review
Milestones in understanding platelet production: a historical overview David J. Kuter Hematology Division, Massachusetts General Hospital, Boston, MA, USA
Summary The discovery of thrombopoietin (TPO, also termed THPO) in 1994 was a major achievement in understanding the regulation of platelet production. In prior decades, physiological studies had demonstrated that platelets were produced from bone marrow megakaryocytes and that the megakaryocytes responded to thrombocytopenia by increasing their number, size and DNA ploidy. In 1958, it was proposed that a ‘thrombopoietin’ must exist that regulated this interaction between the circulating platelet mass and the bone marrow megakaryocytes. After over three decades of effort, TPO was finally purified by five independent laboratories. TPO stimulated megakaryocyte colony-forming cell growth and increased the number, size and ploidy of megakaryocytes. When the genes for TPO or TPO receptor were eliminated in mice, megakaryocytes grew and platelets were made, but at 15% of their normal number. A first generation of recombinant human (rh) TPO molecules [rhTPO and pegylated recombinant human megakaryocyte growth and development factor (PEG-rhMGDF)] rapidly entered clinical trials in 1995 and increased platelet counts in humans undergoing nonmyeloablative chemotherapy but not in those undergoing stem cell transplantation. Antibodies developed against PEGrhMGDF and development of these recombinant thrombopoietins ended. A second generation of TPO receptor agonists (romiplostim and eltrombopag) was then developed. Neither of these TPO receptor agonists demonstrated any significant untoward effects and both are now licensed in many countries for the treatment of immune thrombocytopenia. This review describes the significant experiments that have surrounded the discovery of TPO and its clinical development. Keywords: thrombopoietin, thrombopoietin receptor agonists, thrombocytopenia, haematopoietic growth factors, thrombocytopenia.
Correspondence: David J. Kuter, MD, D Phil, Hematology Division, Massachusetts General Hospital, Yawkey 7858, 55 Fruit Street, Boston, MA 02114, USA. E-mail: [email protected]
First published online 14 February 2014 doi: 10.1111/bjh.12781
Thrombopoietin (TPO, also termed THPO) is the key regulator of platelet production. Over the past two decades, the understanding of TPO function has vastly increased our knowledge of platelet production and its regulation. This has led to the development of thrombopoietic drugs, such as the TPO receptor agonists, romiplostim and eltrombopag, which increase the platelet count and serve all the functions of TPO. The use of these TPO receptor agonists has markedly improved the treatment of immune thrombocytopenia (ITP) and is being explored in a wide variety of other thrombocytopenic conditions. This review will explore the various key experiments and concepts that are the basis for our current understanding of the regulation of platelet production, particularly the concept, discovery and development of the thrombopoietic agents.
Bone marrow megakaryocytes make platelets Although these circulating blood cells had been described by Bizzozero in the 1880s (Bizzozero, 1882; Brewer, 2006), it was not until the seminal work by Wright in 1906 that these cells were demonstrated to arise from bone marrow megakaryocytes (Wright, 1902, 1906, 1910). These discoveries were based upon microscopic analysis of the blood and bone marrow in which platelets and megakaryocytes were found to share common tinctorial properties as well as fortuitous microscopic sections that showed platelets budding from megakaryocytes and entering the circulation (Fig 1). Wright went on to show that a reduction in platelets was associated with bleeding and provided crude semi-quantitative platelet counts using the microscope. Less well appreciated is that at the same time Wright was doing these studies at the Massachusetts General Hospital in Boston, other important observations were made by two of his colleagues at that institution, William Duke and George Minot. William Duke described subjects treated for thrombocytopenia by patient-to-patient whole blood transfusions using arterio-venous cannulation techniques. He showed that in three individuals with low platelet counts, the transfusion of whole blood from a healthy donor would transiently raise the platelet count and reduce bleeding (Duke, 1910). Although George Minot is best known for his discovery of vitamin B12, for which he received a Nobel Prize in 1934, ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review
Fig 1. Megakaryocyte protruding into bone marrow sinusoid and producing platelets. Camera lucida drawing by Wright (1910).
his earlier work focused on ITP, something he called ‘purpura hemorrhagica’ (Minot, 1916). As a young staff physician at Massachusetts General Hospital in 1919 he described the following concept of the pathophysiology of ITP: The low platelet count seemed to have been due to one or both of the following two factors: 1. Some reaction (presumably a specific poison) taking place in the body which destroyed platelets as fast as they were formed. 2. A localized aplasia of the platelet-forming elements of the marrow which might have been due to some toxic phenomenon.
Physiological observations demonstrate that platelet production is regulated From 1910 until approximately 1960, a number of investigators provided important concepts regarding how the platelet count was regulated. These can be summarized as the following five physiological principles: 1 The platelet count is stable in any individual unless affected by disease. Brecher and Cronkite developed early platelet counting techniques and used these to show that for over 18 months, the platelet count in healthy individuals was constant (Brecher & Cronkite, 1950; Brecher et al, 1953). Subsequent studies and clinical practice have validated this finding for periods as long as seven decades 2 The platelet count varies greatly (from 140 9 109/l to 450 9 109/l) between healthy individuals. Although the normal red cell number is very tightly regulated (432 9 1012/l–572 9 1012/l, males), there is a huge difference in platelet counts in healthy populations, as reported in early studies (Sloan, 1951) and current clinical normal ranges. 3 Platelet size is inversely related to the platelet number. Encouraged by the earlier work by Tocantins (1938), von ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Behrens (1972) examined the relationship between platelet size and platelet number in a large cohort of healthy individuals. In his thesis, he was clearly able to show that Anglo-Saxon subjects had a median normal platelet count of 220 9 109/l and median mean platelet volume (MPV) of 10 fl, but that Mediterranean subjects had a median normal platelet count of 89 9 109/l and median MPV of 17 fl. This concept was felt to be an example of ‘phylogenetic canalization’ in that it can be carried over to other species. For example, mice have a platelet count of 1690 (520) 9 109/l and a MPV of 47 (03) fl (Corash et al, 1990) whereas dogs have a platelet count of 286 (101) 9 109/l and a MPV of 106 (12) fl (Zvorc et al, 2010). Subsequent studies have confirmed that this inverse relationship between platelet count and MPV is nonlinear (Bessman et al, 1981). 4 The body defends the platelet mass not the platelet number. Pennington assessed the total number of platelets per animal (platelet mass) in mice that were splenectomized (and had a high platelet count) or with varying degrees of splenomegaly (and proportionally lower platelet counts) compared with healthy mice (Penington, 1967, 1968, 1970; Penington & Olsen, 1970). He showed that in all three groups of mice, the total body platelet mass was the same, albeit with different (and exchangeable) partition between the splenic and circulating platelet pools. Aster subsequently corroborated this concept in humans who had undergone splenectomy (Aster, 1966, 1967). 5 Megakaryocyte size and ploidy are inversely related to the platelet count. Using a variety of techniques to measure megakaryocyte size and ploidy in animals, several studies showed that megakaryocyte size and ploidy increased as the platelet count fell, and both fell when the platelet count was experimentally increased by transfusion (Harker & Finch, 1969; Harker, 1970; Penington & Olsen, 1970; Jackson et al, 1984). Later studies showed that there was an inverse, log-linear relationship between the platelet count and megakaryocyte DNA ploidy (Kuter & Rosenberg, 1990; Kuter, 1996).
