Evaluation of Bivalirudin During Adult Extracorporeal Membrane Oxygenation: A Retrospective Characterization of Dosing, Efficacy and Bleeding

Abstract

Objective: Although heparin is the current standard anticoagulant during venoarterial (VA) and venovenous (VV) extracorporeal membrane oxygenation (ECMO), factors including heparin-induced thrombocytopenia, heparin resistance and drug shortages necessitate alternative anticoagulants such as direct thrombin inhibitors. The aim was to characterize dosing, safety, and efficacy of bivalirudin during ECMO support. Methods: This retrospective single-center study included 24 adults on ECMO support who received ≥6 hours of bivalirudin. The primary endpoint was dose to first therapeutic activated partial thromboplastin time (aPTT). Secondary endpoints included evaluating dosing between ECMO modes, incidence of bleeding and thrombotic events, and time in therapeutic range (TTR). Results: The dose at time of first therapeutic aPTT was bivalirudin 0.05 [0.05-0.1] mg/kg/hour. Bivalirudin dosing requirements were lower in VAECMO compared to VV-ECMO patients and were not impacted by continuous venovenous hemofiltration. Time to therapeutic aPTT was 5.5 [2-13] hours for VA-ECMO and 4.5 [2-8.6] hours for VV-ECMO patients. During any mode of ECMO TTR was 58.3% [39.6-73.1]. Thrombotic events occurred in 3 (13%) patients and major bleeding occurred in 12 (50%) patients. Conclusions: Our findings demonstrated variable bivalirudin dosing requirements based on mode of ECMO and dosing modifications may not be required during CVVH. Factors including mode of ECMO, indication for bivalirudin and concomitant antiplatelet therapy may impact hematologic events. Application of this data can assist with developing a bivalirudin ECMO protocol which provides less variability in initial dosing and TTR.

Introduction

Extracorporeal membrane oxygenation (ECMO) is a form of mechanical circulatory support used for refractory cardiogenic or respiratory failure. A critical component of managing patients receiving ECMO support is balancing the risk of bleeding with the risk of device and systemic thrombosis. Continuous contact of circulating blood with the extracorporeal circuit provokes an inflammatory response thereby activating the coagulation cascade. This places patients at risk for thrombotic complications such as pump or oxygenator thrombosis, fibrin stranding within inflow and outflow cannulas, deep vein thrombosis (DVT), and pulmonary embolism (PE). ¹ Patients on venoarterial (VA) ECMO may also experience blood stasis within the left ventricle (LV) due to reduced LV preload, increased LV afterload, and reduced LV function in the setting of cardiogenic shock. ² Unfractionated heparin (UFH) is the current international standard anticoagulant for maintenance of device function due to widespread availability, reversibility, and clinical familiarity. ³ However, many complications are associated with UFH use with ECMO, including heparin-induced thrombocytopenia (HIT) and acquired antithrombin deficiency resulting from consumption of endogenous antithrombin from exposure to the extracorporeal circuit and heparin. The incidence of HIT during ECMO support is not well characterized; one retrospective analysis of 118 patients reports a prevalence of HIT and heparin induced thrombocytopenia with thrombosis (HITT) of 8.3% and 7.3%, respectively. ⁴ Furthermore, HITT in ECMO patients carries a mortality rate of up to 50%. ⁵

At the same time, bleeding is a common complication with multifactorial causes including von Willebrand factor and platelet dysfunction, consumption of coagulation factors, and fibrinolysis. ⁶ Acquired von Willebrand Syndrome may result from the high-shear environment of extracorporeal support as well as alterations in platelet cell receptor function mediated by altered blood flow.^3,6 Disseminated intravascular coagulation (DIC) may result from severity of illness of the underlying disorder, as well as tissue injury and contact activation in the artificial membrane. Clot deposition within the ECMO circuit can lead to excessive fibrinolysis resulting in coagulopathy-type bleeding. ³ Furthermore, systemic anticoagulation is necessary during ECMO support to combat the risk of thrombosis which may further increase the risk for bleeding complications. These potential complications necessitate the investigation of non-heparin alternatives.

