Introduction
In typical cases, the clinical workflow for brain tumors begins with imaging studies; however, despite the increased sensitivity and capabilities of these methodologies, diagnosing brain tumors requires histopathologic evaluation of tissue. It is imperative for clinicians to understand how brain tumors are classified so that they can better counsel patients at the time of initial presentation, accurately describe prognosis, and prioritize management of medical or neurological comorbidities based on the anticipated behavior of the tumor.
Tumor classification has historically relied primarily upon morphologic features identified by light microscopy. In the past decade, integration of high-throughput genomic testing into routine clinical workflows has refined the approach to diagnosing brain tumors. In this chapter, we explore tumor classification from traditional microscope-based approaches to currently available methodologies. We review genomic profiling and introduce the concept of an integrated diagnosis.
The chapter begins with a discussion of the histologic assessment of brain tumors, which is often the first data to be reported after a tumor surgery. This is followed by a discussion of molecular profiling, which is used to further classify these tumors, guide prognosis, and predict response to therapy. Molecular data typically returns in the weeks following surgery. Finally, we review a series of common case scenarios and integrate the histologic and molecular data into a final integrated diagnosis that clinicians can use to help manage and counsel these patients.
Histologic classification of central nervous system tumors
The primary method of diagnostic neuropathology and brain tumor classification remains microscopic evaluation of hematoxylin and eosin (H&E)–stained tissue sections by light microscopy. This approach has experienced relatively few changes over the past several decades with generations of pathologists still undergoing specialized training focused on learning to read or interpret stained sections on glass slides.
The process of converting a portion of resected brain tissue to an H&E stained section is the first step to achieve a tissue-based diagnosis. Brain tissues removed at the time of surgery can range from small biopsies, such as needle core biopsies, to large resections that often occur during a tumor debulking surgery. Once the fresh tissue reaches the laboratory, the sample is subjected to fixation in formalin-based solutions that serve to preserve tissues and cells.
Formalin-fixed tissues then undergo a series of steps, commonly referred to as tissue processing, which prepares the formalin-fixed tissues for embedding into a block of paraffin wax. Once the tissue fragment is embedded in paraffin and mounted into a cassette, this “block” serves as the final medium where the tissue can be safely stored for years (even decades). Importantly, this processing method preserves the integrity of the tissues for later use. Formalin-fixed paraffin-embedded (FFPE) samples are the mainstay of both traditional microscopic analysis and serve as the starting medium for most tissue-based molecular and genomic assays currently available.
Once a tissue sample is converted to an FFPE block, sections (5 microns thickness) can be cut from the block, placed onto glass slides, and subjected to staining with the dyes hematoxylin and eosin. Hematoxylin stains nucleic acids a deep blue-to-purple color, which readily highlights the nuclei of cells. Eosin stains other cellular components, like proteins, with a bright pink color that contrasts with the blue hematoxylin-stained nuclei—together these two dyes serve as the basis for all H&E-based tumor classification.
Brain tumor classification based on H&E features relies on the fact that the histologic appearance of these tumors is consistent from one patient to another. In fact, the features are so reproducible that they have been codified into the World Health Organization (WHO) Classification of Tumors of the Central Nervous System, which outlines the microscopic features present in all brain tumors and serves as the central resource to ensure all patients around the world are diagnosed and graded using the same criteria.
Tumor Grading . Unlike other tumors, which are staged based on size, lymph node involvement, and presence or extent of metastases (e.g., TNM classification), staging is not used for gliomas, which rarely spread beyond the central nervous system (CNS). Instead, gliomas are graded from grade I to IV and are broadly divided into low- and high-grade tumors. Grade I and II are considered “low grade” and grade III and IV as “high grade”; of these, grade I tumors are generally well circumscribed, whereas grades II–IV are diffusely infiltrating, albeit with some exceptions such as ependymoma and pleomorphic xanthoastrocytoma which tend to be well circumscribed despite conforming to grade II. The histologic features assessed to establish glioma grading include atypia, mitoses, endothelial or microvascular proliferation, and necrosis. Grade I tumors such as pilocytic astrocytomas are most commonly encountered in the pediatric setting. Although such tumors do occur in the adult population, the use of the term “lower-grade gliomas” in adults largely refers to infiltrating grade II tumors such as diffuse astrocytoma and oligodendroglioma. Grade III tumors include anaplastic astrocytoma (AA3) and anaplastic oligodendroglioma (AO3), whereas glioblastoma (GBM) is considered a grade IV astrocytoma. Progression of brain tumors from low to high grade does occur, most commonly in adults from grade II diffuse astrocytoma to higher-grade gliomas such as AA3 or GBM, which are termed “secondary” GBM. Not all GBMs present as progression from a lower-grade neoplasm. In fact, the vast majority of GBMs present as de novo tumors without any prior history of a lower-grade glioma and are referred to as “primary” GBMs. Ependymomas are also considered a subtype of glioma, albeit with combined glial and epithelial lineage, and are graded from I–III; these are generally well circumscribed, with the exception of WHO grade III tumors.
