Pathophysiology and management of chronic immune thrombocytopenia: focusing on what matters

Summary

Immune thrombocytopenia (ITP) is a common autoimmune disease characterized by low platelet counts and an increased risk of bleeding. Antibody-mediated platelet destruction has been the prevailing hypothesis to explain ITP pathogenesis, supported by the efficacy of B-cell depletion therapy; however, the recent success of thrombopoietin receptor agonists lends support to the notion that platelet production is also insufficient. Best practice for the management of chronic ITP has not yet been established because data from comparative trials are lacking. Despite renewed interest in novel drugs capable of increasing platelet counts, ultimate treatment goals for ITP patients must be kept in mind: to improve patients’ health and well-being. In this article, the pathophysiology of ITP is reviewed and key remaining questions about mechanism are explored. A rational approach to the management of ITP in adults is outlined, acknowledging evidence and evidence gaps, and highlighting the need for clinically important endpoints in future clinical trials.

Up until the last decade, not much had changed in the management of patients with chronic immune thrombocytopenia (ITP). Splenectomy was considered the mainstay of treatment, chronic corticosteroid use was generally discouraged and refractory patients were managed in a variety of ways with moderate success. Roughly 10 years ago, rituximab, a monoclonal antibody against CD20, started to be used for some patients with ITP and resulted in a platelet count increase that was often sustained for months (Arnold et al, 2007). Several years later, a new class of drugs, called thrombopoietin (TPO) receptor agonists, was shown to produce a dose-dependent increase in platelet counts even in some patients with refractory ITP (Bussel et al, 2006). Today, although not a licenced indication, rituximab is commonly used to treat ITP, and two TPO receptor agonists have been approved for use in ITP- romiplostim and eltrombopag. These new drug discoveries have led to conceptual advances and additional treatment options for refractory patients. They have also led to the potential for overtreatment in the absence of clinical trials powered on clinically meaningful endpoints.

The objectives of this review are to summarize current understanding of ITP pathophysiology based on lessons learned from recent drug discoveries and to outline a rational approach to the treatment of adults with chronic relapsed ITP.

New concepts in the pathophysiology of ITP

Increased platelet destruction

The prevailing hypothesis to explain thrombocytopenia in ITP has been autoantibody-mediated platelet destruction. An immune basis for ITP fits with several familiar characteristics of the disease including the association with pregnancy; the efficacy of FcR-blocking therapies such as Rh immune globulin (anti-D) and intravenous immune globulin (IVIg) (among other mechanisms attributable to these therapies); and shortened survival of transfused platelets due to their rapid destruction (Buchanan et al, 1977).

The first evidence of a platelet autoantibody in ITP derived from experiments dating back to the 1950s, in which the infusion of blood or plasma from eight of 10 ITP patients caused profound thrombocytopenia in non-ITP controls (Harrington et al, 1951). Subsequently, the immunoglobulin fraction, IgG in particular, was found to be responsible for the anti-platelet activity by Fcγ-mediated platelet destruction in the reticuloendothelial system (Shulman et al, 1965). The association between ITP and human immunodeficiency virus (Bettaieb et al, 1992) or Helicobacter pylori infection (Stasi et al, 2009) provides further evidence for an immune cause of thrombocytopenia due to cross-reactive platelet antibodies (Takahashi et al, 2004). However recent in vitro evidence of H. pylori-induced platelet aggregation suggests that the mechanism of thrombocytopenia may be more complex (Yeh et al, 2010). Platelet destruction in ITP may also be due to direct T-cell mediated lysis (Olsson et al, 2003; Zhang et al, 2006) independent of platelet autoantibodies.

Two recent drug discoveries have provided evidence for and against the autoantibody hypothesis: rituximab and the TPO receptor agonists, respectively. Rituximab is a chimeric monoclonal antibody against CD20 licenced for the treatment of lymphoma and rheumatoid arthritis. A systematic review of rituximab in ITP showed that the drug was effective in inducing a platelet count response in approximately 60% of patients (Arnold et al, 2007). The first randomized trial investigating the effectiveness of rituximab in newly-diagnosed ITP demonstrated that rituximab plus single-course dexamethasone was more effective at achieving a platelet count response than dexamethasone alone (Zaja et al, 2010).