Intravenous direct thrombin inhibitor (DTI), bivalirudin, is an alternative parenteral anticoagulant. Bivalirudin exhibits its anticoagulant effect through direct binding of circulating and fibrin-bound thrombin without requiring an antithrombin intermediate. This differentiating mechanism of action could be advantageous in ECMO patients who exhibit heparin resistance due to congenital or acquired antithrombin deficiency. ⁶ As a result, DTIs may lead to a more consistent and predictable anticoagulant effect while removing the concern for antithrombin depletion and HIT.^1,7 Recent literature comparing heparin to bivalirudin anticoagulation therapy on ECMO have demonstrated better time in therapeutic range, decrease rates of hemorrhagic complications and less transfusion requirements with bivalirudin.^8-10 A disadvantage is the lack of targeted reversal agents for DTIs and bleeding complications can be catastrophic.

At our institution, DTI use in ECMO has historically been reserved for patients with suspected or confirmed HIT or HITT or an established heparin resistance. In 2019, an outbreak of African Swine fever in China resulted in a national shortage of porcine-derived heparin products in the United States, impacting UFH availability and utilization. ¹¹ Bivalirudin became the preferred anticoagulant during VA and venovenous (VV) ECMO support resulting in higher utilization during this time. ¹² Due to the limited evidence surrounding DTI-based anticoagulation in adult ECMO, we aimed to characterize the dosage, efficacy, and safety profiles of bivalirudin in this patient population.

Methods

This analysis was a single-center, retrospective chart review of patients admitted between April 1, 2016 and September 30, 2020 to a 1035-bed acute, tertiary academic medical center. Institutional review board approval was obtained prior to initiation of the study. We included adult patients ≥18 years of age who received greater than or equal to 6 hours of an intravenous bivalirudin while on support with VA-ECMO or VV-ECMO. This time frame ensured steady state bivalirudin concentrations would be achieved even in the setting of renal dysfunction. Additionally, it allowed an adequate washout period if patients received heparin prior to DTI initiation. Patients who received ECMO during the study period were identified through our institution’s prospective ECMO patient database. Medication utilization reports for bivalirudin were obtained through our electronic health record (EHR) Epic (Epic Systems Corporation, Verona, Wisconsin) and cross-screened to identify patients ordered for a bivalirudin infusion during ECMO support. Patients were excluded if they had an admission diagnosis of coronavirus disease 2019 due to altered coagulation and if patients were cannulated for ECMO at an outside hospital due to missing data and details of anticoagulant agents at the time of cannulation.

Electronic medical records were manually reviewed to collect baseline demographics, comorbidities, thrombotic risk factors, and baseline laboratory values to calculate Survival After Veno-Arterial ECMO (SAVE) and Respiratory ECMO Survival Prediction (RESP) scores at the time of cannulation (Supplemental Appendix 1 and 2).^13,14 Acute kidney injury (AKI) was defined in our study as urine output less than 0.5 mL/kg/hour for 6 hours and/or increase in serum creatinine ≥1.5 time the patient’s baseline in the past 7 days. Acute liver injury (ALI) was defined as an International Normalized Ratio (INR) ≥1.5 not due to anticoagulation and/or aminotransferases 3 times the upper limit of normal. Thrombotic risk factors collected included the following: active malignancy, post-partum within 90 days, oral contraceptive or hormone replacement therapy use within 90 days, surgery in the prior 2 months, recent trauma within 90 days, spinal cord injury, active smoking, and nephrotic syndrome. Heparin resistance was defined as patients receiving greater than 35 000 units of heparin in 24 hours to achieve a therapeutic aPTT. ¹⁵ The ECMO mode (VA or VV-ECMO) and indication for ECMO were collected. Patients transitioned between modes were categorized as the mode when bivalirudin was initiated. Support at time of and after ECMO cannulation was attained, including the use of a distal perfuser, CentriMag (Abbott, Abbott Park, Illinois), intra-aortic balloon pump, Impella (Abiomed, Danvers, Massachusetts), durable left ventricular assist device, renal replacement therapy. Medications including aspirin, P2Y12 inhibitor, desmopressin, tranexamic acid or aminocaproic acid, and total amount of blood products (packed red blood cells, fresh frozen plasma, cryoprecipitate, platelets) received during ECMO were collected.

