The lesion now known as pilomyxoid astrocytoma

ORIGINAL ARTICLE Intermediate Pilomyxoid Tumors Michael W. Johnson, MD, PhD,* Charles G. Eberhart, MD, PhD,*w Arie Perry, MD,z Tarik Tihan, MD,y Kenneth J. Cohen, MD,J Marc K. Rosenblum, MD,z Soroush Rais-Bahrami, MD,* Patricia Goldthwaite, MSc,* and Peter C. Burger, MD* Abstract: To define the spectrum of pilomyxoid morphology and to characterize the association between pilomyxoid astrocytoma (PMA) and pilocytic astrocytoma (PA), 84 cases of pediatric astrocytomas with pilomyxoid features were reviewed. Fortytwo of these tumors had coexistent features of PMA and PA ( intermediate ). With the accumulation of more cytoplasm, fibrillar background, microcysts, and thickened blood vessels, these intermediate tumors were more PA-like and less like classic PMA. In the recurrent specimens, PMAs and intermediate tumors sometimes showed more prominent PA features. Both PMA and intermediate tumors involved sites outside the hypothalamus, optic chiasm, and the third ventricle. Both the existence of intermediate tumors and the finding that some PMAs and intermediate tumors mature into PA-like neoplasms over time, provided strong support for a biological relationship between PMA and PA. Additional evidence for a maturational effect was the finding that intermediate tumors occurred in patients with a median age of 36 months compared with the median age of 21 months for those with PMAs (P = 0.017). Features that were often assumed to be poor prognostic indicators in gliomas, that is, necrosis, mitotic figures, and vascular proliferation, were not uncommon in typical PMAs and intermediate lesions. Further follow-up is needed to more accurately determine the prognosis of intermediate tumors. Key Words: pilomyxoid, pilocytic, tumor, astrocytoma, pediatric, suprasellar, brain (Am J Surg Pathol 2010;34:1783 1791) From the Departments of *Pathology; wophthalmology; JPediatrics and Oncology, Johns Hopkins University School of Medicine, Baltimore, MD; zdivision of Neuropathology, Washington University School of Medicine, St Louis, MO; ydepartment of Pathology, University of California at San Francisco, San Francisco, CA; and zdepartment of Pathology, Memorial Sloan-Kettering Cancer Center, NY. Supported in part by the Johns Hopkins Pilocytic Astrocytoma/Pilomyxoid Astrocytoma Research Fund Presented in part at the 98th meeting of the United States and Canadian Academy of Pathology, Boston, MA, March 12, 2009. Correspondence: Peter C. Burger, MD, Pathology Building Room 710, The Johns Hopkins Hospital, 600 N. Wolfe Street, Baltimore, MD (e-mail: pburger@jhmi.edu). Copyright r 2010 by Lippincott Williams & Wilkins The lesion now known as pilomyxoid astrocytoma (PMA), was initially described as a pilocytic astrocytoma (PA) of infancy in 1996. 4 Cytologic monomorphism, cobweb architecture, and abundant myxoid background were features that distinguished it from classic PA. Rosenthal fibers and eosinophilic granular bodies were scant or absent. The sites involved were, in general, those of PAs. An earlier publication had discussed diencephalic PAs in infants, and, although it made no distinction between them and classic PAs, some may well have been PMAs. 6 The issue was revisited in a 1999 study of 18 pediatric astrocytomas that had distinctive clinical and histologic characteristics. 13 Although not discounting the possibility that these were variants of PA, the term pilomyxoid astrocytoma was coined. The rates of recurrence and cerebrospinal fluid dissemination were higher in these tumors compared with classic PA in the same age group and location. Subsequent experience has confirmed an overall, more aggressive biological behavior for this tumor category, but individual PMAs may not behave any differently than classic PAs, and PMAs can occasionally occur in elderly individuals and adults. 5,9,10,11 The 2007 World Health Organization classification considers PMA as a grade 2 variant of PA. 1,12 According to the 2007 World Health Organization classification, PA is defined as a relatively circumscribed, slowly growing, often cystic astrocytoma occurring in children and young adults, histologically characterized by a biphasic pattern with varying proportions of compacted bipolar cells associated with Rosenthal fibers and loosetextured multipolar cells associated with microcysts and eosinophilic granular bodies/hyaline droplets. 