Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The following colleagues provided particularly valuable assistance. Drs. Noah Brown, Sandra Camelo-Piragua, Katherine McFadden, Jeanne Bell, Larry Junck, and Ricardo Lloyd. Peter Ouillette, Sr BSA provided digital slide preparations, and assistance with Aperio. Linda Cisco carefully scanned slides. We thank our Transcription Team for their skill and patience in preparing this chapter. Mr. Mark Deming and Ms. Elizabeth Walker carefully prepared some of the illustrations. Immunohistologists and histopathologists in the University of Michigan Medical Center Pathology Laboratories prepared the fine slides.
The work described in this chapter was supported in part by NIH CA68545 and CA47558 grants awarded by the US Public Health Service.
This chapter focuses on the diagnostic immunohistochemistry (IHC) of nervous system diseases. Concise information about pathologic entities is provided in Tables 20.2 to 20.15 and in the algorithms in Figs. 20.1 to 20.3 . A suspected specific disease can be checked directly in the individual table in which its structural, IHC, and topographic features are listed; the algorithms can also be used. The text and figures elaborate on these features.
Features of unknown diseases may be found in individual algorithms and tables to assist diagnosis. For example, for a mass, specific tables in the chapter summarize differential features of the mass, as follows:
Fibrillar cells: Table 20.7 ; Figs. 20.11, 20.12, and 20.18
Epithelioid cells: Table 20.2 ; Figs. 20.4, 20.22, and 20.39
More than one type of cell: Table 20.8 ; Fig. 20.31
Small anaplastic cells: Table 20.9 ; Figs. 20.24 and 20.41 to 20.44
Syncytial cells: Table 20.11 ; Fig. 20.49
These features are evident on cytologic and histologic preparations. ,
Fig. 20.1 focuses on the differential diagnosis of clear cell lesions, Fig. 20.2 displays the differential diagnosis and IHC of epithelioid tumors, and Fig. 20.3 displays the differential diagnosis of so-called blue tumors. IHC stains that are particularly useful for diagnosis of nervous system diseases are listed in alphabetic order in Table 20.1 .
Primary Antibody, Source, and Dilution a | Principal Lesions and Tissue Components | Antigen Rescue a |
---|---|---|
A6 (CD45RO), Zymed 1:50 or CD3 |
T lymphocytes | Mw 15 min in citrate, pH 6.0 |
Chromogranin A, b 1:160 |
Pituitary adenoma, paraganglioma, neuroendocrine tumors | Mw 15 min in citrate, pH 6.0 |
Collagen type IV, Dako 1:8 |
Fibrosis, abscess, sarcoma, teratoma, fibrous cyst and vessel walls, dura, pituitary fibrovascular stroma | Ventana protease 1, 16 min |
CAM5.2 cytokeratin, BD 1:10 |
Carcinoma, craniopharyngioma, chordoma, epithelia | Ventana protease 2, 16 min |
EMA, Dako 1:50 |
Carcinoma, meningioma, craniopharyngioma, chordoma, epithelia, plasma cells, ependymomas | Mw 15 min in citrate, pH 6.0 |
GFAP, Dako 1:6400 |
Gliosis, gliomas, CNS parenchyma | Ventana protease 2, 16 min |
Herpesvirus antigen, Dako 1:1000 |
Herpes simplex encephalitis, CMV, herpes zoster | None |
IDH1 mutant | Gliomas | — |
JC virus/SV40 viral antigen, Lee Biomolecular 1:500 |
Progressive multifocal leukoencephalopathy | None |
KP1 (CD68), Dako 1:1600 |
Macrophages | Mw 15 min in citrate, pH 6.0 |
Leu-7 (CD57) | Most gliomas, particularly oligodendrogliomas, hemangiopericytomas | — |
L26 (CD20), Dako 1:500 |
B lymphocytes, B lymphoma | Mw 15 min in citrate, pH 6.0 |
MIB-1, Immunotec 1:25 |
Proliferating cells | Mw 15 min in citrate, pH 6.0 |
Neuron-specific enolase | Darkest in medulloblastoma/PNET and hemangioblastoma, most gliomas | — |
NeuN | Neuronal tumors, normal neurons | — |
CD163 | Macrophages and activated macrophages | — |
Neurofilament, Dako 1:50 |
Ganglion cell tumors, neurofibroma, PNET, Alzheimer disease, CNS parenchyma | Mw 15 min in citrate, pH 6.0 |
Neurofilament and Luxol fast blue | Brain: demyelination, necrosis | — |
p53 | Astrocytic gliomas | — |
Phosphohistone H3 | Screening for mitoses | — |
Prealbumin, Dako 1:500 |
Choroid plexus tumors | None |
S100 protein, Dako 1:500 |
Gliomas, PNET, melanoma, schwannoma, neurofibroma, neuronal and chondroid tumors, chordoma, CNS, PNS | None |
Synaptophysin, BioGenex 1:600 |
Neuronal and pineal tumors, PNET, medulloblastoma | Mw 10 min in citrate, pH 6.0 |
Toxoplasma, BioGenex neat | Toxoplasmosis | None |
Vimentin, Dako 1:800 |
Many cells, excessive in meningiomas | Mw 15 min in citrate, pH 6.0 |
a Data from McKeever PE. New methods of brain tumor analysis. In: Mena H, Sandberg G, eds. Dr. Kenneth M. Earle Memorial Neuropathology Review . Washington, DC: Armed Forces Institute of Pathology; 2004, with the expert advice and careful assistance of the immunohistology staff of the Immunoperoxidase Laboratory, Department of Pathology, University of Michigan Medical School.
b Courtesy Dr. Riccardo Lloyd, University of Wisconsin–Madison.
IHC should always be controlled. We prefer selecting a specimen with all of the following features: (1) the lesion of interest; (2) tissue with regions that should react positively to the stain; and (3) tissue with regions that should react negatively to the stain. Regions 2 and 3 serve as internal standard tissue controls. , For example, a specimen stained with glial fibrillary acidic protein (GFAP) could contain regions of gliosis (positive control) and vessels (negative control) in the same block. This approach is better than using a separate tissue control block (STCB) that probably is neither fixed nor treated exactly the same as the one in question. These differences remain when a section of STCB is placed on the same slide with the specimen, and they are aggravated if additional heat is needed to make both sections adhere to one slide.
Although normal and reactive tissues retain their expected immunophenotype, individual neoplasms may not stain for a marker generally representative of that neoplasm. , Because of this, a positive immunostaining result is more meaningful than a negative result. We will emphasize positive features in this chapter.
If a lesion is not identified immediately, a differential diagnosis may be constructed for which a group of appropriate IHC stains is described in the text, algorithms, and tables. The following example describes the application of this approach to an actual case.
Fig. 20.4 shows a tumor from the lumbosacral region of a middle-aged woman. The hematoxylin and eosin (H&E)-stained slide reveals a neoplasm with mainly epithelioid cells and a few clear cells and abundant round to oval nuclei with fine chromatin ( Table 20.2 ; see Fig. 20.4A ). Its IHC is focally positive to epithelial membrane antigen (EMA; see Fig. 20.4B ). Carcinoma, chordoma, craniopharyngioma, pituitary adenoma, and meningioma are EMA positive (see Figs. 20.1 and 20.2 ). The tumor location favors carcinoma, chordoma, and meningioma. This tumor is negative for cytokeratin (CK) CAM5.2 (see Fig. 20.4C ), so it does not fit the IHC profile of carcinoma, chordoma, or craniopharyngioma. It is negative for chromogranin A (see Fig. 20.4D ) and negative for hormones of pituitary adenomas. It was also found to be negative for GFAP, synaptophysin, and HMB-45. Table 20.2 summarizes the differential diagnosis for epithelioid cells, confirms the IHC profile for meningioma, and notes common features and locations. The tumor in the example was observed to have rare whorls and to involve the spinal meninges. In comparison with descriptions of meningiomas in the text, the cells were epithelioid and only focally syncytial. The clear cells were not prominent and lacked the cytoplasmic glycogen found in the clear cell variant. The tumor was a meningothelial meningioma with prominent epithelioid appearance. , Another useful marker for this differential would be IHC for progesterone receptors (PRs) that are present in many meningiomas (see Fig. 20.49 ).
Diagnosis a | Differential Features | ||
---|---|---|---|
Structures | Antibody b | Locations c | |
Gitter cells/xanthogranuloma | Crowded macrophages engorged with lipid vacuoles; eccentric nucleus; noncohesive cells | α-ACT (S), KP1 (+), muramidase (S) | CNS |
Ependymoma/malignant ependymoma | Structures of ependymoma or malignant ependymoma plus epithelioid cells | GFAP (S), cytokeratin (R), EMA (R) | Cerebellum, cerebrum, spinal cord, CNS |
Myxopapillary ependymoma | Cuboidal/columnar epithelium on hyaline fibrovascular papillae, variable fibrillarity | GFAP (S) | Regions of the filum terminale |
Oligodendroglioma | Round cells and nuclei with prominent perinuclear halos, nests of cells between delicate vessels | Leu-7 (+), S100 (+), GFAP (R), IDH1 mutant | Cerebrum, CNS |
Anaplastic oligodendroglioma | Features of oligodendroglioma with mitoses and pleomorphism | Leu-7 (S), S100 (S), GFAP (R), IDH1 mutant | Cerebrum, CNS |
Choroid plexus papilloma | Large mass with structure of choroid plexus | Laminin (+), cytokeratin (+), transthyretin (S), synaptophysin, IGF-2 | Fourth ventricle, lateral ventricle, CP angle, choroid plexus |
Choroid plexus carcinoma | Features of choroid plexus papilloma with anaplasia and mitoses, (necrosis d ) | Cytokeratin (+), CD44, synaptophysin, transthyretin (R) | Same as for choroid plexus papilloma |
Medulloepithelioma | Columnar epithelium with basement membrane on both surfaces, fibrovascular base for papillae and tubules | GFAP, nestin | Deep cerebrum, cauda equina, CNS |
Meningioma | Whorls, psammoma bodies, interdigitating cell processes and desmosomes, (thick collagen) d | Vimentin (+), EMA (S), S100 (R) | Falx, tentorium, meninges, choroid plexus, (extracranial) d |
Chordoma | Masses or cords of physaliphorous cells | Brachyury (+), Cytokeratin (+), S100 (+), EMA (+) | Cauda equina, clivus, spinal canal |
Craniopharyngioma | Squamous, adamantinomatous | Cytokeratin (+) | Suprasellar, sellar |
Carcinoma | Distinct margin with CNS; anaplasia, mitoses, necrosis | Cytokeratin (+), EMA (S) | Cerebrum, cerebellum, meninges, CNS, frequent multiple masses, systemic |
Melanoma | Anaplasia, mitoses, necrosis | HMB-45 (S), S100 (+), SOX10 | CNS, meninges |
a The order of tabulated lesions follows the order of discussion in text.
b Key to staining results: +, almost always strong, diffuse positivity; S, sometimes or focally positive; R, rare cells may be positive, A, absence is expected.
c The most common or most specific location is listed first.
d Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Brain biopsies for nonneoplastic diseases often require IHC combined with additional studies, such as microbiologic culture, polymerase chain reaction (PCR), Western blot, or electron microscopy (EM). Specialized centers are available to assist with interpretations of the results of these studies.
Major categories of lesions of the brain, spinal cord, and meninges—such as solitary and multiple masses, cysts, vascular malformations, and abscesses—are likely to be recognized clinically through the use of computed tomography (CT), magnetic resonance imaging (MRI), or angiography. Methods of using flow voids with CT or MRI have provided noninvasive evaluations of vessels, such as magnetic resonance angiography (MRA). MRI can be targeted on specific central nervous system (CNS) chemicals, such as choline and N-acetylaspartate, to help identify tumor tissue. Diffusion-weighted MRIs can differentiate tissues and lesions by their water movement in three dimensions.
Multiple lesions can be produced by degenerative, vascular, and infectious diseases or by neoplasms. Regarding neoplasms, the M rule for differential diagnosis of common multiple CNS neoplasms includes metastases, malignant lymphoma, melanoma, and medulloblastoma.
Depending upon its age, the tomographic density of hemorrhage is often sufficiently unique to identify it as a major component of a lesion. Calcifications and relationships with the skull are resolved well on CT, whereas gray and white matter, edema, and melanin are better seen on MRI. Vascular abnormalities are frequently defined by MRA, or, if needed, they may be defined angiographically.
Nonneoplastic lesions are often evaluated by a neurologist. Thus a major neurologic symptom (pain, weakness, or visual loss) or category of neurologic disease (e.g., dementia) may focus the differential diagnosis ( Table 20.3 ).
Symptom/Suspected Disease a | Confirmatory Features of Suspected Disease | ||
---|---|---|---|
Structures | Antibody | Locations b | |
Herpes simplex encephalitis | Encephalitis ( Table 20.4 ); Cowdry A amphophilic nuclear inclusions of 90–100 nm target capsids | Herpes simplex virus | Temporal or basilar frontal lobe(s), CNS; frequently bilateral |
Toxoplasmosis | Necrosis containing 3–5 μm tachyzoites; (cysts); (inflammation) c | Toxoplasma | CNS, frequent multiple lesions |
Progressive multifocal leukoencephalopathy | Demyelination; bizarre glia; amphophilic nuclear inclusions of 15–25 nm or 30–40 nm diameter filaments or spheres | JC virus/SV40, myelin, neurofilament, KP1 | Cerebral white matter, CNS |
Dementia/Creutzfeldt-Jakob disease | Cytoplasmic vacuoles indenting nuclei; gliosis | PrP, GFAP | Bilateral cerebral cortex, gray matter |
Small vessel disease | Vasculitis, arterial sclerosis, or congophilic angiopathy | A6, L26, CD31, amyloid, muscle actin, elastin | Cerebrum, CNS, frequent multiple lesions |
Dementia/Alzheimer disease | Argyrophilic plaques, neurofibrillary tangles of bihelical filaments | Neurofilament, tau, ubiquitin, Alz-50 | Bilateral cerebral cortex |
Demyelination | Loss of myelin; gliosis, gitter cells with or without axonal preservation | Myelin, neurofilament, KP1 | Cerebral white matter, CNS |
Epilepsy | Low-grade glioma or ganglioglioma, gliosis, or vascular malformation | GFAP, neurofilament, elastin | Cerebral cortex |
a The order of tabulated lesions follows the order of discussion in text.
b Most common or most specific location is listed first.
c Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Gliosis is a reaction of the CNS to injury of the brain or spinal cord. “Gloios” is the Greek word for “glue.” Although subtle changes occur earlier, gliosis is usually appreciated on H&E or GFAP by 2 to 3 weeks after an injury. Nearly any injury of the CNS can cause gliosis, so its presence is not diagnostic of a specific pathologic entity ( Table 20.4 ).
Diagnosis | Differential Features | ||
---|---|---|---|
Structures | Antibody | Locations a | |
Gliosis b | Cells fibrillar, uncrowded; round/oval nuclei | GFAP in stellate glial processes | CNS |
Macrophages | Cells and nuclei round to elongated; cell content reflects injury | CD68 (+), CD163 (+) | CNS, meninges |
Encephalitis/cerebritis | Perivascular mixture of inflammatory cells | CD3 (+), CD20, CD68 (+), CD163 (+), microorganism | CNS gray matter/CNS |
Hemorrhage | RBCs or macrophages with hemosiderin | Fibrin, CD68 | Deep cerebrum, cerebellum, CNS |
Margin of gliomas c | Cells fibrillar; angular nuclei indent each other; (mitoses) c , d | GFAP | CNS |
Lymphoma | Perivascular noncohesive small round cells | LCA, L26, A6, κ and λ Ig | Deep cerebrum, CNS, meninges |
a Most common or most specific location is listed first.
b Nonspecific reaction to injury.
c Suspicion of margin of glioma on frozen section should be followed by a request for another, more central biopsy. Mitoses suggest margin of a high-grade glioma.
d Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Anti-GFAP immunostain ( Fig. 20.5 ) highlights the dark-brown intense immunoreactivity, relatively low nuclear/cytoplasmic ratio, and distance between individual astrocytes in gliosis. Because gliosis accumulates with aging and varies in different neuroanatomic sites, abnormal gliosis can be hard to distinguish from gliosis within normal limits (WNL). When distinguishing abnormal gliosis from gliosis WNL is critical—for example, in a particular region of white matter in an elderly patient—an age- and site-matched normal control slide can be stained concurrently. Both the number and the density of GFAP-positive cells and cellular processes should be greater in the specimen than in the normal control.
