Second-line therapies in immune thrombocytopenia

Immune thrombocytopenia (ITP), characterized by isolated thrombocytopenia and a risk for hemorrhage, is a heterogeneous disorder with variable clinical symptoms. Bleeding events are unpredictable with currently available laboratory testing. Many patients with ITP, even in the setting of severe thrombocytopenia, do not exhibit significant bleeding. Although bleeding may be the indication for treatment, often hematologists choose to treat patients with pharmacologic therapies for a variety of other reasons, including implications for health-related quality of life (HRQoL), debilitating fatigue, perisurgical planning, or to induce a remission. For this reason, the goal of the therapy or the efficacy measure of a response may differ among patients.

Historic first-line therapies for ITP include observation, steroids, intravenous immunoglobulin (IVIG), and anti-D globulin. These common approaches are used up front in newly diagnosed patients. Steroids, IVIG, and anti-D globulin may also be used periodically throughout the disease course in patients managed primarily with observation, during times of breakthrough bleeding, surgeries, or specific activities. Second-line treatments, which, for this manuscript, include therapies beyond observation, steroids, IVIG, and anti-D globulin, can induce a sustained increase in the platelet count with ongoing treatment and/or may alter the disease.

Studies comparing treatments in ITP are scarce, and the outcome measures used across studies are inconsistent, despite the establishment of standard guidelines for diagnosis and response criteria.^1,2  Novel types of therapies for ITP continue to expand, and selecting second-line treatments remains a standard, yet challenging, aspect of providing ITP care. Shared decision-making is also critical, given that these agents vary considerably with regard to cost, ease of administration, potential adverse effects, and likelihood of remission, all of which may influence patient preference. By comparing second-line treatments with a focus on important patient-related outcomes (Tables 1 and 2), clinicians and patients can make better-informed decisions based on the indication for treatment.

The platelet count has been the historic surrogate primary outcome of response to ITP treatment, because it is objective and easily compared. If the goal of treatment is to raise the platelet count, the aim is often to obtain a “safe” platelet count in which active bleeding or risk of serious bleeding is thought to be low. The International Working Group has defined ITP treatment response on the basis of platelet count and recommends that the cumulative time with a sustained response be calculated.¹  Despite these guidelines, historic clinical trials and case series report platelet response using a wide range of criteria; thus, directly comparing the effect of individual treatments by platelet count, the most objective and used response measure, is challenging (Table 1).

This inconsistency also highlights the fact that clinicians and investigators have varying ideas of the optimal platelet response to treatment. In clinical practice, for providers and patients, a definition of response may reflect a platelet count that achieves a specific primary goal, such as fewer bleeding symptoms or activity restrictions, even if this is below the response threshold set by the International Working Group. Furthermore, the goal of “complete response” may come with more adverse effects, when a lower platelet count “response” would have similarly met the aim of the treatment. Although consistent platelet criteria are needed in clinical trials to allow for comparisons among treatments, in clinical practice, the success of a treatment by platelet count needs to be considered alongside HRQoL, adverse effects, and bleeding response.

Because the overall effect of ITP and its treatments are not adequately measured by the platelet count, HRQoL measures can be used to account for additional variables that affect daily well-being. The pediatric ITP instrument, the Kids’ ITP Tool, a 26-item questionnaire with a child, proxy, and parent version, was the first ITP-specific HRQoL tool to be developed, validated, and incorporated into clinical trials.³  General HRQoL measures, such as the SF-36 and EQ-5D, were initially used in clinical trials for adults with ITP. In 2007, a 50-item ITP-Patient Assessment Questionnaire (ITP-PAQ) was developed and validated in adults.⁴  The ITP-PAQ captures domains of physical health (ITP-related symptoms, fatigue/sleep, bother-physical health, and activity), mental health (psychological distress and fear), work, social activity, reproductive health, and overall quality of life. HRQoL questionnaires, such as the ITP-PAQ, can be used to describe changes over time in a treatment group, but must be interpreted after derivation and in relationship to the smallest difference a patient would identify as important, or the minimal important difference (MID). Overall, the effect of medications on HRQoL is understudied, and greater emphasis should be given to HRQoL in interventional trials and prospective natural history cohorts.

