G-CSF plus azacitidine versus azacitidine alone for patients with high-risk myelodysplastic syndrome: academic, open label, randomized trial

Trial registration GA trial is registered under EudraCT#: 2013-001639-38.


Supplementary Material SM2: Trial design and End points
The study titled "Contribution to the verification of the efficacy of adding granulocyte colonystimulating factor (G-CSF) to 5-Azacitidine therapy in patients with high-risk myelodysplastic syndrome" was approved by the State Institute for Drug Control in the Czech Republic on December 6, 2016 under EudraCT#: 2013-001639-38 in accordance with ethical standards and the Helsinki Declaration of 1975 (as revised in 2008). The institutional ethics committee approved the study (#263/13S-IV) on April 21, 2016. The study included patients with higher risk MDS (Int-2 and high risk according to IPSS criteria (1)), MDS/AML with less than 30% MB and CMML II (according to 2016 WHO criteria), all > 65 years of age, not eligible for allo-HSCT or intensive chemotherapy and not previously treated with HMA. The dose and schedule of AZA administration were standard: 75 mg/m2 for seven days (5+2+2) subcutaneously within a 28-day cycle. G-CSF (Filgrastim, an unpegylated, recombinant cytokine, Neupogen from Amgen or Zarzio from Sandoz GmbH) was administered subcutaneously on day -2 (2 days before the 1st dose of AZA) and day +6 (the day following the 5th dose of AZA) at a dose of 30 MIU in patients with a body weight below 80 kg or 48MIU in patients with a body weight above 80 kg. The application schedule for the GA arm is shown in the figure below: The main objectives were to prolong patient survival, to prolong the transformation to AML, to reduce the incidence of infections and hospitalisation time, and to achieve transfusion independence. Secondary objectives were to evaluate the association of clinical and genetic (somatic variants) variables within a specific study arm. Each patient signed an informed consent form and consented to DNA analysis by sequencing. Exclusion criteria were previous HMA treatment, eligibility for allo-HSCT, adherence to contraceptive measures, known hypersensitivity to study drugs, and failure to follow study guidelines. The full study protocol is available on written request. The study is academic, which means that the academic institutions paid the salaries and wages of the investigators and the insurance of the patients. The molecular studies were sponsored in part by the grants listed in the Acknowledgments section. All authors had full access to all data in the study and accept responsibility for submitting it for publication.

Supplementary Material SM3: Efficacy and safety assessments
Complete blood count (CBC) and blood biochemistry were monitored in each cycle. Restaging analysis and response assessment included bone marrow collection for cytology, flow cytometry and cytogenetic evaluation every four cycles of AZA. BM analysis was also performed when progression was suspected, such as new cytopenias, new need for transfusion, or accumulation of MB in peripheral blood. Bone marrow aspirate analysis included flow cytometry, cytology and cytogenetic evaluation. Histological examination was performed at the time of diagnosis from the trephine biopsy. IPSS-R was determined before initiation of AZA treatment. Randomization was performed by lottery. Adherence to AZA therapy with G-CSF at d-2 and d+6 was designated as arm GA and adherence to standard AZA therapy was designated as arm A. Discontinuation of therapy was allowed at disease progression to AML or at patient request. Response was defined according to the IWG 2006 criteria (2), and in the case of this study, response was defined as either hematological improvement (HI), partial (PR), or complete remission (CR). Clinical and laboratory monitoring of blood counts and baseline biochemistry (urea, creatinine, bilirubin, liver function tests and c-reactive protein) was performed weekly during the first 2 cycles of AZA and monthly after the 3rd cycle of AZA. For grade 4 cytopenias, AZA administration was delayed until recovery. Reduction of the AZA dose to 75% was allowed if the CBC had not recovered after 2 weeks of deferral. The occurrence of febrile neutropenia or any hospitalization was recorded as a serious adverse event (SAE).

Supplementary Material SM4:
Description of the study cohort. Parameters in the left column represnt numbers of patients (N), age, Male/Female ratio, MDS subtype, and tMDS for GA vs A arm.  (3). Such a stochastic framework is extended (4) such that OS and ordered multilevel treatment response are modeled together.

Supplementary Material SM7: Sample size calculation
The sample size calculation is based on power analysis to compare survival curves between the two arms (GA vs A) according to the Cox proportional hazards model for clinical (randomized) trials, which is an implementation of a previously described sample size calculation method (5). This method was originally proposed by Freedman (6). The power to detect the size of the risk ratio is assumed to be 0.90, the significance level (Type I error rate) is assumed to be 0.05, the proportion of patients in the GA arm versus the A arm is assumed to be 1.30, the probability of failure in the GA arm over the maximum study duration is assumed to be 0.30, the probability of failure in the A arm over the maximum study duration is assumed to be 0.45, and the postulated risk ratio is assumed to be 0.25. Then the expected number of patients enrolled in the GA arm is 40, while the expected number of patients in the A arm is 31.

Supplementary Material SM12: Molecular analyses
Upon approval by the ethics committee and informed consent signed by each patient, it was feasible to sequence genomic DNA from BM mononuclear cells by next-generation sequencing (NGS) using the MDS NEB-Next 33 gene panel (New England Biolabs, Ipswich, MA, USA), which contains 33 frequently mutated MDS genes (7). NGS data were processed using an "in house" approach employing data quality assessment (FastQC, FastQ screen, multiQC). After removal of unwanted artificial sequences, mapping to the genome (GRCh38) was performed using alignment tools (BWA MEM). VarScan2 software was used to detect variants (8).
To ensure the quality of variant detection, we only consider a region with a variant coverage of more than 50 reads per base with a variant frequency greater than 10% as variant detection. Subsequently, the identified variants were annotated using the dbSNP and ClinVar databases and their effect on the produced protein was evaluated using the SNPEff tool. A standard FISH-MDS panel was used to assess chromosomal aberrations.

Supplementary Material SM13: G-CSF and its administration in both arms.
In particular, for febrile neutropenia, 21 (54%) patients in the GA arm received an average of 4 doses of G-CSF over and above the 2 planned doses of G-CSF. Similarly, in arm A, 6 (20%) patients received a median of 2 G-CSF injections per AZA cycle in the case of febrile neutropenia. While the difference in total G-CSF consumption was indeed statistically different (p<0.0001, t test), G-CSF consumption for febrile neutropenia was similar between the two arms (p=0.41, t test). In the first year (or 12 cycles), the total number of G-CSF administrations was 968 vs 235 and in the second year (>12 cycles of therapy) 185 vs 71. The relative number (%) of febrile complications requiring antibiotic therapy in the first year was 54% vs 45% whereas in the second year febrile complications were 27% vs 25%.