The GFAP stain helps to identify gliosis (see Table 20.4 ). GFAP-positive cells are uniformly spaced in gliosis (see Fig. 20.5 ). This spacing of individual reactive astrocytes is more uniform than that found in the margin of an infiltrative glioma (see the “Gliomas” section later in the chapter). The nuclear/cytoplasmic ratio of gliosis is less than that of a glioma.
Dual IHC staining for Ki-67 and GFAP helps to identify gliosis by showing a proliferation of less than 4% among the total number of GFAP positive cells (<4 Ki-67+ nuclei among 100 GFAP+ cells) in a medium to high magnification microscopic field.
After 3 days, phagocytic macrophages may be seen in any destruction or irritation ( Fig. 20.6A ; see Tables 20.3 and 20.4 ). Macrophages are rich in enzymes such as α-antichymotrypsin and muramidase, and they possess markers of mononuclear phagocytic cells and thus react with antibodies CD68 (KP1) and MAC387. All of these features can be stained by using IHC. Around hemorrhages or traumatic lesions, macrophages contain iron-positive hemosiderin.
Encephalitis simply means brain inflammation. Many things cause it—from a virus to surgical implants. In cerebritis, meningitis, or encephalitis, the macrophages are pleomorphic cells. Some are thin, and others are loaded with debris (see Fig. 20.6 ); additionally, they may contain yeast and other organisms. Macrophages swollen plump by phagocytosis within the CNS are referred to as granular or gitter cells . They are large and round with a foamy cytoplasm filled with lipid droplets ( Fig. 20.7A ; see Fig. 20.6A ). Macrophages that are small cells with scant cytoplasm participate in the chronic inflammatory infiltrate centered on blood vessels in encephalitides; in glial nodules; around dying neurons (neuronophagia); and in other inflammatory, demyelinating, and degenerative processes. , CD68 stains them well (see Fig. 20.6A ).
Perivascular inflammation consists of small round cells with high nuclear/cytoplasmic ratios. These can be mistaken for lymphoma and also for neuroectodermal clusters, which are particularly common in the brains of children. Leukocyte common antigen (LCA; CD45/45R), CD3-ε, CD5, CD20, and CD79-α markers distinguish the inflammation by highlighting polyclonal reactive lymphocytes (see Fig. 20.6B and C ).
Irritation of the CNS elicits inflammation around blood vessels. CD68-positive macrophages ingest the irritant or injured cellular constituents and move them to the perivascular space. In the absence of classic lymph nodes in the brain, this perivascular region is where cells that respond to antigen intermingle. Depending on the severity and duration of the illness, the perivascular inflammation varies substantially. Old hemorrhage exemplifies a minimal response characterized mainly by perivascular macrophages laden with hemosiderin (see Table 20.4 ). Surgical wounds and implants cause substantial reactions. Viral or allergic encephalitis produces a maximal response with abundant perivascular macrophages and many CD3-ε–positive T lymphocytes.
Some diseases affect mainly veins, such as perivenous encephalitis (PVE). Others affect small arteries, such as cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). Unlike PVE, CADASIL has little or no inflammation (see the “Small Vessel Disease” section later in the chapter). Affected vessels may be distinguished with anti-smooth muscle actin (anti-SMA) or myosin, because cerebral arteries have a thicker circumferential layer of spindle-shaped smooth muscle cells (SMCs) than cerebral veins.
Gliosis is the usual slow reaction to brain injury.
Glial fibrillary acidic protein, the single most important brain immunostain, highlights the low nuclear/cytoplasmic ratios, stellate processes, and evenly separated reactive astrocytes in gliosis.
Polymorphonuclear cells, macrophages, and lymphocytes react to brain injury in a manner similar to the reaction in systemic injury. Without lymph nodes in brain, their interactions are less obvious, and their clusters are perivascular.
Common systemically, fibrosis is rare in brain tissue.
Fibrosis is rare in brain tissue, but it occurs around abscesses (see Fig. 20.7A ), in granulomas, and in desmoplastic and sarcomatous tumors. Fibrosis is more common in meninges. Meningeal fibrosis develops after traumatic injuries, meningitis, vasculitis involving meningeal vessels, and radiation therapy, and as a desmoplastic response to a tumor.
Various constituents of fibrosis can be detected with IHC stains; this includes collagen, fibronectin, and laminin. Type IV collagen works best for most brain and meningeal tissues. For routine identification of fibrosis, standard tinctorial stains such as trichrome rival IHC stains (see Fig. 20.7A ).
Coronavirus is causing a pandemic at the time of this chapter preparation. This has taken a huge toll in human sickness and death, and threatened economies worldwide. This coronavirus pandemic must be addressed. Coronavirus disease 2019 can be associated with meningitis, cerebritis, abscess, vascular thrombosis, and encephalitis. Coronaviruses are discussed in more detail in the “Histopathology” and “Organisms” sections nearby and below.
Infections may produce meningitis, cerebritis, abscess, encephalitis, or encephalopathy. Except for encephalopathy, inflammation is a prominent feature. It proceeds from acute to chronic phases, much like a systemic infection, and infection must be distinguished from lymphoma. Infections cause polyclonal inflammation and often show a prominent T-cell component that stains with IHC markers, including CD45RO, CD3-ε, and CD5 (see Table 20.1 ). Mature, EMA-positive plasma cells signify inflammation when present. On the other hand, large neoplastic cells with malignant nuclei in primary brain lymphomas are usually B cells that stain with CD20 or CD79 α (see the “Hematopoietic and Lymphoid Neoplasms” section later in the chapter). We will describe each histopathologic type of infection here and follow with discussions of the organisms that cause each type.
Meningitis is an inflammation of the meninges that cover the brain and spinal cord. Leptomeningitis affects the thin meninges, the pia and arachnoid. Pachymeningitis affects the thick dura and is less common than leptomeningitis in nonsurgical cases. Organisms access the meninges by local extension from sinuses or from the bloodstream. The perivascular space in the CNS is an extension of the subarachnoid space. Persistent meningitis travels along this space, where it can cause cerebritis or an abscess.
Cerebritis is focal inflammation of brain parenchyma ( myelitis in the spinal cord). Cerebritis precedes abscess formation but requires an early biopsy to be seen (see Table 20.4 ). The inflammatory infiltrate is composed of neutrophils, macrophages, lymphocytes, and plasma cells, with or without parenchymal necrosis. Septic cerebritis is usually caused by bacterial agents, most often streptococci or staphylococci and less commonly by gram-negative organisms, such as Escherichia coli , Pseudomonas , and Haemophilus influenzae . Cerebritis also occurs around neoplasms, ruptured vascular malformations, infarcts, and traumatic lesions.
Granulomatous forms of meningitis and cerebritis are seen in tuberculosis and atypical mycobacterial infections; fungal, parasitic, or spirochetal infections; idiopathic conditions, such as sarcoidosis, systemic lupus erythematosus, and Wegener and lymphomatoid granulomatoses; and histiocytosis X. Some diagnoses are made through biopsy and culture, and others are made through clinical correlation. ,
An abscess combines features of inflammation and fibrosis in response to a suppurative microorganism, often bacterial or fungal. A mixture of polymorphonuclear leukocytes, polyclonal T and B lymphocytes, macrophages, and plasma cells (with or without necrosis) confirms inflammation. Polymorphism of inflammatory components can be verified in difficult cases with CD45RO and CD20 IHC stains for polyclonal T and B lymphocytes, CD68 for macrophages, and EMA plus CD138 for plasma cells (see Table 20.1 ).
The wall of a brain abscess consists of a lining of CD31-positive and CD34-positive vascular tissue and collagen surrounded by highly GFAP-positive reactive gliosis. The adjacent brain is edematous. Because collagen is rare within the CNS, its presence is an important diagnostic feature of an abscess (see Table 20.1 ). Collagen may be difficult to distinguish from fibrillary gliosis on a slide stained with H&E. It can be confirmed histochemically with Masson trichrome stain or immunohistochemically with staining for collagen (see Fig. 20.7A and Table 20.1 ).
Encephalitis is inflammation of the brain tissue (see Fig. 20.6 ), which is often caused by viral or rickettsial organisms that produce a more diffuse inflammation than cerebritis. Most viral infections are self-limited and cause only meningitis or mild meningoencephalitis. The entities emphasized here require surgical attention and are more serious.
COVID-19 disease affects many tissues including brain. Reports available at the time of writing reveal endothelial inflammation (endotheliitis) in brain tissue, particularly in the basal ganglia. Microhemorrhages have been noted. Caspase-3 IHC has shown apoptosis in some endothelial cells.
In COVID-19 infection, cerebrovascular disease is often more spectacular than meningoencephalitis. Cerebrovascular diseases produced necrosis and hemorrhage in nervous tissue. Neuropathologic study of 23 brains from COVID-19 patients revealed widespread microthrombi and acute infarcts in one fourth of these brains. In clinical studies, cerebrovascular diseases occurred in about 5% of 150 COVID-19 patients.
Thus far, encephalitis has been less evident in COVID-19 infection than with many other viruses affecting brain tissue. Meningoencephalitis and microglial nodules are not prominent. Most cases show minimal inflammation. Neuronophagia, perivascular lymphocytes, and increased lymphocytes in brain tissue have been uncommon. An atypical case showed demyelinating disease in a pattern consistent with acute disseminated encephalomyelitis (ADEM).
Encephalopathy (which translates as “brain suffering”) that is caused by infection may show little or no inflammation. This is especially true for the spongiform encephalopathies caused by prions, such as Creutzfeldt-Jakob disease (CJD). Brain cell death followed by gliosis is the common feature of the encephalopathies. A leukoencephalopathy (“white brain suffering”) targets white matter. A myelopathy (“cord-medulla suffering”) generally targets the spinal cord.
Coronavirus disease 2019 (COVID-19) is a severe acute respiratory syndrome (SARS) caused by infection with the “severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)” virus. The International Committee on Taxonomy of Viruses adopted SARS-CoV-2 as the official name of this coronavirus new in 2019. SARS-CoV-2 avoids confusion with the severe acute respiratory syndrome coronaviruses (SARS-CoV) previously linked to the outbreak of SARS in 2002. Perhaps to decrease confusion among non-taxonomists, the SARS-CoV-2 virus has also been called the “COVID-19 virus.”
COVID-19 viral particles are round, ranging from 80 to 110 nm in diameter. Their surfaces show club-shaped projections of glycoprotein spikes called peplomers (peplos: a body-length garment in ancient Greece). These spikes are typical of β-coronaviruses. Spikes are seen around coronavirus particles by electron microscopy (EM), tempting an analogy with the solar corona. COVID-19 mutations have occurred, possibly accelerated by the huge number of cases. Early mutations featured more rapid person-to-person transmission.
Electron microscopy has been reported to show some SARS-CoV-2 viral particles located within dilated vesicles in neuronal cytoplasm, others in brain capillary endothelial cells. Features of coronavirus included the round shape of the virus and a cobbled surface structure having envelope projections and spikes of distinct stalk‐like peplomers. A peplomer is a glycoprotein spike on the viral surface. These protrusions bind only to certain receptors on the host cell. They are essential for viral infectivity.
There is a theory that the SARS-CoV-2 virus, similar to its predecessor SARS-CoV virus, may gain access to human cells via their angiotensin-converting enzyme 2 receptor (ACE2). IHC staining reveals the presence of ACE2 receptors on small vessels in brain tissue. This is consistent with the theory.
Every brain biopsy specimen should be handled so that if inflammation is found at surgery, it will be possible to culture the tissue for bacteria, mycobacteria, and fungi and to use special stains, immunostaining techniques, and EM. My experience has been that for organisms that grow in vitro, microbiologic culture is preferable to histochemical stains, IHC, or PCR assay if sampling of the lesion is uniform. Prior antibiotic treatment or nonuniform sampling of focal infection affects individual cases.
Fungal, bacterial, and parasitic infections are increasingly common in immunocompromised hosts. The common organisms are Cryptococcus neoformans, Listeria monocytogenes, Aspergillus fumigatus, and conventional bacteria such as H. influenzae, Streptococcus pneumoniae, Staphylococcus epidermidis, and Pseudomonas species. Cryptococcal meningitis is the most common form of fungal meningitis, but brain inflammation may be minimal in its presence. Chronic infections with these organisms produce granulomas. The organisms can be cultured or found with special stains, such as periodic acid–Schiff (PAS) and Gomori methenamine silver, but IHC analysis with specific reagents is an option for organisms refractory to culture. , Species identification can be accomplished with immunostaining.
A new threat to the nervous system came recently in bottles of preservative-free methylprednisolone acetate from a single compounding pharmacy. This produced meningitis, posterior stroke, or abscess after injection, and over 25 fatalities resulted. Most isolates were Exserohilum rostratum , a saprophyte that also infects eye and skin in immunocompromised patients (see Fig. 20.7B ). The fungus produces a dark pigment that qualifies it as a “black mold.” The Centers for Disease Control Fungus Reference Laboratory can examine fungal isolates under the microscope and confirm their identification by DNA sequencing methods (visit www.cdc.gov/fungal/index.html or call 800-232-4636).
Tuberculosis can involve any region of the CNS and its coverings. The disease usually causes granulomatous inflammation with or without caseating necrosis, meningitis, or arteritis. The extensive time required to grow mycobacteria invites preliminary testing with IHC, PCR assay, or acid-fast stains.
Syphilis is rising in incidence, predominantly among immunocompromised patients, and it is contributing to the differential diagnosis of granulomatous inflammation. The responsible organism, Treponema pallidum, is refractory to culture. Silver stains for it also stain brain, which produces background staining that confounds detection. IHC offers an alternative and much more specific test.
The most common parasitic infection of the CNS is neurocysticercosis, which prevails in developing countries. If a brain cyst contains a typical Cysticercus with a characteristic invaginated scolex, the disorder can be identified without IHC; analysis using cerebrospinal fluid (CSF) from proven cases as the source of primary antibody is available for mangled or degenerated organisms in cases for which the glycocalyx remains to be found. Schistosomiasis infects the brain and spinal cord. It can be highlighted with the readily available IHC stain for standard high-molecular-weight cytokeratin. Because there is little CK in brain, bits of organism stand out; this is one example of using a surrogate IHC marker when a more specific marker is not readily available. Whenever possible, the more specific marker is preferable.
Whipple disease rarely causes a primary brain disease without gastrointestinal symptoms. The causative bacillus is Tropheryma whipplei . The diagnosis can be made on a brain biopsy specimen evaluated by light microscopy with immunoperoxidase staining for group B streptococci. Histologic features include PAS-positive, diastase-resistant rods in macrophages; microgranulomas; perivascular CD3-ε and CD20-positive lymphocytes; and microglia reactive for CD68.
Lyme disease is caused by a tick-borne spirochete, Borrelia burgdorferi. It involves the CNS and can be detected in CSF.
The most common cause of nonepidemic encephalitis, and the one most often found on biopsy, is herpes simplex virus (HSV; see Table 20.3 ). The process is usually but not always localized to the temporal and frontal lobes. The earliest lesion is an area of vascular engorgement with ischemic changes in neurons, which is positive for HSV when an immunoperoxidase technique is performed on routine paraffin-embedded tissue. Perivascular inflammation is characteristic, composed predominantly of CD45RO-positive and CD20-positive lymphocytes mixed with CD68-positive macrophages, and is accompanied by varying degrees of focal necrosis and hemorrhage. Intranuclear inclusion bodies are consistent with HSV, but may be produced by many viruses, such as cytomegalovirus (CMV), varicella zoster, John Cunningham virus (JCV), and simian virus 40 (SV40).
Cowdry type A bodies of HSV are not easy to demonstrate in small brain biopsy specimens. This argues for sensitive and specific methods of identification such as in situ hybridization (ISH), PCR, and IHC. , EM may demonstrate viral particles within the nuclei or cytoplasm but is less sensitive and less specific. Culture and sequential serologic CSF evaluations are slow but are still the most accurate methods of diagnosis for many viral CNS infections, including HSV.
Enterovirus (Coxsackievirus, echovirus) and arbovirus infections (Eastern equine, St. Louis, West Nile) lack characteristic inclusions. West Nile virus is an arbovirus with mosquito and bird vectors and hosts, which has caused human infection. Its transmission by blood transfusion and organ transplantation has been documented, and some cases are fatal.
The findings include patchy meningitis, encephalitis, and poliomyelitis with variable involvement of the cerebrum, thalamus, basal ganglia, brainstem, and cerebellum. CD3-ε and CD68 highlight perivascular inflammation with evidence of meningoencephalitis, microglial nodules, and neuronophagia. The anterior horn cells are targeted in some patients.
