Clinical Relevance of Intracranial Microembolic Signals in Patients With Left Ventricular Assist Devices
A Prospective Study
Background and Purpose The use of left ventricular assist devices has become an established method in bridging patients with end-stage cardiac failure to heart transplantation. Since thromboembolism is one of the major complications, we undertook this study to evaluate the clinical significance of Doppler microembolic signals (MES) in patients with left ventricular assist devices.
Methods Six patients with left ventricular assist devices were monitored for MES with transcranial Doppler ultrasonography during the first 30 postoperative days. Additionally, repeated (10 per day and patient) and prolonged (3 hours per patient) monitorings were performed to assess the adequacy of the 30-minute recordings. Three observers evaluated 30 randomly assigned monitorings in a blinded fashion to assess the interobserver variability. The relation between MES counts and clinical, radiological, hemostaseological, and pump flow parameters and the predictive value of MES counts regarding the occurrence of embolic events was evaluated.
Results Ten ischemic cerebrovascular accidents and 2 peripheral thromboembolic events occurred during the observation period of 177 days (total incidence, 6.8%). MES were found in 143 of 170 monitorings (84.1%). Their counts were significantly higher on days with clinically manifest embolic events as compared with event-free days (18.5 [3-74] versus 4 [0-52], respectively, median and 95% CI; P<.001, Mann-Whitney). The predictive value of MES counts above 7 per 30 minutes was high (75%). Significant differences in the incidence and counts of MES as well as in the incidence of clinically manifest embolic events were noted among the six patients (all P<.01) without equal differences in anticoagulant treatment or pump flow. Interobserver agreement was high (P=.78 to .89, unpaired Student’s t test). Considerable short- and long-term intrapatient variations of MES counts, without consistent pattern, were noted.
Conclusions Serial monitoring for MES is prognostically superior to single monitorings in patients with left ventricular assist devices. In the future, this new application mode may individually guide anticoagulation strategies and even influence the decision regarding early cardiac transplantation versus long-term use of the assist devices.
Cardiac transplantation has become an established therapy in the treatment of end-stage heart failure. While the number of donor organs has remained constant, the demand has dramatically increased due to aging of the population and increased survival after myocardial infarction.1 Approximately 20% to 30% of suitable transplantation candidates die while awaiting transplantation.2 3 Several different types of mechanical circulatory support, mainly consisting of LVAD, have been introduced to bridge patients to cardiac transplantation.4 Despite considerable improvements both in hemodynamic function and the material these devices are made of,5 thromboembolism remains a major concern during their use.3 6 7 No definite guidelines for anticoagulation therapy have yet been established in patients with LVAD,8 since the pathophysiology of their thromboembolic events is poorly understood. At present, no diagnostic tool is available to estimate the individual risk of thromboembolic complications.
Preliminary data from various stroke-prone cohorts suggest that incidence and numbers of MES detected with TCD provide relevant prognostic information in patients with cerebral occlusive disease9 10 11 and that this information can be used to assess the efficacy of anticoagulant treatment.12 13 To date, no reports have been published concerning the detection of MES in patients with LVAD. We undertook this study to evaluate (1) the incidence and number of MES in LVAD patients and their relation to (2) mode and intensity of anticoagulation, (3) hemostaseological parameters, (4) pump flow, (5) the occurrence of clinical embolic events, and (6) the occurrence of new ischemic lesions on CCT. Finally, the predictive value of the MES counts regarding embolic events was investigated.