The primary endpoint was the weight-based bivalirudin dose to first therapeutic activated partial thromboplastin time (aPTT). Actual body weight at the time of bivalirudin initiation was used for bivalirudin dosing. Weight-based dosing regimens while on ECMO were reported hourly the entire duration of bivalirudin therapy. Dosing and titrations were made according to our institutional standard of practice unless noted (Supplemental Appendix 3). Goal activated partial thromboplastin time (aPTT) range and each aPTT value while on bivalirudin were collected. Secondary endpoint time in therapeutic range (TTR) was calculated by dividing the number of hours in goal aPTT range by the total duration on bivalirudin and expressed as a percentage.

To evaluate safety and efficacy, bleeding and thrombotic events were collected. In congruence to International Society on Thrombosis and Haemostasis (ISTH) criteria, major bleeding was defined as fatal bleeding, symptomatic bleeding in a critical area or organ and/or bleeding resulting in a fall in hemoglobin level of 2 g/dL or leading to transfusion of 2 or more units of whole blood or red cells. Clinically relevant minor bleeding was defined as an acute or subacute overt bleed not meeting major bleed criteria but prompting clinical response including physician guided medical or surgical treatment, change in antithrombotic therapy and/or interruption or discontinuation of anticoagulant drug. ¹⁶ At our institution, general transfusion practices are comparable to the Extracorporeal Life Support Organization Anticoagulation Guidelines (Supplemental Appendix 3). ¹⁷ Thrombotic outcomes included DVT or PE confirmed by radiographic or ultrasound imaging and device thromboses documented as thrombosis in the oxygenator, pump, or circuit, or renal replacement system.

Baseline characteristics were reported using descriptive statistics as median (interquartile range [IQR]) for continuous variables and number (%) for categorical variables. Statistical Package for Social Sciences (Version 25) was used to perform statistical analysis.

Results

A total of 382 adult patients received ECMO support during the study period, of which, 42 patients were simultaneously ordered a DTI and screened for inclusion criteria. Twenty-four patients receiving bivalirudin were included in the final analysis (Figure 1). Of the twenty-four patients, 17 (71%) received VA-ECMO and 7 (29%) VV-ECMO. Additional baseline demographics are detailed in Table 1. The median SAVE score was −9 [−12 to −5] and RESP score was 0 [−2 to 1]. Median duration on ECMO support was 9 [6-13] days and time on bivalirudin while on ECMO was 4 [1-11] days. Seventeen (71%) patients had acute kidney injury and 11 (46%) patients received continuous venovenous hemofiltration (CVVH) while on ECMO support. Fourteen (58%) patients, 11 VA-ECMO and 3 VV-ECMO, were transitioned from a heparin infusion to bivalirudin. All patients transitioned from heparin to bivalirudin had a minimum 2 hours off heparin to the start of bivalirudin infusion. The primary indication for bivalirudin use was HIT/HITT in 17 (71%) patients; additional indications for use and anticoagulation administration at time of ECMO cannulation are detailed in Table 2. Duration of bivalirudin therapy was a median 4 [1-11] days. The median hospital length of stay was 21 [12-46] days with an incidence of both in-hospital and 30-day mortality of 79% (Table 3).

Figure 1.

Inclusion and exclusion.

Note. DTI = direct thrombin inhibitor; ECMO = extracorporeal membrane oxygenation.

Table 1.

