Renal insufficiency and end-stage renal disease in the heart transplant population

Renal Insufficiency and End-Stage Renal Disease in the Heart Transplant Population Jeffrey Rand Rubel, MD,a Edgar Louis Milford, MD,a Dianne Brenda McKay, MD,b and John Adams Jarcho, MDc Background: Renal insufficiency and end-stage renal disease (ESRD) are important problems in the cardiac transplant population, and are associated with significant morbidity, mortality and financial cost. We undertook this study to define pre-operative or early post-operative predictors of subsequent renal insufficiency and ESRD. Methods: We studied 370 patients at Brigham and Women’s Hospital who received heart transplants between 1984 and 1999, with up to 10-year follow-up. We evaluated 2 time-dependent primary outcomes: early reduction in GFR, and development of ESRD at any timepoint. Cox proportional hazards modeling was used in both univariate and multivariate analyses. Results: The mean estimated glomerular filtration rate (GFR) fell 24% within the first post-transplant year, and remained stable thereafter. By actuarial analysis, 23% of patients developed a 50% reduction in GFR by the third year, and 20% developed ESRD by the tenth year of follow-up. In Cox multivariate analysis, significant predictors of post-transplant ESRD included: GFR ⬍50ml/min (hazards ratio [HR] 3.69, p ⫽ 0.024); high mean cyclosporine trough in the first 6 months (HR 5.10, p ⫽ 0.0059); and presence of diabetes (HR 3.53, p ⫽ 0.021). Conclusions about renal insufficiency outcome were limited by difficulties with accurate estimation of GFR and with definition of renal insufficiency. Conclusions: The results of this study underscore the magnitude of the problem of renal insufficiency and ESRD in the heart transplant population. In addition, these data suggest that patients at high risk for these outcomes can be identified early, even preoperatively, to guide post-operative management. J Heart Lung Transplant 2004;23: 289 –300.

R

enal insufficiency and end-stage renal disease (ESRD) are common complications of heart trans-

From the aRenal Division, Brigham and Women’s Hospital, Boston, Massachusetts, USA; bScripps Research Institute, La Jolla, California, USA; and cCardiovascular Division, Brigham and Women’s Hospital, Boston, Massachusetts, USA. Received December 11, 2002; revised March 7, 2003; accepted March 12, 2003. Supported in part by the American Kidney Fund. Reprint requests: Jeffrey Rand Rubel, MD, Renal Division, Tissue Typing Laboratory, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115. Telephone: 617-732-6660. Fax: 617-566-6176. E-mail: [email protected] Copyright © 2004 by the International Society for Heart and Lung Transplantation. 1053-2498/04/$–see front matter doi:10.1016/S1053-2498(03)00191-8

plantation, and are much more common than in the general population. ESRD has been reported in 3% to 10% of transplant recipients.1–7 ESRD is often a long-term complication, occurring several years after transplant, and higher cumulative risk of ESRD has been typically reported in studies with longer follow-up.1,5,7–9 Many patients who undergo cardiac transplantation already have a degree of renal insufficiency prior to their surgery. Contributing factors include poor cardiac output, low effective circulatory volume, hormonal imbalance, and high prevalence of problems such as diabetes, atherosclerosis and hypertension. ESRD is associated with a high mortality in the 289

290

Rubel et al.

general population, with only a 39.5% overall 5-year survival.10 Survival of transplant recipients with ESRD is at least as poor, and likely worse.2,5,7,11–13 ESRD is a costly complication of transplantation. A 1998 study by Hornberger et al found the cumulative cost of ESRD per transplant patient to exceed $13,000.8 In their analysis of 2,088 heart transplant patients, ESRD accounted for ⬎9% of the overall post-procedure costs of transplantation. The Hornberger study was unable to address quality adjustment, but the quality of life of ESRD patients is known to suffer substantially following initiation of dialysis.10 Hypothesized mechanisms for the decline in renal function after transplant include peri-operative renal injury (typically hemodynamic or toxic) and secondary effects of immunosuppressive agents— most notably cyclosporine (CsA). Even with low doses of C.A, decreases in glomerular filtration rate (GFR) of ⱖ45% have been noted, and most studies have reported substantial loss of GFR within the first post-transplant year.4,5,7,9,14 –21 Other potentially detrimental exposures suffered by transplant recipients include routine interval cardiac catheterizations and a variety of new medications. Prior studies have attempted to define the problem of post-transplant renal insufficiency, and some have evaluated risk factors thereof.14,20,22,23 However, studies have been limited by small numbers of patients or outcomes available for analysis, limited long-term follow-up, or imprecise nature of GFR estimation. Few have analyzed the development of ESRD, principally due to limited follow-up, because ESRD typically takes several years to develop posttransplant. We conducted this study to define the natural history of renal insufficiency and ESRD in the transplant population, and to determine predictors of subsequent renal failure. This information may help to define potential contraindications to transplantation, and also to influence clinical practice guidelines for the management of post-operative renal impairment.

MATERIALS AND METHODS Patients We analyzed the cardiac transplant population at Brigham and Women’s Hospital from its first transplant in February 1984 through January 1999 —a total of 370 patients. Patient follow-up varied from 1 to 10 years. The standard immunosuppressive regimen for these patients consisted of CsA, azathioprine and prednisone, with a number of exceptions.

The Journal of Heart and Lung Transplantation March 2004

Prior to 1989, azathioprine was used only in patients who experienced repeated rejection episodes on a regimen of CsA and prednisone alone. Patients with peri-operative or early post-operative renal dysfunction were typically treated with OKT3, with a subsequent 1- to 2-week delay of CsA treatment. In some patients with recurrent episodes of rejection, methotrexate was added to the regimen. More recently, mycophenolate mofetil has been substituted for azathioprine after recurrent rejection. Finally, in a number of patients with exceptionally benign rejection histories, prednisone has been tapered off and discontinued after the first year post-transplant. Use of these patient data in our study was approved by the hospital’s institutional review board.

