Additional Considerations on Aberrant BRG1 (SMARCA4) Expression in Small Cell Carcinoma of the Ovary, Hypercalcemic Type (SCCOHT)

To the Editor:

We have read with great interest the study by Mazibrada et al¹ (American Journal of Surgical Pathology, November 2023) reporting 7 cases of small cell carcinoma of the ovary of hypercalcemic type (SCCOHT) with completely retained BRG1 (SMARCA4) expression by immunohistochemistry (IHC), as opposed to the complete loss of expression which is typically seen in SCCOHT and is used to confirm the diagnosis.^2,3

The authors also mention that, in addition to the retained and strong expression pattern which was the focus of their study, occasional SCCOHT may exhibit weak nuclear staining, at a lower intensity than internal positive controls. We fully agree with the authors that this staining pattern should be upfront recognized as aberrant, as it is indicative of an underlying SMARCA4 gene alteration. We would like to further emphasize this point, since in our experience, such aberrant staining occasionally still remains confusing to pathologists, possibly because BRG1 IHC was initially presented as an “all-or-nothing” type of staining.

We have encountered 3 cases of SCCOHT with aberrant BRG1 staining (1 with moderate expression and 2 with weak). IHC was performed using a Bond III automated immunostainer (Leica Microsystems) and the Bond Polymer Refine detection system, with the anti-BRG1 antibody clone EPR3912 (Abcam, 1:200 dilution, pH9 retrieval, 60 min of incubation) and anti-BRM antibody clone D9E8B (Cell Signaling Technology, 1:1000 dilution, pH9 retrieval, 45 min of incubation).

The clinicopathological characteristics and SMARCA4 somatic gene status are summarized in Table 1 and Figure 1. The clinical and pathological features were consistent with SCCOHT in all cases. The IHC profiles were typical for SCCOHT in cases 1 and 2 (BRM negative and WT1 positive), whereas the phenotype of case 3 was slightly unusual (retained BRM staining in rare tumor cells, no WT1 expression). All cases further showed at least focal positivity for CK AE1/3 and SALL4 and were negative for alpha-inhibin. The 3 tumors were explored at the genomic level using a custom large panel of 571 cancer-related genes known as DRAGON (Detection of Relevant Alterations in Genes involved in Oncogenetics by Next generation sequencing) and commercially available as SureSelect Custom Design panel (Agilent).

TABLE 1.

Clinicopathological Characteristics and SMARCA4 Status of 3 Cases

Case	Age, y	FIGO stage	BRG1 IHC	BRM IHC	SMARCA4 Alterations	VAF, %	Type
1	28	NA	Weak	Negative	c.1614_1619del/p.(Tyr538*) c.3951+2T>A / p.?	43.7 43.2	Stop gain Splice site
2	24	IIIA	Moderate	Negative	c.2318T>C/p.(Leu773Pro) including chromosome 19 LOH through isodisomy	87.4	Missense
3	41	IVB	Weak	Focal	c.1420-2A>T/p.? Near-haploid genome	64.7	Splice site

IHC indicates immunohistochemistry; LOH, loss of heterozygosity; NA, not available; VAF, variant allele frequency.

FIGURE 1.

Histomorphology and IHC findings in 3 cases. Case 1: morphology consistent with SCCOHT (small cell variant), faint BRG1 expression, and absence of BRM expression; Case 2: morphology consistent with SCCOHT (small and large cell variant), moderate BRG1 expression, and absence of BRM expression; Case 3: morphology consistent with SCCOHT (small cell variant), faint BRG1 expression, and retained weak BRM expression in rare tumor cells; “Control SCCOHT”: an SCCOHT case with a truncating biallelic SMARCA4 mutation, demonstrating a complete loss of BRG1 expression and absence of BRM expression (validation of the IHC technique). HES: hematoxylin-eosin-saffron. Scale bars: 50 μm.