Rabies produces round or oval, eosinophilic, 1- to 7-μm cytoplasmic inclusions. Immunostains and PCR assay are available for diagnosis.
Subacute sclerosing panencephalitis and milder encephalitis are caused by measles virus. Focal lymphocytic inflammation in the leptomeninges and perivascular spaces involves the cerebral cortex, with many CD4-positive cells, patchy GFAP-positive fibrillary astrocytosis, and occasional microglial nodules. Diffuse mononuclear inflammation, gliosis, and loss of myelin occur in subcortical white matter. Inclusion bodies are Cowdry type A and may be seen on H&E-stained slides; their specific identification requires IHC.
CNS lesions in acquired immunodeficiency syndrome (AIDS) reflect the entire spectrum of neuropathologic disease, beginning with cerebritis, meningitis, encephalitis, and vascular disease, and ending with degenerative-metabolic changes and neoplasia. The lesions have been summarized in several detailed reviews. , Diseases either are directly caused by human immunodeficiency virus (HIV) or are secondary opportunistic diseases resulting from immunosuppression. CNS diseases were extremely common in the early era of AIDS, with neurologic symptoms at clinical presentation and CNS abnormalities noted in more than 50% of patients. Highly active antiretroviral therapy (HAART) has altered the disease course, so fewer CNS complications occur.
HIV encephalitis can be reliably diagnosed by histologic evaluation. The hallmark of HIV encephalitis is multinucleated giant cells both in parenchyma and around vessels. They are mixed with macrophages and microglia, and they form multiple foci of various sizes within the white matter, deep gray matter, and cortex. IHC detection of HIV p23 and p24 antigen and ISH is useful ( Fig. 20.8 ).
HIV leukoencephalopathy is characterized by diffuse damage to the white matter with loss of myelin, reactive astrogliosis, multinucleated cells, and macrophages. IHC and ISH help confirm the association of HIV with the process. Leukoencephalopathy occasionally manifests as marked vacuolar myelin swelling. This finding is more common in the spinal cord, however, where it forms multiple foci of vacuolar myelopathy that resemble combined systems degeneration without pernicious anemia.
Still another manifestation of HIV infection, lymphocytic meningitis, is remarkable for heavy lymphocytic infiltrates within the leptomeninges and perivascular spaces. HIV cerebral vasculitis and granulomatous angiitis may occur with lymphocytic or lymphoplasmahistiocytic multinucleated giant cell infiltration of cerebral vessel walls, occasionally accompanied by necrosis.
Since the onset of HAART, a new form of HIV encephalitis with severe leukoencephalopathy and intensive perivascular macrophage and lymphocyte infiltrates has been described. It may be a response of the revived immune system.
Opportunistic infections are common in patients with AIDS but may also be found in other immunodeficient patients. Toxoplasmosis is the most common of these infections and manifests as a necrotizing encephalomyelitis characterized by discrete lesions that contain free trophozoites or cysts filled with parasites at the periphery of the necrotic foci. Immunoperoxidase or immunofluorescence stains pinpoint the organism, which is not easily found on routine H&E-stained sections.
CMV infection follows toxoplasmosis in frequency and varies from virtually no associated inflammation to severe necrotizing meningoencephalitis and ependymitis. IHC, ISH, and PCR assays are useful for detecting the virus in paraffin-embedded tissue if bizarre giant cells with nuclear inclusions are not evident.
Severe encephalitis results from coinfection with HIV and JCV, and tuberculosis and neurosyphilis affect patients with AIDS. , Microscopic examination reveals focal lymphoplasmacytic inflammatory infiltrates in a predominantly perivascular arrangement. Exotic CNS infections in patients with AIDS include amebic encephalitis, trypanosomiasis, and strongyloidiasis.
Lesions found to be inflammatory at biopsy should be sent sterile from the operating room to the microbiology laboratory for cultures. Cultures are more sensitive than tissue stains for nearly all microorganisms that grow in vitro .
A variety of serologic and tissue-based assays, which include immunohistochemistry, in situ hybridization, special stains, electron microscopy, and polymerase chain reaction, are available. These should be selected individually on the basis of clinical situation and suspected organisms.
Progressive multifocal leukoencephalopathy (PML) is a disease manifested as multiple discrete foci of destruction of myelin with relative preservation of axons, often with no other evidence of inflammation; radiographically, it may simulate multiple sclerosis (MS) or a mass. PML is caused by DNA papovavirus (predominantly JC papovavirus [JCP] and rarely SV40 virus) in immunodeficient patients (see Table 20.3 ). JCP has nothing in common with prions. Common underlying diseases are leukemia and AIDS. PML also occurs in patients with various types of carcinoma, tuberculosis, systemic lupus erythematosus, and sarcoidosis or after the immunosuppression associated with organ transplantation.
Brain biopsy may show a destructive process within the white matter, with multiple lipid-laden macrophages, frequent large glial nuclei with a ground-glass appearance, and many large, unusual glia with pleomorphic and hyperchromatic nuclei. Perivascular infiltrates of mature lymphocytes are prominent in some cases. The pathology of JCV infection is similar in patients with and without AIDS. However, in patients with AIDS, bizarre astrocytes are less common, and perivascular inflammatory cells are more common.
Glial nuclei are filled with virions in PML. PML should be differentiated from MS, other demyelinating disorders, and astrocytic neoplasia. Random distribution of rather uniformly distorted astrocytes among multiple lipid-laden macrophages is helpful in differentiating this lesion from an astrocytic neoplasm. Bizarre astrocytes and abnormal oligodendrocytes with large nuclei that contain inclusion bodies are characteristic. Diagnosis of PML is confirmed by EM, ISH, immunostaining ( Fig. 20.9 ), or PCR assay for JCV, SV40, and BK virus. ,
Spongiform encephalopathies are characterized by vacuoles (spongiform change) in the gray matter. Vacuoles vary in size up to 30 μm in diameter and larger ( Fig. 20.10 ) and are in the neuropil and cellular perikaryon. Their neuroanatomic distribution varies among specific diseases and in individual cases. Lack of inflammation is usual. Specimens in which a spongiform encephalopathy is suspected should be processed as described later (see the “Dementias” section).
The spongiform encephalopathies include CJD, the much-publicized mad cow disease, scrapie, kuru, Gerstmann-Sträussler-Scheinker (GSS) syndrome, and fatal familial insomnia. , , They are caused by infectious proteins known as prions, which are modified forms of normal counterpart proteins. Hereditary prion diseases, such as familial fatal insomnia, GSS syndrome, and familial CJD, have germline mutations that produce prions. Infectious prion diseases, such as mad cow disease, scrapie, kuru, and spontaneous CJD, are transmitted by intimate contact with prions. As with catalysts, these pathogenic prions propagate by inducing their ubiquitous normal counterparts to refold into the pathogenic conformation. As this cycle continues, a growing percentage of normal counterpart proteins are converted to the pathogenic configuration.
Prions are very difficult to inactivate. Agents that completely denature protein, such as bleach and strong alkali or acid (see the “Dementias” section later in the chapter), are effective, but ultraviolet light, routine formalin fixative, and standard disinfectants fail to eradicate prions.
CJD was a common diagnosis in one evaluation of cerebral biopsy specimens for dementia. Vacuoles in the neuropil and perikaryon of neurons are regionally and temporally variable in CJD (see Figs. 20.5 and 20.9 ). If not prominent, vacuoles can be overlooked or mistaken for artifacts. Spongiform changes usually diminish in late-stage disease (see Table 20.3 ). In contrast, GFAP-positive gliosis gradually increases (see Fig. 20.5 ). Immunostaining with antiprion protein antiserum is a useful tool in the identification of isoforms of this protein for the rapid diagnosis of CJD. Definitive diagnosis can be achieved by Western blot analysis for prion proteins resistant to digestion by proteinase K enzyme.
In 1996, the European Union banned imports of British beef following the mysterious deaths of young “fast food” enthusiasts in 1995 from an atypical variant of CJD. These deaths and the deaths of cattlemen with bovine spongiform encephalopathy in their herds were highly publicized. , Thus emerged mad cow disease. Microscopic plaques that stain immunohistochemically for prion protein are the most striking and consistent neuropathologic features of this atypical variant of CJD. They are even more distinctive when surrounded by spongiform change.
Hemorrhage into brain tissue has many causes and often accompanies other lesions within the CNS. The major role of IHC is to identify certain causes of hemorrhage, such as amyloid and neoplasm. Amyloid angiopathy is a common cause of spontaneous intracerebral hemorrhage in the elderly (see Table 20.3 ). IHC staining with an antibody to β/A4 protein is more sensitive than Congo red stain in demonstrating the extent of vascular amyloid.
Most neoplasms that cause brain hemorrhages are metastatic. Melanoma, renal cell carcinoma (RCC), choriocarcinoma, leukemia, and glioblastoma tend to hemorrhage ( Chapters 5, 7 , and 17 ). Glioblastoma contains GFAP-positive cellular processes and collagen-positive vascular proliferations. Carcinomas express CKs.
Hemorrhages in patients with hypertension often occur within the cerebral hemispheres, especially in the lateral areas of the basal ganglia. Coagulopathy is an important cause of intracerebral hemorrhage, including drug-induced coagulopathy. Saccular aneurysms occasionally rupture into the brain, but radiography reveals their nature. Embolism is an important cause of hemorrhagic cerebral infarcts. Sinus thrombosis followed by venous infarction may occur, usually as a complication of a preexisting infectious or inflammatory disease.
Nontraumatic subarachnoid hemorrhages are usually due to rupture of a saccular aneurysm, most often located at a branch of a major artery or in the circle of Willis. Their source is radiographically apparent.
A subdural hematoma may follow a traumatic event and is seen in elderly patients and in those with systemic cancer and brain tumors. Membranes are formed on both sides of the hematoma, and membrane formation requires several weeks to complete. The membrane on the dural side is usually 2 to 5 vimentin-positive fibroblasts thick in a 5-day-old subdural hematoma. It eventually becomes as thick as normal dura with new collagen that reacts with IHC stains for type IV collagen and fibronectin.
Brain biopsy specimens obtained in search of small vessel disease may require sectioning through the entire block of tissue to yield diagnostic material. Excessively involved vessels may not be recognizable, but CD31 shows them by highlighting their endothelial cells (see Table 20.3 ); CD31 is the endothelial marker of choice for its sensitivity and specificity. Involvement of small cerebral veins that have few spindle-shaped SMCs, compared with arteries of the same diameter with more SMCs, can be assessed with IHC for SMA or myosin. Causes are often cryptogenic in isolated angiitis of the CNS. Systemic vasculitides that may affect brain are associated with lupus erythematosus; illicit drugs, including cocaine, heroin, and amphetamines; infection, such as varicella-zoster virus and meningovascular syphilis; toxins; granulomatous disease; Wegener disease; relapsing polychondritis; and Behçet disease. , , , IHC can aid in the identification of microorganisms and classification of inflammatory cell types.
CADASIL affects small arteries and is a rare disorder that results from NOTCH3 gene mutations (chromosome 19). Characteristic vascular changes can be identified in brain, skin, and muscle biopsy specimens. By light microscopy, the affected vessels have a thickened appearance, and basophilic granular material is seen by H&E stain. This material is PAS positive and displaces the SMCs. The displacement is best seen with an IHC stain for SMA. EM reveals the presence of dark, granular, osmophilic deposits. IHC for NOTCH3 protein deposition is available.
The following five types of vascular malformations are recognized , :
Capillary telangiectasia
Cavernous angioma
Arteriovenous malformation (AVM)
Venous malformation
Sturge-Weber disease (cerebrofacial or cerebrotrigeminal angiomatosis)
Although they may occur anywhere in the CNS, AVMs have a predilection for the cerebral hemispheres ( Table 20.5 ). Elastic stains identify medium to large arteries and their abnormal counterparts. In AVMs, these stains often show abnormal vessels with focal loss or duplication of elastin. A monoclonal antielastin antibody is present, but special stains are typically used, such as Movat pentachrome stain.
Type | Location | Histology |
---|---|---|
Arteriovenous malformation | Cerebral hemispheres, brainstem, cerebellum | Veins and arteries with often poorly formed elastic membrane; gliotic brain tissue |
Venous malformation | Central nervous system, spinal leptomeninges | Veins and gliotic or normal brain tissue; no arteries |
Capillary telangiectasia | Pons, brainstem, central nervous system | Thin-walled dilated capillaries within brain parenchyma |
Cavernous angioma | Central nervous system | Clusters of abnormal, often fibrotic or hyalinized vessels with elastic lamina and without intervening brain tissue |
Abnormal smooth muscle layers can be highlighted with muscle actin. Cerebral veins are reported to have a thinner circumferential layer of SMCs than cerebral arteries, and these features exist in vascular malformations. IHC can be used to identify and localize vascular collagen, fibronectin, myofibroblasts (vimentin and muscle actin), and endothelial cells (CD31).
Primary CNS tumors arise within brain and/or spinal cord tissue itself. They may be called intraaxial tumors in brain, or intramedullary tumors in spinal cord. The most frequent examples of primary CNS tumors are all gliomas, neuronal tumors, medulloblastomas and less common embryonal tumors (ET), and CNS lymphomas.
Different neoplasms of brain predominate in adults and children. More pediatric neoplasms occur in the posterior fossa than in the anterior fossa, and the opposite is true of adult neoplasms. Primary CNS tumors arise within brain and/or spinal cord tissue itself. They may be called intraaxial tumors in brain or intramedullary tumors in spinal cord. The most frequent examples of primary CNS tumors are all gliomas, neuronal tumors, medulloblastomas and less common embryonal tumors (ETs), and CNS lymphomas.
Different brain tumors predominate in adults and children. More pediatric neoplasms occur in the posterior fossa than in the anterior fossa, and the opposite is true of adult neoplasms. Different tumors occur near the midline, surface, and ventricles of brain. IHC and molecular markers have provided insights into the dynamics and prognostic importance of tumors that occur in these various sites.
Many changes in classification of brain tumors, for example incorporation of IDH status in diagnoses of diffuse gliomas, have been recommended in the last few years. This has resulted in new problems classifying CNS tumors. In medicine, not otherwise specified (NOS) separates a general diagnosis from a specific diagnosis. For example, the general diagnosis based on symptoms of “respiratory infection, NOS” would change to the specific diagnosis of “SARS-CoV-2” infection after confirmation by laboratory testing. When there is a definitive test for the disease, it is performed, and it is positive, the specific diagnosis can be made. But when a specific diagnosis cannot be made, the question confronting pathologists is whether NOS is a term too broad to describe all brain tumor interpretations. Perhaps NOS should be confined to situations where the definitive test (or tests) were not done, or a definitive test failed. Other situations where definitive tests were done but they did not match a known diagnosis, might be better to call something else.
The Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy-Not Official WHO (cIMPACT-NOW) addressed these conundrums. The cIMPACT-NOW suggested a new term for situations where definitive tests were done, but they did not match a known diagnosis. Not elsewhere classified (NEC) is proposed for these tumors. The term NEC is proposed as a way to alert clinicians about a tumor that may behave differently than expected from its general diagnosis. It may prompt pathologists to find new markers better for prognostication of patient outcome. In a way, NOS or NEC represent an imperfect diagnosis waiting for new diagnostic approach. Full histologic and molecular data are recommended to better inform clinicians treating the patient.
In summary, NOS has a more narrow meaning if NEC are split from it. For a CNS tumor, NOS also infers that not all tests needed for the most specific diagnosis were done. On the other hand, NEC confirms performance of all tests needed, and adds that these test results do not conform to a recognized diagnosis.
Beginning with the most benign as grade 1, numeric grades 2, 3, and 4 represent increasing tumor aggression.
The World Health Organization (WHO) has established uniform terminology and grading of brain tumors according to histologic criteria. Current guidelines recommend replacing previously used Roman numerals with Arabic numerals for denoting grades. We are grateful that, for whatever reason, the WHO has accepted Arabic numbers. One of us was bitten by the II-III snake a few times in the 1980s. He reported in Arabic for a while to clarify communications, as others may have done.
When helpful, the numerical grades assigned by the WHO classification are included in parentheses after the tumor names in headings in this section. The most important aids to assessing grade of malignancy provided by IHC are as follows , :
Cell type identification with markers ( Tables 20.6 and 20.7 ; see also Tables 20.2 and 20.8 to 20.12 ). Gliomas usually show at least one of these markers: GFAP, S100 protein, a mutation in isocitrate dehydrogenase (IDH; see “Gliomas” section later), or a mutation in the α-thalassemia mental retardation X-linked gene ( ATRX ; see the “Diffuse Astrocytoma” section later). Within the craniospinal vault, four of the markers are relatively specific for gliomas. S100 protein is common among, but not specific for, gliomas. Markers of nonglial tumors vary widely and are noted separately with each tumor.
