Noninvasive Creation of an Atrial Septal Defect by Histotripsy in a Canine Model
Background— The primary objective of this study was to develop an image-guided, noninvasive procedure to create or enlarge an atrial septal defect for the treatment of neonates with hypoplastic left heart syndrome and an intact or restrictive atrial septum. Histotripsy is an innovative ultrasonic technique that produces nonthermal, mechanical tissue fractionation through the use of high-intensity ultrasound pulses. This article reports the pilot in vivo study to create an atrial septal defect through the use of extracardiac application of histotripsy in an open-chest canine model.
Methods and Results— In 10 canines, the atrial septum was exposed to histotripsy by an ultrasound transducer positioned outside the heart. Ultrasound pulses of 6-microsecond duration at a peak negative pressure of 15 MPa and a pulse repetition frequency of 3.3 kHz were generated by a 1-MHz focused transducer. The procedure was guided and monitored by real-time ultrasound imaging. In 9 of 10 canines, an atrial septal defect was produced, and shunting across the atrial septum was visualized. Pathology of the hearts showed atrial septal defects with minimal damage to surrounding tissue. No damage was found on the epicardial surface of the heart or other structures.
Conclusions— Under real-time ultrasound guidance, atrial septal defects were successfully created with extracardiac histotripsy in a live canine model. Although further studies in an intact animal model are needed, these results provide promise of histotripsy becoming a valuable clinical tool.
Received June 24, 2009; accepted December 15, 2009.
Hypoplastic left heart syndrome (HLHS) is a rare and complex congenital heart disease.1 If left untreated, mortality approaches 95% by the first month of life.2 Surgical palliation of patients with a nonrestrictive atrial septum at birth currently increases survival to ≈80% at hospital discharge after stage I palliation.3–5 However, ≈6% of HLHS patients have a completely intact atrial septum at birth, and up to 22% have a severely restrictive atrial septum.6,7 Early survival for this subpopulation of patients has been reported to range from 10% to 48%.7–9 To improve the survival of these neonates, a nonrestrictive atrial septal defect (ASD) is required before surgical palliation.
Clinical Perspective on p 749
Because emergent surgical septectomy usually requires cardiopulmonary bypass with its associated complications, percutaneous catheter approaches have been used. Current percutaneous treatment options include variations of catheter-based septostomies such as Rashkind septostomy, Park blade septostomy, static balloon septoplasty, and atrial septal stent placement.7,8,10 These methods have demonstrated effectiveness in ASD creation and/or enlargement and improved survival of patients with HLHS and a restrictive or intact atrial septum.8,11–13 In certain cases, however, these approaches are limited by anatomic challenges and have increased potential for complications.8 Thus, an alternative noninvasive method such as high-intensity focused ultrasound (HIFU) for creation of ASDs in this patient population may prove to be a beneficial innovation.
HIFU can ablate internal organs noninvasively from an extracorporeal position.14 For cardiac applications, researchers have investigated the feasibility of using extracardiac HIFU to ablate cardiac tissue to treat arrhythmias.15–17 Lesions were created inside the beating heart without damaging the intervening heart wall. The main mechanism of these HIFU studies is that continued or long ultrasound pulses can cause heating in the target tissue, which results in thermal necrosis. However, ASD creation requires tissue removal, which does not occur with HIFU thermal therapy.
Our approach also uses focused ultrasound, but the physical mechanism is different. It depends on precise control of acoustic cavitation to produce nonthermal, mechanical tissue fractionation and removal. This technique has been called histotripsy to reflect the desired end results: soft tissue (“histo”) breakdown (“tripsy”).7–9 Histotripsy uses short pulses (≤20 microseconds) at extremely high pressure to reliably initiate and maintain a cluster of microbubbles (acoustic cavitation) in the target tissue.18 Rapid bubble expansion, contraction, and collapse ultimately fractionate tissue to acellular debris.19 Cavitation is a threshold phenomenon wherein cavitation bubbles are generated only when the ultrasound pressure is above a certain threshold.20 Therefore, tissue damage is limited within the focal zone where the pressure is above the cavitation threshold. Histotripsy suppresses heating by separating short, intense pulses with long cool-down time (at least 20 times longer than the pulse duration). The treatment is guided and monitored in real time by standard ultrasound imaging.21 Previously, histotripsy was demonstrated to create clearly demarcated perforations in excised porcine atrial tissue.22 This article reports a pilot study to create an ASD by histotripsy in a live beating heart using an open-chest canine model.
