Role of gene sequencing in classifying struma ovarii: BRAF p.G469A mutation and TERT promoter alterations favour malignant struma ovarii

Struma ovarii (SO) are rare, accounting for 0.3–1% of ovarian tumours, and include benign and malignant lesions. In most cases, histology is not predictive of clinical outcome and prognosis. The prognosis of histologically malignant thyroid‐type carcinomas can indeed be excellent, while SO, composed of normal thyroid tissue, can recur and are designated highly differentiated follicular carcinoma of the ovary. Clearer diagnostic criteria are therefore required.


Introduction
Struma ovarii (SO) is defined as a mature ovarian teratoma in which thyroid tissue is the predominant or sole component, with both benign and malignant forms. 1 It was previously believed that malignant ovarian struma, representing 5-10% of all SO, could be diagnosed by thyroid gland tumour criteria. 2However, extensive studies with more than 5 years of follow-up [4][5][6] have shown that in 69% of cases, clinically malignant SO are morphologically benign (normal thyroid tissue or follicular adenoma), while clinically benign SO can (in 28% of cases) show nuclear characteristics of papillary carcinoma. 6Furthermore, unequivocal vascular invasion was found to be a very rare feature of otherwise evident clinical malignancy in SO. 4 Capsular invasion is considered a sign of malignancy in follicular neoplasms of the thyroid gland. 7However, most authors do not consider extension through the irregular fibrous thickening of the outer cortex to be true capsular invasion, as the ovary lacks a capsule. 4Thus, according to Robboy et al., except for poorly differentiated cancers, no single histological feature correlates with clinical behaviour in SO. 4 The term 'highly differentiated follicular carcinoma of the ovary' (HDFCO) 8 was coined for SO with benign morphology associated with the so-called 'peritoneal strumosis' (peritoneal extension of benignlooking thyroid tissue) that was initially thought to be benign. 3In other words, any peritoneal extension of a SO is considered a sign of malignancy, even if the SO itself appears benign. 1,9he problem with SO is to define the degree of malignancy and the need for clinical follow-up when the tumour is confined to the ovary.As histology is not predictive of clinical outcomes or prognosis, clear diagnostic criteria are required.

P A T I E N T S E L E C T I O N
All cases coded as struma ovarii between January 2000 and July 2021 were retrospectively retrieved from the records of the pathology department of the Hospices Civils de Lyon (Lyon, France).Sixty cases were retrieved, 30 of which were consultation cases (M.D.S., M.D.P.).Clinical and follow-up data were collected retrospectively.
Tumours were classified in three groups, according to criteria described by Robboy et al. 4 as biologically malignant SO if they showed ovarian surface or extra-ovarian extension at presentation or during follow-up (BMSO, n = 8), or as histologically malignant SO if they showed histological characteristics used to define malignancy in thyroid gland tumours according to WHO classification 2022, 7 without extra-ovarian extension at presentation or during follow-up (HMSO, n = 18) (Figure 1).Tumours without histological characteristics of malignant thyroid tumours or extra-ovarian extension were classified as biologically and histologically benign SO (BSO, n = 32).
Molecular analyses were performed on the most recent cases with available paraffin blocks: 10 cases of HMSO, eight of BMSO (including six with peritoneal implants) and 13 of BSO.A representative formalin-fixed paraffin-embedded (FFPE) tissue block was chosen for molecular investigations in each case.All specimens were analysed by next-generation sequencing (NGS) for gene mutation analyses.RNA sequencing (for fusion detection) was also performed for all eight BMSO and BSO and for HMSO specimens if no mutation was detected by NGS.

M O L E C U L A R C H A R A C T E R I S A T I O N
Total nucleic acids were extracted using the Maxwell 16 LEV RNA FFPE kit on a Maxwell 16 instrument (Promega Corporation, Madison, WI, USA) without DNAse treatment.Areas of interest were selected on the slide by macrodissection.RNA and DNA levels Molecular analyses of struma ovarii 293 were quantified using a Qubit fluorometer (Life Technologies, Carlsbad, CA, USA).Libraries were sequenced on Illumina (TM) NextSeq500â.For DNAseq, we used the Sophia Genetics 'Solid Tumor Solution' pan-cancer panel, which covers alterations of major oncogenes and of tumour suppressor genes that are deregulated in solid tumours (Sophia Genetics, Lausanne, Switzerland) (Supporting information, Data S1).Variants were interpreted using the Sophia DDM version 4 interface with OncoPortal (Sophia Genetics).For RNAseq, we used the Archer Fusion-Plex pan-cancer panel (RNA Seq ARCHER Panel FusionPlex_CHU_Lyon_Pan_Solid_Tumour _Sarcoma_ 17125-v1.0)(Supporting information, Data S1).The data were analysed using Archer Analysis software (version 6.2; ArcherDX, Inc.).

