Erythropoietins: A common mechanism of action

Erythropoiesis 

A primary function of RBCs is to transport O2 from the lungs to O2-dependent tissues. Changes in O2 levels necessitate both acute and long-term physiologic adaptations. Acute adaptations include increases in respiration and heart rate, vasoconstriction, and changes in blood volume; however, these changes cannot be sustained. Erythropoiesis is a longer—term adaptation to boost O2-carrying capacity by increasing the concentration of RBCs, and thus, hemoglobin (Hb) concentration.

RBCs are the most abundant (∼99%) circulating cells in the bloodstream, representing 40% to 45% of total blood volume. In a healthy human with ∼5 L blood, this represents approximately 2.5×1013 cells, a quantity substantial enough to provide the large O2 transport capacity needed to support aerobic respiration. In humans, the RBC lifespan is ∼100 to 120 days, with a daily loss of ∼0.8% to 1.0% of circulating RBCs. To match this loss, the body assumes a normal, prodigious production capacity of ∼2.5×1011 cells/day.

RBC production results from a tightly controlled proliferation and differentiation pathway (Fig. 1). Early hematopoietic progenitors differentiate into burst-forming unit–erythroid cells, in which EPORs appear for the first time; however, EPO is not required at this stage [16]. Burst-forming unit–erythroid cells differentiate into colony-forming unit–erythroid cells, which are dependent on EPO for survival, and there is a corresponding rise in expression of EPORs 17, 18. Continued stimulation with EPO triggers differentiation into erythroblasts, which enucleate to form reticulocytes and after a few days show loss of “reticulin,” resulting in RBCs. Reticulocytes and RBCs stop expressing EPOR and cease being responsive to EPO [18].

View Large Image Download to PowerPoint

Figure 1 Erythropoiesis. BFUe=burst-forming unit–erythroid; CFUe=colony-forming unit–erythroid; Epo=erythropoietin; FLK=fetal liver kinase; FLT=fetal liver tyrosine kinase; G-CSF=granulocyte-colony stimulating factor; GEMM=granulocyte, erythrocyte, monocyte, megakaryocyte; GM-CSF=granulocyte macrophage CSF; IL=interleukin; MIP=macrophage inflammatory protein; SCF=stem cell factor; TGF=transforming growth factor; TPO=thrombopoietin.

Disease states and environmental conditions often alter the tightly controlled balance between RBC production and destruction. When RBC loss exceeds gain, anemia results. Increased RBC loss can occur because of bleeding, enhanced destruction (chemically induced hematotoxicity), or reduced lifespan (sickle cell anemia). Potential causes of insufficient RBC production include defects in O2 sensing, excess of erythropoiesis inhibitors, and inadequate concentrations of ESAs.

EPO 

Erythropoiesis primarily occurs in the kidney, but other organs (liver, brain) also produce EPO. Interstitial fibroblasts produce EPO in the kidney 19, 20, 21, while hepatocytes produce EPO in the liver [22]. Initially, EPO is synthesized as a 193-amino-acid precursor. A 27-amino-acid signal peptide and C-terminal arginine are removed, and carbohydrate is added to three N-linked glycosylation sites and one O-linked glycosylation site [23]. The secreted protein contains 165 amino acids and is heavily glycosylated, with ∼40% of its mass composed of carbohydrate. The structure of rhEPO is a compact globular bundle that contains four α helices (Fig. 2).

View Large Image Download to PowerPoint

Figure 2 The nuclear magnetic resonance minimized average structure of human erythropoietin.

Generally, serum EPO concentrations of 10 to 25 mU/mL [24] maintain Hb levels within the normal range of 12 to 17g/dL [25]. The terminal half-life (t1/2) of EPO is ∼5hours [26], which requires an average EPO production rate of ∼2 U/kg/day. The EPO production rate per cell appears constant [27], with fluctuations in EPO synthesis resulting from changes in the number of cells producing the molecule. In cases of severe anemia, circulating EPO levels can increase up to 1000—fold because of a logarithmic increase in the number of cells producing EPO 24, 27. Other factors affecting EPO levels include iron availability, nutritional status, disease or comorbidities, environmental conditions, and genetic factors (congenital polycythemias).