The procedures described here were reviewed and approved by the University Committee on Use and Care of Animals at the University of Michigan. A total of 10 healthy adult 25- to 40-kg mongrel dogs underwent the surgery. The animals were preanesthetized with acepromazine (3 mg IM), and anesthesia was induced with thiopental (8 to 12 mg/kg IV) after intravenous catheter placement. The dogs were intubated and placed on a surgical table. The animals were maintained on isoflurane (1% to 3%) inhalation anesthesia for the duration of the procedure. Each animal was monitored by a pulse oximeter. A sternotomy incision provided direct access to the heart.
Ultrasound pulses for histotripsy treatment were generated by an external high-power ultrasound therapy transducer. The transducer was submerged in a heated, degassed external water bag, which was coupled to the pericardium with a thin plastic membrane and ultrasound coupling gel. To place the ultrasound focus on the atrial septum, the therapy transducer was moved in the water bag to adjust its standoff distance to the pericardium.
The therapy transducer was designed in our laboratory and fabricated by Imasonic, SA (Besançon, France). The circular transducer has a center frequency of 1 MHz, a geometric focal length of 90 mm, an outer diameter of 100 mm, and a 40-mm inner hole for housing an ultrasound imaging probe (Figure 1a). The transducer was driven by a high-voltage amplifier built in-house. Ultrasound pressure waveforms of the therapy transducer were measured in degassed water with a fiberoptic probe hydrophone.23 The peak negative and positive pressures used in the following experiment were 15 and 61 MPa, respectively (Figure 1b). Taking into account the attenuation caused by the ≈1-cm-thick atrial wall tissue in the pathway,24 the peak negative pressure reaching the atrial septum was estimated to be 12 MPa, resulting in a mechanical index of 12.25 This compares to the maximum mechanical index of 1.9 used in diagnostic ultrasound.25 The ultrasound pulses were 6 microseconds in duration (6 cycles at 1 MHz) and separated by 300 microseconds (ie, pulse repetition frequency, 3.3 kHz). These acoustic parameters have been used successfully to create perforations in the atrial wall tissue in vitro.22
The entire histotripsy procedure was guided by ultrasound imaging. The targeting and monitoring of the procedure were achieved by a 5-MHz phased-array ultrasound imaging probe (GE VingMed System FiVe, GE, Waukesha, Wis) inserted into the central hole of the therapy transducer. Another 10-MHz phased-array imaging probe was used to achieve higher-quality imaging of the atrial septum before and after the procedure by placing it directly on the epicardial surface of the heart.
The first step in the histotripsy procedure was to target the therapy focus to the atrial septum. Before the procedure, histotripsy pulses were applied to a waterbath, resulting in a bubble cloud and a bright (hyperechoic) zone on a 2-dimensional ultrasound image. The front center of the hyperechoic zone was marked as the focal position (Figure 2). Although the hyperechoic bubble zone generated in the water could be 7 to 8 mm wide, the bubble zone created in the blood on the tissue was 2 to 4 mm. The therapeutic transducer was then moved by a 3-axis positioning system (Parker Hannifin, Rohnert Park, Calif) to align the focus marker on the atrial septum surface in the right atrium. To verify the targeting accuracy, a small number of pulses were applied to produce a hyperechoic zone on the atrial septum.
Once the targeting was accomplished, histotripsy ultrasound exposures of 2 minutes were applied to the atrial septum at 1 time. After every 2 minutes of exposure, the atrial septum was scanned by 2-dimensional imaging and Doppler color-flow mapping to identify whether blood flow was seen across the atrial septum. Repetitive 2-minute exposures were delivered to the atrial septum until an ASD was generated.
Within 1 hour after the ASD creation, animals were euthanized with a lethal dose of pentobarbital (100 mg/kg IV). The heart was extracted, fixed in a 10% formalin solution, and dissected after 1 week of fixation. The atrial septum and the rest of the heart were first examined by gross evaluation and then processed with standard hematoxylin and eosin (H&E) staining for pathological evaluation.
