Measurement of Hemostatic Indexes in Conjunction With Transcranial Doppler Sonography in Patients With Ventricular Assist Devices
Background and Purpose—Clinical thromboembolism (TE) remains an impediment to the chronic application of ventricular assist devices (VADs). Microembolic signals (MES) have been detected by transcranial Doppler ultrasound (TCD) in patients with VADs, although their origin and relation to TE remain undefined. We have investigated the hypothesis that hemostatic alterations are related to MES and that MES are associated with TE in a group of 27 VAD patients.
Methods—Indexes of coagulation, fibrinolysis, and cellular activation and aggregation were measured before and during the VAD implantation period in conjunction with TCD. Groups were defined on the basis of presence of MES, degree of MES showering, and incidence of TE.
Results—MES were observed in 67 (58%) of 115 of individual postoperative TCD measurements and in 21 (78%) of 27 patients. Of patients with TE, 10 (83%) of 12 had detectable MES compared with 11 (73%) of 15 patients without TE (P=0.66). MES were significantly associated with elevated thrombin generation during the implantation period, as reflected by plasma prothrombin fragment F1.2. Elevations in indexes of coagulation, platelet activation, and fibrinolysis relative to normal control subjects were found for patients with VADs with and without detected MES.
Conclusions—Although no significant relation between MES and TE in VAD patients was found, the data support the hypothesis that MES are related to increased hemostatic activity in this patient group despite aggressive anticoagulant therapy.
Transcranial Doppler ultrasound (TCD) has been used to measure microembolic signals (MES) during cardiopulmonary bypass and carotid endarterectomy and in patients implanted with prosthetic heart valves and ventricular assist devices (VADs). MES are consistently encountered during cardiopulmonary bypass, particularly after aortic cross-clamp release.1 2 An association between a higher rate of MES during surgery and postsurgical neurological dysfunction has been described in patients undergoing cardiopulmonary bypass.3 MES are frequently observed in patients with carotid endarterectomy, and postoperative cerebral ischemia has been associated with a high rate of MES during surgery and in the immediate postoperative period.4 5 In patients with prosthetic heart valves, asymptomatic MES are frequently detected, with the rate of MES dependent on valve type, number, and placement.6 7 Elevated levels of platelet activation, indicated by elevated plasma levels of the platelet α-granule release product β-thromboglobulin (βTG), have been reported in patients with prosthetic heart valves with MES and recurrent cerebral ischemia.8
TCD observations of MES in patients with VADs are far less numerous in the literature. MES are consistently encountered in patients with VADs, but published results are inconclusive in demonstrating a definitive relation between MES rate and clinical neurological thromboembolism (TE). Several studies have reported elevated MES rates in patients with TE,9 10 whereas others have used TCD in patient groups that remained asymptomatic or that have sample populations too small to offer conclusions.11 12 Changes in hemostasis, fibrinolysis, and cellular aggregation after VAD implantation are well documented.13 14 15 However, no studies to date have measured indexes of coagulation, fibrinolysis, or cellular activation and aggregation simultaneously with TCD in patients with VADs to examine the possible role of hematologic perturbations in MES formation and precipitation of TE. With an incidence between 0% and 30%, dependent on device type and implantation center, TE remains a significant impediment to the chronic application of VADs.16 Considering the demonstrated utility and increasing use of VADs in maintaining patients with cardiomyopathy until heart transplantation, identification of mechanisms of TE is important in improving patient care and potentially decreasing the number of strokes in this patient population.
In the current study, thrombin generation, fibrinolysis, and platelet activation were measured in patients with VADs with the use of enzyme-linked immunosorbent assays (ELISAs). Whole-blood flow cytometry was used to measure platelet activation, monocyte-tissue factor expression (Mono-TF), and leukocyte-platelet conjugate formation in the same patient group. TCD measurements were made on the same postoperative days that blood samples were taken for ELISA and flow cytometry to test the hypothesis that MES in patients with VADs are related to perturbations in coagulation, fibrinolysis, platelet activation, or cellular aggregation caused by blood-biomaterial contact. The hypothesis that MES are related to TE was also tested in this patient group.
