A Population-Based Study of Patients With Small Cell Carcinoma of the Ovary, Hypercalcemic Type, Encompassing a 30-Year Period

We identified all patients with SCCOHT and their relatives, who underwent genetic counseling at the Department of Clinical Cancer Genetics (DCCG), Institute of Oncology Ljubljana, between 2002 and 2021. In families with a germline SMARCA4 PV, we examined pedigrees for confirmed cases of SCCOHT and additional early-onset gynecologic cancers, which could have been unrecognized cases of SCCOHT.

To identify additional cases of SCCOHT in the Slovenian population between 1991 and 2021, we also searched the Slovenian Cancer Registry (SCR) for all cases of ovarian cancers with small cell morphology—morphology code 8041 and cases of undifferentiated ovarian carcinomas—morphology code 8020. The 2 morphology codes were chosen because all confirmed cases of SCCOHT identified from the records of the DCCG were reported in the SCR using 1 of these 2 codes. Because no confirmed cases of SCCOHT have ever been described as occurring after age 56 years, we excluded all who were older than 56 years at diagnosis from further analysis. For all potential SCCOHT cases we obtained tumor tissue samples and performed histopathologic revision with SMARCA4/BRG4 immunostaining. Only cases confirmed to be SCCOHT were included in further analysis. The diagram in Figure 1 illustrates the process of identifying patients for inclusion in our study.

Genetic analysis of tumor and nontumor tissue samples for deceased individuals was performed as part of the research project approved by the Slovenian National Medical Ethics Committee (0120-280/2019/4, date of approval June 14, 2019). Patients who underwent germline genetic testing signed an informed consent form allowing the use of their pseudoanonymized data for research purposes prior to submitting their blood samples.

Tumor samples were all formalin-fixed, paraffin-embedded (FFPE), and were evaluated and prepared as described previously.¹⁵  Whole blood samples were collected in EDTA tubes.

Genomic DNA was extracted either from blood using InnuPREP Master Blood Kit (Analytik Jena, Jena, Germany) or from FFPE tissue sample using MagMax DNA/RNA kit (ThermoFisher Scientific, Austin, Texas).¹⁵ 

Germline analysis was performed using targeted next-generation sequencing (NGS) for SMARCA4 variant identification or Sanger sequencing in cases of known familial PVs. For NGS, DNA libraries were prepared using a Nextera DNA Library Preparation Kit in combination with either TruSight One or TruSight Hereditary Cancer sequencing panels (Illumina, San Diego, California). The samples were sequenced using the Illumina MiSeqDx or the NextSeq 550 Sequencing System (Illumina). Read alignment to the human hg19 reference genome and variant calling was performed using MiSeq Reporter software or BaseSpace Enrichment App (both Illumina). Samples were successfully sequenced if more than 95% of targeted regions were covered at greater than ×40. Targeted analysis of all translated exons and ±25 bp flanking intronic regions of SMARCA4 gene (LRG_878t1) was performed. Variant calling, annotation, and the assessment of sequencing results were performed as previously described.^15,16 

For Sanger sequencing, DNA samples were bidirectionally sequenced using in-house designed primers, BigDye Terminator v3.1 Sequencing Kit and ABI Genetic Analyzer 3500 (both Applied Biosystems), in accordance with standard operating procedures.

For FFPE tumor tissue samples, DNA libraries were prepared using the TruSight Oncology 500 Library Preparation Kit (Illumina) and sequenced on the Illumina NextSeq 550 Sequencing System. Read alignment and variant calling were performed using Local App TruSight Oncology 500 (Illumina). Samples were successfully sequenced if more than 93.5% of targeted regions were covered greater than ×250. Targeted analysis of all translated exons and ±15 bp flanking intronic regions of SMARCA4 gene (LRG_878t1) was performed. Variant calling, annotation, and the assessment of sequencing results were performed as described previously.¹⁵  Tumor mutation burden (TMB) for nonsynonymous variants (TMB(N)) and TMB for both synonymous and nonsynonymous variants (TMB (N+S)) was assessed by Local App TruSight Oncology 500.

