Spontaneous obliteration of cerebral arteriovenous malformations: a meta-analysis of incidence and angioarchitectural correlates.

Spontaneous angiographic obliteration of brain arteriovenous malformations (SpOAVM) is rare, and published incidence estimates are limited by heterogeneous denominators and variable imaging follow-up. We performed a meta-analysis to estimate the incidence of SpOAVM and to describe associated angioarchitectural features. Literature from 1974 to January 2026 was systematically reviewed in accordance with PRISMA. Observational studies reporting complete spontaneous angiographic obliteration of intracranial arteriovenous malformations (AVMs), without prior AVM-directed therapy, were included. The meta-analysis used a linear mixed model with a rare-event-appropriate log-log link. Denominator definition, imaging modality, follow-up duration, reappearance after apparent obliteration, and angioarchitectural variables were extracted. Nine studies contributed 47 eligible SpOAVMs across 7,922 brain AVMs, although some denominators were mixed or estimated. In four cohorts with definitive no-intervention denominators, pooled SpOAVM incidence was 1.93% (95% confidence interval [CI], 1.04-3.44), corresponding to 0.39 per 100 person-years. Across all eligible studies, including cohorts with uncertain denominator purity, the pooled estimate was 1.08% (95% CI, 0.58-3.05). SpOAVMs most commonly followed hemorrhage (37/45, 82%). Exploratory external-cohort comparisons showed that SpOAVMs had a higher proportion of smaller nidus size and single venous drainage. Reappearance after apparent obliteration was reported in 2/35 cases. SpOAVM is rare and occurs far less frequently than hemorrhage during untreated follow-up. Apparent obliteration requires cautious interpretation because imaging modality and follow-up were heterogeneous. Standardized prospective angiographic surveillance is needed to define clinically relevant subgroups.