Pathogenesis of chronic immune thrombocytopenia: increased platelet destruction and/or decreased platelet production


Sherrill J. Slichter, MD, Director, Platelet Transfusion Research, Puget Sound Blood Center, 921 Terry Avenue, Seattle, WA 98104-1256, USA. E-mail:


Chronic immune thrombocytopenia (ITP) is a haematological disorder in which patients predominantly develop skin and mucosal bleeding. Early studies suggested ITP was primarily due to immune-mediated peripheral platelet destruction. However, increasing evidence indicates that an additional component of this disorder is immune-mediated decreased platelet production that cannot keep pace with platelet destruction. Evidence for increased platelet destruction is thrombocytopenia following ITP plasma infusions in normal subjects, in vitro platelet phagocytosis, and decreased platelet survivals in ITP patients that respond to therapies that prevent in vivo platelet phagocytosis; e.g., intravenous immunoglobulin G, anti-D, corticosteroids, and splenectomy. The cause of platelet destruction in most ITP patients appears to be autoantibody-mediated. However, cytotoxic T lymphocyte-mediated platelet (and possibly megakaryocyte) lysis, may also be important. Studies supporting suppressed platelet production include: reduced platelet turnover in over 80% of ITP patients, morphological evidence of megakaryocyte damage, autoantibody-induced suppression of in vitro megakaryocytopoiesis, and increased platelet counts in most ITP patients following treatment with thrombopoietin receptor agonists. This review summarizes data that indicates that the pathogenesis of chronic ITP may be due to both immune-mediated platelet destruction and/or suppressed platelet production. The relative importance of these two mechanisms undoubtedly varies among patients.

Chronic immune thrombocytopenia (ITP) is a haematological disorder that presents clinically with varying degrees of petechiae, purpura, and mucosal bleeding. Symptoms of ITP generally appear when platelet counts drop from normal levels, of 150–450 × 109/l to below 50 × 109/l (Cines & McMillan, 2005). Except for isolated thrombocytopenia, other routine blood counts in patients with ITP are normal, unless there is significant bleeding. Bone marrow examination shows normal to increased numbers of megakaryocytes. Medical investigations of ITP pathogenesis date back to 1915, when Frank (1915) hypothesized that the disorder was due to suppressed platelet production, possibly from a factor produced by the spleen. The following year, Kaznelson (1916) postulated that increased platelet destruction by the spleen was the cause of thrombocytopenia, based on the first report of the potentially curative effect of splenectomy for patients with ITP. Subsequent studies show that both increased destruction and/or decreased rates of platelet production may contribute to the pathogenesis of ITP.

Evidence for increased platelet destruction

ITP blood and plasma infusion studies

In 1950, the focus was shifted to platelet destruction as the major pathogenic mechanism responsible for ITP. When Dr William Harrington was infused with 500 ml of whole blood from a patient with active chronic ITP, his platelet count dropped from a baseline of 800 × 109/l–25 × 109/l within 2 h following infusion, and to zero after 24 h, followed by recovery over the following few days. Subsequent studies showed that transfusion of 500 ml of whole blood or the plasma equivalent from ITP patients into healthy recipients or patients with inoperable neoplasms showed similar results in 16 (61·5%) of 26 recipients (Harrington et al, 1953).

In the 1960s, ITP plasma infusion studies by Shulman et al (1965) showed that: (i) the severity of post-transfusion thrombocytopenia was dose-dependent, (ii) the plasma factor that caused thrombocytopenia could be adsorbed by platelets, (iii) the plasma factor was present in the immunoglobulin G (IgG)–rich fraction after ion-exchange chromatography. The antiplatelet factor reacted with both heterologous and autologous platelets, and a greater amount was required to produce the same degree of thrombocytopenia in recipients who had undergone splenectomy. It was postulated that the antiplatelet factor in ITP was an antiplatelet autoantibody (Harrington et al, 1953; Shulman et al, 1965).

