Predictors and Outcomes of Renal Replacement Therapy After Left Ventricular Assist Device Implantation

ORIGINAL ARTICLE

Predictors and Outcomes of Renal Replacement Therapy After Left Ventricular Assist Device Implantation Rabea Asleh, MD, PhD, MHA; Sarah Schettle, PAC, MS; Alexandros Briasoulis, MD, PhD; Jill M. Killian, BS; John M. Stulak, MD; Naveen L. Pereira, MD; Sudhir S. Kushwaha, MD; Simon Maltais, MD, PhD; and Shannon M. Dunlay, MD, MS Abstract Objective: To examine the frequency and outcomes of patients requiring renal replacement therapy (RRT) early after left ventricular assist device (LVAD) implantation. Patients and Methods: We examined use of in-hospital RRT and outcomes in consecutive adults who underwent continuous-flow LVAD implantation from February 15, 2007, through August 8, 2017. Logistic regression was used to examine predictors of RRT. The associations of RRT with outcomes were examined using Cox proportional hazards regression. Results: Of 354 patients who underwent LVAD implantation, 54 (15%) required in-hospital RRT. Patients receiving RRT had higher preoperative Charlson Comorbidity Index values (median, 5 vs 4; P¼.03), Model for End-Stage Liver Disease scores (mean, 19.0 vs 14.5; P<.001), right atrial pressure (mean, 19.1 vs 13.4 mm Hg; P<.001), and estimated 24-hour urine protein levels (median, 357 vs 174 mg; P<.001) and lower preoperative estimated glomerular filtration rate (eGFR) (median, 43 vs 57 mL/min; P<.001) and measured GFR using 125I-iothalamate clearance (median, 33 vs 51 mL/min; P¼.001) than those who did not require RRT. Approximately 40% of patients with eGFR less than 45 mL/min/1.73 m2 and 24-hour urine protein level greater than 400 mg required RRT vs 6% with eGFR greater than45 mL/min/1.73 m2 and without significant proteinuria. Lower preoperative eGFR, higher estimated 24-hour urine protein level, higher right atrial pressure, and longer cardiopulmonary bypass time were independent predictors of RRT after LVAD implantation. Of patients requiring in-hospital RRT, 18 (33%) had renal recovery, 18 (33%) required outpatient hemodialysis, and 18 (33%) died before hospital discharge. After median (Q1, Q3) follow-up of 24.3 (8.9, 49.6) months, RRT was associated with increased risk of death (adjusted hazard ratio [HR], 2.86; 95% CI, 1.90-4.33; P<.001) and gastrointestinal bleeding (adjusted HR, 4.47; 95% CI, 2.57-7.75; P<.001). Conclusion: In-hospital RRT is associated with poor prognosis after LVAD. A detailed preoperative assessment of renal function before LVAD may be helpful in risk stratification and patient selection. ª 2018 Mayo Foundation for Medical Education and Research

C

ontinuous-flow left ventricular assist devices (LVADs) are increasingly used for patients with advanced heart failure (HF) as a bridge to transplant (BTT) or as destination therapy (DT), and they have been shown to improve survival and quality of life in these settings.1-3 However, the postoperative course of LVAD recipients is often complicated by hospital readmissions for gastrointestinal bleeding, device

n

Mayo Clin Proc. 2019;94(6):1003-1014

infection, thromboembolic events, and right ventricular (RV) failure; hence, proper patient selection is critical to achieve the desired outcomes.4,5 Hemodynamic parameters improve after LVAD implantation, with lower pulmonary arterial and pulmonary capillary wedge pressures and higher cardiac outputs.6 Although cardiorenal syndrome with estimated glomerular filtration rate (eGFR) less than

Mayo Clin Proc. n June 2019;94(6):1003-1014 n https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org n ª 2018 Mayo Foundation for Medical Education and Research

See also page 929 From the Department of Cardiovascular Medicine (R.A., S.S., J.M.S., N.L.P., S.S.K., S.M., S.M.D.), Department of Affiliations continued at the end of this article.

1003

MAYO CLINIC PROCEEDINGS

TABLE 1. Baseline Characteristics of the Study Populationa,b Characteristic

Overall (N¼354)

No RRT (n¼300)

RRT (n¼54)

P value

Age (y), mean  SD

60.412.1

61.011.7

57.113.8

.03

Male sex (No. [%])

279 (78.8)

242 (80.7)

37 (68.5)

.04

White race (No. [%])

313 (88.4)

269 (92.1)

44 (84.6)

.08

Device type (No. [%]) HeartMate II HeartWare HeartMate 3

282 (79.7) 64 (18.1) 8 (2.3)

247 (82.3) 46 (15.3) 7 (2.3)

35 (64.8) 18 (33.3) 1 (1.9)

Device as DT (No. [%])

238 (67.2)

201 (67.0)

37 (68.5)

INTERMACS profile (No. [%])c Class 1 Class 2 Class 3 Class 4 Class 5 Class 6

48 (16.8) 72 (25.3) 58 (20.3) 79 (27.7) 17 (6.0) 11 (3.9)

39 (16.2) 63 (26.1) 48 (19.9) 64 (26.6) 16 (6.6) 11 (4.6)

9 (20.5) 9 (20.5) 10 (22.6) 15 (34.1) 1 (2.3) 0

ICM (No. [%])

165 (46.6)

144 (48.0)

21 (38.9)

.22

4.42.1

4.32.1

4.92.2

.03

143 (40.4)

125 (41.7)

18 (33.3)

.25

.01

.83 .42

Charlson Comorbidity Index score, mean  SD Hypertension (No. [%]) Diabetes mellitus (No. [%])

138 (39.0)

114 (38.0)

24 (44.4)

.37

Hyperlipidemia (No. [%])

147 (41.5)

127 (42.3)

20 (37.0)

.47

Atrial fibrillation (No. [%])

162 (45.8)

130 (43.3)

32 (59.3)

.03

BMI, mean  SD

29.05.8

28.85.7

30.46.0

.05

Redo sternotomy (No. [%])

127 (35.9)

102 (34.0)

25 (46.3)

.08

Left thoracotomy (No. [%])

19 (5.4)

15 (5.0)

4 (7.4)

.47

Preoperative continuous RRT (No. [%])

12 (3.4)

5 (1.7)

7 (13.0)

<.001

ECMO support before LVAD (No. [%])

12 (3.4)

5 (1.7)

