Feasibility of a High-Pitch Protocol for Transcatheter Aortic Valve Replacement Evaluation in Patients not Suitable for ECG-Gated CT

High-Pitch CT for TAVR Evaluation in Patients not Suitable for ECG-Gating The use of transcatheter aortic valve replacement (TAVR) has been established to be non-inferior to surgical aortic valve replacement (SAVR). As a result, in the United States, more patients now undergo TAVR than SAVR. It is recommended that preprocedural CT imaging for aortic valve evaluation and optimal sizing should include an ECG-gated scan of, at least, the aortic root. However, many patients suffer from concomitant tachyarrhythmias such as atrial brillation, which may seriously degrade the diagnostic accuracy of ECG-gated scans. The aim of the present study is to explore whether a high-pitch non-ECG-gated computed tomography angiography (CTA) of the entire aorta can render similar diagnostic preprocedural images as a standard ECG-gated scan. 108 patients were included. Objective image quality parameters such as image noise, CNR and SNR as well as subjective image quality analysis by two different readers were compared. The results showed a signicant increase in image noise at the level of the aortic root with use of the high-pitch protocol (p = 0.001). Otherwise, our study revealed no signicant differences in subjective and objective image quality. Diagnostic image quality was achieved in all patients without a record of inaccurate sizing in the surgical reports or subsequent patient histories.


Introduction
Aortic stenosis (AS) is the most common acquired valve defect 1 . Technological enhancements and procedural simpli cation have increased the use of Transcatheter Aortic Valve Replacement (TAVR) to treat AS. As a result, in the United States, more patients now undergo TAVR than isolated surgical aortic valve replacement 2 . Candidates for this intervention undergo preprocedural contrast-enhanced computed tomography (CT) imaging for annular sizing to guide device selection and assessment of the peripheral access route 3,4 .
The latest recommendations for preprocedural CT issued by the Society of Cardiovascular Computed Tomography (SCCT) suggest that, at least, the aortic root should be imaged with electrocardiogram (ECG)-gated scans to limit motion artifacts 5 . Image quality is best in patients with a slow and regular sinus rhythm 6 . However, more than 9% of patients suffering from AS have concomitant atrial brillation (AF) 7 , which is characterized by a high heart rate (HR) and HR variability 8 . This and other cardiac arrhythmias (CA) may result in signi cant artifacts 9,10 and impair diagnostic performance of CT imaging 11 . Beta-blockers to regulate heartbeat are, however, not recommended in AS because there is concern that antihypertensive treatment may result in hypotension and hemodynamic collapse 9,12 .
In patients with CA, current SCCT guidelines suggest ECG-editing for retrospective ECG-gating to reduce artifacts in these patients 5 . Otherwise, there are no further or speci c recommendations concerning this particular group of patients. In patients with severe tachyarrhythmias, ECG-gating is, however, often not possible. Still, these patients require preprocedural imaging to qualify for TAVR. Thus, the aim of this study was to evaluate image quality and diagnostic adequacy of preprocedural TAVR CT in patients in whom ECG-gating is not possible by using a high-pitch, non-ECG-gated protocol compared with a protocol using prospective ECG-gating.

Patient Characteristics
Patient characteristics are summarized in Table 1. There was no signi cant (n. s.) difference between the two groups with respect to BMI (p = 0.508), sex (p = 0.553), eGFR (p = 0.648), volume of CM administered (p = 0.707), and distribution of CM volumes administered (p = 0.785). Heart rate (HR) was signi cantly faster in Group B with 80.3 ± 17.1 bpm compared to 68.3 ± 11.0 bpm in Group A (p = < 0.001).
Furthermore, signi cantly more patients in Group B had a medical record of chronic CA with 41/56 compared to 11/52 in Group A (p < 0.001). In the patients with CA, persistent AF was by far the most common in both groups with 6/52 in Group A and 19/56 in Group B. Patients in Group B were signi cantly older by a mean of 3.5 years (p = 0.009). Scan time was signi cantly shorter in Group B with 4.5 ± 0.3 s compared to 13.9 ± 1.25 s (p = < 0.001) in Group A. Even though median kilovoltage (kV) was 100 kV in both groups, the median test showed a statistically signi cant difference (p = < 0.001) in kV between both groups. This is explained by the difference in distribution (Group A: 100 kV (n = 51) and 120 kV (n = 1), Group B: 100 kV (n = 38), 120 kV (n = 12) and 135 kV (n = 6)).  Objective Image Analysis The objective image quality scores are summarized in Table 3. Image noise at the level of the aortic root was signi cantly increased in Group B (p = 0.001). Accordingly, SNR and CNR at the level of the aortic root were signi cantly decreased in Group B (p = 0.002 and 0.003). Furthermore, attenuation value for background muscle at the level of the abdominal aorta was signi cantly decreased in Group B (p < 0.001). There were no further statistically signi cant differences between both groups. The subjective image quality scores are summarized in Table 4. There were no statistically signi cant differences between both readers in either of the two groups. The κ w value indicated substantial interrater agreement for all categories. Diagnostic image quality was achieved in all patients without a record of inaccurate sizing in the surgical report or subsequent patient history. Exemplary images of the aortic root are shown in Fig. 1.