Subjects and Methods
Patients and Description of Assist Device
From September 1994 to August 1995, the Novacor LVAD (Novacor Division, Baxter Healthcare Corp) was implanted in six patients awaiting cardiac transplantation in the Münster Department of Thoracic and Cardiovascular Surgery. All patients were listed transplant candidates and fulfilled the selection criteria for LVAD implantation described elsewhere.14
The Novacor N100 LVAD system is a wearable, electromagnetically actuated, implantable pump that consists of a seamless one-piece smooth polyurethane sac bonded to symmetrically opposed dual pusher-plates. The housing is a lightweight shell that incorporates bovine pericardial valves and an energy convertor. A percutaneous lead serves to connect the intracorporeal pump to the extracorporeal control console. In-flow and out-flow synthetic polyester textile fabric grafts, both perforating the diaphragm, connect the pump to the left ventricular apex and to the ascending aorta (Fig 1⇓). The maximum stroke volume of 70 mL provides blood ejection of up to 10 L per minute. The operation of the device in heart-synchronous counterpulsation permits frequencies of as much as 240 beats per minute.15
TCD Monitoring and MES Detection
TCD monitorings were performed using a pulsed Doppler ultrasonograph (Pioneer 4040, EME) with a 2 MHz probe. After identification of the MCA, the probe was fixed on the temporal skull with an elastic band to minimize movement artifact. All monitoring sessions were recorded on digital audiotapes for later reevaluation.
The criteria used to identify an embolus have been described elsewhere16 and had been discussed extensively among the observers. In short, high-intensity signals were characterized as MES if they fulfilled the following conditions: short duration (<0.15 and <0.3 second for MES appearing in systole and diastole, respectively), random appearance within the cardiac cycle, and intensity at least 3 dB above the background signal and characteristic sound.
Interobserver agreement was evaluated by randomly assigning 30 monitoring sessions of 30 minutes’ duration to the three observers participating in the study. Observers were blinded to the patient data and were asked to note both the total MES count and their exact position on the tape.
Additionally, prolonged (3 hours) and repeated (10 monitorings per patient on single days at 6, 8, and 10 am; at noon; at 2, 4, 6, 8, and 10 pm; and at midnight) bilateral monitorings were performed to assess the adequacy of the unilateral 30-minute monitoring sessions.
Preoperative diagnostic screening consisted of (1) evaluation of the extracranial and intracranial vasculatures by means of continuous-wave Doppler, color-coded duplex sonography, and TCD; (2) transesophageal echocardiography; (3) baseline CCT; (4) hematological examinations for preexisting coagulopathies; and (5) baseline bilateral TCD monitoring for MES.
The surgical technique was based on the method described by Portner et al.17 Heparin was initiated after LVAD implantation with a target PTT between 60 and 80 seconds. After stabilization of the patient’s condition (approximately on the 7th to 10th postoperative day), phenprocoumon was administered. Intravenous heparin was discontinued when the INR value exceeded 2.5. Target INR ranged between 2.5 and 4.0.
Postoperative studies were performed during the first 30 postoperative days on the basis of a detailed protocol entailing (1) daily unilateral TCD monitoring at the same time of the day for 30 minutes; (2) daily calculation of cardiac index by averaging the minimal and maximal stroke volumes registered on-line during each monitoring session; (3) transthoracic echocardiography that was routinely performed on the 14th and 30th postoperative days and additionally immediately (within 24 hours) after thromboembolic complications; (4) CCT a few days after LVAD implantation, within 6 hours of transplantation, and 3 to 5 days after the development of acute neurological symptoms; and (5) daily determination of PTT, INR, antithrombin III, fibrinogen, hematocrit, and platelet count, for which venous blood was drawn immediately after each TCD monitoring session. In accordance with the guidelines of our institution, patients were classified as efficiently anticoagulated only on days in which their INR and/or PTT were higher than 2.5 or longer than 60 seconds, respectively. The prevalence of such days was used as an index of the individual overall anticoagulation efficacy (anticoagulation efficacy equals number of days in which the patient was adequately anticoagulated divided by the total number of follow-up days).
Ischemic cerebrovascular accident was defined as focal neurological deficit of sudden onset, after exclusion of intracranial hemorrhage by means of CCT. Systemic embolism was defined as an abrupt vascular insufficiency of the limbs or internal organs associated with clinical or radiological evidence of arterial occlusion, in the absence of previous obstructive disease.
Predictive Value of MES Counts
The specificity and sensitivity of MES counts in predicting the occurrence of an embolic event were evaluated. The hypothesis was that embolic events would occur only when MES counts exceeded a specified “cutoff” value, which was calculated on the basis of ROC curves. These curves show the various trade-offs existing between proportions of true-positive and false-positive responses as the cutoff point is varied,18 thus enabling an accurate definition of the value providing the highest possible combination of sensitivity (number of embolic events predicted divided by total number of embolic events) and specificity (number of event-free days predicted divided by total number of event-free days).
