The increasing number of patients affected by endstage

Left Ventricular Support by Axial Flow Pump: The Echocardiographic Approach to Device Malfunction Emanuele Catena, MD, Filippo Milazzo, MD, Emanuela Montorsi, MD, Giuseppe Bruschi, MD, Aldo Cannata, MD, Claudio Russo, MD, Alberto Barosi, MD, Giuseppe Tarelli, MD, Paolo Tartara, MD, Roberto Paino, MD, and Ettore Vitali, MD, Milan, Italy Axial flow pumps have gained increased acceptance in recent years as a bridge to heart transplantation and, more recently, as destination therapy. As left ventricular (LV) assist device dysfunction will be increasingly prevalent, the aim of our work was to introduce an echocardiographic management protocol as a guide to recognize the causes of pump failure. In this article we describe the echocardiographic approach to 5 episodes of malfunction of an axial flow pump (DeBakey, MicroMed Technology Inc, Houston, Tex) in 4 patients: 4 episodes caused by thrombosis of LV assist device and one caused by abnormal increase of systemic vascular resistance. In our experience, echocardiography played a pivotal role in clinical management of LV assist device failure. It allowed us to: assess patency and position of inflow and outflow cannulae; research the source of thromboembolic material; assess adequate LV filling and unloading; and optimize right ventricular function, volume replacement therapy, and pharmacologic support. (J Am Soc Echocardiogr 2005;18: 1422.e7-e13.) The increasing number of patients affected by endstage cardiac failure refractory to inotropes and intraaortic balloon pump, along with a scarcity of heart donors, encouraged the use of left ventricular (LV) assist device (LVAD) as a bridge to transplantation. 1 In addition to the clinical experience with pulsatile devices, axial flow pumps have gained increased acceptance since their first clinical use in 1998. 2-7 Their peculiarities are small size and absence of a compliance chamber and prosthetic valves. Currently, LVAD use is becoming widespread as a clinically acceptable alternative to cardiac transplantation such that duration of support is becoming extended and pump dysfunction will be increasingly prevalent. 8 Previous reports described the use of transesophageal echocardiography (TEE) as the procedure of choice for evaluation of patients undergoing LVAD support. 9-12 The aim of our work was to introduce an echocardiographic management protocol as a guide to recognize the causes of axial pump failure. In this article we describe the echocardiographic approach to 5 episodes of malfunction of an axial flow pump (DeBakey, MicroMed Technology Inc, Houston, Tex) in 4 patients. PUMP The study LVAD is a small implantable axial flow pump. It is designed to achieve 5 L/min of blood continuous flow. The pump is connected to the LV apex by a titanium inflow cannula and to the aorta by a vascular graft. The pump housing unit contains the impeller (the only moving part) and the motor (Figure 1). LVAD implantation requires median sternotomy and cardiopulmonary bypass. The pump is placed into a small abdominal pocket. A pump motor cable is exteriorized above the right iliac crest and connected to the external control system. The control system consists of a small controller and two batteries. Pump flow, pump speed, power consumption, and current signal values are displayed in a bedside monitor unit that can be used to modify the pump speed. From the Divisions of Cardiothoracic Anesthesia and Intensive Care (E.C., F.M., E.M., R.P.) and Cardiac Surgery, Department of Cardio-Thoracic Surgery A. De Gasperis, Niguarda Cà Granda Hospital. Reprint requests: Emanuele Catena, MD, Via Annunzio Cervi n 4, 20161 Milan, Italy (E-mail: emanuele.catena@tin.it). 0894-7317/$30.00 Copyright 2005 by the American Society of Echocardiography. doi:10.1016/j.echo.2005.04.011 PATIENTS From April 2000 to December 2004, 19 patients with end-stage heart failure underwent to assist device implant procedure at our institution. Two patients died while on LVAD support because of multiple organ failure. The remaining patients recovered rapidly from the operation and were dis- 1422.e7

