Author + information
- Received June 25, 2013
- Revision received September 9, 2013
- Accepted October 3, 2013
- Published online February 1, 2014.
- Chad Kliger, MD∗,
- Vladimir Jelnin, MD∗,
- Sonnit Sharma, MD∗,
- Georgia Panagopoulos, PhD∗,
- Bryce N. Einhorn, BA†,
- Robert Kumar, MD∗,
- Francisco Cuesta, MD∗,
- Leandro Maranan, MD∗,
- Itzhak Kronzon, MD∗,
- Bart Carelsen, PhD‡,
- Howard Cohen, MD†,
- Gila Perk, MD∗,
- Rob Van Den Boomen, B Eng‡,
- Cherif Sahyoun, PhD‡ and
- Carlos E. Ruiz, MD, PhD∗∗ ()
- ∗Department of Cardiovascular Medicine, Division of Structural and Congenital Heart Disease, Lenox Hill Heart and Vascular Institute–North Shore/Long Island Jewish Health System, New York, New York
- †Department of Medicine, Temple University School of Medicine, Philadelphia, Pennsylvania
- ‡Clinical Science Division, Philips Healthcare, Best, the Netherlands
- ↵∗Reprint requests and correspondence:
Dr. Carlos E. Ruiz, Department of Cardiovascular Medicine, Division of Structural and Congenital Heart Disease, Lenox Hill Heart and Vascular Institute of New York, 130 East 77th Street, 9th Floor Black Hall, New York, New York 10021-10075.
Objectives The aim of this proof-of-principle study is to validate the accuracy of fusion imaging for percutaneous transapical access (TA).
Background Structural heart disease interventions, including TA, are commonly obtained under fluoroscopic guidance, which lacks important spatial information. Computed tomographic angiography (CTA)-fluoroscopy fusion imaging can provide the 3-dimensional information necessary for improved accuracy in planning and guidance of these interventions.
Methods Twenty consecutive patients scheduled for percutaneous left ventricular puncture and device closure using CTA-fluoroscopy fusion guidance were prospectively recruited. The HeartNavigator software (Philips Healthcare, Best, the Netherlands) was used to landmark the left ventricular epicardium for TA (planned puncture site [PPS]). The PPS landmark was compared with the position of the TA closure device on post-procedure CTA (actual puncture site). The distance between the PPS and actual puncture site was calculated from 2 fixed reference points (left main ostium and mitral prosthesis center) in 3 planes (x, y, and z). The distance from the left anterior descending artery at the same z-plane was also assessed. TA-related complications associated with fusion imaging were recorded.
Results The median (interquartile range [IQR]) TA distance difference between the PPS and actual puncture site from the referenced left main ostium and mitral prosthesis center was 5.00 mm (IQR: 1.98 to 12.64 mm) and 3.27 mm (IQR: 1.88 to 11.24 mm) in the x-plane, 4.48 mm (IQR: 1.98 to 13.08 mm) and 4.00 mm (IQR: 1.62 to 11.86 mm) in the y-plane, and 5.57 mm (IQR: 3.89 to 13.62 mm) and 4.96 mm (IQR: 1.92 to 11.76 mm) in the z-plane. The mean TA distance to the left anterior descending artery was 15.5 ± 7.8 mm and 22.7 ± 13.7 mm in the x- and y-planes. No TA-related complications were identified, including evidence of coronary artery laceration.
Conclusions With the use of CTA-fluoroscopy fusion imaging to guide TA, the actual puncture site can be approximated near the PPS. Moreover, fusion imaging can help maintain an adequate access distance from the left anterior descending artery, thereby, potentially reducing TA-related complications.
