ASAIO Journal 2015
Adult Circulatory Support
Roller and Centrifugal Pumps: A Retrospective Comparison of Bleeding Complications in Extracorporeal Membrane Oxygenation Ihab Halaweish,* Adam Cole,* Elaine Cooley,* William R. Lynch,† and Jonathan W. Haft*
Centrifugal pumps are increasingly used for extracorporeal membrane oxygenation (ECMO) rather than roller pumps. However, shear forces induced by these types of continuousflow pumps are associated with acquired von Willebrand factor deficiency and bleeding complications. This study was undertaken to compare adverse bleeding complications with the use of centrifugal and roller pumps in patients on prolonged ECMO support. The records of all adult ECMO patients from June 2002 to 2013 were retrospectively reviewed using the University of Michigan Health System database and the Extracorporeal Life Support Organization registry, focusing on patients supported for at least 5 days. Ninety-five ECMO patients met criteria for inclusion (48 roller vs. 47 centrifugal pump). Indications included pulmonary (79%), cardiac (15%), and extracorporeal cardiopulmonary resuscitation (6%), without significant difference between the two groups. Despite lower heparin anticoagulation (10.9 vs. 13.7 IU/kg/hr) with centrifugal pumps, there was a higher incidence of nonsurgical bleeding (gastrointestinal, pulmonary, and neurological) in centrifugal pump patients (26.1 vs. 9.0 events/1,000 patient-days, p = 0.024). In conclusion, in our historical comparison, despite reduced anticoagulation, ECMO support using centrifugal pumps was associated with a higher incidence of nonsurgical bleeding. The mechanisms behind this are multifactorial and require further investigation ASAIO Journal 2015; 61:496–501.
cannulas,6,7 oxygenators,8–11 and blood pumps.12,13 Contemporary centrifugal pumps are magnetically actuated and are associated with lower hemolysis than predicate devices14–19; as such, there has been a recent shift in the past decade from the use of roller pumps to centrifugal pumps.20,21 However, prolonged use of continuous-flow rotary pumps such as ventricular assist devices has been associated with nonsurgical bleeding from the development of an acquired form of von Willebrand factor (vWF) deficiency.22–26 Whether this phenomenon exists with centrifugal pumps in the ECMO application is unknown. This study was undertaken to compare adverse bleeding complications with the use of centrifugal and roller pumps in patients on prolonged ECMO support. Methods Patients Approval was obtained from the University of Michigan Institutional Review Board. From a prospectively collected database, a retrospective review was performed of all patients aged 18 years or older, who underwent ECMO at the University of Michigan from June 2002 to June 2013. Data on demographics, medical history, indication for ECMO, bleeding events, anticoagulation therapy, and blood product transfusion were obtained from the University of Michigan Health System database and the Extracorporeal Life Support Organization (ELSO) registry. Because our hypothesis was that centrifugal pumps were associated with bleeding complications from their impact on von Willebrand function, only patients undergoing support for at least 5 days or more were included to allow time for this effect to result in measurable differences. The indication for ECMO was recorded from the ELSO database.
Key Words: centrifugal, roller, ECMO, von Willebrand factor deficiency
Extracorporeal membrane oxygenation (ECMO) is used to
mechanically support patients with severe cardiac and respiratory failure. The use of ECMO has increased significantly in recent years, in part from the experience with H1N1 influenza1–4 as well as publication of the conventional ventilation or ECMO for severe adult respiratory failure (CESAR) trial demonstrating efficacy in acute respiratory failure.5 In addition, there have been significant technologic advances in
ELSO Registry The ELSO registry is a large, multicentered, international database of more than 40,000 patient cases receiving extracorporeal life support (ECLS) in more than 170 centers. Member institutions voluntarily submit deidentified data to the registry including sex, race, nature of illness, technical details of extracorporeal support used, complications, and outcome. Quality assurance reports are issued to active ELSO member institutions on a semiannual basis. The purpose of the registry is to provide data to the member institutions to improve care. Aggregate data without patient identifiers can be obtained and evaluated by member institutions to support clinical research. Each individual member institution approves data reporting through its local institutional review board.
From the *Department of Cardiac Surgery, University of Michigan, Ann Arbor, Michigan; and †Division of Thoracic Surgery, Department of Surgery, University of Michigan, Ann Arbor, Michigan. Submitted for consideration September 2014; accepted for publication in revised form April 2015. Disclosure: The authors have no conflicts of interest to report. Correspondence: Jonathan W. Haft, Department of Cardiac Surgery, University of Michigan Health System, 1500 E. Medical Center Drive, 5144 CVC/SPC 5864, Ann Arbor, MI 48109. Email: [email protected]. Copyright © 2015 by the American Society for Artificial Internal Organs DOI: 10.1097/MAT.0000000000000243
496
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
497
ECMO BLEEDING COMPLICATIONS
Bleeding Events A “bleeding event” was defined as one of the following complications that occurred during ECMO, as recorded by the ELSO registry: gastrointestinal (GI) hemorrhage (passage of melena, bright red blood per rectum, hematemesis, coffee-ground emesis, or heme-positive stools), cannulation-site hemorrhage, surgical-site hemorrhage, pulmonary hemorrhage (documentation of bleeding during bronchoscopy or copious amounts of hemoptysis or frank blood from the respiratory tract), and neurologic hemorrhage (documented hemorrhage identified by ultrasound or computerized tomography imaging). The presence of these complications was confirmed with our medical records, including information from progress notes and radiology reports. For purposes of analysis, GI hemorrhage, pulmonary hemorrhage, and neurologic hemorrhage were subclassified as “nonsurgical bleeds,” whereas cannulation- and surgical-site bleeding were subclassified as “surgical bleeding.” The incidence rate of bleeding events was calculated as number of bleeding events divided by ECMO patient-days to normalize for different durations of ECMO support. Any bleeding event that occurred within the first 5 days of ECMO was not included for analysis, to eliminate bleeding complications that were likely not related to the pump-mediated bleeding diathesis. Blood Products The amount of blood product transfusions a patient received during ECMO was evaluated to quantify the magnitude of bleeding complications. Data were reported in milliliter per day of packed red blood cells (pRBCs), platelets, frozen fresh plasma, and cryoprecipitate that patients received during ECMO. Anticoagulation Anticoagulation was measured as the average heparin dose in international unit per kilogram per hour while on ECMO. This was calculated by recording the volume of heparin (milliliter) infused during the ECMO run, concentration of heparin in unit per milliliter, and weight of patient (kilogram). Dose was averaged per hour of ECMO run. Four patients in the centrifugal group and one in the roller group developed Heparin induced thrombocytopenia (HIT) and were transitioned to argatroban and thus not included in the anticoagulation analysis. Heparin dose was used as our metric of anticoagulation because the ECMO anticoagulation protocol changed at our institution from point-of-care activated clotting time to activated partial thormboplastin time during the timeframe of our study. Survival Data Survival to decannulation and to discharge was recorded using information from the ELSO database and our hospital records. Survival information is presented as percent of patients alive at end of the given time duration. ECMO Circuits The ECLS equipment was recorded on the ELSO ECLS Registry form at the time of ECMO support. “Roller pump” patients underwent ECMO with the COBE arterial roller pump (COBE Cardiovascular Inc., Arvada, CO) from 2002 to June 2009. “Centrifugal pump” Patients underwent ECMO with the