Germline variants were classified for their clinical importance in accordance with the consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ie, ACMG/AMP guidelines).^17,18  Variants are described according to HGVS v20.05 nomenclature.¹⁹ 

Nuclear expression of SMARCA4/BRG1 was assessed immunohistochemically in FFPE tumor tissue sections using a fully automated immunohistochemical stainer Benchmark Ultra (Ventana ROCHE Inc, Tucson, Arizona). Two 4-μm sections were dried at 56°C for 2 hours. Epitope was retrieved on board employing high pH Cell Conditioning Solution 1 (Ventana ROCHE Inc; catalog No. 950-124) for 48 minutes at 100°C. Retrieved epitope was detected using a primary rabbit monoclonal recombinant anti-BRG1 antibody (clone EPNCIR111A, Abcam, Cambridge, United Kingdom; catalog No. ab110641) diluted at 1:100 and incubated on board for 16 minutes at 37°C. Specifically bound primary antibody was visualized using 3-step multimer detection system OptiView DAB IHC Detection Kit (Ventana ROCHE Inc; catalog No. 760-700) according to the manufacturer’s instructions.

Loss of SMARCA4/BRG1 was defined as the absence of nuclear staining in tumor cells in a background of normal stromal cells, endothelial cells, fibroblasts, and lymphocytes with nuclear staining that served as positive internal controls.

Our analysis identified 7 cases (patients 1–7) of SCCOHT in Slovenia between 1991 and 2021, 4 familial from 2 different families (designated family 1 and family 2), 2 sporadic, and 1 for whom the origin of the SMARCA4 PV could not be determined. Of these, 5 received a correct diagnosis of having SCCOHT and were reported as such in the SCR, even though 2 of these were initially suspected of having a dysgerminoma and a granulosa cell tumor, respectively. Two cases were reported as undifferentiated ovarian carcinomas in the SCR.

A total of 6 of 7 cases were identified from the records kept at the DCCG. Family 1 (Figure 2, A) was first described in a report published in 1995.⁵  An additional female relative in this family was reported as having died of cancer at age 27 years. The diagnosis was registered as undifferentiated ovarian cancer in the SCR (morphology code 8020). The histopathologic report suggested the diagnosis of small cell ovarian carcinoma, pulmonary type or undifferentiated carcinoma. Clinical data for this patient were inaccessible, but her tumor and healthy tissue were obtained for further analysis, which confirmed the diagnosis of SCCOHT. Proband from Family 2 has previously been described in the report by Vivod et al,²⁰  where it is was noted that her mother developed an epithelial ovarian cancer aged 55 years. In fact, the mother received a diagnosis of disseminated disease, originating in her left ovary, at age 28 years. She underwent a hysterectomy, bilateral adnexectomy, omentectomy, and appendectomy, and the histopathologic analysis showed a poorly differentiated adenocarcinoma of the ovary. A review performed at the time indicated the tumor was poorly differentiated but not necessarily an adenocarcinoma. Her serum calcium was normal. She received chemotherapy (cyclophosphamide, Adriamycin, and cisplatin), which was ineffective, and died 6 months after her diagnosis. Her tissue samples are considered lost, so a histopathologic review and genetic analysis could not be undertaken (Figure 2, B).

Based on the data from the SCR 1 additional case of SCCOHT was detected on review of pathology slides. It was reported as an undifferentiated carcinoma in the SCR, even though 1 pathologist did suggest that the differential diagnosis should include SCCOHT at the time of diagnosis.

