Hepcidin as a predictive factor and therapeutic target in erythropoiesis-stimulating agent treatment for anemia of chronic disease in rats

Abstract

Anemia of chronic disease is a multifactorial disorder, resulting mainly from inflammation-driven reticuloendothelial iron retention, impaired erythropoiesis, and reduced biological activity of erythropoietin. Erythropoiesis-stimulating agents have been used for the treatment of anemia of chronic disease, although with varying response rates and potential adverse effects. Serum concentrations of hepcidin, a key regulator of iron homeostasis, are increased in patients with anemia of chronic disease and linked to the pathogenesis of this disease, because hepcidin blocks cellular iron egress, thus limiting availability of iron for erythropoiesis. We tested whether serum hepcidin levels can predict and affect the therapeutic efficacy of erythropoiesis-stimulating agent treatment using a well-established rat model of anemia of chronic disease. We found that high pre-treatment hepcidin levels correlated with an impaired hematologic response to an erythropoiesis-stimulating agent in rats with anemia of chronic disease. Combined treatment with an erythropoiesis-stimulating agent and an inhibitor of hepcidin expression, LDN-193189, significantly reduced serum hepcidin levels, mobilized iron from tissue stores, increased serum iron levels and improved hemoglobin levels more effectively than did the erythropoiesis-stimulating agent or LDN-193189 monotherapy. In parallel, both the erythropoiesis-stimulating agent and erythropoiesis-stimulating agent/LDN-193189 combined reduced the expression of cytokines known to inhibit erythropoiesis. We conclude that serum hepcidin levels can predict the hematologic responsiveness to erythropoiesis-stimulating agent therapy in anemia of chronic disease. Pharmacological inhibition of hepcidin formation improves the erythropoiesis-stimulating agent’s therapeutic efficacy, which may favor a reduction of erythropoiesis-stimulating agent dosages, costs and side effects.

Introduction

Anemia of chronic disease (ACD), also termed anemia of chronic inflammation, is the most prevalent anemia among hospitalized patients21 and is primarily found in subjects suffering from diseases with associated chronic immune activation, such as cancer, auto-immune diseases, chronic infection or dialysis-dependent renal failure.51 A central mechanism by which chronic immune activation causes anemia is the retention of iron in the reticuloendothelial system,96 causing a “functional iron deficiency” and consequently an insufficient iron supply for erythropoiesis.10 In addition, cytokine-mediated effects on erythropoietin functionality and erythrocyte half-life, along with antiproliferative effects of cytokines and radicals on the proliferation and differentiation of erythroid progenitors, further contribute to the pathogenesis of ACD/anemia of chronic inflammation.13112

Hepcidin is a primarily liver-derived peptide that orchestrates body iron homeostasis upon binding to the cellular iron exporter, ferroportin (Fp1), resulting in its internalization and degradation with subsequent reduction of cellular iron egress.14 As hepcidin is an acute phase protein, its levels are increased in inflammation,16157 which is of pivotal importance for the reticuloendothelial iron retention underlying ACD.17109763 Accordingly, neutralization of hepcidin by specific anti-hepcidin antibodies or spiegelmers1918 as well as pharmacological inhibition of hepcidin formation could reverse iron retention in the reticuloendothelial system and/or improve anemia in different animal models of inflammatory anemia.2120 In line with this, serum hepcidin has been suggested as a useful marker for patients with ACD.22

Overexpression of hepcidin in mice or injection of recombinant hepcidin causes a hypochromic anemia and impaired response to endogenous erythropoietin.2310 Of interest, erythropoietin has been shown to decrease hepcidin expression2524 via induction of erythropoiesis.2625 Although erythropoietin treatment decreased liver Hamp (the gene encoding for hepcidin) expression in a mouse model of generalized inflammation,27 erythropoietin treatment could only partially correct anemia in this model.28

As the development of anemia negatively affects a patients’ quality of life and impairs basic physiological functions such as cardiovascular performance.3029 treatment of anemia with erythropoiesis-stimulating agents (ESA; epoetin or darbepoetin alfa) was introduced into clinical practice. This therapy can significantly improve patients’ quality of life3029 but can also result in undesirable side effects, such as thrombophilia, stroke and death, as observed in cancer patients or subjects with end-stage renal disease.3230 Due to the effects of ESA on plasma volume, the risk of thrombosis could depend on the dose of ESA required.30 Interestingly, at least in dialysis patients, the risk of cardiovascular events is associated with impaired hematologic responsiveness to ESA.33 Furthermore, ESA treatment is ineffective in some patients suffering from ACD/anemia of chronic inflammation,42 even if true iron deficiency has been ruled out.34

Clinically useful biomarkers which predict the efficacy of ESA treatment are desirable, but have not been identified thus far. Due to its pivotal role in iron metabolism and the fundamental role of iron in erythropoiesis, hepcidin is a good candidate for such a predictor.