The concept of thrombopoietin In 1958, Professor Endre Kelemen (Fig 2) proposed that a haematopoietic growth factor existed that regulated platelet production in the same way that erythropoietin (EPO) regulated the production of red cells (Kelemen et al, 1958). He suggested that a ‘thrombopoietin’ must exist, which regulated platelet production and was the missing link between the circulating platelet mass and the bone marrow megakaryocyte, as described in the earlier physiological observations (vide supra). Unfortunately, his initial concept of TPO has been misunderstood; rather than a factor appearing during thrombocytopenia, it was his hypothesis that TPO was the substance present in the plasma of thrombocythaemic patients 249
Review
Fig 2. Professor Endre Kelemen (1921–2000) is credited with proposing the name and functional attributes of ‘thrombopoietin’ in 1958. Reprinted from The Lancet, 354, (No author listed), Endre Kelemen, 1220, © 1999, with permission from Elsevier.
with myeloproliferative diseases that stimulated the increase in the platelet count. Plasma and plasma extracts obtained from two patients with myeloproliferative disease and elevated platelet counts were injected into mice and the ‘thrombopoietin’ contained therein stimulated an increase in the platelet count. It was not until 1959 with a single case report (Rak et al, 1959) and in 1963 (Kelemen et al, 1963) with a larger experimental base that his concept of TPO was modified to be the circulating factor in thrombocytopenic plasma that stimulated an increase in platelet production and hastened the recovery from thrombocytopenia.
Early efforts to purify thrombopoietin failed With the concept of TPO clearly embedded in the academic conscience, laboratories run by Kelemen, Odell, McDonald, Murphy and a number of others sought to purify this illusive haematopoietic growth factor (Kelemen et al, 1958, 1960, 1963; McDonald, 1973a,b, 1975, 1981, 1988; McDonald et al, 1974, 1975, 1989; McDonald & Nolan, 1979). Using urine or plasma from thrombocytopenic animals or conditioned medium from human embryonic kidney (HEK) cells, efforts were made to purify TPO using a wide variety of bioassays that had not been validated. Despite over 35 years of heroic efforts by these laboratoriess, no haematopoietic growth factor was purified using techniques that were comparable to those that had been successful for the purification of erythropoietin. Between 1958 and 1994, a number of other haematopoietic growth factors had been discovered and their characteristics defined. These included interleukin (IL) 3, granulocyte-macrophage colony stimulating factor (GM-CSF), c-kit ligand (KITLG), and IL11. In a wide variety of animal and cell culture models, each of these produced a modest increase in megakaryocyte growth or rise in the platelet count. However, in animals in which these haematopoietic growth factors had been ‘knocked out,’ none became thrombocytopenic. Indeed one of these, IL11, increased the platelet count in animals and in 250
humans receiving chemotherapy and was eventually licensed by the US Food and Drug Administration (FDA) for the treatment of chemotherapy-related thrombocytopenia (Tepler et al, 1996). Nonetheless, IL11 was not the true TPO; the platelet count was normal in IL11-deficient animals (Robb et al, 1998). Finally, a single report (Tayrien & Rosenberg, 1987) described the purification of a megakaryocyte-stimulatory factor (MSF) from conditioned medium from HEK cells that increased the synthesis of platelet factor-4 in a promegakaryoblast cell line. Unfortunately, soon after this publication, the authors explored the characteristics of MSF and realized that it had none of the properties that would make it the physiological regulator of platelet production.
Successful efforts to purify thrombopoietin In 1994, five laboratories [Amgen, Genentech, Zymogenetics, Kirin, Massachusetts Institute of Technology (MIT)] almost simultaneously reported the purification of TPO. Most of these laboratory efforts were unknown to each other and had proceeded along totally independent and different lines of investigation to identify and purify this haematopoietic growth factor. Some of these groups had pursued this goal for over a decade. In general, efforts to purify TPO were handicapped by several important factors. The first was to find an adequate source of TPO from which it could be purified. Efforts to purify TPO from thrombocytopenic animal urine had been a failure, unlike the successful efforts used for purifying EPO. In retrospect, this was clearly due to the size of the molecule (~94 kD) because neither TPO nor any of its active degradation products is renally filtered (Li et al, 2001). A related problem that was then unappreciated, is that normal TPO concentrations are 1000-fold lower [39 (range 7–99) pg/ml] (Makar et al, 2013) than those of erythropoietin (32–216 ng/ ml) (Cotes & Bangham, 1966). A final major problem was that there was no validated assay for TPO. Sources containing high concentrations of TPO were probably the easiest challenge to overcome. Animals treated with radiation contained increased amounts of a substance in their thrombocytopenic plasma that stimulated the growth of megakaryocyte colony-forming cells (Meg-CFC) (Solberg, 2013). At certain doses, busulfan produced only thrombocytopenia in mice, rats, rabbits and sheep and eventually allowed the collection of hundreds of litres of thrombocytopenic sheep plasma that contained a substance that increased megakaryocyte number, ploidy and size when added to cultures of bone marrow (Kuter & Rosenberg, 1995; Kuter et al, 1997). Despite the prior 30 years of investigation, no assay had been validated as a measurement of TPO activity. The Kirin and MIT groups both hit upon a novel assay in which rat bone marrow was depleted of identifiable megakaryocytes, cultured for 3 d with plasma fractions containing the putative TPO, and the extent of megakaryocyte growth (number, ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review production. In animals in which both genes for TPO (now termedThpo) or both genes for the TPO receptor (now termed Mpl) were completely ‘knocked out,’ the platelet count and bone marrow megakaryocytes dropped to 12–15% of normal (Fig 3), but the thrombocytopenic animals did not have any major bleeding (Gurney et al, 1994; de Sauvage et al, 1996). However, in animals in which only one of the two TPO receptor genes was genetically knocked out, the platelet count was normal. In contrast, knockout of only one of the two TPO genes produced animals with platelet counts 67% of normal. On one hand, these data clearly showed that megakaryocyte differentiation, growth and platelet production occurred at a basal level independent of TPO. On the other hand, these data showed that TPO was the only key regulator of platelet production and accounted for over 85% of platelet production in normal physiology. Analysis of the bone marrow progenitors of these knock out animals was also of great interest (Carver-Moore et al, 1996). In animals with homozygous knock out of genes for either TPO or the TPO receptor, Meg-CFC were decreased to ~5% or normal (Fig 4). But these animals also had a decrease in erythroid (BFU-E) and myeloid (CFU-GM) colonies to ~40% of normal but without any effect on the red blood cell or white blood cell numbers. These data suggest that TPO serves primarily as a late-acting differentiation factor for megakaryocytes but is also necessary for the growth and viability of cells of all lineages, including the pluripotential stem cell. Although circulating EPO levels are regulated primarily at a transcriptional level by a very efficient sensor of oxygen delivery to the kidney (with red blood cells lacking EPO receptors and thereby not facilitating EPO clearance), studies (Kuter et al, 1994) showed that TPO joined MCSF (Bartocci et al, 1987) and G-CSF (Corbacioglu et al, 2000) in employing a more primitive but equally effective form of regulation of blood cell production in normal physiology. There is no specific ‘sensor’ of the platelet count. Rather TPO is produced constantly in the healthy liver, enters the circulation, and is cleared by high affinity TPO receptors on platelets 1600