Baseline Characteristics. ^a

	All patients N = 24	VA-ECMO N = 17	VV-ECMO N = 7
Age, years	59 [44-68]	65 [58-68]	43 [23-47]
Male	19 (79)	14 (82)	5 (71)
Weight, kg	87 [73-98]	93 [71-100]	81 [71-95]
BMI, kg/m2	27 [24-32]	30 [25-35]	26 [20-28]
Indication for ECMO
Cardiogenic shock	14 (58)	14 (82)	0 (0)
Respiratory failure	7 (29)	0 (0)	7 (100)
Failure to wean from CPB	3 (13)	3 (18)	0 (0)
ECHO characteristics before ECMO
LVEF Percentage	59 [29-75]	58 [22-74]	60 [51-78]
PFO	2 (8)	1 (6)	1 (14)
History of ASCVD equivalent
Coronary artery disease	10 (42)	10 (59)	0 (0)
Peripheral artery disease	2 (8)	2 (12)	0 (0)
Cerebrovascular accident	1 (4)	1 (6)	0 (0)
Co-morbid conditions at baseline
Hypertension	13 (54)	13 (76)	0 (0)
Hyperlipidemia	10 (42)	9 (53)	1 (14)
Diabetes mellitus	9 (38)	7 (41)	2 (28)
Chronic lung disease	3 (13)	0 (0)	3 (43)
Congestive heart failure	2 (8)	2 (12)	0 (0)
Rheumatic disease	2 (8)	2 (12)	0 (0)
Peripheral vascular disease	1 (4)	1 (6)	0 (0)
History of VTE	3 (13)	3 (18)	0 (0)
DVT	2 (8)	2 (12)	0 (0)
PE	1 (4)	1 (6)	0 (0)
Hypercoagulable disorder
Antithrombin deficiency	3 (13)	2 (12)	1 (14)
Thrombotic risk factors on admission	11 (46)	6 (35)	5 (71)
Active malignancy	4 (17)	2 (12)	2 (29)
Active smoking	4 (17)	2 (12)	2 (29)
Surgery within prior 2 months	3 (13)	3 (18)	0 (0)
OCP/HRT use within 90 days	1 (4)	0 (0)	1 (14)
Pre-ECMO Cardiac Arrest	9 (38)	9 (53)	0 (0)
AKI while on ECMO	17 (71)	14 (82)	3 (43)
ALI while on ECMO	17 (71)	12 (71)	5 (71)
CVVH while on ECMO	11 (46)	9 (53)	2 (29)
P2Y12 inhibitor use during ECMO b	4 (17)	4 (24)	0 (0)
Cangrelor	3 (13)	3 (18)	0 (0)
Clopidogrel	1 (4)	1 (6)	0 (0)
Aspirin use during ECMO	10 (42)	10 (59)	0 (0)
≤100 mg	8 (33)	8 (47)	0 (0)

Note. AKI = acute kidney injury; ALI = acute liver injury; ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; CPB = cardiopulmonary bypass; DVT = deep vein thrombosis; ECHO = echocardiogram; ECMO = extracorporeal membrane oxygenation; HRT = hormone replacement therapy; LVEF = left ventricular ejection fraction; OCP = oral contraceptive; PE = pulmonary embolism; PFO = patent foramen ovale; VA = veno-arterial; VTE = venous thromboembolism; VV = veno-venous.

^aData presented as median [interquartile range] or n (%) unless otherwise specified.

^bAll patients receiving a P2Y12 was concomitantly on aspirin.

Table 2.

Characterization of Anticoagulation and Direct Thrombin Inhibitor Use. ^a

	All Patients N = 24	VA-ECMO N = 17	VV-ECMO N = 7
Indication for DTI
HIT or HIT(T)	17 (71)	12 (71)	5 (71)
Heparin resistance	2 (8)	2 (12)	0 (0)
Heparin Shortage	7 (29)	3 (18)	2 (29)
Heparin bolus at cannulation, IU/kg b	88 [44-100]	80 [25-100]	93 [65-110]
Time to anticoagulant infusion initiation from cannulation bolus, hours	7 [0-11]	5 [0-11.5]	7.2 [3.8-9.8]
Time on DTI, days	4 [1-11]	4 [2-18]	1 [1-10]
PTT goal range
55-75 seconds	18 (75)	12 (71)	6 (86)
Other c	6 (25)	5 (29)	1 (14)
DTI percent time in therapeutic range	58.3 [39.6-73.1]	60 [43-72.4]	50 [38-80]
DTI percent time in subtherapeutic range	14.2 [4-36.8]	21.7 [4.2-26.5]	5.6 [0-50]
DTI percent time in supratherapeutic range	13.6 [0.6-31.2]	16.7 [3-29.2]	9.2 [0-55.6]
Time to therapeutic PTT, hours	5.5 [2-12.5]	5.5 [2-13]	4.5 [2-8.6]
Dose to first therapeutic PTT
Bivalirudin (mg/kg/hour)	0.05 [0.05-0.1]	0.05 [0.05-0.07]	0.12 [0.05-0.16]

