Acute Myeloid Leukemia or Myelodysplastic Syndromes and All-Cause Mortality in Randomized Controlled Trials of Cancer Patients Receiving Chemotherapy with or without Granulocyte Colony-Stimulating Factor: A Systematic Review and Meta-Analysis.

The risk of acute myeloid leukemia and myelodysplastic syndromes (AML/MDS) is increased following prior exposure to cytotoxic cancer chemotherapy and radiation therapy. A few retrospective studies have suggested a potential risk for AML/MDS in cancer patients also receiving granulocyte colony-stimulating factor (G-CSF). A systematic review and meta-analysis of randomized controlled clinical trials (RCTs) in patients receiving cancer chemotherapy ± initial G-CSF support was conducted to evaluate the risk of AML/MDS and impact on all-cause mortality.

Electronic databases including Medline, EMBASE, Cochrane Library and meeting proceedings were searched. Eligible studies included RCTs of solid tumor or lymphoma patients receiving chemotherapy ± initial G-CSF and ≥2 years follow-up reporting AML/MDS or second malignancies. Dual blinded data extraction was performed. After evaluating for heterogeneity, relative risk (RR) and absolute risk (AR) difference [95% confidence limits] were estimated by the method of Mantel and Haenszel.

In the 25 eligible RCTs, 6,058 patients were randomized to receive chemotherapy with and 6,746 without G-CSF support. Mean and median follow-up in all studies were 60 and 53 months, respectively. Among 23 RCTs reporting AML/MDS, 22 cases (0.36%) were reported in control- and 43 (0.79%) in G-CSF-treated patients. The RR and AR increase for AML/MDS in G-CSF-supported patients was 1.92 [1.19, 3.07; P=.007] and 0.41% [0.10%, 0.72%; P=.009], respectively. A trend suggesting publication bias was observed for AML/MDS consistent with an underreporting of small negative studies based on funnel plot asymmetry [Egger's regression intercept, P=.039]. No significant difference in reported AML/MDS was found based on source of study funding with RR for AML/MDS for industry and non-industry funded studies of 2.57 and 1.71, respectively. Trends toward greater risk of AML/MDS with higher cumulative chemotherapy dose were observed but none reached statistical significance. Mortality was reported in all 25 trials occurring in 1,845 patients randomized to receive G-CSF and in 2,099 controls. The RR and AR decrease for mortality in G-CSF-supported patients were 0.897 [0.857, 0.938; P<.0001] and 3.40% [2.01%, 4.80%; P<.0001], respectively. Greater RR reduction for mortality was seen for larger study sample size [P=.05]. Greater RR reduction for mortality was also observed with increasing planned [P=.016] as well as greater delivered relative [P=.015] and absolute [P=.027] chemotherapy dose intensity in patients receiving initial G-CSF support compared to controls.

Risk of AML/MDS is increased while all-cause mortality is decreased in patients receiving G-CSF-supported chemotherapy. Greater reductions in risk for all cause mortality were observed in larger studies and with greater planned and delivered chemotherapy dose intensity.

Lyman:Amgen: Research Funding, Speakers Bureau. Crawford:Amgen: Consultancy. Kuderer:Amgen: Husband receives research support. Dale:Amgen Inc.: Consultancy, Honoraria, Research Funding, Speaker.