Canadian Journal of Cardiology 30 (2014) 1073e1079
Clinical Research
Measurement of the Aortic Annulus Diameter Using Transesophageal Echocardiography and Multislice Computed TomographydAre They Truly Comparable? Jean-Michel Serfaty, MD, PhD,a,d,e Dominique Himbert, MD,b Marina Esposito-Farese, PhD,c Nicoletta Pasi, MD,a Vania Tacher, MD,a Jean-Pierre Laissy, MD, PhD,a Florence Tubach, MD, PhD,c Bernard Iung, MD,b,d Alec Vahanian, MD,b,d and David Messika-Zeitoun, MD, PhDb,d,e a b c
Department of Radiology, Assistance PubliqueeHôpitaux de Paris, Paris, France
Department of Cardiology, Assistance PubliqueeHôpitaux de Paris, Paris, France
Department of Biostatistics, Assistance PubliqueeHôpitaux de Paris, Paris, France d e
INSERM U1148, Paris, France
University Paris 7, Bichat Hospital, Paris, France
ABSTRACT
RESUM E
Background: For transcatheter aortic valve implantation (TAVI), transesophageal echocardiography (TEE) and multislice computed tomography (MSCT) measurements of the aortic annulus diameter (AAD) are often regarded as competitive. We evaluated if 1 MSCT method could be interchangeable with TEE measurements. Methods: We compared AAD measurements performed using TEE, MSCT, and transthoracic echocardiography (TTE) in 129 consecutive patients with severe aortic stenosis (AS) who were referred for TAVI. Using MSCT, AAD was measured in the 3-chamber (3C) view and at the level of the virtual basal ring (mean diameter [MD] of the long-axis [LA] and short-axis [SA] diameters, and AAD derived from the crosssectional area [CSA] and the circumference). Correlations with echocardiographic measurements and agreement regarding the TAVI strategy (decision to implant and choice of prosthesis size based on manufacturer cutoff recommendations) were assessed. Results: AAD measured in 3C (intraclass correlation [ICC], 0.79) and MD emphasizing the weight of the SA (MD4 [3 SA þ LA/4]; ICC, 0.76)
ter Introduction : Pour l’implantation valvulaire aortique par cathe (IVAC), les mesures du diamètre de l’anneau aortique (DAA) par chocardiographie transœsophagienne (ETO) et tomodensitome trie e re es comme concurrentes. multicoupe (TDM-MC) sont souvent conside value si une me thodes de TDM-MC pouvait être interNous avons e changeable aux mesures de l’ETO. thodes : Nous avons compare les mesures du DAA re alise es à l’aide Me chocardiographie transthoracique (ETT) de l’ETO, de la TDM-MC et de l’e cutifs souffrant d’une ste nose aortique (SA) grave chez 129 patients conse taient oriente s pour subir une IVAC. À l’aide de la TDM-MC, le DAA a qui e te mesure sur la coupe des 3 cavite s (3C) et au niveau de l’anneau basal e virtuel (diamètre moyen [DM] des diamètres du grand axe [GA] et du petit rive de la coupe transversale [CT] et de la ciraxe [PA], et DAA de rence). Les corre lations avec les mesures e chocardiographiques et confe gie d’IVAC (de cision quant à l’imla concordance concernant la strate plantation et le choix de la taille de la prothèse selon les seuils s par le fabricant) ont e te e value es. recommande
Transcatheter aortic valve implantation (TAVI) is now considered a valuable alternative to conventional surgery for high-risk patients with severe symptomatic aortic stenosis (AS).1-5 Accurate measurements of the aortic annulus diameter (AAD) are crucial for patient selection and procedural
success. Transesophageal echocardiography (TEE), long considered the reference method, has recently been challenged by multislice computed tomography (MSCT). In the absence of randomized studies, the method that should be used remains a matter of debate, and in routine practice both TEE and MSCT are often performed, leaving the clinician facing 2 different measurements. Instead of opposing the 2 methods, we wondered if 1 MSCT method could correspond or be interchangeable with TEE measurements. The aortic annulus (AA) is a complex 3-dimensional crown-like structure with an oval shape.6,7 MSCT measurements can be performed in multiple planes, and AAD can also
Received for publication December 9, 2013. Accepted June 20, 2014. Corresponding author: Dr Messika-Zeitoun, Department of Cardiology, Assistance PubliqueeHôpitaux de Paris, Bichat Hospital, 46 rue Henri Huchard, 75018 Paris, France. Tel.: 01 40 25 86 61; fax: 01 40 25 67 32. E-mail: [email protected] See page 1078 for disclosure information.
http://dx.doi.org/10.1016/j.cjca.2014.06.012 0828-282X/Ó 2014 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved.
1074
Canadian Journal of Cardiology Volume 30 2014
and MD5 [4 SA þ LA/5; ICC, 0.75]) provided the highest correlation and the best agreement with TEE (kappa ¼ 0.47, 0.27, and 0.31 respectively). However, TTE provided a better a correlation and agreement with TEE than all MSCT methods (ICC, 0.87; kappa ¼ 0.66). The agreement between MSCT and TEE varied with AAD eccentricity and degree of aortic valve calcification (AVC), but in all subsets, values observed with MSCT never reached those observed with TTE. Conclusions: MSCT and TEE are measuring different landmarks and consequently MSCT and TEE measurements are not interchangeable. Prospective randomized studies aimed at defining which method provides the best clinical results are clearly needed.
sultats : Le DAA mesure sur la 3C (coefficient de corre lation Re intraclasse [CCI], 0,79), le DM renfonçant le poids du PA (DM4 [3 PA þ GA/4]; CCI, 0,76) et le DM5 [4 PA þ GA/5; CCI, 0,75]) offraient la lation la plus e leve e et la meilleure concordance avec l’ETO corre (kappa ¼ 0,47, 0,27 et 0,31, respectivement). Cependant, l’ETT offrait lation et une meilleure concordance avec l’ETO que une meilleure corre thodes de TDM-MC (CCI, 0,87; kappa ¼ 0,66). La toutes les me du concordance entre la TDM-MC et l’ETO variait avec l’excentricite de calcification de la valve aortique (CVA), mais dans DAA et le degre es avec la TDM-MC n’attous les sous-groupes, les valeurs observe es avec l’ETT. teignaient jamais celles observe Conclusions : La TDM-MC et l’ETO mesurent des points de repère rents et par conse quent les mesures de la TDM-MC et de l’ETO ne diffe tudes prospectives randomise es sont pas interchangeables. Des e terminer quelle me thode fournit les meilleurs ayant pour but de de sultats cliniques sont ne cessaires. re
be derived from multiple measurements (diameter, area, and circumference). In addition, because the AA is oval with a long axis (LA) and a short axis (SA), mathematical transformations emphasizing the weight of the SA may provide closer agreement with TEE than direct measurements of the AAD. Thus, in the present study, we aimed to compare in a series of patients referred for TAVI, direct AAD measurements, AAD measurements derived from circumference/area, and mathematical transformation of direct AAD measurements obtained using MSCT-to-TEE measurements.
Amsterdam, the Netherlands). A scan run consisted of a prospective acquisition of 43-mm-thick contiguous transverse slices. Acquisition time was 0.5 sec/slice, electrocardiographically triggered at 75% of the R-R interval. Degree of aortic valve calcification (AVC) was quantified using the Agatston method and expressed in arbitrary units. Two MSCT runs were performed sequentially with a 1- or 2-mm initial interval. Each run was independently scored and the 2 scores were averaged.8
Methods
Multislice computed tomography Scanning. All examinations were performed using a 64-detector computed tomography scanner (Lightspeed VCTXT, GE Healthcare, Milwaukee, WI) and intravenous injection of 90 mL iobitridol 350 mg/mL. Acquisition was centred at the 75% phase of the R-R cardiac cycle when the heart rate was < 65 beats/min and at 40% when the heart rate was > 65 beats/min to ensure minimum motion artifacts. No bblockers were administered.