Outcomes Two outcomes were analyzed separately in this study. The first outcome represents the development of significant renal insufficiency by the end of the third year post-transplant. Renal insufficiency was defined as a reduction in Cockroft–Gault GFR by ⱖ50% compared with the pre-operative GFR estimate. This threshold value was chosen to represent the clinically substantial change in GFR that would clearly exceed changes due simply to variance of the determinants of the GFR estimation in each patient. The second outcome represents the development of ESRD at any time. This was defined as the requirement of maintenance dialysis for ⱖ3 months.

Estimation of Glomerular Filtration Rate We used the Cockroft–Gault estimation of creatinine clearance (CrCl) as an approximation of GFR: CrCl(ml/min) ⫽

(140 ⫺ age) * weight [Cr] * 72

* 0.85 (if female)

where Cr ⫽ plasma creatinine is expressed in in milligrams per deciliter, weight is expressed in kilograms, and age is expressed in years. We used this formula as our primary method of GFR estimation due to its ease of use in the clinical setting and its fairly widespread acceptance in the literature. This estimation of GFR is not adjusted for body surface area (BSA). The mean BSA of this population was 1.87 m2 with a mean body mass index of 24.5 kg/m2, and this was not considered sufficiently elevated to significantly bias the Cockroft–Gault estimate. As a secondary analysis, we repeated the univariate and multivariate analyses using an estimation of

The Journal of Heart and Lung Transplantation Volume 23, Number 3

GFR derived from the Modification of Diet in Renal Disease (MDRD) study: GFR(ml/min per 1.73 m2) ⫽ 170 * [Cr]⫺0.999 * [age]⫺0.176 * 0.762 (if female) * 1.180 (if black) * [BUN]⫺0.170 * [alb]0.318

where Cr ⫽ plasma creatinine (mg/dl), age is expressed in years, BUN ⫽ blood urea nitrogen (mg/ dl) and alb ⫽ plasma albumin (g/dl). This formula incorporates both demographic and serum variables. It was found to have less variance than the Cockroft–Gault estimate, and has the advantage of being validated with iothalamate GFR.24 The Cockroft–Gault formula has been validated with creatinine clearance, and issues such as tubular secretion of creatinine complicate its interpretation, especially at lower ranges of GFR. The MDRD formula offers a more accurate approximation of GFR and, therefore, of the GFR estimations in this study.

Predictors Table I lists the predictors used in our analysis. This list includes variables we suspected would influence patient outcome, including demographic, historic and laboratory data, as well as surrogates of patient acuity. Mean values are reported for continuous variables, including standard deviation; percentages are reported for categoric or binary variables. Predictors were categorized given their non-linear relationships with each outcome, to maintain statistical power and allow for a reasonable number of outcomes in each category. Information regarding medications used in addition to immunosuppressants was unobtainable, and therefore not included in our analysis. The 12 highest monthly CsA levels were used to calculate the mean level for the first year posttransplant, to represent maximal dose exposure to the drug in a given month. All levels were true morning trough concentrations, measured from whole blood. Serum levels (performed prior to November 1992 at this hospital) were converted to whole blood levels in micrograms per liter using the formula W ⫽ (S ⫹ 34.2)/0.544, where W ⫽ whole blood concentration and S ⫽ concentration of serum.25 Although the linear regression modeling used to determine this conversion formula has limitations in this application.26 we considered this the most appropriate way to represent the data. Cut-offs for CsA level used in categorization were chosen

Rubel et al.

291

based on upper limits of the therapeutic range for the time period in question. Specifically, the target therapeutic range in the first 6 months was a trough level of 250 to 350 ␮g/liter, and in the second 6 months 200 to 250 ␮g/liter. The highest category was chosen using 1 standard deviation above the mean in each 6 month period. Historic diagnoses, including diabetes, peripheral vascular disease, smoking, hypertension and hyperlipidemia, were determined by the medical staff evaluating the patient for transplant. Only confirmed, not “suspected,” diagnoses were included. For the purpose of this study patients were considered to have diabetes if they required any type of medication for adequate blood sugar control. Peripheral vascular disease was defined as a moderate or 50% stenosis in at least one lower extremity by non-invasive study. Smoking was defined as ⱖ10 pack-years or similar assessment by staff. Hypertension was defined as a systolic blood pressure of ⬎140 mm Hg, diastolic blood pressure ⬎85 mm Hg, or use of an anti-hypertensive medication. Hyperlipidemia was defined as a total cholesterol of ⬎200 mg/dl, a low-density lipoprotein level of ⬎100 mg/dl, or use of a lipid-lowering agent.

Statistical Analysis Statistical analysis was performed using SAS for Windows. For each of the 2 outcomes, univariate relationships between each candidate predictor and outcome were calculated using a time-dependent Cox regression model. For categoric predictors, the p-value from the score test was used. Those predictors with p ⱕ 0.20 on univariate analysis were included in the subsequent model selection. A forward selection scheme with a p-value cut-off of 0.05 was used to minimize collinearity, which was expected to be a problem among a number of the covariates. Formal testing was performed of the proportional hazards assumption. Non-significant confounders of the model were included if their addition changed the effect (hazards ratio) of any of the significant variables by ⱖ20%. All covariates used in the model selection scheme were considered as potential confounders (i.e., those with p ⱕ 0.20 on univariate analysis). Age at transplant was also tested as a confounder, regardless of significance. The resulting models were the final Cox proportional hazards regression models used for this study. As previously mentioned, analysis of both primary outcomes was repeated with the same process using the modified MDRD estimate of GFR.

292

Rubel et al.

