (Agatston Score), the calcified volume score (in mm³) or mass score (in mg of CAHA), similar to the assessment of coronary calcifications [38, 43]. Agatston scores exceeding 3,000 are correlated with an increased incidence of paravalvular regurgitation after TAVR [44]. Semiquantitative scores assess the number of affected cusps, homogeneous versus more focal distribution of calcium and involvement of coronary versus non-coronary cusps. Calcium burden is quantified at the aortic annulus and at the SOV at both its inferior portion and superior aspect. Post-deployment balloon dilatation is required more frequently in patients with severe valvular calcification [45].

Distance of the coronary ostia to the aortic annulus plane: Coronary arteries generally arise within the SOV, below the level of the sinotubular junction (STJ). The right coronary artery generally lies higher than the left coronary artery, with average distance between the plane of the aortic annulus and coronary artery ostium (CAO) of 15.5 mm on the left and 17.3 mm on the right [46]. Degenerative aortic stenosis may result in longitudinal remodeling of the SOV with decreased distance from the aortic valve annulus to the CAO (Fig. 18). The combination of a relatively low-lying coronary artery ostium and a large native aortic valve leaflet can therefore obstruct the flow into the coronary arteries during device deployment. The height of the CAO is determined by identifying the origin of the coronary artery from the cross-section MPR and then measuring the vertical distance between the inferior edge of the CAO and the aortic annular plane. If coronary ostia height is more than 10–14 mm in patients with planned THV implantation, the chance of obstruction is low [11].

The width and maximal height of the SOV are important parameters for coronary perfusion with THV, determining whether the THV will be accommodated within the SOV without causing coronary occlusion from displacement of the native aortic valve leaflets [47] (Fig. 19). As for the rest of the aortic valve apparatus, measurements of the aortic sinus diameter and height should be assessed on a double-oblique projection. The SOV width assessment is particularly important for the CoreValve device with a minimum 27 mm required for the 26-mm prosthesis and 29 mm for the 29- and 31-mm prostheses; the minimum SOV height is 15 mm (Table 6) [26].

The diameter of the ascending aorta: For CoreValue devices, orientation of the direction of the device flow occurs when the frame contacts the inner and outer curvature of the aorta. Ascending aortic dilatation of more than 43 mm (when measured 4 cm above the basal aortic annulus plane) precludes the use of the 29- and 31-mm CoreValve device. Dilatation of more than 40 mm precludes the use of the 23- and 26-mm CoreValve device. Size is of less concern for valves that are short in length and confined to the aortic annulus and sinus of Valsalva such as the Edwards SAPIEN valve [11].

Additional features that might interfere with successful implantation: The left ventricular outflow tract (LVOT) is the area between the basal annulus and the anatomic aortoventricular junction. The atrioventricular (AV) node is located adjacent to the membranous septum of the LVOT below the aortic valve. Deep positioning of the THV into the LVOT may cause injury to the interventricular septal and artrioventricular conduction system. The degree of LV upper septal hypertrophy protruding into the LVOT can hinder accurate placement of the valve and present a significant risk of THV repositioning. The aortoventricular angle is the angle between the proximal aorta, aortic annulus and LVOT (Fig. 20). This angle is determined by the angle of the horizontal plane and the angle of the aortic annulus [26]. The aortoventricular angle is an important consideration when using THVs that are long in length and require deployment perpendicular to the native annular plane. Each type of THV has a maximum aortoventricular angle for successful valve deployment, typically 70º for the iliofemoral approach and 30º for the subclavian approach, due to angulation from the right subclavian into the ascending aorta.

Imaging of peripheral access can be obtained with non-gated spiral acquisition to minimise radiation. Peripheral access includes the iliofemoral, transsubclavian, transapical and direct aortic access routes. Transapical and direct aortic access (ascending aorta) routes are assessed with the ECG-gated portion of the MDCT. The common femoral artery should be included in the field of view when assessing femoral access. Assessment of the axillary and subclavian arteries should be obtained with the patient’s arms placed along his/her body to exclude pseudo narrowing of the imaged vasculature.