Platelet count, 109/l
size and ploidy) measured (Kuter et al, 1989, 1992, 1994; Kato et al, 1995). Although initially measured by a very laborious flow cytometry assay that measured the increase in number and ploidy of the megakayocytes, both groups eventually used a more rapid assay that measured the uptake of 3 H-serotonin into the cultures. This was based upon the finding that serotonin is preferentially incorporated into nascent dense granules in megakaryocytes (White, 1971; Schick & Weinstein, 1981; Bricker & Zuckerman, 1984). Assays that assessed the extent of stimulation of Meg-CFC growth were used by other laboratories to guide the purification process. The identification of the putative TPO receptor (c-mpl, now termed MPL) led to a third ‘assay’ for TPO. In 1991 a new haematopoietic growth factor receptor (c-mpl) was identified that possessed all the characteristics of the TPO receptor (and which is now known to be the TPO receptor) (Wendling & Tambourin, 1991). This led to large-scale efforts by many laboratories to use this putative receptor for affinity purification of the ‘c-mpl ligand’, i.e TPO, from thrombocytopenic plasma. Ultimately, the five groups purified TPO. The Kirin group purified ‘thrombopoietin’ from large volumes of thrombocytopenic plasma from irradiated rats guided by the increase in serotonin uptake in bone marrow cultures (Kato et al, 1995). The MIT group purified ‘megapoietin’ directly from thrombocytopenic busulfan-treated sheep plasma guided by the increase in serotonin uptake in rat bone marrow cultures (Kuter et al, 1994). The Genentech group mostly used affinity purification with c-mpl to isolate the ‘c-mpl ligand’ from the plasma of irradiated pigs (de Sauvage et al, 1994; Solberg, 2013). The Amgen group also used c-mpl affinity methods to purify ‘megakaryocyte growth and development factor’ (MGDF) from aplastic canine plasma (Bartley et al, 1994). The Zymogenetics group used a functional cloning strategy, which was initially independent of a TPO source or a TPO assay, to clone directly ‘thrombopoietin’ (Kaushansky et al, 1994; Lok & Foster, 1994; Lok et al, 1994). They took an IL3-dependent cell line that expressed the putative TPO receptor c-mpl, and subjected it to mutagenesis with 2-ethylmethanesulfonate and looked for clones that demonstrated autocrine growth. Cell clones that did not express TPO would die upon withdrawal of IL3 but clones that expressed TPO would not. They thereby identified DNA sequences containing ‘c-mpl ligand’ and then demonstrated that the expressed protein stimulated Meg-CFC, increased megakaryocyte ploidy and increased the platelet count in animals (Kaushansky et al, 1994).
1400 1200 1000 800 600 400 200 0 +/+
Proof that thrombopoietin is the key regulator of platelet production With the cloning of TPO, a number of seminal experiments rapidly demonstrated that the c-mpl ligand was the long sought TPO and that it was the prime regulator of platelet ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
+/–
TPO
–/–
+/+
+/–
–/–
TPO Receptor
Fig 3. Platelet counts (+SD) in mice in which the TPO (de Sauvage et al, 1996) or TPO receptor (Gurney et al, 1994) genes have been eliminated in a homozygous ( / ) or heterozygous (+/ ) fashion, compared with healthy (+/+) mice.
251
Review
Number of cells (percentage of normal)
WT
TPO –/–
TPO-R –/–
100 80 60 40 20 0
Nucleated cells
Meg-CFC
GM-CFC
E-BFU
Fig 4. Effect of homozygous ( / ) elimination of TPO or TPO receptor (TPO-R) genes in mice on the number of nucleated bone marrow cells and the number of megakaryocyte colony-forming cells (Meg-CFC), granulocyte-macrophage colony-forming cells (GMCFC), and erythroid burst-forming units (BFU-E). Data is expressed (+standard deviation) as the percentage of such cells in healthy wildtype (WT) murine bone marrow (Carver-Moore et al, 1996).
(and maybe also megakaryocytes) (Kuter et al, 1994; Fielder et al, 1996; Kuter, 1996; Li et al, 1996; Broudy et al, 1997). TPO levels are inversely proportional to the rate of platelet production (Emmons et al, 1996). There is no factor or clinical setting, other than liver disease, that alters TPO production in the liver (Peck-Radosavljevic et al, 1997, 2000; Yang et al, 1999). This simple system explains why the body regulates the total body platelet mass and not the platelet count. It provides an explanation for the thrombocytopenia of liver disease and why TPO levels are normal in patients with ITP, but elevated in those with bone marrow failure.
Early results with recombinant thrombopoietic agents: successes and failures From 1994 until approximately 2000, a large number of studies were carried out with two recombinant thrombopoietins
(Fig 5), recombinant human thrombopoietin (rhTPO) and pegylated, recombinant human megakaryocyte growth and development factor (PEG-rhMGDF), a molecule composed of the first 153 amino acids of the native sequence of TPO coupled to a polyethylene glycol residue (Kuter & Begley, 2002). These showed successes in multiple areas. In non-myeloablative chemotherapy experiments, administration of rhTPO or rhPEG-MGDF raised the nadir platelet count, decreased the duration of chemotherapy-induced thrombocytopenia and allowed for chemotherapy to be given on time (Basser et al, 1996, 1997; Fanucchi et al, 1997; Vadhan-Raj et al, 2000). In some studies, the need for platelet transfusions was also decreased (Vadhan-Raj et al, 2000). However, despite a wide variety of pre-chemotherapy and post-chemotherapy dosing schedules in myeloablative (e.g, stem cell transplantation, acute myeloid leukaemia induction chemotherapy) chemotherapy settings, neither agent had any effect upon platelet count nadir, duration of thrombocytopenia or time to platelet count recovery. Many of these studies have never been fully reported but, in those that have been published, the only significant effect seen was a clinically irrelevant rebound thrombocytopenia occurring days after recovery from the platelet nadir (Schiffer et al, 2000; Kuter & Begley, 2002). In other small studies in myelodysplastic syndrome (MDS) (Komatsu et al, 2000) and ITP (Rice et al, 2001; Nomura et al, 2002; Kizaki et al, 2003) patients, these recombinant thrombopoietins increased the platelet count and could decrease bleeding. In some MDS patients, there was a multilineage effect of treatment with PEG-rhMGDF (Komatsu et al, 2000). Finally, these recombinant thrombopoietins were shown to be effective in increasing platelet apheresis yields in healthy volunteers and could also be used as a way to harvest additional platelets for patients undergoing chemotherapy who had problems with platelet alloimmunization (Goodnough et al, 2001; Kuter et al, 2001; Vadhan-Raj et al, 2002). In both of these settings, the platelets so harvested were
Fig 5. Structures of the recombinant thrombopoietins, rhTPO and PEG-rhMGDF. Violet arrows indicate areas of the a helix, while orange and blue ‘feathers’ denote glycosylation sites.
252
ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review functional and provided an appropriate increase in platelet count and reduced bleeding when transfused into thrombocytopenic patients. Unfortunately, when PEG-rhMGDF was given monthly for up to three doses to healthy volunteers, thrombocytopenia occurred in 13 of 535. These 13 healthy volunteers developed autoantibodies against PEG-rhMGDF, which then crossreacted with and neutralized endogenous TPO (Li et al, 2001). Two of these subjects required immunosuppressive therapies but all 13 recovered with no apparent sequelae. Nonetheless, development of PEG-rhMGDF and eventually that of rhTPO (despite the lack of any antibody development or other problems with it) was stopped.
