Bioavailability of Iron in Hemodialysis Patients Treated With Erythropoietin: Evidence for the Inhibitory Role of Aluminum Sandra M. Donnelly, MDCM, FRCP(C), MSc, Mahmoud A. M. Ali, MD, PhD, FRCP(C), FACp, and David N. Churchill, MDCM, FRCP(C), MSc • The dose of recombinant human erythropoietin (r-HuEPO) required to correct the anemia of end-stage renal disease (ESRD) varies among patients. The response to r-HuEPO is impaired if absolute or relative iron deficiency exists. Aluminum may cause a microcytic anemia in patients with ESRD, but the mechanism remains incompletely defined. Twenty-two patients in the Canadian Multicentre EPO trial were studied for 6 months. In this randomized double-blind placebo-controlled trial, free erythrocyte protoporphyrin (FEP) was used as an indicator of irondeficient erythropoiesis. The relationship of FEP to the estimates of iron availability (serum iron, transferrin saturation, ferritin) and iron utilization (corrected reticulocyte count, hemoglobin) was evaluated by multiple linear regression analysis. The effect of aluminum on FEP was evaluated by adjusting the statistical model for this variable. All patients were iron replete as assessed by serum ferritin. FEP wa·s not related to serum aluminum before administration of r-HuEPO, but it was significantly correlated with aluminum in the treated group. In hemodialysis patients treated with r-HuEPO, the proportion of the variability explained by the parameters of iron utilization and iron availability was 0.27. The effect of aluminum Increased this to 0.59. In hemodialysis patients not receiving r-HuEPO, the proportion of variability in FEP explained by the model increased from 0.16 to 0.28 by adjusting for aluminum. The data support the hypothesis that aluminum interferes with the bioavailability of stored iron for erythropoiesis and thus may result in a microcytic anemia in patients with ESRD or may blunt their response to r-HuEPO therapy. © 1990 by the National Kidney Foundation, Inc. INDEX WORDS: Erythropoietin; iron; aluminum; free erythrocyte protoporphyrin; hemodialysis.
T
HE DOSE OF recombinant human erythropoietin (r-HuEPO) required to correct the anemia of end-stage renal disease (ESRD) varies among patients. 1-3 The response to r-HuEPO may be impaired if iron deficiency, either absolute or relative, is present. Correlation studies of bone marrow iron and serum ferritin 4 suggest that serum ferritin values less than 30 p,g/L (30 ng/mL) are associated with absolute iron deficiency. During r-HuEPO therapy in anemic ESRD patients, Eschbach et all observed a functional or relative iron deficiency at ferritin levels indicating normal iron stores. This suggested that the rate of erythropoiesis exceeded the rate of iron mobilization from body stores. In addition to the rate of erythropoiesis, aluminum might cause functional iron deficiency. Two observations support this hypothesis. First, animal studies demonstrate plasma aluminum binding to transferrin, 5 and in vitro human erythroid culture studies indicate that transferrin must be present to elicit aluminum inhibition of erythropoiesis. 6 Second, the dose of r-HuEPO required to correct anemia in ESRD is positively correlated with post-deferoxamine serum aluminum levels. 3 Using multiple linear regression? analysis, we investigated whether serum ferritin, serum iron,
transferrin saturation, reticulocyte count, and hemoglobin could predict functional iron deficiency in 22 anemic ESRD patients participating in a randomized clinical trial using r-HuEPO.8 Functional iron deficiency was estimated from the concentration of free erythrocyte protoporphyrin (FEP) in the presence of normal serum ferritins. 9 The analysis was then adjusted for serum aluminum. METHODS
Patients Twenty-two patients, age 18 to 75 years, participating in the Canadian Multicentre EPO trial were studied. They were dialyzed thrice weekly for at least 3 months with a constant dialysis prescription. Hemoglobin values before entry were less than 90 gIL (9.0 g/dL) and less than 95 gIL (9.5 g/dL) posttransfusion. Exclusion criteria were clinical evidence of osteitis fibrosa.
From the Departments of Medicine and Clinical Epidemiology and Biostatistics, Faculty of Health Sciences, McMaster University and St. Joseph's Hospital, Hamilton, Ontario. Canada. Funded in part by a grant from St. Joseph's Hospital Foundation, Hamilton, Ontario. Canada. Address reprint requests to Sandra M. Donnelly, MD, Department of Medicine, Mount Sinai Hospital. 600 University Ave, Toronto. Ontario, Canada M5G 1X5. © 1990 by the National Kidney Foundation, Inc. 0272-6386/90/1605-0007$3.00/0
American Journal of Kidney Diseases, Vol XVI, No 5 (November), 1990: pp 447-451
447
448
DONNELLY, ALI, AND CHURCHILL
aluminum bone disease, prescription of medications known to affect hematopoiesis (eg, estrogens, androgens, cytotoxic drugs) or to predispose to bleeding (eg, anticoagulants), hepatocellular damage, inflammatory disorders, and malignancy.
