Background and aims: Erythropoietin (EPO) is the key regulator of red blood cell production. In response to hypoxia, EPO levels can increase 1000-fold and remain high for weeks to promote hematopoiesis. Therapeutically, the introduction of EPO and erythropoietic stimulating agents into clinical practice has revolutionized the treatment of anemia despite certain concerns regarding the safety of the therapy. Recent studies demonstrate that EPO has activities in addition to hematopoiesis, and modulates bone remodeling by increasing bone resorption and decreasing bone formation, leading to trabecular bone loss. In vitro, EPO directly inhibits murine osteoblast differentiation and mineralization at doses relatively lower than those shown to stimulate osteoclastogenesis. The aim of this study was to investigate the dose-response relationship between EPO dose, hemoglobin (Hgb) levels, and the extent of bone loss, as well as to examine the role of the monocytic and B cell EPO receptor (EPOR) in bone metabolism.

Results: Treating mice for 2 weeks with escalating doses of EPO, ranging from 6-540 IU/week, led to a dose-dependent increase in Hgb accompanied by a more dramatic decrease in trabecular bone mass; regression slopes of Hgb and bone volume/total volume (BV/TV, a measure of bone density) were 0.009 vs -0.09, respectively (p<0.05). These effects were associated with a significant increase in the number of preosteoclasts (CD115+ cells) in the bone marrow (r=0.74, p<0.05). To assess whether the osteoclast lineage contributes to EPO-induced bone loss, we generated mice lacking EPOR in the monocytic lineage (LysM-cre+/+;EPORflox/flox, cKO). At steady state, these mice and their LysM+/+;EPORwt/wt controls exhibited similar levels of Hgb (16.7±0.57 and 16.8 ±0.25 g/dL, respectively) and BV/TV (2.73%± 0.73 and 3.10%±0.76, respectively). Although not completely abolished, the bone loss induced by high EPO doses (540 IU/week) was significantly attenuated in cKO compared to control mice (60%±4.7 reduction versus 40%±13.2 reduction, respectively). At the same time, the levels of osteoclast precursors (CD115+ cells) increased from 3.08%±1.12 to 4.67%±0.92 in EPO-treated control mice, although there was no change in bone marrow preosteoclasts and preosteoblasts (defined as CD11b-/ALP+) in EPO-treated cKO mice. This suggests that osteoclast EPOR is responsible, at least in part, for mediating the effect of EPO on bone mass. Adding to the complexity of EPO's osteoimmunological roles, new findings suggest that EPO also regulates bone resorption via EPOR signaling in B cells. EPO stimulated surface expression of the osteoclastogenic RANKL in B cells (MFI: 2.6%±0.1 to 3.13%±0.09 P<0.05) and we found a higher bone mass in mice with conditional EPOR KO in B cells (MB1-cre+/-;EPORflox/flox) (vBMD, 52.2±15.1 versus 40.8±8.8 mg HA/cm3 in MB1-Cre+/-;EPORwt/wt, p<0.05).

Conclusions: Our data demonstrate the complexity of EPO-induced bone loss mediated at least partly by EPOR signaling in both myeloid and B cell lineages. Furthermore, since patients who require treatment with EPO are prone to osteoporosis, our data suggest that using the lowest effective EPO dose would not only decrease the risk of thromboembolic complications but also minimize adverse skeletal outcomes.

SHB and AK - Equal contribution; YG and DN - Equal contribution

Funded by the German Israel Foundation, Grant # 01021017 to YG, DN, MR and BW and the Israel Science Foundation, Grant # 343/17 to DN.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution