High response rate and improved exercise capacity and quality of life with a new regimen of darbepoetin alfa with or without filgrastim in lower-risk myelodysplastic syndromes: a phase II study by the GFM
- First Online:
- Cite this article as:
- Kelaidi, C., Beyne-Rauzy, O., Braun, T. et al. Ann Hematol (2013) 92: 621. doi:10.1007/s00277-013-1686-4
- 839 Views
Darbepoetin (DAR), with or without granulocyte colony-stimulating factor (G-CSF), has proved effective in treating anemia in patients with lower-risk myelodysplastic syndrome (MDS), but its effects on quality of life (QoL) and exercise functioning are less well established. In this phase II study (no. NCT00443339), lower-risk MDS patients with anemia and endogenous erythropoietin (EPO) level <500 IU/L received DAR 500 μg once every 2 weeks for 12 weeks, with G-CSF added at week 12 in non-responders. Physical performance was assessed with the 6-min walking test and, for fit patients, maximal oxygen consumption (VO2max). QoL was evaluated using SF-36 and FACT-An tests. In 99 patients, erythroid response rate according to IWG 2006 criteria was 48 and 56 % at 12 and 24 weeks, respectively. Addition of G-CSF rescued 22 % of non-responders. In 48 % of the responders, interval between darbepoetin injections could be increased for maintenance treatment. Serum EPO level was the only independent predictive factor of response at 12 weeks, and its most discriminant cutoff value was 100 IU/L. QoL and VO2max showed improvement over time in responders, compared with non-responders. With a median follow-up of 52 months, median response duration was not reached, and 3-year cumulative incidence of acute myeloid leukemia and overall survival (OS) was 14.5 and 70 %, respectively. Baseline transfusion dependence, International Prognostic Score System (IPSS), and Revised IPSS accurately predicted OS from treatment onset. Tolerance of darbepoetin was good. In conclusion, this regimen of darbepoetin every 2 weeks yielded high response rates and prolonged response duration. Objective improvement in exercise testing and in patient-reported QoL confirms the clinical relevance of anemia correction with erythropoiesis-stimulating agents.
KeywordsDarbepoetinMyelodysplastic syndromesAnemiaQuality of lifeExercise capacity
Myelodysplastic syndromes (MDS) are clonal bone marrow stem cell disorders characterized by ineffective hematopoiesis resulting in blood cytopenias and by frequent acute myeloid leukemia (AML) transformation [1, 2]. According to the International Prognostic Score System (IPSS), MDS are grouped in lower-risk (IPSS low and intermediate-1) and higher-risk disease (IPSS intermediate-2 and high) . While AML transformation is the main concern in higher-risk MDS, cytopenias, mostly anemia, represent the main problem in lower-risk MDS patients. The role of anemia in MDS symptoms (e.g., fatigue, dyspnea, angina pectoris) and its impact on quality of life are well established . The effect of anemia on exercise capacity of MDS patients and changes of this capacity with treatment of anemia, remains, however, less well studied .
Erythropoiesis-stimulating agents (ESA), with or without granulocyte colony-stimulating factor (G-CSF), reproducibly induce erythroid response rates of 40 to 50 % in lower-risk MDS patients with anemia, with a median response duration of 24 months [6–12]. Established predictive factors of response include lower serum erythropoietin (EPO) level and limited or no transfusion burden . Previous phase II trials demonstrated the efficacy of darbepoetin (DAR) alfa dosed at 150–300 μg/week [5, 13–19]. We tested in this trial the efficacy of a modified dosage of darbepoetin alfa of 500 μg every 2 weeks with or without G-CSF. This alternative schedule, with less frequent injections, was chosen to improve patient compliance and comfort and reduce healthcare resource utilization while maintaining almost similar weekly dose, compared to our previous experience . Efficacy was evaluated in terms of erythroid response, quality of life, and exercise function (measured by VO2max and the 6-min walking test) at baseline and under treatment.
Patients and methods
This phase II clinical trial (Clinicaltrials NCT00443339) was conducted in 17 centers of the Groupe Francophone des Myélodysplasies (GFM) in France from November 2006 to January 2009. Darbepoetin and filgrastim were provided by Amgen (Thousand Oaks, CA, USA) which, however, did not participate in the study design and in data reporting and analysis. The study protocol was approved by an ethical committee (Comité de Protection des Personnes, n°10, Aulnay sous Bois, France).