7 (13.0)

<.001

IABP support (No. [%])

165 (46.6)

141 (47.0)

24 (44.4)

.73

CPB time (min), mean  SD

111.451.7

106.949.4

136.257.1

<.001

Concomitant surgery (No. [%])

185 (52.3)

146 (48.7)

39 (72.2)

.001

Preoperative hemodynamics, mean  SD MAP (mm Hg) RAP mm Hg mPAP (mm Hg) PCWP (mm Hg) PVR (WU) Cardiac output (L/min) Cardiac index (L/min/m2) RAP/PCWP ratio >0.63 (No. [%])

75.511.0 14.26.7 35.69.4 22.87.5 3.92.7 3.81.2 1.90.5 142 (42.6)

75.911.2 13.46.2 35.69.5 22.57.4 4.02.7 3.81.2 1.90.5 109 (38.2)

72.79.0 19.17.6 35.48.8 24.47.9 3.22.7 3.91.4 1.90.6 33 (68.8)

.14 <.001 .87 .12 .08 .46 .58 <.001

LVEF (%), mean  SD

19.68.8

19.07.6

23.013.3

.003

Preoperative inotropes (No. [%])

226 (64.6)

191 (64.3)

35 (66.0)

.81

LVEDD (mm), mean  SD

71.127.4

72.429.0

63.513.0

.04

ICD (No. [%])

234 (66.1)

194 (64.7)

40 (74.1)

.18

3.80.6

3.80.6

3.60.5

.02

Albumin (g/dL), mean  SD Total bilirubin (mg/dL), mean  SD

1.41.3

1.30.8

2.12.6

<.001

MELD score, mean  SD

15.26.0

14.55.5

19.06.8

<.001

NT-BNP (pg/mL), median (Q1, Q3)

4456 (2583, 8425) 4150 (2352, 7562) 6494 (3249, 14,475)

.01

Continued on next page

1004

Mayo Clin Proc.

n

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org

HEMODIALYSIS AFTER LVAD

TABLE 1. Continued Characteristic

Overall (N¼354)

No RRT (n¼300)

RRT (n¼54)

P value

Hemoglobin (g/dL), mean  SD

11.71.9

11.91.9

10.61.6

<.001

Hemoglobin A1c (%), mean  SD

6.41.4

6.41.4

6.21.4

.32

172.267.4

173.864.8

163.480.6

.30

11.31.1

11.31.1

11.21.1

.57

Platelet count (103/mL), mean  SD Preoperative kidney size (cm), mean  SD

BMI ¼ body mass index; CPB ¼ cardiopulmonary bypass; DT ¼ destination therapy; ECMO ¼ extracorporeal membrane oxygenation; IABP ¼ intra-aortic balloon pump; ICD ¼ implantable cardiac defibrillator; ICM ¼ ischemic cardiomyopathy; INTERMACS ¼ Interagency Registry for Mechanically Assisted Circulatory Support; LVEDD ¼ left ventricular end-diastolic diameter; LVEF ¼ left ventricular ejection fraction; MAP ¼ mean arterial pressure; MELD ¼ Model for End-Stage Liver Disease; mPAP ¼ mean pulmonary arterial pressure; NTBNP ¼ NT pro-Betype natriuretic peptide; PCWP ¼ pulmonary capillary wedge pressure; PVR ¼ pulmonary vascular resistance; RAP ¼ right atrial pressure; RRT ¼ renal replacement therapy; WU ¼ Wood units. b SI conversion factors: To convert albumin values to g/L, multiply by 10; to convert bilirubin values to mmol/L, multiply by 17.104; to convert hemoglobin values to g/L, multiply by 10; to convert platelet counts to 109/L, multiply by 1. c Data are available for 285 patients (241 in the non-RRT group and 44 in the RRT group). a

60 mL/min/1.73 m2 is found in more than half of patients with decompensated HF, renal function improves to eGFR greater than 60 mL/min/1.73 m2 in more than twothirds of patients within a month after LVAD implantation.7 However, early improvement of renal function is not sustained 1 year after LVAD support.8 Moreover, survival and heart transplant rates are significantly lower in patients with renal failure before LVAD implantation,9 and preoperative creatinine level is a strong independent predictor of 90-day mortality.10 More importantly, postoperative renal failure is associated with decreased survival,9 and patients who require renal replacement therapies (RRTs) for acute renal failure after LVAD have lower survival and heart transplant rates.11 A recent retrospective analysis of Medicare beneficiaries with end-stage renal disease (ESRD) before LVAD implantation demonstrated a dismal prognosis as most patients survived less than 3 weeks after surgery.12 Therefore, ESRD requiring RRT is a contraindication for LVAD as DT,13 and preoperative “optimization” with diuresis, inotropic, or temporary mechanical support is often used to improve renal function in patients with cardiorenal syndrome.14 Given the increasing complexity of LVAD candidates and the limited data on the impact of renal function on outcomes after LVAD implantation, we sought to identify independent risk factors for Mayo Clin Proc. n June 2019;94(6):1003-1014 www.mayoclinicproceedings.org

n

postoperative RRT and investigate the clinical outcomes of patients requiring RRT after LVAD implantation. PATIENTS AND METHODS Data Source This was a retrospective study of prospectively collected data from a single large institution. We identified 354 consecutive adult patients (age 18 years) with advanced HF who underwent implantation of continuous-flow (axial or centrifugal) LVADs (HeartMate II and HeartMate 3 [Thoratec], HeartWare [HeartWare Inc]), from February 15, 2007, through August 8, 2017, as either DT or BTT, at Mayo Clinic in Rochester, Minnesota. Patients who required a biventricular assist device or a total artificial heart at the time of surgery were excluded from the study. Patients were censored for LVAD exchange and were not included as separate cases. The study protocol was approved by the Mayo Clinic Institutional Review Board. We excluded 3 patients who declined to provide Minnesota Research Authorization, which allows access to their medical records for retrospective studies. Clinical and Demographic Data Demographic, clinical, echocardiographic, LVAD, and laboratory data were collected from the patients’ electronic medical records. Comorbidity burden was assessed by the

