Article Text

Download PDFPDF

Unanticipated Favorable Effects of Correcting Iron Deficiency in Chronic Hemodialysis Patients
  1. Victor E. Pollak,
  2. Jonathan A. Lorch,
  3. Robert T. Means Jr
  1. From the Department of Medicine (V.E.P.), University of Colorado Health Sciences Center, Denver
  2. MIQS, Inc (V.E.P.), Boulder, Colo
  3. Department of Medicine (J.A.L.), Saint Luke's Hospital and Columbia University College of Physicians and Surgeons, New York
  4. Hematology/Oncology Division (R.T.M.), Medical University of South Carolina and the Ralph H. Johnson Veterans Administration Medical Center, Charleston, SC.
  1. Address correspondence to: Victor E. Pollak, MD, MIQS, Inc, 5621-D Arapahoe Avenue, Boulder, CO 80303. E-mail vpollak{at}
  2. Presented at the 32nd Annual Meeting of the American Society of Nephrology, Miami Beach, Florida, on November 5, 1999.


Background Correction of anemia in hemodialysis patients is seldom completely attained, and the response of parameters other than hemoglobin concentration to anemia correction has not been evaluated in detail.

Methods Laboratory parameters that suggest iron deficiency occurred in 10-15% of 206 recombinant human erythropoietin (rhEPO)-treated patients. Oral iron was given for 9 months and intravenous iron thereafter on a patient-specific basis when iron deficiency was evident. Eighty-seven hemodialysis patients with data for 12 months were followed for another 12 months. A computerized information system enabled data management and analysis.

Results With oral iron, serum ferritin decreased (P<0.001), indicating further iron depletion. With intravenous iron, hemoglobin increased, evidence of iron deficiency decreased, and less rhEPO was needed. Striking macrocytosis appeared. Serum albumin and serum creatinine/kg body weight (an index of muscle mass) increased, while blood pressure decreased. Data were reanalyzed in four mean corpuscular volume (MCV) quartiles and two ferritin subsets at study onset. Iron deficient erythropoiesis (low MCV, mean corpuscular hemoglobin [MCH], and transferrin saturation) was striking in quartile 1; low ferritin was prevalent in all quartiles. With intravenous iron, hemoglobin increased only in quartile 1, the quartile with the greatest decrease (52%) in rhEPO dose. MCV increased in all quartiles (P<0.001). Serum albumin increased in all MCV quartiles and both ferritin subsets, but significant creatinine/kg increase and blood pressure decrease occurred only in the low-ferritin subset.

Conclusions Macrocytosis occurred with intravenous iron replacement. The universal MCV increase suggests unrecognized, inadequately treated, folic acid deficiency unmasked by an adequate iron supply. There was also improved well being. Effects were most clearly evident in patients with deficient iron stores.

Statistics from


In chronic maintenance dialysis, the need for rhEPO is widely recognized1,2as is the need for supplemental iron.1,3–7Yet, the mean hematocrit of hemodialysis patients in the United States in 1993 (0.301) increased only to 0.327 by 1996, while 28% of patients had hematocrits <0.30.8Clinical trials usually focus on selected populations and are not easily translated into practice, as shown recently for rhEPO use in patients treated by dialysis.9Because dialysis patients have diverse underlying diseases and comorbid variables, we postulated that a long-term observational approach might facilitate understanding of anemia management.

This study describes the use of an observational approach to improve timeliness and accuracy of data collection as well as analysis in anemia management in dialysis patients. Objectives were to identify and correct iron deficiency, and to assess effects of interventions on patient well being. Oral iron supplements failed to correct anemia and iron deficiency, but intravenous iron supplements led to greater improvement than anticipated. Analysis of dynamic changes over time made it possible to identify hematologic parameters that predict iron deficiency and response to intravenous iron administration. Unexpectedly, intravenous iron treatment was associated with striking macrocytosis. When patients were divided into quartiles by MCV at study onset, a significant response to intravenous iron occurred only in quartile 1, although macrocytosis developed in all quartiles.