Details concerning clinical features and oncologic treatment are presented in Tables 1 and 2, respectively. No meaningful statistical analysis was feasible because of the small number of patients identified. A total of 6 of 7 patients first presented with abdominal pain, 3 in the setting of ovarian torsion or acute appendicitis. Three of our patients presented with hypercalcemia; there is no information for 1 patient, and for 3 patients serum calcium concentrations at diagnosis are unknown but were not elevated when the disease progressed. Three patients had normal or only somewhat elevated concentrations of the tumor marker CA-125 in the presence of widely disseminated disease. In one case, CA-125 was elevated at diagnosis, and there are no data for the other three patients. Two patients died before undergoing any specific oncologic treatment. Patients with stage I disease appeared to fare somewhat better than the patient with stage III to IV disease. There was no significant response to any of the chemotherapy regimens used, with disease progressing while on chemotherapy in 4 patients. The effect of radiotherapy was limited as well, with clear progression in 2 cases and a limited response to treatment in 1 case. Immunotherapy using pembrolizumab in 1 patient resulted in disease stagnation for almost 12 months. In this patient, programmed death ligand-1 expression was determined to be 10% to 20%, evaluated as tumor proportion score on a fine-needle aspiration biopsy sample using clone 22C3. After disease progression, a combination of nivolumab and ipilimumab was introduced but discontinued because of autoimmune colitis. Later on, she also received a cyclin-dependent kinase 4 (CDK4)/CDK6 inhibitor abemaciclib in combination with nivolumab, but she was on medication for less than 3 months because she progressed while on the treatment.

The results of genetic analysis for germline and somatic SMARCA4 variants are presented in Table 3. In family 1, FFPE tumor tissue testing (patient 1) revealed a novel PV in SMARCA4, LRG_878t1:c.1423_1429delTACCTCA p.(Tyr475Ilefs*24). The variant was confirmed to be germ line by testing the same patient’s FFPE unaffected tissue and was also seen in her healthy sister and the sister’s 2 healthy daughters. The variant was also detected in the FFPE nontumor tissue of the patient’s maternal cousin, who developed ovarian cancer age 27 years (patient 3), thereby indicating that the patient’s mother (patient 2) and the mother’s brother were obligatory carriers of the variant (Figure 2, A). Testing of unaffected and tumor tissue for patient 2 and tumor tissue for patient 3 was either unsuccessful or could not be performed.

In family 2, the proband (patient 4) was shown to be a heterozygous carrier of the LRG_878t1:c.3216-1G>T germ-line PV in SMARCA4. Analysis of her tumor showed high allele frequency for this variant, indicating loss of heterozygosity (Table 3). Her mother’s samples could not be obtained, but the analysis of her grandfather’s tissue revealed he was not a carrier of the SMARCA4 PV. The patient’s father and maternal grandmother could not be tested (Figure 2, B).

For 2 of our patients, tumor tissue testing revealed 2 SMARCA4 PVs in each (2 nonsense variants in one, and a nonsense and a frameshift variant in the other). Testing of unaffected tissue or blood samples showed these variants to be somatic. In 1 patient, genetic testing of her tumor tissue revealed a SMARCA4 frameshift variant with high allele frequency. Healthy tissue testing was not undertaken because the next of kin could not be contacted and notified of the result (Table 3).

In all tested tumors, TMB was low, with SMARCA4 variants being the only significant findings.

A histopathologic revision of available tumor samples with SMARCA4/BRG1 immunostaining was performed for all 7 patients and the results are presented in Figures 3 and 4. In all cases, the diagnosis of SCCOHT was established based on morphologic findings and loss of SMARCA4/BRG1 in the nuclei of tumor cells. Most tumors consisted of small cells with scant cytoplasm. There were, however, notable differences in tumor characteristics between patients, with 2 cases exhibiting the predominance of large cells with eosinophilic cytoplasm (ie, the large cell variant of SCCOHT; Figure 3, B and D). In 1 case, most tumor cells showed weaker staining or complete loss of SMARCA4/BRG1 compared with internal controls (Figure 4, C). The combination of tumor morphology and immunophenotype supports the diagnosis of SCCOHT in this case, rather than that of the neuroendocrine small cell tumor of the ovary, pulmonary type, or undifferentiated carcinoma, as was initially suggested. More detailed information on histopathologic findings and additional immunohistochemistry results are provided in Table 4.