https://doi.org/10.1016/j.mayocp.2018.09.021

1005

MAYO CLINIC PROCEEDINGS

Charlson Comorbidity Index.15 We calculated the Model for End-Stage Liver Disease (MELD) score for each patient based on the following formula: 9.57  ln(serum creatinine [mg/dL]) þ 3.78  ln(total bilirubin [mg/dL]) þ 11.2  ln(INR) þ 6.43.16 To convert creatinine values to mmol/L, multiply by 88.4; to convert bilirubin values to mmol/L, multiply by 17.104. Evaluation of Renal Function Renal function was measured by calculating eGFR using the abbreviated Modification of Diet in Renal Disease equation: GFR ¼ 175  (serum creatinine [mg/dL])1.154  (age [years])0.203  0.742 (if female)  1.21 (if African American).17 Renal function was assessed 1 week before admission for LVAD implantation, the morning before LVAD implantation, and 1, 3, and 6 months after implantation during routine follow-up visits. We estimated 24-hour urine protein concentrations based on a spot urine collection performed in the week before LVAD implantation. Since 2008, LVAD candidates in our institution have also undergone routine evaluation of their renal function by measuring GFR using 125I-iothalamate, an exogenous filtration marker that has been widely used for internal and external validation of eGFR measurements using endogenous filtration markers (such as creatinine and cystatin C). Measured GFR was calculated as previously described18 with patients in a quiet room in a semisupine position. Blood samples were collected, and 125I-iothalamate and 131I-hippurate infusions were given. After a 2-hour stabilization period, baseline measurements started in a steady state of plasma tracer levels and clearance were calculated. Correction for incomplete bladder emptying and dead space was achieved by multiplying the urinary 125Iiothalamate clearances with plasma and urinary 131I-hippurate clearance. Hemodynamic Measurements Hemodynamic parameters, including mean arterial pressure, mean right atrial pressure (RAP), mean pulmonary arterial pressure, pulmonary capillary wedge pressure 1006

Mayo Clin Proc.

n

(PCWP), RV dP/Dt, transpulmonary gradient, cardiac output and cardiac index based on the Fick equation and thermodilution method, pulmonary vascular resistance, RV stroke work index, and RAP/PCWP ratio, were measured during right heart catheterization performed in all patients before LVAD implantation. Outcomes The main outcomes included (1) need for inhospital intermittent hemodialysis (HD) or continuous RRT, (2) all-cause mortality, and (3) length of stay (LOS) after LVAD implantation. Furthermore, we studied LVAD-related outcomes, including (1) RV failure, (2) suspected or confirmed pump thrombosis, (3) thromboembolic stroke or transient ischemic attack, (4) gastrointestinal bleeding, and (5) LVAD driveline or pump infection. All adverse events were defined according to the standard Interagency Registry for Mechanically Assisted Circulatory Support definition. Criteria for suspected pump thrombosis included elevation of lactate dehydrogenase concentration in addition to other clinical findings of hemolysis, including recent arterial thromboembolic event, symptoms of HF confirmed with abnormal hemodynamic findings, and presence of abnormal pump variables or pump function. Survival and clinical events information were obtained from the electronic medical record; outside medical records are routinely sought when patients are hospitalized at another institution after LVAD implantation. Statistical Analyses We expressed normally distributed data as mean  SD and nonnormally distributed data as median with first and third quartiles (Q1, Q3). Patient characteristics were compared between groups using the c2 test for categorical variables, the independent t test for normally distributed continuous variables, and the Mann-Whitney U test for continuous variables with skewed distribution. A logistic regression model, with adjustment for age, eGFR, estimated 24hour urine protein level, mean RAP, and

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org

HEMODIALYSIS AFTER LVAD

TABLE 2. Preoperative Renal Characteristicsa,b,c Renal characteristic

All patients (N¼354)

No RRT (n¼300)

RRT (n¼54)

P value

BUN, day before (mg/dL) (n¼345)

25 (18, 35)

23 (17, 32)

38 (23, 46)

<.001

Creatinine, day before (mg/dL)

1.3 (1.0, 1.6)

1.2 (1.0, 1.5)

1.6 (1.3, 1.9)

<.001

eGFR (MDRD), day before, mL/min/1.73 m2 Estimated 24-h protein (mg) (n¼340) Measured GFR (mL/min/1.73 m2) (n¼181)

55 (42, 73)

57 (45, 75)

43 (36, 54)

<.001

183 (98, 425)

174 (91, 359)

357 (166, 791)

<.001

49 (34, 67)

51 (38, 70)

33 (22, 52)

.001

BUN ¼ blood urea nitrogen; eGFR ¼ estimated glomerular filtration rate; MDRD ¼ Modification of Diet in Renal Disease equation; RRT ¼ renal replacement therapy. b SI conversion factors: To convert BUN values to mmol/L, multiply by 0.357; to convert creatinine values to mmol/L, multiply by 88.4. c Data are expressed as median (quartile 1, quartile 3). a

cardiopulmonary bypass (CPB) time was created to identify predictors of need for RRT. Cox proportional hazards regression models were fit to examine the associations of RRT with mortality, RV failure, pump thrombosis, thromboembolic events, gastrointestinal bleeding, and LVAD infection. Postoperative RRT was treated as a timedependent covariate in all the models; once a patient started RRT, they remained in the RRT cohort until death or last follow-up. Clinically relevant variables adjusted for in the models included age, sex, diabetes, LVAD indication (DT vs BTT), redo sternotomy, and RAP/PCWP ratio greater than 0.63. All significance tests were 2-tailed and were conducted at the 5% significance level. Analyses were performed using SAS version 9.4 and JMP version 8.0 software (SAS Institute Inc).

RESULTS Patient Characteristics Of 354 patients who underwent continuousflow LVAD implantation during the study period, 54 (15%) required in-hospital RRT postoperatively. Baseline demographic, clinical, and hemodynamic characteristics of the study population are presented in Table 1. Patients who required RRT were younger (mean  SD age, 5714 vs 6112 years; P¼.03); more often female (31% vs 19%; P¼.04); more likely to have a HeartWare device (33% vs 15%; P¼.01), a higher burden of comorbidities (mean  SD Charlson Comorbidity Index, 4.92.2 vs 4.32.1; Mayo Clin Proc. n June 2019;94(6):1003-1014 www.mayoclinicproceedings.org