- computed tomographic angiography–fluoroscopy fusion
- fusion imaging
- percutaneous transapical access
- structural heart disease interventions
Percutaneous transapical access (TA) is being increasingly used for a multitude of structural heart interventions. These interventions include closure of ventricular septal defects, left ventricular (LV) pseudoaneurysms, and paravalvular leaks (PVL) and have been more recently used to aid in transcatheter mitral valve-in-valve implantation for degenerative bioprosthetic valves (1–4). TA provides a direct approach to the aortic valve and mitral valve apparatus and has been shown to significantly decrease procedural and fluoroscopy times for complex structural heart cases (1). To maintain TA safety and procedural success, it is important to direct the puncture away from lung parenchyma, epicardial coronary arteries, and papillary muscles and to align the entry with the intended cardiac defect (1,5,6).
TA has been most commonly obtained under fluoroscopic guidance. Nevertheless, fluoroscopy has poor characterization of soft tissue or nonradiopaque structures and provides only 2-dimensional (2D) projections of complex 3-dimensional (3D) anatomy; fluoroscopic images lack the spatial information necessary for complex interventional procedures. The integration of computed tomography angiography (CTA) with fluoroscopy in the catheterization laboratory, also known as CTA-fluoroscopy fusion imaging, now provides an alternative to conventional fluoroscopically guided catheterization (1,3,4,7–9). Fusion imaging has the potential to improve procedural efficacy and safety, while reducing radiation exposure, contrast volume requirements, and procedural duration (1). Importantly, it may provide for accurate TA planning and guidance. However, little data exists on the procedural outcomes assisted by fusion imaging. The aim of this proof-of-principle study is to validate the accuracy of TA performed using CTA-fluoroscopy fusion imaging for structural heart interventions.
Twenty consecutive patients scheduled to undergo percutaneous LV puncture and subsequent closure using CTA-fluoroscopy fusion guidance were prospectively recruited from June 1, 2011, to June 1, 2012 at Lenox Hill Heart and Vascular Institute/North Shore Long Island Jewish Health System. Immediate post-procedural CTA was performed. Written informed consent was obtained prior to the procedure. This study was approved by North Shore/Long Island Jewish Health System's institutional review board. All procedures were performed in a dedicated structural heart catheterization laboratory under general anesthesia.
Computed tomographic angiography–fluoroscopy fusion
Patients underwent cardiac CTA (256-slice iCT scanner, Philips Healthcare, Cleveland, Ohio) using helical scan mode with multiphase acquisition (16 phases, 6.25% RR interval increments) and retrospective electrocardiographic-gating. Nonionic contrast media injection of 60 to 90 ml at a rate of 5 to 6 ml/s was used, depending on patient size and renal function. The timing between contrast injection and the beginning of image acquisition was determined with the aim of peak contrast concentration in the LV. In 2 patients with renal insufficiency (glomerular filtration rate ≤30 ml/min), contrast was not administered on post-CTA follow-up.
The HeartNavigator system (Philips Healthcare, Best, the Netherlands) was the prototype proprietary software used for CTA-fluoroscopy fusion imaging. Pre-procedural CTA slices during end-diastole were 3D volume-rendered and automatically segmented to identify the left atrium, LV, and aorta. Segmentation of the coronary arteries, lungs, and ribs was performed manually. The images were then analyzed and a TA route determined such that entry into the LV was aligned with the cardiac defect and away from lung parenchyma, coronary arteries, and papillary muscles. The true LV apex is typically thin-walled and should be avoided; an adjacent site was chosen for puncture. Landmarks were placed accordingly for skin entry, epicardial left ventricular entry or planned puncture site (PPS), and the defect that required intervention (Figs. 1A to 1D).
The patient's position on the x-ray table was adjusted to achieve the isocenter of the heart in 2 orthogonal fluoroscopic views. The CT images were then coregistered and overlaid onto live fluoroscopy by using either internal markers, such as a prosthetic valve frame, in 2 different views performed 20° apart. Volume-rendered 3D images of the heart were replaced with an outlined view of the prosthetic valve and/or pseudoaneurysm, depending on procedure. Finally, 3D-rendered images and landmarks were displayed with the same perspective as the x-ray system and any rotations of the c-arm were congruent with the CT. Only those patients were included in the study in which these guidance landmarks were followed during the procedure for LV puncture.
