J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofCardiovascularCT.com
Original Research Article
Radiographic and electrocardiography-gated noncontrast cardiac CT assessment of lead perforation: Modality comparison and interobserver agreement Christian Balabanoff MDa,b, Cristopher E. Gaffney MDc,d, Eduard Ghersin MDc, Yoji Okamoto MDa,e, Roger Carrillo MDa, Joel E. Fishman MD, PhDc,* a Department of Surgery, Division of Cardiothoracic Surgery, University of Miami Miller School of Medicine, Miami, FL, USA b Division of Cardiology, Abington Memorial Hospital, Abington, PA, USA c Department of Radiology, Jackson Memorial Hospital, University of Miami Miller School of Medicine, 1611 N.W., 12th Avenue, Miami, FL, USA d Department of Radiology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA e Department of Medicine, Division of Cardiology, Kurashiki Central Hospital, Okayama, Japan
article info
abstract
Article history:
Background: Pacemaker or implantable cardioverter-defibrillator lead extraction may be
Received 4 August 2013
required because of infection, malfunction, or breakage. The preprocedural identification
Received in revised form
of lead tip position may help ensure safe performance of the procedure.
20 June 2014
Objective: To analyze the ability of chest radiography and CT imaging to characterize lead
Accepted 16 August 2014
tip position and identify perforation in a population of patients who underwent lead extraction. Methods: Among patients who underwent lead extraction between November 2008 and
Keywords:
April 2011, a nonrandom subset of 50 patients with 116 leads was selected for retrospective
Pacemaker
analysis. All patients had undergone chest radiography and thin-section electrocardiog-
Artificial
raphy-gated noncontrast cardiac CT. Two radiologists independently evaluated the imag-
Defibrillators
ing studies, using oblique multiplanar image reconstruction techniques for the CT
Tomography
examinations. Beam hardening artifacts were graded (0e3). Likelihood of perforation on
X-ray computed
each imaging study was graded on a 5-point scale.
Radiography
Results: Among 116 leads, 17 were identified as perforated on CT, 12 leads were equivocal,
Thoracic
and 87 were not perforated. Interobserver agreement for CT perforation vs nonperforation was good (k ¼ 0.71); weighted kappa for the entire 5-point scale was moderate (k ¼ 0.54). Beam hardening artifacts were common, with a mean value of 2.1. The 2 observers
Conflict of interest: Roger Carrillo is a consultant for Spectranetics, Medtronic, Inc., St. Jude Medical, Sorin, and Sensormatic and receives grants or research support from St. Jude Medical (NCT # 00940888), Medtronic (NCT # 00893386). The other authors declare no conflicts of interest. * Corresponding author. E-mail address: [email protected] (J.E. Fishman). 1934-5925/$ e see front matter ª 2014 Society of Cardiovascular Computed Tomography. All rights reserved. http://dx.doi.org/10.1016/j.jcct.2014.08.004
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
385
identified perforation on chest radiography with an average sensitivity of 15% compared with CT. The 2 observers did not agree on any cases of chest radiographic perforation (k ¼ 0.1). Conclusion: Electrocardiography-gated noncontrast cardiac CT imaging with oblique multiplanar analysis can identify potential lead perforation with a moderate-to-good level of interobserver agreement. Chest radiography demonstrates poor sensitivity and interobserver agreement compared with CT. ª 2014 Society of Cardiovascular Computed Tomography. All rights reserved.
1.
Introduction
Approximately 67,000 defibrillators, 178,000 pacemakers, 33,000 cardiac resynchronization defibrillators, and 7000 cardiac resynchronization pacemakers were implanted in the United States in 2004.1 Subsequent device and lead extraction may be required for indications such as infection, device malfunction, and lead breakage. Lead extraction may be accomplished by mechanical or energy-assisted means such as thoracotomy, electrocautery, and lasers. In any patient with cardiac leads, early or late lead perforation may occur through the myocardium, into the epicardial space, pericardium, or chest wall. Lead perforation is often clinically occult and only infrequently accompanied by symptoms such as pain. Pre-extraction knowledge of the presence of lead perforation could significantly impact the method and techniques chosen for lead extraction. Chest radiography in 2 views may occasionally demonstrate clear evidence of perforation but is inherently limited because distinction between the ventricular cavity, myocardium, and pericardium cannot be made. Chest CT would seem to have much to recommend it for lead tip identification, and there have been numerous case reports and occasional larger series published to this end.2e6 CT images of leads, however, can be challenging to interpret for 2 main reasons: the heart and leads are in motion, and metallic materials can produce severe streak artifacts.7 Our objective was to analyze a series of chest radiographs and electrocardiography (ECG)-gated noncontrast cardiac CT scans of patients with device leads before extraction to determine the degree of confidence in identifying perforation, calculate interobserver agreement, and compare radiographic results with those of CT.