The development of TPO receptor agonists Even before the development of autoantibodies against PEGrhMGDF, efforts had been underway to identify other TPO compounds that stimulated platelet production and might have improved pharmacological properties. With the TPO receptor now clearly demonstrated, reporter cell lines containing the TPO receptor could be developed and used to screen for peptides and small molecules that would stimulate these TPO-dependent cell lines (Kuter, 2009, 2013). In 1997 a 14 amino acid peptide was identified using a hierarchal screening technique looking for high affinity peptides that bound to the TPO receptor and activated it (Cwirla et al, 1997). A number of peptides were identified that were then significantly modified to provide a 14 amino acid TPO peptide. When dimerized by an alanine bridge this TPO peptide was a very potent TPO agonist and had half maximal stimulatory effect on TPO-dependent cell lines that was the same as TPO. However, like all peptides, this TPO peptide had a very short circulatory half-life and needed to be stabilized to increase its half-life to allow it to be an effective pharmacological agent. Efforts to stabilize this peptide proceeded along two pathways. The first was to insert this peptide into the complementarity-determining region (CDR) of a human IgG Fab (called Fab 59) (Frederickson et al, 2006). This resulted in a very potent thrombopoietic agent that stimulated platelet production in mice but this molecule never entered clinical trials. The second approach was to insert a pair of these TPO peptides into each arm of an IgG type 1 heavy chain molecule using glycine linkers of defined length (Fig 6) (Molineux, 2011). This ‘peptibody’, called romiplostim (AMP-2, AMG-531, Nplate), bound to the TPO receptor (Fig 7) with about 25% of the binding affinity of TPO but was a very potent stimulator of megakaryocyte growth in vitro and increased platelet production in vivo (Broudy & Lin, 2004). Romiplostim is administered subcutaneously once a week and is now FDA approved for the treatment of ITP. A second approach to identifying new TPO agonists was to screen for small non-peptide molecules that could activate the TPO receptor (Erickson-Miller et al, 2005). This ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Fig 6. Structure of romiplostim. This ‘peptibody’ is composed of a dimerized IgG1 Fc domain (blue) into which is inserted via polyglycine linkers (green) four identical 14-amino acid peptides (violet) with the sequence IEGPTLRQWLAARA that bind the TPO-R (Molineux, 2011). The peptide has no sequence homology with TPO. This 60 kD protein is made in E. coli and is recycled by endothelial FcRn (neonatal Fc receptor) receptors to give a ~100-h half-life.
screening technique identified a surprisingly large number of small molecules, some of which were then modified to enhance their biological and pharmacological properties. Of these molecules, eltrombopag (SB497115, Promacta, Revolade) has been successfully developed (Fig 8) and has an interesting biology, which differs from that of TPO and romiplostim. Eltrombopag does not bind the TPO receptor at the same site as TPO (Fig 7); instead, it binds to a transmembrane portion of the TPO receptor at residue 499 and is active in species (only humans and chimps) where a histidine is present at this position. This species restriction presented a significant challenge to initial clinical development for quite a number of years given the inability to test eltrombopag in animal models. Eltrombopag is currently FDA approved for the treatment of ITP and hepatitis C-related thrombocytopenia in patients undergoing antiviral treatment. It is a pill administered orally once a day.
Clinical outcomes with TPO receptor agonists Immune thrombocytopenia was the initial focus of clinical development of the TPO receptor agonists. ITP was chosen because of the perception that the underlying bone marrow was normal and unaffected by disease or prior chemotherapy that might complicate studies in patients with MDS or patients receiving chemotherapy. This decision was later further legitimized by a clearer understanding of the pathophysiology of 253
Review
Fig 7. Activation of the TPO receptor by TPO or TPO receptor agonists. TPO and romiplostim bind to the distal cytokine receptor homology domain (CRH-2) of the preformed, inactive TPO receptor dimer. Eltrombopag binds to the transmembrane region of the TPO receptor. Multiple signal transduction pathways are thereby activated.
Treatment with TPO receptor agonists of the following thrombocytopenic disorders has been studied, but none has received any regulatory approval for their use (Kuter, 2013):
Fig 8. Structure of eltrombopag. This 442 Da molecule binds to the TPO-R at a site different from TPO and is an efficient stimulator of platelet production in healthy subjects.
ITP, which demonstrated that immune processes impaired platelet production as well as destroyed circulating platelets (Gernsheimer et al, 1989; McMillan et al, 2004). Both eltrombopag and romiplostim increased the platelet count in over 80% of ITP patients with minimal adverse effects (Bussel et al, 2006; Kuter et al, 2010; Cheng et al, 2011). ITP patients so treated had less bleeding, needed fewer rescue treatments, could reduce or eliminate their need for corticosteroids, and had an improved health-related quality of life (Cheng et al, 2011; Kuter et al, 2012). Long-term use for at least 3–5 years was effective and without the appearance of new adverse effects (Kuter et al, 2013; Saleh et al, 2013). Potential safety concerns for bone marrow fibrosis, thrombosis and tachyphylaxis affected a very small number of patients. Patients with hepatitis C infection are often thrombocytopenic and upon anti-viral treatment usually have a further 50% decline in platelet count. Eltrombopag has been shown to be effective in mitigating this decline in platelet count, thereby allowing patients be treated more effectively with antiviral agents (McHutchison et al, 2007). Eltrombopag is now FDA approved for this purpose. 254
1 Paediatric ITP patients respond as well as adult ITP patients to eltrombopag and romiplostim with the same or a lesser side effect profile (Bussel et al, 2011). 2 Myosin heavy chain 9-related thrombocytopenia patients have an inherited disorder characterized by moderate thrombocytopenia and modest bleeding risk. Daily treatment with eltrombopag gave a twofold or greater increase in platelet count in 11 of 12 patients and reduced bleeding (Pecci et al, 2010). 3 Aplastic anaemia patients have a low platelet count and highly elevated level of endogenous TPO. Surprisingly, after treatment with eltrombopag daily, 11 of 25 subjects had a haematological response with six subjects ultimately having a trilineage increase as well as increased bone marrow cellularity (Olnes et al, 2012). With already highly elevated TPO levels, it is unclear whether eltrombopag is producing this beneficial effect via the TPO receptor of some other ‘off-target’ effect. 4 MDS patients have thrombocytopenia and this often occurs before other cytopenias (Kantarjian et al, 2007; Bryan et al, 2010; Al Ameri et al, 2011). Treatment of 250 patients with International Prognostic Scoring System low/intermediate-1 MDS with romiplostim (167 patients) produced an increase in platelet count, reduced need for transfusion and reduced bleeding compared with treatment with placebo (83 patients). However, a rise in blast count to >10% occurred in 15% of those on romiplostim versus 36% of those on placebo (Giagounidis et al, 2011; Kantarjian et al, 2012). 5 Liver disease patients often have thrombocytopenia and are at increased risk of bleeding with procedures. Treatment of such patients with platelet counts
Milestones in understanding platelet production: a historical overview David J. Kuter Hematology Division, Massachusetts General Hospital, Boston, MA, USA
Summary The discovery of thrombopoietin (TPO, also termed THPO) in 1994 was a major achievement in understanding the regulation of platelet production. In prior decades, physiological studies had demonstrated that platelets were produced from bone marrow megakaryocytes and that the megakaryocytes responded to thrombocytopenia by increasing their number, size and DNA ploidy. In 1958, it was proposed that a ‘thrombopoietin’ must exist that regulated this interaction between the circulating platelet mass and the bone marrow megakaryocytes. After over three decades of effort, TPO was finally purified by five independent laboratories. TPO stimulated megakaryocyte colony-forming cell growth and increased the number, size and ploidy of megakaryocytes. When the genes for TPO or TPO receptor were eliminated in mice, megakaryocytes grew and platelets were made, but at 15% of their normal number. A first generation of recombinant human (rh) TPO molecules [rhTPO and pegylated recombinant human megakaryocyte growth and development factor (PEG-rhMGDF)] rapidly entered clinical trials in 1995 and increased platelet counts in humans undergoing nonmyeloablative chemotherapy but not in those undergoing stem cell transplantation. Antibodies developed against PEGrhMGDF and development of these recombinant thrombopoietins ended. A second generation of TPO receptor agonists (romiplostim and eltrombopag) was then developed. Neither of these TPO receptor agonists demonstrated any significant untoward effects and both are now licensed in many countries for the treatment of immune thrombocytopenia. This review describes the significant experiments that have surrounded the discovery of TPO and its clinical development. Keywords: thrombopoietin, thrombopoietin receptor agonists, thrombocytopenia, haematopoietic growth factors, thrombocytopenia.
