Serum erythropoietin level: relationships to blood hemoglobin concentration and erythrocytic activity of the bone marrow

Exploring the effects of post operative hyperoxic intermittent stimuli on reticulocyte levels in cancer patients: a randomized controlled study

BACKGROUND

Anemia is common among hospitalized critically ill and surgical oncological patients. The rising incidence of cancer and aggressive treatments has increased the demand for blood products, further strained by a dwindling donor pool. The normobaric oxygen paradox (NOP) has emerged as a potential avenue to increase EPO levels. While some studies support its efficacy, research remains limited in clinical settings. This study aims to assess the effectiveness of a NOP protocol in stimulating erythropoiesis, as measured by changes in reticulocyte counts, in cancer patients undergoing abdominal surgeries.

METHODS

This is a post hoc analysis of a prospective, single-center, controlled, randomized study. A total of 49 patients undergoing abdominal surgery were analyzed at the Institut Jules Bordet. Adult patients admitted to the intensive care unit (ICU) for at least 24 h were enrolled, excluding those with severe renal insufficiency or who received transfusions during the study period. Participants were randomized into two groups: a normobaric oxygen paradox (OXY) group who received 60% oxygen for 2 h on days 1, 3, and 5 post-surgery and a control (CTR) group who received standard care. Data on baseline characteristics, surgical details, and laboratory parameters were collected. Statistical analysis included descriptive statistics, chi-square tests, t-tests, Mann-Whitney tests, and linear and logistic regression.

RESULTS

The final analysis included 33 patients (median age 62 [IQR 58-66], 28 (84.8%) males, with no withdrawals or deaths during the study period. No significant differences were observed in baseline surgical characteristics or perioperative outcomes between the two groups. In the OXY group (n = 16), there was a significant rise (p = 0.0237) in the percentage of reticulocyte levels in comparison to the CTR group (n = 17), with median values of 36.1% (IQR 20.3-57.8) versus - 5.3% (IQR - 19.2-57.8), respectively. The increases in hemoglobin and hematocrit levels did not significantly differ between the groups when compared to their baselines' values.

CONCLUSIONS

This study provides preliminary evidence supporting the potential of normobaric oxygen therapy in stimulating erythropoiesis in cancer patients undergoing abdominal surgeries. While the OXY group resulted in increased reticulocyte counts, further research with larger sample sizes and multi-center trials is warranted to confirm these findings.

TRIAL REGISTRATION

The study was retrospectively registered under NCT number 06321874 on The 10th of April 2024.

Response and prediction of response to recombinant human erythropoietin in patients with solid tumors and platinum-associated anemia

Objective. Anemia is commonly observed in tumor patients and may be worsened by the concomitant administration of chemotherapy with platinum. Recombinant human erythropoietin (rhEPO) is able to improve the chemotherapy-associated anemia and the quality of life in patients with solid tumors. The most effective dose, route, and frequency of administration of rhEPO in anemic cancer patients has not yet been determined.

The purpose of this review is to describe the results obtained up to now concerning the response and prediction of response to rhEPO in anemic patients with solid tumors undergoing platinum-based chemotherapy. In particular, the renal function conditions in this patient group were examined.

Data sources. The literature was reviewed through a MEDLINE search covering 1987 to 1997. Relevant articles were additionally obtained by the systematic examination of article references. We searched the following terms: rhEPO, tumor, anemia, cisplatin, carboplatin, and renal failure.

Study Selection. The current literature was reviewed with regard to both the pathophysiology of platinum-induced anemia and the clinical use, indications, and dosage of rhEPO in anemic tumor patients undergoing platinum therapy. Particular attention was directed to the prediction criteria of this drug.

Conclusions. The rhEPO response rates in patients with solid tumors and platinum-associated anemia vary widely in the current literature and range from 36% to 82%. Unfortunately, no strong predictor of response can be recommended at the present time. Clinical studies support findings concerning a reduction in renal function parameters and a blunted erythropoietin response in patients with platinum-induced anemia. It is possible that a baseline serum erythropoietin concentration in conjunction with the serum creatinine concentration or the creatinine clearance value can serve as a prognostic indicator for rhEPO therapy response in this patient group.

