Author + information
- Received May 17, 2011
- Revision received August 29, 2011
- Accepted September 20, 2011
- Published online January 3, 2012.
- Tal Hasin, MD⁎,
- Yan Topilsky, MD⁎,
- John A. Schirger, MD⁎,
- Zhuo Li, MS†,
- Yanjun Zhao, RPh†,
- Barry A. Boilson, MD⁎,
- Alfredo L. Clavell, MD⁎,
- Richard J. Rodeheffer, MD⁎,
- Robert P. Frantz, MD⁎,
- Brooks S. Edwards, MD⁎,
- Naveen L. Pereira, MD⁎,
- Lyle Joyce, MD†,
- Richard Daly, MD†,
- Soon J. Park, MD† and
- Sudhir S. Kushwaha, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Sudhir S. Kushwaha, Mayo Clinic, Department of Cardiovascular Diseases, 200 First Street, SW, Rochester, Minnesota 55905
Objectives The aim of this study was to determine renal outcomes after left ventricular assist device (LVAD) implantation.
Background Renal dysfunction before LVAD placement is frequent, and it is unclear whether it is due to primary renal disease or to poor perfusion.
Methods A retrospective single-center analysis was conducted in 83 consecutive patients implanted with HeartMate II continuous-flow LVADs (Thoratec Corp., Pleasanton, California). Calculated glomerular filtration rate (GFR) was assessed on admission and 1, 3, and 6 months after implantation. To define predictors for improvement in GFR, clinical variables were examined in patients with decreased renal function (GFR <60 ml/min/1.73 m2) before LVAD, surviving and dialysis-free at 1 month (n = 44).
Results GFR significantly increased from admission (53.2 ± 21.4 ml/min/1.73 m2) to 1 month after LVAD implantation (87.4 ± 27.9 ml/min/1.73 m2) (p < 0.0001). Subsequently, at 3 and 6 months, GFR remained significantly (p < 0.0001) above pre-LVAD values. Of the 51 patients with GFRs <60 ml/min/1.73 m2 before LVAD surviving at 1 month, 34 (67%) improved to GFRs >60 ml/min/1.73 m2. Univariate pre-operative predictors for improvement in renal function at 1 month included younger age (p = 0.049), GFR improvement with optimal medical therapy (p < 0.001), intra-aortic balloon pump use (p = 0.004), kidney length above 10 cm (p = 0.023), no treatment with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers (p = 0.029), higher bilirubin (p = 0.002), higher Lietz-Miller score (p = 0.019), and atrial fibrillation (p = 0.007). Multivariate analysis indicated pre-operative improved GFR (slope = 0.5 U per unit improved; 95% confidence interval: 0.2 to 0.8; p = 0.003), atrial fibrillation (slope = 27; 95% confidence interval: 8 to 46; p = 0.006), and intra-aortic balloon pump use (slope = 14; 95% confidence interval: 2 to 26; p = 0.02) as independent predictors.
Conclusions In most patients with end-stage heart failure considered for LVAD implantation, renal dysfunction is reversible and likely related to poor renal perfusion.
Renal dysfunction (RD) is frequent in patients with chronic congestive heart failure (1). Moderate to severe RD, defined by a glomerular filtration rate (GFR) <60 ml/min/1.73 m2, has been documented in up to 45% of patients with ambulatory heart failure (2) and in 64% of patients hospitalized for decompensated heart failure (3). Compared with patients with heart failure without RD, these patients experience higher morbidity and mortality (4). The optimal management of these patients is unclear, because they have been generally excluded from the major trials of congestive heart failure (5–7).
The question of renal function is especially important in those patients with end-stage heart failure who are considered for advanced therapy such as left ventricular assist device (LVAD) placement or heart transplantation. Cardiac transplant recipients have an increased risk for renal impairment after transplantation (8), which is higher than for other forms of cardiac surgery (9). Subsequently, there is continuous risk related to calcineurin-inhibitor immunosuppressive medication, which also results in a progressive decline in renal function. Indeed, the preferred strategy at many cardiac transplantation centers for the treatment of heart transplantation candidates with significant RD is to consider heart-kidney transplantation.
