Erythropoietin Biology in Cancer

Recombinant human erythropoietin (rEpo) has been used frequently in cancer patient care for the prevention or treatment of anemia associated with cancer. The biological effects of Epo are not limited to the hematopoietic system as many recent studies have shown that Epo is a pleiotropic cytokine that is proangiogenic and also exerts broad tissue-protective effects in diverse nonhematopoietic organs. The ability of rEpo to stimulate physiologic and pathologic angiogenesis and the expression of its cellular receptor EpoR in cancer cells and vascular endothelium have suggested that this hormone may exert direct effects on tumor growth and angiogenesis. Although many clinical trials evaluating the beneficial effects of rEpo in cancer care are positive (1–4), and a recent meta-analysis suggests a favorable effect on survival (5), two recent randomized trials reported that patients in the rEpo arm fared worse than their placebo-treated counterparts with respect to progression-free survival (6, 7). This controversy provides an impetus to step back and review the biology of Epo and EpoR, focusing on potential functional consequences of Epo-mediated signaling in nonhematopoietic tissues, including cancer cells.

Epo is a 30.4 kDa glycoprotein produced primarily in the adult kidney under the control of an oxygen-sensing mechanism and regulates the daily production of 2 × 10¹¹ RBC in the adult bone marrow to maintain the oxygen-carrying capacity of peripheral blood under physiologic conditions (8). Low tissue oxygen tension induces EPO gene expression through both transcriptional activation and mRNA stabilization (9). The hypoxia-dependent up-regulation of EPO is a direct result of hypoxia inducible factor-1 activation, a transcription factor that binds a hypoxia responsive element in the 3′ flanking region of the EPO gene. Besides hypoxia, there are several factors that modulate Epo production, such as hypoglycemia, increased intracellular calcium, insulin release, estrogen, androgenic steroids, and various cytokines (8). Localized Epo production has been shown in neural tissue (10), in the female genital tract (11), in placenta (12), and in testis (13). Epo production by erythroid progenitors has been implicated in an autocrine-paracrine mechanism of erythropoiesis regulation (14). In the uterus and oviduct, Epo production is inducible by estrogen (11, 15).

The biological effects of Epo in hematopoietic cells are mediated through its binding to EpoR, its specific cell surface receptor. EpoR is a member of the type I cytokine receptor family that includes cellular transmembrane receptors for factors such as granulocyte-colony stimulating factor, many of the interleukins, prolactin, and growth hormone. Epo-EpoR signaling is associated with activation of a cytoplasmic, nonreceptor protein tyrosine kinase Jak2 (16) and the downstream signaling molecule Stat5, a cryptic cytoplasmic transcription factor that plays an important role in the regulation of in vivo erythropoiesis (17). Epo-EpoR–induced signaling through pathways other than Jak-Stat, such as phosphatidylinositol 3-kinase-Akt, has been associated with primary erythroblast survival signaling (18), and the activation of the mitogen-activated protein kinase pathway is required for synergistic expansion of erythroid progenitor and precursor cells in response to Epo and stem cell factor (19).

Functional EpoR expression has been documented in many nonhematopoietic cell types, including vascular endothelial cells (20), smooth muscle cells (21), skeletal myoblasts (22), cardiac myocytes (23), neurons (24), retinal photoreceptors (25), liver stromal cells (26), placenta (27), kidney (28), and macrophages (29). EpoR expression and signaling in hematopoietic tissues is essential for normal mammalian erythropoiesis during development. Targeted disruption of the genes encoding either EPO or EPOR in mice has revealed that knockout mouse embryos die in utero due to defective fetal liver erythropoiesis (30). EPO or EPOR knockout embryos also exhibit defects in angiogenesis (31) and cardiac morphogenesis with increased apoptosis in endocardium and myocardium (32). However, a recent study in which erythroid EpoR expression was rescued showed that EPOR expression is not required for normal heart development in the mouse (33).