Erythropoietin guides multipotent hematopoietic progenitor cells toward an erythroid fate

The erythroid stress cytokine erythropoietin (Epo) supports the development of committed erythroid progenitors, but its ability to act on upstream, multipotent cells remains to be established. We observe that high systemic levels of Epo reprogram the transcriptomes of multi- and bipotent hematopoietic stem/progenitor cells in vivo. This induces erythroid lineage bias at all lineage bifurcations known to exist between hematopoietic stem cells (HSCs) and committed erythroid progenitors, leading to increased erythroid and decreased myeloid HSC output. Epo, therefore, has a lineage instructive role in vivo, through suppression of non-erythroid fate options, demonstrating the ability of a cytokine to systematically bias successive lineage choices in favor of the generation of a specific cell type.

Ablation of Gata1 in adult mice results in aplastic crisis, revealing its essential role in steady-state and stress erythropoiesis

The transcription factor Gata1 is expressed in several hematopoietic lineages and plays essential roles in normal hematopoietic development during embryonic stages. The lethality of Gata1-null embryos has precluded determination of its role in adult erythropoiesis. Here we have examined the effects of Gata1 loss in adult erythropoiesis using conditional Gata1 knockout mice expressing either interferon- or tamoxifen-inducible Cre recombinase (Mx-Cre and Tx-Cre, respectively). Mx-Cre-mediated Gata1 recombination, although incomplete, resulted in maturation arrest of Gata1-null erythroid cells at the proerythroblast stage, thrombocytopenia, and excessive proliferation of megakaryocytes in the spleen. Tx-Cre-mediated Gata1 recombination resulted in depletion of the erythroid compartment in bone marrow and spleen. Formation of the early and late erythroid progenitors in bone marrow was significantly reduced in the absence of Gata1. Furthermore, on treatment with a hemolytic agent, these mice failed to activate a stress erythropoietic response, despite the rising erythropoietin levels. These results indicate that, in addition to the requirement of Gata1 in adult megakaryopoiesis, Gata1 is necessary for steady-state erythropoiesis and for erythroid expansion in response to anemia. Thus, ablation of Gata1 in adult mice results in a condition resembling aplastic crisis in human.

Change in erythropoietin pharmacokinetics following hematopoietic transplantation

Pre-clinical studies have demonstrated that bone marrow ablation has a profound effect in decreasing erythropoietin (EPO) elimination. The study's objective was to determine in humans if EPO pharmacokinetics (PKs) are perturbed following bone marrow ablation. EPO PK studies were performed in eight subjects, aged 4 to 61 years, undergoing fully myeloablative hematopoietic stem cell transplantation. Serial PK studies using intravenous injection of recombinant human EPO (92+/-2.0 U/kg) (mean+/-SEM) were carried out during four periods of altered marrow integrity: baseline pre-ablation, post-ablation pre-transplant, early post-transplant pre-engraftment, and late post-transplant full engraftment. Compared with baseline, post-ablation pre-transplant and early post-transplant EPO PKs demonstrated declines in clearance increases in terminal elimination half-life of 36 and 95%, respectively. Clearance and half-life returned to baseline following full engraftment. The association of EPO elimination with decreased bone marrow activity in patients undergoing transplantation conclusively establishes the bone marrow as a key determinant of EPO elimination in humans.

Enhanced hypoxia-stimulated erythropoietin production in mice with depression of erythropoiesis induced by hyperoxia

Current evidence suggests that a modulatory action on O(2)-dependent EPO secretion is exerted by the erythroid/precursor cell population in the erythropoietic organs through a negative feedback system. The hypothesis is based on studies of stimulated-EPO secretion performed in mice in whom the erythropoietic rates were either enhanced or depressed in the presence of normal plasma EPO half-lives. Since erythropoietic depression was elicited by cyclophosphamide administration, which could have altered EPO production directly, the aim of the present investigation was to estimate hypoxia-stimulated EPO secretion in a mouse model of functional depressed erythropoiesis induced by exposure to normobaric hyperoxia. Females CF#1 mice aged 70 d were divided into control (C) and experimental (E) groups. The former was maintained in plastic cages in a normal environment, while the latter was placed in an environment of 60% O(2)/40% N(2) in an 85-dm(3) atmospheric chamber with air flow of 1 L/min. Erythropoiesis was evaluated by either 24-h RBC-(59)Fe uptake or iron kinetics performed 3 h after IV injection of a tracer dose of (59)Fe. Both indexes of the red cell production rate were significantly depressed in E mice. Plasma disappearance of exogenous EPO in C mice, as well as in E mice exposed to hyperoxia for 4 d, was estimated by injecting (125)I-rHuEPO intravenously. Linear regression analysis indicated that neither the differences between the slopes of both curves nor the Y-intercepts were significant. Hypobaric hypoxemia was used as stimulus for EPO production. Plasma immuno-EPO titer after a 4-h exposure to hypobaric air was 73% higher in mice with hyperoxia-induced hypoerythropoiesis than in control mice with normal erythropoiesis. Data support the concept that the rate of erythropoiesis, perhaps through the number of the erythroid progenitor/precursor cell population, modulates O(2)-dependent EPO secretion.