The Journal of Heart and Lung Transplantation March 2004

TABLE I Characteristics of 370 heart transplant patients Mean (ⴞ SD) or number of patients (%) Race White Black Other NEOB classification 1 (mechanical support) 2 (intensive care unit) 3 (in-hospital) 4 (out-of-hospital) UNOS classification 1 (intensive care unit, pressors) 2 (all others) Etiology of heart failure Ischemic heart disease Cardiomyopathy Valvular heart disease Congenital heart disease Other Age at transplant (years) Gender Male Female Serum creatinine (mg/dl)* Estimated GFR (ml/min)*† Serum cholesterol (mg/dl)* Cardiac index (liters/min · m2)* Serum albumin (g/dl)* Mean highest cyclosporine trough (␮g/liter) Months 1–6 post-transplant Months 7–12 post-transplant‡ History of diabetes mellitus§ History of peripheral vascular disease§ History of smoking§ History of hypertension§ History of hyperlipidemia§

344 (93%) 15 (4%) 11 (3%) 46 (13%) 28 (8%) 86 (23%) 208 (56%) 68 (18%) 302 (82%) 158 (43%) 183 (49%) 18 (5%) 4 (1%) 7 (2%) 49.1 ⫾ 11.1 287 (78%) 83 (22%) 1.3 ⫾ 0.64 78 ⫾ 32 170.7 ⫾ 55.0 2.0 ⫾ 0.65 4.0 ⫾ 0.72 397 ⫾ 103 321 ⫾ 79 59 (17%) 67 (33%) 231 (66%) 97 (27%) 115 (33%)

Number of patients with data 370

368

370 370

370 370 329 323 294 315 311 308 351 205 350 353 351

NEOB, New England Organ Bank; UNOS, United Network of Organ Sharing. *Pre-operative values. † Cockroft–Gault estimation. ‡ Refer to Materials and Methods for definition. § Historic conditions determined by evaluating medical staff (see, Materials and Methods section).

RESULTS Descriptive Analyses Characteristics of all 370 patients are illustrated in Table I, and additional characteristics of the 21 patients who subsequently developed ESRD are shown in Table II. The mean estimated GFR of all 370 patients over time is shown in Figure 1, including error bars representing the standard deviation of the mean. The number of patients analyzed at each timepoint is included below the figure. GFR data

were omitted for ESRD patients once they were on dialysis. The estimated GFR fell by 24% from the time of pre-operative evaluation to the end of the first year post-transplant and remained stable thereafter. Of the 370 patients analyzed, 59 developed a 50% reduction in GFR by the end of the third year post-transplant—23.1% by actuarial estimate. Twenty-one developed ESRD in the 10-year posttransplant follow-up, or 20.3% at the tenth year by

The Journal of Heart and Lung Transplantation Volume 23, Number 3

Rubel et al.

293

TABLE II Additional characteristics of the 21 heart transplant patients who developed posttransplant end-stage renal disease (ESRD) Mean (ⴞ SD) or number of patients (%) Age at transplant (years) Male gender White race Ischemic etiology of heart failure Pre-operative creatinine (mg/dl) Pre-operative Cockroft–Gault GFR (ml/min) Pre-operative proteinuria present* Pre-operative active urine sediment† Post-operative proteinuria present* Post-operative active urine sediment† Time to development of urinary abnormalities (years)‡ Time to initiation of dialysis (years)§ Etiology of ESRD㛳 Cyclosporine toxicity Hypertension Diabetes mellitus Acute renal failure Other medication toxicity Miscellaneous Unknown

55.0 ⫾ 6.7 14 (67%) 20 (95%) 9 (43%) 1.3 ⫾ 0.4 69 ⫾ 29 0 (0%) 0 (0%) 16 (76%) 8 (38%) 3.5 ⫾ 2.7 6.3 ⫾ 2.8 17 (81%) 6 (29%) 2 (10%) 4 (19%) 3 (14%) 1 (5%) 1 (5%)

*Proteinuria was defined as 1⫹ or more by urine dipstick. † “Active sediment” was defined as any of the following: 1⫹ blood; 1⫹ leukocytes or 1⫹ nitrites by dipstick; ⱖ5 red blood cells per high-powered field; ⱖ5 white blood cells per high-powered field; or ⱖ1 cellular or granular casts per high-powered field. ‡ Time from transplant to the development of either proteinuria or active sediment as previously defined. § Time from transplant to initiation of dialysis. 㛳 Diagnoses made without biopsy; some patients had more than one etiology of ESRD.

actuarial analysis. Figure 2 is a Kaplan–Meier plot showing the development of various degrees of renal insufficiency over time post-transplant. The actuarial format adjusts for the varying lengths of follow-up available among the patients. Four separate gradations of renal failure are shown: 30%, 40% and 50% reduction in GFR, and the development of ESRD as previously defined. Figure 2 also shows that the patients who ultimately developed ESRD did so several years after transplant (mean 6.3 ⫾ 2.8 years).

FIGURE 1 Mean estimated GFR over time following

cardiac transplantation. Error bars represent 1 standard deviation. The GFR at time ⫽ 0 is the pre-operative estimation. The number of patients with data available for analysis at each year post-transplant is shown below the figure, as is the total number of patients alive at each year. All GFR estimates were calculated with the Cockroft–Gault formula. GFR data were omitted from this figure for ESRD patients once they were on dialysis.

Comparison of GFR Estimates Table III displays the correlations among various pre-operative estimates of GFR: Cockroft–Gault CrCl; GFR estimated with the modified MDRD formula; 24-hour creatinine clearance; and serum creatinine. These were all simultaneous measurements. Table III underscores the fact that no single measurement is ideal and that they correlate poorly. There is significant discrepancy between the various ways of quantifying renal function, an issue common to the general medical literature as well as the heart transplant literature.16,27

Univariate Results Tables IV and V show the results of univariate testing between each separate predictor and the 2 primary outcomes. Table IV displays results using outcome of renal insufficiency (50% reduction in Cockroft–Gault GFR at 3 years post-transplant), and Table V displays results for development of ESRD at any time post-transplant.