Levitronix CentriMag Blood pumping system (Thoratec Inc., Pleasanton, CA) from June 2009 to June 2013. Data Collection/Analysis Data were collected using Excel Software (2007, Microsoft Corp., Redmond, WA). Data analysis was performed with Prism (2013, GraphPad Software Inc., San Diego, CA). Categorical variables were summarized by frequencies and percentages and were analyzed using χ2 test. Continuous variables are presented as mean ± standard error of the mean. Student’s t-test for independent samples was used to test for statistically significant differences between the groups. The primary outcome, bleeding events, was calculated as a bleeding rate of events per 1,000 patient-days. A χ2 test based on a Poisson model was used to determine the differences in bleeding rate outcomes. Results Demographics We analyzed data from 95 adult patients undergoing ECMO for at least 5 days from June 2002 to June 2013. There were 47 patients undergoing ECMO with a roller pump from June 2002 to June 2009 and 48 patients with a centrifugal pump from June 2009 to June 2013. Baseline patient demographics, comorbidities, and select pre-ECMO laboratory values are displayed in Table 1. At baseline, there were a higher proportion of patients with diabetes in the centrifugal group (12.8 vs. 2.1%, p = 0.043). The proportion of patients with documented H1N1 infection was expectedly higher in the centrifugal group (10.6 vs. 0%, p = 0.01) as this virus was first documented in 2009. Centrifugal pump patients had a longer duration of ECMO support (468.6 ± 411 vs. 276.4 ± 148 hours, p = 0.004). Indication for ECMO Respiratory failure was the most common indication for ECMO support (79% of all patients; Table 2). There was no significant difference in the indications for ECMO support between centrifugal and roller pump patients. Hemorrhagic Complications Specific bleeding complications were defined by ELSO as described earlier. There was no difference in the total number of bleeding events between the centrifugal and roller pump patients (74.5 vs. 68.7%, p = 0.5; Table 3). To normalize for ECMO duration, we compared overall bleeding rates in events per 1,000 patient-days and found no significant difference in overall bleeding rate between the centrifugal and roller pump patients (65.4 vs. 79.6 events/1,000 patient-days, p = 0.3; Table 4 and Figure 1). In the centrifugal pump group, there was a higher incidence of neurological hemorrhage (5.5 vs. 0 events/1,000 patient-days, p = 0.083), pulmonary hemorrhage (14.2 vs. 5.5 events/1,000 patient-days, p = 0.12), and GI hemorrhage (6.5 vs. 3.6 events/1000 patient-days, p = 0.462), although not statistically significant. In the roller pump patient population, there was a higher incidence of cannulation-site (34.4 vs. 16.4 events/1,000 patient-days, p = 0.028) and surgical-site bleeding (27.1 vs. 13.1 events/1,000 patient-days, p = 0.054). After grouping specific bleeding complications into surgical bleeds
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
498 HALAWEISH et al. Table 1. Baseline Characteristics of Patients Undergoing ECLS from June 2002 to June 2013 Variable
Roller (n = 48)
Centrifugal (n = 47)
p Value
Age (y) Male White COPD Diabetes Influenza A (H1N1) Weight (kg) Hours on support Cannulation sites per patient Percutaneous cannulation VA ECMO VV or VVDL ECMO O+ Blood type Presupport laboratory values Platelet count (103/μl) INR Creatinine (mg/dl) AST (IU/L) ALT (IU/L) Alk phos (IU/L) Total bilirubin (mg/dl) Lactate (mmol/L)
42.5 ± 12.8 28 (58.3%) 30 (62.5%) 1 (2.1%) 1 (2.1%) 0 (0%) 87.2 ± 22.9 276.4 ± 147.6 2.2 ± 0.7 47 (97.9%) 7 (14.6%) 37 (77.0%) 19 (39.6%)
43.9 ± 15.0 32 (68.1%) 34 (72.3%) 4 (8.5%) 6 (12.8%) 5 (10.6%) 92.0 ± 25.1 468.6 ± 411.1 2.0 ± 0.9 44 (93.6%) 11 (23.4%) 29 (61.7%) 18 (38.3%)
0.630 0.322 0.303 0.161 0.043 0.02 0.333 0.004 0.125 0.30 0.271 0.099 0.898
177.6 ± 123.4 (n = 48) 1.3 ± 0.6 (n = 38) 1.9 ± 1.4 (n = 47) 336.4 ± 931.9 (n = 46) 184.1 ± 520.5 (n = 46) 104.8 ± 95.6 (n = 46) 2.1 ± 1.9 (n = 46) 5.0 ± 3.5 (n = 27)
190.9 ± 112.6 (n = 44) 1.3 ± 0.6 (n=44) 1.4 ± 1.4 (n = 44) 574.3 ± 2385.5 (n = 41) 238.6 ± 883.7 (n = 41) 121.5 ± 77.9 (n = 41) 1.4 ± 1.3 (n = 41) 4.6 ± 5.4 (n = 44)
0.592 0.970 0.161 0.533 0.723 0.372 0.042 0.742
ALT, alanine transaminase; Alk phos, alkaline phosphatise; AST, aspartate transaminase; COPD, chronic obstructive pulmonary disease; ECLS, extracorporeal life support; INR, international normalized ratio; VA, venoarterial; VV, venovenous; VVDL, venovenous double lumen.
(cannulation site and surgical site) and nonsurgical bleeds (pulmonary, neurological, and GI bleeds), there was a higher incidence of nonsurgical bleeding in centrifugal pump patients (26.1 vs. 9.0 events/1,000 patient-days, p = 0.024) and a higher incidence of surgical bleeding in the roller pump patients (61.5 vs. 29.4 events/1,000 patient-days, p = 0.003; Figure 2). Blood Product Use There was a higher rate of pRBC transfusions with the use of roller ECMO pumps (1,003 vs. 637 ml/day, p = 0.016). There were no differences in platelet, frozen fresh plasma, or cryoprecipitate transfusion (Table 5). Anticoagulation Centrifugal pump patients received, on average, lower heparin anticoagulation (10.9 vs. 13.7 IU/kg/hr, p = 0.0103) compared with roller pump patients. This reflects change in practice and change in ECMO circuitry, which, our center believes, allows lower anticoagulation targets. Survival Data Survival to decannulation in roller pump patients was higher compared with centrifugal pump patients (72.9 vs. 53.2%, Table 2. Indication for ECMO Run for Centrifugal, Roller, and All Patients
Respiratory Cardiac ECPR
Total (n = 95)
Roller (n = 48)
Centrifugal (n = 47)
75 (78.9%) 14 (14.7%) 6 (6.3%)
40 (83%) 7 (15%) 1 (2%)
35 (74%) 7 (15%) 5 (11%)
ECPR = ECMO cardiopulmonary resuscitation.
p Value 0.287 0.966 0.084
p = 0.046). Roller pump patients also had higher survival to discharge compared with centrifugal pump patients, although statistically insignificant (52.1 vs. 34.0%, p = 0.076). Discussion In this study, we retrospectively compared bleeding complications at our institution of adult patients undergoing prolonged ECMO using roller versus centrifugal pumps. We have demonstrated that bleeding is a common complication during ECMO, occurring in approximately 70% of patients. Although the overall bleeding event rates were similar between both pump designs, centrifugal pumps were associated with higher nonsurgical bleeding per ECMO patient-days (neurological, pulmonary, and GI) compared with roller pumps, despite lower doses of heparin anticoagulation. We elected to study patients who required ECMO support for at least 5 days to specifically identify bleeding episodes that may be related to altered hemostatic processes created by the ECMO circuit and possibly the centrifugal pump. In patients with continuous-flow rotary left ventricular assist device (CFLVAD), GI bleeding typically occurs after the first month of support, although the reduction in vWF multimers is seen earlier.22 We were surprised to see a significant difference in Table 3. Percentage of Patients Who Experienced a Bleeding Event
Cannulation site Surgical site Neurological Pulmonary Gastrointestinal Overall
Roller (n = 47)
Centrifugal (n = 48)
p Value
19 (39.6%) 15 (31.3%) 0 (0.0%) 3 (6.3%) 2 (4.2%) 33 (68.8%)
15 (31.9%) 12 (25.5%) 5 (10.6%) 13 (27.7%) 6 (12.8%) 35 (74.5%)
0.434 0.536 0.020 0.005 0.131 0.536
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
499
ECMO BLEEDING COMPLICATIONS Table 4. Complication Rate in Events Per 1000 Patient-Days for Specific Bleeding Complications
Cannulation site Surgical site Neurological Pulmonary Gastrointestinal Nonsurgical bleeds Surgical bleeds Overall
Roller (n = 47)
Centrifugal (n = 48)
p -Value
34.4 27.1 0.0 5.4 3.6 9.0 61.5 79.6
16.4 13.1 5.5 14.2 6.5 26.1 29.4 65.4
0.028 0.054 0.083 0.120 0.462 0.024 0.003 0.321
Nonsurgical (gastrointestinal, pulmonary, and neurological) and surgical bleeds (cannulation site and surgical site) represent combinations of the specific bleeding types.
