- Email: [email protected]
Frozen section discrepancy in the evaluation of nonneoplastic central nervous system samples
Available online at www.sciencedirect.com
Annals of Diagnostic Pathology 13 (2009) 359 – 366
Original Contributions
Frozen section discrepancy in the evaluation of nonneoplastic central nervous system samples Thomas P. Plesec, MD, Richard A. Prayson, MD⁎ Cleveland Clinic Lerner College of Medicine, Department of Anatomic Pathology, Cleveland Clinic, Cleveland, Ohio
Abstract
Keywords:
Frozen section (FS) for intraoperative evaluation of central nervous system (CNS) lesions provides the neurosurgeon with a rapid preliminary pathologic diagnosis. Diagnosis of nonneoplastic lesions is particularly challenging in this venue. To highlight common diagnostic pitfalls, we sought to identify discrepancies between FS and final diagnoses among nonneoplastic CNS samples via a retrospective review of 303 FS cases encountered from 1997 to 2006. Thirty-nine (12.9%) discrepant diagnoses were identified, of which 27 were clinically suspected tumors. Final diagnoses in the discrepant group included the following: inflammatory lesions (n = 8, 20.5%), malformation of cortical development-cortical dysplasia (n = 5, 12.8%), gliosis (n = 5, 12.8%), vascular malformations (n = 5, 12.8%), demyelination/progressive multifocal leukoencephalopathy (n = 3, 7.7%), infarct (n = 3, 7.7%), hemorrhage/blood clot (n = 3, 7.7%), and no pathologic changes (n = 3, 7.7%). The remaining 4 (10.2%) discrepant cases involved one case each of amyloid angiopathy, nonspecific vasculopathy, vasculitis, and meningioangiomatosis. Nonneoplastic lesions are often more challenging than neoplastic lesions at FS, particularly because they are less commonly sampled for FS and, therefore, less familiar to pathologists. © 2009 Elsevier Inc. All rights reserved. Frozen section; Nonneoplastic; Intraoperative consultation; Discrepant diagnosis; Central nervous system
1. Introduction Central nervous system (CNS) frozen section (FS) consultation specimens usually target radiographically evident lesions; frequently, a neoplastic process is suspected. The diagnostic yield [1-3] and rate of concordance between FS and permanent diagnoses [1,4-7] have been reported to be consistently greater than 90%. In contrast, most series reporting on nonneoplastic CNS pathologic condition are much smaller with widely variable diagnostic yields that range from 20% to 84% [3,8-11]. The FS discrepancy rate of nonneoplastic lesions is even less established. Brainard et al [9] found 16 (34.8%) of 46 of nonneoplastic cases to be either nondiagnostic (n = 14) or misdiagnosed (n = 2) at the time of FS. One cytology-based study [12] showed a ⁎ Corresponding author. Cleveland Clinic, Department of Anatomic Pathology, L25, Cleveland, Ohio 44195. Tel.: +1 216 444 8805; fax: +1 216 444 6967. E-mail address: [email protected] (R.A. Prayson). 1092-9134/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.anndiagpath.2009.05.006
complete concordance rate between intraoperative consultation and final diagnosis of 37.5% and partial rate of 78% in their series of 112 nonneoplastic cases. These relatively small series suggest that the FS to final diagnosis concordance rate is much lower than similar tumor-based studies. Neither series, however, attempted to categorize or classify the discrepancies. Herein, we present a large single-institution experience of nonneoplastic cases sent for FS consultation. Examination of discrepancies between the FS and final permanent diagnoses adds to the current understanding of the particular challenges that nonneoplastic CNS lesions can present at the time of FS.
2. Materials and methods Upon receiving institutional review board approval, a search was performed to retrospectively identify all nonneoplastic CNS FS cases encountered during a 10-year period. The 303 cases identified were determined
360
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
by the final diagnosis, and an FS neoplastic diagnosis did not preclude inclusion. The FS diagnoses were compared to the final diagnoses, and discrepancies were identified. Very few intraoperative cytologic preparations were performed during the study period, and the interpretations of these procedures were not included in the analysis. In effort to identify sampling differences between the discrepant and nondiscrepant groups, the number of specimens submitted for FS, total specimens submitted, FS slides examined, and total slides examined were tabulated. Statistical comparisons were performed using the Wilcoxon 2-sample test and χ2 test, and a P value of .05 or less was considered statistically significant.
3. Results Thirty-nine (12.9%) discrepant diagnoses were identified of 303 total nonneoplastic diagnoses; patients with these discrepant diagnoses constituted the study group. The mean age of the patients in the study group (22 females, 17 males) was 37.6 years (range, 2-73 years). The patients within the nondiscrepant group (126 females, 138 males) had a mean age of 46.3 years (range, 1-87 years). The discrepant group had a mean of 1.4 FS specimens, 1.5 FS slides, 2.5 total specimens, and 5.4 permanent slides evaluated, and the nondiscrepant group had a mean of 1.6 FS specimens, 1.8 FS slides, 2.3 total specimens, and 4.9 permanent slides evaluated. Although the nondiscrepant group tended to have more FS specimens and slides processed with fewer permanent specimens and slides, none of these differences were statistically significant. The proportion of stereotactic biopsies also did not differ significantly, with 25 (64.1%) of 39 of patients in the discrepant group and 180 (68.2%) of 264 undergoing stereotactic biopsy. All but 2 cases within the discrepant group were examined by a neuropathologist. In the nondiscrepant group, all but 4 cases were seen by a neuropathologist at FS with the final diagnosis rendered by a neuropathologist in all but 2 cases. Final diagnoses in the discrepant case group included the following (Table 1): inflammatory lesions (n = 8; Figs. 1A, B), malformations of cortical development (n = 5; Figs. 1C, D), gliosis (n = 5; Figs. 2A-B), vascular malformations (n = 5; Figs. 2C-D), demyelination/progressive multifocal leukoencephalopathy (n = 3; Figs. 3A-D), infarct (n = 3; Figs. 4A-D), hemorrhage/blood clot (n = 3), and no pathologic changes (n = 3). The remaining 4 discrepant cases involved one case each of amyloid angiopathy, nonspecific vasculopathy, vasculitis, and meningioangiomatosis. Discrepant FS diagnoses were grouped into 3 categories (Table 1): nonspecific findings (n = 23), tumor/suggestion of tumor (n = 11), and misclassified benign lesions (n = 5). Inflammatory lesions (n = 6), vascular malformations (n = 4), and malformations of cortical development (n = 3) were the most frequent final diagnoses in the nonspecific FS diagnosis category. Of the 11 cases diagnosed as or suggestive of a
tumor, gliosis (n = 4) and malformation of cortical development (n = 2) were the most frequent final diagnoses. One of the 11 patients underwent biopsy 7 months later, and this was diagnostic of low-grade glioma. Seven of the remaining 10 patients were observed for 31 to 103 months and have not manifested a CNS neoplasm. Two of the remaining 10 patients were immediately lost to follow-up, and 1 died within days of admission from a myocardial infarct. The 5 misclassified benign lesions most commonly revealed no pathologic abnormality (n = 2). Table 2 summarizes the preoperative clinical impressions for each discrepant category. Thirty-five (90%) cases targeted one or more radiographically evident lesions. Of these, 27 surgeries were performed for suspected tumors, and 8 surgeries were aimed at suspected nonneoplastic processes, mostly vascular malformations. The remaining 4 surgeries were performed to rule out either meningitis or vasculitis.
