A Contemporary Approach to Intraoperative Evaluation in Neuropathology

Although IOCs in neuropathology are colloquially referred to as “frozen sections” or “frozens,” cytologic preparations are also commonly employed. This generally takes the form of a “smear” or “squash smear” in which a small piece of tissue is squashed between 2 slides and smeared in a linear fashion, although a simple squash preparation or touch preparations may also be employed.⁵  Cytologic preparations have historically been used for neuropathology IOC because they can be performed and interpreted rapidly, they provide cytologic information, and they eliminate common artifacts created while freezing CNS tissue due to the higher water content compared to other organs. The frozen section, in contrast, offers the benefit of preserved tissue architecture at the expense of time and the introduction of freezing artifact. Some institutions may perform a cytologic preparation only in routine cases where evaluation of tumor architecture is not necessary, or where preservation of tissue is a priority. While performing a frozen and cytologic preparation together offers at least some diagnostic accuracy as compared to a smear alone, tissue preservation has become critical in the era of molecular diagnostics.^6–9  Therefore, a cytologic preparation alone to confirm the presence of diagnostic tissue may be warranted in many scenarios when tissue is limited.

Understanding institutional practice is a key component of performing an IOC. For example, some neurosurgeons may choose to provide an entire biopsy as a “frozen specimen” and allow the pathologist to sample for the IOC. In other institutions, the expectation may always be that a second specimen will follow, and the frozen specimen will be entirely consumed. Variation may also occur in the degree of diagnostic specificity that a surgeon may expect at the time of IOC. It is crucial that the pathologist understand the surgeon's expectations and what additional tissue, if any, will be provided after the IOC. Although IOCs frequently occur in the setting of a resection, many of the principles in this review are most applicable to a biopsy setting, where a limited additional sample will be provided for permanent section.

The choice to send a sample for IOC is highly variable among institutions and even among surgeons at the same institution. Regardless of practice, all neuropathology IOCs generally involve 2 common indications. The first, and by far the most common, indication is to ensure diagnostic tissue. The second is to provide a diagnosis at the time of IOC. This latter indication often is based more upon identifying a diagnostic category that may drive performing additional surgical resection or triaging whether additional testing is necessary, such as molecular or microbiologic studies. The remainder of this paper will be organized around these 2 indications.

Two factors are important in determining whether an intraoperative sample is “diagnostic” or “lesional.” First, the tissue provided should account for the clinical and radiologic presentation of the patient. Second, the tissue provided must be sufficient to allow the pathologist to make the most specific diagnosis possible on permanent sections, which may require immunohistochemistry, molecular testing, flow cytometry, microbiologic studies, and/or other ancillary tests. In the setting of a suspected neoplasm, there should be enough adequately sampled tissue to classify a tumor according to the 2021 CNS WHO classification system within the limits of the surgical capacity. In the 2021 CNS WHO section on gliomas, glioneuronal tumors, and neuronal tumors, 23 of the 41 tumors include molecular alterations as part of the essential criteria for diagnosis. For an additional 17 entities, it is “desirable.”⁴  Molecular characterization in neuropathology is challenging for several reasons. First, it may be difficult to obtain large amounts of tissue in certain regions of the brain or spinal cord. Second, for diffuse or infiltrating tumors, the tumor content may be low compared to normal CNS constituents and inflammatory cells. Therefore, preserving tissue for ancillary testing should generally have a higher priority than consuming additional tissue to make a more specific diagnosis at the time of IOC.

A wide variety of ancillary testing can be used to assess for diagnostic molecular alterations on permanent sections, which is briefly discussed at the end of this review. In biopsy specimens, the amount of tissue necessary for adequate molecular characterization is highly dependent on the tumor type. Some molecular alterations can be identified by immunohistochemistry, especially on tumors with low tumor cell content; other characterization may require more ample tissue. Additional triaging at the time of IOC may be dependent on factors such as brain banking or clinical trial enrollment. While clinical molecular testing is almost always performed on formalin-fixed, paraffin-embedded tissue, there may be clinical trials or other specific circumstances that require fresh or snap-frozen tissue. Other testing, including flow cytometry or microbiologic studies, should also be anticipated after the IOC diagnosis, and the need for such testing should be directly communicated to the surgeon.

The first step in confirming diagnostic tissue is an understanding of the cytologic and histologic features of nonneoplastic brain tissue. Nonneoplastic brain will typically smear in an even fashion, with an evenly distributed combination of astrocytes, oligodendrocytes, and neurons (Figure 1, a). In cerebellar lesions, cerebellar granular neurons will appear as poorly cohesive small, round cells, which can be mistaken for neoplasm (eg, a small round blue cell tumor) or lymphocytes (Figure 1, b).