Abnormal regulation of Epo-EpoR signaling in hematopoietic cells has been associated with proliferative disorders of the bone marrow characterized by primary erythrocytosis. Primary familial and congenital polycythemia, a disorder characterized by erythrocytosis and Epo hypersensitivity of erythroid progenitor cells, may occur as a consequence of dominant gain-of-function mutations in EpoR. Hematopoietic cells that express these naturally occurring, mutant EpoRs exhibit increased Epo sensitivity associated with deregulation of the rates of Jak2 and Stat5 inactivation (34). More recently, constitutive activation of Jak2 as a consequence of a missense mutation in its pseudokinase domain has been shown to underlie the Epo-independent erythropoiesis that is characteristic of polycythemia vera, a myeloproliferative disorder that results in erythrocytosis (35).

Distinct from its essential role in the regulation of RBC production, Epo is involved in diverse nonhematopoietic biological functions. Epo-EpoR signaling is important for physiologic angiogenesis in the developing mouse embryo (31, 32). Furthermore, a functional Epo-EpoR system has been documented in the female genital tract where endogenous Epo-EpoR signaling was implicated in regulation of physiologic cyclic uterine angiogenesis (11). In another study, Epo-EpoR signaling was shown to be involved in the physiologic wound healing cascade by promoting angiogenesis and granulation tissue formation, an effect that was inhibited by neutralizing anti-Epo antibodies and recombinant soluble EpoR, an antagonist of Epo-EpoR signaling (29). In the central nervous system, where Epo is produced by astrocytes and EpoR is expressed by neurons, administration of soluble EpoR was associated with augmentation of ischemic brain injury in rats, demonstrating that Epo plays an important role in the response of the brain to neuronal injury as part of an innate response to stressors (36). A recent study showed that Epo-EpoR signaling is a potent angiogenic factor involved in the pathologic angiogenesis of diabetic retinopathy (37). The potential role of Epo in tumor angiogenesis, a process that is essential for tumor progression and metastasis, is not established.

The exogenous administration of rEpo has been associated with diverse effects in nonhematopoietic tissues. For instance, expression of EpoR in kidney, muscle cells, and the intestine is associated with the ability of rEpo to induce cellular proliferation (22, 28, 38). In other studies, Epo has proved to be a proangiogenic cytokine in a variety of experimental systems. rEpo stimulates the proliferation and migration of endothelial cells (39), and these proliferative effects are associated with increases in endothelin-1 and cytosolic free calcium (40, 41), as well as tyrosine phosphorylation of intracellular proteins, including Stat5 (42). Ribatti et al. (43) reported that the proangiogenic activity of Epo was similar to that of fibroblast growth factor-2 in the chick chorioallontoic membrane assay, and Jaquet et al. (44) found the angiogenic potential of Epo to be similar to that of vascular endothelial growth factor when stimulating human adult myocardial endothelial cells.

rEpo has emerged as a major tissue-protective cytokine as many studies have documented that administration of rEpo modulates the physiologic response to various forms of tissue injury (45). In the central nervous system, Epo improves neuronal survival by decreasing neuronal apoptosis and inflammation, exhibiting a protective effect in a variety of injury models, including cerebral ischemia, trauma, and excitotoxins (46). Furthermore, a clinical trial of rEpo in acute ischemic stroke patients showed improvements in clinical recovery scores and outcomes (47). In the cardiovascular system, rEpo protects cardiac myocytes against ischemic injury in a number of different preclinical model systems (48). rEpo has also been shown to protect the kidney from both ischemic and toxic injuries (49, 50) as well as exhibiting significant therapeutic value in a rodent model of diabetic neuropathy (51).

Recent reports describing Epo variants, including asialo-Epo (52) and carbamylated Epo, that retain the protective effects of Epo in nonhematopoietic tissues while exhibiting no effect on hematopoietic cells have suggested the presence of fundamental mechanistic differences between Epo-mediated cellular signaling in hematopoietic versus nonhematopoietic cells (53, 54). The tissue-protective effects of Epo in the brain and heart seem to require a second distinct receptor component, the β_c-subunit of the interleukin-3 receptor, also shared by interleukin-5 and granulocyte-macrophage colony stimulating factor. In one study, a physical association between β_c-subunit and EpoR was reported in coimmunoprecipitation experiments (55). The exact structure of the cell surface receptor complex that mediates the biological effects of Epo in nonhematopoietic cells remains to be characterized. The availability of tissue-protective Epo variants devoid of erythropoietic stimulatory effects may prove clinically useful by abrogating concerns over undesirable elevation of the RBC mass.

