Ewing Sarcoma and Atypical Teratoid Rhabdoid Tumor

Introduction

Ewing sarcoma (ES) (primitive neuroectodermal tumor) and atypical teratoid rhabdoid tumor (ATRT) are high-grade pediatric malignancies. ES arises from bone or soft tissue and is potentially curable, especially if there is no sign of metastatic disease at diagnosis. ¹ ES is typically characterized by translocations involving the EWSR1 locus in 22q11.2, most frequently with FLI1 in 11q24; several other fusion partners have also been identified but are less common. ² Additionally, immunohistochemical (IHC) staining of ES for CD99 typically has a characteristic, distinct membranous pattern of staining. The prognosis of ES is better than that of ATRT in the absence of metastasis, provided good local control can be achieved.

ATRT usually occurs in infants and, unlike ES, most frequently develops within the central nervous system. Renal rhabdoid tumor (RRT) and extrarenal rhabdoid tumor (ERRT) involving the bladder, liver, and other organs are histologically and genetically similar to ATRT. The prognosis for ATRT is very poor and the 6-month, 1-year, and 5-year survival rates for these tumors are 65%, 48.8%, and 28.3%, respectively. ³ ATRT, RRT, and ERRT are associated with mutations of the SMARCB1 (INI1) tumor suppressor gene, mapped to chromosome 22q11.23. Mutation and/or deletion of both SMARCB1 alleles results in absence of the SMARCB1 protein; IHC staining with INI1 is therefore negative in these tumors.^4,5

Although Burger et al. ⁶ reported a case of ATRT in which fluorescence in situ hybridization (FISH) using a paint probe to chromosome 22 suggested rearrangement, we did not find any cases in the literature in which a rearrangement of the EWSR1 locus was detected by FISH in ATRT. A recent paper, however, identified 4 cases (poorly differentiated chordoma, ERRT, myoepithelial carcinoma, and epithelioid sarcoma) with deletions of SMARCB1 and concurrent abnormalities detected by EWSR1 FISH. ⁷ In light of the proximity of EWSR1 and SMARCB1 on 22q (SMARCB1 is 5.4 Mb centromeric to EWSR1), ⁸ as well as the frequent CD99 positivity of both tumors, we evaluated archived cases of ES and ATRT, RRT, and ERRT by both IHC and FISH.

Clinical History

A 15-month-old child presented with a 3-week history of right facial weakness and inability to close her right eye, a 1-week history of right esotropia, and head tilt to the left. CT identified a heterogeneously enhancing exophytic 4.2 × 3.5 × 1.8 cm mass in the right cerebellopontine angle, involving the anterior aspect of the right cerebellar hemisphere, pons, and medulla, and extending into the right auditory canal and the posterior aspect of the right cavernous sinus. Based on the cerebral location of the tumor, age < 3 years, clinicopathologic findings, FISH, and molecular testing (see Results below), a diagnosis of embryonal tumor was made and the decision was made to treat the patient with ATRT therapy. Subsequent sequencing of the SMARCB1 gene further supported the diagnosis of ATRT. Despite aggressive chemotherapy, the child died less than 1 year after diagnosis. Results from subsequent testing performed at autopsy were similar to those at diagnosis.

Study Design and Methods

Hematoxylin and eosin-stained slides and IHC stains from 16 ES and 17 ATRT, RRT, and ERRT cases (not including the index case) were reviewed, and demographic features (e.g., patient age and location of tumor) noted. Results of G-banding, FISH, reverse transcriptase polymerase chain reaction (RT-PCR), and sequencing were reviewed when available. In cases for which no diagnostic FISH results were available, it was subsequently performed with a breakapart probe to the EWSR1 locus (Abbott Molecular, Des Plaines, IL, USA). The 5′ portion of this probe (labeled in SpectrumOrange) extends ∼500 Kb toward the centromere of chromosome 22, and the 3′ portion (labeled in SpectrumGreen) extends ∼1100 Kb distally, toward the 22q telomere; there is a small unlabeled gap between these probes that corresponds to most of the EWSR1 gene itself. ⁹ Results were classified as rearranged (one intact fusion and separate orange and green signals) or not rearranged (2 intact fusion signals). IHC was performed using CD99 (Ventana Roche, clone 013) and INI1 (Cell Marque, clone MRQ-27), if previously stained slides were not available in our files. The staining protocols were optimized using the manufacturers’ inserts.^10,11 CD99 staining was characterized as nuclear or cytoplasmic and focal or diffuse; INI1 positivity was defined as diffuse nuclear staining and negativity as absent staining.