nonsurgical bleeding between the two cohorts. It is important to note that we calculated events per 1,000 patient-days to account for the varying duration of ECMO support. Although the overall rates of bleeding were not different in centrifugal pumps and roller pumps, the incidence of cannulation- or surgical-site bleeding was actually higher in the roller pump cohort. We attribute this to evolution of our anticoagulation practices, where we are more likely to withhold heparin when there are bleeding concerns, particularly in the perioperative setting. Given that the rates of percutaneous cannulation were similar in both groups (97.9 vs. 93.6%), it is unlikely that cannulation technique was an important factor. Conversely, despite reduction in anticoagulation, nonsurgical bleeding events were actually increased in the centrifugal pump group; however, we did not explore the temporal relationship between the level of anticoagulation and bleeding because of limitation in ascertaining the exact timing of bleeding events. There were a higher proportion of patients with diabetes in the centrifugal group; however, there are no reports to our knowledge of diabetes as a potential risk factor for bleeding complications. Also, the observed higher rate of pRBC transfusion in the roller pump group is likely because of changes in transfusion protocols with a general trend toward more restrictive strategies in the current era. With the appearance of the H1N1 influenza virus in 2009, there were a higher proportion of patients undergoing centrifugal pump ECMO secondary to H1N1 respiratory failure. The five patients with H1N1 in the centrifugal group accounted for eight hemorrhagic complications including five nonsurgical
Figure 2. Complication rate in events per 1,000 patient-days for surgical (cannulation site and surgical site) and nonsurgical (gastrointestinal, pulmonary, and neurological) bleeds.
bleeds (one GI and four pulmonary) and three surgical bleeds (two cannulation site and one surgical site). This can be potentially a confounding factor as H1N1 has been linked to lower respiratory tract hemorrhage.27 It has been hypothesized that H1N1 infection can aggravate coagulopathy through viralmediated cytokines that enhance the release of unusually large vWF multimers from endothelial cells with consumption of ADAMST1328; however, these data are limited to case reports. It is also possible that the longer duration of ECMO support for patients with H1N1 may lead to higher complication rates; however, in our study, the difference in support length was not statistically significant for patient with respiratory failure secondary to H1N1 infection (629 ± 219 vs. 425 ± 48, p = 0.17). There has been a recent shift in the past decade from the use of roller pumps to centrifugal pumps for ECMO. At our institution, we have completely transitioned to using centrifugal pumps in ECMO since 2009. Centrifugal pumps offer several advantages. Their active suction generates venous flow independent of gravity. This allows miniaturization of the ECMO circuit, both reducing prime and blood transit time and facilitating patient transport. In addition, maximum outflow pressure is substantially less than roller pumps, reducing the risk of “blowout” of the arterial limb of the circuit if distal occlusion occurs. However, early centrifugal pumps were associated with hemolysis and thrombus formation.29,30 The introduction of centrifugal pumps with magnetic levitation and actuation has eliminated these concerns and contributed to their increased utilization.12,14,16 Table 5. Average Amount of Blood Product Transfusions Given to Patients on ECMO for Roller and Centrifugal Pumps
Figure 1. Complication rate in events per 1,000 patient-days for specific bleeding complications.
Packed RBCs (ml) Platelets (ml) FFP (ml) Cryoprecipitate (ml)
Roller (n = 47)
Centrifugal (n = 48)
p Value
1003 ± 906 417.3 ± 265 151 ± 186 2.3 ± 6.5
637 ± 465 373 ± 283 122 ± 164 15 ± 35.7
0.016 0.44 0.422 0.017
FFP, fresh frozen plasma; RBC, red blood cell.
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
500 HALAWEISH et al. In patients with CF-LVADs, there have been reports of increasing bleeding events, as high as 65% in the first year,31,32 with the majority being nonsurgical (GI or nasal mucosal bleeds). This pattern is similar to patients with aortic stenosis and secondary acquired von Willebrand deficiency (AvWD), or Heyde’s syndrome.33 Von Willebrand factor, a 20,000 kDa multimeric glycoprotein, is assembled from 225 kDa monomers. Conformational activation of vWF multimers occurs at sites of vascular injury, turbulent blood flow, and altered shear stress. Subsequently, vWF binds to the subendothelial collagen and platelets to produce the primary hemostatic plug.34,35 It is hypothesized that turbulent flow in CF-LVADs alters shear stress profiles, leading to activation of the enzyme ADAMST13 and subsequent cleavage of vWF multimers, rendering them ineffective in initiating hemostatic response to vascular injury.36–38 Centrifugal ECMO pumps and CF-LVADs share a common pumping mechanism and may result in similar shear forces and their impact on enzymatic cleavage of vWF; however, there is a paucity of data examining acquired vWF deficiency in ECMO. Heilmann et al. examined a series of 32 patients undergoing ECMO and demonstrated that 31 of the 32 patients had reduced vWF multimers, but none in a control population of critically ill patients. We did not measure vWF multimers in our series; however, the increased nonsurgical bleeding is consistent with the syndrome of acquired vWF deficiency.39 The levels of shear required to initiate unfolding of vWF have been reported as low as 2–2.5 Pa with increasing involvement with increasing shear stress.40 These levels are well within the range of shear stress reported for the CentriMag centrifugal pump.41 In fact, the CentriMag pump has been shown to significantly reduce high-molecular-weight vWF after 3 hours in a recirculating in vitro test loop.42 Centrifugal and roller pumps generate flow by different mechanisms. Roller pumps produce peristaltic flow by displacement of fluid from an occlusive rotor applied against a compressible tube. Centrifugal pumps convert rotational energy from a rotor spinning between 2,000 and 4,000 revolutions into kinetic energy as the fluid is thrown radially by “centrifugal” force. This action produces substantially different shear forces on blood, when compared with a peristaltic design pump.43 Although centrifugal pumps for ECMO offer the important advantages of circuit miniaturization and controlled maximum outlet pressures, there may be drawbacks. Shear stresses within centrifugal pumps are known to be design dependent and affected by local features often incorporated for practical reasons such as manufacturability, cost, or performance tradeoff. Opportunities for improvement in pump design of the CentriMag pump remain. Zhang et al.41 reported that despite overall low scalar stress levels in the bulk of the fluid (approximately10 Pa), highest scalar stress levels were found along the outer housing and impeller blade tips (approximately 80 Pa) and moderate stress regions localized in and around the volute-cutwater-outlet region (40 Pa) with recommendations for design improvement made by the authors. Although in vivo comparison of different pumps used in ECMO is needed to identify further mechanistic pathways involved in the different bleeding profiles, reduced shear stress should be an important feature of newly developed pumps in the future. Although our limited data by no means establishes that centrifugal pumps cause more bleeding in ECMO patients, our findings may generate further
