Abstract Background: The addition of eltrombopag (EPAG) to standard immunosuppressive therapy (IST) has improved hematologic response rates in clinical trials, establishing a new standard of care for severe aplastic anemia (SAA). However, the translation of these efficacy findings into real-world effectiveness across large, heterogeneous patient populations remains incompletely characterized. This multicenter, retrospective study aimed to assess the efficacy and safety of adding EPAG to standard IST in a large, real-world cohort of adult patients with SAA. Methods We used TriNetX, a global collaborative database, to conduct a retrospective cohort study comparing outcomes in adult patients (≥ 18 years old) with SAA. Patients with aplastic anemia were identified using ICD-10 codes. SAA was defined by these criteria: absolute neutrophil count (ANC) 500/uL, reticulocytes ≤60,000/uL, or platelets 20,000/uL. Patients were divided into two cohorts: those receiving standard IST alone (consisting of anti-thymocyte globulin with cyclosporine and no EPAG) versus those receiving EPAG with standard IST.. All patients from TriNetX were queried up to August 3rd, 2025. We performed a propensity score outcome analysis at 6 and 12 months. The primary endpoint was overall complete response rate (CRR) and overall partial response rates (PRR) at 6 and 12 months. CRR was defined as ANC 1000/uL, or platelet count 100,000/uL, or hemoglobin 10 g/dL. PRR was defined by patients no longer meeting the criteria for SAA and not meeting the criteria for CRR. Secondary outcomes included all-cause mortality and hepatotoxicity. Results The analysis compared two cohorts: Cohort A (1,112 patients) receiving standard IST alone, and Cohort B (660 patients) receiving EPAG with IST. After propensity score matching, 569 patients remained in each cohort. CRR rates at 6 and 12 months were higher in the IST-alone arm, as surrogated by Hgb levels 10 mg/dL (odds ratio (OR)= 1.942, 95% CI 1.516-2.488), ANC 1000/uL (OR= 0.688, 95% CI 0.524-0.904), and platelet count 100,000/uL (OR= 2.866 (95% CI 2.252-3.647) at 6 months. Hgb levels 10 mg/dL (odds ratio (OR)= 1.67, 95% CI 1.281-2.178), ANC 1000/uL (OR= 0.718, 95% CI 0.543-0.949), and platelet count 100,000/uL (OR= 2.383 (95% CI 1.873-3.032) at 12 months. Conversely, PRR rates at 6 and 12 months were higher in the EPAG arm, surrogated by not meeting two or more of the following criteria: platelet 20,000/uL (OR= 0.409, 95% CI 0.316-0.529), reticulocytes 60,000/uL (OR= 0.674, 95% CI 0.440-1.033), and ANC500/uL (OR= 0.487, 95% CI 0.383-0.620) at 6 months. And platelet 20,000/uL (OR= 0.396, 95% CI 0.304-0.516), reticulocytes 60,000/uL (OR= 0.694, 95% CI 0.471-1.023), and ANC500/uL (OR= 0.487, 95% CI 0.383-0.620) at 12 months. Mortality at 12 months was significantly higher in the EPAG arm (HR 1.416, 95% CI 1.001-2.003), while no significant difference was observed at 6 months (HR 1.209, 95% CI 0.784-1.863). Patients receiving EPAG had significantly higher rates of blood transfusion. In a subgroup analysis excluding all patients who received transfusions (n=195 per group after matching), the increased mortality risk was no longer observed, and the difference became non-significant (HR 0.909, 95% CI 0.506-1.633). Rates of hepatotoxicity were similar between arms. Conclusion This large retrospective analysis adds to a mixed but evolving picture of eltrombopag's real-world effectiveness. The addition of EPAG to IST was associated with significantly higher PRR, a clinically meaningful outcome representing resolution of severe cytopenias. Conversely, CRR were lower in the EPAG arm, differing from some prior real-world reports. The initial observation of a higher one-year mortality risk with EPAG was not significant after accounting for transfusion status, suggesting confounding by indication. These findings indicate that while EPAG improves the depth of hematologic recovery in routine practice, its impact on complete response and survival warrants further prospective evaluation to clarify its definitive role.
Building similarity graph...
Analyzing shared references across papers
Loading...
Saifullah Syed
Blood
Indiana University School of Medicine
Indiana University
University of Indianapolis
Building similarity graph...
Analyzing shared references across papers
Loading...
Saifullah Syed (Mon,) studied this question.
synapsesocial.com/papers/69362f444fa91c937236d58b — DOI: https://doi.org/10.1182/blood-2025-3201