The enigma of the metabolic fate of circulating erythropoietin (Epo) in view of the pharmacokinetics of the recombinant drugs rhEpo and NESP

Recombinant human erythropoietin (rhEpo) is a mainstay in the treatment of anaemia, primarily in renal failure. Because the half-life of circulating rhEpo is relatively short (4-8 h), the drug is usually administered 2-3 times weekly. Recently, a novel erythropoiesis-stimulating protein (NESP) with a longer half-life (24-26 h) has been approved. NESP possesses two additional N-glycans compared to endogenous Epo or rhEpo. The pharmacokinetics of rhEpo and NESP in humans have been investigated in detail. The composition of the N-glycans is clearly important in determining the biological activity and the velocity of the degradation of Epo and its analogues. However, due to the lack of knowledge of the main site and mechanism of the removal of Epo from circulation, the difference in survival of rhEpo and NESP has remained phenomenological. Investigators have implicated the liver, kidneys, and bone marrow as possible sites of the catabolism of Epo. However, while hepatocytes take up desialylated Epo, the liver does not appear to play a major role in the degradation of intact Epo. Likewise, renal Epo clearance is apparently of secondary importance. Studies showing non-linear pharmacokinetics of Epo suggest that Epo is eliminated by saturable mechanisms. The hormone, as well as the recombinant drugs, can be incorporated by erythrocytic progenitors and other tissues expressing the Epo receptor. The affinity of the Epo receptor for rhEpo is 4.3-fold higher than for NESP. Taken together, it seems most likely that native Epo, rhEpo and NESP are degraded following Epo receptor-mediated uptake, mainly in bone marrow.

Erythropoietin

Maintenance of the red cell volume is a fundamental aspect of ensuring oxygen supply to the tissues. The balance between the very dynamic processes of erythropoiesis and erythrocyte loss is precarious and yet normal individuals experience a remarkably constant haematocrit. This is achieved by a very elegant and sensitive homeostatic mechanism which links tissue oxygen delivery to red cell production. The glycoprotein hormone erythropoietin (EPO) is the principle controller of this process. It is now clear that even minor underproduction of EPO will result in anaemia. The most widespread example of this is the anaemia of end-stage renal failure. The pharmacological use of recombinant human EPO (rHuEPO) in this setting is now well established and has had a dramatic impact on the quality of life of patients with renal disease. With the more widespread use of EPO in other clinical conditions and the advent of novel therapeutic approaches, this is an opportune moment to review the physiology and patho-physiology of this fascinating and essential hormone.

Plasma erythropoietin in newborn lambs

In the present study early postnatal changes in erythropoietin (Epo) level and hemoglobin concentration in 8 lambs were examined. Plasma Epo was estimated by a monoclonal enzyme-immunoassay (ELISA), developed for human Epo. In all the lambs, except one, Epo was low, or undetectable, immediately after birth. Within 6-12 h after birth, a marked increase in plasma Epo was found in 4 of the lambs. Within 3-7 days after birth, Epo was back to low levels. In the lamb with high Epo levels at birth, drastic decreases were observed during the next 6 h. There was no obvious correlation between the hemoglobin concentrations and the changes in plasma Epo.