Collection of Human HLHS Patient Data
To evaluate the relevance of the canine study to the applicability of histotripsy to human, important geometric and anatomic data on human HLHS neonatal patients were collected. Available acoustic window position and size; the thickness, size, and motion of the atrial septum; the thickness of overlying tissues in the ultrasound beam pathway; and the distance between these tissues were measured by analyzing patient echocardiography data. A total of 47 neonatal HLHS patients with an intact or restrictive (defined by a left-atrial-to-right-atrial pressure gradient of >5 mm Hg and/or reversal of flow in the pulmonary veins) atrial septum who received ultrasound imaging at Washington University Children’s Hospital during the past 10 years were included. These neonatal patients weighed 2.5 to 3.9 kg (mean, 3.4 kg) and were 1 to 3 days of age. All of the patients underwent a complete 2-dimensional and Doppler study with the commercially available ultrasound imaging system (Philips Sonos 5500, Siemens Sequoia and GE Vivid 7). The above procedures were fully approved by the institutional review board at Washington University Medical School and the University of Michigan.
In 9 of the 10 in vivo canine heart experiments, an ASD was successfully generated in 6 to 16 minutes of histotripsy treatment (Table 1). The ASD generation was verified with ultrasound imaging, color-flow Doppler mapping, gross morphology, and pathological evaluation. All of the animals survived the immediate procedure, and there were no complications such as pericardial effusions or sustained arrhythmias.
During the procedure, generally a ≈2- to 4-mm-wide bubble cloud was created at the focus on the atrial septum. The histotripsy ultrasound pulses went through the right atrium to reach the atrial septum. When the procedure was progressing normally, the bubble cloud appeared as a temporally changing (twinkling) hyperechoic zone on the ultrasound image (Figure 3a). The ultrasound image was disturbed by a small number of lines caused by the interference from the short histotripsy ultrasound pulses, but the image quality was sufficient for real-time monitoring.
The atrial septum was constantly moving during the histotripsy procedure because of the heart contraction and respiration motion. The ultrasound image showed that the atrial septum motion was mostly along the axial direction of the therapy ultrasound beam and perpendicular to the atrial septum, measured to be 8.1±3.3 mm (mean±SD; Table 1). The atrial septum also had motion in and out of the ultrasound image plane in the lateral direction of the ultrasound beam. Unfortunately, the extent of lateral motion could not be evaluated precisely with this setup.
As the treatment progressed, the bubble cloud eroded the atrial septum layer by layer within the treatment zone. Before the completion of the ASD, the erosion appeared as a dark indentation in the atrial septum on the ultrasound image, with the opening facing the right atrium, which was the ultrasound entrance side (Figure 3b). When the atrial septum was perforated, the ASD could be seen as a dark channel through the atrial septum by 2-dimensional ultrasound imaging (Figure 4). The ASD creation was further confirmed by color-flow Doppler mapping revealing left-to-right shunting across the atrial septum (Figure 4). The width of the color jet through the ASD was measured on Doppler images (3.40.9 mm) and is listed in Table 1.
Hemorrhage indicated by dark colorization was observed on both sides of the treated atrial septa where an ASD was created. The location of the ASD was consistent with that indicated on ultrasound image. The diameter of the dark-colored area (including the ASD) was larger on the right atrial side (≈4 to 10 mm in diameter; Figure 5a) and smaller on the left atrial side (≈2 to 5 mm in diameter; Figure 5b). The cross-section of the atrial septa showed cone-shaped lesions extending all the way through the atrial septum, with a larger opening on the right atrial side compared with the left. These results are consistent with the fact that the histotripsy ultrasound pulses reached the atrial septum from the right atrium.
The whole hearts appeared intact from the outside, with no visible damage on the epicardial surfaces (Figure 5c and 5d). After dissection, in 9 of 10 cases, no gross damage was observed inside the heart other than in the treatment area in the atrial septum (Figure 5e). In only 1 of the 9 successful cases, small areas of discoloration were found on the anterior wall of the right atrium (Figure 5f), which was in the pathway of the histotripsy ultrasound pulses, just opposite the created ASD.
