Cerebral Radiation Necrosis: An Analysis of Clinical and Quantitative Imaging and Volumetric Features.

Abstract

Radiation therapy is an effective treatment for primary brain tumors and intracranial metastases but can occasionally precede new enhancing lesions on imaging studies that are difficult to discern between tumor recurrence (TR) and radiation necrosis (RN). The aim of this study was to identify clinical presentation and imaging patterns of RN compared with TR that may obviate the need for invasive definitive biopsy. Patients who received radiation therapy and subsequently presented with a new intracranial lesion were reviewed from 2001 to 2016; 27 patients were identified with adequate records and confirmed pathology to have RN present or TR only. Patient and lesion characteristics were assessed using univariate and multivariate logistic regression analyses. Sensitivity and specificities were calculated for imaging features and quantitatively segmented lesion and edema volumes for identifying RN. Karnofsky performance scale score at presentation significantly predicted pathologic diagnosis on univariate analysis (P= 0.044). Radiation dosage and time from radiation therapy to lesion onset did not differ among pathologic diagnosis groups. No differences existed between RN and TR on quantitative imaging analyses. Multivariate logistic regression found higher Karnofsky performance scale score to be an independent factor associated with TR relative to RN (odds ratio 1.26, 95% confidence interval 1.02-1.56, P= 0.030). Diagnostic imaging can often be inaccurate in detecting RN alone, even with quantitative volume assessment. Functional status on repeat presentation may increase the likelihood of accurate diagnosis before definitive biopsy when neuroimaging remains unclear.