Response to thrombopoietic agents is related to on‐treatment bone marrow megakaryocyte morphology in patients with chronic immune thrombocytopenia

Immune thrombocytopenia (ITP) is an autoimmune disease in which antiplatelet antibodies and T-cells mediate thrombocytopenia by a combination of accelerating platelet destruction and reducing platelet production. It is known that megakaryocytes are usually increased in number in chronic ITP (cITP), although a corollary increase in platelet count is not automatically seen.1 This suggests that the abundant megakaryocytes may be defective in compensatory mechanisms designed to increase platelet production. Megakaryocytes in patients with ITP have traditionally been thought of as large and numerous compared to their normal counterparts, though with various morphologic abnormalities. Recent studies evaluating megakaryocytes in patients with ITP treated with thrombopoietin receptor agonists (TPO-RA) primarily focused on number, although our group and others have reported clustering of megakaryocytes as well as other “myeloproliferative” changes: hypercellularity, atypia, and an increase in reticulin fibrosis.2 This study aimed to establish whether certain bone marrow morphologic characteristics are associated with response to thrombopoietin agents as measured by platelet counts and the absolute immature platelet fraction (A-IPF). The A-IPF, the absolute number of immature platelets per unit volume, reflects the number of immature (reticulated) platelets in circulation and is a marker of platelet production. The A-IPF is of particular interest as it correlates to the acute bleeding score in patients with ITP when platelets are <60 × 109/L, likely due to the increased hemostatic properties of young, reticulated platelets.3 Patients with cITP treated at Weill Cornell Medical College were included in the study which was approved by the Institutional Review Board and consent, as well as assent when appropriate, was obtained from all patients. All study patients underwent bone marrow examination for clinical reasons with aspirate and biopsy, were receiving a TPO-RA at the time of bone marrow biopsy, and had an A-IPF value measured using the Sysmex XE2100 (Sysmex Corporation, Lincolnshire, Illinois). The study primarily used descriptive statistical analysis. Data was analyzed using GraphPad Prism 7.03. Comparison of groups used the Mann-Whitney U test or unpaired t test. Bone marrow biopsy samples were assessed for cellularity, number of megakaryocytes, clustering, and fibrosis. The megakaryocyte count was performed manually for 10 high power fields (HPF) in each sample; the 10 field availability was required for inclusion. Clustering was defined as multiple megakaryocytes nested together as used in assessment of myeloproliferative neoplasms (MPN).4 Reticulin fibrosis was graded per the European consensus system.5 Patients described as TPO-RA responders had platelet increases of at least 20 000/uL to a count >50 000/uL. Patients achieving a response were on variable doses of either romiplostim or eltrombopag. The study group was comprised of 19 patients with a total of 21 bone marrow evaluations and included both children and adults. No patients included in this study had a prior or subsequent diagnosis of myelodysplastic or myeloproliferative syndrome, aplastic anemia, or a thrombocytopenia known not to be ITP. Eight patients had undergone prior splenectomy. Of the 21 bone marrow samples, 14 had megakaryocyte clustering, representing a biologic response to therapy. Responders to TPO-RA tended to be more likely to have clustering of megakaryocytes as shown in Figure 1C (P = .0395). History of splenectomy was associated with higher A-IPF (P < .0001). There was a trend for splenectomized patients to have increased megakaryocytic clustering (P = .066). Increasing marrow reticulin fibrosis showed a trend toward correlation with higher A-IPF as shown in Figure 1D (P = .0651, rs = 0.41). Four patients were graded 0 fibrosis, 15 were graded 1+, one patient grade 2+, and one grade 3+. Pathologic fibrosis (2+ or 3+) in these samples was associated with both history of splenectomy and clustering as well as with increased A-IPF (P = .0052), with the caveat that only two of the 21 patients fell into this category of pathologic fibrosis. There was no significant correlation between fibrosis and clustering (P = .2171). The number of megakaryocytes ranged from 10 to 313 per 10 HPF (median 90, SD 72.8). Total number of megakaryocytes was not significantly correlated to A-IPF (P = .49, rs = 0.1579,) nor was number of small megakaryocytes (P = .754). The platelet count was positively correlated to A-IPF (P = .0015) and tended to be related to splenectomy status (P = .0373), but not to bone marrow characteristics. This pilot, hypothesis-generating study evaluated the relationship of response to TPO-RA and the morphology of megakaryocytes and identified three principle outcomes. First, megakaryocyte clustering was correlated with increased A-IPF. Megakaryocyte clustering is classically associated with MPN, and not observed in normal marrows. It is included in the WHO diagnostic criteria to delineate polycythemia vera, essential thrombocythemia, and primary myelofibrosis.4 The mechanism underlying megakaryocyte clustering remains poorly understood, both in MPN and in ITP patients on TPO-RA. One hypothesis is that clustering occurs in settings of accelerated progenitor development with simultaneous release of multiple megakaryocytes. Second, splenectomy was associated with increased A-IPF, in contrast to previous studies. While patients undergoing but failing splenectomy clearly represent a group with the most severe disease, the increased levels of A-IPF may represent a longer half-life of newly made platelets. However, it is not clear why, if true, this would not result in better platelet count responses to TPO-RA. Last, increasing bone marrow fibrosis was associated with increasing A-IPF, even though fibrosis was identified to be within the normal range in 19/21 marrows. Currently, it is not recommended to obtain routine bone marrow evaluations to monitor fibrosis in ITP patients receiving TPO-RAs.6 Our study is limited by the number of subjects available, due to the rarity of cITP, and was further constrained by requiring a marrow biopsy with 10 interpretable HPF. While a sample size of 21 marrows is small, it was large enough to demonstrate relationships between several of the marrow morphology variables and the A-IPF. In summary, our study demonstrated a correlation with increased A-IPF and megakaryocyte clustering as well as a trend toward increased reticulin fibrosis, suggesting that these are linked to expressions of increased thrombopoietic effect. A substantial increase in A-IPF was also found in patients who had undergone splenectomy despite their lesser platelet count response to TPO agent treatment in previous studies. TPO-RA have clear effects in the marrow space which are best correlated with A-IPF results as opposed to platelet counts per se. Given that clustering and fibrosis are not normally associated with platelet production, this correlation with increased A-IPF may give us further clues about marrow stimulation with these agents. These and other hypotheses generated from the preliminary results of this pilot study deserve further exploration. No funding was specifically provided for this study although it was partially supported by the Children's Cancer and Blood Foundation. J.B. has served as a consultant and/or on advisory boards of Amgen, Novartis, Dova and Shionogi. J.T.G and M.H. have no disclosures. J.T.G. and J.B.B. designed the research study. M.H. and J.T.G. performed the histopathologic analysis of the samples. M.H. carried out statistical analysis. M.H., J.T.G., and J.B.B. all contributed to the writing of the manuscript. We would like to thank the patients and their families for participating in this study.