Abstract
Bleeding is a common complication following left ventricular assist device (LVAD) implantation. The goal of this study was to investigate the incidence, predictors and clinical outcome of early bleeding events in patients after LVAD implantation.
A total of 83 patients (age 50 ± 13 years, 76% men) had an LVAD implanted [77% HeartMate II, 19% HeartMate 3 (Abbott, Chicago, IL, USA)] over a period of 11 years. Patients were included consecutively. An early bleeding event was defined as the need for thoracic surgical re-exploration or transfusion with >4 units of packed red blood cells before discharge.
Overall, 39 (47%) patients (age 50 ± 14 years, 77% men) experienced an early bleeding event [median time 6 days (interquartile range 1–9 days)]. Furthermore, 10 (26%) of these patients had ≥2 bleeding events. Twelve of the 14 (92%) patients with venoarterial extracorporeal membrane oxygenation (ECMO) support before LVAD implantation experienced an early bleeding event versus 27 of the 69 (39%) patients without ECMO support (P < 0.001). No difference was found in early bleeding rates between HeartMate II and HeartMate 3. Predictors for early bleeding events were lower pre- and postimplant platelet counts and ECMO support preimplantation. After multivariable adjustment, early bleeding events were associated with ECMO support preimplantation (odds ratio 6.3, 95% confidence interval 1.2–32.4; P = 0.03) and thrombocytopenia (<150 × 109/l) postimplant (odds ratio 5.9, 95% confidence interval 1.9–18.7; P = 0.002). Patients who experienced an early bleeding event had a significantly worse 90-day survival rate compared to patients who did not (79% vs 96%, P = 0.03).
An early bleeding event needing surgical exploration is highly prevalent after LVAD implantation, especially in patients bridged with ECMO and with pre- and postimplant thrombocytopenia.
INTRODUCTION
Left ventricular assist devices (LVADs) have evolved to an accepted treatment option for patients with end-stage heart failure, either as a bridge to transplant or as destination therapy for patients ineligible for a heart transplant [1, 2]. The use of LVADs has grown exponentially over the last decade, with currently more than 2000 LVADs implanted yearly in the USA [1].
Although an improvement in survival and a reduction in adverse events after LVAD implantation have been reported over time, the adverse event rate remains high [1]. The Interagency Registry for Mechanically Assisted Circulatory Support reports that bleeding is the most common adverse event followed by infection and cardiac arrhythmia [1]. In addition, the recent European Registry for Patients with Mechanical Circulatory Support registry reports that a major bleeding event within the first 3 months is the most frequently observed adverse event after LVAD implantation [6.45 (95% confidence interval 5.62–7.36) events per 100 patient-months] followed by major infection [3]. The main focus, however, is on late gastrointestinal and neurological bleeding events. Consequently, the literature regarding early bleeding events requiring surgical exploration is limited.
Early bleeding may delay recovery and extend the period of hospitalization and can be life-threatening when it leads to haemodynamic compromise. In addition, the abundant use of transfusions may lead to sensitization, right heart failure and a longer waiting time for a heart transplant [4–6]. Therefore, it is important to identify risk factors for bleeding events and predictors to guide protocols for their prevention and diagnosis.
This study was designed to investigate the incidence, predictors and clinical outcome of early bleeding events requiring thoracic surgical re-exploration or transfusion in the postoperative period after LVAD implantation.
METHODS
Study design
We conducted a retrospective cohort study evaluating all HeartMate II (HMII) and HeartMate 3 (HM3) continuous-flow LVAD (Abbott, Chicago, IL, USA) implanted between December 2006 and February 2017 in the Thoraxcenter, Erasmus MC, University Medical Center Rotterdam, Netherlands, a tertiary referral centre. This study was approved by the institutional review board of the Erasmus MC (MEC-2017-1013).
The primary outcome was the occurrence of an early bleeding event, defined as the need for surgical thoracic re-exploration due to bleeding (loss or accumulation) or transfusion with >4 units of packed red blood cells (PRBCs) before discharge. The indication for re-exploration was assessed by a heart team that consisted of a cardiothoracic surgeon or a cardiologist and/or intensivist and was based on one of the following criteria: the need for transfusion(s) despite well-regulated anticoagulation parameters; haemodynamic instability and the need for inotropes or vasopressors due to tamponade/pericardial effusion (diagnosed by ultrasonography); reduced LVAD flow despite optimal filling and fluid status; reduced mixed venous oxygen blood saturation (SvO2) and increased lactate, haemothorax (diagnosed on radiograph or ultrasonography) or persistent excessive thoracic tube production. The secondary outcome was all-cause mortality. Patients were classified into 2 groups based on the occurrence of an early bleeding event.
Data collection and protocol
All data were obtained from electronic patient records. Laboratory values were collected directly preoperatively and postoperatively for all patients and used separately in the analysis. Our target activated partial thromboplastin time (aPTT) in our extracorporeal membrane oxygenation (ECMO) protocols was 60–70 s with a blood flow >3000 ml, 70–80 s with a blood flow >2000 and <3000 ml and 90–100 s with a blood flow <2000 ml. Postoperative transfusion triggers for the transfusion of PRBC were haemoglobin levels <5 mmol/l. Low platelet counts were managed through substitution of platelets, which is indicated in our centre in anticoagulated patients when the patient has a platelet count lower than the target platelet count (>50 000/l in non-bleeders and >100 000/l in patients who have an active bleeding).
In addition, information regarding anticoagulation medication was included if one was used within 7 days prior to LVAD implantation. Depending on the condition and indications [ECMO (n = 14), intra-aortic balloon pump (n = 21), high thrombotic risk (n = 1)], patients were given therapeutic unfractionated heparin (UH) prior to LVAD implantation with a target aPTT of 60–80 s or higher, depending on the ECMO flow. The UH was discontinued 6 h preoperatively. Postimplant, UH was started on Day 1 according to our institutional protocol as follows: Days 1–2, if thoracic drain production is <50 ml/h, start UH 600 IU/h, target aPTT 35–45 s; Days 3–4, target aPTT 40–50 s; Days 5–6, target aPTT 50–65 s; Day 6: add aspirin 80 mg daily; Day 7: start Coumadin; the target international normalized ratio (INR) is 2.0–3.0. Stop UH when the target INR is reached.