Multivariate Results Table VI shows the results of multivariate modeling for the development of ESRD, presented the same way as the univariate outcomes. Of the variables in Table V with p ⱕ 0.20, age, transplant year and

294

Rubel et al.

The Journal of Heart and Lung Transplantation March 2004

other covariates met criteria for confounding in this multivariate analysis.

Secondary Analysis

FIGURE 2 Kaplan–Meier plot showing development of renal insufficiency over time post-transplant. Four separate outcomes are shown, representing different levels of renal dysfunction. All GFR estimates were calculated with the Cockroft–Gault formula. Data for GFR estimation were available at yearly intervals after transplant.

history of hyperlipidemia all met criteria as confounders and were therefore included in the multivariate model. No significant effect modification was found among the 3 significant variables included in Table VI. The likelihood ratio test for the multivariate model had p ⬍ 0.0001. The multivariate results looking at development of renal insufficiency by the end of the third year post-transplant were limited, and therefore were not included in tabular form. The only significant variable that remained in the multivariate model was male gender (hazards ratio 0.41, p ⫽ 0.0010). These data were controlled for pre-operative GFR, but no

The univariate and multivariate testing was repeated using the MDRD study’s estimate of each GFR. Significant univariate predictors of both outcomes were qualitatively the same, with the same variables included in the model selection process. Similarly, the final models included the same covariates and confounders as those based on Cockroft– Gault estimates of GFR. A post hoc analysis was performed to investigate the presence of a relationship between estimated GFR at 1 year post-transplant and subsequent development of ESRD. Prior studies have shown this to be the case.1,9,14,20 In our study, a 1-year estimated GFR of ⬍50 ml/min was associated with a hazards ratio for the development of ESRD of 4.81 [1.75, 13.2], p ⫽ 0.0024. This univariate comparison was even more significant when estimated GFR was used as a continuous variable.

DISCUSSION ESRD is an important problem in the cardiac transplant population and is accompanied by significant morbidity, mortality and financial cost. By actuarial analysis, 20.3% of our patient population developed ESRD by the end of the 10-year followup. This estimate is in keeping with other studies where in long-term follow-up and actuarial analysis were performed.2,20 The mean time to development of ESRD was ⬎6 years in our study, similar to onset times reported in the literature.2,6 – 8 Characteristics

TABLE III Relationships among simultaneous estimates of glomerular filtration rate, measured at evaluation for transplant Serum creatinine Serum creatinine Measured 24hour CrCl Cockroft–Gault estimated CrCl MDRD study estimated GFR

r ⫽ ⫺0.22; p ⫽ 0.0006 r ⫽ ⫺0.65; p ⬍ 0.0001 r ⫽ ⫺0.85; p ⬍ 0.0001

Measured 24-hour CrCl

Cockroft–Gault estimated CrCl

MDRD study estimated GFR

r ⫽ ⫺0.22; p ⫽ 0.0006

r ⫽ ⫺0.65; p ⬍ 0.0001 r ⫽ 0.37; p ⬍ 0.0001

r ⫽ ⫺0.85; p ⬍ 0.0001 r ⫽ 0.36; p ⬍ 0.0001 r ⫽ 0.77; p ⬍ 0.0001

r ⫽ 0.37; p ⬍ 0.0001 r ⫽ 0.36; p ⬍ 0.0001

r ⫽ 0.77; p ⬍ 0.0001

Reported for each correlation are the Spearman correlation coefficient (r) and its associated p-value. CrCl, creatinine clearance; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease.

The Journal of Heart and Lung Transplantation Volume 23, Number 3

Rubel et al.

295

TABLE IV Univariate analyses showing a 50% reduction in glomerular filtration rate (GFR) in the first 3 years post-transplant (referent categories have a hazards ratio of 1.00 by definition) Hazards ratio [95% confidence interval] Pre-operative GFR (ml/min) ⬍50 (vs ⬎90) 50–70 (vs ⬎90) 70–90 (vs ⬎90) White (vs. non-white) NEOB classification 1 or 2 (vs 4) 3 (vs 4) UNOS ICU classification Etiology of heart failure Ischemic (vs non-ischemic) Age at transplant (years) 50–60 (vs ⬍50) ⬎60 (vs ⬍50) Male gender (vs female) Year of transplant 1984–1988 (vs 1994–1999) 1989–1993 (vs 1994–1999) Serum cholesterol ⱖ200 mg/dl Cardiac index ⱕ2 liters/min · m2 Serum albumin ⬍4 g/dl Mean highest CsA trough (␮g/liter) Months 1–6 post-transplant 350–500 (vs ⬍350) ⱖ500 (vs ⬍350) Mean highest CsA trough (␮g/liter) Months 7–12 post-transplant 250–400 (vs ⬍250) ⱖ400 (vs ⬍250) History of diabetes mellitus History of peripheral vascular disease History of smoking History of hypertension History of hyperlipidemia