Intravenous iron administration was also associated with increased serum albumin, decreased blood pressure, and an increase in creatinine/kg body weight (an index of muscle mass in functionally anephric patients). When patients were characterized by serum ferritin10before intravenous iron repletion, these changes in blood pressure and serum creatinine/kg body weight occurred only in the lower ferritin subset.


Patients and Routine Method of Treatment

Upper Manhattan Dialysis Center (UMDC) provides dialysis services for patients from St Luke's, Columbia-Presbyterian and other hospitals in New York City. By January 1996, 206 patients were treated for end-stage renal failure. The 172 treated by hemodialysis (for an average time of 200 minutes thrice weekly) received rhEPO (Epoietin alfa, Epogen®; Amgen, Thousand Oaks, Calif) intravenously with each treatment. A polysaccharide-iron complex containing 150 mg elemental iron, 1 mg folic acid, and 25 μg vitamin B12 (Niferex®-150 Forte; Central Pharmaceuticals, Seymour, Ind) was prescribed for most patients. This is the only oral iron preparation reimbursed by New York State Medicaid. Dose and timing of rhEPO and iron varied as prescribed by eight attending nephrologists.

Computerized Data Collection and Retrieval

In the last trimester of 1994, UMDC installed Disease Manager Plus (MIQS®, Inc, Boulder, Colo), a patient-centered medical information system. The database fully integrates an analyzable computer-based patient record, data repository, and practice management functions. By June 1995 it had replaced the paper chart entirely and included automated electronic download and validation of laboratory test results.

Selection of Patients for Iron Administration

Initial analysis in December 1995 revealed a median hematocrit of 0.309, comparable to that then prevalent in the US dialysis population. Macrocytosis (MCV>98 fL) was found in 8% of patients. Red cells indices suggestive of iron deficiency (MCH<26 pg, MCV<80 fL) were present in 10-15%. In January 1996 it was decided to approach iron administration systematically in all patients. First, iron (Niferex) was prescribed by mouth at night when the stomach was empty and free of phosphate binders. Effects were assessed by examination of hematological data in monthly individual patient reports. By October 1996 it was apparent that this therapy was inadequate. In the planned second study phase, iron dextran (InFeD®; Schein Pharmaceutical, Florham Park, NJ) was administered intravenously to anemic patients with laboratory findings that suggest iron deficiency (transferrin saturation <20% and/or serum ferritin <100 ng/mL). Most received courses of 1 g of elemental iron, consisting of 10 successive 100-mg (2-mL) infusions at the end of treatment sessions. Succeeding courses were given if iron deficiency persisted. Results of routine laboratory tests (complete blood count, tests of iron status, and a serum chemistry panel), done early in the first week of each month, were downloaded electronically, reviewed within 72 hours of testing, and hematinic medications adjusted as required. When hematocrit increased, rhEPO dose was tapered. Because rhEPO reimbursement was at the time disallowed when hematocrit was >0.36,11the striking response after intravenous iron made it difficult to continue the original protocol. The study was initially terminated on June 30, 1997, but patients were followed in detail until December 31, 1997.

Selection of Patients for Longitudinal Analysis

In January 1997, 10 weeks after the first patients received iron intravenously, a preliminary analysis revealed striking changes in hematological and iron parameters in some patients, variation in individual patients though time, and patient-to-patient variation. Although data on all patients were analyzed, a prospective analysis of retrospectively collected data on patients followed for >1 year became a principal analytical focus. With a minimum 12-month prospective follow-up, it was predicted that perspective on treatment effects might be achieved. Patients (n=160) who were alive in January 1997 were initially identified. Patients with AIDS (n=11) or myeloma (n=1) were excluded. The final population was 102 patients with laboratory data from January 1996 or earlier; 87 were undergoing hemodialysis and are the subjects of this analysis. Fifty-one were male. Twenty-four patients were Caucasian, 62 were African American, and one was Native American. Median age was 58.5 years; 65 had Medicare insurance or were Medicare eligible, and 22 had Medicaid only. The median duration of dialysis treatment before the start of the study was 15 months. The cause of end-stage renal disease was hypertensive renal disease in 33, diabetic nephropathy in 28, and glomerulonephritis in 12 subjects. In January 1996, 85 were receiving rhEPO; none was receiving intravenous iron. A dietician participated in the care of all patients and reviewed their nutritional status monthly. Social workers also participated actively in their care and took all necessary steps to ensure access to appropriate nutrition.