Cases reported by Mazibrada et al,¹ some previously published cases of SCCOHT and of rhabdoid tumors,^4–7 and 3 cases presented here, suggest that strong or moderate staining could preferentially be associated with missense SMARCA4 mutations or with in-frame splice or indel events. Conversely, weak/faint staining appears to be more indicative of canonical splice site mutations. Finally, BRG1-negative cases seem to harbor mostly truncating mutations (nonsense or frameshift), and a subset harbor splice site mutations. However, this genotype-phenotype correlation is imperfect, and reflects a tendency rather than a rule, as exemplified by some of the cases published by Mazibrada et al.¹

Although the clinicopathological features of SCCOHT with aberrant or retained BRG1 expression appear to be identical to those of BRG1-negative SCCOHT, it remained unclear whether they present any molecular particularities. Comprehensive genomic testing of our cases showed that cases 1 and 2 harbored remarkably stable, diploid genomes (Figs. 2A, B) and no additional alterations in cancer genes, which are both features typical of SCCOHT.^8,9 Unexpectedly, testing of case 3 unraveled a near-haploid genome (Fig. 2C), confirmed by FISH in selected loci (Fig. 2D). It also revealed pathogenic truncating variants in TP53, RB1, PTEN, and APC, corroborated by IHC (Fig. 2E). Near-haploid genomes are rarely encountered in cancer but have been reported in specific subtypes of glioblastoma,¹⁰ leukemia,¹¹ and mesothelioma.¹² In the latter cancer type, they are associated with SETDB1 alterations, not detected in our case. To our knowledge, haploid genomes have not been reported in SCCOHT. One intriguing hypothesis is the possibility of an oncogenic process affecting a cell undergoing meiotic division, which could be further consistent with the postulated germ cell origin of SCCOHT.¹³ However, because of the other unusual features of this case (additional somatic mutations, focally retained BRM expression, and patient’s age), we cannot exclude the possibility that this tumor may represent a slightly different entity, possibly unreported to date.

FIGURE 2.

Genomic findings in 3 cases. A–C, Allele-specific genomic profiles of the three tumors, showing (A) a diploid, “flat” profile in case 1; (B) a near-diploid profile in case 2, with copy number-neutral LOH (isodisomy) on chromosome 19p, which contains the SMARCA4 gene locus; and (C) a near-haploid profile in case 3. D, Fluorescence in situ hybridization [red: ERBB2 gene, green: chromosome 17 centromeric control probe (CEP17)] in case 3, confirming a haploid status of the tested loci. E, IHC for p53, PTEN and pRb in case 3, showing loss of expression of these proteins in tumor cells, consistent with genomic findings (internal positive controls present in the upper left part of each image); scale bars: 50 μm.

We further explored DNA methylation profiles (Illumina Infinium MethylationEPIC assay) and transcriptomic profiles (whole-transcriptome RNA-sequencing) of the three tumors, and compared them with BRG1-negative SCCOHT, extracranial rhabdoid tumors (ECRTs) with SMARCB1 or SMARCA4 alterations, and ovarian serous carcinoma (TCGA cohort)^7,14 (Fig. 3). Methylation profiles of cases 1 and 2 were consistent with SCCOHT on different types of analyses (Figs. 3A–C). The methylation profile of case 3 could be interpreted as more akin to ECRT (especially SMARCA4-deficient ones), but remained consistent with the ECRT/SCCOHT group. Results of the transcriptomic analysis (Figs. 3D–F) were highly concordant with methylome results.

FIGURE 3.

DNA methylation and RNA-seq findings. A–C, DNA methylation analyses, showing (A) Uniform manifold approximation and projection (UMAP), (B) principal component analysis (PCA), and (C) unsupervised hierarchical clustering using 5000 most variable probes. D–F, Transcriptomic (RNA-seq) analyses: (D) UMAP, (E) PCA, and (F) unsupervised hierarchical clustering using 5000 most variable genes. C and F, Clustering performed using Euclidean distance and Ward´s linkage method; SCCOHT_aberrant_BRG1: 3 cases from this study; SCCOHT_external: BRG1-negative SCCOHT data from a different center; SCCOHT_internal: BRG1-negative SCCOHT data from our center; ECRT_SMARCA4: SMARCA4-deficient extracranial rhabdoid tumors (ECRTs); ECRT_SMARCB1: SMARCB1-deficient ECRTs; OVARIAN_CARCINOMA_TCGA: ovarian high-grade serous carcinoma, TCGA cohort.

In summary, pathologists should be aware that weak or only partly lost BRG1 expression by IHC is an aberrant staining pattern, still consistent with a diagnosis of SCCOHT. Our results further confirm that such tumors show molecular features of SCCOHT. However, comprehensive molecular testing of tumors diagnosed as SCCOHT may occasionally reveal unexpected findings (like in case 3 presented here), which will need to be put into perspective if additional similar cases are identified to better understand their biological and clinical significance.