12 PMA is defined as a piloid neoplasm, closely related to PA, which has a prominent mucoid matrix and angiocentric arrangement of monomorphous, bipolar tumor cells, typically without Rosenthal fibers or eosinophilic granular bodies/hyaline droplets. 12 Thus, the lesions are defined similarly, and in practice, differentiation of the 2 often involves the qualitative and quantitative assessment of mucoid/myxoid matrix and monomorphous cellularity. Pathologists confront the intermediate or mixed neoplasms with features of both PMA and PA. This study focuses largely on these latter tumors with intermediate features to investigate the histologic scope of what may be a nexus between PMA and PA, rather than to create a new subcategory. Am J Surg Pathol Volume 34, Number 12, December 2010 www.ajsp.com 1783

Johnson et al Am J Surg Pathol Volume 34, Number 12, December 2010 MATERIALS AND METHODS One hundred twenty pediatric (18 y or under) cases were retrieved from our in-house and consultation surgical pathology and from the Johns Hopkins University PMA registry (www.hopkinskimmelcancercenter. org/index.cfm/cid/1532) from 1998 to 2008. Cases obtained from Washington University of St Louis (St Louis, MO) and Memorial Sloan-Kettering Cancer Center (NY) were also included. Cases at Washington University were acquired by searching for the term pilomyxoid in the diagnosis, and those at Memorial Sloan-Kettering were all cases that had at least one area of classic PMA. This study was carried out in compliance with the Johns Hopkins Hospital institution review board approval (NA_00015113, PI name: Peter Burger, date of expiration March 1, 2011). Pilomyxoid and infantile PA were the keyword and group keyword search parameters used to query the Johns Hopkins Hospital computerized pathology database for their mention in either the diagnosis section and/or within the descriptive narrative comment portion of the final diagnostic report. The cases that were retrieved by this methodology included specimens diagnosed as PMA, cases with microscopic features suggestive of PMA (such as tumors with pilomyxoid features ), and cases in which a differential within the microscopic description included PMA. Any cases in which the diagnosis explicitly stated that the specimen was not a PMAFa diagnostic strategy not practiced at our institutionfwere not retrieved. At a minimum, hematoxylene and eosin-stained slides were available for all cases. After the cases had been retrieved, all available material was rereviewed in a blinded manner without regard for earlier diagnosis, radiographic information, or patient name, demographics, or clinical outcome. Each case was classified by 2 pathologists (M.W.J. and P.C.B.) as (1) classic PMA, (2) classic PA, (3) neither classic PMA nor classic PA but with some features of both, (4) without features of either, or (5) insufficient material for analysis. Tumors that were classic PA, without features of either, or of insufficient material, were excluded from further evaluation and discussion. PMAs were identified, largely at low magnification, by their monomorphous population of small bipolar cells in which there was a conspicuous lack of compact fibrillar tissue or biphasic architecture. A myxoid/mucoid background was often present, but only optically clear intercellular spaces were present in some cases. Perivascular pseudorosettesfor as some prefer to describe them, angiocentric arrangements of tumor cells so as not to be confused with ependymal differentiationfwere common, but not required. The presence or absence of vascular proliferation, necrosis, and tissue infiltration was also not exclusionary. An occasional microcyst was tolerated, but multiple such structures were generally considered intermediate tumors or PAs unless they remained cytologically monomorphous and were composed exclusively of small bipolar cells. In the absence of other features of PA, neither eosinophilic granular bodies nor Rosenthal fibers themselves precluded the diagnosis of PMA. Intermediate lesions were defined by the presence of some features typical of PMA in combination with features that would be considered more consistent with PA. All intermediate lesions had some combination of pilomyxoid qualities, such as small monomorphous bipolar cells, focal myxoid substance, and variable angiocentric