LVADs are increasingly implanted in patients with end-stage heart failure as a bridge to transplantation (BTT) or as destination treatment (DT) and have dramatically improved both survival and morbidity (10,11). The newer axial-flow and centrifugal-flow devices have the advantage of better durability and smaller size and, therefore, have largely replaced older pulsatile devices for both BTT and DT. Therefore, many patients with heart failure with RD are considered for LVAD implantation. However, whether LVAD implantation improves or worsens renal function is unclear and has not been specifically addressed by the major multicenter LVAD trials (12–14).
The presence of pre-existing RD increases the risk after general cardiac surgery (15) as well as after LVAD implantation (16). Renal function may also deteriorate acutely after cardiac surgery, and post-surgical acute RD has been identified in up to 45% of LVAD recipients. Surgical risk factors including prolonged bypass time, increased intra-operative blood loss, post-operative need for venovenous hemodialysis, and reoperation were identified as negative predictors of poor outcomes (17,18). Acute renal failure (ARF) after LVAD implantation has been identified as a negative predictor of outcomes (10,18,19). Another consideration is whether the presence of a low pulsatility circulation, a result of a continuous-flow device, might cause chronic deterioration of renal function. Therefore, severe RD (need for hemodialysis or serum creatinine >2.5 to 3 mg/dl deemed irreversible) is considered a significant comorbidity and a relative contraindication for LVAD implantation (20,21).
However, it is also possible that improved cardiac output and hemodynamic status with reduction in venous pressure may improve renal function. A recent study demonstrated that 65.3% of patients with RD (GFR <60 ml/min/1.73 m2) actually had improved renal function by 1 month after LVAD implantation, to GFR >60 ml/min/1.73 m2. This improvement was associated with improved survival. The absence of diabetes was the only significant predictor identified for renal recovery (16).
Therefore, the aim of this study was to evaluate the effects of LVAD support on renal function in our cohort of BTT and DT patients implanted with continuous-flow HeartMate II devices (Thoratec Corporation, Pleasanton, California) and to identify pre-operative predictors for improved renal function within this population.
A retrospective analysis of data was performed for our single-center experience of LVAD implantation for 103 patients from February 2007 to June 2010. Only HeartMate II continuous-flow devices were followed (n = 83, 80% of adult LVAD implantations), and both BTT and DT patients were included. Twenty patients with other devices (8 with Jarvik 2000 [Jarvik Heart, Inc., New York, New York], 6 with VentrAssist [Ventracor, Sydney, Australia] and 6 with HeartMate XVE [Thoratec Corporation] were excluded from analysis. This study was approved by the institutional review board.
Renal function was assessed by calculating the GFR using the abbreviated Modification of Diet in Renal Disease equation: GFR = 186 × (serum creatinine [mg/dl])−1.154 × (age [years])−0.203 × 0.742 (if female) (22). The equation has previously been validated in the general, heart failure, (23,24) and heart transplantation (8) populations. Previous studies in the LVAD population have used both the Cockcroft-Gault (25) and the Modification of Diet in Renal Disease equations (16).
Renal function was assessed at admission for LVAD implantation (determined as baseline renal function), the morning before LVAD implantation, and 1 month (range 14 to 46 days), 3 months (range 56 to 120 days), and 6 months (range 150 to 240 days) after implantation during routine follow-up visits. Stages of RD were determined according to calculated GFR in accordance with established guidelines. A GFR cutoff of >60 ml/min/1.73 m2 was used to differentiate mild or normal renal function from more severe RD (26). Data were considered missing if a patient died or underwent transplantation (for GFR after death date or transplantation date) and if no creatinine measurement was available. Patients requiring hemodialysis were considered in stage 5 (GFR <15 ml/min/1.73 m2). Because the actual GFRs in these patients were undetermined, they were considered missing for analysis requiring numerical GFR measurement.
Predictors of RD improvement
Possible pre-operative predictors affecting renal function after LVAD implantation for patients with available GFRs at 1 month (n = 72) were derived from patient files and electronic data. Special attention was given to patients with baseline GFRs <60 ml/min/1.73 m2 for whom GFRs were available at 1 month (n = 44), because there is often uncertainty regarding the recoverability of renal function in these patients going into LVAD implantation. Therefore, it would be helpful to know which factors predict whether renal function is likely to recover.
