Pump thrombosis is a major complication after continuous-flow left ventricular assist device (CF-LVAD) implantation. It is indeed a “multifactorial phenomenon,”1,2 and eventually, has a “wide spectrum of clinical presentation.”3 An algorithm for the diagnosis and management of pump thrombosis has been published recently,1 but the efficacy of this algorithm has not been evaluated by large clinical studies yet. Furthermore, pump thrombosis is associated with an increasing incidence in recent years, and with significant morbidity and mortality rates.4–6 Medical (i.e., direct thrombin inhibitor [DTI],7,8 tissue plasminogen activator [t-PA],9,10 or glycoprotein IIb/IIIa antagonist11) and surgical (i.e., emergent heart transplantation, pump exchange6,12–14) treatment options have been reported, however, best treatment algorithm still remains controversial. In summary, the clinical problem which we face in our daily practice is growing as annual implant numbers and patient volume on CF-LVADs at follow-up continue to increase, and many questions remain unanswered yet.
Hence, with an effort to create a safe and effective treatment algorithm on pump thrombosis in the light of our previous clinical experience, a retrospective study was designed in our center. Herein, we aimed to evaluate the outcomes of patients who had pump thrombosis and underwent various treatment methods, and to create a safer and effective treatment algorithm accordingly.
Between December 2010 and December 2014, 163 consecutive patients (mean age: 50.7 ± 13 years, 84% males, median duration of support: 277 (2–1,077) days) were implanted a CF-LVAD (HeartWare [HVAD; HeartWare Inc, Framingham, MA] in 130, Heartmate-II [HM-II; Thoratec Corp, Pleasanton, CA] in 31 and HeartAssist-5 [Micromed Cardiovascular Inc, Houston, TX] in two patients) in our center because of end-stage heart failure. Prospectively collected data of all patients who had at least one episode of pump thrombosis have been analyzed retrospectively. Median sternotomy and left anterior thoracotomy incisions were used in 147 (90.2%) and in 16 (9.8%) patients, respectively. Ascending aorta was the site of distal anastomosis in the former, whereas descending thoracic aorta in the latter.
Postoperative Anticoagulation-Antiaggregation Protocol
In the absence of continuing bleeding, unfractionated heparin (UFH) was started after 24 hours of implantation, based on coagulation parameters. Activated partial thromboplastin time (aPTT) was targeted at 40–50 seconds on the first postoperative day (POD-1), and at 50–60 seconds thereafter. Acetylsalicylic acid (ASA) and Warfarin were usually initiated on POD-3. Acetylsalicylic acid was given 100 mg/day in hospital, and the dosage was then increased to 300 mg/day before discharge. The ASA dose was regulated according to ASPI activity measured by aggregometer (Roche Multiplate, Roche Diagnostics, Mannheim, Germany). Target international normalized ratio (INR) range was set between 2.5 and 3.0 for all CF-LVADs and for all surgical techniques. This protocol remained consistent throughout the study period. Low molecular weight heparin (LMWH) was added to the treatment in those who had suboptimal INR level (INR < 2.5) during the follow-up.
Diagnosis of Pump Thrombosis
The diagnosis was made by using the following criteria: 1) symptoms (i.e., shortness of breath, fatigue, embolic stroke etc.) and/or findings (i.e., dark-colored urine, jaundice); 2) elevated levels of lactate dehydrogenase (LDH) (≥450 U/L) (reference value: 135–225 U/L); 3) significant alterations of pump parameters (i.e., pump stop-restart, gradual, or sudden increase of power consumption and altered estimated pump flow) beyond the preset limits for each patient. Presence of two or more criteria were sufficient for the diagnosis of pump thrombosis. Presence of only one criterion lead us to further investigate any evidence of pump thrombosis. Ramp test was performed in such cases to evaluate pump function and to reveal any subclinical PT. Besides, computed tomography scan or echocardiography was performed to investigate pump dislocation and inflow or outflow obstruction of the pump in selected cases (Figure 1). Plasma-free hemoglobin and subgroup of LDH were not used for the diagnosis because of unavailability.
Treatment of Pump Thrombosis
Despite the consistency in our postoperative anticoagulation-antiaggregation protocol and definition of pump thrombosis, our treatment protocol showed versatility. The treatment alternatives used at the time of admission are as follows.
Unfractionated heparin infusion was started intravenously with the target aPTT of 50–60 seconds. If the INR was above 3.0 at admission, fresh frozen plasma was given to adjust the INR to the target level.
