Eltrombopag for the treatment of refractory thrombocytopenia associated with connective tissue disease

To assess the efficacy and safety of eltrombopag in connective tissue disease (CTD)-immune thrombocytopenia (ITP), we conducted this single-center retrospective observational study, including patients with refractory CTD-ITP who were treated with eltrombopag between January 2018 and August 2019. The characteristics of patients at baseline, and the efficacy and safety of the drug were analyzed. The predictors for a response were analyzed using a univariate analysis such as Chi-square or nonparametric test and a multiple correspondence analysis (MCA) method. A total of 15 patients with refractory CTD-ITP were included in the study. Their median age at the time of inclusion was 40.6 years. The median platelet count at initiation of eltrombopag was 11.53 × 109/L. The median remission time was 3.42 weeks. The complete remission (CR) and overall response rate decreased with time. The factors that associated with response to eltrombopag in patients with CTD-ITP were protopathy, WBC counts, levels of hemoglobin, and characteristics of bone marrow findings in univariate analysis. In addition, MCA indicated that a poor response to eltrombopag in patients with refractory CTD-ITP was closely associated with a protopathy with SS, medium to severe degree of anemia, leukopenia, and bone marrow aspiration showing aplastic anemia, an absence of megakaryocytes or macrophage activation syndrome (MAS). In conclusion, eltrombopag was effective and well-tolerated in patients with CTD-associated thrombocytopenia. Some factors should be considered in the use of eltrombopag, including the protopathy, blood test, and bone marrow histology.


Materials and methods
A total of 15 hospitalized patients with thrombocytopenia associated to CTD, including 9 patients with SLE, 3 patients with Sjögren's syndrome (SS), and 3 other patients diagnosed with undifferentiated connective tissue disease (UCTD) in the Renji hospital, Shanghai between January 2018 and August 2019 were enrolled in the study. We confirmed that informed consent was obtained from all subjects or, if subjects are under 18, from a parent guardian. All included patients showed a poor or no response to prior high-dose steroids and immunosuppressant, such as cyclosporine A, azathioprine, and rituximab, as defining refractory thrombocytopenia associated with CTD. Inclusion criteria were: (a) subjects who fulfilled the Systemic Lupus International Collaborating Clinics Criteria (2012) for SLE 6 , American College Rheumatology (ACR), and European League Against Rheumatism criteria (EULAR) (2016) for SS 7 or diagnosed with UCTD; (b) the baseline platelet counts ≤ 30 × 10 9 /L. Patients with arterial thrombosis, a history of venous thrombosis necessitating anticoagulation therapy, splenectomy within 12 weeks before the first screening visit, and with active malignancy were excluded from the study.
Response criteria for treatment were as follows: (a) complete remission (CR), platelet count ≥ 100 × 10 9 /L measured on 2 occasions (at least 7 days apart), and the absence of bleeding; (b) partial response(PR), platelet count ≥ 50 × 10 9 /L measured on 2 occasions (at least 7 days apart), and the absence of bleeding 8 ; (c) no response, a platelet count < 50 × 10 9 /L or a less than twofold increase in platelet count from baseline or the presence of bleeding 2 . Patient characteristics and other treatments used for thrombocytopenia and their dose were also noted. The baseline platelet counts, platelet counts change curves, time to response (TTR), duration of treatment, the CR and PR rate during treatment, and adverse effects of eltrombopag were also analyzed. The predictors for a response were analyzed using a univariate analysis such as Chi-square or nonparametric test (Table 3), including age, gender, protopathy, platelet reduction time, prior therapy, anemia, white blood cells (WBCs) counts, reticulocyte counts, level of complement, bone marrow features, and so on. A multiple correspondence analysis (MCA) of the association between variables corresponds to a cross-tabulation of categorical variables and a multidimensional scaling technique. The results of MCA were graphically presented in a 2-dimensional Euclidean space. This study was approved by Renji hospital ethics committee.

