NDT Advance Access first published online on July 7, 2007
This version published online on September 21, 2007
Nephrology Dialysis Transplantation, doi:10.1093/ndt/gfm424
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Factors associated with mortality in patients new to haemodialysis
1Division of Renal Diseases and Hypertension, University of Colorado HSC, Denver, CO, 2Division of Cardiology, Emory University School of Medicine, Atlanta, GA, 3Computer and Statistical Consultants, Inc., Santa Barbara, CA and 4Denver Nephrology, Denver, CO, USA
Correspondence and offprint requests to: David M. Spiegel, MD, University of Colorado Health Sciences Center, 4200 E. 9th Avenue, Box C-281, Denver, CO 80262, USA. Email: david.spiegel{at}uchsc.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Background. Patients receiving dialysis therapy for end-stage kidney failure have a high cardiovascular mortality that can only be partially explained by traditional risk factors.
Methods. This study was a post hoc analysis of a prospectively gathered data set from a randomized trial comparing outcomes in new haemodialysis patients treated with sevelamer or calcium-containing phosphate binders. Patients were followed from the time of enrollment until death or censor on 31 December 2005. Median follow-up was 3.6 years. Demographics, cardiovascular risk factors, laboratory data, medication use and severity of vascular calcification were available at baseline and over the first 18 months of dialysis.
Results. Baseline predictors of mortality included age, creatinine, heart rate, iPTH, C-reactive protein (CRP), coronary and aortic calcium scores and the presence of aortic valve calcification. Over the first 18 months, averages of diastolic blood pressure, BUN, creatinine, albumin, phosphorus, iPTH and CRP were all significantly different between survivors and non-survivors. A stepwise multivariable adjusted Cox regression model demonstrated that low BUN and albumin and high CRP along with the use of calcium-containing phosphate binders (rather than sevelamer) were the strongest predictors of mortality in patients new to haemodialysis.
Conclusions.These findings suggest that non-traditional risk factors, such as inflammation and malnutrition measured during the first 18 months of dialysis, are important determinates of survival in new dialysis patients. In addition, the unique risk factor for dialysis patients, the use of calcium-containing phosphate binders, was associated with a higher mortality rate in patients new to dialysis.
Keywords: ESRD; malnutrition inflammation; mortality; phosphate binders; haemodialysis; vascular calcification
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Patients receiving dialysis therapy for end stage renal disease (ESRD) have a high cardiovascular mortality that can only be partially explained by traditional risk factors such as age, diabetes, hypertension and lipid disorders [1]. Observational analyses of several large dialysis provider databases suggest that non-traditional risk factors such as albumin, creatinine and markers of inflammation are associated with outcome in this patient population [2–4]. In addition, disorders of mineral metabolism have gained increased recognition as potentially modifiable risk factors in ESRD [5]. This includes serum calcium and phosphorus concentrations, active vitamin D sterol use, intact parathyroid hormone levels, as well as the extent of coronary artery and vascular calcification [6–8]. The purpose of the present study was to evaluate the association between baseline demographic, laboratory data and coronary and aortic calcification, as well as laboratory data and medication use in the first 18 months of dialysis and mortality in a cohort of patients new to haemodialysis who had been randomized to sevelamer or calcium-containing phosphate binders. Results from this post hoc analysis are intended to be hypothesis generating and statistically significant findings need to be confirmed by prospective trials.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
One hundred and twenty-seven patients enrolled in a prospective trial of sevelamer vs calcium-containing phosphate binders who underwent baseline electron beam computed tomography (EBCT) scans served as the study population. The baseline characteristics of the study population, the effects of phosphate binder choice on the progression of vascular calcification, and the results of the prospective randomized trial on all-cause mortality have been previously reported [9–11].
Subjects
Adults >18 years of age who were new to haemodialysis therapy were enrolled from September 2000 to December 2002. Exclusion criteria were prior history of dialysis, kidney transplant, coronary artery bypass graft, weight >300 lbs, atrial fibrillation, or life expectancy <6 months. Written informed consent was obtained from all subjects for the prospective trial and Investigational/Ethical Review Board approval was obtained for the follow-up mortality data in adherence with the Declaration of Helsinki guidelines regarding ethical principles for medical research involving human subjects. The study was closed on 31 December 2005 at which time patients were recorded as alive, dead, censored due to transplantation or lost to follow-up [11].