Anemia is an independent prognostic factor for survival in patients with cancer (56). The European Cancer Anaemia Study recently reported that 72% of patients with hematologic malignancies and 66% of solid tumor patients were anemic at some point during the course of their disease although the exact prevalence of anemia varies according to the type of neoplasm (57). The pathophysiology of malignancy-related anemia can be multifactorial. In addition to anemia of chronic disease, nutritional deficiencies, bleeding, hemolysis, bone marrow involvement with malignant cells, and chemoradiotherapy can all contribute to anemia.

There is evidence that anemia may be associated with a diminished response to radiotherapy (58), chemotherapy (59), and surgery (60). Although a direct relationship between acute anemia and intratumoral hypoxia has been reported (61), the effect of chronic anemia is more difficult to interpret (62). Brizel et al. (63) found only a weak correlation between anemia and poor tumor oxygenation in head and neck cancer patients and many nonanemic patients had hypoxic tumors. Tumor hypoxia is common among a wide variety of malignancies and is a factor associated with treatment resistance, aggressive clinical phenotype, and poor prognosis (64, 65).

Beyond correcting anemia in preclinical anemic tumor models, rEpo has been shown to restore radiosensitivity, increase tumor growth delay (66–70), and increase cytotoxicity of chemotherapeutics (71). In a preclinical model, Blackwell et al. (72) found that systemic rEpo administration to nonanemic rats bearing mammary adenocarcinoma flank tumors improved tumor oxygenation independent of the effects on hemoglobin levels. Other studies have similarly shown a decrease in the hypoxic fraction of solid tumors with systemic rEpo treatment: one in an i.p. hemorrhagic ascites model of anemia in mice (73) and another in a total body irradiation anemia model in rats in which radiotherapy efficacy was also increased by rEpo administration (67). Epo alone was not associated with modulation of microvessel density or tumor growth rate in these studies.

The available rEpo preparations include epoetin-α, epoetin-β, and the longer-acting darbepoetin-α, and are effective in increasing hemoglobin levels, decreasing the need for red cell transfusions, and improving quality of life in cancer patients receiving therapy (1–3, 74). In several clinical trials, improvements in survival were suggested with rEpo administration to anemic cancer patients. In a retrospective study, significant improvements in response, control, and survival rates were observed in patients with squamous cell carcinomas of the oral cavity and oropharynx treated with chemoradiation and rEpo (75). Trends toward improved survival with rEpo treatment of anemic cancer patients were also reported in a randomized, double-blind, placebo-controlled trial of nonmyeloid malignancies (4) and a study of lung cancer patients receiving chemotherapy (74), although the latter was not sufficiently powered to evaluate survival. In contrast, two recent clinical trials reported adverse outcome associated with rEpo therapy of metastatic breast cancer patients undergoing chemotherapy (7) and head and neck cancer patients undergoing radiotherapy (6). The unfavorable survival rate in rEpo-treated patients was associated, at least in part, with disease progression in both trials. Additionally, rEpo administration seemed to be associated with an increased incidence of deep venous thrombosis in the breast cancer trial, as well as in a trial of concurrent chemoradiation in cervical cancer patients (76). Several design- and treatment-related issues have been pointed out in the breast and head-neck cancer trials, making their interpretation difficult. In contrast to the results of these trials, a recent meta-analysis provided suggestive, but inconclusive, evidence that rEpo may improve overall survival in anemic cancer patients (5). Further prospective, randomized, and controlled studies will be required to investigate the effects, if any, of rEpo therapy on disease progression and survival of cancer patients.

