Centrally Reduced Diffusion Sign for Differentiation between Treatment-Related Lesions and Glioma Progression: A Validation Study

Abstract

Background and purpose: Differentiating between treatment-related lesions and tumor progression remains one of the greatest dilemmas in neuro-oncology. Diffusion MR imaging characteristics may provide useful information to help make this distinction. The aim of the study was to assess the diagnostic accuracy of the centrally reduced diffusion sign for differentiation of treatment-related lesions and true tumor progression in patients with suspected glioma recurrence.

Materials and methods: The images of 231 patients who underwent an operation for suspected glioma recurrence were reviewed. Patients with susceptibility artifacts or without central necrosis were excluded. The final diagnosis was established according to histopathology reports. Two neuroradiologists classified the diffusion patterns on preoperative MR imaging as the following: 1) reduced diffusion in the solid component only, 2) reduced diffusion mainly in the solid component, 3) no reduced diffusion, 4) reduced diffusion mainly in the central necrosis, and 5) reduced diffusion in the central necrosis only. Diagnostic accuracy metrics and the area under the receiver operating characteristic curve were estimated for the diffusion patterns.

Results: One hundred three patients were included (22 with treatment-related lesions and 81 with tumor progression). The diagnostic accuracy results for the centrally reduced diffusion pattern as a predictor of treatment-related lesions ("mainly central" and "exclusively central" patterns versus all other patterns) were as follows: 64% sensitivity (95% CI, 41%-83%), 84% specificity (95% CI, 74%-91%), 52% positive predictive value (95% CI, 37%-66%), and 89% negative predictive value (95% CI, 83%-94%).

Conclusions: The centrally reduced diffusion sign is associated with the presence of treatment effect. The probability of a histologic diagnosis of a treatment-related lesion is low (11%) in the absence of centrally reduced diffusion.

FIG 1.

Standards for the Reporting of Diagnostic Accuracy Studies flow diagram. The thickness of the arrows is proportional to the number of subjects.

FIG 2.

Distribution of the different diffusion patterns in the treatment-related lesion and recurrent-tumor groups.

FIG 3.

Preoperative MR imaging of 6 patients with suspected glioma recurrence. In 3 cases (1, 2, and 3), histopathology revealed treatment-related lesions, and in the other 3 cases (4, 5, and 6), recurrent tumor. T1-weighted postcontrast (A) and DWI (B) are shown in each case. Treatment-related lesions (1, 2, and 3) show reduced diffusion predominantly in the central necrotic region. Lesions corresponding to progressive tumor (4, 5, and 6), on the other side, show reduced diffusion mainly in the solid-lesion components.

FIG 4.

ROC curve for diffusion patterns. AUC indicates area under the curve.

FIG 5.

The box-and-whisker plot shows the distribution of the percentages of concurrent treatment effect in patients with recurrent tumor by diffusion pattern. The whiskers mark the maximum and minimum values, and the box edges indicate the first and third quartiles. The horizontal line inside the box corresponds to the median. The median percentage of treatment effect was higher in the group with centrally reduced diffusion than in the group with the other diffusion patterns, but differences were not statistically significant (P = .076).