Spontaneous Regression of Atypical Teratoid Rhabdoid Tumor Without Therapy in a Patient With Uncommon Regional Inactivation of SMARCB1 ( hSNF5/INI1 )

Introduction

Atypical teratoid/rhabdoid tumor (ATRT) is the most common malignant central nervous system tumor in children less than 1 year of age. ¹ Loss of nuclear staining for INI1 is nearly pathognomonic for ATRT ² and is caused by the inactivation of SMARCB1 (hSNF5/INI1), with the remaining minority of cases having loss of function of SMARCA4, both components of the switch/sucrose nonfermenting (SWI/SNF) chromatin-remodeling complex. ³ Infants with this tumor have a poor prognosis even with intensive multimodal therapy. ⁴ We report a case of congenital ATRT with regional loss of nuclear immunostaining for INI1, and with fluorescence in situ hybridization (FISH) and DNA sequencing revealing SMARCB1 inactivation only in the INI1-negative cells. To the best of our knowledge, we present the first report of a patient with congenital ATRT with molecular heterogeneity for SMARCB1 inactivation who had spontaneous tumor regression without therapy and remains disease-free over 4 years after diagnosis.

Case Report

The patient was a female infant of dizygotic twin gestation who, following detection of ventriculomegaly on routine prenatal ultrasonography, was found to have a congenital right cerebral hemispheric tumor on fetal MRI at 32 weeks’ gestation. An MRI of the brain, following delivery at 36 weeks’ gestation, showed a complex solid and cystic mass, centered in the right frontal lobe (Figure 1(A)). A biopsy of the tumor was performed at 5 days of life. Following intraoperative consultation, tumor tissue was paraffin-embedded, formalin-fixed, and stained with hematoxylin and eosin (H&E). Immunohistochemical staining was performed with glial fibrillary acidic protein (GFAP), epithelial membrane antigen (EMA), vimentin, pancytokeratin, INI1, and SMARCA4/BRG1. Laser microdissection was performed using the Arcturus XT system, on INI1-stained slides, to separate tumor cells that were INI1-negative from those that had intact nuclear staining. For both tumor cell populations, reverse transcription polymerase chain reaction DNA amplification followed by DNA sequencing of the INI1 locus on chromosome 22, as well as FISH on formalin-fixed paraffin-embedded tissue sections (5 µm) using the Abbott Molecular pretreatment kit, were performed. Two test probes specific to the SMARCB1 locus on chromosome 22 and a reference probe telomeric to chromosome 22 were used. A SMARCB1 test probe to reference probe ratio of less than 0.8 was considered a deletion.

OPEN IN VIEWER

Figure 1.

Imaging findings. T1-weighted MRI of the brain at diagnosis (A), showed a complex solid and cystic mass centered in the right frontal lobe, and 5 months later (B) showed near complete resolution of the tumor, with loculated hydrocephalus and severe enlargement of the lateral ventricle.

Intraoperative squash preparations of the biopsy showed both papillary architecture and collections of discohesive cells. The cytology was variable, with cells that had eosinophilic cytoplasmic globules with eccentric nuclei, and others that had a vesicular nuclear chromatin pattern (Figure 2(A)). Mitotic activity was frequent and cell wrapping was present. The histology showed the same rhabdoid cells with eosinophilic cytoplasm and eccentric nuclei (Figure 2(B)) as well as frequent perivascular arrangements (Figure 2(C) and (D)). Cytoplasmic vacuolization was also seen. Immunohistochemical staining for GFAP, EMA, and pancytokeratin were all focally positive. The MIB-1 labeling index was markedly elevated and focally greater than 90%. Immunohistochemical staining for INI1 showed heterogeneous staining with regions of retained nuclear positivity adjacent to those that were negative (Figure 2(E)). Multiple additional INI1-stained slides also showed the same pattern of immunostaining. Immunohistochemical staining for vimentin was homogenously positive, even in the areas of INI1 negativity. Careful review of the H&E-stained slides comparing the 2 populations identified on INI1 staining did not yield any distinguishing histologic features between the 2 populations (Figure 2 (C) and (D)), nor were any identifiable differences noted between the 2 populations on additional review of all other immunohistochemical stains. Immunostaining for SMARCA4/BRG1 showed homogenous nuclear staining.

OPEN IN VIEWER

Figure 2.

Histopathology. Hematoxylin and eosin smears (A) and formalin-fixed paraffin-embedded tissue sections (B) revealed collections of cells with eosinophilic globules and eccentric nuclei. Frequent perivascular arrangements were seen, both in regions of retained (C) and deficient (D) INI1 staining. Immunohistochemical staining for INI1 showed heterogeneous staining, with regions of retained nuclear positivity adjacent to those that lack nuclear positivity (E). Note: Color image is available in the online version of the article.