Fatigue is an important domain of HRQoL, although ITP-associated fatigue is not well understood.⁵  Hypothetical causes of fatigue include the effect of activity restrictions, immune activation and pro-inflammatory state, toxicity of medical therapy, and existing comorbidities. Although fatigue symptoms appear to correlate with severity of thrombocytopenia, the effect of pharmacologic treatments in raising the platelet count, and thereby decreasing ITP-related fatigue, has not been well established.⁶  The Fatigue Impact Scale assesses the effect of fatigue on psychosocial, cognitive, and physical function, and approximately 22% to 39% of surveyed patients with ITP report significant fatigue (Fatigue Impact Scale ≥40). Additional validated measures for ITP-associated fatigue include the physical health domain in the ITP-PAQ and the Functional Assessment of Chronic Illness Therapy-Fatigue questionnaire. When measures of fatigue have been integrated as a secondary outcome measure in clinical trials, the baseline assessment of fatigue in ITP is worse than in the general US population and, in some studies, is the most severely affected of all the HRQoL domains.^7,8  Given its prevalence in ITP, changes in fatigue must be considered in addition to other HRQoL measures in clinical trials of ITP treatments.

Because the platelet count is a poor correlate for bleeding, bleeding scales have been developed to quantitate bleeding symptoms and provide an objective outcome.⁹  One of the first ITP-specific tools, developed by Bolton-Maggs and Moon, uses an ordered categorical scale and measures interference of symptoms with daily living. The Buchanan and Adix scale, with an overall bleeding score of 0 to 5, measured during the last 24 hours, is easy to apply in pediatric clinical practice. The Page scale is more complex, addressing bleeding at 9 anatomical sites during the prior week, both by history and exam. A more detailed and extensive bleeding assessment tool has been proposed for clinical trials; however, the complex nature of this measure, with 11 site-specific assessments across 1 to 3 bleeding grades, is more onerous, and improved precision with the use of this tool has not been demonstrated. Despite the existence of ITP-specific bleeding tools, the World Health Organization bleeding scale, a 5-point scale, has most commonly been used in clinical trials. Although capturing and reporting bleeding events is a highly relevant outcome, it is unlikely that bleeding will ever be the primary outcome for clinical trials. Because of the rarity of severe bleeding events, an unreasonably large number of patients would be required to adequately power a trial aimed at demonstrating a reduction of severe bleeding events.

The thrombopoietin receptor agonists (TPO-RAs) bind to and stimulate the TPO receptor and increase platelet production. Two recombinant TPO-RAs, eltrombopag and romiplostim, are approved by the US Food and Drug Administration for adults with chronic ITP. Eltrombopag is also approved for children older than 1 year who have chronic ITP. These agents are not considered curative based on their mechanism of action. The literature supports that there is no cross-resistance between the 2 agents, and therefore, if 1 agent is not effective, trialing the other agent may be beneficial.¹⁰  Among the second-line treatments, the most robust data exist for these agents with regard to their effect on platelet count, bleeding, fatigue, and HRQoL.

Randomized trials in children and adults have demonstrated an acute platelet response with romiplostim of approximately 79% to 88%, and a durable response with ongoing treatment of 38% to 52% (Table 1).^11,12  Reports of adult and pediatric real-world use are similar to reported clinical trial results, although response patterns are more variable.¹³  A case series demonstrated remission after treatment; however, larger studies are needed to confirm whether the incidence of remission is greater than the expected spontaneous remission rate.¹⁴ 

Using a composite outcome for bleeding, including actual bleeding events or use of rescue medication to prevent bleeding, the effect of a treatment on actual bleeding is less clear. Romiplostim has been shown to reduce composite bleeding episodes; however, this effect is not seen if actual bleeding events alone are examined (3.7 vs 5.6 per 100 patient-weeks; P = .66).¹⁵  Furthermore, bleeding events may not be an independent outcome, as no effect on bleeding is seen in overall romiplostim cohorts, but all bleeding events occur at a platelet count lower than 20 × 10⁹/L.