By targeting and destroying B-lymphocytes, the only human cells known to express CD20 (Jazirehi & Bonavida, 2005), rituximab effectively removes the cells that are responsible for pathogenic autoantibody production. Upon self-renewal, CD20⁺ B lymphocytes may either resume autoantibody production (relapse), or lose their pathogenic potential (remission). Alternatively, pathogenic autoantibodies may be produced by short-lived plasma cells or plasmablasts that are not replenished after B-cell depletion (Stasi, 2010). These hypotheses remain speculative: While rapid and profound B-cell depletion has been well documented following rituximab treatment (Stasi et al, 2001; Roll et al, 2006), the correlation with platelet autoantibody levels or with clinical response have not yet been demonstrated.

Besides its effect on autoantibody production, rituximab might improve ITP through its effect on T-cells. Compared with normal controls, ITP patients showed baseline T-cell abnormalities including high Th1/Th2 and Tc1/Tc2 ratios and oligoclonal T-cell expansion, which reverted to normal in some patients who responded to rituximab (Stasi et al, 2007). Similarly, reductions in the number and function of T-regulatory cells have been shown to improve following treatment in some patients (Stasi et al, 2008). Thus, B-cell depletion therapy may cause an increase in platelet count either by removing the autoantibody-producing cells, by normalizing dysfunctional T cells, or both.

Relative impairment in platelet production

The second discovery that forced a significant change in the understanding of ITP pathogenesis was TPO receptor agonists. This class of drugs works by stimulating the TPO receptor (c-Mpl) on haematopoietic stem cells and megakaryocytes (de Sauvage et al, 1996) causing an increase in the production of megakaryocytes and platelets (Italiano et al, 1999). Two agents in this class of medications are now licenced for the treatment of ITP: romiplsotim and eltrombopag. In randomized trials compared with placebo, each has been associated with a platelet count response in 60–70% of patients (Kuter et al, 2008; Bussel et al, 2009a). The remarkable success of TPO receptor agonists is consistent with earlier labelled autologous platelet studies that showed an impairment in platelet turnover in some patients with ITP (Stoll et al, 1985; Heyns et al, 1986; Ballem et al, 1987); yet is somewhat incongruous with the autoantibody theory because newly formed platelets would be expected to be rapidly destroyed. Thus, insufficient platelet production is another mechanism of thrombocytopenia in ITP.

Reconciling platelet destruction and insufficient platelet production

Increased destruction and insufficient production appear to be unrelated, paradoxical mechanisms for the development of thrombocytopenia in ITP. But in fact, they may be linked. Reconciling these mechanisms may help address several unanswered questions about ITP pathogenesis including: 1) Why are anti-platelet autoantibodies so elusive? 2) Why are circulating TPO levels reduced in ITP? and 3) Why do TPO receptor agonists work in the face of platelet autoantibodies?

Why are anti-platelet autoantibodies so elusive in ITP?

Early studies found that platelet-associated IgG was elevated in the majority of patients with ITP (Kelton et al, 1982); however, platelet-associated IgG was due to non-specific binding and was found in many patients with non-immune thrombocytopenia. Subsequently, antigen-specific assays were developed to measure antibodies against platelet glycoproteins (GP). Direct tests (measuring antibody on platelets) for anti-GP IIb-IIIa and/or anti-GP Ib-IX had improved specificity (78–93%), but moderate sensitivity (49–66%) in prospective studies (Warner et al, 1999; McMillan, 2003, 2005). Indirect tests (measuring circulating antibody) were only rarely positive.

The sensitivity of platelet antibody testing may be low because of the heterogeneity of the patient population in which they were studied. Alternatively, platelet autoantibodies may be missed by current monoclonal-based assays that only detect antibodies with known specificity (typically, GPIIb-IIIa and GPIb-IX). Finally, circulating autoantibodies may be undetectable because they are sequestered on other tissues and/or cells; particularly megakaryocytes. Indeed, in vitro cell culture studies have shown that ITP antibodies can impede megakaryocyte growth (Chang et al, 2003; McMillan et al, 2004), which might also explain why thrombopoiesis is impaired, and which would fit the autoantibody and underproduction hypotheses.