In a region of nuclear INI1 immunoreactivity, FISH showed an appropriate ratio of the SMARCB1 test probe to the reference probe, consistent with an intact SMARCB1 locus (Figure 3(A)), and DNA sequencing demonstrated the presence of the wild-type SMARCB1 gene (Figure 3(A), top inset). However, in a region of negative INI1 immunostaining, FISH showed a test to reference probe ratio of 0.6, representing heterozygous loss of SMARCB1 (Figure 3(B)), and DNA sequencing of the INI1 locus on chromosome 22 showed a point mutation in exon 7 that substitutes a thymine for cytosine, resulting in a change from coding for glutamic acid to a stop codon (SMARCB1:NM_003073:c.967C > T; p.Q323*) (Figure 3(B), top inset).

OPEN IN VIEWER

Figure 3.

Molecular analysis. In an area with intact INI1 nuclear staining (A), FISH testing showed 2 copies of SMARCB1 (reference probe is orange and test probe is green), consistent with an intact SMARCB1 locus, and DNA sequencing demonstrated the wild-type SMARCB1 gene (A—top inset), while in an area that was INI1-negative (B), FISH showed only 1 copy of SMARCB1 (white arrow), representing heterozygous loss of SMARCB1 and DNA sequencing of the INI1 locus on chromosome 22 revealed a point mutation in exon 7 (B—top inset). Note: Color image is available in the online version of the article.

The patient’s tumor encased several major arteries and bled briskly with the initial biopsy, rendering further surgical resection highly risky. Given the poor prognosis following a histologic diagnosis of ATRT, the parents declined further debulking surgery and chemotherapy and pursued hospice care. At 5 months of age, without any treatment for her tumor, repeat MRI showed complete resolution of solid tumor, with multiseptated cysts, loculated hydrocephalus, and severe enlargement of the lateral ventricles with evidence of old hemorrhage (Figure 1(B)). A ventriculo-peritoneal shunt was placed, and biopsies of the cyst walls were performed, which showed only fibrous tissue with a dense macrophage infiltrate and no viable tumor. Germline testing was negative for deletion/duplication of the INI1/SMARCB1 locus in chromosome band 22q11.2, and coding sequence mutations in exons 1 to 9 of the promoter region of the INI1 gene. At last follow-up at 4 years of age, she had no evidence of tumor recurrence on her last MRI of the brain and spine. Due to intraventricular hemorrhage, mass effect from loculated cysts, chronic cortical thinning, and hydrocephalus, the patient has medically refractory seizures, disabling left hemiparesis, multiple cranial neuropathies, and severe global developmental delay.

Discussion

In the WHO Classification of Tumors of the Central Nervous System (2016, Revised 4th Edition), the loss of nuclear staining for INI1 is essential to making the diagnosis of ATRT. ⁵ These tumors are caused most commonly by inactivation of the SMARCB1 (hSNF5/INI1) tumor suppressor gene^6,7 and in a minority of cases by SMARCA4 inactivation. ⁸ A component of the SWI/SNF chromatin-remodeling complex, the SMARCB1 gene is located on chromosome 22q11.2 and is inactivated by a variety of mechanisms including deletions, loss of heterozygosity, and truncating frameshift or nonsense mutations. ⁹ Other tumor entities reported to have heterogeneous loss of INI-1 protein expression, include cribriform neuroepithelial tumor (CRINET)^10,11 and rhabdoid glioblastoma multiforme (GBM). ¹² ATRT has also been reported to evolve from other tumor entities like optic pathway ganglioglioma and pleomorphic xanthoastrocytoma.^13,14 However, our patient’s tumor had rhabdoid cells typical of ATRT, with the absence of cribriform strands and trabeculae composed of small undifferentiated cells, as is typical of CRINET, ¹⁰ and a lack of cells typical of GBM or other gliomas. ¹² Although our patient’s tumor lacked the typical histopathological features of CRINET, the favorable clinical course was more consistent with that tumor entity, than ATRT, which is an aggressive tumor with poor outcomes despite multimodal therapy, with median progression-free survival less than 6 months and overall survival less than 1 year. ¹⁵ To our knowledge, there are no previous reports of the spontaneous resolution of any INI1 deficient tumors (ATRT and CRINET). One hypothesis for this is extensive tumor ischemia and infarction secondary to profuse bleeding during surgery. A more plausible explanation is that despite the presence of high-grade histologic features suggestive of an aggressive biology, this heterogeneously INI1-deficient tumor might represent a yet undefined entity that lies at the favorable end of a spectrum of INI1 deficient tumors.