Correspondence: David J. Kuter, MD, D Phil, Hematology Division, Massachusetts General Hospital, Yawkey 7858, 55 Fruit Street, Boston, MA 02114, USA. E-mail: [email protected]
First published online 14 February 2014 doi: 10.1111/bjh.12781
Thrombopoietin (TPO, also termed THPO) is the key regulator of platelet production. Over the past two decades, the understanding of TPO function has vastly increased our knowledge of platelet production and its regulation. This has led to the development of thrombopoietic drugs, such as the TPO receptor agonists, romiplostim and eltrombopag, which increase the platelet count and serve all the functions of TPO. The use of these TPO receptor agonists has markedly improved the treatment of immune thrombocytopenia (ITP) and is being explored in a wide variety of other thrombocytopenic conditions. This review will explore the various key experiments and concepts that are the basis for our current understanding of the regulation of platelet production, particularly the concept, discovery and development of the thrombopoietic agents.
Bone marrow megakaryocytes make platelets Although these circulating blood cells had been described by Bizzozero in the 1880s (Bizzozero, 1882; Brewer, 2006), it was not until the seminal work by Wright in 1906 that these cells were demonstrated to arise from bone marrow megakaryocytes (Wright, 1902, 1906, 1910). These discoveries were based upon microscopic analysis of the blood and bone marrow in which platelets and megakaryocytes were found to share common tinctorial properties as well as fortuitous microscopic sections that showed platelets budding from megakaryocytes and entering the circulation (Fig 1). Wright went on to show that a reduction in platelets was associated with bleeding and provided crude semi-quantitative platelet counts using the microscope. Less well appreciated is that at the same time Wright was doing these studies at the Massachusetts General Hospital in Boston, other important observations were made by two of his colleagues at that institution, William Duke and George Minot. William Duke described subjects treated for thrombocytopenia by patient-to-patient whole blood transfusions using arterio-venous cannulation techniques. He showed that in three individuals with low platelet counts, the transfusion of whole blood from a healthy donor would transiently raise the platelet count and reduce bleeding (Duke, 1910). Although George Minot is best known for his discovery of vitamin B12, for which he received a Nobel Prize in 1934, ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review
Fig 1. Megakaryocyte protruding into bone marrow sinusoid and producing platelets. Camera lucida drawing by Wright (1910).
his earlier work focused on ITP, something he called ‘purpura hemorrhagica’ (Minot, 1916). As a young staff physician at Massachusetts General Hospital in 1919 he described the following concept of the pathophysiology of ITP: The low platelet count seemed to have been due to one or both of the following two factors: 1. Some reaction (presumably a specific poison) taking place in the body which destroyed platelets as fast as they were formed. 2. A localized aplasia of the platelet-forming elements of the marrow which might have been due to some toxic phenomenon.
Physiological observations demonstrate that platelet production is regulated From 1910 until approximately 1960, a number of investigators provided important concepts regarding how the platelet count was regulated. These can be summarized as the following five physiological principles: 1 The platelet count is stable in any individual unless affected by disease. Brecher and Cronkite developed early platelet counting techniques and used these to show that for over 18 months, the platelet count in healthy individuals was constant (Brecher & Cronkite, 1950; Brecher et al, 1953). Subsequent studies and clinical practice have validated this finding for periods as long as seven decades 2 The platelet count varies greatly (from 140 9 109/l to 450 9 109/l) between healthy individuals. Although the normal red cell number is very tightly regulated (432 9 1012/l–572 9 1012/l, males), there is a huge difference in platelet counts in healthy populations, as reported in early studies (Sloan, 1951) and current clinical normal ranges. 3 Platelet size is inversely related to the platelet number. Encouraged by the earlier work by Tocantins (1938), von ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Behrens (1972) examined the relationship between platelet size and platelet number in a large cohort of healthy individuals. In his thesis, he was clearly able to show that Anglo-Saxon subjects had a median normal platelet count of 220 9 109/l and median mean platelet volume (MPV) of 10 fl, but that Mediterranean subjects had a median normal platelet count of 89 9 109/l and median MPV of 17 fl. This concept was felt to be an example of ‘phylogenetic canalization’ in that it can be carried over to other species. For example, mice have a platelet count of 1690 (520) 9 109/l and a MPV of 47 (03) fl (Corash et al, 1990) whereas dogs have a platelet count of 286 (101) 9 109/l and a MPV of 106 (12) fl (Zvorc et al, 2010). Subsequent studies have confirmed that this inverse relationship between platelet count and MPV is nonlinear (Bessman et al, 1981). 4 The body defends the platelet mass not the platelet number. Pennington assessed the total number of platelets per animal (platelet mass) in mice that were splenectomized (and had a high platelet count) or with varying degrees of splenomegaly (and proportionally lower platelet counts) compared with healthy mice (Penington, 1967, 1968, 1970; Penington & Olsen, 1970). He showed that in all three groups of mice, the total body platelet mass was the same, albeit with different (and exchangeable) partition between the splenic and circulating platelet pools. Aster subsequently corroborated this concept in humans who had undergone splenectomy (Aster, 1966, 1967). 5 Megakaryocyte size and ploidy are inversely related to the platelet count. Using a variety of techniques to measure megakaryocyte size and ploidy in animals, several studies showed that megakaryocyte size and ploidy increased as the platelet count fell, and both fell when the platelet count was experimentally increased by transfusion (Harker & Finch, 1969; Harker, 1970; Penington & Olsen, 1970; Jackson et al, 1984). Later studies showed that there was an inverse, log-linear relationship between the platelet count and megakaryocyte DNA ploidy (Kuter & Rosenberg, 1990; Kuter, 1996).