0.24 [0.085, 0.69] 0.26 [0.13, 0.52] 0.42 [0.22, 0.79] 1.39 [0.50, 3.84] 1.49 [0.81, 2.73] 0.90 [0.47, 1.72] 1.63 [0.91, 2.92] 0.69 [0.40, 1.18] 1.27 [0.73, 2.21] 1.12 [0.50, 2.50] 0.49 [0.29, 0.84] 0.98 [0.33, 2.91] 1.70 [0.96, 3.01] 1.16 [0.64, 2.10] 0.75 [0.44, 1.29] 1.14 [0.68, 1.91] 1.62 [0.90, 2.89] 1.85 [0.90, 3.81] 1.42 [0.60, 3.35] 2.55 [0.98, 6.63] 1.51 [0.83, 2.75] 0.64 [0.29, 1.40] 0.95 [0.56, 1.61] 1.34 [0.77, 2.33] 1.05 [0.61, 1.78]

p-value ⬍0.0001*

0.52 0.32 0.10* 0.17* 0.69 0.0074* 0.14* 0.62 0.30 0.63 0.16*

0.078*

0.18* 0.26 0.84 0.30 0.87

NEOB, New England Organ Bank; UNOS, United Network of Organ Sharing. *Covariate significant at p ⱕ 0.20.

of our patient population are similar to those in prior studies, including one large study of Medicare beneficiaries.8 Prior studies of ESRD in transplant recipients have likely underestimated the magnitude of the problem of ESRD due to follow-up periods that are often shorter than those typically required to see the emergence of ESRD in the heart transplant population.2,4,15–17,19,21,22,28,29 Studies with longer follow-up have reported a higher incidence of ESRD.1,5,7–9 Hornberger’s large-scale study confirmed the suspicion that the burden of ESRD is substantially larger than previously described.8 As more studies include long-term follow-up, we antic-

ipate that the numbers represented in our study are likely to be reproduced. The estimated GFR in our patient population fell substantially (24%) between the measurement before transplant and that estimated at the end of the first year of follow-up. This early and substantial decline in renal function has been reported in virtually all studies on renal function in heart transplant recipients.4,5,7,9,14 –21 After this initial decline some studies have demonstrated a continuing but decelerated decline,4,7,9,20 –22 whereas others have shown renal function to remain stable.5,16 –19,30 –32 The estimated mean GFR in our study population did not decline over time. As shown in Figure 1, the

296

Rubel et al.

The Journal of Heart and Lung Transplantation March 2004

TABLE V Univariate analyses showing development of end-stage renal disease (ESRD) at any time posttransplant (referent categories have a hazards ratio of 1.00 by definition) Hazards ratio [95% confidence interval] Pre-operative GFR (ml/min) ⬍50 (vs ⬎90) 50–70 (vs ⬎90) 70–90 (vs ⬎90) White (vs non-white) NEOB classification 1 or 2 (vs 4) 3 (vs 4) UNOS ICU classification Etiology of heart failure Ischemic (vs non-ischemic) Age at transplant (years) 50–60 (vs ⬍50) ⬎60 (vs ⬍50) Male gender (vs female) Year of transplant 1984–1988 (vs 1994–1999) 1989–1993 (vs 1994–1999) Serum cholesterol ⱖ200 mg/dliters Cardiac index ⱕ2 liters/min · m2 Serum albumin ⬍4 g/dliters Mean highest CsA trough (␮g/liter) Months 1–6 post-transplant 350–500 (vs ⬍350) ⱖ500 (vs ⬍350) Mean highest CsA trough (␮g/liter) Months 7–12 post-transplant 250–400 (vs ⬍250) ⱖ400 (vs ⬍250) History of diabetes mellitus History of peripheral vascular disease History of smoking History of hypertension History of hyperlipidemia

3.76 [1.06, 13.3] 1.05 [0.26, 4.21] 1.18 [0.32, 4.41] 1.25 [0.17, 9.42] 0.66 [0.15, 2.95] 1.52 [0.57, 4.06] 0.65 [0.15, 2.83] 1.66 [0.68, 4.06] 6.53 [1.86, 23.0] 7.85 [1.73, 35.7] 0.70 [0.27, 1.82] 0.10 [0.013, 0.81] 0.33 [0.058, 1.85] 1.80 [0.62, 5.19] 0.52 [0.18, 1.51] 0.93 [0.37, 2.31] 0.76 [0.18, 3.19] 2.43 [0.63, 9.35] 0.69 [0.15, 3.10] 0.90 [0.17, 4.74] 3.71 [1.46, 9.43] 1.61 [0.54, 4.83] 1.29 [0.46, 3.62] 1.68 [0.70, 4.08] 3.07 [1.26, 7.50]

p-value 0.063*

0.83 0.52 0.57 0.26 0.0019* 0.46 0.055* 0.27 0.22 0.87 0.050*

0.81

0.0032* 0.39 0.63 0.24 0.0097*

CsA, cyclosporine; NEOB, New England Organ Bank; ICU, intensive care unit; UNOS, United Network of Organ Sharing. *Covariate significant at p ⱕ 0.020.

TABLE VI Multivariate analysis showing development of ESRD at any time post-transplant* Pre-operative GFR (ml/min) at ⬍50 (vs ⬎50) Mean highest CsA trough (␮g/ liter) at Months 1–6 posttransplant at ⱖ500 (vs ⬍500) History of diabetes mellitus

Hazards ratio [95% confidence interval]

p-value

3.69 [1.19, 11.4]

0.024

5.10 [1.60, 16.3]

0.0059

3.53 [1.21, 10.2]

0.021

*Results controlled for age, year of transplantation and history of hyperlipidemia. Referent categories have a hazards ratio of 1.00 by definition.