Several recent studies have reported expression of EpoR in tumor cell lines as well as primary cancers (Table 1). Many tumors were also reported to express Epo mRNA transcripts and protein, suggesting the potential for generation of an autocrine or paracrine growth-stimulatory Epo-EpoR loop in cancer cells. The expression of Epo in tumor cell lines was reported to be hypoxia inducible (77, 78). The expression of EpoR splice variants was also shown in cancer cell lines, which could be important in modulating the cellular effects of Epo (79). Furthermore, the expression of EpoR in vascular endothelium in tumors has suggested potential effects of Epo on the tumor microenvironment, such as the stimulation of tumor angiogenesis. In addition, it has been suggested that the potential for direct activation of Epo-EpoR signaling pathways in cancer cells may be associated with modulation of various aspects of tumor biology, including cellular proliferation, apoptosis, and sensitivity to chemoradiation therapy (Fig. 1).

There have been in vitro studies that reported increased proliferation of tumor cells in response to exogenous rEpo. Acs et al. (77) reported a ∼25% increase in the proliferation of monolayer cultures of MCF7 and BT-549 breast cancer cells. Similar findings were reported with Caki-2, 786-0, and RAG renal cell carcinoma cells (80). A limited rEpo-mediated induction of proliferation and stimulation of invasion was reported in human head and neck squamous cell carcinoma cell lines (81). Further, Mohyeldin et al. (82) reported a correlation between disease progression and EpoR expression in head and neck cancer biopsies, and an increase in invasiveness in head and neck cancer cells in response to rEpo. In contrast, however, Westphal et al. (83) reported no proliferative effects of exogenous rEpo on six EpoR-positive tumor cell lines, and Selzer et al. (84) reported that EpoR-positive human melanoma cells did not proliferate in response to rEpo. Similarly, in a study of 22 tumor cell lines derived from a range of human solid tumors, rEpo exerted no growth-modulating effects (85). Taken together, these findings suggest that exogenous Epo does not consistently provide a proliferative stimulus for cancer cells. Whether the contrasting data with respect to the proliferative effects of Epo are due to differences in the cancer cell types tested or the experimental models is unclear. Further studies in xenograft models of human cancer will be required to evaluate the effects of Epo on tumor growth rates in the presence or absence of concomitant chemoradiotherapy.

There is an accumulating body of experimental evidence for the presence of functional endogenous Epo-EpoR signaling in tumors from studies that used strategies to block Epo signaling pathways. Yasuda et al. (86) reported that Epo signaling contributes to the growth and angiogenesis of female reproductive tract tumors. Blockade of Epo signaling with local soluble EpoR or anti-Epo antibody resulted in tumor cell destruction and reduction of vascularity in ovarian and uterine cancer xenografts, associated with an increase in apoptotic death of both tumor cells and vascular endothelial cells. It was also shown that i.p. injections of an EpoR antagonist blocked Stat5 phosphorylation and inhibited melanoma and stomach choriocarcinoma tumor cell survival and angiogenesis (87). In another study, Arcasoy et al. (88) found that administration of Epo-EpoR inhibitors, including neutralizing anti-Epo antibody, soluble EpoR, or pharmacologic Jak2 inhibitor in rat mammary adenocarcinoma tumors in a tumor chamber model, resulted in significant tumor growth delay. These preclinical data suggest that the exploration of strategies to block Epo-EpoR function to target tumor growth and angiogenesis may be warranted.