Patients received venoarterial ECMO by bifemoral cannulation (using 21–29 Fr multistage venous and 17–21 Fr arterial cannula) via percutaneous (Seldinger) or surgical access. A permanent life support or Cardiohelp set (Maquet Cardiopulmonary, Rastatt, Germany) was used to reach a blood flow of 3.5–5.0 l/min. The goal of the treatment was stabilization of the haemodynamics (SvO2 >60%, mean arterial pressure >60 mmHg, low lactate level and diminishing need for vasopressors) with regular aortic valve opening.
Statistical analysis
Continuous parameters were expressed as the median and interquartile range or as the mean and standard deviation and compared with the Student’s t-test results unless the data were not normally distributed (Kolmogorov–Smirnov test); in these instances, the Mann–Whitney U-test was used. Categorical parameters were expressed as number of patients and percentage and compared using Pearson’s χ2 test, or if a group had fewer than 5 members, by the Fisher’s exact test. Kaplan–Meier curves stratified by group were constructed to evaluate the incidence of early bleeding and the number of deaths in the first 90 days and 1 year postimplant. Patients were censored at the time of a heart transplant. Differences in 90-day and 1st-year survival rates pooled over strata were compared by the log-rank test with the Breslow rule for handling ties. In addition, the Kaplan–Meier curve was also used to evaluate the time to discharge, with patients censored at time of death or heart transplant. Multivariable logistic regression analysis was performed for identification of parameters associated with an early bleeding event after continuous-flow LVAD implant. Variables with a P-value <0.10 in the univariable analysis were included in the multivariable regression analysis model. Finally, through a stepwise variable selection, excluding variables with P > 0.05, the final model was constructed. Two-tailed P-values <0.05 were considered statistically significant. Receiver operating characteristic curves were generated to assess the ability of independent associated variables and the final model to predict the primary outcome. Analyses were performed using statistical software SPSS, version 20.0 for Mac (SPSS Inc., Chicago, IL, USA).
RESULTS
Overall, 83 patients were implanted with a continuous-flow LVAD during the study period. Baseline characteristics of the patient population are reported in Table 1. The mean age was 50.7 ± 12.8 years, 76% were men and 54% had ischaemic cardiomyopathy as the primary cardiac diagnosis. In the majority of the patients, the continuous-flow LVAD was implanted as a bridge to transplant (89%), and most received an HMII (77%). There were no significant differences in demographic characteristics, indications for an LVAD, device type or Interagency Registry for Mechanically Assisted Circulatory Support class between the 2 groups based on the occurrence of an early bleed. However, patients who experienced an early bleeding event had significantly lower rates of implantable cardioverter defibrillators (P = 0.03) and low platelet counts (P = 0.006) at baseline. In addition, patients who experienced an early bleeding event were more often supported by ECMO preimplantation (29% vs 13, P = 0.002).
Baseline characteristics of the study population
| All (n = 83) | No early bleeding event (n = 38) | Early bleeding event (n = 45) | |
|---|---|---|---|
| Age (years) | 50.7 ± 12.8 | 50.6 ± 11.8 | 50.8 ± 13.6 |
| Male gender | 63 (76) | 30 (79) | 33 (73) |
| Body mass index (kg/m2) | 23.4 ± 3.9 | 24.0 ± 4.3 | 22.9 ± 3.4 |
| Aetiology | |||
| Non-ischaemic CMP | 45 (46) | 18 (47) | 27 (60) |
| Ischaemic CMP | 38 (54) | 20 (53) | 18 (40) |
| Comorbidities | |||
| Diabetes mellitus | 9 (11) | 3 (8) | 6 (13) |
| Hypertension | 13 (16) | 8 (21) | 5 (11) |
| ICD/PM | 55 (66) | 30 (79) | 25 (56)a |
| Myocardial infarction | 37 (45) | 21 (55) | 16 (37) |
| CABG | 6 (7) | 1 (3) | 5 (11) |
| CVA | 5 (6) | 4 (11) | 1 (2) |
| IABP | 31 (37) | 15 (40) | 16 (36) |
| ECMO | 14 (17) | 1 (3) | 13 (29)a |
| Days on ECMO | 6 (3–12) | 8 | 4 (3–13) |
| Bridge to transplantation | 74 (89) | 36 (95) | 38 (84) |
| Device type | |||
| HeartMate II | 64 (77) | 29 (76) | 35 (78) |
| HeartMate 3 | 19 (23) | 9 (24) | 10 (22) |
| INTERMACS | |||
| Class 1 | 16 (19) | 5 (13) | 11 (24) |
| Class 2 | 30 (36) | 13 (34) | 17 (38) |
