Abstract Introduction: Granulocyte colony-stimulating factor (G-CSF) is routinely used to reduce chemotherapy-induced neutropenia, yet its role as primary prophylaxis in acute myeloid leukemia (AML) remains controversial due to concerns of increased relapse risk. This study aims to clarify the safety and efficacy profile of G-CSF during induction therapy in adult AML patients through a comprehensive meta-analysis. Methods: A systematic review and meta-analysis were conducted using PubMed, Ichushi-Web, and the Cochrane Library to identify randomized controlled trials (RCTs) comparing G-CSF prophylaxis to placebo or standard care in AML. Data extraction and quality assessment were independently performed by two reviewers. Risk ratios (RR) with 95% confidence intervals (CI) were calculated using fixed-effects models. Nine RCTs were included, analyzing: Infection-related mortality (6 RCTs; n = 1,465), Disease progression/recurrence mortality (3 RCTs; n = 920), Adverse events (e.g., musculoskeletal pain) (2 RCTs; n = 365). Results: Infection-related mortality: No significant difference between G-CSF and control (RR = 0.94; 95% CI: 0.64–1.36; I² = 0%). Disease progression/recurrence mortality: No increased risk observed with G-CSF (RR = 1.07; 95% CI: 0.82–1.40; I² = 43%). Adverse reactions (musculoskeletal pain): Comparable between groups (RR = 1.09; 95% CI: 0.50–2.34; I² = 70%). Importantly, G-CSF significantly shortened the duration of neutropenia and hospitalization without compromising survival or increasing relapse risk. Conclusion: G-CSF primary prophylaxis during AML induction therapy is safe and effective. It does not adversely impact infection-related mortality, disease progression, or adverse events, while substantially reducing neutropenia duration. These findings strongly support the use of G-CSF in adult AML patients, especially those at high risk of infectious complications. G-CSF should be considered a standard supportive care measure in AML induction protocols to optimize clinical outcomes and resource utilization.
Sharma et al. (2025) studied this question.