Eligibility criteria included (a) MDS according to WHO 2001 classification; (b) anemia, with hemoglobin level <10 g/dL with or without red blood cell (RBC) transfusion need; (c) IPSS low or intermediate-1; (d) serum EPO level lower than 500 IU/L; and (e) ability to carry physical function tests. Patients with chronic myelomonocytic leukemia (CMML) were eligible if they had <10 % bone marrow blasts and WBC count ≤13 G/L. Patients with any of the following conditions were excluded: treatment-related MDS; uncontrolled arterial hypertension; congestive heart failure, arrhythmia; serum creatinine level ≥120 % upper normal value; nutritional deficiency-related anemia or additional causes of anemia (hemolysis, hemorrhage, hemoglobinopathy); pregnancy; history of seizures or of a thrombotic event; and treatment with an ESA in the previous 8 weeks. All patients gave written informed consent.
Patients received darbepoetin alfa 500 μg every 2 weeks subcutaneously for 12 weeks. Non-responders at 12 weeks were to continue darbepoetin at the same dosing for an additional 12 weeks, but with the addition of filgrastim, initially dosed at 300 μg twice weekly and subsequently adjusted to maintain WBC counts between 5 and 10 G/L. Darbepoetin was definitively discontinued after 24 weeks of treatment in non-responders to this combination.
In responders, darbepoetin dosing was adjusted to maintain hemoglobin levels between 11 and 12 g/dL. According to the adjustment algorithm, darbepoetin was discontinued if Hb raised above 12 g/dL, until Hb level fell below 11 g/dL, and then resumed by increasing intervals by 1 week between injections. The study duration was 52 weeks, after which patients still responding could continue treatment. RBC transfusions were to be made according to ANSM (French Health authority) recommendations at hemoglobin level ≤8 g/dL or at higher levels if dictated by limited cardiopulmonary reserve or severe symptoms .
Follow-up and assessment of outcomes
The primary endpoint was the proportion of patients with erythroid response according to IWG 2006 criteria after 12 weeks of treatment . Secondary endpoints were erythroid response rate at 24 weeks (after addition of G-CSF in non-responders), safety, quality of life, physical functioning, and patient long-term outcome, including response duration, cumulative incidence of AML, and overall survival (OS). Progression was defined according to IWG 2006 criteria. The recently proposed Revised International Prognostic Score System (IPSS-R) was also assessed for erythroid response prediction and association with patient outcomes .
For physical functioning assessment, maximal and/or submaximal exercise tests were performed, respectively VO2max, for physically fit patients, and a 6-min walking test (measuring the distance walked in 6 min) in patients incapable of performing maximal exercise testing [23, 24]. The Short Physical Performance Battery test, which consists in three simple tasks (walking speed through a 4-m distance, time to perform five chair stands and equilibrium), was used in frail patients, unable to undergo either exercise test, e.g., VO2max or the 6-min walking test . VO2max measurement was made on a cycle ergometer, according to the French Society of Sport Medicine protocol . Symptom-limited VO2max was measured instead of VO2max when only a peak VO2 value, rather than the maximal theoretical value, could be reached by the patient. The Borg scale was used to assess dyspnea. Contra-indications to VO2max measurement were a history of coronary artery disease or cardiac failure, tachycardia or hypertension on the day of testing, a physical handicap precluding cycling (rheumatologic disease, peripheral artery disease, respiratory insufficiency), and hemoglobin level <8 g/dL. In patients with Hb < 8 g/dL, baseline VO2 max measurement was made approximately at the midpoint between consecutive RBC transfusions. A cardiology visit with echocardiography including left ventricular mass measurement was systematically performed before each VO2max measurement.
Quality of life was assessed by the SF-36 and the Anemia version of the Functional Assessment of Cancer Therapy (FACT-An) questionnaires [27, 28]. Physical functioning and quality of life assessments were scheduled at baseline and after 12 and 24 weeks of treatment.
A sample size of 99 patients was chosen to provide a power of 95 % to detect a 10 % increment in response rate, as compared with response rates of about 50 % in previous studies with darbepoetin alfa. Statistical analysis was based on the intention-to-treat principle. Safety was considered in patients who received at least one dose of treatment. Quality of life and physical functioning were analyzed in those patients who filled in at least one questionnaire or performed at least one of the physical tests at baseline and during follow-up.