We, therefore, used a well-established arthritis model in rats, which has features resembling human ACD,357 and studied the biological interaction between hepcidin and ESA treatment in a prospective fashion.

Methods

Animals

Female Lewis rats were kept on a standard rodent diet until they reached an age of 6 to 8 weeks and a body weight of 140 to 160 g. All treatments were performed by intraperitoneal (i.p.) injection. Chronic inflammation (arthritis) causing ACD was induced as described elsewhere,21 using group A streptococcal peptidoglycan-polysaccharide (PG-APS) (Lee Laboratories, Grayson, GA, USA) at a total dose of 15 μg rhamnose/g body weight.

For a short-term ESA experiment, untreated ACD rats were compared to ACD rats treated with 500 U of recombinant human erythropoietin (rhEPO, epoetin alfa, EPREX, Janssen Cilag). One group received the rhEPO treatment once (on day 16 after PG-APS injection), the other group thrice on three consecutive days, starting from day 14 after PG-APS injection. Rats were sacrificed on day 17 after PG-APS injections.

In all the following experiments, rats were euthanized 42 days after the induction of arthritis. In these long-term experiments, darbepoetin alfa (Aranesp, Amgen) was used, at a dose of 10 μg/kg body weight, because of its longer half life. Darbepoetin alfa treatment was given weekly for the indicated periods.

Another treatment modality was the administration of LDN-193189 (LDN, Axon Medchem, Groningen, the Netherlands), which was used at a dose of 3 mg/kg body weight and given every second day.

First, rats were treated with darbepoetin alfa for 14, 21, 28 or 35 days before euthanasia. One group of ACD rats received no darbepoetin and control rats were not injected with PG-APS. In another experimental setup, ACD rats either received no drug or were treated with darbepoetin alfa alone or in combination with LDN for 21 days before euthanasia. In a further series of independent experiments, ACD animals were injected with LDN alone.

For flow cytometry analysis of bone marrows, we compared healthy controls, untreated ACD rats and ACD rats treated with darbepoetin alfa, LDN or both for 21 days before euthanasia.

For determination of hemoglobin and hepcidin levels over time, small blood samples (300 μL) were taken weekly by tail vein puncture from every animal.

Analysis of rat specimens

Details on complete blood counts, serum hepcidin7 and serum iron measurements, western blotting,367 real-time reverse transcriptase polymerase chain reaction analysis,37 macrophage iron export measurements,7 immunofluorescence, bone marrow smears and bone marrow flow cytometry are provided in the Online Supplement.

Statistics

Comparisons between multiple groups in this study were performed by ANOVA with the Bonferroni-Holm correction or the Dunnett test for multiple comparisons. A corrected P value less than 0.05 was regarded as statistically significant.

Further details on the methods used for this study can be found in the Online Supplement.

Results

Erythropoiesis-stimulating agent therapy inhibits hepatic Hamp mRNA expression in vivo in a rodent model of anemia of chronic disease

Following our working hypothesis, we first tested whether ESA treatment affects hepcidin expression in a rat model of ACD. We, therefore, injected rhEPO (500 U i.p.) into rats once or on three consecutive days starting 2 weeks after induction of ACD by PG-APS injection. As shown in Figure 1A, both a single as well as three consecutive doses of rhEPO, significantly decreased liver Hamp mRNA expression as compared to that in untreated animals. Importantly, this was paralleled by reduced serum hepcidin levels in the respective groups (Figure 1B).