Study population Consecutive patients with severe AS (aortic valve area < 1 cm2 or mean gradient 40 mm Hg, or both) who were referred for TAVI and underwent TTE, TEE, and MSCT at our institution within 1 month were enrolled in the present study. All examinations were clinically indicated as workup for TAVI. Exclusion criteria included patients with bicuspid aortic valves and suboptimal TEE or MSCT images precluding accurate AAD measurements. Patients with atrial fibrillation were not excluded. The study was approved by our local ethics committee, and all patients gave written informed consent. Two-dimensional echocardiography Comprehensive transthoracic echocardiography (TTE) and TEE were performed by experienced echocardiographers using high-quality commercially available ultrasound systems (IE33; Philips Medical Systems, Cleveland, OH and Vivid 7; General Electric, Vingmed, Horten, Norway). AAD was measured using the zoom mode at the insertion of the leaflets in midsystole from the parasternal LA view in TTE or from the 120 -140 LA view (3-chamber [3C] view) in TEE. Measurements were averaged from 3-5 beats. Evaluation of AS severity was based on mean transaortic gradient and the aortic valve area calculated using the continuity equation. Calcium scoring scanning MSCT was performed using a 16-detector computed tomography scanner (MX 8000 IDT 16; Philips Healthcare,
Image analysis. As previously described,9 we used a double oblique multiplanar reconstruction to obtain 1 slice perpendicular to the aortic root including the 3 basal points of attachment of the annulus (transverse view corresponding to the virtual basal ring) and 1 slice including the aortic root, the left ventricular outflow tract, and the left atrium and ventricle (3C view), similar to the LA echocardiographic plane. On the 3C view, we measured the distance between the hinge points of the leaflets. On the transverse view, we measured SA and LA diameters of the annulus, the cross-sectional area (CSA), and the circumference (C). From these measurements, we calculated the mean diameter (MD ¼ [SA þ LA]/2), the diameter derived from the CSA (CSAeD ¼ O(4 * CSA/p)) and the diameter derived from the circumference (CirceD ¼ C/p) considering the AA to be circular. Because the AA is oval and the SA underestimates the AAD measured using TEE, whereas the LA and MD tend to overestimate it, we assessed a range of modified MD calculations enhancing the weight of SA: MD2 ¼ (2 * SA þ LA)/3, MD3 ¼ (3 * SA þ LA)/4, MD4 ¼ (4 * SA þ LA)/5, MD5 ¼ (5 * SA þ LA)/6. The eccentricity index (EI) was calculated as follows: EI ¼ 1 e SA/LA. MSCT
Serfaty et al. Aortic Annulus Diameter in TAVI
measurements were performed blindly from echocardiographic measurements. Statistical analysis Continuous variables are expressed as mean standard deviation. Comparisons between TEE and MSCT or TTE measurements were assessed as described by Bland-Altman analysis and single-measure ICC. Interobserver and intraobserver variability of AAD measurements was calculated in 30 patients as absolute difference standard deviation. MSCT variability required new reconstructions, whereas echocardiographic measurements required new acquisitions. The Edwards-Sapien valve (Edwards Lifesciences, Irvine, CA) is currently available in 3 sizes: 23, 26, and 29 mm. According to the manufacturer’s recommendations, a 23-mm prosthesis should be implanted if the annulus is > 18 mm and 21 mm, a 26-mm prosthesis should be implanted if the annulus diameter is > 21 and 25 mm, and a 29-mm prosthesis should be implanted if the annulus diameter is > 25 and 27 mm; the procedure should not be performed if the annulus diameter is 18 or > 27 mm. A contingency analysis was performed to assess the agreement between the different imaging modalities regarding the decision to implant a prosthesis and the choice of prosthesis size expressed by the kappa value. P < 0.05 was considered statistically significant. Results Population One hundred twenty-nine patients who underwent TTE, TEE, and MSCT in our hospital within 1 month with imaging quality compatible with accurate AA analysis constituted our study population. Mean age was 80 10 years, 69 patients (54%) were men, and 19% had atrial fibrillation. Mean aortic valve area was 0.68 0.16 cm2 and mean gradient was 51 15 mm Hg. Mean left ventricular ejection fraction was 49% 15% and < 50% in 48 patients (41%). Mean logistic euroSCORE and Society of Thoracic Surgeons score were 28% 15% and 15% 9% respectively. Comparison between echocardiographic and MSCT measurements Mean AAD was 23.4 2.0 mm using TEE and 23.0 2.0 mm using TTE. TTE and TEE were closely correlated (ICC, 0.87; 95% confidence interval [CI] 0.83-0.91), and the quality control plots using the Altman-Bland method showed that there was no trend for underestimation or overestimation (mean difference, 0.3 1.0 mm) (Table 1 and Fig. 1). The absolute difference between methods was 0.72 0.70 mm. MSCT measurements are reported in Table 1. The 3C method was also well correlated with TEE (ICC, 0.79; 95% CI, 0.73-0.83) with no trend for underestimation or overestimation (mean difference, 0 1.5 mm). Compared with TEE measurements, SA was significantly smaller and LA was significantly larger. The MD provided significantly larger AAD measurements (mean difference, 1.5 1.5 mm) than TEE, with an ICC lower than the 3C method (ICC, 0.62; 95% CI, 0.53-0.69). Mean difference decreased from MD to MD5 and was null for MD4. With this latter method, the
1075
ICC was also good (ICC, 0.76; 95% CI, 0.69-0.81). AAD derived from the CSA was approximately 1 mm larger than with TEE, whereas AAD derived from the circumference was markedly larger (mean difference, 7.0 1.6 mm). Thus, the 3C and MD4 methods provided the better correlations with TEE measurements, with no trend for underestimation or overestimation. The worst intraobserver and interobserver variability was observed with the 3C and circumference methods (Table 1). Agreement between methods regarding the TAVI strategy We evaluated the theoretical impact of the method of AAD measurement on TAVI strategy (choice of prosthesis size and decision to implant) based on cutoffs proposed by the manufacturer using TEE as the reference method (Table 2). TTE provided the best agreement with TEE (concordance of TAVI strategy in 83% of patients; kappa, 0.66), and all MSCT methods provided a lower agreement with kappa values between 0 and 0.47. The best computed tomographic methods were 3C and MD5. The number of patients in whom the prosthesis size and decision to implant changed is presented in Table 2. We further evaluated whether the agreement between TEE and MSCT was influenced by the degree of eccentricity of the AA (expressed as the EI) or the degree of AVC (calcium score measured using MSCT; available in 91 patients). The mean EI was 0.19 0.07 (range, 0-0.34) and the mean calcium score was 3351 1598 (range, 759 to 9030). The kappa values varied according to tertile of EI or calcium scoring but never reached the values observed with TTE (Supplemental Table S1). Combination of TTE and MSCT We then explored whether a combination of TTE and MSCT could improve the agreement with TEE (Supplemental Table S2). The 3C, MD4, and MD5, which provided the best agreement with TEE, were evaluated. The population was divided into 2 groups according to the absolute difference between TTE and MSCT (> 1 mm or 1 mm). Agreement between the 3 MSCT methods and TEE was better in the subgroup of patients in whom the difference between MSCT and TTE measurements was 1 mm than in the subgroup with a > 1 mm difference (kappa, 0.50-0.67 vs 0.11-0.33) but remained lower than the agreement between TTE and TEE (kappa, 0.63-0.78). Discussion In the present study, in a cohort of patients with severe AS referred for TAVI, we showed that (1) no direct or indirect MSCT measurements were more closely correlated with TEE than TTE and, consequently, (2) all MSCT methods had a lower agreement with TEE than TTE regarding the TAVI strategy, (3) the degree of agreement between TEE and MSCT was influenced by the degree of eccentricity or the degree of AVC but remained lower than with TTE in all subgroups, and (4) combining TTE and MSCT did not provide a better agreement with TEE than did the single use of TTE.
