Predictors of future haemorrhage from cerebral cavernous malformations: a retrospective cohort study

Cerebral cavernous malformations (CCMs) are commonly diagnosed, with a low reported rate of haemorrhage on long-term follow-up. The identification of factors predictive of future haemorrhage risk would assist in guiding the management of patients with CCM. The aim of this study was to identify variables associated with haemorrhage, and calculate haemorrhage risk in CCM. We conducted a retrospective study of patients diagnosed with a CCM, managed at a specialist tertiary neuroscience centre (2007-2019). The primary outcome was symptomatic haemorrhage, and secondary outcomes were variables associated with increased risk of haemorrhage, using multivariable Cox regression analysis. Included were 545 patients, with 734 confirmed cavernomas. Median age at diagnosis was 47 (interquartile range [IQR] 35-60), with a median follow-up duration after diagnosis of 46 months (IQR 19-85). Of the patients, 15.0% had multiple lesions (N = 82/545). Symptomatic presentation was observed in 52.5% of patients (N = 286/545). The annual haemorrhage rate was 1.00% per lesion-year (25 events in 2512 lesion-years), and higher in those with symptoms at presentation (1.50% per lesion-year, 22 events vs 0.29%, 3 events, P < 0.001). The variables associated with symptomatic haemorrhage were increased size (hazard ratio [HR] 1.04, 95% confidence interval [CI] 1.01-1.07, P = 0.004), eloquent location (HR 2.63, 95% CI 1.12-6.16, P = 0.026), and symptomatic haemorrhage at presentation (HR 5.37, 95% CI 2.40-11.99, P < 0.001). This study demonstrated that CCMs have a low haemorrhage rate. Increased size, eloquent location, and haemorrhage at presentation appear to be predictive of a higher risk of haemorrhage, and could be used to stratify management protocols.

Automated summary

This study aimed to identify variables associated with haemorrhage and calculate haemorrhage risk in CCM. The results revealed that CCMs have a low haemorrhage rate, but increased size, eloquent location, and haemorrhage at presentation may predict a higher risk of haemorrhage.