n

P¼.03), and longer CPB time (mean  SD, 136.257.1 vs 106.949.4 minutes; P<.001); and more likely to be receiving extracorporeal membrane oxygenation support than patients who did not require RRT. Moreover, patients requiring RRT had higher preoperative RAP (mean  SD, 19.17.6 vs 13.46.2 mm Hg; P<.001) and RAP/PCWP ratio than patients not receiving RRT after LVAD implantation. Mean  SD preoperative MELD scores (19.06.8 vs 14.55.5; P<.001) and total bilirubin levels (2.12.6 vs 1.30.8 mg/dL; P<.001) were significantly higher, whereas mean  SD albumin (3.60.5 vs 3.80.6 g/dL; P¼.02) (to convert to g/L, multiply by 10) and hemoglobin (10.61.6 vs 11.91.9 g/dL; P<.001) (to convert to g/L, multiply by 10) levels were significantly lower in patients requiring RRT. Preoperative Renal Characteristics Patients requiring RRT after LVAD had higher pre-LVAD blood urea nitrogen (BUN) (P<.001), creatinine (P<.001), and estimated 24-hour protein (P<.001) levels compared with patients who did not require RRT (Table 2). The median (Q1, Q3) baseline eGFR was significantly lower in patients who required post-LVAD RRT (43 [36, 54] vs 57 [45, 75] mL/min/1.73 m2; P<.001). Of patients who completed the measured GFR study (n¼181, 51%), those receiving RRT had significantly lower GFR than those who did not require RRT (33 [22, 52] vs 51 [38, 70] mL/min/1.73 m2; P¼.001). In the overall cohort, BUN and serum creatinine

https://doi.org/10.1016/j.mayocp.2018.09.021

1007

MAYO CLINIC PROCEEDINGS

60

2.5

2.0

40 1.5 30 1.0 20 0.5

10 0

Creatinine (mg/dL)

BUN (mg/dL)

50

BUN Creatinine 1 wk before

0.0 1d before

1 mo after

3 mo after

6 mo after

FIGURE 1. Change in renal function with time before and during left ventricular assist device (LVAD) support. Creatinine and blood urea nitrogen (BUN) measurements are shown for the whole group from 1 week before to 6 months after LVAD. Data are presented as mean  SD. The number of patients with creatinine assessed at each time point shown are as follows: 1 week before (n¼288), 1 day before (n¼354), 1 month after (n¼326), 3 months after (n¼299), and 6 months after (n¼224) LVAD implantation. SI conversion factors: To convert creatinine values to mmol/L, multiply by 88.4; to convert BUN values to mmol/L, multiply by 0.357.

levels improved 1 month postoperatively (paired P<.001), but this beneficial effect markedly attenuated after 6 months, resulting in average values similar to those observed before LVAD (Figure 1).

Need for RRT After LVAD Implantation The median (Q1, Q3) time from LVAD implantation to RRT was 3 (1, 10) days. Of the 54 patients who required RRT, 44 (81%) were started on continuous RRT and 10 were started on intermittent HD. Eventually, 46 patients (85%) were transitioned or maintained on intermittent HD during their in-hospital stay after LVAD implantation. Forty-five patients (83%) who required RRT had a preoperative eGFR less than 60 mL/min/1.73 m2 (Figure 2A), whereas 37 patients (69%) who required RRT had proteinuria, with an estimated 24-hour urine protein level greater than 200 mg/d (Figure 2B). Of the 105 patients with a GFR less than 45 mL/min/1.73 m2, 30 (29%) required RRT (Figure 2A). Similarly, of the 1008

Mayo Clin Proc.

n

89 patients with a 24-hour protein level of 400 mg/d or more, 25 (28%) required inhospital RRT (Figure 2B). Of the 32 patients with both preoperative eGFR less than 45 mL/min/1.73 m2 and 24-hour urine protein level greater than 400 mg/d, 12 (37.5%) required RRT (Figure 2C), and the presence of both low eGFR (<45 mL/min/1.73 m2) and high urine protein levels (>400 mg/d) was associated with a significantly higher likelihood of RRT (odds ratio [OR], 9.4; 95% CI, 3.7-24.0; P<.001) compared with patients with none or just 1 of these adverse parameters. A thorough review of the 9 patients with normal preoperative eGFR (>60 mL/min/1.173 m2) that required postLVAD RRT revealed that they had complicated intraoperative (ie, concomitant cardiac surgeries, prolonged CPB time) and postoperative (ie, RV failure) courses that may have contributed to the need for RRT. Younger age (P¼.03), higher creatinine level the day before LVAD (P<.001), lower eGFR (P<.001), higher BUN level (P<.001), estimated 24-hour urine protein level greater than 400 mg (P<.001), longer CPB time (P<.001), and lower measured GFR (P¼.006) were associated with the need for RRT on univariate analysis (Table 3). In a multivariable model, lower eGFR (OR per 1-mL/min decrease, 1.05; 95% CI, 1.03-1.07; P<.001), estimated 24hour urine protein level greater than 400 mg (OR, 2.59; 95% CI, 1.22-5.48; P¼.01), higher mean RAP (OR, 1.10; 95% CI, 1.041.16; P<.001), and longer CPB time (OR, 1.01; 95% CI, 1.01-1.02; P<.001) were independent predictors of RRT (Table 4). Outcomes Of the 54 patients requiring in-hospital RRT, 18 (33.3%) had renal recovery and did not require outpatient HD, 18 (33.3%) required outpatient HD, and 18 (33.3%) died before hospital discharge. By comparison, inhospital mortality was 4.7% in those who did not require RRT after LVAD. For patients who had renal recovery before hospital discharge, the median (Q1, Q3) time from starting RRT to renal recovery that no longer required RRT was 16.5 (5.3, 42.3) days. The

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org

DISCUSSION The main findings of this study are as follows: (1) higher preoperative creatinine levels, 24-hour urine protein levels, and mean RAP and longer CPB time were independent predictors of RRT after LVAD; (2) one-third of patients requiring RRT did not survive to hospital discharge, one-third had renal recovery, and one-third required outpatient HD; and (3) RRT was associated with longer in-hospital LOS, higher Mayo Clin Proc. n June 2019;94(6):1003-1014 www.mayoclinicproceedings.org

n

35 24/83

30 20

15/100

15 10

9/149

5 0

>60

>45-60 >30-45 eGFR by MDRD 1 d before LVAD

Patients requiring dialysis (%)

30

≤30

25/89

25 20

12/72 13/93

15 10 5 0

4/87 <100

B

C

6/22

25

A

Patients requiring dialysis (%)