Data were collected into a JMP file and analyzed using SAS version 9.2 (SAS Institute Inc., Cary, North Carolina) for association with renal function independently by a coauthor of this report (Z.L.). Descriptive statistics for categorical variables are reported as frequencies and percentages, while continuous variables are reported as mean ± SD or as median (range) as appropriate. Change of GFR from 1 time point to another was tested using paired t tests or Wilcoxon signed rank tests. The Bonferroni method was used to assess the significance of multiple comparisons. Linear regression models were used to find the univariate and multivariate predictors of change in GFR at 1 month. GFR was evaluated as a continuous variable, as previously described (27). The multivariate model considered pre-operative univariately significant variables (p < 0.05), excluding variables with more than 20% missing values. We used model selection using the stepwise backward and forward selection. All statistical tests were 2-sided, with the alpha level set at 0.05 for statistical significance.
Baseline characteristics for the 83 LVAD patients are shown in Table 1. The mean age was 63 ± 12 years, and the majority of patients were men (82%). The majority of LVADs (70%) were implanted as DT, and about half of the patients had ischemic etiology. The main reason for DT in these patients was older age (median 70 vs. 55 years for patients bridged to transplantation, p < 0.0001) as well as the presence of other comorbidities. About one-half of the patients had chronic RD. Table 1 also shows the baseline characteristics for the subset of 54 LVAD patients with GFRs on admission <60 ml/min/1.73 m2 in whom predictors of improvement in renal function were subsequently evaluated. This cohort had a mean admission GFR of 40 ± 12 ml/min/1.73 m2. Compared with patients with preserved renal function (GFR ≥60 ml/min/1.73 m2), those with low baseline GFRs were significantly older, with more chronic kidney disease and longer pre-operative hospital stays. These patients also had more right ventricular dysfunction and higher wedge pressures.
Renal function after LVAD implantation
GFR over the follow-up time period is shown in Figure 1A. GFR measurements at 1 month were available for 72 patients (87%) and at 6 months were available for 57 patients (69%). At 6 months, 12 patients had died, 6 patients had undergone transplantation, 5 patients had no follow-up creatinine measurements, and 3 patients were alive on chronic dialysis. Overall GFR significantly improved 1 month after LVAD implantation (from 53.2 ± 21.4 to 87.4 ± 27.9 ml/min/1.73 m2, p < 0.0001). Thereafter, GFR partially declined at 3 months in 41 of 66 patients with GFR estimates at both time points (77.6 ± 22.8 ml/min/1.73 m2, p = 0.0001, compared to 1 month). Between 3 and 6 months, GFR further declined in 36 of 55 patients with available GFR estimates (71.2 ± 21.0 ml/min/1.73 m2, p = 0.0032, compared to 3 months). Only 6 patients had continuous GFR improvements without any decline over the study period (67 had some decline or no recovery of RD, and data were missing for 10 patients). Overall, GFR remained significantly higher at 6 months compared to pre-operative GFR (p < 0.0001). Figure 1B shows the changes in GFR over time for the subsets of patients with pre-operative RD (GFR <60 ml/min/1.73 m2) and preserved renal function (GFR ≥60 ml/min/1.73 m2). Overall, both subsets followed the same pattern of change in GFR as the entire cohort. The subset with pre-operative RD had a significant increase at 1 month (p < 0.0001), a partial decline (1 to 3 months, p = 0.0059; 3 to 6 months, p = 0.0258), and overall improvement over 6 months (p < 0.0001). The distribution of the stages of RD before and during follow-up after LVAD implantation is illustrated in Figure 2A. Consistent with the findings in Figure 1, an improved stage distribution was noted after LVAD implantation. Similar trends were noted for the large subset of patients in stage 3 before implantation, as illustrated in Figure 2B. Indeed, for patients with available renal staging at 1 month, 57 patients (72%) improved their RD stages or remained at stage 1. Fourteen patients (18%) remained in their pre-operative renal stages (10 in stage 2, 4 in stage 3), and 8 (10%) deteriorated (2 from stage 4 to 5, 5 from stage 3 to 5, and 1 from stage 2 to 3).
