Von Willebrand Factor-GP1bα Interactions in Venoarterial Extracorporeal Membrane Oxygenation Patients

ARTICLE IN PRESS Journal of Cardiothoracic and Vascular Anesthesia 000 (2018) 18

Contents lists available at ScienceDirect

Journal of Cardiothoracic and Vascular Anesthesia journal homepage: www.jcvaonline.com

Original Article

Von Willebrand Factor-GP1ba Interactions in Venoarterial Extracorporeal Membrane Oxygenation Patients 1

D3X XMichael Mazzeffi, D4X XMD, MPH, MSc*, , D5X XShaheer Hasan, D6X XMSc*, D7X XEzeldeen Abuelkasem, D8X XMDy, D9X XMichael Meyer, D10X XMScz, D1X XKristopher Deatrick, D12X XMDx, D13X XBradley Taylor, D14X XMD, MPHx, D15X XZachary Kon, D16X XMD||, D17X XDaniel Herr, MDD{18X X , D19X XKenichi Tanaka, D20X XMD, MSc* *

Department of Anesthesiology, University of Maryland School of Medicine, Baltimore, MD Department of Anesthesiology, University of Pittsburgh School of Medicine, Pittsburgh, PA z Institute for Transfusion Medicine, Pittsburgh, PA x Department of Surgery, Division of Cardiothoracic Surgery, University of Maryland School of Medicine, Baltimore, MD || Department of Surgery, Division of Cardiothoracic Surgery, Langone Health, New York University School of Medicine, New York, NY { Department of Medicine, Shock Trauma Critical Care, University of Maryland, Baltimore, MD y

Objective: Evaluate extracorporeal membrane oxygenation (ECMO) patients’ platelet adhesion and aggregation under shear stress and determine whether addition of von Willebrand factor (VWF) concentrate improves platelet function. Also, explore whether reduced platelet adhesion and aggregation is associated with clinical bleeding during ECMO. Design: Prospective observational cohort study with translational component. Setting: Academic medical center. Participants: Consecutive venoarterial (VA) ECMO patients were screened and 20 patients enrolled. Interventions: VWF multimers, VWF antigen, ristocetin cofactor activity, and plasma glycocalicin levels were measured and values were compared at study points: ECMO day 1 or 2, day 3, and day 5. Platelet adhesion and aggregation were measured in vitro using the total thrombus analysis system. Platelet function was expressed as area under the flow-pressure curve (AUC). VWF concentrate was added in vitro and the AUC after VWF supplementation (VWF AUC) was compared with baseline AUC. Further, baseline AUCs and VWF AUCs were compared between patients who experienced bleeding during ECMO and those who did not. Measurements and Main Results: ECMO patients had high VWF antigen levels, high ristocetin cofactor activity, and large VWF multimer loss. Platelet counts fell over the first 5 days on ECMO, and plasma glycocalicin levels were elevated mildly. ECMO patients had severely low platelet adhesion and aggregation in vitro: median AUC = 5.8 (3.5-9.7) ECMO day 1 or 2, median AUC = 6.3 (5.3-11.1) day 3, and median AUC = 5.5 (4.1-8.1) day 5. There was no significant change in AUC over time (p = 0.47). Addition of VWF concentrate increased the AUC compared to baseline at each point (all p < 0.05), but VWF AUC values remained low. Patients with bleeding during ECMO had a low VWF AUC at all points, whereas those without bleeding had a higher VWF AUC on ECMO day 3. Conclusions: VA ECMO patients have severely impaired platelet function, which improved but did not normalize with VWF concentrate. The data suggest that GP1ba receptor loss of dysfunction also contributes to impaired platelet adhesion and aggregation

This study was funded by a Society of Cardiovascular Anesthesiologists Starter Grant, which was awarded in 2016. 1 Address reprint requests to Michael Mazzeffi, MD, MPH, MSc, Department of Anesthesiology, University of Maryland School of Medicine, 22 South Greene Street, Baltimore, MD 21201. E-mail address: [email protected] (M. Mazzeffi). https://doi.org/10.1053/j.jvca.2018.11.031 1053-0770/Ó 2018 Elsevier Inc. All rights reserved.

