An analysis of image texture, tumor location, and MGMT promoter methylation in glioblastoma using magnetic resonance imaging
Introduction
Glioblastoma (GBM) is the most common primary brain tumor in adults. Despite multimodality treatment, GBM patients survive only 15 months on average. Standard treatment now includes a DNA alkylating agent called temozolomide (TMZ). TMZ is the only chemotherapeutic that prolongs survival in this disease (Stupp et al., 2005). Interestingly, benefit from TMZ may be predictable through a test for methylation of the O6-methylguanine-DNA methyltransferase (MGMT) gene promoter. Methylation of the MGMT promoter inhibits the repair of therapeutic DNA damage induced by TMZ rendering a drug-resistant cancer more sensitive to chemotherapy (Hau et al., 2007). For unknown reasons, MGMT is methylated (silenced) in 50% of newly diagnosed GBMs (Hegi et al., 2005). Unfortunately, testing for MGMT promoter methylation status by methylation-specific polymerase chain reaction (MS-PCR) requires a large tissue sample and the best results are obtained with cryopreserved tumor tissue. Accurate determinations of MGMT methylation are often not possible from small biopsy specimens or after formalin fixation, both of which are common scenarios for GBM (Preusser et al., 2007). Other methods of assessing MGMT methylation such as activity assays or immunohistochemical detection of MGMT protein have technical shortcomings (Hegi et al., 2005). These and other issues limit the potential utility of testing for MGMT status to customize treatment for GBM.
For these reasons, we are examining the potential of novel MR image analysis methods to predict MGMT methylation status noninvasively. In this study, we test the hypothesis that image texture or tumor location is associated with patterns of MGMT methylation. “Texture analysis” has already been used by our group with promising initial results to identify molecular changes in oligodendrogliomas, a cancer related to GBM (Brown et al. 2008). Others have also begun to explore the potential of imaging to predict MGMT methylation status in GBM. Eoli et al. (2007) have shown a correlation between MGMT promoter methylation status and tumor location assessed visually by expert judgment: more methylated tumors were found in the parietal and occipital lobes and more unmethylated ones in the temporal lobes. Moreover, unmethylated tumors were more necrotic and more likely to be ring enhancing. The analysis of image features by expert judgment can lead to increased interobserver and intercenter variability with inaccurate prediction of the molecular marker. Here, we report a comparison of MGMT promoter methylation status and the look and location of GBM tumors. Unlike earlier reports, texture was analyzed not only visually (by expert observation of tumor borders, presence of cysts, pattern of contrast enhancement, and appearance of tumor signal in T2-weighted image) but also by applying a novel, quantitative, space–frequency transform. Location of the tumors was analyzed by an automatized image segmentation and registration software using a common anatomical space both within and between patients instead of using classification by expert observation. By using these techniques, we are able to evaluate subtleties not seen by an observer (for both texture and location) while moving forward towards the standardization of MR imaging analysis in gliomas, decreasing the potential for interobserver bias, and increasing the likelihood of an accurate noninvasive prediction of MGMT status.
Section snippets
Patients
Patients with newly diagnosed GBM (astrocytoma grade IV, WHO classification) treated at the Tom Baker Cancer Centre in Calgary, Alberta, between January 1, 2004, and December 31, 2006, were identified. Inclusion criteria included: age ≥ 18 years, preoperative MR images in the Picture Archiving and Communication System [PACS; T2, FLAIR, and T1 postgadolinium were used for texture analysis and T1 postcontrast for assessing location] and paraffin-embedded GBM tissue from the first surgery.
Texture analysis by expert observation
The results of this texture assessment are shown in Table 1. Most of the enhancing tumors had undefined margins (71%) and only 6/31 (19%) had a cystic component. Ring enhancement that was present in 61% of the cases was significantly associated with unmethylated MGMT status (P = 0.006). When evaluating T2 sequence images, most of the tumors (23/31—74%) showed a heterogeneous appearance. MGMT status was not significantly associated with any other visually assessed MR characteristics (tumor
Rationale and relevance of the texture analysis
Image texture refers to the local characteristic pattern of image intensity that may be used to identify a tissue. Texture, by definition, also determines local spectral or frequency content in an image; changes in local texture will cause changes in the local spatial frequency. Aspects of texture in an MR image can be quantified by assessing the local spatial frequency content using a space–frequency transform: strong low frequencies appear as homogenous smooth regions, while strong high
Conflicts of interest
S. Drabycz, J.R. Mitchell, and J.G. Cairncross have a financial interest in Calgary Scientific, Inc., which holds a patent on medical imaging applications of the S-transform.
Acknowledgments
The authors wish to acknowledge Michael Eliasziw of the University of Calgary for assistance with the statistical analysis of the data.
This research was supported by the Hotchkiss Brain Institute, the Alberta Cancer Research Institute, and the Alberta Informatics Circle of Research Excellence (iCORE). J.R.M. is supported by the Alberta Heritage Foundation for Medical Research (AHFMR). S.D. was supported by AHFMR, iCORE, and the Alberta Ingenuity Foundation.
Grant support: Alberta Informatics
References (22)
- et al.
Texture analysis of medical images
Clin. Radiol.
(2004) - et al.
A major improvement to the network algorithm for Fisher's exact test in 2×c contingency tables
Comput. Stat. Data Anal.
(2006) - et al.
An overlap invariant entropy measure of 3D medical image alignment
Pattern Recognit.
(1999) - et al.
The use of magnetic resonance imaging to noninvasively detect genetic signatures in oligodendroglioma
Clin. Cancer Res.
(2008) - et al.
BrainWeb: online interface to a 3D MRI simulated brain database
NeuroImage
(1997) - et al.
Design and construction of a realistic digital brain phantom
IEEE Trans. Med. Imaging
(1998) Stem cells and brain tumours
Nature
(2006)- et al.
Texture quantification of medical images using a novel complex space–frequency transform
Int. J. CARS
(2008) - et al.
Methylation of O6-methylguanine DNA methyltransferase and loss of heterozygosity on 19q and/or 17p are overlapping features of secondary glioblastomas with prolonged survival
Clin. Cancer Res.
(2007) - et al.
Tumorigenesis in the brain: location, location, location
Cancer Res.
(2007)
Intratumoral homogeneity of MGMT promoter hypermethylation as demonstrated in serial stereotactic specimens from anaplastic astrocytomas and glioblastomas
Int. J. Cancer
Cited by (178)
Grading diffuse glioma based on 2021 WHO grade using self-attention-base deep learning architecture: variable Vision Transformer (vViT)
2024, Biomedical Signal Processing and ControlPost-treatment imaging of gliomas: challenging the existing dogmas
2024, Clinical RadiologyGlioma radiogenomics and artificial intelligence: road to precision cancer medicine
2023, Clinical RadiologyImmune-related lncRNAs signature and radiomics signature predict the prognosis and immune microenvironment of glioblastoma multiforme
2024, Journal of Translational Medicine
- 1
Co-senior authors on this work.