Renal failure after LVAD implantation
Eight patients (10%) developed ARF after LVAD implantation necessitating acute hemodialysis. Of these, 2 died in the early post-operative period, and 2 recovered renal function. Four patients (5%) continued with chronic hemodialysis. The clinical characteristics and post-operative courses of these patients are summarized in Tables 2 to 4.⇓⇓⇓ Predisposing comorbidities included severe RD necessitating acute dialysis before transplantation (Patients #1, #5, and #7), prior renal transplantation with severe RD for Fabry disease (Patient #2), multiple comorbidities (Patients #3 and #6), and insulin-dependent diabetes mellitus (Patients #4 and #8). Most patients (except Patient #8) had low GFRs before LVAD implantation. Patient #3 underwent emergent implantation for previous device (HeartMate XVE) failure. All patients had significant post-operative complications, including right ventricular dysfunction, infections, bleeding, and need for prolonged inotropic support. Renal function recovered in 2 patients. Four patients (Patients #2, #5, #6, and #8) died during the initial hospital stay after prolonged complicated courses. The 2 other patients survived longer (1.5 and 2.5 years) and died of nonrenal causes.
Predictors of improved GFR 1 month after LVAD implantation
We used a univariate model for predicting an increase in GFR (as a continuous variable) for the 72 patients with available GFR measurements at 1 month. Significant positive predictors associated with GFR improvement were the use of an intra-aortic balloon pump (IABP) before surgery (slope = 17, p = 0.003), higher bilirubin (slope = 12.1, p = 0.002) alanine transaminase (slope = 0.03, p = 0.041), Lietz-Miller score (slope = 1.5, p = 0.003), and higher right atrial pressure (slope = 0.97, p = 0.048). An increase in GFR with optimal medical treatment before surgery was associated with further improvement 1 month after surgery (slope = 0.6 per ml/min/1.73 m2, p < 0.001). Higher pump speed at discharge (slope = 5.9 per 200-rpm increase, p = 0.01) was also associated with improved GFR. Negative predictors were having at least 1 kidney smaller than 10 cm on ultrasound (slope = −23.7, p = 0.01) and treatment with an angiotensin pathway inhibitor (angiotensin-converting enzyme inhibitor or angiotensin receptor blocker) before surgery (slope = −15.7, p = 0.006). The multivariate model suggested that having at least 1 kidney smaller than 10 cm (slope = −21; 95% confidence interval [CI]: −37.7 to −4.6; p = 0.012), use of an IABP (slope = 11.8; 95% CI: 0.8 to 22.8; p = 0.035) and Lietz-Miller score (slope = 1.2 per unit increase; 95% CI: 0.25 to 2.11; p = 0.013) were independent predictors.
To determine the predictors of improved GFR after LVAD implantation in patients with significant pre-operative RD, we focused our analysis on patients with baseline GFRs <60 ml/min/1.73 m2 (stages 3 to 5). For this subset of 54 patients, data were available at 1 month for 44 patients, as 3 died and 7 were on hemodialysis. Overall, GFR improved from 40 ± 12 to 80 ± 27 ml/min/1.73 m2, as shown in Figure 1B. For patients with available staging, 34 (67%) had improved GFRs (to >60 ml/min/1.73 m2) at 1 month, and 17 (33%) did not. A univariate model for prediction of increase in GFR (as a continuous variable) is shown in Table 5. Positive predictors associated with GFR improvement included atrial fibrillation at baseline, higher bilirubin and Lietz-Miller score, an increase in GFR (from 40 ± 12 to 55 ± 18 ml/min/1.73 m2) with optimal medical treatment before surgery, the use of an IABP before surgery, higher LVAD pulsatility index at discharge, and pump speed setting in the higher range (>9,200 rpm). Negative predictors included older age, having at least 1 kidney smaller than 10 cm (on ultrasound before operation), and treatment with an angiotensin pathway inhibitor (angiotensin-converting enzyme inhibitor or angiotensin receptor blocker) before surgery. The multivariate model suggested that the increase in GFR with optimal treatment before surgery (slope = 0.5 U per unit increase; 95% CI: 0.2 to 0.8; p = 0.0028), atrial fibrillation at baseline (slope = 27; 95% CI: 7.7 to 46.4; p = 0.0062), and the use of an IABP (slope = 13.9; 95% CI: 1.8 to 26.1; p = 0.0244) were independent predictors.