1076
Canadian Journal of Cardiology Volume 30 2014
Table 1. Comparison between echocardiographic and MSCT measurements using transesophageal echocardiography as reference
Method TEE TTE MSCT-3C MSCT-SA MSCT-LA MSCT-MD MSCT-MD2 MSCT-MD3 MSCT-MD4 MSCT-MD5 MSCT-CSA-D MSCT-Circ-D
Mean diameter 23.4 23.0 23.4 22.3 27.6 25.0 24.1 23.6 23.4 23.2 24.6 29.4
2.0 2.0 2.6 2.6 2.9 2.5 2.4 2.5 2.5 2.5 2.3 2.4
Difference Mean e 0.3 0 1.1 4.2 0.5 0.7 0.2 0 0.2 1.2 7.0
1.0 1.5 1.8 2 1.5 1.5 1.5 1.6 1.6 1.6 1.6
Absolute 0.72 1.21 1.51 4.24 1.83 1.40 1.30 1.30 1.29 1.58 6.01
e
0.70 0.91 1.36 1.95 1.17 0.86 0.86 0.87 0.91 1.04 1.54
ICC (Lower-higher 95% CI) 0.87 0.79 0.64 0.28 0.62 0.74 0.76 0.76 0.75 0.62 0.17
e (0.83-0.91) (0.73-0.83) (0.53-0.73) (0.22-0.33) (0.53-0.69) (0.68-0.79) (0.70-0.82) (0.69-0.81) (0.67-0.81) (0.52-0.71) (0.13-0.2)
Variability Intraobserver 0.60 0.60 0.96 0.84 0.91 0.60 0.64 0.66 0.68 0.71 0.62 0.90
0.50 0.80 0.92 0.68 0.72 0.63 0.62 0.64 0.65 0.65 0.43 0.75
Interobserver
NA
0 4 2 0 0 0 0 0 0 0 0 3
0.60 1.00 1.48 0.33 0.94 0.67 0.77 0.84 0.88 0.91 0.67 2.05
1.00 0.70 1.15 1.51 1.17 0.61 0.64 0.71 0.78 0.83 0.54 0.82
Results are presented as mean value standard deviation. 3C, three-chamber; CircD, diameter derived from the annulus circumference; CSA-D, diameter derived from aortic cross-sectional area; ICC, intraclass correlation coefficient; LA, long axis; MD, mean diameter; MSCT, multislice computed tomography; NA, measurements not available for analysis; SA, short axis; TTE, transthoracic echocardiography; TEE, transesophageal echocardiography.
In contrast to surgical aortic valve replacement, sizing in TAVI is based on imaging. Accurate measurements of the AA size are thus crucial for procedural success and to minimize complications. Undersizing may lead to paravalvular regurgitation or prosthesis migration, whereas oversizing may lead to annulus rupture.10 Current recommendations for valve sizing
have been established by and for echocardiography, and despite limitations, TEE was first considered as the reference method before being challenged by MSCT.11-13 The AAD can be measured using multiple MSCT methods. The 3C method mimics the echocardiographic LA view, and consequently this method provided the highest
Figure 1. Quality control plots using Bland-Altman analysis compared with transesophageal echocardiography (TEE): (A) Transthoracic echocardiography (TTE), (B) 3-chamber (3C), (C) mean of the long-axis (LA) and short-axis (SA) mean diameter (MD), (D) mean of the LA and SA diameter overpowered for the SA (MD4), (E) diameter derived from the cross-sectional area of the virtual basal ring (CSA-D), (F) diameter derived from the circumference of the virtual basal ring (CircD). The middle line represents the mean and the upper and lower lines 2 standard deviations.
Serfaty et al. Aortic Annulus Diameter in TAVI
1077
Table 2. Theoretical impact of the method of annulus diameter measurements on TAVI strategy (decision to perform the procedure and choice of the prosthesis size) using transesophageal echocardiography as reference TAVI strategy Method TEE TTE MSCT-3C MSCT-SA MSCT-LA MSCT-MD MSCT-MD2 MSCT-MD3 MSCT-MD4 MSCT-MD5 MSCT-CSA-D MSCTeCirc-D
Agreement with TEE
23-mm prosthesis
26-mm prosthesis
29-mm prosthesis
No implantation
n (%)
28 25 23 39 0 8 17 17 18 20 10 0
78 86 69 65 28 57 68 72 77 77 64 2
20 11 24 15 30 37 28 24 19 19 35 17
3 3 11 10 71 27 16 16 15 13 20 107
129 103 (83) 86 (68) 80 (62) 16 (12) 52 (40) 68 (53) 68 (53) 74 (57) 77 (60) 63 (49) 9 (7)
No agreement with TEE
Kappa
Smaller prosthesis size
Larger prosthesis size
New implantation
No implantation
e 0.66 0.47 0.38 0 0.11 0.23 0.21 0.27 0.31 0.18 0
e 13 9 28 0 1 7 10 11 13 2 0
e 7 20 10 45 50 37 34 28 25 45 13
e 1 2 2 0 1 2 2 2 2 1 0
e 1 10 9 68 25 15 15 14 12 18 104
Data are presented as number of patients (percent). 3C, three-chamber; CircD, diameter derived from the annulus circumference; ICC, intraclass correlation coefficient; LA, long axis; MD, mean diameter; MSCT, multislice computed tomography; NA, measurements not available for analysis; SA, short axis; SurfD, diameter derived from the annulus area; TAVI, transcatheter aortic valve transplantation; TTE, transthoracic echocardiography; TEE, transesophageal echocardiography.
correlation with TEE measurements and the highest agreement regarding the TAVI strategy. Nevertheless, the agreement between the 3C method and TEE remained lower (kappa, 0.47) than the agreement observed with TTE (kappa, 0.66). In addition, the 3C method was hindered by modest intraobserver and interobserver variability. The AA is oval rather than circular.7,9 Because the SA systematically underestimates and the LA systematically overestimates the AAD measured using TEE, the weight of the SA was progressively overpowered using mathematical transformations (MD2-MD5). These mathematical transformations were associated with an improvement of the correlations between MSCT and TEE measurements and a parallel improvement in the agreement between these 2 methods regarding the TAVI strategy. However, the kappa values remained in a modest range (between 0.23 and 0.31). The AA is a complex crown-like structure, and TEE measurements are performed between the basal attachments of 1 leaflet to the opposing point across the lumen toward the intercommissural triangle (Fig. 2) and are, therefore, also dependent on the diameter of the aortic root. In contrast, SA and LA are measured at the level of the so-called virtual basal ring. Thus, MSCT and TEE definitively measure diameters located in completely different 3-dimensional planes. We have previously shown that the use of the mean diameter instead of TEE measurements significantly changes the TAVI strategy.9 We extend these findings to mathematical transformation of MSCT emphasizing the weight of the SA. It is worth noting that the agreement between TEE and MSCT methods was only mildly influenced by the EI, reinforcing the fact that the intrinsic anatomical differences in the diameters measured by both methods cannot be compensated by any mathematical formula. We also evaluated diameters derived from the CSA and the annulus circumference of the virtual basal ring. Diameter derived from the CSA was significantly larger than TEE measurements. In our study, the difference between both methods (1.2 1.6 mm) was very similar to that reported in a recent article (1.2 1.7 mm).10 The ICC between both methods was correct (ICC, 0.62), but the agreement was poor (kappa, 0.18). The larger diameter observed with this MSCT
method translated into the use of a larger prosthesis in one third of the population and cancellation of the procedure in 15%. One may hypothesize that if the AA becomes circular after TAVI implantation with CSA changes, whereas the circumference may remain unchanged, use of diameter measurements derived from the circumference may be the ideal method. In the present study, diameters derived from circumference were dramatically larger than TEE measurements (7.0 1.6 mm), with a poor ICC and poor agreement with TEE. Interestingly, use of the AAD derived from the circumference would have led to cancellation of the procedure in more than 80% of the population. A lower agreement using diameters derived from the circumference has been previously reported.14 Therefore, this method seems inadequate for selection of patients for TAVI. The degree of AVC has been associated with the rate of procedural success and more specifically the incidence of paravalvular regurgitation.15,16 In the present study, the agreement between MSCT and TEE was influenced by the degree of AVC but remained lower than that between TTE and TEE. In addition, leaflet and annular calcifications are often asymmetrical, which may add difficulty to any method for accurately assessing the shape and size of the AA. Finally, we also evaluated whether a combination of TTE and MSCT could avoid the performance of TEE. A difference < 1 mm between TTE and MSCT was observed in half of the patients in whom such a strategy could have been an alternative. Unfortunately, MSCT and TTE were not synergistic enough, and even in this subset, MSCT did not perform better than TTE alone. The results of the present study may have important clinical implications. Because TEE measurements are often smaller than most MSCT measurements, some authors have suggested systematically shifting from the 1/1.5-mm cutoff values provided by manufacturers.17 The present clinical evaluation clearly demonstrates that MSCT measurements are not interchangeable, even using mathematical formulas. Thus, TEE and MSCT are intrinsically different, and because they measure different landmarks, the question of TAVI sizing from a clinical point of view should not be asked in terms of “which method is the true one” but as “which one provides
1078
Canadian Journal of Cardiology Volume 30 2014
Figure 2. Schematic representation of the aortic annulus and the diameters measured using transesophageal echocardiography (red) and multislice computed tomography (long and short axis in yellow) at the level of the virtual basal ring (blue). Transesophageal diameter is in a different plane than the virtual basal ring.