The concept of thrombopoietin In 1958, Professor Endre Kelemen (Fig 2) proposed that a haematopoietic growth factor existed that regulated platelet production in the same way that erythropoietin (EPO) regulated the production of red cells (Kelemen et al, 1958). He suggested that a ‘thrombopoietin’ must exist, which regulated platelet production and was the missing link between the circulating platelet mass and the bone marrow megakaryocyte, as described in the earlier physiological observations (vide supra). Unfortunately, his initial concept of TPO has been misunderstood; rather than a factor appearing during thrombocytopenia, it was his hypothesis that TPO was the substance present in the plasma of thrombocythaemic patients 249
Review
Fig 2. Professor Endre Kelemen (1921–2000) is credited with proposing the name and functional attributes of ‘thrombopoietin’ in 1958. Reprinted from The Lancet, 354, (No author listed), Endre Kelemen, 1220, © 1999, with permission from Elsevier.
with myeloproliferative diseases that stimulated the increase in the platelet count. Plasma and plasma extracts obtained from two patients with myeloproliferative disease and elevated platelet counts were injected into mice and the ‘thrombopoietin’ contained therein stimulated an increase in the platelet count. It was not until 1959 with a single case report (Rak et al, 1959) and in 1963 (Kelemen et al, 1963) with a larger experimental base that his concept of TPO was modified to be the circulating factor in thrombocytopenic plasma that stimulated an increase in platelet production and hastened the recovery from thrombocytopenia.
Early efforts to purify thrombopoietin failed With the concept of TPO clearly embedded in the academic conscience, laboratories run by Kelemen, Odell, McDonald, Murphy and a number of others sought to purify this illusive haematopoietic growth factor (Kelemen et al, 1958, 1960, 1963; McDonald, 1973a,b, 1975, 1981, 1988; McDonald et al, 1974, 1975, 1989; McDonald & Nolan, 1979). Using urine or plasma from thrombocytopenic animals or conditioned medium from human embryonic kidney (HEK) cells, efforts were made to purify TPO using a wide variety of bioassays that had not been validated. Despite over 35 years of heroic efforts by these laboratoriess, no haematopoietic growth factor was purified using techniques that were comparable to those that had been successful for the purification of erythropoietin. Between 1958 and 1994, a number of other haematopoietic growth factors had been discovered and their characteristics defined. These included interleukin (IL) 3, granulocyte-macrophage colony stimulating factor (GM-CSF), c-kit ligand (KITLG), and IL11. In a wide variety of animal and cell culture models, each of these produced a modest increase in megakaryocyte growth or rise in the platelet count. However, in animals in which these haematopoietic growth factors had been ‘knocked out,’ none became thrombocytopenic. Indeed one of these, IL11, increased the platelet count in animals and in 250
humans receiving chemotherapy and was eventually licensed by the US Food and Drug Administration (FDA) for the treatment of chemotherapy-related thrombocytopenia (Tepler et al, 1996). Nonetheless, IL11 was not the true TPO; the platelet count was normal in IL11-deficient animals (Robb et al, 1998). Finally, a single report (Tayrien & Rosenberg, 1987) described the purification of a megakaryocyte-stimulatory factor (MSF) from conditioned medium from HEK cells that increased the synthesis of platelet factor-4 in a promegakaryoblast cell line. Unfortunately, soon after this publication, the authors explored the characteristics of MSF and realized that it had none of the properties that would make it the physiological regulator of platelet production.
Successful efforts to purify thrombopoietin In 1994, five laboratories [Amgen, Genentech, Zymogenetics, Kirin, Massachusetts Institute of Technology (MIT)] almost simultaneously reported the purification of TPO. Most of these laboratory efforts were unknown to each other and had proceeded along totally independent and different lines of investigation to identify and purify this haematopoietic growth factor. Some of these groups had pursued this goal for over a decade. In general, efforts to purify TPO were handicapped by several important factors. The first was to find an adequate source of TPO from which it could be purified. Efforts to purify TPO from thrombocytopenic animal urine had been a failure, unlike the successful efforts used for purifying EPO. In retrospect, this was clearly due to the size of the molecule (~94 kD) because neither TPO nor any of its active degradation products is renally filtered (Li et al, 2001). A related problem that was then unappreciated, is that normal TPO concentrations are 1000-fold lower [39 (range 7–99) pg/ml] (Makar et al, 2013) than those of erythropoietin (32–216 ng/ ml) (Cotes & Bangham, 1966). A final major problem was that there was no validated assay for TPO. Sources containing high concentrations of TPO were probably the easiest challenge to overcome. Animals treated with radiation contained increased amounts of a substance in their thrombocytopenic plasma that stimulated the growth of megakaryocyte colony-forming cells (Meg-CFC) (Solberg, 2013). At certain doses, busulfan produced only thrombocytopenia in mice, rats, rabbits and sheep and eventually allowed the collection of hundreds of litres of thrombocytopenic sheep plasma that contained a substance that increased megakaryocyte number, ploidy and size when added to cultures of bone marrow (Kuter & Rosenberg, 1995; Kuter et al, 1997). Despite the prior 30 years of investigation, no assay had been validated as a measurement of TPO activity. The Kirin and MIT groups both hit upon a novel assay in which rat bone marrow was depleted of identifiable megakaryocytes, cultured for 3 d with plasma fractions containing the putative TPO, and the extent of megakaryocyte growth (number, ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review production. In animals in which both genes for TPO (now termedThpo) or both genes for the TPO receptor (now termed Mpl) were completely ‘knocked out,’ the platelet count and bone marrow megakaryocytes dropped to 12–15% of normal (Fig 3), but the thrombocytopenic animals did not have any major bleeding (Gurney et al, 1994; de Sauvage et al, 1996). However, in animals in which only one of the two TPO receptor genes was genetically knocked out, the platelet count was normal. In contrast, knockout of only one of the two TPO genes produced animals with platelet counts 67% of normal. On one hand, these data clearly showed that megakaryocyte differentiation, growth and platelet production occurred at a basal level independent of TPO. On the other hand, these data showed that TPO was the only key regulator of platelet production and accounted for over 85% of platelet production in normal physiology. Analysis of the bone marrow progenitors of these knock out animals was also of great interest (Carver-Moore et al, 1996). In animals with homozygous knock out of genes for either TPO or the TPO receptor, Meg-CFC were decreased to ~5% or normal (Fig 4). But these animals also had a decrease in erythroid (BFU-E) and myeloid (CFU-GM) colonies to ~40% of normal but without any effect on the red blood cell or white blood cell numbers. These data suggest that TPO serves primarily as a late-acting differentiation factor for megakaryocytes but is also necessary for the growth and viability of cells of all lineages, including the pluripotential stem cell. Although circulating EPO levels are regulated primarily at a transcriptional level by a very efficient sensor of oxygen delivery to the kidney (with red blood cells lacking EPO receptors and thereby not facilitating EPO clearance), studies (Kuter et al, 1994) showed that TPO joined MCSF (Bartocci et al, 1987) and G-CSF (Corbacioglu et al, 2000) in employing a more primitive but equally effective form of regulation of blood cell production in normal physiology. There is no specific ‘sensor’ of the platelet count. Rather TPO is produced constantly in the healthy liver, enters the circulation, and is cleared by high affinity TPO receptors on platelets 1600