EPO-RBC time–response relationship 

A direct correlation exists between RBC production and serum EPO concentrations 28, 29. However, the rate of erythropoiesis change (∼4—fold) [29]) is small compared to the larger change in EPO concentrations (∼1000—fold) [30]. Thus, the magnitude of increase in RBC concentration is primarily controlled by the length of time EPO concentrations are maintained, and not by the EPO concentration level per se (Fig. 3). Increased EPO synthesis has a prolonged effect due to the disproportionate relationship between EPO t1/2 and RBC lifespan. Thirty minutes of hypoxia can result in production of EPO (t1/2 ∼5hours) [26]. In turn, EPO stimulates formation of enucleated reticulocytes (t1/2=1–5 days) 31, 32, which rapidly mature into RBCs that have a long lifespan (100–120 days) [33]. Thus, a short duration of EPO exposure results in a prolonged increase in RBC concentration.

View Large Image Download to PowerPoint

Figure 3 Disproportion between half-life of recombinant human erythropoietin and lifespan of red blood cells. MEC=minimum effective dose. Adapted from Molineux [108] with permission.

EPO and EPOR 

The mechanism of action by which EPO stimulates erythropoiesis has been under extensive investigation. Early evidence indicated that EPO interacted with a protein on the cell surface, triggering activation of the JAK-signal transducers and activators of transcription, phosphatidylinositol 3 kinase, and mitogen-activated protein kinase pathways (Fig. 4), resulting in the proliferation and terminal differentiation of erythroid precursor cells and providing protection from apoptosis [4]. The EPO-binding component on cells was first detected by measuring physical attachment of radiolabeled EPO to erythroid precursor cells 18, 34. The EPOR gene was subsequently identified by expression cloning and found to be a single gene with no apparent homologs 35, 36.

View Large Image Download to PowerPoint

Figure 4 Signal transduction pathways of the erythropoietin receptor. Binding of erythropoietin (EPO) causes conformational changes to the EPO receptor, transphosphorylation of associated JAK2 molecules, phosphorylation of tyrosine residues in the cytoplasmic tail of the receptor, and phosphorylation or activation of signaling molecules. Phosphorylation of signal transducers and activators of transcription (STAT) 5 transcription factor (TF) causes homodimerization, translocation to the nucleus, and activation of genes for antiapoptotic molecules. Phosphorylated phosphatidylinositol 3-kinase (PI-3 kinase) phosphorylates protein kinase B (PKB)/Akt. PKB/Akt: 1) phosphorylates and inactivates proapoptotic molecules (Bad, caspase 9 or glycogen synthase kinase-3b [GSK-3b]); 2) phosphorylates FOXO TF, inhibiting translocation to the nucleus and activation of target genes (Fas ligand, Bim); and 3) phosphorylates IκB, allowing the release of the transcription factor nuclear factor (NF)-κB that then translocates into the nucleus and activates target genes encoding antiapoptotic molecules (XIAP, c-IAP2). Binding of EPO to its receptor also activates Hsp70, which binds to and inactivates proapoptotic molecules (apoptosis protease-activating factor-1 [Apaf-1], apoptosis-inducing factor [AIF]).

While other components may mediate affinity or aid in signal transduction, the activation of signal transduction is initiated by an early, direct interaction of EPO with EPOR. Activation of EPOR occurs following cross-linking of two EPORs via one EPO ligand 37, 38, 39, 40, which induces a conformational change in the receptor, triggering downstream signal transduction 41, 42, 43.

Anemia treatment and development of rhEPO 

Before development of rhEPO, blood transfusion was the most common treatment for patients with anemia. However, blood transfusions carry inherent risks, including risk of transmission of infectious agents and iron overload. Additionally, the blood supply is limited, and immune reactions developed after transfusion can make organ transplantation more problematic [50]. Iron supplementation was largely ineffective as a stand-alone treatment for anemia. The need for an effective anemia treatment option was obvious, and attempts to make and test rhEPO via cloning of the human EPO gene began.