Reactive changes can be observed adjacent to any neoplasm, as part of chronic seizure or injury, as well as postradiation, among other scenarios. The greatest challenge in establishing diagnostic tissue occurs in distinguishing reactive gliosis from the infiltrating edge of a glial neoplasm. The challenge of distinguishing gliosis from glioma has been extensively discussed in the literature.^7,9,10  Tissue undergoing reactive gliosis may appear mildly hypercellular, including not only reactive astrocytes but additional lymphocytes, activated microglia, and macrophages, all of which may resemble an infiltrating glial neoplasm. Reactive astrocytes can be recognized by abundant eosinophilic cytoplasm and numerous long glial processes (Figure 1, c). Piloid gliosis, characterized by a predominance of bundles of glial intermediate filaments referred to as Rosenthal fibers, can be seen in areas of long-standing injury and should not be mistaken for a pilocytic astrocytoma (Figure 1, d). The helpful features that would favor neoplastic tissue rather than reactive gliosis include the degree of cellularity; the presence of cells with angulated, hyperchromatic nuclei; the absence of significant fibrillary processes; and the loss of even spacing between astrocytes. The so-called “secondary structures of Scherer,” used to describe patterns of infiltrative growth, are particularly useful. These include tumor cell growth around neurons and vessels (perineuronal and perivascular satellitosis), distribution along the pial surface (subpial spread), and growth along white matter tracts (interfascicular spread).¹¹  These features are generally best appreciated on frozen section, which can be a useful adjunct to cytologic preparation in this setting (Figures 1, e and f).

It is useful to have a structured approach for these diagnoses to convey the appropriate information to the surgeon. For example, a diagnosis of “reactive gliosis” will typically convey no suspicion for tumor in the sample. This may indicate to the surgeon that the tissue is entirely nondiagnostic or representative of a nonneoplastic specimen, if the surgeon is confident the lesion has been sampled. The presence of reactive gliosis alone may be diagnostic in certain settings, such as one in which the clinical differential diagnosis includes radiation change or encephalitis. “Reactive gliosis with atypical glial cells” conveys that the tissue is potentially near an infiltrating edge of a glioma but is not yet diagnostic. Even in the presence of atypical glial cells, a few circumstances may raise the possibility of other diagnoses. Abundant necrosis can be seen in infarcts, radiation necrosis, and infectious causes and, even if a tumor was sampled, the presence of only necrotic tumor cells will limit diagnosis. Inflammatory cells or macrophages should raise the possibility of demyelination, infection, infarct, or partially treated lymphoma. Analyzing the cells adjacent and admixed with the inflammation will help inform the most likely diagnosis. The differential diagnosis of these nonneoplastic entities is beyond the scope of this current review.

If the pathologist is convinced of the presence of an infiltrating glioma, it is helpful to assess whether the tumor appears to be at an infiltrating edge or in dense tumor, to convey whether more tissue is necessary for ancillary testing. A small, paucicellular sample of tumor may preclude the appropriate immunohistochemical or molecular evaluation on permanent sections. In cases where an apparently nondiagnostic specimen is obtained, either because a hypocellular infiltrating edge was sampled, or because it was not surgically possible to obtain more material, a diagnosis can sometimes be salvaged by molecular testing on the infiltrating edge of tumor to prove the presence of neoplastic cells. Next-generation sequencing (NGS)-based testing is especially sensitive in detecting alterations in such cases, and in our experience, has sometimes identified diagnostic molecular alterations in cases with a relatively small number of atypical cells.

The second indication for intraoperative diagnosis is establishing a diagnosis, which may include providing a grade, where appropriate. Placing a tumor in the most accurate possible diagnostic category is desirable and particularly critical in 2 specific circumstances. First, delineating circumscribed tumors, particularly ependymomas, from infiltrating gliomas may be important in triggering surgical management of maximal resection for cure. Second, it is important to identify specific nonneoplastic lesions or tumors, such as lymphomas and germinomas, which should receive medical therapy rather than surgical resection.

Briefly, a general approach to providing a grade at the time of IOC is provided, followed by a review of several important broad diagnostic categories: (1) gliomas; (2) glioneuronal tumors; (3) ependymomas; (4) meningiomas; (5) poorly differentiated or small blue cell tumors. These categories are not intended to be comprehensive but to provide a general diagnostic framework for evaluating common CNS neoplasms.