2.
Methods
This project was approved by the University of Miami institutional review board with waiver of informed consent. A cardiothoracic surgeon (R.C.) has been performing laserassisted and operative lead extraction at the University of Miami Hospital since September 2008. As part of clinical protocol, ECG-gated noncontrast cardiac CT scanning was performed before the extraction. Among all 329 patients who underwent lead extraction between November 2008 and April 2011, a nonrandom convenience subset of patients was selected for retrospective radiographic and CT analysis. To capture as many patients with potentially perforated leads as possible, the subset included all patients (n ¼ 16) who had a preprocedure suspicion of possible lead perforation based on
the radiographic and/or CT report generated at the time of the study. A larger group (n ¼ 34) of randomly selected patients whose imaging did not suggest perforation was added for a total group size of 50 nonrandom patients with 116 leads. The obtained images included single anteroposterior (n ¼ 25) or 2view (n ¼ 25) chest radiography. Chest CT was performed on a Siemens ASþ 128-slice, single-source scanner (Siemens Healthcare, Malvern, PA) using ECG gating at 120 kVP, 150 to 190 mAs, and a mixture of both prospective and retrospective technique. Prospectively gated studies were centered at 70% of the R-R interval. The “best diastolic” phase selected by the scanner was used for retrospectively gated studies, which ranged from 66% to 86% of the R-R interval. Multiphase images were isovolumetrically (0.6 0.6 mm) reconstructed with no overlap using a B26f smooth cardiac kernel and interpreted using oblique multiplanar reformat (MPR) in TeraRecon iNtuition (version 4.4.7; Foster City, CA). Heart rate was not recorded and no beta blockers were administered. Subsequently, patients underwent laser lead extraction with concurrent transesophageal ECG of a total of 106 leads. Ten leads were not extracted. Each extracted lead was transected proximally and a laser sheath was inserted over the lead, advanced in the direction of the lead tip, and activated at points of lead adhesion. Lasering was halted 1 cm from the lead tip in the myocardium unless the CT scan demonstrated lead perforation, in which case the laser sheath was halted at a prudent distance from the tip to prevent a larger perforation. One patient whose preprocedure CT demonstrated lead perforation had a left anterior minithoracotomy performed to assist in safe removal of the lead, which was directly observed extending out the right ventricular chamber. Two other patients without perforation also had thoracotomies performed for other reasons. There were 4 deaths among the 50 patients during their hospitalizations, all due to sepsis and resulting multiorgan failure in patients whose indication for lead removal was infection. There were no deaths attributable to the lead extraction procedure itself. One patient had a surgical site hematoma requiring drainage, constituting a minor complication of lead removal.
2.1.
Image analysis
Two radiologists (E.G., J.F.), who were not involved in the initial clinical interpretation of the CT scans and who were unaware of whether there was suspected lead perforation, independently and randomly evaluated the imaging studies. Using an image reconstructed in mid-diastole, orthogonal oblique MPRs were created to demonstrate the length of the lead tip in 2 orthogonal views with the third view being a
386
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
cross-section of the lead. Dual-window and level Hounsfield unit (HU) settings were routinely applied. The first setting was designed for optimal demarcation of the lead tip while minimizing streak artifacts, to most closely demonstrate the architecture of the lead tip as seen on the CT scout view (Fig. 1AeC). This used a relatively wide window (w1400) with a center of approximately 300 HU, equivalent to a modified bone setting. As on nonecontrast-enhanced CT the density of the right ventricle blood pool is generally within 100 HU of epicardial fat, a higher contrast window and level setting was then applied for optimal demarcation of the interface between the right ventricular myocardium and blood pool and the epicardial fat. This was similar to a mediastinal setting, using a center around 50 HU and a narrower width of approximately 400 HU (Fig. 1D). Using these images, the presence of perforation was graded on a 5-point scale according to the following. Grade 1: tip clearly within the region of the blood pool and myocardium. Grade 2: tip abutting the blood pool and myocardiumepicardial fat interface but does not measurably cross the interface.