Tumor Lineage. Tumor lineage is another important concept that goes hand in hand with grading. Lineage refers to the presumed cell of origin to which a tumor morphologically resembles. This designation then leads to broad categories of tumors such as astrocytoma, oligodendroglioma, and ependymoma among others.
When reviewing H&Es for tumor classification, the first step is determining whether tumor cells are present. An initial feature to assess is the overall cellularity of the tissue compartment (white matter, cortex, thalamus, cerebellum) being analyzed. This cellularity should then be compared to what is normally expected for these tissues. When a tumor is moderate-to-densely cellular, this step can be trivial; however, in instances of low tumor content or scattered tumor cells infiltrating white matter (e.g., gliomatosis pattern), this step can be surprisingly difficult, as these tumors can be indistinguishable from reactive processes (gliosis) seen in the brain.
Once tumor cells are recognized, a grade must be assigned. Among diffuse gliomas, this is accomplished by cataloging the presence or absence of low- and high-grade histologic features, including nuclear atypia, presence/absence of mitotic figures, microvascular proliferation, and necrosis. Grade II diffuse gliomas are typically moderately cellular tumors that can occur as both solid lesions with diffuse infiltration of adjacent brain tissues or as largely diffusely infiltrating tumors. Grade II diffuse gliomas by definition lack mitoses, microvascular proliferation, and necrosis but show increased cellularity and nuclear atypia. The presence of mitoses (a marker of cell division) in a diffuse glioma warrants upgrading to an anaplastic glioma (grade III), and if necrosis and/or microvascular proliferation are present in astrocytic lineage tumors, a diagnosis of GBM (grade IV) would be warranted.
It should be noted that there are differences in grading based on tumor lineage. For example, oligodendrogliomas are restricted to grade II and III tumors, with no grade I or IV oligodendroglioma categories; in contrast, astrocytomas are graded on a scale of I to IV. Oligodendrogliomas with necrosis and microvascular proliferation are grade III, whereas astrocytomas with the same features are considered grade IV.
In addition to features important to grading, other histologic features readily apparent on H&E sections include tumor cell morphology and growth features, which are characteristic of certain lineages. Oligodendroglioma tumor cells are classically associated with round nuclei within cells that have clear cytoplasm, resulting in a “fried egg” appearance. These tumor cells are often distributed within a network of fine capillary-like vessels reminiscent of a “chicken wire” pattern. Astrocytomas often have irregularly shaped nuclei that may be eccentrically displaced (or pushed to the side of the cell); when associated with abundant amounts of pink cytoplasm, it is commonly referred to as gemistocytic morphology.
For many decades, the assessment of histologic features by H&E has been the foundation by which tumors were defined and served as the basis for much of our current understanding of natural history of diseases. Although this approach has proven invaluable, it is not without known limitations. Such challenges include tumors that demonstrate overlapping features between different lineages, thereby creating a challenge when attempting to assign a diagnosis and grade. Additionally, classification is entirely dependent on the tissue sampled, which introduces the concept of “under sampling.” In some cases wherein neuroimaging studies highlight tumors that likely represent high-grade gliomas, on H&E sections the resected tissue may only demonstrate low-grade glioma characteristics. The discrepancy frequently results from undersampling of the more overtly malignant portions of the tumor.
In practice, H&E analysis is paired with immunohistochemistry (IHC) to support tumor classification particularly for lineage assignment and, more recently, to identify specific mutations that can be detected at the protein level. IHC is a method of applying antibodies against specific protein antigens that are evaluated using light microscopy. Common markers that are assessed by IHC include GFAP (glial fibrillary acidic protein), a cytoplasmic marker of astrocytes and glial cells, NeuN (neuronal nuclei), a marker of neuronal differentiation, and OLIG2 (oligodendrocyte transcription factor 2), a nuclear marker of glial lineage. In recent years, antibodies capable of detecting specific mutant proteins, such as H3F3A(K27M), BRAF(V600E), and IDH1(R132H), have become widely available for routine clinical use. Integrating IHC results with histologic features provides neuropathologists with critical information that is used to accurately assign a tumor lineage and grade. Additionally, as we will discuss below, determining mutational status can also provide therapeutic and prognostic information for the patient and clinical providers.