Improvement of HRQoL while receiving romiplostim has been reported in several adult and pediatric studies. The ITP-PAQ was administered to adults randomly assigned to receive romiplostim vs nonromiplostim medical therapy over the course of 52 weeks.¹⁶  Romiplostim significantly improved report of symptoms, activity, psychological health, and overall HRQoL compared with baseline.⁷  These improvements exceeded the MID estimates, indicating a clinically significant improvement. However, when compared with baseline, the nonromiplostim group also had significant improvements in HRQoL, and the difference between romiplostim and nonromiplostim medical therapy did not exceed the MID value. In children, only parental burden is significantly reduced when receiving romiplostim in comparison with placebo.¹⁷ 

Adults with a platelet count response to romiplostim had a significant improvement in fatigue, as measured by the IT-PAQ, but not above the MID estimate.¹⁶  This finding has been consistent across studies in which treatment has not led to a consistent or clinically significant improvement in fatigue. If fatigue in ITP is related to immune dysregulation or activation, one would not expect the TPO-RAs to improve fatigue, even in those with a platelet response.⁵ 

Randomized trials in children and adults have demonstrated an initial platelet response with eltrombopag of 59% to 75%, and a durable response with continued treatment of 62% (Table 1).^18,19  Similar to romiplostim, eltrombopag is not thought to induce remission of ITP, but several case series report patients with remission after eltrombopag.

In the RAISE trial, the odds of clinically significant bleeding measured by the World Health Organization scale were 65% lower among treated patients compared with those in the placebo group. An analysis of 5 eltrombopag trials reported a decrease in bleeding from 50% to 73% at baseline to 26% to 39% at week 2 in treated patients, which was maintained throughout the study period.²⁰  The PETIT 1/2 pediatric trials have similarly shown that eltrombopag reduces bleeding compared with placebo.²¹ 

Although early trials did not show a change of SF-36-measured HRQoL in adults treated with eltrombopag compared with those receiving placebo, in the RAISE trial, significantly greater improvements from baseline in the SF-36v2 (physical, emotional, and mental function) and Functional Assessment in Cancer (FACT)-thrombocytopenia scale were seen in eltrombopag-treated patients.¹⁸  Patients receiving eltrombopag also demonstrate an improvement in vitality, a marker of fatigue on the SF-36v2, compared with those receiving placebo. Of note, however, fatigue has also been reported as an adverse event in eltrombopag trials. In PETIT1, children receiving eltrombopag had a small improvement in the Kids’ ITP Tool score, but did not meet criteria for a MID.²¹  Thus, the effect of eltrombopag on HRQoL in children with ITP is not clear.

Long-term follow-up data are available for both agents (romiplostim, >5 years; eltrombopag, ∼3 years). Both are generally well-tolerated (Table 2). However, in the United States, romiplostim is a weekly subcutaneous injection requiring weekly visits to the physician’s office. This adds to missed school or work and decreased flexibility with travel, and is a difficult mode of delivery for some patients. Eltrombopag absorption is affected by food and dairy, which makes administration more difficult for some patients, in that it must be taken 1 hour before or 2 hours after a meal and at least 2 hours before dairy. Major adverse effect concerns include bone marrow reticulin formation (approximately 2%) and thrombotic events (approximately 4%).^22-24  Data suggest bone marrow reticulin may be reversible with discontinuation of the drug; however, only a small number of patients with reticulin have been evaluated off-therapy.^22,25  Neutralizing antibodies have been reported with loss of treatment response to romiplostim, and eltrombopag can be associated with hepatotoxicity 2% (∼10% rate of transaminitis) and cataract formation (approximately 5%), which occurred almost exclusively in patients with prior chronic corticosteroid use.^19,26,27 

Splenectomy is the most predictable method for achieving a durable remission. The spleen is the major site of platelet destruction in most patients and is an important organ for B-cell development and diversification of the T-cell repertoire. Consideration of splenectomy is given most strongly to those with life-threatening bleeding (ie, intracranial hemorrhage), adults who fail corticosteroids, and patients with chronic primary ITP.

