Effect of recombinant human erythropoietin combined with granulocyte/ macrophage colony-stimulating factor in the treatment of patients with myelodysplastic syndrome

This was a multicenter, randomized, double-blinded, placebo-controlled, parallel-group study in patients with a diagnosis of MDS and refractory anemia (RA), refractory anemia with ringed sideroblasts (RARS), or refractory anemia with excess blasts (RAEB). At the screening examination, patients provided written informed consent and full medical histories, and they underwent physical examinations that included hematologic assessment. Eligible patients were neutropenic (absolute neutrophil count [ANC] less than 1500 cells/mL) and anemic (hemoglobin less than or equal to 10 g/dL), and they had platelet counts (unrelated to transfusion) greater than or equal to 15,000/mL. Patients had to be transfusion dependent (requiring at least 4 U RBC in the previous 3 months) and older than 17 years of age, and they had to have a performance status less than or equal to 2 (on the Zubrod scale) and normal renal and hepatic function. Patients were not to have undergone cytotoxic treatment for MDS for at least 30 days before the study. Criteria for exclusion included history of malignancy (except basal or squamous cell skin cancer or cervical carcinoma in situ); recent (within 1 year) history of thromboembolic disease; acute leukemia or classification of chronic myelomonocytic leukemia; more than 20% bone marrow blasts or more than 5% blasts in peripheral blood or Auer rods on histologic examination of the bone marrow; serious bacterial or fungal infection within 30 days before the study; receipt of more than 10 mg/d corticosteroid or androgen therapy within the preceding 30 days; evidence of folate or vitamin B₁₂deficiency or untreated iron deficiency; exposure to any cytokine or any experimental drug within the preceding 30 days.

Sixty-six patients were enrolled at 13 centers in the United States. Patients were stratified at entry according to their endogenous serum erythropoietin levels (less than or equal to 500 mU/mL or more than 500 mU/mL) and were randomly assigned, in a 2:1 ratio, to receive by subcutaneous injection either GM-CSF (Schering Plough, Kenilworth, NJ) (0.3-5.0 mg/kg·d) plus epoetin alfa (PROCRIT) (supplied by Ortho Biotech, Raritan, NJ) (150 IU/kg 3 times/wk) or GM-CSF (0.3-5.0 mg/kg·d) plus placebo (3 times/wk). Treatment was administered in a double-blinded fashion for 12 weeks. GM-CSF and epoetin alfa or placebo were injected at different sites to allow the differentiation of site reactions. The initial dose of GM-CSF was dependent on the ANC: 0.3 mg/kg·d with ANC more than or equal to 500 cells/mL or 1.0 mg/kg·d with ANC less than 500 cells/mL. The dose was increased as necessary by a maximum of 0.5 mg/kg·wk so as to achieve and maintain an ANC of 2000-5000 cells/mL. The maximum permissible dose was 5.0 mg/kg·d. If the hemoglobin level reached 13.0 g/dL (unrelated to transfusion), no further doses of epoetin-alfa/placebo were given until it returned to 12.0 g/dL.

On days 1 (the first day of medication administration), 3, and 5 and weekly for 12 weeks, hematologic variables, vital signs, concurrent therapy, transfusion information, adverse events, febrile events, and infections were recorded. Patients were weighed weekly before the administration of study medication, and blood chemistry, serum iron profiles, and liver and spleen size evaluations were repeated monthly (days 29, 57, and 85). On the last day of the study (day 85), bone marrow aspirates and biopsy procedures were repeated. Transfusions were given to patients whose hematocrit was less than or equal to 25% and, at the investigator's discretion, to patients whose hematocrit was between 25% and 30%. Serum erythropoietin levels were measured using a modified competitive radioimmunoassay with a limit of quantitation of 5 mU/mL.

Criteria for withdrawal included a World Health Organization grade 3 adverse event13 attributable to GM-CSF and the development of acute leukemia or refractory anemia with excess blasts in transformation (RAEB-t). Two patients (1 from each treatment group) with French–American–British classifications (FAB) of RAEB-t were allowed to continue in the study because their conditions remained stable for a considerable time.

Primary efficacy measures were change in hemoglobin from baseline to endpoint (last value recorded), RBC transfusion rate in months 2 and 3, and proportion of patients requiring RBC transfusion during study. Secondary measures included determining the proportion of patients showing increases in hemoglobin (unrelated to transfusion) greater than or equal to 2 g/dL above the baseline value (hemoglobin responders), the proportion of patients becoming ANC correctors (ANC greater than or equal to 2000 cells/mL), and a change in platelet count from baseline to endpoint.