| Class 3 | 17 (21) | 9 (24) | 8 (18) |
| Class ≥4 | 20 (24) | 11 (29) | 9 (20) |
| Medication | |||
| Vitamin K antagonist | 50 (60) | 27 (71) | 23 (51) |
| Heparin | 36 (43) | 14 (37) | 22 (49) |
| Baseline laboratory data | |||
| Platelet count (109/l) | 216 ± 76 | 240 ± 71 | 196 ± 76a |
| Creatinine (µmol/l) | 145 ± 70 | 155 ± 64 | 136 ± 74 |
| Haemoglobin (mmol/l) | 7.4 ± 1.3 | 7.4 ± 1.1 | 7.4 ± 1.4 |
| CRP (mg/l) | 43 ± 59 | 39 ± 54 | 45 ± 63 |
| INR | 1.8 ± 0.8 | 1.9 ± 0.9 | 1.8 ± 0.7 |
| aPTT (s) | 59 ± 28 | 57 ± 20 | 62 ± 33 |
| All (n = 83) | No early bleeding event (n = 38) | Early bleeding event (n = 45) | |
|---|---|---|---|
| Age (years) | 50.7 ± 12.8 | 50.6 ± 11.8 | 50.8 ± 13.6 |
| Male gender | 63 (76) | 30 (79) | 33 (73) |
| Body mass index (kg/m2) | 23.4 ± 3.9 | 24.0 ± 4.3 | 22.9 ± 3.4 |
| Aetiology | |||
| Non-ischaemic CMP | 45 (46) | 18 (47) | 27 (60) |
| Ischaemic CMP | 38 (54) | 20 (53) | 18 (40) |
| Comorbidities | |||
| Diabetes mellitus | 9 (11) | 3 (8) | 6 (13) |
| Hypertension | 13 (16) | 8 (21) | 5 (11) |
| ICD/PM | 55 (66) | 30 (79) | 25 (56)a |
| Myocardial infarction | 37 (45) | 21 (55) | 16 (37) |
| CABG | 6 (7) | 1 (3) | 5 (11) |
| CVA | 5 (6) | 4 (11) | 1 (2) |
| IABP | 31 (37) | 15 (40) | 16 (36) |
| ECMO | 14 (17) | 1 (3) | 13 (29)a |
| Days on ECMO | 6 (3–12) | 8 | 4 (3–13) |
| Bridge to transplantation | 74 (89) | 36 (95) | 38 (84) |
| Device type | |||
| HeartMate II | 64 (77) | 29 (76) | 35 (78) |
| HeartMate 3 | 19 (23) | 9 (24) | 10 (22) |
| INTERMACS | |||
| Class 1 | 16 (19) | 5 (13) | 11 (24) |
| Class 2 | 30 (36) | 13 (34) | 17 (38) |
| Class 3 | 17 (21) | 9 (24) | 8 (18) |
| Class ≥4 | 20 (24) | 11 (29) | 9 (20) |
| Medication | |||
| Vitamin K antagonist | 50 (60) | 27 (71) | 23 (51) |
| Heparin | 36 (43) | 14 (37) | 22 (49) |
| Baseline laboratory data | |||
| Platelet count (109/l) | 216 ± 76 | 240 ± 71 | 196 ± 76a |
| Creatinine (µmol/l) | 145 ± 70 | 155 ± 64 | 136 ± 74 |
| Haemoglobin (mmol/l) | 7.4 ± 1.3 | 7.4 ± 1.1 | 7.4 ± 1.4 |
| CRP (mg/l) | 43 ± 59 | 39 ± 54 | 45 ± 63 |
| INR | 1.8 ± 0.8 | 1.9 ± 0.9 | 1.8 ± 0.7 |
| aPTT (s) | 59 ± 28 | 57 ± 20 | 62 ± 33 |
Continuous variables are presented as mean ± standard deviation or median (interquartile range); categorical variables are presented as n (%).
For P-value <0.05.
aPTT: activated partial thromboplastin time; CABG: coronary artery bypass graft; CMP: cardiomyopathy; CRP: C-reactive protein; CVA: cerebrovascular accident; ECMO: extracorporeal membrane oxygenation; IABP: intra-aortic balloon pump; ICD: implantable cardioverter defibrillator; INTERMACS: Interagency Registry for Mechanically Assisted Circulatory Support; INR: international normalized ratio; PM: pacemaker.
Baseline characteristics of the study population
| All (n = 83) | No early bleeding event (n = 38) | Early bleeding event (n = 45) | |
|---|---|---|---|
| Age (years) | 50.7 ± 12.8 | 50.6 ± 11.8 | 50.8 ± 13.6 |
| Male gender | 63 (76) | 30 (79) | 33 (73) |
| Body mass index (kg/m2) | 23.4 ± 3.9 | 24.0 ± 4.3 | 22.9 ± 3.4 |
| Aetiology | |||
| Non-ischaemic CMP | 45 (46) | 18 (47) | 27 (60) |
| Ischaemic CMP | 38 (54) | 20 (53) | 18 (40) |
| Comorbidities | |||
| Diabetes mellitus | 9 (11) | 3 (8) | 6 (13) |
| Hypertension | 13 (16) | 8 (21) | 5 (11) |
| ICD/PM | 55 (66) | 30 (79) | 25 (56)a |
| Myocardial infarction | 37 (45) | 21 (55) | 16 (37) |
| CABG | 6 (7) | 1 (3) | 5 (11) |
| CVA | 5 (6) | 4 (11) | 1 (2) |
| IABP | 31 (37) | 15 (40) | 16 (36) |
| ECMO | 14 (17) | 1 (3) | 13 (29)a |
| Days on ECMO | 6 (3–12) | 8 | 4 (3–13) |
| Bridge to transplantation | 74 (89) | 36 (95) | 38 (84) |
| Device type | |||
| HeartMate II | 64 (77) | 29 (76) | 35 (78) |
| HeartMate 3 | 19 (23) | 9 (24) | 10 (22) |
| INTERMACS | |||
| Class 1 | 16 (19) | 5 (13) | 11 (24) |
| Class 2 | 30 (36) | 13 (34) | 17 (38) |
| Class 3 | 17 (21) | 9 (24) | 8 (18) |
| Class ≥4 | 20 (24) | 11 (29) | 9 (20) |
| Medication | |||
| Vitamin K antagonist | 50 (60) | 27 (71) | 23 (51) |
| Heparin | 36 (43) | 14 (37) | 22 (49) |
| Baseline laboratory data | |||
| Platelet count (109/l) | 216 ± 76 | 240 ± 71 | 196 ± 76a |
| Creatinine (µmol/l) | 145 ± 70 | 155 ± 64 | 136 ± 74 |
| Haemoglobin (mmol/l) | 7.4 ± 1.3 | 7.4 ± 1.1 | 7.4 ± 1.4 |
| CRP (mg/l) | 43 ± 59 | 39 ± 54 | 45 ± 63 |
| INR | 1.8 ± 0.8 | 1.9 ± 0.9 | 1.8 ± 0.7 |
| aPTT (s) | 59 ± 28 | 57 ± 20 | 62 ± 33 |
