Listen to a discussion of this article through our AI-generated podcast.
Beginning with our September 2024 issue, CAP TODAY now offers the option to listen to a discussion of our articles with voices reminiscent of a Siri or Garmin speaker. Since this AI-generated podcast uses technology that is still in its early stages of development, there may be some errors in pronunciation and the tone may sound too upbeat for the material covered.
Amy Carpenter
September 2024—Leomar Ballester, MD, PhD, of the University of Texas MD Anderson Cancer Center has simple advice for those called to provide an intraoperative neuropathologic consultation: review the imaging and know the entities.
“If you don’t review the imaging first, you’re making your job more difficult and increasing the chances of a misdiagnosis.” And “it’s impossible to diagnose a tumor that you don’t know exists,” says Dr. Ballester, associate professor of neuropathology and molecular genetic pathology in the anatomic pathology and translational molecular pathology departments, Division of Pathology and Laboratory Medicine.
When he was in training, he debated whether to pursue neuropathology because of the frozen sections and how that would make his life “slightly stressful.” Now it’s the part of the day he likes most. “It’s enjoyable when you feel comfortable about an answer,” he says.
For pathologists not yet fully comfortable because they’re still in training or because they do neuropathology consultations infrequently, he provided a few tips and cases in a CAP23 session last fall.
Age, sex, histories of cancer and cancer predisposition syndrome, tumor location, and imaging characteristics should be known before an intraoperative consultation, he noted. “I spend a lot of time reviewing this before I go for an intraoperative consultation. Many times you already have a good idea or know exactly what the diagnosis is going to be, and that removes a lot of the stress associated with doing an intraoperative consultation.”
Of all brain tumors, metastasis is the most common diagnosis, he said. Meningiomas are the second most common diagnoses, accounting for 40 percent of primary brain tumors, and pituitary adenomas and glioblastomas (infiltrating gliomas) constitute 17 percent and 14 percent of primary brain tumors, respectively. The “other” category, at 21 percent, “is a small percentage of all the patients we see, but hundreds of tumors fall into this category,” Dr. Ballester said. His approach is to be prepared to recognize and diagnose metastases, meningioma, pituitary adenomas, and glioblastoma tumors efficiently, “so then I have enough time to focus on and work up less frequently presented tumors when they appear.”
Tumor location is part of how he builds the differential diagnosis before an intraoperative consultation or his review of the case; for example, a pineal region mass raises a very specific differential diagnosis, he said. In this region are the pineal parenchymal tumors, pineocytomas, pineal parenchymal tumors of intermediate differentiation (grades two or three), pineoblastomas (grades one and four), germinomas, and papillary tumors of the pineal region.
Diffuse midline glioma H3K27-altered is found in the brainstem, thalamus, and spinal cord and more common in pediatric patients.
In the setting of a cyst with a mural nodule in the cerebellum, pilocytic astrocytomas or hemangioblastomas would be the most common lesions, Dr. Ballester said. “In an adult patient, a tumor in the cerebellum is most likely a metastasis. And for any pediatric patient, we have to think about medulloblastoma. If it involves the fourth ventricle, think of an ependymoma.”
Tumors in the pituitary region are the pituitary adenoma, craniopharyngioma (adamantinomatous or papillary), germinoma, Rathke’s cyst, and meningioma. Hemispheric tumors include glioblastomas, astrocytomas, oligodendrogliomas, and metastasis.
Common tumors presenting hemorrhagic metastases are renal cell carcinomas, melanomas, and choriocarcinomas. When receiving a specimen for an intracranial hemorrhage, Dr. Ballester said, “consider metastases in the differential as well. I’ve been surprised a few times with an unexpected metastasis in the setting of an intracerebral hemorrhage.” A cerebellar contrast-enhancing lesion in an adult patient is a common presentation and most likely a metastasis.
Subclassification of diffuse gliomas requires molecular characterization, and Dr. Ballester says he’s happy to leave behind the days when pathologists were forced to distinguish between astrocytoma, oligodendroglioma, and glioblastoma based on histology alone. At the time of frozen section, there’s no need to make a diagnosis of glioblastoma, oligodendroglioma, astrocytoma. “Diffuse glioma is sufficient; if there are high-grade histology features, then high-grade diffuse glioma, and that is enough to guide the surgery,” he said.