Statistical analysis. A descriptive statistical analysis was developed in Excel (Microsoft Corp). Normally
and non-normally distributed continuous variables were respectively summarized as the median and interquartile range (IQR). Discrete variables were summarized as percentages. In order to identify predictors for response, we performed a univariate analysis such as Chi-square or nonparametric test of baseline variables between responder and non-responder of eltrombopag in patients with CTD-ITP. To visualize homogeneous clusters of patients with distinct response to eltrombopag, we included all baseline variables (Table 3) showing p-values < 0.1 in univariate analysis by MCA as previously described 9,10 . We stated that all methods were carried out in accordance with relevant guidelines and regulations.

Results
Patient characteristics. A total of 15 female patients with refractory CTD-ITP were enrolled in this study.
There are 9 SLE patients, 3 SS patients, 3 patients diagnosed with UCTD. Only one patient was APS (tested positive for aPL and β2-GPI). Their median age at the time of inclusion was 40.6 years (IQR, 16-63 years). The median protopathy duration was 6.87 years (IQR, 0-31 years), the severe thrombocytopenia was 4.8 years (IQR, 0-16 years), and the median disease duration before the onset of severe thrombocytopenia was 2.07 years. In 13 patients, severe thrombocytopenia was the presenting manifestation of protopathy. The lowest platelet counts were less than 15 × 10 9 /L in all patients. All patients were refractory to a high dose of corticosteroids (at least 1 mg/kg day) and prior immunosuppressive treatment with thrombocytopenia, including cyclosporine A (86.7%, 13/15), IVIG (53.3%, 8/15), TPO (33.3%, 5/13), rituximab (13.3%, 2/15) and other therapy (e.g., azathioprine, cyclophosphamide, mycophenolate mofetil, hydroxychloroquine). Among 15 patients with CTD-ITP, only one patient was suffering from hepatitis B cirrhosis, splenomegaly, and hyperthyroidism; 1 patient had ovarian cancer. Among 9 patients with SLE, the disease activity of one patient was severe, while most of the patients were stable or mild as measured by systemic lupus erythematosus disease activity index (SLEDAI) score.
Bone marrow aspiration was performed in 11 patients, where 5 patients showed a typical ITP. Patient characteristics, prior immunosuppressive treatments, and bone marrow feature are summarized in Table 1.
Efficacy of eltrombopag on refractory CTD-ITP. The median platelet count at initiation of eltrombopag was 11.53 × 10 9 /L (IQR, 1-30 × 10 9 ); 12 (80%) patients had a baseline platelet count 15 × 10 9 /L or less. All subjects received eltrombopag at a dose of 25 mg/day, except one patient who received 50 mg/day. Nearly all the patients (93.33%, 14/15) completed 4 weeks treatment with eltrombopag; one patient discontinued the treatment of eltrombopag for complete remission, while 3 patients discontinued as lacking efficacy after 4 weeks treatment. The baseline platelet counts before treatment and platelet counts change curves at each time point are shown in Fig. 1A.
The median duration of therapy was 21.2 weeks (2-104 weeks), and the median remission time was 3.42 weeks (1-12 weeks) (Fig. 1B). Most of the patients (80%, 12/15) reposed well to eltrombopag. As shown in Fig. 1C, the CR and overall response rate was 26.67% and 66.67% 4 weeks later, respectively. However, the CR and overall response rate decreased with time; the CR rate was 13.33%, and the overall response rate was 40% 6 months later. The severe adverse events were seldom, except in one patient who stopped taking eltrombopag 4 weeks Identification of predictive factors for eltrombopag. To investigate the predictive factors for eltrombopag in refractory CTD-ITP, we first compared baseline variables between responder and non-responder of eltrombopag in patients with CTD-ITP by univariate analysis. Some potentially baseline variables (defined as p-values < 0.1) for the efficacy of eltrombopag among patients with CTD-ITP were shown in Table 3, including protopathy (p = 0.071), WBCs counts (p = 0.004), levels of hemoglobin (p = 0.002), and characteristics of bone marrow findings (p = 0.01). To graphically represent the association of categorical variables in a 2-dimensional Euclidean space, we then performed a MCA based on the baseline clinical manifestations and laboratory findings. With a Cronbach's alpha of 0.905 and variance of 77.7% for the modeling, the dimension 1 (the horizontal axis) seemed to separate two distinct clusters with favorable or unfavorable factors for the use of eltrombopag in patients with refractory CTD-ITP. As shown in Fig. 2, medium to severe anemia, leukopenia, bone marrow aspiration showing aplastic anemia (AA), an absence of megakaryocytes or macrophage activation syndrome (MAS), and a protopathy with SS have been associated with the poor response to eltrombopag in patients with refractory CTD-ITP. While, patients with normal WBCs counts, no or slight anemia, and relative normal bone marrow aspiration have been associated with a good response to eltrombopag.