| All (n = 83) | No early bleeding event (n = 38) | Early bleeding event (n = 45) | |
|---|---|---|---|
| Age (years) | 50.7 ± 12.8 | 50.6 ± 11.8 | 50.8 ± 13.6 |
| Male gender | 63 (76) | 30 (79) | 33 (73) |
| Body mass index (kg/m2) | 23.4 ± 3.9 | 24.0 ± 4.3 | 22.9 ± 3.4 |
| Aetiology | |||
| Non-ischaemic CMP | 45 (46) | 18 (47) | 27 (60) |
| Ischaemic CMP | 38 (54) | 20 (53) | 18 (40) |
| Comorbidities | |||
| Diabetes mellitus | 9 (11) | 3 (8) | 6 (13) |
| Hypertension | 13 (16) | 8 (21) | 5 (11) |
| ICD/PM | 55 (66) | 30 (79) | 25 (56)a |
| Myocardial infarction | 37 (45) | 21 (55) | 16 (37) |
| CABG | 6 (7) | 1 (3) | 5 (11) |
| CVA | 5 (6) | 4 (11) | 1 (2) |
| IABP | 31 (37) | 15 (40) | 16 (36) |
| ECMO | 14 (17) | 1 (3) | 13 (29)a |
| Days on ECMO | 6 (3–12) | 8 | 4 (3–13) |
| Bridge to transplantation | 74 (89) | 36 (95) | 38 (84) |
| Device type | |||
| HeartMate II | 64 (77) | 29 (76) | 35 (78) |
| HeartMate 3 | 19 (23) | 9 (24) | 10 (22) |
| INTERMACS | |||
| Class 1 | 16 (19) | 5 (13) | 11 (24) |
| Class 2 | 30 (36) | 13 (34) | 17 (38) |
| Class 3 | 17 (21) | 9 (24) | 8 (18) |
| Class ≥4 | 20 (24) | 11 (29) | 9 (20) |
| Medication | |||
| Vitamin K antagonist | 50 (60) | 27 (71) | 23 (51) |
| Heparin | 36 (43) | 14 (37) | 22 (49) |
| Baseline laboratory data | |||
| Platelet count (109/l) | 216 ± 76 | 240 ± 71 | 196 ± 76a |
| Creatinine (µmol/l) | 145 ± 70 | 155 ± 64 | 136 ± 74 |
| Haemoglobin (mmol/l) | 7.4 ± 1.3 | 7.4 ± 1.1 | 7.4 ± 1.4 |
| CRP (mg/l) | 43 ± 59 | 39 ± 54 | 45 ± 63 |
| INR | 1.8 ± 0.8 | 1.9 ± 0.9 | 1.8 ± 0.7 |
| aPTT (s) | 59 ± 28 | 57 ± 20 | 62 ± 33 |
Continuous variables are presented as mean ± standard deviation or median (interquartile range); categorical variables are presented as n (%).
For P-value <0.05.
aPTT: activated partial thromboplastin time; CABG: coronary artery bypass graft; CMP: cardiomyopathy; CRP: C-reactive protein; CVA: cerebrovascular accident; ECMO: extracorporeal membrane oxygenation; IABP: intra-aortic balloon pump; ICD: implantable cardioverter defibrillator; INTERMACS: Interagency Registry for Mechanically Assisted Circulatory Support; INR: international normalized ratio; PM: pacemaker.
Forty-five (54%) patients experienced an early bleeding event requiring reoperation in 39 (47%) cases and >4 units PRBC in 6 (7%) cases. The median time to a bleeding event was 5 (interquartile range 3–7) days (Fig. 1A). Among these patients, 69% of the events occurred within 7 days following LVAD implantation and 91%, within 14 days. Patients who experienced an early bleeding event were significantly more often transfused with PRBC (7.8 ± 7.1 vs 1.3 ± 1.4, P < 0.001), fresh frozen plasma (0.9 ± 2.0 vs 0.3 ± 0.8, P = 0.04) and platelets (0.6 ± 1.6 vs 0.03 ± 0.17, P = 0.02) compared to patients who did not experience an early bleeding event, respectively. No difference was found in the early bleeding rate between HMII and HM3 (P = 0.87). When stratified by ECMO, patients who were on ECMO support had a significantly higher rate of early bleeding events compared to the non-ECMO group (100% vs 46%, P = 0.002, Fig. 1B). A subanalysis excluding ECMO patients, presented in Supplementary Material, Tables S1 and S2, includes baseline characteristics and surgical exploration summaries. The median time to discharge was 29 (interquartile range 26–32) days vs 46 (38–54) days for the non-bleeder versus the bleeders group (P = 0.009), respectively.
Cumulative incidence rate of early bleeding events. (A) Overall cumulative incidence rate of early bleeding events. (B) Cumulative incidence rate of early bleeding in patients on ECMO versus no ECMO support. ECMO: extracorporeal membrane oxygenation.
A description of the surgical exploration procedure is presented in Table 2. Of the 39 patients requiring re-exploration, no bleeding focus was found in 17 (44%) patients; in 8 (21%) patients, the end-to-side anastomosis of the outflow cannula to the ascending aorta was oozing or leaking. In 9 (23%) cases, there was pleural blood accumulation. In approximately half of the cases (n = 18.46%), no specific action was taken, and only a thoracic lavage with isotonic saline was done. In 12 (31%) cases, an additional stitch to the end-to-side anastomosis of the aorta or of a substernal bleeding focus was necessary. Overall, 10 (12%) patients required a 2nd and 4 (5%) patients, a 3rd surgical re-exploration for bleeding before discharge.