Note. DTI = direct thrombin inhibitor; ECMO = extracorporeal membrane oxygenation; HIT = heparin-induced thrombocytopenia; HITT = heparin-induced thrombocytopenia with thrombosis; PTT = partial thromboplastin time; VA = veno-arterial; VV = veno-venous.

^aData presented as median [interquartile range] or n (%) unless otherwise specified.

^bAll patients received heparin bolus at ECMO cannulation.

^cPatient-specific aPTT goal per hematology consult service.

Table 3.

Outcomes. ^a

	All patients N = 24	VA-ECMO N = 17	VV-ECMO N = 7
Median time on ECMO support, days	9 [6-13]	8 [5-12]	12 [8-42]
Any thrombotic event	3 (13)	3 (18)	0 (0)
Circuit thrombosis	1 (4)	1 (22)	0 (0)
Ischemic stroke	2 (8)	2 (12)	0 (0)
Number of ECMO circuit changes	1	1	0
Major bleeding event	12 (50)	11 (65)	1 (14)
Decrease Hb ≥2 g/dL and/or ≥2 units PRBC	8 (33)	7 (35)	1 (14)
GI hemorrhage	2 (8)	1 (6)	1 (14)
ICH	2 (8)	2 (12)	0 (0)
Pulmonary hemorrhage	1 (4)	1 (6)	0 (0)
Blood products received on ECMO
PRBC, units	12 [6-20]	12 [6-25]	11 [4-14]
FFP	2 [1-7]	2 [1-8]	0 [0-3]
PLT	3 [0-11]	4 [1-18]	0 [0-7]
Cryoprecipitate	0 [0-1]	0 [0-1]	0 [0-0]
Hemostatic Medications
Desmopressin	5 (21)	5 (29)	0 (0)
Aminocaproic acid	8 (33)	6 (35)	2 (29)
Topical	4 (17)	2 (12)	2 (29)
Systemic	5 (21)	4 (24)	1 (14)
TXA	5 (21)	2 (12)	2 (29)
Nebulize	2 (8)	0 (0)	2 (29)
Systemic	2 (8)	2 (12)	0 (0)
Hospital length of stay, days	21 [12-46]	23 [12-49]	19 [13-44]
Inpatient mortality	19 (79)	13 (76)	6 (86)
30-day mortality	19 (79)	13 (76)	6 (86)

Note. DTI = direct thrombin inhibitor; DVT = deep vein thrombosis; ECMO = extracorporeal membrane oxygenation; FFP = fresh frozen plasma; GI = gastrointestinal; ICH = intracerebral hemorrhage; PRBC = packed red blood cell; PLT = platelet; PE = pulmonary embolism; RRT = renal replacement therapy; TXA = tranexamic acid; VA = veno-arterial; VTE = venous thromboembolism; VV = veno-venous.

^aData presented as median [interquartile range] or n (%) unless otherwise specified.

To characterize dosing requirements, the median bivalirudin dose at time of first therapeutic aPTT was 0.05 [0.05-0.1] mg/kg/hour. Patients on VA-ECMO required a median bivalirudin dose of 0.05 [0.05-0.07] mg/kg/hour, in comparison patients on VV-ECMO required a median dose of 0.12 [0.05-0.16] mg/kg/hour. All patients with a bivalirudin starting dose of 0.025 mg/kg/hour or less were subtherapeutic on first aPTT check. All patients on concomitant CVVH demonstrated a median dose to therapeutic aPTT of 0.05 [0.04-0.09] mg/kg/hour.

