Literature Review Paper

www.elsevier.com/locate/humpath

Human Pathology (2012) 43, 1668–1676

Original contribution

Infant brain tumors: a neuropathologic population-based institutional reappraisal☆,☆☆

Christopher DunhamMD, FRCPCa,⁎, Shibu Pillai MBBS,MChb, PaulSteinbok MBBS, FRCSCb

aDivision of Anatomic Pathology, Department of Pathology and Laboratory Medicine, Vancouver, British Columbia bDivision of Pediatric Neurosurgery, Department of Surgery, British Columbia's Children's Hospital (BCCH) and the University of British Columbia (UBC), Vancouver, British Columbia

Received 27 September 2011; revised 6 December 2011; accepted 8 December 2011

C v A N M

C

0 d

Keywords: Infant brain tumor; Atypical teratoid rhabdoid tumor (ATRT); Neuropathology; Cribriform neuroepithelial tumor (CRINET)

Summary The factors that impact the long-term functional outcome for infants with brain tumor are unclear. The clinicopathologic features of all infant brain tumors occurring at our institution (1982- 2005) were reexamined to explore the factors influencing prognosis. The details of the neuropathologic review are reported herein. Thirty-five cases were identified and included 7 astrocytomas (6 low grade and 1 glioblastoma), 6 atypical teratoid rhabdoid tumors, 5 choroid plexus papillomas, 4 ependymomas (3 anaplastic), 4 teratomas (3 immature), 2 supratentorial primitive neuroectodermal tumors, 2 ganglio- gliomas, 2 desmoplastic tumors of infancy, and 1 each of “medulloblastoma with extensive nodularity,” adamantinomatous craniopharyngioma, and 1 “malignancy not otherwise specified.” The original diagnosis was changed in 8 cases (23%), and atypical teratoid rhabdoid tumors was the most common revision (n = 5). Case 9 was unusual in that both the patient and her 2-year-old sister displayed INI-1 immunonegative posterior fossa tumors and extended survival. Tumor grade was altered in 6 cases (17%), the most significant instance being the downgrading from the World Health Organization grade IV to I (case 18: supratentorial primitive neuroectodermal tumors to desmoplastic tumors of infancy). As opposed to other reports in the literature, our cohort contained a substantially higher frequency of atypical teratoid rhabdoid tumors and a lower frequency of medulloblastoma. Changes in the histologic diagnosis/grade in a significant subset of cases most likely reflect the continual evolution of brain tumor classification schemes. INI-1 immunohistochemistry was instrumental in the pathologic assessment of select cases and raised the possibility that atypical teratoid rhabdoid tumors may be the most common infant brain malignancy. © 2012 Elsevier Inc. All rights reserved.

☆ Financial interests: Nothing to disclose. ☆☆ Disclosures: These data were presented in part at the 51st Annual

anadian Association of Neuropathologists' (CANP) Meeting in Vancou- er, British Columbia, Canada, at the 39th Annual Meeting of the American ssociation of Neurological Surgeons (AANS)/CNS Section on Pediatric eurological Surgery, Cleveland, Ohio, USA and at the 39th Annual eeting of the International Society for Pediatric Neurosurgery, Goa, India. ⁎ Corresponding author. UBC 4500 Oak St. Vancouver, British

olumbia, Canada, V6H 3N1. E-mail address: [email protected] (C. Dunham).

046-8177/$ – see front matter © 2012 Elsevier Inc. All rights reserved. oi:10.1016/j.humpath.2011.12.011

1. Introduction

According to the National Cancer Institute's “Surveil- lance, Epidemiology, and End Results (SEER) Program,” the incidence of infant brain tumors (IBTs) (ie, tumors occurring before 12 months of age) from 1986 to 2006 was 3.5 of 100 000 [1]. IBTs usually present in a somewhat vague manner typified by macrocrania, signs of increased intracranial

1669Infant brain tumors

pressure, and “nonspecific findings” (eg, failure to thrive, irritability, lethargy, delay in developmental milestones, etc) [2]. The resulting delay in diagnosis and treatment of many IBTs may be caused by the fact that these typically large tumors most often originate in the more expansive supra- tentorial space [3].

The meta-analysis of Larouche et al [2] detailed the his- toric treatment and outcomes for IBTs. Most cases under- went surgery, with rates of gross total resection varying from 6% to 55%. Chemotherapy, involving varying combinations of agents, was undertaken in 4% to 34% of cases. Radiation therapy was delivered in 5% to 39%. Notably, SEER data demonstrated a marked reduction in the use of radiation therapy from 2.1 of 100 000 to less than 0.2 of 100 000 cases from 1973-1986 to 1986-2006 [1]. With respect to outcome, 5-year survivals ranged from 21% to 81% across series, suggesting a highly varied experience among institutions.

