The next step in understanding the prognosis of cerebral cavernous malformations.

Neurology 2012;78:614–615 Cerebral cavernous malformations (CCM) are comprised of vascular sinusoids lacking smooth muscle cells, elastic lamina, and tight junctions between endothelial cells, which renders them prone to intracranial hemorrhage.1 The magnitude of the risk of CCM hemorrhage and its predictors are of considerable concern to patients2 and care providers (Angioma Alliance [www.angiomaalliance.org] and Cavernoma Alliance UK [www.cavernoma.org.uk]). However, despite the prevalence of CCM, which affects 1 in 200–600 people,3,4 knowledge about the risk of CCM hemorrhage is incomplete.1 The untreated clinical course of CCM has been described in mostly small retrospective hospitalbased case series susceptible to selection bias, with different inception points, average lengths of follow-up ranging from 2 to 5 years, and variable outcome definitions.5 In these studies, the estimated annual risk of first CCM hemorrhage (0.4% to 2.0%) has been less than the estimated annual risk of recurrent CCM hemorrhage (3.8% to 22.9%), although 3 studies have found that the annual risk of recurrent hemorrhage diminishes over time.6 Studies have been inconsistent about whether sex, CCM location, or CCM multiplicity influences the risk of hemorrhage. In this issue of Neurology, Flemming et al.7 take the next step in furthering our understanding of the untreated clinical course of CCM, with the largest study of CCM natural history to date. This study benefited from the Mayo Clinic’s unique medical records system,8 which enabled the identification of a retrospective cohort of 292 children and adults seen with CCM at the Mayo Clinic between 1989 and 1999, with 2,035 patient-years of follow-up through 2000 –2003. The primary outcome measure was clinically symptomatic hemorrhage during followup, for which radiographic confirmation was available in most cases. The authors confirmed that the annual rate of symptomatic hemorrhage was low after an incidental CCM diagnosis and that the annual risk of recurrent hemorrhage was higher (declining from 18% in the first year after hemorrhage to 3% by 5–10 years after hemorrhage). In a multivariable analysis, initial CCM presentation with hemorrhage was the strongest independent predictor of hemorrhage during follow-up (5-fold hazard ratio, although with a wide confidence interval). Male sex and the existence of multiple CCM (both with 2-fold hazard ratios, with wide confidence intervals) also predicted hemorrhage during follow-up. However, CCM present several challenges for researchers dedicated to describing their long-term outcome in the setting of everyday clinical practice. Because MRI techniques evolve during long studies, the diagnostic utility of MRI may vary for identification of both solitary and multiple CCM, especially with sensitive gradient-recalled echo sequences and other susceptibility-weighted imaging techniques. Patients with multiple CCM attributable to genetic mutations may have a different clinical course depending on the underlying genetic mutation, but genetic testing is not undertaken for every patient with multiple CCM in clinical practice. CCM multiplicity also creates statistical problems when trying to study the association of CCM characteristics, which vary within one patient, with patient-level characteristics in survival analyses. Finally, detection, diagnostic, misclassification, and reporting biases are likely to affect the identification of symptomatic CCM hemorrhage. The imaging protocol of any study will be unable to fully account for the recognized phenomena of new acute CCM hemorrhage on radiographic imaging when a patient manifests only a seizure or headache or no symptoms at all. Fortunately, some of these biases can be mitigated by using recommended definitions for outcome events.5 In particular, reliable recognition of acute CCM hemorrhage requires timely imaging using the appropriate modality with interpretation by neuroradiologists. Patients with CCM may also experience “focal neurologic deficits” attributable to the anatomic location