Contemporary management of primary immune thrombocytopenia in adults


Adam Cuker, Hospital of the University of Pennsylvania, 3 Dulles, 3400 Spruce Street, Philadelphia, PA 19104, USA.
Tel.: +1 215 615 6555; fax: +1 215 615 6599.


Summary.  Immune thrombocytopenia (ITP) comprises a syndrome of diverse disorders that have in common immune-mediated thrombocytopenia, but that differ with respect to pathogenesis, natural history and response to therapy. ITP may occur in the absence of an evident predisposing etiology (primary ITP) or as a sequela of a growing list of associated conditions (secondary ITP). Primary ITP remains a diagnosis of exclusion and must be differentiated from non-autoimmune etiologies of thrombocytopenia and secondary causes of ITP. The traditional objective of management is to provide a hemostatic platelet count (> 20–30 × 109 L−1 in most cases) while minimizing treatment-related toxicity, although treatment goals should be tailored to the individual patient and clinical setting. Corticosteroids, supplemented with either intravenous immune globulin G or anti-Rh(D) as needed, are used as upfront therapy to stop bleeding and raise the platelet count acutely in patients with newly diagnosed or newly relapsed disease. Although most adults with primary ITP respond to first-line therapy, the majority relapse after treatment is tapered and require a second-line approach to maintain a hemostatic platelet count. Standard second-line options include splenectomy, rituximab and the thrombopoietin receptor agonists, romiplostim and eltrombopag. Studies that directly compare the efficacy, safety and cost-effectiveness of these approaches are lacking. In the absence of such data, we do not favor a single second-line approach for all patients. Rather, we consider the pros and cons of each option with our patients and engage them in the decision-making process.


Immune thrombocytopenia (ITP) comprises a heterogeneous group of disorders characterized by autoimmune-mediated platelet destruction and impairment of platelet production and a variable bleeding tendency. ITP may occur in the absence of an apparent predisposing etiology (primary ITP) or as a sequela of an associated condition (secondary ITP) [1]. The objective of this article is to review the diagnosis and treatment of primary ITP in adults. The dearth of high-quality evidence on this topic has been noted [2]. We cite existing evidence to support our recommendations where available. Where evidence is lacking, we base our suggestions on opinion and clinical experience and endeavor to describe the range of existing expert opinion, pointing out where it departs from our own practice. We conclude with a summary of some of the remarkable advances in ITP over the last 30 years and the unanswered questions of today that require investigation. The pathogenesis, natural history, diagnosis, and management of childhood and secondary ITP are distinct from primary adult ITP and are reviewed elsewhere [1–4].


Administrative database studies suggest an incidence of ITP of 1.6–3.9 per 100 000 person-years [5,6]. Because ITP assumes a chronic course in most adults, prevalence estimates are somewhat higher (9.5–23.6 per 100 000 persons) [6,7]. Adult-onset ITP was once thought to be largely a disease of young women. It is now appreciated that the incidence distribution is bimodal with a smaller peak during young adulthood and a larger peak in the elderly. Among young adults, ITP is approximately twice as common in females as in males. In contrast, incident ITP in persons > 65 years of age affects both genders equally [5,6].


ITP may be primary or secondary. The etiology of primary ITP is unknown. Predisposing polymorphisms in various cytokines and Fcγ-receptors have been reported [8,9]. A reduction in T-regulatory cell function, a skewed ratio of type 1 to type 2 T-helper cells and an increase in B-cell activating factor have been described in active disease and may predispose to loss of tolerance and production of high-affinity anti-platelet antibodies [10–12]. Mechanisms of autoantibody production may differ in secondary ITP. Causes of secondary ITP include broader autoimmune disorders, infections, vaccinations, lymphoproliferative disorders, congenital immune deficiencies and drugs (Table 1). We are able to identify a secondary cause in approximately 20% of our patients [4]. This proportion may be greater in parts of the world where predisposing infections such as Helicobacter pylori are endemic, and is likely to grow as new etiologies are elucidated [13].