Meningioma (Grade 1) | Atypical Meningioma (Grade 2) | Anaplastic Meningioma (Grade 3) |
---|---|---|
<4 mitotic figures per 10 hpf (0.16 mm ) | Increased mitotic activity: 4–19 per 10 hpf (0.16 mm ) | Increased mitotic activity: >19 per 10 hpf (0.16 mm ) |
Fails to meet diagnostic criteria to right | Or >2 of the following:
|
Or malignant and/or anaplastic cytologic appearance (e.g., resembling sarcoma, carcinoma, melanoma) |
Fits into a well-differentiated subtype | (WHO grade 2 is assigned to clear cell and chordoid meningiomas) | (WHO grade 3 is assigned to rhabdoid and papillary meningiomas) |
Diagnosis a | Differential Features | ||
---|---|---|---|
Structures | Antibody b | Locations c | |
Fibrosis | Spindle cells of meningeal or perivascular origin | Collagens (+), Vimentin (+) | Meninges, CNS |
Granuloma | Like fibrosis with “whorls” and inflammation | Microorganisms (see Table 20.1 ) | Basal meninges, CNS |
Pilocytic astrocytoma | Hypercellularity, hair-like fibrillarity, Rosenthal fibers, microcysts | GFAP (+), S100, α B-crystallin (S), ubiquitin (+) | Cerebellum, thalamus/hypothalamus, optic nerve, CNS |
Astrocytoma | Hypercellularity; angular nuclei cluster and indent one another; infiltrates CNS | GFAP (S), S100, p53, IDH1 mutant (+) ATRX (A) |
Cerebrum, brainstem, spinal cord, CNS |
Anaplastic astrocytoma | Increase in above features; mitoses | GFAP (S), p53, IDH1 mutant (S) | Cerebrum, brainstem, CNS |
Gemistocytic astrocytoma | Hypercellularity; cells swollen with hyaline pink cytoplasm and eccentric pleomorphic nuclei; infiltrates CNS | GFAP (S), p53, IDH1 mutant (S) | Cerebrum |
Giant cell astrocytoma | Giant astrocytes with thick fibrils; large round/oval nuclei | GFAP (S) | Lateral ventricle, subependymal |
Astroblastoma | Perivascular rosettes with expanded glial cell processes | Nonfibrillar GFAP (S) | Cerebrum, CNS |
Pleomorphic xanthoastrocytoma | Pleomorphic cells often vacuolated | GFAP (S), BRAF, collagen IV (S) | Leptomeninges, cerebral cortex |
Ependymoma | Hypercellularity, ependymal or perivascular rosettes or both, round/oval nuclei, cilia, basal bodies | GFAP (+), EMA (R) | Cerebrum, cerebellum, spinal cord, CNS |
Tanycytic ependymoma/ subependymoma | Combination of astrocytoma and ependymoma; round/oval nuclei cluster among fibrillar mats | GFAP (+) | Spinal cord, fourth ventricle, subependymal, CNS |
Anaplastic ependymoma | Preceding features with mitoses; necrosis | GFAP (S), EMA (R) | Cerebrum, cerebellum |
Glioblastoma multiforme | Regions of coagulation necrosis; mitoses; pleomorphism; endothelial proliferation | GFAP (S), S100, p53, IDH1 mutant (S), EGFR (+), ATRX | Cerebrum, CNS |
Gliosarcoma | Glioblastoma plus fibrosarcoma intermixed | GFAP (S), collagens (S), vimentin (S) | Cerebrum |
Ganglion cell tumors | Binucleated & pleomorphic neurons; Dx depends on glia and neurons | GFAP (S), synaptophysin (S), PGP9.5, neurofilament (S), collagen IV | Cerebrum, CNS |
Central neurocytoma | Round cells and nuclei, thin fibrils near vessels | Synaptophysin (+), neuN (S), neurofilament (R) | Septum pellucidum, lateral ventricles |
Pineocytoma | Normal pineal structures | Synaptophysin (+), neurofilament (R) | Pineal |
Fibroblastic meningioma | Spindle cells, interdigitating cell processes and desmosomes, (thick collagen), (whorls) d | Vimentin (+), EMA (S), PR (S), S100 (R) | Falx, tentorium, meninges, choroid plexus |
Fibrosarcoma/undifferentiated pleomorphic sarcoma | Hypercellular; pleomorphic spindle cells and nuclei, mitoses, necrosis | Vimentin, collagen | Meninges, CNS |
Schwannoma | Verocay bodies, Antoni A and B; thin pericellular basement membrane | S100 (+), SOX10 (+), collagen IV (+), GFAP (R) | Eighth cranial nerve, spinal roots, PNS |
Neurofibroma | Multiple cell types spread axons | Neurofilament (R), EMA, S100 (+), Leu-7 | Spinal root, PNS, cranial nerve |
Histiocytosis | Sheet-like pattern of macrophages, fibroblasts, and leukocytes | α-ACT, S100 (S) | Parasellar, CNS, systemic |
Melanoma | Anaplasia, mitoses, necrosis | HMB-45 (S), S100 (+) | CNS/meninges, frequent multiple metastases, systemic |
a The order of tabulated lesions follows the order of discussion in text.
b Key to staining results: +, almost always strong, diffuse positivity; S, sometimes or focally positive; R, rare cells may be positive.
c Most common or most specific location is listed first.
d Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Diagnosis a | Differential Features | ||
---|---|---|---|
Structures | Antibody b | Locations c | |
Glioblastoma/gliosarcoma with epithelial metaplasia | Structures of glioblastoma/gliosarcoma ( Table 20.7 ) plus epithelial regions | GFAP (S), S100 (S), cytokeratin (S), EMA (S), p53, IDH1 mutant, EGFR, ATRX | Cerebrum, CNS |
Ganglion cell tumors | Binucleated and pleomorphic neurons plus glioma ( Table 20.7 ) plus fibrosis plus inflammation | GFAP (S), synaptophysin (S), neuN (S), neurofilament (R), type IV collagen (R) | Cerebrum, CNS |
Desmoplastic medulloblastoma | Regions of medulloblastoma and fibrosis | Synaptophysin (+), S100, reticulin (+), neurofilament (R) | Lateral cerebellum, CNS, meninges, (extraaxial) d |
Transitional meningioma | Regions of fibrous ( Table 20.7 ) and syncytial ( Table 20.11 ) meningioma | Vimentin (+), PR (S), EMA (S), S100 (R) | Falx, tentorium, meninges, choroid plexus |
Hemangioblastoma | Multivacuolated stromal cells between many capillaries; hypervascularity; (fibrillarity is frozen section artifact) d | CD31 (S), factor VIII (S), NSE (S) | Cerebellum, spinal cord, CNS |
Desmoplastic carcinoma | Regions of carcinoma ( Table 20.2 ) and fibrosis ( Table 20.7 ); occasional inflammation | Cytokeratin (S), EMA (S) | Cerebrum, cerebellum, meninges, CNS, frequent multiple masses, systemic |
Melanoma | Regions of fibrillar and epithelioid melanoma ( Tables 20.7, 20.2 ) | HMB-45 (S), S100 (+) | Cerebrum, cerebellum, meninges, CNS, frequently multiple masses, systemic |
a The order of tabulated lesions follows the order of discussion in text.
b Key to staining results: +, almost always strong, diffuse positivity; S, sometimes or focally positive; or mixed cell populations; R, rare cells may be positive.
c Most common or most specific location is listed first.
d Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Diagnosis | Differential Features | ||
---|---|---|---|
Structures | Antibody a | Locations b | |
Embryonal tumor with multilayered rosettes (Ependymoblastoma) | Like PNET; ribbons/cords of cells, true ependymal rosettes | Vimentin (S), GFAP (R) | Cerebrum, cerebellum |
Medulloblastoma/pineoblastoma/neuroblastoma/PNET | Slight fibrillarity, (Homer Wright rosettes), (palisades), “carrot” nuclei, (neural or glial foci) c | Synaptophysin (+), neuN (S), S100, neurofilament (R), GFAP (R) | Cerebellum, brainstem, pineal gland, CNS, (extraaxial) c |
Rhabdomyosarcoma/medullomyoblastoma | Like PNET; muscle striations | Desmin (S), muscle-specific actin | Pineal, cerebellum, CNS |
Atypical teratoid- rhabdoid tumor | Like PNET; more cytoplasm | INI1 = SMARCB1(A), Vimentin (+), GFAP (S), cytokeratin (S), EMA (S) | Cerebellum, brain |
Hemangiopericytoma | Hypercellularity, thick pericellular matrix, mitoses | Vimentin (+) | Falx, tentorium meninges, (extracranial) c |
Lymphoma | Noncohesive round cells, vascular wall invasion | L26 (+), LCA (S), monoclonal κ and λ Ig CD20 (+) |
Deep cerebrum, CNS, meninges; may be multiple |
Small cell carcinoma | Cohesive cells, (epithelioid), (desmosomes) c | Cytokeratin (+), EMA, synaptophysin (S) | CNS, meninges; frequent multiple masses; systemic |
a Key to staining results: +, almost always strong, diffuse positivity; S, sometimes or focally positive; R, rare cells may be positive.
b Most common or most specific location is listed first.
c Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Tumor Type | WHO Grade | Subgroup | Common Genetic Alterations | Valid IHC for Altered Gene |
---|---|---|---|---|
Glial Tumors | ||||
Astrocytic Tumors | ||||
Glioblastomas—adults | 4 | Proneur CIMP+ | IDH1/2, TP53, ATRX (Secondary GBM) | IDH1 (R132H), ATRX |
4 | Proneur CIMP- | PDGFRA, TP53 (Primary GBM) | — | |
4 | Classic | EGFR, TP53 (Primary GBM) | — | |
4 | Mesenchymal | NF1, TP53 (Primary GBM) | — | |
4 | Neural | EGFR, TP53(Primary GBM) | — | |
4 | No subgroup | PTEN, PI3K, RB1, CDKN, FGFR, MDM2, MET, TERT promoter | — | |
Epithelioid glioblastoma | 4 | — | BRAFV600E | BRAF V600E |
Glioblastomas—children/young adults | 4 | — | H3K27M, ACVR, MYC, H3G34R/V, TP53, ATRX | H3K27M, H3G34R/V, ATRX |
Anaplastic astrocytoma | 3 | — | IDH1/2, TP53, ATRX | IDH1 (R132H), ATRX |
Diffuse astrocytoma | 2 | — | IDH1/2, TP53, ATRX | IDH1 (R132H), ATRX |
Low-grade gliomas—child/young adult | 2 | — | BRAF, FGFR1, MYB | — |
Pleomorphic xanthoastrocytoma | 2 | — | BRAFV600E | BRAF V600E |
Pilocytic astrocytoma | 1 | — | BRAF-KIAA1549, other BRAF fusions | — |
Oligodendroglial Tumors | ||||
Anaplastic oligodendroglioma | 3 | — | 1p/19q del, IDH1/2, TERT promoter | IDH1 (R132H) |
Oligodendroglioma | 2 | — | 1p/19q del, IDH1/2, TERT promoter | IDH1 (R132H) |
Ependymal Tumors | ||||
Supratentorial ependymoma | 2/3 | Fusion | C11Orf95-RELA fusions YAP1 fusions |
— |
2/3 | Non fusion | — | — | |
Posterior fossa ependymoma | 2/3 | GroupA | Global reduction in H3K37me3 | H3K27me3 |
2/3 | GroupB | Retained H3K27me3 | H3K27me3 | |
Neuroglial and Neuroepithelial Tumors | ||||
Ganglioglioma | 1 | — | BRAFV600E | BRAF V600E |
Dysembryoplastic neuroepithelial tumor | 1 | — | FGFR1, BRAFV600E | BRAFV600E |
Rosette-forming glioneuronal tumor | 1 | — | FGFR1, PI3KCA | — |
Angiocentric glioma | 1 | — | MYB | — |
Pineal Tumor | ||||
Pineoblastoma | 4 | — | DICER1 | — |
Papillary tumors of the pineal region | 2/ 3 | — | PTEN | — |
Embryonal Tumors | ||||
Medulloblastoma | 4 | SHH | PTCH, SUFU, GLI2, TP53 | GAB1, YAP1 |
4 | WNT | CTNNB1, TP53 | β-Catenin | |
4 | Group 3 | MYC, GFI1/1B | — | |
4 | Group 4 | MYCN, CDK6, SNCAp1 | — | |
CNS PNET | 4 | CNS neuroblast | FOXR2 | — |
4 | CNS Ewing | CIC | — | |
4 | CNS hi-gr NET | MN1 | — | |
4 | CNS hi-gr NET | BCOR | — | |
Atypical teratoid rhabdoid tumor | 4 | Grp1, Grp 2 | SMARCB1, SMARCB4 | SMARCB1, SMARCB4 |
Meningiomas | I/2/ 3 | — | NF2, TRAF7, KLF4, AKT1, SMO, FAM109B, TPRXL | — |
Craniopharyngioma | 1 | — | BRAF, CTNNB1 | — |
Diagnosis | Differential Features | ||
---|---|---|---|
Structures | Antibody a | Locations b | |
Meningiomas | Whorls, psammoma bodies, interdigitating cell processes and desmosomes, (thick collagen) c | Vimentin (+), EMA (S), PR (S), S100 (R) | Falx, tentorium, meninges, choroid plexus, (extracranial) c |
Anaplastic meningioma | Decrease in preceding features, mitoses, necrosis, CNS invasion | Vimentin (+), EMA (R), S100 (R) | Same as above |
Hemangiopericytoma | Hypercellularity, thick pericellular matrix, mitoses | Vimentin (+) STAT6 |
Same as above |
a Key to staining results: S, sometimes positive; R, rare cells may be positive; +, almost always strong, diffuse positivity.
b The most common or most specific location is listed first.
c Parentheses around a differential feature indicate an uncommon feature that is very useful in differential diagnosis when found.
Disease | Characteristic Histology | CD68 (KP1) | S100 | CD1a | Birbeck Granules (EM) |
---|---|---|---|---|---|
Macrophage | Foamy epithelioid, multinucleated giant cells | + | – | – | – |
Erdheim-Chester | Touton giant cells | + | S a | – | – |
Rosai-Dorfman | Emperipolesis | + | + | – | – |
Langerhans histiocytosis | Reniform nuclei, eosinophilic cytoplasm | + | + | + | + |
a S100 protein has been positive in some, but not all, cases of Erdheim-Chester disease.
Identification of tumor infiltration. A rule of thumb for grading is that primary brain tumors that have a distinct margin with CNS tissue tend to be grade 1. In contrast, infiltrating primary brain tumors invade CNS tissue and are called “diffuse” tumors. They tend to be grade 2 and higher, and include grade 4 medulloblastomas and CNS lymphomas. Neurofilament (NF) and synaptophysin stains aid assessment of tumor infiltration of CNS by staining preexisting axons, especially in white matter, and preexisting synapses in gray matter. With a good hematoxylin nuclear counterstain, infiltrating neoplastic cells are evident with such stains. Eosin can be applied in addition to hematoxylin if needed.
Identification of vascular tumors and microvascular proliferations in gliomas can be done with stains for collagen like Masson trichrome, Movat, and collagen type IV IHC. CD31, CD34, Ulex europaeus ( Ulex ), or factor VIII (FVIII) highlight endothelial markers. Anti-SMA IHC localizes vascular smooth muscle cells in media.
Mitotic activity is key to grading certain tumors. For now, H&E is the standard for certifying mitoses. Over 5 mitoses per 10 high-power fields (HPF) determine grade 3 for oligodendroglioma. Meningioma grade 1 has less than 4 mitoses per HPF. A standard grade 2 atypical meningioma has more than 3 mitoses per HPF, while a grade 3 meningioma has over 19 mitoses per HPF. These and more brain tumor diagnoses rely heavily on mitotic indices.
A marker, phosphohistone H3 (PHH3), stains mitotic figures, but it often stains questionable chromatin in cells or interphase nuclei. Consecutive serial sections (CSS) of tissue stained like a proliferation sandwich: CSS1-H&E, CSS2-PHH3, CSS3-H&E take advantage of the low-to-medium power screening capability of PHH3 for proliferating cell reconnaissance, and of the medium-to-high power finer structural resolution of H&E for confirmation of mitotic figures.