At the time of IOC, neurosurgeons are commonly interested in having a general understanding of grading. For radiologically unusual lesions or pediatric patients, this may sometimes be a greater priority than knowing a general diagnostic category. The intraoperative pathologist should have a general grading framework and be aware of how recent changes, particularly with the 2021 CNS WHO, may drive ultimate grading that is discrepant from the histology seen at the time of IOC. Dating back to 1979 with the institution of the WHO tumor grading scheme, traditional tumor classification in neuropathology has divided CNS lesions into 4 tiers, corresponding to biological behavior. While grade 1 tumors are well-characterized tumors with benign behavior, curable with resection, grades from 2 to 4 are more complex depending on the tumor type. Uncommonly, a precise grade can be provided at the time of IOC; more frequently, a lesion can be classified as low-grade (corresponding to CNS WHO grade 1 or 2) or high-grade (corresponding to WHO grade 3 or 4). While some combination of mitotic activity, cellularity, nuclear atypia, microvascular proliferation (MVP), and/or necrosis may justify classification as “high-grade” at the time of IOC, these features are highly variable among tumor type and will be discussed below.

In the recent 2021 CNS WHO classification, a small number of newly described entities are ungraded as their biologic potential is still not clearly understood. These entities are rare. Nonetheless, for unusual gliomas or glioneuronal tumors that do not appear to clearly fit in a diagnostic category, it is best to defer on grading at the time of IOC. Additionally, many entities now incorporate molecular criteria as part of final grading. In such cases, it may be warranted to clarify that an IOC grade is only based on histologic criteria or, again, to defer on grading. These entities are described in greater detail in the diagnostic categories below.³ 

Gliomas are the most commonly encountered primary CNS neoplasms and have seen significant changes in grading, classification, and diagnostic criteria in recent years. Gliomas have a wide variety of features, but will typically be defined by the presence of tumor cells with eosinophilic, fibrillary cytoplasm and variable degrees of processes, usually best established on a cytologic preparation. Variations on glial morphology, including oligodendroglial morphology (cells with round nuclei, distinct nucleoli, short glial processes, and perinuclear halos), astrocytic morphology (abundant eosinophilic cytoplasm and often oval nuclei, which may be peripherally arranged), and piloid morphology (cells with long bipolar fibrillary processes), may also occur (Figure 2, a).

The 2021 CNS WHO separates gliomas into “diffuse” (synonymous with “infiltrative”) and “circumscribed.” Infiltrating gliomas are much more common. It is helpful, when possible, to identify infiltrative growth. Evidence of secondary structures of Scherer and the presence of axons or nonneoplastic neurons in the background of the neoplastic glial cells helps identify the lesion as infiltrating. If these features are present at IOC, an interpretation of “infiltrating glioma” is appropriate. In small biopsies, it may not be possible to evaluate these features. Some high-grade infiltrating gliomas, including gliosarcomas, epithelioid glioblastomas, and giant-cell glioblastomas, may also grow in a more well-circumscribed fashion, although these are rare variants.^12–14 

The 2021 CNS WHO classification further divides infiltrating gliomas into adult-type and pediatric-type infiltrating gliomas. The 2021 CNS WHO criteria has simplified adult-type infiltrating gliomas into 3 categories based on the presence or absence of mutation in the genes coding for isocitrate dehydrogenase (IDH), and these 3 will compose the majority of infiltrating gliomas encountered at IOC: (1) glioblastoma, IDH–wild type; (2) astrocytoma, IDH-mutant; and (3) oligodendroglioma, IDH-mutant.

Glioblastomas are the most common infiltrating gliomas. These tumors will typically appear as enhancing supratentorial masses in older adults. Classically, at the time of IOC they will be hypercellular lesions, composed of cells with angulated, hyperchromatic nuclei, with irregular nuclear contours and varying degrees of nuclear pleomorphism, accompanied by mitotic activity, necrosis, and/or MVP (Figure 2, b).

It is not uncommon for an intraoperative specimen in an older adult with an enhancing lesion to show an infiltrating glial neoplasm with “low-grade” histologic features. This is often due to undersampling (eg, peripheral infiltrating edge). In such cases, it is important to refrain from using “low-grade glial neoplasm” and instead use “glioma” or “infiltrating glioma, with no high-grade features seen.” If the lesion is enhancing, the latter communicates that the specimen examined does not fully explain the clinical and radiologic presentation. In such situations, the surgeon may wish to collect additional material to identify diagnostic features. Where diagnostic features are not sampled, molecular testing can establish a diagnosis of glioblastoma. The molecular characteristics of either a TERT promoter mutation, EGFR amplification, or gain of chromosome 7 and loss of chromosome 10 are now diagnostic of glioblastoma, IDH–wild type, CNS WHO grade 4, even in the setting of low-grade histology.³  Lower-grade gliomas do rarely occur in older adults (>55 years of age) and can be considered in nonenhancing lesions. Given the rarity of this situation, a diagnosis of “low-grade glioma” should still be avoided in nonenhancing lesions. Some IDH–wild type tumors present without enhancement and have low-grade histology, and only a TERT promoter mutation is identified with molecular testing. The biologic behavior of these lesions is uncertain,^15,16  but this circumstance highlights that even nonenhancing lesions with “low-grade” radiology may not behave as low-grade tumors. In these cases, a deferral on grading and preservation of tissue for molecular testing is of paramount importance.