Intravascular platelet survival in ITP

Platelets are normally removed from the circulation by the reticuloendothelial system, mainly because of senescence (Aster & Jandl, 1964). Recent studies in mice suggest that an intrinsic programme of apoptosis controls platelet lifespan (Mason et al, 2007). Based on the results of several large studies, reduced intravascular platelet survival was noted in essentially all patients with active ITP, with the major sites of platelet destruction appearing to be in the liver and spleen (Najean et al, 1967, 1997; Aster & Keene, 1969; Harker, 1970; Branehog et al, 1974, 1975; Ridell & Branehog, 1976; Kernoff et al, 1980). In most studies, there was a direct relationship between the circulating platelet counts and platelet lifespan, regardless of whether autologous or allogeneic platelets were used. However, in a study of 27 patients with thrombocytopenia secondary to marrow aplasia, a direct relationship between the platelet count and platelet survival was also observed at platelet counts of <100 × 109/l (Hanson & Slichter, 1985). Besides senescent loss of platelets from circulation, a fixed fraction of platelets (7·1 × 109/l) are lost randomly apparently in an endothelial supportive function. This random platelet loss becomes more apparent at lower platelet counts, resulting in reduced platelet survivals. Thus, reduced platelet survival in a thrombocytopenic patient does not necessarily indicate accelerated platelet destruction, although platelet survival times in ITP patients with similar platelet counts are usually much shorter than in patients with marrow aplasia (Tomer et al, 1991). These platelet survival data support the conclusion that thrombocytopenia in ITP patients is caused, at least in part, by increased platelet destruction.

Clinical response of ITP patients to therapy that inhibits phagocytosis

The importance of platelet destruction in ITP is further supported by the demonstration that splenectomy results in a complete, unmaintained remission in about two-thirds of ITP patients (Branehog, 1975; Gernsheimer et al, 1989; Louwes et al, 2001). In addition, agents such as intravenous IgG, anti-D and corticosteroids, which work, in part, by interfering with phagocytosis, may result in a temporary normalization of the platelet count (Cines & McMillan, 2005).

Causes of platelet destruction

Autoantibody-induced platelet destruction in ITP. In vivo infusion studies suggested that the antiplatelet factor in ITP plasma was an autoantibody. Early attempts to demonstrate antiplatelet antibodies in ITP, such as tests of platelet agglutination, platelet factor 3 release, serotonin release, and complement activation, were largely unsuccessful because of poor sensitivity and/or specificity of the assays. In the 1970s, a variety of studies showed that platelet-associated IgG was elevated in more than 90% of ITP patients (McMillan, 1983). However, the assays had poor specificity and were unable to distinguish patients with ITP from those with non-immune thrombocytopenia (Mueller-Eckhardt et al, 1980). In the 1980s, several laboratories developed antigen-specific assays that directly measured antibodies specific for several platelet surface proteins, notably glycoprotein (GP) IIb/IIIa and GPIb/IX (McMillan, 2005). These newer assays demonstrated a higher degree of specificity for patients with immune thrombocytopenia and are positive in about 60% of patients (Brighton et al, 1996; Warner et al, 1999; McMillan et al, 2003).

Antibody-induced platelet destruction occurs primarily by phagocytosis in the reticuloendothelial system. Platelet phagocytosis by splenic macrophages and peripheral blood neutrophils in ITP has been documented morphologically (Firkin et al, 1969; Handin & Stossel, 1974), and antibody-dependent platelet phagocytosis has been demonstrated in vitro (McMillan et al, 1974; Tsubakio et al, 1983). There is also in vitro evidence of autoantibody-induced complement activation with subsequent platelet lysis, suggesting that this mechanism may be important in some ITP cases (Tsubakio et al, 1986). Additional support for antibody-induced phagocytosis, as an important mechanism in ITP, is the demonstration that the infusion of a monoclonal antibody (3G8) against an Fcγ receptor into an ITP patient with severe thrombocytopenia resulted in temporary normalization of the platelet count (Clarkson et al, 1986). Furthermore, there is evidence that certain polymorphisms in the Fcγ-RIIIa gene (FCGR3A) are differentially represented in ITP (Fujimoto et al, 2007).

T lymphocyte-mediated platelet lysis.  The lack of detectable platelet-associated autoantibodies in about 40% of ITP patients, and the absence of thrombocytopenia in almost the same percentage of ITP plasma recipients in Harrington’s early studies (Harrington et al, 1953), suggest that alternative mechanisms for platelet destruction may exist. A recent study provided evidence that cytotoxic T lymphocytes from some ITP patients can induce platelet lysis (Olsson et al, 2003). Stimulated T lymphocytes taken from some patients with active ITP, but not from ITP patients in remission or from healthy subjects, caused significant lysis of autologous platelets. As platelets and megakaryocytes share many surface antigens, it seems likely that cytotoxic T lymphocytes may also damage megakaryocytes. This possibility is supported by a recent study showing an increase in the surface expression of VLA-4 and CX3Cr1 (factors involved in T-cell homing) in bone marrow T cells in ITP, as well as an increased number of CD3+ T cells in ITP bone marrow when compared to controls (Olsson et al, 2008).