The cooperative survey of the Educational Committee of the International Society of Pediatric Neurosurgery, led by Di Rocco et al [3], reviewed 886 IBTs and found the most common histologic diagnoses to be astrocytoma (31%), medulloblastoma (13%), ependymoma (12%), choroid plexus papillomas (CPP) (12%), “primitive neuroectodermal tumors” (PNETs) (7%), and teratomas (6%) [3]. The frequency of some of the foregoing entities was slightly different in the study of Oi et al [4], whose 262 cases originated in the far east of Asia. In particular, Oi et al found higher percentages of medulloblastomas (17%) and terato- mas (8%) but a lower percentage of astrocytoma (23%), suggesting certain geographic/racial differences in the incidence of some IBTs. Although informative, the use of these large, older studies is somewhat limited. First, and as illustrated by Larouche et al [2], most of the reported infant astrocytomas (316/395, or 80%) were not graded. The latter is especially problematic because “astrocytoma,” being the most common category of IBT, now represents a large col- lection of entities varying from glioblastoma (World Health Organization [WHO] grade IV) to pilocytic astrocytoma (WHO grade I), whose treatment and ultimate prognosis is strongly dependent on grade [5]. Second, these older studies do not benefit from recent revisions in our brain tumor classification schemes. For example, entities such as “atypical teratoid rhabdoid tumor” (ATRT) and desmoplastic tumor of infancy (DTI) were largely unrecognized in the meta-analysis of Larouche et al [2,6]. Because recent work has shown that many IBTs exhibit unique molecular under- pinnings (eg, BRAF-KIAA1549 tandem duplication and fusion of pilocytic astrocytomas, V600E BRAF mutation in a variety of low-grade astrocytomas and glioneuronal neo- plasms, etc), their exact classification is crucial to modern targeted therapeutic regimes [7,8].

The most common individual IBT is most likely pilocytic astrocytoma [1,2,5,9]. Gliomas represented 42.5% (254/597) IBTs in the study of Bishop et al [1], of which 161 were graded. “Low” grades (ie, WHO grade I or II) were rendered for 104 (65%) of the 161 cases; nearly half were histolo-

gically designated as pilocytic astrocytoma, whereas all remaining astrocytomatous subtypes were represented in the remaining half [1]. In the meta-analysis of Larouche et al of IBTs, 58 (73%) of 79 graded cases were “low” [2]. The latter were most commonly associated with the hypothala- mus/optic pathway, a finding that would usually be most consistent with pilocytic astrocytoma (or potentially its so-called infantile variant, pilomyxoid astrocytoma) versus a diffusely infiltrating WHO grade II astrocytoma [9].

Advances in central nervous system (CNS) tumor classi- fication have dramatically evolved from the early pioneering work of Harvey Cushing and Percival Bailey to the WHO 2007 schema that is most widely accepted today [5,10,11]. ATRTs typify the evolution of a how a particular neoplasm is classified and emphasize the need for expert neuropathologic analysis of historic cases. First detailed in 1996 by Rorke et al [6] as a highly aggressive form of neoplasia that pri- marily affects infants, ATRTs were eventually found to harbor genetic alterations of the INI-1 at 22q11.2 [6,12]. Subsequently, Judkins et al [13] developed an immunohis- tochemical (IHC) stain against the INI-1 protein product to assist in the identification of ATRTs. Detecting the loss of staining of tumor nuclei via INI-1 IHC is now considered virtually diagnostic and has become a standard procedure in the workup of ATRTs and their common mimics that include medulloblastoma, supratentorial PNET (sPNET), and choroid plexus carcinoma (CPC) [14,15].

The long-term functional outcomes for survivors of an IBT, and the determinants thereof, are particularly poorly understood. Families of the newly diagnosed are often left with many challenging questions ranging from initial management (eg, “Is surgery appropriate?”) to their child's eventual quality of life. The primary goal of our main cli- nically based study was to elucidate the long-term prognos- tic factors for infants with brain tumors; these results have been presented elsewhere [16]. As part of the foregoing study, the initial tumor diagnoses were reexamined by an expert pediatric neuropathologist, and the details of this histologic review are the subject of this publication.