the best clinical results?” Some centres are already considering CSA-D in clinical practice, and retrospective analyses suggest a lower paravalvular aortic regurgitation rate using MSCT measurements,12,13 which may be associated with a better outcome. In our centre, echocardiographic and MSCT measurements are considered complementary and are integrated with clinical data for final sizing decisions. Dedicated strategies based on MSCT measurements should be developed and results of such strategies compared with results of TEE-based strategies including end points such as rate of aortic regurgitation and annulus rupture as well as the number of patients who would be offered TAVI. In a recent study, a MSCT algorithm was associated with a lower rate of paravalvular aortic regurgitation, but in as many as 20% of patients, MSCT measurements were finally disregarded.18 In our opinion, these results reinforce the idea that TEE and MSCT measurements should not be systematically opposed, but because they are measuring different landmarks, they should be regarded as complementary, and our results as well as those of the previously mentioned study, are a strong argument for an integrative approach. The present study deserves several comments. First, we did not aim to define whether MSCT was better or worse than TEE regarding the rate of procedural success, which would ideally require a randomized study on TEE- vs MSCT strategye based implantation. There is no gold standard method for AAD measurements and we “only” performed comparisons between methods. Second, the agreement between methods was assessed based on cutoff values used for the balloonexpandable Edwards-Sapien prosthesis. However, despite different annulus diameter ranges and cutoffs, as well as different prosthesis sizes with the CoreValve Revalving System (Medtronic, Minneapolis, MN), our conclusions with
modest agreement between TEE and computed tomography are not dependent on the type of device. Computed tomographic, and echocardiographic measurements were not exactly performed at the time (end-systole/mid-diastole vs midsystole, respectively). However, variations of the annulus size during the cardiac cycle are limited,7 especially in patients with severe AS with severe calcifications. Finally, this was not a feasibility study, and patients with suboptimal echocardiographic or MSCT examinations were excluded. Conclusions In this series of patients with severe symptomatic AS who were referred for TAVI and underwent both TEE and MSCT at our institution, we demonstrated using multiple MSCT methods that MSCT and TEE measurements are not interchangeable. Because MSCT and TEE are intrinsically different and measure different landmarks, they should not be regarded as competitive but more probably as complementary. Prospective randomized studies aiming at defining which method provides the best clinical results are clearly needed and should compare not only each method separately but also an integrative approach based on both measurements. Funding Sources The present work was partially funded by the COFRASA and GENERAC studies (Assistance PubliqueeHôpitaux de Paris; PHRC National 2005 and PHRC regional 2007, respectively). Disclosures See Supplementary Material for disclosures.
Serfaty et al. Aortic Annulus Diameter in TAVI
References 1. Vahanian A, Alfieri O, Andreotti F, et al. Guidelines on the management of valvular heart disease (version 2012): The Joint Task Force on the Management of Valvular Heart Disease of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS). Eur Heart J 2012;33:2451-96. 2. Leon MB, Smith CR, Mack M, et al. Transcatheter aortic-valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med 2010;363:1597-607.
1079
regurgitation: comparison among cardiovascular magnetic resonance, cardiac computed tomography, and echocardiography. J Am Coll Cardiol 2011;58:2165-73. 12. Jilaihawi H, Kashif M, Fontana G, et al. Cross-sectional computed tomographic assessment improves accuracy of aortic annular sizing for transcatheter aortic valve replacement and reduces the incidence of paravalvular aortic regurgitation. J Am Coll Cardiol 2012;59:1275-86.
3. Smith CR, Leon MB, Mack MJ, et al. Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med 2011;364: 2187-98.
13. Willson AB, Webb JG, Labounty TM, et al. 3-dimensional aortic annular assessment by multidetector computed tomography predicts moderate or severe paravalvular regurgitation after transcatheter aortic valve replacement: a multicenter retrospective analysis. J Am Coll Cardiol 2012;59:1287-94.
4. Boothroyd LJ, Spaziano M, Guertin JR, et al. Transcatheter aortic valve implantation: recommendations for practice based on a multidisciplinary review including cost-effectiveness and ethical and organizational issues. Can J Cardiol 2013;29:718-26.
14. Schultz CJ, Moelker A, Piazza N, et al. Three dimensional evaluation of the aortic annulus using multislice computer tomography: are manufacturer’s guidelines for sizing for percutaneous aortic valve replacement helpful? Eur Heart J 2010;31:849-56.
5. Webb J, Rodes-Cabau J, Fremes S, et al. Transcatheter aortic valve implantation: a Canadian Cardiovascular Society position statement. Can J Cardiol 2012;28:520-8. 6. Piazza N, De Jaegere P, Schultz C, et al. Anatomy of the aortic valvar complex and its implications for transcatheter implantation of the aortic valve. Circ Cardiovasc Interv 2008;1:74-81. 7. Tops LF, Wood DA, Delgado V, et al. Noninvasive evaluation of the aortic root with multislice computed tomography implications for transcatheter aortic valve replacement. JACC Cardiovasc Imaging 2008;1: 321-30. 8. Cueff C, Serfaty JM, Cimadevilla C, et al. Measurement of aortic valve calcification using multislice computed tomography: correlation with haemodynamic severity of aortic stenosis and clinical implication for patients with low ejection fraction. Heart 2011;97:721-6. 9. Messika-Zeitoun D, Serfaty JM, Brochet E, et al. Multimodal assessment of the aortic annulus diameter: implications for transcatheter aortic valve implantation. J Am Coll Cardiol 2010;55:186-94. 10. Blanke P, Reinohl J, Schlensak C, et al. Prosthesis oversizing in balloonexpandable transcatheter aortic valve implantation is associated with contained rupture of the aortic root. Circ Cardiovasc Interv 2012;5: 540-8. 11. Jabbour A, Ismail TF, Moat N, et al. Multimodality imaging in transcatheter aortic valve implantation and post-procedural aortic
15. Ewe SH, Ng AC, Schuijf JD, et al. Location and severity of aortic valve calcium and implications for aortic regurgitation after transcatheter aortic valve implantation. Am J Cardiol 2011;108:1470-7. 16. John D, Buellesfeld L, Yuecel S, et al. Correlation of device landing zone calcification and acute procedural success in patients undergoing transcatheter aortic valve implantations with the self-expanding CoreValve prosthesis. JACC Cardiovasc Interv 2010;3:233-43. 17. Gurvitch R, Webb JG, Yuan R, et al. Aortic annulus diameter determination by multidetector computed tomography: reproducibility, applicability, and implications for transcatheter aortic valve implantation. JACC Cardiovasc Interv 2011;4:1235-45. 18. Binder RK, Webb JG, Willson AB, et al. The impact of integration of a multidetector computed tomography annulus area sizing algorithm on outcomes of transcatheter aortic valve replacement: a prospective, multicenter, controlled trial. J Am Coll Cardiol 2013;62:431-8.
Supplementary Material To access the supplementary material accompanying this article, visit the online version of the Canadian Journal of Cardiology at www.onlinecjc.ca and at http://dx.doi.org/10. 1016/j.cjca.2014.06.012.