The issue of RD after LVAD implantation is important for a number of reasons. First, the possible need for chronic dialysis after LVAD implantation may influence the decision about whether to proceed. Second, the question of reversibility of RD is important regarding the need for additional renal transplantation in addition to cardiac transplantation in BTT candidates. Therefore, it would be helpful to be able to predict whether renal function is likely to improve or worsen after axial-flow LVAD implantation.
Our study cohort reflects the current trends in LVAD implantation as recently documented in the third Interagency Registry for Mechanically Assisted Circulatory Support report, with increasing proportions of DT implantations and continuous-flow devices (28). In some respects, however, our cohort may represent a more chronically ill population, as it includes older patients and fewer BTT candidates (30% vs. 40%). Given that continuous-flow pumps currently account for more than 98% of adult primary LVAD implantations in the United States, our choice of pump is in keeping with national trends. The HeartMate II pump is currently the only device approved by the U.S. Food and Drug Administration for the rapidly expanding DT population, and therefore we felt that restricting our analysis to HeartMate II implantations was reasonable.
The major finding of this study is the dramatic improvement in renal function with LVAD support. The small but significant improvement in renal function before operative intervention can be attributed to improved hemodynamic management and consequent better renal perfusion. However, despite this initial improvement, we observed continued improvement after LVAD implantation, suggesting that medical and IABP intervention was not sufficient to restore optimal renal perfusion. Previous studies have focused on the occurrence of ARF early after LVAD implantation, with incidence as high as 32% or 45% (17,18) and 16% in the Thoratec HeartMate II LVAS (Left Ventricular Assist System) for Destination Therapy study (14). In the present study, the development of ARF was less frequent (10%). We agree with previous reports that the development of ARF may result from an unfavorable early post-operative course and early hemodynamic instability (17). The majority of patients in our cohort actually had improved renal function with LVAD support. GFR significantly increased from 53.2 ± 21.4 ml/min/1.73 m2 at baseline to 87.4 ± 27.9 ml/min/1.73 m2 at 1 month and to 71.2 ± 21.0 ml/min/1.73 m2 at 6 months. Among patients with pre-operative RD, 34 (67%) had improved GFRs to above 60 ml/min/1.73 m2 at 1 month. These results are in agreement with previous reports for the pulsatile pump (25) as well as continuous-flow implants in the BTT population (16). Our findings therefore validate previous observations showing improved renal function in the current population of patients with LVAD implants with continuous-flow pumps.
The reversibility of RD in the cardiorenal syndrome is a major issue in patients being considered for heart transplantation. RD is the most common modifiable contraindication to heart transplantation in LVAD patients (28). Indeed, in the face of progressive RD, heart transplantation may be deferred, or heart-kidney instead of heart transplantation alone may be considered. In a recent analysis, a cutoff GFR <33 ml/min/1.73 m2 was suggested for this strategy, with 1-year results better than for heart transplantation alone (29). Another report identified less post-transplantation RD in a small group of patients supported with LVADs as BTT compared with inotropic support (30). Our findings suggest that LVAD support before heart transplantation may help differentiate between reversible and nonreversible kidney disease.
Another interesting observation is that although GFR generally improves with LVAD support, a gradual partial but significant decline was observed at 3 and at 6 months compared with 1 month after the operation. These findings are also consistent with previous reports (16), and 1 explanation may be that the very low muscle mass after surgery may decrease creatinine levels and result in overestimation of GFR at 1 month. In our analysis, low pre-operative albumin as a marker for cachexia showed a trend for predicting higher GFR (lower creatinine). The progressive expected increase in muscle mass at 3 and 6 months may correct for that bias. Resuming treatment with angiotensin pathway inhibitors could influence the late decline in renal function. However, this did not come up as a significant predictor in our analysis (data not shown). Another explanation is that lack of pulsatile flow may result in gradual renal deterioration over time. Previous studies comparing non-pulsatile-flow and pulsatile-flow LVADs were not supportive of this suggestion (31). In the present study, higher pump speed settings were associated with improved 1-month renal function. In contrast, a higher pump pulsatility index was also associated with improved GFR. Further study is needed to evaluate for long-term effects of axial-flow LVAD support on renal function as well as ways to facilitate optimal cardiac output, venous pressures, and arterial pulsatility with a continuous-flow pump.
