A 2-Year, Longitudinal, Prospective Study of the Effects of Eltrombopag on Bone Marrow in Patients with Chronic Immune Thrombocytopenia

The acquired disorder immune thrombocytopenia (ITP) results in low platelet counts due to an increase in platelet clearance [1], as well as an inhibition of megakaryocyte platelet release [2]. Mild reticulin has been shown to be present in the bone marrow (BM) of approximately 27-31% of hematologically normal patients [3, 4] and 28-52% of patients diagnosed with primary ITP [5, 6].

Eltrombopag, an oral nonpeptide thrombopoietin receptor agonist (TPO-RA), increases platelet counts in patients with chronic ITP [7, 8, 9, 10]. However, TPO-RAs have been associated with increases in reticulin fibers in the BM [11, 12, 13]. Whether the risk of BM fibrosis is due to the TPO-RAs themselves or to the underlying disease remains unclear. While BM biopsies can detect an increase in reticulin deposition in the BM, they are not routinely performed in the management of patients with chronic ITP. Although some studies have retrospectively reviewed biopsy specimens, prospectively collected data in patients with chronic ITP are limited. Additionally, inconsistencies in specimen preparation, staining, and pathologic analysis across institutions in previous studies have further confounded conclusions.

The purpose of this study was to prospectively investigate the potential effects of eltrombopag on the BM. Specifically, the study was designed to characterize any change in BM fibrosis in patients with previously treated persistent/chronic ITP from baseline to 1 and 2 years after initiating treatment with eltrombopag.

This phase IV, open-label safety study was conducted at 28 centers in 11 countries from May 6, 2010, to May 6, 2014. The study was conducted according to the principles set forth in the 2008 Declaration of Helsinki. Written informed consent was obtained from each patient prior to the performance of any study-specific procedures. The study protocol was reviewed and approved by national, regional, or investigational center ethics committees or institutional review boards as appropriate. The primary objective was to assess the levels of BM reticulin and/or collagen fibers at baseline and to detect any changes after 1 and 2 years of treatment with eltrombopag in adult patients with persistent/chronic ITP. The secondary objective was the long-term safety and tolerability of oral eltrombopag treatment in patients with persistent/chronic ITP.

Adult patients with a diagnosis of persistent ITP (for ≥6 months) or chronic ITP (for ≥12 months) based on the American Society of Hematology/British Committee for Standards in Haematology guidelines [14, 15, 16], with no evidence of other disease that could cause thrombocytopenia, were eligible for study inclusion. Patients who had previously received eltrombopag or romiplostim must have completed treatment with these agents ≥6 months before the screening BM biopsy. Patients treated with any other TPO-RA were not eligible. Patients were not eligible if their baseline BM biopsy had a marrow fibrosis (MF) reticulin grade of MF-3 (the highest grade according to the European Consensus scale) [17].

The enrollment of patients previously exposed to a TPO-RA was limited to no more than 20% of the total study population. The duration of the screening period was up to 8 weeks. Eltrombopag was administered for 2 years (±8 weeks) followed by a 4-week follow-up period. The total study duration was up to 116 weeks. The starting dose of eltrombopag was 50 mg/day, or 25 mg/day for patients of East Asian ancestry. The dosing regimen of eltrombopag was then adjusted to a lower dose or a maximum dose of 75 mg for each patient to maintain platelet counts within the clinically indicated range.

Biopsy specimens were collected at baseline (before treatment) and after 1 and 2 years of treatment with eltrombopag, or at early withdrawal of treatment if the patient withdrew >3 months after the last BM biopsy. Fresh trephine biopsy was used for all assessments with the exception of the screening biopsy, which used paraffin block, if available. An independent central laboratory processed and stained the BM biopsies with hematoxylin and eosin, and for reticulin (Gordon and Sweet method) and collagen (Masson trichrome) to reduce preanalytical variation. Two hematopathologists (R.K.B. and A.O.) independently conducted the morphology review and assessed specimens for cellularity, megakaryocyte, erythroid, and myeloid quantity and appearance, trabecular bone quality, reticulin grade (MF-0 through MF-3) [17], and the presence of collagen. Differences in scoring were reconciled by rereview and discussion.

Of 186 patients screened, 167 were enrolled and received at least 1 dose of the study drug. Five patients from one center were excluded due to protocol/Good Clinical Practices violations. Of the 162 remaining patients, 118 (73%) completed the study and 113 (70%) completed >22 months of treatment with eltrombopag. Of the 48/162 (30%) patients who discontinued eltrombopag, 22 (14%) discontinued due to adverse events (AEs), 11 (7%) due to lack of efficacy, 7 (4%) due to investigator discretion, and 8 (5%) due to patient decision.

Of the AEs leading to discontinuation, only increased alanine aminotransferase (ALT) levels (n =3), toxic hepatitis (n =2), cerebral hemorrhage (n =2), and increased blood bilirubin (n =2) occurred in more than 1 patient each (Table 1). Information on the completion status was not available for 1 (<1%) patient.