Subjects and Methods
Over a 4-year period beginning in November 1993, a group of 27 patients (21 male, 6 female) scheduled for VAD implantation were entered into the study after informed consent was given (University of Pittsburgh Medical Center institutional review board No. 9505108 and No. 9506117). Twelve patients received the Novacor Left Ventricular Assist System (Baxter Healthcare) and 15 the Thoratec System (Thoratec Laboratories). Six of the patients with the Thoratec received both left and right ventricular support (BiVAD). The median time on device was 84 days (range 11 to 236) and the median age of participants was 49 years (range 15 to 68). The anticoagulation regimen after VAD implantation consisted of dextran at 6 to 8 hours after surgery. Heparin was started at 12 to 24 hours to maintain the activated partial thromboplastin time (PTT) at a target of 1.5 times the upper normal range (45 to 55 seconds); after chest tube removal and extubation (5 to 7 days after implantation), oral coumarin was started to maintain the international normalized ratio (INR) within a target range of 3.0 to 4.0. Oral antiplatelet agents were given to 25 out of the 27 patients in the study. Four patients received aspirin at 325 mg/d prophylactically, 12 received 81 mg/d prophylactically, and 6 were started on 81 mg/d after the first postoperative week. Three patients received 225 mg/d of persantine, and 2 patients were not given antiplatelet agents subsequent to a history of thoracic and gastrointestinal bleeding and epistaxis, respectively. The mean PTT for each postoperative day during the first week after implantation (while on heparin) and the mean INR for each postoperative day for the time period after the first week (after initiation of coumarin) were recorded. Suspected thromboembolic neurological events were confirmed by neurological consultation and, if necessary, computed tomographic scan and recording.
Transcranial Doppler Technique
Transcranial Doppler measurements were performed before VAD implantation and on postoperative days 1 to 5, 7, 10, 15, 21, 28, 45, and 60. The right and left middle cerebral arteries were isonated for 10 to 15 minutes each at a depth of 45 to 55 mm with a 2-MHz probe (Medasonics Transpect-2000), with data recorded onto videocassette for visual analysis by counting the total number of emboli and calculating the mean MES rate (MES/min). The criteria for identification of MES were those recommended by the Consensus Committee of the Ninth International Cerebral Hemodynamic Symposium.17 Briefly, MES were identified by their short duration (<300 ms) and greater amplitude than the background blood flow. They had a random appearance within the cardiac cycle and the associated characteristic “snap” or “chirping” sound. All patients had at least 1 TCD measurement, for a total of 123 measurements (8 preimplantation and 115 postimplantation). Patient compliance was an issue with TCD measurements, with several patients choosing not to continue in the study after only a few measurements. Since some patients were discharged with the device during the implantation period, and transplants occurred throughout this time, all but 2 of the TCD measurements occurred within the first 30 days after VAD implantation.
Commercial ELISA kits were used to quantify thrombin generation, fibrinolysis, and platelet activation by measurement of plasma concentrations of thrombin-antithrombin III complex (TAT; Behring Diagnostics), prothrombin fragment 1.2 (F1.2; Behring Diagnostics), d-dimer, βTG, and platelet factor 4 (PF4; the latter all from American Bioproducts). Blood samples were obtained from peripheral arterial catheter (while in intensive care) or through venous access lines or peripheral venipuncture (after transfer to patient units) after the TCD measurement schedule. Samples for βTG and PF4 were drawn into 3- or 5-mL Vacutainer tubes (Becton-Dickinson) containing an anticoagulant and antiplatelet cocktail consisting of citric acid (0.11 mol/L), theophylline (15 mmol/L), adenosine (3.7 mmol/L), and dipyridamole (0.198 mmol/L).18 Samples for the remaining tests were drawn into 3-mL Monovette blood collection tubes containing 3.8% citrate (Sarstedt). On several occasions, sample volumes obtained were not adequate to permit all tests to be performed.