Studies have also shown that T lymphocytes from ITP patients express abnormal levels of gene products that are associated with apoptosis (Olsson et al, 2005; Zhang et al, 2006). Furthermore, cytotoxic T lymphocytes from ITP patients with active disease are resistant to dexamethasone-induced suppression, while T lymphocytes from ITP patients who are in remission are more susceptible (Olsson et al, 2005). These findings suggest that the autoreactive T lymphocytes in patients with active ITP are not cleared by normal apoptotic mechanisms, resulting in the continued destruction of platelets. Regulatory T lymphocytes, which function to prevent the activation and proliferation of autoreactive T lymphocytes, are also reduced in patients with ITP (Sakakura et al, 2007) and may further contribute to the dysregulation of T lymphocytes in ITP. T cell-mediated platelet and/or megakaryocyte lysis may well be an important mechanism of thrombocytopenia in some patients with ITP, but its existence is only based on ‘in vitro’ rather than ‘in vivo’ evidence.

Evidence for decreased platelet production

Morphological studies

In studies dating back to the 1940s, bone marrow from ITP patients examined by light microscopy showed morphological evidence of: (i) abnormal thrombopoiesis, including normal or increased megakaryocyte numbers with a larger percentage of younger forms lacking cytoplasmic granularity or evidence of platelet formation, (ii) degenerative changes in both nuclei and cytoplasm (Dameshek & Miller, 1946; Diggs & Hewlitt, 1948). Phase contrast studies in the 1950s confirmed these findings and also showed that infusing healthy controls with plasma from ITP patients produced these same abnormalities in megakaryocytes (Pisciotta et al, 1953). In the 1980s, electron microscopy studies confirmed the presence of abnormal megakaryocytes, as shown by markedly distended demarcation membranes, vacuolized cytoplasm, swollen mitochondria, and disrupted peripheral zone. In some cases, these damaged cells were attached to monocytes that appeared to phagocytose fragments of the megakaryocytes (Stahl et al, 1986).

Recently, Houwerzijl et al (2004) conducted ultrastructural studies of bone marrow from 11 ITP patients and reported morphological abnormalities in 78 ± 14% [standard deviation (SD)] of the megakaryocytes. Abnormalities were more common as the cells matured: 37 ± 37% in stage I (immature), 65 ± 37% in stage II (maturing), and 84 ± 16% in stage III (mature). These morphological alterations showed features of para-apoptosis, including mitochondrial swelling with cytoplasmic vacuolization, distention of the demarcation membranes, and chromatin condensation within the nucleus. Affected megakaryocytes had an intact but thickened peripheral zone. Many abnormal ITP megakaryocytes were surrounded by neutrophils and macrophages, some undergoing phagocytosis. Morphological features of apoptosis were noted in four specimens (36%) but only in stage III megakaryocytes. In addition, immunohistochemical staining for activated caspase-3 was positive in about one-quarter of the mature megakaryocytes from these patients, whereas staining of control megakaryocytes was negative. Para-apoptotic abnormalities similar to those found in the ITP megakaryocytes could be induced in megakaryocytes produced from normal CD34+ cells cultured with megakaryocytes in the presence of ITP plasma, suggesting that antiplatelet antibodies might play a role in initiating the cascade of programmed cell death.

Effects of autoantibodies on megakaryocytopoiesis

Animal models developed in the late 1960s to study normal and aberrant megakaryocytopoiesis provided the first opportunity to test the hypothesis that antiplatelet antibodies might not only cause peripheral destruction of mature platelets, but also directly affect megakaryocyte maturation in ITP. Using a rat model, Rolovic et al (1970) demonstrated a striking alteration in megakaryocytopoiesis in rats with thrombocytopenia induced by antiplatelet serum compared to those undergoing repeated thrombocytopheresis. Although there was an increase in the number of mature megakaryocytes compared to normal in both the antiplatelet and platelet pheresis groups, maturation of megakaryocytes from stage I to stage III was severely impaired only in the antiplatelet group.

Subsequent rapid advances were made in identifying the specific antigens targeted by antiplatelet antibodies (van Leeuwen et al, 1982; McMillan et al, 1987; Kiefel et al, 1991). As previously noted, the two most frequent platelet membrane glycoprotein targets, GPIIb/IIIa heterodimer and GPIb/IX complex, are also expressed on megakaryocytes during the early stages of differentiation. ITP autoantibodies were also found to bind to megakaryocytes (McKenna et al, 1962; McMillan et al, 1978).

