Left ventricular assist devices (LVADs) acutely decrease left ventricular wall stress. Thus, early postoperative levels of N-terminal pro–brain natriuretic peptide (NT-proBNP) should decrease. This study investigated postoperative changes in NT-proBNP levels, the parameters related to changes, and the possible association with complications by performing a retrospective analysis of changes in daily NT-proBNP (pg/ml) levels from admission to discharge both before and after LVAD implantation in a tertiary referral center. For 72 patients implanted with HeartMate II LVADs, baseline NT-proBNP levels were elevated at 3,943 ng/ml (interquartile range 1,956 to 12,964). Preoperative stabilization led to marked decreases in NT-proBNP. Levels peaked 3 days after surgery and subsequently decreased. Patients with complicated postoperative courses had higher early postoperative elevations. By discharge, NT-proBNP decreased markedly but was still 2.83 (1.60 to 5.76) times the age-based upper limit of normal. The 26% reduction in NT-proBNP between admission and discharge was due mostly to the preoperative reductions and not those induced by the LVAD itself. The decrease was not associated with decreases in LV volume. In conclusion, preoperative treatment reduces NT-proBNP values. The magnitude of early postoperative changes is related to the clinical course. Levels at discharge remain markedly elevated and similar to values after preoperative stabilization despite presumptive acute LV unloading.
Highlights
- •
Preoperative treatment in advanced heart failure reduces N-terminal pro–brain natriuretic peptide (NT-proBNP) values.
- •
Magnitude of early postoperative NT-proBNP changes is related to clinical course.
- •
Discharge NT-proBNP values after left ventricular assist device implantation remain elevated despite acute left ventricular unloading.
Left ventricular assist devices (LVADs) are increasingly implanted in patients with advanced heart failure. LVAD support acutely reduces left ventricular diastolic pressure and improves cardiac output. Because LVADs directly drain the left ventricle (LV), they should decrease LV dimensions and stretch and thus natriuretic peptide (NP) levels. However, the chronically maladapted NP system may not respond as anticipated. Adverse postoperative processes such as right ventricular (RV) dysfunction (a possible Achilles heel of LVADs), volume overload, inflammation, and renal dysfunction might influence the NP response. Postoperative NP values thus may be associated with clinical course and outcomes. The time course of changes in NP, their extent, and clinical correlates after LVAD have not been well characterized. The present study sought to evaluate changes in NT-proBNP before and after LVAD implantation, to evaluate the clinical implications of values, and to investigate possible mechanistic determinants associated with NT-proBNP changes.
Methods
Patients implanted with a HeartMate II axial flow LVAD (Thoratec, Pleasanton, California) from February 2007 to May 2011 in whom NT-proBNP levels were available were evaluated. Samples were obtained before operation and frequently thereafter based on clinical judgment. Patients with at least 5 postoperative NT-proBNP values, at least one of which within 5 days of surgery, were included. Ten patients with restrictive cardiomyopathy were excluded because the correlation between LV dimensions and filling pressures in these patients was deemed less predictable, leaving 72 patients for analysis.
Postoperative hospital courses were categorized as “complicated” or “uncomplicated” as previously published. “Uncomplicated” patients were in the intensive care unit for ≤5 days without readmission, on ventilator support ≤2 days with a total hospital stay ≤14 days. For patients who died in the hospital, duration of intensive care unit stay and ventilator support were not evaluated. We compared NT-proBNP trends in this “uncomplicated” group with those with a more “complicated” course. The study was approved by the institutional review board.
NT-proBNP was measured by Electrochemiluminescence Immunoassay from Roche Diagnostics on the Elecsys analyzer (both from Indianapolis, Indiana).
The “baseline” NT-proBNP value was the measurement closest to admission. NT-proBNP was measured the day before surgery (preoperative) after admission for stabilization, which included inotropic therapy and an intra-aortic balloon pump. Postoperative NT-proBNP “peak” values were defined as the maximal postoperative level within 5 days of surgery. NT-proBNP values were obtained daily until discharge. After discharge, NT-proBNP levels were measured during visits at 1 to 4 months (mean 3 months).
The NT-proBNP response to LVAD implantation was calculated as the difference between values at baseline and discharge. An NT-proBNP “responder” was defined as a patient with NT-proBNP <1,000 pg/ml at discharge or a decrease from admission of >80%. Covariates for differences in NT-proBNP between baseline and discharge were probed. Regression models were used to calculate trends before and after implantation. We created “adjusted” NT-proBNP variables at baseline, preoperatively, and at discharge by regressing the log values (to meet distributional assumptions) based on age, gender, and estimated glomerular filtration rate (eGFR) values from the same time as the NT-proBNP values after exponentiating the residuals to obtain numbers on the original scale. If a patient was on dialysis and the eGFR was missing, a value of 10 ml/min/m 2 was assigned. The upper limits of normal associated with heart failure were NT-proBNP of 450 pg/ml for patients aged <50 years, 900 pg/ml for those 50 to 75 years and 1,800 pg/ml for patients >75 years.
Two-dimensional transthoracic echocardiography was performed in a standard manner before implantation, after as clinically indicated, before discharge, and during follow-up. RV function, tricuspid regurgitation, and mitral regurgitation severity were qualitatively graded using a 4-point scale (normal, mild, moderate, or severe). RV function was assessed using tissue Doppler assessment of lateral tricuspid annular motion, systolic tricuspid regurgitation duration, and the RV index of myocardial performance and was characterized as normal, mildly, moderately, or severely dysfunctional.
Variables are described using plots, medians, and interquartile ranges (twenty-fifth and seventy-fifth percentiles) for continuous variables and counts and percentages for categorical variables. Serial values were examined using trajectory plots and summaries of follow-up times and described using spline smoothing for correlated data, assuming an autoregressive moving average correlation structure. Comparisons were performed using Wilcoxon rank sum tests for continuous variables and Fisher’s exact tests (for small sample sizes) for categorical variables. Paired comparisons were performed with Wilcoxon signed rank tests. Because the study was exploratory, statistical correction for multiple comparisons was not done. Logistic regression was used to predict complicated versus uncomplicated status. Distributions of NT-proBNP values varied widely and contained outliers. Thus, models used ranks of the values to stabilize the distributions. Stepwise selection of predictors was used to select variables for regression models. For each model, we calculated odds ratios, profile likelihood confidence intervals, and likelihood ratio test p values. Spearman’s rank correlations and associated p values were used to describe associations between changes in left ventricular end-diastolic diameter (LVEDD) and NT-proBNP. Linear regression models were used to predict NT-proBNP at discharge and changes in values. Continuous NT-proBNP values were log transformed to meet assumptions for linear regression. Regression model residuals indicated that analyses had appropriate distributions. Stepwise selection of predictors was used to select variables for regression models. Variables were included if p <0.05 in the univariate model with a minimum of missing data (n ≥80%). For each model, regression coefficients (β), their associated standard errors, and p values from F tests were calculated. Analyses were conducted using SAS, version 9.2 (SAS Institute, Cary, NC), and R, version 2.14.0 (R Foundation for Statistical Computing, Vienna, Austria). Spline smoothing was run with the gss package in R.
Results
The patient characteristics are listed in Table 1 . Median INTERMACS score was 4 (twenty-fifth to seventy-fifth percentile 2 to 4). After LVAD implantation, 4 patients died, 21 were discharged home, and 47 to rehabilitation. Including the 4 patients who died, only 2 had follow-up times <20 days and 4 had <70 days. The operative and postoperative characteristics are listed in Table 2 .
| Variable | All (n = 72) | Uncomplicated (n = 20) | Complicated (n = 52) | p-Value |
|---|---|---|---|---|
| Age (years) | 63 (53; 69) | 63.5 (50; 68) | 62.5 (53; 70) | 0.767 |
| Men | 63 (87%) | 16 (80%) | 47 (90%) | 0.252 |
| Hypertension | 24 (33%) | 8 (40%) | 16 (31%) | 0.578 |
| Diabetes mellitus | 25 (35%) | 5 (25%) | 20 (38%) | 0.408 |
| Chronic kidney disease | 44 (61%) | 10 (50%) | 34 (65%) | 0.284 |
| Bridge to transplant | 21 (29%) | 9 (45%) | 12 (23%) | 0.086 |
| Ischemic etiology | 36 (50%) | 11 (55%) | 25 (48%) | 0.793 |
| Redo sternotomy | 37 (51%) | 9 (45%) | 28 (54%) | 0.602 |
| Weight (kg) | 87.9 (76.0; 101.0) | 93 (75.3; 101.5) | 87.4 (76.2; 101.3) | 0.995 |
| Systolic blood pressure (mm Hg) | 98 (90; 110) | 100 (92; 110) | 96 (90; 110) | 0.224 |
| NYHA class 4 | 56 (78%) | 13 (65%) | 43 (83%) | 0.123 |