Erythropietin The r-HuEPO (Ortho Pharmaceutical, Canada) is immunologically and biologically indistinguishable from human EPO.
Hematology and Biochemistry Hemoglobin concentration and red blood cell indices were measured by Coulter counter. Serum ferritin was measured by radioimmunoassay, 10 and serum iron and iron binding capacity by the method of Brozovic and Purcell." PEPs were determined by the method described by Piomelli et al. 12 Serum aluminum was analyzed by atomic absorption. 13
Protocol Design The study was a prospective randomized double-blind trial with three treatment groups: (1) placebo, (2) medium target (hemoglobin 95 to 110 gIL [9.5 to 11.0 g/dL]), and (3) high target (hemoglobin 115 to 130 gIL [11.5 to 13.0 g/dL]). The duration of the trial was 6 months. r-HuEPO was administered intravenously thrice weekly postdialysis in doses adjusted to achieve and to maintain the target hemoglobin levels. The hematological and biochemical outcomes were measured weekly, except aluminum, which was measured once within the last 2 months of the trial.
Statistical Analysis Results were calculated with the general linear model of the Statistical Analysis System. ,. With this method, a model was constructed in which PEp, the dependent variable was explained by independent variables reflecting iron supply (ferritin, serum iron, and transferrin saturation) and iron utilization (reticulocyte count, hemoglobin, and dose of r-HuEPO). The model was then adjusted for serum aluminum to evaluate its effect on iron availability. The R2 value for each multiple regression model is the fraction of the total variability in the dependent variable that is explained by the set of independent variables included in the model. The increment in R2 resulting from the addition of a new variable refects the importance of the variable. The possibility that the new factor exerts its effect through one of the variables already in the regression model is evaluated by assessing the change in the regression coefficient of these variables resulting from the addition of the new factor.
RESULTS
The clinical and biochemical features of the 22 patients, as well as their total r-HuEPO dose, are summarized in Table 1. Patient 9 experienced grand mal seizures at week 12 and patient 16 developed hypertension at week 12 and they were withdrawn from the study at that time. The remaining patients were observed for 26 weeks. All patients are included in the analysis and 19 patients are included in the adjusted model. The data were analyzed in a multiple regression
model to determine the influence of the variables reflecting iron supply and utilization on FEP and then to evaluate the effect of aluminum. Table 2 indicates the R2 value for the models, both unadjusted and adjusted for aluminum, in the r-HuEPO-treated and the placebo groups. Each model is statistically significant at a P value < 0.001. The effect of aluminum on the availability of iron is demonstrated by the dramatic increase in R2 from 0.27 to 0.59 in the r-HuEPO group and from 0.16 to 0.28 in the placebo group. As shown in Table 3, ferritin, serum iron, iron saturation, and dose of r-HuEPO were highly significant predictors of the availability of iron. Furthermore, the influence of aluminum changes only the relationship of serum iron and transferrin saturation to FEP, as indicated by the change in their regression coefficients in the adjusted model. DISCUSSION
Iron stores are proportional to ferritin in dialysis patients without hepatocellular or inflammatory disease. 4 The functional iron deficiency during rHuEPO-stimulated erythropoiesis reflects inadequate mobilization of iron from the reticuloendothelial system I or abnormal incorporation of iron by the erythron. In vitro data suggest that aluminum might impair iron bioavailability by occupying iron binding sites on transferrin. 6 Although the anemia of chronic disease is also marked by impaired bioavailability of stored iron, 15 the principal factor responsible for the anemia of ESRD is distinct from that of the anemia of chronic inflammation and is due to the lack of rHuEPO.16 The role of aluminum remains incompletely defined. FEP, considered the best marker of functional iron deficiency at the level of the erythron, was used to estimate functional iron deficiency. 9 An increase in FEP, independent of bone marrow iron stores, has been reported in uremic patients. 17 Since the degree of uremia was not altered during our study, its effects on FEP level were assumed to be constant for individual patients during this study. There is no significant correlation of serum aluminum with FEP when the degree of uremia is controlled. IS When adjusted for aluminum, the proportion of variability in the marker of relative iron deficiency (FEP) increases from 0.27 to 0.59. The relation of serum iron and transferrin saturation to FEP is
449
ALUMINUM, IRON, AND R-HuEPO IN ESRD PATIENTS
Table 1_ Patient No.