Initially, technical challenges with megakaryocyte colony assays, sampling error, and patient-to-patient variability resulted in reports of contradictory results in the evaluation of platelet production in ITP. In an effort to develop a standardized assay, Chang et al (2003) used thrombopoietin (TPO)-induced CD34+ stem cells from cord blood in liquid culture to measure the effects of ITP plasma or monoclonal antibodies on normal megakaryocyte production. With this abundant, cost-effective source of healthy stem cells, the investigators were able to compare multiple patients with controls in a single run or compare a number of plasma samples from different time points on the same stem cell product. In this study of 53 paediatric ITP patients, there was significant suppression of megakaryocyte production in vitro when the ITP plasma samples contained anti-GPIb/IX antibodies alone or in combination with anti-GPIIb/IIIa antibodies (Chang et al, 2003).

Platelet adsorption studies further demonstrated the importance of antibodies in suppressing megakaryocyte production. Plasma from ITP patients was adsorbed with fresh platelets to remove anti-GPIb autoantibodies (Chang et al, 2003). Megakaryocyte production was doubled in the presence of the adsorbed ITP plasma compared to the same plasma without adsorption, whereas platelet adsorption of normal control plasma had no effect on megakaryocyte yield.

McMillan et al (2004) compared the effect of plasma from 18 adult patients with severe chronic ITP to that of 28 control subjects on the in vitro production and maturation of megakaryocytes. Plasma from 12 ITP patients (67%) significantly suppressed in vitro megakaryocyte production, ranging from 26% to 95% suppression, without any observable effect on the number of non-megakaryocytic cells. Among these 12 ITP plasma samples, six (50%) had anti-GPIIb/IIIa antibodies, three (25%) had anti-GPIb/IX antibodies, and three (25%) had both anti-GPIIb/IIIa and anti-GP1b/IX antibodies. As seen in studies of paediatric patients (Chang et al, 2003), the role of adult ITP autoantibody in the suppression of megakaryocyte production was supported by the finding that IgG from ITP patients inhibited megakaryocyte production when compared to control IgG, and adsorption of autoantibody with immobilized antigen resulted in less suppression of megakaryocyte production compared to plasma that was not adsorbed. In addition, the effect of ITP plasma on ploidy distribution was studied in six plasma samples that showed suppressed megakaryocyte production and one that did not. Those showing suppressed megakaryocyte production also had impaired maturation (McMillan et al, 2004), similar to that originally described in the rat model of ITP (Rolovic et al, 1970).

The in vitro assays described also provided a new opportunity to study the effect of autoantibodies that are specific for novel antigen targets. To examine the effects of autoantibodies on the TPO receptor, cMpl, Kuwana et al (2002) compared plasma from patients with ITP (n = 84) or systemic lupus erythematosus (SLE) (n = 69) with healthy controls (n = 60). Binding of plasma antibodies to cMpl was 8·3%, 11·6%, and 0% for ITP, SLE, and controls respectively. ITP patients with anti-cMpl antibodies had significantly greater TPO levels than ITP patients who did not have antibodies (= 0·004) and healthy controls (< 0·001). Suppression of megakaryocyte proliferation by anti-cMpl autoantibodies was demonstrated using IgG fractions isolated from the plasma of thrombocytopenic ITP or SLE patients in megakaryocyte colony-forming assays.

Direct inhibition of cMpl by autoantibodies potentially represents immune-mediated interference with megakaryocytopoiesis; however, this has not been observed in most patients with ITP, and limited data are available to determine its relative importance to their thrombocytopenia. The majority of patients have bone marrow histopathology consistent with an ‘apoptosis-like’ programmed cell death in their megakaryocytes, the immune mechanism for which remains largely unknown (Houwerzijl et al, 2006). The further evolution of large-scale, reproducible, in vitro megakaryocytic stem cell assays may provide valuable tools for understanding the mechanisms by which autoantibodies or other plasma factors inhibit megakaryocyte expansion and maturation or prematurely trigger para-apoptosis and death.

Platelet turnover in ITP patients

Platelet kinetic studies enable not only determination of intravascular survival, but also calculation of platelet turnover as a measure of platelet production. Harker and Finch (1969) correlated radiolabeled thrombokinetic measurements with studies of megakaryocyte mass (cell number and volume) to define thrombocytopenic disorders. Platelet recovery was calculated from the platelet recovery at zero time (by extrapolating the platelet survival curve to time zero), multiplied by the estimated blood volume and divided by the 51Cr activity injected. At stable platelet counts, the rate of platelet production equals platelet destruction (i.e., platelet turnover). Platelet turnover (× 109/l per day) is calculated as:


The last term in the equation is used to correct for splenic pooling. In asplenic individuals, about 90% of the transfused platelets circulate and the remaining 10% are considered to have been damaged during radiolabelling, resulting in their immediate removal after transfusion (Aster, 1966).