Heparin and tirofiban infusion
Unfractionated heparin infusion was started intravenously with the target aPTT of 40–50 seconds. In addition, Tirofiban infusion was started according to a standard dose scheme (0.4 μg/kg/min within the first 30 minutes, and then 0.1 μg/kg/min maintenance dose for 24 hours).
Some patients without a history of bleeding were selected for t-PA infusion. If the INR was above 3.0 at admission, fresh frozen plasma was given to adjust the INR to the target level. Then, a pigtail catheter was inserted via femoral artery, and advanced into the left ventricle through the aortic valve, locating the tip of the catheter near the inflow cannula under fluoroscopic guidance in five events. In two events, t-PA was administered through central venous catheter. Thirty-to-fifty mg of t-PA (tenectaplase or alteplase) was infused via the catheter in 3–5 minutes, and a second dose was infused if pump parameters did not return to normal within 30 minutes after the initial dose.
All patients were listed for urgent transplantation (status-1A) as soon as the diagnosis has been confirmed according to the algorithm defined above, and were transplanted if a suitable donor could be matched.
If the pump thrombosis did not resolve and/or a suitable donor was not matched within a reasonable period (depending on the patient’s clinical status and hemodynamics), malfunctioning pump was exchanged. Pump exchange was performed by left anterior thoracotomy for HVAD15 and by subcostal chevron incision for HM-II.13 Cardiopulmonary bypass was not used except in one patient who had hemodynamic instability. Femoral artery and vein were the cannulation sites for cardiopulmonary bypass.
Complete resolution of pump thrombosis after treatment was defined by restoration of pump parameters to normal levels, disappearance of symptoms and findings and normalization of LDH levels after treatment, for at least 72 consecutive hours. Procedural success was defined as survival to discharge after treatment procedure with complete resolution (for those who had medical treatment) or with normal hemodynamics (for those who had pump exchange or heart transplantation).
Continuous variables were provided as mean ± standard deviation and median (range). Categorical variables were provided as frequency (%).
During the study period, a total of 21 PT events were observed in 15 patients (9.2% of the cohort, 0.137 events per patient-year [EPPY]), one of which had a second pump thrombosis event late after pump exchange because of a new pump thrombus. Median duration of support at the time of first PT event was 259 (8–585) days for the entire cohort, and 293 (25–585) days and 67 (8–418) days for HVAD and HM-II, respectively. The incidence of pump thrombosis in patients with HVAD and HM-II were 9.2% (12/130, 0.14 EPPY) and 9.7 % (3/31, 0.10 EPPY), respectively.
As for diagnostic criteria, symptoms and/or findings related with pump thrombosis, LDH ≥ 450 U/L and significant alterations of pump parameters were present in 9 (42.9%), 19 (90.4%), and 20 (95.2%) events, respectively. The most common symptoms and findings were fatigue, dark-colored urine, dyspnea, jaundice, and embolic stroke. Median INR level at the time of hospital admission was 2.3 (range: 1.0–4.4). Suboptimal INR level within one month before the first PT episode was detected in 86.7% (13/15) of patients. Of these, one patient had replacement of warfarin with LMWH because of gastrointestinal bleeding, and another one had driveline infection. No predisposing factor was determined in the remaining. Mean LDH and total bilirubin levels were 1,271 ± 1,016 U/L (range: 237–4,146), and 2.07 ± 1.1 mg/dl (range: 0.81–5.08), respectively. As for antiplatelet therapy, ASA dose was 100 mg/day before 5 events and 300 mg/day before eight events. Three patients had only clopidogrel 75 mg/day and three patients with recurrent PT had ASA 300 mg/day in addition to clopidogrel 75 mg/day. One patient with previous hemorrhagic stroke did not have any antiplatelet agent before two PT events. The course of all pump thrombosis events has been depicted in detail in Figure 2.
Recurrent pump thrombosis events were observed five times in three patients (twice in two patients, and once in one patient). Of these, one patient had two events on POD-225 and POD-529, both successfully treated by UFH and UFH–tirofiban combination followed by thrombolytics. A third pump thrombosis event occurred on POD-888, and was treated with UFH infusion followed by thrombolytics. Although complete thrombus resolution was achieved, the patient died because of intracranial hemorrhage shortly afterwards. Another patient had also three pump thrombosis events; first event on POD-118 was treated successfully by UFH–tirofiban combination followed by thrombolytics, second event on POD-139 by UFH–tirofiban combination, and third event on POD-162 by UFH only. This patient survived all three events. Finally, another patient had two events on POD-293 and on POD-302, both treated by UFH–tirofiban combination. Although complete resolution was achieved after the first event, the patient died shortly after the second event because of stroke. Using the echocardiography and computed tomography scan in selected cases, pump malposition was never diagnosed in our series.