ARTICLE IN PRESS 2

M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

during ECMO. Based on these findings, clinical bleeding in ECMO patients is unlikely to be correctable with VWF supplementation alone. Ó 2018 Elsevier Inc. All rights reserved. Key Words: bleeding; ECMO; coagulation; VWF

THE USE OF adult extracorporeal membrane oxygenation (ECMO) has increased substantially during the past 20 years. According to the Extracorporeal Life Support Organization registry, there were approximately 1,800 ECMO cases in 2000, which grew to 10,000 cases in 2017. During the same period, the number of ECMO centers tripled.1 Despite this rapid growth in ECMO use, survival to discharge remains around 40% for venoarterial (VA) ECMO patients with cardiogenic shock. One reason for this relatively low survival is that patients on VA ECMO continue to have high complication rates, particularly bleeding. In a contemporary meta-analysis of 1,900 VA ECMO patients, 40% had bleeding complications.2 In another detailed epidemiologic study of bleeding complications during ECMO, the incidence of bleeding events was 19 per 100 VA ECMO days with 69% of patients having at least 1 bleeding event.3 There are a number of putative mechanisms that contribute to bleeding during ECMO, including impaired platelet function, fibrinolysis, and acquired von Willebrand disease (VWD), which is thought to be related to loss of large von Willebrand factor (VWF) multimers during mechanical circulatory support.47 When blood is exposed to high shear stress, the conformation of VWF changes, facilitating cleavage of large multimers by the metalloproteinase ADAMTS13 at the A2 domain.810 Although loss of large VWF multimers is well described, there is a paucity of data on glycoprotein 1b alpha (GP1ba) receptor function during ECMO. The GP1ba receptor is highly prevalent on platelet surfaces, approximately 25,000 copies per platelet, and is critical for platelet binding to exposed collagen via the A1 domain of VWF.1112 In one study of 20 ECMO patients, 80% had loss of large VWF multimers and there was also significant shedding of the GP1ba ectodomain (glycocalicin), suggesting that both acquired VWD and acquired Bernard-Soulier syndrome occur during ECMO.13 The primary study hypothesis was that VA ECMO patients have abnormal platelet adhesion and aggregation under shear stress. Secondarily, the authors hypothesized that in vitro supplementation with VWF concentrate would restore normal platelet adhesion and aggregation and that patients with clinical bleeding during ECMO would have lower platelet adhesion and aggregation compared to those without bleeding during ECMO.

Methods Subjects The institutional review board at the University of Maryland, Baltimore approved the study, and written informed

consent was obtained from all patients or their legally authorized representative. The Strengthening The Reporting of Observational Studies in Epidemiology statement checklist for cohort studies was referenced in preparing the manuscript. Consecutive VA ECMO patients were screened for the study, and 20 adult VA ECMO patients with cardiogenic shock were enrolled. All patients were enrolled during their first 48 hours on ECMO. Patients who were not willing to accept allogeneic blood transfusion were excluded because of the risk of worsening anemia with repeated blood draws. Study Data For all subjects, the authors collected demographic data and medical history: hypertension, diabetes mellitus, congestive heart failure, left ventricular ejection fraction, acute liver injury, antiplatelet drug use within 5 days of initiating ECMO, previous stroke, and baseline creatinine level. The authors also collected details about ECMO: (total ECMO days, etiology of shock, platelet count, and blood transfusion data) and outcome data (acute renal failure requiring renal replacement therapy, bleeding complications, thrombotic complications, and in-hospital mortality). Bleeding complications were defined as bleeding that required surgical treatment or transfusion of at least 2 red blood cell units. Evaluation of VWF Antigen, VWF Ristocetin Cofactor Activity, and VWF Multimers All patients had VWF antigen (VWF Ag), VWF ristocetin cofactor activity, and VWF multimer analysis performed on ECMO days 1 or 2, day 3, and day 5. Plasma was collected from whole blood in citrated tubes (3.2%) by double centrifugation at 1500xg and stored at ¡80˚C until analysis. Plasmabased coagulation assays were performed using commercially available kits at the Institute for Transfusion Medicine (Pittsburgh, PA). VWF Ag levels and ristocetin cofactor activity were determined using STA-Liatest VWF:Ag reagent (Diagnostica Stago, Asnieres sur, Seine, France) and BC VWF reagent (Siemens, Marburg, Germany), respectively, according to the manufacturer’s directions. Both assays were performed on a Siemens BCS XP coagulation analyzer. For VWF multimer analysis, VWF was separated by Western blotting on low-resolution agarose gel (1.2% agarose), and blotted on a polyvinylidene difluoride membrane as outlined by Raines et al.14 VWF multimers were detected by appropriate primary and secondary antibodies, and chemiluminescence as described previously. The amount of sample loaded for all patients and the normal pooled plasma control were