Table 1.   Causes of secondary immune thrombocytopenia (ITP)
Broader autoimmune disordersSystemic lupus erythematosus
Antiphospholipid syndrome
Evans syndrome
InfectionsHuman immunodeficiency virus
Hepatitis C virus
Helicobacter pylori
VaccinationsMeasles, Mumps and Rubella
Lymphoproliferative disordersChronic lymphocytic leukemia
Congenital immune deficienciesCommon variable immune deficiency
Autoimmune lymphoproliferative syndrome
Purine analogues


It has long been recognized that the platelet lifespan is reduced in ITP as a consequence of antibody-mediated clearance of platelets by tissue macrophages [14]. Several lines of evidence suggest that platelet production is also impaired in ITP. In contrast to hypoplastic thrombocytopenic disorders, thrombopoietin levels are normal or only minimally elevated in patients with ITP [15]. Platelet kinetic studies demonstrate a corresponding attenuation in thrombopoietic response [16]. Ultrastructural studies show evidence of megakaryocyte injury and apoptosis in patients with ITP [17]. Addition of ITP plasma to healthy megakaryocytes recapitulates these changes and inhibits megakaryopoiesis in vitro [18,19]. Antiplatelet antibodies are not detected in all individuals with ITP and some patients do not respond to pharmacologic or surgical inhibition of antibody-mediated platelet clearance or B-cell suppression, suggesting the possible involvement of other pathogenic mechanisms such as antigen shedding or T-cell-mediated platelet destruction or marrow suppression in a subset of patients [20,21].

Clinical presentation


The most important clinical manifestation of ITP is an increased risk of bleeding. As with other disorders of primary hemostasis, bleeding is primarily mucocutaneous. Petechiae, ecchymoses, epistaxis, gingival bleeding and menorrhagia are common. Major gastrointestinal hemorrhage and hematuria are less frequent. Intracranial hemorrhage (ICH), the most dreaded complication of ITP, is rare. Compared with the general population, the relative risk of ICH in adults with chronic primary ITP was 3.2 (95% confidence interval [CI]: 1.2–9.0) in a recent Danish case–control study [22]. Standardized bleeding assessment scales have been developed, but remain to be validated in large prospective studies [23–25].

The platelet count is a fairly crude but widely used predictor of bleeding. In the absence of a hemostatic comorbidity, trauma, surgery, or concomitant anti-thrombotic therapy, major bleeding is rare when the platelet count exceeds 20–30 × 109 L–1 [26]. Spontaneous ICH occurs almost exclusively in patients with platelet counts < 10–20 × 109 L–1 [27]. Nevertheless, there is substantial interindividual variation in bleeding phenotype. Some severely thrombocytopenic patients (< 10 × 109 L–1) may be asymptomatic, whereas others with milder thrombocytopenia may suffer from frequent bleeding symptoms. Bleeding that is disproportionate to the degree of thrombocytopenia should prompt an investigation for other hemostatic disorders.

Advanced age and prior bleeding are established independent risk factors for hemorrhage in ITP. In a retrospective cohort study and pooled analysis of a case series of patients with ITP and platelet counts < 30 × 109 L–1, the risk of fatal hemorrhage and major non-fatal bleeding was approximately 30-fold greater in individuals older than 60 than in those < 40 years of age [28,29]. Other predictors that underlie the interindividual variation in bleeding tendency, including measurements of platelet function [30,31], are under investigation and may improve individualization of therapy.

Non-hemorrhagic manifestations

Patients with ITP commonly suffer from disabling fatigue, a fear of bleeding, withdrawal from professional and social activities and poor quality of life (QOL) [32]. In a systematic QOL assessment using a standardized scale for measuring functional health, adults with chronic ITP endorsed a health-related QOL that was significantly lower than the general population and intermediate between patients with cancer and heart failure [33].

Epidemiologic evidence suggests that adults with chronic ITP may also be at mildly increased risk of infection, thromboembolism and hematologic malignancy [22,34–38]. Further studies are needed to confirm these associations and their relationship to therapy. Approximately 5% of patients presenting with primary ITP develop systemic lupus erythematosus or another autoimmune disorder over time [39,40].


A platelet count < 100 × 109 L–1 (rather than 150 × 109 L–1) is recommended as the cut-off for diagnosis [1]. This threshold is based on a prospective cohort of otherwise healthy individuals with a platelet count between 100 and 150 × 109 L–1, only 6.9% of whom developed more severe thrombocytopenia over the ensuing 10 years [41].

ITP remains a diagnosis of exclusion. There is no gold standard. The single most compelling evidence of its presence is a response to ITP-specific therapy. Initial evaluation of a patient with suspected primary ITP involves exclusion of non-autoimmune causes of thrombocytopenia (Table 2) and exclusion of secondary causes of ITP (Table 1). At a minimum, this evaluation must include a careful history, physical examination, complete blood cell count and evaluation of the peripheral blood smear by an expert hematologist or hematopathologist.