Successfully cloning the EPO gene was difficult, as low circulating EPO levels made protein purification difficult, a primary source of EPO mRNA was not obvious, and mRNA was difficult to obtain. Once small quantities of purified human EPO became available (10mg from 1000 L urine from human patients with aplastic anemia), oligonucleotide probes for EPO were designed. Two different probes were used to screen a λ phage library containing sheared human genomic DNA, and the human EPO gene was cloned [5].

rhEPO—The first ESA 

Successful cloning of the EPO gene in 1983 [5] allowed for the large-scale production of rhEPO and its subsequent clinical use [29]. Epoetin alfa (Epogen; Amgen Inc., Thousand Oaks, CA, USA; Procrit; Ortho Biotech Products, L.P., Bridgewater, NJ, USA) was the first rhEPO commercialized in the United States, followed by a second epoetin α (Eprex; Ortho Biotech Products) and epoetin beta (NeoRecormon; F. Hoffmann-La Roche Ltd., Basel, Switzerland) in Europe. Epoetins alfa and beta, both produced by Chinese hamster ovary cells, have minor structural differences but the same physiological effects [51]. An epoetin produced in baby hamster kidney cells, epoetin omega, differs from previous epoetins in the glycosylation profile. More recently, epoetin delta (Dynepo, Shire plc, Basingstoke, UK), produced from an engineered human fibrosarcoma cell line HT1080, has been described [52]. As patents for epoetins α and β expire, follow-on or biosimilar epoetins (Binocrit [Sandoz International GmbH, Holzkirchen, Germany]; epoetin α HEXAL [Hexal Biotech Forschungs GmbH, Oberhaching, Germany], and Abseamed [Medice Arzneimittel Puetter GmbH & Co. KG, Iserlohn, Germany]), have been approved in Europe.

Development of additional ESAs 

Introduction of epoetins into clinical practice represented a milestone in anemia treatment, yet opportunities for further improvement in certain patient populations and clinical settings remained. The initial TIW dosing regimen for hemodialysis patients proved to be burdensome and inconvenient in other clinical settings such as CKD and oncology. Thus, research and development of additional treatment opportunities continued.

Design considerations 

While all ESAs have the same mechanism regarding EPOR activation, they differ in molecular structure, receptor binding affinity, serum t1/2, clearance, bioavailability, and in vivo potency. Together, these characteristics shape the clinical efficacy and safety of these agents, as well as their versatility, especially in terms of dosing schedules.

The clinical advantages of extended dosing regimens led to development of newer ESAs with a longer duration of action, but comparable clinical benefit. For example, extending the t1/2 may allow for extended-dosing schedules that offer patients and caregivers, especially those with CKD not on dialysis and cancer patients, the convenience of less frequent visits. Increased molecular stability may also decrease degradation products and the risk of immunogenicity, as well as allow for alternate storage opportunities or delivery systems (e.g., oral administration or slow delivery via pumps).

Potential limitations or trade-offs need to be considered. Too long a t1/2 can result in loss of Hb control or Hb cycling, while too short a t1/2 may limit the range of extended dosing regimens. Too high an affinity for the EPOR may affect the dose-response relationship, while insufficient binding may result in a slower or incomplete response. Additionally, differences in IV vs SC potency may result in the need for an increased drug dose with the less-efficient route. These factors must be considered when developing new ESAs and will vary with patient needs, indication, region, and clinical practice. Ultimately, these newer molecules should offer patients efficacy and safety at least comparable with the original ESAs.

Potential strategies 

Researchers have taken several approaches to reengineer and thereby extend treatment options with original ESAs. Prolonging the time of erythropoietic stimulation can increase response, which may allow for extending the dosing interval. Increasing the dose of faster—clearing molecules is one strategy, but this can be inefficient because of the high doses required. A more dose-conserving strategy is to increase serum t1/2 7, 9, 72. Administration of a long-acting ESA may extend serum residence time, resulting in an increased relative biological response over time [28].

Early attempts to improve rhEPO included enhancing molecular stability or increasing affinity for EPOR [73]. However, increased affinity did not necessarily increase in vivo biological potency, because serum t1/2 was the primary driver of the degree of response 9, 28, 74. The exception was modifications that yielded molecules with an affinity that was too low, resulting in ineffective EPOR dimerization and activation.