Grade 3: tip passes no more than 2 mm past the blood pool and myocardium-epicardial interface, 2 mm being chosen because that is the approximate length of the lead tip screw. Grade 3 was considered equivocal for perforation. Grade 4: tip crosses >2 mm and no greater than 4 mm past the blood pool and myocardium-epicardial fat interface. Grade 5: tip crosses >4 mm past the blood pool and myocardium-epicardial fat interface. After all independent readings, a consensus reading was performed when the independent readings were not either grade 1 or 2, grade 3, or grade 4 or 5. Motion artifacts were noted when sufficient to degrade interpretation of lead position (absent ¼ 0, present ¼ 1), and metallic streak artifacts were graded on a 4-point scale (from none ¼ 0 to severe ¼ 3). Chest radiographs were reviewed using a picture archiving and communication system workstation (Philips iSite, Andover, MA). A 5-point scale was used to judge lead position (1 ¼ definitely not perforated, 2 ¼ probably not perforated, 3 ¼ equivocal, 4 ¼ probably perforated, and 5 ¼ definitely perforated). Consensus interpretations were not performed in cases of disagreement.
Fig. 1 e Perforated right ventricular implantable cardioverter-defibrillator lead. (A) CT scout image demonstrating the architecture of the lead. (B, C) orthogonal multiplanar reformat images using modified bone settings redemonstrate the lead tip architecture. (D) Same image as (C) using mediastinal settings better demonstrates grade-5 perforation of the lead tip (arrow) past the pericardium (arrowheads).
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
2.2.
Statistical analysis
Continuous and categorical variables were analyzed using SAS, version 9.1 (SAS Institute Inc., Cary, NC). Continuous variables are analyzed as mean standard deviation. Categorical variables are analyzed as numbers or percentages. The Fisher exact test was used for 2 2 contingency tables. P values
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofCardiovascularCT.com
Original Research Article
Radiographic and electrocardiography-gated noncontrast cardiac CT assessment of lead perforation: Modality comparison and interobserver agreement Christian Balabanoff MDa,b, Cristopher E. Gaffney MDc,d, Eduard Ghersin MDc, Yoji Okamoto MDa,e, Roger Carrillo MDa, Joel E. Fishman MD, PhDc,* a Department of Surgery, Division of Cardiothoracic Surgery, University of Miami Miller School of Medicine, Miami, FL, USA b Division of Cardiology, Abington Memorial Hospital, Abington, PA, USA c Department of Radiology, Jackson Memorial Hospital, University of Miami Miller School of Medicine, 1611 N.W., 12th Avenue, Miami, FL, USA d Department of Radiology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA e Department of Medicine, Division of Cardiology, Kurashiki Central Hospital, Okayama, Japan
article info
abstract
Article history:
Background: Pacemaker or implantable cardioverter-defibrillator lead extraction may be
Received 4 August 2013
required because of infection, malfunction, or breakage. The preprocedural identification
Received in revised form
of lead tip position may help ensure safe performance of the procedure.
20 June 2014
Objective: To analyze the ability of chest radiography and CT imaging to characterize lead
Accepted 16 August 2014
tip position and identify perforation in a population of patients who underwent lead extraction. Methods: Among patients who underwent lead extraction between November 2008 and
Keywords:
April 2011, a nonrandom subset of 50 patients with 116 leads was selected for retrospective
Pacemaker
analysis. All patients had undergone chest radiography and thin-section electrocardiog-
Artificial
raphy-gated noncontrast cardiac CT. Two radiologists independently evaluated the imag-
Defibrillators
ing studies, using oblique multiplanar image reconstruction techniques for the CT
Tomography
examinations. Beam hardening artifacts were graded (0e3). Likelihood of perforation on
X-ray computed
each imaging study was graded on a 5-point scale.