In 15 healthy subjects studied using autologous radiolabeled platelets, platelet counts averaged 250 ± 35 × 109/l, platelet recoveries 65 ± 4%, platelet survivals 9·9 ± 0·6 d, and platelet turnovers 35 ± 400 × 109 platelets/l per day (Harker & Finch, 1969). Thrombocytopenic patients demonstrating a direct correlation between decreased platelet turnovers and decreased megakaryocyte mass were defined as having hypoproliferative thrombocytopenia. Patients with ineffective thrombopoiesis had platelet turnovers less than their megakaryocyte mass. Splenomegaly resulted in reduced platelet recoveries, normal platelet survivals, and increased platelet turnovers with corresponding increases in megakaryocyte mass. Increased platelet destruction, whether immunological (ITP) or consumptive (e.g., consumptive coagulopathy, haemolytic uremia syndrome, vasculitis), was characterized by accelerated platelet turnovers with corresponding increases in megakaryocyte mass.

Table I shows platelet turnover data based on autologous thrombokinetic measurements from seven studies involving 218 patients with untreated ITP (Branehog, 1975; Branehog et al, 1975; Stoll et al, 1985; du P Heyns et al, 1986; Ballem et al, 1987; Isaka et al, 1990; Louwes et al, 1999). These results showed that increased platelet production rates are uncommon in ITP patients, with most exhibiting either depressed or normal platelet production rates. A key question raised by this observation is why platelet production rates in ITP patients were rarely increased in these studies compared to prior studies, in which they were almost uniformly increased (Harker, 1970; Branehog & Weinfeld, 1974).

Table I.   Platelet turnover rates in untreated ITP patients based on autologous platelet survival measurements.
StudyPublication dateNIsotopePlatelet turnover*
Decreased n (%)Normal n (%)Increased n (%)
  1. *Platelet turnover was considered to be decreased if it was more than two standard deviations (SD) below the average turnover value in each study’s normal control group; normal if it was within two SDs above or below the normal value; and increased if it was more than two SDs above normal.

  2. †One SD was used to define the groups.

Branehog et al (1975)19751051Cr08 (80)2 (20)
Branehog et al (1975)1975851Cr06 (75)2 (25)
Stoll et al (1985)19852051Cr4 (20)14 (70)2 (10)
du P Heyns et al (1986)198610111In5 (50)1 (10)4 (40)
Ballem et al (1987)19871751Cr or 111In7 (41)9 (53)1 (6)
Isaka et al (1990)19901251Cr8 (67)4 (33)0
Louwes et al (1999)199914151Cr60 (43)47 (33)34 (24)

Although the reasons for discrepancies between results in early and later thrombokinetic studies of ITP patients are not clear from the available data, they are most likely due to the source of platelets used in each study. Radiolabeled allogeneic platelets were predominantly used in early studies when only 51Cr was available as a radiolabel, whereas autologous platelets were used in later studies when 111In became available. The better labelling characteristics of 111In versus51Cr allowed adequate radiolabelling with low platelet counts. This hypothesis is supported by survival data from paired autologous and allogeneic platelet survival measurements in 24 patients with ITP (Fig 1) (Abrahamsen, 1970; Branehog & Weinfeld, 1974; du P Heyns et al, 1986; Ballem et al, 1987). In these studies, when autologous platelet survival measured no more than 1 d (n = 13), there was a significant and positive relationship (r = 0·79; = 0·001) between the two survival measurements (mean autologous platelet survival: 0·46 ± 0·33 d; mean allogeneic platelet survival: 0·40 ± 0·30 d). However, when autologous platelet survival was longer than 1 d (n = 11), the corresponding allogeneic platelet survival times were significantly lower than when autologous platelet survival was ≤1 d (mean autologous platelet survival: 2·48 ± 1·23 d; mean allogeneic platelet survival: 1·27 ± 0·6 d; = 0·01). These data are consistent with Dornhorst’s original hypothesis (Dornhorst, 1951), which states that data using either allogeneic or autologous platelets give comparable survivals if the survivals are substantially reduced, but with long autologous platelet survival times, allogeneic platelets do not comparably reproduce the circulating population of autologous platelets. If the survival of autologous platelets is relatively long, it suggests that these platelets are at least partially resistant to the effects of the autoantibody. In contrast, transfused allogeneic platelets that represent a heterogeneous population of platelets may be rapidly removed when exposed to the antibody. A similar situation of longer autologous compared to allogeneic red cell survivals has been observed in patients with autoimmune haemolytic anaemia (Mollison, 1979).