Several studies reported Epo-mediated modulation of tumor cell sensitivity to apoptosis and investigated the ability of Epo to influence the effects of chemoradiotherapy with contrasting results, in both in vitro or in vivo experimental models. Batra et al. (78) showed increased expression of antiapoptotic genes (bcl-x_L, bcl-2, and mcl-1), as well as an increase in nuclear factor-κB DNA binding activity in Ewing sarcoma family tumor and neuroblastoma cell lines in response to rEpo. This finding was associated with increased secretion of proangiogenic cytokines, which promoted endothelial cell proliferation and chemotaxis. In human melanoma cells, incubation with rEpo increased tumor cell resistance to hypoxia-induced cell death (89). These cells also showed increased mitochondrial membrane potential, a possible mechanism for the observed antiapoptotic effect. In a study of HeLa uterine cervix carcinoma cells, Acs et al. (90) reported rEpo-mediated, dose-dependent reduction in apoptosis and increase in the surviving fraction of tumor cells treated with cisplatin. In another study, overexpression of EpoR in HeLa cells was associated with rEpo-mediated activation of nuclear factor-κB and a 50% increase in plating efficiency of HeLa cells, although this effect was not associated with a change in radiation sensitivity (91). In human melanoma cells, rEpo increased cell viability during treatment with varying concentrations of dacarbazine and cisplatin, although rEpo-induced resistance to dacarbazine was higher than to cisplatin (89). Using U87 human malignant glioma and HT100 primary cervical cancer cell lines, Belenkov et al. (92) reported an increase in both radiation and cisplatin resistance in response to exogenous rEpo, an effect that was reversible upon addition of a Jak2 inhibitor.

Although these data suggest that rEpo may protect cancer cells against the effects of chemoradiotherapy, other studies have reported contrasting findings. For instance, human renal cell carcinoma and myelomonocytic leukemia cell lines treated with rEpo exhibited an increase in apoptosis in response to daunorubicin and vinblastine, which was associated with inactivation of the nuclear factor-κB pathway (93). In a preclinical myeloma model, rEpo was associated with stimulation of antitumor immune responses resulting in tumor regression (94). In a more recent study, rEpo administration was associated with remodeling of tumor microvessels and increased chemosensitivity to 5-fluorouracil treatment of human squamous cell and colorectal carcinoma xenografts but without a stimulatory effect on the in vitro proliferation of cancer cells (95). Further well-controlled studies in xenograft models of different types of cancer are required to characterize the ability of rEpo to modulate tumor growth, angiogenesis, chemoradiation sensitivity, and apoptosis.

Epo is a pleiotropic cytokine that exerts diverse biological effects in many nonhematopoietic tissues. Epo is involved in the wound-healing cascade, functions as a proangiogenic cytokine during physiologic angiogenesis in the embryo and uterus, and exerts tissue-protective effects as part of an innate response to stressors. The recent characterization of Epo variants, such as asialo-Epo and carbamylated-Epo, that retain nonhematopoietic, tissue-protective properties of Epo without stimulating erythropoiesis has uncovered new areas of research into mechanisms of Epo-mediated signaling in nonhematopoietic tissues as well as novel clinical applications for rEpo and its derivatives in disorders other than anemia. rEpo has been widely used in the prevention and treatment of cancer-related anemia, leading to increased hemoglobin levels, reduction of RBC transfusion requirements, and improvement of quality of life. Despite these beneficial effects and some studies suggesting improved survival trend in Epo-treated patients with cancer, two recent prospective, randomized clinical trials involving head-neck and breast cancer patients have raised concerns over potential adverse effects of rEpo in cancer patients. The expression of EpoR in cancer cells has suggested the possibility that exogenous rEpo may exert direct effects on tumor cells associated with stimulation of proliferation, inhibition of apoptosis, or modulation of sensitivity to chemoradiation therapy. The presence of an autocrine-paracrine Epo-EpoR system in tumors and possible effects of Epo on tumor microenvironment and angiogenesis are consistent with a complex biology for Epo-EpoR signaling in cancer that requires further research. The overall direct effect of Epo-EpoR signaling is clearly not a straightforward one, as signaling can potentially activate several pathways important to tumor behavior and response to treatment. Further, the expression of EpoR alone is not always sufficient to modulate these pathways as shown by the variable effect of rEpo on chemoradiation sensitivity and the absence of a consistent proliferative effect in different types of cancer cells. The potential development of erythropoiesis-stimulating agents that are devoid of the biological effects of Epo in nonhematopoietic cells may constitute a future strategy for selective targeting of erythropoietic agent therapy. Characterization of the mechanisms of the biological effects of Epo on nonhematopoietic cells, in particular cancer cells, should continue to be explored to optimize the use of Epo in anticancer therapy.