Data Abstraction and Analysis

All analyses were done by one author (V.E.P.), who was ignorant of treatment details, knowing only that most patients were treated with rhEPO and had received oral iron for a long period before being switched to intravenous iron dextran. Data abstracted for 1- to 12-month periods were exported into a tab-delimited text file and imported for further analysis into SAS JMP® (version 3.2; SAS Institute, Inc, Cary, NC). Laboratory data were tested for normality of distribution with the Shapiro Wilk W test. Those conforming to a normal distribution after logarithmic transformation were analyzed after logarithmic transformation. To provide clinically relevant information on oral iron treatment in individual patients, mean values for each were calculated for the first 3 months (January-March 1996) and for two successive trimesters (April-June and July-September 1996) during oral iron administration. Here, mean values in the last trimester of oral iron treatment (July-September 1996) are compared with those in the first (January-March 1996). To assess the effect of intravenous iron treatment in individual patients, mean values for each were calculated for the last initially planned trimester (April-June 1997) of intravenous iron and compared with those of the last trimester of oral iron treatment (July-September 1996). Another comparison was made with data obtained in the final trimester (October-December 1997). Mean values were compared using the paired t test for normally distributed data, or with the Wilcoxon signed-rank test for data not normally distributed. Other analyses were done by analysis of variance. Mean values of non-paired data were compared using Student's t test.


Selection ensured that all 87 patients who entered the study on January 1, 1996, were alive in early January 1997; 67 were still followed on December 31, 1997. Of the other 20, one received a kidney transplant, seven transferred to other dialysis units, and 12 (mean age, 66 years; range, 51-86 years) died.

Treatment With Oral Iron

There were statistically significant but very small and poorly sustained increases in hemoglobin concentration (+2.58 g/L) and erythrocyte count (+0.16 1012/L). Mean MCV and MCH decreased slightly (P<0.05 and P<0.01, respectively). Transferrin saturation changed little, but median serum ferritin decreased by 23% from 111 ng/mL (P<0.001). During oral iron treatment, the amount of blood processed increased significantly by 0.08 L/kg from 1.04 L/kg. Postdialysis weight, serum creatinine, serum albumin, mean arterial pressure (MAP), urea reduction ratio, and Kt/V changed little.

Treatment With Intravenous Iron

Between October 1996 and June 1997, 85 patients received iron dextran intravenously. The mean total dose of elemental iron per patient was 2960 mg (range, 900-5400 mg given during a 9-month period). Mean hematocrit increased to 0.366, mean hemoglobin concentration to 110.5 g/L (P<0.01), and erythrocyte count to 3.88 1012/L (P<0.001). MCV increased continuously to 95.6 fL (P<0.001), greater than one standard deviation above the normal mean (Table 1). Concurrently, mean corpuscular hemoglobin concentration (MCHC) decreased to 301 g/L (P<0.001). Improvement was sustained over the next 6 months, ending with the October through December 1997 trimester. Iron saturation and ferritin increased significantly (P<0.001 for each) (Table 1).

Table 1

Differences between hematological values in the last trimester of oral iron administration and at 6-9 and 12-15 months after intravenous iron administration was started in patients studied longitudinally.