Stated another way, the PHH3 assists in screening all candidates, and the better chromosomal detail in the H&E stain certifies mitoses and helps to cull possible mimics.
Proliferation markers , such as molecular immunology Borstel 1 (MIB-1), the first marker of Ki-67 that worked on paraffin sections, supplement mitotic activity by stain ing mitotic and nonmitotic proliferating cells ( Fig. 20.11 ). Since Ki-67 shows all proliferating cells, it is especially useful for interpreting low-grade tumors that show few or no mitotic figures, or are diffusely invasive with neoplastic cells hidden among normal and reactive cells. Various companies now supply antibodies to Ki-67 that work on paraffin sections, and the term Ki-67 has gained renewed popularity.
Ki-67 shows proliferative cells in all phases of the cell cycle. A proliferation index (PI) can be derived from them. , The PI is the number of antigen-positive cells divided by the total number of cells in sampled microscopic areas of the tumor.
Properly standardized, the Ki-67 PI helps to predict outcomes of patients who have specific types of tumors. , , Tissue processed, cells sampled, and PI assessed the same way from a group of the same type of tumor is needed to establish a baseline to glean the best prognostic information from the PI of an individual patient’s tumor. Laboratories vary in their range of staining and PI assessments, which limits the value of comparison with published data.
Ki-67 can be used in a manner similar to PHH3 to find mitoses: Consecutively stained serial sections (CSS) of tissue such as CSS1-H&E, CSS2-Ki67, CSS3-H&E take advantage of the Ki-67 low-to-medium power screening capability for proliferating cells, and the medium-to-high power finer structural resolution of H&E for reconnaissance and confirmations of mitotic figures.
Apoptosis is the programmed death of cells. From cytologic and cytochemical assays the pathologist can determine an apoptotic index analogous to the PI for proliferation. The balance between cell proliferation and cell death affects tumor growth.
Glioma is a general term that describes a glial tumor: astrocytoma, glioblastoma, ependymoma, oligodendroglioma, and their various subtypes and combinations. An important general rule is that gliomas tend to contain GFAP and to lack CK, collagen, reticulin, and fibronectin in their parenchyma, which distinguishes most of them from nonglial neoplasms. , Uncommon variants such as pleomorphic xanthoastrocytoma (PXA) may have parenchymal reticulin (see Table 20.7 ). Oligodendroglioma cells are more variable in their GFAP expression, and they uniformly express only glial proteins of low specificity, such as Leu-7 and S100 protein (see “Oligodendroglioma,” “Theranostic,” and “Genomic Applications”). , Gliomas lack widespread CK in their parenchyma but have been misinterpreted because of cross-reactivity between some anti-CK antibodies and GFAP.
Diffuse glioma is a term that mainly restricted to astrocytic and oligodendroglial tumors, grades 2 to 4, that infiltrate CNS tissue. Glioblastomas and certain other gliomas with lethal genetic abnormalities like H3 K27M are considered grade 4. People sometimes focus on astrocytic and oligodendroglial diffuse gliomas grades 2 to 4 when they speak of “gliomas.” This is incorrect because it ignores ependymomas and grade 1 gliomas that show macroscopically distinct margins with CNS tissue. A glioma is a brain tumor that likely began in a CNS glial cell, including an ependymal cell.
The histologic term low grade, as applied to astrocytomas and other gliomas, does not necessarily imply a benign neoplasm or even a favorable prognosis. A benign designation, which implies that the glioma will not recur once removed, is frequently encountered only among WHO grade 1 astrocytoma, neuronal tumor, and ependymoma subtypes. Even these tumors need to be in favorable locations where they can be completely resected, thus giving the patient a good chance for cure.
Studies of the molecular biology of gliomas show that some grade I gliomas and gliomas that do not infiltrate brain tend to show genetic alterations in the BRAF gene (v-raf murine sarcoma viral oncogene homolog B1). These gliomas do not show IDH mutations, over-express TP53 gene product, or over-express epidermal growth factor receptors (EGFRs). In contrast, high-grade and infiltrative gliomas tend to over-express at least one of these. ,
Postoperative systemic thromboses are a major complication of brain tumor surgery. The pathologist can identify patients likely to encounter this difficulty by reporting the tumors, usually malignant gliomas, that contain thrombosed vessels.
Discoveries of major importance are the frequency of mutations in IDHs and their effect on prognosis in gliomas. The IDH enzymes normally catalyze the oxidative carboxylation of isocitrate to α-ketoglutarate, which results in the formation of nicotinamide adenine dinucleotide phosphate from NADP, an important antioxidant. For some reason, IDH genes mutate frequently in diffuse gliomas in younger patients (30 to 40 years). Approximately 70% of intermediate grade (2 and 3) astrocytomas, oligodendrogliomas, and 5% to 10% of glioblastoma (GBM; termed secondary GBM ) exhibit IDH mutations. There are three isoforms of IDH enzymes: IDH1 is cytoplasmic, whereas IDH2 and IDH3 are mitochondrial. IDH1 and IDH2 mutations occur in gliomas with IDH1 approximately 10 times more often than IDH2 mutations. The R132H mutation is the most common mutation in IDH1 . An antibody was developed to this particular mutant because of its high frequency and potential diagnostic and prognostic utility. The IHC stain for IDH1 R132H highlights an IDH1 mutation about 90% of the time among gliomas that have IDH mutations. PCR for other IDH1/2 mutations finds nearly 100% of all IDH mutations when the percentage of glioma cells in the specimen is adequate. IDH1/2 mutation distinguishes many gliomas from gliosis. The presence of an IDH1 mutation in a glioma confers better patient survival.
Major changes in classification of gliomas have been driven by the need for more precise prediction of patient outcomes (i.e., which patients do better). Gliomas that look the same under the microscope can have favorable gene abnormalities that impact patient outcome. For example, adults with a diffuse astrocytic glioma of a given grade (2, 3, or 4) and with a mutation in the IDH gene tend to do better than adults with that same grade of astrocytic glioma, but no mutation in IDH (IDH-wild-type). A recent classification has separate categories for each grade of common diffuse astrocytic glioma with an IDH mutation and without an IDH mutation.
Tumor progression has been documented in diffuse gliomas in adults. Progression is associated with an increase in the grade of malignancy from grade 3 glioma up to grade 4 “secondary” glioblastoma (secondary to a lower-grade glioma). Tumor progression is often associated with mutations in TP53, ATRX, and IDH genes. Although progression results in a higher-grade glioma, overall patient survival including time spent with the lower grade-glioma tends to be long. In addition to this, glioblastomas with IDH gene mutations tend to occur in young patients and have a better prognosis than the majority of glioblastoma patients.
It is important to recognize two types of specimens of glioma ( Fig. 20.12 ). The first type is the tumor itself (see Tables 20.2, 20.7, and 20.8 ), which has cellular density exceeding that of surrounding brain (see Fig. 20.12B ). This tumor nidus is optimal for histopathologic classification. ,
The second type of specimen is brain tissue infiltrated by the margin of the glioma, and it is a product of the infiltrative nature of many gliomas. IHC stains for brain neuroanatomic components are helpful in identifying this brain tissue. NF protein localizes axons in white matter, where axons are neuroanatomically oriented in parallel, and also in gray matter. The extent that glioma cells infiltrate this axonal meshwork in brain tissue is evident from the hematoxylin counterstain in IHC preparations ( Fig. 20.13 ). Synaptophysin stains a finely pixelated “carpet” of synapses in gray matter; glioma cells disrupt this carpet.
If only the margin is available for examination, it is often impossible to determine the histologic grade and type of glioma giving rise to an infiltrative margin of neoplastic glia. Further from the glioma itself, neoplastic glia in CNS parenchyma are difficult to distinguish from gliosis ( Fig. 20.14 ; see Fig. 20.12A ). Nonetheless, GFAP can help identify gliosis by showing excess cytoplasmic GFAP and regular spacing between cells in gliosis (see Fig. 20.5 ).
Proper surgical sampling is needed for accurate classification and grading of diffuse or heterogeneous gliomas. Intraoperative consultation of pathologist and surgeon optimizes sampling.
Most grade 2 to 4 gliomas infiltrate CNS tissue, which makes total resection difficult to impossible. Exceptions include certain ependymomas and xanthoastrocytomas.
The proliferation marker Ki-67 augments grading and prediction of patient outcome.
The current update on the WHO classifications of CNS tumors recommends including the IDH mutational status in the diagnosis of adult gliomas.
Astrocytomas are among the most fibrillar of the CNS neoplasms, more fibrillar than other gliomas except tanycytic ependymomas and subependymomas (see Table 20.7 ). Astrocytomas nearly always contain GFAP ( Fig. 20.15A ; see also Fig. 20.12A ), although the amount is variable. GFAP is the single most important IHC marker and distinguishes astrocytomas from nearly all nonglial neoplasms. Nerve sheath tumors occasionally show focal GFAP in substantially lesser amounts than the fibrillary astrocytomas that resemble them. Many astrocytomas express vimentin, and when they do, this feature distinguishes them from vimentin-negative oligodendrogliomas.
Pilocytic is defined as “composed of hair cells,” which is a major feature of the pilocytic astrocytoma. Parallel bundles of elongated, fibrillar, cytoplasmic processes resemble mats of hair (see Fig. 20.15A ). These hair-like processes contain large amounts of glial fibrils, which stain well with immunoperoxidase for GFAP (see Table 20.7 ).
A diagnosis of pilocytic astrocytoma is good news within the group of astrocytomas. This tumor has a better prognosis than its diffuse counterparts, especially when it occurs in the cerebellum rather than its other common location near the third ventricle. , It is critical to distinguish pilocytic astrocytoma from fibrillary grade 2 astrocytoma, which has a poorer prognosis. Even its better prognosis is tempered, because adequate surgical removal of a pilocytic astrocytoma depends on its location, and some develop as multicentric disease. The 10-year survival rate of patients with supratentorial tumors is 100% after gross total resection and 74% after subtotal resection or biopsy. Pilocytic astrocytomas rarely manifest malignant degeneration, indicated by hypercellularity, mitoses, and necrosis.
Pilocytic astrocytomas have a well-demarcated MRI appearance; microscopically, some have discrete margins, but many incorporate elements of brain at their margins. Still, diffuse grade 2 astrocytomas (see discussion later in this section) infiltrate brain to a much greater extent than pilocytic astrocytomas. The extent of microscopic infiltration can be evaluated by comparing GFAP staining in serial sections to identify the edge of the highly GFAP-positive tumor and NF protein staining to identify axons of brain tissue at the edge of the tumor. Pilocytic astrocytomas show few axons in the tumor side of their margins, but grade 2 astrocytomas show many. A nearly even mix of axons and neoplastic cells signals a grade 2 astrocytoma. Evaluate the center of the largest tissue pieces in a scalpel biopsy. This is where a pilocytic astrocytoma has rare to no axons, whereas a diffuse grade 2 astrocytoma infiltrates between axons ( Fig. 20.16 ).
Most, but not all, pilocytic astrocytomas occur in children and young adults. They are most abundant in the posterior fossa and around the third ventricle, thalamus, hypothalamus, neurohypophysis, and optic nerve. Cerebral hemispheric pilocytic astrocytomas are less common, but it is important to recognize them to ensure appropriate treatment.
Rosenthal fibers (RFs) are highly eosinophilic hyaline structures that are round, oval, or beaded and have slightly irregular margins. Their beaded appearance results from their formation within glial processes. In comparison with erythrocytes, they are pink, rather than orange, and have greater variation in size and shape. In CNS, gliosis with RFs is called “piloid gliosis.” Although RFs assist in distinguishing the pilocytic astrocytoma from other astrocytoma variants, they are hard to use for differentiating diffuse astrocytomas from gliosis in CNS, because they occur in both abnormalities. If a generous specimen is intact, a halo of piloid gliosis in CNS around a RF-negative tumor may support lack of neoplastic RFs. RFs are common in gliosis around craniopharyngiomas and hemangioblastomas. RFs are also seen in spinal syrinx and Alexander disease.
IHC reveals that RFs contain α B-crystallin, stain with ubiquitin, and are centrally GFAP negative (see Fig. 20.15 ). α B-crystallin is a lens protein in the small heat-shock protein family. When abundant RFs are easily appreciated on H&E stain, IHC is not needed for other than confirmation. IHC is most needed diagnostically when RFs are scarce, and the ideal marker for scarce RFs would be both sensitive and specific. α B-crystallin IHC has improved to rival the sensitivity of ubiquitin, a sensitive but not as specific a marker as α B-crystallin. We use ubiquitin to screen for scarce RF, which should be confirmed with consecutive sections stained for α B-crystallin and H&E. Structural features and color on H&E are often sufficient to identify RFs.
Eosinophilic granular bodies (EGBs) are droplets of protein often found in pilocytic astrocytomas. These eosinophilic protein droplets are usually smaller and more aggregated than RFs and are usually intracellular, but occasionally are extracellular, and they are up to 40 μm in diameter. They are PAS positive (see Fig. 20.15B ). Both EGB and RFs are immunoreactive with α B-crystallin, which also is reported to stain cortical Lewy bodies, other astrocytomas, schwannomas, hemangioblastomas, and chordomas.
Observations of subtypes of S100 protein suggest that they distinguish pilocytic astrocytomas from WHO grade 2 to 4 astrocytomas. Pilocytic astrocytomas (PA) lack IDH and TP53 mutations and corresponding proteins. , PA tend to lack EGFR abnormalities. These features may discourage PA in some diagnostic differentials, or suggest further testing for BRAF gene abnormalities.
Cystic cerebellar pilocytic astrocytomas resemble hemangioblastomas, which may have focally GFAP-positive cells and a GFAP-positive cyst wall. Unlike hemangioblastoma, the mural nodule of a pilocytic astrocytoma contains highly fibrillar and abundantly GFAP-positive neoplastic astrocytes without clear vacuoles from lipids. CD31 and other endothelial cell markers show less abundant capillaries in the astrocytoma than in the hemangioblastoma.
Surgery is a primary treatment for pilocytic astrocytomas. Patients with cerebellar and other pilocytic astrocytomas that can be totally removed are likely to do well.
Pilocytic astrocytomas (grade 1) show tandem duplications at chromosome 7q34 forming a fusion gene between the kinase domain of BRAF and KIAA1549 ( BRAF-KIAA1549 fusion gene). BRAF is member of the Raf kinase family of proteins involved in the mitogen-activated protein (MAP) kinase pathway, whereas the function of K1AA1549 is poorly understood. Rare BRAF fusions such as BRAF - FAM131B, BRAF-RNF130, BRAF-CLCN6, BRAF-MKRN1, and BRAF-GNAI1 have also been reported, but at a much lower frequency. Pilocytic astrocytomas show variable frequency of BRAF fusion depending on their location in the CNS. For example, approximately 75% of tumors in the cerebellum demonstrate BRAF - KIAA1549 fusions. Supratentorial astrocytomas exhibit BRAF fusions in only 33% of cases, whereas approximately 50% of optic nerve pilocytic astrocytomas contain BRAF fusions. Although there is no specific immunostain for BRAF fusions, many practices have incorporated molecular testing based on PCR or fluorescent ISH (FISH)-based technologies.
Vemurafenib (PLX4032) is a potent inhibitor of mutated BRAF. It has marked antitumor effects against melanoma cell lines with the BRAF V600E mutation, but not against cells with wild-type BRAF. This or a next-generation drug may be effective in treating pilocytic astrocytomas or PXAs with this mutation. Pilocytic astrocytomas with BRAF fusions are more complex. These may need therapy targeted to specific fusions to generate a treatment response.
Fangusaro et al. found that the MEK1/2 inhibitor selumetinib was active in pediatric patients with recurrent, refractory, or progressive pilocytic astrocytoma with common BRAF aberrations and neurofibromatosis type 1 (NF1)-associated low-grade glioma. This new finding in 2019 needs large-scale confirmation, but it suggests MEK1/2 inhibition may help to manage cases not cured by surgery.
The fibrillary astrocytoma is more common than the protoplasmic astrocytoma. Fibrillary astrocytomas are a mixture of cellular processes (fibrils) and nuclei of greater angularity and density than normal or reactive astrocytes (see Fig. 20.16 ; see also Fig. 20.14B ). They contain more intracytoplasmic fibrils, and their cellular processes are longer than those in protoplasmic astrocytomas. Thus only the fibrillary astrocytoma stains well with phosphotungstic acid hematoxylin (PTAH), which stains fibrillar protein arrays, whereas both astrocytomas contain GFAP that can be stained immunohistochemically.