Many different approaches to extend the t1/2 have been considered, including addition of polyethylene glycol (PEG), hyperglycosylation, dimerization, and physical fusion of the peptide portion of the ESA to other molecules, such as antibodies or other proteins (e.g., albumin) 7, 75, 76. Successful approaches took advantage of the well-established safety and efficacy profile of rhEPO by modifying the “nonfunctional” regions without loss of structure or substantial reductions in affinity for EPOR. Any structural loss is of particular concern because of the increased potential for degradation or molecular instability of the drug with long-term storage, and hence, reduced potency and potentially increased risk of immunogenicity. Reduced affinity for EPOR may occur due to steric hindrance from the attachments (e.g., PEG 77, 78) or changes in electrostatic properties of the molecule (hyperglycosylation) 28, 79, which directly or indirectly inhibit binding.

Hyperglycosylation 

Both endogenous and recombinant EPO have microheterogeneity in carbohydrate structures with variation in sialic acid content—up to 14 attached sialic acid residues 74, 80, 81, 82. The importance of sialic acid content for clearance was noted in experiments with rhEPO isoforms, revealing a direct and positive relationship between sialic acid content and in vivo potency [74]. The molecules with increased sialic acid content had reduced affinity for the EPOR and increased serum t1/2, suggesting that a longer t1/2 was a stronger determinant of potency than was receptor affinity. Consequently, the mixture of glycoforms was important in defining the biologic properties of the particular rhEPO products. This observation also led to the hypothesis that “reengineering” of the EPO molecule, by adding more carbohydrate chains and increasing the sialic acid content (>14 residues), might prolong the serum t1/2 and enhance potency compared with rhEPO [7].

Darbepoetin α 

Darbepoetin α is a hyperglycosylated analog that contains two additional N-linked carbohydrate chains at positions 30 and 88, as a result of five amino acid substitutions (Ala30Asn, His32Thr, Pro87Val, Trp88Asn, Pro90Thr) (Fig. 5) [7]. The new carbohydrates did not directly interfere with receptor binding or disrupt the structure or stability of the molecule. The carbohydrate portion increased from 40% to 51%, and the approximate molecular weight increased from 30kDa to 37kDa. The theoretical maximum number of sialic acid residues increased from 14 to 22.

View Large Image Download to PowerPoint

Figure 5 Molecular structures of rhEPO (A) and darbepoetin alfa (B). Reprinted by permission from Macmillan Publishers Ltd: Nature Biotechnology [7], 2003. rhEPO=recombinant human erythropoietin.

Compared with epoetin α, darbepoetin α had a threefold longer serum t1/2, a lower receptor-binding affinity, and enhanced in vivo bioactivity in multiple species 7, 28, 72, 74. In a phase I trial in rhEPO-naïve patients receiving peritoneal dialysis [83], darbepoetin α administered IV had an approximately threefold longer mean terminal t1/2 compared with rhEPO (25.3 vs 8.5hours, respectively), a more than twofold greater area under the curve (291±8 vs 132±88 ng h/mL), and a fourfold lower biphasic clearance (1.6±0.3 vs 4.0±0.3mL/h/kg). The volume of distribution was similar for the two molecules (52.4±2.0 and 48.7±2.1mL/kg, respectively). The mean terminal t1/2 for darbepoetin α SC was longer than that for IV administration (∼49hours); Cmax averaged about 10% of the IV value, Tmax averaged 54±5hours, and mean bioavailability was 37%.

Relative potency of darbepoetin α compared with rhEPO increased when the dosing interval was extended. At TIW dosing, three times more rhEPO was needed than darbepoetin α to maintain or elicit a similar erythropoietic response in normal mice 9, 72. With QW administration, the difference increased 13-fold. With a single injection, the rhEPO dose needed to be 30 to 40 times higher to match the effect of the lower dose of darbepoetin α. In humans receiving rhEPO or darbepoetin α QW, the dose of rhEPO needed to be 45% higher than that of darbepoetin α to maintain Hb levels in CKD patients [84]. The differences in relative potencies likely resulted from the threefold longer serum t1/2 of darbepoetin α, allowing for a prolonged period of time for erythroid cell exposure to the ESA [28].