4. Discussion Classifying the discrepant FS diagnoses into 3 broad categories emphasizes important factors that can contribute to FS discrepancy in neuropathologic cases—sampling, bias, and misinterpretation. Given the discrepancy rate of nearly 5 times that of neoplastic CNS lesions reported from our own institution (12.9% vs 2.7%) [13], these factors may present more of a pitfall when confronted with nonneoplastic lesions. Many of the 23 cases marked by nonspecific findings at FS (56% of discrepant cases) illustrate the sampling problems inherent to FS. Diagnostic yields for nonneoplastic CNS pathologic condition are reported to range from 20% to 84% [3,8-11]. Although most FS (90%) were targeted at specific lesions, a subset represented diffuse or potentially multifocal processes, which often pose even greater diagnostic difficulty at FS. The surgeon also may be occasionally off target, resulting in nondiagnostic tissue. It is also well recognized that additional deeper sections generated from the same paraffin blocks may uncover pathologic condition not evident on the first tissue sections. In addition, not all available tissue sent for FS is subjected to FS processing, as it is advisable to reserve some unaltered tissue for paraffin embedding. This ensures good quality well-fixed specimens for evaluation and final diagnosis but does potentially contribute to sampling error. Similarly, not all sampled tissue is designated for FS by the surgeon. Sampling error likely played a key role in the final diagnosis group of inflammatory lesions, which were all signed out with nonspecific FS diagnoses. Although not inconsistent with the final diagnoses, key features were not identified at the time of FS, such as chronic inflammation or granulomas. The 11 cases diagnosed as tumors or suggestive of tumor (30.8% of discrepant cases) expose the possibilities of bias and misinterpretation in FS diagnosis. One of the tenets of surgical neuropathologic condition is to correlate the
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
361
Table 1 Summary of discrepant diagnoses Permanent diagnosis Inflammation Perivascular chronic inflammation Encephalitis consistent with Rasmussen's Perivascular chronic inflammation and microglial cell proliferation Perivascular, infiltrating lymphocytes, and gliosis Perivascular chronic inflammation and gliosis Granulomatous inflammation with gliosis Granulomatous inflammation Granulomatous inflammation Malformations of cortical development Cortical dysplasia Changes suggestive of cortical dysplasia Cortical dysplasia Suggestive of cortical dysplasia Cortical dysplasia Gliosis Focally hypercellular tissue Gliosis Hypercellular parenchyma with atypical astrocytes Hypercellular white matter with reactive astrocytes Mildly hypercellular white matter Vascular malformation Cavernous angioma Benign vascular malformation Benign vascular malformation Hemangioma Favor cavernous angioma Demyelination/PML Consistent with PML Demyelinating disease Suggestive of demyelinating disease Infarct Consistent with subacute to remote infarction Suggestive of resolving infarction Small focus of infarcted neural parenchyma Hemorrhage Acute blood clot and thrombus Hemorrhage with organization and necrosis Blood, fibrin, histiocytes, and hemosiderin No abnormality Benign epithelium and fibrovascular tissue Minute CNS tissue with no abnormality Benign choroid plexus and cerebellum Other Amyloid angiopathy Meningioangiomatosis Vasculitis with focal subacute infarction Focal necrosis with vascular sclerosis
Frozen diagnosis
Discrepancy
Cortex with focal acute ischemic changes Gliosis, no encephalitis seen Gray matter with no definite abnormality Reactive astrocytosis Gliosis Lymphocytic infiltrate with gliosis Leptomeningeal chronic inflammation Normal dura and cerebellum
Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific
Rare atypical astrocytes Gliosis No evidence of neoplasm No evidence of tumor Low-grade glioma
Tumor Nonspecific Nonspecific Nonspecific Tumor
Favor low-grade glioma Suggestive of vascular malformation Low-grade glioma Favor low-grade glioma Low-grade astrocytoma
Tumor Misclassification Tumor Tumor Tumor
Organizing hematoma Favor inflammatory process Chronic inflammation with hemorrhage Blood vessel with clot Benign fibrovascular tissue
Nonspecific Misclassification Nonspecific Nonspecific Nonspecific
Reactive astrocytosis Favor astrocytic neoplasm Reactive astrocytes and macrophages
Nonspecific Tumor Nonspecific
Necrosis, gliosis, focus of atypical cells Gliosis Blood clot
Tumor Nonspecific Nonspecific
Gliosis and perivascular hypercellularity Contents of vascular malformation Necrosis, probably necrotic tumor
Nonspecific Nonspecific Tumor
Consistent with encephalocele Gliosis Low-grade neuroectodermal neoplasm
Misclassification Misclassification Tumor
Blood clot and rare blood vessel Low-grade glioneuronal neoplasm No specific pathologic condition Amorphous material
Nonspecific Tumor Nonspecific Nonspecific
PML indicates progressive multifocal leukoencephalopathy. Nonspecific discrepancies include those cases that received a descriptive diagnosis at FS without commitment to a specific process and those requiring ancillary testing (eg, demyelinating disease or amyloid angiopathy). These errors were likely sampling related. Tumor discrepancies include those cases diagnosed as neoplastic at FS but were considered nonneoplastic upon review of permanent sections. These errors were likely misinterpretations. Misclassification discrepancies include one nonneoplastic process misdiagnosed as another nonneoplastic process and normal tissue misdiagnosed as abnormal. The errors were likely misinterpretations.
pathologic findings with clinical and radiologic information; however, this can be a source of error if the clinical impression unduly sways the pathologic interpretation. Of the 11 cases in this category, 10 (90.9%) had a preoperative impression of tumor. In addition, 4 of 5 cases with a final diagnosis of gliosis were diagnosed as tumor at the time of FS. This underscores the challenge of the gliosis vs glioma
differential diagnosis and the caution one needs to exercise to avoid overdiagnosis. It is helpful to remember that gliosis tends to demonstrate evenly distributed cellularity with low nuclear-to-cytoplasmic (N/C) ratios and relatively uniform nuclear features, whereas gliomas are marked by an unevenly distributed and increased cellularity, high N/C ratios, nuclear angularity and enlargement, and hyperchromasia [14].
362
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
Fig. 1. (A-C) Thirty-one year old male with diffuse lemptomeningeal enhancement. (A, B) FS: chronic leptomeningeal inflammation. (C) Permanents: nonnecrotizing granulomas; the patient was diagnosed with neurosarcoidosis. The error was likely due to sampling at FS. (D-F) Twenty-three year old female with imaging suggestive of low-grade glioma. (D, E) FS: hypercellular white matter with hyperchromatic, angulated glial nuclei, consistent with low-grade glioma. (F) Permanents: disordered neuronal architecture with dysmorphic neurons and balloon cells, consistent with cortical dysplasia. The error was likely due to misinterpretation.
Fig. 2. (A-C) Forty-four year old female with imaging suggestive of a low-grade glioma. (A, B) FS: hyperchromatic, angulated glial nuclei, consistent with lowgrade glioma. (C) Permanents: evenly spaced, regular nuclei without significant atypia, and gliosis was diagnosed. The error was likely due to misinterpretation. (D-F) Twelve year old female with heterogeneous left frontal mass. (D, E) FS: abundant hemosiderin fibrin, and numerous capillary-type vessels, consistent with organizing hemorrhage/thrombus. (F) Permanents: anastomosing thin- and thick-walled blood vessels without intervening neural parenchyma, changes consistent with cavernous angioma. The error was likely due to sampling.