Agel Sex
60 M 2
64 F
3
21 M
4
30 M
5
33 M
6
57 M
7
34 M
8
44 F
9
62 F
10
54 M
11
30 F
12
43 F
13
55 F
14
66 M
15
59 M
16
60 M
17
67 M
18
64 M
19
46 M
20
23 M
21
26 M
22
26 M
Characteristics of Patients: Data at Baseline and 6 Months
Hgb gIL (g/dL) Pre
Post
66 (6.6) 84 (8.4) 69 (6.9) 57 (5.7) 65 (6.5) 62 (6.2) 67 (6.7) 72 (7.2) 71 (7.1) 71 (7.1) 54 (5.4) 68 (6.8) 73 (7.3) 59 (5.9) 82 (8.2) 90 (9.0) 79 (7.9) 86 (8.6) 83 (8.3) 70 (7.0) 63 (6.3) 64 (6.4)
60 (6.0) 59 (5.9) 75 (7.5) 71 (7.1 ) 63 (6.3) 62 (6.2) 62 (6.2) 71 (7.1) 113 (11.3) 111 (11.1 ) 120 (12.0) 89 (8.9) 102 (10.2) 102 (10.2) 123 (12.3) 112 (11.2) 114 (11.4) 124 (12.4) 133 (13.3) 128 (12.8) 136 (13.6) 130 (13.0)
Ferritin "gIL (or ngfmL) Pre
Post
92
174
58
122
578
235
137
354
90
318
1,903
1,405
1,413
866
33
331
532
253
438
170
1,978
1,120
111
149·
51
138
127
347
405
195
132
190
1,159
442
57
215
67
205
55
126
233
267
406
88
TIBC "mollL tl'gldL)
FEP "moI/LoRBC tl'9/dLoRBC)
Pre
Post
Pre
Post
AI nmollL tl'g/L)
r-HuEPO (u/kg)
48 (268) 53 (296) 45 (251) 46 (257) 39 (218) 30 (168) 40 (223) 47 (262) 33 (184) 38 (212) 29 (162) 50 (279) 37 (206) 43 (240) 31 (173) 36 (201) 24 (134) 43 (240) 40 (223) 48 (268) 40 (223) 32. (178)
41 (229) 31 (173) 36 (201) 39 (218) 34 (190) 32 (178) 43 (240) 36 (201) 32 (179) 39 (218) 24 (134) 47 (262) 20 (112) 34 (190) 24 (134) 27 (151) 24 (134) 35 (195) 30 (168) 35 (195) 30 (168) 36 (201)
0.77 (43.5) 1.1 (62.1 ) 0.8 (45.2) 0.7 (39.5) 1.0 (56.4) 1.2 (67.8) 0.6 (33.9) 0.6 (33.9) 0.5 (28.2) 0.6 (33.9) 0.5 (28.2) 1.6 (90.4) 1.2 (67.8) 0.6 (33.9) 0.5 (28.2) 1.21 (68.4) 0.6 (33.9) 1.1 (62.1 ) 0.6 (33.9) 0.7 (39.5) 0.6 (33.9) 0.5 (28.2)
1.1 (62.1 ) 0.7 (39.5) 0.6 (33.9) 0.5 (28.2) 0.6 (33.9) 1.2 (67.8) 0.5 (16.9) 0.3 (16.9) NA
2,447 (66.0) 2,189 (59.1) 598 (16.1 ) 2,779 (75.0) 1,867 (50.4) 268 (7.2) 616 (16.6) 597 (16.1) NA
0
NA
1.5 (84.7) 0.9 (58.8) 2.0 (113.0) 1.4 (79.1) 0.5 (28.2) 1.1 (62.1 ) 1.9 (107.3) 1.7 (50.8) 0.8 (45.2) 0.9 (50.8) 0.7 (39.5) 0.7 (39.5) 0.8 (45.2)
NA
6,400
301 (8.1) 10,562 (285) 1,524 (41.1 ) 2,052 (55.4) 603 (16.3) NA
6,800
6,116 (39.8) 1,458 (39.3) 1,477 (39.8) 3,728 (100.6) 1,572 (42.4) 691 (18.6)
0 0 0 0 0 0 0
14,100 11,450 3,750 5,200 NA 8,400 11,100 8,325 5,650 12,200 8,275
Abbreviation: NA, not available.
Table 2. R2 for the Models Unadjusted and Adjusted for Aluminum Group
r-HuEPO Placebo *P < 0.001.
Unadjusted Rt
Adjusted for Aluminum R2
.27* .16*
.59* .28*
modified by aluminum and thus the effect of aluminum on iron availability is mediated, in part, through these factors. These are the same factors, which in the presence of a serum ferritin indicative of normal iron stores, characterize the functional iron deficiency as described by Eschbach. 16 These data support the observation that higher
450
DONNELLY, ALI, AND CHURCHILL Table 3.