Table 2.   Selected non-autoimmune causes of thrombocytopenia
  1. MDS, myelodysplastic syndrome.

Disorders of decreased platelet production
 • Congenital thrombocytopenias
 • Myelosuppressive therapy (e.g. chemotherapy and radiation)
 • Ethanol toxicity
 • Folate or vitamin B12 deficiency
 • Primary bone marrow disorders (e.g. MDS, myelofibrosis, leukemias and lymphomas)
 • Infiltrative diseases of the bone marrow
 • Certain viral infections
Disorders of decreased platelet survival
 • Certain drugs (e.g. heparin and quinine)
 • Alloimmune thrombocytopenias (e.g. post-transfusion purpura)
 • Disseminated intravascular coagulation
 • Thrombotic thrombocytopenic purpura/hemolytic uremic syndrome
 • Cardiopulmonary bypass
 • Severe infection/sepsis
Splenic sequestration
 • Portal hypertension
 • Infiltrative diseases of the spleen
Dilutional thrombocytopenia

Exclusion of non-autoimmune causes of thrombocytopenia

The archetypal presentation of primary ITP is a generally healthy individual with isolated thrombocytopenia, an otherwise unremarkable blood smear and a physical examination notable only for evidence of bleeding commensurate with the platelet count. Any deviation from this paradigm should prompt a focused investigation for other causes of thrombocytopenia (Table 2).

Congenital thrombocytopenias, in particular, must be borne in mind in the evaluation of isolated thrombocytopenia. Documentation of a previous normal platelet count, a focused family history and, when necessary, platelet counts of family members, should be sought to exclude this group of disorders. A meticulous history of recent exposure to prescription, over-the-counter and recreational drugs as well as herbal supplements should be performed to rule out drug-induced thrombocytopenia.

Bone marrow aspiration and biopsy is often unnecessary in individuals presenting in typical fashion, but should be performed in patients that do not respond to standard first-line therapy. Bone marrow examination may also be considered prior to splenectomy or initiation of a thrombopoietin receptor agonist (TRA) or intense immunosuppression, if not previously performed. Some experts recommend a bone marrow biopsy in individuals over the age of 60 years when the incidence of myelodysplastic syndrome becomes significant [42], although this is not our practice unless atypical features are present.

Assays for platelet antigen-specific antibodies lack sufficient sensitivity (49–66%) to exclude ITP and are hampered by poor reproducibility [43,44]. Although the presence of such antibodies is suggestive of ITP, antibodies are also detected in 10–20% of individuals with non-immune causes of thrombocytopenia such as myelodysplastic syndrome and chronic liver disease [43]. We seldom request platelet antibody testing in our practice.

Exclusion of secondary ITP

After non-autoimmune causes of thrombocytopenia have been excluded, a thorough history and physical examination should be conducted to assess the likelihood of a predisposing infection, malignancy, autoimmune disorder, congenital immune deficiency, vaccine or drug exposure (Table 1). Distinguishing primary ITP from secondary causes may have critical implications with respect to prognosis and treatment. For instance, in accordance with recent guidelines [2] we request HIV and hepatitis C testing in all patients with ITP because of the frequency of these conditions in the adult population, the high incidence of ITP remission with antiviral therapy and concerns regarding the effects of corticosteroids in hepatitis C.

In some parts of the world where H. pylori infection is endemic such as Japan and Italy, microbial eradication produces platelet responses in more than half of infected individuals with ITP. In these regions, it may be advisable to test routinely for the organism or to offer empiric eradication therapy to all patients with newly diagnosed ITP. In the US, H. pylori infection is less common and antimicrobial therapy seldom induces sustained platelet responses [45]. It is therefore our practice to test only individuals with gastrointestinal symptoms as well as those from parts of the world where H. pylori infection is endemic. Some experts advocate testing in all ITP patients, irrespective of locale [42].

Quantitative measurement of immunoglobulins to screen for congenital immune deficiencies in patients with ITP is recommended by an international working group [42]. The pick-up rate and cost-effectiveness of this approach has not been studied. We do not perform such testing unless there is a history of frequent infection or other manifestations to suggest an underlying immune deficiency state. Similarly, we do not routinely request antinuclear or antiphospholipid antibody testing unless features suggestive of systemic lupus erythematosus or antiphospholipid syndrome are present.

Natural history

ITP may be classified according to disease duration as newly diagnosed (< 3 months), persistent (3–12 months) or chronic (> 12 months) [1]. Primary ITP in adults assumes a chronic course in most cases. Approximately 80–90% of individuals respond to standard first-line therapy [46], but the majority relapse and require additional treatment [47]. The likelihood of achieving a sustained remission lessens as disease duration increases, although rare individuals may show spontaneous improvement, even after years of severe disease [48].