Patient characteristics and response rates were compared using Fisher’s exact test. Logistic regression was used for multivariate analysis of putative predictors of response. Time from diagnosis to onset of treatment was not pre-defined as a model covariate because it was considered as correlated with other variables like age, disease severity, and time to referral and therefore potentially misleading. Response duration was calculated from time of response achievement, assessed either at 12 or 24 weeks. Response duration and OS were measured by the Kaplan–Meier method . OS differences between patient groups were compared using the log-rank test . Cumulative incidences (CI) of AML transformation were estimated taking into account the competing risk of death and compared using the Gray test . Repeated-measures analysis of variance was used to test the within-subject and between-subject differences of quality of life and physical functioning scores, with time as the within-subject factor (from baseline to weeks 12 and 24) and response to darbepoetin as the between-subject factor (responders vs. non-responders). All calculations were performed using R version 2.8.1.
Baseline patient characteristics
N = 95
% or median, IQR
% bone marrow blasts
Isolated del 20q
Isolated del 5q
Cytogenetic risk (IPSS)
Cytogenetic risk (IPSS-R)
Serum EPO level
% RBC transfusion dependent
Nb of RBC units transfused during the previous 6 months (in transfused patients)
Median ferritin level (ng/mL)
Forty-six of the 95 patients (48 %) (95 % confidence intervals 38–58 %) reached the primary endpoint of erythroid response at 12 weeks of treatment, according to IWG 2006 criteria. Forty (82 %) of the 49 non-responders at 12 weeks had addition of filgrastim to darbepoetin alfa, and nine of them (22 %) were responders after 12 weeks of combined treatment, increasing the overall response rate to 56 % at 24 weeks. According to the previously used IWG 2000 criteria , the response rate was 60 and 59 % at 12 and 24 weeks of treatment, and major responses were seen in 41 and 49 % of all cases at 12 and 24 weeks of treatment, respectively.
Median time to response was 5 weeks (range 1–20), including eight patients without baseline transfusion requirement who reached response criteria (i.e., increase in Hb level by at least 1.5 g/dL) within the first 2 weeks of treatment. Hemoglobin level exceeded at some point 12 g/dL in 59 % of the responders. Three patients had an increase in Hb greater than 2 g/dL after the first darbepoetin injection, reaching a Hb level of 11.6, 12.3, and 12.8 g/dL, respectively, 2 weeks after the first injection. Serum EPO level was <100 IU/L in all of them. None of these patients had reduced renal function. No clinical adverse events were associated with those rapid early responses. During maintenance treatment, the interval between darbepoetin injections of 500 μg was greater than 2 weeks in 48 % of the responders, range 2.4–20 weeks, with nine patients requiring injections every 12 weeks or more.
Prognostic factors of response (according to IWG 2006 criteria) at 12 and 24 weeks of treatment by univariate analysis
Response rate (%)
Response rate (%)
% BM blasts
Cytogenetic risk (IPSS)
Cytogenetic risk (IPSS-R)
Serum EPO level
Nb RBC transfused during the previous 6 months
The proportion of non-responders rescued by G-CSF did not differ with WHO classification, IPSS, IPSS-R, and baseline serum EPO level. However, none of the ten non-responders with marrow blasts >2 % responded to G-CSF addition vs. eight of the 13 (39 %) non-responders with marrow blasts ≤2 % (P = 0.01).
Quality of life
Physical Functioning and Bodily Pain, which evaluate physical health, and Vitality, reflecting both physical and mental health, were the SF-36 scales whose evolution over time was best correlated with erythroid response (mean difference over time in responders vs. non-responders for Physical Functioning 9.6 vs. −11.2, P = 0.0002 for the interaction between group and time; Bodily Pain 7.9 vs. −7.8, P = 0.04; Vitality 11.1 vs. −7.1, P < 0.0001) (Fig. 1a). By contrast, scales evaluating mental health were not significantly correlated with erythroid response. Likewise, the Physical Component Summary was improved over time in responders, as compared with non-responders (mean difference 3 vs. 0.4, P = 0.04), but not the Mental Component Summary (mean difference 3.5 vs. 0.7, P = 0.23).