The term “glioblastoma,” which previously described any diffuse glioma with necrosis or MVP, now only refers to a specific IDH–wild-type high-grade glioma, which usually occurs in adults older than 55 years of age and is typically, but not always, associated with specific molecular features (EGFR amplification, +7/−10, and/or TERT promoter mutations). Specifically, IDH-mutant tumors are no longer ever referred to as glioblastomas; the previous term “glioblastoma, IDH-mutant” has been renamed “astrocytoma, IDH-mutant, CNS WHO grade 4.” Therefore, it is not possible to distinguish a glioblastoma, IDH–wild-type, CNS WHO grade 4 from an astrocytoma, IDH-mutant, CNS WHO grade 4 on histology alone. While it is tempting to render an intraoperative diagnosis of “glioblastoma” for a high-grade glial neoplasm with MVP and/or necrosis, the term “glioblastoma” should be avoided at the time of frozen section. Instead, terms such as “high-grade glioma” or “high-grade astrocytoma” are more accurate and should be used. In older patients with classic histologic features, it is appropriate to state “favor glioblastoma” or “likely glioblastoma” upon request, but such specificity is usually not necessary.

IDH-mutant gliomas—astrocytomas and oligodendrogliomas—should be considered at IOC in adults younger than age 55, especially in the age range of 30 to 50 years with a supratentorial infiltrating glioma. These tumors generally present as large, bulky masses; usually the absence of contrast enhancement indicates a lower-grade tumor, and contrast enhancement suggests higher-grade progression. Classically, astrocytomas and oligodendrogliomas were distinguished on cellular morphology. Astrocytomas often have enlarged, angulated irregular nuclei with coarse chromatin and abundant fibrillar processes, while oligodendrogliomas have rounded uniform nuclei with fine chromatin and fewer glial processes and may have microcalcifications. Previously, the diagnosis of “oligoastrocytoma” was commonly utilized in cases of overlapping morphologic features, but all IDH-mutant tumors are now classified as either astrocytoma or oligodendroglioma. Astrocytoma, IDH-mutant, often harbors TP53 and ATRX mutations, while oligodendroglioma, IDH-mutant, harbors TERT promoter mutations and codeletion of chromosomes 1p and 19q. The latter feature is, by definition, absent in astrocytomas.⁴ 

As the distinction between oligodendroglioma and astrocytoma is now a molecular diagnosis, it is generally not helpful to attempt to distinguish the 2 at the time of IOC. Although the classic histologic features commonly correlate with the molecular diagnosis, they fail to correlate in many cases.¹⁷  For this reason, an attempt should not be made to distinguish these 2 entities as an intraoperative diagnosis, even when clear histologic features are present. A diagnosis of “infiltrating glioma” should generally be rendered.

The grading criteria for “low-grade” and “high-grade” also varies between astrocytoma, IDH-mutant, and oligodendroglioma, IDH-mutant, and therefore caution is warranted in grading these tumors at the time of IOC. In astrocytoma, IDH-mutant, a grade of 3 (corresponding to “high-grade”) is given for increased mitotic activity, with 1 mitotic figure justifying this diagnosis in a small biopsy.^4,18  In contrast, for oligodendroglioma, IDH-mutant, a combination of increased mitotic activity (usually greater than 5 mitotic figures per 10 high-power fields¹⁹), MVP, necrosis, or an increased Ki-67 proliferation index are used to warrant a grade of 3. In general, the term “high-grade infiltrating” glioma should be reserved for tumors in which there is significantly increased mitotic activity. Since the histologic features of a grade 3 or “high grade” lesion differ between the 2 entities, grading should be deferred in cases with borderline mitotic activity and no MVP or necrosis. No grade 4 classification exists for oligodendroglioma, and therefore a specific grade should not be provided in this scenario.

Careful correlation with radiology is important, particularly if there are areas of enhancement concerning for a higher-grade tumor or in settings of previously diagnosed IDH-mutant tumors where radiology is concerning for progression in grade. In this setting, it is important to evaluate for high-grade histologic features and report them during the IOC. Ultimately, this can inform the surgeon that truly diagnostic tissue for accurate grading was identified. If such biopsies still only show low-grade histology, a conversation with the neurosurgeon may be warranted to ensure that the areas of greatest suspicion have been sampled or to request additional tissue for permanent analysis. If a previous diagnosis of astrocytoma, IDH-mutant, has been established, it may be warranted to provide a CNS WHO grade of 4 on the basis of MVP and/or necrosis. However, prior radiation may limit interpretation of necrosis, and care should be taken to not overinterpret radiation necrosis as tumor-type necrosis.