Figure 1.ESA therapy inhibits hepatic Hamp mRNA induction in vivo in a rodent model of ACD. ACD was induced in female Lewis rats by i.p. injection of PG-APS. Animals were then treated i.p. with 500 U erythropoietin alfa (ESA) on day 16 after PG-APS (ACD, 24h ESA group) or on days 14, 15 and 16 after PG-APS injection (ACD, 72h ESA group). All animals (n=9 per group) were sacrificed on day 17. Hamp mRNA levels in the livers were measured by real-time RT-PCR (A). Data are normalized to the expression of the housekeeping gene Gusb and expressed as abundance relative to the mean of control (arbitrary units) and depicted as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers). Hepcidin peptide was measured in serum (B). ESA groups were compared against ACD using ANOVA and the Bonferroni-Holm correction for multiple comparisons.

Long-term effect of erythropoiesis-stimulating agent treatment in anemia of chronic disease

As the short-term experiments proved the effectiveness of rhEPO treatment at reducing the hepatic expression of hepcidin in this specific model, we conducted long-term experiments to study the effects of such interventions on the correction of ACD. Figure 2A shows hemoglobin concentrations of ACD animals at day 42 after injection of PG-APS. Depending on the assigned group, animals received ESA treatment for 14, 21, 28 or 35 days. Darbepoetin alfa (10 μg/kg) was used for this experiment because of its longer half-life.

Figure 2.Blood hemoglobin, serum hepcidin and iron levels in ACD rats treated with darbepoetin alfa. ACD was induced by injection of PG-APS in rats on day 0. On day 42 all animals were euthanized. Controls received no PG-APS. Groups of ACD rats were treated with darbepoetin alfa (ESA) for either 35, 28, 21 or 14 days before euthanasia. Endpoint data are depicted as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers). Statistical significances of differences were calculated using ANOVA and the Bonferroni-Holm correction for multiple comparisons. All groups were compared to each other. Only significant differences are depicted. * indicates a significant difference versus each single other group (P<0.001). (A) Blood hemoglobin, (B) serum hepcidin and (C) serum iron are shown; n=10 for controls, n=6 for ACD and n=5 to 6 for each ACD-ESA group.

Figure 2A shows the trend towards higher hemoglobin concentrations in ESA-treated groups, although this change became significant only for the group treated with the ESA for 28 days (P<0.01). All ACD groups, whether ESA-treated or untreated, had significantly lower hemoglobin levels at the end of the observation period than had healthy controls (P<0.001, Figure 2A).

We found that serum hepcidin was reduced by ESA treatment (Figure 2B), although statistical significance was not reached for all ESA groups. However, serum hepcidin levels in the ESA-treated animals were not significantly different from those in controls, while untreated ACD animals had increased serum hepcidin concentrations (P<0.01, Figure 2B). Interestingly, the ESA-mediated reduction of hepcidin levels did not translate into significant changes of serum iron concentrations (Figure 2C), and serum iron remained below control levels in all ACD groups (P<0.001, Figure 2C).

Differences between the single ESA treatment groups (n=5 to 6 per group) were small and not significant for hemoglobin, serum hepcidin and serum iron. However, on the basis of pooled data from the ACD rats from all treatment periods, the overall effect of ESA was a significant increase of hemoglobin and serum iron levels (P<0.001, Online Supplementary Figure S2)

When performing western blot analysis of the spleen after the termination of the experiment (day 42), we found that Fp1 protein levels (Figure 3A) were lower in ACD rats than in control rats, and that this reduction was partially antagonized by ESA treatment in a time-dependent fashion (Figure 3A). Accordingly, splenic ferritin levels were higher in ACD rats than in control animals, whereas ferritin expression declined with prolongation of ESA therapy (Figure 3A). These observations in the spleen were paralleled by corresponding changes in the liver (Online Supplementary Figure S3).

Figure 3.Spleen and macrophage iron metabolism following ESA treatment. (A) ACD was induced by injection of PG-APS in rats on day 0. On day 42 all animals were euthanized. Controls received no PG-APS. Groups of ACD rats were treated with darbepoetin alfa (ESA) for either 35, 28, 21 or 14 days before euthanasia. One representative blot out of at least three western blots of spleen Fp1 and ferritin are shown. Beta actin was used as a loading control (n=10 for controls, n=6 for ACD and n=5 to 6 for each single ACD-ESA group, see Figure 2). In another experiment (B–C), peritoneal macrophages were taken from control rats, ACD rats (42 days after PG-APS injection), and ESA-treated ACD rats (35 days of weekly ESA treatment). The cells were loaded with Fe⁵⁹-citrate for 4 h. Afterwards, the radioactivity was measured in the washed, harvested cells (B). Other cells were washed and incubated for 2 h in Fe⁵⁹ free medium. At the end, gamma counting was performed on supernatants to determine iron release (C). Data are depicted as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers). Data were normalized to control (arbitrary units). Statistical significances were calculated using ANOVA and the Bonferroni-Holm correction for multiple comparisons (n=12 for controls and n=8 for ACD and ACD-ESA for macrophage experiments).