The apparent paradox—that darbepoetin α has lower receptor-binding affinity but increased in vivo activity—was explained by the counteracting effects of sialic acid–containing carbohydrate on clearance 9, 28. At extended time intervals, the darbepoetin α relative concentration greatly exceeded that of rhEPO. Thus, prolonged exposure more than compensated for the reduced receptor-binding affinity, resulting in increased in vivo activity.

Unlike rhEPO, studies with darbepoetin α in humans demonstrated that the dose efficiency after SC and IV administration was approximately equivalent, with similar doses providing similar efficacy (i.e., similar Hb increases) [85]. The apparent t1/2 of darbepoetin α administered SC is extended approximately twofold relative to IV administration, which compensates for decreased bioavailability [83].

PEGylation 

PEGylation, the addition of PEG to reactive regions of proteins or carbohydrates by either solution or solid-phase chemistry, has been used successfully to extend the serum t1/2 of many different recombinant proteins [86]. PEGylated molecules have an increased hydrodynamic size because a “water shell” surrounds the molecule. The increased hydrodynamic size can result in reduced clearance because of a reduced rate of translocation from blood to extravascular tissues [86]. PEG is thought to be relatively inert and nonimmunogenic and, thus, to be a suitable starting material for protein conjugate therapeutics. However, PEGylation of a molecule does not guarantee protection against an immune response [87].

One issue with drugs made by solution or solid-phase chemistry is poor specificity of conjugation in the chemical reaction, with generation of undesirable byproducts. Current chemistries typically target the reactive amino groups on lysine or the amino terminal amine [88]. Consequently, it is difficult to target specific reactive amine groups. PEGylated rhEPO molecules typically contain mixtures of molecules with PEG attached to different reactive amines, each of which may have differential effects on activity and protein folding. For example, if lysines are proximal to active sites involved in receptor binding, PEGylation can reduce binding activity and, consequently, in vitro activity, either by altering essential amino acids important in EPOR binding or by steric hindrance 89, 90, 91.

Low affinity for EPOR associated with amine chemistry may be overcome via other PEGylation strategies. Cysteine substitutions at targeted “nonfunctional” regions can allow addition of the conjugate with high specificity to the sulfhydryl group with retention of in vitro activity 89, 92, 93. Another strategy is to produce PEGylated EPO synthetically. During synthesis, a PEG—conjugated amino acid is introduced in place of the unconjugated amino acid, allowing targeting of particular amino acid positions for PEG attachment, such as the glycosylation sites, and reducing the potential for loss of in vitro activity 77, 94. Both strategies may allow for extended serum t1/2. However, it is unclear if these particular molecules will have similar stability, in vivo activity, and lack of immunogenicity to their glycosylated counterparts.

Impact of PEG ylation and hyperglycosylation on clearance 

Both hyperglycosylation and PEGylation of ESAs can reduce receptor-binding affinity, although the reduction is substantially greater for PEGylated vs hyperglycosylated rhEPO (50- to 100-fold vs 5-fold) 7, 9, 28, 78, 90, 91. The reduced EPOR binding activity results in corresponding decreases in in vitro potency with these molecules. It has been suggested that the reduced clearance of PEGylated and hyperglycosylated ESAs is due to their effect on receptor-mediated endocytosis and degradation, which has been reported in vitro 64, 91. However, clearance studies with rhEPO analogs with altered receptor-binding characteristics suggested PEGylated and hyperglycosylated ESAs have reduced clearance due to their impact on non-EPOR–mediated clearance pathways 58, 89, 94.

Hyperglycosylated or PEGylated rhEPO and analogs have increased hydrodynamic size due to attached hydrated carbohydrate or PEG [86]. Thus, the reduced clearance of these molecules may be partially explained by steric factors. For example, the transport of ESAs from the blood to the clearing organs may be reduced as the size of the molecule increases, as was reported for unconjugated PEG [86].