Diffuse infiltrating gliomas often require significant molecular testing. Molecular testing often includes immunohistochemistry for IDH1 R132H, ATRX, and P53. In patients younger than 55 years of age, in whom an IDH1 mutation cannot be demonstrated by immunohistochemistry, molecular testing by polymerase chain reaction or NGS may be required for less common IDH1 or IDH2 mutations. For potential oligodendrogliomas, testing for 1p/19q codeletion and TERT promoter mutation might be necessary. Testing for homozygous deletion of CDKN2A/B genes is commonly performed in astrocytoma, IDH-mutant, as this alteration has been shown to confer a worse prognosis and warrant a diagnosis of astrocytoma, IDH-mutant, CNS WHO grade 4.^18,20  Testing for MGMT promoter methylation is standard of care for all glioblastomas and may be performed in other grade 4 tumors.²¹  Various assays can be used for these many biomarkers, and familiarity with the testing method that will be used may be useful in triaging small biopsies.

Most high-grade glial neoplasms have been reclassified in young adults, children, and infants. While traditional glioblastomas can still arise in these clinical settings, most of these tumors have been reclassified in the recent 2021 CNS WHO into specific categories based on age, location, and molecular characteristics. These cases will not be specifically diagnosed at the time of IOC, but an awareness of new entities can help guide intraoperative judgment. Among hemispheric tumors, many in children and young adults will ultimately be classified as “diffuse pediatric-type high-grade glioma” or “diffuse hemispheric glioma,” depending on molecular characteristics. While both tumors may histologically resemble glioblastomas, they may show primitive neuronal features, including cells with high nuclear to cytoplasmic ratios and rosette formation.²²  These tumors should not be mistakenly classified as “embryonal tumors.” In infants, many high-grade gliomas will be classified as “infant-type hemispheric glioma.” These tumors sometimes appear with spindled or fascicular architecture and may show more circumscribed growth. Many such cases are associated with specific fusions, and appropriate molecular testing should be anticipated.²³  In children and young adults, when a tumor is located in a midline structure (thalamus, brainstem, cerebellum, or spinal cord), the diffuse midline glioma, H3K27-altered, should be considered. Importantly, diffuse midline gliomas can sometimes show lower-grade histology, but these tumors are still graded as WHO grade 4 in the setting of the appropriate molecular alterations. Therefore, a diagnosis of “low-grade glioma” is not appropriate in such cases.

Several circumscribed gliomas are common enough to be considered at the time of IOC, particularly in children and young adults. Pilocytic astrocytoma (PA) usually presents as a cystic mass with an enhancing mural nodule in the cerebellum of a child or young adult, but may also occur in the supratentorial compartment.²⁴  Cytologic preparations at the time of IOC are extremely helpful in highlighting characteristic features of bipolar piloid glial processes, Rosenthal fibers, and eosinophilic granular bodies (EGBs), although these features may also be appreciated in frozen sections. Additional histologic features include biphasic appearance (solid and cystic areas), round, monotonous, oligodendrocyte-like cells, and hyalinized vessels.²⁵  Where radiology is classic and all histologic features are present, an intraoperative diagnosis of “pilocytic astrocytoma” can be rendered (Figure 2, c).

PAs present several unique pitfalls at the time of IOC. PAs may have areas of MVP that may erroneously lead to the diagnosis of a “high-grade glioma,” and this feature should not be overinterpreted in the setting of Rosenthal fibers, EGBs, prominent piloid features, and/or circumscribed growth. An abundance of Rosenthal fibers relative to astrocytes should raise the consideration that piloid gliosis was sampled, a reactive change that can be seen in long-standing lesions and is particularly prominent near hemangioblastomas, craniopharyngiomas, ependymomas, and cysts.⁷  Where infiltrative growth is noted, it may be appropriate to offer a more general diagnosis such as “low-grade glioma,” with the caveat that a more general diagnosis of “glioma” is appropriate if a tumor is markedly infiltrative and may be a diffuse midline glioma. A spectrum of newly defined diffuse pediatric low-grade gliomas can overlap with PA and should be considered in the differential diagnosis.²⁶  When the oligodendrocyte-like component is prominent, PA may be difficult to distinguish from other glioneuronal tumors (discussed below), especially if the tumor is supratentorial. In these cases, we prefer more general terminology such as “glial or glioneuronal tumor” to encompass these possibilities.^24,27–29  Finally, marked nuclear anaplasia, necrosis, and mitotic activity should trigger a more general diagnosis of “glioma” or, if prominent, “high-grade glioma,” even in the presence of clear piloid morphology and Rosenthal fibers. Such findings raise the possibility of a high-grade glioma with piloid features, a tumor that requires extensive molecular testing for definitive diagnosis.³⁰ 