The traditional goal of management is to provide a hemostatic platelet count (> 20–30 × 109 L−1 for most patients and clinical scenarios) while minimizing treatment-related toxicity. Therapy is generally not indicated for patients with platelet counts above this threshold.

Treatment should be tailored to the individual patient and clinical setting. Individuals with extensive bleeding symptoms, comorbidities or lifestyles that predispose to bleeding and those that require concomitant antithrombotic therapy or an invasive procedure may benefit from a higher platelet count. Conversely, a lower target platelet count may be acceptable in patients who bleed little and tolerate treatment poorly.

First-line therapy

Corticosteroids, supplemented with either intravenous immune globulin G (IVIG) or anti-Rh(D) as needed, are used as upfront therapy to stop bleeding and raise the platelet count acutely in patients with newly diagnosed or newly relapsed disease. Dosing, time to response and selected toxicities of these agents are shown in Table 3. Most (80–90%) patients evince a partial or complete response to first-line therapy [46]. A non-response should prompt reconsideration of the diagnosis and an investigation for alternative causes of thrombocytopenia.

Table 3.   First-line treatment options for immune thrombocytopenia (ITP)
AgentTypical dosingTime to responseSelected toxicities
  1. IVIG, intravenous immune globulin G.

Prednis(ol)one0.5–2 mg kg−1 day−1 × 2–4 weeks followed by slow taperSeveral days to several weeksMood swings, insomnia, anxiety, psychosis, weight gain, Cushingoid facies, hyperglycemia, decreased bone density, hypertension, skin changes, gastrointestinal distress and ulceration, avascular necrosis, increased susceptibility to infections, cataracts, adrenal insufficiency
Methylprednisolone30 mg kg−1 day−1 × 7 days2–7 days
Dexamethasone40 mg day−1 for 4 days every 2–4 weeks for 1–4 cyclesSeveral days to several weeks
IVIG0.4 g kg−1 day−1 × 5 days
1 g kg−1 day−1 × 1–2 days
1–4 daysHeadache, aseptic meningitis, renal insufficiency, fever, chills, nausea, thromboembolism, anaphylactoid reactions in patients with IgA-deficiency
Anti-Rh(D)50–75 μ kg−11–5 daysHemolytic anemia, fever, chills. Rarely, intravascular hemolysis, DIC, and renal failure

Several uncontrolled studies suggest that high-dose dexamethasone for one to four cycles as initial therapy increases response rates and prolongs remission without additional toxicity [49,50]. However, a recent randomized controlled trial of a single cycle of high-dose dexamethasone vs. standard-dose prednisolone did not corroborate these findings [51]. Zaja and colleagues randomized patients to receive one cycle of high-dose dexamethasone, alone or in combination with rituximab, as initial therapy. Patients in the rituximab arm showed a higher response rate at 6 months (63% vs. 36%, P = 0.004), but this difference eroded over extended follow-up and grade 3 and 4 toxicities were more frequent in patients treated with combination therapy [52]. In a recent pilot randomized controlled trial of non-splenectomized patients with newly diagnosed or newly relapsed ITP, the addition of rituximab to standard therapy did not improve outcomes at 6 months [53]. Further controlled trials with long-term follow-up are needed to determine whether aggressive therapy at the onset of disease can ameliorate the natural history of ITP.

Anti-Rh(D) is indicated in Rh(D)-positive non-splenectomized patients. It is rarely associated with life-threatening intravascular hemolysis, disseminated intravascular coagulation (DIC) and acute renal failure, and should be avoided in patients with pre-existing hemolysis. Some experts recommend avoidance of anti-Rh(D) in all patients with a positive direct antiglobulin test not as a result of previous therapy [54].

Second-line therapy

Although a majority of patients respond to first-line therapy, most ultimately relapse after treatment is tapered and require a second-line approach to maintain a hemostatic platelet count [47]. Standard second-line options include a splenectomy, rituximab and the TRAs, romiplostim and eltrombopag (Table 4).