Clinical Research
Measurement of the Aortic Annulus Diameter Using Transesophageal Echocardiography and Multislice Computed TomographydAre They Truly Comparable? Jean-Michel Serfaty, MD, PhD,a,d,e Dominique Himbert, MD,b Marina Esposito-Farese, PhD,c Nicoletta Pasi, MD,a Vania Tacher, MD,a Jean-Pierre Laissy, MD, PhD,a Florence Tubach, MD, PhD,c Bernard Iung, MD,b,d Alec Vahanian, MD,b,d and David Messika-Zeitoun, MD, PhDb,d,e a b c
Department of Radiology, Assistance PubliqueeHôpitaux de Paris, Paris, France
Department of Cardiology, Assistance PubliqueeHôpitaux de Paris, Paris, France
Department of Biostatistics, Assistance PubliqueeHôpitaux de Paris, Paris, France d e
INSERM U1148, Paris, France
University Paris 7, Bichat Hospital, Paris, France
ABSTRACT
RESUM E
Background: For transcatheter aortic valve implantation (TAVI), transesophageal echocardiography (TEE) and multislice computed tomography (MSCT) measurements of the aortic annulus diameter (AAD) are often regarded as competitive. We evaluated if 1 MSCT method could be interchangeable with TEE measurements. Methods: We compared AAD measurements performed using TEE, MSCT, and transthoracic echocardiography (TTE) in 129 consecutive patients with severe aortic stenosis (AS) who were referred for TAVI. Using MSCT, AAD was measured in the 3-chamber (3C) view and at the level of the virtual basal ring (mean diameter [MD] of the long-axis [LA] and short-axis [SA] diameters, and AAD derived from the crosssectional area [CSA] and the circumference). Correlations with echocardiographic measurements and agreement regarding the TAVI strategy (decision to implant and choice of prosthesis size based on manufacturer cutoff recommendations) were assessed. Results: AAD measured in 3C (intraclass correlation [ICC], 0.79) and MD emphasizing the weight of the SA (MD4 [3 SA þ LA/4]; ICC, 0.76)
ter Introduction : Pour l’implantation valvulaire aortique par cathe (IVAC), les mesures du diamètre de l’anneau aortique (DAA) par chocardiographie transœsophagienne (ETO) et tomodensitome trie e re es comme concurrentes. multicoupe (TDM-MC) sont souvent conside value si une me thodes de TDM-MC pouvait être interNous avons e changeable aux mesures de l’ETO. thodes : Nous avons compare les mesures du DAA re alise es à l’aide Me chocardiographie transthoracique (ETT) de l’ETO, de la TDM-MC et de l’e cutifs souffrant d’une ste nose aortique (SA) grave chez 129 patients conse taient oriente s pour subir une IVAC. À l’aide de la TDM-MC, le DAA a qui e te mesure sur la coupe des 3 cavite s (3C) et au niveau de l’anneau basal e virtuel (diamètre moyen [DM] des diamètres du grand axe [GA] et du petit rive de la coupe transversale [CT] et de la ciraxe [PA], et DAA de rence). Les corre lations avec les mesures e chocardiographiques et confe gie d’IVAC (de cision quant à l’imla concordance concernant la strate plantation et le choix de la taille de la prothèse selon les seuils s par le fabricant) ont e te e value es. recommande
Transcatheter aortic valve implantation (TAVI) is now considered a valuable alternative to conventional surgery for high-risk patients with severe symptomatic aortic stenosis (AS).1-5 Accurate measurements of the aortic annulus diameter (AAD) are crucial for patient selection and procedural
success. Transesophageal echocardiography (TEE), long considered the reference method, has recently been challenged by multislice computed tomography (MSCT). In the absence of randomized studies, the method that should be used remains a matter of debate, and in routine practice both TEE and MSCT are often performed, leaving the clinician facing 2 different measurements. Instead of opposing the 2 methods, we wondered if 1 MSCT method could correspond or be interchangeable with TEE measurements. The aortic annulus (AA) is a complex 3-dimensional crown-like structure with an oval shape.6,7 MSCT measurements can be performed in multiple planes, and AAD can also
Received for publication December 9, 2013. Accepted June 20, 2014. Corresponding author: Dr Messika-Zeitoun, Department of Cardiology, Assistance PubliqueeHôpitaux de Paris, Bichat Hospital, 46 rue Henri Huchard, 75018 Paris, France. Tel.: 01 40 25 86 61; fax: 01 40 25 67 32. E-mail: [email protected] See page 1078 for disclosure information.
http://dx.doi.org/10.1016/j.cjca.2014.06.012 0828-282X/Ó 2014 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved.
1074
Canadian Journal of Cardiology Volume 30 2014
and MD5 [4 SA þ LA/5; ICC, 0.75]) provided the highest correlation and the best agreement with TEE (kappa ¼ 0.47, 0.27, and 0.31 respectively). However, TTE provided a better a correlation and agreement with TEE than all MSCT methods (ICC, 0.87; kappa ¼ 0.66). The agreement between MSCT and TEE varied with AAD eccentricity and degree of aortic valve calcification (AVC), but in all subsets, values observed with MSCT never reached those observed with TTE. Conclusions: MSCT and TEE are measuring different landmarks and consequently MSCT and TEE measurements are not interchangeable. Prospective randomized studies aimed at defining which method provides the best clinical results are clearly needed.
sultats : Le DAA mesure sur la 3C (coefficient de corre lation Re intraclasse [CCI], 0,79), le DM renfonçant le poids du PA (DM4 [3 PA þ GA/4]; CCI, 0,76) et le DM5 [4 PA þ GA/5; CCI, 0,75]) offraient la lation la plus e leve e et la meilleure concordance avec l’ETO corre (kappa ¼ 0,47, 0,27 et 0,31, respectivement). Cependant, l’ETT offrait lation et une meilleure concordance avec l’ETO que une meilleure corre thodes de TDM-MC (CCI, 0,87; kappa ¼ 0,66). La toutes les me du concordance entre la TDM-MC et l’ETO variait avec l’excentricite de calcification de la valve aortique (CVA), mais dans DAA et le degre es avec la TDM-MC n’attous les sous-groupes, les valeurs observe es avec l’ETT. teignaient jamais celles observe Conclusions : La TDM-MC et l’ETO mesurent des points de repère rents et par conse quent les mesures de la TDM-MC et de l’ETO ne diffe tudes prospectives randomise es sont pas interchangeables. Des e terminer quelle me thode fournit les meilleurs ayant pour but de de sultats cliniques sont ne cessaires. re
be derived from multiple measurements (diameter, area, and circumference). In addition, because the AA is oval with a long axis (LA) and a short axis (SA), mathematical transformations emphasizing the weight of the SA may provide closer agreement with TEE than direct measurements of the AAD. Thus, in the present study, we aimed to compare in a series of patients referred for TAVI, direct AAD measurements, AAD measurements derived from circumference/area, and mathematical transformation of direct AAD measurements obtained using MSCT-to-TEE measurements.
Amsterdam, the Netherlands). A scan run consisted of a prospective acquisition of 43-mm-thick contiguous transverse slices. Acquisition time was 0.5 sec/slice, electrocardiographically triggered at 75% of the R-R interval. Degree of aortic valve calcification (AVC) was quantified using the Agatston method and expressed in arbitrary units. Two MSCT runs were performed sequentially with a 1- or 2-mm initial interval. Each run was independently scored and the 2 scores were averaged.8
Methods
Multislice computed tomography Scanning. All examinations were performed using a 64-detector computed tomography scanner (Lightspeed VCTXT, GE Healthcare, Milwaukee, WI) and intravenous injection of 90 mL iobitridol 350 mg/mL. Acquisition was centred at the 75% phase of the R-R cardiac cycle when the heart rate was < 65 beats/min and at 40% when the heart rate was > 65 beats/min to ensure minimum motion artifacts. No bblockers were administered.
