Paroxysmal nocturnal hemoglobinuria‐phenotype cells predict a good response to eltrombopag in patients with refractory aplastic anemia

Abstract To identify factors affecting responsiveness to eltrombopag (EPAG), we retrospectively analyzed 38 aplastic anemia patients treated with EPAG who were refractory (n = 29) or showed an inadequate response (n = 9) to conventional therapies. The efficacy was evaluated at 16 weeks after starting EPAG and at any given time when the best response was achieved. Hematologic responses were observed in 15 patients (39%) at week 16 and in 25 (66%) at any given time. Ten of 19 (53%) achieved transfusion independence. A univariate analysis revealed the presence of PNH‐phenotype cells and the relatively higher platelet counts as associated with a good response to EPAG.

1 year (RAA), and nine had only shown an inadequate response to IST (IrAA), which was defined as a reduction for red blood cell transfusions, improvement of blood counts, and not meeting the response criteria according to the previous publication by Gluckman et al [2] (Table 1a).
The efficacy was evaluated at 16 weeks after starting EPAG and at any given time when patients achieved response (best overall response). The response was defined according to the criteria by Desmond et al. [3] EPAG was continued in non-responders at 3 months to anticipate further improvement when the patients showed some signs of improvement that did not meet the response criteria.
The clinical characteristics of the patients who responded and those who failed to respond to EPAG were compared using Fisher's exact test for categorical variables and the t-test for continuous variables. P-values were two-sided, and outcomes were considered significant when P-values were <.05. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University), which is a graphical user interface for the R software program (The

Patients' characteristics
The median duration from the AA diagnosis to the start of EPAG  The median time from starting EPAG to a response in at least one lineage was 9 weeks (range: 1-80 weeks, Figure 1).

A comparison between responders and non-responders to EPAG
In a univariate analysis, the platelet count (0.9-7.3 × 10 9 /L, median 3.2 × 10 9 /L) at the start of EPAG in 15 responders at 16 weeks was significantly higher than that (0.6-6.5 × 10 9 /L, median 1.8 × 10 9 /L) in 23 non-responders (P = .013, Table 1c). When the response was evaluated according to the best overall response, the response rate in eight patients positive for PNH-phenotype cells (PNH[+], 100%) was significantly higher than that (55%) of the patients (16/29) negative for PNHphenotype (P = .032). In addition, male patients tended to respond to EPAG better than female patients (P = .086). There were no marked differences in the age, time from the diagnosis to the start of EPAG, number of treatment regimens prior to EPAG, neutrophils/Hb/reticulocytes or WT1 mRNA copy number at the start of treatment between the responders and non-responders.

DISCUSSION
There are only a few reports on the efficacy of EPAG in patients with RAA and IrAA based on large cohorts. Olnes et al.(n = 25) and Desmond et al (n = 43) reported that higher reticulocyte counts at the start of EPAG were a predictor of the EPAG response. [5], [3] Although no such tendency was observed in our study, higher platelet counts and PNH-phenotype cells at the start of EPAG were factors for predicting the EPAG response. Small populations of PNH-phenotype cells are detected in approximately 50% of patients with newly diagnosed AA, [6] and are thought to represent benign bone marrow failure. [7] In this study focusing on refractory AA cases, such PNH(+) patients accounted for only 22% of our population. The favorable response to EPAG in PNH(+) patients suggests that even though they failed to respond to IST, hematopoietic stem/progenitor cells of PNH(+) patients may be healthier than those of patients without increased PNH-phenotype cells. Fattizzo et al. reported a higher response rate to EPAG in 32 PNH(+) cases than in 17 PNH(-) cases. However, the study included seven treatment-naive patients, and the response rate to EPAG in their 42 relapse/refractory patients was 15%. [8] It is unclear how many responders in the relapse/refractory cases had PNH clones.
Our study is therefore first to demonstrate a predictive value of PNH clones for a good response to EPAG in RAA or IrAA patients.
In this study, the maximum dose of EPAG was 100 mg, which is equivalent to 200 mg for non-East Asian subjects [9]. The therapeutic effect of EPAG was evaluated at 2 points: 16 weeks and at any given time when patients achieved the best response. While the Japanese medical insurance system recommends that physicians decide whether or not to continue EPAG by week 16 after starting the treatment, in the present study, 40% of patients responded to EPAG at 17 weeks or later. When the laboratory data at the start of EPAG treatment were compared between 15 responders at 16 weeks and 10 responders at 17 weeks or later, the platelet counts in the former group was significantly higher than that in the latter (Table 1c.; 32.0 × 10 9 /L vs 18.0 × 10 9 /L, P = .013). These findings support the report from NIH that the prolonged administration of EPAG is needed to achieve response in a cohort of SAA patients. [10] Current study describes the ability to predict response to EPAG, based on higher platelet counts, which is also a reflection of severity of the disease, for instance, less severe and more chance to respond. Among the six patients whose platelet counts at the start of EPAG was <10.0 × 10 9 /L, three (50%) responded by 19 weeks, while none showed a response beyond 20 weeks from the start. Based on these findings, changing the treatment may need to be considered if patients with platelet counts <10.0 × 10 9 /L at the start of EPAG do not show signs of a response by 20 weeks.

CONFLICT OF INTEREST
KI, HY, and SN received honoraria fees and lecture fees from Novartis Pharma K.K.