Molecular characterization of central nervous system tumors
Understanding the genomic drivers of each patient’s tumor is important to achieving the goals of precision medicine. Traditional methodologies for understanding oncogenic drivers have been limited to single gene sequencing methods that interrogate hotspot mutations in specific genes such as BRAF or IDH1/2 . Moreover, fluorescence in situ hybridization (FISH) analyses allow for gene level or arm-level evaluation to identify copy number alterations commonly associated with CNS tumors such as EGFR amplification in GBMs or chromosome 1p/19q co-deletion in oligodendrogliomas. Although these methodologies provide powerful insights into the molecular mechanisms driving gliomagenesis, they require significant tumor input if multiple probes need to be tested, and need specialized personnel for interpretation, all of which can limit the availability of testing.
The ability to extract nucleic acids (DNA and RNA) from FFPE samples has revolutionized integration of genomic data into diagnostic algorithms. Extracted FFPE DNA can now be used in massively paralleled or next-generation sequencing (NGS) assays that have largely replaced the need for single gene assays or FISH testing in brain tumors. Common approaches to NGS testing of DNA range from targeted panels that interrogate 30–500 cancer-related genes to whole-exome (WES) or whole-genome sequencing (WGS). RNA testing options range from gene expression profiling to whole transcriptome RNA sequencing. In many cases, the amount of tissue (and DNA) needed to perform a single gene assay is equivalent to that needed for a large panel that includes over 300 genes. The benefit of panel-based sequencing or WES lies in the ability to identify single nucleotide variants, small insertion-deletions (indels), rearrangements, and even, in some cases, copy number alterations in hundreds of genes in parallel compared to single gene assays.
Application of these advanced technologies has identified key genomic signatures of specific tumor lineages. Seminal studies from The Cancer Genome Atlas demonstrated that adult gliomas are driven more by copy number alterations than mutations contrary to what is observed in other malignant tumors such as lung, colon, or breast carcinomas. Indeed, adult GBMs are typically characterized by polysomy of chromosome 7, EGFR amplification, CDKN2A/B deletion, and monosomy of chromosome 10 (leading to single copy loss of PTEN ). Conversely, oligodendroglial lineage tumors harbor IDH1/2 mutations in association with chromosome 1p/19q co-deletion and frequent alterations involving CIC and FUBP1 . Young adult diffuse astrocytomas are characterized by a combination of IDH1/2 , TP53 , and ATRX mutations. In the pediatric setting, grade I pilocytic astrocytomas are frequently characterized by BRAF alterations, including point mutations ( BRAF V600E) or gene fusions (KIAA1549-BRAF) . Pediatric midline high-grade gliomas are now categorized based on their mutational profile. The presence of an H3 mutation (e.g., H3F3A K27M) in a midline diffuse glioma now warrants a grade IV designation independent of other histologic features. Focused analysis of rare glioma subtypes has also identified novel oncogenic drivers, as is the case with angiocentric gliomas, which are essentially defined by the presence of MYB alterations such as MYB-QKI fusions.
By recognizing that many brain tumors can be readily distinguished from each other based on genomic signatures, tumor classification has been refined to incorporate these data. The field is moving toward a model of an “integrated diagnosis” that combines histologic findings with those of genomic/molecular data. In this approach, tumors are classified and graded first based on H&E and IHC features which yields a histopathologic diagnosis. In parallel, a portion of tissue is subjected to NGS testing that yields genomic characteristics such as mutations, fusions, and copy number alterations identified in the tumor. The histologic and genomic findings are then collected together into a unified final and integrated diagnosis—often several weeks after a histopathologic diagnosis has been rendered.
Given the time required to perform genomic testing, treatment is typically initiated based on the histopathologic diagnosis; however, the value of additional genomic and molecular data is to enhance diagnostic accuracy, provide prognostic biomarkers (e.g., IDH1/2 mutational status), and refine the treatment plan to better align with the patient’s genomic profile. In this approach, the integrated diagnosis serves to overcome limitations associated with traditional histopathology diagnoses and ensures optimal clinical management.
Clinical cases
In the following case presentations, we will explore how histopathologic findings can be combined with genomic data to support diagnosis and prognosis, and guide clinical decision-making.
