Genes and molecules

In this Educational Lecture, I will introduce you to the classification of gliomas, how this system has changed since the introduction of the World Health Organization (WHO) Brain Tumor Classification in 2016, and what that means for MR imaging.

A glioma is comprised of cells that are characterized by combinations of genetic alterations that cause them to proliferate at an uncontrolled rate and invade healthy tissues.^1,2 Gliomas can arise from several types of cells including stem cells, progenitor cells, and more mature cells; however, because mature cells are more resistant to transformation, it is currently thought that stem cells play the primary role in gliomagenesis.³ Many different genetic alterations play a role in gliomagenesis, but their main effects are quite similar: production of abnormal proteins and dysregulation of metabolic and other molecular pathways within cells, which lead to cell cycle progression, proliferation, defense against the immune system, and survival. Together, this combination of proteins and dysregulated pathways is called the molecular phenotype.⁴

Historically, gliomas were classified according to their characteristic features on light microscopy.⁵ While relatively easy to perform, it became apparent that in clinical practice even gliomas with cells with a similar structure – and therefore diagnosed as the same glioma type – sometimes had different clinical features or tumor behavior. MR imaging was even more limited, since it could only visualize the macroscopic consequences of the microscopic tumor cell characteristics; for instance, fields of poorly differentiated tumor cells resembling astrocytes within areas of necrosis (i.e., a GBM) were seen macroscopically as a rather nonspecific mass with a T₁-hypointense necrotic center and an enhancing rim. This classification, however, has undergone a drastic paradigm change with the introduction of the WHO 2016 Brain Tumor Classification, largely influenced by results from the Human Genome Project.⁶ Instead of focusing on histopathology alone, this classification combines histopathology with molecular and genetic information. The reason behind this is that genetic alterations occurring inside glioma cells affect (amongst others) how fast and aggressive a glioma grows, and how resistant it will be against which treatment, ultimately influencing patient prognosis.^7,8

For the sake of clarity, in this lecture I will focus only on the classification of diffuse gliomas. In the 2016 WHO Classification, their common denominator is the IDH mutation (or lack thereof). Diffuse gliomas have cells that either look like astrocytes, oligodendrocytes or a combination, although the latter is rare. The presence or absence of an IDH mutation subdivides the tumors into IDH-mutant and IDH-wildtype. In case they also have a 1p/19q codeletion, they are called oligodendroglioma. GBM is the highest grade of diffuse glioma, and is similarly subdivided into IDH-mutant or IDH-wildtype.^7,8 Below are some of the main examples how this classification is used to guide treatment and predict prognosis in clinical practice:^7-9

1. IDH-mutated gliomas have a better prognosis than IDH-wildtype gliomas.
2. 1p/19q co-deleted gliomas (i.e. oligodendrogliomas) have a better prognosis as well as a better response to certain chemotherapy.
3. A gain of chromosome 7, loss of chromosome 10, EGFR amplification or TERT mutation classify otherwise lower-grade IDH-wildtype gliomas (based on histopathology) as GBM.
4. Gliomas with CDKN2A/B deletion have a worse prognosis.
5. MGMT-methylated gliomas are more sensitive to chemotherapy.

WHO Classification & MRI

Classic imaging features of gliomas have of course not changed much since the introduction of the 2016 WHO Classification. However, we should now also focus on genetic and molecular features of glioma cells, or the molecular phenotype of gliomas, and try to visualize these with MRI. The most obvious example is detection of 2HG with proton spectroscopy, but IDH-mutant gliomas can also be identified through the T₂-FLAIR mismatch sign, which is a high signal intensity within a glioma on T₂-weighted images and near complete suppression of the signal of FLAIR images that seems to be very specific to IDH-mutant, 1p19q co-deleted gliomas.^7,8 Other examples of using MRI to define certain aspects of the WHO 2016 classification are determining MGMT methylation with radiomics¹⁰, susceptibility-weighted imaging to differentiate IDH-mutant from IDH-wildtype gliomas¹¹, and detection of cystathionine with MR spectroscopy, which is associated with the 1p/19q codeletion¹².
Next to aiding diagnosis of glioma types in a noninvasive way, we could also use our knowledge on genetic and molecular features to define imaging biomarkers that are specific of gliomas. This might be even more clinically relevant than diagnosing these tumors, as many will be surgically resected providing plenty of tissue for pathologists to perform all kinds of genetic and molecular tests that in sheer number can never be trumped by MR imaging markers. Some examples of the use of MRI when we look beyond making an initial diagnosis are the differentiation of recurrent GBM from radiation necrosis using APT-CEST¹³, and using sodium MRI to identify tumor infiltration in the non-enhancing T₂-hyperintense area of a glioma, where it is often difficult if not impossible to differentiate between edema and infiltrating tumor cells¹⁴.

References

¹Weinberg (2014) Cell 157:267-271; ²Fouad et al. (2017) Am J Cancer Res 7:1016-1036; ³Azzarelli et al. (2018) Development 145(10):dev162693; ⁴Dagogo-Jack et al. (2018) Nat Rev Clin Oncol 15:81-94; ⁵Wesseling et al. (2011) Diagn Histopathol 17:486-494; ⁶Wheeler et al. (2013) Genome Res 23:1054-1062; ⁷Johnson et al. (2017) Radiographics 37:2164-2180; ⁸Louis et al. (2016) Acta Neuropathol 131:803-820; ⁹Gonzalez Castro et al. (2021) Neurooncol Pract 8:4-10; ¹⁰Yogananda et al. (2021) AJNR Mar 4 (online ahead of print); ¹¹Grabner et al. (2017) Eur Radiol 27:1556-1567; ¹²Branzoli et al. (2019) Neuro Oncol 21:765-774; ¹³Park et al. (2018) Eur Radiol 28:3285-3295; ¹⁴Regnery et al. (2020) Neuroimage Clin 28:102427.