Significance was set at 0.05, and most tests were 1-sided. Tests of interaction were 2-sided and had a significance level of 0.10. Linear model analyses (least-squares mean) were performed to assess the effects of possible covariables (baseline hemoglobin, baseline RBC transfusion rate, and FAB classification of MDS at diagnosis) for hemoglobin level, RBC cumulative transfusion rate, platelet cumulative transfusion rate, and RBC transfusion rate in months 2 and 3. Two-sample t tests were used to test differences in monthly platelet transfusion rates. Differences in the proportion of patients transfused with RBCs and platelets were assessed using the 1-sided Fisher exact test. For the secondary variables, summary statistics were provided for weekly ANC and white blood cell counts, and 2-sample t tests were used to analyze changes in hemoglobin and platelet count. Proportions of patients showing increased (greater than or equal to 2 g/dL) hemoglobin or becoming ANC correctors were compared using the 1-sided Fisher exact test. Analysis of efficacy was based on the intent-to-treat population. Analysis of safety included data from all patients who received study medication.

Of the 66 patients enrolled, 45 received epoetin alfa and 21 received placebo. Demographic and baseline characteristics were similar in both treatment groups (Tables 1 and2). Thirty-seven patients (25 epoetin alfa, 12 placebo) had baseline endogenous erythropoietin levels less than or equal to 500 mU/mL. Baseline endogenous erythropoietin levels were greater than 500 mU/ml in 29 patients (20 epoetin alfa and 9 placebo). Mean baseline hemoglobin levels for both baseline endogenous erythropoietin groups (treatment groups combined) were 9.4 and 8.6 g/dL, respectively. Forty-eight patients completed the study (34 in the epoetin alfa group and 14 in the placebo group).

Changes in hemoglobin levels over the course of the study are shown in Table 3. In patients treated with GM-CSF and placebo, hemoglobin levels tended to decrease during the study despite transfusion, whereas hemoglobin levels were maintained at approximately baseline levels in patients on GM-CSF and epoetin alfa. For patients with more than 500 mU/mL endogenous erythropoietin, the mean change in hemoglobin level from baseline to endpoint was greater in the epoetin alfa group than in the placebo group. The adjusted mean change in hemoglobin level was 0.07 g/dL in patients who received epoetin alfa compared with −0.96 g/dL in those who received placebo (P = .048).

Hemoglobin responses occurred in 4 of 45 (9%) patients in the GM-CSF + epoetin alfa group, compared to 1 of 21 (5%) patients in the GM-CSF + placebo group (P = NS). All 5 hemoglobin responders had baseline endogenous erythropoietin concentrations less than or equal to 500 mU/ml. All 4 hemoglobin responders who received epoetin alfa had RAEB, whereas the hemoglobin responder who received placebo had RARS. The durations of hemoglobin response were 9, 13, 13, 13, and 75 weeks.

The percentage of patients transfused during the study and the mean number of units of blood transfused during months 2 and 3 are shown in Table 3. In the less than or equal to 500 mU/mL endogenous erythropoietin group, only 15 of 25 (60%) epoetin alfa patients were transfused with RBCs compared with 11 of 12 (92%) placebo patients. This difference approached statistical significance (P = .051). In the more than 500 mU/mL endogenous erythropoietin group, 19 of 20 (95%) epoetin alfa patients and 8 of 9 (89%) placebo patients were transfused. Among patients who received epoetin alfa, the percentage transfused with RBCs was significantly lower in the less than or equal to 500 mU/mL endogenous erythropoietin group (60%) than in the more than 500 mU/mL endogenous erythropoietin group (95%), P = .0069.

For all treatment groups, the overall mean ANC rose from 948 cells/μL at baseline to 3831 cells/μL by week 12. In the less than or equal to 500 mU/mL endogenous erythropoietin group, 83% of epoetin alfa patients and 92% of placebo patients were classified as ANC correctors. The corresponding values for the more than 500 mU/mL endogenous erythropoietin subgroup were 80% and 75%, respectively.

Analysis of mean change in platelet count from baseline to endpoint revealed no significant differences between the 2 treatment groups. The mean pretreatment platelet count was 108 000/μL in the epoetin alfa group and 120 000/μL in the placebo group. In the low endogenous erythropoietin group, the mean change in platelet count from baseline to endpoint was −14 640/μL in the epoetin alfa patients and −32 920/μL in the placebo patients (P = .12). In the high endogenous erythropoietin group, the mean change in platelet count from baseline to endpoint was −17 750/μL in the epoetin alfa patients and −12 330/μL in the placebo patients (P = .62).

A total of 243 adverse events were reported, 165 in the 45 epoetin alfa-treated patients and 78 in the 21 placebo-treated patients. Injection site reactions were most common (62% epoetin alfa and 86% placebo). Of the adverse events, 57 (36 epoetin alfa and 21 placebo) were deemed to have a probable or definite relationship to study medication (Table 4). Nine of the adverse events were rated as severe—5 in the epoetin alfa patients (1 pain in extremities, 3 thrombocytopenias, and 1 stroke) and 4 in the placebo patients (1 myalgia, 1 erythema, and 2 injection-site reactions).

At the doses used, combination therapy with GM-CSF and epoetin alfa was well tolerated in most patients. However, 18 of 66 patients withdrew during the course of therapy. In 8 of these 18 patients, the reason for withdrawal was thought to be disease related: infection (3), evolution to acute leukemia (2), second malignancy (1), gait disturbance (1), and asthenia (1). The other 10 patients who withdrew did so because of side effects that were thought to be drug related: thrombocytopenia (3), myalgia (2), skin erythema (2), fever (1), pericarditis (1), and stroke (1).