Figure 1.

 Paired autologous versus allogeneic platelet survival measurements in ITP subjects. With autologous platelet survivals of ≤1 d (n = 13), there is a statistically significant relationship with allogeneic platelet survivals (r = 0·79, = 0·001). However, in 11 patients with autologous platelet survivals of >1 d, allogeneic platelet survivals were always significantly shorter (= 0·01). The line represents a theoretical 1:1 relationship between the two platelet survival measurements. Data are from Abrahamsen (1970); Branehog et al (1974); du P Heyns et al (1986); Ballem et al (1987).

Thrombopoietin in ITP

Megakaryocytopoiesis is affected by many different cytokines, including stem cell factor; interleukins 1β, 3, 6, and 11; granulocyte, macrophage, and granulocyte-macrophage colony-stimulating factors, and, in some cases, erythropoietin (Nichol, 1997). However, none are specific or potent regulators of megakaryocyte development or platelet production. In the mid-1990s, cMpl, the major cytokine receptor involved in megakaryocytopoiesis, was discovered (Vigon et al, 1992), along with its protein ligand (Bartley et al, 1994; Kuter et al, 1994; Lok et al, 1994; de Sauvage et al, 1994; Kato et al, 1995). In its native form, the cMpl ligand, more commonly known as TPO, is the primary regulator of megakaryocytopoiesis and thrombopoiesis. Endogenous TPO (eTPO) stimulates the growth of megakaryocyte progenitors and megakaryocyte maturation, leading to platelet production. Based on studies of patients with severe hepatic diseases, it appears that eTPO is produced primarily in the liver, with some studies suggesting secondary production in the spleen, kidneys, bone marrow, and central nervous system (Dame et al, 2003). In the liver, TPO is produced at a constant rate, and circulating levels are thought to be regulated by receptor-mediated clearance by platelets and megakaryocytes (Karpatkin, 1980; Li et al, 1999).

Studies using different assay systems to measure eTPO levels in serum and plasma can be compared by normalizing the data to values reported in the same assay system for healthy individuals. eTPO concentrations measured by enzyme immunoassay in normal subjects are generally lower than 200 pg/ml (Emmons et al, 1996; Marsh et al, 1996; Tahara et al, 1996) and range from 20 to 218 pg/ml (in plasma) with normal platelet counts (150–450 × 109/l) (Nichol, 1998). Several studies have reported higher concentrations of eTPO in serum compared to matching plasma samples (Folman et al, 1997; Nichol, 1997; Wang et al, 1997). Overall, eTPO concentrations in patients compared with healthy individuals are generally consistent across studies (Nichol, 1998). It is important that patient and control specimens are collected in the same manner, and that the platelet count – which affects circulating levels of TPO – be considered when comparing results (Nichol, 1998).

eTPO concentrations as great as 10–25 times that of normal have been observed in disorders in which there is diminished platelet production due to bone marrow failure or ablation, such as aplastic anaemia (Emmons et al, 1996; Kosugi et al, 1996; Nichol, 1997; Haznedaroglu et al, 1998; Hirayama et al, 1998; Koike et al, 1998; Kosar et al, 1998). eTPO deficiencies fall into two main categories: (i) primary (e.g., decreased eTPO production from hepatic impairment), (ii) functional (e.g., eTPO deficiencies due to increased eTPO clearance by cMpl expressing cells, such as platelets, megakaryocytes, or leukemic blast cells) (Emmons et al, 1996; Ichikawa et al, 1996; Kunishima et al, 1996; Usuki et al, 1996; Hiyoyama et al, 1997; Nichol, 1997). eTPO levels and platelet counts for patients with ITP or aplastic anaemia and for healthy controls are presented in Fig 2 (Nichol, 1998). Patients with ITP characteristically exhibit lower-than-expected eTPO levels (Nichol, 1998).

Figure 2.

 Platelet counts and TPO levels in subjects with ITP, aplastic anaemia, and healthy controls. TPO levels are normal in 75% of ITP patients despite lower than normal circulating platelet numbers. Data are from Nichol (1998).