Why are circulating TPO levels low in ITP?

TPO is constitutively secreted from the liver (and, to a lesser degree, other tissues), meaning that the amount of free TPO released into circulation is constant at all times. Once TPO binds to c-Mpl it is internalized, degraded and removed from circulation. Thus, levels of free TPO are regulated by the number of circulating platelets and the megakaryocyte mass: When platelet counts are low, excess freeTPO is available for binding to megakaryocytes causing an increase in thrombopoiesis; and when platelet counts are high, less free TPO is available for binding (Kuter & Begley, 2002). TPO levels in ITP are usually normal or low, and not high as might be expected (Kosugi et al, 1996). Possible explanations are that excess free TPO is sequestered by an expanded megakaryocyte mass, but that does not explain the common finding of normal megakaryocyte number on bone marrow examinations; or that TPO is rapidly cleared from the circulation due to increased platelet turnover. Autoantibodies that target cellular events occurring at a late stage of megakaryocytopoiesis might also fit this model (Takahashi et al, 1999).

Why do TPO receptor agonists work in the face of platelet autoantibodies?

If pathogenic platelet autoantibodies cause accelerated platelet destruction in ITP, how do platelet growth factors (TPO receptor agonists) effectively cause a platelet count rise? One would expect that newly produced platelets would be rapidly cleared by the autoantibody just as they are after typical platelet transfusions in patients with ITP (Buchanan et al, 1977). One possible explanation is mass effect whereby sufficient c-Mpl stimulation may increase the rate of thrombopoiesis enough to overcome the autoantibody. Support for this hypothesis derives from the dose-dependent response observed with TPO receptor agonists and from observations that very high doses of transfused platelets can temporarily increase the platelet count in ITP (Salama et al, 2008).

Rational approach to treatment of chronic ITP

The definitions of chronic and refractory ITP have recently been revised (Rodeghiero et al, 2008). Chronic ITP refers to patients who have had the illness for 12 months or more, and refractory ITP refers to patients with severe symptoms who have failed previous therapies including splenectomy (the definition of refractory may be different for children). A variety of treatment options are available for adults with chronic, refractory ITP; however few randomized trials have compared these head to head. Thus treatment recommendations in ITP tend to be based on expert opinion. A recent consensus report (Provan et al, 2010) differs from previous guidelines (George et al, 1996; British Committee for Standards in Haematology Taskforce, 2003) in that it highlights how TPO receptor agonists and rituximab could be positioned, and emphasizes the need to consider patient preference in treatment decision-making. Cost is another important consideration for treatment choices in ITP, especially at a policy level (Xie et al, 2009); however, a full discussion on cost-effectiveness of ITP treatments is outside the scope of this review.

Initial Treatment of ITP

The prognosis for patients with ITP is generally good, with much of the morbidity due to complications of treatment rather than the illness itself (Portielje et al, 2001). Corticosteroid-based treatments are the accepted first-line therapy with most patients achieving a short term response (Provan et al, 2010). Corticosteroids work by reducing the phagocytic capacity of the reticuloendothelial system and interrupting autoantibody production. Several longitudinal studies in adults report 20% (Stasi et al, 1995; Andres et al, 2003) to 47% (Leung et al, 2001) 6-month remission rates (platelet count > 100 × 10⁹/l) following corticosteroids, but with longer follow-up the risk of relapse is high and responses are generally not sustained (DiFino et al, 1980; Dan et al, 1992). High doses of corticosteroids, usually dexamathasone 40 mg/d for 4 d every 2–4 weeks for up to 6 cycles results in rapid platelet count responses and may be associated with prolonged remissions when given as first-line therapy (Cheng et al, 2003; Mazzucconi et al, 2007).

For patients failing initial treatment, splenectomy is associated with the most durable remissions. In a systematic review 456 (64%) of 707 patients had a complete response with a median follow-up of 7·25 years (range, 5–12·75 years) (Kojouri et al, 2004). Peri-operative thrombosis, infection and bleeding and post-splenectomy infection are recognized complications of the procedure. In addition, vascular complications including atherothrombosis, venous thrombosis and pulmonary arterial hypertention can occur as a result of complex mechanisms (Crary & Buchanan, 2009). A tendency towards splenectomy avoidance (Rodeghiero & Ruggeri, 2008) has recently been observed, resulting in its use later in the disease course. Furthermore, the uptake of splenectomy is largely driven by patient (and physician) preference and not by evidence from comparative trials. Rituximab may also induce lasting remissions, but these tend not to be as durable as splenectomy (Godeau et al, 2008).