Surgical exploration summary
| Early bleeding (n = 45) | |
|---|---|
| Treated conservatively, n (%) | 6 (13) |
| Required re-exploration, n (%) | 39 (87) |
| Time to re-exploration (days), median (IQR) | 5 (3–7) |
| Bleeding focus during re-exploration (n=39) , n (%) | |
| No surgical site found | 17 (44) |
| End-to-side anastomosis aorta | 8 (21) |
| Substernal | 7 (18) |
| Diffuse bleeding | 4 (10) |
| Not reported | 3 (8) |
| Pleural blood accumulation, n (%) | 9 (23) |
| Treatment, n (%) | |
| Thoracic lavage with isotonic saline | 18 (46) |
| Suture | 12 (31) |
| Diathermy | 5 (13) |
| Not reported | 3 (8) |
| Pericardial window | 1 (3) |
| Second re-exploration, n (%) | 9 (23) |
| No surgical site found | 3 |
| End-to-side anastomosis aorta—suture | 1 |
| Substernal | |
| Suture | 1 |
| Diathermy | 2 |
| Driveline tunnel bleeding—fibrin sealant | 1 |
| Diaphragm fissure—suture | 1 |
| Third re-exploration, n (%) | 4 (10) |
| Surgical site not found | 2 |
| End-to-side anastomosis aorta—suture | 1 |
| Small fissure peritoneum—suture | 1 |
| Early bleeding (n = 45) | |
|---|---|
| Treated conservatively, n (%) | 6 (13) |
| Required re-exploration, n (%) | 39 (87) |
| Time to re-exploration (days), median (IQR) | 5 (3–7) |
| Bleeding focus during re-exploration (n=39) , n (%) | |
| No surgical site found | 17 (44) |
| End-to-side anastomosis aorta | 8 (21) |
| Substernal | 7 (18) |
| Diffuse bleeding | 4 (10) |
| Not reported | 3 (8) |
| Pleural blood accumulation, n (%) | 9 (23) |
| Treatment, n (%) | |
| Thoracic lavage with isotonic saline | 18 (46) |
| Suture | 12 (31) |
| Diathermy | 5 (13) |
| Not reported | 3 (8) |
| Pericardial window | 1 (3) |
| Second re-exploration, n (%) | 9 (23) |
| No surgical site found | 3 |
| End-to-side anastomosis aorta—suture | 1 |
| Substernal | |
| Suture | 1 |
| Diathermy | 2 |
| Driveline tunnel bleeding—fibrin sealant | 1 |
| Diaphragm fissure—suture | 1 |
| Third re-exploration, n (%) | 4 (10) |
| Surgical site not found | 2 |
| End-to-side anastomosis aorta—suture | 1 |
| Small fissure peritoneum—suture | 1 |
IQR: interquartile range.
Surgical exploration summary
| Early bleeding (n = 45) | |
|---|---|
| Treated conservatively, n (%) | 6 (13) |
| Required re-exploration, n (%) | 39 (87) |
| Time to re-exploration (days), median (IQR) | 5 (3–7) |
| Bleeding focus during re-exploration (n=39) , n (%) | |
| No surgical site found | 17 (44) |
| End-to-side anastomosis aorta | 8 (21) |
| Substernal | 7 (18) |
| Diffuse bleeding | 4 (10) |
| Not reported | 3 (8) |
| Pleural blood accumulation, n (%) | 9 (23) |
| Treatment, n (%) | |
| Thoracic lavage with isotonic saline | 18 (46) |
| Suture | 12 (31) |
| Diathermy | 5 (13) |
| Not reported | 3 (8) |
| Pericardial window | 1 (3) |
| Second re-exploration, n (%) | 9 (23) |
| No surgical site found | 3 |
| End-to-side anastomosis aorta—suture | 1 |
| Substernal | |
| Suture | 1 |
| Diathermy | 2 |
| Driveline tunnel bleeding—fibrin sealant | 1 |
| Diaphragm fissure—suture | 1 |
| Third re-exploration, n (%) | 4 (10) |
| Surgical site not found | 2 |
| End-to-side anastomosis aorta—suture | 1 |
| Small fissure peritoneum—suture | 1 |
| Early bleeding (n = 45) | |
|---|---|
| Treated conservatively, n (%) | 6 (13) |
| Required re-exploration, n (%) | 39 (87) |
| Time to re-exploration (days), median (IQR) | 5 (3–7) |
| Bleeding focus during re-exploration (n=39) , n (%) | |
| No surgical site found | 17 (44) |
| End-to-side anastomosis aorta | 8 (21) |
| Substernal | 7 (18) |
| Diffuse bleeding | 4 (10) |
| Not reported | 3 (8) |
| Pleural blood accumulation, n (%) | 9 (23) |
| Treatment, n (%) | |
| Thoracic lavage with isotonic saline | 18 (46) |
| Suture | 12 (31) |
| Diathermy | 5 (13) |
| Not reported | 3 (8) |
| Pericardial window | 1 (3) |
| Second re-exploration, n (%) | 9 (23) |
| No surgical site found | 3 |
| End-to-side anastomosis aorta—suture | 1 |
| Substernal | |
| Suture | 1 |
| Diathermy | 2 |
| Driveline tunnel bleeding—fibrin sealant | 1 |
| Diaphragm fissure—suture | 1 |
| Third re-exploration, n (%) | 4 (10) |
| Surgical site not found | 2 |
| End-to-side anastomosis aorta—suture | 1 |
| Small fissure peritoneum—suture | 1 |
IQR: interquartile range.
Univariable analysis showed that ECMO preimplantation (P = 0.011) and lower platelet counts directly pre- and postimplant (P = 0.013, P = 0.012, respectively) were associated with a higher probability for an early bleeding event (Table 3). A separate univariable analysis excluding patients on ECMO is included in Supplementary Material, Table S3. In the multivariable analysis, ECMO preimplantation and a lower platelet count postimplant were independently associated with a higher probability for an early bleeding event (Model 1, Table 4). The C-statistic of Model 1 was 0.73 with a sensitivity of 82% and a specificity of 53%. Receiver operating characteristic analysis showed that the optimal cut-off value for the postimplant platelet count was 150 × 109/l. Incorporated in the multivariable analysis (Model 2), patients with a postimplant platelet count <150 × 109/l, which is the lower limit of normal in our institution, had a 4.5 times higher probability and patients on ECMO preimplantation had a 9.6 times higher probability for an early bleeding event.