2. Methods

A retrospective clinicopathologic review of all IBTs treated at British Columbia's Children's Hospital (BCCH) from 1982 to 2005 was undertaken after appropriate local Research Ethics Board approval (no. H10-01253). In this study, “infants” were defined as patients younger than 1 year at the time of diagnosis. Individual treatments, including surgical intervention, chemotherapy, and radiation therapy, were recorded. The extent of surgical resection was based on the contrast-enhanced immediate postoperative magnetic resonance imaging or computed tomographic (CT) scan and was recorded as “gross total” (GTR, 100%), “near total” (NTR, b100% but N90%), “subtotal” (STR, b90%), or biopsy only. Cases occurring after 2005 were excluded from

Table 1 Clinicopathologic summary of n = 35 infant brain tumors

Study no. a

Age b, sex

Review Dx, grade c Original Dx, grade d Site Sx e CTx RTx Status f

6 11, F Pilocytic astrocytoma, I

Optic/hypothalamic glioma, WHO I

Hypothalamus/CN2 STR X X D, 11 mo

37 11, M Pilocytic astrocytoma, I

Astrocytoma, low grade

Optic nerve STR Ac, V, E, Ma 5040 cGy in 28 Fr; cranial

D, 11 y

35 11, F Pilomyxoid astrocytoma, II

Astrocytoma, low grade

Suprasellar NTR Da, V X D, 2 y

5 2, M Low-grade astrocytoma, I/II

Astrocytoma, low grade

Cervicomedullary NTR X 5500 cGy × 30 Fr; craniospinal

D, 7 y

22 6, F Low-grade astrocytoma, I/II

Astrocytoma, low grade

Hypothalamic B CCG 9952 # X D, 5 y

23 6, M Low-grade astrocytoma I/II

Astrocytoma, low grade

Subependymal B COG A9952 reg B then A

X A, 9 y

38 7, F Glioblastoma, IV

Glioblastoma, small cell

L frontoparietal B X X D, 8 mo

9 2, F ATRT, IV Ependymoma, malignant

Posterior fossa STR 8 in 1 reg g 5400 cGy; CS A, 25 y

24 7, F ATRT, IV Medulloblastoma Posterior fossa GTR HS III then HS II reg A #

X D, 19 mo.

31 9, M ATRT, IV ATRT Posterior fossa NTR X X D, 11 mo 32 11, F ATRT, IV Medulloblastoma Cerebellum STR HS III X D, 21 mo 34 11, F ATRT, IV Pineoblastoma Pineal gland STR X X D, 16 mo 39 3, F ATRT, IV Medulloblastoma Posterior fossa NTR CCG 9921 X D, 10 mo 2 2, F CPP-atypical, II CPP-atypical Intraventricular GTR X X A, 11 y 14 3, F CPP, I CPP Pineal GTR X X A, 18 y 25 6, M CPP, I CPP 3rd ventricle GTR X X A, 15 y 29 6, F CPP, I CPP Choroid plexus,

NOS GTR X X A, 18 y

40 6, M CPP, I CPP Lateral ventricle GTR X X A,17 y 19 4, F Ependymoma, II Ependymoma,

focal anaplasia Intraventricular GTR CIS, V, E, C,

MTX # X D, 14 mo

7 3, M Anaplastic ependymoma, III

Ependymoma Posterior fossa Au X X D, 2 mo

10 2, F Anaplastic ependymoma, III

Ependymoma, malignant

Posterior fossa GTR HS II # 5400 cGy; local A, 9 y

27 7, M Anaplastic ependymoma, III

Ependymoma Posterior fossa GTR CCG 9921 reg A

5400 cGy in 30 Fr D, 3 y

1 b1, M Immature teratoma Immature teratoma Cerebral cortex STR X X D, b1 mo 12 3, F Immature teratoma Teratoma (benign) Suprasellar STR X X A, 16 y 16 3, M Immature teratoma Teratoma-immature,

grade II Brain, NOS GTR X X A, 16 y

15 4, F Mature teratoma Teratoma, with focal GNB areas

Intraventricular GTR X X A, 17 y

3 b1, M Supratentorial/ central PNET, IV

PNET with neuronal differentiation

Left frontal lobe NTR h POG 9233; HS II #

X A, 6 y

17 4, M Supratentorial/ central PNET, IV

PNET Right temporal lobe

GTR CCG 9921 reg B

X D, 9 mo

13 2, M Ganglioglioma, I Ganglioglioma Temporal lobe STR X X A, 11 y 30 8, M Ganglioglioma, I Ganglioglioma Temporal lobe GTR X X A, 17 y

1670 C. Dunham et al.

Table 1 (continued)