Platelet count, 109/l
size and ploidy) measured (Kuter et al, 1989, 1992, 1994; Kato et al, 1995). Although initially measured by a very laborious flow cytometry assay that measured the increase in number and ploidy of the megakayocytes, both groups eventually used a more rapid assay that measured the uptake of 3 H-serotonin into the cultures. This was based upon the finding that serotonin is preferentially incorporated into nascent dense granules in megakaryocytes (White, 1971; Schick & Weinstein, 1981; Bricker & Zuckerman, 1984). Assays that assessed the extent of stimulation of Meg-CFC growth were used by other laboratories to guide the purification process. The identification of the putative TPO receptor (c-mpl, now termed MPL) led to a third ‘assay’ for TPO. In 1991 a new haematopoietic growth factor receptor (c-mpl) was identified that possessed all the characteristics of the TPO receptor (and which is now known to be the TPO receptor) (Wendling & Tambourin, 1991). This led to large-scale efforts by many laboratories to use this putative receptor for affinity purification of the ‘c-mpl ligand’, i.e TPO, from thrombocytopenic plasma. Ultimately, the five groups purified TPO. The Kirin group purified ‘thrombopoietin’ from large volumes of thrombocytopenic plasma from irradiated rats guided by the increase in serotonin uptake in bone marrow cultures (Kato et al, 1995). The MIT group purified ‘megapoietin’ directly from thrombocytopenic busulfan-treated sheep plasma guided by the increase in serotonin uptake in rat bone marrow cultures (Kuter et al, 1994). The Genentech group mostly used affinity purification with c-mpl to isolate the ‘c-mpl ligand’ from the plasma of irradiated pigs (de Sauvage et al, 1994; Solberg, 2013). The Amgen group also used c-mpl affinity methods to purify ‘megakaryocyte growth and development factor’ (MGDF) from aplastic canine plasma (Bartley et al, 1994). The Zymogenetics group used a functional cloning strategy, which was initially independent of a TPO source or a TPO assay, to clone directly ‘thrombopoietin’ (Kaushansky et al, 1994; Lok & Foster, 1994; Lok et al, 1994). They took an IL3-dependent cell line that expressed the putative TPO receptor c-mpl, and subjected it to mutagenesis with 2-ethylmethanesulfonate and looked for clones that demonstrated autocrine growth. Cell clones that did not express TPO would die upon withdrawal of IL3 but clones that expressed TPO would not. They thereby identified DNA sequences containing ‘c-mpl ligand’ and then demonstrated that the expressed protein stimulated Meg-CFC, increased megakaryocyte ploidy and increased the platelet count in animals (Kaushansky et al, 1994).
1400 1200 1000 800 600 400 200 0 +/+
Proof that thrombopoietin is the key regulator of platelet production With the cloning of TPO, a number of seminal experiments rapidly demonstrated that the c-mpl ligand was the long sought TPO and that it was the prime regulator of platelet ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
+/–
TPO
–/–
+/+
+/–
–/–
TPO Receptor
Fig 3. Platelet counts (+SD) in mice in which the TPO (de Sauvage et al, 1996) or TPO receptor (Gurney et al, 1994) genes have been eliminated in a homozygous ( / ) or heterozygous (+/ ) fashion, compared with healthy (+/+) mice.
251
Review
Number of cells (percentage of normal)
WT
TPO –/–
TPO-R –/–
100 80 60 40 20 0
Nucleated cells
Meg-CFC
GM-CFC
E-BFU
Fig 4. Effect of homozygous ( / ) elimination of TPO or TPO receptor (TPO-R) genes in mice on the number of nucleated bone marrow cells and the number of megakaryocyte colony-forming cells (Meg-CFC), granulocyte-macrophage colony-forming cells (GMCFC), and erythroid burst-forming units (BFU-E). Data is expressed (+standard deviation) as the percentage of such cells in healthy wildtype (WT) murine bone marrow (Carver-Moore et al, 1996).
(and maybe also megakaryocytes) (Kuter et al, 1994; Fielder et al, 1996; Kuter, 1996; Li et al, 1996; Broudy et al, 1997). TPO levels are inversely proportional to the rate of platelet production (Emmons et al, 1996). There is no factor or clinical setting, other than liver disease, that alters TPO production in the liver (Peck-Radosavljevic et al, 1997, 2000; Yang et al, 1999). This simple system explains why the body regulates the total body platelet mass and not the platelet count. It provides an explanation for the thrombocytopenia of liver disease and why TPO levels are normal in patients with ITP, but elevated in those with bone marrow failure.
Early results with recombinant thrombopoietic agents: successes and failures From 1994 until approximately 2000, a large number of studies were carried out with two recombinant thrombopoietins
(Fig 5), recombinant human thrombopoietin (rhTPO) and pegylated, recombinant human megakaryocyte growth and development factor (PEG-rhMGDF), a molecule composed of the first 153 amino acids of the native sequence of TPO coupled to a polyethylene glycol residue (Kuter & Begley, 2002). These showed successes in multiple areas. In non-myeloablative chemotherapy experiments, administration of rhTPO or rhPEG-MGDF raised the nadir platelet count, decreased the duration of chemotherapy-induced thrombocytopenia and allowed for chemotherapy to be given on time (Basser et al, 1996, 1997; Fanucchi et al, 1997; Vadhan-Raj et al, 2000). In some studies, the need for platelet transfusions was also decreased (Vadhan-Raj et al, 2000). However, despite a wide variety of pre-chemotherapy and post-chemotherapy dosing schedules in myeloablative (e.g, stem cell transplantation, acute myeloid leukaemia induction chemotherapy) chemotherapy settings, neither agent had any effect upon platelet count nadir, duration of thrombocytopenia or time to platelet count recovery. Many of these studies have never been fully reported but, in those that have been published, the only significant effect seen was a clinically irrelevant rebound thrombocytopenia occurring days after recovery from the platelet nadir (Schiffer et al, 2000; Kuter & Begley, 2002). In other small studies in myelodysplastic syndrome (MDS) (Komatsu et al, 2000) and ITP (Rice et al, 2001; Nomura et al, 2002; Kizaki et al, 2003) patients, these recombinant thrombopoietins increased the platelet count and could decrease bleeding. In some MDS patients, there was a multilineage effect of treatment with PEG-rhMGDF (Komatsu et al, 2000). Finally, these recombinant thrombopoietins were shown to be effective in increasing platelet apheresis yields in healthy volunteers and could also be used as a way to harvest additional platelets for patients undergoing chemotherapy who had problems with platelet alloimmunization (Goodnough et al, 2001; Kuter et al, 2001; Vadhan-Raj et al, 2002). In both of these settings, the platelets so harvested were
Fig 5. Structures of the recombinant thrombopoietins, rhTPO and PEG-rhMGDF. Violet arrows indicate areas of the a helix, while orange and blue ‘feathers’ denote glycosylation sites.
252
ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Review functional and provided an appropriate increase in platelet count and reduced bleeding when transfused into thrombocytopenic patients. Unfortunately, when PEG-rhMGDF was given monthly for up to three doses to healthy volunteers, thrombocytopenia occurred in 13 of 535. These 13 healthy volunteers developed autoantibodies against PEG-rhMGDF, which then crossreacted with and neutralized endogenous TPO (Li et al, 2001). Two of these subjects required immunosuppressive therapies but all 13 recovered with no apparent sequelae. Nonetheless, development of PEG-rhMGDF and eventually that of rhTPO (despite the lack of any antibody development or other problems with it) was stopped.