Laboratory and imaging techniques
Routine biochemical laboratory measurements were obtained at baseline (study entry) and at monthly or quarterly intervals as previously reported [10]. EBCT scans for imaging of coronary artery, aorta and valvular calcification were performed within the first 90 days of starting dialysis as previously described [10]. Coronary artery calcium scores (CCS), aorta calcium scores (ACS) and valvular calcium scores were calculated according to a standard method as previously described [10].
Statistical analysis
Baseline and the first 18 month characteristics were compared between the patients alive/censored at the conclusion of the study vs those who died using a two-sided t-test for continuously distributed data and the Fisher's exact test for categorical data. Since the EBCT CCS and ACS were not normally distributed, median scores were used for comparisons using the non-parametric Mann–Whitney U-test. Univariate Cox modelling was performed on individual variables. Due to the large number of significant variables and the small number of events (34), a factor analysis was performed on the 16 potential covariates. Five factors were retained, using a minimum eigenvalue of 1.0 as the criterion for retention. The five factors were then rotated to a varimax criterion of simple structure. One variable was selected from each factor for inclusion in the proportional hazards model. For those factors defined by more than one variable, the selection was made on the basis of the magnitude of the variable's loading on the factor. The factor analysis was followed by a stepwise Cox regression model with the selected variables and the binder randomization. Modelling results were scaled to clinically meaningful increments, for example, age per 10 years rather than age per year. Scaling does not change the statistical significance or P-value of a variable, only its relative hazard ratio.
Missing data points were imputed as the mean value. All P-values were two-tailed. P-values < 0.05 were considered statistically significant. Analyses were performed using GraphPad Prism version 4.02 (San Diego, CA, USA) and SAS version 8.2 (Cary, NC, USA).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Baseline data
Baseline demographics, laboratory data and CCS and ACS based on outcome are shown in Table 1. Baseline characteristics based on binder randomization have been previously published [11]. Patients who died were older (66 ± 11 vs 54 ± 15 years, P < 0.0001), had lower serum creatinine (480 ± 183 vs 587 ± 229 µmol/l; P = 0.015), lower iPTH (25 ± 23 vs 42 ± 35 pmol/l, P = 0.0048) and higher high-sensitivity C-reactive protein (hsCRP) (6.62 vs 4.08, P = 0.002) and heart rate (85 ± 14 vs 78 ± 13, P = 0.0121) than those alive/censored. Median CCS and median ACS were significantly higher in those who died compared with patients alive at follow-up (CCS: dead vs alive, 472 vs 12, P < 0.0001; ACS: dead vs alive, 1107 vs 72, P = 0.0002). The prevalence of aortic valve calcification was also higher in non-survivors.
|
Follow-up data
Outcome data were available for all 127 patients. The median follow-up was 3.6 years. Nine patients in both the sevelamer and calcium-treated groups were censored at the time of kidney transplant. Two patients randomized to sevelamer and one patient randomized to calcium were lost to follow-up. Laboratory data and dialysis treatment information on patients over the first 18 months of dialysis therapy are shown in Table 2. Diastolic blood pressure, serum albumin, serum creatinine, BUN, serum phosphorus and iPTH were all lower in patients who died compared with those alive/censored. The hsCRP was significantly greater in patients who died than in those alive at the end of study (P = 0.014). There was a significantly greater use of calcium-containing phosphate binders in patients who died vs those alive/censored (68 vs 47%, P = 0.047). The use of vitamin D analogues was borderline significant between groups with 83% use in the survivors vs 65% use in non-survivors (P = 0.051).
|
Predictive models
Predictors of all-cause mortality in a univariate Cox model are shown in Table 3. Among baseline characteristics, older age (P = 0.0020), lower BUN (P = 0.0256), higher hsCRP (P = 0.0474), higher CCS (P < 0.0001) and the presence of aortic valve calcification (P = 0.0403) were associated with mortality. During follow-up, lower time averaged serum levels of albumin, BUN, creatinine, phosphorus, iPTH and higher hsCRP as well as lower diastolic blood pressure were associated with increased mortality. The choice of phosphate binder and the use of vitamin D analogues were borderline significant in this model. Results of the rotated factor solution are shown in Table 4. Stepwise multivariate Cox regression model including binder randomization (sevelamer vs calcium) and the five factors identified by the size of the variable's loading in the factor model are shown in Table 5. The analysis demonstrated that lower BUN and serum albumin and higher hsCRP averaged over the first 18 months of dialysis, along with the use of calcium-containing phosphate binders, were the four variables predictive of mortality.