Study no. a

Age b, sex

Review Dx, grade c Original Dx, grade d Site Sx e CTx RTx Status f

18 4, F DTI, I PNET Frontoparietal GTR 8 in 1 reg X A, 5 y 33 11, M DTI, I Astrocytoma,

low grade temporal lobe GTR X X A, 15 y

8 2, F MBEN, IV Medulloblastoma Cerebellum GTR X X D, 2 mo 28 6, M Adamantinomatous

CPG, I Adamantinomatous CPG

(Suprasellar) NTR X X A, 14 y

26 6, M Malignancy NOS, high

Disseminated CNS tumor, NOS

NA CSF X X D, 7 mo

Abbreviations: A, alive; Ac, Actinomycin-d; Au, autopsy; B, biopsy; C, cyclophosphamide; carbo, carboplatin; cGy, centigray; CSF, cerebrospinal fluid; CCG, Children's Cancer Group; CCNU, lomustine; CIS, cisplatin; COG, Children's Oncology Group; CPG, craniopharyngioma; CTx, chemotherapy; D, dead; Da, dactinomycin; Dx, diagnosis; F, female; Fr, fraction; GNB, ganglioneuroblastomatous; GTR, gross total resection; HS, HeadStart; M, male; Ma, mannitol; MBEN, medulloblastoma with extensive nodularity; MTX, methotrexate; NA, not applicable; NTR, near total resection; P, procarbazine; PCV, procarbazine + CCNU + vincristine; PNET, primitive neuroectodermal tumor; R, resection; reg, regimen; RTx, radiotherapy; STR, subtotal resection; Sx, surgery; V, vincristine; X, not received; ∼, approximately; #, stem cell transplant.

a The cases were assigned a number in a previous clinical study [16] and are maintained in the present study. b Age in months. c World Health Organization grade. d Grade listed if provided. e Please see text for definitions of extent of surgical resection. f Age at death or last follow-up if alive g Vincristine, methylprednisolone, procarbazine, CCNU, hydroxyurea, cytosine arabinoside, cis-platinum and cyclophosphamide. h Subsequent surgeries in 2006 for recurrence and 2011 for epilepsy.

1671Infant brain tumors

further analysis to facilitate assessment of outcomes. For the purposes of this neuropathologic study and completeness, the outcome data recorded herein were limited to the age at last follow-up if alive, or the age at death.

All original pathologic materials, including reports and slides, were reviewed by an expert pediatric neuropathologist (C. D.). Initial diagnoses were recorded. Cases wherein the original slides/paraffin block could not be located were excluded from further analysis. In addition, cases wherein the pathologic diagnosis was incompatible with neoplasia were also excluded. Review diagnoses were based on the WHO 2007 classification of CNS tumors and were made in a blinded fashion with respect to the outcome data [5]. Notably, the 2007 version of the WHO classification of brain tumors does not assign grades to intracranial germ cell tumors (GCTs), including teratomas [5]. Nonetheless, for the purposes of outcome assessment in our main clinically based study, an attempt was made to roughly grade GCTs into “low-grade” (germinoma and “mature” teratoma) and “high-grade” (all other GCTs including “immature” terato- ma) groups based on the existing literature [5,17,18].

All cases that were considered to represent a “small round blue cell tumor” (ie, those CNS tumors residing within the embryonal category) were assessed by INI-1 IHC to evaluate for the possibility of an ATRT. Purified murine Anti-INI-1 antibody (BD Transduction Laboratories, Mississauga, Ontario, Canada) was used at a dilution of 1:100, and slides were stained using the Ventana BenchMark XT automated stainer (Ventana Medical Systems, Inc, Tucson, AZ). Antigen retrieval was performed as per Ventana's protocol. Loss of the normally ubiquitous INI-1 nuclear staining was

considered to be truly negative only if appropriate internal positive control staining (ie, endothelial cells and lympho- cytes) was present within the tumor.

3. Results

A summary of the clinicopathologic data is provided in Table 1. Forty cases were originally identified as part of the main study, of which 5 cases were excluded from further analysis. Thirty-five IBTs were thus available for neuro- pathologic review. There were 18 females and 17 males. The patients were between 3 days and 11 months of age (mean age, 5.4 months). Of 35, 25 (71%) were supratentorial in location. Of 35 patients, 30 (86%) patients received a form of surgical resection, including 16 GTRs, 6 NTRs, and 8 STRs (see Table 1). There were 3 biopsies, and in a single case, only cerebrospinal fluid was available for histologic exami- nation. There was 1 case with only autopsy tissue available for review. Of 35, 14 (40%) received chemotherapy, and 5 (14%) received radiation therapy. Survival data are reported in Table 1 (please note that the detailed survival data and analysis thereof have been presented elsewhere [16]).