growth; and at least some pilocytic features, such as microcysts (multiple), cells with thicker, longer processes, compact finely fibrillar tumor tissue, and Rosenthal fibers. PAs were operationally defined as gliomas with variable ratios of fibrillar and spongy/microcystic tissue. A myxoid background was permitted but there had to be some areas without intervening myxoid substance. Although often present, neither eosinophilic granular bodies nor Rosenthal fibers were required. Piloid tumors that were solidfthat is, without microcystsfwere also accepted if they were compact and contained Rosenthal fibers. Thesimplepresenceorabsence,orsemiquantitative measures of 11 specific histologic features were recorded: biphasic architecture, calcification, myxoid substance, microcysts, perivascular pseudorosettes, eosinophilic granular bodies, Rosenthal fibers, oligodendroglioma-like features, glomeruloid vascular proliferation, neuropil infiltration, and necrosis. The mitotic rates were determined for 10 HPF ( 40 objective; representing a manually calibrated field area of 0.193 mm 2 ). The results of immunohistochemical staining for glial fibrillary acidic protein (GFAP), synaptophysin, and Ki-67 (DAKO, Mib-1) were recorded when the tissue was available for such studies. There were immunohistochemical preparations for GFAP and synaptophysin in 37 and 25 cases, respectively. The Ki-67 (Mib-1) rate was approximated by visually matching observed fields with a diagram of computer-generated/ simulated fields depicting varying degrees of cellularity and Ki-67 indices. Histologic features were compared in initial and recurrent lesions in the 9 cases in which specimens of both were available. TABLE 1. Cases Pilomyxoid Astrocytoma Intermediate Tumors No. cases 42 42 Average age (mo) 41* 66* Median age (mo) 21** 36** Sex Male 30 26 Female 12 16 Location Suprasellar 23 24 Supratentorial 11 9 Infratentorial (5) (5) Cerebellum 2 Brainstem 2 Posterior fossa unspecified 3 3 Spinal 3 4 *P = 0.018 by a Student t test. **P = 0.017 by Wilcoxon Signed-ranks test. 1784 www.ajsp.com r 2010 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 34, Number 12, December 2010 The mean and median ages of patients with classic PMA and intermediate cases were compared by the Student t test (using Microsoft Office Excel software) and Wilcoxon signed-ranks test (VassarStats; http://faculty. vassar.edu/lowry/wilcoxon.html), respectively. These were 1-sided analyses because we hypothesized that the intermediate tumors would be in an older patient population than PMAs. Significance was valued at P<0.05. The P values for each statistical test are reported with the means and medians (Table 1). RESULTS There were 42 PMAs and 42 intermediate or mixed lesions, for a total of 84 cases. These 84 cases included the 18 presented by Tihan et al. 13 The remaining 36 cases FIGURE 1. Spectrum of pilomyxoid astrocytomas (PMAs)/pilocytic astrocytomas (PAs). A to L, The histologic continuum and variations along the PMA-PA spectrum. The loosely organized tumors in panels A and B have abundant myxoid background and cells with variable bipolarity. Some tumor cells relate to vessels. Better-formed perivascular formations resembling those of ependymomas are present in panels C and D. Tumors shown in panel E, F, G, H, and I are variably loose and piloid, yet without the dense fibrillarity or spongy microcystic architecture of PA. Intermediate tumors with small monomorphous cells typical of PMA, but also microcysts (J) emphasize the difficulty in some cases of assigning a lesion to either PMA or PA category. K and L, Classic PA, with biphasic architecture. r 2010 Lippincott Williams & Wilkins www.ajsp.com 1785

Johnson et al Am J Surg Pathol Volume 34, Number 12, December 2010 were excluded from the study because 10 cases were classic PAsFwithout pilomyxoid featuresfaccording to the criteria mentioned above, 9 cases lacked specific features of either PMA or PA, 13 cases had insufficient pathology material for definitive diagnosis, and in 4 cases, patient demographic and/or topological information was missing. Of the 9 cases determined to be neither PMA nor PA, 5 cases could only be described as infiltrating gliomas, 3 cases were most consistent with the diagnosis of angiocentric glioma, and 1 case seemed to be a dysembryoplastic neuroepithelial tumor-like lesion of the septum