Our study presents all known cases of SCCOHT during a 30-year period in a population of 2 million. Family 1 (Figure 2, A) is of historical interest because this was first presented in one of the early reports, which first suggested SCCOHT could sometimes be familial.⁵  When a genetic cause of SCCOHT was discovered in 2013/2014, genetic analysis of FFPE tumor tissue samples for patients 1 and 2 was performed by another research group but was unsuccessful.⁷  We now demonstrate that the familial occurrence of SCCOHT in family 1 is due to a novel frameshift variant in SMARCA4, and we provide data on an additional case of SCCOHT in this family. Also, we identified a splice-site variant in SMARCA4 in family 2 (Figure 2, B) that, to our knowledge, has not been previously reported as germline.

The penetrance of germline SMARCA4 PVs is not well defined.²¹  Very few cases of female carriers living to old age with no evidence of SCCOHT have been published thus far.²²  In cases of unremarkable family histories of SMARCA4 PV carriers, the variant was often inherited from the father or, rarely, shown to be de novo.²³  The penetrance of SMARCA4 PVs therefore appears to be high, but such an impression could be a spurious result of ascertainment bias, because the SMARCA4 gene is unlikely to be analyzed unless a diagnosis of SCCOHT/RTPS2 is suspected. High-penetrance PVs in published cases could therefore be outliers, with most SMARCA4 PV carriers at low risk of developing SCCOHT. A reliable estimate of penetrance would be of great value for healthy female carriers of SMARCA4 PVs, for whom a bilateral salpingo-oophorectomy is the only viable preventive option. The procedure is, however, associated with considerable morbidity and greatly reduces the patient’s reproductive options. Also, optimal timing of surgery is difficult to define because the disease sometimes occurs in children.^24,25  It is also worth noting that female patients with the developmental disorder Coffin-Siris syndrome, who carry SMARCA4 PVs, very rarely develop SCCOHT, but PVs seen in Coffin-Siris syndrome are for the most part missense/gain-of-function, as opposed to inactivating variants seen in SCCOHT.²⁶  In our family 1, 3 of 6 known SMARCA4 PV carriers developed SCCOHT (Figure 2, A). Of the 3 healthy carriers, 2 are in their twenties. They opted for gynecologic surveillance because they wished to preserve their fertility. Based on her family history of ovarian cancer, their mother underwent a prophylactic bilateral salpingo-oophorectomy at age 45 years, which showed foci of endometriosis but no other abnormalities. Other elderly and healthy female members of this family were unavailable for testing, precluding a valid estimate of penetrance for the familial SMARCA4 PV. Based on the available data, it appears to be reasonably high. In family 2, the proband is the only confirmed case of SCCOHT. Nevertheless, considering her mother’s early diagnosis of ovarian malignancy, tumor morphology, and clinical course, it is likely the mother’s diagnosis was SCCOHT, and also due to the familial PV. If that was the case, the mother would have inherited the PV from her healthy mother, or, alternatively, it could have arisen de novo. Intriguingly, the grandmother’s niece died age 7 months because of an aggressive abdominal tumor (Figure 2, B). Such a tumor could have been a SMARCA4-related rhabdoid tumor because cases of ovarian and rhabdoid tumors due to the same familial SMARCA4 variant have been described in the past.²³  Unfortunately, we were unable to confirm or refute this hypothesis because the child’s tissue samples were unavailable. No cases of SMARCA4-deficient undifferentiated uterine sarcoma, which can also arise due to SMARCA4 germline PVs, were detected in our families.²⁷ 

To identify potential SMARCA4 PV carriers with no personal or family history of SCCOHT, we performed additional analyses of genetic testing results in all individuals tested with NGS panels, which included SMARCA4, at our institution. In 2474 individuals tested between August 2017 and December 2021, no SMARCA4 PVs were detected, indicating these variants are not a common finding in individuals with a personal or family history of cancers other than SCCOHT in the Slovenian population. Of note, SMARCA4-predicted loss-of-function variants are rare in the Genome Aggregation Database (gnomAD). The probability of SMARCA4 being loss-of-function intolerant is estimated to be 1 based on the gnomAD data set, indicating protein-truncating variants might be selected against.²⁸  Notwithstanding the rarity of SCCOHT, we would argue available data indicate germline SMARCA4 loss-of-function variants are unlikely to be low penetrance.

