Imaging biomarkers of brain tumour margin and tumour invasion

Abstract

Invasion of tumour cells into the normal brain is one of the major reasons of treatment failure for gliomas. Although there is a good understanding of the molecular and cellular processes that occur during this invasion, it is not possible to detect the extent of the tumour with conventional imaging. However, there is an understanding that the degree of invasion differs with individual tumours, and yet they are all treated the same. Newer imaging techniques that probe the pathological changes within tumours may be suitable biomarkers for invasion. Imaging methods are now available that can detect subtle changes in white matter organisation (diffusion tensor imaging), tumour metabolism and cellular proliferation (using MR spectroscopy and positron emission tomography) occurring in regions of tumour that cannot be detected by conventional imaging. The role of such biomarkers of invasion should allow better delineation of tumour margins, which should improve treatment planning (especially surgery and radiotherapy) and provide information on the invasiveness of an individual tumour to help select the most appropriate therapy and help stratify patients for clinical trials.

Figure 1

Example of the use of diffusion tensor imaging (DTI) to understand the effect of the tumour on white matter (WM) tracts. The upper row shows a tumour deviating a tract (arrows). In this case the fractional anisotropy (FA) values are similar to the contralateral side but the directionally encoded colour (DEC) maps shows the tract colour is different to the contralateral side. The middle row shows tumour invasion of a tract (arrows). The tract can still be seen but with reduced FA and hue on DEC. The lowest row shows tumour disruption (arrows), where no tract can be identified on either method.

Figure 2

Visual differences in fractional anisotropy between a metastasis (upper row) and a glioblastoma. With the metastasis there is still intact white matter surrounding the tumour (arrows) whereas in the glioblastoma there is disruption of the white matter over a larger area. This difference, despite the marked oedema from both tumours, has been suggested as a result of tumour invasion.

Figure 3

An example of the proton spectra (echo time=30 ms) within and surrounding a World Health Organization grade III anaplastic astrocytoma. The first region is within normal brain and the normal spectra can be seen with high N-acetylaspartate (NAA) peaks and a lower choline peak. The second spectrum is taken from the peritumoural tissue where there is a much higher choline peak and a lower NAA peak. NAA is still detectable suggesting viable neurons are still present. Within the tumour there is a very high choline peak but NAA is undetectable.

Figure 4

An example of a 18F-fluorothymidine (FLT) positron emission tomography (PET) image overlying a contrast-enhanced T₁ weighted image of a glioblastoma. Although the FLT uptake is largely in areas of contrast enhancement, the uptake extends into the surrounding brain. Histological analysis of image-guided biopsies, however, showed that the FLT uptake underestimated the region of tumour invasion in this case.