Corticosteroid-Dependent or Corticosteroid-Unresponsive Primary ITP

This month, Waleed Ghanima, MD, PhD, Terry Gernsheimer, MD, and David J. Kuter, MD, DPhil, review treatment options for adult patients with primary immune thrombocytopenia (ITP) that is dependent on or unresponsive to corticosteroids. This material is repurposed from “How I treat primary ITP in adult patients who are unresponsive to or dependent on corticosteroid treatment,” published in the May 20, 2021, edition of Blood.

  • Approximately 80% of adult patients with ITP have treatment failure with corticosteroids (the standard frontline therapy) or become dependent on them and require second-line therapy.
  • Corticosteroids, at one time given for many months, are now strongly recommended to be discontinued by six to eight weeks.
  • Splenectomy should be delayed for at least 12 to 24 months, allowing for more patients to achieve remission on medical therapies before considering surgery.
  • Of the three medical options, both TPO-RAs and fostamatinib are maintenance therapies, whereas rituximab is given for a limited number of doses.
  • Although the platelet response is usually maintained with a TPO-RA or fostamatinib, half of rituximab responders will no longer respond one to two years after administration and require retreatment or other therapy.

Immune thrombocytopenia (ITP) is characterized by an isolated platelet count <100×109/L, which increases patient bleeding risk. ITP treatment attempts to reduce platelet and megakaryocyte destruction through immune modulation and splenectomy or to increase platelet production by megakaryocytes with thrombopoietin receptor agonists (TPO-RAs). Corticosteroids are the standard first-line treatment and, although >75% of adults with ITP respond to corticosteroids, approximately 80% of adult patients with ITP will experience treatment failure with corticosteroids or become dependent on them and require second-line therapy.

Several new and effective therapies have been introduced during the past decade and the understanding of ITP’s disease burden and its effect on quality of life has expanded. Guidelines now recommend that splenectomy, the standard second-line therapy for decades, be delayed for at least one to two years, allowing for more patients to achieve remission on medical therapies before considering surgery.

The available treatment options have variable efficacy, safety profiles, mechanisms of action, and modes of administration. Unfortunately, there are no reliable biomarkers that help in treatment selection, mandating an individualized approach to treatment where patient involvement, preferences, and values have become central to the process of choosing the appropriate therapy. Here, we describe our approach to personalizing therapy for adults who have ITP that is unresponsive to or dependent on corticosteroids.

Patient 1: Young Female Patient

A 23-year-old woman presented with generalized mucocutaneous bleeding and a platelet count of 3×109/L. A diagnosis of ITP was made, and she was treated with dexamethasone 40 mg/day for four days. Her platelet count began increasing by day 4, reaching 112×109/L at day 14, but then declined by day 25 to 11×109/L. Because she responded to the first cycle of dexamethasone and treatment was well tolerated, a second cycle was initiated. After a second response, the ITP again relapsed. At a follow-up visit 10 weeks after diagnosis, she had a platelet count of 9×109/L, cutaneous bruises, and heavy menstrual bleeding. She noted increasing fatigue and anxiety, fearing the consequences of disease, and worried whether she would be able to safely become pregnant and carry a child in the future.

Treatment Selection for Patient 1

The treatment algorithm for corticosteroid-dependent or -unresponsive ITP is illustrated in the FIGURE. Current clinical guidelines recommend treating patients with a platelet count of <20 to 30×109/L. With no life-threatening bleeding, there was no need for rapid platelet count elevation. Had her menstrual bleeding become more significant, the addition of hormonal contraception or tranexamic acid could have been considered. A TPO-RA, rituximab, or fostamatinib were possible options. Depending on when she was considering pregnancy, she may have wished to avoid TPO-RAs and fostamatinib. Given evidence indicating that females <40 years of age may have a 73% response rate (56% complete response), rituximab may be the most appropriate choice.

TPO-RAs are usually considered before rituximab because they are well tolerated and very effective; up to 30% of patients taper off therapy in the first year and maintain safe platelet counts. While fostamatinib is effective as a second-line therapy, we would consider other medications first because of fostamatinib’s gastrointestinal side effects. After discussing treatment advantages and disadvantages and evaluating her values and preferences, rituximab was chosen.

Patient 2: Older Patient

During a routine check-up, a 78-year-old woman had an abnormal complete blood count and a blood smear remarkable only for the decreased number of platelets. She had hypertension, diabetes, and osteoporosis, and was taking aspirin. Aspirin was discontinued, and prednisone was initiated at 1 mg/kg per day. Platelet count increased to 62×109/L after two weeks. Prednisone was gradually tapered by 5 to 10 mg/day weekly, but the platelet count fell to pretreatment levels at 10 mg/day, prompting a return to prednisone dose of 30 mg/day. Her blood sugar became persistently elevated, requiring addition of another antidiabetic agent. At a follow-up visit 12 weeks after diagnosis, she was still receiving prednisone 20 mg/day with a platelet count of 34×109/L.

Treatment Selection for Patient 2

Current clinical guidelines recommend limiting the maximum duration of corticosteroid treatment to six to eight weeks. The patient had already been treated with corticosteroids for three months and still required a relatively high prednisone dose to maintain an adequate platelet count. Her diabetes worsened and required additional antidiabetic agents. Prednisone was tapered and the therapeutic options were discussed with the patient.