Flow Cytometric Analysis
Whole-blood flow cytometry was used to measure platelet surface expression of p-selectin, Mono-TF, and formation of monocyte-platelet (Mono-plt) and granulocyte-platelet (Gran-plt) conjugates. Samples were drawn into 3-mL Monovette tubes containing 3.8% sodium citrate on the same schedule as ELISA samples and TCD measurements. Immediately after sample collection, aliquots were incubated for 40 minutes with monoclonal antibodies specific for platelet glycoprotein Ib (CD42b; Gentrak or Becton-Dickinson Immunocytometry Systems), platelet p-selectin (CD62), the monocyte marker CD14, the granulocyte marker CD15 (all from Becton-Dickinson), tissue factor (TF; American Diagnostica), or immunoglobulin G (IgG) isotype controls (Becton-Dickinson). Anti-CD42b, anti-TF, anti-CD15, and their IgG isotype controls were obtained conjugated to fluorescein-isothiocyanate (FITC), and p-selectin, anti-CD14, and an additional anti-CD42b and their IgG isotype controls were obtained conjugated to phycoerythrin (PE) for 2-color flow cytometric analysis.
After incubation, samples were fixed with 1% paraformaldehyde and analyzed within 24 hours on a FACScan flow cytometer (Becton-Dickinson Immunocytometry Systems). For each sample, 5000 events were acquired by live gating on events positive for CD42b-FITC, CD14-PE, or CD15-FITC with a positive fluorescence threshold determined by excluding 98% of matched isotype control fluorescence. Single platelets were identified by positive fluorescence for CD42b and characteristic forward scatter profile, and platelet activation was quantified by the percentage of the CD42b-positive population also positive for CD62. Monocytes and granulocytes were identified by positive fluorescence for either CD14 or CD15 and characteristic forward and side scatter profiles. Mono-plt and Gran-plt conjugates were defined as the percentage of the CD14- or CD15-positive populations, respectively, which were also positive for the platelet marker CD42b, and Mono-TF expression was similarly defined as the percentage of the CD14-positive population also positive for tissue factor.
Control blood samples for all ELISA and flow cytometric indexes, collected over a 3-year period, were obtained after informed consent (IRB No. 9505133) by antecubital venipuncture from 23 healthy, nonsmoking, nonpregnant donors 19 to 31 years of age who were medication free for at least 2 weeks before sampling. Normal ranges represent the mean±SEM of control data.
Fisher’s exact test was used to test the significance of differences in the occurrence of MES between patients with and those without thromboembolic events and between patients with a Novacor VAD, a Thoratec VAD, and Thoratec BiVADs. Mean MES rates (MES/min) for the different patient groups are presented for comparison to MES rates previously reported in the literature. Formal statistical comparison of MES rates between patient groups was not pursued because of differing numbers of measurements per patient and a strong association of time with MES. The variation of the mean MES rate over time was evaluated with the use of an ANOVA with post-hoc Neuman-Keuls testing for differences between postimplantation MES rates and preoperative levels.
To evaluate the association of MES showering with measured hemostatic parameters MES showering data were grouped based on the presence of MES (any number of MES detected vs no MES detected) and the degree of MES (no MES showering vs light MES showering vs heavy MES showering). Light MES showering was defined as an observed rate of >0 and <10 MES/min. Heavy MES showering was defined as ≥10 MES/min.
Repeated-measures ANOVA was used for the primary analysis to account for the correlation between measurements on the same patients. Two approaches were applied for the evaluation of continuous outcomes (hemostatic variables). The first, termed the mixed models approach, implemented a random patient effect to introduce this correlation in the model as described by Laird and Ware.19 This was accomplished by use of the PROC MIXED routine in the SAS programming language. The robustness and stability of these analyses were also examined by use of a generalized estimating equation approach in which an exchangeable correlation matrix was implemented to model the repeated measures structure.20 A SAS IML program was used for this technique21 and MES data obtained in the first 28 days of VAD implantation was evaluated. For maximum generality, the effect of time, which observationally appeared quite complex, was modeled as arbitrary (different mean for each time point) rather than assuming a polynomial or other underlying effect in both techniques. To compare hemostatic variables between VAD patients and normal control subjects, the random-effect model approach was also used with data grouped by the presence of MES showering and the time of the measurement in the implantation period (before or after 5 days).