Discussion
Cardiac arrhythmias are a frequent phenomenon in patients suffering from AS and substantially degrade CT image quality as inconsistent R-R intervals may result in stair-step artifacts or extensive blurring within a single axial slab 9,16 , which may make it necessary to repeat the CT scan or examine the patient with alternative imaging modalities 17 . Some sources thus advise not to examine patients with AF on 16or 64-slice CT scanners, reporting that diagnostic image quality is frequently not achieved in these patients 9,17 . As a result, patients with severe CA such as tachyarrhythmias including AF and/or HR variability/HR higher than 65 bpm have often been excluded from research trials of cardiac CT imaging 10 . Therefore, we investigated the performance of a high-pitch non-ECG-gated protocol for preprocedural TAVR imaging in patients in whom ECG-gating was precluded based on the presence of severe CA at the time of CT planning compared to a group of patients with sinus rhythm imaged with prospective ECGgating.
The ndings of our study show that the non-ECG-gated high-pitch protocol is a feasible alternative to the ECG-gated protocol as both methods result in diagnostic images. This was achieved despite CA identi ed by ECG monitoring immediately before CT and a signi cantly higher proportion of patients with a medical record of chronic CA in the group examined with the non-ECG-gated high-pitch protocol. Nonetheless, the non-ECG-gated protocol does signi cantly increase image noise and decrease SNR and CNR in the aortic root. Notably, these changes in objective image quality parameters did not result in a signi cant difference in subjective image quality ratings for either reader. Furthermore, none of the examinations with either the high-pitch or ECG-gated protocol resulted in artifacts that signi cantly degraded assessment of the aortic root or aorta. Therefore, precise aortic annulus measurement was always possible in both groups with no report of inaccurate sizing in the surgeon's report or subsequent patient history. Moreover, the high-pitch protocol resulted in a signi cantly lower radiation exposure by 55% compared to the ECG-gated protocol (from 13.44 mSv to 6.07 mSv). Dose reduction may appear secondary as TAVR is currently generally performed in an elderly population 18 . Nevertheless, two recent clinical trials published in The New England Journal of Medicine in 2019 have established the noninferiority of TAVR compared to SAVR in low-risk patients. In these trials, TAVR had a clear early safety bene t over SAVR in low-risk patients and was associated with faster discharge from hospital, faster recovery and fewer rehospitalizations 2,19 . This has set the stage for a new wave of TAVR indications in a younger patient populace. Furthermore, since patients will already receive a signi cant radiation dose from procedural uoroscopy, the ALARA ("as-low-as-reasonably-achievable") principle should be taken seriously.
An unexpected result of our analysis was that attenuation of background muscle at the level of the abdominal aorta was signi cantly lower in Group B compared to Group A. This might be explained by the statistically signi cantly older age of patients in Group B compared to Group A.
In interpreting these data, some alternatives should be considered. First, as an alternative to prospective ECG-gating, which is generally prone to artifacts in arrhythmic patients at the borders of the acquired slabs, retrospective ECG-gated CT acquisition allows image reconstruction at numerous points within the R-R cycle using online or o ine ECG-editing tools offered by vendors of selected CT scanners 9,20 .
Second, third-generation CT scanners with wide scan coverage along the patient Z-axis and 16-cm detector coverage enable prospective ECG-gated scanning of the entire heart within a single heartbeat. Therefore, as scans that require image acquisition during more than one cardiac cycle are inherently sensitive to CA regardless of whether prospective or retrospective ECG-gating is used, a volume CT scanner covering the entire heart in an axial snapshot may yield more robust images while at the same time allowing further dose reduction 20 . While the rst alternative, namely retrospective ECG-gating, comes with an increase in radiation dose 21 , the latter requires CT scanners that are not readily available. Nonetheless, a study performed by Annoni et al. in 2018 showed such state-of-the-art CT hardware to reduce potential artifacts in preprocedural TAVR imaging, even in patients with AF (n = 15/115) 22 . However, the authors also acknowledged that their study population had a lower incidence of AF in comparison with previous studies, which might have in uenced their results 22 .
Our study has some limitations. Because our protocol acquired images without ECG-gating, the aortic root was imaged in a random phase of the cardiac cycle. However, multiple studies agree that the annulus is larger in systole, and thus annular measurement for TAVR is typically derived from midsystolic images 23 . Nonetheless, as previously mentioned, we found no evidence of inaccurate sizing in surgical reports or subsequent patient histories in either of the two groups investigated.
In conclusion, the ungated high-pitch CT protocol allows accurate preprocedural assessment for TAVR with excellent subjective image quality in a patient population suffering from CA. Therefore, it is a feasible and dose-saving alternative in patients not suitable for ECG-gated CT imaging.