The term diffuse appropriately describes an astrocytoma whose margin gradually diminishes in cellularity. Within the extensive margin, neoplastic cells intermingle with brain parenchyma. NF staining highlights the axons of brain infiltrated by neoplastic astrocytes (see Fig. 20.16 ). Diffuse invasion of brain may also be evident as formations of secondary structures of Scherer, which are described in the “Gliosis Versus Glioma” section later in the chapter.
The diffuse nature of the growth and infiltration of low-grade astrocytomas demonstrates why they are so seldom cured despite their relatively benign histologic features. Postoperative survival is highly variable but usually is 3 to 10 years. The extreme variation in prognosis among grade 2 diffuse astrocytomas places a premium on better measures of outcome for individual patients. A low Ki-67 PI identifies patients with good prognosis ( Fig. 20.17 ; see Fig. 20.12 ).
Gemistocytes are cells swollen with hyaline pink cytoplasm that is reactive for GFAP ( Fig. 20.18A ; see Table 20.7 ). Their hyperchromatic and angulated nuclei are at the rim of the cells and produce a bizarre caricature of a reactive astrocyte. Astrocytomas with at least 20% gemistocytes may be considered gemistocytic astrocytomas, which are considered more aggressive than their nongemistocytic counterparts. Whereas gemistocytic astrocytomas are particularly likely to progress to a higher grade, those without high-grade features are considered grade 2 astrocytomas before progression. Gemistocytic astrocytomas are characterized by frequent TP53 and IDH mutations.
Gemistocytic astrocytomas are histologically distinguished from oligodendrogliomas with microgemistocytes by their more angulated and pleomorphic nuclei and their longer GFAP-positive cellular processes; their lack of synaptophysin-positive neoplastic neurons distinguishes them from gangliogliomas, and their smaller and more angulated nuclei and greater tendency to infiltrate brain distinguish them from subependymal giant cell tumors.
The anaplastic designation emphasizes the high grade of malignancy of the anaplastic astrocytoma. Variants of diffuse astrocytomas like gemistocytic astrocytomas may also be anaplastic. Features shared by high-grade gliomas are mitotic activity, as well as increases in cellular density, and nuclear pleomorphism (see Fig. 20.18 ) and hyperchromatism. Anaplastic astrocytomas retain GFAP-positive cellular processes and GFAP reactivity around their anaplastic nuclei (see Fig. 20.18A ); this important feature distinguishes them from reactive astrocytes trapped in other tumors. Their Ki-67 LIs are intermediate among gliomas.
The combined lack of foci of coagulation necrosis and lack of conspicuous vascular proliferation in an astrocytic glioma distinguish anaplastic astrocytomas from glioblastomas (see Table 20.7 ). Average survival of patients with anaplastic astrocytoma is slightly more than 2 years. In pediatric patients, a low Ki-67 L1 identifies a group of patients who have a better prognosis.
In adults, ∼70% of intermediate grade (including grade 2 diffuse astrocytomas and grade 3 anaplastic astrocytomas) are characterized by mutations in IDH , TP53 , and ATRX genes. The TP53 gene encoding the p53 protein involved in the cell cycle is often mutated in diffuse astrocytomas. Direct detection of mutations in p53 requires special procedures not generally used, but loss of normal p53 function usually results in increased expression of wild-type p53. This over-expression of p53 is so excessive that the p53 can be stained by IHC in many astrocytoma nuclei. Over-expression of p53 helps to identify many astrocytomas of grades 2 to 4. Over-expression of p53 is associated with tumor progression to glioblastoma. Strong P53 immunostaining in greater than 10% tumor nuclei has been suggested to be a surrogate TP53 mutation in IDH1/3-mutant gliomas, the detection of strong and diffuse p53 immunopositivity can be used as a surrogate for TP53 mutations and in support of the diagnosis of IDH-mutant astrocytoma. However, genetic alterations resulting in degradation of p53 or truncating mutations can manifest as complete loss of nuclear staining.
The histopathologic feature that best distinguishes diffuse astrocytomas (WHO CNS grade 2) from anaplastic astrocytoma (WHO grade 3) is the detection of mitotic activity. The threshold or numbers of mitotic figures required to establish this can vary with the size of the biopsy due to sampling error. A single mitotic figure may be sufficient in a very small biopsy, but not in a larger resection specimen. Similarly, distinguishing anaplastic astrocytoma (WHO grade 3) from glioblastoma (WHO grade 4) relies on the presence of microvascular proliferation and tumor necrosis that may not be detected in a small biopsy specimen due to sampling error. To help in better tumor prognostication and circumvent histologic sampling bias, molecular features that correlate with aggressive behavior have been recommended to help establish tumor grade in both IDH-mutant and IDH wild-type grade 2 and grade 3 astrocytomas.
Most anaplastic astrocytomas have an IDH mutation. They have been called anaplastic astrocytoma, IDH-mutant. Astrocytomas with an IDH mutation are now thought to be distinct entities from IDH wild-type astrocytomas. As a group, astrocytomas with an IDH mutation have better prognoses than their wild-type counterparts. And so, to further distinguish IDH mutant from wild-type astrocytomas, changing the names of diffusely infiltrating IDH mutants to astrocytoma IDH mutant grade 2, astrocytoma IDH mutant grade 3, and astrocytoma IDH mutant grade 4 has been proposed. The terms “anaplastic” and “glioblastoma” then only might be used for wild-type astrocytomas.
The product of ATRX is DNA helicase and chromatin remodeling protein ATRX. , Germline loss-of-function mutations in ATRX are associated with ATRX-linked syndrome. A primary function of ATRX protein is incorporation of histone H3.3 monomers into chromatin in collaboration with the histone chaperone protein DAXX (death-associated protein 6). , ATRX mutations are mutually exclusive with 1p/19q codeletion in glioma and are strongly associated TP53 mutation. ATRX mutations can be assessed by loss of staining in tumor cells using an ATRX protein specific antibody and can thus be used to differentiate tumors of astrocytic and oligodendroglial lineage. It is critical to evaluate ATRX staining in surrounding non-neoplastic tissue including neurons and endothelial, which serves as an internal positive control.
While ∼70% of intermediate grade (including grades 2 and 3) astrocytomas bear an IDH1 or 2 mutation, the clinical behavior of these tumors correlates better with molecular features than histologic grade alone. CDKN2A/B homozygous deletion is associated with worse outcome in IDH mutant gliomas independent of histologic grade. Current cIMPACT-NOW guidelines recommend that IDH-mutant astrocytomas with CDKN2A/B be assigned WHO grade 4, independent of their histologic grade.
Molecular alterations in IDH-wild-type astrocytomas (that do not demonstrate classic histologic features of glioblastoma-microvascular proliferation and tumor necrosis) including (1) the presence of TERT promoter mutations; (2) detection of EGFR gene amplification; and (3) presence of whole chromosome 7 gain accompanied with loss of whole chromosome 10 (designated +7/−10) copy number changes have been recommended for prognostication and assignment of grade. An IDH-wild-type astrocytoma with one or more of these molecular features is recommended to be assigned a WHO grade 4 designation.
Diffuse astrocytomas in adults and children have different molecular characteristics. In children alterations in MYB (V-Myb Avian Myeloblastosis Viral Oncogene Homolog), MYBL1 (V-Myb Avian Myeloblastosis Viral Oncogene Homolog-Like 1), BRAF, and FGFR1 (fibroblast growth factor receptor1 gene) are noted. , , Many of these pathways are thought to drive activation of the MAP kinase pathway.
Subependymal giant cell astrocytoma (GCA or SEGA) has a distinctive location, histology, and association with tuberous sclerosis (TS). The suppressor gene product associated with TS, tuberin, is predictably lost in GCA associated with TS. The tumor arises from the medial portion of the floor of the lateral ventricle in the region where the subependymal nodules of giant astrocytes in TS, known as candle guttering, are frequently found (see Table 20.7 ). Tumors are composed of giant astrocytes with large nuclei and prominent nucleoli ( Fig. 20.19 ). Although these tumors are pleomorphic, most nuclei lack sharp angulations, and the giant cells are not crowded. These cells may contain glial filaments variably positive for GFAP (see Fig. 20.19A ).
IHC has revealed partial neuronal differentiation in some GCAs (see Fig. 20.19B ), which complicates their classification as astrocytomas (vs. gangliogliomas). These giant astrocytes and their characteristically thick cytoplasmic processes have a tendency to form disoriented fascicles. It is very important to recognize this histologic entity, because their pleomorphism is at variance with their relatively benign behavior and WHO grade 1, and many GCAs are associated with TS.
Astroblastoma is rare. Astroblastic rosettes resemble perivascular pseudorosettes of ependymomas, except that the astroblastic processes remain thick the entire distance from cell body to adventitia of the vessel. This is a key feature of astroblastoma. Foot processes may even thicken near the adventitia. Vascular hyalinization is a helpful feature when present. IHC stain used in conjunction with routine histochemical stain helps to define this neoplasm. Although astroblastomas express focal GFAP, they do not stain with PTAH, a dichotomy that may be due to expression of a nonfibrillar form of the GFAP molecule, which is different from the fibrils of ependymoma and astrocytoma that stain for both.
Astroblastomas demonstrate alterations in the gene MN1 including structural rearrangements and fusions with partners such as BEND2 and CXXC5.
PXA is a supratentorial astrocytoma that often involves both the leptomeninges and cerebral cortex (see Table 20.7 ). It has a more distinct margin with brain than most astrocytomas (see Fig. 20.13B ). Its fibrillarity and pleomorphic, hyaline, lipid-laden, and multinucleated cells are clues to the diagnosis ( Fig. 20.20 ). Intracellular lipid content and protein granular degeneration vary from abundant to absent in individual tumors.
PXA may assume a clear cell appearance and thus may require identification by a panel of IHC reagents (see Fig. 20.1 ). Astrocytes are identified from their characteristic strongly GFAP-positive cells, often with coexpression of α-1–antitrypsin. Sparse lipid droplets are conspicuously negative for GFAP. These cells may be surrounded by reticulin fibers and basement membranes positive for type IV collagen, breaking a general rule that glioma cells lack reticulin. Neuronal elements occur in some tumors, suggesting that a PXA may be the glial portion of a ganglioglioma. The grade 2 PXA has been confused with the grade 4 glioblastoma; both are very pleomorphic. The pathologist must look for low Ki-67, virtual lack of mitoses, EGBs, little invasion (see the “Pilocytic Astrocytoma” section earlier in this chapter), and little or no over-expression of EGFR or p53 to distinguish PXA from diffuse astrocytomas and especially from glioblastomas ( Fig. 20.21 ). ,
Nonsynonymous point mutations in BRAF resulting in a valine to glutamic acid substitution at position 600 (V600E) are most frequent in PXAs (∼70%), gangliogliomas (∼20%), and at lower frequencies in pilocytic astrocytomas, diffuse astrocytomas, and pilomyxoid astrocytomas. , BRAF V600E mutant low-grade gliomas show a trend toward increased risk for progression and in gangliogliomas were associated with shorter recurrence-free survival. A BRAF V600E specific antibody can detect this mutation. , Astroblastomas demonstrate alterations in the gene MN1 including structural rearrangements and fusions with partners such as BEND2 and CXXC5.
The term theranostic has been used to define different things. One definition is diagnostic testing of tissue that goes beyond standard histology to select targeted therapy. Theranostic applications enable personalized (a.k.a. precision or stratified) medical treatments of specific patients.
In neuropathologic tumor classifications, theranostic and standard histologic data have lately been combined to reclassify gliomas (see “Diffuse Astrocytoma,” “Anaplastic Astrocytoma,” “Oligodendroglioma,” and “Glioblastoma”).
Among groups of patients with grade 2 astrocytomas, Ki-67 distinguishes tumors with good prognosis by their low PI (see Fig. 20.11A ). Ki-67 PI differentiates between grade 2 and grade 3 gliomas.
Thirty-three to 50% of diffuse astrocytomas have p53 protein abnormalities. TP53 gene mutations and/or p53 over-expression are early changes in astrocytomas. These p53 abnormalities are used to distinguish between astrocytomas and oligodendrogliomas.
Approximately 5% of gliomas are familial. A fraction of these are associated with Li-Fraumeni syndrome (LFS), the result of a germline mutation of the TP53 gene on the short arm of chromosome 17 (17p13). Data suggest that approximately 1% of gliomas occur in patients with LFS. More than 50% of these are astrocytic gliomas, including astrocytoma grades 2 to 4 (diffuse low-grade astrocytoma through glioblastoma). Clinical situations that should arouse suspicion of LFS in a glioma patient include any second neoplasm, a family history of sarcoma or choroid plexus (CP) tumor, or a young (<45 years) close blood relative of the patient with a cancer, lymphoma, or brain tumor.
GFAP with a good hematoxylin counterstain is a very important IHC stain to show neoplastic nuclei surrounded by GFAP and the fibrillar GFAP-positive processes of astrocytomas. Ki-67/GFAP dual stain helps to distinguish reactive from neoplastic astrocytes.
Neurofilament stain both tests for an uncommon ganglionic tumor component (see the “Neuronal Tumors” section later in this chapter) and often shows axons of brain tissue that existed before the tumor. The latter helps assess infiltration by neoplastic cells.
Noninfiltrating astrocytomas that lack both p53 and EGFR over-expression tend to be grade 1 tumors.
Ki-67 is important in assessing proliferation, particularly among astrocytomas with inconspicuous mitoses.
IDH1 (R132H) and ATRX mutations assessed using IHC can together strengthen the diagnosis of astrocytic tumors and help in differentiating these tumors from oligodendrogliomas. Additional molecular diagnostic assays are available for IDH mutations less common than IDH1 (R132H).
IDH mutant astrocytomas with CDKN2A/2B homozygous deletion are assigned WHO grade 4, independent of histologic grading. Current cIMPACT-NOW guidelines recommend that CDKN2A/B homozygous deletion cause outcomes in IDH mutant gliomas bad enough to warrant a WHO grade 4 regardless of their histologic grade.
IDH wild-type astrocytomas with one or more of the following molecular alterations, including presence of (1) TERT promoter mutations; (2) EGFR gene amplification; and (3) whole chromosome 7+/10− is recommended to be assigned a WHO grade 4.
Posterior fossa ependymomas are frequently found in children, whereas supratentorial, and spinal cord ependymomas are more common in adults. ,
The cellular conformations of ependymomas vary between fibrillar and epithelial, posing special problems of differentiation from not only other gliomas but also carcinomas and meningiomas (see Tables 20.2 and 20.7 ). These differentiations are facilitated by the understanding that even epithelioid and clear cell ependymomas stain with anti-GFAP.
Epithelial membrane antigen (EMA) is not as helpful in confirming ependymoma as was hoped, partly because it usually shows nothing or tiny cytoplastic dots with no further structure. EMA rarely stains recognizable ependymal structures in tumors difficult to interpret. Instead, it stains a fraction of well-differentiated ependymomas easily recognized on H&E. Subtle and ambiguous cases often show few or no cytoplasmic “dot-like” structures, imitated by other tumors including gliomas. The more actual structure the EMA reveals, the more useful it will be, like revealing cytoplasmic structures rather than dots, or groups of positive epithelioid cells. Rarely, EMA positive material shows more structure, sometimes “donut-like” structures.
Ependymomas often contain perivascular rosettes, less often show luminal rosettes, but may only show a few cells with fibrillar processes ( Fig. 20.22 ; see Figs. 20.1 and 20.2 ). The anti-GFAP stain highlights these fibrillar processes and facilitates their identification (see Fig. 20.22A and B ). A good place to look for these fibrillar processes is around vessels. Ependymomas commonly show round or oval nuclei with scattered dark and light chromatin called “salt and pepper” nuclei.
Ependymomas usually lack the marker of astrocytomas and oligodendrogliomas, OLIG2, but not always. In one study, over 90% of astrocytomas and oligodendrogliomas were positive for OLIG2, and only 11% of ependymomas were positive for OLIG2.
In contrast to nonglial neoplasms, aggregated ependymoma cells in tumor parenchyma lack a basement membrane. Immunostaining shows no collagen or fibronectin in these aggregates.
CP papilloma cells grow on a collagenous fibrovascular core apparent in aggregates of papillae. Although rare ependymomas have sparse CAM 5.2 CK immunoreactivity, even these ependymomas have less CAM5.2 than CP papillomas and carcinomas. Ependymomas lack transthyretin, whereas CP papillomas have little or no GFAP. GFAP, CAM 5.2, trichrome, and transthyretin are recommended to distinguish ependymomas from CP papillomas.