Dialysis parameters are shown in Table 2. Postdialysis weight decreased significantly by the last study trimester. Serum creatinine and serum albumin both increased significantly. There was a very small increase in urea reduction ratio, Kt/V, and the amount of blood processed. MAP increased slightly in approximately the first 4 months of intravenous iron treatment. The mean change in MAP in the fourth trimester was +1.23 mm Hg pre-dialysis, and -0.22 mm Hg postdialysis; in the fifth trimester, the mean changes were +1.04 and +0.92 mm Hg, respectively. Thereafter, there was a slow decline in MAP, more clearly delineated in the posthemodialysis MAP (Table 2). The decline in postdialysis MAP was significant in the sixth (3%, P<0.05), seventh (3.4%, P<0.01), and eighth trimesters (4.5%, P<0.001).

Table 2

Differences between findings in the last trimester of oral iron administration and at 6-9 and 12-15 months after intravenous iron administration was started in patients studied longitudinally.

Changes in Mean Corpuscular Volume

Increase of MCV from low to mid-normal levels was expected in iron deficient patients undergoing iron repletion. Shortly after intravenous iron treatment began, it became apparent that the effect on MCV in many patients was greater than expected. One particular patient (Figure 1) drew attention to the large MCV changes. This 37-year-old man with renal failure caused by Type 1 diabetes mellitus started chronic hemodialysis treatment on November 1, 1995. He was severely anemic and profoundly iron deficient (hemoglobin 66 g/L, MCV 73 fL, MCH 20.4 pg, transferrin saturation 4%, serum ferritin 29 ng/mL). As oral iron failed to correct the problem, intravenous iron was started on November 8, 1996. He received 1000 mg in November, 1700 mg between January and March 1997, 900 mg in June 1997. Hemoglobin, MCV, MCH, and transferrin saturation increased promptly. Hemoglobin and MCV increased to maximum levels of 176 g/L and 118 fL, respectively, in April 1997. MCV was >100 fL in all 23 samples measured from January to December 1997.

Figure 1

Findings in a single patient over 24 months. Hemoglobin (Hb) and mean corpuscular volume (MCV) rose rapidly from low to above normal levels after intravenous iron was started, despite a great decrease in the dose of erythropoietin (EPO). MCH = mean corpuscular hemoglobin.

This observation prompted systematic assessment of MCV after intravenous iron was given. MCV was >98 fL in 4.3% of samples from January to November 1996, in 15% in December,1996, and in 21% in January 1997. Each month from February to December 1997, MCV was >98 fL in >30% of samples.

Mean Corpuscular Volume in Other Dialysis Patients

The longitudinally studied patients were compared with 84 UMDC patients not participating in the longitudinal study and with 710 patients in four other dialysis units that used the patient-centered medical information system (Tables 3 and 4). In dialysis unit no. 4, the mean MCV was in the expected range, and only 12.3% of samples had an MCV >98 fL. In dialysis unit nos. 1, 2, and 3, mean MCV was elevated, and MCV was >98 fL in 32.7% of 2274 samples; the 3-month patient mean was elevated in 29.2% of 658 patients.

Table 3

Occurrence of an elevated MCV (>98 fL) in five dialysis units.

Table 4

Selected findings in the last trimester of oral iron administration and at 6-9 and 12-15 months after intravenous iron administration was started in patients not studied longitudinally.

These findings led us to the postulate that a deficiency or disorder of metabolism of folate might be more common than hitherto thought. As MCV is the red cell index primarily affected by iron and folic acid deficiency, the data of the longitudinally studied patients were reanalyzed as a function of MCV at study start. The January 1996 findings were used to divide patients by four MCV quartiles (quartile 1, MCV 73-82.9 fL; quartile 2, MCV 83-86.9 fL; quartile 3, MCV 87-92.9 fL; quartile 4, MCV 93-103 fL).

Quartiles at Start of Study

There were proportionately more males in quartile 1 than in the other three (χ2 [1]=4.70, P<0.05). Age, race, and insurance were distributed similarly in all quartiles. Duration of prior dialysis treatment was similar in all quartiles. There was no difference in distribution of primary disease. Patients in quartile 1 had the most striking evidence of iron deficiency (Table 5) at outset, and MCH and transferrin saturation were low. Transferrin saturation in 12 patients was <16%, at which level effective iron transport would not be expected.