| IABP pre-op | 33 (46%) | 7 (35%) | 26 (50%) | 0.299 |
| Need for inotropes pre-surgery | 46 (65%) n = 71 | 13 (65%) | 33 (65%) n = 51 | 1.000 |
| Angiotensin inhibitor | 47 (65%) | 17 (85%) | 30 (58%) | 0.051 |
| Beta receptor blocker | 60 (83%) | 18 (90%) | 42 (81%) | 0.490 |
| Aldosterone inhibitor | 33 (46%) | 7 (35%) | 26 (50%) | 0.299 |
| Hemoglobin (gm%) | 11.4 (10.2; 13.0) | 11.8 (10.6; 13.2) | 11.3 (10.1; 12.9) | 0.525 |
| Platelets (×1000) | 153.0 (123.5; 199.5) | 160.5 (132.3; 239.0) | 148.5 (117.3; 185.8) | 0.195 |
| GFR, admission (ml/min/1.72 m 2 ) | 56.8 (41.4; 66.3) | 64.6 (54.6; 76.1) | 47.7 (38.8; 63.3) | 0.003 |
| GFR pre-implant (ml/min/1.72 m 2 ) | 62.2 (47.0; 87.1) | 70.6 (54.9; 90.3) | 57.4 (43.3; 86.0) | 0.150 |
| Bilirubin total (mg%) | 1.0 (0.8; 1.7) n = 71 | 0.9 (0.7; 1.6) | 1.1 (0.9; 1.7) n = 51 | 0.144 |
| AST (u/L) | 32 (25; 49) | 28.5 (22.3; 37.3) | 37.0 (27.3; 51.8) | 0.030 |
| Albumin (grm%) | 3.8 (3.4; 4.2) n = 68 | 4.1 (3.6; 4.4) | 3.8 (3.3; 4.1) n = 48 | 0.062 |
| RV dysfunction >moderate ∗ | 30 (42%) n = 71 | 8 (40%) | 22 (43%) n = 51 | 1.000 |
| LVEDD ∗ (mm) | 70 (64; 76) n = 69 | 74 (67; 80) | 67 (62; 75) n = 49 | 0.016 |
| Ejection fraction ∗ (%) | 17 (15; 22) n = 45 | 15 (15; 20) n = 17 | 19.5 (15; 23) n = 28 | 0.248 |
| RA mean pressure (mm Hg) | 15 (11; 21) n = 64 | 15 (10; 18) | 17 (11; 24) n = 44 | 0.128 |
| PA mean pressure (mm Hg) | 37 (31; 43) n = 64 | 38 (32; 45) | 36 (30; 42) n = 44 | 0.400 |
| Pulmonary capillary wedge (mm Hg) | 24 (18; 28) n = 63 | 24 (19; 35) | 24 (18; 27) n = 43 | 0.535 |
| RVSWI (grm/m 2 /beat) | 7.4 (4.5; 9.3) n = 64 | 8.4 (7.2; 9.9) | 6.2 (4.3; 9.2) n = 44 | 0.064 |
| Cardiac index (L/min/m 2 ) | 1.9 (1.6; 2.4) n = 64 | 1.9 (1.5; 2.5) | 1.9 (1.6; 2.4) n = 44 | 0.811 |
| Variable | All (n = 72) | Uncomplicated (n = 20) | Complicated (n = 52) | p-Value |
|---|---|---|---|---|
| Bypass time (min) | 114 (84; 168) n = 71 | 90 (66; 126) n = 19 | 124 (97; 179) | 0.002 |
| Days in the ICU | 1 (1; 2) n = 69 | 1 (1; 1) | 2 (1; 2.5) n = 49 | <0.0001 |
| Days on ventilation | 1 (1; 2) n = 69 | 1 (1; 1) | 2 (1; 2.5) n = 49 | <0.0001 |
| Duration of inotropic support (Hr) | 82 (72; 156) | 49 (31; 93) | 120 (72; 234) | 0.0005 |
| Postoperative dialysis | 10 (14%) | 0 (0%) | 10 (19%) | 0.053 |
| RA mean pressure (mm Hg) post-operative ∗ | 13 (11; 15) n = 63 | 12 (9; 15) n = 17 | 14 (12; 16) n = 46 | 0.104 |
| PA systolic pressure (mm Hg) post-operative ∗ | 43 (36; 47) n = 62 | 46 (37; 52) n = 17 | 42 (36; 46) n = 45 | 0.286 |
| RV dysfunction >moderate by post operative echo | 24 (37%) n = 65 | 3 (17%) n = 18 | 21 (45%) n = 47 | 0.046 |
| RA pressure by pre-discharge echo (mm Hg) | 10 (10; 15) n = 49 | 10 (5; 14) n = 13 | 10 (10; 19) n = 36 | 0.066 |
| RV systolic pressure by pre-discharge echo (mm Hg) | 39 (33; 44) n = 42 | 39 (33; 41) n = 12 | 38 (34; 48) n = 30 | 0.468 |
| LVEDD at discharge (mm) | 58 (51; 62) n = 69 | 65 (60; 71) | 55 (49; 59) n = 49 | 0.0001 |
| LVEDD at 1 month (mm) | 56 (51; 65) n = 65 | 61 (55; 70) n = 19 | 55 (50; 60) n = 46 | 0.044 |
| LVEDD decrease baseline to discharge (mm) | 10 (6; 17) n = 68 | 7 (3; 17) | 11 (7; 20) n = 48 | 0.062 |
| LVEDD decrease baseline to one month (mm) | 13 (8; 20) n = 64 | 12 (5; 22) n = 19 | 14 (8; 20) n = 45 | 0.889 |
| Pump speed at discharge (RPM) | 9400 (9200; 9600) n = 61 | 9400 (9200; 9600) n = 17 | 9300 (9200; 9600) n = 44 | 0.751 |
| Pump flow at discharge | 5.3 (4.7; 5.5) n = 60 | 5.2 (4.8; 5.5) n = 18 | 5.3 (4.7; 5.5) n = 42 | 0.703 |
| Pump pulsatility index at discharge | 5.0 (4.6; 5.5) n = 58 | 5.1 (4.4; 5.5) n = 18 | 5.0 (4.6; 5.5) n = 40 | 0.762 |
| GFR, discharge (ml/min/1.72 m 2 ) | 84 (53; 113) n = 70 | 90 (73; 117) | 75 (44; 113) n = 50 | 0.114 |
| Days in the hospital (implant to discharge) | 14 (10; 24) | 9 (8; 12) | 18 (14; 29) | <0.0001 |
∗ Last postoperative measurement recorded with a pulmonary artery catheter.