Regression Coefficients of the Independent Variables Unadjusted
Group
EPO
Placebo
Variables in the Model
Regression Coefficient ± SEE
Adjusted for Aluminum Regression Coefficient ± SEE
P Value'
Ferritin Iron Saturation Hemoglobin Reticulocytes r-HuEPO dose
-2.9 -5.2 1.8 -2.7 -2.4 1.0
± ± ± ± ± ±
1.0 1.4 0.5 1.6 5.9 0.2
x 10-4 x 10-2
Ferritin Iron Saturation Hemoglobin Reticulocytes
-2.0 -1.7 3.7 3.4 -0.8
± ± ± ± ±
0.5 0.8 3.6 3.5 1.0
x x x x x
P Value'
-2.9 -8.3 2.8 -2.4 -1 .0 1.0
± ± ± ± ± ±
0.9 1.1 0.4 1.3 4.6 0.1
x 10-4 x 10-2
x 10-3 x 10-4 x 10-3
0.0147 0.0004 0.0008 0.1013 0.6910 0.0001
10-4 10-2 10-1 10-3 10-3
0.0003 0.0515 0.2972 0.3214 0.4061
-3.4 -2.1 8.2 8.1 -0.3
± ± ± ± ±
0.7 0.8 3.4 3.5 1.1
x x x x x
x 10-3 x 10-4 x 10-3
0.0016 0.0001 0.0001 0.0682 0.8175 0.0001
10-4 10-2 10-1 10-3 10-3
0.0001 0.0091 0.0177 0.0236 0.8110
• P values indicate the probability that the regression coefficient equals zero. Abbreviation: SEE, standard error of the estimate.
serum aluminum levels post-deferoxamine correlate with a higher dose of r-HuEPO for the correction of anemia in ESRD. 3 Furthermore, it suggests that the pathophysiological mechanism for the effect on erythropoiesis is, in part, an interference of aluminum with the bioavailability of iron. Finally, the clinical response to aluminum chelation therapy in one of these patients, which demonstrates an improved availability of iron in spite of decreasing serum iron levels, supports this hypothesis. 19 The clinical importance is that aluminum may blunt the effect of and thus greatly increase the dose (and cost) of r-HuEPO required to treat
ESRD patients. The critical body burden of aluminum at which this effect is clinically important and requires DFO treatment remains to be determined. In conclusion, the functional iron deficiency observed in patients treated with r-HuEPO is due, in part, to the effects of aluminum on iron kinetics .
ACKNOWLEDGMENT We thank Ortho Pharmaceuticals and the Canadian Erythropoietin Study Group for providing the framework and drug to execute this study, L. Spicer for assisting in data tabulation and analysis, and A. O'Sullivan for preparing the manuscript.
REFERENCES I. Eschbach JW, Egrie JC, Downing MR, et al: Correction of the anemia of end-stage renal disease with recombinant human erythropoietin . N Engl J Med 316:73-78, 1987
2. Winearls CG, Oliver DO, Pippard MJ, et al: Effect of human erythropoietin derived from recombinant DNA on the anaemia of patients maintained by chronic haemodialysis. Lancet 2:1175-1177,1986 3. Casati S, Passerini P, Campise MR, et al: Benefits and risks of protracted treatment with human recombinant erythropoietin in patients having haemodialysis. Br Med J 295:10171020, 1987 4. Aljama P, Ward MK, Pierides AM, et al: Serum ferritin concentration: A reliable guide to iron overload in uremic and hemodialyzed patients. Clin Nephrol 10:101-104, 1978 5. Trapp GA: Plasma aluminum is bound to transferrin. Life Sci 33:311-316, 1983
6. Mladenovic J: Aluminum inhibits erythropoiesis in vitro. J Clin Invest 81: 1661-1665, 1988 7. Kleinbaum DG, Kupper LL, Muller KE: Applied Regression Analysis and Other Multivariable Methods (ed 2) . Boston, MA, PWS Kent, 1988, pp 102-144 8. Canadian Erythropoietin Study Group: Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving haemodialysis. Br Med J 300:573-578, 1990 9. Langer EE, Haining RG, Labbe RF, et al: Erythrocyte protoporphyrin. Blood 40: 112-128, 1972 10. Luxton AW, Walker WHC, Goldie J, et al: A radioimmunoassay for serum ferritin. Clin Chern 23 :683-689, 1977 11. Brozovic B, Purcell Y: An automated micro method for measuring iron concentration in serum using thioglycollic acid and bathophenantroline sulphonate. J Clin Pathol 27 :222-226, 1974
ALUMINUM, IRON, AND R·HuEPO IN ESRD PATIENTS
451
12. Piomelli S, Young P, Gay G: A micro method for free erythrocyte protoporphyrin: The FEP test. J Lab Clin Med 81 :932-940, 1973 13. Brodie KG, Routh MW: Trace analysis of lead in blood, aluminum and manganese in serum and chromium in urine by graphite furnace atomic absorption spectrometry. Clin Biochem 17:19-26,1984 14. SAS Institute Inc.: SAS User's Guide: Statistics, 1982 edition. Cary, NC, SAS Institute, 1982 15. Cartwright GE, Lee GK: Anaemia of chronic disorders. Br J HaematoI21:147-152, 1971 16. Eschbach JW: The anemia of chronic renal failure: Pathophysiology and the effects of recombinant erythropoietin. Kidney Int 35: 134-148, 1989