ARTICLE IN PRESS M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

standardized to approximately 1.0 U/mL of VWF Ag. A normal pooled plasma control and a commercially available type 2B VWD control (George King Bio-Medical, Overland Park, KS) were included with each VWF multimers analysis as reference points. The VWD sample was used as a control so that large multimer loss in ECMO patients could be compared against that of a patient with known type 2B VWD. Lengths of multimer bands in ECMO patients were compared against band lengths in pooled plasma controls to quantify the degree of multimer loss. Measurement of Plasma Glycocalicin Levels Glycocalicin is a large proteolytic fragment of the GP1ba receptor ectodomain that can be measured in human plasma and is indicative of GP1ba proteolysis.15 Glycocalicin levels were measured using an enzyme-linked immunosorbent sandwich assay (MyBioSource, Vancouver, Canada). Patient plasma (100 mL) was added to 96 well plates precoated with anti-glycocalicin antibody and incubated for 2 hours at 37˚C. After removal of liquid from the wells, Biotin antibody was added to each well and the plate was incubated for 1 hour at 37˚C. Wells were aspirated and washed twice with buffer solution per the manufacturer’s instructions. Then 100 uL of horseradish peroxidase-avidin was added to each well and the plate was incubated for an additional hour at 37˚C. After washing, tetramethylbenzidine substrate was added to the wells and the plate was incubated for 30 minutes at 37˚C. Stop solution was added and the optical density of each well was measured within 5 minutes using an EMax microplate reader (Molecular Devices, San Jose, CA) set to 450 nm. Glycocalicin levels were calculated using a standard curve. The glycocalicin index (GI) was calculated for each sample to normalize glycocalicin levels for the platelet count. The GI was calculated using a previously described formula: GI = glycocalicin concentration as percent of normal  (platelet count/250,000 platelets per microliter). A normal plasma glycocalicin level was assumed to be 2 mg/mL for this calculation.16 This level was selected based on previously published levels in healthy volunteers.16 In Vitro Evaluation of VWF-GP1ba Interactions VWF-GP1ba interactions were studied in vitro using the Total Thrombus-formation Analysis System (T-TAS) (Fujimori Kogyo Ltd., Tokyo, Japan).1718 For each assay, the platelet chip, which contains 25 capillary channels coated with type 1 collagen, was used. Whole blood was added to the system from a hirudin blood tube (330 mL), and mineral oil was used as a carrier to create blood flow at a shear stress rate of 2000 s¡1 or 24 mL/min. Platelet adhesion and aggregation was expressed as an area under the flow-pressure curve (AUC), with a high AUC indicating more robust platelet adhesion and aggregation. At each study point (ECMO days 1 or 2, 3, and 5), patients had a baseline assay run and also a VWFtreated assay where VWF concentrate (Wilate, Octapharma AG, Lachen, Switzerland) was added to whole blood (final concentration 1.5 IU/mL). This concentration equates to

3

approximately a 50 IU/kg dose in a 70-kg patient. Videos were stored of the baseline assay and VWF-treated assay for all study points in each patient. Statistical Analysis Statistical analysis was performed using SAS 9.3 (SAS Corp., Cary, NC). A sample size of 20 patients was determined to be sufficient to detect an increase in AUC of 100 with VWF concentrate supplementation, assuming a standard deviation of 100 for AUC, 80% power, and an alpha of 0.05. Continuous study variables were summarized as the mean + standard deviation or median (Q1-Q3) depending on normality. Categorical data were summarized as the number and percentage of patients. VWF Ag, VWF ristocetin cofactor activity, platelet count, and plasma glycocalicin levels were summarized for each study point (ECMO day 1 or 2, day 3, and day 5). Temporal trends in these variables were analyzed using either repeated measures analysis of variance or Friedman’s test depending on normality. A p value less than 0.05 was considered statistically significant. Flow-mediated platelet adhesion and aggregation (expressed as AUC) was summarized as median (Q1-Q3) for each study point both at baseline and after treatment with VWF concentrate (VWF AUC). The difference in AUC between baseline samples and VWF-treated samples was tested at each point using the Wilcoxon rank sum test. The authors also tested for differences in AUC and VWF AUC over time using Friedman’s test. To explore the relationships among clinical bleeding, AUC, and VWF AUC, the authors stratified patients in the cohort into 2 groups based on whether they had clinical bleeding at any time during ECMO and compared AUC and VWF AUC values between the 2 groups using the Wilcoxon rank sum test. Results Twenty VA ECMO patients with cardiogenic shock were enrolled in the study. Patient characteristics are shown in Table 1. Most patients (n = 10, 50.0%) were on ECMO for submassive or massive pulmonary embolism. The median duration of ECMO was 9 days. Nine patients in the cohort (45.0%) had bleeding complications, and in-hospital mortality was 30.0% for the cohort. Median red blood cell transfusion was 7 units, fresh frozen plasma transfusion was 1 unit, platelet transfusion was 2 units, and no patient in the cohort was transfused cryoprecipitate during ECMO. There was no desmopressin use in any patient. Table 2 shows VWF Ag levels, VWF ristocetin cofactor activity, platelet counts, and plasma glycocalicin levels at the 3 study points. VWF Ag levels were high at all 3 points compared to assay normal values, as were ristocetin cofactor activity levels. There were no obvious temporal trends in either parameter. Platelet counts decreased significantly over the 5 days on ECMO (p = 0.003) from a mean value of 108 £ 109 on day 1 or 2 to a mean value of 84 £ 109 on day 5. Plasma glycocalicin levels were elevated mildly on ECMO day 1 or 2 and day 5 as was the GI, but there was no apparent temporal