Mann-Whitney and Kruskal-Wallis tests were used for nonnormally distributed data and unpaired Student’s t test for normally distributed data. Distribution of frequencies was evaluated with the χ2 test. Interobserver agreement was assessed by performing unpaired t tests between all possible observer pairs. Spearman-rank test was used to evaluate correlations. Significance was declared at a level of P<.05.
The only pathological findings disclosed on preoperative screening were a mild stenosis (50%) of the left internal carotid artery and an old lacunar infarction in the left basal ganglia on CCT scan in patient 2. Hematological and cardiological examinations provided no evidence of preexisting coagulopathy or intracardiac thrombus, respectively. No MES were detected preoperatively in any of the six patients.
Twelve embolic events occurred during the observation period of 177 days (overall incidence, 6.8%), 10 of which affected the cerebrovascular system and 2 the peripheral arteries (Table 1⇓). Of the 10 cerebral embolic events, 7 affected the anterior circulation (MCA, n=6; ophthalmic artery, n=1) and 3 the posterior circulation lodging in the basilar (n=2) or posterior cerebral artery (n=1). All cerebral embolic events were transient (duration of symptoms between 30 minutes and 8 hours) except for 1 fatal basilar embolism that occurred in patient 4 on the 24th postoperative day. Peripheral embolic events consisted of 2 transient occlusions of the popliteal artery. Management in both cases was conservative. CCT scan of patient 4 on the 25th postoperative day revealed bilateral thalamic and left-sided occipital infarctions. CCT scans of the remaining patients failed to disclose new ischemic lesions.
MES were detected in 143 of 170 monitoring sessions (84.1%). Their incidence and counts were significantly different among patients (total P<.01, Kruskal-Wallis; Table 1⇑). The number of MES was significantly higher on days with clinically manifest embolic events as compared with event-free days (18.5 [3-74] versus 4 [0-52] MES, respectively, median and 95% CI; P<.001, Mann-Whitney). No correlation between the incidence of embolic events and median MES counts was noted (r=.21, P>.05).
According to the ROC curves, the best cutoff value was 7 MES. Use of this value provided a sensitivity of 75% (9 of 12 embolic events occurred while the count of MES was higher than 7) and a specificity of 74% (MES counts were lower than 7 in 124 of the 165 event-free days) in predicting the occurrence of embolic events (Fig 2⇓).
Significant differences in the efficacy of anticoagulation were noted among the 6 patients (P<.01, χ2 test). No relation between these differences and MES counts, or incidence of embolic events, could be established. Six of the 12 embolic events (50%) occurred while patients were adequately anticoagulated. Significant differences among the 6 patients were also noted in hemostaseological parameters (days in therapeutic range, hematocrit, platelet counts, fibrinogen, and antithrombin III) and pump flow, again without apparent relation to clinical course or MES incidence and counts (Table 1⇑).
Repeated and prolonged monitorings revealed a considerable intraindividual variability of MES counts (Table 2⇓ and Fig 3⇓). SDs ranged between 27% and 120% of the corresponding mean, which were higher in patients with lower MES counts. No significant differences were found between MES counts over the right and left MCAs in single cases (P=.3 to .85, Mann-Whitney) and in the overall assessment (MES counts 2.5 [2-4] and 2.5 [1.5-3.5] in the left and right MCAs, respectively, median and 95% CI; P=.47, Mann-Whitney). Statistical evaluation of the day-profiles disclosed no significant differences in MES counts among the different monitoring times (P=.95, Kruskal-Wallis).
Interobserver agreement was high (observer 1 versus observer 2, P=.88; observer 1 versus observer 3, P=.89; and observer 2 versus observer 3, P=.78; all unpaired t tests).