The Journal of Heart and Lung Transplantation Volume 23, Number 3

apparent increase in mean GFR over the follow-up period, although statistically significant, likely represents survivorship bias rather than a meaningful physiologic improvement. Despite the stability of our population’s mean estimated GFR, however, Figure 2 demonstrates that renal insufficiency and failure developed at increasing rates post-transplant, regardless of the percent decline used to define renal insufficiency. The actuarial format underscores the magnitude of renal insufficiency almost immediately post-transplant. Interestingly, our Kaplan–Meier plot of ESRD over time is virtually identical to that published in the Hornberger study at the end of their 6-year follow-up.8 Analysis of ESRD and its predictors is fairly new to the heart transplant literature, although one study reported an association between GFR at evaluation and elevated creatinine post-transplant.14 We found low pre-operative GFR to be a significant predictor of subsequent ESRD. This finding is not surprising, although it has not been described previously.5 The significance of poor pre-operative GFR as a predictor of subsequent ESRD was maintained on multivariate analysis as well, with an associated hazards ratio of 3.69. Estimated GFR at 1 year post-transplant also significantly predicted the development of ESRD in our study, when included in post hoc analysis. This is consistent with previous studies linking a poor renal outcome to low GFR in the first year post-transplant, and provides valuable information about subsequent ESRD risk.1,9,14,20 Our study is the first to demonstrate a strong, significant relationship between cyclosporine level and development of post-transplant ESRD. Some studies have shown a relationship between cyclosporine dose and/or level and decline in renal function, but not to ESRD.1,4,14,15,19,33–35 Studies looking at advanced renal failure or ESRD have not shown such an association using either dose or level of cyclosporine.5,20,22 We focused on drug levels in the first year because both target and actual levels are the highest over the course of the first year posttransplant, and cyclosporine-mediated renal injury is only rarely thought to be reversible after the first year post-transplant despite dose reduction or complete discontinuation.1,36 Advanced age and diabetes are risk factors for ESRD in the general population, and it is not surprising that they are markers of risk in the heart transplant recipients studied here. The association of hyperlipidemia to ESRD in the transplant popu-

Rubel et al.

297

lation is less clear, and prior studies have failed to demonstrate a significant relationship.14 In our study, this significant univariate predictor ultimately failed to maintain statistical significance in multivariate analysis due to collinearity with both cyclosporine level and diabetes. The fact that diabetes remains a strong independent risk factor for development of ESRD may suggest a sub-group of transplant patients whose higher risk for ESRD should lead to closer follow-up of post-operative renal function. Earlier year of transplant was associated with lower risk of ESRD on univariate analysis, although it lost significance in the multivariate model. The most likely explanation for this association is the growing trend, seen in our transplant program and many others, to accept as candidates patients with greater risk of potential complications. More patients with diabetes, pre-existing renal insufficiency and advanced age have been accepted for transplant compared with the 1990s. This trend is now reversing in many centers as the true long-term frequency of ESRD becomes apparent. Another explanation for our finding is that many more ESRD-free person-years exist in our database for the earlier transplants. The few patients in our database who were transplanted from 1994 to 1999 and developed ESRD did so much more rapidly than those transplanted in earlier years. Although this shorter time to ESRD of the later transplant patients probably represents random variation, it likely contributed to the apparent lower risk of ESRD associated with an earlier transplant. The second outcome analyzed in our study was a 50% reduction in estimated GFR by the end of the third year post-transplant. Information about shortterm renal risk can be clinically useful, and can potentially guide patient management shortly after the transplant. However, of the variables we tested, the only one to achieve statistical significance on univariate and multivariate analyses was male gender. Male gender was associated with an apparent protective effect, with a hazards ratio of 0.49 on univariate analysis, and 0.41 on multivariate analysis. One possible explanation for this is the smaller kidney size and associated lower nephron number in women compared with men. However, the physiologic basis for this observation is not clear, and no such relationship has been described previously in the cardiac transplant literature. One of the goals of this study was to determine if pre-operative risk factors existed that were important enough in magnitude and significance to alter

298

Rubel et al.

organ allocation practices. In this study we were unable to identify risk factors of sufficient influence to recommend any such change, although high-risk patients, such as those with diabetes and low preoperative GFR, pose a significant challenge. Nevertheless, we were able to gain valuable insight into patient risk of renal dysfunction and ESRD that could impact patient management both before and after surgery. Cyclosporine toxicity is an important contributor to renal dysfunction, one that has pathologic correlates on renal biopsy.1,14,37,38 Other etiologies of renal insufficiency and failure exist in this population. Interstitial nephritis (often drug-related), atheroembolic disease, renovascular disease, contrast nephropathy and acute tubular necrosis are also important clinical entities in heart transplant recipients. Many of these factors, including cyclosporine toxicity, can develop insidiously with a heterogeneous presentation and variable activity in the urine sediment (see Table II for representative data from our study). Unfortunately, patients with renal dysfunction after transplant are not frequently seen by nephrologists, and renal biopsies are very rarely performed. Data from this study suggest that high-risk groups should be followed even more closely than usual, such as those with diabetes, poor pre-operative renal function, high early cyclosporine levels, and those with renal dysfunction at 1 year post-transplant. More frequent follow-up in these high-risk patients can provide data about rate of progression of renal dysfunction, activity of the urine sediment, or presence and progression of proteinuria. One reasonable way to achieve this is through formal evaluation and treatment by a nephrologist. Early referral has been associated with significantly better survival in the general population, potentially due to attention to complications of renal failure, which are independent of the etiology.39,40 No definitive literature exists to support modification of a transplant patient’s medical regimen to reduce risk of renal progression. However, there is a growing body of literature that provides insight into possibilities for renal risk reduction. Dose reduction of CsA can reduce its hemodynamic effects, although it may be less effective at reducing the fibrosis associated with its long-term use.1,3,15,22,30,37,38,41 Data from our study suggest that maintaining lower trough levels would reduce the risk of subsequent ESRD, especially early in the post-transplant period. There is evidence that tacrolimus may be less nephrotoxic than CsA,29,42 and its use has become increasingly common in solid