1422.e8 Catena et al December 2005 Figure 1 Left ventricular (LV) assist device. Axial flow pump is connected to LV apex by titanium inflow cannula (left) and to aorta by vascular outflow graft (right). charged from the intensive care department (ICU) to the regular ward. After physical retraining, 8 patients were able to return to their normal daily activities. During LVAD support 4 patients (21%) experienced 5 episodes of device failure: 4 patients had a suggested thrombosis of the pump, one patient had a first episode of pump failure caused by an abnormal increase of systemic vascular resistance, and a following one caused by suggested thrombosis. In all patients we adopted the coagulation management and anticoagulation treatment based on a previously proposed protocol. 13 Case 1 The first patient, on the 112th day of support, was admitted to our institution for fever and increasing dyspnea. The pump controller indicated an abnormal increase in current (0.6-1.4 A) and power (7.8-15 W), changes suggestive of outflow LVAD obstruction or increasing friction on the pump impeller possibly caused by pump thrombosis. We started intravenous heparin administration promptly. TEE showed no endoventricular thrombus image or other sources of thromboembolic material. Obstruction or kinking of the outflow graft was excluded (Figure 2). A normal low-flow velocity through inflow and outflow cannulae was detected (Figure 3). The patient s condition continued to deteriorate and LVAD flow rate decreased from 4.5 to 2.2 L/min. Therefore, a thrombosis of impeller was suggested. Then, we started thrombolysis by direct delivery of recombinant tissue plasminogen activator (12, 10, and 10 mg over 15 minutes) into the LV. After 20 minutes, flow rate began to increase. After 3 hours, current and power values reached normal ranges, with flow rate normalization and complete hemodynamic recovery. Figure 2 Transesophageal long-axis view shows nonobstructed outflow graft stitched on ascending aorta. Case 2 The second patient, on the 105th day of support, after a 7-day history of fever, was admitted to the ICU with low pump flow and high current consumption, respiratory failure, anuria, and poor peripheral perfusion. His mean arterial pressure was greater than 130 mm Hg. An urgent TEE examination showed poor LV emptying by LVAD, moderate mitral regurgitation, and minimal aortic valve opening. No endoventricular thrombus or other causes of obstruction were observed. Pulsed wave Doppler interrogation documented continuous flow through outflow cannula with an increased pulsatility (peak velocity 2.2 m/s) suggestive of abnormal afterload increase (Figure 4). High-dose vasodilators (nitroprusside 0.55 g/kg/min) were instituted and after

Volume 18 Number 12 Catena et al 1422.e9 Figure 3 Color Doppler demonstration of normal low-velocity flow through inflow (left) and outflow (right) cannula (top) and continuous (left) and pulsed (right) wave Doppler flow profile (bottom). Figure 4 Pulsed wave Doppler demonstration of continuous flow through outflow graft with increased pulsatility (peak velocity 2.2 m/s) typical of abnormal afterload increase. 30 minutes pump flow increased with a rapid improvement of the patient s hemodynamics and peripheral perfusion. The same patient, on the 150th postoperative day, was admitted to the ICU because of pump arrest and cardiogenic shock. A few days before, the pump controller indicated short and sudden abnormal increases of current and power. Mechanical ventilation and inotropic support with high-dose epinephrine were instituted. TEE showed a very poor fractional shortening with minimal aortic valve opening. Color and pulsed wave Doppler demonstrated a pulsatile low-velocity retrograde flow from aorta (through outflow cannula) to LV (through inflow cannula) (Figure 5). Velocitytime integral of retrograde systolic flow was calculated for 3 cycles running with sample volume Figure 5 Pulsed wave Doppler demonstration of pulsatile low-velocity retrograde flow through outflow (top) and inflow (bottom) cannula during pump arrest.