Table 4.   Second-line treatment options for immune thrombocytopenia (ITP)
ApproachTypical dosingResponse rateTime to responseSelected toxicities
SplenectomyN/ATwo-thirds of patients achieve long-term remission0–24 daysAdverse effects of surgery and anesthesia, increased risk of infection, long-term vascular complications
Rituximab375 mg m−2 weekly × 4 weeks (lower doses may be effective)40% at 1 year; 20–25% at 5 years1–8 weeksInfusion reactions, reactivation of hepatitis B infection, rare cases of progressive multifocal leukoencephalopathy
Eltrombopag12.5–75 mg PO daily> 80%. Most responses are sustained for up to 3–5 years with continual administration1–4 weeksIncreased bone marrow reticulin, rebound thrombocytopenia, thrombosis. Eltrombopag also associated with liver function test abnormalities
Romiplostim1–10 μg kg−1 SC weekly

A splenectomy offers the best chance for prolonged remission. Two-thirds of patients attain a durable long-term remission and another 10–20% attain a partial response that may permit a reduction in the use of concomitant ITP medications [55]. Distant relapse five or more years after splenectomy has been reported in a minority of responders [56]. Response rates are lower in older patients [55]. Apart from age, there are no reliable predictors of a response in individual patients with the possible exception of 111In-labeled autologous platelet scanning, which is neither widely available nor well standardized [57]. Open and laparoscopic splenectomy are equally effective and operative risk is low in experienced hands [55]. The relative risk of post-splenectomy sepsis is 1.4 (95% CI: 1.0–2.0) in the first year after surgery and remains elevated throughout life [58]. This risk is reduced by adhering to recommended vaccination protocols and initiating antibiotics at the first sign of a febrile illness [59]. All patients with newly diagnosed ITP should be vaccinated in anticipation of the possible need for splenectomy in the future. Recent concerns about vascular complications after splenectomy including thrombosis, atherosclerosis and pulmonary hypertension require further study [60]. The risks of splenectomy depend, in part, on the nature of the underlying disorder. A greater incidence of post-splenectomy infection and thrombosis has been observed in patients with congenital and acquired hemolytic anemia than in individuals with ITP [61].

Rituximab, an anti-CD20 monoclonal antibody, induces complete remissions at standard doses (375 mg m−2 weekly × 4 weeks) in 40% of patients at 1 year [62]. Approximately half of such individuals enjoy sustained responses lasting five or more years [63]. A small uncontrolled study of low-dose rituximab (100 mg weekly × 4 weeks) showed similar efficacy [64]. An initial rise in platelet count often occurs within 1–2 weeks of the first infusion, suggesting an effect on platelet clearance, but more durable responses may not be observed for several weeks to months [62]. Individuals that achieve a complete remission at 1 year and subsequently relapse often respond to retreatment. Partial responders, in contrast, often relapse within 1 year and generally do not achieve sustained responses with retreatment [65]. Rituximab is contraindicated in patients with active hepatitis B infection owing to the risk of fulminant hepatitis. More than 50 cases of progressive multifocal leukoencephalopathy have been reported in HIV-negative patients treated with rituximab, one of whom had ITP [66]. The contribution of rituximab to the acquisition of progressive multifocal leukoencephalopathy in these heavily pretreated patients is unclear.

The TRAs, romiplostim and eltrombopag, are approved by the US Food and Drug Administration for use in patients with primary ITP who require treatment after an initial course of corticosteroids. In some jurisdictions (e.g. European Union), approval is restricted to patients who have previously undergone a splenectomy or have a contraindication to surgery. Romiplostim is composed of an Fc fragment fused to four identical peptides that bind to the thrombopoietin receptor [67]. It is delivered as a weekly subcutaneous injection. Current labeling requires administration in a physician’s office, although the feasibility of home injection has been demonstrated in selected patients [68]. Eltrombopag, a small molecule that binds to the transmembrane domain of the thrombopoietin receptor at a distance from the thrombopoietin binding site [69], is formulated for daily oral administration. It must be taken several hours removed from meals and medicinal products containing polyvalent cations, which interfere with drug absorption. An initial 50% dose reduction is necessary in individuals of East Asian ancestry.