pellucidum. The patients ranged from 7 weeks to 18 years of age. Patients with PMAs were younger at the time of initial diagnosis, (mean age, 41 mo; P = 0.018; median age, 21 mo, P = 0.017) than were those with patients with intermediate tumors (mean age, 66 mo; median age 36 mo) (Table 1). Forty-seven tumors were suprasellar/hypothalamic, 20 tumors were supratentorial but nonsuprasellar (cerebral cortical), 10 tumors were infratentorial, and 7 tumors were intraspinal. The PMA and intermediate tumors were distributed similarly (Table 1). Intermediate tumors formed a broad spectrum from those that were more similar to PMA than to those that were more similar to PA. At one extreme, some of the intermediate lesions were prominently myxoid and monomorphous with vague pseudorosettes/angiocentric arrangements, but also cells with thicker processes and at least focal, dense fibrillar tissue. At the other end of the spectrum, the tumors had more classic features of PA, such as a dense growth pattern, extensive fibrillarity, occasional eosinophilic granular bodies, but also focal myxoid areas with small bipolar cells with little cytoplasm. Tumors lying between the 2 extremes varied quantitatively and qualitatively in the proportion of small bipolar cells, the degree of eosinophilia produced by compact areas of cells processes, and the extent of vascular hyalinization (Fig. 1). Eosinophilic granular bodies were present in some cases with exclusive PMA features, but Rosenthal fibers were present only in compact eosinophilic fibrillar tissue in the intermediate category. Thickened vessels seemed more prominent in intermediate tumors than in PMA. The frequencies of specific histologic features in PMAs and intermediate lesions are given in Table 2. PMAs lacked biphasic architecture and Rosenthal fibers, and were less likely to have microcysts by definition. Rosenthal fibers were restricted to the intermediate group. Vascular proliferation (Figs. 2A, B) was present in approximately the same percentage of PMAs (38%) and intermediate lesions (36%). Coagulative necrosis (Fig. 2C) affecting both vessels and parenchyma without surrounding palisading tumor cells was present within both the PMAs (17%) and the intermediate lesions (21%). For the most part, mitotic activity in both categories was low. Zero mitosis per 10 HPF was seen in 32 PMAs and 32 intermediate tumors. Five PMAs and 8 intermediate tumors contained 1 mitosis per 10 HPF. Two mitoses per 10 HPF were seen in 3 PMA and 2 intermediate tumors. Greater than 2 mitoses per 10 HPF TABLE 2. Histologic Features Pilomyxoid Astrocytoma (n = 42) Intermediate Tumors (n = 42) Perivascular pseudorosettes 32 (76%) 30 (71%) Myxoid substance 28 (67) 30 (71) Microcysts 18 (43) 31 (74) Biphasic architecture 0 (0) 8 (19) EGBs 1 (2) 10 (24) Rosenthal fibers 0 (0) 7 (17) Calcification 1 (2) 2 (5) Oligodendroglioma like 5 (12) 8 (19) Vascular proliferation 16 (38) 15 (36) Infiltrative nature 9 (21) 7 (17) Necrosis 7 (17) 9 (21) EGB indicates eosinophilic granular bodies. were seen only in 2 PMAs, each of which showed 4 mitoses per 10 HPF. Ki-67 (MIB-1) preparations were available in 15 PMAs and 14 of the intermediate lesions. Staining indices ranged from less than 1% to 15% (the last seen in one of the intermediate cases). However, the mean Ki-67 indices were similar in PMA (mean, 3.4%; median, 3.0%) and intermediate groups (mean, 3.4%; median, 2.0%). Immunostains for GFAP were available in 16 cases in the PMA group and 21 cases in the intermediate group. All tumors, in both categories, showed strong positive staining with GFAP (Fig. 3A). Ten of 16 PMAs were synaptophysin immunoreactive as were 9 of the 11 intermediate lesions. The extent of synaptophysin reactivity was variable and was either diffusely cytoplasmic (Fig. 3B) or focal and punctuate. Immunostained slides were those from the original institutions; we did not repeat them. Slides were available in 9 cases from both the initial and at least 1 subsequent operation (Table 3). Two tumors in the PMA group were histologically identical after 6 months and 6 years of initial surgery, respectively. Three other tumors in the PMA group subsequently had histologic features of intermediate tumors approximately 9 months, 3 years, and 7 years later, respectively. One tumor had histologic features of PMA initially and histologic features of PA at second