College of American Pathologists

2012 — April Case of the Month

Posted April 10, 2012

CLINICAL SUMMARY: Central nervous system  

CAP Foundation April Online Case of the Month

Click Slide Image to View Case with DigitalScope

After reading the summary, try answering the three related multiple-choice questions below.

A 55-year-old man presented with a one year history of insidiously progressive neurobehavioral deficits, most recently confusion, mild left hemiparesis, and an abnormal gait. MRI of the brain showed extensive hyperintensity of the frontal white matter, somewhat more prominent on the right, extending to the temporal lobes, diencephalon, midbrain, and pons. A diagnostic biopsy was performed. The patient succumbed within one year.

Archive Case and Diagnosis: This case first appeared as Performance Improvement Program in Surgical Pathology (PIP) 2009, case 11, and is gliomatosis cerebri.

Criteria for Diagnosis and Comments:
Sections show a highly infiltrative glial neoplasm that, by neuroimaging studies, was shown to involve a large proportion of the central nervous system. The diagnosis is gliomatosis cerebri.

In the recent 2007 World Health Organization (WHO) classification, gliomatosis cerebri (GC) is described as a glial neoplastic process that involves “at least three cerebral lobes, usually with bilateral involvement of the cerebral hemispheres and/or deep gray matter, and frequent extension to the brain stem, cerebellum and even the spinal cord.” This reflects a change from the previous edition in which GC was considered a “neuroepithelial tumour of uncertain origin.” Currently, GC is recognized as a variant of astrocytoma with a "pattern of particularly extensive glioma infiltration."

Fewer than 200 cases of GC have been reported in the literature, and in the past, most examples were diagnosed postmortem. Today, improved neuroimaging studies show that diffuse, multilobe, bilateral infiltration of brain by gliomas is not as rare as was previously appreciated. The cerebrum is the most common anatomic site involved, and individual neoplastic cells have a predilection for diffuse permeation of the cerebral white matter, especially the centrum semiovale and corpus callosum. Although the neoplastic cells may also spread to cortical and subcortical gray matter, only 19% of cases show infiltration of the cerebral cortical gray matter, 43% of cases involve thalamus, 34% spread to basal ganglia, and 29% to cerebellum. The centrum semiovale of the cerebral white matter is the most common site of involvement and more than three-fourths of cases are bilateral.

Two types of GC have been distinguished by some authors: type I (the classical lesion), which shows diffuse infiltration of tumor without formation of a cohesive tumor mass (at least at the time of initial clinical presentation) and type II, which demonstrates extensive tumor involvement of multiple lobes of the brain by ill-defined infiltrative tumor but with an additional, superimposed neoplastic mass lesion. Patients of any age can be affected. As with most gliomas, prognosis correlates with patient age and grade.

Histologically, GC shows elongated individual glial cells permeating the background neuropil. No necrosis, microcyst formation, microvascular proliferation, calcification, or microglial clusters are present. Single tumor cells preferentially infiltrate the white matter parallel to myelinated fibers in the white matter, yielding the impression that they are aligned in a linear fashion.

Unlike reactive microglial cells, the cells in GC are not clustered into tight microglial “stars,” and are not associated with neuronophagia (i.e., they are not responding to individual neuronal cell death), or perivascular chronic inflammation, as in viral infections of the CNS. In addition, tumor nuclei are significantly more hyperchromatic and variable in size and shape than reactive microglial cells.

Many tumor cells are devoid of visible eosinophilic cytoplasm on hematoxylin and eosin staining (H&E), making their interpretation as glial cells less obvious. Usually a wide range of cell morphology can be discerned if enough tissue is present on the biopsy, and in most cases at least some cells show immunoreactivity for glial fibrillary acidic protein (GFAP) or S100 protein. Rarely GC can be composed of oligodendroglial tumor cells; the usual or predominant astrocytic phenotype of the tumor, however, allowed it to be classified as an astrocytic neoplasm in the 2007 WHO classification. Proliferation in GC correlates with tumor grade. Some cases have easily identifiable mitoses and cell cycle labeling indices of up to 30%; others have few mitoses and low proliferation indices.

As in all gliomas, and especially in high-grade or previously treated gliomas with necrosis, microglial cells and macrophages may be present. These cells will show the expected immunoreactivity for CD68, CD45 (LCA), and/or HAM56. Microglial cells, however, are not the main cell type in GC and cannot account for the hypercellularity, considerable cytological atypia, and hyperchromaticity in the lesion. Hence, pathological diagnosis of GC often rests on the identification of the neoplastic cytologic features; indiscriminate use of immunohistochemistry out of context with the H&E findings can lead to an incorrect diagnostic interpretation.