17. Moreb J, Popovizer MM, et al: Evaluation of iron status in patients on chronic hemodialysis: Relative usefulness of bone marrow hemosiderin, serum ferritin, transferrin saturation, mean corpuscular volume and red cell protoporphyrin. Nephron 35 :196-200, 1983 18. Buchet Jp, Lauwerys R, Hassoun A, et al: Effect of aluminum on porphyrin metabolism in hemodialyzed patients. Nephron 46:360-363, 1987 19. Donnelly SM, Smith EKM: The role of aluminum in the functional iron deficiency of patients treated with erythropoietin: Clinical characteristics and response to treatment. A case report. Am J Kidney Dis 16:487-490, 1990
T
HE DOSE OF recombinant human erythropoietin (r-HuEPO) required to correct the anemia of end-stage renal disease (ESRD) varies among patients. 1-3 The response to r-HuEPO may be impaired if iron deficiency, either absolute or relative, is present. Correlation studies of bone marrow iron and serum ferritin 4 suggest that serum ferritin values less than 30 p,g/L (30 ng/mL) are associated with absolute iron deficiency. During r-HuEPO therapy in anemic ESRD patients, Eschbach et all observed a functional or relative iron deficiency at ferritin levels indicating normal iron stores. This suggested that the rate of erythropoiesis exceeded the rate of iron mobilization from body stores. In addition to the rate of erythropoiesis, aluminum might cause functional iron deficiency. Two observations support this hypothesis. First, animal studies demonstrate plasma aluminum binding to transferrin, 5 and in vitro human erythroid culture studies indicate that transferrin must be present to elicit aluminum inhibition of erythropoiesis. 6 Second, the dose of r-HuEPO required to correct anemia in ESRD is positively correlated with post-deferoxamine serum aluminum levels. 3 Using multiple linear regression? analysis, we investigated whether serum ferritin, serum iron,
transferrin saturation, reticulocyte count, and hemoglobin could predict functional iron deficiency in 22 anemic ESRD patients participating in a randomized clinical trial using r-HuEPO.8 Functional iron deficiency was estimated from the concentration of free erythrocyte protoporphyrin (FEP) in the presence of normal serum ferritins. 9 The analysis was then adjusted for serum aluminum. METHODS
Patients Twenty-two patients, age 18 to 75 years, participating in the Canadian Multicentre EPO trial were studied. They were dialyzed thrice weekly for at least 3 months with a constant dialysis prescription. Hemoglobin values before entry were less than 90 gIL (9.0 g/dL) and less than 95 gIL (9.5 g/dL) posttransfusion. Exclusion criteria were clinical evidence of osteitis fibrosa.
From the Departments of Medicine and Clinical Epidemiology and Biostatistics, Faculty of Health Sciences, McMaster University and St. Joseph's Hospital, Hamilton, Ontario. Canada. Funded in part by a grant from St. Joseph's Hospital Foundation, Hamilton, Ontario. Canada. Address reprint requests to Sandra M. Donnelly, MD, Department of Medicine, Mount Sinai Hospital. 600 University Ave, Toronto. Ontario, Canada M5G 1X5. © 1990 by the National Kidney Foundation, Inc. 0272-6386/90/1605-0007$3.00/0
American Journal of Kidney Diseases, Vol XVI, No 5 (November), 1990: pp 447-451
447
448
DONNELLY, ALI, AND CHURCHILL
aluminum bone disease, prescription of medications known to affect hematopoiesis (eg, estrogens, androgens, cytotoxic drugs) or to predispose to bleeding (eg, anticoagulants), hepatocellular damage, inflammatory disorders, and malignancy.
Erythropietin The r-HuEPO (Ortho Pharmaceutical, Canada) is immunologically and biologically indistinguishable from human EPO.
Hematology and Biochemistry Hemoglobin concentration and red blood cell indices were measured by Coulter counter. Serum ferritin was measured by radioimmunoassay, 10 and serum iron and iron binding capacity by the method of Brozovic and Purcell." PEPs were determined by the method described by Piomelli et al. 12 Serum aluminum was analyzed by atomic absorption. 13
Protocol Design The study was a prospective randomized double-blind trial with three treatment groups: (1) placebo, (2) medium target (hemoglobin 95 to 110 gIL [9.5 to 11.0 g/dL]), and (3) high target (hemoglobin 115 to 130 gIL [11.5 to 13.0 g/dL]). The duration of the trial was 6 months. r-HuEPO was administered intravenously thrice weekly postdialysis in doses adjusted to achieve and to maintain the target hemoglobin levels. The hematological and biochemical outcomes were measured weekly, except aluminum, which was measured once within the last 2 months of the trial.
