Franz Hutzschenreuter, Ina Monsef, Karl‐Anton Kreuzer, Andreas Engert, Nicole Skoetz

Myelodysplastic syndromes (MDS) are a heterogeneous group of haematological diseases which are characterised by a uni- or multilineage dysplasia of haematological stem cells. Standard treatment is supportive care of the arising symptoms including red blood cell transfusions or the administration of erythropoiesis-stimulating agents (ESAs) in the case of anaemia or the treatment with granulocyte (G-CSF) and granulocyte-macrophage colony stimulating factors (GM-CSF) in cases of neutropenia. The objective of this review is to assess the evidence for the treatment of patients with MDS with G-CSF and GM-CSF in addition to standard therapy in comparison to the same standard therapy or the same standard therapy and placebo. We searched MEDLINE (from 1950 to 3 December 2015) and CENTRAL (Cochrane Central Register of Controlled Trials until 3 December 2015), as well as conference proceedings (American Society of Hematology, American Society of Clinical Oncology, European Hematology Association, European Society of Medical Oncology) for randomised controlled trials (RCTs). Two review authors independently screened search results. We included RCTs examining G-CSF or GM-CSF in addition to standard therapy in patients with newly diagnosed MDS. We used hazard ratios (HR) as effect measure for overall survival (OS), progression-free survival (PFS) and time to progression, and risk ratios for response rates, adverse events, antibiotic use and hospitalisation. Two independent review authors extracted data and assessed risk of bias. Investigators of two trials were contacted for subgroup information, however, no further data were provided. G-CSF and GM-CSF were analysed separately. We screened a total of 566 records. Seven RCTs involving 486 patients were identified, but we could only meta-analyse the two evaluating GM-CSF. We judged the potential risk of bias of these trials as unclear, mostly due to missing information. All trials were randomised and open-label studies. However, three trials were published as abstracts only, therefore we were not able to assess the potential risk of bias for these trials in detail. Overall, data were not reported in a comparable way and patient-related outcomes like survival, time to progression to acute myeloid leukaemia (AML) or the incidence of infections was reported in two trials only.Five RCTs (N = 337) assessed the efficacy of G-CSF in combination with standard therapy (supportive care, chemotherapy or erythropoietin). We were not able to perform meta-analyses for any of the pre-planned outcomes due to inconsistent and insufficient reporting of data. There is no evidence for a difference for overall survival (hazard ratio (HR) 0.80, 95% confidence interval (CI) 0.44 to 1.47), progression-free survival (only P value provided), progression to AML, incidence of infections and number of red blood transfusions (average number of 12 red blood cell transfusions in each arm). We judged the quality of evidence for all these outcomes as very low, due to very high imprecision and potential publication bias, as three trials were published as abstracts only. Data about quality of life and serious adverse events were not reported in any of the included trials.Two RCTs (N = 149) evaluated GM-CSF in addition to standard therapy (chemotherapy). For mortality (two RCTs; HR 0.88, 95% CI 0.62 to 1.26), we found no evidence for a difference (low-quality evidence). Data for progression-free survival and serious adverse events were not comparable across both studies, without evidence for a difference between both arms (low-quality evidence). For infections, red blood cell and platelet transfusions, we found no evidence for a difference, however, these outcomes were reported by one trial only (low-quality evidence). Time to progression to AML and quality of life were not reported at all.Moreover, we identified two cross-over trials, including 244 patients and evaluating GM-CSF versus placebo, without publishing results for each arm before crossing over. In addition, we identified two ongoing studies, one of which was discontinued due to withdrawal of pharmaceutical support, the other was terminated early, both without publishing results. Although we identified seven trials with a total number of 486 patients, and two unpublished, prematurely finished studies, this systematic review mainly shows that there is a substantial lack of data, which might inform the use of G-CSF and GM-CSF for the prevention of infections, prolonging of survival and improvement of quality of life. The impact on progression to AML remains unclear.