After neuropathologic review of the 35 cases, 17 were deemed high grade, and 18, low grade. The most common diagnosis was “astrocytoma” in 7 (20%) of 35; of these, there was a single high-grade tumor (case 38, glioblastoma). The remaining 6 cases were low grade and included 2 pilocytic, 1 pilomyxoid, and 3 “low-grade astrocytomas not otherwise specified (NOS)” (Fig. 1). The next most common diagnosis, and the most common individual diagnosis, was ATRT in

Fig. 1 A and B, Case 35, pilomyxoid astrocytoma. Note the monotonous cell population that is embedded in a prominent mucoid matrix. Tumor cell angiocentricity is conspicuous (hema- toxylin and eosin, original magnification ×100 [A] and ×400 [B]).

Fig. 2 Case 32, ATRT. A, A “small round blue cell” phenotype predominates in this field (the arrows mark mitotic figures) (hematoxylin and eosin, original magnification ×400). B, INI-1 immunohistochemistry demonstrates negativity in tumor nuclei in the context of appropriate internal control positivity within endothelial cells (original magnification, ×400).

1672 C. Dunham et al.

6 (17%) of 35. In each of these 6 cases, INI-1 IHC revealed a lack of staining in tumor nuclei. Generally, a “small round blue cell” phenotype predominated, and rhabdoid cells were rare or absent in these ATRTs (Fig. 2). ATRT was the most common malignant/high-grade form of infant brain neopla- sia overall. The next most common diagnosis was CPP in 5 (14%) of 35; although one of these tumors was deemed to meet “atypical” WHO criteria, no cases of CPC were iden- tified. Ependymoma was diagnosed in 4 (11%) of 35 cases, of which 3 were considered “anaplastic.” Likewise, teratoma was noted in 4 (11%) of 35 of cases; among these, 3 were considered “immature” based on the presence of fetal-type tissues. Two cases each of sPNET, ganglioglioma, and DTI were seen (Fig. 3). With regard to the latter, additional IHC was not performed to confirm a “neuronal” component. Finally, single cases of “medulloblastoma with extensive nodularity,” adamantinomatous craniopharyngioma, and “malignancy NOS” were diagnosed.

The initial tumor diagnoses and grades were compared with those arising from neuropathologic review. A different

diagnosis was rendered in 8 (23%) of 35 of cases. The most common revised diagnosis was ATRT in 5 instances; 4 of 5 cases were initially diagnosed as a form of central PNET (including 2 sPNET, 1 medulloblastoma, and 1 pineoblas- toma), and 1 case was initially diagnosed as malignant ependymoma (case 9). Therefore, of the 6 ATRTs identified in total after review, only 1 case initially carried this diagnosis.

Of these 6 ATRTs identified on review, the clinicopath- ologic details from case 9 were especially unusual. This girl presented at 2 months of age with a rapidly enlarging head, and CT scanning revealed a large, partially calcified fourth ventricular tumor causing hydrocephalus. An STR was obtained, and the “8 in 1” chemotherapeutic regimen was undertaken (see Table 1). The original tumor diagnosis was “malignant ependymoma,” but on review, a diagnosis of ATRT was rendered (Fig. 4A-C). Rosettes (including “true” and perivascular forms) were absent. Although solid areas were noted, trabeculae and small microcysts were appreciated

Fig. 3 Case 18, DTI. A, Several plump gemistocytic tumor cells are admixed with a mesenchymal spindled-cell population (hematoxylin and eosin, original magnification ×200). B, Mitotically active, primitive, “small round blue cells” focally mimic PNET (hematoxylin and eosin, original magnification ×400). C, Immunostaining highlights the gemistocytic tumor cells (glial fibrillary acidic protein, original magnification ×200). D, A dense mesenchymal element situated between gemistocytes is highlighted green via Masson's Trichrome stain (original magnification ×200).

1673Infant brain tumors

among this mitotically active tumor. A second surgery was performed at 4 years of age for tumor recurrence. At 19 years of age, a cervical spinal cord tumor was resected at a different facility and was designated as a “benign ependy- moma.” Slides from the latter were retrieved, and the his- tology was reviewed. Not only were classic “perivascular pseudorosettes” not appreciated, but interestingly, INI-1 immunohistochemistry revealed a lack of staining in the tumor nuclei (Fig. 4D-E). At the last follow-up, 25 years after her initial presentation, the patient was considered relatively well aside from some residual difficulties with balance, hearing, and a mild learning disability. The patient's younger sister was also diagnosed with what was deemed to be a histologically identical left cerebellopontine angle tumor (ie, malignant ependymoma) at 2 years of age. This younger sister's tumor displayed monosomy 22. Neuropathologic review of this case was most notable for lack of INI-1 immunostaining in the tumor nuclei. Unfor- tunately, this sister died 14 years later because of a sarcoma

of the neck that was presumably radiation induced. Finally, a maternal male cousin was diagnosed with “ependymoma” of the right lateral ventricle at 13 years of age; despite surgery and radiotherapy, this cousin died 2 years later secondary to tumor progression. The cousin's tumor was not available for review. Notably, these 3 “ependymomas” were previously reported [19].