To specifically investigate the functional effects of ESA therapy on iron homeostasis on a cellular basis, we studied peritoneal macrophages from healthy controls and from ACD rats, with or without 35 days of darbepoetin alfa treatment, in vitro. We found that Fe release from macrophages derived from ACD rats was significantly lower than that from macrophages of control animals (P<0.001, Figure 3B), while ESA treatment of ACD rats significantly increased macrophage iron release (P<0.05, Figure 3B), which, however, remained lower than in control animals (P<0.001, Figure 3B). Importantly, these alterations of macrophage iron export corresponded nicely with the changes in spleen Fp1 expression observed in ACD animals with and without ESA treatment (Figure 3A,B). The effects seen on Fe release (Figure 3B) were not caused by changes in Fe uptake, because uptake was increased in ACD (P<0.001, Figure 3C) and not significantly changed by ESA treatment, with a trend towards a decrease (Figure 3C).

Hepcidin levels predict response to erythropoiesis-stimulating agent treatment

Based on these observations and being aware of the regulatory effects of hepcidin on Fp1 expression,14 which largely determines circulating iron levels and iron availability for erythropoiesis,21 and because we observed variations in the responsiveness of rats to ESA treatment (Figure 2A, Online Supplementary Figure S1), we questioned whether pre-treatment serum hepcidin levels may determine the responsiveness of ACD rats to ESA therapy. Mean serum levels of hepcidin in ACD rats at 7, 14 and 21 days after induction of inflammation did not differ significantly (48.4±12.5 ng/mL, 62.7±18.8 ng/mL, and 50.2±22.2 ng/mL, mean ± standard deviation, respectively), indicating that there was no general change over time in pretreatment serum hepcidin levels before initiation of ESA and/or LDN therapy.

We thus performed a correlation analysis between serum hepcidin levels in ACD rats immediately before the start of ESA therapy in the different treatment groups and the consecutive change of hemoglobin concentrations over 2 weeks, because it took at least 2 weeks until a significant change in hemoglobin levels upon ESA treatment could be observed (Online Supplementary Figure S1).

Performing Spearman rank correlation analysis, we observed a highly significant negative correlation between pre-treatment serum hepcidin levels and the relative increase of hemoglobin after 2 weeks of ESA therapy (Figure 4A, rho=−0.557, P<0.01). In individuals with pretreatment serum hepcidin levels above 40 ng/mL, blood hemoglobin levels on average declined even with ESA treatment over 2 weeks of follow up (Figure 4B), while lower pre-treatment hepcidin levels (below 40 ng/mg) predicted a better hematologic response to ESA therapy (P<0.01, Figure 4B). This indicates that high pre-treatment hepcidin levels in rats with ACD are associated with a poor hematologic response to ESA therapy.

Figure 4.Correlation of pre-treatment serum hepcidin levels with changes of hemoglobin concentrations following ESA treatment in ACD rats. ACD was induced by injection of PG-APS in rats on day 0. On day 42 all animals were euthanized. Groups of ACD rats were treated with darbepoetin alfa (ESA) for either 35, 28, 21 or 14 days before euthanasia. Data are from the same ACD animals as in Figure 2 and Figure 3A. Serum hepcidin concentrations prior to onset of ESA treatment of ACD animals are plotted against the relative change of hemoglobin (DHb) over the first 2 weeks of ESA treatment (hemoglobin after 2 weeks of ESA treatment minus hemoglobin concentrations before ESA treatment). (A) The linear regression is shown. Spearman rank correlation reveals a highly significant negative correlation between these two parameters (rho=−0.557, P<0.01). (B) Animals are divided into two groups according to low (≤40 ng/mL) and high (>40 ng/mL) pre-treatment serum hepcidin levels. Data are depicted as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers). The statistical significance of the difference in hemoglobin change between the two groups was calculated by the Student t-test (P<0.01).