Standard institutional aPTT goals were targeted in 19 (79%) patients on bivalirudin. Five (21%) patients had a custom aPTT goal range, ranging from 40 to 80 seconds, due to bleeding and thrombotic risks determined by the hematology consult service. To assess controllability on bivalirudin, the median TTR was 58.3% [39.6-73.1], 60% [43-72.4] for VA-ECMO and 50% [38-80] for VV-ECMO; median time in subtherapeutic and supratherapeutic range can be found in Table 2. The time to reach a therapeutic aPTT was 5.5 [2-12.5] and 4.5 [2-8.6] hours for patients on VA-ECMO and VV-ECMO respectively (Table 2).

Thrombotic events on bivalirudin during ECMO occurred in 3 (13%) patients, 2 (8%) ischemic stroke events and 1 (4%) was due to circuit thrombosis requiring a circuit change. All thrombotic events were in patients receiving VA-ECMO and were titrated to standard institutional aPTT goals. All patients with a thrombotic event had an aPTT within therapeutic range at the time of event and time in a subtherapeutic aPTT range was 21.7% [5.3-40].

A major bleeding event occurred in 12 (50%) patients. Major bleeding events were primarily due to a drop in hemoglobin by 2 g/dL and/or requiring transfusion of 2 or more pack red blood cells (PRBC) followed by gastrointestinal bleeding, intracerebral hemorrhage and pulmonary hemorrhage (Table 3). The incidence of major bleeding events was higher in VA-ECMO at 65% compared to 14% in the VV-ECMO patients. Nine (75%) of the patients who bled were titrated to the standard institutional aPTT goals and 3 (25%) patients titrated to an aPTT goal range of 60 to 80 seconds. The median time patients who experience a major bleeding event had a supratherapeutic aPTT range was 11.7% [1-21.3]. Seven (58%) patients with a major bleeding event were on concomitant antiplatelet; 3 patients were on aspirin and 4 patients were on aspirin and either cangrelor or clopidogrel. Details on blood products received during ECMO and adjunct hemostatic medications are included in Table 3.

Hospital mortality occurred in 19 (79%) patients. Hematological complications contributed to 4 (21%) patients’ death; 2 patients with intracerebral hemorrhages and 2 patients with fatal embolic/ischemic strokes. The remainder 15 patients cause of death was due to deterioration of the patients underlying cardiac or respiratory disease state.

Discussion

Anticoagulation during ECMO support remains necessary to prevent circuit thrombosis and thromboembolic events. Use of non-heparin anticoagulants are essential when intravenous heparin is ineffective, contraindicated, or not available. As data comparing DTIs to heparin anticoagulation in ECMO grows there is an ongoing need to further characterize DTI dosing, safety, and efficacy in adult patients on ECMO.

Few studies have evaluated dosing characteristics of bivalirudin during ECMO support, this study being the largest cohort composed of HIT/HITT patients to date. Currently, bivalirudin dosing requirements in the literature are heterogenous; bolus doses ranging from 0.04 to 2.5 mg/kg maintenance infusion rates ranging between 0.025 and 2.5 mg/kg/hour with increased dosing requirements during continuous renal replacement therapy (CRRT). ¹ In our study, a median bivalirudin dose to achieve goal aPTT was 0.05 mg/kg/hour which is lower than doses previously reported for ECMO patient populations. ¹ A retrospective analysis of 14 adult patients receiving ECMO, 86% VV-ECMO, by Walker et al found the median initial bivalirudin dose to achieve target aPTT was 0.15 mg/kg/h [range 0.04-0.26 mg/kg/hour]. Dosing requirements to maintain goal aPTT increased when CRRT was used concomitantly, with a median dose of 0.21 mg/kg/hour. Average time spent within target aPTT range was 76%, however, 36% of patients required circuit change due to clotting or failing oxygenation while 28.6% of patients experienced significant bleeding requiring reduction in aPTT goals or a temporary hold of bivalirudin. ¹⁸ Converse to the findings by Walker et al , the 10 patients in our cohort receiving ECMO and CVVH required the same median bivalirudin dose to therapeutic aPTT as ECMO patients without CVVH. Since most of our patient population had a diagnosis of AKI it is possible the patients not receiving CVVH could have had similar bivalirudin clearance as CVVH. Notably, we observed VV-ECMO patients required a higher dose of bivalirudin compared to VA-ECMO. A study conducted in patients receiving bivalirudin for heparin induced thrombocytopenia on ECMO, 50% VA-ECMO, the 7 patients receiving concomitant CRRT had an initial therapeutic dose of 0.05 mg/kg/hour [0.05-0.065 mg/kg/hour]. ¹⁹ When comparing our findings to studies conducted by Berei et al ²⁰ and Ranucci et al, ⁹ comprised of 91% and 100% VA-ECMO patients respectively, similar dosing ranges of 0.03 to 0.05 mg/kg/hour were observed. Overall, our cohort found similar dosing requirements to previous literature when reviewing VA-ECMO and VV-ECMO independently but found no dosing difference when patients were receiving CRRT.