Aside from the 5 ATRTs, there were 3 other cases where the neuropathologic review diagnosis was discrepant with the initial diagnosis. In 2 cases, DTI was diagnosed on review; initial diagnoses included sPNET and “low-grade astrocyto- ma, NOS.” The remaining case, which was initially diagnosed as “low-grade astrocytoma, NOS,” was more precisely designated as pilomyxoid astrocytoma upon review.

Tumor grade was revised in 6 (17%) of 35 cases. The most significant change involved case 18, wherein the initial diagnosis of sPNET was changed to DTI, and accordingly, the WHO grade was reduced from IV to I (for further grade changes, please see Table 1).

Fig. 4 Case 9, originally diagnosed as “malignant ependymoma” and rediagnosed as “ATRT” on review. A, Note the somewhat trabecular and microcystic architecture of this tumor (hematoxylin and eosin, original magnification ×100). B, On high magnification, numerous mitoses are noted among primitive tumor cells (hematoxylin and eosin, original magnification ×400). C, INI-1 immunohistochemistry demonstrates negativity in tumor nuclei in the context of appropriate internal control positivity within endothelial cells (original magnification ×400). D, Scattered acellular foci are seen among the tumor cells that display round to oval nuclei and little cytoplasm; this section is taken from the cervical spinal cord tumor resected at 19 years of age and originally called benign ependymoma (original magnification ×200). E, INI-1 immunohistochemistry performed on the tumor shown in 4D again demonstrates INI-1 immunonegativity in tumor nuclei, whereas nonneoplastic endothelial cells and lymphocytes retain nuclear positivity (original magnification ×400).

1674 C. Dunham et al.

1675Infant brain tumors

4. Discussion

Our retrospective analysis of 35 IBTs included a detailed neuropathologic review that used standard staining practices, including INI-1 IHC. Upon review, the most common tumor category identified was “astrocytoma” and featured primarily low-grade entities. ATRT was the second most frequent tumor and the most common “malignant”/high-grade entity. Although the size of the present population-based study is somewhat limited, the findings herein raise some doubt as to whether medulloblastoma is truly the most common infant brain malignancy [2].

Our ATRT incidence of 17% upon review was substantially higher than that documented in prior reports. Three studies in the meta-analysis of Larouche et al [2] included ATRT cases, but they amounted to less than 4% (4/117) in total. The study of Geyer et al [20] of malignant IBTs (n = 72) featured 12 ATRTs (17%); this percentage is far lower than reported in our series when only malignant cases are accounted for (6/17, or 35%). Serowka et al [21] documented 2 cases of ATRT among their 33 IBTs (6%), as did Mehrotra et al (2/18, or 11%) [12,21,22]. The incidence of ATRT in these recent reviews is likely artifactually low because: (1) both studies failed to use INI-1 IHC to aid in the detection of ATRT and (2) no neuropathologic review of the histology was performed by Mehrotra et al [21] (personal corre- spondence with Dr M. Vassilyadi). Given the frequency of diagnoses that may have mimicked ATRT in these studies, it is reasonable to hypothesize that many ATRTs initially escaped identification in the Mehrotra et al (including 2 each of sPNET, CPC, and medulloblastoma) and Serowka et al (including 5 sPNETs, 3 medulloblastomas, and 2 CPCs) reports [21,22].

Case 9 was a very unusual neoplasm that was originally diagnosed as malignant ependymoma but was reclassified as ATRT on blinded neuropathologic review. Notably, this patient had a sister that was likewise diagnosed at a similar young age. Despite the fact that both siblings had INI-1 immunonegative tumors, their long survival is somewhat unusual for the diagnosis of ATRT [23]. The familial pattern of disease in this case could be accounted for by a germline mutation in INI-1, as is seen “rhabdoid tumor predisposition syndrome,” but the latter is usually associated with very poor prognosis [24]. However, Hasselblatt et al [25] recently reported the loss of nuclear INI-1 immunopositivity in the brain tumor cells of 2 young children who experienced extended survival. The histology in these 2 cases displayed a unique cribriform and trabecular architecture with well-defined epithelial membrane antigen-immunopositive surfaces. Notably, one patient demonstrated a homozygous 4-bp duplication in exon 4 (492duplCCTT) of INI-1 on sequencing. Based on these 2 cases, Hasselblatt et al. suggested the existence of a new entity called “cribriform neuroepithelial tumor” (CRINET). After correlating the pathology in case 9 with

the clinical features (especially the extended survival), the possibility of familial CRINET (which has not been recognized to date) should be considered.