1422.e10 Catena et al December 2005 Figure 6 Transesophageal 4-chamber view shows thrombotic material in small pocket next to apical orifice of inflow cannula and interventricular septum. placed at inflow cannula orfice (11.1 cm). Crosssectional area of inflow cannula was well known (2.0 cm 2 ). By means of the following formula, we calculated flow retrograde flow (2.6 L/min): flow (cm 3 /min) cross-sectional area (cm 2 ) velocitytime integral (in centimeters) heart rate (L/min). Critical condition persisted despite maximal inotropic support. Thrombolysis by direct delivery of recombinant tissue plasminogen activator into the LV was unsuccessful so that pump replacement was required. Unfortunately, the hemodynamic status progressively worsened and the patient died in the operating department during device replacement. Case 3 The third patient, on the 244th postoperative day, was admitted to our institution because urinary tract infection was diagnosed and the pump controller indicated a continuous abnormal increase in current (1.5 A) and power (16 W). During TEE control, thrombotic material had been detected in the small pocket next to the apical orifice of inflow cannula (Figure 6). We decided to increase the heparinemia levels despite an adequate anticoagulation profile. After 4 hours, current and power values reached normal ranges. The patient was placed on high urgency transplant list and successfully transplanted after 48 hours. The presence of apical thrombotic material was confirmed by intraoperative inspection. Case 4 The fourth patient, on the 122th postoperative day, was admitted to the ICU for pump arrest after a few hours with power and current increases. Because of sudden hemodynamic deterioration, the patient was transferred to the operating department with profound hypotension (mean arterial pressure 40 mm Figure 7 Transesophageal 4-chamber view shows right ventricle (RV) enlarged and left ventricular (LV) chamber collapsed around inflow cannula. Hg), respiratory failure, and anuria, and he underwent emergency LVAD replacement. Early postoperative course was complicated by persisting critical conditions. TEE evidenced a severe right ventricle (RV) hypokinesia and dilation with a very low RV fractional area change ( 15%) unresponsive to maximal inotropic support (Figure 7). The patient developed refractory shock and multiorgan failure and died on the second postoperative day. ECHOCARDIOGRAPHIC ASSESSMENT TEE is an essential tool in the care of patients undergoing LVAD. We adopted TEE for serial evaluation before, during, and after LVAD implantation. Before implantation, echocardiography is usually applied to accomplish the following: (1) search for atrial septal defects or patent foramen ovale and aortic regurgitation because the former may create right-to-left shunt, systemic desaturation, or paradoxical embolus and the latter impedes LV emptying; (2) diagnosis of mitral stenosis that would limit LVAD filling; (3) evaluation of ascending aorta to exclude aneurysmal dilatations and atherosclerotic damage of the aortic wall where pump outflow cannula has to be connected; (4) search for apical thrombotic material into LV apex where pump inflow cannula has to be inserted; and (5) evaluation of RV function by anterior free wall motion and RV fractional area change. These parameters are reliable indexes of RV function and of great clinical significance because adequate LVAD circulatory support must be warranted by adequate transpulmonary blood flow.

Volume 18 Number 12 Catena et al 1422.e11 In the operating department, after implantation, TEE allows the documentation of LVAD function. Echographic signs of good LVAD function are the following: (1) neutral septum position indicating adequate LV filling; (2) inflow cannula correctly oriented toward the mitral valve without abutting any wall; (3) slight or absent mitral regurgitation; (4) closed aortic valve; and (5) detection of unidirectional continuous low-velocity flow by means of color Doppler and continuous flow with slight pulsatility by means of pulsed wave Doppler through inflow and outflow cannulae. On the other side, markers of LVAD malfunction are: poor LV emptying by LVAD as indicated by rightward septum shift, mitral valve regurgitation degree /IV or more, minimal aortic valve opening, and spontaneous contrast echo into the left atrium and LV. Achieving hemodynamic stability is the main goal during postoperative intensive care of patients with LVAD. The most common reasons for hemodynamic instability are: hypovolemia, RV dysfunction, and cardiac tamponade. All the above-mentioned events may be well documented by TEE examination. The importance of serial evaluation of RV function cannot be overemphasized because adequate LVAD function depends on normal or high LV filling pressure that, in turn, depends on RV output. Qualitative and quantitative echographic assessment of RV function is crucial in optimizing volume replacement therapy and titrating pharmacologic support. During a long-term course it is imperative that the cause of pump malfunction be identified and corrected as soon as possible and all attempts to restart the pump must be made immediately. TEE is an invaluable tool to urgently assess cardiac function and LVAD components during malfunction because of its ability to image anatomic structures at bedside. At our institution, we developed an echocardiographic management protocol for fast evaluation of patients showing axial flow LVAD malfunction, coupled with device controller informations. Low Pump Flow with Increase in Current and Power Values When the controller shows an increase in current and power values with decreases in flow, an obstruction (of outflow cannula or impeller) or an abnormal afterload increase must be suggested. Despite technical and clinical improvements, thrombosis and thromboembolism are still reported for patients on mechanical support. 14 The hypothesis is that LVAD support induces a state where procoagulant and prothrombotic forces delicately balance one another. Investigators have described the induction of activated coagulation and fibrinolytic cascades despite normal coagulation values and platelet counts during LVAD support. 15,16 Moreover, if a stress situation is imposed, such as infection, sepsis, Figure 8 Top, Small pocket next to left ventricular assist device inflow cannula orifice and interventricular septuminferior wall. Bottom, Pulsed wave Doppler interrogation with simple volume placed into this area shows very low velocity ( 20 cm/s). ongoing bleeding, or an operation, the balance can be tipped and result in thromboembolic events. This complication was previously reported in patients with a ventricular assist device, and in some cases, the device was urgently replaced. 4 In our experience, 2 patients had pump dysfunction caused by thromboembolic material and two others cases were suspected. In all cases the premonitory signs of pump malfunction were increases in current and power along with sudden and short episodes of decreases in flow rate. It is noteworthy that 3 patients showed infection some days before pump failure. In the event that the LVAD malfunctions or stops operating, TEE is mandatory to research outflow cannula obstruction and any source of thromboembolic materials. Intracavity thrombus research. Left atrial appendage and LV apex must be examined carefully. In our experience, thrombus was detected (case 3) in the small pocket next to the LVAD inflow cannula orifice and interventricular septum-inferior wall. In all patients pulsed wave Doppler interrogation with simple volume placed into this area showed very low blood velocity (Figure 8).