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
363
Fig. 3. A 45-year-old male with lesions suggestive of lymphoma or multifocal glioma. (A, B) FS: atypical astrocytes, favor astrocytoma. Permanents: foamy macrophages (C). Myelin disruption on Luxol fast blue stain (D), and preservation of axons on Bodian stain (E), diagnostic of demyelinating disease. The patient was diagnosed with multiple sclerosis. The error was likely due to misinterpretation.
Malformation of cortical development (MCD) or cortical dysplasia also may pose difficulty in the tumor vs nontumor differential diagnosis, and 2 of 5 cases were misinterpreted as neoplastic. MCDs most commonly are examined by FS in patients with intractable seizures [15]. The clinical history can be misleading in that tumors are also a well-recognized cause of chronic epilepsy, and some forms of MCD can appear radiographically as a localized lesion [16,17]. MCDs represent a histologic spectrum from mild architectural abnormalities to hypercellular lesions marked by atypical cells resembling such tumors as ganglioglioma. In addition, gangliogliomas and dysembryoplastic neuroepithelial tumor
may coexist with MCD. Because some of these lesions may be characterized by subtle architectural alterations, a fragmented specimen or biopsy may be insufficient to make a diagnosis, particularly at FS. Similarly, a limited sample of certain MCDs at FS, particularly those marked by dysmorphic neurons and balloon cells, may be difficult to distinguish from a ganglioglioma. It has been our experience that most cases of MCD are not reliably diagnosed at FS but require more extensive sampling with properly oriented sections for definitive diagnosis. Although the typical demyelinating disease, such as multiple sclerosis, usually presents with multifocal white
Fig. 4. A 36-year-old female with a left frontal lobe lesion. (A, B) FS: reactive astrocytosis. Permanents: macrophages and reactive astrocytes (C), myelin disruption on Luxol fast blue stain (D), and nearly absent axons on Bodain stain (E), features of infarct. The error was likely due to misinterpretation.
364
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
Table 2 Categories of discrepancies in relation to clinical suspicion Categories of discrepancies
Inflammation (n = 8) Malformation of cortical development (n = 5) Gliosis (n = 5) Vascular malformation (n = 5) Demyelination/PML (n = 3) Infarct (n = 3) Hemorrhage (n = 3) No abnormality (n = 3) Amyloid angiopathy (n = 1) Vasculopathy (n = 1) Vasculitis (n = 1) Meningioangiomatosis (n = 1)
Clinical suspicion Tumor (n = 27)
Nonneoplastic (n = 8)
No discrete lesion (n = 4)
4 5
1
3
4 3 3 2 2 2
1 2 1 1 1 1
1 1 1
PML indicates progressive multifocal leukoencephalopathy.
matter lesions on imaging, an occasional patient may present with a solitary lesion, resembling a tumor on radiographic studies [18-21]. These atypical, “tumorlike” presentations often lead to stereotactic biopsy and difficulty for the pathologist [17]. The usual demyelinating lesion consists of macrophages, reactive astrocytes, and benign perivascular lymphocytes. It is critical to not to confuse macrophages with malignant cells; cytologic preparations may help in identifying the characteristic cell borders, uniform nuclei, and abundant foamy cytoplasm of macrophages [16,22]. Demyelinating lesions predominantly are restricted to the white matter, lack necrosis, and demonstrate reactive astrocytes throughout the lesion—all features that can help differentiate these lesions from infarct. Diligent search for viral inclusions is important in recognizing progressive multifocal leukoencephalopathy and subacute sclerosing panencephalitis—2 important infectious causes of demyelination. Progressive multifocal leukoencephalopathy typically presents in immunocompromised hosts with multiple asymmetrically distributed white matter lesions and is histologically characterized by demyelination with dense “ground glass” intranuclear inclusions within oligodendroglial cells [16,23]. Subacute sclerosing panencephalitis usually presents in the first 2 decades of life, often with a history of antecedent measles infection. Intranuclear inclusions can be identified within oligodendroglial and neuronal nuclei [24]. Interestingly, 15 of 39, or 38%, discrepant cases were vascular related. These are common causes of nonneoplastic CNS pathologic condition, as about 1 of 4 of all the diagnoses rendered in this study were vascular related (data not shown). In addition, the discrepancy rate of vascular lesions (15/39, 38%) was 1.6 times that of the rate nondiscrepant vascular lesions (62/264, 23%), suggesting that these are disproportionately difficult. Infarcts may involve the white matter, gray matter, or both. Infarcts, in addition to demyelinating lesions, are an
important diagnosis to consider when one encounters numerous tissue macrophages. The macrophages may be seen as early as 24 hours after the infarct and may persist for months [17,25]. Again, it is critical not to confuse macrophages with neoplastic cells. In the absence of prior radiotherapy, macrophages are not commonly identified in most malignancies. Touch/squash preparations may be particularly useful in this setting [16]. Central necrosis and eventual cavitation with peripheral gliosis are the key features of infarct and helpful in distinguishing it from a demyelinating lesion. In addition, infarcts demonstrate acidophilic neurons, characterized by densely eosinophilic cytoplasm and pyknotic nuclei, so-called red and dead neurons, and less chronic inflammation than demyelinating lesions [16,17]. When confronted with increased numbers of blood vessels at FS, diagnostic considerations include granulation tissue, tumors, or vascular malformation [26]. Granulation tissue is composed of small capillary-type vessels embedded in an inflamed stroma. Increased vascularity is often seen in primary tumors such as fibrillary and pilocytic astrocytomas, hemangioblastoma, and meningioma as well as certain metastatic lesions, including renal cell carcinoma and melanoma. Once the neoplastic diagnostic possibilities have been ruled out, one must consider the possibility of vascular malformation. Arteriovenous malformations and cavernous angiomas are the most often excised CNS vascular malformations. Arteriovenous malformations have a relatively high risk of hemorrhage [17,27], and arteriovenous malformations are distinguished from all other malformations by an arterial component admixed with both thin-walled and thick-walled veins and intervening neural parenchyma. Cavernous angiomas are composed of a collection of thinwalled blood vessels and never display arteries or intervening neural parenchyma [26]. The final 4 discrepant cases also represent examples of vascular-related pathologic conditions. Meningioangiomatosis is a rare hamartomatous lesion that histologically consists of an assortment of small vessels surrounded by spindled meningothelial cells with gliotic intervening neural parenchyma. They tend to present as plaquelike lesions and have a typically benign course [28]. If one focuses on the intervening gliotic parenchyma and considers the vascular proliferation secondary, the lesion may be misinterpreted as low-grade glioneuronal or glial neoplasm. Most cases of abnormally thickened vessel walls are related to hypertension and atherosclerotic changes. Radiation is another important cause and often induces sclerosis of the vessels in addition to parenchymal necrosis and dramatic cytologic atypia. Although these histologic changes can be quite suggestive of radiation injury, it is always prudent to correlate the morphologic findings with the clinical history. Amyloid angiopathy is a relatively common cause of normotensive intracerebral hemorrhage and accounts for 5% to 20% of spontaneous cerebral hemorrhages in the elderly
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
[29]. Although one must exert caution at diagnosing amyloid at FS without the aid of special stains, abnormally thickened blood vessels with amorphous eosinophilic material certainly raises the possibility, and this can be communicated to the surgeon. Targeted biopsies to diagnose vasculitis are occasionally performed, as this remains the most reliable method of diagnosis [30]. Vasculitis is defined as inflammation of the vessel wall, often accompanied by vessel wall damage. Optimally, biopsies sample the dura, leptomeninges, cortex, and white matter [30]. The highest diagnostic yield is from the superficial cortex and leptomeninges; however, only about two thirds of cases are diagnostic [26]. Given the patchiness of the process, it is not surprising that a nondiagnostic FS may be accompanied by permanent sections that yield the diagnosis. One must always remember to consider lymphoma in the differential of vasculitis, as the pattern of lymphocytic infiltration may be similar. At the time of FS, this distinction is best made by scrutiny for cytologic atypia. The lymphocytes in vasculitis should be small, round, and regular, whereas malignant lymphoma cells are most commonly large and somewhat pleomorphic with prominent nucleoli [13]. We have presented a large series of nonneoplastic CNS FS discrepancies generated from one institution; the design, however, presented several limitations to the study. First, the single-institution patient population may not be representative of the public at-large, especially because most of the samples from this series were targeted at suspected tumors, and our institution had a large cohort of chronic epilepsyrelated cases. Second, we use very few intraoperative cytologic preparations, which may skew the types of discrepancies encountered. Third, most cases were examined by 1 of 3 neuropathologists, who may bring along their own personal biases. Fourth, the retrospective chart review focus on the discrepant diagnoses provided, in our opinion, more useful information for the practicing pathologist but prevented a more definitive assessment of the sources of error for each individual case and forced us to comment on error in more general terms. Nevertheless, this study is helpful to the practicing pathologist on several fronts. Often nonneoplastic FS diagnoses are nonspecific, which is most likely related to the inherent sampling difficulties encountered at FS. It is also important for the pathologist to remember that, occasionally, nonneoplastic disorders are confused with neoplasms preoperatively. If neoplastic tissue is not definitively present, then additional tissue should be requested, or the diagnosis should be deferred. In addition, several lesions presented recurring difficulties at FS, including gliosis, MCD, demyelinating diseases, and vascular-related lesions, all of which are important entities to consider when one is presented with a difficult nonneoplastic CNS FS. Open communication with the neurosurgeon is especially important in evaluating these difficult neuropathologic FS cases.