surgery after 3 years (Fig. 4). Three intermediate tumors did not appreciably alter their histologic features after 4 months, 4 months, and 3 years, respectively. Significant clinical follow-up information was available in 20 of the 84 cases (Table 4). These included patients with tumors of which 12 were classic PMAs and 8 were intermediate in appearance at the time of initial diagnosis (Table 3). One patient with a classic PMA of the optic chiasm is deceased. Three patients, with intermediate tumors in the suprasellar region, lateral ventricle, and brainstem, are deceased. Although death certificates for the deceased patients were not available to us, per communication with medical staff at the treating institutions, the deaths resulted from complications of the tumors of the central nervous system. 1786 www.ajsp.com r 2010 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 34, Number 12, December 2010 FIGURE 2. Mitoses, vascular proliferation, necrosis, and invasion. Some pilomyxoid astrocytomas (PMAs) and intermediate tumors had features that in some contexts are indices of higher grade: mitoses (A), vascular proliferation (A, B), and necrosis (C, left). Parenchymal invasion, although not unknown in pilocytic astrocytomas, was especially prominent in some tumors in PMAs and intermediate lesions (D). Tumor cells diffusely infiltrate along vessels (E). On reoperation 3 years later, the tumor shown in panel D was pilocytic astrocytoma with microcysts and Rosenthal fibers (Figs. 4C, D). DISCUSSION The purpose of this study was to review retrospectively a large group of tumors that had initially been described as having features of PMA to (1) further our understanding of the range and scope of those histologic features within such tumors; (2) determine whether histologic features of PMA and PA coexist in pediatric tumors; (3) confirm that features suggestive of high grade r 2010 Lippincott Williams & Wilkins www.ajsp.com 1787

Johnson et al Am J Surg Pathol Volume 34, Number 12, December 2010 FIGURE 3. Glial fibrillary acidic protein (GFAP) and synaptophysin staining. Both PMA and intermediate tumors were positive for GFAP (A). The majority of tumors were also immunoreactive for synaptophysin staining. The latter varied from diffusely cytoplasmic or punctuate to membranous (B). in other types of tumors of the central nervous system, are often present in both PMA and PA; and (4) investigate the GFAP and synaptophysin immunohistochemical profiles of the lesions. As we expected, given anecdotal difficulties in differentiating PA from PMA, we found a wide variation in a number of histologic features. Of the 84 cases discussed here, half of the cases (42 cases) had intermediate features of both PA and PMA. Such tumors seemed to comprise varying combinations of both diagnostic entities; therefore, we designated them as intermediate tumors. We did not do this to coin a new diagnostic category, but rather to designate lesions that could not be strictly confined to either the PMA or PA category. It is not a well-defined group; however, and even after reviewing 84 cases, it was sometimes difficult to decide whether a lesion, on rereview, was best considered PMA, intermediate tumor, or PA, much less to do it reproducibly. Where to draw the line between PMA and intermediate lesions in a given case is thus not always clear, if not impossible, in tumors in which the 2 tissue patterns coexist. Intermediate cases represent a transition or link between PMAs and PAs, and are also supported by the fact that PMA and intermediate lesions occur in similar locations as PAs in general (Table 1). Comparisons of specific histologic features between PMAs and the intermediate group also showed a few surprises. Fibrillar areas were more common in intermediate lesions than in PMAs, by definition as such tissue was not accepted in PMAs. The incidence of biphasic architecture was also different as this required fibrillar areas that were operationally excluded from the PMA group. Rosenthal fibers were more frequent in the intermediate group, in a sense by definition as they occurred in fibrillar areas that would not be present in PMAs. Eosinophilic granular bodies, in contrast, were more common in intermediate lesions. Calcifications were also more frequent in intermediate lesions than PMAs, but few cases in either group were mineralized. Both PMA