The use of 2 statistical approaches for determining the relation between the continuous hemostatic variables and MES was used in an effort to address some of the limitations associated with the relatively small number of subjects and the high incidence of missing data. Specifically, the small data set prevented rigorous evaluation of the assumption of a normal error structure in both techniques, and the standard errors for coefficients and resulting probability value were based on asymptotic theory that may not be highly accurate because of the data set size. The small number of subjects and missing data necessitated imposing a correlation structure in both techniques in which the degree of association was identical between all within-subject measurements. This was clearly an approximation because measurements closer in time might be expected to be more highly correlated. Because of these limitations, results from both statistical models are presented and emphasis is placed on results found to have a substantial association with MES as determined by both techniques.
MES Rates Between Patients With and Those Without Thromboembolic Neurological Events
TE events were classified as transient ischemic attacks (TIA, symptoms resolving within 24 hours) or cerebrovascular accidents (CVA, permanent deficit or symptom duration >24 hours). Thirteen TE events, 5 CVA and 8 TIA, were recorded in 12 of the patients. Temporally, 7 of the events occurred within the first 4 weeks after implantation, with the remaining 6 events occurring between weeks 5 and 12. Six of the patients with the Novacor VAD (3 CVA, 3 TIA), 5 of the patients with the Thoratec VAD (2 CVA, 4 TIA), and 1 of the patients with Thoratec BiVADs (1 TIA) had TE. One of the patients with the Thoratec VAD had 2 TIAs.
MES were not observed in any TCD measurements made before VAD implantation. After implantation, the MES rate increased between postoperative days 4 to 7, reaching a maximum of 25±16 MES/min on postoperative day 7 (Figure 1⇓). After postoperative day 7, the rate of MES decreased but remained elevated over preoperative levels. The mean MES/min for all postoperative measurements was 8±3. Table 1⇓ summarizes the incidence of MES detection (by patients and by measurement) for all patients and in the subgroup that had TE. Of the 12 patients with Novacors, one half had MES detected compared with 100% of patients with Thoratec VADs (P=0.001 vs Novacor) and 100% of patients with Thoratec BiVADs (P=0.014 vs Novacor). In line with these significant differences, the mean MES rate for all TCD measurements from patients with Novacors was 0.17±0.07 MES/min compared with 9±2 MES/min for all Thoratec VAD measurements and 21±9 MES/min for all Thoratec BiVAD measurements. Across the 3 devices, MES were detected in 10 (83%) of 12 patients with TE and 11 (73%) of 15 without TE (P=0.66 vs patients with TE). The mean MES rate was 11±5 MES/min for patients with TE and 4±1 MES/min for patients without TE.
Hemostatic Indexes and the Presence of MES Showering
The generation of thrombin in vivo as reflected in plasma prothrombin fragment F1.2 was found to be related in a highly significant manner to the presence of MES showering in patients with VADs in both statistical models (Table 2⇓). The parameters predicted by both models indicate a significant elevation in F1.2 (estimated to be increased by 1.4 nmol/mL) in measurements in which MES were detected relative to those without MES. A marker for circulating inactivated thrombin, TAT, did not reach the criteria for significance (P=0.08 in one model) but reflected a similar trend toward a positive relation between coagulation activity and MES signals.
Circulating Gran-plt conjugates were also found to be significantly related to the presence of MES in patients with VADs. Here there was an inverse relation in which the presence of MES was associated with a decrease in the percentage of circulating granulocytes harboring attached platelets. Both models demonstrated this effect to be significant.
Hemostatic Indexes and the Degree of MES Showering
In Tables 3⇓ and 4⇓ the relation between the degree of MES showering and the measured hemostatic parameters are presented for the 2 statistical models, respectively. Prothrombin fragment F1.2 was found to be significantly higher for light versus no MES showering in both statistical approaches. In Figure 2⇓, F1.2 measurements made over the course of VAD implantation are presented for the light and no MES showering groups. The consistent elevation in F1.2 measurements in which light MES showering was detected is apparent here. In one model (Table 4⇓), a significant difference between F1.2 measurements was also found between heavy versus no MES showering. Circulating TAT approached significance in one of the models (Table 4⇓), with a trend supporting a positive relation between coagulation activity and light or heavy MES showering. No significant differences were detected for heavy versus light MES showering for either F1.2 or TAT.
The negative relation between circulating Gran-plt conjugates observed in the earlier analysis also held for light versus no MES showering in both models. No significant differences were detected for heavy versus no or light MES showering. Other significant differences found when grouping the data on the degree of MES showering were detected using one of the two statistical models. Of note here were the relations found between a lower PTT within the first 7 days of VAD implantation and heavy MES showering (Table 3⇑).