According to a systematic review of 2623 splenectomized adults, a 66% complete response was seen after a median length of follow-up of 29 months.²⁸  In a second review of 1223 patients with laparoscopic splenectomies for ITP, 92% had an early platelet count response rate with a 72% response rate at 5 years.²⁹  Indium-labeled autologous platelet scanning has been proposed to predict the likelihood of response to splenectomy, in that those with splenic clearance are theoretically more likely to respond to splenectomy than those with hepatic or mixed patterns. A pooled analysis found that this technique predicts response to splenectomy with an odds ratio of 15.4 (CI, 9.6-24.8) in those with a platelet sequestration pattern.³⁰  Since that time, several additional analyses have shown similar findings, whereas others demonstrate poor predictability.^31-33  Importantly, in all studies, many patients with hepatic clearance or mixed patterns have a complete response to splenectomy. Furthermore, clearance patterns may change after splenectomy, which could explain why the platelet response can wane over time. Availability and expertise of the indium scan is currently limited.

Systematic studies of bleeding response to splenectomy have not been performed.

HRQoL was studied using ITP-PAQ scores in 1 small subanalysis of adults with ITP pre- and postsplenectomy (n = 13), and no change was found.¹⁶  A Web-based support group survey found that patients with splenectomy had lower ITP-PAQ scores for the domains of work, bother, psychological, fear, and social activity compared with nonsplenectomized individuals; however, the overall response to splenectomy was not reported, and the cohort may have been biased by being derived from a support network.³⁴  No studies report the effect of splenectomy on fatigue.

The reported mortality rate for splenectomy for ITP is less than 1%, mainly as a result of perisurgical bleeding.²⁸  Short-term complications of splenectomy include bleeding, local infection, and thrombosis. Splenectomy is associated with a lifelong risk for life-threatening infection, historically at a rate of 0.73 per 1000 patient-years. The current risk is likely lower, given modern-day immunizations and sepsis precautions. However, evaluation for underlying immunologic disorders, such as autoimmune lymphoproliferative disorder, and secondary ITP is critical before splenectomy as a result of the increased risk of overwhelming sepsis and death in these patients.³⁵  Splenectomy may also cause an increased risk for thrombosis, including vascular concerns for cardiovascular disease and pulmonary hypertension.³⁶  However, this is less well studied in ITP than in conditions associated with hemolysis. Splenectomy is also irreversible, and often treating physicians and/or patients do not wish to trade the risk of bleeding resulting from a low platelet count for the risk for infection.

Rituximab is an anti-CD-20 monoclonal antibody, used in a wide variety of autoimmune conditions including ITP, and causes a rapid depletion of CD-20-positive B cells responsible for antibody production, a decrease in serum immunoglobulin levels, and an increased number of T-regulatory cells. Dosing was first established in oncology trials, and therefore ideal dosing for autoimmune conditions such as ITP remains unclear.

A meta-analysis of 5 randomized trials (467 nonsplenectomized adults) of rituximab vs corticosteroids, IVIG, and/or anti-D globulin demonstrated a platelet count of 100 × 10⁹/L or higher at 6 months in 46.8% of those treated with rituximab vs 32.5% in the other medical therapy group (P = .002).³⁷  For some patients, the chance of a durable remission is the reason for selecting rituximab; however, response wanes over time with durable remission rates of 21% to 26% at 5 years.³⁸  This calls into question whether rituximab is truly a curative therapy.

Studies comparing bleeding symptoms between rituximab vs corticosteroids, IVIG, and/or anti-D globulin or placebo found no significant difference (9.2% vs 5.2%; P = .44).³⁷ 

A substudy of 16 adults with chronic ITP treated with rituximab found no change in the HRQoL, using ITP-PAQ scale scores, before and after treatment.¹⁶  A randomized trial found no effect of rituximab on physical or mental quality of life compared with placebo, using the SF-36.³⁹  In addition, 22% of patients who received rituximab reported fatigue compared with 8% of patients who received placebo.

For many patients, such as those in whom adherence is challenging, the absence of a daily or home medication is of major benefit. Rituximab may be less tolerated than other therapies, with infusion reactions in up to 18% of patients. Severe or life-threatening adverse events are reported in up to 4% of patients, although this is likely an overestimate resulting from a lack of a causal relationship between rituximab and certain fatal events such as intracranial hemorrhage.⁴⁰  Recommended screening labs before infusion, monitoring, and selected adverse effects and toxicities are listed in Table 2.