Univariable logistic regression analysis for early bleeding
| Baseline variables | OR | 95% CI | P-value |
|---|---|---|---|
| Age (years) | 1.001 | 0.97–1.04 | 0.938 |
| BMI (kg/m2) | 0.925 | 0.82–1.04 | 0.191 |
| Female gender | 1.364 | 0.49–3.80 | 0.552 |
| Aetiology (non-ischaemic) | 1.667 | 0.70–4.00 | 0.251 |
| Diabetes | 1.795 | 0.42–7.72 | 0.432 |
| Hypertension | 0.469 | 0.14–1.58 | 0.221 |
| CABG | 4.625 | 0.52–41.4 | 0.171 |
| IABP | 0.846 | 0.35–2.06 | 0.713 |
| ECMO | 15.03 | 1.86–121.3 | 0.011a |
| Device (HeartMate III) | 0.921 | 0.33–2.57 | 0.875 |
| Laboratory values, preimplantation | |||
| Platelet count (109/l) | 0.992 | 0.99–1.00 | 0.013a |
| Creatinine (µmol/l) | 0.996 | 0.99–1.00 | 0.237 |
| Haemoglobin (mmol/l) | 0.994 | 0.71–1.40 | 0.974 |
| CRP (mg/l) | 1.002 | 0.99–1.01 | 0.693 |
| INR | 0.783 | 0.43–1.42 | 0.422 |
| aPTT (s) | 1.007 | 0.99–1.03 | 0.441 |
| Laboratory values, postimplant | |||
| Platelet count (109/l) | 0.985 | 0.97–1.00 | 0.012a |
| Creatinine (µmol/l) | 1.000 | 0.99–1.00 | 0.356 |
| Haemoglobin (mmol/l) | 1.562 | 0.96–2.55 | 0.073 |
| CRP (mg/l) | 1.005 | 0.99–1.02 | 0.449 |
| INR | 0.531 | 0.16–1.73 | 0.293 |
| aPTT (s) | 1.011 | 0.98–1.05 | 0.543 |
| Baseline variables | OR | 95% CI | P-value |
|---|---|---|---|
| Age (years) | 1.001 | 0.97–1.04 | 0.938 |
| BMI (kg/m2) | 0.925 | 0.82–1.04 | 0.191 |
| Female gender | 1.364 | 0.49–3.80 | 0.552 |
| Aetiology (non-ischaemic) | 1.667 | 0.70–4.00 | 0.251 |
| Diabetes | 1.795 | 0.42–7.72 | 0.432 |
| Hypertension | 0.469 | 0.14–1.58 | 0.221 |
| CABG | 4.625 | 0.52–41.4 | 0.171 |
| IABP | 0.846 | 0.35–2.06 | 0.713 |
| ECMO | 15.03 | 1.86–121.3 | 0.011a |
| Device (HeartMate III) | 0.921 | 0.33–2.57 | 0.875 |
| Laboratory values, preimplantation | |||
| Platelet count (109/l) | 0.992 | 0.99–1.00 | 0.013a |
| Creatinine (µmol/l) | 0.996 | 0.99–1.00 | 0.237 |
| Haemoglobin (mmol/l) | 0.994 | 0.71–1.40 | 0.974 |
| CRP (mg/l) | 1.002 | 0.99–1.01 | 0.693 |
| INR | 0.783 | 0.43–1.42 | 0.422 |
| aPTT (s) | 1.007 | 0.99–1.03 | 0.441 |
| Laboratory values, postimplant | |||
| Platelet count (109/l) | 0.985 | 0.97–1.00 | 0.012a |
| Creatinine (µmol/l) | 1.000 | 0.99–1.00 | 0.356 |
| Haemoglobin (mmol/l) | 1.562 | 0.96–2.55 | 0.073 |
| CRP (mg/l) | 1.005 | 0.99–1.02 | 0.449 |
| INR | 0.531 | 0.16–1.73 | 0.293 |
| aPTT (s) | 1.011 | 0.98–1.05 | 0.543 |
P < 0.05.
aPTT: activated partial thromboplastin time; BMI: body mass index; CABG: coronary artery bypass graft; CI: confidence interval; CRP: C-reactive protein; ECMO: extracorporeal membrane oxygenation; IABP: intra-aortic balloon pump; INR: international normalized ratio; OR: odds ratio.
Univariable logistic regression analysis for early bleeding
| Baseline variables | OR | 95% CI | P-value |
|---|---|---|---|
| Age (years) | 1.001 | 0.97–1.04 | 0.938 |
| BMI (kg/m2) | 0.925 | 0.82–1.04 | 0.191 |
| Female gender | 1.364 | 0.49–3.80 | 0.552 |
| Aetiology (non-ischaemic) | 1.667 | 0.70–4.00 | 0.251 |
| Diabetes | 1.795 | 0.42–7.72 | 0.432 |
| Hypertension | 0.469 | 0.14–1.58 | 0.221 |
| CABG | 4.625 | 0.52–41.4 | 0.171 |
| IABP | 0.846 | 0.35–2.06 | 0.713 |
| ECMO | 15.03 | 1.86–121.3 | 0.011a |
| Device (HeartMate III) | 0.921 | 0.33–2.57 | 0.875 |
| Laboratory values, preimplantation | |||
| Platelet count (109/l) | 0.992 | 0.99–1.00 | 0.013a |
| Creatinine (µmol/l) | 0.996 | 0.99–1.00 | 0.237 |
| Haemoglobin (mmol/l) | 0.994 | 0.71–1.40 | 0.974 |
| CRP (mg/l) | 1.002 | 0.99–1.01 | 0.693 |
| INR | 0.783 | 0.43–1.42 | 0.422 |
| aPTT (s) | 1.007 | 0.99–1.03 | 0.441 |
| Laboratory values, postimplant | |||
| Platelet count (109/l) | 0.985 | 0.97–1.00 | 0.012a |
| Creatinine (µmol/l) | 1.000 | 0.99–1.00 | 0.356 |
| Haemoglobin (mmol/l) | 1.562 | 0.96–2.55 | 0.073 |
| CRP (mg/l) | 1.005 | 0.99–1.02 | 0.449 |
| INR | 0.531 | 0.16–1.73 | 0.293 |
| aPTT (s) | 1.011 | 0.98–1.05 | 0.543 |
| Baseline variables | OR | 95% CI | P-value |
|---|---|---|---|
| Age (years) | 1.001 | 0.97–1.04 | 0.938 |
| BMI (kg/m2) | 0.925 | 0.82–1.04 | 0.191 |
| Female gender | 1.364 | 0.49–3.80 | 0.552 |
| Aetiology (non-ischaemic) | 1.667 | 0.70–4.00 | 0.251 |
| Diabetes | 1.795 | 0.42–7.72 | 0.432 |
| Hypertension | 0.469 | 0.14–1.58 | 0.221 |
| CABG | 4.625 | 0.52–41.4 | 0.171 |
| IABP | 0.846 | 0.35–2.06 | 0.713 |
| ECMO | 15.03 | 1.86–121.3 | 0.011a |
| Device (HeartMate III) | 0.921 | 0.33–2.57 | 0.875 |
| Laboratory values, preimplantation | |||
| Platelet count (109/l) | 0.992 | 0.99–1.00 | 0.013a |
| Creatinine (µmol/l) | 0.996 | 0.99–1.00 | 0.237 |
| Haemoglobin (mmol/l) | 0.994 | 0.71–1.40 | 0.974 |
| CRP (mg/l) | 1.002 | 0.99–1.01 | 0.693 |
| INR | 0.783 | 0.43–1.42 | 0.422 |
| aPTT (s) | 1.007 | 0.99–1.03 | 0.441 |
| Laboratory values, postimplant | |||
| Platelet count (109/l) | 0.985 | 0.97–1.00 | 0.012a |
| Creatinine (µmol/l) | 1.000 | 0.99–1.00 | 0.356 |
| Haemoglobin (mmol/l) | 1.562 | 0.96–2.55 | 0.073 |
| CRP (mg/l) | 1.005 | 0.99–1.02 | 0.449 |
| INR | 0.531 | 0.16–1.73 | 0.293 |
| aPTT (s) | 1.011 | 0.98–1.05 | 0.543 |
P < 0.05.