Mild renal dysfunction is sufficient to induce erythropoietin deficiency in patients with unexplained anaemia

Current guidelines suggest that anaemia due to erythropoietin deficiency almost exclusively occurs with creatinine concentrations of at least 177 micromol/l or above. The aim of this prospective case control pilot study was to evaluate whether borderline renal function or mild renal dysfunction with creatinine concentrations well below 177 micromol/l is sufficient to induce inadequate erythropoietin secretion. Patients referred for work-up of otherwise unexplained anaemia with mildly abnormal creatinine concentrations (104-129 micromol/l; study group: eight patients) and patients referred for work-up or therapy of other diseases who also presented with anaemia but normal creatinine levels (<100 micromol/l; control group: nine patients matched for gender, age and degree of anaemia) were included. All but two patients in the control group had bone marrow biopsies to exclude other pathologies. Mild renal dysfunction (as evidenced by creatinine concentrations between 100 and 140 micromol/l, median concentration 112 micromol/l) was found to be sufficient to induce inadequate erythropoietin secretion. The physiologic hemoglobin-dependent erythropoietin regulation demonstrated in the control group was abolished in the study group. Patients with mild renal dysfunction and unexplained anaemia should be investigated for erythropoietin concentration. If the erythropoietin concentration is found to be inadequate for the degree of anaemia, substitution therapy should be considered.

Serum erythropoietin and interleukin-6 levels in hemodialysis patients with hepatitis virus infection

The influence of hepatitis B (HBV) and hepatitis C virus (HCV) infection on blood hemoglobin (Hb) and serum erythropoietin (Epo) and interleukin-6 (IL-6) concentrations was studied in 48 anemic patients on regular hemodialysis. They were grouped as follows: (I) 19 patients whose Hb values improved after infection (Hb > 85 g/L), (II) 10 patients with persisting anemia after infection (Hb < 75 g/L), and, without hepatitis virus markers (III) 8 patients with Hb > 85 g/L and (IV) 11 patients with Hb < 75 g/L. Serum immunoreactive Epo levels were significantly higher in group I (34.4+/-47.1 U/L) than in the other groups (II, 10.8+/-6.0; III, 7.9+/-3.2; IV, 8.4+/-4.3). Serum IL-6 was higher in group I than group III (7.7+/-7.8 pg/ml vs. 3.6+/-2.4; p = 0.05) but similar to the other groups. Hb levels in group I were maximal at the time of serum alanine aminotransferase normalization. Red cell production increases as a result of elevated circulating Epo during hepatic regeneration after HBV or HCV infection.

Are current methods of measurement of erythropoietin (EPO) in human plasma or serum adequate for the diagnosis of polycythaemia vera and the assessment of EPO deficiency?

Current methodology for the immunoassay of erythropoietin (EPO) in human plasma or serum is reviewed, with an emphasis on measurement of EPO concentrations in the low and normal ranges, analytical interference and blood sampling requirements. In only 2 out of 8 research or in-house immunoassays reported since 1987 was there evidence that patients with polycythaemia vera (PV) could be identified, PV being an EPO-independent form of polycythaemia in which EPO concentrations are low in untreated cases. The same was true for only 1 out of 13 currently available kit methods. Remarkable differences in sample stability have been observed with different methods. Measurement of EPO in serum is recommended in most published articles. However, only EDTA plasma seems to be acceptable for the one generally available method with proven high diagnostic sensitivity for PV. It is concluded that most EPO assay methods have not been shown to be adequate for the measurement of the low EPO concentrations, and thus have poor diagnostic sensitivity for PV. It is inferred that they might not be appropriate to assess states of EPO deficiency. Only when a sufficiently sensitive diagnostic method becomes generally available will it be possible to define the various causes of low EPO concentrations. As in other fields of polypeptide hormone measurement, further developments in the field of EPO assay may be expected to be important in diagnostic medicine.

A tracer interaction method for nonlinear pharmacokinetics analysis: application to evaluation of nonlinear elimination

A drug tracer is most commonly applied to get information about the pharmacokinetics (PK) of a drug that is not confounded by an endogenously produced drug or an unknown drug input. An equally important use of tracers that has not been fully recognized is their use in the study of nonlinear PK behavior. In the present study a system analysis is applied to examine the interaction between drug molecules characteristic and intrinsic to any nonlinear process which enables the nonlinearity to be identified and modeled using a drug tracer. The proposed Tracer Interaction Methodology (TIM) forms a general developmental framework for novel methods for examining nonlinear phenomena. Such methods are potentially much more sensitive and accurate than previous methods not exploiting the tracer principle. The methodology proposed is demonstrated in a simulation study and with real data in a specific implementation aimed at determining the Michaelis-Menten (MM) parameters of nonlinear drug elimination while accounting for drug distribution effects. The simulation study establishes that the TIM-based, MM parameter evaluation produces substantially more accurate parameter estimates than a nontracer (NT) conventional method. In test simulations the accuracy of the TIM was in many cases an order of magnitude better than the NT method without evidence of bias. The TIM-based, MM parameter estimation methodology proposed is ideally suitable for dynamic, non-steady-state conditions. Thus, it offers greater applicability and avoids the many problems specific to steady state evaluations previously proposed. TIM is demonstrated in an evaluation of the nonlinear elimination behavior of erythropoietin, a process that likely takes place via receptor-based endocytosis. Due to its high sensitivity, accuracy, and intrinsic nonlinearity the TIM may be suitable for in-vivo studies of receptor binding of the many biotechnology produced peptide drugs and endogenous compounds displaying receptor-mediated elimination.