Pleomorphic xanthoastrocytoma (PXA) is a cortically based circumscribed glioma that typically occurs in young people and is characterized by the presence of marked nuclear pleomorphism, so-called lipidized cells, and soft histologic features such as EGBs and perivascular lymphocytic cuffing. The presence of EGBs and relative circumscription in the setting of marked nuclear pleomorphism in a young adult should prompt consideration of PXA more than a high-grade glioma in a young adult. These features may be present in variable degrees, even on permanent sections, and the diagnosis of PXA is challenging without molecular testing. Additionally, PXA may show marked radiologic and histologic overlap with PA and ganglioglioma (discussed below), particularly where ganglion-cell differentiation is prominent. As PXA warrants a CNS WHO grade of 2, a diagnosis of “low-grade glial or glioneuronal tumor” may be appropriate. However, mitotic activity, MVP, and necrosis may warrant a CNS WHO grade of 3. Such cases may raise the differential diagnosis of anaplastic PXA (a CNS WHO grade 3 tumor) and epithelioid glioblastoma (a CNS WHO grade 4 tumor). As these entities are also challenging to distinguish without molecular testing, a diagnosis of “high-grade glioma” should be rendered.¹² 

At the time of IOC, a glioneuronal tumor should be considered in the clinical differential diagnosis of adolescents and young adults with radiologically well-circumscribed, cystic, or calcified lesions. These tumors may appear superficial or cortically based, as in the dysembryoplastic neuroepithelial tumor or ganglioglioma, while others have characteristic locations, including the septum pellucidum (central neurocytoma, myxoid glioneuronal tumor) or fourth ventricle (rosette-forming glioneuronal tumor). The WHO categorization of low-grade gliomas and glioneuronal tumors consists of numerous entities that show overlapping histologic features, which may require extensive immunohistochemistry and molecular testing to confirm a diagnosis. Oligodendrocyte-like morphology is a common feature of many tumors, which may also overlap with pilocytic astrocytoma or with other low-grade infiltrating gliomas.^24,27–29  While some of these tumors do have classic histologic features and location may aid in diagnosis, these may be variable and not always sampled at the time of IOC. Therefore, a general diagnosis such as “low-grade glial or glioneuronal tumor” or, if the neuronal component is evident, “low-grade glioneuronal tumor” should be rendered, rather than attempting specific subclassification intraoperatively.

The prototypic glioneuronal tumor is the ganglioglioma, which is defined by the presence of a neoplastic glial and neuronal population. Typically, frozen sections and cytologic preparations for a ganglioglioma may demonstrate some combination of neoplastic spindled or angulated glial cells with fibrillary processes or round cells with minimal cytoplasm. When neurons are identified within a tumor, it is necessary to distinguish neoplastic from nonneoplastic neurons, for which a cytologic preparation is particularly useful. Abnormal dysmorphic forms, neuronal cytoplasmic vacuoles, binucleation, and abnormal clustering are features suggestive of a neoplastic neuronal population. Perivascular lymphocytic cuffing and the presence of EGBs are other features supportive of ganglioglioma. As mentioned, given the substantial overlap between ganglioglioma, PA, and PXA, ganglioglioma may be difficult to diagnose at IOC unless the neoplastic ganglion-cell population is prominent.

As glioneuronal tumors are commonly well circumscribed and often cortically based, there is usually ample tissue for permanent section diagnosis, but tissue may be limited in smaller tumors or tumors in challenging locations like the fourth ventricle. Extensive molecular testing, including testing for fusions, point mutations, and/or DNA methylation profiling, might be needed in permanent sections for accurate classification and to provide prognosis and therapeutic targets.

Ependymoma poses a particularly critical juncture at the time of IOC. Even with the growing importance of molecular biomarkers, extent of resection remains a critical factor in the prognosis of ependymomas.^31,32  In general, ependymomas should be in the differential diagnosis of well-circumscribed intraventricular, periventricular, posterior fossa, or spinal intramedullary lesions. Although usually occurring in children and young adults, ependymomas can occur at any age. At IOC, ependymomas are classically composed of cells with glial fibrillary cytoplasm and round, monomorphic nuclei with speckled chromatin that are best appreciated on a cytologic preparation. Perivascular pseudorosettes are commonly present in both frozen section and cytologic preparation, where they may appear as glial cells clinging to vessels (Figure 2, d). Less common, but more specific, are true ependymal rosettes or ependymal canals. Ependymomas should be well circumscribed but may show reactive changes in the adjacent CNS parenchyma, including piloid gliosis. Any evidence of infiltrative growth should prompt a more general diagnosis such as “glial neoplasm.”