Case. A 69-year-old male presented to the emergency department with new onset seizures and rapidly progressive right sided weakness. Brain MRI with contrast revealed a centrally necrotic, multifocal ring-enhancing lesion in the left posterior frontal lobe ( Fig. 1.1 ).
Histology. Microscopic examination of H&E sections ( Fig. 1.2 A,B ) demonstrates a densely cellular glial neoplasm composed of cells with irregular and hyperchromatic nuclei associated with moderate amounts of eosinophilic cytoplasm. Mitoses are readily identified. Multifocal microvascular proliferation is present, as well as areas of multifocal necrosis including palisading necrosis ( Fig. 1.2B , black arrows). Immunohistochemistry reveals positive staining in tumor cells for GFAP ( Fig. 1.2F ), retained expression of ATRX ( Fig. 1.2D ), and negative staining for IDH1(R132H) ( Fig. 1.2C ).
Genomic and molecular analyses. Genomic profiling of tumor DNA by NGS reveals EGFR vIII amplification, TERT promoter mutation and homozygous deletion of CDKN2A/B . Alterations were not detected in the following genes: IDH1 , IDH2 , BRAF , ATRX, TP53, PDGFRA . Arm-level copy number analysis revealed polysomy of chromosome 7 and monosomy of chromosome 10. There was no evidence of 1p/19q co-deletion. Additionally, MGMT promoter methylation analysis demonstrated no evidence of methylation.
Integrated Diagnosis : GLIOBLASTOMA, WHO Grade IV, IDH1/2 wild-type, EGFRvIII amplified; MGMT promoter unmethylated
Teaching Points. This case illustrates the classic clinical, histologic, and genomic presentation of adult glioblastoma. Patients present with neurologic symptoms, typically nonspecific but which can include new-onset seizures in the fifth to seventh decades of life. Imaging studies reveal important characteristics such as location, size, enhancement, and involvement of adjacent brain tissues. The histologic analysis of the tumor in this case demonstrated characteristic features of GBM including increased cellularity, cells with irregular nuclei, and high-grade features (mitoses, microvascular proliferation, and necrosis). Immunohistochemical stains for GFAP, OLIG2, and SOX2 support a glial lineage. The genomic profile of this tumor also demonstrates the classic alterations associated with adult GBM, notably gains of chromosome 7, single copy loss of chromosome 10, and homozygous deletion of CDKN2A/B. Amplification of EGFR is detected in approximately 40% of adult GBMs and the EGFR vIII variant is seen in half of these cases. The EGFRvIII variant results from deletion of exons 2–7, which generates a constitutively activated, ligand- independent kinase that promotes cell proliferation through the PI3 kinase, RAS, and MAPK signaling pathways. Although no specific therapies have demonstrated utility against EGFRvIII , it is an active area of investigation and serves as a key player of tumor growth. Loss of one copy of chromosome 10 is the most frequently detected genomic alteration in GBM, occurring in 60–80% of cases. The tumor suppressor PTEN is located at 10q23.3 and considered a critical candidate gene in gliomagenesis as it is an important negative regulator of the PI3 kinase pathway. TERT promoter mutations are detected in 70–80% of GBMs and are associated with a worse prognosis in the absence of IDH1/2 mutations.
As standard of care, patients diagnosed with GBM are treated with combination chemo- and radiation (adjuvant) therapy (see Chapter 12, Chapter 4 for further discussion of treatment regimens for GBM). Analysis of the MGMT promoter is typically performed to evaluate the tumor’s sensitivity to the alkylating agent temozolomide. O6-methylguanine methyltransferase (MGMT) is a DNA repair enzyme that is associated with resistance to alkylating agents when expressed. Methylation of the MGMT promoter leads to reduced protein expression in tumor cells allowing for accumulation of alkylating-induced DNA damage that promotes cell death. In this tumor, the absence of MGMT promoter methylation (unmethylated) indicates decreased responsiveness to temozolomide.
In summary, this case highlights the classic pathologic features of adult GBM and illustrates how the genomics can inform the signaling pathways driving gliomagenesis and provide potential therapeutic targets and important prognostic information that can inform clinical decision-making.
Clinical Pearls
- 1.
Microvascular proliferation and palisading necrosis are pathologic hallmarks of glioblastoma and, when present, establish the diagnosis of GBM.
- 2.
IDH1/2 gene mutation is rare in glioblastoma; EGFR amplification and MGMT promoter methylation are observed in approximately 40% of GBMs.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