Changes in laboratory test values from baseline to endpoint were similar for epoetin alfa-treated and placebo-treated patients. Systolic and diastolic blood pressure values remained fairly constant in both groups. Most patients did not experience changes in liver or spleen size. Of the 36 patients on epoetin alfa who did not have liver enlargement at baseline, 3 (8.3%) experienced liver enlargement during the study. Three of 19 (15.8%) patients in the placebo group experienced liver enlargement during the study (complete liver and spleen data were available for all; none had liver enlargement at baseline). Two of 38 (5.2%) patients in the epoetin alfa group with normal spleens at baseline experienced spleen enlargement during the study, in comparison with 4 of 19 (21.1%) patients on placebo; 1 patient on epoetin alfa who had an enlarged spleen at baseline experienced a reduction in spleen size.

Transformation to acute leukemia occurred in 1 patient in the GM-CSF + epoetin alfa group and in 1 patient in the GM-CSF + placebo group. There were 3 deaths on the study (all in the GM-CSF + epoetin alfa group). The causes of death were pericarditis, stroke, and thrombocytopenia.

The objectives of this randomized, double-blinded, placebo-controlled study were to assess the efficacy (in terms of restoring normal hematopoiesis) and tolerability of combined therapy with GM-CSF and epoetin alfa in anemic, neutropenic patients with MDS and to assess the effect of pretreatment endogenous erythropoietin level on response to therapy. Hemoglobin response (an increase in hemoglobin of at least 2 g/dL unrelated to transfusion) occurred in 4 of 45 (9%) patients in the GM-CSF + epoetin alfa group, compared with 1 of 21 (5%) patients in the GM-CSF + placebo group, a difference that was not statistically significant. Although all the hemoglobin responders had levels of endogenous EPO less than or equal to 500 mU/mL, no statistically significant effect of endogenous EPO on hemoglobin response was demonstrated. The median duration of hemoglobin response was 13 weeks.

GM-CSF in combination with epoetin alfa appeared to offer benefits over GM-CSF and placebo in terms of maintaining hemoglobin levels and in reducing transfusion requirements in patients with low baseline endogenous erythropoietin levels. Low (less than or equal to 500 mU/mL) endogenous erythropoietin levels have previously been suggested to be predictive of response to epoetin alfa in patients with MDS.6,7 This is consistent with our finding that, among previously transfusion-dependent patients treated with GM-CSF and epoetin alfa, those with low erythropoietin levels were less likely to require RBC transfusion than those with high (more than 500 mU/mL) endogenous erythropoietin levels. However, the apparent effect of epoetin alfa was not observed solely in the low endogenous erythropoietin group. The difference in mean hemoglobin over the course of the study in the epoetin alfa versus placebo recipients reached statistical significance only in the more than 500 mU/ml endogenous erythropoietin group.

Other studies have shown that GM-CSF can increase leukocyte counts in patients with MDS.14-16 In agreement with these findings, our study indicated that in anemic and neutropenic patients with MDS, treatment with GM-CSF, with or without concurrent epoetin alfa, increased the ANC approximately 3-fold. An ANC increase of this magnitude may help overcome infection. Moreover, GM-CSF, with or without epoetin alfa, had no apparent effect on mean platelet counts.

Eighteen patients withdrew from the study. Withdrawal was for disease-related reasons in 8 of them. Withdrawal from the protocol for disease-related reasons in 8 of 66 patients is not surprising in this population of neutropenic, transfusion-dependent patients with MDS. The other 10 patients who withdrew did so because of side effects, including thrombocytopenia, myalgia, skin erythema, fever, and pericarditis, that were thought to be drug related. The occurrence of myalgia, skin erythema, fever, and pericarditis was likely to be GM-CSF-related because they have been reported in other studies of GM-CSF treatment and they not common side effects of epoetin alfa. The thrombocytopenia observed was likely related to both GM-CSF and epoetin alfa therapy. The 3 cases of treatment withdrawal because of thrombocytopenia occurred in the GM-CSF + epoetin alfa group, whereas there were no cases of treatment withdrawal because of thrombocytopenia in the GM-CSF + placebo group. Six patients had documented enlargement of the spleen during the study, a side effect that has previously been reported with GM-CSF therapy. The development of splenomegaly may contribute to worsening thrombocytopenia in this patient population.

In conclusion, combination therapy with GM-CSF and epoetin alfa offers benefit over GM-CSF alone with respect to maintenance of hemoglobin levels and reduction of transfusion requirements in patients with MDS who have less than or equal to 500 mU/mL baseline endogenous erythropoietin levels. GM-CSF, with or without epoetin alfa, can also substantially increase ANC in anemic, neutropenic patients with MDS. For transfusion-dependent patients with MDS who are not candidates for bone marrow transplantation, a therapeutic trial of GM-CSF and epoetin alfa for 3 months may be indicated to determine the response to cytokine therapy.