Hemostatic Indexes in Patients With VADs Relative to Normal Control Subjects
Comparing the measured hemostatic indexes from patients with VAD with those from normal control blood donors in Table 5⇓, it is seen that marked elevations occurred in the VAD patient group with and without MES showering and in the early implantation period (≤5 days of device implantation) as well as later measurements. Elevations in platelet activation (evidenced by PF4 and βTG and the percentage of circulating platelets expressing p-selectin), coagulation (prothrombin fragment F1.2 and TAT), fibrinolysis (d-dimer), and leukocyte-platelet conjugates (Gran-plts) were present in the implantation period after 5 days, regardless of the detection of MES showering.
Reports in the literature present differing conclusions regarding an association between MES and TE in patients with VADs. Slater et al11 studied 8 patients with Thermo Cardiosystems HeartMate VADs and observed MES in 28% of their TCD measurements. They report a relatively low mean MES rate of 0.52±1.00 MES per 30-minute session for all of their TCD studies, and none of their patients examined with TCD exhibited clinically apparent TE. Moazami et al9 studied 14 patients with HeartMates and observed MES in 26% of their TCD measurements. They found an elevated rate of MES in one patient who had a thromboembolic neurological event (0.03 MES/min) relative to asymptomatic patients (0.016 MES/min). Nabavi et al10 studied 6 patients with Novacor VADs and observed MES in 84% of their TCD measurements. They observed a significant elevation in MES on days of clinically apparent thromboembolic events (18.5 MES per 30-minute session) versus days without events (4 MES per 30-minute session, P<0.001). We report here an elevation in MES after VAD implantation, observations of MES in 58% of our TCD measurements, and a trend toward increased MES rates in patients with clinically apparent thromboembolic neurological events. The percentage of MES-positive TCD measurements in our patients with Novacor VADs (18%) was comparable to the incidence of MES reported by Moazami et al and Slater et al for the HeartMate VADs and much lower than the 84% rate reported by Nabavi et al in their Novacor VAD patient group. Of note, Nabavi et al reported a higher incidence of thromboembolic neurological events in their 6 patients, with 67% having multiple TE, whereas we report TE in 6 of 12 patients with Novacors, all with single events.
MES are frequently observed in prosthetic heart valve patients, although the relation between MES, hemostatic alterations, and thromboembolic neurological events remains poorly defined. Braekken et al22 found elevated MES rates up to 5 years after implantation in a group of bileaflet mechanical valve patients with a history of cerebrovascular symptoms (1 MES/min) when compared with patients who remained asymptomatic (0.02 MES/min, P=0.04), whereas Georgiadis et al6 23 and Sliwka and Georgiadis7 report no significant differences between patients with and those without TE. Sturzenegger et al8 found significantly elevated βTG compared with normal control subjects in a group of 5 patients with mechanical prosthetic heart valves with a history of recurrent TE and suggested that MES are composed in part of activated platelets. However, that study did not find elevations in TAT, fibrinopeptide A, or d-dimer versus normal control subjects. Georgiadis et al measured plasma d-dimer, antithrombin III, and TAT simultaneously with TCD in a group of 100 patients with prosthetic heart valves and did not find significant relations between MES and their measured hematologic parameters.24 They report that the incidence of MES was significantly elevated in patients with mechanical valves compared with patients with bioprosthetic prosthetic heart valves and suggest that observed MES were not the result of thrombotic material but platelet aggregates or gaseous material. In a recent study, Sliwka and Georgiadis7 report significantly elevated MES counts (MES/h) in patients with mechanical valves versus patients with bioprosthetic heart valves and suggest that MES in patients with bioprosthetic valve are caused by formed material resulting from blood-biomaterial interactions.