The Journal of Heart and Lung Transplantation March 2004

organ transplantation. It is unclear how tacrolimus should fit into current immunosuppressive practice in heart transplant patients, but the drug is clearly worthy of further research efforts, and is increasingly used being in the heart transplant population. We were not able to control for use of these medications in our study, but the conflicting nature of these data make it difficult to predict whether anti-hypertensive drugs or other agents would have a positive or negative impact on long-term renal function, if any impact at all. However, although the magnitude and direction of such potential confounding is unclear, it is unlikely to explain the magnitude of effect seen here with the significant multivariate predictors. Strategies for renal risk reduction that are used in the general population may also be useful in heart transplant patients, including hydration and medications such as acetylcysteine.43– 45 Evidence from this and other studies suggests that aggressive lipid lowering may reduce renal risk.14,18 Data on angiotensin-converting enzyme inhibitors are sparse, although they have not been shown to reduce the fibrosis associated with CsA use.9,20 Calcium channel-blocking agents have not been adequately studied in transplant recipients, but they have been shown to improve GFR in short-term studies and deserve further investigation.9,46 The patients in our study were similar to most contemporary heart transplant recipients, and the present results are thus reasonably generalizable to the United States heart transplant population. However, our results must be interpreted in the context of the study design. The appropriate methods for estimating GFR and for defining an appropriate surrogate for progression of renal insufficiency are controversial. Therefore, the outcome of 50% reduction in estimated GFR at 3 years is difficult to interpret. We performed post hoc analyses using the MDRD study’s GFR estimate for this reason, but the implications of short-term outcome are nonetheless limited. We were unable to control for post-operative hypertension, which develops in a large percentage of patients post-transplant, and may contribute to the observed decline in renal function.1,14,19 Our analysis also did not address intra- or peri-operative factors, such as ischemic time, time on cardiopulmonary bypass, or pressor use, all of which may influence renal function. ESRD is a much more straightforward outcome to define, and therefore much less subjective. Our analysis of ESRD is limited more by number of outcomes than by its definition. One goal of future

The Journal of Heart and Lung Transplantation Volume 23, Number 3

research will be to better define events in the peri-operative period and the first year post-transplant. It is highly likely that such events influence subsequent renal function, and can provide valuable information about risk assessment. To more completely characterize the problem of ESRD in the heart transplant population, studies must be undertaken with more patients and statistical power, with follow-up periods of sufficient duration observe the development of ESRD as a more common outcome. Multicenter and prospective studies offer the potential to provide such information. Given the burden of ESRD in this population, its reduction has the potential to significantly diminish the overall morbidity and mortality of cardiac transplant recipients. The authors thank Sui Tsang from the Brigham and Women’s Cardiac Transplant Office for her invaluable contributions to this work. We also thank the Brigham and Women’s cardiac transplant staff for their flexibility and assistance.

REFERENCES 1. Myers BD, Sibley R, Newton L, et al. The long-term course of cyclosporine associated chronic nephropathy. Kidney Int 1988;33:590 –600. 2. Jayasena SD, Riaz A, Lewis CM, Neild GH, Thompson FD, Woolfson RG. Outcome in patients with end-stage renal disease following heart or heart–lung transplantation receiving peritoneal dialysis. Nephrol Dial Transplant 2001;16: 1681–5. 3. Woolfson RG, Neild GH. Cyclosporine nephrotoxicity following cardiac transplantation. Nephrol Dial Transplant 1997;12:2054 –6. 4. Greenberg A, Thompson ME, Griffith BJ, et al. Cyclosporine nephrotoxicity in cardiac allograft patients—a seven year follow-up. Transplantation 1990;50:589 –93. 5. van Gelder T, Balk AHMM, Zietse R, Hesse C, Mochtar B, Weimar W. Renal insufficiency after heart transplantation: a case– control study. Nephrol Dial Transplant 1998;13:2322–6. 6. Parameshwar J, Schofield P, Large S. Long-term complications of cardiac transplantation. Br Heart J 1995;74:341–2. 7. Goldstein DJ, Zuech N, Sehgal V, Weinberg AD, Drusin R, Cohen D. Cyclosporine-associated end-stage nephropathy after cardiac transplantation: incidence and progression. Transplantation 1997;63:664 –8. 8. Hornberger J, Best J, Geppert J, McClellan M. Risks and costs of end-stage renal disease after heart transplantation. Transplantation 1998;66:1763–70. 9. Herlitz H, Lindelow B. Renal failure following cardiac transplantation. Nephrol Dial Transplant 2000;15:311–4. 10. U.S. Renal Data System. USRDS 2002 annual data report: atlas of end-stage renal disease in the United States. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2002. 11. Bernardini J, Piraino B, Kormos RL. Patient survival with renal replacement therapy in heart transplantation patients. ASAIO J 1998;44:546 –8. 12. Frimat L, Villemot J-P, Cormier L, et al. Treatment of