E T H I C S A P P R O V A L A N D D A T A A V A I L A B I L I T Y
The study was conducted according to the Declaration of Helsinki and approved by and registered with the local ethics committee (no.22_5652).Data sharing is not applicable to this article, as no data sets were generated or analysed during the current study.

Results
Thirty-one patients with available paraffin blocks were included in the study (Table 2).The median follow-up time was 45 months for seven of the HMSO and 49.5 months for the BSO, none of which recurred during follow-up.The median follow-up time for the BMSO patients was 81 months (range = 28.8-132.2months; Table 2).Among the six patients with BMSO with peritoneal spread, at the time of presentation three had extra-ovarian spread into the pelvic peritoneum.Peritoneal implants in the remaining three BMSO with peritoneal spread occurred, on average, 5.2 years after diagnosis (range = 3-10.6years).
The eight cases of BMSO included six with peritoneal spread, of which 66% (four of six) had an ovarian thyroid carcinomatous component with columnar cell papillary carcinoma morphology (one of four) or a follicular variant of papillary thyroid carcinoma (FV-PTC) morphology (three of four).The peritoneal implants of these papillary carcinomas had the same morphology as the ovarian tumours, except for two cases where the implants were composed of normal thyroid tissue.Two BMSO (BMSO 4 and 5) with peritoneal spread corresponded to HDFCO, because the ovarian tumours and their implants were composed of normal thyroid tissue with no morphological signs of thyroid carcinoma (Figure 1).
The two BMSO without peritoneal implant extended into the ovarian surface, with grossly visible nodular projections on the surface of the ovary (Figure 1D).One had a solid/trabecular thyroid papillary carcinoma morphology and one was a poorly differentiated thyroid carcinoma, without nuclear features of papillary carcinoma, showing insular growth pattern with 9.28 mitoses per 2 mm 2 and necrosis.

HMSO histology (n = 10)
PTC was diagnosed in eight cases of HMSO (80%), with papillary carcinoma-like cytology.The remaining two HMSO (20%) were poorly differentiated carcinoma, with a trabecular and insular growth pattern and without nuclear features of PTC, showing 7.6 and 4.22 mitoses per 2 mm 2 without necrosis and with intravascular extension.

BSO histology (n = 13)
Five cases corresponded to follicular adenoma, with dense follicles packed together without nuclear atypia, or any capsular or vascular invasion.The eight other cases corresponded to normal thyroid tissue composed of follicles of different sizes, some of which had a hyperplastic thyroid morphology.
NGS failed in one case of BMSO in the primary ovarian tumour, while its peritoneal metastasis showed an NRAS Q61R mutation.Excluding this case, all BMSO with peritoneal spread had a molecular abnormality in the primary ovarian tumour.Taken together, 87.5% of the BMSO (seven of eight) and all (six of six) BMSO with peritoneal spread had a molecular abnormality, either in the primary ovarian tumour or in the peritoneal implant (for the case in which sequencing failed in the primary tumour).For the BMSO with peritoneal spread, the same molecular alteration was observed in the ovarian tumour as in the peritoneal metastases, except for two cases of peritoneal implants with an HDFCO morphology that did not have the NRAS mutation and TERT deletion detected in the corresponding ovarian tumours.However, in one case of columnar cell PTC, in addition to the TERT promoter mutation seen in the primary ovarian tumour, the peritoneal nodule also harboured a TP53 mutation.The morphology was nevertheless identical in both sites (Table 2).
Among HMSO, 70% (seven of 10) had either BRAF (six of 10) or NRAS mutations (one of 10).Most of the detected molecular abnormalities were gene mutations.Only two gene rearrangements were observed, one involving duplication of the BRAF tyrosine kinase domain in a poorly differentiated stage IA HMSO, and the other, EWSR1::PATZ1 fusion in a case of BMSO with peritoneal spread, in both the primary tumour and the peritoneal implant.
TERT promoter alterations were detected in two cases of BMSO with peritoneal implants, including a case of HDFCO, and were not detected in any HMSO or BSO.