ARTICLE IN PRESS 4

M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

Table 1 Patient Data Variable

Mean + SD, Median (Q1-Q3), or n (%)

Age Sex (% male) Blood type A B AB O Indication for ECMO Pulmonary embolism Post cardiotomy shock Myocardial infarction Cardiac arrest Other ECPR Hypertension Diabetes mellitus Congestive heart failure LVEF (%) Shock liver Antiplatelet drugs within 5 days of ECMO initiation* Previous stroke Baseline creatinine (mg/dL) Total ECMO days RBC transfusion (units) FFP transfusion (units) Platelet transfusion (units) Acute renal failure requiring RRT Bleeding complication Thrombotic complicationy In-hospital mortality

58 + 9 12 (60.0) 11 (55.0) 2 (10.0) 1 (5.0) 6 (30.0) 10 (50.0) 1 (5.0) 3 (15.0) 3 (15.0) 3 (15.0) 4 (20.0) 12 (60.0) 3 (15.0) 12 (60.0) 60 (40, 65) 7 (35.0) 8 (40.0) 2 (10.0) 1.3 (1.1, 1.8) 9 (7, 11) 7 (4, 14) 1 (0, 2) 2 (0, 4) 5 (25.0) 9 (45.0) 1 (5.0) 6 (30.0)

Abbreviations: ECMO, extracorporeal membrane oxygenation; ECPR, extracorporeal cardiopulmonary resuscitation; FFP, fresh frozen plasma; LVEF, left ventricular ejection fraction; RBC, red blood cell; RRT, renal replacement therapy; SD, standard deviation. * 5 patients taking aspirin only and 3 taking aspirin and clopidogrel. y Thrombotic complication was a clot in the ECMO circuit.

trend in either parameter. Plasma glycocalicin levels in normal healthy volunteers are typically in the range of 2.0 mg/mL, and the GI is in the range of 2.2.15

A decrease was observed in the largest VWF multimer bands on Western blot. Figure 1 shows electrophoresis of VWF multimers in normal pooled plasma, a study patient at the 3 ECMO points, and also a type 2B VWD control. The ECMO patient exhibited loss of the largest VWF multimer bands, but it was not to the same degree as the type 2B VWD control patient. On day 1 or 2, VA ECMO patient multimer bands were 89.7% + 6.1% of control pooled plasma length; on day 3, ECMO patient bands were 90.6% + 5.7% of control pooled plasma length; and on day 5, ECMO patient bands were 91.3% + 5.2% of control pooled plasma length. Figure 2 shows results of the T-TAS experiments. The AUC was decreased severely at all 3 study points in ECMO patients, as a normal AUC in healthy volunteers is approximately 358 + 111.17 On ECMO day 1 or 2, patients had a median AUC of 5.8 (3.5-9.7), which is approximately 2% of the AUC reported in healthy volunteers in prior studies.17 On ECMO day 3 median AUC was 6.3 (5.3-11.1), and on day 5 it was 5.5 (4.1-8.1). Addition of VWF concentrate improved the AUC significantly at all points (p = 0.0001 for day 1 or 2, p = 0.002 for day 3, and p = 0.001 for day 5). There was no difference in the AUC (p = 0.47) or the VWF AUC (p = 0.99) over time. Table 3 shows the AUC and VWF AUC in patients with and without bleeding at any time during ECMO. Patients who had bleeding during ECMO had a minimal increase in AUC with VWF supplementation at all study points, whereas patients who did not have bleeding during ECMO appeared to have more early response to VWF. On ECMO day 3, patients who had bleeding during ECMO had a lower VWF AUC than patients who did not have bleeding during ECMO (p = 0.04). Video 1 shows a T-TAS experiment. The clip shows increased platelet adhesion and aggregation with the addition of VWF concentrate on ECMO day 2 and in the same patient a minimal increase in platelet adhesion and aggregation with VWF concentrate on ECMO day 5. Of note, on day 5 the chip channels do not occlude completely with platelet plugs and platelet adhesion is mainly at the tips of the collagen-coated channels. The patient whose platelets are shown in the clip did not have any bleeding complications during ECMO. Figure 3 shows a still image of the same T-TAS experiment from the video at 8 minutes demonstrating these findings.