Results

Biopsy of the cerebellar portion of the mass revealed a cellular tumor with mild pleomorphism, nuclei surrounded by cytoplasmic clearing, and occasional nucleoli (Figure 1(A)). Initial IHC staining with CD99 showed membranous positivity (Figure 1(B)); INI1 was difficult to interpret due to numerous background lymphocytes (Figure 1(C) and (D)). Because the differential diagnosis included ES, FISH was performed with a breakapart probe to the EWSR1 locus, which showed 99.6% of interphase cells to have 1–2 intact EWSR1 signals plus an additional 3′ (telomeric) signal (Figure 2). As there was no apparent size difference between the separate green signal and the green component of the fusion signal, it was concluded that the rearrangement most likely disrupted the EWSR1 gene itself. RT-PCR analysis was negative for the presence of an EWSR1/FLI1 gene fusion; no additional RT-PCR for other EWSR1 partners was performed. IHC staining for INI1 was repeated and more clearly showed the tumor cells to be negative. Sequencing of SMARCB1 exons 1–9 and gene deletion/duplication testing showed a deletion within the long arm of one copy of chromosome 22 including the SMARCB1 locus; no second deletion or mutation was identified. At autopsy, the tumor showed distinct membranous staining with CD99, INI1 negativity, and variable staining with pancytokeratin, desmin, and glial fibrillary acidic protein. Repeat FISH using the Abbot breakapart probe to EWSR1 showed 85.5% of cells with a similar signal pattern indicative of an unbalanced rearrangement, with 1–2 intact EWSR1 signals and an additional 3′ signal.

OPEN IN VIEWER

Figure 1.

All photos taken at 40×. (A) H&E of tumor at diagnosis. (B) CD99 staining. (C) INI1 staining demonstrating positive staining in intervening lymphocytes. (D) CD45 staining highlighting lymphocytes.

OPEN IN VIEWER

Figure 2.

Fluorescence in situ hybridization of index tumor at diagnosis with EWSR1 probe showing 1–2 intact EWSR1 signals (red = 5′ [centromeric], green = 3′ [telomeric]) plus an additional 3′ (telomeric) signal.

All 16 of the ES cases reviewed were positive for INI1 nuclear staining and for CD99 (Table 1). Although most cases showed the typical membranous pattern of CD99 positivity, 4 cases had “dirty” staining for CD99 (in which it was difficult to determine if the staining was membranous or cytoplasmic) and 1 case had cytoplasmic CD99 staining. FISH was positive for EWSR1 rearrangement in all but 2 of these cases; one of these cases was positive by RT-PCR for an EWSR1-FLI1 gene fusion. This apparent discordance of the FISH and RT-PCR results may reflect a cytogenetically cryptic rearrangement such as a small insertion that would not be detectable by FISH. All 17 ATRT, RRT, and ERRT cases showed no INI1 staining. CD99 staining varied from completely negative to positive in both membranous and cytoplasmic patterns. None of the ATRT, RRT, or ERRT cases showed EWSR1 rearrangement by FISH (Table 2).

OPEN IN VIEWER

Table 1.

Ewing Sarcoma Data.

Case	Age	Location of Tumor	INI1	CD99	FISH
1	8 year	chest mass	+	+ mem dirty	R
2	16 year	pleural mass	+	+ mem	R
3	9 year	chest wall	+	+ mem dirty	R
4	17 year	Lung	+	+ mem	R
5	8 year	Hemithorax	+	+ mem dirty	NR
6	17 year	Pelvis	+	+ mem	R
7	16 year	chest wall	+	+ mem	R
8	6 year	Lung	+	+ focal cyto	R
9	22 year	Lung	+	+ mem	R
10	14 year	rib	+	+ mem	R
11	8 year	chest mass	+	+ mem	NR
12	11 year	skull mass	+	+ mem	R
13	2 year	axilla	+	+ mem	R
14	17 year	paraspinal	+	+ mem dirty	R
15	5 year	chest wall	+	+ mem	R
16	12 year	chest mass	+	+ mem	R

OPEN IN VIEWER

Table 2.

Atypical TeratoidRhabdoid Tumor and Rhabdoid Tumor Data.