Table 5

Hematological values (mean±SEM) in the initial 3-month period, January-March, 1996, for patients studied longitudinally and arrayed by four MCV quartiles.

Hemoglobin was slightly higher in each successive quartile, as were MCH and transferrin saturation. By contrast, serum ferritin was similar in each quartile. When quartiles 3 and 4 were compared with quartiles 1 and 2, the proportion of patients with MCH and transferrin saturation values below the lower limits of normal differed significantly (MCH χ2 [1]=16.72, P<0.001; transferrin saturation χ2 [1]=12.22, P<0.001), but the proportion with low-ferritin values did not (χ2 [1]=0.015, P>0.90).

When the effect of oral iron was examined by individual quartiles, there was a decrease in MCV and MCH only in quartile 4, whereas serum ferritin decreased in all. The small increment in hemoglobin was associated with a slight decrease in red cell size and hemoglobin content, and with a significant decrease in iron stores. When the effect of intravenous iron administration was examined by individual quartiles, hemoglobin and red blood cell increases were significant only in quartile 1 at the 3- to 6-month and 9- to 12-month periods (Figure 2). Increase in transferrin saturation was greatest in quartile 1, minimal in quartile 4. MCV increased significantly (P<0.001) and MCHC decreased significantly (P<0.001) in all quartiles. MCH decreased significantly (P<0.001) in quartile 4, and at both time periods.

Figure 2

Changes from the last trimester (July-September 1996) of oral iron to the 6th (April-June 1997) and 8th trimesters (October-December 1997), during which patients received iron intravenously. Asterisks denote significance of the changes: * P *<0.05; ** P *<0.01; *** P *<0.001. MCV = mean corpuscular volume; EPO = erythropoietin.

Iron Stores and the Effects of Intravenous Iron

In chronic hemodialysis patients, serum ferritin reflects total body iron stores. To explore the effects of iron repletion, the patients were divided into two equal-sized subsets based on median serum ferritin in the 9-month period of January though September 1996, when iron was given by mouth. In the low-ferritin subset, median serum ferritin ranged from 15 to 88 ng/mL, in the high subset from 89 to 853 ng/mL. Hemoglobin, serum albumin, urea reduction ratio, Kt/V and the amount of blood processed in each treatment were identical in the two subsets, as was MAP (P>0.5, for each). Serum creatinine was lower in the low- (12.03±0.56 μmol/L/kg) than in the high- (14.33±0.81 μmol/L/kg) ferritin subset (t[81]=2.33, P<0.05). The total amount of iron given intravenously was similar in low- and high-ferritin subsets (Figure 3). A significant increase in serum albumin occurred in both ferritin subsets. The increase in serum creatinine was significant in the low-ferritin subset only. Also, the decrease in diastolic and mean arterial pressures occurred only in the low-ferritin subset.

Figure 3

Changes from the last trimester (July-September 1996) of oral iron to the 6th (April-June 1997) and 8th trimesters (October-December 1997), during which patients received iron intravenously. Asterisks denote significance of the changes: * P *<0.05; ** P *<0.01; *** P *<0.001. HD = hemodialysis.

Potential Adverse Effects

Hemosiderosis, usually preceded by prolonged high serum ferritin elevation, is associated with repeated parenteral iron administration. Prolonged high elevation of serum ferritin was infrequent. Three or more values >1000 ng/mL occurred in nine patients, three of whom had high serum ferritin levels before intravenous iron. Overall, serum ferritin levels >1000 ng/mL were recorded in 3.26% of 1791 specimens in 20 patients. Reactions associated with iron administration by vein were not encountered. One patient developed gastrointestinal bleeding during the oral iron replacement phase, and a second during intravenous iron replacement. Of 109 stool examinations for occult blood during this study, only one was positive.

