Serum erythropoietin levels in children with leukemia

Our aim was to test the hypothesis that, in leukemic children, serum erythropoietin (EPO) levels vary inversely with hemoglobin.

Twenty-four children (15 males, nine females) with an age range of 1-16 years (mean, 7.7 years) diagnosed with acute leukemia (22 acute lymphocytic, two acute myeloid) were studied over 4 months. Serum EPO and hemoglobin were measured simultaneously at multiple time points in the course of their disease, and a multiple regression analysis was performed to describe the EPO-hemoglobin relationship.

In a model adjusted for individual subject, there was a significant correlation between hemoglobin and logEPO in these leukemic children (r = -0.55, P < .01, n = 100). When measurements at hemoglobins less than 10.0 were analyzed the correlation increased significantly (r = -0.88, P < .01, n = 21). However, approximately 20% of the observations fell into one of two groups: an inappropriately low EPO for hemoglobin or an inappropriately elevated EPO for hemoglobin. The clinical characteristics of the children at each of these determinations were not different in any manner from the determinations which fell within the 95% confidence intervals for predicted mean EPO value: each of the outlying points came from a patient who at other times had an appropriate EPO for hemoglobin.

There existed a significant inverse relationship between hemoglobin and EPO, suggesting that the feedback mechanism for EPO is intact. Reasons for inappropriately high or low EPO, for level of hemoglobin, are not clear and may be reflective of other aspects of bone marrow or EPO metabolism.