Are cerebral cavernomas truly nonenhancing lesions and thereby distinguishable from arteriovenous malformations? MRI findings and histopathological correlation Katja Pinker a , Ioannis Stavrou b , Engelbert Knosp b , Siegfried Trattnig a, 4 a Department of Radiology/MR Centre of Excellence, Medical University Vienna/General Hospital Vienna, 1090 Vienna, Austria b Department of Neurosurgery, Medical University Vienna/General Hospital Vienna, 1090 Vienna, Austria Received 15 October 2005; accepted 15 October 2005 Abstract Purpose: The aim of this study was to determine whether cerebral cavernomas are truly nonenhancing lesions on magnetic resonance imaging (MRI), whether they can be distinguished from arteriovenous malformations (AVM) on that basis and to evaluate the incidence of their association with developmental venous anomalies (DVA). Patients and Methods: Thirty-two patients who underwent neurosurgical operation for a cerebral vascular malformation and had a standard MRI conclusive of cerebral cavernoma were retrospectively evaluated for size of the lesions, contrast enhancement of the lesion and the coexistence of DVA. The contrast uptake of these lesions was investigated, and contrast enhancement was classified as none, moderate or marked. The incidence of an associated DVA was also investigated. The radiological findings were subsequently correlated with neurohistopathological findings. Results: No difference was found between the contrast enhancement of cavernomas and AVMs. Cross tables were calculated for contrast enhancement and size, which demonstrated no statistically significant correlation. Cross tables were calculated for contrast enhancement and histopathological diagnosis, which revealed that both entities presented variable degrees of contrast enhancement and were thereby not distinguishable from each other on the basis of contrast enhancement. We found an association of cavernoma with DVA in 30% of cases. Conclusion: Neither a correlation between the absence of contrast enhancement and the histopathological diagnosis of cavernoma nor the size and contrast enhancement was found. We conclude that cavernomas present with variable degrees of contrast enhancement on MRI and, thus, are definitely not distinguishable from AVM on the basis of contrast enhanced MRI. We found an association between cavernomas and DVA in approximately one third of patients. D 2006 Elsevier Inc. All rights reserved. Keywords: Cerebral cavernoma; Magnetic resonance imaging; Contrast enhancement; Arteriovenous malformation 1. Introduction Cerebral cavernomas, also called angiographically occult vascular malformations or cavernous angiomas, occur at an incidence of 0.4 – 0.9% of the general population [1,2]. They represent approximately 5 – 10% of central nervous system vascular malformations [3,4]. Most commonly, lesions are found supratentorially (75%) and rarely in the brainstem, accounting for about 20% of all intracerebral cavernomas [4]. These lesions occur most often in patients between 30 and 40 years of age. Symptomatic lesions usually present with evidence of recent hemorrhage combined with sei- zures, when located supratentorially, and with focal neuro- logical deficits or, sometimes, even death when located in the posterior fossa. Nevertheless, patients harboring a lesion may also be completely asymptomatic [5–7]. In contrast to extraaxial cavernous hemangiomas, which typically present with homogenous contrast enhancement, intraaxial cavernomas should not enhance significantly [8]. Therefore, the term slow-flow vascular malformation has been suggested in the literature in order to distinguish cerebral cavernous angiomas from their extraaxial counter- parts, as well as from the intraaxial, high-flow arteriovenous malformations (AVM) [9–11]. By definition, the perfusion 0730-725X/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2005.10.037 This paper was presented at the ISMRM 2005, Miami, FL, USA. 4 Corresponding author. E-mail address: siegfried.trattnig@meduniwien.ac.at (S. Trattnig). Magnetic Resonance Imaging 24 (2006) 631 – 637