Discussion
Little is known about the genetic characteristics of SO, particularly for HDFCO and malignant SO.To our knowledge, the molecular characteristics of primary SO and peritoneal metastases (so-called 'peritoneal Ó 2023 The Authors.Histopathology published by John Wiley & Sons Ltd., Histopathology, 84, 291-300.strumosis') have never previously been compared.We report the most extensive series of SO with DNA and RNA sequencing, consisting of eight BMSO, six with peritoneal metastases, 10 HMSO and 13 BSO.We found molecular alterations in 78% of biologically or histologically malignant SO (14 of 18), while only one of 13 BSO (7.7%) had a genetic abnormality, a statistically significant difference between the two groups.Our results confirm that SO are genetically similar to primary thyroid carcinomas, with the same type of molecular alterations as previously reported in 53 cases of SO. [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27] As in thyroid gland carcinomas, 7 the most common genetic alterations in malignant SO are BRAF abnormalities (observed in 57% of cases in our cohort (eight of 14) and 27% (10 of 37) in the literature), 11,12,19,21,25,27 which have not been observed in BSO either previously (13 cases) 10,11,13 or in the present study (13 cases).In the thyroid gland, the p.V600E BRAF mutation is highly specific for malignancy.It is detected in 45-60% of papillary carcinomas, and is particularly common in high-risk PTC subtypes such as the classical, tall cell and hobnail variants. 7This mutation is rarely seen in malignant SO, however, and accounted for just one of the eight BRAF alterations in this study and three of 10 of those described in the literature. 11,27The most Molecular analyses of struma ovarii 297 common BRAF mutation in SO is p.G469A, which we observed in four histologically or biologically malignant SO, including one case with peritoneal metastases that had the same mutation.9][30] The p.G469A BRAF mutation seems to be characteristic of ovarian SO, and appears to be associated with malignancy.In this study, all the cases of this mutation involved follicular-pattern carcinomas, including FV-PTC and HDFCO.In contrast, the BRAF K601E mutation is rare in malignant SO (one case of FV-PTC in the present study and three cases in the literature 11,19,25 ).In the thyroid gland, the BRAF K601E mutation is considered a RAS-like mutation because of its prevalence in follicular tumours, its association with benign non-invasive neoplasms and its different mechanism of activation of the MAPK pathways from that of the BRAF V600E mutation. 31Somatic mutations of RAS family members (NRAS, KRAS, HRAS) are not a reliable indicator of malignancy in thyroid nodules because they are recurrently found in follicular tumours, ranging from follicular adenoma to follicular carcinoma and FV-PTC and, more recently, to NIFTP (non-invasive follicular thyroid neoplasm with papillary-like features). 32,33Thus, it is not surprising that one BSO (7.7%) and 17% of FV-PTC and HDFCO in our study, as well as eight (22%) of the FV-PTC type malignant SO described in the literature, 14,15,18,20,21,23,27 showed RAS mutations.Among the BSO in this study, eight had a normal/hyperplastic thyroid tissue morphology, none of which had a gene mutation or fusion.The single BSO with KRAS mutation had follicular adenoma histology.A limitation of our study is that the follow-up period of the BSO with KRAS mutation was very short (3 years), and the long-term behaviour of this tumour is unclear.