Hepcidin inhibition improves the therapeutic efficacy of erythropoiesis-stimulating agent treatment

Based on the observations made thus far, we questioned whether inhibition of hepcidin formation may increase the therapeutic efficacy of ESA. We thus performed a new series of experiments, and investigated the combined effects of LDN, a small-molecule inhibitor of bone morphogenetic protein (BMP) type I receptors, which reduces hepcidin expression in models of ACD,21 and ESA therapy. Therefore, anemic rats were given either darbepoetin alone on a weekly basis or a combination of darbepoetin plus LDN. Importantly, the combined treatment with ESA/LDN resulted in a faster and more sustained increase of hemoglobin levels as compared to treatment with ESA alone (Figure 5A).

Figure 5.Effects of ESA and/or LDN-193189 on hemoglobin (Hb) levels and erythropoiesis in rats with inflammatory anemia. ACD was induced by injection of PG-APS in rats on day 0. Groups of ACD rats were treated from day 21 onwards with darbepoetin alfa (ESA) or with ESA and LDN in combination On day 42 all animals were euthanized. Accordingly, the treatment duration was always 21 days. Data are depicted as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers) (n=7 for each group). The trend of blood hemoglobin over the complete treatment period is shown in (A). Groups are ACD (gray), ACD-ESA (diagonal lines), ACD-ESA/LDN (dots and lines). All other figures (B–J) show the endpoint of treatment. Liver Hamp mRNA (B), as measured by real-time reverse transcriptase polymerase chain reaction, was normalized to the expression of the housekeeping gene Gusb and expressed as abundance relative to the mean of ACD (arbitrary units). Serum hepcidin (C) was measured by mass spectrometry and serum iron (D) by colorimetry. In (E–F) age-matched, untreated, non-ACD controls (n=10), and treated only with LDN (n=6, also treated from day 21) are included for comparison. Bone marrow cellularity and the ratio of granulopoiesis to erythropoiesis are shown in (H, I). The SMAD1/5/8 complex was detected by western blotting in liver nuclear extracts (J). TATA binding protein (1TBP18 antibody) was detected in the respective samples as a nuclear loading control. One representative blot out of three is shown. Statistical significance of differences between groups was calculated using ANOVA with the Bonferroni-Holm correction for multiple testing.

To see whether the combined effects of ESA and LDN treatments differed from those observed after sole LDN treatment, we analyzed the changes of hemoglobin levels following sole LDN treatment of ACD rats. Interestingly, hemoglobin levels of ACD rats were similar after 3 weeks of either ESA or LDN treatment (Figure 5E), but neither regimen could normalize hemoglobin levels to the values observed in untreated, healthy controls (Figure 5E, P<0.01). Of note, combined ESA/LDN treatment resulted in significantly higher hemoglobin levels than treatment with LDN alone (Figure 5E, P<0.05), providing further evidence for synergistic effects of LDN and ESA in the reversal of inflammatory anemia.

At the end of the treatment period, both liver Hamp mRNA (Figure 5B) and serum hepcidin levels (Figure 5C), were significantly lower in the ESA/LDN-treated ACD animals than in the untreated ACD rats (P<0.05 and P<0.01). In line with the predicted effect of LDN toward BMP type I receptor-mediated SMAD phosphorylation we found a strong reduction of SMAD1,5,8 phosphorylation in the livers of ACD rats (Figure 5J) after combined treatment with ESA/LDN, whereas sole ESA treatment had little effect, which resembles the differences in the reduction of hepcidin expression between the two treatment regimens (Figure 5B,C,J).

The functional relevance of serum hepcidin changes was supported by immunofluorescence analysis of the spleen (Online Supplementary Figure S4) and duodenum (Online Supplementary Figure S5). The cell surface expression of Fp1 in spleen macrophages and duodenal enterocytes is strongly reduced in ACD as compared to the expression in control animals, whereas treatment with ESA or LDN and, most prominently, combined ESA/LDN treatment enhanced Fp1 protein cell surface expression in these two organs (Online Supplementary Figures S4 and S5).

Serum iron levels were not altered by sole ESA treatment, whereas the combination of ESA with LDN resulted in a significant increase of serum iron concentrations (P<0.05, Figure 5D).