Final diagnosis of ependymoma will incorporate location and molecular features that will inform biologic behavior. The WHO classification continues to allow for histologic grading; therefore, a diagnosis of “high-grade” may still be appropriate if brisk mitotic activity, necrosis, and/or MVP are abundant at the time of IOC.⁴  The location of the lesion is of primary importance in anticipating the potential workup and diagnosis. Some degree of molecular testing should be anticipated in all ependymomas, and appropriate tissue should be preserved. Unique molecular classifications have been used for delineating supratentorial ependymomas (including ZFTA fusions),^33–36  spinal ependymomas (MYCN amplification),³⁷  and posterior fossa ependymomas (PFA and PFB subtypes), the latter of which are characterized by a DNA methylation profile rather than specific genetic alterations.³⁸  While molecular classification takes precedence, some histologic correlates can be recognized at the time of IOC. ZFTA-fused ependymomas sometimes present as clear-cell neoplasms with few or no appreciable perivascular pseudorosettes, and it is critical to keep supratentorial ependymoma in the differential diagnosis in such a scenario.³⁹  Generally, PFA ependymomas and MYCN-amplified spinal ependymomas show high-grade features, including nuclear hyperchromasia, increased cellularity, mitotic activity, necrosis, and/or MVP. Therefore, notable high-grade histology may be appropriate to comment on during an intraoperative diagnosis, but it should be recognized that the ultimate diagnosis is dependent on molecular characterization.³⁷ 

Subependymomas are well-circumscribed nodular, firm lesions located in the lateral ventricles or fourth ventricle, which often do not cause clinical symptoms unless they obstruct cerebrospinal fluid flow and correspond to a CNS WHO grade 1 tumor. These tumors show distinct histologic features, with clusters of small nuclei separated by nuclear-free fibrillary areas with hyalinized vessels and focal hemosiderin deposition and microcalcifications. Tumors with mixed subependymoma-ependymoma histology should be recognized; these cases require more comprehensive molecular testing as some subtypes may show more aggressive biologic behavior.⁴⁰  “Ependymal neoplasm” or “mixed ependymoma-subependymoma” is an appropriate intraoperative diagnosis in these cases.

Meningiomas typically present as extra-axial dural-based masses, which may appear anywhere throughout the CNS. A cytologic preparation is of utmost importance in the diagnosis of meningioma, as meningothelial cytologic features are evidently appreciated on such preparations. The presence of these cells, which have characteristic translucent cytoplasm that folds upon itself, giving the impression of a curtain or veil blowing in the wind, can warrant definitive diagnosis. Intranuclear pseudoinclusions, psammomatous calcifications, and meningothelial whorls are helpful features (Figure 2, e). The primary differential diagnosis for meningioma will often be carcinoma or schwannoma (if in the cerebello-pontine angle); a smear preparation is particularly useful in distinguishing the latter. Less common entities to consider include gliosarcoma, solitary fibrous tumor, and other intracranial sarcomas.

Most meningiomas are grade 1. Grading may be challenging at the time of IOC. Higher-grade features such as mitotic activity, necrosis, hypercellularity, and macronucleoli may be readily recognizable and should be mentioned but typically should not be used for definitive grading at the time of IOC unless the features are unequivocal. Close evaluation for brain invasion is helpful as this warrants a CNS WHO grade of 2 and may alter surgical management.³  Molecular testing for TERT promoter mutation and CDKN2A/B homozygous deletion can be anticipated in cases with borderline features, as these biomarkers have been shown to harbor a worse prognosis and warrant a CNS WHO grade of 3.^41,42  Preservation of tissue is usually not of concern in meningiomas, but small specimens may result in difficult-to-resect locations.

A wide variety of tumors may present as poorly differentiated neoplasms with a poorly cohesive appearance on smear preparation and cells with high nuclear to cytoplasmic ratios and no appreciable diagnostic evidence of lineage (Figure 2, f). These tumors are commonly referred to as “small round blue cell tumors” or “small blue cell tumors,” but it may be helpful to provide a differential diagnosis as the significance of this term may not necessarily be evident to the neurosurgeon. If malignant features are appreciated, such as abundant mitotic activity, then a diagnosis of “high-grade neoplasm” or “malignant neoplasm” should be considered. Age and location may permit favoring a certain neoplasm, such as a germ cell tumor (sellar or pineal mass in a pediatric patient or young adult) or medulloblastoma (posterior fossa mass). Other considerations include high-grade glioma with primitive neuronal or small-cell change, CNS embryonal tumors (such as atypical teratoid/rhabdoid tumor), hematolymphoid neoplasms, metastatic neuroendocrine (small-cell) carcinoma, metastatic melanoma, or sarcomas. Depending on the clinical scenario, an extensive immunohistochemical workup and potential molecular testing, including DNA methylation profiling, should be anticipated.

