Efficacy and safety of cell-based therapies for drug-resistant epilepsy: a systematic review of human studies

Drug-resistant epilepsy affects roughly one-third of people with epilepsy and carries substantial morbidity and mortality. Cell-based therapies are hypothesized to modulate neuroinflammation and network dysfunction. We systematically reviewed human studies to appraise efficacy and safety and to identify priorities for future trials. Following the PRISMA 2020 guideline, we searched PubMed, Embase, Scopus, Web of Science, and the Cochrane Library from inception to 31 October 2025, without language or year limits. Eligible designs were empirical studies evaluating any cell-based therapy given with therapeutic intent to patients with refractory/drug-resistant epilepsy. Risk of bias was assessed using design-specific tools (RoB 2 for randomized trials, ROBINS-I for non-randomized comparative studies, the NIH/NHLBI Before–After tool for single-arm pre-post studies, and Joanna Briggs Institute checklists for case series and case reports) and harmonized onto a common four-level scale using prespecified mapping rules. Given substantial clinical, methodological, and product-related heterogeneity, we used a structured narrative synthesis informed by the Synthesis Without Meta-analysis reporting conventions and graded certainty with GRADE. Of 2264 records screened, 21 full texts were assessed and 12 reports met eligibility. After an explicit audit of overlapping reports, two pairs were consolidated into their most complete publications, yielding 10 unique studies with 102 treated participants. Designs comprised one open-label randomized controlled trial, several prospective single-arm cohorts/case series, a historical case series, and four case reports. Interventions included autologous bone marrow-derived mesenchymal stromal cells (intravenous and/or intrathecal), intra-arterial bone marrow mononuclear cells, intrathecal CD271+ BM-MSCs, intrathecal adipose-derived regenerative cells, olfactory mucosa mesenchymal stromal cells, allogeneic embryonic neural tissue grafts, and allogeneic umbilical cord blood transplantation. Reported seizure reductions were derived predominantly from open-label, uncontrolled, single-arm data in which co-interventions were frequent and often dominant, limiting causal attribution to the cell product. Overall risk of bias was Serious or High in 9 of 10 studies. GRADE certainty was very low for all efficacy outcomes and for long-term safety, and low for early harms. Short-term tolerability appeared acceptable in small, selected samples, but harms were incompletely and unsystematically ascertained. Across heterogeneous, predominantly uncontrolled studies, reported outcomes suggest possible seizure, EEG, and patient-important benefits following cell-based therapy in DRE. These signals remain hypothesis-generating. Very low-certainty evidence precludes firm conclusions about effectiveness, and safety remains uncertain beyond short-term tolerability. Cell-based therapy should remain strictly investigational. Adequately powered, blinded, controlled trials with standardized outcome sets, preregistered analytic plans, manufacturing transparency, etiology-tailored routes and dosing, and standardized long-term pharmacovigilance are required.