Hospitalization and Mortality

Hospitalization and mortality rates are important outcome measures, but the data must be interpreted with caution because of the small number of patients and relatively short follow-up period. All 102 patients were included in the calculations, ie, 87 treated by hemodialysis and 15 treated by peritoneal dialysis. Patients studied longitudinally were in the hospital for an average of 14.9 days per year at risk during the 33-month period January 1, 1997, to September 30, 1999. This is 19.5% lower than the 1996 rate of 18.5 days. A similar trend was observed when the rate was calculated for all patients treated at UMDC (studied longitudinally and cross-sectionally), excluding those with AIDS. The average hospital stay of 7.7 days was 29% lower than the historical average of 10.9 days. Twenty-eight patients died in the 33-month period, a mortality rate of 128.7 per 1000 years at risk. This is 31% lower than the facility's historical 1995-1996 mortality rate of 186.6 per 1000 years.

Patients Not Studied Longitudinally

The study patients had been treated by dialysis for a median of 15 months before the start of the study. The remaining dialysis patients included many in whom hemodialysis treatment was started recently. To test whether the findings and trends were similar in this patient population, not studied longitudinally, the data were analyzed for all other patients treated by hemodialysis (excluding those with AIDS). Selected results are shown in Table 4. Differences between the time periods were analyzed using Student's t test for unpaired observations. Trends were similar in direction to those in patients studied longitudinally. There was a small increment in the number of erythrocytes and in serum creatinine/kg. Of particular note was the statistically significant increase in serum albumin and in MCV, confirming the observations in the study group.



The requirement for iron to sustain the erythroid response to rhEPO was observed more than 10 years ago.1,2The present study provides further evidence for failure of oral iron to correct the iron deficiency in hemodialysis patients.4,7Nephrologists have long used intravenous iron in dialysis patients. Although intravenous iron continues to be recommended in dialysis patients on rhEPO,11,12comparison of the present data with the mean hemoglobin of US dialysis patients (noted above) suggests that this recommendation has not been translated effectively into daily practice.

During oral iron administration, a very small hemoglobin increase was accompanied by a decrease in MCV and MCH and was related inversely to change in ferritin (F [2,79]=6.76, P<0.002), indicating further depletion of iron stores as well as iron-deficient erythropoiesis. The improved hematopoietic response after intravenous iron re-emphasizes the importance of impaired iron availability as a component of anemia in dialysis patients.3–7,13

Dividing patients at outset into four MCV quartiles facilitated understanding of responses to intravenous iron administration. One quartile with an unequivocal hemoglobin response to intravenous iron repletion was highlighted. MCV increased with intravenous iron administration in all quartiles, whether erythrocytes were small, normal, or large at outset (Figure 2). This increase was expected in iron-deficient patients with small erythrocytes. Although a slight increase in red cell size was reported in hemodialysis patients treated with parenteral iron but not with rhEPO,14the rapid development and persistence of macrocytosis in some patients was not anticipated. Its persistence makes it unlikely that macrocytosis was caused by a burst of reticulocytes, although this parameter was not measured prospectively.

On the basis primarily of changes in transferrin saturation and MCHC, it has been proposed that rhEPO induces functional iron deficiency in dialysis patients.15We observed these changes but suggest a different interpretation. The MCHC decline, with stable MCH, primarily reflects the striking MCV increase (Figure 4), a feature not characteristic of iron deficiency. Given the association of folate deficiency with anemia in dialysis patients,16,17we suggest that the MCV increase is a marker of inadequately treated folate deficiency unmasked by an adequate iron supply18and that decreased MCHC is consequent on this MCV change. Serum folate (n=176) and vitamin B12 and red blood cell folate (n=15) levels were normal or high. This does not preclude a potential role for folate, as increased hematocrit and a concurrent decrease in high MCV may occur when 10 mg of folate is administered daily for 3-4 months to dialysis patients with normal or high serum folate levels.17These levels might be explained by endogenous or exogenous inhibitors of folic acid in blood in renal failure.19There is clearly need for systematic evaluation of the effects of large doses of folate, particularly in light of the recent report that concluded that elevated plasma homocysteine levels are common in dialysis patients.20Doses of folate greater than 1 mg/d were not administered in this study.