The discrepancy between the growing number of patients awaiting cardiac transplantation and the scarcity of donor organs has led to the increased use of mechanical cardiac assist devices to bridge prolonged waiting periods.3 6 19 20 These devices, originally used short-term in cases of postoperative cardiac failure,21 22 are currently being implanted for as long as 1.5 years in individual cases.23 Besides infection, thromboembolism and bleeding are the main complications of LVAD use that preclude transplantation.5 24 Since the occurrence of the latter is strongly influenced by mode and intensity of anticoagulation, this observation emphasizes the need for clinically based anticoagulation guidelines, which are still lacking. Concerning the Novacor LVAD, several authors recommend an effective anticoagulation3 19 20 ; in contrast, others assess this regimen as excessive due to the containment of bioprosthetic valves in this device.8 No diagnostic test can assess the individual embolic risk of LVAD patients. The identification of high-risk patients could lead to more-intensive anticoagulation or even indicate their need for early transplantation. This is of particular interest in these patients, since due to limited organ preservation time, the selection of organ recipients is not solely based on the degree of compatibility of the major histocompatibility gene complex.25
Our initial results suggest that serially performed MES detection by means of TCD can provide crucial prognostic information on the individual embolic risk of LVAD patients. The predictive value of MES counts was astonishingly high (75%).
The embolic risk of LVAD patients has been associated with various technical or clinical parameters, including pump flow,26 fibrinogen concentration,27 factor XIII concentration,27 and anticoagulation regimen.28 In our study, however, 6 of the 12 embolic events occurred while the patients were adequately anticoagulated (at least according to the therapeutic regimen applied in our department). Additionally, no relation between clinically manifest embolic events and hemostaseological parameters or pump flow was noted. These data indicate that these parameters alone are insufficient surveillance markers in patients with LVAD. The lacking relation between hemostaseological parameters, anticoagulant treatment, and MES count further suggests that at least some of the detected MES are not caused by full blood clots. d-dimer levels would provide more conclusive evidence on this matter. We have already included d-dimer as another parameter in our ongoing study. Platelet counts did not influence the amount of MES in our study. Since platelet activation studies were not performed, a pathogenetic role of platelet aggregates cannot be excluded in this setting.
Since our repeated and prolonged measurements revealed considerable short-term and long-term intrapatient variabilities of MES counts, single monitorings would probably have failed to provide conclusive information on the individual embolic risk. Thus, our study clearly demonstrated the superiority of serial over single monitoring sessions. The advantages of this promising MES detection mode should therefore be investigated in other stroke-prone patient groups. Serial and long-term monitorings could be facilitated by the application of smaller, more lightweight TCD devices and reliable automated detection methods29 30 31 to avoid loss of patient compliance and limit personnel requirements, respectively.
In conclusion, our preliminary results suggest that serial TCD monitoring for MES provides prognostic information in patients with LVAD. In the future, this new application mode could be used to monitor anticoagulant treatment and guide therapeutic decisions, in particular concerning early transplantation or prolonged LVAD use. Still, these results remain to be evaluated by large-scale studies.
Selected Abbreviations and Acronyms
|CCT||=||cerebral computed tomography|
|INR||=||International Normalized Ratio|
|LVAD||=||left ventricular assist devices|
|MCA||=||middle cerebral artery|
|PTT||=||partial thromboplastin time|
|TCD||=||transcranial Doppler ultrasound|
- Received December 18, 1995.
- Revision received February 12, 1996.
- Accepted February 15, 1996.
- Copyright © 1996 by American Heart Association
Annual report of the US Scientific Registry for organ transplantation and the organ procurement and transplantation network—1990. Es 8.-9. Washington DC: US Department of Health and Human Services.
Rowles JR, Mortimer BJ, Olsen DB. Ventricular assist and total artificial heart devices for clinical use in 1993. ASIAO Transactions. 1993;39:840-855.
Watson JT. Innovative ventricular assist system. ASIAO Transactions. 1994;40:M902.
Kanter KR, McBride LR, Pennington DG, Swartz M, Ruzevich SA, Miller LW, Willman VL. Bridging to cardiac transplantation with pulsatile ventricular assist devices. Ann Thorac Surg. 1988;46:130-140.
Mason RG, Mohammad SF, Chuang HYK. Artificial devices in clinical practice. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and Thrombosis: Principles and Clinical Practice. Philadelphia, Pa: JB Lippincott Co, 1982:944-953.