For FACT-An score, there was a steady improvement of all scales over time in responders as compared with non-responders, i.e., in FACT-General (mean difference over 6 months in responders vs. non-responders 4.1 vs. −5.6, P = 0.007), FACT-An Trial Outcome Index, a composite score of Physical Well-Being, Functional Well-Being and Anemia subscales (mean difference over 6 months in responders vs. non-responders 14.4 vs. −5.5, P = 0.001), and FACT-Anemia total score (mean difference over 6 months in responders vs. non-responders 13 vs. −8.1, P = 0.002) (Fig. 1b).
VO2max was serially measured in 15 patients, nine responders and six non-responders. The difference over time between responders and non-responders was significant (mean difference between baseline and 6 months of 212.1 and −89.5 mL/min in responders and non-responders, respectively, P = 0.01 for the interaction between group and time), with VO2max increasing in responders and decreasing in non-responders (Fig. 2b). Echographically measured left ventricular mass index and ejection fraction did not show any significant modification over time in either group of patients (data not shown).
Response duration, progression to AML, and survival
The 3-year CI of AML was 14.5 %. Three-year OS was 70 %, and median OS was not reached (Fig. 4b, c). The 3-year mortality considered related to MDS (i.e., by hemorrhage, infection, AML transformation) was 9.9 % in responders and 27.9 % in non-responders whereas no differences between responders and non-responders were found for other causes of death (cardiovascular, neoplasia, others) (not shown).
Serious adverse events included non-fatal pulmonary embolism, non-fatal stroke, and coma from unknown cause leading to death, reported in one patient each (all responders) after 10, 6, and 9 months of treatment, respectively, with corresponding Hb levels of 11.5, 12.6, and 11.5 g/dL. One fatal stroke was seen in a patient with severe cardiovascular comorbidity, 36 days after the first injection of darbepoetin. In that patient, treatment had been discontinued for injection-site reaction and the last Hb value before death was <10 g/dL. Other adverse events included rash, injection-site reactions, and filgrastim-related arthralgia/myalgia and hyperleucocytosis in 3, 2, 3, and 1 patient, respectively. One patient withdrew his consent.
Studies with darbepoetin in MDS
Type of study
Nb of patients
% transfused patients
Response rate: IWG 2000 (%)
Response rate: IWG 2006 (%)
Musto, 2005 
Stasi, 2005 
Giraldo, 2006 
Mannone, 2006 
Gabrilove, 2006 
500 μg/3 weeks
Gotlib, 2009 
250–1,100 μg/week (median 390 μg/week)
Oliva, 2010 
Villegas, 2011 
Nilsson-Ehle, 2011 
500 μg/2 weeks
Serum EPO level was the only independent predictive factor of response at 12 weeks of treatment, as in previous studies. The “best” threshold serum EPO level for predicting response to ESAs is often disputed, varying from 100 to 500 IU/L in various series. In the present series, 100 IU/L was the most discriminant one, and the response rate was as low as 30 % in patients with a level ≥100 IU/L. On the other hand, in our recent experience, 60 % of patients with lower risk MDS and hemoglobin ≤10 g/dL had a serum EPO level <100 IU/L , whereas that percentage exceeded 80 % in newly diagnosed patients included in the EU low risk registry (T. De Witte, unpublished data), suggesting that a majority of lower risk MDS with anemia may be candidates to ESAs as first-line treatment.
We found that early relapses occurred in patients with RAEB-1 and with IPSS-R high. Conversely, all but two of the 11 patients with very low IPSS-R were responders at 24 weeks.
The percentage of non-responders rescued by the addition of G-CSF (22 %) was relatively small and similar to that of our previous study on darbepoetin dosed at 300 μg weekly (20 %) . Another work suggested that G-CSF addition is less effective when the ESA dose is high . On the other hand, the addition of G-CSF in the Stanford study yielded response in 47 % of cases, essentially in RARS . By contrast, in our study, RARS and RCMD-RS did not have lower response rates to DAR alone, and did not particularly benefit from the addition of G-CSF (although regarding the latter point, patient numbers precluded any answer).
