Accuracy of apparent diffusion coefficient in differentiation of glioblastoma from metastasis

Abstract

Background: Brain metastasis and glioblastoma multiforme are two of the most common malignant brain neoplasms. There are many difficulties in distinguishing these diseases from each other.

Purpose: The purpose of this study was to determine whether the mean apparent diffusion coefficient and absolute standard deviation derived from apparent diffusion coefficient measurements can be used to differentiate glioblastoma multiforme from brain metastasis based on cellularity levels.

Material and methods: Magnetic resonance images of 34 patients with histologically verified brain tumors were evaluated retrospectively. Apparent diffusion coefficient and standard deviation values were measured in the enhancing tumor, peritumoral region, and contralateral healthy white matter. Then, to determine whether there was a statistical difference between brain metastasis and glioblastoma multiforme, we analyzed different variables between the two groups.

Results: Neither mean apparent diffusion coefficient values and ratios nor standard deviation values and ratios were significantly different between glioblastoma multiforme and brain metastasis. Receiver operating characteristic curve analysis of the logistic model with backward stepwise feature selection yielded an area under the curve of 0.77, a specificity of 84%, a sensitivity of 67%, a positive predictive value of 83.33%, and a negative predictive value of 78.26% for distinguishing between glioblastoma multiforme and brain metastasis. The absolute standard deviation and standard deviation ratios were significantly higher in the peritumoral edema compared to the tumor region in each case.

Conclusion: Apparent diffusion coefficient values and ratios, as well as standard deviation values and ratios in peritumoral edema, cannot be used to differentiate edema with infiltration of tumor cells from vasogenic edema. However, standard deviation values could successfully characterize areas of peritumoral edema from the tumoral region in each case.

Keywords: Apparent diffusion coefficient; brain metastasis; diffusion weighted imaging; glioblastoma multiforme.

Figure 1.

Glioblastoma multiforme (GBM) case evaluation and postprocessing. ADC: apparent diffusion coefficient; SD: standard deviation.

Figure 2.

Brain metastasis case evaluation and postprocessing.

Figure 3.

The receiver-operating characteristic (ROC) curve for the logistic model with the area under curve (AUC) (95% confidence interval) as well as the sensitivity and specificity values for different thresholds (the threshold for best trade-off between sensitivity and specificity, the threshold where the sensitivity is 100%, and the threshold where the specificity is 100%).