ESA use in clinical practice 

ESAs have been used successfully to treat anemia in many different patient populations, helping to prevent or minimize use of blood transfusions. ESAs have been used primarily to treat anemia associated with kidney failure and chemotherapy-induced anemia. Additional indications have been considered, including anemia associated with cancer (including myelodysplasia), anemia of chronic disease, anemia associated with AIDS treatment, and treatment in perisurgery indications. In some disease settings (e.g., end-stage renal disease and chronic kidney disease without dialysis), there is a consistent relationship between ESA treatment, increased hemoglobin, and improved quality of life 98, 99, 100, 101, 102, 103. This is particularly true for measures related to energy, physical function, and exercise. While results are mixed in the oncology setting, some studies suggest aspects of quality of life, such as fatigue, may be improved by the resultant increases in Hb levels [104].

Practice patterns vary with the indication and geographical region. Dosing frequency is often associated with visits to health care professionals and the duration of action of the ESA. When rhEPO first became available, it was used to treat anemia in patients on hemodialysis. Such patients may attend either a dialysis unit or a hospital twice or thrice weekly for hemodialysis, and frequent administration of ESAs that coincided with hemodialysis sessions was deemed practical. For CKD patients who are not on dialysis and for cancer patients receiving chemotherapy, extended dosing regimens may be more convenient and, in some cases, more cost efficient. Thus, subsequent clinical studies have investigated different dosing schedules (QW, Q2W, Q3W, and QM) across different clinical settings 11, 12, 13, 14, 105, 106. Current practice patterns, even when limited to the nephrology setting, vary considerably. Data from the Dialysis Outcomes Practice Patterns trial show a broad spectrum of ESA dosing patterns and hemoglobin outcomes in different countries (Table 1) [107].

Table 1. Hemoglobin and erythropoiesis-stimulating agent (ESA) therapy in prevalent hemodialysis patients from the Dialysis Outcomes and Practice Patterns Studya

		Hemoglobin (median)		ESA therapy (mean±SEM)			Doses per week (% of patients)
	Country	n	Hb (g/dL)	n	% of Patients treated SC	Dose (units/kg/week)	n	3×	2×	1×
	France	521	11.1	581	75±1.8	102±4.81	442	47	38	14
	Germany	565	10.9	586	77±1.73	86±2.92	467	71	17	11
	Italy	556	10.8	595	86±1.44	132±4.04	527	50	30	18
	Spain	508	11.6	551	91±1.25	114±3.7	510	49	36	15
	UK	440	11.0	454	92±1.3	107±3.6	422	60	31	8
	Total	2590	11.1	2767	84±0.71	109±1.76	2368	55	30	14

ESA=erythropoiesis-stimulating agent; Hb, hemoglobin; SC=subcutaneous; SEM, standard error of the mean.

a Modified from Locatelli et al. [107].

In the oncology setting, patients were initially dosed twice or thrice weekly primarily because of the short serum t1/2 of rhEPO and the common use of these regimens in the nephrology setting. However, dose frequencies were subsequently extended to QW to minimize needle-sticks and unnecessary visits to health care professionals. Less frequent administration of longer-lived ESAs (e.g., darbepoetin α) that coincide with QW, Q2W, and Q3W chemotherapy cycles have also been explored.

ESA therapy has transformed anemia management in both the renal and oncology settings. The major advantages of ESAs have been the treatment of anemia and avoidance or minimization of blood transfusions, as well as possible improvements in quality of life associated with higher Hb levels. However, there may be an Hb limit beyond which the risk-to-benefit ratio is no longer favorable for the patient. The frequency of dosing of the various ESAs is driven by a number of factors, including practice patterns and the duration of action of the ESA. While extending the dosing interval of ESA therapy is often possible, the major considerations from the clinical and pharmacoeconomic viewpoints are whether Hb levels can be effectively maintained within a target range, e.g., 11 to 12g/dL, and whether there is any dose penalty (i.e., need for dose increases greater than those calculated by simple multiplication of time intervals) for extending the dosing-frequency schedule. We have expanded our understanding of the ways in which molecules stimulate erythropoiesis, including the interaction of EPO with its receptor and the effect on in vitro and in vivo activity. Further refinement of this understanding to probe the trade-off between the receptor affinity of ESAs and the circulating t1/2 of the molecule indicated that there is a balance between these biologic properties that translates into the clinical arena. It is to be hoped that greater appreciation of this phenomenon will foster high-quality randomized controlled trials to determine the optimal use for each ESA.