365
References [1] Colbassani HJ, Nishio S, Sweeney KM, et al. CT-assisted stereotactic brain biopsy: value of intraoperative frozen section diagnosis. J Neurol Neurosurg Psychiatry 1988;51:332-41. [2] Taratuto AL, Sevlever G, Piccardo P. Clues and pitfalls in stereotactic biopsy of the central nervous system. Arch Pathol Lab Med 1991;115: 596-602. [3] Gomez H, Barnett GH, Estes ML, et al. Stereotactic and computerassisted neurosurgery at the Cleveland Clinic: review of 501 consecutive cases. Cleve Clin J Med 1993;60:399-410. [4] Martinez AJ, Pollack I, Hall WA, et al. Touch preparations in the rapid intraoperative diagnosis of central nervous system lesions. A comparison with frozen sections and paraffin-embedded sections. Mod Pathol 1988;1:378-84. [5] Di Stefano D, Scucchi LF, Cosentino L, et al. Intraoperative diagnosis of nervous system lesions. Acta Cytol 1998;42:346-56. [6] Shah AB, Muzumdar GA, Chitale AR, et al. Squash preparation and frozen section in intraoperative diagnosis of central nervous system tumors. Acta Cytol 1998;42:1149-54. [7] Savargaonkar P, Farmer PM. Utility of intra-operative consultations for the diagnosis of central nervous system lesions. Ann Clin Lab Sci 2001;31:133-9. [8] Whiting DM, Barnett GH, Estes ML, et al. Sterotactic biopsy of nonneoplastic lesions in adults. Cleve Clin J Med 1992;59:48-55. [9] Brainard JA, Prayson RA, Barnett GH. Frozen section evaluation of stereotactic brain biopsies: diagnostic yield at the stereotactic target position in 188 cases. Arch Pathol Lab Med 1997;121: 481-4. [10] Javedan SP, Tamargo RJ. Diagnostic yield of brain biopsy in neurodegenerative disorders. Neurosurgery 1997;41:823-8. [11] Pulhorn H, Quigley DG, Bosma JJ, et al. Impact of brain biopsy on the management of patients with nonneoplastic undiagnosed neurological disorders. Neurosurgery 2008;62:833-7. [12] Firlik KS, Martinez AJ, Lunsford LD. Use of cytological preparations for the intraoperative diagnosis of sterotactically obtained brain biopsies: a 19-year experience and survey of neuropathologists. J Neursurg 1999;91:454-8. [13] Plesec TP, Prason RA. Frozen section discrepancy in the evaluation of central nervous system tumors. Arch Pathol Lab Med 2007;131: 1532-40. [14] Burger PC, Vogel FS. Frozen section interpretation in surgical neuropathology. I. Intracranial lesions. Am J Surg Pathol 1977;1: 323-47. [15] Sisodiya SM. Surgery for malformations of cortical development causing epilepsy. Brain 2000;123:1075-91. [16] Berger PC, Scheithauer B, Vogel FS, editors. Surgical pathology of the nervous system and its coverings. 4th Ed. New York: Churchill Livingstone; 2002. p. 379-89, 414-7. [17] Love S, editor. Greenfield's neuropathology, 8th Ed.London, England: Hodder Arnold; 2007. [18] Kepes JJ. Large focal tumor-like demyelinating lesions of the brain: intermediate entity between multiple sclerosis and acute disseminated encephalomyelitis? A study of 31 patients. Ann Neurol 1993;33: 18-27. [19] Capello E, Roccatagliata L, Pagano F, et al. Tumor-like multiple sclerosis (MS) lesions: neuropathological clues. Neurol Sci 2001;22 (Suppl 2):S113-6. [20] Heyman D, Delhaye M, Fournier D, et al. Pseudotumoral demyelination: a diagnosis pitfall (report of three cases). J Neurooncol 2001;54: 71-6. [21] Love S. Demyelinating diseases. J Clin Pathol 2006;59:1151-9. [22] Burger PC. Use of cytological preparations in the frozen section diagnosis of central nervous system neoplasia. Am J Surg Pathol 1985; 9:344-54. [23] Berger JR. Progressive multifocal leukoencephalopathy. Curr Neurol Neurosci Rep 2007;7:461-9.
366
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
[24] Garg RK. Subacute sclerosing panencephalitis. Postgrad Med J 2002; 78:63-70. [25] Prayson RA, editor. Neuropathology. Philadelphia: Elsevier ChurchillLivingstone; 2005. p. 37-72. [26] Prayson RA, Kleinschmidt-DeMasters BK. An algorithmic approach to the brain biopsy—part II. Arch Pathol Lab Med 2006;130:1639-48. [27] Brown Jr RD. Unruptured brain AVMs: to treat or not to treat. Lancet Neurol 2008;7:195-6.
[28] Perry A, Kurtkaya-Yapicier O, Scheithauer BW, et al. Insights into meningioangiomatosis with and without meningioma: a clinicopathologic and genetic series of 24 cases with review of the literature. Brain Pathol 2005;15:55-65. [29] Maia LF, Mackenzie IR, Feldman HH. Clinical phenotypes of cerebral amyloid angiopathy. J Neurol Sci 2007;257:23-30. [30] Schmidley JW. 10 questions on central nervous system vasculitis. Neurologist 2008;14:138-9.
