Dynamic nature of cavernous malformations: a prospective magnetic resonance imaging study with volumetric analysis

Object. Although cavernous malformations (CMs) are not detected in angiogra phic studies, they have a characteristic appearance on magnetic resonance ( MR) images. A number of reports published in the last decade have focused o n the behavior of these lesions within the clinical environment. However, l ittle has been published about the evolution of CMs over time, as observed in imaging studies. To understand imaging-documented changes in CMs over ti me, we analyzed MR images of 114 cavernous malformations in 68 patients who were followed prospectively. Methods. For each CM the location, volume, and MR imaging signal characteri stics were recorded. Volume data were available for 107 lesions from initia l images. The mean volume of these 107 CMs was 2779 mm(3). The lesions rang ed in size from 0.5 to 46.533 mm(3) (46.5 cm(3)). Volume data from a second set of images were available for 76 CMs (mean interval from first imaging session 26 months), and from a third set of images for 24 lesions (mean int erval from second imaging session 18 months). Over the first follow-up inte rval, the mean volume change was -991 mm(3) (a decrease of approximately 1 cm(3)) and over the second interval the mean volume change was -642 mm(3). Although these mean volume changes appear modest, volume changes in single lesions during follow-up intervals were more dramatic, with decreases as la rge as 45,629 mm(3) (45.6 cm(3)) and increases as large as 6,074 mm(3) (6 c m(3)). Serial examinations of the MR imaging signal characteristics of thes e CMs demonstrate a trend for maturation of blood products from a subacute, to a mixed, and finally to a chronic appearance. Three lesions appeared de novo during the follow-up period. Conclusions. On the basis of their analysis, the authors conclude that CMs exhibit a range of dynamic behaviors including enlargement, regression, and de novo formation, as well as progression through a series of characterist ic MR imaging appearances.