Multiple other agents are used as second-line treatments for ITP, including mycophenolate mofetil, dapsone, sirolimus, vincristine, 6-mercaptopurine, azathioprine, and danazol. In addition, multiple treatment combinations have been effective and published in case reports and series. Although most of these agents have been available and used to treat ITP for decades, no research has rigorously reported on bleeding, HRQoL, and fatigue. Although these treatments may be safer or possibly equally as effective as better-studied therapies, selection of these agents is more challenging in the absence of data. Highlighted platelet response and safety data on mycophenolate mofetil and dapsone, both of which have been investigated in larger, more modern, cohorts, is outlined here. Novel treatments currently in clinical trials are shown in Table 3.

Mycophenolate mofetil is a pro-drug that is metabolized to mycophenolic acid, a noncompetitive inhibitor of a key enzyme involved in purine biosynthesis that, when inhibited, leads to cell cycle inhibition with selective inhibition of lymphocyte proliferation. Small cohort studies of patients with refractory ITP have shown the benefit of this immunosuppressant agent.

A study of 21 refractory patients reported a response rate of 53% with a platelet count higher than 50 × 10⁹/L within 12 weeks of initiation and with continued response up to a median of 24 weeks.⁴¹  A similar-sized cohort study replicated this finding with a 69% response rate.⁴²  Most recently, a 46-adult-patient cohort with primary and secondary ITP was treated with a 52% response rate of a platelet count higher than 30 × 10⁹/L.⁴³ 

Although possible adverse effects include an increased risk for infection and malignancy, common adverse effects are nausea, diarrhea, and headaches. Potential risks exist to the fetus in pregnant women. In the case series of patients with ITP, it has been a reportedly well-tolerated therapy.^41,42 

Dapsone is an inexpensive antibacterial sulfonamide with poorly understood anti-inflammatory properties. Several case series of adults with refractory ITP have shown a platelet count response to this medication. The mechanism of action of effect in ITP is not clear, possibly because of hemolysis, which leads to erythrophagocytosis by the reticuloendothelial system, preventing destruction of platelets, or a result of downregulation of Mac-1 expression, leading to decreased complement-opsonized platelet clearance by macrophages.⁴⁴ 

A review of published case series in children and adults with ITP reports an overall platelet response to dapsone of 40% to 62%.⁴⁵  The median time to response is approximately 1-2 months, and response tends to be sustained if dapsone is continued.

Dapsone is contraindicated in patients with glucose-6-phosphate dehydrogenase deficiency. In the absence of glucose-6-phosphate dehydrogenase deficiency, dose-dependent hemolysis is reported in up to 20% of patients, although in most cases it is mild. Methemoglobinemia, agranulocytosis, and hypersensitivity reactions can also occur. Although infrequently selected as a second-line treatment, dapsone represents one of the most cost-effective treatments available.

Published studies of second-line treatments for ITP are limited by varying outcome measures and definitions, efficacy endpoints, and lack of comparative trials. Ultimately, to derive a precision medicine approach to the management of ITP, we must be able to identify individual biologic profiles, which elucidate the mechanism of disease and, thereby, predict the natural history and therapy response. A personalized approach requires an understanding of the reason for treatment and, therefore, the measure of a clinically meaningful success. The effect of each treatment on that measure, including fatigue, bleeding, HRQoL, and platelet count, must then be assessed. Furthermore, these outcomes must be considered together as the post-treatment response of a patient with a platelet count of 50 × 10⁹/L or higher, but debilitating fatigue should not be considered a success.

Given the impracticality of randomized comparative trials resulting from significant differences in treatment approaches (surgical vs pharmacologic) and duration (ongoing vs intermittent), small patient numbers, and cost, investigators should consider additional trial designs such as noncomparative trials. Comparative effectiveness research can help account for patient preferences, various outcome measures, cost, and potential adverse effects. Until these data are available, clinicians must be mindful of the areas of insufficient data and the implications of this for their patients, and continued shared decision-making between patients and their providers will be critical to treatment selection and success. As therapies for ITP expand, research trials must be powered to detect differences among treatments and important outcomes to patients.

Rachael Grace, Dana-Farber/Boston Children’s Cancer and Blood Disorder Center, 450 Brookline Ave, D3-106, Boston, MA 02215; e-mail: Rachael.Grace@childrens.harvard.edu.