Annals of Diagnostic Pathology 13 (2009) 359 – 366
Original Contributions
Frozen section discrepancy in the evaluation of nonneoplastic central nervous system samples Thomas P. Plesec, MD, Richard A. Prayson, MD⁎ Cleveland Clinic Lerner College of Medicine, Department of Anatomic Pathology, Cleveland Clinic, Cleveland, Ohio
Abstract
Keywords:
Frozen section (FS) for intraoperative evaluation of central nervous system (CNS) lesions provides the neurosurgeon with a rapid preliminary pathologic diagnosis. Diagnosis of nonneoplastic lesions is particularly challenging in this venue. To highlight common diagnostic pitfalls, we sought to identify discrepancies between FS and final diagnoses among nonneoplastic CNS samples via a retrospective review of 303 FS cases encountered from 1997 to 2006. Thirty-nine (12.9%) discrepant diagnoses were identified, of which 27 were clinically suspected tumors. Final diagnoses in the discrepant group included the following: inflammatory lesions (n = 8, 20.5%), malformation of cortical development-cortical dysplasia (n = 5, 12.8%), gliosis (n = 5, 12.8%), vascular malformations (n = 5, 12.8%), demyelination/progressive multifocal leukoencephalopathy (n = 3, 7.7%), infarct (n = 3, 7.7%), hemorrhage/blood clot (n = 3, 7.7%), and no pathologic changes (n = 3, 7.7%). The remaining 4 (10.2%) discrepant cases involved one case each of amyloid angiopathy, nonspecific vasculopathy, vasculitis, and meningioangiomatosis. Nonneoplastic lesions are often more challenging than neoplastic lesions at FS, particularly because they are less commonly sampled for FS and, therefore, less familiar to pathologists. © 2009 Elsevier Inc. All rights reserved. Frozen section; Nonneoplastic; Intraoperative consultation; Discrepant diagnosis; Central nervous system
1. Introduction Central nervous system (CNS) frozen section (FS) consultation specimens usually target radiographically evident lesions; frequently, a neoplastic process is suspected. The diagnostic yield [1-3] and rate of concordance between FS and permanent diagnoses [1,4-7] have been reported to be consistently greater than 90%. In contrast, most series reporting on nonneoplastic CNS pathologic condition are much smaller with widely variable diagnostic yields that range from 20% to 84% [3,8-11]. The FS discrepancy rate of nonneoplastic lesions is even less established. Brainard et al [9] found 16 (34.8%) of 46 of nonneoplastic cases to be either nondiagnostic (n = 14) or misdiagnosed (n = 2) at the time of FS. One cytology-based study [12] showed a ⁎ Corresponding author. Cleveland Clinic, Department of Anatomic Pathology, L25, Cleveland, Ohio 44195. Tel.: +1 216 444 8805; fax: +1 216 444 6967. E-mail address: [email protected] (R.A. Prayson). 1092-9134/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.anndiagpath.2009.05.006
complete concordance rate between intraoperative consultation and final diagnosis of 37.5% and partial rate of 78% in their series of 112 nonneoplastic cases. These relatively small series suggest that the FS to final diagnosis concordance rate is much lower than similar tumor-based studies. Neither series, however, attempted to categorize or classify the discrepancies. Herein, we present a large single-institution experience of nonneoplastic cases sent for FS consultation. Examination of discrepancies between the FS and final permanent diagnoses adds to the current understanding of the particular challenges that nonneoplastic CNS lesions can present at the time of FS.
2. Materials and methods Upon receiving institutional review board approval, a search was performed to retrospectively identify all nonneoplastic CNS FS cases encountered during a 10-year period. The 303 cases identified were determined
360
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
by the final diagnosis, and an FS neoplastic diagnosis did not preclude inclusion. The FS diagnoses were compared to the final diagnoses, and discrepancies were identified. Very few intraoperative cytologic preparations were performed during the study period, and the interpretations of these procedures were not included in the analysis. In effort to identify sampling differences between the discrepant and nondiscrepant groups, the number of specimens submitted for FS, total specimens submitted, FS slides examined, and total slides examined were tabulated. Statistical comparisons were performed using the Wilcoxon 2-sample test and χ2 test, and a P value of .05 or less was considered statistically significant.
3. Results Thirty-nine (12.9%) discrepant diagnoses were identified of 303 total nonneoplastic diagnoses; patients with these discrepant diagnoses constituted the study group. The mean age of the patients in the study group (22 females, 17 males) was 37.6 years (range, 2-73 years). The patients within the nondiscrepant group (126 females, 138 males) had a mean age of 46.3 years (range, 1-87 years). The discrepant group had a mean of 1.4 FS specimens, 1.5 FS slides, 2.5 total specimens, and 5.4 permanent slides evaluated, and the nondiscrepant group had a mean of 1.6 FS specimens, 1.8 FS slides, 2.3 total specimens, and 4.9 permanent slides evaluated. Although the nondiscrepant group tended to have more FS specimens and slides processed with fewer permanent specimens and slides, none of these differences were statistically significant. The proportion of stereotactic biopsies also did not differ significantly, with 25 (64.1%) of 39 of patients in the discrepant group and 180 (68.2%) of 264 undergoing stereotactic biopsy. All but 2 cases within the discrepant group were examined by a neuropathologist. In the nondiscrepant group, all but 4 cases were seen by a neuropathologist at FS with the final diagnosis rendered by a neuropathologist in all but 2 cases. Final diagnoses in the discrepant case group included the following (Table 1): inflammatory lesions (n = 8; Figs. 1A, B), malformations of cortical development (n = 5; Figs. 1C, D), gliosis (n = 5; Figs. 2A-B), vascular malformations (n = 5; Figs. 2C-D), demyelination/progressive multifocal leukoencephalopathy (n = 3; Figs. 3A-D), infarct (n = 3; Figs. 4A-D), hemorrhage/blood clot (n = 3), and no pathologic changes (n = 3). The remaining 4 discrepant cases involved one case each of amyloid angiopathy, nonspecific vasculopathy, vasculitis, and meningioangiomatosis. Discrepant FS diagnoses were grouped into 3 categories (Table 1): nonspecific findings (n = 23), tumor/suggestion of tumor (n = 11), and misclassified benign lesions (n = 5). Inflammatory lesions (n = 6), vascular malformations (n = 4), and malformations of cortical development (n = 3) were the most frequent final diagnoses in the nonspecific FS diagnosis category. Of the 11 cases diagnosed as or suggestive of a
tumor, gliosis (n = 4) and malformation of cortical development (n = 2) were the most frequent final diagnoses. One of the 11 patients underwent biopsy 7 months later, and this was diagnostic of low-grade glioma. Seven of the remaining 10 patients were observed for 31 to 103 months and have not manifested a CNS neoplasm. Two of the remaining 10 patients were immediately lost to follow-up, and 1 died within days of admission from a myocardial infarct. The 5 misclassified benign lesions most commonly revealed no pathologic abnormality (n = 2). Table 2 summarizes the preoperative clinical impressions for each discrepant category. Thirty-five (90%) cases targeted one or more radiographically evident lesions. Of these, 27 surgeries were performed for suspected tumors, and 8 surgeries were aimed at suspected nonneoplastic processes, mostly vascular malformations. The remaining 4 surgeries were performed to rule out either meningitis or vasculitis.