A basic familiarity with common molecular assays can help anticipate triaging of tissue at the time of frozen section assessment. Robust immunohistochemical markers exist for many molecular alterations, including antibodies specific to mutant proteins for IDH1 R132H, BRAF V600E, H3K27M, and H3G34R/V mutations, as well as antibodies to p53, ATRX, and H3K27me3, and in such cases a diagnosis may be made on small amounts of tissue. In some specific settings, such as a high suspicion for a diffuse midline glioma or an IDH-mutant tumor, a small number of immunohistochemical stains may be capable of establishing a diagnosis. A few previously discussed chromosomal abnormalities, particularly 1p/19q codeletion and CDKN2A/B homozygous deletion, or concurrent gain of chromosome 7 and loss of chromosome 10, are important in the classification of glial neoplasms. These can be evaluated by fluorescence in situ hybridization testing, but may also be evaluated by chromosomal microarray analysis, which typically requires more tissue. Most DNA sequencing alterations can be evaluated by NGS or polymerase chain reaction, and these tests may offer superior sensitivity in small amounts of tissue. For pediatric patients or tumors with characteristic gene fusions, RNA extraction for analysis may also be necessary. Lastly, DNA methylation profiling has become an additional ancillary testing method to assist in diagnosing a variety of tumors and is required by 2021 CNS WHO criteria for some entities.³  General suggestions for DNA methylation profiling suggest that up to 70% of tumor cell content is appropriate for testing. In general, more complex and extensive molecular testing should be anticipated for tumors in pediatric patients and poorly differentiated tumors.

Telepathology has been utilized for several decades to aid in intraoperative diagnosis of neurosurgical specimens. A few features render neuropathology uniquely posed to utilize telepathology: the tradition of having dedicated neuropathologists, when available, to review neuropathology specimens; a relatively low case volume, making it unfeasible to always have a dedicated neuropathologist available; and a relative lack of gross complexity of specimens, rendering techniques such as on-site gross assessment and margin evaluation less important. Multiple modalities have been used for intraoperative evaluations in teleneuropathology, with rapid whole-slide imaging and robotic microscopy being the 2 most well-characterized methods. In the setting of general IOC, both modalities have shown high concordance rates with in-person diagnosis.⁴³  These findings have been replicated in the setting of neuropathology intraoperative diagnoses. Michigan Medicine implemented teleneuropathology for neuropathology intraoperative diagnoses in 2016 with success and good concordance with in-person diagnosis.

The prevalence of telepathology for neuropathology intraoperative diagnosis will likely continue to grow. The move toward more extensive implementation of these modalities was accelerated by the COVID-19 pandemic.⁴⁴  As neuropathologic diagnosis becomes more complex, teleneuropathology offers better access to neuropathologic expertise during IOC. However, as described in the current review, the complexity of molecular diagnosis in neuropathology may have paradoxically rendered intraoperative diagnosis simpler, as fewer specific diagnosis can be made at the time of surgery. Therefore, intraoperative neuropathologic diagnosis may be as accessible to the general surgical pathologist as ever.

Other novel modalities have yet to become widely adopted in clinical practice. Some institutions have the capacity to perform “rapid” molecular testing at the time of IOC to determine features such as IDH mutational status at the time of IOC.^45–47  While this may not have clear clinical implications currently, the development of more targeted therapies may make such testing more useful in coming years and may also aid in clinical trial enrollment. Stimulated Raman histology is an optical imaging method that can rapidly produce digital images of fresh tissue for review within minutes. It has been proven that interpretation of these digital images is feasible by pathologists and produces diagnostic concordance similar to that of whole-slide imaging.^48,49  In addition to rapid imaging, such methods offer the advantage of not consuming the tissue and not having any freezing artifact. Since the tissue is not consumed, it can later be used for additional ancillary testing. Lastly, techniques of artificial intelligence and machine learning have been shown to perform well when compared to neuropathologists in the setting of intraoperative diagnosis, but are as yet limited by, among other factors, a restriction to trained diagnostic categories and challenges in appropriately classifying rare and unusual neoplasms.^50,51  While these modalities remain in relative infancy, they may augment neuropathologic diagnosis in coming years.

The field of intraoperative neurosurgical pathology continues to evolve. While the subclassification of tumors of the CNS has become substantially more complex, the basic approach to IOC has not substantially changed in the last century. However, the move from diagnoses based solely on histology to diagnoses aided by immunohistochemistry and molecular testing has shifted the priorities and terminology used at the time of IOC. Having a standard approach to identifying diagnostic or lesional tissue remains of greatest importance. Once diagnostic tissue is established, the priority shifts to establishing a general diagnostic category with an eye toward appropriately triaging tissue for potential workup and classification. The use of telepathology, artificial intelligence, and other intraoperative diagnostic techniques will continue to shape the field in the coming decades.