Kaplan-Meier estimated 1-year survival was 47.4% (95% CI, 33.6%-61.2%) for patients who required RRT compared with 84.6% (95% CI, 80.3%-88.7%) for those who did not require RRT. Heart transplant occurred in 20% of patients who did not require RRT and in 13% of RRT patients (P¼.20). The median (Q1, Q3) LOS was significantly longer in patients who required RRT (53.5 [39, 83] vs 20 [15, 29] days; P<.001). For patients discharged with a plan for intermittent outpatient HD, the median (Q1, Q3) LOS was markedly prolonged at 69 (42.5, 118) days. Median (Q1, Q3) follow-up was 24.3 (8.9, 49.6) months for the overall cohort. In total, 170 patients (45%) died during follow-up. The estimated median survival was 51.2 months (95% CI, 39.2-61.2 months). All-cause mortality was increased in patients who required post-LVAD RRT (hazard ratio [HR], 2.7; 95% CI, 1.9-4.0; P<.001) (Figure 3). In patients who required RRT and survived to hospital discharge (n¼36), all-cause mortality was similar in patients with (n¼18) and without (n¼18) renal recovery (HR, 0.71; 95% CI, 0.29-1.75; P¼.45). In multivariable analysis, the need for in-hospital RRT (adjusted HR, 2.86; 95% CI, 1.90-4.33; P<.001), diabetes mellitus (adjusted HR, 1.55; 95% CI, 1.122.14; P¼.008), and DT LVAD indication (adjusted HR, 1.95; 95% CI, 1.26-3.02; P¼.003) were independent predictors of all-cause mortality (Table 5). After adjustment for potential confounders, RRT was also associated with an increased risk of gastrointestinal bleeding (adjusted HR, 4.47; 95% CI, 2.57-7.75; P<.001) (Table 6).

Patients requiring dialysis (%)

HEMODIALYSIS AFTER LVAD

40 35 30 25 20 15 10 5 0

100-<200 200-<400 Estimated 24-h protein before LVAD

≥400

12/32 18/68 13/57

11/183 eGFR>45, protein<400

eGFR>45, protein≥400

eGFR≤45, protein<400

eGFR≤45, protein≥400

FIGURE 2. Proportion of patients requiring in-hospital renal replacement therapy (RRT) according to estimated glomerular filtration rate (eGFR) and 24-hour urine protein levels before left ventricular assist device (LVAD) implantation. A, Patients stratified by eGFR (mL/min/1.73 m2). In-hospital RRT was required in 28.3%, 15%, and 6% of patients with eGFR <45, 46 to 60, and >60 mL/min/1.73 m2, respectively (P<.001). B, Patients stratified by estimated 24-hour urine protein level (milligrams). There were significantly more patients requiring RRT in association with more pronounced proteinuria, particularly among those with estimated 24-hour urine protein levels greater than 400 mg (P<.001 for differences among the 4 protein level groups). C, Patients stratified by both eGFR (using a cutoff value of 45 mL/ min/1.73 m2) and 24-hour urine protein levels (using a cutoff value of 400 mg). Of patients with low GFR and high urine protein levels, 37.5% required in-hospital dialysis compared with only 6% of patients without these adverse renal biomarkers. MDRD ¼ Modification of Diet in Renal Disease.

mortality rates, and an increased risk of major gastrointestinal bleeding. Acute kidney injury after LVAD implantation is common, with a reported incidence

https://doi.org/10.1016/j.mayocp.2018.09.021

1009

MAYO CLINIC PROCEEDINGS

TABLE 3. Univariable Predictors of In-hospital Dialysis After Left Ventricular Assist Device Implantationa,b Predictor

Odds ratio (95% CI)

P value

Age (per 1-y increase)

0.98 (0.95-1.00)

.03

Creatinine, day before (per 1-mg/dL increase)

4.02 (2.26-7.15)

<.001

eGFR (MDRD), 1 day before (per - mL/min decrease)

1.04 (1.02-1.06)

<.001

BUN, 1 day before (per 1-mg/dL increase)

1.04 (1.02-1.06)

<.001

Elevated 24-h protein (>400 mg)

3.03 (1.65-5.55)

<.001

Measured GFR (per 1-mL/min increase)

0.97 (0.95-0.99)

.006

RAP (per 1emm Hg increase)

1.13 (1.08-1.19)

<.001

CPB time (per 1-min increase)

1.01 (1.00-1.02)

<.001

BUN ¼ blood urea nitrogen; CPB ¼ cardiopulmonary bypass; eGFR ¼ estimated glomerular filtration rate; MDRD ¼ Modification of Diet in Renal Disease equation; RAP ¼ right atrial pressure. b SI conversion factors: To convert creatinine values to mmol/L, multiply by 88.4; to convert BUN values to mmol/L, multiply by 0.357. a

of 15% to 45%.19 In the present study, 15% of all patients required RRT in the hospital after LVAD. Consistent with previous work, we found that worse preoperative renal function20,21 and suboptimal hemodynamics22 were strong predictors of the need for post-LVAD RRT. The present study also revealed that both lower eGFR and the presence of proteinuria are independently associated with increased risk of RRT, and the combination of both identifies a particularly high-risk subgroup. In fact, more than 40% of those with eGFR less than 45 mL/ min/1.73 m2 and estimated 24-hour urine protein level greater than 400 mg required post-LVAD RRT. These data are consistent with those of Topkara et al,20 who found that spot urine protein to creatinine ratio

TABLE 4. Multivariable Predictors of In-hospital Renal Replacement Therapy After Left Ventricular Assist Device Implantation Predictor

Odds ratio (95% CI)

P value

Age (per 1-y increase)

0.98 (0.95-1.01)

.11

eGFR (MDRD), 1 day before (per 1-mL/min decrease)

1.05 (1.03-1.07)

<.001

Elevated 24-h protein (>400 mg)

2.59 (1.22-5.48)

.01

RAP (per 1emm Hg increase)

1.10 (1.04-1.16)

<.001

CPB time (per 1-min increase)

1.01 (1.01-1.02)

<.001

CPB ¼ cardiopulmonary bypass; eGFR ¼ estimated glomerular filtration rate; MDRD ¼ Modification of Diet in Renal Disease equation; RAP ¼ right atrial pressure.