MDCT findings crucial in the peripheral vascular pathway assessment are summarised in Tables 1, 2 and 6. MDCT selection of patients with appropriate peripheral access has substantially reduced the risk of vascular complications from 30.1 % [12] to 8 % [48]. A sheath-to-femoral artery ratio ≥1.05 has been shown to be associated with increased vascular complications and 30-day mortality [36, 48]. A standardised approach for peripheral access has been developed, adding significant reduction in vascular injury rates [48]. Accepted practice is to evaluate peripheral access with several reconstruction techniques: curved multiplanar reconstructions (MPR), 3D volume-rendered images and, if necessary, maximum intensity projection (MIP) images (Fig. 21). These multiple measurements are taken along the common iliac, external iliac, common femoral, innominate, subclavian and proximal axillary arteries bilaterally.

Vessel tortuosity in the iliofemoral or brachiocephalic/subclavian arteries can be optimally assessed using 3D reconstructions, with multiple oblique views seen along the 360° circle in addition to the anterior-posterior view. Precise angle measurements are obtained using PACS or workstation angle measurement tools. Note that vessel angulation of 90° or more may be a contraindication for large-bore catheter insertion (Fig. 22). The burden and pattern of vascular calcifications and presence of circumferential or horseshoe calcification can be better assessed on curved multiplanar and maximum intensity projection (MIP) reconstructions. The combination of relatively small vessel caliber or stenotic segments with these types of vascular calcifications is a relative contraindication for a transfemoral approach. However, a relatively straight segment with no substantial calcification or atheroma can be cannulated even if marginally smaller in diameter than the intended sheath [49]. The presence of dissection or a complex atheroma is assessed on both axial and curved reconstruction images, with such findings again considered contraindications for a transfemoral approach. If the iliofemoral approach is not an option for a given patient, transapical, direct aortic, direct brachiocephalic or subclavian artery approaches should be assessed (Figs. 23 and 24).

For all TAVR approaches, assessment of the entire aorta is done routinely as part of the pre-procedure protocol using curved MPR with subsequent creation of true perpendicular axial images for precise patent lumen diameter assessment. For the transfemoral approach, the aorta is assessed to exclude the presence of an abdominal aneurysm, elongation with kinking of the aorta, dissection or large thrombi protruding into the lumen and/or complex atheromas, as these are contraindications for transfemoral approach [11]. For the direct aortic approach, assessment of the ascending aorta for the presence and pattern of calcification is of paramount importance, as calcium may interfere with direct puncture of the aorta. In these cases, the ascending aorta from the aortic annulus plane to the origin of the brachiocephalic artery is assessed for the presence of aneurysmatic dilatation or circumferential/anterior calcifications (Fig. 25). The relationship between the aorta and chest is important for planning thoracic chest wall incisions. It is important to assess the retrosternal area for the presence of masses particularly in post-CABG patients to avoid bypass graft injury. After median sternotomy or a right parasternal incision has been performed for the direct aortic approach, the aorta is punctured either through its anterior or right lateral aspects (accordingly), usually within the most proximal 8.5 cm, to allow sufficient room for CoreValve device expansion. MDCT reconstructions routinely include 3D VR reconstructions with osseous structures providing anatomic landmarks for future thoracotomy. For transapical access, assessment of the entire aorta is less crucial, unless a change in the approach is anticipated during the procedure. The crucial data are the position of the left ventricle (LV) apex relative to the chest wall and alignment of the LV axis with the LV outflow tract. Also, left atrial (LA) and/or LV thrombi can be a source of embolic complications that will need to be identified and reported. As with the direct aortic approach, chest deformities and mediastinal lesions are important and should be reported and reviewed with the team prior to the procedure.

Historically, patient eligibility for transcatheter valve therapy and sizing of the prosthesis were largely based upon aortic annulus measurements on 2D echocardiography (transthoracic or transesophageal) and occasionally on angiography. The major disadvantage of these techniques [50‐52] is the underestimation of the oval and not circular shape of the aortic annulus (Figs. 26 and 27) [20]. Direct comparison of TEE and MDCT aortic annulus measurements is problematic because of different parameter measures by each of the modalities. Aortic annulus diameter on coronal view by MDCT has been shown to be significantly larger than that obtained on sagittal view by MDCT, TTE or TEE [53]. Long-axis measurement of the aortic annulus by TTE and TEE approximates the minor axis of the elliptically shaped annulus as measured by MDCT (Fig. 28). Thus, consistently larger measurements obtained with MDCT as compared with 2D echocardiography most likely reflect the measurements of the largest of the two diameters of this oval-shaped annulus as opposed to the smaller one measured by echocardiography [54‐56]. The 2D echocardiography usually underestimates the aortic annulus average diameter by 1 ± 1.7 mm [26, 52]. As a result of the eccentric geometry of the oval aortic annulus, correlative studies have shown a systematic underestimation of annular sizing by 2D echocardiography alone [57].