Overall mortality was 40% (6/15), and overall procedural success was 71.4% (15/21) in our entire cohort. The cause of mortality was hemorrhagic stroke in those who had medical treatment (n = 5), and sepsis and right ventricular failure in the other who had pump exchange. Pump exchange was performed in 5 patients after 2, 3, 7, 38 days, and 10 weeks of unsuccessful medical treatment, and was associated with a procedural success of 80%. In addition, two patients were treated successfully by heart transplantation, with a procedural success of 100% (Figure 2). All CF-LVADs explanted at the time of pump exchange or transplant (n = 7) had macroscopic thrombus within the pump components.
The CF-LVADs provide favorable survival rates up to four years after implantation, although device complications are still a clinical challenge to the clinicians.16 Among these, pump thrombosis is a potentially life-threatening one, which may occur despite optimal anticoagulation and be further complicated by stroke or device failure. It is indeed a “multifactorial phenomenon.”2 A group of experts have recently published a detailed list of these factors, simply and rightfully classifying them as pump-, patient-, and management-related factors.1 Although this classification may help clinicians understand the underlying possible causes and mechanisms of thrombus formation within the CF-LVADs, it clearly demonstrates the complexity of the problem. Severity of each factor in a given patient may vary considerably, and presence of one factor may not necessarily prevent the co-existence of another or others. Numerous combinations of causative factors may further complicate the clinical presentation and the course as well. Najjar et al5 summarized the three patterns of clinical presentation as acute occlusion associated with suction events, subtotal occlusion associated with a precipitous alterations in flow and power consumption, and insidious propagation of thrombus formation associated with gradual alterations of flow and power consumption. At the time of diagnosis, clinical picture may be quite divergent, ranging from slightly abnormal laboratory markers (i.e., elevated LDH, plasma-free hemoglobin) or pump parameters without any concomitant symptoms to profound cardiogenic shock and/or stroke leading to death.1,3 Despite the diversity of etiology and clinical presentation, in fact, the definition of pump thrombosis is simple and clear; “development of clot within the flow path of any or all of the components that constitute the pump, including the titanium inflow cannula, the outflow graft, and the pump housing that contains the rotor” as stated by Goldstein et al.1 However, considerable controversy regarding the diagnosis and treatment of pump thrombosis as well as reporting outcome measures still exists in the field.
Our paper reports a pump thrombosis rate of 0.137 EPPY for the entire cohort, and 9.2% (12/130, 0.14 EPPY) and 9.7 % (3/31, 0.10 EPPY) for HVAD and HM-II, respectively. The incidence of pump thrombosis in our series is slightly higher than that of recent reports,2,4–6 this is mainly because of two reasons: 1) some groups report only “confirmed” pump thrombosis, defined as demonstration of thrombus material at pump exchange, transplantation or autopsy4 and 2) the optimal threshold level of LDH for early detection of pump thrombosis is still unclear, and our threshold for diagnosis of pump thrombosis regarding LDH elevation is well below than that of other groups.1,9,17
Our diagnostic criteria consist of clinical parameters, and focus on patient management rather than demonstration of thrombus material visually. We believe that the distinction between currently used “suspected” and “confirmed” pump thrombosis is artificial and clinically irrelevant in its present form. For this aim, objective criteria derived from clinical parameters are needed. Shah et al.18 searched for an LDH threshold level, and reported that an LDH level of 600 UI/L (2.5 times the upper limit normal) provided a sensitivity and specificity of 78% and 97%, respectively. Similarly, the group from the Columbia University, New York recently suggested an LDH threshold of 2.5 times the upper limit normal for earlier detection of pump thrombus,2 in contrast to their previous report in which the threshold was set at four times the upper limit normal (1,103 UI/L).17 In our series, the threshold was set at 450 UI/L (two times the upper limit normal), and the diagnosis was confirmed with an LDH level between 450 and 750 U/L in four events, in which only two criteria were met.