The development of TPO receptor agonists Even before the development of autoantibodies against PEGrhMGDF, efforts had been underway to identify other TPO compounds that stimulated platelet production and might have improved pharmacological properties. With the TPO receptor now clearly demonstrated, reporter cell lines containing the TPO receptor could be developed and used to screen for peptides and small molecules that would stimulate these TPO-dependent cell lines (Kuter, 2009, 2013). In 1997 a 14 amino acid peptide was identified using a hierarchal screening technique looking for high affinity peptides that bound to the TPO receptor and activated it (Cwirla et al, 1997). A number of peptides were identified that were then significantly modified to provide a 14 amino acid TPO peptide. When dimerized by an alanine bridge this TPO peptide was a very potent TPO agonist and had half maximal stimulatory effect on TPO-dependent cell lines that was the same as TPO. However, like all peptides, this TPO peptide had a very short circulatory half-life and needed to be stabilized to increase its half-life to allow it to be an effective pharmacological agent. Efforts to stabilize this peptide proceeded along two pathways. The first was to insert this peptide into the complementarity-determining region (CDR) of a human IgG Fab (called Fab 59) (Frederickson et al, 2006). This resulted in a very potent thrombopoietic agent that stimulated platelet production in mice but this molecule never entered clinical trials. The second approach was to insert a pair of these TPO peptides into each arm of an IgG type 1 heavy chain molecule using glycine linkers of defined length (Fig 6) (Molineux, 2011). This ‘peptibody’, called romiplostim (AMP-2, AMG-531, Nplate), bound to the TPO receptor (Fig 7) with about 25% of the binding affinity of TPO but was a very potent stimulator of megakaryocyte growth in vitro and increased platelet production in vivo (Broudy & Lin, 2004). Romiplostim is administered subcutaneously once a week and is now FDA approved for the treatment of ITP. A second approach to identifying new TPO agonists was to screen for small non-peptide molecules that could activate the TPO receptor (Erickson-Miller et al, 2005). This ª 2014 John Wiley & Sons Ltd British Journal of Haematology, 2014, 165, 248–258
Fig 6. Structure of romiplostim. This ‘peptibody’ is composed of a dimerized IgG1 Fc domain (blue) into which is inserted via polyglycine linkers (green) four identical 14-amino acid peptides (violet) with the sequence IEGPTLRQWLAARA that bind the TPO-R (Molineux, 2011). The peptide has no sequence homology with TPO. This 60 kD protein is made in E. coli and is recycled by endothelial FcRn (neonatal Fc receptor) receptors to give a ~100-h half-life.
screening technique identified a surprisingly large number of small molecules, some of which were then modified to enhance their biological and pharmacological properties. Of these molecules, eltrombopag (SB497115, Promacta, Revolade) has been successfully developed (Fig 8) and has an interesting biology, which differs from that of TPO and romiplostim. Eltrombopag does not bind the TPO receptor at the same site as TPO (Fig 7); instead, it binds to a transmembrane portion of the TPO receptor at residue 499 and is active in species (only humans and chimps) where a histidine is present at this position. This species restriction presented a significant challenge to initial clinical development for quite a number of years given the inability to test eltrombopag in animal models. Eltrombopag is currently FDA approved for the treatment of ITP and hepatitis C-related thrombocytopenia in patients undergoing antiviral treatment. It is a pill administered orally once a day.
Clinical outcomes with TPO receptor agonists Immune thrombocytopenia was the initial focus of clinical development of the TPO receptor agonists. ITP was chosen because of the perception that the underlying bone marrow was normal and unaffected by disease or prior chemotherapy that might complicate studies in patients with MDS or patients receiving chemotherapy. This decision was later further legitimized by a clearer understanding of the pathophysiology of 253
Review
Fig 7. Activation of the TPO receptor by TPO or TPO receptor agonists. TPO and romiplostim bind to the distal cytokine receptor homology domain (CRH-2) of the preformed, inactive TPO receptor dimer. Eltrombopag binds to the transmembrane region of the TPO receptor. Multiple signal transduction pathways are thereby activated.
Treatment with TPO receptor agonists of the following thrombocytopenic disorders has been studied, but none has received any regulatory approval for their use (Kuter, 2013):
Fig 8. Structure of eltrombopag. This 442 Da molecule binds to the TPO-R at a site different from TPO and is an efficient stimulator of platelet production in healthy subjects.
ITP, which demonstrated that immune processes impaired platelet production as well as destroyed circulating platelets (Gernsheimer et al, 1989; McMillan et al, 2004). Both eltrombopag and romiplostim increased the platelet count in over 80% of ITP patients with minimal adverse effects (Bussel et al, 2006; Kuter et al, 2010; Cheng et al, 2011). ITP patients so treated had less bleeding, needed fewer rescue treatments, could reduce or eliminate their need for corticosteroids, and had an improved health-related quality of life (Cheng et al, 2011; Kuter et al, 2012). Long-term use for at least 3–5 years was effective and without the appearance of new adverse effects (Kuter et al, 2013; Saleh et al, 2013). Potential safety concerns for bone marrow fibrosis, thrombosis and tachyphylaxis affected a very small number of patients. Patients with hepatitis C infection are often thrombocytopenic and upon anti-viral treatment usually have a further 50% decline in platelet count. Eltrombopag has been shown to be effective in mitigating this decline in platelet count, thereby allowing patients be treated more effectively with antiviral agents (McHutchison et al, 2007). Eltrombopag is now FDA approved for this purpose. 254
1 Paediatric ITP patients respond as well as adult ITP patients to eltrombopag and romiplostim with the same or a lesser side effect profile (Bussel et al, 2011). 2 Myosin heavy chain 9-related thrombocytopenia patients have an inherited disorder characterized by moderate thrombocytopenia and modest bleeding risk. Daily treatment with eltrombopag gave a twofold or greater increase in platelet count in 11 of 12 patients and reduced bleeding (Pecci et al, 2010). 3 Aplastic anaemia patients have a low platelet count and highly elevated level of endogenous TPO. Surprisingly, after treatment with eltrombopag daily, 11 of 25 subjects had a haematological response with six subjects ultimately having a trilineage increase as well as increased bone marrow cellularity (Olnes et al, 2012). With already highly elevated TPO levels, it is unclear whether eltrombopag is producing this beneficial effect via the TPO receptor of some other ‘off-target’ effect. 4 MDS patients have thrombocytopenia and this often occurs before other cytopenias (Kantarjian et al, 2007; Bryan et al, 2010; Al Ameri et al, 2011). Treatment of 250 patients with International Prognostic Scoring System low/intermediate-1 MDS with romiplostim (167 patients) produced an increase in platelet count, reduced need for transfusion and reduced bleeding compared with treatment with placebo (83 patients). However, a rise in blast count to >10% occurred in 15% of those on romiplostim versus 36% of those on placebo (Giagounidis et al, 2011; Kantarjian et al, 2012). 5 Liver disease patients often have thrombocytopenia and are at increased risk of bleeding with procedures. Treatment of such patients with platelet counts