Adult-type diffuse gliomas consist of astrocytoma, IDH-mutant, CNS WHO grades two–four; oligodendroglioma, IDH-mutant, 1p/19q-codeleted, CNS WHO grades two–three; and glioblastoma, IDH-wildtype, CNS WHO grade four.
Dr. Ballester presented a case in which the patient’s lesion had contrast enhancing and non-enhancing components, and histology pointed to a diffuse glioma.
In the predominant non-enhancing component of the glioma, the histology shows minimal mitotic activity and no vascular proliferation or necrosis, and it would be a CNS WHO grade two tumor (Fig. 1). The minor enhancing component looks more aggressive, with vascular proliferation, necrosis, and elevated mitotic activity (Fig. 2). “This would make the tumor a grade four,” Dr. Ballester said.
“This brings up the issue of sampling and why it’s also important to review the imaging, so you can recognize when the tissue you have may not be entirely representative of the lesion,” he said.
A classic triad of molecular alterations—IDH1 or IDH2 mutation, ATRX loss of expression, and TP53 mutation—is associated with IDH-mutant diffuse astrocytic disease. “And we have immunohistochemistry for the most common mutations in IDH1, ATRX, and TP53.”
Fig. 3 is an example of an IDH1 R132H-mutant tumor by diffuse expression of the mutant protein. In Fig. 4, ATRX is seen with the normal cells retaining expression of the ATRX protein, while the tumor cells are negative. In this tumor with an ATRX mutation, the normal glia and endothelial cells retain ATRX expression and don’t have the mutation, he said. TP53 can be variable but was not positive in this case. The final diagnosis was an astrocytoma, IDH-mutant, CNS WHO grade four.
In the second case, the imaging suggests glioblastoma by contrast-enhancing lesion (Fig. 5). The histology shows vascular proliferation and necrosis with pseudopalisading (Fig. 6), with the cells running away from the necrotic area and the vascular proliferation increasing the blood supply. “Classic histology for high-grade gliomas,” Dr. Ballester said.
This lesion in 2007 would have been “a glioblastoma, WHO grade four, no additional testing needed,” he said. Today the same histology can be any of five entities: astrocytoma, IDH-mutant, CNS WHO grade four; glioblastoma, IDH-wildtype, CNS WHO grade four; diffuse midline glioma, H3K27-altered, CNS WHO grade four; diffuse hemispheric glioma, G34-mutant, CNS WHO grade four; diffuse pediatric-type high-grade glioma, H3-wildtype and IDH-wildtype, CNS WHO grade four. “We cannot distinguish that on histology alone, so we need further testing,” Dr. Ballester said. Two immuno-stains for this scenario are the IDH1 R132H-mutant protein and the antibody against the H3K27M-mutant protein.
“Now some of these entities, like the G34-mutant tumors, require sequencing,” he said, “and it’s important to know when you require sequencing to make the final diagnosis.”
In his third case, Dr. Ballester showed a non-enhancing lesion (Fig. 7) for which a glioma is suspected, he said, though not a high-grade glioma because it is enhancement that correlates with the histologic characteristics of high-grade gliomas—vascular proliferation and necrosis.
The histology is consistent with a diffuse glioma (Fig. 8) with no elevated mitotic activity, no microvascular proliferation, and no necrosis. “And still the differential diagnosis in this case today could be an astrocytoma, IDH-mutant, grade two, three, or four, or a glioblastoma, IDH-wildtype, grade four, even without vascular proliferation or necrosis,” Dr. Ballester said. “It could be a diffuse hemispheric glioma, G34-mutant, grade four, or a diffuse low-grade glioma, MAPK-altered.
“The reason for this is that we don’t need vascular proliferation or necrosis to make a diagnosis of a high-grade glioma these days because there are molecular alterations that correlate with grade four behavior,” he continued. “So what looks in histology as a grade two tumor could very well be a grade four lesion, and that’s important to recognize in terms of the age of the patient, the location of the tumor, and when it’s appropriate to do molecular characterization.”
The final case was a non-enhancing lesion involving the pons (Fig. 9). Here too the histology is consistent with a diffuse glioma, “completely different just based on the location and the molecular characteristics of the tumor” (Fig. 10).