Time in therapeutic range is one quality measure to compare anticoagulation dose management between agents and across studies. ²¹ When comparing bivalirudin to heparin, studies in adult patients receiving ECMO support have found a statistically significant improvement in aPTT percent TTR, 85.7%versus 50% and 86%versus 33% , respectively.^10,22 Despite improved TTR this did not translate into differences in thrombotic or hemorrhagic clinical outcomes.^10,22 In the largest cohort of ECMO patients receiving a DTI, Rivosecchi et al demonstrated a significant difference in TTR between UFH and bivalirudin in TTR (37%vs 56%, P = .01) and difference in bleeding events (40.7%vs 11.7%, P < .001). ⁸ This study was limited to VV-ECMO patients, utilizing bivalirudin as their primary agent for anticoagulation. Despite a median percent time in supratherapeutic aPTT of 13.6% and TTR at 58.3%, we observed a higher percentage of major bleeding complications and PRBC transfusions. This is likely attributed to our population primarily including VA-ECMO patients as well as a high proportion receiving antiplatelet therapy. A retrospective study by Hanna et al evaluated 12 patients on ECMO with HIT who received bivalirudin and reported a 66.7% incidence of major bleeding events. This population was primarily VA-ECMO and all patients who received antiplatelet therapy had a major bleeding event. ¹⁹ Similar to this study, 7 patients on antiplatelet therapy, and all 4 patients on dual antiplatelet therapy with aspirin and clopidogrel or cangrelor, experienced a major bleeding event. When assessing aPTT at time of the bleeding event, none of the patients had a supratherapeutic aPTT. Since the aPTT is performed in the absence of platelets, this laboratory value does not assess platelet activity or impact on clot development. These findings suggest percentage TTR is not the only factor to account for when assessing causality of clinical outcomes across single center studies and the development of hematological complications is often multifactorial in this patient population.

When evaluating our findings including the lower percent TTR compared to what has been previously reported, difference in dosing requirements based on mode of ECMO and time to therapeutic aPTT pose an opportunity to evaluate our bivalirudin dosing nomogram. Currently our bivalirudin dosing nomograms do not specify starting doses between modes of ECMO and dose titrations are a discrete number value based on the selected nomogram starting dose (Supplemental Appendix 3). As other institutions titrate bivalirudin based on a percentage of the current dosing rate, this could be an area to improve our TTR.^10,23 Using a discrete number eliminates the need to calculate each dose titration, reducing the risk of mathematical errors, but this method requires selecting the appropriate starting dose to ensure the correct dosing titrations. Selecting the incorrect initial dose could thereby impact both time to therapeutic aPTT and TTR. From our observations we found all patients started on a bivalirudin dose of 0.025 mg/kg/hour were subtherapeutic upon first aPTT check. Notably, 2 patients in acute liver failure did require maintenance infusion less than 0.025 mg/kg/hr to maintain a therapeutic aPTT and were concomitantly receiving CVVH with concern for disseminated intravascular coagulation documented on chart review. Therefore, based on the median dose to therapeutic aPTT, the standard starting rates should be 0.05 mg/kg/hour for VA-ECMO and 0.12 mg/kg/hour for VV-ECMO with no dosing adjustments required for CVVH. Only in the setting of high bleeding risk including receiving dual antiplatelet therapy, concomitant hepatic and renal failure could a starting dose of 0.025 mg/kg/hour be considered.

