Abstract
Background/Objectives: Leucovorin (folinic acid) is used off-label for seizures across the cerebral folate deficiency (CFD) spectrum and is approved by the US Food and Drug Administration (FDA) for FOLR1-related cerebral folate transport deficiency. We synthesized patient-level seizure outcomes by etiology. Methods: PRISMA 2020 patient-level systematic review and Bayesian meta-analysis (PROSPERO CRD420261422063). PubMed/MEDLINE, Embase, CENTRAL and Web of Science were searched up to 15 August 2026. Two authors screened records independently; one author extracted patient-level outcomes and the second verified them against the source publications, and both appraised risk of bias (Cochrane RoB 2, JBI, Murad). A Bayesian hierarchical logistic model with publication-level random effects estimated cluster-conditional response probabilities per stratum. Results: The analytic pool comprised 38 publications and 79 patients across eight etiologic strata. The model-estimated, publication-cluster-conditional probability of any seizure reduction was highest in pyridoxine-dependent epilepsy/folinic acid-responsive seizures (0.94, 95% HDI 0.79–1.00) and FOLR1-related CFD (0.91, 0.80–1.00) and lowest in Rett syndrome and PNPO deficiency (both 0.31; 95% HDIs 0.00–0.78 and 0.00–0.88), with wide uncertainty in the smallest strata. No adverse event or death was attributed to leucovorin by the source authors, but safety reporting was incomplete, and the absence of reported attribution does not establish the absence of events. GRADE certainty was very low throughout. Conclusions: Seizure reduction was most consistent in FOLR1-related and PDE/FARS epilepsies and least in Rett syndrome and PNPO deficiency; evidence was insufficient in strata with ≤3 evaluable patients (mitochondrial, PCDH19, secondary/genetic). Prospective controlled trials are needed.