Diagnostic challenges of radiation necrosis.

progression or recurrence (TPR) in its clinical presentation and imaging findings in brain tumor patients. In this issue, Korchi et al describe imaging appearances of five consecutive cases of RN from 73 skull-based tumors treated with proton-beam radiotherapy (RT) in Switzerland.1 An important feature of this study is the selection of patients with extra-axial tumors that tend to recur locally. As such, intra-axial lesions detected following RT more likely represent radiation injury rather than TPR—an interesting model to study imaging findings of RN. Their findings are concordant with those seen with photon RT suggesting that the process of RN may be partly independent of underlying pathology and radiation modality. This study, however, does not help us understand how to distinguish these lesions from TPR. Although challenging, differentiating TPR from RN is becoming more relevant with increasing availability of newer targeted therapies and salvage RT for TPR. Conversely, patients that are accurately diagnosed with RN may not only be treated conservatively with steroids, but also offered hyperbaric oxygen or bevacizumab in refractory cases. A recent double-blinded randomized controlled trial showed improvement in neurological symptoms or signs in all bevacizumab-treated patients with RN.2 Although surgical resection has been the historical gold standard for diagnosis, non-invasive techniques have become increasingly preferred. Several reviews of the use of MRI, magnetic resonance spectroscopy (MRS), positron emission tomography (PET) and single photon emission CT (SPECT) in this setting have been recently published for patients with glioma.3-5 Shah et al systematically reviewed 17 studies (7 prospective, 10 retrospective) published between 2007 and 2012.3 Histology or clinical/radiological follow-up confirmed TPR in 282 cases (69%) and RN in 100 cases (24%). Discovery of the recurrent lesion was most often by MRI, and at a mean interval of 13.2 months following RT. Two studies (n=104) of standard MRI protocols comparing lesion enhancement to the contralateral hemisphere demonstrate a cumulative sensitivity of 89% but poor specificity of 33%, indicating that MRI alone appears insufficient for reliably differentiating TPR from RN. Specialized protocols may improve performance of MRI. Dynamic susceptibility contrast enhanced (DCE) perfusion MRI measures relative cerebral blood volume (rCBV). Radiation necrosis shows low perfusion, whereas the high metabolic activity and angiogenesis of TPR results in high rCBV. Pseudoresponse from bevacizumab, an antiangiogenic agent used in the management of glioblastoma, may limit interpretability of perfusion MRI. Six studies (n=136) measuring rCBV within the lesion yield cumulative sensitivity and specificity of 80% and 77%, respectively.3 Vascular permeability measured as Ktrans yields unimpressive results.3 Diffusion-