MDCT can provide better estimates of both the long- and short-axis diameter of the aortic annulus, surface area and perimeter measurements. Extensive work has been done to establish the role of MDCT in aortic annular sizing with Edward Sapiens transcatheter heart valves (THV) [11, 26, 46]. While exaggerated oversizing of the THV and excessive calcification of the native aortic valve can result in aortic annular rupture, some THV oversizing is required for both types of THV to prevent PAR. For the CoreValve device the recommended device/annulus oversizing is 15 % of the aortic basal ring perimeter. For the Edwards Sapiens valve, the recommended device/annulus oversizing is 15–25 % of the area and 7–12 % of the mean diameter. This degree of oversizing of the THV appears to provide the best risk-benefit ratio in terms of PAR reduction and conduction disorders [57, 58]. Undersizing of the THV can lead to increased PAR and greater likelihood of valve ‘pop-out’ or migration [10, 59].

Post-implantation imaging can be divided into immediately post-procedure and long-term follow-up. For the immediate assessment of valve position and haemodynamic status, including the gradients and effective valve area, the modality of choice is TEE, which can be done in the hybrid procedure room. Paravalvular and transvalvular regurgitation can also be estimated in real time, allowing appropriate steps to be taken to minimise complications and optimise device positioning [16, 54].

For long-term follow-up of TAVR ECG-gated MDCT is useful in diagnosing prosthesis misplacement by careful inspection of the exact positioning of the device in relation to the aortic annulus plane [55].

Significantly low implantation may result in severe paravalvular regurgitation, residual aortic valve stenosis, mitral valve insufficiency, conduction abnormalities and, in extreme cases, device drop into the left ventricular cavity. Considerably high implantation may result in paravalvular regurgitation, coronary flow obstruction and device embolisation into the thoracic aorta [10, 60].

The exact definition of misplacement is specific to the TAVR device type. The optimal deployment location of the Medtronic CoreValve® device requires that the most inferior edge of the device, the inflow portion, should be positioned approximately 4–6 mm below the aortic annulus plane [61, 62] (Fig. 29). The optimal deployment location of the Edwards SAPIEN device is likely gained when 50 % of its height is below and 50 % of its height above the aortic annulus plane [10, 60]. (Fig. 30).

TAVR is a novel procedure that provides unique opportunities for multimodality research comparing different imaging strategies and different THV devices. Given the exponential rise in the number of procedures being performed with varying local expertise, there is a need to standardise imaging protocols as part of pre- and post-procedure assessment of TAVR patients. Implementation of standardised protocols is crucial not only for precise assessment of the aortic valve apparatus and vascular access with different scanners in different institutions, but also for creating opportunities for further multicentre research. The SCCT expert consensus document devoted to this topic suggests potential algorithms and techniques for imaging before TAVR. Although there is growing evidence of the importance of MDCT in the pre-procedure assessment, one of the current challenges is to find the precise role of MDCT in valve assessment and sizing. Thus, developing a rational algorithm for imaging use both before and after implant should become a priority to avoid substantial redundancy in imaging. Another potential role for imaging is the assessment of outcome prediction, based on the diagnosis of both cardiac and non-cardiac comorbidities. If successful, this should allow for optimised patient selection for the currently costly procedure. Overall, current trends strongly suggest that imaging, in particular MDCT, will play an increasingly important role in all aspects related to the TAVR procedure.

During the last decade transcatheter aortic valve replacement has become widely used in many centres across the world with good clinical outcomes. The planning and performance of the procedure are based on a multidisciplinary team approach, with imaging proving crucial for pre-procedure planning. MDCT plays an important role in assessing the aortic valve, aortic root and ascending aorta, as well as the access root for the procedure (i.e., ilio-femoral, direct aortic or subclavian) utilising the combination of multiplanar reconstruction and 3D imaging. TAVR technology and equipment are rapidly advancing, with increased utilisation of advanced MDCT imaging contributing to continuing outcome improvement and broadening of procedure applications.