In bone marrow failure, levels of circulating hematopoietic growth factors (HGFs) are extremely elevated, in compensation for low blood counts. In general, administration of recombinant HGFs is not effective in aplastic anemia (AA) and related syndromes. We have established the utility of eltrombopag, a thrombopoietin (Tpo) mimetic, in severe AA (SAA) refractory to immunosuppression (IST) (Olnes, N Engl J Med, 2012; Desmond, Blood, 2014). Eltrombopag in combination with standard IST also appears to increase the rate and speed of recovery in treatment-naive SAA (Townsley, EHA. 2015). These clinical results are surprising, as plasma Tpo is markedly increased in SAA (Feng, Haematologica, 2011). In investigating possible mechanisms of action of eltrombopag, we studied HGF dynamics over time in SAA patients undergoing various treatment regimens and manifesting a range of hematologic responses. In a cohort of 37 treatment-naive SAA patients who were treated with either standard IST (horse antithymocyte globulin/cyclosporine A:ATG/CsA, n = 10) alone or with the addition of eltrombopag (n = 27), among whom 8 were non-responders and 29 responders at 6 months, we measured plasma Tpo, granulocyte-colony stimulating factor (G-CSF) and erythropoietin (Epo) using magnetic multiplex assays. Concentrations of these HGFs were greatly elevated before treatment. In a majority of patients, G-CSF declined rapidly to the normal range by 6 months, with no difference between responders and non-responders, and no correlation between G-CSF and absolute neutrophil counts. Epo concentrations decreased to the normal range at 6, 12, and 24 months post treatment in responders, to a significantly greater degree than in non-responders (p = 0.001, 0.0012, 0.038, respectively). There was a negative correlation between Epo and hemoglobin (r2 = 0.3126, p<0.0001). Tpo levels did decrease by 6 months and later, but remained much higher than the normal range, however Tpo levels in responders were significantly lower than in non-responders at 6, 12, and 24 months, respectively (p = 0.0009, 0.0477, 0.0036), with Tpo in non-responders remaining at high baseline levels (Figure 1A). These data suggest that a decrease in Tpo and/or Epo levels following initial IST therapy is an indicator of hematologic response. Linear regression analysis reveals a negative correlation between Tpo levels and platelet counts (r2 = 0.3243, P < 0.0001), consistent with previous reports. In a cohort of 7 SAA patients refractory to IST but who achieved a clinical response to eltrombopag, Epo and G-CSF levels in plasma decreased to the normal range at 3-4 months post initiation of eltrombopag treatment, but Tpo levels remained significantly elevated over the normal range for up to 5 years of follow-up, despite sustained clinical response (Figure 1B). Regression analysis revealed a negative correlation between Tpo and platelet counts (r2 = 0.3696, p < 0.0001). We asked if persistently elevated levels of Tpo even in responders are due to incomplete recovery of platelets or megakaryocytes, and thus followed up Tpo levels up to 7 years post horse ATG/CsA treatment in a cohort of 9 patients who obtained stable complete remission with platelet counts greater than 100 K/uL. Tpo levels in these patients had declined by 3 and 6 months post treatment, gradually decreased, then stabilized at levels still higher than healthy controls (Figure 1C) despite platelet counts of 102-314 K/uL (median, 174 K/uL) by 6 months and later post treatment, and recovery of bone marrow megakaryocyte frequency to normal or only mildly decreased. Tpo levels in these patients negatively correlated with platelet counts (r2 = 0.3904, p < 0.0001. Figure 1D). In conclusion, Epo and G-CSF levels fall promptly in patients with SAA who recover blood counts. In contrast, circulating Tpo levels remain markedly elevated for long periods even after complete response to treatment. Likely, Tpo is high due to low megakaryocyte mass. The consequences of sustained high Tpo levels on primitive progenitors and stem cells in these circumstances are unclear, and may play a role in stem cell exhaustion and clonal evolution. [Display omitted] DisclosuresTownsley:Novartis: Research Funding; GSK: Research Funding. Winkler:Novartis: Research Funding; GSK: Research Funding. Dumitriu:Novartis: Research Funding; GSK: Research Funding. Young:Novartis: Research Funding; GSK: Research Funding.