Despite majority of our study population receiving VA-ECMO, in which circuit clots are more common, and the primary indication for bivalirudin use was HIT/HITT, our study had a low rate of thrombotic complications. ³ Thrombotic rates reported in this patient population have ranged from 0% to 22.7% during bivalirudin therapy.^8,9,18-20 Thrombosis onset can occur with prolonged time in subtherapeutic aPTT range, one retrospective study found a significantly lower aPTT (47 ± 12.3vs 53.6 ± 12.5 seconds, P = .037) in ECMO patients diagnosed with venous thromboembolism. ²⁴ None of our patients who experienced a thrombotic event had subtherapeutic aPTTs at time of thrombus diagnosis. Given the retrospective design it is possible the onset of thrombosis was not accurately captured; therefore, the aPTT collected at the time of diagnosis may not be reflecting the actual thrombotic event. Additionally, we reported 29% of patients received systemic antifibrinolytic therapy with aminocaproic acid or tranexamic acid, in which the data on thrombosis risk with these agents in ECMO is limited.

Across various studies evaluating bivalirudin in ECMO, there is large variability in the incidence of major hemorrhagic complications reported. These differences could be contributed to the design of previous studies primarily being single-center retrospective, the indication for receiving a DTI, and differences in how major hemorrhagic complications are defined. Additionally, risk factors for developing a hemorrhagic complication during ECMO include patients receiving VA-ECMO, duration of ECMO support, strategies for monitoring the degree of anticoagulation, and anticoagulation target ranges. Although our major hemorrhagic complication rate on bivalirudin is high at 50%, this is comparable to rates seen in the VA-ECMO patient population. Hanna et al found a 66.7% rate of major hemorrhagic complications in ECMO, 50% VA-ECMO, patients receiving bivalirudin for HIT. ¹⁹ Additionally, the study by Berei et al comprised of 91% VA-ECMO found 45.5% of patients experienced a major hemorrhagic complication on bivalirudin. ²⁰ When compared to the study conducted by Rivosecchi et al in solely VV-ECMO patients, the bivalirudin group had a 11.7% major hemorrhagic complication rate.

Our study is limited by its retrospective design in which bleeding and thrombotic outcomes were manually extracted from the EHR. Given events were manually extracted, there were difficulties in capturing the exact time of bleeding and thrombotic events and may limit the utility of associating the aPTT to time of event. The single-center small sample size of 24 patients, predominately male, may limit the generalizability of our dosing characterization to a wider patient population. This study was limited to bivalirudin anticoagulation during ECMO and was not compared to our current standard anticoagulation therapy with heparin. The only anticoagulation monitoring parameter used to adjust and assess anticoagulation therapy was aPTT; other methods including activated clotting time, dilute thrombin time, rotational thromboelastometry or thromboelastography might have provided additional perspective regarding our TTR, thrombotic and bleeding events. Patients transitioned from heparin to bivalirudin had a minimum 4 hours off heparin until the first aPTT check on bivalirudin was performed, it is possible residual heparin effects were observed. This residual effect could impact both the dose to first therapeutic aPTT and time to therapeutic aPTT. To ensure minimal impact on dose to first therapeutic aPTT, it was confirmed no bivalirudin dose changes occurred after the second aPTT check in patients with an initial therapeutic aPTT; this allowed a minimum 6-hour heparin wash-out period.

Conclusion

Current literature describing DTI therapy, including dosing requirements, in the ECMO patient population is limited. Our study, with a predominately HIT/HITT indication for use, found bivalirudin dosing requirements could be variable based on mode of ECMO, dosing modifications for CVVH may not be required and major bleeding could be higher in patients receiving concomitant antiplatelet therapy with bivalirudin on ECMO. These variabilities highlight the importance of institutionally evaluating ECMO anticoagulation practices. Application of this data can assist with developing a bivalirudin ECMO protocol which provides less variability in initial dosing and TTR. The overall high rate of major bleeding complications in our study warrants further larger multi-center studies to evaluate dosing, bleeding, and thrombotic complications with DTIs in this high-risk patient population.