investigation. The incidence of GI bleeding in patients with CF-LVAD was unexpected and has had a substantial impact on the effectiveness and utilization of long-term mechanical support of heart failure patients. There are several important limitations to this study. This is a retrospective and single-institution review with a limited number of patients compared during different time periods. In any historical comparison, there will be limitations related to a variety of changes during the evolution of practices over time. In our program, there have been numerous changes in patient selection, timing of ECMO initiation, anticoagulation practices, oxygenators used, and duration of ECMO support. Although these changes could also have contributed to the differences in nonsurgical bleeding observed, we present our interesting observation in light of the growing knowledge of how rotary pumps can cause bleeding diathesis. The historical difference between the two groups of patients could lead to a change in outcomes because of inclusion of sicker patients in the current ECMO era as evidenced by worse survival to decannulation and discharge in the centrifugal group. Also, we only compared one type of roller and centrifugal pump; although these are commonly used pumps, they are not representative of all pump designs available. Furthermore, identification and classification of bleeding events is limited by the quality of medical records, and we may have underestimated the number and different types of bleeding events. Given the nature of the study, blood analysis of vWF levels was not feasible, and thus, we cannot draw a direct link between the type of bleeding observed in our centrifugal pump group and AvWD. Despite the aforementioned limitations, this study is one of the first to identify an increase in nonsurgical bleeding events with centrifugal pumps. Further analysis of larger databases, including the ELSO registry, may shed more light on this topic. With the increasingly sick profile of patients in the current ECMO era, the identification of patients at risk for bleeding complications may allow early diagnosis, possibly through measurement of vWF levels and biological activity and clinical intervention. It may also be important to take into consideration the type of pump used during ECMO support based on the patient’s risk for hemorrhagic complications. References 1. Zangrillo A, Biondi-Zoccai G, Landoni G, et al: Extracorporeal membrane oxygenation (ECMO) in patients with H1N1 influenza infection: A systematic review and meta-analysis including 8 studies and 266 patients receiving ECMO. Crit Care 17: R30, 2013. 2. Davies A, Jones D, Bailey M, et al: Extracorporeal membrane oxygenation for 2009 Influenza A(H1N1) acute respiratory distress syndrome. JAMA 302: 1888–1895, 2009. 3. Beurtheret S, Mastroianni C, Pozzi M, et al: Extracorporeal membrane oxygenation for 2009 influenza A (H1N1) acute respiratory distress syndrome: Single-centre experience with 1-year follow-up. Eur J Cardiothorac Surg 41: 691–695, 2012. 4. Hou X, Guo L, Zhan Q, et al: Extracorporeal membrane oxygenation for critically ill patients with 2009 influenza A (H1N1)-related acute respiratory distress syndrome: Preliminary experience from a single center. Artif Organs 36: 780–786, 2012. 5. Peek GJ, Mugford M, Tiruvoipati R, et al; CESAR Trial Collaboration: Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): A multicentre randomised controlled trial. Lancet 374: 1351–1363, 2009. 6. Rais-Bahrami K, Walton DM, Sell JE, Rivera O, Mikesell GT, Short BL: Improved oxygenation with reduced recirculation during
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
ECMO BLEEDING COMPLICATIONS
venovenous ECMO: Comparison of two catheters. Perfusion 17: 415–419, 2002. 7. Mangoush O, Purkayastha S, Haj-Yahia S, et al: Heparin-bonded circuits versus nonheparin-bonded circuits: An evaluation of their effect on clinical outcomes. Eur J Cardiothorac Surg 31: 1058–1069, 2007. 8. Motomura T, Maeda T, Kawahito S, et al: Development of silicone rubber hollow fiber membrane oxygenator for ECMO. Artif Organs 27: 1050–1053, 2003. 9. Iwahashi H, Yuri K, Nosé Y: Development of the oxygenator: Past, present, and future. J Artif Organs 7: 111–120, 2004. 10. Haworth WS: The development of the modern oxygenator. Ann Thorac Surg 76: S2216–S2219, 2003. 11. Kawahito S, Motomura T, Glueck J, Nosé Y: Development of a new hollow fiber silicone membrane oxygenator for ECMO: The recent progress. Ann Thorac Cardiovasc Surg 8: 268–274, 2002. 12. Mendler N, Podechtl F, Feil G, Hiltmann P, Sebening F: Seal-less centrifugal blood pump with magnetically suspended rotor: Rot-a-flot. Artif Organs 19: 620–624, 1995. 13. Tamari Y, Lee-Sensiba K, King S, Hall MH: An improved bladder for pump control during ECMO procedures. J Extra Corpor Technol 31: 84–90, 1999. 14. Nakazawa T, Makinouchi K, Takami Y, Glueck J, Takatani S, Nosé Y: The effect of the impeller-driver magnetic coupling distance on hemolysis in a compact centrifugal pump. Artif Organs 20: 252–257, 1996. 15. Bottrell S, Bennett M, Augustin S, et al: A comparison study of haemolysis production in three contemporary centrifugal pumps. Perfusion 29: 411–416, 2014. 16. Akamatsu T, Tsukiya T, Nishimura K, Park CH, Nakazeki T: Recent studies of the centrifugal blood pump with a magnetically suspended impeller. Artif Organs 19: 631–634, 1995. 17. Bennett M, Horton S, Thuys C, Augustin S, Rosenberg M, Brizard C: Pump-induced haemolysis: A comparison of short-term ventricular assist devices. Perfusion 19: 107–111, 2004. 18. Thiara AP, Hoel TN, Kristiansen F, Karlsen HM, Fiane AE, Svennevig JL: Evaluation of oxygenators and centrifugal pumps for long-term pediatric extracorporeal membrane oxygenation. Perfusion 22: 323–326, 2007. 19. Yu K, Long C, Hei F, et al: Clinical evaluation of two different extracorporeal membrane oxygenation systems: A single center report. Artif Organs 35: 733–737, 2011. 20. Hirschl RB: Devices, in Zwischenberger JB, Bartlett RH (eds), ECMO: Extracorporeal Cardiopulmonary Support in Critical Care. Ann Arbor, MI, Extracorporeal Life Support Organization, 1995, pp. 150–190. 21. Lawson S, Ellis C, Butler K, McRobb C, Mejak B: Neonatal extracorporeal membrane oxygenation devices, techniques and team roles: 2011 survey results of the United States’ Extracorporeal Life Support Organization centers. J Extra Corpor Technol 43: 236–244, 2011. 22. Uriel N, Pak SW, Jorde UP, et al: Acquired von Willebrand syndrome after continuous-flow mechanical device support contributes to a high prevalence of bleeding during long-term support and at the time of transplantation. J Am Coll Cardiol 56: 1207–1213, 2010. 23. Malehsa D, Meyer AL, Bara C, Strüber M: Acquired von Willebrand syndrome after exchange of the HeartMate XVE to the HeartMate II ventricular assist device. Eur J Cardiothorac Surg 35: 1091–1093, 2009. 24. Crow S, Chen D, Milano C, et al: Acquired von Willebrand syndrome in continuous-flow ventricular assist device recipients. Ann Thorac Surg 90: 1263–1269, 2010; discussion 1269.
501
25. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired von Willebrand syndrome in patients with ventricular assist device or total artificial heart. Thromb Haemost 103: 962–967, 2010. 26. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired Von Willebrand syndrome is an early-onset problem in ventricular assist device patients. Eur J Cardiothorac Surg 40: 1328–1333, 2011; discussion 1233. 27. Kennedy ED, Roy M, Norris J, et al; 2009 Pandemic H1N1 Influenza-Associated Lower Respiratory Tract Hemorrhage Working Group: Lower respiratory tract hemorrhage associated with 2009 pandemic influenza A (H1N1) virus infection. Influenza Other Respir Viruses 7: 761–765, 2013. 28. Akiyama R, Komori I, Hiramoto R, Isonishi A, Matsumoto M, Fujimura Y: H1N1 influenza (swine flu)-associated thrombotic microangiopathy with a markedly high plasma ratio of von Willebrand factor to ADAMTS13. Intern Med 50: 643–647, 2011. 29. Tamari Y, Lee-Sensiba K, Leonard EF, Parnell V, Tortolani AJ: The effects of pressure and flow on hemolysis caused by BioMedicus centrifugal pumps and roller pumps. Guidelines for choosing a blood pump. J Thorac Cardiovasc Surg 106: 997–1007, 1993. 30. Kawahito K, Nosé Y: Hemolysis in different centrifugal pumps. Artif Organs 21: 323–326, 1997. 31. Slaughter MS, Rogers JG, Milano CA, et al; HeartMate II Investigators: Advanced heart failure treated with continuous-flow left ventricular assist device. N Engl J Med 361: 2241–2251, 2009. 32. Crow S, John R, Boyle A, et al: Gastrointestinal bleeding rates in recipients of nonpulsatile and pulsatile left ventricular assist devices. J Thorac Cardiovasc Surg 137: 208–215, 2009. 33. Warkentin TE, Moore JC, Morgan DG: Aortic stenosis and bleeding gastrointestinal angiodysplasia: Is acquired von Willebrand’s disease the link? Lancet 340: 35–37, 1992. 34. Ruggeri ZM: von Willebrand factor. J Clin Invest 100 (11 suppl): S41–S46, 1997. 35. Ruggeri ZM, Ware J: von Willebrand factor. FASEB J 7: 308–316, 1993. 36. Zhang X, Halvorsen K, Zhang CZ, et al: Mechanoenzymatic cleavage of the ultralarge vascular protein von Willebrand factor. Science 324: 1330–1334, 2009. 37. Baldauf C, Schneppenheim R, Stacklies W, et al: Shear-induced unfolding activates von Willebrand factor A2 domain for proteolysis. J Thromb Haemost 7: 2096–2105, 2009. 38. Tsai HM: Shear stress and von Willebrand factor in health and disease. Semin Thromb Hemost 29: 479–488, 2003. 39. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired von Willebrand syndrome in patients with extracorporeal life support (ECLS). Intensive Care Med 38: 62–68, 2012. 40. Vergauwe RM, Uji-i H, De Ceunynck K, Vermant J, Vanhoorelbeke K, Hofkens J: Shear-stress-induced conformational changes of von Willebrand factor in a water-glycerol mixture observed with single molecule microscopy. J Phys Chem B 118: 5660–5669, 2014. 41. Zhang J, Gellman B, Koert A, et al: Computational and experimental evaluation of the fluid dynamics and hemocompatibility of the CentriMag blood pump. Artif Organs 30: 168–177, 2006. 42. Chan CH, Pieper IL, Hambly R, et al: The CentriMag centrifugal blood pump as a benchmark for in vitro testing of hemocompatibility in implantable ventricular assist devices. Artif Organs 39: 93–101, 2015. 43. Ghosh S, Falter F, Cook DJ: Cardiopulmonary Bypass. Cambridge University Press, New York, NY, 2009.