Treatment of chronic refractory ITP

Patients with refractory ITP post-splenectomy have the highest risk of death and disease-related or treatment-related complications (McMillan & Durette, 2004). Treatments are aimed at preventing serious bleeding, improving quality of life (QoL) and achieving a safe (but not necessarily normal) platelet count. The choice of therapy depends on the goal of treatment (Table I) which may be to achieve a rapid rise in platelet counts; to maintain a stable, haemostatic platelet count; or to induce remission. For patients with chronic refractory ITP, remission induction is rarely achieved, although providers should keep in mind that spontaneous remissions may occur. The following section focuses on management of patients with refractory ITP because this patient group is most controversial. We describe a goal-directed approach to treatment.

Table I.

Goal-directed therapy for patients with chronic refractory immune thrombocytopenia.

Goal of therapy	Clinical indication	Treatment options
To achieve an immediate rise (24–48 h) in platelet count	Treatment of serious bleeding, in preparation for emergency surgery	IVIg, high dose corticosteroids, high dose platelet transfusions.
To achieve a rapid rise (1–2 weeks) in platelet count	In advance of planned surgical operations or invasive procedures	TPO receptor agonists, vincristine, high dose corticosteroids.
To induce remission	Patients at a relatively early stage of disease	Splenectomy, rituximab.
To maintain a stable, haemostatic platelet count	Refractory patients with frequent bleeding complications, or reduced quality of life	TPO receptor agonists, combination immunosuppressant drugs.

IVIg, intravenous immunoglobulin; TPO, thrombopoietin.

Goal: To achieve a rapid rise in platelet count

Transient elevations in platelet counts are often necessary as emergency treatment in preparation for surgical procedures or to treat acute bleeding episodes. High dose corticosteroids, IVIg, TPO receptor agonists or platelet transfusions are useful for this purpose. In patients who have failed previous ITP treatments, including splenectomy, high dose dexamethasone may be useful to achieve rapid (within weeks) platelet count rises but responses are generally not durable (Gutierrez-Espindola et al, 2003) and side effects may be limiting (Warner et al, 1997). In an observational study of 10 patients who had failed at least two prior ITP treatments, including six who had failed splenectomy, treatment with high dose dexamethasone (40 mg/d for 4 d, given every 4 weeks for 6 cycles) achieved a rapid platelet count increase that was maintained for up to 6 months in most patients (Andersen, 1994). Other investigators have observed transient (Stasi et al, 2000) or no responses (Warner et al, 1997) with this treatment. IVIg administered at a dose of 1–2 g/kg causes a rapid transient increases in platelet counts (within 1–2 d) in over 80% of patients (Godeau et al, 1999); however platelet counts generally return to pretreatment levels within 4 weeks. Rhesus immune globulin (RhIg/anti-D) is felt to be less effective for patients post-splenectomy (Bussel et al, 1991). Repeated doses of IVIg at regular intervals may be useful as maintenance therapy for patients who require ongoing treatment because of bleeding. Treatment with vinca alkaloids, such as vincristine, can also be used to produce a rapid (within 1–3 weeks) yet temporary rise in platelet count (Sikorska et al, 2004). Platelet transfusions, especially in high doses, may also be used if rapid platelet count increments are needed to treat bleeding (Salama et al, 2008). Short courses of TPO receptor agonists may be effective in increasing the platelet count within 1–2 weeks in preparation for surgery.

Goal: To induce remission

Remission induction is generally not a realistic goal of therapy for patients with refractory ITP; however spontaneous remissions can occur. Durable remissions have been achieved with splenectomy and, to a lesser extent, with rituximab when administered to patients at an early stage of their disease (Godeau et al, 2008; Zaja et al, 2010). The main safety concern of rituximab is its effect on immune function. Bacterial infections may be marginally increased (Arnold et al, 2007) and reactivation of the hepatitis B virus, cytomegalovirus, varicella zoster virus, and others are recognized complications (Coiffier, 2006; Aksoy et al, 2007). Rituximab has also been linked to a rare life-threatening opportunistic infection called progressive multifocal leucoencephalopathy (PML) (Carson et al, 2009), which results from reactivation of latent JC virus and has a high case-fatality.