Study population Consecutive patients with severe AS (aortic valve area < 1 cm2 or mean gradient 40 mm Hg, or both) who were referred for TAVI and underwent TTE, TEE, and MSCT at our institution within 1 month were enrolled in the present study. All examinations were clinically indicated as workup for TAVI. Exclusion criteria included patients with bicuspid aortic valves and suboptimal TEE or MSCT images precluding accurate AAD measurements. Patients with atrial fibrillation were not excluded. The study was approved by our local ethics committee, and all patients gave written informed consent. Two-dimensional echocardiography Comprehensive transthoracic echocardiography (TTE) and TEE were performed by experienced echocardiographers using high-quality commercially available ultrasound systems (IE33; Philips Medical Systems, Cleveland, OH and Vivid 7; General Electric, Vingmed, Horten, Norway). AAD was measured using the zoom mode at the insertion of the leaflets in midsystole from the parasternal LA view in TTE or from the 120 -140 LA view (3-chamber [3C] view) in TEE. Measurements were averaged from 3-5 beats. Evaluation of AS severity was based on mean transaortic gradient and the aortic valve area calculated using the continuity equation. Calcium scoring scanning MSCT was performed using a 16-detector computed tomography scanner (MX 8000 IDT 16; Philips Healthcare,
Image analysis. As previously described,9 we used a double oblique multiplanar reconstruction to obtain 1 slice perpendicular to the aortic root including the 3 basal points of attachment of the annulus (transverse view corresponding to the virtual basal ring) and 1 slice including the aortic root, the left ventricular outflow tract, and the left atrium and ventricle (3C view), similar to the LA echocardiographic plane. On the 3C view, we measured the distance between the hinge points of the leaflets. On the transverse view, we measured SA and LA diameters of the annulus, the cross-sectional area (CSA), and the circumference (C). From these measurements, we calculated the mean diameter (MD ¼ [SA þ LA]/2), the diameter derived from the CSA (CSAeD ¼ O(4 * CSA/p)) and the diameter derived from the circumference (CirceD ¼ C/p) considering the AA to be circular. Because the AA is oval and the SA underestimates the AAD measured using TEE, whereas the LA and MD tend to overestimate it, we assessed a range of modified MD calculations enhancing the weight of SA: MD2 ¼ (2 * SA þ LA)/3, MD3 ¼ (3 * SA þ LA)/4, MD4 ¼ (4 * SA þ LA)/5, MD5 ¼ (5 * SA þ LA)/6. The eccentricity index (EI) was calculated as follows: EI ¼ 1 e SA/LA. MSCT
Serfaty et al. Aortic Annulus Diameter in TAVI
measurements were performed blindly from echocardiographic measurements. Statistical analysis Continuous variables are expressed as mean standard deviation. Comparisons between TEE and MSCT or TTE measurements were assessed as described by Bland-Altman analysis and single-measure ICC. Interobserver and intraobserver variability of AAD measurements was calculated in 30 patients as absolute difference standard deviation. MSCT variability required new reconstructions, whereas echocardiographic measurements required new acquisitions. The Edwards-Sapien valve (Edwards Lifesciences, Irvine, CA) is currently available in 3 sizes: 23, 26, and 29 mm. According to the manufacturer’s recommendations, a 23-mm prosthesis should be implanted if the annulus is > 18 mm and 21 mm, a 26-mm prosthesis should be implanted if the annulus diameter is > 21 and 25 mm, and a 29-mm prosthesis should be implanted if the annulus diameter is > 25 and 27 mm; the procedure should not be performed if the annulus diameter is 18 or > 27 mm. A contingency analysis was performed to assess the agreement between the different imaging modalities regarding the decision to implant a prosthesis and the choice of prosthesis size expressed by the kappa value. P < 0.05 was considered statistically significant. Results Population One hundred twenty-nine patients who underwent TTE, TEE, and MSCT in our hospital within 1 month with imaging quality compatible with accurate AA analysis constituted our study population. Mean age was 80 10 years, 69 patients (54%) were men, and 19% had atrial fibrillation. Mean aortic valve area was 0.68 0.16 cm2 and mean gradient was 51 15 mm Hg. Mean left ventricular ejection fraction was 49% 15% and < 50% in 48 patients (41%). Mean logistic euroSCORE and Society of Thoracic Surgeons score were 28% 15% and 15% 9% respectively. Comparison between echocardiographic and MSCT measurements Mean AAD was 23.4 2.0 mm using TEE and 23.0 2.0 mm using TTE. TTE and TEE were closely correlated (ICC, 0.87; 95% confidence interval [CI] 0.83-0.91), and the quality control plots using the Altman-Bland method showed that there was no trend for underestimation or overestimation (mean difference, 0.3 1.0 mm) (Table 1 and Fig. 1). The absolute difference between methods was 0.72 0.70 mm. MSCT measurements are reported in Table 1. The 3C method was also well correlated with TEE (ICC, 0.79; 95% CI, 0.73-0.83) with no trend for underestimation or overestimation (mean difference, 0 1.5 mm). Compared with TEE measurements, SA was significantly smaller and LA was significantly larger. The MD provided significantly larger AAD measurements (mean difference, 1.5 1.5 mm) than TEE, with an ICC lower than the 3C method (ICC, 0.62; 95% CI, 0.53-0.69). Mean difference decreased from MD to MD5 and was null for MD4. With this latter method, the
1075
ICC was also good (ICC, 0.76; 95% CI, 0.69-0.81). AAD derived from the CSA was approximately 1 mm larger than with TEE, whereas AAD derived from the circumference was markedly larger (mean difference, 7.0 1.6 mm). Thus, the 3C and MD4 methods provided the better correlations with TEE measurements, with no trend for underestimation or overestimation. The worst intraobserver and interobserver variability was observed with the 3C and circumference methods (Table 1). Agreement between methods regarding the TAVI strategy We evaluated the theoretical impact of the method of AAD measurement on TAVI strategy (choice of prosthesis size and decision to implant) based on cutoffs proposed by the manufacturer using TEE as the reference method (Table 2). TTE provided the best agreement with TEE (concordance of TAVI strategy in 83% of patients; kappa, 0.66), and all MSCT methods provided a lower agreement with kappa values between 0 and 0.47. The best computed tomographic methods were 3C and MD5. The number of patients in whom the prosthesis size and decision to implant changed is presented in Table 2. We further evaluated whether the agreement between TEE and MSCT was influenced by the degree of eccentricity of the AA (expressed as the EI) or the degree of AVC (calcium score measured using MSCT; available in 91 patients). The mean EI was 0.19 0.07 (range, 0-0.34) and the mean calcium score was 3351 1598 (range, 759 to 9030). The kappa values varied according to tertile of EI or calcium scoring but never reached the values observed with TTE (Supplemental Table S1). Combination of TTE and MSCT We then explored whether a combination of TTE and MSCT could improve the agreement with TEE (Supplemental Table S2). The 3C, MD4, and MD5, which provided the best agreement with TEE, were evaluated. The population was divided into 2 groups according to the absolute difference between TTE and MSCT (> 1 mm or 1 mm). Agreement between the 3 MSCT methods and TEE was better in the subgroup of patients in whom the difference between MSCT and TTE measurements was 1 mm than in the subgroup with a > 1 mm difference (kappa, 0.50-0.67 vs 0.11-0.33) but remained lower than the agreement between TTE and TEE (kappa, 0.63-0.78). Discussion In the present study, in a cohort of patients with severe AS referred for TAVI, we showed that (1) no direct or indirect MSCT measurements were more closely correlated with TEE than TTE and, consequently, (2) all MSCT methods had a lower agreement with TEE than TTE regarding the TAVI strategy, (3) the degree of agreement between TEE and MSCT was influenced by the degree of eccentricity or the degree of AVC but remained lower than with TTE in all subgroups, and (4) combining TTE and MSCT did not provide a better agreement with TEE than did the single use of TTE.