Both approved TRAs have demonstrated high response rates in randomized clinical trials, even among splenectomized and refractory patients. Sixty-three splenectomized and 62 non-splenectomized patients were randomized to receive romiplostim or placebo for 6 months. The primary endpoint, a platelet count of 50 × 109 L−1 or greater for at least 6 of the last 8 weeks of the study in the absence of rescue therapy, was achieved in 61% of non-splenectomized and 38% of splenectomized subjects receiving romiplostim and in only 1 of 42 placebo-treated patients [70]. An open-label trial tested the utility of romiplostim as a means of delaying or avoiding a splenectomy. Non-splenectomized individuals (n = 234) were randomly allocated to receive romiplostim or standard of care for 12 months. The incidence of treatment failure (11% vs. 30%, P < 0.001) and splenectomy (9% vs. 36%, P < 0.001) were significantly lower in patients receiving romiplostim [71]. Similar efficacy has been observed in trials of eltrombopag. In a 6-month phase III study, 197 subjects were randomized to eltrombopag or placebo. The primary endpoint, a platelet count between 50 and 400 × 109 L−1 at one or more time points during the study, was achieved in 79% and 28% of eltrombopag- and placebo-treated subjects, respectively (P < 0.0001) [72]. Sustained responses of up to 5 years have been reported in open-label extension studies of both romiplostim and eltrombopag [73,74]. Patients treated during study with both drugs experienced less need for corticosteroids and other rescue medications, improved QOL and less overall bleeding as compared with placebo-treated subjects [71–76]. However, a recent meta-analysis failed to show a reduction in major (WHO grade III and IV) bleeding with these agents [77].

Romiplostim and eltrombopag are generally well tolerated [78]. Increased bone marrow reticulin has been observed in < 5% of patients receiving treatment for at least 12 months. In most cases, reticulin deposition was not associated with a loss of response or a myelopthistic picture and was reversible with discontinuation of drug [79,80]. Although not our practice, some clinicians perform bone marrow biopsies on all patients before initiating treatment with a TRA to establish a baseline. Bone marrow examination should be performed if there is loss of response to treatment or new abnormalities appear in the peripheral blood cell counts or smear. It remains unclear whether TRAs increase the risk of thrombosis in patients with ITP [81]. The incidence rate of thromboembolism across studies of eltrombopag and romiplostim was 4.6 and 5.2 events per 100 patient-years [82,83], respectively, figures in line with the reported incidence rate of thrombosis in the general ITP population [35]. Both venous and arterial events were observed, most occurring at normal or subnormal platelet counts [82,83]. Almost all patients who developed thrombosis during a study had pre-existing thrombotic risk factors. Patients at highest risk (e.g. those with a history of venous thromboembolism or known atherosclerosis) were excluded from some TRA trials. We favor other second-line treatment options in such patients. Rebound thrombocytopenia to levels below those at the onset of treatment has been observed in approximately 10% of patients who discontinued either romiplostim or eltrombopag in clinical trials [81]. Tapering of the dose, careful monitoring and/or preemptive introduction of concomitant therapy at the time of TRA cessation may prevent or alleviate this toxicity. Hepatobiliary laboratory abnormalities were detected in 13% of eltrombopag-treated patients in clinical studies, prompting a black box warning that calls for regular monitoring of liver function tests. In most patients, laboratory abnormalities are non-progressive or resolve without clinical sequelae [76]. Eltrombopag stimulated cataract formation in rodents at doses five times the human clinical exposure, but has not been shown to promote cataract formation or progression in patients with ITP [84].

TRAs are expensive. Romiplostim costs approximately $4 per microgram and the average wholesale acquisition cost of a 30-day supply of eltrombopag 25 mg tablets is $1650 (USD) [85]. The annualized cost for a 70-kg patient treated at a weekly dose of 5.9 μ kg−1 (the mean weekly dose in the romiplostim extension study [68]) is approximately $86 000. The yearly cost for a daily dose of eltrombopag 50 mg (the median dose in the 6-month phase III trial [72]) is about $40 000. Neither of these estimates includes the expenses of physician visits and laboratory monitoring associated with TRA use. Appraisals undertaken by the National Institute for Health and Clinical Excellence (NICE) in the United Kingdom noted an absence of evidence for long-term cost-effectiveness of these drugs relative to appropriate comparators such as splenectomy and rituximab [86,87].

The optimal treatment for patients requiring second-line therapy is controversial. Some authorities favor splenectomy in the absence of a contraindication to surgery [2,61]; others do not express a preference among second-line options [42] or note a dearth of evidence necessary for prioritization of such options [61]. Studies that directly compare the efficacy, safety, impact on QOL and cost-effectiveness of second-line options are needed to guide management of patients failing or relapsing after first-line therapy. Until comparative data become available, we remain unconvinced that there is a clear ‘winner’ among splenectomy, rituximab and the TRAs for all patients. Barring a contraindication, we discuss the pros and cons of each option (Table 5) with our patients within the context of the individual’s age, comorbidities, lifestyle, financial considerations, values and preferences. The goals of this conversation are to empower patients to actively participate in the decision-making process and to arrive at an individualized treatment plan with which they feel most comfortable. Because long-term remissions are most likely to occur in the first year after diagnosis [46,88], we generally recommend deferral of splenectomy until a chronic disease course has been established. Increasingly, patients in our practice and elsewhere are opting for medical therapy over splenectomy [89], although the use of TRAs prior to splenectomy remains off-label in some jurisdictions (e.g. European Union).