Unexpected hypoxia-dependent erythropoietin secretion during experimental conditions not affecting tissue oxygen supply/demand ratio

Although a great deal of evidence supports the hypothesis that plasma erythropoietin (EPO) levels of mammals are related to the oxygen supply to the tissues relative to their oxygen needs, several observation millitate against its inherent simplicity. This study presents our results obtained from in vivo experiments that suggest that hypoxia-dependent EPO production can be altered by conditions which apparently do not modify the tissue oxygen supply/demand ratio. Hypoxia-dependent EPO production rate (EPO-PR), derived from plasma EPO titers and plasma EPO half-lives, were estimated in both transfused-polycythemic and normocythemic mouse models subjected to different treatments. From calculations of the O2 carrying capacity of blood and body O2 consumption, it was assumed that the tissue supply/demand ratios were similar in both experimental and control mice of the same model at the time of induction of EPO production. The following observations were worth noting: (1) EPO-PRs in transfused polycythemic mice whose erythropoietic rates were stimulated by intermittent exposure to hypobaria (0.5 atm, 18 hr/day x 3 weeks), phenylhydrazine administration (40 mg/kg at weekly intervals x 3 weeks) or repeated rh-EPO injections (1500 U/kg 3 times a week x 3 weeks) before transfusion were more than five times high than in comparabily polycythemic mice whose erythropoietic rates were not stimulated previously; and (2) EPO-PR in response to hypobaric hypoxia was 2.08 times normal in normocythemic mice with cyclophosphamide (100 mg/kg) induced depression of erythropoiesis, and 0.33 times normal in normocythemic mice with rh-EPO (400 U/kg x 2) induced enhancement of erythropoiesis. Although the results obtained in polycythemic mice are difficult to explain, those from normocythemic mice suggest the existence of a feedback mechanism between EPO-responsive cells and EPO-producing cells. Both demonstrate the existence of experimental conditions in which modulation of the hypoxia-dependent expression of the EPO gene appears to occur. This modulation would be dependent on factors other than oxygen.

Serum erythropoietin and serum transferrin receptor levels in aplastic anaemia

Serum erythropoietin (EPO) and soluble transferrin receptor levels were serially measured in 74 patients with aplastic anaemia (AA). As control groups we investigated healthy controls (n = 24) and patients with iron-deficiency (n = 23) or haemolytic anaemia (n = 16). There was a significant negative correlation of log EPO on haematocrit both in AA patients and in the anaemic control group. However, for the same degree of anaemia, log EPO levels in AA were significantly higher than in iron-deficiency or haemolytic anaemia. EPO levels at diagnosis did not correlate with severity of aplastic anaemia, nor did they predict outcome after immunosuppression. During immunosuppressive treatment of AA with anti-thymocyte globulin and cyclosporine A, EPO levels were significantly lower compared with pre-treatment values without a corresponding change in haematocrit. This impaired EPO response to anaemia during immunosuppression might affect recovery of erythropoiesis. In AA patients, EPO levels declined with haemopoietic recovery. However, compared with normal controls, EPO levels in remission patients were still higher with respect to their haematocrit. Results of this study argue against the model of a simple feedback regulation of EPO via hypoxic anaemia. Our data support the hypothesis that cytokines and the erythropoietic progenitor pool are involved in the regulation of EPO production. The results illustrate that serial measurements of EPO along with therapeutic interventions are necessary to identify patients who might benefit from treatment with exogenous recombinant human EPO.