Discussion
Eltrombopag is a thrombopoietin-receptor agonist (TPO-RA) that mimics the action of endogenous TPO on megakaryocytes and megakaryocyte precursors, promoting their growth and differentiation, thus, increasing platelet production 11 . In 2018, guidelines for ITP recommended TPO-RAs for patients who did not respond well to glucocorticoids or rituximab, or those with splenectomy 12 . In this study, we performed a real-world investigation of eltrombopag treatment for patients with CTD-ITP. Our results demonstrated that eltrombopag might be an effective therapy for patients with ITP secondary to autoimmune diseases who did not respond well to standard ITP treatment.
In this study, 80% of patients responded well to eltrombopag. Most of our patients received a dose of eltrombopag of 25 mg/day, which has been recommended for patients of East Asian descent. Eltrombopag was well tolerated, except for severe oral ulcers, which were observed in one patient. The median time taken for initial response to eltrombopag was 3.42 weeks, and peak response rate occurred after a one-month treatment. Many positive results in terms of effectiveness and maintained platelet response rate of eltrombopag have been reported in the treatment of chronic ITP 13,14 . Interestingly, we found that the response rate gradually declined until reaching the plateau stage. It might be related to the small dose of eltrombopag administrated on the CTD-ITP patients, and also the clinical severity of the primary disease for CTD-ITP. Indeed, several clinical trials observed the similar plateau phenomenon in more than half of robust responders of eltrombopag in refractory severe aplastic www.nature.com/scientificreports/ anemia 15 or patients with chronic immune thrombocytopenia 16 . Furthermore, the effect of eltrombopag on cell proliferation rates may also have a plateau phenomenon 17 .
A number of studies have indicated that TPO-RAs is associated with thrombotic risk when treating patients with chronic liver disease and those having antiphospholipid antibodies (aPL) [18][19][20][21][22] . In our study, one patient had hepatitis B cirrhosis, and another one was aPL and β2-GPI positive; yet, no thromboembolic events were observed after 26 weeks and 16 weeks treatments, respectively.
Many ITP real-world studies evaluated clinical and laboratory variables associated with response to eltrombopag and the relative factors, including age, anemia, bone marrow features, and so on 11 . In our study, we found that anemia, decreased WBCs counts, bone marrow aspiration showing aplastic anemia, an absence of megakaryocytes or macrophage activation syndrome (MAS) were predictors of worse outcome. Bone marrow characteristics such as megakaryocytopenia, bone marrow hypocellularity, and dyserythropoiesis were predictors of worse outcome 23 . Eltrombopag is effective in ITP, aplastic anemia (AA), and myelodysplastic syndromes (MDS), which is probably due to a sort of spillover specificity on other growth factor receptors, as well as to the action at the stem-cell level 11 . On the whole, although responses are observed in MDS and AA, eltrombopag is more effective in classic ITP, confirming the importance of megakaryocytic reservoir 23,24 . Indeed, marrow megakaryocyte count could be a response predictor of severe thrombocytopenia in patients with SLE and SS 25,26 . However, more prospective studies are warranted to address whether SS-ITP patients have a poor response to eltrombopag than other CTD-ITP patients.
This study has a few limitations. First, the sample size was small. Second, because this is a retrospective study, possible recall and selection errors cannot be excluded.

Conclusions
Eltrombopag is effective and well-tolerated in CTD-associated thrombocytopenia. Some factors should be considered in the use of eltrombopag, including the disease of protopathy, blood test, and bone marrow aspiration. Additional large studies would be needed to confirm the results. www.nature.com/scientificreports/