Szukalski EA, Reedy JE, Pennington DG, Swartz MT, McBride LR, Miller LW. Oral anticoagulation in patients with ventricular assist devices. ASIAO Transactions. 1990;36:M700-M703.
Siebler M, Nachtmann A, Sitzer M, Rose G, Kleinschmidt A, Rademacher J, Steinmetz H. Cerebral microemoblism and the risk of ischemia in asymptomatic high-grade internal carotid artery stenosis. Stroke. 1995;26:2184-2186.
Babikian VL, Hyde C, Pochay V, Winter MR. Clinical correlates of high-intensity transient signals detected on transcranial Doppler sonography in patients with cerebrovascular disease. Stroke. 1994;25:1570-1573.
Tegeler CH, Burke GL, Dalley GM, Stump DA. Carotid emboli predict poor outcome in stroke. Stroke. 1993;24:186. Abstract.
Georgiadis D, Hill M, Zunker P, Stögbauer F, Ringelstein EB. Anticoagulation monitoring with transcranial Doppler. Lancet. 1994;344:1373-1374. Letter.
Consensus Committee of the Ninth International Symposium on Cerebral Hemodynamics. Basic identification criteria of microembolic signals. Stroke. 1995;26:1123.
Swets JA, Pickett RM, Whitehead SF, Getty DJ, Schnur JA, Swets JB, Freeman BA. Assessment of diagnostic technologies. Science. 1979;205:753-759.
Vetter OH, Kaulbach HG, Schmitz C, Forst A, Überfuhr P, Kreuzer E, Pfeiffer M, Brenner P, Oliver D, Reichart P. Experience with the Novacor left ventricular assist system as a bridge to cardiac transplantation, including the new wearable system. J Thorac Cardiovasc Surg. 1995;109:74-80.
Pae WE, Rosenberg G, Donachy JH, Phillips WM, Parr GL, Prophet GA, Pierge WS. Mechanical circulatory assistance for postoperative cardiogenic shock: a three-year experience. ASIAO Transactions. 1980;26:256-260.
Spencer RC, Eisman B, Trinkle JK, Rossi NP. Assisted circulation for cardiac failure following intracardiac surgery with cardiopulmonary bypass. J Thorac Cardiovasc Surg. 1965;49:56-72.
Meyers TJ, Dasse KA, Macris MP, Poirier VL, Cloy MJ, Frazier OH. Use of a left ventricular assist device in an outpatient setting. ASIAO Transactions. 1994;40:M471-M475.
Termulen DF, Swartz MT, Pennington DG, McBridge LR, Szukalski EA, Reedy JA, Ruzevich SA. Thrombembolic complications with the Pierce-Donachy ventricular assist device. ASIAO Transactions. 1989;35:616-618.
Kormos RL, Borovetz HS, Pristas JM, Armitrage JM, Stuart RS, Marrone GC, Hardesty RL, Griffith BP. LVAS pump performance following initiation of left ventricular assistance. ASIAO Transactions. 1990;36:M703-M705.
Solen KA, Mohammad SF, Burns GL, Pantalos GM, Peng YKY, Pitt WG, Reynolds LO, Olsen DB. Markers of thromboembolisation in a bovine ex vivo left ventricular assist device model. ASIAO Transactions. 1994;40:M602-M608.
Takahama T, Kanai F, Hiraishi M, Onishi K, Yamazaki Z, Naruse Y, Furuse A, Yoshitake T. Combined administration of protease inhibitor and thromboxane A2 synthetase inhibitor for anticoagulation of a left ventricular assist device. ASIAO Transactions. 1990;36:M141-M144.
Markus HS, Loh A, Brown MM. Computerized detection of cerebral emboli and discrimination from artifact using Doppler ultrasound. Stroke. 1993;24:1667-1672.
Georgiadis D, Goeke J, Hill M, König M, Nabavi DG, Stögbauer F, Zunker P, Ringelstein EB. A novel technique for identification of Doppler microembolic signals based on the coincidence method: in vitro and in vivo evaluation. Stroke. 1996;27:683-686.