and intermediate tumors were similarly likely to have a myxoid background, pseudorosette formation/angiocentric pattern, and rare features, such as oligodendroglial-like infiltrative components. Vascular proliferation and necrosis, although seen in less than half of the tumors, were present with a similar frequency in both PMA and intermediate tumors. TABLE 3. Patients With Multiple Specimens Age Location No. Surgeries* Op Interval(s) Therapyw Clinical Impression Tumor Typez 7 y Parietal-temporal 2 6 y? Recurrence with growth on MRI PMA-PMA 6 m Temporal-thalamic 2 6 mo No Completion of subtotal (95%) resection PMA-PMA 10 y Temporal 2 9 mo No Recurrence after gross total resection PMA-Intermediate 7 y Lateral ventricle 3 7 y/2 y Yes Recurrences 2 times with growth on MRI PMA-Intermediate 5 m Suprasellar 4 9 mo/2 y Yes Recurrences 3 times with growth on MRI PMA-Intermediate 9 y Suprasellar 2 4 mo No Subtotal resection with growth on MRI Intermediate-Intermediate 13 m Suprasellar 3 3 mo/2 wk No Incomplete resections due to complications Intermediate-Intermediate 9 y Suprasellar 2 3 y Yes Incomplete resection with growth on MRI Intermediate-Intermediate 14 m Midbrain 2 3 y Yes Incomplete resection with growth on MRI PMA-PA *These were the number of surgeries from which surgical specimens were available. The 5-month-old patient had a subsequent third repeat resection that was not available for review. wchemotherapies and/or radiotherapies occurred during the interval between surgeries. Chemotherapeutic regimens included carboplatin and vincristine. Most patients received additional therapies afterresection/debulking, but complete oncologic follow-up after resections were not uniformly available. zon the left of the arrow is the diagnosis of the initial biopsy or resection. On the right of the arrow is the diagnosis in subsequent resection(s).?information not available. MRI indicates magnetic resonance imaging; Op Interval, interval between surgeries; PA, pilocytic astrocytoma; PMA, pilomyxoid astrocytoma. 1788 www.ajsp.com r 2010 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 34, Number 12, December 2010 FIGURE 4. Maturation of a pilomyxoid astrocytoma. Shown are 4 images from 2 resections of a child with a midbrain tumor. A, First resection specimen with the typical monomorphism and myxoid background of PMA. The tumor was infiltrative (B). The recurrent tumor was resected 4 years later (C, D) at which time it was dominated by compact architecture and microcysts (C). Rosenthal fibers were present (C, D). The terminology for the spectrum and grading for the intermediate lesions is unclear, especially as their outcomes vis-a -vis those of PMA and PA are unknown. The number of years over which our cases were accrued, dispersion of patients, multiplicity of referring institutions, short length of follow-up in some of the recent cases, and the lack of clinical information from patients lost to other regional treatment facilities limited our ability to analyze outcomes more completely and reliably. Given these uncertainties, perhaps one should refer to these as PA with pilomyxoid features, with an explanation that the biological significance of such a lesion is unclear. The intermediate lesions should not automatically be considered benign, however, as several intermediate tumors in our series recurred and required additional therapy. In addition, several of the patients with tumors in our intermediate category have died. Given these anecdotal cases, it might be useful to provide a comment as to the extent of pilomyxoid features to convey the fact that such a change is extensive or only focal. Despite the inability to correlate histology and follow-up, several observations are pertinent. Patients with intermediate lesions were significantly older than those with PMAs (median age 36 mo vs. 21 mo), in accordance with the tendency of PMAs to occur in younger patients. This is also consistent with the concept that PMAs can mature through intermediate forms into PA. In our case group, 9 patients had consecutive resections of which 4 showed histologic evolution toward the pilocytic end of the spectrum. Three were PMAs at diagnosis but had intermediate histologic features on subsequent resection for recurrence. One patient who was originally diagnosed with PMA developed a recurrence after chemotherapy and radiotherapy that was histologically identical to a classic PA. Given the rarity of maturation, it is impossible to say with certainty whether this is a natural phenomenon or is induced by chemotherapy and/or radiotherapy. It is possible that this perceived maturation is an artifact of sampling from heterogeneous tumors. However, maturation or evolution has been noted histologically by others, 2,3,4,7 and r 2010 Lippincott Williams & Wilkins www.ajsp.com 1789