High-dose cyclophosphamide with autologous lymphocyte-depleted peripheral blood stem cell transplantation has been investigated as a means of inducting remission. In an uncontrolled series of 14 patients with refractory ITP, six patients achieved a complete response lasting for 9–42 months; however toxicities were substantial (Huhn et al, 2003).

Goal: To maintain a stable, haemostatic platelet count

TPO receptor agonists (romiplostim and eltrombopag), immunosuppressant drugs or combination therapy using a variety of agents are reasonable options to maintain a haemostatic platelet count for refractory patients. Romiplostim is a peptibody (Fc-linked peptide) that binds to c-Mpl at the same site as endogenous TPO. It is administered by weekly subcutaneous injection (1–10 μg/kg). Eltrombopag is a small non-peptide molecule that binds to the transmembrane portion of c-Mpl and is administered as on oral daily pill (25–75 mg/d). In a randomized placebo-controlled trial of 125 patients, romiplostim achieved a platelet count of 50 × 10⁹/l or greater lasting at least six of the last 8 weeks of treatment in 38% of splenectomized patients compared with 0% with placebo (P = 0·0013) (Kuter et al, 2008). Higher responses were observed in non-splenectomized patients. Similarly, in a randomized placebo-controlled trial of 114 patients with chronic ITP (49% splenectomized), eltrombopag induced a platelet count response (above 50 × 10⁹/l) in 59% compared with 16% on placebo (P < 0·0001) (Bussel et al, 2009a). With both agents, responses were maintained as long as the drug was continued. Collectively, these data suggest that TPO receptor agonists can maintain platelet increases in a significant proportion of refractory ITP patients while on therapy.

TPO receptor agonists have rarely been associated with an increase in bone marrow reticulin in patients with ITP (Bussel et al, 2009b; Dmytrijuk et al, 2009). In a retrospective study of 271 patients treated with romiplostim, 11 patients had a bone marrow examination for a variety of reasons and of those, 10 demonstrated some degree of reticulin staining (Kuter et al, 2009). In a small prospective study of six patients with bone marrow examinations performed before and after romiplostim, one patient showed a 1-grade increase in reticulin that was still within the normal range (Kuter et al, 2009). Reticulin staining resolved once the drug was discontinued. Thromboembolic events have been rarely reported in patients treated with either romiplostim or eltrombopag, which did not correlate with a platelet count rise. Hepatotoxicity has been observed with eltrombopag (Dmytrijuk et al, 2009).

Combination immunosuppressant and/or cytoreductive therapy may be another effective means of maintaining adequate platelet counts in patients with refractory ITP. In a retrospective study of 19 refractory ITP patients, combined use of azathioprine, mycophenolate and cyclosporine was well tolerated and resulted in a platelet count response in 73·7% of patients (Arnold et al, 2010). The combination of cyclophosphamide, vincristine, procarbazine, vincristine, etoposide and prednisone was associated with a platelet count response in six of 12 patients with severe refractory ITP (Figueroa et al, 1993) of whom four maintained remission for 60–150 months (McMillan, 2001). In a prospective study, the combination of IVIg, corticosteroids, vincristine and anti-D was associated with remission in 25 of 35 (71%) patients with chronic ITP, and the combination of azathioprine and danazol was able to maintain remission in 13 of 17 (76·5%) patients (19 of 35 were splenectomized) (Boruchov et al, 2007). Danazol and dapsone are other maintenance treatment options; however, the likelihood of success is low with either drug used as monotherapy.

Adverse events with combination immunosuppressant use included transient and mild leucopenia, mild to severe infection, and gum hypertrophy and tremors associated with cyclosporine. Reports of side effects from the use of combination chemotherapy in patients with refractory ITP included neutropenia, nausea, alopecia and malaise.