Radiography
Results: Among 116 leads, 17 were identified as perforated on CT, 12 leads were equivocal,
Thoracic
and 87 were not perforated. Interobserver agreement for CT perforation vs nonperforation was good (k ¼ 0.71); weighted kappa for the entire 5-point scale was moderate (k ¼ 0.54). Beam hardening artifacts were common, with a mean value of 2.1. The 2 observers
Conflict of interest: Roger Carrillo is a consultant for Spectranetics, Medtronic, Inc., St. Jude Medical, Sorin, and Sensormatic and receives grants or research support from St. Jude Medical (NCT # 00940888), Medtronic (NCT # 00893386). The other authors declare no conflicts of interest. * Corresponding author. E-mail address: [email protected] (J.E. Fishman). 1934-5925/$ e see front matter ª 2014 Society of Cardiovascular Computed Tomography. All rights reserved. http://dx.doi.org/10.1016/j.jcct.2014.08.004
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
385
identified perforation on chest radiography with an average sensitivity of 15% compared with CT. The 2 observers did not agree on any cases of chest radiographic perforation (k ¼ 0.1). Conclusion: Electrocardiography-gated noncontrast cardiac CT imaging with oblique multiplanar analysis can identify potential lead perforation with a moderate-to-good level of interobserver agreement. Chest radiography demonstrates poor sensitivity and interobserver agreement compared with CT. ª 2014 Society of Cardiovascular Computed Tomography. All rights reserved.
1.
Introduction
Approximately 67,000 defibrillators, 178,000 pacemakers, 33,000 cardiac resynchronization defibrillators, and 7000 cardiac resynchronization pacemakers were implanted in the United States in 2004.1 Subsequent device and lead extraction may be required for indications such as infection, device malfunction, and lead breakage. Lead extraction may be accomplished by mechanical or energy-assisted means such as thoracotomy, electrocautery, and lasers. In any patient with cardiac leads, early or late lead perforation may occur through the myocardium, into the epicardial space, pericardium, or chest wall. Lead perforation is often clinically occult and only infrequently accompanied by symptoms such as pain. Pre-extraction knowledge of the presence of lead perforation could significantly impact the method and techniques chosen for lead extraction. Chest radiography in 2 views may occasionally demonstrate clear evidence of perforation but is inherently limited because distinction between the ventricular cavity, myocardium, and pericardium cannot be made. Chest CT would seem to have much to recommend it for lead tip identification, and there have been numerous case reports and occasional larger series published to this end.2e6 CT images of leads, however, can be challenging to interpret for 2 main reasons: the heart and leads are in motion, and metallic materials can produce severe streak artifacts.7 Our objective was to analyze a series of chest radiographs and electrocardiography (ECG)-gated noncontrast cardiac CT scans of patients with device leads before extraction to determine the degree of confidence in identifying perforation, calculate interobserver agreement, and compare radiographic results with those of CT.
2.
Methods
This project was approved by the University of Miami institutional review board with waiver of informed consent. A cardiothoracic surgeon (R.C.) has been performing laserassisted and operative lead extraction at the University of Miami Hospital since September 2008. As part of clinical protocol, ECG-gated noncontrast cardiac CT scanning was performed before the extraction. Among all 329 patients who underwent lead extraction between November 2008 and April 2011, a nonrandom convenience subset of patients was selected for retrospective radiographic and CT analysis. To capture as many patients with potentially perforated leads as possible, the subset included all patients (n ¼ 16) who had a preprocedure suspicion of possible lead perforation based on
the radiographic and/or CT report generated at the time of the study. A larger group (n ¼ 34) of randomly selected patients whose imaging did not suggest perforation was added for a total group size of 50 nonrandom patients with 116 leads. The obtained images included single anteroposterior (n ¼ 25) or 2view (n ¼ 25) chest radiography. Chest CT was performed on a Siemens ASþ 128-slice, single-source scanner (Siemens Healthcare, Malvern, PA) using ECG gating at 120 kVP, 150 to 190 mAs, and a mixture of both prospective and retrospective technique. Prospectively gated studies were centered at 70% of the R-R interval. The “best diastolic” phase selected by the scanner was used for retrospectively gated studies, which ranged from 66% to 86% of the R-R interval. Multiphase images were isovolumetrically (0.6 0.6 mm) reconstructed with no overlap using a B26f smooth cardiac kernel and interpreted using oblique multiplanar reformat (MPR) in TeraRecon iNtuition (version 4.4.7; Foster City, CA). Heart rate was not recorded and no beta blockers were administered. Subsequently, patients underwent laser lead extraction with concurrent transesophageal ECG of a total of 106 leads. Ten leads were not extracted. Each extracted lead was transected proximally and a laser sheath was inserted over the lead, advanced in the direction of the lead tip, and activated at points of lead adhesion. Lasering was halted 1 cm from the lead tip in the myocardium unless the CT scan demonstrated lead perforation, in which case the laser sheath was halted at a prudent distance from the tip to prevent a larger perforation. One patient whose preprocedure CT demonstrated lead perforation had a left anterior minithoracotomy performed to assist in safe removal of the lead, which was directly observed extending out the right ventricular chamber. Two other patients without perforation also had thoracotomies performed for other reasons. There were 4 deaths among the 50 patients during their hospitalizations, all due to sepsis and resulting multiorgan failure in patients whose indication for lead removal was infection. There were no deaths attributable to the lead extraction procedure itself. One patient had a surgical site hematoma requiring drainage, constituting a minor complication of lead removal.