Table 2 VWF Ag Levels, VWF Ristocetin Cofactor Activity, Platelet Counts, and Plasma Glycocalicin Levels Blood Parameter

Day 1 or 2

Day 3

Day 5

p Value

VWF Ag U/mL* VWF:RCo U/mLy Platelet count £ 109/L Glycocalicin levels (mg/mL)z Glycocalicin indexx

3.0 (2.9-3.0) 1.8 (1.5-2.4) 108 + 48 2.5 (1.5-4.3) 2.9 (2.8-7.6)

3.0 (3.0-3.0) 1.7 (1.5-2.4) 91 + 43 8 1.7 (1.2-3.6) 3.4 (1.2-7.3)

3.0 (3.0-3.0) 2.0 (1.4-2.3) 84 + 30 2.5 (2.4-3.8) 3.4 (1.7-7.4)

1.0 1.0 0.003 1.0 1.0

NOTE. Values are mean + SD, median (Q1-Q3), or n (%). Abbreviations: AG, antigen; SD, standard deviation; VWF, von Willebrand factor. * Normal range is 0.50-1.60 U/mL; upper limit of assay is 3 U/mL. y Normal range is 0.50-1.50 U/mL. z Levels in healthy volunteers are 2.0 mg/mL + 0.5D.1X X x Levels in healthy volunteers are 2.2 + 0.7D.2X X

ARTICLE IN PRESS M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

5

Discussion

Fig 1. Western blot showing VWF multimers of different sizes. A pooled plasma control is shown along with ECMO patient samples and samples from a type 2B VWD patient. Band lengths at study points were compared against the pooled plasma control band length. ECMO, extracorporeal membrane oxygenation; MWM, molecular weight multimer; VFW, von Willebrand factor.

In a cohort of 20 VA ECMO patients, the authors demonstrated severely impaired in vitro platelet adhesion and aggregation under shear stress using a flow chamber model. Both VWF antigen levels and VWF ristocetin cofactor activity were normal to high, suggesting that quantitative loss of large VWF multimers does not explain fully acquired bleeding in ECMO patients. The authors believe that plasma VWF antigen levels most likely were elevated due to the inflammatory state induced by ECMO. Further, this study data showed that in vitro supplementation with VWF concentrate increases platelet adhesion and aggregation in vitro, but does not restore it to normal levels. Acquired VWD is well described in patients on mechanical circulatory support. When VWF is exposed to high shear stress from centrifugal pumps, it leads to a change in VWF A2 domain, which is thought to facilitate cleavage of large VWF multimers by ADAMTS13.4-10 A number of prior studies have described quantitative loss of large VWF multimers as the primary acquired bleeding disorder in ECMO patients (VWD

Fig 2. Results of flow chamber experiments showing the baseline AUC and VWF AUC in ECMO patients at 3 study points. Addition of VWF concentrate improved the AUC significantly at all points (p = 0.0001 for day 1 or 2, p = 0.002 for day 3, and p = 0.001 for day 5). Neither the baseline AUC (p = 0.47) or VWF AUC changed significantly over the study time period (p = 0.99). AUC, area under the flow-pressure curve; VWF AUC, AUC after von Willebrand factor supplementation.