Three patients who discontinued eltrombopag had increased reticulin formation at the time of discontinuation (Table 2). A biopsy from patient 1, who discontinued due to lack of efficacy, was graded MF-2 with no collagen. She had no AEs related to BM function, although there was no subsequent biopsy performed. Patient 2, who discontinued on day 525 because of an ALT level that was more than 5 times the upper limit of normal, had a biopsy graded MF-2; a biopsy on day 568 showed MF-1. Finally, a biopsy was graded MF-3 with the presence of collagen for patient 3, who discontinued treatment because she wanted to become pregnant. The follow-up biopsy 182 days later, on day 516, was graded MF-0 and negative for collagen. Biopsies from the remaining 2 patients with findings of MF-2/MF-3 and collagen who were not withdrawn from the study are also shown in Table 2.

Demographic and baseline characteristics of the 162 patients are summarized in Table 3. The majority of patients (63%) were between 18 and 49 years old, and the median age was 42 years (range 18-80). The overall study population comprised 50% Caucasian and 49% Asian patients (20% East Asian, 29% Central/South Asian), and 1% had missing ethnicity data. A total of 155 patients (96%) had an ITP diagnosis for ≥6 months, and 124 (77%) had an ITP diagnosis for ≥12 months at baseline. A total of 114 patients (70%) had received prior therapy for ITP, and 93 (57%) were receiving ITP medication at baseline. Thirteen patients (8%) had prior exposure to a TPO-RA, of whom 7 received eltrombopag and 4 received romiplostim ≥6 months before enrollment. Approximately three quarters (76%) of the patients had a baseline platelet count of ≤50 × 10⁹/L, and 23% of patients were splenectomized.

The median daily dose of eltrombopag was 49.7 mg (range 5-75 mg), and the median time on treatment was 104 weeks (range 2.4-113 weeks). The majority of patients (61%) completed at least 24 months of eltrombopag treatment, and 73% completed more than 18 months of treatment.

Baseline BM biopsies for 3 patients were not assessable, 2 because the sample was inadequate and 1 because of a procedural error. These patients were not included in the BM analysis. Of the 159 patients assessed at baseline, 150 (94%) had MF-0 and 9 (6%) had MF-1 (Fig. 1). No baseline biopsies were positive for collagen. Most patients' (96%) on-treatment biopsies were normal or had minimally increased reticulin (graded as either MF-0 or MF-1, respectively).

Overall, 127 patients completed the baseline and 1-year BM biopsies, and 93 patients completed all 3 BM biopsies (i.e., baseline and 1 and 2 years). At 1 year (n =127), biopsies of 87 patients (69%) had a grade of MF-0, 35 (28%) had MF-1, 3 (2%) had MF-2, and 2 (2%) had MF-3 (Fig. 1). At 2 years (n =93), 83 patients (89%) had a grade of MF-0, 10 (11%) had MF-1, and no patients had MF-2 or MF-3. Biopsies of 5 patients (4%) at 1 year and 1 patient (1%) at 2 years were positive for collagen.

Eight patients had either ≥MF-2 only (n =2), the presence of collagen only (n =3), or both (n =3; Table 2). None of these patients had clinical symptoms typical of BM dysfunction or abnormalities of clinical concern reported in the white blood cell count or peripheral blood smear.

During the first year, 2 patients (2%) had shifts from MF-0 to MF-2, and 2 patients (2%) had shifts from MF-0 to MF-3. There was also 1 patient (<1%) with a shift from MF-1 to MF-2. One patient was graded as MF-2 at 1 year and remained at MF-2 at the posttreatment day 22 assessment. One patient was graded as MF-3 at 1 year, but an on-treatment biopsy 1 month later was graded as MF-0, as was a biopsy at 2 years.

Compared with baseline, 79 out of 93 patients (85%) had MF-0 at 2 years, 9 (10%) had a 1-grade increase, 2 (2%) had a 1-grade decrease, 1 (1%) remained MF-1, and none had 2- or 3-grade increases (Table 4). Data were missing for 2 (2%) patients. None of the on-treatment biopsies was prompted by an abnormal peripheral blood smear or performed at the investigator's discretion for clinical symptoms suggestive of BM dysfunction.

In addition to grading for reticulin and collagen, BM specimens were assessed for cellularity, quantity, appearance of megakaryocytes, erythroid, and myeloid cells, and trabecular bone quality (Table 5). Megakaryocyte hyperplasia was observed in the majority of patients (91% at baseline and 98% at both years 1 and 2). Erythroid hyperplasia was observed in 10-12% of patients, and myeloid hyperplasia was found in 10-21%. Megakaryocyte, myeloid, and erythroid morphology was normal in most patients.