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
Adult Circulatory Support
Roller and Centrifugal Pumps: A Retrospective Comparison of Bleeding Complications in Extracorporeal Membrane Oxygenation Ihab Halaweish,* Adam Cole,* Elaine Cooley,* William R. Lynch,† and Jonathan W. Haft*
Centrifugal pumps are increasingly used for extracorporeal membrane oxygenation (ECMO) rather than roller pumps. However, shear forces induced by these types of continuousflow pumps are associated with acquired von Willebrand factor deficiency and bleeding complications. This study was undertaken to compare adverse bleeding complications with the use of centrifugal and roller pumps in patients on prolonged ECMO support. The records of all adult ECMO patients from June 2002 to 2013 were retrospectively reviewed using the University of Michigan Health System database and the Extracorporeal Life Support Organization registry, focusing on patients supported for at least 5 days. Ninety-five ECMO patients met criteria for inclusion (48 roller vs. 47 centrifugal pump). Indications included pulmonary (79%), cardiac (15%), and extracorporeal cardiopulmonary resuscitation (6%), without significant difference between the two groups. Despite lower heparin anticoagulation (10.9 vs. 13.7 IU/kg/hr) with centrifugal pumps, there was a higher incidence of nonsurgical bleeding (gastrointestinal, pulmonary, and neurological) in centrifugal pump patients (26.1 vs. 9.0 events/1,000 patient-days, p = 0.024). In conclusion, in our historical comparison, despite reduced anticoagulation, ECMO support using centrifugal pumps was associated with a higher incidence of nonsurgical bleeding. The mechanisms behind this are multifactorial and require further investigation ASAIO Journal 2015; 61:496–501.
cannulas,6,7 oxygenators,8–11 and blood pumps.12,13 Contemporary centrifugal pumps are magnetically actuated and are associated with lower hemolysis than predicate devices14–19; as such, there has been a recent shift in the past decade from the use of roller pumps to centrifugal pumps.20,21 However, prolonged use of continuous-flow rotary pumps such as ventricular assist devices has been associated with nonsurgical bleeding from the development of an acquired form of von Willebrand factor (vWF) deficiency.22–26 Whether this phenomenon exists with centrifugal pumps in the ECMO application is unknown. This study was undertaken to compare adverse bleeding complications with the use of centrifugal and roller pumps in patients on prolonged ECMO support. Methods Patients Approval was obtained from the University of Michigan Institutional Review Board. From a prospectively collected database, a retrospective review was performed of all patients aged 18 years or older, who underwent ECMO at the University of Michigan from June 2002 to June 2013. Data on demographics, medical history, indication for ECMO, bleeding events, anticoagulation therapy, and blood product transfusion were obtained from the University of Michigan Health System database and the Extracorporeal Life Support Organization (ELSO) registry. Because our hypothesis was that centrifugal pumps were associated with bleeding complications from their impact on von Willebrand function, only patients undergoing support for at least 5 days or more were included to allow time for this effect to result in measurable differences. The indication for ECMO was recorded from the ELSO database.
Key Words: centrifugal, roller, ECMO, von Willebrand factor deficiency
Extracorporeal membrane oxygenation (ECMO) is used to
mechanically support patients with severe cardiac and respiratory failure. The use of ECMO has increased significantly in recent years, in part from the experience with H1N1 influenza1–4 as well as publication of the conventional ventilation or ECMO for severe adult respiratory failure (CESAR) trial demonstrating efficacy in acute respiratory failure.5 In addition, there have been significant technologic advances in
ELSO Registry The ELSO registry is a large, multicentered, international database of more than 40,000 patient cases receiving extracorporeal life support (ECLS) in more than 170 centers. Member institutions voluntarily submit deidentified data to the registry including sex, race, nature of illness, technical details of extracorporeal support used, complications, and outcome. Quality assurance reports are issued to active ELSO member institutions on a semiannual basis. The purpose of the registry is to provide data to the member institutions to improve care. Aggregate data without patient identifiers can be obtained and evaluated by member institutions to support clinical research. Each individual member institution approves data reporting through its local institutional review board.
From the *Department of Cardiac Surgery, University of Michigan, Ann Arbor, Michigan; and †Division of Thoracic Surgery, Department of Surgery, University of Michigan, Ann Arbor, Michigan. Submitted for consideration September 2014; accepted for publication in revised form April 2015. Disclosure: The authors have no conflicts of interest to report. Correspondence: Jonathan W. Haft, Department of Cardiac Surgery, University of Michigan Health System, 1500 E. Medical Center Drive, 5144 CVC/SPC 5864, Ann Arbor, MI 48109. Email: [email protected]. Copyright © 2015 by the American Society for Artificial Internal Organs DOI: 10.1097/MAT.0000000000000243
496
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
497
ECMO BLEEDING COMPLICATIONS
Bleeding Events A “bleeding event” was defined as one of the following complications that occurred during ECMO, as recorded by the ELSO registry: gastrointestinal (GI) hemorrhage (passage of melena, bright red blood per rectum, hematemesis, coffee-ground emesis, or heme-positive stools), cannulation-site hemorrhage, surgical-site hemorrhage, pulmonary hemorrhage (documentation of bleeding during bronchoscopy or copious amounts of hemoptysis or frank blood from the respiratory tract), and neurologic hemorrhage (documented hemorrhage identified by ultrasound or computerized tomography imaging). The presence of these complications was confirmed with our medical records, including information from progress notes and radiology reports. For purposes of analysis, GI hemorrhage, pulmonary hemorrhage, and neurologic hemorrhage were subclassified as “nonsurgical bleeds,” whereas cannulation- and surgical-site bleeding were subclassified as “surgical bleeding.” The incidence rate of bleeding events was calculated as number of bleeding events divided by ECMO patient-days to normalize for different durations of ECMO support. Any bleeding event that occurred within the first 5 days of ECMO was not included for analysis, to eliminate bleeding complications that were likely not related to the pump-mediated bleeding diathesis. Blood Products The amount of blood product transfusions a patient received during ECMO was evaluated to quantify the magnitude of bleeding complications. Data were reported in milliliter per day of packed red blood cells (pRBCs), platelets, frozen fresh plasma, and cryoprecipitate that patients received during ECMO. Anticoagulation Anticoagulation was measured as the average heparin dose in international unit per kilogram per hour while on ECMO. This was calculated by recording the volume of heparin (milliliter) infused during the ECMO run, concentration of heparin in unit per milliliter, and weight of patient (kilogram). Dose was averaged per hour of ECMO run. Four patients in the centrifugal group and one in the roller group developed Heparin induced thrombocytopenia (HIT) and were transitioned to argatroban and thus not included in the anticoagulation analysis. Heparin dose was used as our metric of anticoagulation because the ECMO anticoagulation protocol changed at our institution from point-of-care activated clotting time to activated partial thormboplastin time during the timeframe of our study. Survival Data Survival to decannulation and to discharge was recorded using information from the ELSO database and our hospital records. Survival information is presented as percent of patients alive at end of the given time duration. ECMO Circuits The ECLS equipment was recorded on the ELSO ECLS Registry form at the time of ECMO support. “Roller pump” patients underwent ECMO with the COBE arterial roller pump (COBE Cardiovascular Inc., Arvada, CO) from 2002 to June 2009. “Centrifugal pump” Patients underwent ECMO with the
