Clinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient data

Abstract

Background: Cerebral cavernous malformations (CCMs) can cause symptomatic intracranial haemorrhage (ICH), but the estimated risks are imprecise and predictors remain uncertain. We aimed to obtain precise estimates and predictors of the risk of ICH during untreated follow-up in an individual patient data meta-analysis.

Methods: We invited investigators of published cohorts of people aged at least 16 years, identified by a systematic review of Ovid MEDLINE and Embase from inception to April 30, 2015, to provide individual patient data on clinical course from CCM diagnosis until first CCM treatment or last available follow-up. We used survival analysis to estimate the 5-year risk of symptomatic ICH due to CCMs (primary outcome), multivariable Cox regression to identify baseline predictors of outcome, and random-effects models to pool estimates in a meta-analysis.

Findings: Among 1620 people in seven cohorts from six studies, 204 experienced ICH during 5197 person-years of follow-up (Kaplan-Meier estimated 5-year risk 15·8%, 95% CI 13·7-17·9). The primary outcome of ICH within 5 years of CCM diagnosis was associated with clinical presentation with ICH or new focal neurological deficit (FND) without brain imaging evidence of recent haemorrhage versus other modes of presentation (hazard ratio 5·6, 95% CI 3·2-9·7) and with brainstem CCM location versus other locations (4·4, 2·3-8·6), but age, sex, and CCM multiplicity did not add independent prognostic information. The 5-year estimated risk of ICH during untreated follow-up was 3·8% (95% CI 2·1-5·5) for 718 people with non-brainstem CCM presenting without ICH or FND, 8·0% (0·1-15·9) for 80 people with brainstem CCM presenting without ICH or FND, 18·4% (13·3-23·5) for 327 people with non-brainstem CCM presenting with ICH or FND, and 30·8% (26·3-35·2) for 495 people with brainstem CCM presenting with ICH or FND.

Interpretation: Mode of clinical presentation and CCM location are independently associated with ICH within 5 years of CCM diagnosis. These findings can inform decisions about CCM treatment.

Funding: UK Medical Research Council, Chief Scientist Office of the Scottish Government, and UK Stroke Association.

Copyright © 2016 Horne et al. Open Access article distributed under the terms of CC BY. Published by Elsevier Ltd.. All rights reserved.

Figures

Figure 1

Study flowchart CCM=cerebral cavernous malformation. *See appendix pp 18–19. †See appendix pp 19–20 for references. ‡One eligible study provided data from two time periods, which are included as two separate cohorts. §See appendix p 20.

Figure 2

Forest plots of associations between mode of presentation and cerebral cavernous malformation location with primary and secondary outcomes Plots show cohort-level and pooled estimates of associations between ICH or FND at presentation (A and C) or brainstem CCM location (B and D) and outcome during 5 years of follow-up. The area of each shaded box is proportional to the weight of the cohort it represents. CCM=cerebral cavernous malformation. FND=focal neurological deficit. HR=hazard ratio. ICH=intracranial haemorrhage.

Figure 3

Kaplan-Meier plots of progression to intracranial haemorrhage or to intracranial haemorrhage or focal neurological deficit Plots show the proportion of people progressing to ICH (A) or ICH or FND (B) during follow-up, stratified by ICH or FND presentation from brainstem CCMs, ICH or FND presentation from non-brainstem CCMs, other presentation from brainstem CCMs, and other presentation from non-brainstem CCMs. CCM=cerebral cavernous malformation. FND=focal neurological deficit. HR=hazard ratio. ICH=intracranial haemorrhage.