ARTICLE IN PRESS 6

M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

Table 3 Platelet Count, AUC, and VWF AUC in Patients With Bleeding and Without Bleeding During ECMO Variable

Bleeding

No Bleeding

p Value

PLT count day 1 or 2* AUC day 1 or 2 VWF AUC day 1 or 2 PLT count day 3 AUC day 3 VWF AUC day 3 PLT count day 5 AUC day 5 VWF AUC day 5

99 + 35 6.8 (3.4-10.7) 18.0 (13.2-41.7) 83 + 34 5.5 (3.5-9.7) 17.1 (8.4-19.2) 78 + 31 4.8 (3.7-7.1) 15.3 (9.7-48.7)

115 + 58 5.6 (3.5-8.7) 41.4 (13.3-156.1) 97 + 1 6.5 (6.1-48.3) 26.6 (19.8-111.4) 89 + 37 6.2 (5.1-9.0) 30.9 (13.4-54.1)

0.47 0.71 0.39 0.47 0.27 0.04 0.41 0.22 0.54

NOTE. Data are mean + SD or median (Q1-Q3). Abbreviations: AUC, area under the curve; ECMO, extracorporeal membrane oxygenation; PLT, platelet; SD, standard deviation; VWF, von Willebrand factor. * PLT counts are £ 109/L.

type 2A phenomenon). The current data confirm that ECMO patients have loss of large VWF multimers, but further demonstrate that GP1ba receptor loss of dysfunction appears to contribute to the bleeding diathesis during ECMO. Western blots of VWF multimers in this study showed loss of the largest VWF multimers in ECMO patients. However, ristocetin cofactor activity was normal in all patients during the first 5 days on ECMO. This finding strongly suggests that despite some multimer loss, ECMO patient plasma contains adequate VWF

multimers to induce normal platelet aggregation in the presence of normal platelets. VWF ristocetin cofactor activity is tested using plateletpoor plasma from the patient and exogenous formalin-fixed platelets, which cannot secrete VWF. Ristocetin mediates agglutination between the patient’s intrinsic VWF and exogenous platelet GP1ba.19 If loss of large VWF multimers was the only mechanism contributing to abnormal VWF-GP1ba interaction in ECMO patients, ristocetin cofactor activity would have been abnormal in ECMO patients. The flow chamber data also support a paradigm of GP1ba receptor dysfunction or loss in VA ECMO patients. In the flow chamber experiments, the AUC in ECMO patients was approximately 2% to 4% of that in normal patients. Low platelet counts during ECMO partially may explain these findings, but cannot account fully for the severely low AUC values observed in this study. The AUC at 10 minutes is reported to be 358 + 111 in normal healthy volunteers and the median AUC was 5 to 7 in VA ECMO patients.17 When VWF concentrate was added to patient samples in vitro, the AUC increased significantly at all study points, but the increases were modest, suggesting that GP1ba receptor dysfunction probably also plays a role in abnormal platelet adhesion and aggregation during ECMO. Interestingly, patients with bleeding complications during ECMO had a minimal response to VWF concentrate at all 3 study points, whereas patients without bleeding complications had a more robust early response to VWF concentrate. The authors believe that progressive loss or dysfunction of the

Fig 3. Figure shows a T-TAS experiment. Blood samples are shown from a VA ECMO patient on ECMO day 2 and ECMO day 5. The top frame shows the baseline sample and the bottom frame shows the VWF treated sample. The image is shown at 8 minutes into the assay. On day 2, treatment with VWF leads to platelet plugs in all channels. On ECMO day 5, treatment with VWF concentrate did not lead to occlusion of the capillary channels by platelet plugs. PLT, platelet; RBC, red blood cell; T-TAS, Total Thrombus-formation Analysis System; VA ECMO, venoarterial extracorporeal membrane oxygenation; VWF, von Willebrand factor.