Levitronix CentriMag Blood pumping system (Thoratec Inc., Pleasanton, CA) from June 2009 to June 2013. Data Collection/Analysis Data were collected using Excel Software (2007, Microsoft Corp., Redmond, WA). Data analysis was performed with Prism (2013, GraphPad Software Inc., San Diego, CA). Categorical variables were summarized by frequencies and percentages and were analyzed using χ2 test. Continuous variables are presented as mean ± standard error of the mean. Student’s t-test for independent samples was used to test for statistically significant differences between the groups. The primary outcome, bleeding events, was calculated as a bleeding rate of events per 1,000 patient-days. A χ2 test based on a Poisson model was used to determine the differences in bleeding rate outcomes. Results Demographics We analyzed data from 95 adult patients undergoing ECMO for at least 5 days from June 2002 to June 2013. There were 47 patients undergoing ECMO with a roller pump from June 2002 to June 2009 and 48 patients with a centrifugal pump from June 2009 to June 2013. Baseline patient demographics, comorbidities, and select pre-ECMO laboratory values are displayed in Table 1. At baseline, there were a higher proportion of patients with diabetes in the centrifugal group (12.8 vs. 2.1%, p = 0.043). The proportion of patients with documented H1N1 infection was expectedly higher in the centrifugal group (10.6 vs. 0%, p = 0.01) as this virus was first documented in 2009. Centrifugal pump patients had a longer duration of ECMO support (468.6 ± 411 vs. 276.4 ± 148 hours, p = 0.004). Indication for ECMO Respiratory failure was the most common indication for ECMO support (79% of all patients; Table 2). There was no significant difference in the indications for ECMO support between centrifugal and roller pump patients. Hemorrhagic Complications Specific bleeding complications were defined by ELSO as described earlier. There was no difference in the total number of bleeding events between the centrifugal and roller pump patients (74.5 vs. 68.7%, p = 0.5; Table 3). To normalize for ECMO duration, we compared overall bleeding rates in events per 1,000 patient-days and found no significant difference in overall bleeding rate between the centrifugal and roller pump patients (65.4 vs. 79.6 events/1,000 patient-days, p = 0.3; Table 4 and Figure 1). In the centrifugal pump group, there was a higher incidence of neurological hemorrhage (5.5 vs. 0 events/1,000 patient-days, p = 0.083), pulmonary hemorrhage (14.2 vs. 5.5 events/1,000 patient-days, p = 0.12), and GI hemorrhage (6.5 vs. 3.6 events/1000 patient-days, p = 0.462), although not statistically significant. In the roller pump patient population, there was a higher incidence of cannulation-site (34.4 vs. 16.4 events/1,000 patient-days, p = 0.028) and surgical-site bleeding (27.1 vs. 13.1 events/1,000 patient-days, p = 0.054). After grouping specific bleeding complications into surgical bleeds
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
498 HALAWEISH et al. Table 1. Baseline Characteristics of Patients Undergoing ECLS from June 2002 to June 2013 Variable
Roller (n = 48)
Centrifugal (n = 47)
p Value
Age (y) Male White COPD Diabetes Influenza A (H1N1) Weight (kg) Hours on support Cannulation sites per patient Percutaneous cannulation VA ECMO VV or VVDL ECMO O+ Blood type Presupport laboratory values Platelet count (103/μl) INR Creatinine (mg/dl) AST (IU/L) ALT (IU/L) Alk phos (IU/L) Total bilirubin (mg/dl) Lactate (mmol/L)
42.5 ± 12.8 28 (58.3%) 30 (62.5%) 1 (2.1%) 1 (2.1%) 0 (0%) 87.2 ± 22.9 276.4 ± 147.6 2.2 ± 0.7 47 (97.9%) 7 (14.6%) 37 (77.0%) 19 (39.6%)
43.9 ± 15.0 32 (68.1%) 34 (72.3%) 4 (8.5%) 6 (12.8%) 5 (10.6%) 92.0 ± 25.1 468.6 ± 411.1 2.0 ± 0.9 44 (93.6%) 11 (23.4%) 29 (61.7%) 18 (38.3%)
0.630 0.322 0.303 0.161 0.043 0.02 0.333 0.004 0.125 0.30 0.271 0.099 0.898
177.6 ± 123.4 (n = 48) 1.3 ± 0.6 (n = 38) 1.9 ± 1.4 (n = 47) 336.4 ± 931.9 (n = 46) 184.1 ± 520.5 (n = 46) 104.8 ± 95.6 (n = 46) 2.1 ± 1.9 (n = 46) 5.0 ± 3.5 (n = 27)
190.9 ± 112.6 (n = 44) 1.3 ± 0.6 (n=44) 1.4 ± 1.4 (n = 44) 574.3 ± 2385.5 (n = 41) 238.6 ± 883.7 (n = 41) 121.5 ± 77.9 (n = 41) 1.4 ± 1.3 (n = 41) 4.6 ± 5.4 (n = 44)
0.592 0.970 0.161 0.533 0.723 0.372 0.042 0.742
ALT, alanine transaminase; Alk phos, alkaline phosphatise; AST, aspartate transaminase; COPD, chronic obstructive pulmonary disease; ECLS, extracorporeal life support; INR, international normalized ratio; VA, venoarterial; VV, venovenous; VVDL, venovenous double lumen.
(cannulation site and surgical site) and nonsurgical bleeds (pulmonary, neurological, and GI bleeds), there was a higher incidence of nonsurgical bleeding in centrifugal pump patients (26.1 vs. 9.0 events/1,000 patient-days, p = 0.024) and a higher incidence of surgical bleeding in the roller pump patients (61.5 vs. 29.4 events/1,000 patient-days, p = 0.003; Figure 2). Blood Product Use There was a higher rate of pRBC transfusions with the use of roller ECMO pumps (1,003 vs. 637 ml/day, p = 0.016). There were no differences in platelet, frozen fresh plasma, or cryoprecipitate transfusion (Table 5). Anticoagulation Centrifugal pump patients received, on average, lower heparin anticoagulation (10.9 vs. 13.7 IU/kg/hr, p = 0.0103) compared with roller pump patients. This reflects change in practice and change in ECMO circuitry, which, our center believes, allows lower anticoagulation targets. Survival Data Survival to decannulation in roller pump patients was higher compared with centrifugal pump patients (72.9 vs. 53.2%, Table 2. Indication for ECMO Run for Centrifugal, Roller, and All Patients
Respiratory Cardiac ECPR
Total (n = 95)
Roller (n = 48)
Centrifugal (n = 47)
75 (78.9%) 14 (14.7%) 6 (6.3%)
40 (83%) 7 (15%) 1 (2%)
35 (74%) 7 (15%) 5 (11%)
ECPR = ECMO cardiopulmonary resuscitation.
p Value 0.287 0.966 0.084
p = 0.046). Roller pump patients also had higher survival to discharge compared with centrifugal pump patients, although statistically insignificant (52.1 vs. 34.0%, p = 0.076). Discussion In this study, we retrospectively compared bleeding complications at our institution of adult patients undergoing prolonged ECMO using roller versus centrifugal pumps. We have demonstrated that bleeding is a common complication during ECMO, occurring in approximately 70% of patients. Although the overall bleeding event rates were similar between both pump designs, centrifugal pumps were associated with higher nonsurgical bleeding per ECMO patient-days (neurological, pulmonary, and GI) compared with roller pumps, despite lower doses of heparin anticoagulation. We elected to study patients who required ECMO support for at least 5 days to specifically identify bleeding episodes that may be related to altered hemostatic processes created by the ECMO circuit and possibly the centrifugal pump. In patients with continuous-flow rotary left ventricular assist device (CFLVAD), GI bleeding typically occurs after the first month of support, although the reduction in vWF multimers is seen earlier.22 We were surprised to see a significant difference in Table 3. Percentage of Patients Who Experienced a Bleeding Event
Cannulation site Surgical site Neurological Pulmonary Gastrointestinal Overall
Roller (n = 47)
Centrifugal (n = 48)
p Value
19 (39.6%) 15 (31.3%) 0 (0.0%) 3 (6.3%) 2 (4.2%) 33 (68.8%)
15 (31.9%) 12 (25.5%) 5 (10.6%) 13 (27.7%) 6 (12.8%) 35 (74.5%)
0.434 0.536 0.020 0.005 0.131 0.536
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
499
ECMO BLEEDING COMPLICATIONS Table 4. Complication Rate in Events Per 1000 Patient-Days for Specific Bleeding Complications
Cannulation site Surgical site Neurological Pulmonary Gastrointestinal Nonsurgical bleeds Surgical bleeds Overall
Roller (n = 47)
Centrifugal (n = 48)
p -Value
34.4 27.1 0.0 5.4 3.6 9.0 61.5 79.6
16.4 13.1 5.5 14.2 6.5 26.1 29.4 65.4
0.028 0.054 0.083 0.120 0.462 0.024 0.003 0.321
Nonsurgical (gastrointestinal, pulmonary, and neurological) and surgical bleeds (cannulation site and surgical site) represent combinations of the specific bleeding types.