4. Discussion Classifying the discrepant FS diagnoses into 3 broad categories emphasizes important factors that can contribute to FS discrepancy in neuropathologic cases—sampling, bias, and misinterpretation. Given the discrepancy rate of nearly 5 times that of neoplastic CNS lesions reported from our own institution (12.9% vs 2.7%) [13], these factors may present more of a pitfall when confronted with nonneoplastic lesions. Many of the 23 cases marked by nonspecific findings at FS (56% of discrepant cases) illustrate the sampling problems inherent to FS. Diagnostic yields for nonneoplastic CNS pathologic condition are reported to range from 20% to 84% [3,8-11]. Although most FS (90%) were targeted at specific lesions, a subset represented diffuse or potentially multifocal processes, which often pose even greater diagnostic difficulty at FS. The surgeon also may be occasionally off target, resulting in nondiagnostic tissue. It is also well recognized that additional deeper sections generated from the same paraffin blocks may uncover pathologic condition not evident on the first tissue sections. In addition, not all available tissue sent for FS is subjected to FS processing, as it is advisable to reserve some unaltered tissue for paraffin embedding. This ensures good quality well-fixed specimens for evaluation and final diagnosis but does potentially contribute to sampling error. Similarly, not all sampled tissue is designated for FS by the surgeon. Sampling error likely played a key role in the final diagnosis group of inflammatory lesions, which were all signed out with nonspecific FS diagnoses. Although not inconsistent with the final diagnoses, key features were not identified at the time of FS, such as chronic inflammation or granulomas. The 11 cases diagnosed as tumors or suggestive of tumor (30.8% of discrepant cases) expose the possibilities of bias and misinterpretation in FS diagnosis. One of the tenets of surgical neuropathologic condition is to correlate the
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
361
Table 1 Summary of discrepant diagnoses Permanent diagnosis Inflammation Perivascular chronic inflammation Encephalitis consistent with Rasmussen's Perivascular chronic inflammation and microglial cell proliferation Perivascular, infiltrating lymphocytes, and gliosis Perivascular chronic inflammation and gliosis Granulomatous inflammation with gliosis Granulomatous inflammation Granulomatous inflammation Malformations of cortical development Cortical dysplasia Changes suggestive of cortical dysplasia Cortical dysplasia Suggestive of cortical dysplasia Cortical dysplasia Gliosis Focally hypercellular tissue Gliosis Hypercellular parenchyma with atypical astrocytes Hypercellular white matter with reactive astrocytes Mildly hypercellular white matter Vascular malformation Cavernous angioma Benign vascular malformation Benign vascular malformation Hemangioma Favor cavernous angioma Demyelination/PML Consistent with PML Demyelinating disease Suggestive of demyelinating disease Infarct Consistent with subacute to remote infarction Suggestive of resolving infarction Small focus of infarcted neural parenchyma Hemorrhage Acute blood clot and thrombus Hemorrhage with organization and necrosis Blood, fibrin, histiocytes, and hemosiderin No abnormality Benign epithelium and fibrovascular tissue Minute CNS tissue with no abnormality Benign choroid plexus and cerebellum Other Amyloid angiopathy Meningioangiomatosis Vasculitis with focal subacute infarction Focal necrosis with vascular sclerosis
Frozen diagnosis
Discrepancy
Cortex with focal acute ischemic changes Gliosis, no encephalitis seen Gray matter with no definite abnormality Reactive astrocytosis Gliosis Lymphocytic infiltrate with gliosis Leptomeningeal chronic inflammation Normal dura and cerebellum
Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific
Rare atypical astrocytes Gliosis No evidence of neoplasm No evidence of tumor Low-grade glioma
Tumor Nonspecific Nonspecific Nonspecific Tumor
Favor low-grade glioma Suggestive of vascular malformation Low-grade glioma Favor low-grade glioma Low-grade astrocytoma
Tumor Misclassification Tumor Tumor Tumor
Organizing hematoma Favor inflammatory process Chronic inflammation with hemorrhage Blood vessel with clot Benign fibrovascular tissue
Nonspecific Misclassification Nonspecific Nonspecific Nonspecific
Reactive astrocytosis Favor astrocytic neoplasm Reactive astrocytes and macrophages
Nonspecific Tumor Nonspecific
Necrosis, gliosis, focus of atypical cells Gliosis Blood clot
Tumor Nonspecific Nonspecific
Gliosis and perivascular hypercellularity Contents of vascular malformation Necrosis, probably necrotic tumor
Nonspecific Nonspecific Tumor
Consistent with encephalocele Gliosis Low-grade neuroectodermal neoplasm
Misclassification Misclassification Tumor
Blood clot and rare blood vessel Low-grade glioneuronal neoplasm No specific pathologic condition Amorphous material
Nonspecific Tumor Nonspecific Nonspecific
PML indicates progressive multifocal leukoencephalopathy. Nonspecific discrepancies include those cases that received a descriptive diagnosis at FS without commitment to a specific process and those requiring ancillary testing (eg, demyelinating disease or amyloid angiopathy). These errors were likely sampling related. Tumor discrepancies include those cases diagnosed as neoplastic at FS but were considered nonneoplastic upon review of permanent sections. These errors were likely misinterpretations. Misclassification discrepancies include one nonneoplastic process misdiagnosed as another nonneoplastic process and normal tissue misdiagnosed as abnormal. The errors were likely misinterpretations.
pathologic findings with clinical and radiologic information; however, this can be a source of error if the clinical impression unduly sways the pathologic interpretation. Of the 11 cases in this category, 10 (90.9%) had a preoperative impression of tumor. In addition, 4 of 5 cases with a final diagnosis of gliosis were diagnosed as tumor at the time of FS. This underscores the challenge of the gliosis vs glioma
differential diagnosis and the caution one needs to exercise to avoid overdiagnosis. It is helpful to remember that gliosis tends to demonstrate evenly distributed cellularity with low nuclear-to-cytoplasmic (N/C) ratios and relatively uniform nuclear features, whereas gliomas are marked by an unevenly distributed and increased cellularity, high N/C ratios, nuclear angularity and enlargement, and hyperchromasia [14].
362
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
Fig. 1. (A-C) Thirty-one year old male with diffuse lemptomeningeal enhancement. (A, B) FS: chronic leptomeningeal inflammation. (C) Permanents: nonnecrotizing granulomas; the patient was diagnosed with neurosarcoidosis. The error was likely due to sampling at FS. (D-F) Twenty-three year old female with imaging suggestive of low-grade glioma. (D, E) FS: hypercellular white matter with hyperchromatic, angulated glial nuclei, consistent with low-grade glioma. (F) Permanents: disordered neuronal architecture with dysmorphic neurons and balloon cells, consistent with cortical dysplasia. The error was likely due to misinterpretation.
Fig. 2. (A-C) Forty-four year old female with imaging suggestive of a low-grade glioma. (A, B) FS: hyperchromatic, angulated glial nuclei, consistent with lowgrade glioma. (C) Permanents: evenly spaced, regular nuclei without significant atypia, and gliosis was diagnosed. The error was likely due to misinterpretation. (D-F) Twelve year old female with heterogeneous left frontal mass. (D, E) FS: abundant hemosiderin fibrin, and numerous capillary-type vessels, consistent with organizing hemorrhage/thrombus. (F) Permanents: anastomosing thin- and thick-walled blood vessels without intervening neural parenchyma, changes consistent with cavernous angioma. The error was likely due to sampling.