Rubel et al.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

299

end-stage renal failure after heart transplantation. Nephrol Dial Transplant 1998;13:2905–8. Anguita M, Arizon JM, Valles F, et al. Influence on survival after heart transplantation of contraindications seen in transplant recipients. J Heart Lung Transplant 1992;11:708 –15. Sehgal V, Radhakrishnan J, Appel GB, Valeri A, Cohen DJ. Progressive renal insufficiency following cardiac transplantation: cyclosporine, lipids, and hypertension. Am J Kidney Dis 1995;26:193–201. Greenberg A, Egel JW, Thompson ME, et al. Early and late forms of cyclosporine nephrotoxicity: studies in cardiac transplant recipients. Am J Kidney Dis 1987;9:12–22. Goral S, Ynares C, Shyr Y, Yeoh TK, Johnson HK. Longterm renal function in heart transplant recipients receiving cyclosporine therapy. J Heart Lung Transplant 1997;16:1106. Ruggenenti P, Perico N, Amuchastegui CS, Ferrazzi P, Mamprin F, Remuzzi G. Following an initial decline, glomerular filtration rate stabilizes in heart transplant patients on chronic cyclosporin. Am J Kidney Dis 1994;24:549 –53. Hartmann A, Andereassen AK, Holdaas H, Simonsen S, Geiran O, Berg KJ. Five years’ follow-up of renal glomerular and tubular functions in heart transplant recipients. J Heart Lung Transplant 1996;15:972–9. Tinawi M, Miller L, Bastani B. Renal function in cardiac transplant recipients: retrospective analysis of 133 consecutive patients in a single center. Clin Transplant 1997;11:1–8. Lindelow B, Bergh C-H, Herlitz H, Waagstein F. Predictors and evolution of renal function during 9 years following heart transplantation. J Am Soc Nephrol 2000;11:951–7. Hakim M, Wallwork J, English T. Cyclosporin A in cardiac transplantation: medium term results in 62 patients. Ann Thorac Surg 1988;46:495–501. Zietse R, Balk AH, Vanden Dorpel MA, Meeter K, Bos E, Weimar W. Time course of the decline in renal function in cyclosporine-treated heart transplant recipients. Am J Nephrol 1994;14:1–5. Macris MP, Ford EG, Van Buren CT, Frazier OH. Predictors of severe renal dysfunction after heart transplantation and intravenous cyclosporine therapy. J Heart Transplant 1989; 8:444 –8. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999;130:461–70. Wenk M, Follath F, Abisch E. Temperature dependency of apparent cyclosporine A concentrations in plasma. Clin Chem 1983;29:1865. Agarwal RP, McPherson RA, Threatte GA. Assessment of cyclosporine A in whole blood and plasma in five patients with different hematocrits. Ther Drug Monit 1985;7:61–5. Cantarovich M, Giannetti N, Cecere R. Correlation between serum creatinine, creatinine clearance, the calculated creatinine clearance, and the glomerular filtration rate in heart transplant patients. J Heart Lung Transplant 2002;21:815–7. Spratt P, Esmore D, Baron D, Shanahan MX, Farnsworth AE, Chang VP. Effectiveness of minimal dosage cyclosporine in limiting toxicity and rejection. J Heart Lung Transplant 1986;5:8 –11. Pham SM, Kormos RL, Kawai A, et al. Tacrolimus (FK506) in clinical cardiac transplantation: a five year experience. Transplant Proc 1996;28:1002–4. Waser M, Maggiorini M, Binswanger U, et al. Irreversibility of

300

31.

32.

33.

34.

35.

36.

37.

38.

Rubel et al.

cyclosporine-induced renal function impairment in heart transplant recipients. J Heart Lung Transplant 1993;12:846–50. Gonwa TA, Mai ML, Pilcher J, et al. Stability of long-term renal function in heart transplant patients treated with induction therapy and low-dose cyclosporine. J Heart Lung Transplant 1992;11:926 –8. Lewis RM, Van Buren CT, Radovancevic B, et al. Impact of long term cyclosporine immunosuppression therapy on native kidneys versus renal allografts: serial renal function in heart and kidney transplant recipients. J Heart Lung Transplant 1991;10:63–70. Feutren G, Mihatsch MJ. Risk factors for cyclosporineinduced nephropathy in patients with autoimmune diseases. N Engl J Med 1992;326:1654 –60. Cantarovich M, Fitchett D, Latter DA. Cyclosporine trough levels, acute rejection, and renal dysfunction after heart transplantation. Transplantation 1995;59:444 –7. Hornung TS, de Goede CGEL, O’Brien C, Moghal NE, Dark JH, O’Sullivan JJ. Renal function after pediatric cardiac transplantation: the effect of early cyclosporine dosage. Pediatrics 2001;107:1346 –50. Young EW, Ellis CN, Messana JM, et al. A prospective study of renal structure and function in psoriasis patients treated with cyclosporine. Kidney Int 1994;45:1216 –22. Griffiths MH, Crowe AV, Papadaki L, et al. Cyclosporine nephrotoxicity in heart and lung transplant patients. Q J Med 1996;89:751–63. Bertani T, Ferrazzi P, Schieppati A, et al. Nature and extent

The Journal of Heart and Lung Transplantation March 2004

39.

40. 41.

42.

43.

44.

45.

46.

of glomerular injury induced by cyclosporine in heart transplant patients. Kidney Int 1991;40:243–50. Avorn J, Bohn RL, Levy E, Levin R, Owen WF, Jr, Winkelmayer WC, Glynn RJ. Nephrologist care and mortality in patients with chronic renal insufficiency. Arch Intern Med 2002;162:2002–6. Levin A. Consequences of late referral on patient outcomes. Nephrol Dial Transplant 2000;15(Suppl 3):8 –13. Falkenhain ME, Cosio FG, Sedmak DD. Progressive histologic injury in kidneys from heart and liver transplant recipients receiving cyclosporine. Transplantation 1996;62:364–70. Pham SM, Kormos RL, Hattler BG, et al. A prospective trial of tacrolimus (FK506) in clnical heart transplantation: intermediate-term results. J Thorac Cardiovasc Surg 1996;111:764–72. Diaz-Sandoval LJ, Kosowsky BD, Losordo DW. Acetylcysteine to prevent angiography-related renal tissue injury (the APART trial). Am J Cardiol 2002;89:356 –8. Shyu K, Cheng J, Kuan P. Acetylcysteine protects against acute renal damage in patients with abnormal renal function undergoing a coronary procedure. J Am Coll Cardiol 2002;40:1383–8. Denys BG, Reddy PS, Urestsky BF. The use of ionic and nonionic contrast agents and the effects of hydration in the post cardiac transplant patient with moderate renal insufficiency. Angiology 1991;3:218 –23. Chan C, Maurer J, Cardella C, Cattran D, Pei Y. A randomized controlled trial of verapamil on cyclosporine nephrotoxicity in heart and lung transplant recipients. Transplantation 1997;63:1435–40.