Abstract / Summary
Febrile neutropenia (FN) is an important complication of first-line immunochemotherapy for diffuse large B-cell lymphoma (DLBCL). Thresholds for primary granulocyte colony-stimulating factor (G-CSF) prophylaxis are risk-based, but the FN burden in routine practice is uncertain, and whether trial-derived estimates describe it is unknown.
Systematic review and meta-analysis of proportions. MEDLINE, CENTRAL, ClinicalTrials.gov, Europe PMC, Epistemonikos, Lens.org, Google Scholar, and citation tracking were searched from inception to 8 September 2026. Eligible studies reported FN incidence in adults with DLBCL receiving first-line R-CHOP or R-CHOP-like therapy. Analyses were stratified by design, because observational cohorts and trials measure FN under materially different conditions. Risk of bias and certainty were assessed using ROBINS-I/RoB 2 and GRADE.
Fifteen studies (5440 patients; 885 FN events) were included. Across ten observational cohorts (4020 patients), pooled FN incidence was 19.0% (95% CI 17.1-21.1; I2 = 34.0%), prediction interval 15.3-23.3%, and was stable across all sensitivity and leave-one-out analyses. Five clinical trials (1,420 patients) reported 11.7% (95% CI 5.7-22.7) but were markedly heterogeneous (I2 = 79.4%; estimates 2.4-20.0%); the difference between designs was not significant (p = 0.051) and depended substantially on one trial.
In routine practice, approximately one in five adults with DLBCL receiving first-line R-CHOP-based therapy develops FN-an incidence at, rather than below, the conventional 20% threshold for primary G-CSF prophylaxis. Trial-derived estimates should not be assumed to represent this burden.