1076
Canadian Journal of Cardiology Volume 30 2014
Table 1. Comparison between echocardiographic and MSCT measurements using transesophageal echocardiography as reference
Method TEE TTE MSCT-3C MSCT-SA MSCT-LA MSCT-MD MSCT-MD2 MSCT-MD3 MSCT-MD4 MSCT-MD5 MSCT-CSA-D MSCT-Circ-D
Mean diameter 23.4 23.0 23.4 22.3 27.6 25.0 24.1 23.6 23.4 23.2 24.6 29.4
2.0 2.0 2.6 2.6 2.9 2.5 2.4 2.5 2.5 2.5 2.3 2.4
Difference Mean e 0.3 0 1.1 4.2 0.5 0.7 0.2 0 0.2 1.2 7.0
1.0 1.5 1.8 2 1.5 1.5 1.5 1.6 1.6 1.6 1.6
Absolute 0.72 1.21 1.51 4.24 1.83 1.40 1.30 1.30 1.29 1.58 6.01
e
0.70 0.91 1.36 1.95 1.17 0.86 0.86 0.87 0.91 1.04 1.54
ICC (Lower-higher 95% CI) 0.87 0.79 0.64 0.28 0.62 0.74 0.76 0.76 0.75 0.62 0.17
e (0.83-0.91) (0.73-0.83) (0.53-0.73) (0.22-0.33) (0.53-0.69) (0.68-0.79) (0.70-0.82) (0.69-0.81) (0.67-0.81) (0.52-0.71) (0.13-0.2)
Variability Intraobserver 0.60 0.60 0.96 0.84 0.91 0.60 0.64 0.66 0.68 0.71 0.62 0.90
0.50 0.80 0.92 0.68 0.72 0.63 0.62 0.64 0.65 0.65 0.43 0.75
Interobserver
NA
0 4 2 0 0 0 0 0 0 0 0 3
0.60 1.00 1.48 0.33 0.94 0.67 0.77 0.84 0.88 0.91 0.67 2.05
1.00 0.70 1.15 1.51 1.17 0.61 0.64 0.71 0.78 0.83 0.54 0.82
Results are presented as mean value standard deviation. 3C, three-chamber; CircD, diameter derived from the annulus circumference; CSA-D, diameter derived from aortic cross-sectional area; ICC, intraclass correlation coefficient; LA, long axis; MD, mean diameter; MSCT, multislice computed tomography; NA, measurements not available for analysis; SA, short axis; TTE, transthoracic echocardiography; TEE, transesophageal echocardiography.
In contrast to surgical aortic valve replacement, sizing in TAVI is based on imaging. Accurate measurements of the AA size are thus crucial for procedural success and to minimize complications. Undersizing may lead to paravalvular regurgitation or prosthesis migration, whereas oversizing may lead to annulus rupture.10 Current recommendations for valve sizing
have been established by and for echocardiography, and despite limitations, TEE was first considered as the reference method before being challenged by MSCT.11-13 The AAD can be measured using multiple MSCT methods. The 3C method mimics the echocardiographic LA view, and consequently this method provided the highest
Figure 1. Quality control plots using Bland-Altman analysis compared with transesophageal echocardiography (TEE): (A) Transthoracic echocardiography (TTE), (B) 3-chamber (3C), (C) mean of the long-axis (LA) and short-axis (SA) mean diameter (MD), (D) mean of the LA and SA diameter overpowered for the SA (MD4), (E) diameter derived from the cross-sectional area of the virtual basal ring (CSA-D), (F) diameter derived from the circumference of the virtual basal ring (CircD). The middle line represents the mean and the upper and lower lines 2 standard deviations.
Serfaty et al. Aortic Annulus Diameter in TAVI
1077
Table 2. Theoretical impact of the method of annulus diameter measurements on TAVI strategy (decision to perform the procedure and choice of the prosthesis size) using transesophageal echocardiography as reference TAVI strategy Method TEE TTE MSCT-3C MSCT-SA MSCT-LA MSCT-MD MSCT-MD2 MSCT-MD3 MSCT-MD4 MSCT-MD5 MSCT-CSA-D MSCTeCirc-D
Agreement with TEE
23-mm prosthesis
26-mm prosthesis
29-mm prosthesis
No implantation
n (%)
28 25 23 39 0 8 17 17 18 20 10 0
78 86 69 65 28 57 68 72 77 77 64 2
20 11 24 15 30 37 28 24 19 19 35 17
3 3 11 10 71 27 16 16 15 13 20 107
129 103 (83) 86 (68) 80 (62) 16 (12) 52 (40) 68 (53) 68 (53) 74 (57) 77 (60) 63 (49) 9 (7)
No agreement with TEE
Kappa
Smaller prosthesis size
Larger prosthesis size
New implantation
No implantation
e 0.66 0.47 0.38 0 0.11 0.23 0.21 0.27 0.31 0.18 0
e 13 9 28 0 1 7 10 11 13 2 0
e 7 20 10 45 50 37 34 28 25 45 13
e 1 2 2 0 1 2 2 2 2 1 0
e 1 10 9 68 25 15 15 14 12 18 104
Data are presented as number of patients (percent). 3C, three-chamber; CircD, diameter derived from the annulus circumference; ICC, intraclass correlation coefficient; LA, long axis; MD, mean diameter; MSCT, multislice computed tomography; NA, measurements not available for analysis; SA, short axis; SurfD, diameter derived from the annulus area; TAVI, transcatheter aortic valve transplantation; TTE, transthoracic echocardiography; TEE, transesophageal echocardiography.
correlation with TEE measurements and the highest agreement regarding the TAVI strategy. Nevertheless, the agreement between the 3C method and TEE remained lower (kappa, 0.47) than the agreement observed with TTE (kappa, 0.66). In addition, the 3C method was hindered by modest intraobserver and interobserver variability. The AA is oval rather than circular.7,9 Because the SA systematically underestimates and the LA systematically overestimates the AAD measured using TEE, the weight of the SA was progressively overpowered using mathematical transformations (MD2-MD5). These mathematical transformations were associated with an improvement of the correlations between MSCT and TEE measurements and a parallel improvement in the agreement between these 2 methods regarding the TAVI strategy. However, the kappa values remained in a modest range (between 0.23 and 0.31). The AA is a complex crown-like structure, and TEE measurements are performed between the basal attachments of 1 leaflet to the opposing point across the lumen toward the intercommissural triangle (Fig. 2) and are, therefore, also dependent on the diameter of the aortic root. In contrast, SA and LA are measured at the level of the so-called virtual basal ring. Thus, MSCT and TEE definitively measure diameters located in completely different 3-dimensional planes. We have previously shown that the use of the mean diameter instead of TEE measurements significantly changes the TAVI strategy.9 We extend these findings to mathematical transformation of MSCT emphasizing the weight of the SA. It is worth noting that the agreement between TEE and MSCT methods was only mildly influenced by the EI, reinforcing the fact that the intrinsic anatomical differences in the diameters measured by both methods cannot be compensated by any mathematical formula. We also evaluated diameters derived from the CSA and the annulus circumference of the virtual basal ring. Diameter derived from the CSA was significantly larger than TEE measurements. In our study, the difference between both methods (1.2 1.6 mm) was very similar to that reported in a recent article (1.2 1.7 mm).10 The ICC between both methods was correct (ICC, 0.62), but the agreement was poor (kappa, 0.18). The larger diameter observed with this MSCT
method translated into the use of a larger prosthesis in one third of the population and cancellation of the procedure in 15%. One may hypothesize that if the AA becomes circular after TAVI implantation with CSA changes, whereas the circumference may remain unchanged, use of diameter measurements derived from the circumference may be the ideal method. In the present study, diameters derived from circumference were dramatically larger than TEE measurements (7.0 1.6 mm), with a poor ICC and poor agreement with TEE. Interestingly, use of the AAD derived from the circumference would have led to cancellation of the procedure in more than 80% of the population. A lower agreement using diameters derived from the circumference has been previously reported.14 Therefore, this method seems inadequate for selection of patients for TAVI. The degree of AVC has been associated with the rate of procedural success and more specifically the incidence of paravalvular regurgitation.15,16 In the present study, the agreement between MSCT and TEE was influenced by the degree of AVC but remained lower than that between TTE and TEE. In addition, leaflet and annular calcifications are often asymmetrical, which may add difficulty to any method for accurately assessing the shape and size of the AA. Finally, we also evaluated whether a combination of TTE and MSCT could avoid the performance of TEE. A difference < 1 mm between TTE and MSCT was observed in half of the patients in whom such a strategy could have been an alternative. Unfortunately, MSCT and TTE were not synergistic enough, and even in this subset, MSCT did not perform better than TTE alone. The results of the present study may have important clinical implications. Because TEE measurements are often smaller than most MSCT measurements, some authors have suggested systematically shifting from the 1/1.5-mm cutoff values provided by manufacturers.17 The present clinical evaluation clearly demonstrates that MSCT measurements are not interchangeable, even using mathematical formulas. Thus, TEE and MSCT are intrinsically different, and because they measure different landmarks, the question of TAVI sizing from a clinical point of view should not be asked in terms of “which method is the true one” but as “which one provides
1078
Canadian Journal of Cardiology Volume 30 2014
Figure 2. Schematic representation of the aortic annulus and the diameters measured using transesophageal echocardiography (red) and multislice computed tomography (long and short axis in yellow) at the level of the virtual basal ring (blue). Transesophageal diameter is in a different plane than the virtual basal ring.