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
This is the first study to evaluate factors present at the initiation of haemodialysis including coronary artery and aorta calcification, laboratory data and medication use during the first 18 months of dialysis care to determine their association with all-cause mortality. This study finds that binder choice and the first 18 month time-averaged hsCRP and BUN are predictive of mortality. A low serum albumin was also predictive of mortality while the baseline aortic calcification score and the use of vitamin D analogues were dropped from the model. In the baseline analysis [9], ACS was inversely correlated with the baseline serum albumin and this may explain this finding. These findings are consistent with previously published literature suggesting that malnutrition and inflammation are important determinants of outcome in dialysis patients. [2–4]. The finding that the use of sevelamer was associated with a survival advantage over calcium-containing phosphate binders after adjustment for the first 18 months of laboratory values including hsCRP, BUN and albumin, strengthens the previously published findings of the randomized prospective trial showing improved survival in new dialysis patients treated with sevelamer compared with calcium-containing phosphate binders which adjusted only for differences in baseline characteristics [11].
It is interesting to note that a large number of variables at baseline or during the first 18 months of dialysis were associated with poor survival in univariate analysis. Many of these variables were not significant when other factors were included in the model including age, lower diastolic blood pressure, phosphorus and iPTH, and the administration of vitamin D analogues for a least 1 month during their first 18 months of dialysis. The large number of factors predictive of poor outcome in univariate analysis that were not significant in multivariable analysis is consistent with multiple reports in the literature and warns against the oversimplification of our understanding of factors responsible for the high mortality rate in ESRD patients. This notion again underlines the need for randomized controlled interventional trials in ESRD.
Limitations
This study was a post hoc analysis of a well-characterized cohort of new dialysis patients [9]. Due to the small sample size and small number of events (deaths) a factor analysis was utilized to identify five reasonably independent covariates for use in the Cox modelling. Furthermore, the relatively small size of the study may have resulted in negative results where true differences existed. This is particularly true for some of the traditional mortality risk factors such as diabetes which has been shown to correlate with outcome in many large database analyses [1,2]. For example, our sample size provides us with an 80% probability of finding a significant association with mortality if diabetes conferred a 2-fold increased risk of death in this incident dialysis population. The fact that diabetes did not show significance suggests a possible type II statistical error, not finding a difference when one truly exits. This is a known problem with small sample sizes and may explain some of the negative findings in this study.
Our study design measured laboratory values only for the first 18 months of dialysis. A potential limitation of this trial is that variables may change over time resulting in increased or decreased risk. However, the finding that variables measured early in the dialysis lifetime of patients are predictive of future events gives added weight to these variables.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
We demonstrated that many non-traditional risk factors such as malnutrition (low time-averaged BUN and low albumin) and inflammation (high time-averaged hsCRP and low albumin) appear to play a role in outcomes of new dialysis patients. Finally, the use of sevelamer and the avoidance of calcium-containing phosphate binders offered a survival benefit to new dialysis patients. The finding that variables present at baseline and during the first 18 months of dialysis are predictive of mortality suggests that trials designed to intervene early in the dialysis treatment course may prove more effective than interventions started after several years of dialysis treatments. This study further suggests that randomized trials designed to reduce inflammation, improve nutrition and minimize calcium loading in new dialysis patients may prove beneficial for the long-term survival of dialysis patients.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
The authors thank the clinical coordinators who dedicated themselves to the successful completion of the randomized trial. We also thank our patients who continue to participate in important clinical studies designed to advance our knowledge. We thank the dedicated dialysis staff who provides day-to-day care for our patients.
The parent study was supported by Genzyme, Corp. All aspects of the trial design, conduct and analysis including ownership of the data were the sole authority of the authors without any restrictions on publication.
