Hemorrhage from cerebral cavernous malformations: a systematic pooled analysis

OBJECTIVE The aim of this paper is to define an overall cavernous malformation (CM) hemorrhage rate and risk factors for hemorrhage. METHODS The authors performed a systematic, pooled analysis via the PubMed database through October 2015 using the terms "cavernoma," "cavernous malformation," "natural history," "bleeding," and "hemorrhage." English-language studies providing annual rates and/or risk factors for CM hemorrhage were included. Data extraction, performed independently by the authors, included demographic data, hemorrhage rates, and hemorrhage risk factors. RESULTS Across 12 natural history studies with 1610 patients, the mean age at presentation was 42.7 years old and 52% of patients (95% CI 49%-55%) were female. Presentation modality was seizure in 30% (95% CI 25%-35%), hemorrhage in 26% (95% CI 17%-37%), incidental in 17% (95% CI 9%-31%), and focal deficits only in 16% of cases (95% CI 11%-23%). CM location was lobar in 66% (95% CI 61%-70%), brainstem in 18% (95% CI 13%-24%), deep supratentorial in 8% (95% CI 6%-10%), and cerebellar in 8% (95% CI 5%-11%). Pooling 7 studies that did not assume CM presence since birth, the annual hemorrhage rate was 2.5% per patient-year over 5081.2 patient-years of follow-up (95% CI 1.3%-5.1%). Pooling hazard ratios across 5 studies that evaluated hemorrhage risk factors, prior CM hemorrhage was a significant risk factor for hemorrhage (HR 3.73, 95% CI 1.26-11.1; p = 0.02) while younger age, female sex, deep location, size, multiplicity, and associated developmental venous anomalies (DVAs) were not. CONCLUSIONS Although limited by the heterogeneity of incorporated reports and selection bias, this study found prior hemorrhage to be a significant risk factor for CM bleeding, while age, sex, CM location, size, multiplicity, and associated DVAs were not. Future natural history studies should compound annual hemorrhage rate with prospective seizure and nonhemorrhagic neurological deficit rates.