Cellularity was normal in 80, 80, and 76% of patients at baseline and the 1- and 2-year assessments, respectively (Table 6). Trabecular bone thinning was found in 50% of patients at baseline, 50% at year 1, and 38% at year 2. No pathologic patterns of changes or clinically relevant changes in lineage features were identified in the on-treatment biopsies compared with baseline.

There were no BM fibrosis-related AEs reported. The most common AEs occurring in >10% of patients were headache, cough, arthralgia, diarrhea, nausea, upper respiratory tract infection, and pyrexia (Table 7).

This open-label safety study was designed to prospectively determine baseline levels of BM fibrosis in previously treated adults with persistent/chronic ITP and to evaluate the long-term effect of eltrombopag treatment on the formation of BM reticulin and/or collagen fibers. Six percent of patients had a BM fibrosis score of MF-1 at baseline. After 2 years of treatment, 86% of patients had no increase in grade of reticulin, and 10% had an increase of 1 grade in reticulin. No patient with an increase to MF-2 or the presence of collagen had clinical signs or symptoms indicative of BM dysfunction.

BM biopsies are not routinely performed in the diagnosis or management of chronic ITP. Although some studies have retrospectively reviewed biopsy specimens, prospectively collected data in patients with chronic ITP are limited. A Danish study of 187 patients with chronic ITP never treated with TPO-RAs showed that 60% had MF-0, 39% had MF-1, and approximately 2% had MF-2 or MF-3 [18]. Whether the risk of BM fibrosis is due to the TPO-RAs themselves or the underlying disease remains unclear. Even in the absence of TPO-RAs, the results of several studies have revealed the presence of background BM reticulin fibrosis in patients with a diagnosis of ITP [6, 11, 18, 19, 20]. Furthermore, increased BM reticulin formation has been described in patients with refractory ITP, as well as in those with other autoimmune diseases [21, 22].

TPO-RAs have been associated with increases in the formation of reticulin fibers [11, 13, 19], but the lack of prospectively collected and consistently stained and analyzed pretreatment BM specimens has limited the robustness of the conclusions. An open-label study was conducted to assess the incidence of fibrosis after 3 years of treatment with romiplostim using the Bauermeister scale, and it also included a central review by 2 hematopathologists [23]. Patients were required to have normal BM at baseline and were assigned to 1 of 3 cohorts, with BM biopsies scheduled after 1, 2, or 3 years of treatment rather than annual assessments in the same patients, as in the current study. After 2 and 3 years of treatment, 7 patients had increases in reticulin of ≥2 grades, and specimens from 2 patients showed the presence of fibrosis. After discontinuing romiplostim, 3 out of 9 patients with BM changes had follow-up biopsies that showed normal BM.

In this prospective, longitudinal study, we found that most patients with chronic ITP had MF-0 at baseline, and most patients experienced no change or a mild increase in reticulin over 2 years of treatment with eltrombopag. Furthermore, our findings indicate that collagen fibrosis was reversible.

In conclusion, these data suggest that treatment with eltrombopag is generally not associated with clinically relevant increases in BM reticulin or collagen.

This work was supported by GlaxoSmithKline (GSK); eltrombopag is an asset of Novartis AG as of March 2, 2015. The authors wish to thank Christine K. Bailey, former employee of GSK at the time of the study, for her assistance with the data analysis for the manuscript and her critical review to previous drafts. Editorial support (assembling tables and figures, collating author comments, copyediting, fact checking, and referencing) and graphic services were provided by Prasad Kulkarni, PhD, William Perlman, PhD, and Nancy Price, PhD, of AOI Communications, L.P., and were funded by GSK and Novartis Pharmaceuticals Corporation.

R.K.B. contributed to the conception and design of the study, acquisition of the data, and data analysis and interpretation; R.S.M.W. and A.O. contributed to the acquisition of the data and data analysis and interpretation; M.M.T. contributed to the acquisition of the data; K.K.B. contributed to the conception and design of the study and data analysis and interpretation; and P.B. and D.T. contributed to the data analysis and interpretation. All authors provided critical review of the manuscript and approval of the final version of the manuscript. All listed authors meet the criteria for authorship set forth by the International Committee for Medical Journal Editors.

R.K.B. and A.O. received research funding from GlaxoSmithKline (GSK). R.S.M.W. received research funding from Bayer, Biogen-Idec, Bristol-Myers Squibb, GSK, Johnson & Johnson, Merck Sharp & Dohme, Novartis, Pfizer, and Roche; he serves as a consultant for Bayer, Biogen-Idec, and Novartis, serves on the speaker's bureau for Novartis, and has participated in advisory boards for Biogen-Idec and Novartis. K.K.B. and D.T. were employees of GSK during the time in which the study was conducted. P.B. was an employee of GSK during the study and initial publication development, and is currently an employee at Novartis. K.K.B., P.B., and D.T. hold equity ownership in GSK.

Clinicaltrials.gov identifier: NCT01098487.