Technique of transapical access and closure
Fused overlay imaging was used to guide needle entry following the appropriate landmarks (1). A 21-gauge micropuncture needle (Cook Medical, Bloomington, Indiana) was used for access. Entry was determined by using live fluoroscopy and injecting diluted contrast into the LV (Fig. 1E). The procedure was also monitored with transesophageal echocardiography to assess for the presence and degree of pleural and pericardial effusions. After establishing LV entry, an 0.018-inch guidewire was advanced through the needle, and the needle was exchanged for a 5-F or 6-F radial sheath (Cook Medical) (Fig. 1F). If necessary, the radial sheath was exchanged for alternative sizes or delivery sheaths ranging from 6-F to 12-F.
TA was closed using a 6/4-mm Amplatzer Duct Occluder (ADO) (St. Jude Medical, Minneapolis, Minnesota); this is an “off-label” use (Fig. 1G). Implantation of the closure device was performed under fluoroscopic guidance. The device was introduced from the delivery sheath and the distal disk of the device was opened into the LV cavity. The device was pulled back and a small amount of contrast was injected to identify the endocardial surface. The device was released after the disk conformed to the endocardial surface and there was elongation of the device body in the myocardium with systolic compression.
Data collection and analysis
TA distance difference was evaluated by the difference between the PPS as determined by the pre-CTA–fluoroscopic fusion landmark and the actual puncture site (APS) as determined by the epicardial surface of the ADO on the post-CTA. CTA image sets, both pre- and post-procedure, were chosen from the same diastolic phase of the cardiac cycle. In addition, to reduce measurement uncertainty, index points of fixed cardiac structures were determined including the ostium of left main coronary artery (oLMCA) and the center of the mitral valve prosthesis (VAL). These points were chosen because of their reliability of measurement and their stability around or within the heart.
ImageJ software (National Institutes of Health, Bethesda, Maryland) was used to perform analysis. Coordinates of pre-procedure landmarks that were placed from the HeartNavigator software for guidance were transcribed onto the pre-procedural CTA using ImageJ. Distance measurements were calculated by subtracting the coordinates from the fixed cardiac structure to both the PPS and APS in 3 planes, lateral (x), anteroposterior (y), and craniocaudal (z) (Figs. 2A to 2D). All coordinates were confirmed in the sagittal, coronal, and transverse planes simultaneously. The following measurements were obtained: distance of PPS to oLMCA and PPS to center of VAL in pre-procedural CTA and APS to oLMCA and APS to center of VAL in post-procedural CTA. Additional measurements obtained were the distances from the left anterior descending (LAD) artery to the PPS and APS on their respective imaging studies at the same z-plane. The distance in 2D and 3D were calculated using the Pythagorean theorem. A diagrammatic representation of the measurements is shown in Figure 3.
Two independent readers, blinded to both the patients and each other, performed measurements. Readers were trained on the appropriate study software and standardized measurements were performed on index cases prior to initiation of the study. The average of the 2 readers was calculated and used to determine the distance difference between the PPS and APS in each of the 3 planes to better estimate the true distance.
Historical control subjects, who had LV puncture/closure attempted prior to fusion imaging guidance, matched to age and sex were collected and compared to assess differences in success rates and clinical endpoints. Procedural data such as contrast volume and fluoroscopy time were recorded for the entire procedure. Technical success was defined as successful TA with introduction of a sheath into the LV and closure of the entry site as described herein. Clinical endpoints chosen included TA-related complications related to fusion imaging: coronary artery laceration (particularly of the LAD); hemothorax; pericardial effusion/tamponade; closure device embolization/migration; pneumothorax; new wall motion abnormality; bleeding with hemodynamic compromise requiring surgical closure; or death.