Statistical Analysis Results were calculated with the general linear model of the Statistical Analysis System. ,. With this method, a model was constructed in which PEp, the dependent variable was explained by independent variables reflecting iron supply (ferritin, serum iron, and transferrin saturation) and iron utilization (reticulocyte count, hemoglobin, and dose of r-HuEPO). The model was then adjusted for serum aluminum to evaluate its effect on iron availability. The R2 value for each multiple regression model is the fraction of the total variability in the dependent variable that is explained by the set of independent variables included in the model. The increment in R2 resulting from the addition of a new variable refects the importance of the variable. The possibility that the new factor exerts its effect through one of the variables already in the regression model is evaluated by assessing the change in the regression coefficient of these variables resulting from the addition of the new factor.
RESULTS
The clinical and biochemical features of the 22 patients, as well as their total r-HuEPO dose, are summarized in Table 1. Patient 9 experienced grand mal seizures at week 12 and patient 16 developed hypertension at week 12 and they were withdrawn from the study at that time. The remaining patients were observed for 26 weeks. All patients are included in the analysis and 19 patients are included in the adjusted model. The data were analyzed in a multiple regression
model to determine the influence of the variables reflecting iron supply and utilization on FEP and then to evaluate the effect of aluminum. Table 2 indicates the R2 value for the models, both unadjusted and adjusted for aluminum, in the r-HuEPO-treated and the placebo groups. Each model is statistically significant at a P value < 0.001. The effect of aluminum on the availability of iron is demonstrated by the dramatic increase in R2 from 0.27 to 0.59 in the r-HuEPO group and from 0.16 to 0.28 in the placebo group. As shown in Table 3, ferritin, serum iron, iron saturation, and dose of r-HuEPO were highly significant predictors of the availability of iron. Furthermore, the influence of aluminum changes only the relationship of serum iron and transferrin saturation to FEP, as indicated by the change in their regression coefficients in the adjusted model. DISCUSSION
Iron stores are proportional to ferritin in dialysis patients without hepatocellular or inflammatory disease. 4 The functional iron deficiency during rHuEPO-stimulated erythropoiesis reflects inadequate mobilization of iron from the reticuloendothelial system I or abnormal incorporation of iron by the erythron. In vitro data suggest that aluminum might impair iron bioavailability by occupying iron binding sites on transferrin. 6 Although the anemia of chronic disease is also marked by impaired bioavailability of stored iron, 15 the principal factor responsible for the anemia of ESRD is distinct from that of the anemia of chronic inflammation and is due to the lack of rHuEPO.16 The role of aluminum remains incompletely defined. FEP, considered the best marker of functional iron deficiency at the level of the erythron, was used to estimate functional iron deficiency. 9 An increase in FEP, independent of bone marrow iron stores, has been reported in uremic patients. 17 Since the degree of uremia was not altered during our study, its effects on FEP level were assumed to be constant for individual patients during this study. There is no significant correlation of serum aluminum with FEP when the degree of uremia is controlled. IS When adjusted for aluminum, the proportion of variability in the marker of relative iron deficiency (FEP) increases from 0.27 to 0.59. The relation of serum iron and transferrin saturation to FEP is
449
ALUMINUM, IRON, AND R-HuEPO IN ESRD PATIENTS
Table 1_ Patient No.