For the pathologist unfamiliar with the entity of GC, the elongate shape of the tumor cells could prompt consideration of pilocytic astrocytoma. However, cells in GC lack the easily visible, elongate “piloid,” bipolar, hair-like cytoplasmic processes characteristic of pilocytic astrocytoma. They also lack the additional architectural features of pilocytic astrocytoma such as microcyst formation, calcifications, and glomeruloid vasculature.

The histologic features of the biopsy allow diagnosis of a glial neoplasm and negate an inflammatory or degenerative process. Clearly, pathological-neuroimaging correlation is necessary to appreciate the full extent of brain involvement, especially when a surgical biopsy, rather than an autopsy brain and spinal cord specimen, is available to the pathologist.

The clinical features of GC can be confusing. A presentation with cognitive deficits often leads to a diagnosis of a neurodegenerative or neuropsychological disorder until more focal findings (such as hemiparesis or seizures) manifest themselves. Once neuroimaging studies are obtained, most neurodegenerative disorders can be eliminated from consideration. Biopsies of GC also lack spongiform change or neuronal loss, neurofibrillary tangles, and neuritic plaques (on H&E or special silver stains). Hence, pathologic examination excludes a diagnosis of Creutzfeldt-Jakob disease or Alzheimer disease, respectively.

Unlike most glial neoplasms that form a nidus of high-cell density tumor growth with destruction of underlying brain tissue, in GC the architecture of the brain remains largely intact. Myelin is mainly affected, and axons are less injured, at least initially, accounting for this structural preservation when small numbers of tumor cells infiltrate any given area of the brain. This relatively greater myelin loss correlates with MRI changes, especially on T2-weighted images, and occasionally can lead to neuroimaging consideration of demyelinating disorders such as multiple sclerosis or leukoencephalopathy until biopsy or autopsy demonstrates the cytologic atypia of the infiltrating tumor cells in GC.

For pathologists and clinicians, this extensive white matter damage explains why neurobehavioral dysfunction may be the most prominent feature of GC at all stages of the disease. Artigas and colleagues, in their review of the literature, noted that the most common presenting features of GC were mental and behavioral changes (40%), as opposed to seizures (29%) or headache (29%). Mental and behavioral changes were also the most common persistent clinical features. While corticospinal tract deficits often are listed as the single most common clinical symptom of GC (58%), the neurobehavioral features, when considered collectively, are actually more common (dementia in 44%, lethargy and/or obtundation in 20%, and behavioral changes/psychosis in 19% of patients). These statistics argue that GC is a neoplasm with predominantly neurobehavioral manifestations that parallel its tendency to cause cerebral white matter damage. Both clinicians and pathologists need to be aware of the unusual clinical and morphological features of this entity.

Supplementary Questions: For each of the following, select the most likely diagnosis from the diagnostic set (an answer may be used once, more than once, or not at all).

Question Diagnostic Set
1. The most commonly involved anatomic location for the presented entity is: A. Basal ganglia
B. Cerebellar vermis
C. Cerebral white matter
D. Hippocampus
E. Optic nerves
2. How is this process currently classified? A. Extensively infiltrative glial neoplasm
B. Highly reactive inflammatory condition
C. Low-grade glioma
D. Neurodegenerative disorder
E. Uniformly fatal infectious condition
3. The single key histologic feature that allows accurate diagnosis is which of the following? A. Cytologic atypia
B. Demyelination
C. Microglial clusters
D. Spongiform change
E. Viral inclusions


  1. Fuller, GN, Kros, JM. Gliomatosis cerebri. In: Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, eds. WHO Classification of Tumours of the Central Nervous System. Lyon, France: IARC Press; 2007:50 52.
  2. Artigas J, Cervos-Navarro J, Iglesias JR, Ebhardt G. Gliomatosis cerebri: Clinical and histological findings. Clin Neuropathol. 1985;4:135 148.
  3. Lantos PL, Bruner JM. Gliomatosis cerebri. In: Kleihues P, Cavenee WK, eds. Pathology and Genetics of Tumours of the Nervous System. Lyon, France: IARC Press; 2000:92 93.
  4. Jennings MT, Frenchman M, Shehab T, et al. Gliomatosis cerebri presenting as intractable epilepsy during early childhood. J Child Neurol.1995;10:37-45.
  5. Filley CM, Kleinschmidt-DeMasters BK, Lillehei KO, et al. Gliomatosis cerebri: Neurobehavioral and neuropathological observations. Cogn Behav Neurol. 2003;16:149 59.

B.K. DeMasters, MD FCAP
Neuropathology Committee
University of Colorado
Aurora, CO

Anthony T. Yachnis, MD FCAP
Neuropathology Committee
Shands Hospital at University of Florida
Gainesville, FL