Median response duration, not reached after a median follow-up of more than 4 years, was unexpectedly long. Response duration had not been documented beyond 1 year of follow-up in other registered studies using high-dose darbepoetin, with the exception of the Stanford study where median response duration was 19 months, but in a relatively small series. It is currently unknown whether darbepoetin, with its inherent pharmacokinetic characteristics, could induce longer responses than EPO α or β.
The cumulative incidence of AML was approximately 15 % and median overall survival not reached after a median of 4.5 years of follow-up, consistent with the low-risk profile of the population. In particular, patients still responding after 2 years of treatment had a particularly favorable subsequent outcome (only one death out of 33 patients). Non-responders had an excess of mortality mainly attributed to MDS-related causes. Two retrospective studies and one prospective study showed that responders to ESAs have better long-term outcomes than non-responders [9–11]. Those results could also suggest that response to ESAs selects patients with intrinsically better prognosis, whereas non-responders are deemed to have more MDS-related complications, especially AML transformation.
Anemia is commonly associated with fatigue and impaired quality of life and exercise capacity. However, improvement of quality of life by ESAs in anemia due to MDS has been disputed . Here, using two different tests (i.e., SF-36, the most widely used psychometric tool for quality of life across disciplines, and a general oncological test, FACT, in its general and anemia-related versions), we found increasing scores over time of approximately 10 points for physical components of SF-36 in responders vs. decreasing scores in non-responders, whereas the increase was more modest for mental components. Likewise, increase in anemia-specific components of FACT exceeded 10 points in responders as compared with inverse time-score relationships in non-responders. A quality of life improvement using FACT-An was also found by Spiriti et al. in responders compared to non-responders and in a randomized trial between treated and non-treated patients (though without significant difference) [8, 11]. On the other hand, the absence of significant improvement in quality of life in the treated groups of a randomized GFM and a phase II Nordic MDS group trial of EPO alpha was possibly due to small patient numbers and “dilution” of the effect in responders by that of non-responders [5, 7]. Very intensive RBC transfusion support in the control arm, not allowing Hb level to fall below 12 g/dL, but rarely achieved in routine practice, could also have explained the lack of significant difference in the Nordic study . In other situations, the durable rise in hemoglobin level obtained in responders to ESAs may improve quality of life compared to variable hemoglobin levels associated to repeated RBC transfusion.
The increase of blood oxygen-carrying capacity enhances maximal oxygen uptake. We could demonstrate using VO2max, despite the small number of patients able to undergo serial measures of effort spirometry, that, in parallel with quality of life, physical effort capacity was improved in responders as compared with non-responders despite regular RBC transfusions in the latter group. In the Nordic group study, using an intensive transfusion program, there was no specific pattern in maximal oxygen uptake with hemoglobin maintained ≥12 g/dL . However, the significant difference over time between responders and non-responders in our study suggests that improving hemoglobin values by transfusion alone in routine practice may be insufficient to induce sustained increase in cardiorespiratory fitness. In contrast, the 6-min walking test and the short battery physical performance test proved less sensitive in this non-institutionalized elderly population, selected on the basis of being able to undergo at least some exercise tests. Whether those widely used geriatric tests can be effectively used by large MDS patient numbers merits further investigation.
As in previous studies, darbepoetin was well tolerated. Venous and arterial thromboembolic events attributable to darbepoetin have been reported, though using either higher doses or with a hemoglobin target higher than 12 g/dL. Close monitoring of hemoglobin values according to current recommendations should guarantee safe delivery of high dose schedules.
In the present study, responses occurred within the first 2 weeks of treatment in 20 % of the responders, with Hb increasing above 12 g/dL after the first darbepoetin injection in three patients. None of these patients had reduced renal function. Although there is no formal contraindication to use high dose ESA in elderly MDS patients with reduced renal function, extra caution is necessary with respect to abrupt Hb increases (>2 g/dL in 2 weeks) and dose adjustment rules in such patients.
In conclusion, darbepoetin 500 μg every 2 weeks ±G-CSF was an effective and safe induction regimen for anemia in lower-risk myelodysplastic syndromes, associated with favorable long-term clinical outcomes in responders, including patient-reported quality of life and objectively measured exercise capacity. Alternative treatments should be rapidly considered in non-responders due to excessive MDS-related mortality.
Conflict of interest
The authors declare no conflict of interest.