Case	Age	Location of Tumor	INI1	CD99	FISH for EWSR1
1.	4 weeks	neck	neg	focal pos mem and dot	NR
2.	14 year	pelvis	neg	neg	NR
3.	4 months	brain	neg	focal cyto and mem	NR
4.	3 months	kidney	neg	mem	NR
5.	10 weeks	kidney	neg	neg	NR
6.	9 year	neck	neg	neg	FISH for BCR 22q11.2 showed 22q deletion or monosomy 22 with loss of both marker signals
7.	8 months	posterior fossa	neg	focal cyto and mem	NR
8.	1 year	pineal	neg	focal mem	NR
9.	1 year	brain	neg	rare pos mem	NR
10.	8 months	brain	neg	pos mem	NR
11.	2 year	paraspinal	neg	focal weak mem	FISH for BCR 22q11.2 not rearranged
12.	3 months	liver	neg	neg	NR
13.	5 year	brain	neg	neg	NR
14.	5 months	third ventricle	neg	rare pos	NR
15.	1 year	thalamic	neg	pos mem	NR
16.	5 weeks	kidney	neg	pos mem	NR
17.	11 weeks	sacrum	neg	neg	NR

INI1, INI1 immunstaining; neg, negative; pos, positive; mem, membranous staining; cyto, cytoplasmic staining; NR, EWSR1not rearranged.

Discussion

This case is unusual in that FISH showed a signal pattern consistent with rearrangement of EWSR1, with gain of an additional 3′ (telomeric) signal. Although abnormalities in FISH have recently been reported in poorly differentiated chordoma, ERRT, myoepithelial carcinoma, and epithelioid sarcoma, ⁷ this is the first reported case of an ATRT with an apparent unbalanced translocation in EWSR1 by FISH. Unlike the majority of ATRTs,^12,13 sequencing of the present tumor revealed only a single deletion in the long arm of chromosome 22, with no concomitant mutation of the remaining SMARCB1 homolog. Sequencing may have revealed a mutation only in one of the alleles if the second mutation existed within a non-coding region. Zhang et al. ¹⁴ have previously shown that increased methylation of the promoter does not result in decreased SMARCB1 protein production. Alternatively, it is possible that a gene up- or downstream of SMARCB1 affects its expression in some cases. In one study of 79 brain tumors, sequencing uncovered mutations in the coding region of 63 (80%) central nervous system tumors; although RRTs had similar rates of sequencing mutations when compared to brain tumors, the percentage in ERRTs has been found to be smaller. ¹³ In a study of rhabdoid tumors in which both SMARCB1 mutations were identified, the mutations were the result of deletions, duplications, frame shifts, point mutations, splice site mutations, and/or loss of heterozygosity. ¹⁵

Although our study did not identify any additional cases of ATRT, RRT, or ERRT with aberrant FISH results using the EWSR1 probe, results from this patient’s tumor show that the EWSR1 probe in cases of high-grade central nervous system tumors in very young children may yield abnormal results of unclear significance. In light of the fact that SMARCB1 is just 5.4 Mb centromeric to EWSR1, it is possible that a structural abnormality of 22q undetectable by G-banding resulted in both deletion of part of the SMARCB1 gene and rearrangement of the EWSR1 locus. In the present case, FISH showed gain of an additional copy of the 3′ (telomeric) signal, suggestive of an unbalanced translocation with a breakpoint within EWSR1; of note, it is the 5′ (centromeric) portion of EWSR1, not the 3′, that is the critical partner in the various translocations associated with ES. Additionally, as EWSR1 is telomeric to SMARCB1, this rearrangement would not be expected to result in a similar gain of SMARCB1.

Given the small sample size, it is not unexpected that we did not identify another ES or ATRT, RRT, or ERRT case with similar abnormalities. Because SMARCB1 is a tumor suppressor gene, it is unusual to identify only a deletion without a concomitant mutation of the second homolog. Nonetheless, the CNS location, early age of onset, INI1 negativity, and rapid progression despite aggressive therapy is most consistent with a diagnosis of ATRT. We show that CD99 positivity is not uncommon in ATRT, RRT, and ERRT and should not be interpreted as a specific marker for ES, especially in the setting of a tumor that lacks INI1 staining.

Although the diagnosis of ATRT is frequently considered in very young patients with central nervous system tumors, RRT and especially ERRT are often lower in the differential diagnosis of many pathologists. The histology of RRT and ERRT can be quite similar to that of ES, and given the potential CD99 positivity of all of these tumors, RRT and ERRT could be overlooked. Together with aberrant FISH results such as those in our case, the similarities among these tumors could result in misdiagnosis. The use of INI1 immunohistochemistry should be considered in young pediatric patients with tumors suspicious for ES to exclude the diagnosis of RRT and ERRT.