Neuropathologic review led to the revision of 23% (n = 8) of tumor diagnoses. The identification of 5 previously unrecognized ATRTs is attributed to the evolution of this tumor's diagnostic criteria, which routinely implements INI-1 IHC and recognizes a “small round blue cell” pre- dominant phenotype [13,15]. The frequency of diagnostic revision in this study differs markedly from the similar study of Serowka et al [21], who reported only one instance of diagnostic disconcordance in 33 cases (3%). The high degree of diagnostic agreement in Serowka et al after review is especially surprising because: (1) retrospective analysis dates back to 1979, a time since which tumor diagnostic criteria have significantly changed; (2) INI-1 IHC was not used to facilitate diagnosis of ATRT; and (3) there is a complete lack of representation of DTI, pilomyxoid astrocytoma, and likely the most common single IBT, pilocytic astrocytoma (WHO grade I), in their series. In fact, of the 11 astrocytomas diagnosed, an unusually high number, 8, were considered “astrocytoma, WHO grade II.” When closely examined, 6 of these latter 8 were described as having tumor epicenters that would seemingly be more compatible with pilocytic/ pilomyxoid astrocyto- ma (ie, hypothalamus, optic pathway, midbrain, posterior fossa, etc) [5]. Moreover, 4 of these 11 astrocytomas were not histologically reviewed (cases 11, 12, 23, and 24). These observations argue for the importance of histologic reviews of historic IBT cases and implementation of the most current WHO criteria.

After review, the tumor grade was changed in 17% (n = 6) of cases. Likely, the most significant alteration involved the downgrading from WHO IV to I in case 18 (sPNET to DTI), a commonly recognized pitfall that needs to be avoided given the more favorable prognosis associated with DTI [5]. The upgrading of 2 ependymomas (cases 7 and 27) is likely attributable to the progress that has been made in the grading ependymomas. To the latter, it is now generally appreciated that mitotic/proliferative activity, microvascular prolifera- tion, and marked hypercellularity play crucial roles in the grading of ependymomas [26].

In conclusion, this study highlights the importance of expert neuropathologic review of historic IBT cases given the evolution of diagnostic and grading schemes over time. Precise histologic assessment will likely become even more critical in this emerging age of molecularly targeted therapy. The identification of ATRT as potentially the most common malignant IBT is intriguing but requires analyses of addi- tional institutional series.

Acknowledgments

The authors would like to acknowledge Ms Theresa Sturbys for her technical assistance in performing the INI-1

1676 C. Dunham et al.

immunohistochemistry on the pathologic specimens and Ms Mary Metrie for her work in collecting and analyzing the clinical data.

References

[1] Bishop AJ, McDonald MW, Chang AL, Esiashvili N. Infant brain tumors: incidence, survival, and the role of radiation based on Surveil- lance, Epidemiology, and End Results (SEER) data. Int J Radiat Oncol Biol Phys 2012;82:341-7.

[2] Larouche V, Huang A, Bartels U, Bouffet E. Tumors of the central nervous system in the first year of life. Pediatr Blood Cancer 2007;49: 1074-82.

[3] Di Rocco C, Iannelli A, Ceddia A. Intracranial tumors of the first year of life. A cooperative survey of the 1986-1987 Education Committee of the ISPN. Childs Nerv Syst 1991;7:150-3.

[4] Oi S, Matsumoto S, Choi JU, et al. Brain tumors diagnosed in the first year of life in five Far-Eastern countries. Statistical analysis of 307 cases. Childs Nerv Syst 1990;6:79-85.

[5] Louis DN. International Agency for Research on Cancer, World Health Organization. WHO classification of tumours of the central nervous system. Lyon, France: International Agency for Research on Cancer; 2007.

[6] Rorke LB, Packer RJ, Biegel JA. Central nervous system atypical teratoid/rhabdoid tumors of infancy and childhood: definition of an entity. J Neurosurg 1996;85:56-65.

[7] Dougherty MJ, Santi M, Brose MS, et al. Activating mutations in BRAF characterize a spectrum of pediatric low-grade gliomas. Neuro Oncol 2010;12:621-30.

