Adding Eltrombopag to Immunosuppression Improves Hematologic Response in Patients With Severe Aplastic Anemia

Acquired aplastic anemia (AA), which results from immune-mediated destruction of the bone marrow, can often be treated with immunosuppressive therapies. Whether the efficacy of this approach can be improved with the use of growth factors to stimulate the residual hematopoietic stem cells characteristic of AA is not known.

Based on earlier research showing that the thrombopoietin-receptor agonist eltrombopag improved hematologic response as a single agent in patients with AA refractory to immunosuppression, Danielle M. Townsley, MD, from the National Heart, Lung, and Blood Institute, and co-authors hypothesized that adding eltrombopag to immunosuppressive therapy as a firstline treatment would increase rates of complete response (CR) and improve long-term outcomes in patients with AA.

According to the study’s findings, published in the New England Journal of Medicine, this approach led to more than one-third of patients achieving a CR (primary endpoint) by 6 months. “Stimulating remaining stem cells with a drug that mimics the actions of a natural growth substance for marrow stem cells – while suppressing the immune system – improves the likelihood, quality, and speed of recovery of these seriously ill patients,” Dr. Townsley told ASH Clinical News. “A much higher number of patients responded to eltrombopag combined with immunosuppressive therapy, compared with prior results in these patients using immunosuppression alone.”

The investigator-initiated, non-randomized, historically-controlled, phase I/II study enrolled 92 consecutive patients (≥2 years old) with previously untreated severe AA between June 2012 and November 2015. Patients with Fanconi anemia, severe liver impairment, or evidence of a clonal myeloid disorder (per a cytogenetic test performed within 12 weeks prior to enrollment) were excluded.

All patients (median age = 32 years; range = 3-82 years) received immunosuppression with horse antithymocyte globulin (ATG; days 1-4) and cyclosporine daily for 6 months plus eltrombopag (150 mg daily for patients ≥12 years old; 75 mg daily for patients 6-11 years old; and 2.5 mg/kg of body weight per day in patients 2-5 years old). Patients were split into three cohorts based on the initiation and duration of eltrombopag dosing:

  • cohort 1: from day 14 to 6 months
  • cohort 2: from day 14 to 3 months
  • cohort 3: from day 1 to 6 months

The median baseline absolute neutrophil count (ANC) was 310/mm3 (range = 0-1,810/mm3), and the median baseline platelet count was 9,000/mm3 (range = 0-37,000/mm3).

Investigators performed serial bone marrow biopsies and measured blood counts to assess hematologic response; CR was defined as an ANC of >1,000 mm3, hemoglobin level of ≥10 g/dL, and platelet count of ≥100,000 mm3; patients whose blood counts no longer met the criteria for severe AA, but did not meet the criteria for CR, were considered to have a partial response (PR).

Sixty-nine English- and Spanish-speaking adult patients also completed self-reported health outcome questionnaires to measure health-related quality of life.

At 6 months, the overall response rate (ORR; defined as those reaching PR or CR) was 87 percent and, Dr. Townsley added, “almost all responders achieved independence from transfusions, with 39 percent showing markedly better, near-normal blood counts.”

More patients in cohort 3 achieved CR, compared with the other cohorts (TABLE). “In cohort 3, in which eltrombopag was given immediately along with immunosuppression and continued for 6 months, 94 percent of patients responded, with 58 percent having near-normal or normal blood counts by 6 months,” Dr. Townsley explained.

Among all responders, the median time to transfusion-independence was 32 days (range = 12-38 days) for platelets and 39 days (range = 10-73 days) for red blood cells.

In multivariate regression analyses, only longer telomere length of leukocytes and younger age were associated with response (p<0.01 and p=0.01, respectively), while no previously reported baseline hematologic characteristics predictive of response to immunosuppression were associated with response.

Seventeen patients discontinued eltrombopag because platelet counts increased to >200,000/mm3. Seven patients briefly discontinued eltrombopag during the first 2 weeks of treatment because of transient elevations in liver enzyme levels. Severe cutaneous eruptions occurred in two patients (grade 2 and grade 3) and resulted in eltrombopag discontinuation at 4 and 6 weeks, respectively. One patient died during the study because of a non-hematologic disorder.