aPTT: activated partial thromboplastin time; BMI: body mass index; CABG: coronary artery bypass graft; CI: confidence interval; CRP: C-reactive protein; ECMO: extracorporeal membrane oxygenation; IABP: intra-aortic balloon pump; INR: international normalized ratio; OR: odds ratio.
Multivariable logistic regression analysis for early bleeding
| OR | 95% CI | P-value | |
|---|---|---|---|
| Model 1: multivariable analysis including platelet count (continue) | |||
| Postimplant platelet count (lower) | 1.01 | 1.00–1.03 | 0.04 |
| Preimplant ECMO support | 11.34 | 1.38–93.2 | 0.02 |
| Model 2: multivariable analysis including platelet count (dichotomized) | |||
| Postimplant platelet count <150 × 109/l | 4.46 | 1.57–12.7 | 0.005 |
| Preimplant ECMO support | 9.64 | 1.15–81.1 | 0.04 |
| OR | 95% CI | P-value | |
|---|---|---|---|
| Model 1: multivariable analysis including platelet count (continue) | |||
| Postimplant platelet count (lower) | 1.01 | 1.00–1.03 | 0.04 |
| Preimplant ECMO support | 11.34 | 1.38–93.2 | 0.02 |
| Model 2: multivariable analysis including platelet count (dichotomized) | |||
| Postimplant platelet count <150 × 109/l | 4.46 | 1.57–12.7 | 0.005 |
| Preimplant ECMO support | 9.64 | 1.15–81.1 | 0.04 |
CI: confidence interval; ECMO: extracorporeal membrane oxygenation; OR: odds ratio.
Multivariable logistic regression analysis for early bleeding
| OR | 95% CI | P-value | |
|---|---|---|---|
| Model 1: multivariable analysis including platelet count (continue) | |||
| Postimplant platelet count (lower) | 1.01 | 1.00–1.03 | 0.04 |
| Preimplant ECMO support | 11.34 | 1.38–93.2 | 0.02 |
| Model 2: multivariable analysis including platelet count (dichotomized) | |||
| Postimplant platelet count <150 × 109/l | 4.46 | 1.57–12.7 | 0.005 |
| Preimplant ECMO support | 9.64 | 1.15–81.1 | 0.04 |
| OR | 95% CI | P-value | |
|---|---|---|---|
| Model 1: multivariable analysis including platelet count (continue) | |||
| Postimplant platelet count (lower) | 1.01 | 1.00–1.03 | 0.04 |
| Preimplant ECMO support | 11.34 | 1.38–93.2 | 0.02 |
| Model 2: multivariable analysis including platelet count (dichotomized) | |||
| Postimplant platelet count <150 × 109/l | 4.46 | 1.57–12.7 | 0.005 |
| Preimplant ECMO support | 9.64 | 1.15–81.1 | 0.04 |
CI: confidence interval; ECMO: extracorporeal membrane oxygenation; OR: odds ratio.
During the 1st year after implantation, 14 (17%) patients died and 19 (23%) underwent heart transplants. The Kaplan–Meier survival curve, stratified by early bleeding event, is presented in Fig. 2. Patients who experienced an early bleeding event had a significant lower survival rate at 90 days compared to patients who did not (80% vs 97%, P = 0.02). In addition, there was a significant difference in survival rate between patients with and without an early bleeding event at 1 year (75% vs 91%, log-rank P = 0.04).
Kaplan–Meier survival curve by early bleeding event. (A) Early survival by early bleeding event. (B) First-year survival by early bleeding event.
DISCUSSION
This study addressed the incidence, predictors and outcome of early bleeding events requiring surgical exploration after implantation of an LVAD. This study showed that (i) overall incidence of an early bleeding event was high after LVAD implantation, with a median time of 5 days and the majority of the events occurring within 14 days; (ii) the occurrence of an early bleeding event was related to ECMO support pre-LVAD implantation and a low platelet count pre- and postimplant and (iii) patients experiencing an early bleeding event had impaired survival compared to patients who did not experience this event. This result suggests that the patients previously on ECMO or with a low platelet count should be carefully monitored during the initial admission in the intensive care unit postimplant, in order for the medical team to intervene adequately and on time.