One of our major goals was to ascertain pre-operative predictors for improved renal function in patients with LVAD support. We focused our analysis on patients with pre-operative RD (grade 3 and below) who had the potential for improved renal function. Interestingly, our univariate analysis identified high bilirubin, atrial fibrillation, and partial reversibility of RD with optimal treatment before operation as well as the use of an IABP as positive predictors of improved GFR. Not surprisingly, older age and small kidney size were negative predictors. We found the negative association of angiotensin inhibitors surprising, although previous reports of kidney dysfunction after heart transplantation suggested similar results (8). The predictors that were significant in the multivariate model were RD partial reversibility, the use of an IABP, and atrial fibrillation. Previous studies have identified only diabetes mellitus as a significant predictor (16). Despite a similar proportion of patients with diabetes in the present study, we did not find this to be a significant predictor. Recently, a new nomenclature for the cardiorenal syndromes was developed, differentiating between primarily cardiac or renal etiology (7). The positive predictors found in our analysis (atrial fibrillation, high right atrial pressure) may represent stronger cardiac-derived cardiorenal syndrome (type 2), while the negative predictors (small kidney, older age) represent less primary renal (type 4) cardiorenal syndrome.
The use of LVADs has improved our ability to understand the physiology of complex multiple-organ syndromes involving the heart, because they acutely improve cardiac output and hemodynamic status. We observed that in our population of patients with advanced heart failure, prompt assistance of left ventricular function resulted in reversibility of established cardiorenal syndrome. Therefore, our results suggest that although long-standing renal hypoperfusion may cause RD, irreversible pathological tissue damage is limited and restoration of perfusion is likely to improve function.
Our study had the limitations of a retrospective analysis. A further limitation was the declining number of patients in whom renal function was evaluated during follow-up (because of patient death, transplantation, missing measurements, and renal deterioration leading to hemodialysis). Therefore, the GFR estimates later in the post-implantation course might be biased to healthier patients. We therefore examined GFR at various time points, used paired analysis for trends in GFR, and selected the 1-month GFR (with less patient dropout) for our analysis of predictors. Because of the limited data points, only 3 parameters could be fitted into the multivariate model. This potentially could result in the underestimation of other potential independent predictors beyond the ones evaluated. Baseline comparison between the 72 patients with GFR results at 1 month and the 11 patients missing indicates that missing patients were sicker (more chronic kidney disease, lower GFRs on admission, lower hemoglobin, higher aspartate transaminase, higher international normalized ratios). They also had shorter tricuspid regurgitation times and smaller left ventricles. Operations were longer, with more need for blood products, and their hospital stays were longer as well. Detailed investigation of the duration, course, and contributing factors of pre-operative RD was not available in our analysis and might have added predictive variables to our predictive model.
In most patients with end-stage heart failure considered for LVAD, renal dysfunction is reversible. Acute renal failure after LVAD is less common than previously reported and is associated with a complicated post-operative course. Prediction of post-operative improvement in renal dysfunction should take into account the likely contribution of renal hypoperfusion and congestion, irreversible renal injury and response to medical treatment pre-implant.
Drs. Hasin, Park, and Kushwaha received an unrestricted grant from Thoratec Corporation for a different unrelated project. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute renal failure
- bridge to transplantation
- confidence interval
- destination treatment
- glomerular filtration rate
- intra-aortic balloon pump
- left ventricular assist device
- renal dysfunction
- Received May 17, 2011.
- Revision received August 29, 2011.
- Accepted September 20, 2011.
- American College of Cardiology Foundation
- Heywood J.T.,
- Fonarow G.C.,
- Costanzo M.R.,
- Mathur V.S.,
- Wigneswaran J.R.,
- Wynne J.
- Nohria A.,
- Hasselblad V.,
- Stebbins A.,
- et al.
- Ronco C.,
- McCullough P.,
- Anker S.D.,
- et al.
- Pagani F.D.,
- Miller L.W.,
- Russell S.D.,
- et al.
- Filsoufi F.,
- Rahmanian P.B.,
- Castillo J.G.,
- Chikwe J.,
- Carpentier A.,
- Adams D.H.
- Smilde T.D.,
- van Veldhuisen D.J.,
- Navis G.,
- Voors A.A.,
- Hillege H.L.