nonsurgical bleeding between the two cohorts. It is important to note that we calculated events per 1,000 patient-days to account for the varying duration of ECMO support. Although the overall rates of bleeding were not different in centrifugal pumps and roller pumps, the incidence of cannulation- or surgical-site bleeding was actually higher in the roller pump cohort. We attribute this to evolution of our anticoagulation practices, where we are more likely to withhold heparin when there are bleeding concerns, particularly in the perioperative setting. Given that the rates of percutaneous cannulation were similar in both groups (97.9 vs. 93.6%), it is unlikely that cannulation technique was an important factor. Conversely, despite reduction in anticoagulation, nonsurgical bleeding events were actually increased in the centrifugal pump group; however, we did not explore the temporal relationship between the level of anticoagulation and bleeding because of limitation in ascertaining the exact timing of bleeding events. There were a higher proportion of patients with diabetes in the centrifugal group; however, there are no reports to our knowledge of diabetes as a potential risk factor for bleeding complications. Also, the observed higher rate of pRBC transfusion in the roller pump group is likely because of changes in transfusion protocols with a general trend toward more restrictive strategies in the current era. With the appearance of the H1N1 influenza virus in 2009, there were a higher proportion of patients undergoing centrifugal pump ECMO secondary to H1N1 respiratory failure. The five patients with H1N1 in the centrifugal group accounted for eight hemorrhagic complications including five nonsurgical
Figure 2. Complication rate in events per 1,000 patient-days for surgical (cannulation site and surgical site) and nonsurgical (gastrointestinal, pulmonary, and neurological) bleeds.
bleeds (one GI and four pulmonary) and three surgical bleeds (two cannulation site and one surgical site). This can be potentially a confounding factor as H1N1 has been linked to lower respiratory tract hemorrhage.27 It has been hypothesized that H1N1 infection can aggravate coagulopathy through viralmediated cytokines that enhance the release of unusually large vWF multimers from endothelial cells with consumption of ADAMST1328; however, these data are limited to case reports. It is also possible that the longer duration of ECMO support for patients with H1N1 may lead to higher complication rates; however, in our study, the difference in support length was not statistically significant for patient with respiratory failure secondary to H1N1 infection (629 ± 219 vs. 425 ± 48, p = 0.17). There has been a recent shift in the past decade from the use of roller pumps to centrifugal pumps for ECMO. At our institution, we have completely transitioned to using centrifugal pumps in ECMO since 2009. Centrifugal pumps offer several advantages. Their active suction generates venous flow independent of gravity. This allows miniaturization of the ECMO circuit, both reducing prime and blood transit time and facilitating patient transport. In addition, maximum outflow pressure is substantially less than roller pumps, reducing the risk of “blowout” of the arterial limb of the circuit if distal occlusion occurs. However, early centrifugal pumps were associated with hemolysis and thrombus formation.29,30 The introduction of centrifugal pumps with magnetic levitation and actuation has eliminated these concerns and contributed to their increased utilization.12,14,16 Table 5. Average Amount of Blood Product Transfusions Given to Patients on ECMO for Roller and Centrifugal Pumps
Figure 1. Complication rate in events per 1,000 patient-days for specific bleeding complications.
Packed RBCs (ml) Platelets (ml) FFP (ml) Cryoprecipitate (ml)
Roller (n = 47)
Centrifugal (n = 48)
p Value
1003 ± 906 417.3 ± 265 151 ± 186 2.3 ± 6.5
637 ± 465 373 ± 283 122 ± 164 15 ± 35.7
0.016 0.44 0.422 0.017
FFP, fresh frozen plasma; RBC, red blood cell.
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
500 HALAWEISH et al. In patients with CF-LVADs, there have been reports of increasing bleeding events, as high as 65% in the first year,31,32 with the majority being nonsurgical (GI or nasal mucosal bleeds). This pattern is similar to patients with aortic stenosis and secondary acquired von Willebrand deficiency (AvWD), or Heyde’s syndrome.33 Von Willebrand factor, a 20,000 kDa multimeric glycoprotein, is assembled from 225 kDa monomers. Conformational activation of vWF multimers occurs at sites of vascular injury, turbulent blood flow, and altered shear stress. Subsequently, vWF binds to the subendothelial collagen and platelets to produce the primary hemostatic plug.34,35 It is hypothesized that turbulent flow in CF-LVADs alters shear stress profiles, leading to activation of the enzyme ADAMST13 and subsequent cleavage of vWF multimers, rendering them ineffective in initiating hemostatic response to vascular injury.36–38 Centrifugal ECMO pumps and CF-LVADs share a common pumping mechanism and may result in similar shear forces and their impact on enzymatic cleavage of vWF; however, there is a paucity of data examining acquired vWF deficiency in ECMO. Heilmann et al. examined a series of 32 patients undergoing ECMO and demonstrated that 31 of the 32 patients had reduced vWF multimers, but none in a control population of critically ill patients. We did not measure vWF multimers in our series; however, the increased nonsurgical bleeding is consistent with the syndrome of acquired vWF deficiency.39 The levels of shear required to initiate unfolding of vWF have been reported as low as 2–2.5 Pa with increasing involvement with increasing shear stress.40 These levels are well within the range of shear stress reported for the CentriMag centrifugal pump.41 In fact, the CentriMag pump has been shown to significantly reduce high-molecular-weight vWF after 3 hours in a recirculating in vitro test loop.42 Centrifugal and roller pumps generate flow by different mechanisms. Roller pumps produce peristaltic flow by displacement of fluid from an occlusive rotor applied against a compressible tube. Centrifugal pumps convert rotational energy from a rotor spinning between 2,000 and 4,000 revolutions into kinetic energy as the fluid is thrown radially by “centrifugal” force. This action produces substantially different shear forces on blood, when compared with a peristaltic design pump.43 Although centrifugal pumps for ECMO offer the important advantages of circuit miniaturization and controlled maximum outlet pressures, there may be drawbacks. Shear stresses within centrifugal pumps are known to be design dependent and affected by local features often incorporated for practical reasons such as manufacturability, cost, or performance tradeoff. Opportunities for improvement in pump design of the CentriMag pump remain. Zhang et al.41 reported that despite overall low scalar stress levels in the bulk of the fluid (approximately10 Pa), highest scalar stress levels were found along the outer housing and impeller blade tips (approximately 80 Pa) and moderate stress regions localized in and around the volute-cutwater-outlet region (40 Pa) with recommendations for design improvement made by the authors. Although in vivo comparison of different pumps used in ECMO is needed to identify further mechanistic pathways involved in the different bleeding profiles, reduced shear stress should be an important feature of newly developed pumps in the future. Although our limited data by no means establishes that centrifugal pumps cause more bleeding in ECMO patients, our findings may generate further
investigation. The incidence of GI bleeding in patients with CF-LVAD was unexpected and has had a substantial impact on the effectiveness and utilization of long-term mechanical support of heart failure patients. There are several important limitations to this study. This is a retrospective and single-institution review with a limited number of patients compared during different time periods. In any historical comparison, there will be limitations related to a variety of changes during the evolution of practices over time. In our program, there have been numerous changes in patient selection, timing of ECMO initiation, anticoagulation practices, oxygenators used, and duration of ECMO support. Although these changes could also have contributed to the differences in nonsurgical bleeding observed, we present our interesting observation in light of the growing knowledge of how rotary pumps can cause bleeding diathesis. The historical difference between the two groups of patients could lead to a change in outcomes because of inclusion of sicker patients in the current ECMO era as evidenced by worse survival to decannulation and discharge in the centrifugal group. Also, we only compared one type of roller and centrifugal pump; although these are commonly used pumps, they are not representative of all pump designs available. Furthermore, identification and classification of bleeding events is limited by the quality of medical records, and we may have underestimated the number and different types of bleeding events. Given the nature of the study, blood analysis of vWF levels was not feasible, and thus, we cannot draw a direct link between the type of bleeding observed in our centrifugal pump group and AvWD. Despite the aforementioned limitations, this study is one of the first to identify an increase in nonsurgical bleeding events with centrifugal pumps. Further analysis of larger databases, including the ELSO registry, may shed more light on this topic. With the increasingly sick profile of patients in the current ECMO era, the identification of patients at risk for bleeding complications may allow early diagnosis, possibly through measurement of vWF levels and biological activity and clinical intervention. It may also be important to take into consideration the type of pump used during ECMO support based on the patient’s risk for hemorrhagic complications. References 1. Zangrillo A, Biondi-Zoccai G, Landoni G, et al: Extracorporeal membrane oxygenation (ECMO) in patients with H1N1 influenza infection: A systematic review and meta-analysis including 8 studies and 266 patients receiving ECMO. Crit Care 17: R30, 2013. 