Figure 4

Inverse relationship between mean corpuscular volume (MCV) and mean corpuscular hemoglobin concentration (MCHC). The values are means for each of the 24 months of study.

We also suggest that decreased transferrin saturation does not solely reflect a reduction in the iron available for erythropoiesis but is a consequence of improved protein synthesis. Transferrin is a marker of nutritional status in dialysis patients.21Both transferrin and albumin increased into normal ranges following intravenous iron. Decreased transferrin saturation may, therefore, reflect, at least in part, an enhanced anabolic state and an increment in transferrin concentration rather than a decrement in iron availability.

There was no clear association of response to iron with any demographic or comorbid factors. Several other findings known to affect bone marrow function in renal failure were analyzed. There was no relationship between hemoglobin increment following intravenous iron and serum calcium, phosphorus, calcium-phosphorus product, alkaline phosphatase, intact parathyroid hormone, or aluminum levels (P>0.2 for each).

Although patients with a higher MCV had a much smaller increment in hemoglobin in response to iron, they also showed the MCV increment that was observed in all patients in the study. This implies that some impairment of erythrocyte iron metabolism is present in virtually all chronic renal failure patients.

Diagnosis of Iron Deficiency Anemia

Absence of stainable iron on a Prussian blue-stained marrow aspirate is the traditional gold standard for diagnosis of iron deficiency. In the present study, we used hemoglobin response to intravenous iron to operationally decide whether iron deficiency anemia was present at outset. That this is a valid assumption is supported by the hemoglobin response to intravenous iron. It was significant and sustained only in quartile 1, despite a 52% decrease in weekly rhEPO (Figure 2).

Low MCV and low transferrin saturation are traditional diagnostic criteria for iron deficiency anemia, but it has been suggested that low serum ferritin and low transferrin saturation may be more valid as criteria in chronic renal failure.11Nevertheless, the lower limit of normal for serum ferritin in dialysis patients should probably be set higher than that for otherwise healthy individuals.22,23Because peak hemoglobin, and presumably peak intravenous iron response, may have occurred before or at study close, we calculated the maximum mean hemoglobin for each trimester of 1997. The maximum increase in any trimester was chosen as the end point for measuring the response.

Age, sex, race, and diabetes mellitus had no effect on response (P> 0.4 for each). Only three findings at the outset were predictive of subsequent hemoglobin response to intravenous iron: MCV quartile (P<0.005); MCH (P=0.008); and transferrin saturation (P<0.001); The first two are indicators of erythrocyte iron metabolism; the third indicates iron availability for erythropoiesis. Serum ferritin, a predictor of iron stores, was a poor predictor of subsequent hemoglobin response (P=0.08).

Low serum ferritin levels appear to provide very strong evidence for decreased iron stores in these patients. Dividing patients at outset into two equal-sized subsets facilitated understanding of the role of iron stores. The division occurred at a serum ferritin level of 88 ng/mL, approximating the proposed lower limit of normal (100 ng/mL) for patients treated by dialysis.22,23At the outset, the two subsets were similar in most respects. Serum creatinine, measured as μmol/L/kg, was less in the lower-ferritin subset, suggesting less muscle mass. A striking increment in serum creatinine occurred in this subset only.

