Comparative efficacy and safety of four immunomodulatory agents for acute spinal cord injury: a systematic review and Bayesian network meta-analysis.

STUDY

DESIGN

Systematic review and Bayesian network meta-analysis.

OBJECTIVE

To compare the efficacy and safety of methylprednisolone (MP), erythropoietin (EPO), granulocyte colony-stimulating factor (G-CSF), and minocycline for the treatment of acute spinal cord injury (ASCI).

METHODS

MEDLINE, Embase, the Cochrane Library, and ClinicalTrials.gov were searched from inception to 15 June 2025. Randomized and nonrandomized comparative studies were eligible. Bayesian random-effects network meta-analyses compared MP, EPO, EPO plus MP, G-CSF, and minocycline. Treatment effects were reported as odds ratios (ORs) or mean differences (MDs) with 95% credible intervals (CrIs). Model convergence and fit, heterogeneity, predictive intervals, and global and local inconsistency were evaluated. Follow-up-stratified and exploratory restriction analyses were performed where feasible. Formal comparison-level Confidence in Network Meta-Analysis (CINeMA) assessment was not feasible; certainty was appraised qualitatively across its six domains.

RESULTS

Twenty comparative reports were included: nine randomized and 11 nonrandomized reports. For neurological grade improvement, G-CSF was associated with higher odds of improvement than the common control node (OR = 5.12, 95% CrI 1.15-32.42). Estimates for EPO (OR = 4.87, 95% CrI 0.44-135.90), EPO plus MP (OR = 3.44, 95% CrI 0.09-164.02), and MP (OR = 1.31, 95% CrI 0.30-5.72) were imprecise and included the null. For change in ASIA Motor Score, G-CSF had the largest point estimate (MD = 11.49, 95% CrI -0.44 to 23.49), followed by minocycline (MD = 6.83, -15.24-28.75), EPO (MD = 6.75, -13.16-26.42), and MP (MD = 2.28, -6.24-10.94); all credible intervals crossed zero. For overall adverse events, the estimate for MP was imprecise (OR = 1.13, 95% CrI 0.58-2.34), while the G-CSF effect was not reliably estimable because near-complete event occurrence produced a highly prior-sensitive posterior. Other safety estimates were imprecise.

CONCLUSION

G-CSF showed a potential beneficial signal for neurological grade improvement. However, the limited evidence base, inclusion of nonrandomized and nonconcurrent comparisons, substantial heterogeneity, wide credible and predictive intervals, and uncertain transitivity preclude definitive conclusions about comparative efficacy or safety. Treatment rankings should be regarded as exploratory and do not establish the superiority of any intervention.

SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/, identifier CRD420251027856.