Robert T. Means, Jr, Hematology/Oncology Division, Medical University of South Carolina, 903 CSB, 96 Jonathan Lucas Street, Charleston, SC 29425, USA. E-mail: email@example.com
Summary. The hepatic antimicrobial protein, hepcidin, is implicated in duodenal iron absorption and mobilization. Overexpression of the hepcidin gene is associated with a hypoferraemic, microcytic, iron-refractory anaemia. On the basis of these observations, it has been proposed that hepcidin is a mediator of the common clinical syndrome, anaemia of chronic disease (ACD), and recent findings evaluating urinary hepcidin production in patients support this hypothesis. In the present report, serum hepcidin concentrations were measured in 55 specimens submitted for ferritin determination, and in 37 specimens collected from anaemic patients undergoing diagnostic bone marrow examination. The serum hepcidin concentration exhibited a statistically significant correlation with serum ferritin concentrations in both patient subsets. No statistically significant correlations were observed between serum hepcidin and other laboratory markers of iron status or anaemia diagnosis. Serum hepcidin does not appear to correlate as well with clinical diagnosis as urinary hepcidin, suggesting that a better understanding of the clearance and metabolism of this protein is required to understand fully its potential contribution to the pathogenesis of ACD.
Anaemia of chronic disease (ACD) is one of the most common syndromes observed in clinical medicine, and is probably second only to blood loss with consequent iron deficiency as an aetiology of anaemia (Lee, 1983). ACD is observed in disorders associated with the activation of cytokines that mediate the immune and inflammatory response, and it has generally been considered that ACD is a consequence of the effects of these cytokines (Means & Krantz, 1992; Means, 1999).
ACD is associated with abnormalities of systemic iron handling, typically characterized by the decreased availability of iron for erythropoiesis despite adequate reticuloendothelial iron stores. This has also been considered as a consequence of cytokine activation, possibly mediated through second messengers such as nitric oxide (Domachowske, 1997). More recently, it has been suggested that hepcidin, a hepatic antimicrobial protein, may be a major contributor to the pathogenesis of ACD (Fleming & Sly, 2001; Ganz, 2002). Also known as liver-expressed antimicrobial peptide 1 (LEAP-1), hepcidin can be detected in both human blood and urine (Krause et al, 2000; Park et al, 2001). Mice with iron overload overexpress a protein homologous to hepcidin (Pigeon et al, 2001), and knockout mice lacking the hepcidin gene exhibit severe parenchymal iron deposition (Nicolas et al, 2001). Potentially more relevant to ACD, humans with hepatic adenomas expressing hepcidin mRNA at high levels exhibit a severe microcytic, hypoferraemic anaemia, which resolves after adenoma resection (Weinstein et al, 2002). A similar syndrome is observed in transgenic mice overexpressing the hepcidin gene (Nicolas et al, 2002).
In a recent publication, Nemeth et al (2003) determined urinary hepcidin excretion in a variety of patients with iron overload, iron deficiency and either infectious or inflammatory diseases. Urinary hepcidin excretion was increased 10- to 100-fold in patients with iron overload or infection/inflammation, compared with normal individuals. Urinary hepcidin excretion showed a statistically significant correlation with the serum ferritin concentration.
In order to assess further the contribution of hepcidin to anaemia and to clinical iron metabolism more generally, we applied a Western blot assay for hepcidin to serum specimens from 55 patients undergoing evaluation of iron status, and from 37 anaemic patients undergoing a clinically indicated bone marrow examination.
Patients and methods
Patient material. Serum specimens and data were collected under protocols approved by the institutional review board of the University of Cincinnati Medical Center. Serum specimens and data were collected with informed consent from 37 anaemic patients undergoing diagnostic marrow examination as part of another study (Means et al, 1999), and from specimens submitted to the Diagnostic Hematology Laboratory of the University of Cincinnati for ferritin analysis (55 samples). These specimens were selected to reflect the range of serum ferritin values observed in that laboratory, and without prior knowledge of haematological, clinical or other laboratory values. The use of these specimens and data for the current study was approved by the Medical University of South Carolina (MUSC) institutional review board and by the Research & Development Committee of the Ralph H. Johnson VA Medical Center.