2.1.
Image analysis
Two radiologists (E.G., J.F.), who were not involved in the initial clinical interpretation of the CT scans and who were unaware of whether there was suspected lead perforation, independently and randomly evaluated the imaging studies. Using an image reconstructed in mid-diastole, orthogonal oblique MPRs were created to demonstrate the length of the lead tip in 2 orthogonal views with the third view being a
386
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
cross-section of the lead. Dual-window and level Hounsfield unit (HU) settings were routinely applied. The first setting was designed for optimal demarcation of the lead tip while minimizing streak artifacts, to most closely demonstrate the architecture of the lead tip as seen on the CT scout view (Fig. 1AeC). This used a relatively wide window (w1400) with a center of approximately 300 HU, equivalent to a modified bone setting. As on nonecontrast-enhanced CT the density of the right ventricle blood pool is generally within 100 HU of epicardial fat, a higher contrast window and level setting was then applied for optimal demarcation of the interface between the right ventricular myocardium and blood pool and the epicardial fat. This was similar to a mediastinal setting, using a center around 50 HU and a narrower width of approximately 400 HU (Fig. 1D). Using these images, the presence of perforation was graded on a 5-point scale according to the following. Grade 1: tip clearly within the region of the blood pool and myocardium. Grade 2: tip abutting the blood pool and myocardiumepicardial fat interface but does not measurably cross the interface.
Grade 3: tip passes no more than 2 mm past the blood pool and myocardium-epicardial interface, 2 mm being chosen because that is the approximate length of the lead tip screw. Grade 3 was considered equivocal for perforation. Grade 4: tip crosses >2 mm and no greater than 4 mm past the blood pool and myocardium-epicardial fat interface. Grade 5: tip crosses >4 mm past the blood pool and myocardium-epicardial fat interface. After all independent readings, a consensus reading was performed when the independent readings were not either grade 1 or 2, grade 3, or grade 4 or 5. Motion artifacts were noted when sufficient to degrade interpretation of lead position (absent ¼ 0, present ¼ 1), and metallic streak artifacts were graded on a 4-point scale (from none ¼ 0 to severe ¼ 3). Chest radiographs were reviewed using a picture archiving and communication system workstation (Philips iSite, Andover, MA). A 5-point scale was used to judge lead position (1 ¼ definitely not perforated, 2 ¼ probably not perforated, 3 ¼ equivocal, 4 ¼ probably perforated, and 5 ¼ definitely perforated). Consensus interpretations were not performed in cases of disagreement.
Fig. 1 e Perforated right ventricular implantable cardioverter-defibrillator lead. (A) CT scout image demonstrating the architecture of the lead. (B, C) orthogonal multiplanar reformat images using modified bone settings redemonstrate the lead tip architecture. (D) Same image as (C) using mediastinal settings better demonstrates grade-5 perforation of the lead tip (arrow) past the pericardium (arrowheads).
J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 3 8 4 e3 9 0
2.2.
Statistical analysis
Continuous and categorical variables were analyzed using SAS, version 9.1 (SAS Institute Inc., Cary, NC). Continuous variables are analyzed as mean standard deviation. Categorical variables are analyzed as numbers or percentages. The Fisher exact test was used for 2 2 contingency tables. P values