Achieving haemostasis following LVAD implantation can be challenging, if not difficult, because of fragile tissues, coagulation disorders and extensive blood loss. In the recently published non-blinded randomized controlled trial comparing the HMII with the HM3, the overall bleeding rate was 33% and 39% for patients with the HM3 and the HMII, respectively, with 10% of the patients with the HM3 and 14% of those with the HMII experiencing a bleeding event that required surgery [7]. In the HMII LVAD bridge-to-transplant trial, the rate of bleeding events requiring reoperation was 31%, with 53% of patients being transfused with at least 2 units of red blood cells [8]. Finally, in the HMII versus HeartMate XVE trial, the reoperation rate for bleeding was 30% postimplant in patients with the HMII, with 81% of the cohort requiring blood transfusions for bleeding [2]. The incidence of bleeding (55%) was higher in this study than in those mentioned previously. This finding is likely due to the fact that the present study also included patients needing ECMO support preoperatively. If we excluded the patients on ECMO from the analysis, the incidence for early bleeding would be similar to that found in the literature. However, the high risk of early bleeding events after LVAD implant in this subgroup is an important finding of the present study, especially considering the fact that the use of ECMO is increasing globally and recognition of this high-risk group is of paramount importance to decrease the mortality and morbidity of patients with have an LVAD [9].
Although thromboembolic events and acute pump thrombosis are potentially life-threatening in patients with an LVAD, the risk of bleeding is higher than the risk of thrombosis in the early phase [10]. This situation underscores the importance of a structural approach to prevent and manage this complication. ECMO support preimplantation and a lower platelet count postimplant were independent predictors of early bleeding events after LVAD implantation. ECMO has been used with success as a bridge to an LVAD or a heart transplant [11, 12]. Nevertheless, the use of ECMO is accompanied by a high risk of complications as a consequence of heparinization and acquired coagulopathies, with 19% of the patients experiencing cannulation site haemorrhage and 20%, surgical site haemorrhage [9]. Factors including acquired von Willebrand factor deficiency, haemolysis and thrombocytopenia contribute to the bleeding risk during ECMO support [13, 14]. In addition, these conditions remain present after LVAD implantation, despite the removal of ECMO, and might be more severe due to blood loss and heparinization during surgery [15]. Recent findings have confirmed that nearly all patients with an LVAD, regardless of device, experience a loss in large von Willebrand multimers, subsequently resulting in reduced von Willebrand factor activity and an acquired form of von Willebrand deficiency [16], also known as acquired von Willebrand syndrome. This loss in activity is observed early in the postoperative period and persists during support with an LVAD [17, 18]. Furthermore, there is an increased risk of heparin-induced thrombocytopenia and disseminated intravascular coagulation, which all add to the burden of haematological complications during ECMO support and the perioperative period [14, 19]. All these factors and intrinsic changes could explain the high risk of bleeding in patients bridged with ECMO compared to patients who are not bridged with an ECMO.
In this study, platelet count was the only laboratory value independently associated with an early bleeding event, with the postimplant platelet count having the highest sensitivity. In addition, we found that patients with a platelet count <150 × 109/l had a nearly 5-fold higher risk of an early bleeding event. The current standard of care for the monitoring of perioperative coagulation consists of platelet count, prothrombin time/INR and aPTT. The use of laboratory data obtained directly pre- and postoperatively may explain the lack of the association between prothrombin time/INR, aPTT and bleeding in our study. Longitudinal changes in these variables and their association with bleeding in patients with an LVAD have yet to be determined.
Platelet function tests, point-of-care thromboelastography- and (rotational) thromboelastometry-based coagulation management have been found to significantly reduce the re-exploration rate in patients having cardiac surgery [20]. However, articles regarding their clinical use in patients on LVAD support are scarce. We are currently performing a study to investigate whether rotational thromboelastometry can be used as a predictor of bleeding events after implantation of an LVAD. Furthermore, more proactive use of the echocardiographic assessment of pericardial effusions could detect a cardiac tamponade earlier on. Therefore, perioperative monitoring of platelet count and coagulation parameters in combination with frequent postoperative echocardiographic follow-up scans could be helpful for prevention and detection of a bleeding event and provide time for a planned intervention.
In a recent study, Rojas et al. [21] reported that no postoperative bleeding was observed in 26 destination therapy patients who received an LVAD through less invasive surgery. However, studies regarding less invasive surgery for LVAD implants are limited by their study design and sample size. Larger randomized studies are needed to confirm if less invasive surgery is superior to conventional surgery. In addition, formal evaluation of the efficacy and efficiency is required to determine the absolute benefit.
In the present study, an early bleeding event requiring reoperation was associated with impaired survival at 90 days and at 1 year. This result is in line with previously published data; Genovese et al. [22] reported a significantly higher mortality rate among patients who required a reoperation 1 year after receiving an LVAD implant.
Limitations
This study has several limitations. The retrospective single-centre study design and centre-specific protocol for patient selection and management may weaken the applicability of our findings. Furthermore, data on platelet function or von Willebrand factors were not available; these data are essential in order to understand the underlying mechanisms that contribute to the higher risk of bleeding in patients on an LVAD. Future prospective research is needed to determine the relation between these factors and the risk of bleeding in patients on an LVAD. Because postoperative echocardiographic evaluation was not available systematically in all patients, assessment of the predictive value of echocardiograms and a comparison between the groups were not possible. Finally, the number of HM3 devices is low; therefore, direct comparisons between devices is of limited power. The strengths of our study include the complete follow-up, the pre- and postimplant laboratory monitoring and the multivariate models, all of which lead to a clinically useful and accurate result.
CONCLUSIONS
Early bleeding events are highly prevalent after implantation of an LVAD, especially in patients bridged with ECMO and with low platelet counts. These patients had a significantly worse survival rate compared to patients who did not experience a bleeding event. No difference was found in the rates of early bleeding events in patients with different types of LVADs.
SUPPLEMENTARY MATERIAL
Supplementary material is available at EJCTS online.
Funding
This work was supported by a NWO Veni grant from the Netherlands Organisation for Scientific Research.
Conflict of interest: Frank W.G. Leebeek is a consultant for UniQure, Shire and NovoNordisk and has received unrestricted research grants from CSL Behring and Shire, all of which are outside the scope of the submitted work.