ARTICLE IN PRESS M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

GP1ba receptor in some ECMO patients could explain these findings, as patients with a greater degree of GP1ba receptor loss or dysfunction are less likely to respond to intrinsic or extrinsic VWF and probably have increased bleeding risk. The GP1ba receptor is found in high concentrations on platelet surfaces and is part of the GP1b-V-IX complex.20 Discovery and characterization of the receptor took place initially in the 1960s and was tied to work on Bernard-Soulier syndrome.20 In its complex, GP1b is noncovalently bonded with GPIX in a 1:1 ratio and with GPV in a 2:1 ratio. The GP1ba receptor is a leucine-rich repeat protein that contains several regions, including an anionic sulfated tyrosine sequence, a sialomucin domain, a transmembrane domain, and a cytoplasmic tail.20 The sialomucin domain has N-acetylglucosamine units that have galactose, manose, and sialic acid caps, which are essential for normal receptor function.20 One region of the GP1ba receptor between the transmembrane domain and sialomucin domain is susceptible to cleavage by ADAMTS17.12 This region has been described as “unstable,” and exposure of blood to high shear stress could affect its susceptibility to cleavage.12 Although this study did not evaluate specifically the concentration of GP1ba receptors on platelet surfaces, the authors speculate that augmented GP1ba proteolysis or perhaps reduced receptor production may occur in ECMO patients. Patients in this study had elevated glycocalicin levels compared to what has been reported previously in healthy volunteers, but glycocalicin levels can be increased in a number of disease states, including end-stage renal disease and cirrhosis.15 The data suggest that there is at least some increased GP1ba proteolysis in ECMO patients. On top of this, ECMO patients also may have decreased production of GP1ba on platelet surfaces. In a previous study of cardiopulmonary bypass (CPB) patients, Murase et al. demonstrated that GP1ba expression was decreased after CPB and intraplatelet BAX levels were associated with decreased receptor expression.21 Since ECMO exposes platelets to comparable shear stress as CPB, a similar mechanism could explain reduced expression of GP1ba receptors on platelet surfaces in ECMO patients. The authors hope to test these hypotheses in future studies. One final possibility is that ECMO patients develop a mild type 2B VWD phenomenon rather than type 2A. In type 2B VWD, there is a “gain of function” mutation in VWF that leads to increased affinity between the A1 VWF domain and GP1ba receptor.2224 Paradoxically, this can cause a bleeding phenotype as GP1ba receptors become occupied constitutively with VWF. Type 2B VWD also is associated with thrombocytopenia and shortened platelet life span, which are seen in ECMO patients. The elevated GIs in the authors’ patients suggest that a shortened platelet lifespan occurs during ECMO. Patients with type 2B VWD can have normal VWF antigen levels, which is consistent with the study data.23 The electrophoresis analysis performed in this study did not show the same pattern of intermediate and large multimer loss in ECMO patients that a type 2B VWD control patient had, but perhaps ECMO patients have a type 2B variant with isolated large multimer loss.

7

Alternatively, the pattern of high VWF antigen and high ristocetin cofactor activity observed in this study could represent a completely new qualitative variant of VWD. The median ratio of VWF antigen to ristocetin cofactor activity in the authors’ patients was 1.5. This pattern is not consistent with any of the previously described type 2 VWD variants in the published literature.25 This study had several limitations. First, the cohort was relatively small, containing 20 VA ECMO patients, most of whom had pulmonary embolism. Some patients had underlying hypercoagulable states, which may limit the generalizability of the data. Second, although the data are highly suggestive of GP1ba receptor loss or dysfunction in VA ECMO patients, they do not determine definitely whether patients on ECMO have reduced GP1ba receptor production, increased proteolysis, qualitative receptor dysfunction, or a type 2B VWD phenomenon. Third, almost half of the patients in the cohort received antiplatelet drugs within 5 days of ECMO, and this could have affected platelet adhesion and aggregation in the flow chamber experiments. Both aspirin and clopidogrel can affect AUC as has been reported previously.26 Fourth, the authors did not obtain a pre-ECMO or post-ECMO blood sample for comparison of platelet adhesion and aggregation. Finally, the authors examined platelet adhesion and aggregation in vitro, and the experiments may not represent fully in vivo coagulation and platelet function in ECMO patients. In summary, in a cohort of 20 VA ECMO patients, the authors demonstrated severely reduced platelet adhesion and aggregation under shear stress, which was only partially correctable with VWF concentrate supplementation. The clinical implication of these findings is that VWF supplementation alone is unlikely to adequately treat bleeding in ECMO patients. Treatment with desmopressin or VWF concentrate may have to be combined with platelet transfusion so that both large VWF multimers and functional GP1ba receptors are restored. Future studies are needed to confirm these findings and elucidate the specific pathophysiology of impaired VWFGP1ba interactions in ECMO patients. Supplementary materials Supplementary material associated with this article can be found in the online version at doi:10.1053/j.jvca.2018.11.031. References 1 Extracorporeal Life Support Organization registry report. Available at: https://www.elso.org/Registry/Statistics.aspx. Accessed 10 August 2018. 2 Cheng R, Hachamovitch R, Kittleson M, et al. Complications of extracorporeal membrane oxygenation for treatment of cardiogenic shock and cardiac arrest: A meta-analysis of 1,866 adult patients. Ann Thorac Surg 2014;97:610–6. 3 Mazzeffi M, Greenwood J, Tanaka K, et al. Bleeding, transfusion, and mortality on extracorporeal life support: ECLS working group on thrombosis and hemostasis. Ann Thorac Surg 2016;101:682–9. 4 Cheung PY, Sawicki G, Salas E, et al. The mechanisms of platelet dysfunction during extracorporeal membrane oxygenation. Crit Care Med 2000;28:2584–90.