Agel Sex
60 M 2
64 F
3
21 M
4
30 M
5
33 M
6
57 M
7
34 M
8
44 F
9
62 F
10
54 M
11
30 F
12
43 F
13
55 F
14
66 M
15
59 M
16
60 M
17
67 M
18
64 M
19
46 M
20
23 M
21
26 M
22
26 M
Characteristics of Patients: Data at Baseline and 6 Months
Hgb gIL (g/dL) Pre
Post
66 (6.6) 84 (8.4) 69 (6.9) 57 (5.7) 65 (6.5) 62 (6.2) 67 (6.7) 72 (7.2) 71 (7.1) 71 (7.1) 54 (5.4) 68 (6.8) 73 (7.3) 59 (5.9) 82 (8.2) 90 (9.0) 79 (7.9) 86 (8.6) 83 (8.3) 70 (7.0) 63 (6.3) 64 (6.4)
60 (6.0) 59 (5.9) 75 (7.5) 71 (7.1 ) 63 (6.3) 62 (6.2) 62 (6.2) 71 (7.1) 113 (11.3) 111 (11.1 ) 120 (12.0) 89 (8.9) 102 (10.2) 102 (10.2) 123 (12.3) 112 (11.2) 114 (11.4) 124 (12.4) 133 (13.3) 128 (12.8) 136 (13.6) 130 (13.0)
Ferritin "gIL (or ngfmL) Pre
Post
92
174
58
122
578
235
137
354
90
318
1,903
1,405
1,413
866
33
331
532
253
438
170
1,978
1,120
111
149·
51
138
127
347
405
195
132
190
1,159
442
57
215
67
205
55
126
233
267
406
88
TIBC "mollL tl'gldL)
FEP "moI/LoRBC tl'9/dLoRBC)
Pre
Post
Pre
Post
AI nmollL tl'g/L)
r-HuEPO (u/kg)
48 (268) 53 (296) 45 (251) 46 (257) 39 (218) 30 (168) 40 (223) 47 (262) 33 (184) 38 (212) 29 (162) 50 (279) 37 (206) 43 (240) 31 (173) 36 (201) 24 (134) 43 (240) 40 (223) 48 (268) 40 (223) 32. (178)
41 (229) 31 (173) 36 (201) 39 (218) 34 (190) 32 (178) 43 (240) 36 (201) 32 (179) 39 (218) 24 (134) 47 (262) 20 (112) 34 (190) 24 (134) 27 (151) 24 (134) 35 (195) 30 (168) 35 (195) 30 (168) 36 (201)
0.77 (43.5) 1.1 (62.1 ) 0.8 (45.2) 0.7 (39.5) 1.0 (56.4) 1.2 (67.8) 0.6 (33.9) 0.6 (33.9) 0.5 (28.2) 0.6 (33.9) 0.5 (28.2) 1.6 (90.4) 1.2 (67.8) 0.6 (33.9) 0.5 (28.2) 1.21 (68.4) 0.6 (33.9) 1.1 (62.1 ) 0.6 (33.9) 0.7 (39.5) 0.6 (33.9) 0.5 (28.2)
1.1 (62.1 ) 0.7 (39.5) 0.6 (33.9) 0.5 (28.2) 0.6 (33.9) 1.2 (67.8) 0.5 (16.9) 0.3 (16.9) NA
2,447 (66.0) 2,189 (59.1) 598 (16.1 ) 2,779 (75.0) 1,867 (50.4) 268 (7.2) 616 (16.6) 597 (16.1) NA
0
NA
1.5 (84.7) 0.9 (58.8) 2.0 (113.0) 1.4 (79.1) 0.5 (28.2) 1.1 (62.1 ) 1.9 (107.3) 1.7 (50.8) 0.8 (45.2) 0.9 (50.8) 0.7 (39.5) 0.7 (39.5) 0.8 (45.2)
NA
6,400
301 (8.1) 10,562 (285) 1,524 (41.1 ) 2,052 (55.4) 603 (16.3) NA
6,800
6,116 (39.8) 1,458 (39.3) 1,477 (39.8) 3,728 (100.6) 1,572 (42.4) 691 (18.6)
0 0 0 0 0 0 0
14,100 11,450 3,750 5,200 NA 8,400 11,100 8,325 5,650 12,200 8,275
Abbreviation: NA, not available.
Table 2. R2 for the Models Unadjusted and Adjusted for Aluminum Group
r-HuEPO Placebo *P < 0.001.
Unadjusted Rt
Adjusted for Aluminum R2
.27* .16*
.59* .28*
modified by aluminum and thus the effect of aluminum on iron availability is mediated, in part, through these factors. These are the same factors, which in the presence of a serum ferritin indicative of normal iron stores, characterize the functional iron deficiency as described by Eschbach. 16 These data support the observation that higher
450
DONNELLY, ALI, AND CHURCHILL Table 3.
Regression Coefficients of the Independent Variables Unadjusted
Group
EPO
Placebo
Variables in the Model
Regression Coefficient ± SEE
Adjusted for Aluminum Regression Coefficient ± SEE
P Value'
Ferritin Iron Saturation Hemoglobin Reticulocytes r-HuEPO dose
-2.9 -5.2 1.8 -2.7 -2.4 1.0
± ± ± ± ± ±
1.0 1.4 0.5 1.6 5.9 0.2
x 10-4 x 10-2
Ferritin Iron Saturation Hemoglobin Reticulocytes
-2.0 -1.7 3.7 3.4 -0.8
± ± ± ± ±
0.5 0.8 3.6 3.5 1.0
x x x x x
P Value'
-2.9 -8.3 2.8 -2.4 -1 .0 1.0
± ± ± ± ± ±
0.9 1.1 0.4 1.3 4.6 0.1
x 10-4 x 10-2
x 10-3 x 10-4 x 10-3
0.0147 0.0004 0.0008 0.1013 0.6910 0.0001
10-4 10-2 10-1 10-3 10-3
0.0003 0.0515 0.2972 0.3214 0.4061
-3.4 -2.1 8.2 8.1 -0.3
± ± ± ± ±
0.7 0.8 3.4 3.5 1.1
x x x x x
x 10-3 x 10-4 x 10-3
0.0016 0.0001 0.0001 0.0682 0.8175 0.0001
10-4 10-2 10-1 10-3 10-3
0.0001 0.0091 0.0177 0.0236 0.8110
• P values indicate the probability that the regression coefficient equals zero. Abbreviation: SEE, standard error of the estimate.