NT-proBNP trends are shown in Figure 1 and Table 3 . Adjusted values for NT-proBNP by age, gender, and eGFR correlated with unadjusted values (Spearman r = 0.93 at baseline, 0.86 before implantation, and 0.95 at discharge). Thus, unadjusted values are reported. After admission but before implantation, NT-proBNP decreased (p = 0.004, 38 patients). Early postoperative levels increased, peaking at 3 days, and then decreasing. Overall, NT-proBNP decreased from baseline to discharge (p = 0.048) but not from the time of implantation to discharge (p = 0.576). All values at discharge remained greater than the upper limits of normal (median 2.8 times greater, p <0.0001) and were 74% (median) of baseline values. Baseline NT-proBNP values correlated with values at discharge (Spearman r = 0.46, p = 0.0002).
| NT-ProBNP (pg/ml) | All (n = 72) | Uncomplicated (n = 20) | Complicated (n = 52) | p-Value |
|---|---|---|---|---|
| Baseline | 3943 (1956; 12964) n = 62 | 3199 (869; 11803) n = 17 | 4786 (2232; 13790) n = 45 | 0.224 |
| Baseline adjusted ∗ | 1.14 (0.45; 2.32) n = 61 | 1.09 (0.32; 3.30) n = 16 | 1.14 (0.46; 2.16) n = 45 | 0.725 |
| Pre-implant | 3431 (1486; 6034) n = 55 | 3431 (932; 5113) n = 17 | 3446 (1801; 8101) n = 38 | 0.187 |
| Pre-implant adjusted ∗ | 1.05 (0.53; 2.53) n = 54 | 0.98 (0.49; 2.16) n = 16 | 1.11 (0.52; 2.72) n = 38 | 0.410 |
| Baseline to pre-implant | −726 (−5071; 0) n = 38 | −2406 (−10416; 0) n = 9 | −365 (−4585; 0) n = 29 | 0.426 |
| Peak (early post-implant) | 13183 (7122; 22556) | 8147 (4287; 17346) | 13497 (7947; 27419) | 0.023 |
| Time to peak value (days) | 3.0 (2.4; 4.0) | 2.0 (1.25; 2.0) | 3.0 (2.0; 4.0) | 0.002 |
| Early increase post surgery | 7881 (3190; 15464) n = 55 | 4051 (1718; 11917) n = 17 | 9644 (4276; 18775) n = 38 | 0.013 |
| Discharge | 3006 (2022; 5818) n = 67 | 2939 (1487; 4235) | 3252 (2224; 6187) n = 47 | 0.276 |
| Discharge adjusted ∗ | 0.94 (0.61; 1.51) n = 67 | 0.95 (0.53; 1.24) | 0.93 (0.66; 1.6) n = 47 | 0.515 |
| Discharge (times upper normal) | 2.83 (1.60; 5.76) n = 67 | 2.56 (1.33; 4.99) | 3.09 (1.60; 6.30) n = 47 | 0.447 |
| Baseline to discharge | −930 (−7047; 539) n = 58 | −1437 (−4849; 830) n = 17 | −883 (−8208; 639) n = 41 | 0.864 |
| Percent decrease baseline to discharge | −26% (−64; 44) n = 58 | −25% (−59; 114) n = 17 | −28% (−70; 34) n = 41 | 0.585 |
| Follow-up (1-4 months) | 1696 (1051; 3251) n = 64 | 1157 (625; 2139) n = 18 | 2382 (1192; 3383) n = 46 | 0.019 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