ARTICLE IN PRESS 8

M. Mazzeffi et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2018) 18

5 Mutlak H, Reyher C, Meybohm P, et al. Multiple electrode aggregometry for the assessment of acquired platelet dysfunction during extracorporeal membrane oxygenation. J Thorac Cardiovasc Surg 2015;63:21–7. 6 Hundalani SG, Nguyen KT, Soundar E, et al. Age-based difference in activation markers of coagulation and fibrinolysis in extracorporeal membrane oxygenation. Pediatr Crit Care Med 2014;15:e198–205. 7 Tauber H, Ott H, Streif W, et al. Extracorporeal membrane oxygenation induces short term loss of high molecular weight von Willebrand factor multimers. Anesth Analg 2015;120:730–6. 8 Huck V, Schneider MF, Gorzelanny C, et al. The various states of von Willebrand factor and their function in physiology and pathophysiology. Thromb Haemost 2014;111:598–609. 9 Chan CH, Pieper IL, Fleming S, et al. The effect of shear stress on size, structure, and function of human von Willebrand factor. Artif Organs 2014;38:741–50. 10 Jilma-Stohlawetz P, Quehenberger P, Schima H, et al. Acquired von Willebrand factor deficiency caused by LVAD is ADAMTS 13 and platelet dependent. Thromb Res 2016;137:196–201. 11 Coller BS, Peerschke EI, Scudder LE, et al. Studies with a murine monoclonal antibody that abolishes ristocetin-induced binding of von Willebrand factor to platelets: Additional evidence in support of GP1b as a platelet receptor for von Willebrand factor. Blood 1983;61:99–110. 12 Andrews RK, Gardiner EE. Metalloproteolytic receptor shedding. . .platelets “acting their age”. Platelets 2016;27:512–8. 13 Lukito P, Wong A, Jing J, et al. Mechanical circulatory support is associated with platelet receptors glycoprotein Ib alpha and glycoprotein VI. J Thromb Hemost 2016;14:2253–60. 14 Raines G, Aumann H, Sykes S, et al. Multimeric analysis of Von Willebrand Factor by molecular sieving electrophoresis in sodium dodecyl sulphate agarose gel. Thromb Res 1990;60:201–12. 15 Beer JH, Buchi L, Steiner B. Glycocalicin: A new assay  the normal plasma levels and its potential usefulness in selected diseases. Blood 1994;83:691–702.

16 Steinberg MH, Kelton JG, Coller BS. Plasma glycocalicin. An aid in the classification of thrombocytopenic disorders. N Engl J Med 1987;317: 1037–42. 17 Arima Y, Kaikita K, Ishii M, et al. Assessment of platelet derived thrombogenicity with the total thrombus-formation analysis system in coronary artery disease patients receiving antiplatelet therapy. J Thromb Haemost 2016;14:850–9. 18 Hosokawa K, Ohnishi T, Sameshima H, et al. Analysing responses to aspirin and clopidogrel by measuring platelet thrombus formation under arterial flow conditions. Thromb Haemost 2013;109:102–11. 19 Redaelli R, Corno AR, Borroni L, et al. Von Willebrand factor ristocetin cofactor (VWF:RCo) assay: Implementation on an automated coagulometer (ACL). J Thromb Haemost 2005;3:2684–8. 20 Solum NO, Clemetson KJ. The discovery and characterization of platelet GP1b. J Thromb Haemost 2005;3:1125–32. 21 Murase M, Nakayama Y, Sessler DI, et al. Changes in platelet Bax levels contribute to impaired platelet response to thrombin after cardiopulmonary bypass: Prospective observational clinical and laboratory investigation. Br J Anaesth 2017;119:1118–26. 22 Miura S, Chester LQ, Cao Z, et al. Interaction of von Willebrand factor domain A1 with platelet glycoprotein Iba-(1-289). J Biol Chem 2000;275: 7539–46. 23 Bryckaert M, Rosa JP, Denis CV, et al. Of von Willebrand factor and platelets. Cell Mol Life Sci 2015;72:307–26. 24 Madabhushi SR, Zhang C, Kelkar A, et al. Platelet Gp1ba binding to von Willebrand factor under fluid shear: Contributions of the D’D3-Domain, A1-Domain flanking peptide and O-linked glycans. J Am Heart Assoc 2014;3:e001420. 25 Ng C, Motto DG, Di Paola J. Diagnostic approach to von Willebrand disease. Blood 2015;125:2029–37. 26 Hosokawa K, Ohnishi T, Fukasawa M, et al. A microchip flow-chamber system for quantitative assessment of the platelet thrombus formation process. Microvasc Res 2012;83:154–61.