Molecular analyses of struma ovarii 295
This study is the first, to our knowledge, to report TERT alterations in malignant SO, including TERT promoter mutation in one BMSO and TERT deletion in another case of BMSO.Herein, TERT alterations are only detected in aggressive carcinomas with peritoneal dissemination, despite their well-differentiated and apparently benign morphology.][36] Here, the patient with TERT promoter mutation in the primary ovarian and the peritoneal tumour died of her disease.Moreover, other alterations involving the TERT gene, such as gain of TERT locus, aberrant TERT promoter methylation, TERT mRNA overexpression and TERT deletion (as detected here in one case of HDFCO) are associated with adverse outcomes in well-differentiated follicular cell-derived thyroid carcinomas. 37,38he teratomatous nature of extra-ovarian HDFCO has been demonstrated by evidence of genetic homozygosity in three cases in the literature, 26 as previously reported for mature ovarian teratomas. 39,40he histogenesis of peritoneal extension of HDFCO, either from seeding and implantation secondary to tumour rupture or representing true tumour spread, may be debatable, although the review from Roth and Karseladze 8 favours true peritoneal metastases.None of the five HDFCO studied previously showed any of the gene mutations or fusions typical of thyroid cancer. 17,24,27We report NRAS Q61R and BRAF G469A mutations and TERT deletion for the first time in three separate cases of HDFCO, including two cases of HDFCO morphology in the peritoneal spread (the so-called 'peritoneal strumosis').BRAF and TERT alterations are thought to be indicators of malignancy in thyroid carcinomas, 37,38 and they were not observed in any of the BSO studied here or in the literature. 10,13Our findings thus confirm the malignant nature of HDFCO (including the so-called 'peritoneal strumosis') and their potential aggressiveness, despite lacking the typical histological features of thyroid carcinoma.It is worth noting that, among BMSO with peritoneal spread, none of the HDFCO recurred during follow-up, while the two cases with papillary carcinoma features in the peritoneal implants recurred.Although malignant, HDFCO have a better prognosis than histologically malignant thyroid-type carcinoma with extra-ovarian extension. 5he molecular alterations were comparable in the primary tumour and the peritoneal metastases of our BMSO with peritoneal spread, confirming the carcinogenetic relationship between the primary ovarian tumour and the peritoneal extension, despite possible differences in morphology between the two sites (in two cases).Among the BMSO with peritoneal recurrence, one was an HDFCO with sequencing failure in the primary tumour but with an NRAS Q61R mutation in the peritoneal nodule, and one was an FV-PTC with KRAS G12R mutation in both the primary tumour and the peritoneal recurrence.In a third case, the BMSO was a columnar cell PTC with TERT promoter mutation and EWSR1::PATZ1 fusion in the ovarian tumour.In this case, the peritoneal recurrence 10.6 years later showed the same alteration as the primary ovarian tumour, in addition to TP53 Ó 2023 The Authors.Histopathology published by John Wiley & Sons Ltd., Histopathology, 84, 291-300.mutation, showing that molecular alterations can accumulate during the evolution of an SO.The remaining three BMSO with peritoneal spread at diagnosis were also associated with molecular alterations.In one case of FV-PTC, despite the primary tumour and the peritoneal nodule having different morphologies (FV-PTC and HDFCO, respectively), the same BRAF G469A mutation was detected in both sites.In two cases (BMSO2 and BMSO4), the peritoneal nodules that were very well differentiated without histologically malignant morphology (i.e.HDFCO) did not harbour the NRAS mutation and TERT deletion of the primary ovarian tumour, indicating that the so-called 'peritoneal strumosis' is not always caused by tumour rupture.The patient with TERT deletion in a HDFCO had previously had a benign 3.5-cm SO in the contralateral ovary without molecular alterations 8 years before the HDFCO with peritoneal extension.In the initial description of HDFCO by Roth et al. in 2008, 8 their literature review of 18 additional cases included two bilateral metachronous struma with peritoneal dissemination associated with the second struma, one of which developed SO in the contralateral ovary associated with peritoneal spread 11 years after excision of the first SO, identical to the case we describe here.We interpreted our case of HDFCO with TERT deletion as being the tumour responsible for the peritoneal spread.However, as the initial SO eight years earlier did not show any molecular alteration comparable to the findings of the peritoneal nodule, we cannot exclude the possibility that the peritoneal nodule was derived from the initial tumour.

Conclusion
This study demonstrates the clinical utility of molecular sequencing in struma ovarii.In this limited number of cases, BRAF mutations are suggestive of a malignant process.However, this needs confirmation in a larger series, probably through a multicentric study.Our results also indicate that the p.G469A BRAF mutation is more common in struma ovarii than in eutopic thyroid neoplasia, where the p.V600E variant is more frequent.Also, TERT promoter alterations seem to be indicative of more aggressive tumours, such as seen in thyroid gland neoplasms.However, as the prognosis of all struma ovarii is excellent, even for those with thyroid carcinoma morphology, more extensive molecular studies are required with more than 10 years of follow-up to draw any conclusions on the disease courses associated with these genetic alterations.

Figure 1 .
Figure1.Histological characteristics of struma ovarii (haematoxylin and eosin, H&E).A, Benign struma ovarii (BSO) with KRAS G12V mutation composed of variable-sized thyroid follicles filled with colloid material and showing adenomatous changes.The nuclei are bland, round and small.B, Histologically malignant struma ovarii (HMSO) with BRAF V600E mutation showing a classical papillary thyroid carcinoma (PTC) with distinctive nuclear features, including nuclear enlargement, irregular and overlapping nuclei with chromatin clearing, margination and glassy nuclei.C, Histologically malignant poorly differentiated struma ovarii (HMSO) with NRAS Q12K mutation: poorly delimited tumour with trabecular, insular and solid patterns.The tumour cells are small, with round and monotonous hyperchromatic nuclei (insert).D, Biologically malignant struma ovarii (BMSO) with no molecular alteration: a papillary thyroid carcinoma with a trabecular and microfollicular pattern protruding through the surface of the ovary (arrow).E, Highly differentiated follicular carcinoma of the ovary (HDFCO) with peritoneal extension (F), showing the same histological features in the ovary and the peritoneum: follicles of different sizes filled with colloid and lined by small cells with small round bland nuclei.The primary ovarian tumour showed a TERT alteration (del c.-162 to c.-141), while the peritoneal implant showed no molecular alteration.