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
363
Fig. 3. A 45-year-old male with lesions suggestive of lymphoma or multifocal glioma. (A, B) FS: atypical astrocytes, favor astrocytoma. Permanents: foamy macrophages (C). Myelin disruption on Luxol fast blue stain (D), and preservation of axons on Bodian stain (E), diagnostic of demyelinating disease. The patient was diagnosed with multiple sclerosis. The error was likely due to misinterpretation.
Malformation of cortical development (MCD) or cortical dysplasia also may pose difficulty in the tumor vs nontumor differential diagnosis, and 2 of 5 cases were misinterpreted as neoplastic. MCDs most commonly are examined by FS in patients with intractable seizures [15]. The clinical history can be misleading in that tumors are also a well-recognized cause of chronic epilepsy, and some forms of MCD can appear radiographically as a localized lesion [16,17]. MCDs represent a histologic spectrum from mild architectural abnormalities to hypercellular lesions marked by atypical cells resembling such tumors as ganglioglioma. In addition, gangliogliomas and dysembryoplastic neuroepithelial tumor
may coexist with MCD. Because some of these lesions may be characterized by subtle architectural alterations, a fragmented specimen or biopsy may be insufficient to make a diagnosis, particularly at FS. Similarly, a limited sample of certain MCDs at FS, particularly those marked by dysmorphic neurons and balloon cells, may be difficult to distinguish from a ganglioglioma. It has been our experience that most cases of MCD are not reliably diagnosed at FS but require more extensive sampling with properly oriented sections for definitive diagnosis. Although the typical demyelinating disease, such as multiple sclerosis, usually presents with multifocal white
Fig. 4. A 36-year-old female with a left frontal lobe lesion. (A, B) FS: reactive astrocytosis. Permanents: macrophages and reactive astrocytes (C), myelin disruption on Luxol fast blue stain (D), and nearly absent axons on Bodain stain (E), features of infarct. The error was likely due to misinterpretation.
364
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
Table 2 Categories of discrepancies in relation to clinical suspicion Categories of discrepancies
Inflammation (n = 8) Malformation of cortical development (n = 5) Gliosis (n = 5) Vascular malformation (n = 5) Demyelination/PML (n = 3) Infarct (n = 3) Hemorrhage (n = 3) No abnormality (n = 3) Amyloid angiopathy (n = 1) Vasculopathy (n = 1) Vasculitis (n = 1) Meningioangiomatosis (n = 1)
Clinical suspicion Tumor (n = 27)
Nonneoplastic (n = 8)
No discrete lesion (n = 4)
4 5
1
3
4 3 3 2 2 2
1 2 1 1 1 1
1 1 1
PML indicates progressive multifocal leukoencephalopathy.
matter lesions on imaging, an occasional patient may present with a solitary lesion, resembling a tumor on radiographic studies [18-21]. These atypical, “tumorlike” presentations often lead to stereotactic biopsy and difficulty for the pathologist [17]. The usual demyelinating lesion consists of macrophages, reactive astrocytes, and benign perivascular lymphocytes. It is critical to not to confuse macrophages with malignant cells; cytologic preparations may help in identifying the characteristic cell borders, uniform nuclei, and abundant foamy cytoplasm of macrophages [16,22]. Demyelinating lesions predominantly are restricted to the white matter, lack necrosis, and demonstrate reactive astrocytes throughout the lesion—all features that can help differentiate these lesions from infarct. Diligent search for viral inclusions is important in recognizing progressive multifocal leukoencephalopathy and subacute sclerosing panencephalitis—2 important infectious causes of demyelination. Progressive multifocal leukoencephalopathy typically presents in immunocompromised hosts with multiple asymmetrically distributed white matter lesions and is histologically characterized by demyelination with dense “ground glass” intranuclear inclusions within oligodendroglial cells [16,23]. Subacute sclerosing panencephalitis usually presents in the first 2 decades of life, often with a history of antecedent measles infection. Intranuclear inclusions can be identified within oligodendroglial and neuronal nuclei [24]. Interestingly, 15 of 39, or 38%, discrepant cases were vascular related. These are common causes of nonneoplastic CNS pathologic condition, as about 1 of 4 of all the diagnoses rendered in this study were vascular related (data not shown). In addition, the discrepancy rate of vascular lesions (15/39, 38%) was 1.6 times that of the rate nondiscrepant vascular lesions (62/264, 23%), suggesting that these are disproportionately difficult. Infarcts may involve the white matter, gray matter, or both. Infarcts, in addition to demyelinating lesions, are an
important diagnosis to consider when one encounters numerous tissue macrophages. The macrophages may be seen as early as 24 hours after the infarct and may persist for months [17,25]. Again, it is critical not to confuse macrophages with neoplastic cells. In the absence of prior radiotherapy, macrophages are not commonly identified in most malignancies. Touch/squash preparations may be particularly useful in this setting [16]. Central necrosis and eventual cavitation with peripheral gliosis are the key features of infarct and helpful in distinguishing it from a demyelinating lesion. In addition, infarcts demonstrate acidophilic neurons, characterized by densely eosinophilic cytoplasm and pyknotic nuclei, so-called red and dead neurons, and less chronic inflammation than demyelinating lesions [16,17]. When confronted with increased numbers of blood vessels at FS, diagnostic considerations include granulation tissue, tumors, or vascular malformation [26]. Granulation tissue is composed of small capillary-type vessels embedded in an inflamed stroma. Increased vascularity is often seen in primary tumors such as fibrillary and pilocytic astrocytomas, hemangioblastoma, and meningioma as well as certain metastatic lesions, including renal cell carcinoma and melanoma. Once the neoplastic diagnostic possibilities have been ruled out, one must consider the possibility of vascular malformation. Arteriovenous malformations and cavernous angiomas are the most often excised CNS vascular malformations. Arteriovenous malformations have a relatively high risk of hemorrhage [17,27], and arteriovenous malformations are distinguished from all other malformations by an arterial component admixed with both thin-walled and thick-walled veins and intervening neural parenchyma. Cavernous angiomas are composed of a collection of thinwalled blood vessels and never display arteries or intervening neural parenchyma [26]. The final 4 discrepant cases also represent examples of vascular-related pathologic conditions. Meningioangiomatosis is a rare hamartomatous lesion that histologically consists of an assortment of small vessels surrounded by spindled meningothelial cells with gliotic intervening neural parenchyma. They tend to present as plaquelike lesions and have a typically benign course [28]. If one focuses on the intervening gliotic parenchyma and considers the vascular proliferation secondary, the lesion may be misinterpreted as low-grade glioneuronal or glial neoplasm. Most cases of abnormally thickened vessel walls are related to hypertension and atherosclerotic changes. Radiation is another important cause and often induces sclerosis of the vessels in addition to parenchymal necrosis and dramatic cytologic atypia. Although these histologic changes can be quite suggestive of radiation injury, it is always prudent to correlate the morphologic findings with the clinical history. Amyloid angiopathy is a relatively common cause of normotensive intracerebral hemorrhage and accounts for 5% to 20% of spontaneous cerebral hemorrhages in the elderly
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
[29]. Although one must exert caution at diagnosing amyloid at FS without the aid of special stains, abnormally thickened blood vessels with amorphous eosinophilic material certainly raises the possibility, and this can be communicated to the surgeon. Targeted biopsies to diagnose vasculitis are occasionally performed, as this remains the most reliable method of diagnosis [30]. Vasculitis is defined as inflammation of the vessel wall, often accompanied by vessel wall damage. Optimally, biopsies sample the dura, leptomeninges, cortex, and white matter [30]. The highest diagnostic yield is from the superficial cortex and leptomeninges; however, only about two thirds of cases are diagnostic [26]. Given the patchiness of the process, it is not surprising that a nondiagnostic FS may be accompanied by permanent sections that yield the diagnosis. One must always remember to consider lymphoma in the differential of vasculitis, as the pattern of lymphocytic infiltration may be similar. At the time of FS, this distinction is best made by scrutiny for cytologic atypia. The lymphocytes in vasculitis should be small, round, and regular, whereas malignant lymphoma cells are most commonly large and somewhat pleomorphic with prominent nucleoli [13]. We have presented a large series of nonneoplastic CNS FS discrepancies generated from one institution; the design, however, presented several limitations to the study. First, the single-institution patient population may not be representative of the public at-large, especially because most of the samples from this series were targeted at suspected tumors, and our institution had a large cohort of chronic epilepsyrelated cases. Second, we use very few intraoperative cytologic preparations, which may skew the types of discrepancies encountered. Third, most cases were examined by 1 of 3 neuropathologists, who may bring along their own personal biases. Fourth, the retrospective chart review focus on the discrepant diagnoses provided, in our opinion, more useful information for the practicing pathologist but prevented a more definitive assessment of the sources of error for each individual case and forced us to comment on error in more general terms. Nevertheless, this study is helpful to the practicing pathologist on several fronts. Often nonneoplastic FS diagnoses are nonspecific, which is most likely related to the inherent sampling difficulties encountered at FS. It is also important for the pathologist to remember that, occasionally, nonneoplastic disorders are confused with neoplasms preoperatively. If neoplastic tissue is not definitively present, then additional tissue should be requested, or the diagnosis should be deferred. In addition, several lesions presented recurring difficulties at FS, including gliosis, MCD, demyelinating diseases, and vascular-related lesions, all of which are important entities to consider when one is presented with a difficult nonneoplastic CNS FS. Open communication with the neurosurgeon is especially important in evaluating these difficult neuropathologic FS cases.