1422.e12 Catena et al December 2005 Outflow line assessment. Midesophageal longitudinal view at 100 to 120 degrees shows outlet cannula perpendicularly stitched on the anterior wall of the ascending aorta. Normal blood outflow appears as a low-velocity flow at color Doppler and pulsed wave Doppler documents an unidirectional slightly pulsatile flow with peak velocity usually ranging from 1.0 to 2.0 m/s. Cannula obstruction can be assessed by means of color Doppler as high-velocity aliased flow at the cannula orifice with manifest convergence area. Cannula kinking causes the loss of Doppler signal in any echocardiographic view. Impeller assessment. The pump s echogenicity does not allow the assessment of its interior with echocardiographic examination. Intravascular ultrasound 17 cannot be used to visualize the interior of axial blood pump because the catheter cannot be advanced into the device through the inflow port without interfering with the pump function. After TEE exclusion of thrombi in the LV, in the vascular graft, or at the aortic anastomosis, it is mandatory to suspect impeller thrombosis and eventually to perform immediate administration of low-dose thrombolysis into LV. This approach eliminates the risks associated to high-dose systemic thrombolysis before device replacement. In the event that the LVAD stops operating, echocardiography allows documentation and calculation of retrograde flow from the ascending aorta, through the ouflow cannula pump inflow cannula, to the left ventricle. Afterload assessment. Intense vasoconstriction is often a compensatory mechanism in high oxygen demand states (eg, infection, anemia) to maintain central perfusion. It can result in sudden pump failure and poor peripheral perfusion. In this case, TEE shows poor LV emptying by LVAD as indicated by rightward septum shift, mitral valve regurgitation (degree /IV), and minimal aortic valve opening, and pulsed wave Doppler interrogation documents continuous flow with an abnormal increased pulsatility (peak velocity 2.0 m/s). Increases of systemic vascular resistance have to be avoided during axial flow pump support. Hypertensive crisis may be treated with a short-acting vasodilator (nitroprusside) to maintain mean arterial blood pressure less than 85 to 90 mm Hg until complete hemodynamic recovery. Low Pump Flow with Normal Current and Power Values When the controller shows low pump flow with normal values of current and power, the following items must be assessed. Preload assessment. LV filling can be compromised by hypovolemia or RV failure. When RV failure is severe, the RV is depicted as enlarged in the 4-chamber view and LV chamber may collapse around the inflow cannula. During LVAD assistance adequacy of RV performance is crucial. The onset of RV dysfunction in these patients is heralded by poor filling of the LV and a severe decrease of pump flow. Adequate and stable RV function must be achieved by titrating drug therapy according to wall-motion analysis, RV fractional area change, modifications of end-diastolic RV diameters/volumes, and tricuspidal regurgitation. Leftward septum shift may suggest poor RV performance needing an increase in inotropic support and pulmonary vasodilators. Inflow line assessment. TEE 4-chamber midesophageal view is essential to assess the correct positioning of inflow conduit. The inlet cannula, introduced through the apex, has to stay central into the LV, not abutting any wall. When pump achieves 5 L/min of blood flow with a motor speed of 10,000 rpm with a proper blood drainage, detection of blood flow through inlet cannula by means of color Doppler shows unidirectional continuous low-velocity flow and pulsed wave Doppler a continuous flow with slight pulsatility. In our experience peak filling velocity patterns are variable (1.0-2.0 m/s) according to preload and to the remaining pumping action of the patient s heart. Cannula obstruction can be assessed by means