Johnson et al Am J Surg Pathol Volume 34, Number 12, December 2010 TABLE 4. Patients With Significant Follow-up Histologic Features Tumor Type Age at Diagnosis/Sex Year of Diagnosis Tumor Location Necrosis VP Ki-67 Ca ++ Patient Status PMA* 7 y/f 2000 Parietaltemporal Yes Yes <1% No Initial diagnosis 2000; reoperated in 2006; patient with persistent disease PMA 21 m/m 2002 Suprasellar No No NA No Chemo-radiotherapy for recurrence age 5 y; patient stable without recurrence PMA* 7 y/m 1995 Lateral Ventricle No No NA Yes Multiple recurrences despite therapy; patient is deceased Intermediate 6 m/f 1995 Suprasellar No No <1% No No recurrence Intermediate 6 y/m 1990 Suprasellar No No NA No No recurrence Intermediate* 9 y/m 1997 Suprasellar No No <1% No No recurrence after completion resection Intermediate* 9 y/f 1995 Suprasellar No No 2% No No recurrence after completion resection and chemo-radiotherapy PMA 14 y/m 2000 Temporal No No NA No No recurrence PMA 6 y/m 2000 Spinal cord No Yes NA No No recurrence PMA 13 m/m 1988 Suprasellar No No <1% No Pituitary dysfunction, but no recurrence Intermediate 7 y/m 2005 Spinal cord No No <1% No Stable persistant disease s/p chemotherapy PMA 2 y/m 2006 Spinal cord No No 2% No Gross total resection; recurrence; stable persistent disease s/p chemotherapy Intermediate 8 m/f 2006 Suprasellar No No NA No Patient is deceased Intermediate 11 m/f 2005 Suprasellar No No NA No Patient with stable persistent disease PMA 2 y/m 1994 Suprasellar No Yes NA No Multiple shunt revisions; CVA; no tumor recurrence PMA 3 m/f 1994 Suprasellar No No NA No Patient is deceased Intermediate 8 y/f 1998 Brainstem No No <1% No Patient is deceased after hemorrhage s/p chemoradiation. PMA 16 m/m 1983 Suprasellar No No NA No No recurrence PMA 6 m/m 1992 Cerebral Yes Yes <1% No Stable persistent disease Cortex PMA 12 y/f 1993 Suprasellar No No NA No Stable persistent disease NA: Ki-67 preparations were not available. *In patients in which the histologic appearance of the tumor changed in subsequent resections (PMA to intermediate or pilocytic astrocytoma), the initial diagnosis is listed (see Table 3 for comparison). CVA indicates cerebrovascular accident; F, female; M, male; s/p, status post; PMA, pilomyxoid astrocytoma; VP, vascular proliferation. radiologic evolution of a PMA from noncystic to cystic architecture has also been shown. 8 The incidence and timing of this evolution are unknown. We did not, within this study or in our general experience, see the reverse phenomenon (a classically pilocytic tumor becoming PMA). The incidence of tissue infiltration present in some cases of both PMA and intermediate tumors was of interest. Although this is often used to help distinguish infiltrating, diffuse astrocytomas from pilocytic types, it is thus not an absolute discriminant. The occurrence of vascular proliferation in PMAs or intermediate tumors similarly adds uncertainty, although neither is prima facie evidence of a high-grade infiltrating astrocytoma. We confirmed the largely anecdotal experience that PMAs and intermediate lesions, and even some PAs, may be immunoreactive for synaptophysin. As cells in PMAs are isolated, with little cell-cell contact, it is difficult to exclude an artifactual edge effect, but the staining appeared to be real or at least intense and localized to tumor cells in a manner that might create diagnostic problems. The small bipolar cells in a myxoid background are unlikely to be confused with neurons by any other criteria, however. The significance of such staining, in what is usually an intensely GFAP-positive glioma, is unclear. What is clear, however, is that the staining is repeatable with different preparations. Therefore, a surgical pathologist should not use synpatophysin staining, focal or otherwise, to exclude intermediate tumors or PMAs. Although approximately half of the cases reviewed for this study were intermediate, one should not