1010

Mayo Clin Proc.

n

and eGFR preoperatively predict postLVAD RRT. Because RV failure is also a common adverse outcome after LVAD and shares risk factors and pathophysiology with acute kidney injury,23,24 it is not surprising that hemodynamics reflective of poor RV function before LVAD (higher RAP, higher RAP/PCWP ratio) were associated with increased risk of post-LVAD RRT. These data underscore the importance of optimization of RV function before LVAD with diuresis, inotropic support, and hemodynamic RV support if required to minimize the risk of post-LVAD acute kidney injury requiring RRT. An additional finding of the present study is that longer CPB time was found to be an independent predictor of post-LVAD RRT. Although previous studies have shown that patients with longer CPB duration had significantly more pronounced renal failure requiring RRT,25,26 no studies have shown this association in the LVAD population. It is plausible to assume that similar mechanisms, including the lower perfusion pressures, reduced oxygen tension, oxidative stress, and activation of inflammatory cascades may all contribute to alterations in renal function, particularly in patients with significant underlying kidney injury. These findings are suggestive of a deleterious effect caused by longer CPB time and underscore the need for preoperative optimization of renal function for cases in which longer CPB duration is anticipated, such as in patients requiring concomitant valvular or coronary artery bypass grafting surgery. Among those who required RRT after LVAD, onethird died before discharge, one-third had recovery of renal function such that they no longer required RRT, and one-third continued RRT in the outpatient setting. Although becoming increasingly common, it can be challenging to identify a local dialysis center that is willing to accept a patient with an LVAD. As such, the need for outpatient dialysis can represent a substantial barrier to hospital discharge. Dialysis centers unfamiliar with LVAD pumps may be concerned about the safety of HD with an LVAD and the need to manage the LVAD

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org

HEMODIALYSIS AFTER LVAD

Mayo Clin Proc. n June 2019;94(6):1003-1014 www.mayoclinicproceedings.org

n

100 Dialysis Nondialysis

90 80

log-rank test P<.001

70 Survival (%)

in the case of alarms or emergencies. However, limited available data suggest that HD is safe with an LVAD. One analysis of 281 outpatient HD sessions in 10 LVAD recipients showed a low interruption rate of HD of approximately 5%, very low incidence of low-flow alarms, and no serious adverse events or deaths.27 Intradialytic hypotension was the main reason to terminate the HD session, and some of the hypotensive episodes were precipitated by catheter-related sepsis. Our experience also supports that outpatient HD is safe and feasible in adequately trained HD centers. Although previous concerns have been raised regarding maturation of arteriovenous fistulas in the setting of continuous flow, recent reports suggest normal maturation without hemodynamic or infectious complications.28,29 Peritoneal dialysis may be an option for LVAD recipients who require long-term outpatient RRT. It offers the advantage of continuous filtration, with lower risk of systemic infection. Reports have suggested feasibility of peritoneal dialysis in LVAD recipients.30 However, risk of driveline infection, hyperglycemia, and peritoneal albumin loss leading to malnutrition may limit the use of this dialysis modality in this population. Patients who require RRT after LVAD implantation have substantially increased morbidity and mortality. A recent Medicare analysis demonstrated poor survival among LVAD recipients with ESRD at the time of implantation as most patients survived for less than 3 weeks.12 Based on these and other clinical data, ESRD requiring permanent dialysis before LVAD is considered a contraindication to proceeding as DT.13 However, even among individuals who are not receiving RRT before LVAD, the need for post-LVAD RRT is associated with increased mortality.21,31 In the present study, after adjusting for potential confounders, patients requiring post-LVAD RRT had a nearly 3-fold increased risk of long-term mortality. Interestingly, recovery of renal function had no effect on prognosis in patients requiring RRT who survived to hospital discharge. These data suggest that

60 50 40 30 20 10 0 0

No. at risk Dialysis Nondialysis

10

20

30 40 50 60 70 80 90 100 110 120 Time since LVAD implantation (mo)

54 24 18 12 8 7 300 231 179 137 104 78

3 57

3 46

2 36

1 24

0 10

2

FIGURE 3. Kaplan-Meier estimates of survival after left ventricular assist device (LVAD) implantation in patients requiring in-hospital dialysis compared with patients not requiring in-hospital dialysis (P<.001 by logrank test).

patients who require post-LVAD RRT remain at high risk for adverse outcomes even if their renal function recovers enough to no longer require RRT. We also found that RRT was associated with an increase in the risk of gastrointestinal bleeding. Gastrointestinal bleeding remains a major source of morbidity after LVAD and affects approximately 20% of

TABLE 5. Predictors of All-cause Mortality After LVAD Implantation Hazard ratio (95% CI)

P value

Age (per 1-y increase)

1.02 (1.00-1.03)

.05

Female sex

0.98 (0.65-1.49)

.92

Predictor

RRT (time-dependent)

2.86 (1.90-4.33)

<.001

Diabetes mellitus

1.55 (1.12-2.14)

.008

Redo sternotomy

1.32 (0.94-1.86)

.11

DT LVAD

1.95 (1.26-3.02)

.003

RAP/PCWP ratio >0.63

1.28 (0.92-1.78)

.15

DT ¼ destination therapy; LVAD ¼ left ventricular assist device; PCWP ¼ pulmonary capillary wedge pressure; RAP ¼ right atrial pressure; RRT ¼ renal replacement therapy.

https://doi.org/10.1016/j.mayocp.2018.09.021

1011

MAYO CLINIC PROCEEDINGS

TABLE 6. Associations of Time-Dependent RRT With Post-LVAD Adverse Outcomesa Outcome

HR (95% CI) for time-dependent RRT

P value

Device infection Unadjusted Adjustedb

1.17 (0.47-2.94) 1.16 (0.45-2.96)

.73 .76

Stroke/transient ischemic attack Unadjusted 0.73 (0.18-3.08) 0.68 (0.16-2.97) Adjustedb

.67 .61

Pump thrombosis Unadjusted Adjustedb

.46 .49

1.34 (0.61-2.93) 1.33 (0.60-2.94)

Gastrointestinal bleeding Unadjusted 3.26 (1.95-5.46) 4.47 (2.57-7.75) Adjustedb RV failure Unadjusted Adjustedb

<.001 <.001

2.54 (1.05-6.12) 1.88 (0.76-4.67)