Hepcidin Western blot assay. After determination of the protein concentration, serum specimens were diluted with phosphate-buffered saline (PBS) to 15 µg of protein in 14 µl aliquots. A synthesized peptide of the C-terminal, 25-amino-acid, secreted form of human hepcidin (Alpha Diagnostic International, San Antonio, TX, USA) was used as the standard and diluted to the desired concentrations with PBS. The Invitrogen NuPAGE® gel system (Invitrogen, Carlsbad, CA, USA) was used for Western blotting. Samples and standards were heated for 10 min at 70°C before separation on 4–12% bis-(2-hydroxyethyl)imino-tris(hydroxymethyl)methan (Bis-Tris) gels and electrophoretically transferred to 0·2 µmol/l nitrocellulose membranes. Membranes were blocked with 3% bovine serum albumin (BSA) (ICN Biomedical, Aurora, OH, USA) in PBS/0·1% Tween-20 (PBS-T/BSA) (Tween-20 from Sigma; PBS from Gibco) for 1 h, washed for 5 min in 0·1% Tween-20–PBS (PBS-T) six times and incubated at 4°C overnight with a rabbit antibody directed against a 13-amino-acid sequence in the circulating form of human hepcidin (Alpha Diagnostic International), 1:2000 in PBS-T/BSA. Membranes were washed as indicated and incubated for 1 h with anti-rabbit immunoglobulin (Ig) G (Sigma Chemical, St Louis, MO, USA) in 1:4000 PBS-T/BSA, followed by washing. Immunocomplexes were visualized using chemiluminescence reagents (Amersham Pharmacia Biotech, Piscataway, NJ, USA) and a 5-min exposure. Films were scanned with a personal computer-connected 16-bit scanner at 300 d.p.i. and band intensity determined with scion image program (Scion Co., Frederick, MD, USA). Molecular weights were determined with gel analysis software (Alpha Innotech Corporation). Determinations of the hepcidin concentration were calculated from a linear fit to the hepcidin standards.
Western blot analysis of serum specimens showed a discrete band at 9 kDa, as well as a number of less discrete bands at molecular weights > 60 kDa. However, when the hepcidin standard was added to specimens, only the 9 kDa band was enhanced, indicating that this band represented hepcidin-specific antibody binding (Fig 1).
In order to assess the stability of specimens over time, serum hepcidin concentrations were determined using serum specimens collected from five normal donors and retained at −80°C for > 5 years and specimens collected recently from five other normal individuals. Hepcidin concentrations were not significantly different between the two groups (Wilcoxon signed rank test, P = 0·50).
Statistical analysis. Statistical analysis was performed using the Statistical Package for the Social Sciences version 10·0 (SPSS, Chicago, IL, USA). Correlations were made using the Pearson product moment test.
The laboratory characteristics of the two patient subsets studied are shown in Table I. All parameters overlapped considerably, although the distributions of serum iron and serum ferritin values were significantly different between the two subsets (P < 0·01; Wilcoxon signed rank). Because the two patient subsets represent individuals with disparate clinical syndromes, they will not be compared further.
Table I. Clinical laboratory parameters of specimens evaluated.
Ferritin evaluation patients
Anaemia evaluation patients
Fe, serum iron; TIBC, serum total iron-binding capacity; TIBC sat, serum total iron-binding capacity saturation; MCV, mean red cell corpuscular volume; MCH, mean red cell haemoglobin; TfR, serum soluble transferrin receptor; Hgb, haemoglobin.
Serum hepcidin concentrations from five normal subjects were < 0·40 mg/ml.
TIBC sat (%)
Correlations between the serum hepcidin concentration and clinical laboratory parameters of iron status observed in the 55 specimens submitted for ferritin determination are shown in Table II. The serum ferritin concentration (Fig 2) was correlated with the serum hepcidin concentration (P < 0·01). A significant inverse correlation was also observed between the serum total iron-binding capacity and serum hepcidin (P = 0·02), but this correlation lost significance if it was controlled for ferritin concentration (P = 0·36). Inverse correlations, which did not achieve statistical significance, were observed between serum hepcidin and the haemoglobin concentration, serum iron and the serum soluble transferrin receptor concentration.
Table II. The correlation between clinical laboratory parameters and serum hepcidin in specimens from patients undergoing the evaluation of iron parameters and from anaemic patients undergoing bone marrow examination.
Ferritin evaluation patients
Anaemia evaluation patients
CCorr, Pearson correlation coefficient. Other abbreviations as in Table I.
Table II also shows the correlations between clinical laboratory parameters and the serum hepcidin concentrations in the 37 patients undergoing evaluation for anaemia. The serum ferritin concentration was significantly correlated with the serum hepcidin concentration (P < 0·01; Fig 2). The correlation coefficients and regression lines obtained for serum ferritin and hepcidin concentrations were very similar in both patient subsets. There was an inverse correlation between the serum total iron-binding capacity and serum hepcidin, which did not achieve statistical significance. The inverse correlations observed in the ferritin patient subset between hepcidin and iron, soluble transferrin receptor concentration and haemoglobin were not seen with the anaemia subset. There was no statistically significant relationship between the presence of stainable iron in the bone marrow aspirate and serum hepcidin observed in anaemic patients.
Before determination of serum hepcidin concentrations in the 37 specimens obtained from anaemic patients undergoing clinically indicated marrow examination, these specimens were classified diagnostically in two ways. First, they were classified purely according to laboratory criteria as ACD (hypoferraemia with iron stores present on marrow aspirate; n = 4), iron deficiency (absence of Prussian blue-stainable iron on a marrow aspirate; n = 10) or anaemias of other aetiologies (either a different specific diagnosis or not meeting the criteria above for either ACD or iron deficiency; n = 23). Secondly, these specimens were also classified according to clinical criteria as ACD (ACD as defined above or an undefined, iron-replete anaemia in a patient with a syndrome associated with ACD; n = 8), iron deficiency (as defined above; n = 10) or anaemias of other aetiologies (as defined previously; n = 19). The clinical syndromes associated with clinical ACD were human immunodeficiency virus infection (four cases), malignancy (two cases), sarcoidosis (one case) and cellulitis (one case). Serum hepcidin concentrations did not differ significantly between the three diagnostic subsets, regardless of how ACD was defined. In contrast to the findings reported for urine hepcidin excretion (Nemeth et al, 2003), the mean serum hepcidin concentration for ACD patients was lower than that observed for iron deficiency anaemia (1·09 mg/ml versus 1·44 mg/ml and 0·77 mg/ml versus 1·65 mg/ml for ACD versus iron deficiency in the clinical and laboratory definition groups respectively).