Many clinical studies have compared eTPO levels and platelet counts in patients with ITP (Emmons et al, 1996; Kosugi et al, 1996; Marsh et al, 1996; Meng et al, 1996; Mukai et al, 1996; Tahara et al, 1996; Hiyoyama et al, 1997; Nichol, 1997; Haznedaroglu et al, 1998; Hirayama et al, 1998; Hou et al, 1998; Koike et al, 1998; Kosar et al, 1998; Sakane et al, 1998; Fabris et al, 2000; Gouin-Thibault et al, 2001; Kappers-Klunne et al, 2001; Gu et al, 2002; Aledort et al, 2004; Kuwana et al, 2005). Most studies reported no significant difference between ITP and normal eTPO concentrations (Kunishima et al, 1996; Hiyoyama et al, 1997; Hou et al, 1998; Koike et al, 1998; Kappers-Klunne et al, 2001; Kuwana et al, 2005), although some reported significantly greater levels of eTPO among ITP patients than among healthy controls (Kosugi et al, 1996; Mukai et al, 1996; Hirayama et al, 1998; Fabris et al, 2000), while others reported lower levels (Haznedaroglu et al, 1998; Kosar et al, 1998; Gu et al, 2002). However, despite similar platelet counts, eTPO levels in ITP patients were consistently lower than those observed in patients with aplastic anaemia (Emmons et al, 1996; Kosugi et al, 1996; Hirayama et al, 1998; Koike et al, 1998; Kosar et al, 1998; Gu et al, 2002).

In most studies of ITP, eTPO levels did not correlate with platelet counts (Kosugi et al, 1996; Mukai et al, 1996; Koike et al, 1998; Gouin-Thibault et al, 2001; Gu et al, 2002; Aledort et al, 2004). However, one study (Hiyoyama et al, 1997) reported a significant inverse (negative) correlation; another (Kappers-Klunne et al, 2001) reported a significant negative correlation among patients with platelet counts no higher than 20 × 109/l but a positive correlation among all patients between eTPO and the logarithm of the platelet turnover. Two ITP studies conducted correlation analyses between eTPO and megakaryocyte count: Hirayama et al (1998) reported a significant positive correlation, and Sakane et al (1998) reported a significant negative correlation (among all ITP patients). The observation of reduced TPO levels in the presence of low platelet counts suggests impaired platelet production and failure to compensate for the thrombocytopenia. The rationale for studying Mpl agonists in the treatment of ITP was driven by both the apparent functional TPO deficiency in some patients with ITP and the potential to increase platelet production to a rate that exceeds platelet destruction by the immune system.

Clinical response of ITP patients to TPO receptor agonists.  Until recently, treatment options for ITP have been intended to inhibit the immune response or interfere with platelet destruction. Treatments have included corticosteroids, splenectomy, danazol, immunoglobulins, and a variety of chemotherapeutic and immunosuppressive agents (Cines & McMillan, 2005). However, emerging trends in treatment are aimed toward augmenting functional TPO interaction with megakaryocytes and precursor cells to boost platelet production. Early studies showed that pegylated recombinant human megakaryocyte growth and development factor (Peg-rHuMGDF) stimulated platelet production in ITP patients (Nomura et al, 2002). In other studies in healthy individuals who received Peg-rHuMGDF, some healthy subjects and one cancer patient developed antibodies that cross-reacted with eTPO, ultimately resulting in the discontinuation of the clinical development programme.

Recently published clinical trial results in patients with ITP show the promise of TPO receptor agonists lacking structural homology with eTPO (romiplostim and eltrombopag). Clinical trials with these agents in patients with chronic ITP show that treatment with these agents results in a significant increase in platelet counts in the majority of patients with ITP which can be maintained for several months with continued therapy. Several reports on the safety and efficacy of these products in patients with ITP are available (Bussel et al, 2006, 2007, 2009; Newland et al, 2006; Jenkins et al, 2007; Kuter et al, 2008). These results provide further evidence for the suppression of platelet production in patients with chronic ITP.


This article reviews the evidence that two pathogenic immune-mediated mechanisms may be operative in patients with chronic ITP: (i) increased platelet destruction; (ii) decreased platelet production (data summarized in Table II). Autoantibody-mediated accelerated platelet destruction with subsequent clearance in the reticulo-endothelial (RE) system can be reduced by agents that prevent RE phagocytosis, such as IV IgG, anti-D, or corticosteroids, or may be permanently resolved by splenectomy. Studies of platelet turnover have demonstrated decreased production rates in a proportion of ITP patients, and evidence indicates several contributing mechanisms. Antiplatelet autoantibodies are not only an important mediator of platelet destruction, but also suppress megakaryocytopoiesis in patients with ITP. Recent data showing platelet lysis by cytotoxic T lymphocytes suggest that these cells may also damage megakaryocytes, resulting in suppression of platelet production. TPO levels in ITP patients are lower than expected in a disease of reduced platelet number. In addition, the effectiveness of TPO receptor agonists in the treatment of patients with chronic or refractory ITP further demonstrates that insufficient platelet production is a major mechanism in the pathogenesis of thrombocytopenia in some ITP patients. However, it should be recognized that the relative importance of these two mechanisms of reducing platelet counts may vary widely among ITP patients.