of color Doppler as high-velocity aliased flow at the cannula orifice with manifest convergence area. Cannula kinking causes the loss of Doppler signal in any echocardiographic views. Conclusion TEE is an invaluable tool for treatment of patients with axial flow pump malfunction. It allows research for the source of thromboembolic materials and assessment of patency and position of inflowoutflow cannulae. Moreover, it allows assessment of LV unloading and RV function to optimize volume replacement therapy and pharmacologic support. Echocardiography is pivotal because it offers rapidity of device performance assessment and quantification. REFERENCES 1. Boehmer JP. Device therapy for heart failure. Am J Cardiol 2003;91:53-9. 2. Wieselthaler GM, Schima H, Heismayr M, Pacher R, Laufer G, Noon GP, et al. First clinical experience with DeBakey LVAD continuous-axial-flow pump for bridge to transplantation. Circulation 2000;101:356-9. 3. Noon GP, Morlry D, Irwin S, Benkowski R. Development and clinical application of the MicroMed DeBakey LVAD. Curr Opin Cardiol 2000;15:166-71. 4. Noon GP, Morlry D, Irwin S, Abdelsayed SV, Benkowski RJ, Lynch BF. Clinical experience with the MicroMed DeBakey ventricular assist device. Ann Thorac Surg 2001;71:133-8. 5. Grinda JM, Latremouille CH, Chevalier P, D Attelis N, Boughenou F, Guillemain R, et al. Bridge to transplantation

Volume 18 Number 12 Catena et al 1422.e13 with the DeBakey VAD axial pump: a single center report. Eur J Cardiothorac Surg 2002;22:965-70. 6. Vitali E, Lanfranconi M, Ribera E, Bruschi G, Colombo T, Frigerio M, et al. Successful experience in bridging to heart transplantation with the MicroMed DeBakey ventricular assist device. Ann Thorac Surg 2003;75:1200-4. 7. Goldstein DJ. Worldwide experience with the Micromed De- Bakey ventricular assist device as a bridge to transplantation. Circulation 2003;108:272-7. 8. Horton SC, Khodaverdian R, Powers A, Revenaugh J, Renlund DG, Moore SA, et al. Left ventricular assist device malfunction: a systematic approach to diagnosis. J Am Coll Cardiol 2004;43:1574-83. 9. Scalia GM, McCarthy PM, Savage RM, Smedira NG, Thomas JD. Clinical utility of echocardiography in the management of implantable ventricular assist devices. J Am Soc Echocardiogr 2000;13:754-63. 10. Barzilai B, Davila-Roman VG, Eaton NM, Rosenbloom M, Spray TL, Wareing TH, et al. Transesophageal echocardiography predicts successful withdrawal of ventricular assist devices. J Thorac Cardiovasc Surg 1992;104:1410-6. 11. Simon P, Owen AN, Moriz A, Rokitansky A, Laczovics A, Wolner E, et al. Transesophageal echocardiographic evaluation in mechanically assisted circulation. Eur J Cardiothorac Surg 1991;5:492-7. 12. Davila-Roman VG, Barzilai B. Transesophageal echocardiographic evaluation of patients receiving mechanical assistance from ventricular assist devices. Echocardiography 1997;14: 505-11. 13. Szefner J. Control and treatment of hemostasis in cardiovascular surgery: the experience of La Pitie Hospital with patients on total artificial heart. Int J Artif Organs 1995;18:633-48. 14. Stevenson LW, Kormos RL, Bourge RC, Gelijns A, Griffith BP, Hersberger RE, et al. Mechanical cardiac support 2000: current applications and future trial design. J Am Coll Cardiol 2001;37:340-70. 15. Livingston ER, Fisher CA, Bibidakis EJ, Pathak AS, Todd BA, Furukawa S, et al. Increased activation of the coagulation and fibrinolytic systems leads to hemorrhagic complications during left ventricular assist implantation. Circulation 1996;94: 227-34. 16. Spanier T, Oz M, Levin H, Weinberg A, Stamatis K, Stern D, et al. Activation of coagulation and fibrinolytic pathways in patients with left ventricular assist devices. J Thorac Cardiovasc Surg 1996;112:1090-7. 17. Leibowitz DW, Levin HR, Weslow RG, Slater J, Di Tullio MR, Schwartz A, et al. Intravascular ultrasound imaging of the heart-mate 1000 IP left ventricular assist device. J Am Soc Echocardiogr 1994;7:624-30.