For patients with a diagnosis of iron deficiency, decreased hepcidin production would be expected if it was part of the response to changes in iron stores. Hepcidin is a negative regulator of duodenal iron absorption (Fleming & Sly, 2001), which is increased in iron deficiency (Baynes et al, 1987). Increasing hepatic iron stores in mice are associated with a downregulation of hepcidin mRNA (Pigeon et al, 2001). These findings are also consistent with the iron overload observed in knockout mice unable to express the hepcidin gene (Nicolas et al, 2001).
Similarly, the hypothesis that hepcidin is the mediator of the iron abnormalities that characterize ACD would require an elevated hepcidin production to be observed in this syndrome. This hypothesis is based on the observations that hepcidin mRNA is induced by lipopolysaccaride (Pigeon et al, 2001), as would be anticipated for a mediator of ACD, that both transgenic mice and human patients with hepatic adenomas overexpressing the hepcidin gene exhibit a severe hypochromic microcytic anaemia resistant to iron, and that the anaemia observed in adenoma patients can resolve with resection of the hepcidin-overexpressing tumour (Nicolas et al, 2002; Weinstein et al, 2002). A role for hepcidin would also be consistent with the reported characteristics of intestinal iron metabolism in ACD: normal absorption with increased retention of iron in duodenal reticuloendothelial cells (Schade, 1972; Hershko et al, 1974).
The studies of urinary hepcidin excretion reported by Nemeth et al (2003) are consistent with the hypothesized role of hepcidin in the iron abnormalities of ACD, as described above, although there is no mention of the inverse correlation with serum iron that would have been expected from both the clinical characteristics of ACD and studies of patients or animals overexpressing the hepcidin gene (Nicolas et al, 2002; Weinstein et al, 2002). The serum hepcidin results reported in the present study also show a correlation with serum ferritin, in both anaemic individuals undergoing marrow examination and patients undergoing ferritin determination for a variety of reasons. However, there are no significant differences in serum hepcidin concentration between individuals with different aetiologies of anaemia.
There are several possible explanations for this apparent discrepancy. Hepcidin is a low-molecular-weight molecule and may be rapidly cleared from the circulation: urinary hepcidin content, particularly when normalized to a marker of glomerular filtration such as creatinine, may be a more accurate measure of its production. The estimated molecular weight of the circulating hepcidin peptide is ≈ 2700 kDa, whereas the antibody used in this assay binds to a complex with a molecular weight of ≈ 9000 kDa. It is possible that there is both a carrier-bound and ‘free’ form of hepcidin, with the latter showing a better correlation with different disease states and being reflected more accurately in the fraction excreted in urine. Nemeth et al (2003) demonstrated that urinary hepcidin excretion declines with resolving inflammation; it is possible, although unlikely, that a significant proportion of specimens from ACD patients were collected late in the clinical course.
Alternatively, hepcidin in either serum or urine may be primarily associated with the regulation of ferritin production, rather than being an actual mediator of ACD. The microcytic anaemia syndrome observed in patients with hepatic adenomas overexpressing hepcidin mRNA is not strictly identical to ACD, as the authors of the report make clear (Weinstein et al, 2002). Although the degree of anaemia and hypoferraemia observed in most patients was compatible with ACD, the degree of microcytosis was greater, and the ferritin values reported were lower than is typical in ACD. ACD is characterized by the cytokine-mediated suppression of erythropoiesis and has a strong pathogenetic association with interleukin-1 (IL-1) (Eastgate et al, 1988; Maury et al, 1988; Means, 1999); increased hepcidin expression is associated with IL-6 and not with IL-1 (Nemeth et al, 2003). Although IL-6 administration is associated with anaemia, this anaemia is dilutional rather than a consequence of erythroid suppression (Atkins et al, 1995).
The findings observed in studies of transgenic animals and in patients with abnormalities of hepatic hepcidin gene expression, as well as the correlations observed between serum ferritin and hepcidin in either serum or urine, provide evidence that hepcidin plays a significant role in the regulation of iron metabolism under particular clinical circumstances. The serum hepcidin concentration does not appear to correlate as well with clinical diagnosis as urinary hepcidin, suggesting that a better understanding of the clearance and metabolism of this protein is required to understand fully its potential contribution to the pathogenesis of ACD.
This study was supported by funds from the Medical Research Service, US Department of Veterans Affairs, and by grant HL69418-01 from the US National Institutes of Health.