Electron microscopy (EM) remains the ultimate standard of ependymoma confirmation. EM exceeds chemical and IHC stains in features that confirm difficult ependymomas (see Fig. 20.22C ). EM shows cilia, basal bodies, cytoplasmic inclusions of microvilli, and elongated intercellular junctions.
A tumor with a specific genetic abnormality that superficially resembles ependymoma (i.e., possible rosettes, GFAP positive epithelial cells, “dot-like” perinuclear EMA, some round or oval nuclei with scattered dark and light chromatin) should be examined with EM to confirm its diagnosis. If ultrastructural features of an “odd” ependymoma were not sought, ependymoma, NOS might be considered. Or a descriptive diagnosis and a comment about the prognostic value of the molecular abnormality might be appropriate. On the other hand, if quality EM were done, and no ultrastructural features of ependymoma were found, other diagnoses could be considered, like embryonal tumor or glioma, NEC.
The major problem we face with EM today is properly saving wet tissue for possible EM if needed. “Rescuing” tissue from paraffin generally produces unacceptable results. Surgeons will usually wait longer for EM on a glioma that may be an ependymoma if the situation is explained.
The general features of ependymoma just described are useful in identifying the types of ependymoma discussed in this section.
Classic ependymomas are solid tumors with prominent rosettes. The pathologist should look for rosettes to confirm suspicion of any suspected ependymoma (see Tables 20.2 and 20.7 ). Perivascular rosettes (aka, perivascular “pseudorosettes”) are most useful, because they occur in most well-sampled ependymomas. Perivascular rosettes have a fibrillar zone without nuclei that is at least three erythrocyte diameters wide around central vessels. Anti-GFAP stains the fibrillar zone, making subtle perivascular rosettes easier to find (see Fig. 20.22B ). The processes taper to become very thin as they radiate from the cells to the vascular adventitia, distinguishing them from the thick processes of astroblastic formations.
True ependymal rosettes (luminal rosettes) are characteristic of ependymoma, but many samples of ependymoma lack them. The ependymal rosette consists of ependymal cells spaced around a lumen ( Fig. 20.23 ). Some tumors show expanded ependymal rosettes, and others have long ependymal linings that do not close into rosettes. Some embryonal tumors have luminal rosettes that are not quite the same as ependymal rosettes (see “Embryonal Tumors” below).
Many ependymomas have a relatively discrete margin with brain compared with other gliomas. This margin is revealed best in white matter with the NF IHC stain, which shows an abrupt border between the NF-positive axons abundant in white matter and the NF-negative ependymoma. Synaptophysin shows a corresponding distinct border between positively reacting neuropil and negatively reacting tumor. This distinct margin is likely one reason that resection of ependymomas is a major factor associated with better survival.
The low-grade designation is often dropped from the name for this group of tumors, which are referred to simply as ependymoma. The features just described of low-grade ependymomas distinguish them from other tumors. Round and oval nuclei with finely dispersed chromatin distinguish ependymomas from most primary brain tumors other than meningiomas (see Figs. 20.22A and 20.23 ). In the parenchyma away from rosettes, ependymoma nuclei tend to be more uniformly crowded (see Fig. 20.22 ) than nuclei in low-grade astrocytomas (see Figs. 20.11 and 20.14 ) and less crowded than in medulloblastomas ( Fig. 20.24 ) and primitive neuroectodermal tumors (PNETs).
Epithelioid ependymomas occasionally have remarkably distinct margins with brain that imitate margins of nonglial neoplasms (see Fig. 20.2 ). Anti-GFAP stain for glial filaments is extremely helpful in differentiating these ependymomas from carcinomas, pituitary adenomas, craniopharyngiomas, and meningiomas (see Fig. 20.22B and Table 20.2 ). The stain accentuates fibrillar cellular processes, which distinguish the ependymoma.
Papillary ependymomas closely resemble CP papillomas. Solid regions of ependymoma parenchyma, where GFAP-positive neoplastic cells grow on one another, rather than on fibrovascular stroma, can be appreciated from their lack of collagen and fibronectin with IHC staining.
Clear cell ependymoma (see Fig. 20.22A ) resembles oligodendroglioma and central neurocytoma (CN; Fig. 20.25A and 20.27A ). It is an epithelioid ependymoma that also has clear perinuclear halos. IHC staining for GFAP may reveal ependymal features, such as perivascular fibrils (see Fig. 20.22B ). The clear cell appearance of these ependymomas requires a panel of IHC reagents or, often, EM, to differentiate them from other clear cell tumors (see Fig. 20.1 and 20.22C ).
Tanycytic ependymomas are found within the brain and particularly in the spinal cord. Their round to oval nuclei with distinctly light and dark regions of chromatin resemble those in ependymoma, and abundant GFAP-positive cellular processes resemble those in astrocytomas. They form structures replete with nuclei next to zones of fibrillar cellular processes. These structures are distinguished from Verocay bodies by their diffuse, extensive GFAP positivity and their lack of type IV collagen. They are not limited to surrounding GFAP-negative vessels, such as the perivascular rosettes of other ependymomas. The margins of tanycytic ependymomas with surrounding parenchyma tend to be discrete, to exclude NF-positive axons of spinal cord tracts, and to be potentially resectable. Diffuse astrocytoma cells infiltrate between NF-positive axons. ,
The subependymoma protrudes from the wall of a ventricle into the ventricular space. Its histologic and IHC features closely resemble those of tanycytic ependymoma (see Table 20.7 ). Subependymoma is usually benign.
The myxopapillary ependymoma (MXPE) appears the least glial in H&E-stained slides. It is nearly always found in the region of the filum terminale, cauda equina, sacrum, and adjacent extravertebral soft tissues (see Fig. 20.23 and Table 20.2 ). This ependymoma differs from others in its amount of mucin production. Its hallmark is parenchymal and perivascular mucin produced by ependymal cells (see Fig. 20.23 ). MXPE is often papillary but may be solid.
Although the differential features described in the general discussion of ependymomas can be useful, the peculiar morphology and growth of MXPEs pose unique problems. Individual tumors vary dramatically between epithelial and fibrillar cells. The most difficult variants of MXPE to recognize are those that are nearly all myxoid or all epithelial. The highly myxoid variety may produce cords of cells in a mucoid matrix that resembles chordoma, a neoplasm found in the same location. The presence of GFAP is the key IHC feature distinguishing MXPE from chordoma.
Fibrillary MXPE may be confused with fibrous meningioma and schwannoma. The epithelial and papillary variants may resemble carcinoma or meningioma, although a positive GFAP stain response differentiates MXPE from GFAP-negative carcinoma and meningioma. MXPE lacks type IV collagen and fibronectin-positive basement membranes around each cell, a feature of schwannoma (see Fig. 20.55A ).
In contrast to metastatic carcinoma, MXPEs lack malignant cytology, have a lower Ki-67 LI, and are focally fibrillar. Paragangliomas may mimic MXPEs, but MXPEs lack chromogranin A and express GFAP.
Anaplastic ependymomas are those with malignant features, including conspicuous brisk mitotic activity, nuclear and cellular pleomorphism, multinucleated and giant cells, high cellular density, conspicuous necrosis, and many vascular proliferations (see Tables 20.2 and 20.7 ).
Histologic grade is less predictive of survival in ependymoma than in astrocytoma. Classic ependymoma is a grade 2, and anaplastic ependymoma is a grade 3 by WHO criteria. No association between this difference in ependymoma grade and patient survival has been definitively established.
Poor prediction by standard features of malignancy may be improved by IHC: In posterior fossa ependymomas, EZHIP and H3K27me3 status may predict different patient outcomes and treatments. The combination of increased vimentin expression and decreased GFAP expression may predict poor survival in infratentorial ependymomas. Anaplastic ependymomas are more likely to over-express p53 or EGFR protein than are low-grade ependymomas.
Radiographic evidence of residual disease after surgery predicts markedly reduced survival, putting a premium on correct intraoperative interpretation and total removal of the tumor. In this regard, the notion that all myxopapillary ependymomas should be elevated to grade 2 might consider grade 1 for subgroups of MXPEs presenting without spread into other tissues, that were totally resected upon first surgery.
Ependymomas develop throughout the neuraxis, but molecular genetics seems to depend on the region of origin: posterior fossa, supratentorial, and spinal cord. Supratentorial ependymomas have been grouped into fusion positive and negative subgroups. Supratentorial ependymomas with poor prognosis bear C11orf95 (also called ZFTA)-RELA (part of the NF-kappa B signaling pathway detected by PCR or FISH-based assays) fusions. , , C11orf95-RELA fusion ependymomas show nuclear p65-RELA and cytoplasmic L1CAM expression, which have been suggested as molecular surrogates for these tumors. However, the sensitivity and specificity for these stains is variable. Another subset of supratentorial ependymomas bear fusions of the YAP1 gene with MAMLD1 or other genes partners. ,
Posterior fossa (PF) ependymomas have been subgrouped into PF-group A and PF-group B tumors based on gene expression profiles and DNA CpG island methylation (detected by DNA methylation arrays) phenotype. , , PF-group A ependymomas mainly occur in children, have a distinct gene expression profile, exhibit CpG island hypermethylation, are more invasive, and bear a worse prognosis. In contrast, PF-group B ependymomas occur in older individuals and younger adults, do not show CpG island methylation, are less invasive, and carry a better prognosis. , , PFA ependymomas exhibit global reduction of the histone repressive mark H3K27me3, which can be assessed by immunohistochemistry. Retained H3K27me3 immunoreactivity in endothelial and immune cells within the tumor microenvironment cells can be used as an internal control. PFA ependymomas show global reduction in H3K27me3 in tumor cells, but variability in the proportion of immunonegative cells can be encountered. PFB ependymomas do not demonstrate H3K27me3 reduction and 80% immunopositive cells has been proposed as a cutoff value above which an ependymoma is more likely to fall into the PFB molecular group. , , H3K27me3 reduction is mediated by overexpression of a protein termed Enhancer of Zeste Homologs inhibitory protein (EZHIP), that can be detected by immunohistochemistry. , EZHIP is also called CXORF67.
A subset of spinal cord ependymomas with classic ependymal histology demonstrate MYCN amplification and are associated with a poor prognosis. MYCN protein expression can be detected by immunohistochemistry. ,
Because a major prognostic factor of ependymomas is extent of resection, the pathologist must warn the surgeon during intraoperative consultation when histopathology suggests ependymoma.
GFAP and EMA are the best immunohistochemical markers for ependymoma at this time. GFAP often highlights fibrils in perivascular rosettes. EMA does not always distinguish ependymoma from other tumors.
When ependymoma is in the differential, small pieces of the tumor should be saved in fixative for possible electron microscopy (EM). The specimen is best taken as fresh unfrozen tissue at biopsy and immediately put in glutaraldehyde. If this was not done, remaining unfrozen fixed and wet tissue can be post-fixed in glutaraldehyde and processed for possible EM.
Molecular features including the presence of C11orf95/ZFTA-RELA and YAP1-MAMLD1 fusions have been demonstrated in supratentorial ependymomas.
Posterior fossa ependymomas are classified into PFA (globally reduced for nuclear H3K27me3 immunostaining in tumor cells) and PFB (preserved nuclear H3K27me3 staining in tumor cells) tumors.
Spinal cord ependymomas with MYCN amplification bear a poor prognosis.
Oligodendroglioma are defined by the presence of IDH mutations and 1p/19q-codeletions according to current guidelines. Oligodendroglioma differ from other gliomas, except for a few ependymomas, in having an epithelioid, rather than a fibrillar, appearance (see Table 20.2 ). This appearance is most evident within the central portion of the neoplasm, which is most crowded with neoplastic cells. Perinuclear halos are an important feature of formalin-fixed paraffin sections of oligodendroglioma (see Fig. 20.25 ). Well-differentiated oligodendrogliomas have remarkably round and regular nuclei centrally placed within cells, which resemble fried eggs. Their vessels are usually numerous, fine, CD31-positive capillaries that sometimes segregate the parenchyma into small lobules.
Microgemistocytes have been considered oligodendroglial and are distinguished from gemistocytic astrocytoma cells by their round nuclei and short processes. Microgemistocytes have a ball of cytoplasmic GFAP immunoreactivity near their nucleus and have shorter cellular processes than gemistocytic astrocytes and astrocytomas (see Fig. 20.25B ).
The epithelioid appearance of pure oligodendrogliomas (see Fig. 20.25A ) imitates that of true epithelial neoplasms (see Table 20.2 ). Suprasellar oligodendrogliomas may be mistaken for pituitary adenomas (see Fig. 20.2 ). Oligodendrogliomas may be confused with clear cell and other ependymomas and meningiomas. OLIG2 IHC is characteristic of astrocytoma and oligodendroglioma, and usually absent in ependymomas and meningiomas. The occasional presence of OLIG2 in ependymomas, and difficulty using negative results to validate an interpretation, suggest caution using negative OLIG2 as a marker of ependymoma.
Grade 3 oligodendrogliomas simulate metastatic carcinoma, particularly RCC. GFAP-positive tumor cells (see Fig. 20.25B ) distinguish the oligodendroglioma from these other tumors, but GFAP-positive neoplastic cells are inconspicuous in some oligodendrogliomas. In this case, try OLIG2 IHC. And inspect the tumor margin with brain (see Fig. 20.1 ). Oligodendrogliomas are diffuse gliomas. They have diffuse margins (see Fig. 20.13A ) that not only invade CNS, but also allow reactive gliosis to enter the tumor. And so, NF-positive axons or GFAP-positive reactive astrocytes can be seen within an oligodendroglioma (see Fig. 20.25A ), a pattern usually not seen in meningioma, adenoma, or carcinoma.
Even macroscopically discrete oligodendrogliomas show a more diffuse margin with brain than adenomas, carcinomas, and meningiomas. Either NF or synaptophysin used as a brain tissue marker or GFAP-positive gliosis delineates a sharp margin with brain in these other tumors, even carcinomas that engulf chunks of brain (see Figs. 20.28 and 20.29B ). Among these tumors, secondary structures are typical only with the glioma (see the sections on astrocytomas and gliomas). Precise localization of the biopsy specimens is helpful, because oligodendrogliomas do not originate from the adenohypophysis or dura and rarely invade them. A panel of immunostains for chromogranin and pituitary hormones can identify adenomas, but no specific marker for oligodendroglia withstands paraffin embedding. The discovery of such a marker would be a major contribution to neuropathology.
The broad specificity among gliomas of Leu-7 and S100 protein limits their use for IHC analysis of oligodendroglioma. However, an oligodendroglioma can be differentiated from a meningioma by its positivity with the OLIG2 and Leu-7 antibodies, because meningiomas are typically negative for both markers. Most meningiomas have distinctive ultrastructure on EM not seen in oligodendroglioma.
The expression of synaptophysin by as many as 18% of oligodendrogliomas should be noted so that they are not confused with glioneuronal tumors, central neurocytomas, and dysembryoplastic neuroepithelial tumors (DNTs). Synaptophysin-positive cells are scattered sparsely when present. Both central neurocytomas and DNTs have distinct margins with brain, in contrast to the diffuse margin of oligodendroglioma.
Many tumors can exhibit mixed histologic features including classic oligodendroglial morphology accompanied with areas that resemble astrocytic tumors. These gliomas were previously designated oligoastrocytomas. Seminal molecular studies now suggest that astrocytomas and oligodendrogliomas show distinct molecular features, and the term oligoastrocytoma is generally discouraged as a diagnosis. Oligodendrogliomas are now defined based on the presence of IDH mutations and codeletions on the short arm of chromosome 1 (1p) and on the long arm of chromosome 19 (19q). , , The correlation between 1p/19q codeletions and better survival has elevated the combined 1p/19q chromosomal marker to be the major diagnostic criterion for oligodendrogliomas along with IDH mutations. Chromosomal deletions in 1p and 19q can be detected either by single nucleotide polymorphism (SNP) chromosome microarray or FISH ( Fig. 20.26D ). SNP is recommended for its better detection of loss of the 1p and 19q whole arms.
The utility of chemotherapy when codeletions are present in 1p and 19q is thought to be higher than when deletions are not present. Studies of the potential utility of chemotherapy in anaplastic oligoastrocytomas and its relation to 1p/19q status are available. , Mutations in CIC (homolog of the Drosophila gene capicua) on chromosome 19q and FUBP1 (FUSE binding protein 1) on chromosome 1p have been recently described in oligodendrogliomas. , Consistent with a loss-of-function phenotype, CIC mutations are distributed throughout the coding region of the gene (with a predilection for exon 5) and include nonsense, insertions/deletions, missense, and frame shift variants. Also consistent with loss-of-function, FUBP1 mutations are mainly frameshift and nonsense variants, and occur at lower frequencies (14% to 22%) than CIC mutations in low-grade oligodendrogliomas (2/14, 2/9, 3/21, 3/17 ).