Table II.   Summary of evidence supporting either increased platelet destruction or decreased platelet production as mechanism of disease in ITP.
ThemeEvidence of increased platelet destructionEvidence of decreased platelet production
Early hypothesesSplenectomy potentially curative, suggesting increased platelet destruction in the spleenKaznelson (1916)Splenic factor caused decreased platelet production by megakaryocytesFrank (1915)
Plasma transfusion studiesFactor capable of causing thrombocytopenia transmitted by transfusion of plasma from patients with ITPHarrington et al (1953)
Plasma factor causing thrombocytopenia adsorbed by platelets and precipitated with immunoglobulins, suggesting it is an antiplatelet antibodyShulman et al (1965)
Antiplatelet antibodiesElevated platelet-associated IgG found in more than 90% of ITP patientsMcMillan (1983)Impaired megakaryocyte maturation in rats given antiplatelet serum to induce thrombocytopeniaRolovic et al (1970)
Detection by new assays of antibodies specific for several platelet surface proteins, including GPIIb/IIIa and GPIb/IXBrighton et al (1996); Warner et al (1999); McMillan et al (2003)Detection of ITP autoantibodies bound to megakaryocytesMcKenna et al (1962); McMillan et al (1978)
Suppression of megakaryocyte production in vitro by ITP plasma containing anti-GPIb/IX or anti-GPIIb/IIIa antibodiesChang et al (2003); McMillan et al (2004)
ITP TreatmentClinical response to therapy that prevents platelet phagocytosis; e.g., steroids, IV IgG, anti-D, and splenectomyBranehog (1975); Gernsheimer et al (1989); Louwes et al (2001); Cines and McMillan (2005)TPO mimetics increase platelet countsNomura et al (2002); Bussel et al (2006); Newland et al (2006); Bussel et al (2007); Jenkins et al (2007); Bussel et al (2009); Kuter et al (2008)
MorphologyMorphological demonstration of antibody-induced platelet phagocytosis by splenic macrophagesFirkin et al (1969); Handin and Stossel (1974); McMillan et al (1974)Light and electron microscopy studies of ITP patient samples show presence of immature and abnormal megakaryocytesDameshek and Miller (1946)
Morphological features of apoptosis found in ITP patient megakaryocytesHouwerzijl et al (2004)
Platelet survival studiesIntravascular platelet survival is reduced in patients with ITP. Platelet destruction occurred in the spleen and liverNajean et al (1967); Aster and Keene (1969); Harker (1970); Branehog et al (1974, 1975); Ridell and Branehog (1976); Kernoff et al (1980); Najean et al (1997)
Platelet kinetics studiesStudies using allogenic platelets show increased platelet production rates in ITP patientsHarker and Finch (1969), (Harker (1970), (Branehog and Weinfeld (1974)Studies using autologous platelets show decreased or normal platelet production rates in most untreated ITP patientsBranehog (1975); Branehog et al (1975); Stoll et al (1985); du P Heyns et al (1986); Ballem et al (1987); Isaka et al (1990); Louwes et al (1999)
Allogenic and autologous platelet survival times are only comparable when platelet survival times are ≤1 dAbrahamsen (1970); Branehog et al (1974); du P Heyns et al (1986); Ballem et al (1987)
TPO  eTPO levels normal in patients with ITP, suggesting failure to compensate for thrombocytopeniaReviewed in Nichol (1998)
Clincal studies show ITP patients respond to treatment with TPO receptor agonistsBussel et al (2007, 2009); Kuter et al (2008)


Medical writing assistance was provided by Julie Gage, who was funded by Amgen Inc.

Author contribution

DN, RM, JLN, and SS wrote the paper. All authors reviewed and revised the paper, and provided approval of the final version.

Conflict of interest

DN has no conflicts-of-interest to declare. RM has served on ad hoc advisory boards for Amgen and Galaxo-Smith-Kline (GSK) and owns stock in GSK. JLN is a former employee of Amgen Inc., SS has previously been a consultant to Amgen and received research support.