[8] Schindler G, Capper D, Meyer J, et al. Analysis of BRAF V600E mutation in 1,320 nervous system tumors reveals high mutation fre- quencies in pleomorphic xanthoastrocytoma, ganglioglioma and extra- cerebellar pilocytic astrocytoma. Acta Neuropathol 2011;121:397-405.

[9] Johnson MW, Eberhart CG, Perry A, et al. Spectrum of pilomyxoid astrocytomas: intermediate pilomyxoid tumors. Am J Surg Pathol 2010;34:1783-91.

[10] Kunschner LJ. Harvey Cushing and medulloblastoma. Arch Neurol 2002;59:642-5.

[11] Bailey P, Cushing H. A classification of the tumors of the glioma group on a histogenic basis with a correlated study of prognosis. Philadelphia, PA: JB Lippincott Co; 1926. p. 1-174.

[12] Biegel JA, Tan L, Zhang F, Wainwright L, Russo P, Rorke LB. Alterations of the hSNF5/INI1 gene in central nervous system atypical teratoid/rhabdoid tumors and renal and extrarenal rhabdoid tumors. Clin Cancer Res 2002;8:3461-7.

[13] Judkins AR, Mauger J, Ht A, Rorke LB, Biegel JA. Immunohisto- chemical analysis of hSNF5/INI1 in pediatric CNS neoplasms. Am J Surg Pathol 2004;28:644-50.

[14] Bourdeaut F, Freneaux P, Thuille B, et al. hSNF5/INI1-deficient tumours and rhabdoid tumours are convergent but not fully over- lapping entities. J Pathol 2007;211:323-30.

[15] Haberler C, Laggner U, Slavc I, et al. Immunohistochemical analysis of INI1 protein in malignant pediatric CNS tumors: lack of INI1 in atypical teratoid/rhabdoid tumors and in a fraction of primitive neuro- ectodermal tumors without rhabdoid phenotype. Am J Surg Pathol 2006;30:1462-8.

[16] Pillai S, Metrie M, Dunham C, Sargent M, Hukin J, Steinbok P. Intracranial tumors in infants: long-term functional outcome, survival, and its predictors. Childs Nerv Syst 2012;28:547-55.

[17] Lee YH, Park EK, Park YS, Shim KW, Choi JU, Kim DS. Treatment and outcomes of primary intracranial teratoma. Childs Nerv Syst 2009;25:1581-7.

[18] Matsutani M, Sano K, Takakura K, et al. Primary intracranial germ cell tumors: a clinical analysis of 153 histologically verified cases. J Neurosurg 1997;86:446-55.

[19] Savard ML, Gilchrist DM. Ependymomas in two sisters and a maternal male cousin with mosaicism with monosomy 22 in tumour. Pediatr Neurosci 1989;15:80-4.

[20] Geyer JR, Sposto R, Jennings M, et al. Multiagent chemotherapy and deferred radiotherapy in infants with malignant brain tumors: a report from the Children's Cancer Group. J Clin Oncol 2005;23: 7621-31.

[21] Serowka K, Chiu Y, Gonzalez I, et al. Central nervous system (CNS) tumors in the first six months of life: the Children's Hospital Los Angeles experience, 1979-2005. Pediatr Hematol Oncol 2010;27: 90-102.

[22] Mehrotra N, Shamji MF, Vassilyadi M, Ventureyra EC. Intracranial tumors in first year of life: the CHEO experience. Childs Nerv Syst 2009;25:1563-9.

[23] Hilden JM, Meerbaum S, Burger P, et al. Central nervous system atypical teratoid/rhabdoid tumor: results of therapy in children enrolled in a registry. J Clin Oncol 2004;22:2877-84.

[24] Kordes U, Gesk S, Fruhwald MC, et al. Clinical and molecular fea- tures in patients with atypical teratoid rhabdoid tumor or malignant rhabdoid tumor. Genes Chromosomes Cancer 2010;49:176-81.

[25] Hasselblatt M, Oyen F, Gesk S, et al. Cribriform neuroepithelial tumor (CRINET): a nonrhabdoid ventricular tumor with INI1 loss and relatively favorable prognosis. J Neuropathol Exp Neurol 2009;68: 1249-55.

[26] Tihan T, Zhou T, Holmes E, Burger PC, Ozuysal S, Rushing EJ. The prognostic value of histological grading of posterior fossa ependymo- mas in children: a Children's Oncology Group study and a review of prognostic factors. Mod Pathol 2008;21:165-77.