Patients who responded to eltrombopag treatment also self-reported “improvements in their physical health and overall health-related quality of life by 6 months,” the authors wrote.

After a median follow-up of 23 months (range = 84-1,422 days), the overall survival rate at 2 years was 97 percent (95% CI 94-100).

Twelve patients underwent allogeneic hematopoietic cell transplantation (alloHCT): Six had not responded to eltrombopag, three had a relapse (defined as declining blood counts that warranted the reintroduction of full-dose cyclosporine), and three had clonal evolution (defined as the development of the myelodysplastic syndromes and acute myeloid leukemia from AA). Two patients died following alloHCT. When the researchers censored the data for alloHCT, the overall survival rate increased to 99 percent (95% CI 97-100).

Twenty-five of the 78 patients (32%) who responded to eltrombopag/immunosuppressive treatment relapsed after 6 months. The researchers amended the study protocol to allow the continuation of low-dose cyclosporine from 6 months to 2 years, and this reduced the frequency of relapse: 14 percent of patients (n=6/43) with a response who continued therapy beyond 6 months relapsed, compared with 54 percent of patients (n=19/35) with a response who stopped cyclosporine at 6 months. Reinstituting cyclosporine effectively reversed relapses and increased blood counts in 13 of 25 patients, and the addition of eltrombopag in combination with cyclosporine reversed relapse in an additional 10 patients.

Clonal evolution “was a concern” for the researchers, but they noted that the seven patients who experienced chromosomal changes “were within the range that would be expected with immunosuppression alone at 2 years,” the authors wrote.

The study is limited by its small patient population, short duration of follow-up, and non-randomized design. Also, the authors were not able to determine the exact mechanism by which eltrombopag acts in the context of bone marrow failure. A large, randomized, placebo-controlled study is underway in Europe to confirm the results of the current trial.

“The median time to clonal evolution with immunosuppressive therapy is 4 to 6 years, which is longer than the median follow-up of 2 years [in this study],” the authors wrote.

Drs. Townsley and Winkler report research funding from GlaxoSmithKline and Novartis; Dr. Scheinberg reports personal fees from Novartis.


Townsley DM, Scheinberg P, Winkler T, et al. Eltrombopag added to standard immunosuppression for aplastic anemia. N Engl J Med. 2017;376:1540-50.

TABLE. Hematologic Response to Immunosuppression Plus Eltrombopag

Rate at 3 Months

Rate at 6 Months

P Value

All Cohorts (n=92)

Overall response†

74 (80%)

(95% CI 72-89)

80 (87%)

(95% CI 80-94)


Partial response

46 (50%)

(95% CI 40-60)

44 (48%)

(95% CI 37-58)

Complete response

28 (30%)

(95% CI 21-40)

36 (39%)

(95% CI 29-49)


Cohort 1 (n=30)

Overall response

23 (77%)

(95% CI 61-93)

24 (80%)

(95% CI 65-95)

Partial response

18 (60%)

(95% CI 41-79)

14 (47%)

(95% CI 28-66)

Complete response

5 (17%)

(95% CI 3-31)

10 (33%)

(95% CI 15-31)


Cohort 2 (n=31)
Overall response

24 (77%)

(95% CI 62-93)

27 (87%)

(95% CI 75-100)

Partial response

16 (52%)

(95% CI 33-70)

19 (61%)

(95% CI 43-79)

Complete response

8 (26%)

(95% CI 9-42)

8 (26%)

(95% CI 9-42)


Cohort 3 (n=31)
Overall response

27 (87%)

(95% CI 75-100)

29 (94%)

(95% CI 84-103)

Partial response

12 (39%)

(95% CI 21-57)

11 (35%)

(95% CI 18-53)

Complete response

15 (48%)

(95% CI 30-67)

18 (58%)

(95% CI 40-76)


†The P value is for testing the comparison of the overall response rate to the overall response rate in a historic cohort (67 of 102 patients [66%]).