Table 5.   Major pros and cons of second-line treatment options
SplenectomyBest chance for long-term remission
Extensive long-term safety data available
Need for surgery
Increased risk of infection
Limited ability to predict whether an individual patient will respond
RituximabCompletion of treatment after 4 weeks
Generally well-tolerated
Least chance for long-term response
Longer median time to response than other options
Low but non-negligible risk of serious infection
 receptor agonists
High response rates
Generally well-tolerated
Need for ongoing treatment and monitoring in most patients
Encouraging but limited long-term safety data
More costly than other options
Romiplostim requires weekly SC injection in a physician’s office
Eltrombopag must be taken 1–2 h removed from meals and 4 h removed from foods or supplements containing polyvalent cations

Special populations and clinical settings

Refractory ITP

Refractory disease is defined as ITP requiring platelet-raising therapy after a splenectomy, due either to non-response or relapse [1]. In the past, many patients with refractory ITP required treatment with third-line agents (Table 6), either alone or in combination. These drugs are limited by modest response rates and narrow therapeutic indices. Evidence supporting their use in ITP is generally limited to uncontrolled case series [90,91] and data comparing one regimen to another are lacking. With the availability of rituximab and TRAs, which are associated with response rates of 40–80% in splenectomized and otherwise refractory patients [62,70,72], fewer individuals in our practice now require third-line agents.

Table 6.   Third-line treatment options for ITP
AgentTypical dosingSelected toxicities
  1. G6PD, glucose-6-phosphate dehydrogenase; i.v., intravenous.

Azathioprine1–2 mg kg−1 day−1 (maximum 150 mg day−1)Liver function abnormalities, neutropenia, anemia, infection
Cyclosporine5 mg kg−1 day−1 × 6 days, then 2.5–3 mg kg−1 day−1 (titrated to blood levels of 100–200 ng mL−1)Renal failure, hypertension, tremor, infection
Cyclophosphamide1–2 mg kg−1 PO daily or 0.3–1 g m−2 i.v. every 2–4 weeks × 1–3 dosesMyelosuppression, infection, secondary malignancy
Danazol200 mg 2–4 times per dayAcne, hirsutism, dyslipidemia, amenorrhea, liver function abnormalities
Dapsone75–100 mg dailyHemolytic anemia in patients with G6PD deficiency, rash, methemoglobinemia
Mycophenolate mofetil1000 mg twice dailyHeadache, back pain, infection
Vincristine1–2 mg i.v. weekly (total dose 6 mg)Neuropathy, constipation, cytopenias, thrombophlebitis at the infusion site

Hospitalization and emergency therapy

Patients should be hospitalized for major bleeding. Admission should also be considered for patients with platelet counts < 10–20 × 109 L−1, particularly if there is a history of significant bleeding, non-compliance, or if responsiveness to therapy has not been established or cannot be easily monitored in an outpatient setting. Hospitalized patients are generally treated with standard first-line therapy and discharged when bleeding has ceased and the platelet count rises above 20–30 × 109 L−1. In the pre-TRA era, multiagent induction with methylprednisolone, IVIG, anti-Rh(D) (in Rh-positive, non-splenectomized patients) and vinca alkaloid demonstrated efficacy in hospitalized patients failing first-line therapy [92]. We now generally add a TRA to first-line agents before resorting to the addition of third-line therapies and intense immunosuppression, although this approach has not been systematically investigated.

Platelet transfusion, delivered in combination with medical therapy, may be useful for life- or organ-threatening hemorrhage. Recombinant factor VIIa is reserved for the rare patient unresponsive to other modalities in whom an immediate hemostatic response is necessary. We use tranexamic acid or ε-aminocaproic acid as adjuncts to manage mucosal hemorrhage and hormonal contraceptives to control menstrual bleeding. We also employ general medical measures such as blood pressure control and avoidance of drugs that impair hemostasis to reduce bleeding risk [42].