On the H&E-stained slides, the central damage zone of the ASD contained acellular tissue debris and red blood cells (Figure 6a and 6b). Other than the red blood cells, no discernible intact cellular structures remained. It appeared that variable degrees of thrombus formed within the ASD (Figure 6b). Because all of the ASDs were patent with blood flow up to the time of euthanasia confirmed by Doppler, the thrombus is likely postmortem; however, the possibility of premortem thrombus formation partially obstructing the defect cannot be excluded. This necrotic zone extended from the right atrial surface all the way to the left, suggesting complete penetration through the atrial septum and creation of an ASD. The cross-section of the acellular zone was generally cone shaped. The width of the acellular zone measured on H&E cross-section slides was 5.0±1.1 mm on the right atrial surface and 2.6±0.4 mm on the left (Table 1).
Immediately outside the acellular zone was a rim of hemorrhage and structurally intact myocytes with a more eosinophilic hue compared with the normal myocytes, suggesting early contraction necrosis (Figure 6c). These myocytes were likely affected by histotripsy. Beyond this rim of damaged cells were normal-appearing myocytes with no further discernible damage (Figure 6d). The width of the eosinophilic and hemorrhagic rim surrounding the acellular ASD zone was measured from the outermost edge of the acellular zone to the outermost edge of the eosinophilic myocyte and hemorrhage zone on either side of the ASD. For each ASD, this measurement was performed for both sides of the ASD for the right and left atrial surfaces and using multiple slides across the same ASD, and the maximal value of the measurements is reported as the hemorrhage width in Table 1.
Histology of the atrial wall tissues in the ultrasound pathway appeared normal. In the 1 case in which the anterior wall of the right atrial wall had slight discoloration on morphology, H&E slides showed intact and normal-appearing myocytes with slight hemorrhage (Figure 7).
The Failure Case
In the 1 treatment in which an ASD was not generated, ultrasound went through first the right ventricle and then the right atrium to reach the atrial septum. This pathway is different from our other treatments in which ultrasound reaches the atrial septum through the right atrium directly. With the pathway used in the failure case, the atrial septum has a significantly larger motion with respect to the ultrasound beam (17.1-mm axial motion). In this case, the atrial septum was eroded but not perforated. This treatment was also performed early in the series of experiments, when the treatment procedures were not well developed.
Collection of Human HLHS Patient Data
The measurements based on echocardiography of 47 human HLHS neonatal patients with an intact or restrictive atrial septum showed a subcostal acoustic window with a length of 5.1 to 6.7 cm. From the subcostal window, ultrasound can access the atrial septum without bone or lung obstruction. To reach the atrial septum, the ultrasound beam would go through the abdominal skin, the right atrial wall, and sometimes a small portion of the liver. The ultrasound beam could be positioned perpendicular to the atrial septum. Measurements of the thickness and size of the atrial septum, the motion of the atrial septum, the thickness of overlying tissues in the ultrasound beam pathway, and the distance between these tissues are reported in Table 2.
This study has demonstrated the feasibility of creating an ASD through epicardial application of histotripsy. The treatment targeting and monitoring are guided in real time by ultrasound imaging. The ASD creation is precise and reproducible.
Ultrasound Imaging Guidance
The treatment targeting is guided by a temporally changing, hyperechoic bubble cloud zone on ultrasound image. The location of the bubble cloud indicates the focal position and is used to target the focus to a preselected tissue volume such as the atrial septum. The hyperechoic bubble cloud also provides real-time feedback for monitoring the initiation and progress of the entire histotripsy procedure. The location and size of the ASD creation evaluated by ultrasound Doppler match those found by pathology. This real-time image feedback is unique for histotripsy and an essential feature for this noninvasive therapy technique.
Currently, the imaging probe is housed in the center hole of the therapy transducer, which has a 9-cm focal length. With this setup, the imaging probe needs to be 9 cm away from the target during the treatment. Because of the long standoff distance, a lower-frequency (5-MHz) imaging probe was used. As a result, the imaging quality during the treatment is less than desired. To use a higher-frequency imaging probe and improve the imaging quality, a therapy transducer with a shorter focal length is needed and is being developed.