.04 .18

HR ¼ hazard ratio; LVAD ¼ left ventricular assist device; RRT ¼ renal replacement therapy; RV ¼ right ventricular. b Adjusted for age, sex, LVAD type, and LVAD as destination therapy. a

patients.32 An RV failure results in hepatic congestion, coagulopathy, decreased pulsatility, and elevated RAP, which, in turn, increases venous pressure in the mesenteric circulation and leads to elevated shear stress and higher risk of bleeding from existing angiodysplasias.33 After LVAD implantation, decreased renal venous congestion, augmented cardiac output, reduced sympathetic tone, and downregulation of the renin-angiotensin-aldosterone system34,35 all may contribute to improvement of renal function in some of these patients, as suggested in the present study. However, patients with preexisting renal disease with lower creatinine clearance and proteinuria may be less likely to experience improvement in renal function despite hemodynamic LVAD support. There are important clinical implications of these data. Given the high incidence of postoperative RRT and predictive value of baseline renal function, all potential LVAD candidates should undergo a comprehensive renal function evaluation to include an assessment of creatinine clearance and 1012

Mayo Clin Proc.

n

screening for proteinuria. In patients with advanced renal dysfunction that does not improve preoperatively despite aggressive hemodynamic optimization (diuresis, inotropic, and temporary hemodynamic support), LVAD implantation may not be a lifeprolonging and quality-of-lifeeimproving strategy. As such, baseline renal function should be considered in interdisciplinary discussions of LVAD candidacy. Furthermore, the risk of post-LVAD RRT and prognostic implications should be discussed with patients with significant renal dysfunction and their families. This is of particular importance in those considering LVAD as DT because concomitant renal transplant may be a potential option in patients with an LVAD as BTT who require RRT. This study is limited by its retrospective design and tendency toward selection bias. In addition, it represents the experience of a single center, where most patients underwent implantation of a HeartMate II LVAD as DT, thus results might not be generalized to all LVAD centers and populations. Despite these limitations, this study has several important strengths. We report findings in a large number of patients, including a high percentage implanted with LVAD as DT. Furthermore, the presence of a detailed renal function assessment before LVAD implantation allowed us to examine various renal function parameters and their associations with the risks of in-hospital RRT and adverse clinical outcomes. CONCLUSION According to these findings, 15% of LVAD recipients required in-hospital RRT, which was associated with prolonged LOS, increased risk of all-cause mortality, and gastrointestinal bleeding after LVAD implantation. One-third of patients who required RRT died before hospital discharge, onethird required outpatient HD, and only one-third had renal recovery. Preoperative measures suggesting advanced renal dysfunction (eGFR <40 mL/min/1.73 m2 and 24-hour urine protein level >400 mg), elevated RAP, and longer CPB time were independent predictors of RRT. A detailed

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org

HEMODIALYSIS AFTER LVAD

renal function assessment preoperatively using a multibiomarker approach is essential in risk stratification and patient selection for LVAD. Further research should focus on identifying effective ways to optimize hemodynamics and renal function before LVAD to modify post-LVAD outcomes. Abbreviations and Acronyms: BMI = body mass index; BTT = bridge to transplant; BUN = blood urea nitrogen; CPB = cardiopulmonary bypass; DT = destination therapy; ECMO = extracorporeal membrane oxygenation; eGFR = estimated glomerular filtration rate; ESRD = end-stage renal disease; HD = hemodialysis; HF = heart failure; HR = hazard ratio; IABP = intra-aortic balloon pump; ICD = implantable cardiac defibrillator; ICM = ischemic cardiomyopathy; INTERMACS = Interagency Registry for Mechanically Assisted Circulatory Support; LOS = length of stay; LVAD = left ventricular assist device; LVEDD = left ventricular enddiastolic diameter; LVEF = left ventricular ejection fraction; MAP = mean arterial pressure; MDRD = Modification of Diet in Renal Disease equation; MELD = Model for EndStage Liver Disease; mPAP = mean pulmonary arterial pressure; NTBNP = NT pro-Betype natriuretic peptide; OR = odds ratio; PCWP = pulmonary capillary wedge pressure; PVR = pulmonary vascular resistance; RAP = right atrial pressure; RRT = renal replacement therapy; RV = right ventricular; WU = Wood units Affiliations (Continued from the first page of this article.): Cardiovascular Surgery (R.A., S.S., J.M.S., N.L.P., S.S.K., S.M., S.M.D.), and Department of Health Sciences Research (J.M.K., S.M.D.), Mayo Clinic, Rochester, MN; and Division of Cardiovascular Diseases, University of Iowa Hospitals and Clinics, Iowa City (A.B.). Grant Support: The work was supported by grants K23 HL116643 and R03 HL135225 (S.M.D.) from the National Institutes of Health. Potential Competing Interests: The authors report no competing interests. Correspondence: Address to Shannon M. Dunlay, MD, MS, Department of Cardiovascular Diseases, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (dunlay.shannon@ mayo.edu).

REFERENCES 1. Kirklin JK, Pagani FD, Kormos RL, et al. Eighth annual INTERMACS report: special focus on framing the impact of adverse events. J Heart Lung Transplant. 2017;36(10):1080-1086. 2. Miller LW, Pagani FD, Russell SD, et al; HeartMate II Clinical Investigators. Use of a continuous-flow device in patients awaiting heart transplantation. N Engl J Med. 2007;357(9): 885-896. 3. Aaronson KD, Slaughter MS, Miller LW, et al; HeartWare Ventricular Assist Device (HVAD) Bridge to Transplant ADVANCE Trial Investigators. Use of an intrapericardial, continuous-flow, centrifugal pump in patients awaiting heart transplantation. Circulation. 2012;125(25):3191-3200.