necessarily conclude that intermediate tumors and classic PMAs are found with equal frequency in the general population. The large majority of cases reviewed for this study were acquired as a result of the process of outside consultation. Focal areas suggestive of PMA in a PA can cause diagnostic difficulties that lead to requested expert input. Thus, such cases may have been self-selected from somewhere in the midst of the PA-PMA spectrum. The initial goal of this study was to better understand PMA and their relationship with PA. Future review of a larger group including both pilocytic and pilomyxoid tumors may provide more information about the frequency of tumors with intermediate histologic features. However, whatever the proportion of classic PMA and tumors with intermediate features, we wish to convey several points to surgical pathologists: (1) tumors with intermediate PMA/ PA features exist; (2) such intermediate tumors are not rare; and (3) at this point in our understanding of these tumors we cannot predict the behavior of individual patients based on the quantity of the admixed histologic 1790 www.ajsp.com r 2010 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 34, Number 12, December 2010 elements. We do not know whether intermediate tumors behave like PA or like PMA, although some can be aggressive, as can occasional classic PAs. Pathologists therefore should note a PMA component in tumors whenever it is present. In summary, we have illustrated the complex pathologic spectrum of lesions with pilomyxoid features. Although as discussed, patient follow-up at the time of this study was insufficient to evaluate the relative clinical behavior of these tumors with pilomyxoid features, the process of continued follow-up in these patients is underway through the Pilomyxoid Astrocytoma Registry (The Sidney Kimmel Comprehensive Cancer Center, www. hopkinskimmelcancercenter.org/index.cfm/cid/1759). REFERENCES 1. Brat DJ, Scheithauer BW, Fuller GN, et al. Newly codified glial neoplasms of the 2007 WHO classification of tumours of the central nervous system: angiocentric glioma, pilomyxoid astrocytoma and pituicytoma. Brain Pathol. 2007;17:319 324. 2. Ceppa EP, Bouffet E, Griebel R, et al. The pilomyxoid astrocytoma and its relationship to pilocytic astrocytoma: report of a case and critical review of the entity. J Neurooncol. 2007;81:191 196. 3. Chikai K, Ohnishi A, Kato T, et al. Clinico-pathological features of pilomyxoid astrocytoma of the optic pathway. Acta Neuropathol. 2004;108:109 114. 4. Cottingham SL, Boesel CP, Yates AJ. Pilocytic astrocytoma in infants: a distinctive histological pattern. J Neuropathol Exp Neurol. 1996;55:654. 5. Fernandez C, Figarella-Branger D, Girard N, et al. Pilocytic astrocytomas in children: prognostic factorsfa retrospective study of 80 cases. Neurosurgery. 2003;53:544 553. 6. Janisch W, Schreiber D, Martin H, et al. Diencephalic pilocytic with clinical onset in infancy: biological behavior and pathomorphological findings in 11 children. Zentralbl Allg Pathol. 1985;130:31 43. 7. Jeon Y-K, Cheon J-E, Kim S-K, et al. Clinicopathological features and global genomic copy number alterations of pilomyxoid astrocytoma in the hypothalamus/optic pathway: comparative analysis with pilocytic astrocytoma using array-based comparative genomic hybridization. Mod Pathol. 2008;21:1345 1356. 8. Kleinschmidt-Demasters BK, Foreman N, Handler M. Pilomyxoid tumors with a chronic relapsing course may become more cystic and mature to pilocytic histology: abstract from the International Symposium on Pediatric Neuro-oncology. J Neurooncol. 2008;10: 460 461. 9. Komotar RJ, Burger PC, Carson BS, et al. Pilocytic and pilomyxoid hypothalamic/chiasmatic astrocytomas. Neurosurgery. 2004;54:72 79. 10. Komotar RJ, Mocco J, Carson BS, et al. Pilomyxoid astrocytoma: a review. MedGenMed. 2004;6:42 52. 11. Komotar RJ, Mocco J, Zacharia BE, et al. Astrocytoma with pilomyxoid features presenting in an adult. Neuropathol. 2006;26:89 93. 12. Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO Classification of Tumours of the Central Nervous System. Acta Neuropathol. 2007;114:97 109. 13. Tihan T, Fisher PG, Kepner JL, et al. Pediatric astrocytomas with monomorphous pilomyxoid features and a less favorable outcome. J Neuropathol Exp Neurol. 1999;58:1061 1068. r 2010 Lippincott Williams & Wilkins www.ajsp.com 1791