2. Davies A, Jones D, Bailey M, et al: Extracorporeal membrane oxygenation for 2009 Influenza A(H1N1) acute respiratory distress syndrome. JAMA 302: 1888–1895, 2009. 3. Beurtheret S, Mastroianni C, Pozzi M, et al: Extracorporeal membrane oxygenation for 2009 influenza A (H1N1) acute respiratory distress syndrome: Single-centre experience with 1-year follow-up. Eur J Cardiothorac Surg 41: 691–695, 2012. 4. Hou X, Guo L, Zhan Q, et al: Extracorporeal membrane oxygenation for critically ill patients with 2009 influenza A (H1N1)-related acute respiratory distress syndrome: Preliminary experience from a single center. Artif Organs 36: 780–786, 2012. 5. Peek GJ, Mugford M, Tiruvoipati R, et al; CESAR Trial Collaboration: Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): A multicentre randomised controlled trial. Lancet 374: 1351–1363, 2009. 6. Rais-Bahrami K, Walton DM, Sell JE, Rivera O, Mikesell GT, Short BL: Improved oxygenation with reduced recirculation during
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
ECMO BLEEDING COMPLICATIONS
venovenous ECMO: Comparison of two catheters. Perfusion 17: 415–419, 2002. 7. Mangoush O, Purkayastha S, Haj-Yahia S, et al: Heparin-bonded circuits versus nonheparin-bonded circuits: An evaluation of their effect on clinical outcomes. Eur J Cardiothorac Surg 31: 1058–1069, 2007. 8. Motomura T, Maeda T, Kawahito S, et al: Development of silicone rubber hollow fiber membrane oxygenator for ECMO. Artif Organs 27: 1050–1053, 2003. 9. Iwahashi H, Yuri K, Nosé Y: Development of the oxygenator: Past, present, and future. J Artif Organs 7: 111–120, 2004. 10. Haworth WS: The development of the modern oxygenator. Ann Thorac Surg 76: S2216–S2219, 2003. 11. Kawahito S, Motomura T, Glueck J, Nosé Y: Development of a new hollow fiber silicone membrane oxygenator for ECMO: The recent progress. Ann Thorac Cardiovasc Surg 8: 268–274, 2002. 12. Mendler N, Podechtl F, Feil G, Hiltmann P, Sebening F: Seal-less centrifugal blood pump with magnetically suspended rotor: Rot-a-flot. Artif Organs 19: 620–624, 1995. 13. Tamari Y, Lee-Sensiba K, King S, Hall MH: An improved bladder for pump control during ECMO procedures. J Extra Corpor Technol 31: 84–90, 1999. 14. Nakazawa T, Makinouchi K, Takami Y, Glueck J, Takatani S, Nosé Y: The effect of the impeller-driver magnetic coupling distance on hemolysis in a compact centrifugal pump. Artif Organs 20: 252–257, 1996. 15. Bottrell S, Bennett M, Augustin S, et al: A comparison study of haemolysis production in three contemporary centrifugal pumps. Perfusion 29: 411–416, 2014. 16. Akamatsu T, Tsukiya T, Nishimura K, Park CH, Nakazeki T: Recent studies of the centrifugal blood pump with a magnetically suspended impeller. Artif Organs 19: 631–634, 1995. 17. Bennett M, Horton S, Thuys C, Augustin S, Rosenberg M, Brizard C: Pump-induced haemolysis: A comparison of short-term ventricular assist devices. Perfusion 19: 107–111, 2004. 18. Thiara AP, Hoel TN, Kristiansen F, Karlsen HM, Fiane AE, Svennevig JL: Evaluation of oxygenators and centrifugal pumps for long-term pediatric extracorporeal membrane oxygenation. Perfusion 22: 323–326, 2007. 19. Yu K, Long C, Hei F, et al: Clinical evaluation of two different extracorporeal membrane oxygenation systems: A single center report. Artif Organs 35: 733–737, 2011. 20. Hirschl RB: Devices, in Zwischenberger JB, Bartlett RH (eds), ECMO: Extracorporeal Cardiopulmonary Support in Critical Care. Ann Arbor, MI, Extracorporeal Life Support Organization, 1995, pp. 150–190. 21. Lawson S, Ellis C, Butler K, McRobb C, Mejak B: Neonatal extracorporeal membrane oxygenation devices, techniques and team roles: 2011 survey results of the United States’ Extracorporeal Life Support Organization centers. J Extra Corpor Technol 43: 236–244, 2011. 22. Uriel N, Pak SW, Jorde UP, et al: Acquired von Willebrand syndrome after continuous-flow mechanical device support contributes to a high prevalence of bleeding during long-term support and at the time of transplantation. J Am Coll Cardiol 56: 1207–1213, 2010. 23. Malehsa D, Meyer AL, Bara C, Strüber M: Acquired von Willebrand syndrome after exchange of the HeartMate XVE to the HeartMate II ventricular assist device. Eur J Cardiothorac Surg 35: 1091–1093, 2009. 24. Crow S, Chen D, Milano C, et al: Acquired von Willebrand syndrome in continuous-flow ventricular assist device recipients. Ann Thorac Surg 90: 1263–1269, 2010; discussion 1269.
501
25. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired von Willebrand syndrome in patients with ventricular assist device or total artificial heart. Thromb Haemost 103: 962–967, 2010. 26. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired Von Willebrand syndrome is an early-onset problem in ventricular assist device patients. Eur J Cardiothorac Surg 40: 1328–1333, 2011; discussion 1233. 27. Kennedy ED, Roy M, Norris J, et al; 2009 Pandemic H1N1 Influenza-Associated Lower Respiratory Tract Hemorrhage Working Group: Lower respiratory tract hemorrhage associated with 2009 pandemic influenza A (H1N1) virus infection. Influenza Other Respir Viruses 7: 761–765, 2013. 28. Akiyama R, Komori I, Hiramoto R, Isonishi A, Matsumoto M, Fujimura Y: H1N1 influenza (swine flu)-associated thrombotic microangiopathy with a markedly high plasma ratio of von Willebrand factor to ADAMTS13. Intern Med 50: 643–647, 2011. 29. Tamari Y, Lee-Sensiba K, Leonard EF, Parnell V, Tortolani AJ: The effects of pressure and flow on hemolysis caused by BioMedicus centrifugal pumps and roller pumps. Guidelines for choosing a blood pump. J Thorac Cardiovasc Surg 106: 997–1007, 1993. 30. Kawahito K, Nosé Y: Hemolysis in different centrifugal pumps. Artif Organs 21: 323–326, 1997. 31. Slaughter MS, Rogers JG, Milano CA, et al; HeartMate II Investigators: Advanced heart failure treated with continuous-flow left ventricular assist device. N Engl J Med 361: 2241–2251, 2009. 32. Crow S, John R, Boyle A, et al: Gastrointestinal bleeding rates in recipients of nonpulsatile and pulsatile left ventricular assist devices. J Thorac Cardiovasc Surg 137: 208–215, 2009. 33. Warkentin TE, Moore JC, Morgan DG: Aortic stenosis and bleeding gastrointestinal angiodysplasia: Is acquired von Willebrand’s disease the link? Lancet 340: 35–37, 1992. 34. Ruggeri ZM: von Willebrand factor. J Clin Invest 100 (11 suppl): S41–S46, 1997. 35. Ruggeri ZM, Ware J: von Willebrand factor. FASEB J 7: 308–316, 1993. 36. Zhang X, Halvorsen K, Zhang CZ, et al: Mechanoenzymatic cleavage of the ultralarge vascular protein von Willebrand factor. Science 324: 1330–1334, 2009. 37. Baldauf C, Schneppenheim R, Stacklies W, et al: Shear-induced unfolding activates von Willebrand factor A2 domain for proteolysis. J Thromb Haemost 7: 2096–2105, 2009. 38. Tsai HM: Shear stress and von Willebrand factor in health and disease. Semin Thromb Hemost 29: 479–488, 2003. 39. Heilmann C, Geisen U, Beyersdorf F, et al: Acquired von Willebrand syndrome in patients with extracorporeal life support (ECLS). Intensive Care Med 38: 62–68, 2012. 40. Vergauwe RM, Uji-i H, De Ceunynck K, Vermant J, Vanhoorelbeke K, Hofkens J: Shear-stress-induced conformational changes of von Willebrand factor in a water-glycerol mixture observed with single molecule microscopy. J Phys Chem B 118: 5660–5669, 2014. 41. Zhang J, Gellman B, Koert A, et al: Computational and experimental evaluation of the fluid dynamics and hemocompatibility of the CentriMag blood pump. Artif Organs 30: 168–177, 2006. 42. Chan CH, Pieper IL, Hambly R, et al: The CentriMag centrifugal blood pump as a benchmark for in vitro testing of hemocompatibility in implantable ventricular assist devices. Artif Organs 39: 93–101, 2015. 43. Ghosh S, Falter F, Cook DJ: Cardiopulmonary Bypass. Cambridge University Press, New York, NY, 2009.
Copyright © American Society of Artificial Internal Organs. Unauthorized reproduction of this article is prohibited.
