Is Current Imaging Good Enough to Differentiate Radiation-Induced Brain Injury from Tumor Recurrence?

Differentiating radiation-induced brain injury from tumor recurrence is a challenging problem. This is important in the current medical environment, where therapeutic strategies for CNS disorders are moving toward minimally invasive procedures. On many clinical occasions, imaging is the only reference for choosing an appropriate therapeutic strategy and for evaluating the therapeutic result. Gliomas, particularly those of high grade, contain heterogeneous tissue components with tumor necrosis. Their clinical and pathologic behaviors are different from that of radiation-induced necrosis. They may share similar imaging characters, although the management strategy and outcome of tumor necrosis and radiation necrosis are obviously different.

In the current issue of AJNR, Asao et al present a series of 20 brain tumor lesions in 17 patients and in most (14/17) they were initially high grade (grade III and IV astrocytic tumors). At the time of study, 12 lesions in 10 patients developed radiation necrosis, and the other eight lesions of seven patients were tumor recurrence. By drawing five regions of interest on trace DWI and apparent diffusion coefficients (ADC) maps within these lesions, they observed that radiation necrosis usually showed heterogeneity on DWI, and often included spotty, marked hypointensity visually on DWI. The maximum ADC values were significantly lower for the recurrence group than for the necrosis group. With this observation, the authors concluded that DWI was useful in differentiating recurrent neoplasm from radiation necrosis.

The message conveyed in the article is similar to that from an article published last year in the AJNR (1). Clinically, the results may be applicable to many tumors with classic imaging manifestations, or to tumors with massive necrosis either radiation-induced or nonradiation-induced. For cases with mixture of tissue components or with an evolving necrotic process or tumor progression, the dilemma, however, remains. Unfortunately, that is the most common situation.

After reading these articles, some questions remain. Can overall signal intensities and their corresponding ADC values represent tissue characteristics? Is the spatial resolution of our current imaging technique (DWI, PWI, MR spectroscopy) high enough to deal with the tissue heterogeneity? Does current metabolic imaging provide enough specificity (2, 3)? Although we have many sophisticated imaging tools at our disposal, the questions we face are even more complicated. It is clear that the complexity of different tissue component in brain tumors and the relative ineffectiveness of current therapies make these questions even more problematic. The real challenge is to find a biomarker of tumor recurrence that can be identified in its early stage.

• Copyright © American Society of Neuroradiology