In contrast to the recent literature reporting “an alarming increase in the rate of pump thrombosis after CF-LVAD implantation,”4–6,19 we have not observed a significant increase of pump thrombosis events throughout the study period. Mehra et al.20 attributed this increase to the reduced pump speeds in an effort to facilitate aortic valve opening to preclude late aortic valve problems (i.e., aortic root thrombosis, de novo aortic insufficiency) and gastrointestinal bleeding, and relaxed anticoagulant regimes encouraged by the low thromboembolic rates. In our experience, although lower pump speed strategy was adopted parallel to the literature, neither postoperative prophylaxis protocol nor target INR level has been changed throughout the study period. We believe that intensive anticoagulation protocols including bridging with UFH should be implemented early after implantation, and continued thereafter.
Although emergent heart transplantation is the treatment of choice in patients with complicated CF-LVADs, it can be performed in limited number of patients because of donor shortage worldwide. Medical agents including UFH, DTIs, t-PA, and glycoprotein IIb/IIIa antagonists have been used in clinics in an attempt to treat pump thrombosis. However, all currently available medications have limited procedural success rates, and are associated with substantial complication rates.9,11,20 When analyzed in depth, our results provide some interesting but important findings for the treatment. First, use of UFH infusion with a target aPPT of 50–60 seconds immediately after diagnosis seemed beneficial, and added no further risk associated. However, adjunct of a glycoprotein IIb/IIIa inhibitor to the initial UFH treatment provided no additional benefit, but increased the risk of lethal complications. Tellor et al.11 reported a similar experience with a different glycoprotein IIb/IIIa inhibitor, and concluded that the risk of eptifibatide outweighed its benefit. Therefore, glycoprotein IIb/IIIa inhibitors seem not to have a role, and UFH should be the first step for the medical treatment of pump thrombus. Second, in patients who did not benefit from UFH infusion, intravenous administration of thrombolytics provided increased rate of complete thrombus resolution, at the expense of increased risk of mortality with current doses. Similar findings with the use of thrombolytics were reported previously.9,10 Although thrombolytics are potent agents with lethal side effects, we believe that they may still have a role in the medical treatment to resolve the thrombus completely and save some patients from pump exchange, but preferably with lower doses to decrease the risk of bleeding, stroke, and death. Third, pump exchange was the ultimate option for those who did not benefit from the medical treatment and were not able to have a heart transplant in our series. Despite the fact that various antiplatelet and anticoagulant agents were administered before the pump exchange surgery, a procedural success rate of 80% was achieved. In one patient, pump exchange was performed as an attempt for bail out after 10 weeks of unsuccessful medical treatment, but the patient died 7 days after surgery. Previous reports demonstrated that pump exchange is associated with low early mortality and no adverse effect on late survival.12 On the other hand, it is an invasive surgical method associated with increased infection and neurologic event rates postoperatively.6 The cost of the new pump should also be taken into account. Thus, we believe that pump exchange is an effective treatment option in which complete thrombus resolution was not achieved by medical treatment. Decision to move toward pump exchange should not be delayed if medical treatment fails within a reasonable period. Prolonged medical treatment and delayed pump exchange may increase the risk of right ventricular failure and decreased perfusion because of suboptimal pump performance, and may eventually increase the risk of mortality and morbidity. In the light of our findings, we have built a simple algorithm for the treatment of pump thrombus (Figure 3).
Expectedly, the current study has some limitations. First of all, it is a retrospective analysis of a single center experience composed of relatively small number of patients, but with satisfactory outcomes and complete follow-up. Second, because of its retrospective design, some valuable data that might influence the outcomes such as pump rates, systemic arterial pressures, and left ventricle and aortic valve functions are missing. However, it provides valuable data on a patient cohort who received a uniform postoperative anticoagulation and antiaggregation protocol throughout the study period. Third, because of the lack of a standard diagnostic algorithm at the time of study period, a center-specific diagnostic algorithm was established in our center.
In conclusion, pump thrombosis has a high mortality and morbidity rate despite meticulous treatment. The diagnosis should be based on clinical parameters, including a lower threshold (2–2.5 times the upper limit of the normal) for LDH level. Glycoprotein IIb/IIIa antagonists such as tirofiban does not have a beneficial role in medical treatment of pump thrombosis. To salvage patients from surgical trauma and its consequences, medical treatment should be the first treatment option. Unfractionated heparin infusion alone may completely solve the problem in some patients. Intravenous administration of thrombolytic agents with strict blood pressure control can be added when UFH infusion fails, however further research is necessary for optimal dosing to decrease the rate of side effects such as bleeding. In patients where medical treatment does not result in complete thrombus resolution within a reasonable period, pump exchange is the ultimate solution. Less invasive techniques such as left anterior thoracotomy and subcostal chevron incision might be preferred for pump exchange.