Statistical analyses were performed using SPSS statistical package (version 20, IBM SPSS Inc., Chicago, Illinois). Continuous variables were reported as mean ± SD or as median (interquartile range [IQR]) where appropriate. Unpaired Student t test and z test were used for the comparison of continuous, normally distributed variables and to compare proportions between the study group and historical control subjects, respectively. Kolmogorov-Smirnov and Shapiro-Wilk tests were performed to assess whether measurements were normally distributed. Interobserver variability between readers was assessed using the Bland-Altman method and the Pearson rank correlation coefficient with limits of agreement reported. Statistical significance was defined as p < 0.05.
Patient demographics are presented in Table 1. Patients had a mean age of 70 ± 12 years and body surface area of 1.9 ± 0.2 m2. All patients had a mitral valve prosthesis that could be used for reference. The types of procedures performed that required TA were mitral PVL closure in 70% (n = 14), mitral PVL closure and transcatheter valve-in-valve implantation in 20% (n = 4), LV pseudoaneurysm closure in 5% (n = 1), and aortic PVL and Gerbode ventricular septal defect closure in 5% (n = 1).
The median TA distance between APS and PPS in reference to the oLMCA was 5.00 mm (IQR: 1.98 to 12.64 mm) in the x-plane, 4.48 mm (IQR: 1.98 to 13.08 mm) in the y-plane, and 5.57 mm (IQR: 3.89 to 13.62 mm) in the z-plane. In reference to the VAL, the median TA distance between APS and PPS was 3.27 mm (IQR: 1.88 to 11.24 mm) in the x-plane, 4.00 mm (IQR: 1.62 to 11.86 mm) in the y-plane, and 4.96 mm (IQR: 1.92 to 11.76 mm) in the z-plane (Table 2). The median TA 3D distance was 15.57 mm (IQR: 7.67 to 20.51 mm) and 14.11 mm (IQR: 9.08 to 18.05 mm) to the oLMCA and VAL, respectively. The median TA distance between APS and PPS in reference to the LAD was 5.68 mm (IQR: 3.12 to 7.43 mm) in the x-plane and 5.20 mm (IQR: 2.57 to 11.82 mm) in the y-plane, with a median 2D distance of 9.11 mm (IQR: 5.70 to 13.46 mm). The mean TA distance to the LAD was 15.5 ± 7.8 mm and 22.7 ± 13.7 mm in x- and y-planes.
Distances measured between each of the 3 fixed cardiac structures (oLMCA, VAL, and LAD) and the PPS or APS by the 2 independent cardiac readers are reported in Table 3. Mean difference between the 2 readers is also shown, ranging from 0.54 mm to 2.88 mm. The maximum observed mean difference between the readers was 2.88 mm ± 0.48 mm, across all measurement in 3 planes. Pearson correlation coefficient (Table 3) and Bland-Altman plots (Fig. 4) in reference to each measurement revealed agreement between observers with plots shown. The overall mean difference between Reader #1 and Reader #2 was small.
Historical control subjects matched to age and sex were similar in patient demographics to study patients, except for significantly less body surface area (1.8 ± 0.2 m2 vs. 1.9 ± 0.2 m2; p = 0.03) and body mass index (25.1 ± 3.9 kg/m2 vs. 28.4 ± 6.2 kg/m2; p = 0.05). There were no differences in procedural data, contrast volume (32 ± 19 cc vs. 41 ± 54 cc; p = 0.4) and fluoroscopy time (29.4 ± 22 min vs. 35.2 ± 21 min; p = 0.4). Technical success was similar between groups with 100% of patients having successful TA. The study patients had no evidence of coronary artery laceration identified on immediate post-procedural CTA. Furthermore, no other acute complications associated with percutaneous TA were recognized. Historical control subjects, however, had a higher incidence of complications (20% vs. 0%; p = 0.03). Two patients developed pericardial effusions/tamponade requiring drainage; 1 patient developed a hemothorax requiring drainage; and 1 patient died, intraprocedurally, in the setting of suprasystemic pulmonary hypertension after developing pulseless electrical activity after TA.