Effects of chemotherapeutic and immunosuppressive drugs on the production of erythropoietin in human hepatoma cultures

The effects of agents used in combination chemotherapy were studied on erythropoietin synthesis in cultures of the human hepatoma cell line, HepG2. Erythropoietin was measured by radioimmunoassay in the culture media after 24-h treatment periods. The RNA synthesis-inhibiting drugs daunorubicin, cyclophosphamide, ifosfamide, and cis-diamine-dichloroplatinum produced a dose-dependent decrease of the production of erythropoietin. Inhibition was also induced by the tubulin-binding agents vincristine and colchicine and the immunosuppressive agents azathioprine and cyclosporin A. The DNA synthesis-inhibiting drugs methotrexate and cytosine arabinoside, the antimycotics 5-fluorocytosine and amphotericin B, and glucocorticoids did not inhibit. Viability studies showed that the inhibition of erythropoietin production was partly correlated with cytotoxicity. These data could be relevant with respect to recombinant erythropoietin replacement therapy in tumor-associated anemia.

A sensitive sandwich ELISA for measuring erythropoietin in human serum

A sandwich, non-competitive enzyme-linked immunosorbent assay (ELISA) for erythropoietin (EPO) is described. The ELISA utilizes a monospecific, polyclonal antibody raised in rabbits against human recombinant EPO (rhu EPO) and purified over a rhu EPO affinity chromatography column. The ELISA procedure can be summarized as follows: Anti-EPO is coated onto 96-well ELISA microtitre plates; standard EPO or sample is added and left to bind to this catching antibody; this is followed by the addition of the same antibody which has been biotinylated; finally, anti-biotin conjugated to alkaline phosphatase is added and the enzyme reaction developed and read at 405 nm. All parameters of the assay have been optimized. Recombinant human EPO was standardized against the World Health Organization 2nd International Reference Preparation for erythropoietin. The minimal detectable concentration of rhu EPO was 0.3-0.5 mU/ml, which corresponded to 1.2-2 mU/ml of EPO in serum (serum diluted 1:4). No reaction was obtained with a variety of blood components and cytokines, indicating that the anti-EPO antibody did not cross-react with those substances to produce false-positive results. The intra-assay variation ranged from 3% to 10%, while the inter-assay variation ranged from 8.5% to 24%. Serum dose-response curves were parallel to the standard dose-response curve. The assay is easy to use, rapid, reproducible, but above all quantitative, specific and sensitive to measure the EPO content in all serum samples.

Serum erythropoietin levels in hemodialysed patients after administration of recombinant human erythropoietin

In anemic patients on regular hemodialysis (HD), correction of anemia with recombinant human erythropoietin (rHuEpo) administered intravenously (iv) or subcutaneously (sc) was followed over a 2-month period. Monitoring serum Epo post-dose concentrations after the first iv rHuEpo injection and following another regular injection after 2 months of therapy with rHuEpo iv in 9 patients showed that the Epo elimination half-life was reduced from 7.48 h to 4.68 h. In the same patients the initially low percentage of erythroblasts and mature erythroid progenitors increased during 2 months of rHuEpo therapy. Because Epo molecules bound to Epo receptors are internalized in target cells we suggest that the expansion of the Epo responsive cell pool could explain the shorted Epo elimination time after 2 months of rHuEpo treatment. By monitoring serum Epo concentration following sc rHuEpo injection in 7 HD patients it was found that the modest increase in serum Epo levels (30-60 mU/ml) was sufficient to correct anemia.

Changes in serum erythropoietin levels during allogeneic bone marrow transplantation

Serial serum erythropoietin levels were measured in 10 consecutive patients undergoing allogeneic bone marrow transplantation. Observed erythropoietin levels are compared with those predicted from a large control population of anaemic patients not receiving chemotherapy. There was an initial acute rise in serum erythropoietin, peaking between days 1 and 4 after marrow transfusion, which was unrelated to changes in haemoglobin concentration. Patients maintained serum erythropoietin concentrations at around twice the predicted level for the first 2 weeks following transplantation, with a gradual fall into the expected range by wk 3. Erythropoietin levels did not change with episodes of bacterial infection or acute graft-versus-host disease. A patient with severe aplastic anaemia had initial successful engraftment with normalisation of erythropoietin levels, but showed a marked and amplified rise in erythropoietin 2 wk before falling peripheral blood counts indicated failure of the bone marrow graft.