There is no vascular proliferation or necrosis, but this tumor was positive for the H3K27M mutation, and the tumor cells are positive for the mutant protein. “The endothelial cells are not; they don’t have the mutation.”
Among the diagnostic pitfalls is that the IDH1 R132H immunostain doesn’t capture all the clinically relevant IDH1 and IDH2 mutations, so in some cases sequencing is needed, particularly in patients under age 55, Dr. Ballester said.
“It’s important to exclude the non-canonical mutations in IDH1 and IDH2 which the antibody doesn’t recognize. It recognizes only the most common mutation, which is R132H.”
Two antibodies look at the lysine 27 in the histone 3 protein—one against the mutant protein, H3K27M, and one against the trimethylated lysine 27, which is the wildtype scenario. In Fig. 11 (top), the tumor does not have a mutation and is H3K27-wildtype, and the antibody against the mutant protein is negative. In the antibody against the trimethylated lysine 27, the signal is retained in all the cells—tumor and endothelial. “All the cells are positive.”
Fig. 11 (bottom) also shows an H3K27M-mutant tumor and expression of the mutant protein with the antibody in the tumor cells. “Endothelial cells do not have the mutation and do not express the mutant protein,” Dr. Ballester said. “This antibody against the trimethylated lysine 27 is negative in the tumor cells, but the expression is retained in epithelial cells because these cells do not have the mutation.”
Recognizing that the two different antibodies serve two different purposes is important, he said. “I’ve seen diagnoses where people have interpreted this as the mutant protein and made a diagnosis of diffuse midline glioma, K27M-mutant, because it’s positive in all the cells.”
Small changes in nomenclature can cause confusion, he noted. In histone 3, lysine 27, “we later discovered that the amino acid methionine at the beginning of the protein was being cleaved. So the numbers changed, and lysine 27 is in fact lysine 28. And the gene name changed from H3F3A to H3-3A. So H3F3A p.K27M-mutant is the same as an H3-3A p.K28M mutation”—important to be aware of, he added, because sequencing reports may vary.
A small number of these tumors don’t have methionine, Dr. Ballester said. “The mutation is to a different amino acid not recognized by the antibody. So they would be negative for the K27M-mutant antibody but will show the K27me3 loss. “That tells you something’s wrong at the position, and that is one way of recognizing these other tumors that don’t necessarily have the K27M mutation but still fall under this diagnosis of diffuse midline glioma, H3K27-altered.”
Oligodendroglioma is defined as a diffuse glioma with IDH1/IDH2 mutation and codeletion of 1p/19q (Ballester LY, et al. Hum Pathol. 2017;69:15–22). “However,” Dr. Ballester said, “there is a small number of glioblastomas, about five to 10 percent, that will show 1p/19q codeletion by FISH. But those tumors are IDH1/IDH2 wildtype, so they’re not oligodendrogliomas, and they sometimes create confusion” in a case that looks like a glioblastoma but is 1p/19q codeleted by FISH (if FISH is ordered). “It raises the question, ‘Is this an oligodendroglioma?’ and the answer is no.”
FISH probes are very small in relationship to the chromosome (1 and 19), and small partial losses can interfere with the regions where the probes bind and give the appearance of 1p/19q codeletion by FISH, he said. “That happens in a percentage of glioblastomas.” But that is not the entire loss of chromosome 1p and 19q seen in oligodendrogliomas. An array will not show 1p/19q codeletion, he said. “It’s a limitation of the FISH assay.”
Dr. Ballester refers routinely to the 2021 WHO classification of central nervous system tumors, which he says took to the next level the molecular classification of brain tumors introduced in the 2016 edition.
“We even have new tumor entities based on novel diagnostic technologies like DNA methylome profiling. Some tumors can be diagnosed with certainty only after methylation analysis because they are defined by their unique methylation profile,” Dr. Ballester said. Although the majority of brain tumors today don’t require methylation profiling, “it’s important to recognize when you’re dealing with one of these tumors that requires methylation analysis and to know that technique is available.” In some cases, he said, it’s the most appropriate test to diagnose a particular tumor.
Amy Carpenter is CAP TODAY senior editor.