365
References [1] Colbassani HJ, Nishio S, Sweeney KM, et al. CT-assisted stereotactic brain biopsy: value of intraoperative frozen section diagnosis. J Neurol Neurosurg Psychiatry 1988;51:332-41. [2] Taratuto AL, Sevlever G, Piccardo P. Clues and pitfalls in stereotactic biopsy of the central nervous system. Arch Pathol Lab Med 1991;115: 596-602. [3] Gomez H, Barnett GH, Estes ML, et al. Stereotactic and computerassisted neurosurgery at the Cleveland Clinic: review of 501 consecutive cases. Cleve Clin J Med 1993;60:399-410. [4] Martinez AJ, Pollack I, Hall WA, et al. Touch preparations in the rapid intraoperative diagnosis of central nervous system lesions. A comparison with frozen sections and paraffin-embedded sections. Mod Pathol 1988;1:378-84. [5] Di Stefano D, Scucchi LF, Cosentino L, et al. Intraoperative diagnosis of nervous system lesions. Acta Cytol 1998;42:346-56. [6] Shah AB, Muzumdar GA, Chitale AR, et al. Squash preparation and frozen section in intraoperative diagnosis of central nervous system tumors. Acta Cytol 1998;42:1149-54. [7] Savargaonkar P, Farmer PM. Utility of intra-operative consultations for the diagnosis of central nervous system lesions. Ann Clin Lab Sci 2001;31:133-9. [8] Whiting DM, Barnett GH, Estes ML, et al. Sterotactic biopsy of nonneoplastic lesions in adults. Cleve Clin J Med 1992;59:48-55. [9] Brainard JA, Prayson RA, Barnett GH. Frozen section evaluation of stereotactic brain biopsies: diagnostic yield at the stereotactic target position in 188 cases. Arch Pathol Lab Med 1997;121: 481-4. [10] Javedan SP, Tamargo RJ. Diagnostic yield of brain biopsy in neurodegenerative disorders. Neurosurgery 1997;41:823-8. [11] Pulhorn H, Quigley DG, Bosma JJ, et al. Impact of brain biopsy on the management of patients with nonneoplastic undiagnosed neurological disorders. Neurosurgery 2008;62:833-7. [12] Firlik KS, Martinez AJ, Lunsford LD. Use of cytological preparations for the intraoperative diagnosis of sterotactically obtained brain biopsies: a 19-year experience and survey of neuropathologists. J Neursurg 1999;91:454-8. [13] Plesec TP, Prason RA. Frozen section discrepancy in the evaluation of central nervous system tumors. Arch Pathol Lab Med 2007;131: 1532-40. [14] Burger PC, Vogel FS. Frozen section interpretation in surgical neuropathology. I. Intracranial lesions. Am J Surg Pathol 1977;1: 323-47. [15] Sisodiya SM. Surgery for malformations of cortical development causing epilepsy. Brain 2000;123:1075-91. [16] Berger PC, Scheithauer B, Vogel FS, editors. Surgical pathology of the nervous system and its coverings. 4th Ed. New York: Churchill Livingstone; 2002. p. 379-89, 414-7. [17] Love S, editor. Greenfield's neuropathology, 8th Ed.London, England: Hodder Arnold; 2007. [18] Kepes JJ. Large focal tumor-like demyelinating lesions of the brain: intermediate entity between multiple sclerosis and acute disseminated encephalomyelitis? A study of 31 patients. Ann Neurol 1993;33: 18-27. [19] Capello E, Roccatagliata L, Pagano F, et al. Tumor-like multiple sclerosis (MS) lesions: neuropathological clues. Neurol Sci 2001;22 (Suppl 2):S113-6. [20] Heyman D, Delhaye M, Fournier D, et al. Pseudotumoral demyelination: a diagnosis pitfall (report of three cases). J Neurooncol 2001;54: 71-6. [21] Love S. Demyelinating diseases. J Clin Pathol 2006;59:1151-9. [22] Burger PC. Use of cytological preparations in the frozen section diagnosis of central nervous system neoplasia. Am J Surg Pathol 1985; 9:344-54. [23] Berger JR. Progressive multifocal leukoencephalopathy. Curr Neurol Neurosci Rep 2007;7:461-9.
366
T.P. Plesec, R.A. Prayson / Annals of Diagnostic Pathology 13 (2009) 359–366
[24] Garg RK. Subacute sclerosing panencephalitis. Postgrad Med J 2002; 78:63-70. [25] Prayson RA, editor. Neuropathology. Philadelphia: Elsevier ChurchillLivingstone; 2005. p. 37-72. [26] Prayson RA, Kleinschmidt-DeMasters BK. An algorithmic approach to the brain biopsy—part II. Arch Pathol Lab Med 2006;130:1639-48. [27] Brown Jr RD. Unruptured brain AVMs: to treat or not to treat. Lancet Neurol 2008;7:195-6.
[28] Perry A, Kurtkaya-Yapicier O, Scheithauer BW, et al. Insights into meningioangiomatosis with and without meningioma: a clinicopathologic and genetic series of 24 cases with review of the literature. Brain Pathol 2005;15:55-65. [29] Maia LF, Mackenzie IR, Feldman HH. Clinical phenotypes of cerebral amyloid angiopathy. J Neurol Sci 2007;257:23-30. [30] Schmidley JW. 10 questions on central nervous system vasculitis. Neurologist 2008;14:138-9.