The incidence of SCCOHT is currently unknown, with 1 German study suggesting it might be expected to develop in 1 in 80 million children and adolescents per year.²⁹  Our study identified 7 patients with confirmed SCCOHT in Slovenia between the years 1991 and 2021. Because of difficulties in establishing the diagnosis before SMARCA4/BRG1 immunostaining was introduced in 2014, we presume there could have been additional cases, particularly prior to 2014, that were misdiagnosed. Ovarian cancer cases reported using morphology codes other than 8041 and 8020 in the SCR would not have been included in our study. We excluded patients older than 56 years from our analysis. Diagnoses of SCCOHT in elderly patients with ovarian tumors should be considered with caution even in cases of loss of SMARCA4/BRG1 staining because SWI/SNF-deficient undifferentiated ovarian carcinoma is a more likely diagnosis in this group.³⁰  We did not identify any childhood-onset cases, and it is possible these would have been diagnosed as abdominal, not ovarian, tumors. Hence, we consider the number of patients we identified as the minimal number of SCCOHT cases between 1991 and 2021. In this period, the population of Slovenia as registered by the Statistical Office of the Republic of Slovenia was relatively stable, with a moderate increase from 1 999 068 to 2 108 708. We therefore estimate the minimal annual incidence of SCCOHT for our population at approximately 0.12 per 1 million. Considering the small number of patients in our study, and the disproportional number of cases from a single family, our estimate could also be too high. In a population as small as Slovenia’s, 1 large familial cluster of cases could easily result in skewed data. Studies in larger populations are needed to assess the incidence with any level of confidence.

The rarity of SCCOHT is probably the principal reason why there is no standard of care for these patients and why survival is poor, even when diagnosed early.³¹  Outcomes in our SCCOHT patients were abysmal, with 6 of 7 dying of their disease within 27 months from diagnosis (Table 2). The last patient to receive a diagnosis did, however, benefit from recent developments, which have established immunotherapy as the most promising nonstandard treatment for SCCOHT.⁴  Recent studies have demonstrated that SMARCA4-deficient SCCOHT cells are sensitive to inhibition of CDK4/6 due to cyclin D1 downregulation, and there is anecdotal evidence that CDK4/6 inhibitors in conjunction with immunotherapy can elicit durable responses in SCCOHT.^4,32  Our only patient who received abemaciclib in combination with immunotherapy nivolumab was on medication for less than 3 months and progressed while receiving it. Recent work exploring epigenetic deregulation in SCCOHT and targeted treatments such as inhibitors of the catalytic subunit of the polycomb repressor complex, EZH2, has shown SCCOHT cells to be sensitive to selective EZH2 inhibition.³³  None of our patients received EZH2 inhibitors or other epigenetic therapies.

We consider our ability to obtain tissue samples and patients’ medical records even for cases treated decades ago to be one of the strengths of our work. Also, because notification of cancer has been compulsory in Slovenia since the foundation of the SCR in 1950 and is prescribed by law, we are confident we were able to identify all correctly diagnosed cases of SCCOHT. The small number of cases is our study’s principal weakness.

In conclusion, we present germline and somatic genetic testing results, and histopathologic and clinical characteristics for 7 cases of SCCOHT. Novel germline SMARCA4 variants are reported in 2 families, possibly associated with high penetrance. Our tentative estimate of SCCOHT incidence suggests the disease, albeit very rare, might be somewhat more common than previously assumed.

The authors would like to thank all the patients and their family members who participated in this study. They would also like to thank Janez Lamovec, MD, PhD, for his work, which formed the basis of this study. Also, they gratefully acknowledge the contribution made by the staff of the Departments of Clinical Cancer Genetics, Molecular Diagnostics, and Pathology at the Institute of Oncology, Ljubljana.