Efficacy, safety, and certainty of AAV-based shortened dystrophin gene therapy in duchenne muscular dystrophy: a construct-specific systematic review and GRADE appraisal

Duchenne muscular dystrophy (DMD) has driven the development of adeno-associated virus (AAV)-based shortened dystrophin gene therapies, including approved delandistrogene moxeparvovec and discontinued fordadistrogene movaparvovec. A multi-construct synthesis incorporating both EMBARK and CIFFREO with construct-specific certainty assessment was lacking. This PROSPERO-registered systematic review with exploratory meta-analysis (CRD420261346035) searched PubMed, Scopus, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) to 24 March 2026. Eligible studies evaluated AAV-based shortened dystrophin constructs in DMD. Evidence was synthesized separately by construct, and cross-construct pooling was not performed. Exploratory random-effects meta-analysis was restricted to comparable randomized delandistrogene data. Certainty was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. Twelve reports representing seven clinical trial datasets and five companion reports were included. EMBARK did not meet its primary North Star Ambulatory Assessment (NSAA) endpoint (between-group difference 0.65 points, 95% confidence interval [CI] − 0.45 to 1.74; p = 0.2441). CIFFREO was also negative for NSAA (difference 0.09 points, 95% CI − 1.46 to 1.64; p = 0.91) and reported higher adverse-event and serious-adverse-event frequencies with fordadistrogene than placebo. Exploratory delandistrogene pooling identified a low-certainty signal for Time to Rise (mean difference − 0.65 s, 95% CI − 1.06 to − 0.24; I 2 = 0.0%), whereas pooled NSAA was not significant and 10-Meter Walk/Run and 100-Meter Walk/Run estimates were heterogeneous. Post-authorization fatal liver failure cases prompted a U.S. Food and Drug Administration (FDA) boxed warning and restriction of the delandistrogene indication. GRADE certainty ranged from moderate to very low across construct-specific randomized outcomes. Both phase 3 programs failed to demonstrate benefit on their primary functional endpoint. Current evidence supports biologic target engagement but does not establish consistent 1-year functional efficacy. These findings support construct-specific interpretation, longer follow-up, standardized functional assessment, and continued safety surveillance.