the best clinical results?” Some centres are already considering CSA-D in clinical practice, and retrospective analyses suggest a lower paravalvular aortic regurgitation rate using MSCT measurements,12,13 which may be associated with a better outcome. In our centre, echocardiographic and MSCT measurements are considered complementary and are integrated with clinical data for final sizing decisions. Dedicated strategies based on MSCT measurements should be developed and results of such strategies compared with results of TEE-based strategies including end points such as rate of aortic regurgitation and annulus rupture as well as the number of patients who would be offered TAVI. In a recent study, a MSCT algorithm was associated with a lower rate of paravalvular aortic regurgitation, but in as many as 20% of patients, MSCT measurements were finally disregarded.18 In our opinion, these results reinforce the idea that TEE and MSCT measurements should not be systematically opposed, but because they are measuring different landmarks, they should be regarded as complementary, and our results as well as those of the previously mentioned study, are a strong argument for an integrative approach. The present study deserves several comments. First, we did not aim to define whether MSCT was better or worse than TEE regarding the rate of procedural success, which would ideally require a randomized study on TEE- vs MSCT strategye based implantation. There is no gold standard method for AAD measurements and we “only” performed comparisons between methods. Second, the agreement between methods was assessed based on cutoff values used for the balloonexpandable Edwards-Sapien prosthesis. However, despite different annulus diameter ranges and cutoffs, as well as different prosthesis sizes with the CoreValve Revalving System (Medtronic, Minneapolis, MN), our conclusions with
modest agreement between TEE and computed tomography are not dependent on the type of device. Computed tomographic, and echocardiographic measurements were not exactly performed at the time (end-systole/mid-diastole vs midsystole, respectively). However, variations of the annulus size during the cardiac cycle are limited,7 especially in patients with severe AS with severe calcifications. Finally, this was not a feasibility study, and patients with suboptimal echocardiographic or MSCT examinations were excluded. Conclusions In this series of patients with severe symptomatic AS who were referred for TAVI and underwent both TEE and MSCT at our institution, we demonstrated using multiple MSCT methods that MSCT and TEE measurements are not interchangeable. Because MSCT and TEE are intrinsically different and measure different landmarks, they should not be regarded as competitive but more probably as complementary. Prospective randomized studies aiming at defining which method provides the best clinical results are clearly needed and should compare not only each method separately but also an integrative approach based on both measurements. Funding Sources The present work was partially funded by the COFRASA and GENERAC studies (Assistance PubliqueeHôpitaux de Paris; PHRC National 2005 and PHRC regional 2007, respectively). Disclosures See Supplementary Material for disclosures.
Serfaty et al. Aortic Annulus Diameter in TAVI
References 1. Vahanian A, Alfieri O, Andreotti F, et al. Guidelines on the management of valvular heart disease (version 2012): The Joint Task Force on the Management of Valvular Heart Disease of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS). Eur Heart J 2012;33:2451-96. 2. Leon MB, Smith CR, Mack M, et al. Transcatheter aortic-valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med 2010;363:1597-607.
1079
regurgitation: comparison among cardiovascular magnetic resonance, cardiac computed tomography, and echocardiography. J Am Coll Cardiol 2011;58:2165-73. 12. Jilaihawi H, Kashif M, Fontana G, et al. Cross-sectional computed tomographic assessment improves accuracy of aortic annular sizing for transcatheter aortic valve replacement and reduces the incidence of paravalvular aortic regurgitation. J Am Coll Cardiol 2012;59:1275-86.
3. Smith CR, Leon MB, Mack MJ, et al. Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med 2011;364: 2187-98.
13. Willson AB, Webb JG, Labounty TM, et al. 3-dimensional aortic annular assessment by multidetector computed tomography predicts moderate or severe paravalvular regurgitation after transcatheter aortic valve replacement: a multicenter retrospective analysis. J Am Coll Cardiol 2012;59:1287-94.
4. Boothroyd LJ, Spaziano M, Guertin JR, et al. Transcatheter aortic valve implantation: recommendations for practice based on a multidisciplinary review including cost-effectiveness and ethical and organizational issues. Can J Cardiol 2013;29:718-26.
14. Schultz CJ, Moelker A, Piazza N, et al. Three dimensional evaluation of the aortic annulus using multislice computer tomography: are manufacturer’s guidelines for sizing for percutaneous aortic valve replacement helpful? Eur Heart J 2010;31:849-56.
5. Webb J, Rodes-Cabau J, Fremes S, et al. Transcatheter aortic valve implantation: a Canadian Cardiovascular Society position statement. Can J Cardiol 2012;28:520-8. 6. Piazza N, De Jaegere P, Schultz C, et al. Anatomy of the aortic valvar complex and its implications for transcatheter implantation of the aortic valve. Circ Cardiovasc Interv 2008;1:74-81. 7. Tops LF, Wood DA, Delgado V, et al. Noninvasive evaluation of the aortic root with multislice computed tomography implications for transcatheter aortic valve replacement. JACC Cardiovasc Imaging 2008;1: 321-30. 8. Cueff C, Serfaty JM, Cimadevilla C, et al. Measurement of aortic valve calcification using multislice computed tomography: correlation with haemodynamic severity of aortic stenosis and clinical implication for patients with low ejection fraction. Heart 2011;97:721-6. 9. Messika-Zeitoun D, Serfaty JM, Brochet E, et al. Multimodal assessment of the aortic annulus diameter: implications for transcatheter aortic valve implantation. J Am Coll Cardiol 2010;55:186-94. 10. Blanke P, Reinohl J, Schlensak C, et al. Prosthesis oversizing in balloonexpandable transcatheter aortic valve implantation is associated with contained rupture of the aortic root. Circ Cardiovasc Interv 2012;5: 540-8. 11. Jabbour A, Ismail TF, Moat N, et al. Multimodality imaging in transcatheter aortic valve implantation and post-procedural aortic
15. Ewe SH, Ng AC, Schuijf JD, et al. Location and severity of aortic valve calcium and implications for aortic regurgitation after transcatheter aortic valve implantation. Am J Cardiol 2011;108:1470-7. 16. John D, Buellesfeld L, Yuecel S, et al. Correlation of device landing zone calcification and acute procedural success in patients undergoing transcatheter aortic valve implantations with the self-expanding CoreValve prosthesis. JACC Cardiovasc Interv 2010;3:233-43. 17. Gurvitch R, Webb JG, Yuan R, et al. Aortic annulus diameter determination by multidetector computed tomography: reproducibility, applicability, and implications for transcatheter aortic valve implantation. JACC Cardiovasc Interv 2011;4:1235-45. 18. Binder RK, Webb JG, Willson AB, et al. The impact of integration of a multidetector computed tomography annulus area sizing algorithm on outcomes of transcatheter aortic valve replacement: a prospective, multicenter, controlled trial. J Am Coll Cardiol 2013;62:431-8.
Supplementary Material To access the supplementary material accompanying this article, visit the online version of the Canadian Journal of Cardiology at www.onlinecjc.ca and at http://dx.doi.org/10. 1016/j.cjca.2014.06.012.