Study Population
A total of 108 patients suffering from AS who were referred from our cardiology department for preprocedural CT for TAVR over a period of 21 months were considered for inclusion in this retrospective study. These patients were categorized into two groups. Group A (n = 52) included patients with sinus rhythm at the time of CT planning who were examined with a prospective ECG-gated protocol. Group B (n = 56) included patients whose ECG signal indicated substantial tachyarrhythmia at the time of CT planning who were examined using a high-pitch non-ECG-gated protocol.
The institutional review board of Charité -University Medicine Berlin (Charité Ethics Committee) approved this study (EA4/140/17) and waived informed consent due to the retrospective data acquisition. This study was conducted in accordance with local laws and regulations, the Declaration of Helsinki and Good Clinical Practice.

CT Protocol
No premedication to control heart rate was added to the patient's baseline medication before the CT scan.
All imaging was performed on an 80-detector-row CT scanner (Aquilion PRIME, Canon Medical Systems, Otawara, Japan) adapted from a previously used protocol 21 . Scan parameters were as follows: temporal resolution of 175 ms using half-scan reconstruction, automated tube voltage selection (min = 100 kV in Group A and Group B, max = 120 kV in Group A and 100 kV in Group B), automated tube current modulation based on two scanned projection radiographs (ATCM, min = 40 mA, max = 600 mA, with selected image noise of pixel values in the reconstructed image serving as image quality reference parameter, 0.5 mm thickness, 0.5 mm increment, 40 × 0.5 collimation, 0.35 s rotation time, 512 × 512 matrix). Axial images were reconstructed from raw data using Canon's integrated adaptive iterative dose reduction (AIDR-3D) reconstruction algorithm with 0.5 mm slice thickness.
In Group A, images were acquired with a non-ECG-gated scan of the upper thoracic aperture at a pitch of 0.813, followed by a prospective ECG-gated acquisition of the heart at a pitch of 0.267 and a subsequent non-ECG-gated abdominal/pelvic scan with a pitch of 0.813 reconstructed as one volume as previously described 21 . In Group B, a high pitch of 1.388 was set for the entire non-ECG-gated single acquisition of the entire aorta.
All patients were administered an intravenous contrast agent (CA) bolus of iomeprol (400 mg iodine/ml; Imeron-400 MCT, Bracco, Milan, Italy) followed by a 60-ml saline ush using an automatic power injector (Accutron CT-D, Medtron AG, Saarbrücken, Germany). The CA dose and administration rate were adjusted to the estimated glomerular ltration rate (eGFR) as follows: 60-ml CA bolus at a rate of 3.0 ml/s for After acquisition, Vital's Vitrea™ Advanced 6.2 TAVR software (Vital Images Inc., Minnetonka, USA) was used for TAVR evaluation including semiautomatic identi cation and measurement of the area of the aortic annulus valve plane.