Use of 1p/19q codeletions to define oligodendrogliomas has nearly eliminated the classification of oligoastrocytoma. However, there are rare oligoastrocytomas that have some regions of oligodendroglioma with 1p/19q codeletions (usually associated with retained ATRX staining and lack of p53 overexpression), and other regions of astrocytoma without 1p/19q codeletions (usually associated with loss of ATRX staining and presence of p53 over-expression).
Molecular markers to test for IDH mutation and complete whole-arm codeletions of 1p and 19q signal better prognosis and chemotherapy sensitivity. They are now the recommended diagnostic criterion of oligodendrogliomas of both grades 2 and 3.
Criteria for anaplastic oligodendroglioma are different than those for anaplastic astrocytoma. An occasional mitosis and limited vascular proliferation may be found in a grade 2 oligodendroglioma.
A few otherwise typical and diffusely invasive oligodendrogliomas may show some cells with positive neuronal markers, particularly synaptophysin. These tumors are still considered oligodendrogliomas.
Years of frustration using structural features alone to decide whether an infiltrating glioma (IG) is an astrocytoma, oligoastrocytoma, or oligodendroglioma have resulted in the declaration that the combination of both IDH mutation and whole-arm chromosomal codeletions of 1p and 19q (“1p/19q codeleted”) define an IG to be an oligodendroglioma. This has worked well for the most important people: our patients. Studies show that patients with 1p/19q codeleted gliomas have improved survival and response to treatment than patients who lack this codeletion. And so, clinical needs have expanded the pathologist’s role in the interpretation of gliomas.
Procarbazine and chloroethylcyclohexylnitrosourea (CCNU) and vincristine (PCV) chemotherapy plus radiotherapy improved survival of patients with 1p/19q codeleted gliomas. , Temozolomide is often substituted for PCV chemotherapy on patients with these tumors for its preferred administration. Molecular characterizations of oligodendrogliomas are now being used both as guides to therapy and as diagnostic criteria. Whole-arm chromosomal codeletions of 1p and 19q are used by neuropathologists for diagnosis of oligodendrogliomas. Chromosomal deletions in 1p and 19q can be detected by SNP array or FISH (see Fig. 20.26D ).
Virtually all 1p/19q codeleted gliomas have an IDH mutation. , , This association serves as an important check on the structural diagnosis of oligodendroglioma. If a glioma resembles an oligodendroglioma but the common IDH R132H mutation is not seen by IHC, then assay, such as PCR or NGS, for all IDH1 and IDH2 mutations should be done. If this is also negative for IDH mutation, but FISH appears to show 1p/19q codeletion, then confirmation of FISH with whole-arm analysis of 1p and 19q is needed. FISH probes do not identify every nucleotide sequence in the aforementioned short and long arms, and may miss small sequences of 1p or 19q. SNP array provides whole-arm analysis of chromosomes. Partial loss of 1p and/or 19q in a glioma is not associated with the improved survival of a glioma.
If 1p/19q and IDH analysis of a glioma that structurally resembles an oligodendroglioma shows lack of both 1p/19q codeletion and IDH mutation, then a “pediatric oligodendroglioma” should be considered. Histologic progression is reported to be rare in these tumors.
Clinical needs have expanded the pathologist’s role in the interpretation of gliomas. For example, the effectiveness of procarbazine and chloroethylcyclohexylnitrosourea (CCNU) and vincristine (PCV) chemotherapy plus radiotherapy on improving survival of patients who have oligodendroglial tumors with 1p and 19q whole-arm chromosomal codeletions (del 1p/19q) , has increased the value of recognizing these tumors. Temozolomide is also effective on patients with these tumors and simpler to take. The strong relationship between del 1p/19q and IDH mutations with improved survival in oligodendrogliomas and anaplastic oligodendrogliomas requires recognition of 1p/19q codeleted and IDH mutant tumors. , ,
Molecular characterizations of tumors with oligodendroglioma components are now being used both as guides to therapy and as diagnostic criteria. The molecular definition of the common adult oligodendroglioma includes presence of both an IDH mutation and whole-arm chromosomal codeletions of 1p and 19q. , , If a glioma resembles an oligodendroglioma but the common IDH R132H mutation is not seen by IHC, then assay, such as PCR or NGS, for all IDH1 and IDH2 mutations should be done. If this is also negative for IDH mutation, but FISH appears to show 1p/19q codeletion, then confirmation of FISH with whole-arm analysis of 1p and 19q is needed. FISH probes do not identify every nucleotide sequence in the aforementioned short and long arms, and may miss small pieces of 1p or 19q. SNP array provides whole-arm analysis of chromosomes and shows small pieces of 1p and 19q. Such pieces are important because partial loss of 1p and/or 19q in a glioma is not associated with improved survival.
The histologic features of anaplastic transformation in oligodendroglioma are similar to those in other gliomas (see Figs. 20.22 and 20.26 ). These include high cellularity, cytologic atypia and pleomorphism, high mitotic activity (6 or higher mitoses per 10 HPF), presence of microvascular proliferation, and tumor necrosis. However, limited amounts of vascular proliferation are frequently present in oligodendroglioma, and in isolation, they cannot be considered evidence of malignant transformation (see Table 20.2 ). Vascular proliferation is highlighted by vimentin and CD31 stains. Anaplastic and low-grade oligodendrogliomas may express neuronal markers and vimentin. Synaptophysin-positive cells are scattered sparsely when present.
The molecular definition of the common adult oligodendroglioma includes presence of both an IDH mutation and whole-arm chromosomal codeletions of 1p and 19q. , , The presence of IDH1 mutations carries a strong favorable prognostic significance for overall survival in anaplastic oligodendroglial tumors and in other gliomas.
MIB-1 (Ki-67) is useful in predicting good outcome among patients with oligodendrogliomas. A Ki-67 PI less than or equal to 5% has been found to correlate with better survival. As always, a standard method of sampling and counting the PI must be employed. Sampling a region with a conspicuously high density of positive nuclei, and using a hundred total nuclei surrounding this “hot spot” for the denominator is recommended. A glioma’s proliferative potential should be assessed by its highest proliferation, just like a glioma is graded by its worst histologic features, even if focal. Vimentin expression and molecular alterations in CDKN2A, PTEN, and EGFR genes correlate with poor prognosis. ,
Partly due to the difficulty of defining an oligoastrocytoma by histologic and conventional phenotypic markers and to the prognostic value of 1p/19q codeletions in predicting good patient outcome and response to therapy, presence or absence 1p/19q codeletions has become the critical determinant, the gold standard, of whether a glioma is an oligodendroglioma or astrocytoma.
On the other hand, there do seem to be rare oligoastrocytomas that have some regions of oligodendroglioma with 1p/19q codeletions (usually associated with retained ATRX staining and lack of p53 over-expression), and other regions of astrocytoma without 1p/19q codeletions (usually associated with loss of ATRX staining and presence of p53 over-expression).
In large part as a result of revelations about its frequent expression of GFAP-positive fibrillar cellular processes, rather than being classified as an embryonal neuroglial malignancy, glioblastoma is now considered the most malignant of astrocytomas. “Multiforme” once referred to its heterogeneity of tissue features including glioma, necrosis, vascularity, and hemorrhage. In tweet, a glioblastoma multiforme is a GbM. A GbM often contains focal astrocytoma, less often oligodendroglioma, and, rarely, ependymoma.
The most common glioblastoma in adults, and particularly older adults, is negative for IDH mutation. The lack of this mutation is normal, and so it is recognized as an IDH-wild-type glioblastoma. These IDH-wild-type GbM are also known as “primary” glioblastomas because they do not progress through lower grade 2 and 3 gliomas, but appear to arise de novo .
The diagnostic criteria for glioblastoma were relaxed in the 1990s. Formerly, the cytologic criteria of anaplastic astrocytoma—mitotic activity, hypercellularity, pleomorphism, and nuclear hyperchromasia—plus both microvascular proliferation (MVP) and spontaneous necrosis were required (see Fig. 20.12B ). Now, either MVP or spontaneous necrosis is a sufficient addition to anaplasia for the diagnosis of GbM (see the “Anaplastic Astrocytoma” section for a comparison).
Spontaneous necrosis is presumed when the necrosis is surrounded by tumor and the patient has not had previous treatment, particularly not radiation therapy. Spontaneous necrosis is coagulation necrosis in tumor of at least moderate cellular density that shows eosinophilic remnants of tumor cells and vessels. At low magnification on H&E stain, it shows patterns of pink necrosis in blue live tumor that have been called “geographic” because they often resemble natural regions on a map. Spontaneous necrosis may have a band of increased cellular density around its periphery called “pseudopalisades,” more likely to be appreciated in large specimens. These features are seen in high-grade (at least grade 3) gliomas but are especially conspicuous in GbM. Spontaneous tumor necrosis usually lacks both fibrinoid change and calcifications seen in radiation necrosis (see “Pitfalls: Treatment Effect vs. Recurrent Tumor”).
MVPs are presumed when increased cells in the walls of tumor vessels with small lumens are found. If necrosis is absent, MVP identification should be unequivocal to confirm glioblastoma. MVP can be highlighted with collagen type IV, or fibronectin stain. Individual cells within MVP can be stained with CD34 or vascular endothelial growth factor (VEGF), SMA, and vimentin. For unequivocal confirmation of MVP, we find the dual stain for Ki67 and GFAP (Ki67/GFAP) helpful. Because vessels are GFAP negative, entire vascular walls from lumen to adventitia are easy to identify, and Ki67 reveals whether their vascular wall cells are proliferating.
Other malignant features of glioblastomas are bizarre nuclei, multinucleated cells, mitoses, and extreme pleomorphism (see Figs. 20.12B and 20.29A ). Unfortunately, the heterogeneity of glioblastomas for these histologic features compromises diagnoses obtained on small specimens, such as those obtained with stereotactic needle biopsy, and this jeopardizes accurate grading (see Fig. 20.12 ).
Giant cell glioblastomas were called monstrocellular sarcomas before recognition of their GFAP reactivity ( Figs. 20.27 to 20.29 ). Giant cell glioblastomas are negative for IDH mutation. They are a variant of IDH-wild-type glioblastoma, grade 4.
Confusion arises in distinguishing glioblastoma from malignant meningioma, sarcoma, and carcinoma. Unlike this terrible triad, GbM contain fibrillar neoplastic cells that express GFAP in their cellular processes ( Tables 20.8 and 20.9 ; see Figs. 20.4 , 20.12 , 20.27–20.29 and Table 20.2 ). The nuclear and cytologic features of the GFAP-positive cells should be checked for anaplasia to confirm glioblastoma, because nonglial malignancies can trap islands of CNS parenchyma and stimulate gliosis (see Figs. 20.27 and 20.28 ). Sarcoma is easily confused with glioblastoma on H&E staining, but GFAP reveals the glioblastoma (see Fig. 20.29 ). The dual Ki67/GFAP stain is very helpful in distinguishing whether or not the malignant component is glioblastoma. Although the rapid growth of a glioblastoma may produce a pseudocapsule, neoplastic glia are evident beyond this margin infiltrating within brain tissue (see Fig. 20.29 ).
Epithelioid glioblastoma is distinguished from other glioblastomas by a predominant collection of crowded epithelioid cells more than seen in focal epithelioid metaplasia, and some rhabdoid cells. Epithelial metaplasia is more focal and differentiated (glands or squamous metaplasia) than epithelioid glioblastoma cells. Epithelioid glioblastomas are SMARCB1 deficient. Epithelioid glioblastomas lack IDH and often (about 50%) express a BFAF V600E mutation. Patients with epithelioid glioblastomas have a poor prognosis (about 6 months, rarely over 2 years).
A gliosarcoma is composed of cells that resemble glioblastoma and other cells that resemble sarcoma ( Fig. 20.30 ). Regions of collagen-positive and GFAP-negative cells bridge the glioblastoma in a marbled configuration (see Table 20.7 ). Differences in collagen messenger RNA (mRNA) and total nuclear DNA content indicate the extent of variation between these regions. , Gliosarcomas can metastasize. Tumor progression from gliosarcoma to pure sarcoma lacking GFAP-positive cells can occur. The glial and mesenchymal elements have similar genetic alterations. ,
Gliosarcomas are negative for IDH mutation. They are a variant of IDH-wild-type glioblastoma, grade 4.
Rarely, IDH-wild-type gliosarcomas and glioblastomas produce adenoid formations or epithelial foci with squamous differentiation and keratin pearls. These regions stain immunohistochemically for CK and EMA (see Table 20.8 and Fig. 20.30D ). To avoid confusion of this tumor with carcinoma, it is necessary to obtain adequate sampling and to be aware that these regions are focal and that other regions will show the familiar fibrillar, GFAP-positive neoplastic cells of a glioblastoma. It is important to remember that carcinoma cells are GFAP negative.
Glioblastomas manifest other peculiar features. Rare glioblastomas occur with granular cell tumors. Some glioblastomas contain diffuse cytoplasmic lipids.
Two structurally similar varieties of glioblastoma are called primary and secondary GbM. Primary glioblastomas are those that arise de novo in older patients, who have higher MIB-1 proliferation indices and shorter survival than their younger counterparts. Primary glioblastomas are often associated with CDKN2A deletions, PTEN alterations, and EGFR gene amplification plus EGFR over-expression.
Loss of genetic material in chromosome 10 is the most frequent genetic abnormality in glioblastomas and occurs in approximately 66%. Loss of genetic material in the short arm of chromosome 10 (10p) is nearly always associated with primary glioblastomas. Loss in the long arm (10q) is found in both primary and secondary GbM. These losses can be observed with comparative genomic hybridization, FISH, ISH enhanced with immunostaining, or PCR assay. Amplifications of chromosome 7 DNA are also demonstrated by modifications of these techniques.
Primary glioblastomas are the most common GbM. In adults they have been extensively studied and constitute the first tumor type to be characterized by The Cancer Genomic Atlas. Four main transcriptional subclasses are defined and include the proneural, mesenchymal, neural, and classic subtypes (the fourth neuronal group initially described is now less well defined). , Next-generation sequencing of GbMs have revealed a spectrum of molecular abnormalities in these tumors. The proneural group is characterized by alterations in PDGFRA, IDH1/2, MET, CDK6, and CDK4. Classical subtype is enriched for EGFR amplification, PTEN deletions, and CDKN2A loss. EGFR amplification can be confirmed by FISH. EGFR over-expression can be defined immunohistochemically ( Fig. 20.31 ). The mesenchymal subtype shows alterations in NF1, TP53, and CDKN2A. Mechanistically these complex genetic lesions fall into three core groups irrespective of the subgroup classification: (1) More than 90% of GbM exhibit aberrant receptor tyrosine kinase/RAS/phosphatidylinositol-3-kinase (RTK/RAS/PI3K) signaling; (2) 86% of GbM show altered p53 pathway activation; and (3) RB signaling is deregulated in 79% of GbM.
Gene products detected by IHC including p53 and EGFR are usually proteins. IHC stains for p53 and EGFR are titrated to not detect normal levels of these proteins. And so, p53 or EGFR brown (or red) positive staining seen on the slide actually indicates over-expression of p53 or EGFR protein. Over-expression of p53 or EGFR protein gene product reflects the corresponding abnormal gene in the GbM cells. The GbM staining of its overexpressed p53 or EGFR protein exceeds that of normal structures including vessels with undetected p53 and EGFR (see Fig. 20.31 ).
In adults, the most common molecular signature seen on IHC staining of primary glioblastomas is IDH-wild-type (IDH stain/PCR negative), ATRX-wild-type (ATRX positive), p53 over-expressed often, but not always, in few cells; and EGFR strongly and diffusely over-expressed (see Figs. 20.31 and 20.32 ). IHC for IDH1 R132H is always recommended. Because most primary glioblastomas are IDH-wild-type (i.e., have no IDH mutation), most do not stain for IDH1 R132H. In younger patients and people with prior lower-grade glioma, absent ATRX and strong p53, PCR for all IDH mutations is critical despite cost if their IHC for IDH1 R132H is negative. This PCR is sometimes skipped in IHC-negative patients over 54 years old. They have a less than 1% chance of having a non-R132H IDH mutation. The IHC stain for ATRX stains the normal (i.e., wild-type) ATRX gene product. Because most primary glioblastomas are ATRX-wild-type, they retain ATRX staining, and the glioblastoma cell staining matches staining of vessels and brain.
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here