Invasive procedures and antithrombotic therapy

There is a paucity of evidence to guide the management of ITP in the peri-operative setting. An expert panel recommends platelet counts of ≥ 20–30 × 109 L−1 for dental cleaning, regional dental block and simple tooth extraction; ≥ 50 × 109 L−1 for complex dental extraction and minor surgery; ≥ 80 × 109 L−1 for major surgery; and ≥ 100 × 109 L−1 for neurosurgery [42]. First-line therapies (Table 3) are generally used to increase the platelet count for surgery in responding patients. Antifibrinolytic agents and topical thrombin and fibrin glues are useful adjuncts for prevention of bleeding with dental procedures. The use of TRAs for the peri-operative management of ITP has been proposed, but requires investigation.

There is likewise little evidence to inform management of patients requiring antithrombotic therapy. In accord with others, we generally strive to maintain a platelet count of 50 × 109 L−1 or higher in patients on antiplatelet or anticoagulant drugs [93].


ITP accounts for 3% of thrombocytopenia in pregnant women [94]. ITP arising during pregnancy must be differentiated from usual causes of non-autoimmune thrombocytopenia (Table 1) and thrombocytopenic conditions unique to pregnancy such as pre-eclampsia, eclampsia, HELLP (Hemolysis, Elevated Liver enzymes, Low Platelets) syndrome and gestational thrombocytopenia. It is particularly important to distinguish ITP from the last of these entities, a common condition characterized by mild thrombocytopenia that generally requires no treatment and resolves after delivery [95].

Pregnancy has a variable effect on ITP. Thrombocytopenia may worsen, remain stable or improve during gestation. In the first two trimesters, the goals of therapy are similar to the non-pregnant state: maintenance of a platelet count of 20–30 × 109 L−1 and prevention of bleeding. As delivery approaches, the target platelet count increases to 50 × 109 L−1 in preparation for the potential need for urgent cesarean section [42]. Neuraxial anesthesia may require a platelet count of 75 × 109 L−1 or higher depending on the preference of the anesthetist [96].

First-line treatment options for ITP during pregnancy include corticosteroids and IVIG at standard dosing (Table 3) [2]. Splenectomy is reserved for refractory cases and is most safely performed in the second trimester [42]. TRAs, anti-Rh(D) and rituximab are of uncertain risk in pregnancy (risk category C) and should generally be avoided, although the safe use of the latter two agents during gestation has been reported [97,98].

As a result of placental transfer of maternal anti-platelet IgG, approximately 10% of neonates born to mothers with ITP manifest severe thrombocytopenia (< 50 × 109 L−1). Fortunately, the risk of ICH among thrombocytopenic neonates is rare (< 1%) [99–101]. There is no evidence that this risk is reduced by cesarean delivery. Therefore, the decision to pursue a cesarean section should be based on maternal indications alone. Fetal procedures associated with hemorrhagic risk such as cordocentesis, fetal scalp blood sampling or electrode placement and forceps- or vacuum-assisted delivery should be avoided. Neonatal platelet count monitoring should be initiated shortly after birth under the care of a pediatrician and continued for 1–2 weeks. Some pediatricians advocate transcranial ultrasonography in all neonates with severe thrombocytopenia to exclude silent ICH [42].


Over the past three decades, our understanding of ITP and its management has been turned on its head. A disease once thought to be an affliction of young woman is now known to present with greatest frequency in the elderly. A disease formerly thought to be mediated exclusively by antibody-induced platelet clearance is now known to involve impaired platelet production as well. A disease once termed idiopathic now has a number of recognized causes. And a disease once managed with corticosteroids and splenectomy now has a growing list of effective first- and second-line therapies (Tables 3–4).

These remarkable advances have generated new questions and challenges with respect to management: What factors underlie interindividual variation in bleeding phenotype? Can these factors be used as a complement to the platelet count to estimate bleeding risk and individualize goals of therapy? Does aggressive medical intervention at the onset of disease have the potential to alter its natural history? Is the burden of ITP adequately captured by bleeding and drug toxicity, or should functional health and QOL play a greater role in defining treatment goals? What is the relative cost-effectiveness of available therapies and how should this figure in management decisions? If the same energy, resources and ingenuity that have driven the paradigm shifts of the last 30 years are applied to the questions of today, continued advances for the betterment of patients with ITP will be assured.


A. Cuker is supported by K23HL112903.

Disclosure of Conflict of Interests

A. Cuker has provided consulting services to Bayer, Biogen-Idec, Canyon, CSL Behring, and Genzyme; has received research funding from Baxter, Bayer, and Novo Nordisk; and has provided expert witness testimony relating to ITP. S. Lakshmanan has no conflict of interest to disclose.