1. ALBUMIN IS AN EFFECT MODIFIER IN THE RELATIONSHIP BETWEEN ERYTHROPOIETIN STIMULATING AGENT AND MORTALITY IN HEMODIALYSIS PATIENTS.
Satoshi Mikami1, Takayuki Hamano2, Osamu Iba1, Takuya Inoue1,
Mitsunobu Toki1, Hiroshi Mikami1, Yoshihiro Takamitsu1, Masamitsu Fujii1
Department of Internal medicine Higashikohri Hospital
Clinical Epidemiology Unit, Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine

2. Introduction
Anemia is a major complication of hemodialysis patients affecting future outcome. Erythropoiesis-stimulating agents (ESAs) have contributed to the care of anemia in this population. The response of hemoglobin to ESAs is dose-dependent, however, it varies between individuals. Although some studies have shown that this difference affects the prognosis of hemodialysis patients, whether the impact of ESA responsiveness on mortality is modified by other factors has not yet been well understood.

3. Aim
In this research, we studied the impact of ESA responsiveness on mortality in hemodialysis patients.
Subjects
We performed a retrospective observational study. The study period was from November, 2005 to December, We enrolled 195 patients in our facility, who had received hemodialysis for more than 1 year. All ESA users had received for at least 6 months. We excluded those who were followed up for 2 weeks or less in this study.

4. Method TSAT= ESAI= Serum Fe (mg/dL) ×100
The outcome of interest was all-cause death. We converted the dose of Darbepoetin into EPO by multiplying 200 times. To evaluate the response to ESA treatment we used the ESA index (ESAI).
Covariates were age, hemodialysis vintage, hemoglobin (Hb), hematocrit (Ht), total cholesterol (T-chol), C-reactive protein (CRP), creatinine (Cr), blood urea nitrogen (BUN), albumin (Alb), ferritin, transferrin saturation (TSAT) and ESAI. TSAT and ESAI were calculated by the following formula.
Serum Fe (mg/dL)
TSAT=
×100
Total Iron Binding Capacity (mg/dL)
Weekly dose of ESA (U)/W 1)
ESAI=
Dry weight (kg)×Hb (g/dL)
1) López-Gómez JM, et al. Kidney Int Suppl Dec;(111):S75-81

5. Statistical analyses
Variables with non-normal distribution (e.g. Ferritin and CRP) were logarithmically transformed prior to analyses.
Kap/// curves were drown to compared the survival rate followed by the Log-rank method.
In multivariable analyses, we employed cox proportional hazards model and formally tested the interaction between ESAI and covariates (ESAI*covariate). When interaction was significant, we performed stratified analyses by the covariate.
Covariates included in multivariable analyses, were decided based on univariate analysis. P values less than 0.05 were considered significant. With regard to interaction term, P values less than 0.15 were considered significant.

6. Results ①
Sixty-five out of 195 (33%) patients died during the observational period. The ESAI ranged from 0 to 41.0 with the median value of 8.03 U/kg・Hb, by which we classified the subjects into two groups, the baseline characteristic of High ESAI Group demonstrated significantly higher proportion of female, with older age, higher dose of EPO, lower Hb, lower Cr, and lower Alb levels. (Table 1)
Kaplan-Meier curves showed a significant higher mortality in the High ESAI Group. When divided by Alb level at 3.5 g/dL and by Hb at 10.0 g/dL, Low Alb and Low Hb Group demonstrated significantly worse outcome, respectively. (Figure 1)
Univariate analysis showed that mortality was associated with older age, lower Hb, lower Cr, lower BUN, lower Alb, higher CRP, and higher ESAI. (Table 2)

7. Results ②
Because ESAI has Hb in its formula and has higher HR in univariate analysis, we selected ESAI instead of Hb as independent variable in the multivariable analysis. In the same reason, we introduced Cr instead of BUN into the model.
The multivariate analysis showed that higher ESAI, as a continuous variable, was significant contributor to mortality, as well as older age, lower Alb, and lower Cr. As there was a significant interaction between ESAI and Alb, we divided the subjects into Low and High-Alb Groups. In High-Alb Group, low ESAI was associated with longer patient survival. However, no relationship was found between ESAI and mortality in Low-Alb Group. (Table 3)
We then stratified the subjects into four groups by combination of ESAI (High / Low) and Alb (High / Low). Low-ESAI and High-Alb Group had a significantly lower mortality than the other three groups. (Figure 2)

8. Table 1: baseline characteristic
Compared between High and Low-ESAI group.
Data are mean±S.D. and median [range]
* vs Low ESAI p<0.05

9. Figure 1:Survival rate of patients according to ESAI, Hb and Alb.
0.0
0.2
0.4
0.6
0.8
1.0 10 20 30 40 50 60
High ESAI
Low ESAI
Survival rate
Month
High Alb
Low Alb
0.0
0.2
0.4
0.6
0.8
1.0 10 20 30 40 50 60
Survival rate
Month ** **
High Hb
Low Hb
0.0
0.2
0.4
0.6
0.8
1.0 10 20 30 40 50 60
Survival rate
Month
**Log-rank P<0.05 **

10. Table 2: Univariate Analysis for mortality
Hazard ratio (95% CI) P
Age (1year)
Log (HD vintage)
gender (0=M, 1=F)
ESAI/SD
Hb/SD
Cr/SD
BUN/SD
T-chol (mg/dL)
Alb (g/dL)
Log (CRP)
Log (Ferritin)
TSAT (%)
1.06 ( )
0.85 ( )
1.57 ( )
1.74 ( )
0.57 ( )
0.42 ( )
0.54 ( )
1.00 ( )
0.15 ( )
1.46 ( )
1.17 ( )
0.99 ( )
<0.01
n.s.

11. Table 3: Multivariable Analysis for mortality after divided into Low and High-Alb Group.
Total
Hazard ratio (95% CI) P
Age (1year)
gender (0=M, 1=F) Cr
Alb
ESAI
Alb * ESAI
1.03 ( )
1.07 ( )
0.90 ( )
0.31 ( )
1.06 ( )
<0.01
n.s.
<0.05
0.12
High-Alb Group
Hazard ratio (95% CI) P
Age (1year)
gender (0=M, 1=F) Cr
ESAI
1.04 ( )
1.18 ( )
0.87 ( )
1.10 ( )
<0.05
n.s.
<0.01
Low-Alb Group
Hazard ratio (95% CI) P
Age (1year)
gender (0=M, 1=F) Cr
ESAI
1.04 ( )
0.77 ( )
0.87 ( )
1.02 ( )
<0.05
n.s.

12. Figure 2: Mortality of patients in 4 group by ESAI and Alb.
1.0 **
0.8
0.6
0.4
0.2
**Log-rank P<0.05
0.0 10 20 30 40 50 60
Month
Low-ESAI and High-Alb
Low-ESAI and Low-Alb
High-ESAI and High-Alb
High-ESAI and Low-Alb

13. Conclusion
Our data suggest that Alb is an effect modifier in the relationship between ESAI and mortality, and that ESAI can be used in risk stratification in patients without malnutrition.
Iron deficiency and the inflammatory state, as the causes of impaired response to ESA2,3). However, ESAI was not related to the parameter of ferrokinetics or inflammation, because there were many missing values.
2) Horl WH et al. Nephrol Dial Transplant 2000; l5(Suppl 4): 43–50.
3) Macdougall IC et al. Nephrol Dial Transplant 2002; l7(Suppl I): 48–52.