Among the three medical therapies, a TPO-RA was favored over rituximab and fostamatinib based on their durable, six-month response rates. A possible increased thromboembolism risk with TPO-RAs should be considered.

The patient opted for a TPO-RA because of the high response rate and few side effects. Because she preferred an oral agent with no dietary restrictions, avatrombopag 20 mg/day was initiated. Two weeks after initiation of avatrombopag therapy, her platelet count increased to 72×109/L, and during the next 12 months, it fluctuated between 59 and 88×109/L. Given that the patient had a stable platelet count for six to 12 months, avatrombopag was tapered (every other week) to 20 mg every other day, then to 20 mg every third day. At that dose, the platelet count declined to 30×109/L, prompting a return to 20 mg every other day.

Patient 3: Patient Needing Elective Surgery

A 39-year-old man with a platelet count of 27×109/L was referred for a presurgical evaluation before an elective laminectomy. Three years earlier, a platelet count of 22×109/L was noted on a routine physical examination, with no bleeding and no other potential causes. At that time, he was given prednisone 60 mg/day, and within eight days, his platelet count rose to 46×109/L, but his blood sugar rose to 350 mg/dL. Prednisone was discontinued. Since then, he has been monitored without therapy, with platelets 25 to 38×109/L. He has no symptoms or other medical problems except for obesity.

Treatment Selection for Patient 3

The risk of bleeding in a relatively young, healthy, and asymptomatic patient, with platelet counts of 20 to 30×109/L, is very low, and a watch-and-wait approach is consistent with all current clinical guidelines. It is often helpful to educate such a patient at the microscope and show him that, although his platelet number is low, his platelets are large and well granulated, both indicators of good platelet function.

However, even large platelets may not fully protect him from bleeding with surgery. Corticosteroids would certainly work in this situation, but prednisone had previously precipitated diabetes. Intravenous immunoglobulin may rapidly increase the platelet count in emergency procedures, but it is expensive and may require multiple infusions.

This patient was started on romiplostim 3 mg/kg each week for two weeks, based on a retrospective study that showed that this dose increased the platelet count to >100×109/L in 79% of patients within 14 days. His platelet count rose to 175×109/L by the time of his operation, 17 days after initiation of romiplostim. His surgery was uneventful, and three weeks later, his platelet count had returned to 37×109/L.

Patient 4: Patient Seeking a Cure

An otherwise healthy 46-year-old airline pilot presented with easy bruising, petechiae, and a platelet count of 12×109/L. He had Hashimoto thyroiditis and took only thyroid replacement medication. Three years ago, he had a platelet count of 6×109/L and hematuria and had been treated with prednisone 60 mg/day, with his platelet count rising to 181×109/L. Corticosteroids were tapered over nine months, and his platelet count remained at 105 to 135×9/L. He was frustrated by the recurrence of ITP because he had been told the corticosteroids had “cured” his disease.

Treatment Selection for Patient 4

Although current clinical guidelines eschew using corticosteroids for nine months, as this patient had experienced, he was one of ~30% of adults who have a corticosteroid remission lasting for three years. No clear triggering event precipitated his relapse. He was opposed to further corticosteroids, given his previous poor quality of life during the treatment. Because of his occupation as a pilot with irregular working hours, the patient was not enthusiastic about a TPO-RA and sought a treatment that would provide the highest immediate rate of successful remission.

He opted for splenectomy because of his desire not to take medications for a prolonged time, the high remission rate, and possibly the reduced need for platelet count monitoring and doctor visits. He received pneumococcal, meningococcal, and Haemophilus influenzae type-B vaccines, and laparoscopic splenectomy was scheduled after six weeks. In the meantime, romiplostim was initiated as a bridging therapy before splenectomy to provide a preoperative platelet count of >50×109/L. His platelet count increased to 72×109/L on the day of splenectomy and was 223×109/L six days later when romiplostim was discontinued. However, over the next four months, his platelet count returned to 12×109/L.

Unfortunately, there is no reliable, readily available, noninvasive means of predicting the response to splenectomy. The high response and remission rates for splenectomy are based on old studies when splenectomy was often performed in the first few months of disease, thereby including up to 30% of patients who would have had a remission in the first year. Therefore, response rates in patients for whom second-line medical therapies have already failed may be 70% to 80%. Next, the patient was treated with avatrombopag with a good response.


Corticosteroids, at one time given for many months, are now strongly recommended to be discontinued by six to eight weeks. Splenectomy, the standard second-line therapy for decades, is now recommended to be delayed for at least one to two years, allowing for more patients to achieve remission with medical therapies before considering a surgical option. As newer medications are continuously being developed and evaluated in randomized clinical trials, it is highly recommended that therapies with evidence from randomized controlled trials be tried first, leaving therapies with less validation for later use.

The number of patients unresponsive to or dependent on corticosteroids may be reduced in the future by more aggressive initial treatments, such as adding recombinant TPO or rituximab, but those treatments have not yet been incorporated into any clinical guidance documents or into our practice.