Radiation Dose
Radiation dose exposure was estimated and compared using dose-length product (DLP) in mGy*cm, effective dose (E) in mSv, and size-speci c dose estimates (SSDE) in mGy. DLP was recorded from an automatically generated protocol, based on the CT dose index (CTDI). E was calculated from DLP according to the method and conversion coe cients (k) presented in the European Guidelines on Quality Criteria for Computed Tomography 13,14 : we used a k of 0.017, which is the mean of the k of the chest (0.017 mSv/mGy*cm), abdomen (0.015 mSv/mGy*cm), and pelvis (0.019 mSv/mGy*cm), and the following formula: E = k ⋅ DLP. SSDE was calculated, as described in another study 15 , by multiplying conversion coe cients as a function of the sum of the lateral and anteroposterior dimensions with CTDI.

Objective Image Analysis
For quanti cation of objective image quality, circular ROIs were placed in the aortic lumen and the closest adjacent muscle at two anatomical levels in axial images: 1) the aortic root and 2) the abdominal aorta just proximal to the aortic bifurcation. The following parameters were measured: a) CT attenuation number of the artery and b) image noise, de ned as the SD of the CT attenuation number of the artery.
The following parameters were calculated: c) signal-to-noise ratio (SNR), de ned as mean attenuation of the artery divided by image noise of the CT attenuation value of the artery and d) contrast-to-noise ratio (CNR), de ned as the difference between mean attenuation of the artery and mean attenuation of the closest adjacent muscle, divided by image noise of the CT attenuation value of the artery 13,24−27 . All aortic ROIs were drawn as large as possible while avoiding calci cations or metallic artifacts to exclude partial volume effects. Muscle ROIs were made the same size as the corresponding vessel ROIs.
Subjective Image Analysis was performed according to a previously published manuscript 21 . Two medical doctors with different levels of experience (rater 1: 15 years; rater 2: 3 years) rated image quality of the aortic root and the aortoiliac pathway with respect to the following features: 1) clear identi cation of the annulus plane; 2) clear depiction of valve lea ets; 3) arterial wall sharpness; and 4) conspicuity of arterial wall calci cations. Image quality was rated on a 4-point Likert scale (1: excellent, 2: good, 3: su cient, 4: poor) to avoid a midway option. Image datasets of both groups were blindly evaluated in random order using a hanging protocol on RA1000 PACS (GE Healthcare, Waukesha, USA) with a preset CT Angio window (W: 600 L: 300 HU) and 1-mm slice thickness. Raters only used axial images for reading but were allowed to change window settings.
Statistical Analysis All data were tested for normal distribution using the Shapiro-Wilk test. Differences in body mass index (BMI), heart rate (HR), and eGFR were tested for signi cance with an unpaired Student's t-test. To compare the distribution of male and female patients, the distribution of CM volume, and the distribution of patients with a record of chronic CA a chi-squared test (χ 2 ) was used. Differences in patient age, scan time, CM volume, radiation dose, artery attenuation, image noise, SNR, CNR, and subjective image quality scores between the two groups were tested for signi cance using the Mann-Whitney U-test. Differences in median kilovoltage (kV) were tested for signi cance using the median test. Interrater agreement of subjective image quality scores between the two readers was compared using Cohen's weighted kappa coe cient (κ w ). κ w was interpreted as follows: 0.00-0.20: poor, 0.21-0.40: slight, 0.41-0.60: moderate, 0.61-0.80: substantial, and 0.81-1.00: almost perfect agreement 28,29 . A p-value below 0.05 was considered statistically signi cant. Values are presented as mean ± SD unless speci ed otherwise. SPSS (SPSS® Mac, v. 20.0; IBM Corp., New York, NY) was used for all statistical analyses.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Author Contribution
Figures Figure 1 Sample images of the aortic valve of representative patients examined in Group A (left) and Group B (right). Even though objectively image noise was signi cantly increased in the aortic root in Group B, subjective image parameters showed similar excellent image quality. Furthermore, precise aortic annulus measurement was always possible in both groups with no report of inaccurate sizing in the surgeon's report or subsequent patient history.