Cavitation microbubbles are hypothesized to be responsible for observed elevations in MES rates in patients with mechanical heart valves.7 25 Hyperbaric oxygenation has been reported to result in decreased MES rates, believed to be the result of displacement of less soluble nitrogen with more soluble oxygen in the blood, thus reducing the rate of cavitation by gaseous nitrogen.26 27 On the other hand, the extremely small size and short duration of cavitation bubbles and observations of cavitation in vitro at supernormal as opposed to physiological pressure loading suggest that cavitation is not the sole source for MES in vivo.25 28 Shu et al28 report that TCD was unable to detect gaseous cavitation microbubbles in vitro under known cavitation generating conditions and suggest that this was due to cavitation bubbles being too small, too few, or of too short a duration. It is interesting to speculate that our observed differences in MES rates between the Thoratec VAD and BiVAD (mechanical valves) and the Novacor system (bioprosthetic valves) may be due at least in part to cavitation effects. However, our observed trends of elevated thrombin generation with increasing MES rates suggest a thrombotic origin for MES.
Prosthetic heart valve implantation is similar to VAD implantation in that patient blood contacts artificial surfaces and disturbed flow fields for an extended implant period. However, VAD implantation results in much more extensive blood-biomaterial contact than prosthetic heart valve implantation. The expected alterations in coagulation, fibrinolysis, and cellular activation after VAD implantation have been described for several devices. Spanier et al14 report elevated thrombin generation and fibrinolysis in HeartMate VAD patients, indicated by elevated plasma levels of TAT, F1.2, and d-dimer. Livingston et al29 found higher levels of thrombin generation (F1.2, TAT), platelet activation (PF4, βTG), and fibrinolysis (d-dimer) in patients with HeartMates immediately after surgery when compared with patients with non-VAD bypass. A previous study by our group found significant increases in thrombin generation (TAT), monocyte-tissue factor expression, and circulating levels of monocyte-platelet and granulocyte platelet conjugates compared with normal control subjects for the duration of device support in a group of 17 patients with Novacor VADs and 13 with Thoratec VADs (5 BiVADs).15 These previous studies provide a foundation for hypothesizing that hemostatic alterations associated with VAD implantation may provide a thrombotic source for emboli subsequently detected as MES.
We have suggested in prior studies that thrombotic deposition may preferentially occur early in the implantation period within the VAD and that this deposition may decrease over time as fibrinolytic pathways dominate.30 The relation that was found in one of the statistical models between heavy MES showering and lower PTT in the first week of implantation would support this model. The INR after the first week of implantation was not found to be different among any MES-based groupings. With regard to the use of anticoagulation monitoring, it is worthwhile to note that elevations in F1.2 and TAT in all patients with VADs and the relation between F1.2 and MES appeared in the face of a relatively aggressive anticoagulation regimen.
We have shown that MES are related to VAD implantation and remain significantly elevated for at least the first 4 weeks of VAD support. Although we did not find a direct and significant statistical relation between thromboembolic neurological events and MES, the trend in MES data suggests that there may be an important role for MES in mediating TE in patients with VADs. It is possible that the mechanisms responsible for generating the small asymptomatic emboli detected as MES may not be related to the generation of larger emboli that precipitate clinical thromboembolism. Further, cavitation may be partly responsible for MES, particularly in patients supported by devices with mechanical valves. However, the hematologic indexes suggest that subclinical coagulation may be an important mediator in the formation of MES. Whereas the utility of TCD in predicting increased risk for stroke in VAD patients remains debatable, TCD in conjunction with sensitive assays of coagulation, fibrinolysis, and cellular activation may provide a methodology for identifying potential thrombotic mechanisms that result in MES. Additional studies into the possible relation between MES and stroke and between hematologic alterations and MES are warranted, as identification of patients or time periods of increased risk for stroke would greatly improve patient treatment and quality of life.
This study was supported by a grant-in-aid (W.R.W.) from the American Heart Association. The authors would like to thank Stephen Winowich, Carla Nastala, and Eileen Stanford of the University of Pittsburgh Medical Center Artificial Heart Program for their clinical support in this study.
- Received August 27, 1999.
- Accepted September 20, 1999.
- Copyright © 1999 by American Heart Association
Barbut D, Hinton RB, Szatrowski TP, Hartman GS, Bruefach M, Williams-Russo P, Charlson ME, Gold JP. Cerebral emboli detected during bypass surgery are associated with clamp removal. Stroke. 1994;25:2398–2402.