1. Goldstein DJ, John R, Salerno C, et al: Algorithm for the diagnosis and management of suspected pump thrombus. J Heart Lung Transplant 2013.32: 667–670.
2. Uriel N, Han J, Morrison KA, et al: Device thrombosis in Heartmate II continuous-flow left ventricular assist devices: A multifactorial phenomenon. J Heart Lung Transplant 2014.33: 51–59.
3. Birati EY, Quiaoit Y, Wald J, et al: Ventricular assist device thrombosis: A wide spectrum of clinical presentation. J Heart Lung Transplant 2015.34: 613–615.
4. Starling RC, Moazami N, Silvestry SC, et al: Unexpected abrupt increase in left ventricular assist device thrombosis. N Engl J Med 2014.370: 33–40.
5. Najjar SS, Slaughter MS, Pagani FD, et al: An analysis of pump thrombus events in patients in the HeartWare ADVANCE Bridge to transplant and continued access protocol trial. J Heart Lung Transplant 2014.33: 23–34.
6. Kirklin JK, Naftel DC, Kormos RL, et al: Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) analysis of pump thrombosis in the HeartMate II left ventricular assist device. J Heart Lung Transplant 2014.33: 12–22.
7. Sylvia LM, Ordway L, Pham DT, et al: Bivalirudin for treatment of LVAD thrombosis: A case series. ASAIO J 2014.60: 744–747.
8. Badiye A, Hernandez GA, Chaparro S: Argatroban as novel therapy for suspected thrombosis in patients with continuous-flow left ventricle assist device and hemolysis. ASAIO J 2014.60: 361–365.
9. Schlendorf K, Patel CB, Gehrig T, et al: Thrombolytic therapy for thrombosis of continuous flow ventricular assist devices. J Card Fail 2014.20: 91–97.
10. Muthiah K, Robson D, Macdonald PS, et al: Thrombolysis for suspected intrapump thrombosis in patients with continuous flow centrifugal left ventricular assist device. Artificial Organs 2013.37: 313–322.
11. Tellor BR, Smith JR, Prasad SM, et al: The use of eptifibatide for suspected pump thrombus or thrombosis in patients with left ventricular assist devices. J Heart Lung Transplant 2014.33: 94–101.
12. Stulak JM, Cowger J, Haft JW, et al: Device exchange after primary left ventricular assist device implantation: Indications and outcomes. Ann Thorac Surg 2013.95: 1262–1268.
13. Ota T, Yerebakan H, Akashi H, et al: Continuous-flow left ventricular assist device exchange: clinical outcomes. J Heart Lung Transplant 2014.33: 65–70.
14. Anand J, Singh SK, Hernández R, et al: Continuous-flow ventricular assist device exchange is safe and effective in prolonging support time in patients with end-stage heart failure. J Thorac Cardiovasc Surg 2015.149: 267–75, 278.e1.
15. Rojas SV, Avsar M, Khalpey Z, et al: Minimally invasive off-pump left ventricular assist device exchange: Anterolateral thoracotomy. Artificial Organs 2014.38: 539–542.
16. Kirklin JK, Naftel DC, Pagani FD, et al: Sixth INTERMACS annual report: a 10,000-patient database. J Heart Lung Transplant 2014.33: 555–564.
17. Uriel N, Morrison KA, Garan AR, et al: Development of a novel echocardiography ramp test for speed optimization and diagnosis of device thrombosis in continuous-flow left ventricular assist devices: The Columbia ramp study. J Am Coll Cardiol 2012.60: 1764–1775.
18. Shah P, Mehta VM, Cowger JA, et al: Lactate dehydrogenase is superior to serum free hemoglobin as a marker of pump thrombosis in left ventricular assist devices. J Heart Lung Transplant 2013.32: S37.
19. Wang JX, Lee EH, Bonde P: Over 400% increase in LVAD thrombosis reported to the FDA’s manufacturer and user facility device experience (MAUDE) database from 2010 to 2012. J Heart Lung Transplant 2014.33: S9–S10.
20. Mehra MR, Stewart GC, Uber PA: The vexing problem of thrombosis in long-term mechanical circulatory support. J Heart Lung Transplant 2014.33: 1–11.
left ventricular assist device; pump thrombosis; treatment algorithm; pump exchange; thrombolytics