The ASD generated by histotripsy in this study has a diameter of 2.6±0.4 mm on the left atrial surface and 5.0±1.1 mm on the right atrial surface, with an ≈1- to 4-mm rim of injury surrounding the ASD. We did not observe any clear trend for the dependence of the ASD diameter and the width of surrounding injury on the treatment time or motion of the atrial septum; however, only 1 acoustic parameter set was used here. Our previous in vitro experiments have shown that acoustic parameters can significantly affect the diameter and boundary condition of ASD.22,26 For example, shorter pulse duration and lower-duty cycle produce smaller ASD and smoother boundary. Further studies are planned to explore the acoustic parameter sets that can minimize the surrounding injury. In this study, the injury to the surrounding tissue may be underestimated because a 1-hour waiting time between treatment completion and euthanasia of the animal may not be sufficient for all potential inflammatory effects and myocardial injury to become recognizable by light microscopy. We also plan to perform short-term and long-term studies to evaluate whether this injury is reversible.
Although no clear trend was observed on the effect of septal motion on the surrounding injury, the septal motion probably contributes to generating the collateral damage by exposing surrounding septal tissue to the histotripsy ultrasound beam. In the canine model, the movement of atrial septum along the axial ultrasound beam is 8.1±3.3 mm. There is also movement along the lateral beam estimated up to 4 to 5 mm. In comparison, the axial and lateral movement of human HLHS patients is measured to be ≈1 to 2 mm. With less movement of the atrial septum in human neonates, it is expected that less collateral damage would be produced. At the same time, the critical structures in the neonatal heart are in closer proximity, and higher accuracy will be required.
To reduce the effect of septal motion and to further increase treatment accuracy, there are 2 possible solutions. First, one could synchronize the histotripsy treatment pulses with the patient’s heartbeat and respiration, with the pulses applied only when the atrial septum moves to the same location. This synchronization approach is straightforward to achieve and has been used for HIFU epicardiac ablation,15,17 but the treatment is prolonged because time is wasted when the atrial septum moves out of that location. The second approach uses motion tracking to maximize the treatment accuracy while minimizing the treatment time. The therapy focal position is steered electronically to track the atrial septum movement in real time27 and is technically challenging. We have designed and constructed phased-array therapeutic transducers that can electronically steer the transducer focus instantaneously. Other researchers have demonstrated that the movement of the atrial septum can be obtained from ultrasound imaging.28 We hope to develop and incorporate a motion-tracking algorithm in our future ASD applications.
In 1 case, slight hemorrhage was produced in the right free atrial wall. It occurred probably because some cavitation was induced on the atrial wall when the heart moved significantly. With less motion of the atrial septum in neonatal animals and motion-tracking method, we hope to eliminate such events.
Because histotripsy mechanically erodes the atrial septum, there is concern that the eroded tissue debris may become hazardous emboli, especially in single-ventricle physiology. We measured the size distribution of the tissue debris particles in a separate in vitro study.19 Measurements showed no particles >60 μm in diameter, and >80% of the particles were <6 μm. Such particles are unlikely to form hazardous emboli because 100-μm mechanical filters have been successful at preventing embolization in catheter-based thrombolysis procedures.29 There may, however, be a high risk for portions of newly developing mural thrombi at the site of tissue injury or necrotic tissue surrounding the ASD to embolize over time. To investigate the risk of embolization during treatment and over time, we plan to use transcranial Doppler during treatments and to perform postprocedure diffusion-weighted magnetic resonance imaging and pathology of the brain, lungs, and kidneys in our future short-term and long-term studies.
In the canine experiments studied here, open-chest surgery was performed, and the transducer was positioned outside the heart. Ultimately, to treat neonatal patients noninvasively, the ASD will be created with an extracorporeal application. Because ultrasound does not penetrate dense bone or air found in lungs, an ideal acoustic window to access the atrial septum would have to avoid these structures. Echocardiography from neonatal patients shows that such an acoustic window exists subcostally. The atrial septal length in infants with HLHS ranges from 11.2 to 17.4 mm (Table 2). This approach, however, requires a smaller therapy transducer with a shorter focal distance than the current transducer. The shorter standoff distance would allow a higher-frequency imaging probe and would provide better imaging quality for treatment targeting and monitoring. We are in the process of developing a 6×4-cm therapy transducer with a 5-cm focal distance. This therapy transducer will house an 8- to 12-MHz imaging probe and should be suitable for human neonatal use.