Barbut D, Yao FS, Hager DN, Kavanaugh P, Trifiletti RR, Gold JP. Comparison of transcranial Doppler ultrasonography and transesophageal echocardiography to monitor emboli during coronary artery bypass surgery. Stroke. 1996;27:87–90.
Levi CR, O’Malley HM, Roberts AK, Hoare MC, Royle JP, Chan A, Beiles BC, Chambers BR, Bladin CF, Donnan GA. Transcranial Doppler detected cerebral microembolism following carotid endarterectomy. Brain. 1997;120:621–629.
Spencer M. Transcranial doppler monitoring and causes of stroke from carotid endarterectomy. Stroke. 1997;28:685–691.
Georgiadis D, Kaps M, Berg J, Mackay TG, Dapper F, Wheatley DJ, Lees KR. Transcranial Doppler detection of microemboli in prosthetic heart valve patients: dependency upon valve type. Eur J Cardiothorac Surg. 1996;10:253–258.
Sliwka U, Georgiadis D. Clinical correlations of Doppler microembolic signals in patients with prosthetic cardiac valves. Stroke. 1998;29:140–143.
Sturzenegger M, Beer JH, Rihs F. Monitoring combined antithrombotic treatments in patients with prosthetic heart valves using transcranial Doppler and coagulation markers. Stroke. 1995;26:63–69.
Nabavi DG, Georgiadis D, Mumme T, Schmid C, Mackay TG, Scheld HH, Ringelstein EB. Clinical relevance of intracranial microembolic signals in patients with left ventricular assist devices. Stroke. 1996;27:891–896.
Slater JP, Roes EA, Levin HR, Frazier OH, Roberts JK, Weinberg AD, Oz MC. Low thromboembolic risk without anticoagulation using advanced-design left ventricular assist devices. Ann Thorac Surg. 1996;32:1321–1328.
Roberts JK, Mohr JP, Moazami N, Oz MC. Microembolic signals in patients with left ventricular assist devices. Stroke. 1996;27:1915–1916. Letter.
Wagner WR, Johnson PC. Therapeutic techniques for reducing thrombosis after cardiovascular implants. In: Klitzman B, ed. Problems in General Surgery. Philadelphia, Pa: JB Lippincott Co; 1994;11:241–264.
Consensus Committee of the Ninth International Cerebral Hemodynamic Symposium. Basic identification criteria of Doppler microembolic signals. Stroke. 1995;26:1123.
Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika.. 1986;73:13–22.
Groemping U. GEE: An SAS Macro for Longitudinal Data Analysis. Universitaet Dortmund, Germany: Fachbereich Statistik; 1994.
Braekken KS, Russell D, Brucher R, Svennevig J. Incidence and frequency of cerebral embolic signals in patients with a similar bileaflet mechanical valve. Stroke. 1995;26:1225–1230.
Georgiadis D, Grosset DG, Kelman AW, Faichney A, Lees KR. Prevalence and characteristics of intracranial microembolic signals in patients with different types of prosthetic cardiac valves. Stroke. 1994;25:587–592.
Georgiadis D, Mallinson A, Grosset DG, Lees KR. Coagulation activity and emboli counts in patients with prosthetic cardiac valves. Stroke. 1994;25:1121–1214.
Kaps M, Hansen J, Weiher M, Tiffert K, Kayser I, Droste DW. Clinically silent microemboli in patients with artificial prosthetic aortic valves are predominantly gaseous and not solid. Stroke. 1997;28:322–325.
Georgiadis D, Wenzel A, Lehmann D, Lindner A, Zerkowski HR, Zierz S, Spencer MP. Influence of oxygen ventilation on Doppler microembolic signals in patients with artificial heart valves. Stroke. 1997;28:2189–2194.
Livingston ER, Fisher CA, Bibidakis EJ, Pathak AS, Todd BA, Furukawa S, McCluren JB, Addonizio VP, Jeenanandam V. Increased activation of the coagulation and fibrinolytic systems leads to hemorrhagic complications during left ventricular assist implantation. Circulation. 1996;94(suppl II):II-227–II-234.
Wagner WR, Johnson PC, Kormos RL, Griffith BP. Evaluation of bioprosthetic valve-associated thrombus in ventricular assist device patients. Circulation. 1993;88:2023–2029.