Mayo Clin Proc. n June 2019;94(6):1003-1014 www.mayoclinicproceedings.org

n

4. Estep JD, Starling RC, Horstmanshof DA, et al; ROADMAP Study Investigators. Risk assessment and comparative effectiveness of left ventricular assist device and medical management in ambulatory heart failure patients: results from the ROADMAP study. J Am Coll Cardiol. 2015;66(16):1747-1761. 5. Shah KB, Starling RC, Rogers JG, et al; ROADMAP Investigators. Left ventricular assist devices versus medical management in ambulatory heart failure patients: an analysis of INTERMACS Profiles 4 and 5 to 7 from the ROADMAP study. J Heart Lung Transplant. 2018;37(6):706-714. 6. Haft J, Armstrong W, Dyke DB, et al. Hemodynamic and exercise performance with pulsatile and continuous-flow left ventricular assist devices. Circulation. 2007;116(11 suppl):I8-I15. 7. Hasin T, Topilsky Y, Schirger JA, et al. Changes in renal function after implantation of continuous-flow left ventricular assist devices. J Am Coll Cardiol. 2012;59(1):26-36. 8. Hasin T, Grupper A, Dillon JJ, et al. Early gains in renal function following implantation of HeartMate II left ventricular assist devices may not persist to one year. ASAIO J. 2017;63(4):401-407. 9. Sandner SE, Zimpfer D, Zrunek P, et al. Renal function and outcome after continuous flow left ventricular assist device implantation. Ann Thorac Surg. 2009;87(4):1072-1078. 10. Cowger J, Sundareswaran K, Rogers JG, et al. Predicting survival in patients receiving continuous flow left ventricular assist devices: the HeartMate II risk score. J Am Coll Cardiol. 2013; 61(3):313-321. 11. Kaltenmaier B, Pommer W, Kaufmann F, Hennig E, Molzahn M, Hetzer R. Outcome of patients with ventricular assist devices and acute renal failure requiring renal replacement therapy. ASAIO J. 2000;46(3):330-333. 12. Bansal N, Hailpern SM, Katz R, et al. Outcomes associated with left ventricular assist devices among recipients with and without end-stage renal disease. JAMA Intern Med. 2018;178(2):204-209. 13. Feldman D, Pamboukian SV, Teuteberg JJ, et al; International Society for Heart and Lung Transplantation. The 2013 International Society for Heart and Lung Transplantation Guidelines for mechanical circulatory support: executive summary. J Heart Lung Transplant. 2013;32(2):157-187. 14. Kirklin JK, Naftel DC, Kormos RL, et al. Quantifying the effect of cardiorenal syndrome on mortality after left ventricular assist device implant. J Heart Lung Transplant. 2013;32(12): 1205-1213. 15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5): 373-383. 16. Boone MD, Celi LA, Ho BG, et al. Model for end-stage liver disease score predicts mortality in critically ill cirrhotic patients. J Crit Care. 2014;29(5):881.e7-881.e13. 17. Levey AS, Coresh J, Greene T, et al; Chronic Kidney Disease Epidemiology Collaboration. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006;145(4):247-254. 18. Apperloo AJ, De Zeeuw D, Donker AJ, De Jong PE. Precision of glomerular filtration rate determinations for long-term slope calculations is improved by simultaneous infusion of 125I-iothalamate and 131I-hippuran. J Am Soc Nephrol. 1996;7(4):567-572. 19. Ross DW, Stevens GR, Wanchoo R, et al. Left ventricular assist devices and the kidney. Clin J Am Soc Nephrol. 2018;13(2):348-355. 20. Topkara VK, Coromilas EJ, Garan AR, et al. Preoperative proteinuria and reduced glomerular filtration rate predicts renal replacement therapy in patients supported with continuousflow left ventricular assist devices. Circ Heart Fail. 2016;9(12): e002897. 21. Raichlin E, Baibhav B, Lowes BD, et al. Outcomes in patients with severe preexisting renal dysfunction after continuousflow left ventricular assist device implantation. ASAIO J. 2016; 62(3):261-267.

https://doi.org/10.1016/j.mayocp.2018.09.021

1013

MAYO CLINIC PROCEEDINGS

22. Topkara VK, Dang NC, Barili F, et al. Predictors and outcomes of continuous veno-venous hemodialysis use after implantation of a left ventricular assist device. J Heart Lung Transplant. 2006; 25(4):404-408. 23. Bellavia D, Iacovoni A, Scardulla C, et al. Prediction of right ventricular failure after ventricular assist device implant: systematic review and meta-analysis of observational studies. Eur J Heart Fail. 2017;19(7):926-946. 24. Borgi J, Tsiouris A, Hodari A, et al. Significance of postoperative acute renal failure after continuous-flow left ventricular assist device implantation. Ann Thorac Surg. 2013;95(1):163-169. 25. Boldt J, Brenner T, Lehmann A, et al. Is kidney function altered by the duration of cardiopulmonary bypass? Ann Thorac Surg. 2003;75(3):906-912. 26. Del Duca D, Iqbal S, Rahme E, Goldberg P, de Varennes B. Renal failure after cardiac surgery: timing of cardiac catheterization and other perioperative risk factors. Ann Thorac Surg. 2007; 84(4):1264-1271. 27. Quader MA, Kumar D, Shah KB, Fatani YI, Katlaps G, Kasirajan V. Safety analysis of intermittent hemodialysis in patients with continuous flow left ventricular assist devices. Hemodial Int. 2014;18(1):205-209. 28. Calenda BW, Smietana J, Casagrande L. Long-term hemodialysis via arteriovenous fistula in patients with continuous-flow left ventricular assist devices. Artif Organs. 2016;40(7):712.

1014

Mayo Clin Proc.

n

29. Schaefers JF, Ertmer C. Native arteriovenous fistula placement in three patients after implantation of a left ventricular assist device with non-pulsatile blood flow. Hemodial Int. 2017;21(3): E54-E57. 30. Guglielmi AA, Guglielmi KE, Bhat G, Siemeck R, Tatooles AJ. Peritoneal dialysis after left ventricular assist device placement. ASAIO J. 2014;60(1):127-128. 31. Muslem R, Caliskan K, Akin S, et al. Pre-operative proteinuria in left ventricular assist devices and clinical outcome. J Heart Lung Transplant. 2018;37(1):124-130. 32. Demirozu ZT, Radovancevic R, Hochman LF, et al. Arteriovenous malformation and gastrointestinal bleeding in patients with the HeartMate II left ventricular assist device. J Heart Lung Transplant. 2011;30(8):849-853. 33. Sparrow CT, Nassif ME, Raymer DS, et al. Pre-operative right ventricular dysfunction is associated with gastrointestinal bleeding in patients supported with continuous-flow left ventricular assist devices. JACC Heart Fail. 2015;3(12):956-964. 34. James KB, McCarthy PM, Thomas JD, et al. Effect of the implantable left ventricular assist device on neuroendocrine activation in heart failure. Circulation. 1995;92(9 suppl):II191-II195. 35. James KB, McCarthy PM, Jaalouk S, et al. Plasma volume and its regulatory factors in congestive heart failure after implantation of long-term left ventricular assist devices. Circulation. 1996; 93(8):1515-1519.

June 2019;94(6):1003-1014

n

https://doi.org/10.1016/j.mayocp.2018.09.021 www.mayoclinicproceedings.org