Treating what matters

Renewed interest in ITP has led to an acceleration of research, including several multinational trials, consensus documents and ongoing drug development. However, such enthusiasm has the potential to lead to overtreatment unless practice is guided by clinical trials designed to address what really matters to patients. Platelet count, bleeding and QoL are common endpoints in ITP clinical trials; the merits and shortcomings of each are discussed below and summarized in Table II.

Table II.

Outcomes in immune thrombocytopenia (ITP) clinical trials.

Outcome	Advantages	Disadvantages
Platelet count	Efficiency	Surrogate outcome whose correlation with bleeding or other clinical outcomes has not been well defined.
Serious bleeding	Clinically important. Non-serious bleeding (oral purpura) may be a useful surrogate	Rare event, methodologically difficult to measure, lack of validated bleeding measurement tool for ITP.
Quality of life	Clinically important, validated measurement tools for adults and children available	Cumbersome, ambiguous definition of minimal clinically important difference.

Platelet count

Most trials in ITP have used platelet count thresholds as their primary endpoint (Kuter et al, 2008; Bussel et al, 2009a; Zaja et al, 2010). While efficient and responsive, the platelet count is a surrogate endpoint for serious bleeding. The use of a surrogate in any clinical trial may be appropriate if it correlates with, predicts, and fully captures the net effect of the outcome of true importance (Arnold & Lim, 2008). A low platelet count correlates with serious bleeding (Lacey & Penner, 1977); however, the nature of this relationship has not yet been fully established and may be modifiable by certain risk factors including age (Cohen et al, 2000). Bleeding is rare when the platelet count is 30 × 10⁹/l or higher and even with very low platelet counts, bleeding symptoms may be minimal in ITP due to the production of young haemostatically effective platelets. Overall, platelet count does not consistently predict bleeding nor does it capture its net effect (Neunert et al, 2009), and thus may not be a valid surrogate marker for bleeding. Nevertheless, targeting a safe platelet count is a reasonable treatment goal for patients with chronic ITP but should be balanced against toxicities of treatment and the need for frequent clinic visits (Neunert et al, 2009).

Bleeding

Severe bleeding is a rare event in patients with ITP. The estimated risk of intracranial haemorrhage was 1% among adults with refractory ITP in one retrospective study (Portielje et al, 2001) and may be even lower in children (Psaila et al, 2009). Thus, a clinical trial powered on severe (intracerebral) bleeding would probably not be feasible due to the large sample size and rare patient population. Other bleeding events, such as haemorrhagic oral purpura or extensive petechiae may be more feasible outcomes to measure, but only valid if they correlated well with more severe events. The measurement of bleeding itself poses another methodological challenge and requires prospective assessments by raters (Heddle et al, 2003) or patients (Webert et al, 2006). Tools to measure bleeding require validation in ITP populations, and meaningful bleeding outcome events need to be carefully defined.

Quality of life

QoL is a clinically important endpoint in ITP clinical trials (Mathias et al, 2007a). General (McMillan et al, 2008) and disease-specific tools for adults (Mathias et al, 2007b) and children (von Mackensen et al, 2006; Klaassen et al, 2007) have been shown to be reliable and responsive (Mathias et al, 2007a; Klaassen & Young, 2010). QoL measurements have been correlated with platelet count responses (Mathias et al, 2007a) and capture patients’ feelings of anxiety; fatigue; the worry of bleeding; and the inconvenience of frequent therapies. As an outcome in clinical trials, QoL measurements are time consuming and require careful justification of minimal clinically important differences (Klaassen & Young, 2010).

Conclusion

The mechanism of ITP involves insufficient platelet production and autoantibody-mediated platelet destruction. The success of the TPO receptor agonists suggests a fundamental concept, that increased platelet production by megakaryocytes might overwhelm autoantibody-mediated platelet destruction. Still, some key questions about ITP pathogenesis remain unresolved, including how to reconcile autoantibody-mediated platelet destruction and insufficient platelet production. Management of patients with chronic refractory ITP is challenging and goals of therapy, whether short term or long term, should guide treatment decisions. The advent of new drugs, especially the TPO receptor agonists, has advanced our understanding of ITP pathogenesis but has also led to the potential for overtreatment in the absence of clinical trials powered on clinically important endpoints.