Conflicts of interest statement. D.M.S. is involved with the Speakers Bureau, Genzyme Corp., P.R. is an advisor and consultant to Genzyme Corp. and G.A.B. is an advisor and consultant to Genzyme Corp.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
- Foley RN, Parfrey PS, Sarank MJ. Clinical epidemiology of cardiovascular disease in chronic renal disease. Am J Kidney Dis (1998) 32:S112–S119.[Web of Science][Medline]
- Kakiya R, Shoji T, Tsujimoto Y, et al. Body fat mass and lean mass as predictors of survival in hemodialysis patients. Kidney Int (2006) 70:549–556.[Web of Science][Medline]
- Kalantar-Zadeh K, Kopple JD, Block G, Humphreys MH. A malnutrition-inflammation score is correlated with morbidity and mortality in maintenance hemodialysis patients. Am J Kidney Dis (2001) 38:1251–1263.[Web of Science][Medline]
- Zimmerman J, Herrlinger S, Pruy A, Metzger T, Wanner C. Inflammation enhances cardiovascular risk and mortality in hemodialysis patients. Kidney Int (1999) 55:648–658.[CrossRef][Web of Science][Medline]
- Block GA, Klassen PS, Lazarus JM, Ofsthun N, Lowrie EG, Chertow GM. Mineral metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc Nephol (2004) 15:2208–2218.
[Abstract/Free Full Text] - Kalantar-Zadeh K, Kuwae N, Regidor DL, et al. Survival predictability of time-varying indicators of bone disease in maintenance hemodialysis patients. Kidney Int (2006) 70:771–780.[CrossRef][Web of Science][Medline]
- Teng M, Wolf M, Ofsthun MN, et al. Activated injectable vitamin D and hemodialysis survival: a historical cohort study. J Am Soc Nephol (2005) 16:1115–1125.
[Abstract/Free Full Text] - Matsuoka M, Iseki K, Tamashiro M, et al. Impact of high coronary artery calcification score on survival in patients on chronic hemodialysis. Clin Exp Nephrol (2004) 8:54–58.[CrossRef][Medline]
- Spiegel DM, Raggi P, Mehta R, et al. Coronary and aorta calcifications in patients new to dialysis. Hemo Int (2004) 8:265–272.[CrossRef]
- Block GA, Spiegel DM, Ehrlich J, et al. Effects of sevelamer and calcium on coronary artery calcification in patients new to hemodialysis. Kidney Int (2005) 68:1815–1824.[CrossRef][Web of Science][Medline]
- Block GA, Raggi P, Bellasi A, Kooienga L, Spiegel DM. Mortality effect of coronary calcification and phosphate binder choice in incident hemodialysis patients. Kidney Int (2007) 71:438–441.[CrossRef][Web of Science][Medline]
Accepted in revised form: 5. 6.07
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. A. Bakkaloglu, A. Saygili, L. Sever, A. Noyan, S. Akman, M. Ekim, N. Aksu, B. Doganay, N. Yildiz, A. Duzova, et al. Assessment of cardiovascular risk in paediatric peritoneal dialysis patients: a Turkish Pediatric Peritoneal Dialysis Study Group (TUPEPD) report Nephrol. Dial. Transplant., November 1, 2009; 24(11): 3525 - 3532. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Labriola, P. Wallemacq, B. Gulbis, and M. Jadoul The impact of the assay for measuring albumin on corrected ('adjusted') calcium concentrations Nephrol. Dial. Transplant., June 1, 2009; 24(6): 1834 - 1838. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. DeLoach, M. M. Joffe, X. Mai, S. Goral, and S. E. Rosas Aortic calcification predicts cardiovascular events and all-cause mortality in renal transplantation Nephrol. Dial. Transplant., April 1, 2009; 24(4): 1314 - 1319. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. D. Cohen, P. L. Kimmel, R. Neff, L. Agodoa, and K. C. Abbott Association of Incident Gout and Mortality in Dialysis Patients J. Am. Soc. Nephrol., November 1, 2008; 19(11): 2204 - 2210. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Stenvinkel, J. J. Carrero, J. Axelsson, B. Lindholm, O. Heimburger, and Z. Massy Emerging Biomarkers for Evaluating Cardiovascular Risk in the Chronic Kidney Disease Patient: How Do New Pieces Fit into the Uremic Puzzle? Clin. J. Am. Soc. Nephrol., March 1, 2008; 3(2): 505 - 521. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||