Improvement in serum albumin and muscle mass with intravenous iron has not (to our knowledge) been reported previously in adults. The increment in serum creatinine was statistically significant only in the low-ferritin subset; ie, the subset with a low serum creatinine before intravenous iron was started. Iron is essential for oxidation-reduction catalysis and bioenergetics,24and activity of a variety of iron-containing enzymes of intermediary metabolism can be impaired in iron deficiency.25Improvement in muscle mass may be a consequence of correction of such a problem. Hypoalbuminemia occurs occasionally in iron-deficient children and is also rectified by iron replacement.26These findings are typically attributed to malabsorption owing to sloughing of intestinal mucosa during iron deficiency, but this mechanism has not been definitively confirmed.27Children have increased nutritional demands because they are in a growth phase. No patient in this study had any evidence of a disorder associated with malabsorption, such as regional enteritis. Malnutrition is frequent in chronic maintenance dialysis patients21,28–31who, in addition to problems with absorption from the gastrointestinal tract, suffer considerable loss of essential nutrients, such as amino acids and vitamins, in the dialysate. On the basis of studies in nonuremic rats that have undergone intestinal resection and reanastomosis, it has recently been proposed that rhEPO exerts anabolic effects.32However, such effects, even if they occur in humans, cannot explain the findings observed in our patients, because the patients who exhibited the increase in muscle mass received significantly less rhEPO.

That dialysis patients suffer from deficiency of iron stores is not surprising. The prevalence of iron deficiency in the present study closely approximates the reported rate of 50% in dialysis patients in the United States.33The present study strongly suggests that effective iron repletion can lead not only to improvement in anemia but also to considerable improvement in nutrition and well being.

Serum albumin, an independent predictor of mortality,34increased by 2.19 g/L during intravenous iron administration, and by 4.00 g/L in July through September 1997 (P<0.0001), which suggested further improvement in well being. The reason for this striking increase is obscure. It seems reasonable to assume that increased well being that occurred with anemia correction may be associated with better appetite and, consequently, better nutrition. The decrease in blood pressure was unexpected, particularly as attention in literature has centered around the increased blood pressure observed when anemia of dialysis patients improves through rhEPO treatment.35Improvement in anemia was greatest in patients with low MCV at the outset and was independent of initial serum ferritin level. Improvement in blood pressure, on the other hand, was greater in patients with lower serum ferritins, suggesting an association with iron repletion in patients with deficient iron stores. This differential improvement is not explained by a difference in the antihypertensive drugs prescribed. In the 2-year period, 38% of the low-ferritin and 18% of the high-ferritin subset received no antihypertensives. When the amount received in the sixth and eighth trimesters was compared with that in the third trimester, 74.4% and 77.8% of those in the low-ferritin subset and 59.4% and 69.7% of those in the high-ferritin subset received the same or lower doses of antihypertensive drugs. The hypertension observed in dialysis patients on rhEPO has been attributed to both a hydrodynamic effect of the expansion of total blood volume that results from an increased red cell mass and from a postulated pressor effect of rhEPO.36,37For patients in the low-ferritin subset, the reduction in blood pressure may be caused by decreased rhEPO use.

Observations reported herein illustrate the advantages of long-term detailed follow-up of patients treated by dialysis, in whom changes in well being manifest slowly. The changes in serum albumin, serum creatinine, and blood pressure became evident and statistically significant only 4-6 months after intravenous iron treatment was started and were statistically significant after 6-12 months. These are changes that have a favorable effect on morbidity and mortality,34,38and they were associated with an apparent decrease in morbidity and mortality in patients included in the present study.


We thank staff of the Upper Manhattan Dialysis Center, whose work and devotion to the care of their patients enabled these observations; John P. Flynn, MIQS, Inc, who designed the software and query tools that made the analyses possible; C.R. Buncher, ScD, J.I.E. Hoffman, MD, Richard W. Vilter, MD, and Lawrence Frohmann, MD, for constructive suggestions about the manuscript. We also thank Keith Shearlock, MD, Pat DiGiovanni, RN, and Jonathan Lorch, MD, for kindly allowing access to mean corpuscular-volume data from patients in dialysis units of Renal Care Group of the Southeast, Pensacola, Fla, Englewood Hospital, Englewood, NJ, and The Rogosin Institute, New York, NY.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.