When analyzing hematologic parameters in the different treatment groups we found that mean corpuscular volume (MCV) and mean cell hemoglobin (MCH) were not affected by ESA treatment alone, while the administration of LDN (P<0.001, Figure 5F,G) and the combination of ESA/LDN (P<0.001, Figure 5F,G) increased these indices.

Erythropoiesis-stimulating agent/LDN treatment ameliorates erythropoiesis in rats with anemia of chronic disease

Upon investigation of bone marrow composition (for sample bone marrow micrographs see Online Supplementary Figure S6) we found an increased cellularity after treatment with ESA alone (P<0.01, Figure 5H); this increase in cellularity was even more prominent after combined ESA/LDN treatment (P<0.001, Figure 5H). This was paralleled by a pronounced and highly significant (P<0.001, Figure 5I) shift in the ratio of granulopoiesis to erythropoiesis upon both ESA and ESA/LDN treatment. These data demonstrate an expansion of the erythropoietic lineage in the bone marrow which was more pronounced with ESA/LDN treatment than with ESA treatment alone.

To further investigate the effects of combined treatment on erythropoiesis, we performed FACS analysis on bone marrow samples. We studied the percentage of erythroid cells (CD71+CD11b−) (Online Supplementary Figure S7B) and further analyzed maturation stages of erythroid differentiation using anti-rat-erythroid-cells and anti-CD44 antibodies (Online Supplementary Figure S7A).

In agreement with Richardson et al.38 we found dramatically fewer in CD71CD11b erythroid cells in the bone marrow of ACD rats than in control rats (P<0.001, Figure 6A,C). When studying erythroid differentiation, we found a significantly lower number of orthochromatic cells/reticulocytes in the ACD rats than in the control animals (P<0.001, Figure 6B,D), while other precursors were not reduced (Figure 6B).

Figure 6.ESA and LDN treatment ameliorates defective marrow erythropoiesis in ACD rats. ACD was induced by injection of PG-APS in rats on day 0 and treatment with darbepoetin alfa (ESA), LDN or a combination of both was started on day 21. Controls received no PG-APS. On day 42 all animals were euthanized for marrow analysis by flow cytometry (n=7/group). All details on the gating strategy are shown in Online Supplementary Figure S7. Representative blots for control and ACD rats are shown in (A) and (B). The numbers in (A) and (B) are the mean percentages of all animals for each gate ± standard error of the mean (SEM). The total number of marrow erythroid cells (CD71⁺CD11b⁻) per femur (C–D) and the total number of orthochromatic cells/reticulocytes (CD44^dim,FCS^low) per femur (E–F) are shown as mean ± SEM. The Student t-test was applied for pairwise comparison (C,E) and ANOVA with the Dunnett test for multiple comparisons versus ACD (D,F). *P<0.05, **P<0.01.

Both ESA and LDN treatment resulted in a trend towards increased numbers of CD71CD11b erythroid cells (Figure 6E) and orthochromatic cells/reticulocytes (Figure 6F) in the bone marrow, but the combination of ESA and LDN resulted in a more sustained and significant increase in the number of CD71CD11b erythroid cells and orthochromatic cells/reticulocytes (P<0.05, Figure 6E,F). The FACS results are in accordance with the results obtained from the bone marrow smears (Figure 5H–I)

The complete blood count data from ESA/LDN experiments are presented in Online Supplementary Tables S1 and S2. Of note, neither treatment resulted in a change in total white blood cell counts (Online Supplementary Table S2) as compared to those in ACD animals.

Anti-inflammatory effect of erythropoiesis-stimulating agent/LDN treatment

As erythropoietin has recently been shown to inhibit pro-inflammatory immune effector pathways39 and because hepcidin exerts immune modulatory effects4140 we questioned whether part of the erythropoiesis-stimulating activity of ESA and/or LDN could be traced back to alterations of immune activation states. Indeed, we found that ESA significantly reduced TNF-alpha mRNA expression in the spleen (P<0.05, Figure 7A), and this effect was even more pronounced when combined ESA/LDN was administered (P<0.001, Figure 7A). Similarly, ESA/LDN treatment significantly inhibited spleen IFN-gamma (P<0.01, Figure 7B) and IL-6 (P<0.01, Figure 7C) mRNA expression. To rule out that this effect was due to substitution of macrophages by hematopoietic cells in the spleen we also studied TNF-alpha (Online Supplementary Figure S8A) and IL-6 (Online Supplementary Figure S8B) mRNA expression in the liver and found comparable alterations of their expression following treatment with ESA and LDN.