serum aluminum levels post-deferoxamine correlate with a higher dose of r-HuEPO for the correction of anemia in ESRD. 3 Furthermore, it suggests that the pathophysiological mechanism for the effect on erythropoiesis is, in part, an interference of aluminum with the bioavailability of iron. Finally, the clinical response to aluminum chelation therapy in one of these patients, which demonstrates an improved availability of iron in spite of decreasing serum iron levels, supports this hypothesis. 19 The clinical importance is that aluminum may blunt the effect of and thus greatly increase the dose (and cost) of r-HuEPO required to treat
ESRD patients. The critical body burden of aluminum at which this effect is clinically important and requires DFO treatment remains to be determined. In conclusion, the functional iron deficiency observed in patients treated with r-HuEPO is due, in part, to the effects of aluminum on iron kinetics .
ACKNOWLEDGMENT We thank Ortho Pharmaceuticals and the Canadian Erythropoietin Study Group for providing the framework and drug to execute this study, L. Spicer for assisting in data tabulation and analysis, and A. O'Sullivan for preparing the manuscript.
REFERENCES I. Eschbach JW, Egrie JC, Downing MR, et al: Correction of the anemia of end-stage renal disease with recombinant human erythropoietin . N Engl J Med 316:73-78, 1987
2. Winearls CG, Oliver DO, Pippard MJ, et al: Effect of human erythropoietin derived from recombinant DNA on the anaemia of patients maintained by chronic haemodialysis. Lancet 2:1175-1177,1986 3. Casati S, Passerini P, Campise MR, et al: Benefits and risks of protracted treatment with human recombinant erythropoietin in patients having haemodialysis. Br Med J 295:10171020, 1987 4. Aljama P, Ward MK, Pierides AM, et al: Serum ferritin concentration: A reliable guide to iron overload in uremic and hemodialyzed patients. Clin Nephrol 10:101-104, 1978 5. Trapp GA: Plasma aluminum is bound to transferrin. Life Sci 33:311-316, 1983
6. Mladenovic J: Aluminum inhibits erythropoiesis in vitro. J Clin Invest 81: 1661-1665, 1988 7. Kleinbaum DG, Kupper LL, Muller KE: Applied Regression Analysis and Other Multivariable Methods (ed 2) . Boston, MA, PWS Kent, 1988, pp 102-144 8. Canadian Erythropoietin Study Group: Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving haemodialysis. Br Med J 300:573-578, 1990 9. Langer EE, Haining RG, Labbe RF, et al: Erythrocyte protoporphyrin. Blood 40: 112-128, 1972 10. Luxton AW, Walker WHC, Goldie J, et al: A radioimmunoassay for serum ferritin. Clin Chern 23 :683-689, 1977 11. Brozovic B, Purcell Y: An automated micro method for measuring iron concentration in serum using thioglycollic acid and bathophenantroline sulphonate. J Clin Pathol 27 :222-226, 1974
ALUMINUM, IRON, AND R·HuEPO IN ESRD PATIENTS
451
12. Piomelli S, Young P, Gay G: A micro method for free erythrocyte protoporphyrin: The FEP test. J Lab Clin Med 81 :932-940, 1973 13. Brodie KG, Routh MW: Trace analysis of lead in blood, aluminum and manganese in serum and chromium in urine by graphite furnace atomic absorption spectrometry. Clin Biochem 17:19-26,1984 14. SAS Institute Inc.: SAS User's Guide: Statistics, 1982 edition. Cary, NC, SAS Institute, 1982 15. Cartwright GE, Lee GK: Anaemia of chronic disorders. Br J HaematoI21:147-152, 1971 16. Eschbach JW: The anemia of chronic renal failure: Pathophysiology and the effects of recombinant erythropoietin. Kidney Int 35: 134-148, 1989
17. Moreb J, Popovizer MM, et al: Evaluation of iron status in patients on chronic hemodialysis: Relative usefulness of bone marrow hemosiderin, serum ferritin, transferrin saturation, mean corpuscular volume and red cell protoporphyrin. Nephron 35 :196-200, 1983 18. Buchet Jp, Lauwerys R, Hassoun A, et al: Effect of aluminum on porphyrin metabolism in hemodialyzed patients. Nephron 46:360-363, 1987 19. Donnelly SM, Smith EKM: The role of aluminum in the functional iron deficiency of patients treated with erythropoietin: Clinical characteristics and response to treatment. A case report. Am J Kidney Dis 16:487-490, 1990