TA provides a direct approach to both the aortic valve and mitral valve apparatus. Traditionally, TA was limited to diagnostic procedures due to increased rates of complications with larger sheath sizes and anticoagulation required for intervention (5,6,10,11). The use of fluoroscopy alone has limited the ability of the interventionalist to reliably achieve a direct puncture at a safe distance from lung and coronary arteries while also being in direct line with the cardiac defect of interest, increasing the risk of bleeding, pneumothorax, and coronary laceration and decreasing the rates of procedural success with increased procedural times. Improved pre-procedural planning and intraprocedural guidance, with the use of fusion imaging, provide an alternative means to guide a more accurate and safe TA (1).
The results of this study demonstrate that, with the use of CTA-fluoroscopy fusion imaging to guide TA, the APS can be approximated near the intended or planned puncture site, despite cardiac motion. The planned and actual transapical puncture sites were within a median distance ranging from 3.27 mm to 5.57 mm of each other, across all 3 planes (x, y, and z). In addition, placement of landmarks on the pre-procedural CTA and segmentation of the cardiac structures including the coronary arteries, aided in avoidance of the coronary vasculature.
The actual transapical puncture site was able to be maintained at a safe distance away from the LAD, mean distance of approximately 1 to 2 cm (x- and y-planes). Due to the ability to label the coronary arteries, the location of the LAD with respect to TA was determined without the need for coronary angiography. Although complications have been well reported with the TA approach, the potential for improved safety using fusion imaging guidance is further supported with this study (1).
Although attempts were made to reduce uncertainty in measurement, there may exist differences that could not be accounted for between the pre- and post-procedural CTA. Changes in the index point position due to respiration could affect measurement. In addition, all pre-procedural CTA were performed with an inspiratory breath-hold; for TA, however, needle puncture was performed during cessation of ventilation in order to deflate the lung away from puncture site. Furthermore, small differences in volume status, cardiac chamber size, and ejection fraction between studies are difficult to assess and were not corrected for.
Current fusion imaging software also does not provide for motion compensation, specifically cardiac motion. The potential fusion of multiple cardiac phases, true 4-dimensional overlay, may further improve the accuracy of this technology. Two patients did not receive contrast for the post-procedural CTA, reducing the number of patients for whom the distance of the APS to the LAD could be assessed. Additionally, the sample size was small. This may be attributed to the less than optimal variance noted between readers.
In addition, the study results may also be operator-dependent, requiring operator experience with both percutaneous TA and fusion imaging technology. Over 50 transapical procedures had been performed at our center prior to enrollment in the study and may have contributed to the technical results reported. Additional CT training is required for the acquisition and evaluation of structural heart patients and separate instruction is necessary for the fusion imaging technology (12). Nonetheless, with the automation of many steps in the fusion software, the learning curve is not steep.
CTA-fluoroscopy fusion imaging is useful for the guidance of TA with the ability to accurately puncture the LV while maintaining an adequate distance from the LAD, thereby, potentially reducing TA-related complications. This work should promote further study on the use of CTA-fluoroscopy fusion imaging technology to guide complex structural heart interventions.
Dr. Kliger has received honoraria from St. Jude Medical. Dr. Kronzon has received consulting fees from St. Jude Medical and honoraria from Philips Healthcare. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- Amplatzer ductal occluder
- actual puncture site
- computed tomography angiography
- interquartile range
- left anterior descending
- left ventricle
- ostium of the left main coronary artery
- planned puncture site
- paravalvular leak
- transapical access
- center of mitral valve prosthesis
- Received June 25, 2013.
- Revision received September 9, 2013.
- Accepted October 3, 2013.
- American College of Cardiology Foundation
- Ruiz C.E.,
- Jelnin V.,
- Kronzon I.,
- et al.
- Dudiy Y.,
- Jelnin V.,
- Einhorn B.N.,
- Kronzon I.,
- Cohen H.A.,
- Ruiz C.E.
- Brock R.,
- Milstein B.B.,
- Ross D.N.
- Auricchio A.,
- Sorgente A.,
- Soubelet E.,
- et al.
- Budoff M.J.,
- Cohen M.C.,
- Garcia M.J.,
- et al.