Figure 7.Effects of ESA or ESA/LDN treatment on spleen cytokine expression in ACD rats. ACD was induced by injection of PG-APS in rats on day 0 and treatment with darbepoetin alfa (ESA) or a combination of ESA and LDN was started on day 21. On day 42 all animals were euthanized (n=7/group). Groups are ACD (gray), ACD-ESA (diagonal lines) and ACD-ESA/LDN (dots and lines). Data are presented as lower quartile, median and upper quartile (boxes) and minimum/maximum ranges (whiskers) (n=7 for each group). Spleen TNF-alpha mRNA (A), spleen IFN-gamma mRNA (B) and spleen IL-6 mRNA (C), as measured by real-time reverse transcriptase polymerase chain reaction, were normalized to the expression of the housekeeping gene Gusb and expressed as abundance relative to the mean of ACD (arbitrary units). Statistical significance of differences between groups was calculated using ANOVA with the Bonferroni-Holm correction for multiple testing. Data were logarithmized to reach homogeneity of variance.

Discussion

Here we provide evidence of two important, clinically relevant functions of the master regulator of iron homeostasis, hepcidin. First, in predicting the response to ESA therapy in inflammatory anemia and second in showing that inhibition of hepcidin formation can accelerate and potentiate the hematologic response to ESA resulting in normalization of hemoglobin levels even in the presence of severe inflammation.

Erythropoiesis is the main consumer of iron in the human body, and the majority of iron needed for this process originates from macrophages which take up and degrade senescent erythrocytes resulting in re-utilization of iron.44426 The transfer of iron from macrophages to the circulation is largely controlled by the interaction of hepcidin with the iron export protein, Fp1. Thus, the high hepcidin levels found in ACD block macrophage iron egress and contribute to iron-restricted erythropoiesis.2210973

Here we demonstrate in a rat model of ACD that higher serum hepcidin levels predict a poorer hematologic response to ESA treatment. However, the benefit/risk-ratio of ESA therapy of anemia is still a matter of discussion,3230 which makes the availability of predictive diagnostic markers desirable.

The negative correlation between high circulating hepcidin levels and the response to ESA therapy is most likely related to the fact that higher hepcidin activity is associated with a reduced availability of iron for erythropoiesis. Our observation is in agreement with the observation by Prentice et al., who found that hepcidin is the major predictor of iron incorporation into red cells in anemic children in Africa.45 As hepcidin is regulated by various stimuli such as hypoxia, anemia, iron and inflammation,17153 it is important to consider which regulatory mechanism underlies the poor predictive effect of high hepcidin levels for the response to ESA therapy.

Importantly, hepcidin is differently regulated by iron and inflammation and recent evidence suggests that iron deficiency dominates over inflammation-mediated regulation of hepcidin expression in animal models of inflammatory anemia and in patients with inflammation or undergoing phlebotomy.47462721

Thus, the negative association between ESA responsiveness and hepcidin levels may be due to enhanced inflammation and cytokine activities which promote macrophage iron retention and erythrophagocytosis, but also block ESA activity via inhibition of erythropoietin receptor expression and/or functionality on erythroid progenitor cells and presumably reduce the expression/activity of erythroid regulators of hematopoiesis such as GDF1548 along with apoptosis-inducing effects of several cytokines toward erythroid progenitor cells.112

However, serum hepcidin might prove to be a good marker for identifying those ACD patients who may respond to ESA therapy, thereby avoiding unnecessary treatment cycles. Nonetheless, the predictive diagnostic value of hepcidin may be different in other disorders such as chronic renal failure, in which higher hepcidin levels are also a consequence of reduced renal clearance.22

We demonstrated that ESA treatment reduced serum hepcidin and liver Hamp mRNA expression in rats with ACD. In accordance with the changes in hepcidin levels, ESA treatment of ACD rats reduced iron storage in the spleen as evidenced by reduced ferritin levels and increased iron mobilization from macrophages of ACD animals.