In the open-chest model, ultrasound has to penetrate only ≈1 cm of the right atrial free wall. In the human neonatal patients, using the subcostal window, ultrasound needs to go through ≈3- to 4-cm-deep tissue, including abdominal skin, liver, and the right atrial free wall, to reach the atrial septum (Table 2). Multiple layers of tissue with various acoustic properties would increase the ultrasound attenuation and aberration and affect the focal volume size and beam profile. Previous studies have shown that histotripsy can produce sufficiently high pressure through 3- to 4-cm overlying tissue and create accurate lesions in organs such as prostate.30 We have also started experiments in an intact neonatal pig model; early results show that histotripsy can produce accurate lesions through multiple layers of tissue, including neonatal ribs. We plan to continue our preclinical investigations and development of histotripsy in intact animal models.
We thank Dr Kimberly Ives, DVM, for assistance in performing the animal experiments. We also thank Dr Timothy L. Hall for help with construction of the therapy transducer driving system.
Sources of Funding
This research has been funded by grants from the National Institutes of Health R01-HL077629 and The Hartwell Foundation.
Abu-Harb M, Hey E, Wren C. Death in infancy from unrecognised congenital heart disease. Arch Dis Child. 1994; 71: 3–7.
Mahle W, Spray T, Wernovsky G, Gaynor J, Cark B. Staged reconstruction for hypoplastic left heart syndrome: a 15-year experience from a single institution. Circulation. 2000; 102: 136–141.
Tweddell JS, Hoffman GM, Mussatto KA, Fedderly RT, Berger S, Jaquiss RD, Ghanayem NS, Frisbee SJ, Litwin SB. Improved survival of patients undergoing palliation of hypoplastic left heart syndrome: lessons learned from 115 consecutive patients. Circulation. 2002; 106: 82–89.
Forbess JM, Cook N, Roth SJ, Serraf A, Mayer JE Jr, Jonas RA. Ten-year institutional experience with palliative surgery for hypoplastic left heart syndrome: risk factors related to stage I mortality. Circulation. 1995; 92: 262–266.
Vlahos AP, Lock JE, McElhinney DB, van der Velde ME. Hypoplastic left heart syndrome with intact or highly restrictive atrial septum: outcome after neonatal transcatheter atrial septostomy. Circulation. 2004; 109: 2326–2330.
Strickberger SA, Tokano T, Kluiwstra JU, Morady F, Cain CA. Extracardiac ablation of the canine atrioventricular junction using high intensity focused ultrasound. Circulation. 1999; 100: 203–208.
Ninet J, Roques X, Seitelberger R, Deville C, Pomar JL, Robin J, Jegaden O, Wellens F, Wolner E, Vedrinne C, Gottardi R, Orrit J, Billes MA, Hoffmann DA, Cox JL, Champsaur GL. Surgical ablation of atrial fibrillation with off-pump, epicardial, high-intensity focused ultrasound: results of a multicenter trial. J Thorac Cardiovasc Surg. 2005; 130: 803–809.
Otsuka R, Fujikura K, Abe Y, Okajima K, Pulerwitz T, Engel DJ, Muratore R, Ketterling JA, Kalisz A, Sciacca R, Marboe C, Yi G, Wang J, Homma S. Extracardiac ablation of the left ventricular septum in beating canine hearts using high-intensity focused ultrasound. J Am Soc Echocardiogr. 2007; 20: 1400–1406.
Duck FA. Physical Properties of Tissue. Burlington, Mass: Academic Press; 1990.
AIUM. Acoustic Output Labeling Standard for Diagnostic Ultrasound Equipment: A Standard for How Manufacturers Should Specify Acoustic Output Data, Revision 1. Laurel, Md: American Institute of Ultrasound in Medicine; 2008.
Although previous studies have reported successful cardiac tissue ablation with high-intensity focused ultrasound using a mechanism of thermal necrosis, our study is the first to report targeted tissue fractionation and creation of an atrial septal defect in a beating heart using extracardiac ultrasound through controlled acoustic cavitation, or histotripsy. In an open-chest canine model, histotripsy generated demarcated tissue destruction through the targeted region of atrial septum with no insult to other cardiac structures and modest damage to adjacent septal tissue. Continued advances in probe design and motion-tracking algorithms should minimize this collateral damage. Although further studies are necessary to determine the efficacy and safety of this technology in an intact neonatal animal model, this initial report provides the first evidence that histotripsy may be a useful, minimally invasive clinical tool to create atrial septal defects or other intracardiac communications in infants with congenital heart diseases.