However, this ESA-driven iron mobilization was insufficient, as demonstrated by a non-significant increase of serum iron levels and unaffected MCV and MCH, which also translated into minimum improvement of anemia. Importantly, high hepcidin levels were associated with a reduced therapeutic efficacy of ESA, which prompted us to study the combined effects of ESA and the pharmacological inhibitor of hepcidin expression, LDN.2120

Compared to both LDN and ESA monotherapy, the combination of LDN and ESA resulted in an improved and faster hematologic response, and hemoglobin levels reached the threshold observed in non-inflamed control animals. The LDN/ESA combination therapy was also superior to the monotherapies in increasing MCV and MCH and expanding the number of erythroid cells and orthochromatic cells/reticulocytes in the bone marrow.

This is compatible with the observation that hepcidin inhibition by shRNA-mediated knock down prior to induction of acute anemia upon injection of heat-inactivated Brucella melitensis into mice or application of anti-hepcidin antibodies could partly prevent anemia development upon concomitant injection of ESA.19 However, in this model neither an effect of hepcidin inhibition alone nor an effect of the combined treatments toward improvement of an existing anemia could be demonstrated. Our results thus provide novel evidence of the potential therapeutic efficacy of this combined treatment in a rodent model which resembles the pathophysiology of human ACD.

This leads to questions regarding the mechanism underlying the improved therapeutic efficacy of combination treatment.

Combining ESA/LDN treatment improved the hemoglobin concentration and significantly affected serum iron, MCV and MCH, indicating that ESA therapy benefits from supportive iron mobilization through hepcidin antagonism in ACD. Furthermore, ESA/LDN treatment reduced the expression of pro-inflammatory cytokines, such as IL-6, TNF-α and IFN-γ. These proteins contribute to inflammatory anemia by promoting macrophage iron retention,8 blocking duodenal iron absorption,49 reducing erythrocyte half-life,50 impairing erythroid progenitor proliferation and differentiation112 and/or blocking the biological activity of erythropoietin.1312 IFN-γ, a T-helper cytokine, causes anemia by reducing erythrocyte half-life,50 by blocking the proliferation of erythroid progenitor cells,112 or by reducing macrophage iron egress via inhibition of Fp1 transcription.8 Mechanistically, part of these anti-inflammatory effects can be traced back to erythropoietin-mediated inhibition of pro-inflammatory nuclear factor-κB-driven immune effector pathways in macrophages.39 Accordingly, erythropoietin has been demonstrated to improve the clinical course of inflammation-driven auto-immune disorders by this pathway.39

Interestingly, we found a more sustained reduction of pro-inflammatory cytokines with combined ESA/LDN treatment than with ESA alone. The blockage of hepcidin expression and/or the BMP pathway by LDN may thus exert anti-inflammatory effects. This is supported by the finding that hepcidin by itself causes immune modulatory effects.41 A BMP-4-dependent stress erythropoiesis was shown to be partly responsible for erythropoietin-induced improvement of hemoglobin levels in zymosan-induced anemia.28 However, the fact that the BMP receptor inhibitor LDN strongly improves the erythropoietin response rules out that this mechanism is responsible for the anti-inflammatory effects observed in our model.

Apart from reducing hepcidin expression by LDN, other therapeutic strategies to block the biological activity of hepcidin, such as soluble hemojuvelin, anti-hepcidin antibodies or spiegelmers may also prove efficient at improving the therapeutic response to ESA therapy.2118

In summary, we provide evidence that predicting the response to ESA therapy by measuring pre-treatment hepcidin levels and applying a combined treatment of ESA and hepcidin-antagonizing agents can improve the therapeutic efficacy of ESA therapy. Secondary benefits could be a reduction of dosages, costs and, most importantly, undesired or serious adverse effects of ESA treatment. Once anti-hepcidin strategies become clinically available this strategy of combination therapy would have to be studied in a randomized prospective fashion also keeping in mind that it would be important to evaluate the effect of correcting the anemia on the course of the diseases underlying ACD.

Acknowledgments

The authors would like to thank the Austrian Science Fund (FWF) for its support for GW(TRP-188) and IT (P-24749-B13), the Tiroler Wissenschaftsfonds (TWF) for supporting MT (UNI-0404/765) and the “Verein zur Förderung von Forschung und Weiterbildung in Infektiologie und Immunologie, Innsbruck” (GW).

Footnotes

• The online version of this article has a Supplementary Appendix.
• Authorship and Disclosures Information on authorship, contributions, and financial & other disclosures was provided by the authors and is available with the online version of this article at www.haematologica.org.

• Received October 10, 2013.
• Accepted May 21, 2014.