Our study provided a clinicopathological, immunohistochemical, and molecular reappraisal of ovarian SCLEC. We found that most SCLECs show an admixture of sertoliform and granulosa-like architecture, highlighting the frequent nuclear β-catenin and CDX2 expression, luminal CD10 positivity, and CTNNB1 mutations.

In agreement with the results of previous studies, patients with SCLEC in our series showed a highly variable age at diagnosis (range 50–89 years; mean 65.7). Such range overlaps with both conventional EC and adult granulosa cell tumor [13, 14], while it differs from Sertoli-Leydig cell tumor, which mostly occurs at a young age (mean age ~ 25 years) [15]; this may be helpful in differential diagnosis. Similar to conventional low-grade EC, most SCLEC cases show relatively favorable prognosis, with no extension beyond ovary [3]. However, a minority of cases may show advanced stage and/or aggressive behavior [16]. In our series, 3/8 cases showed extraovarian extension; one of these cases was p53-abnormal.

While previous studies on SCLEC mainly included sertoliform EC [2,3,4], in our series the granulosa cell tumor-like pattern and sertoliform pattern had similar frequencies. In 5/9 cases of our series and in most previously published cases, SCLEC showed low-grade features with no high nuclear pleomorphism [1, 2]; four cases showed eosinophilic changes with enlarged nuclei to a variable extent, with areas of serous-like pleomorphism. While striking pleomorphism might potentially represent a diagnostic pitfall, it was not diffusely present in any of the included cases. Interestingly, all and only the cases exhibiting cytoplasmic eosinophilia and enlarged nuclei harbored CTNNB1 c.101G > T; p.Gly34Val. mutation. It is unclear whether this association is meaningful.

Given the wide range of adnexal tumors exhibiting a sex cord-like architecture [17], the presence of EC morphological features in SCLEC can be crucial for a correct diagnosis. While most previously published cases of SCLEC were accompanied by an overt EC glandular component [1, 2], the latter was present in only three cases of our series. Squamous/morular differentiation was observed in four cases. Remarkably, three cases lacked unequivocal EC features and were diagnosed by a combination of morphological, immunohistochemical, and molecular features; STK11 alterations were also analyzed and excluded in these tumors. In ambiguous cases, immunohistochemistry may help identifying the case as a SCLEC by showing positivity for epithelial markers (CK7, EMA) and negativity for sex cord markers (inhibin, calretinin, SF1) [6,7,8]. In our series, all cases were positive for EMA and 8/9 for CK7, while only one case was focally positive for inhibin. An advanced age and the absence of hormonal manifestation would also favor SCLEC [2].

As previously described, the differential diagnosis of SCLEC is broad and includes not only sex cord tumors but also neuroendocrine, Wolffian/mesonephric-like neoplasms, and STK11 adnexal tumors [7, 8]. Neuroendocrine tumors may be excluded based on negativity for neuroendocrine markers. In our series, only one case was focally positive for chromogranin and two were focally/zonally positive for synaptophysin (which is less specific) [18].

FATWO is a rare entity which may show overlapping morphology and immunophenotype with SCLEC. Unlike SCLEC, FATWO is mostly negative or focally positive for EMA and often shows a non-diffuse expression of sex cord markers; however, the immunophenotype of FATWO has been shown to be inconsistent. PAX8 is reportedly more common in SCLEC than in FATWO, but in our series only three SCLECs expressed PAX8. Moreover, upper mesonephric remnants (from which FATWO is thought to derive) are positive for PAX8 [8]. We have therefore some concerns about the accuracy of this marker in the differential diagnosis. Morphological features of FATWO that may be helpful in the differential diagnosis are para-tubal localization, multilobate architecture, expansile borders, bland nuclear features, and low mitotic index [8, 19].

Mesonephric-like carcinoma is an aggressive entity which exhibits a wide range of morphological patterns. The typical immunophenotype of mesonephric-like carcinoma includes CD10 luminal expression; PAX8, GATA3, and TTF1 nuclear positivity, negativity, or focal positivity for ER and PR; retained MMR expression; wild-type p53 pattern; and non-diffuse p16 expression [8, 17, 20]. In our series, most cases showed a luminal expression of CD10, which might raise the concern of a mesonephric-like carcinoma. However, all SCLECs in our series were negative for GATA3 and TTF1 and positive for ER and PR, allowing us to exclude mesonephric-like carcinoma. Moreover, 6/9 SCLECs were negative for PAX8.

STK11 adnexal tumor is a recently described entity, previously included in the FATWO category. In most cases of our series, morphological and immunophenotypical features did not appear consistent with STK11 adnexal tumors. In fact, the latter mostly show paratubal localization and a peculiar architecture characterized by interanastomosing cords and trabeculae immersed in a myxoid matrix, and basophilic rather than eosinophilic intraluminal secretion. Unlike our cases, most STK11 adnexal tumors showed diffuse WT1 and calretinin expression and focal CK7 and CD10 positivity, while EMA is uncommonly expressed [8, 21]. In our series, we only tested STK11 in two tumors which lacked obvious endometrioid features. This was useful to make a diagnosis of SCLEC by exclusion. We excluded another case of putative SCLEC from our study because it showed morphological and immunophenotypical features suggestive for STK11 adnexal tumor on pathological review; molecular analysis confirmed the presence of STK11 alteration in that case.

Another novelty of our study is the finding of nuclear β-catenin positivity in 8/9 SCLECs. The presence of nuclear β-catenin accumulation in SCLEC might suggest that it is involved in the development of sex cord-like features. In fact, ECs with nuclear β-catenin often show altered differentiation, such as morular metaplasia, “corded and hyalinized” pattern, and pilomatrix carcinoma-like morphology [22,23,24,25]. As observed in morular metaplasia, in our series nuclear β-catenin was accompanied by CDX2 positivity [26]. Since CDX2 is used as a marker of gastrointestinal differentiation [27,28,29], its expression may be a potential pitfall if the differential diagnosis includes a metastasis from an extraovarian site, especially if Müllerian marker PAX8 is negative. In these cases, the moderate-to-strong positivity for ER and PR, combined with the negativity for the breast epithelial marker GATA3, may suggest the gynecological origin of the neoplasm; however, a comprehensive immunohistochemical evaluation remains crucial for a correct diagnosis.

Other possible differential diagnoses may include malignant Brenner tumor (which, unlike SCLEC, is typically ER/PR-negative and GATA3-positive) and EC with a transitional-like pattern (another morphological variant of EC with seemingly no clinical significance, more common than SCLEC) [17, 30].

Our study also attempted to provide the first TCGA-based assessment of SCLEC. We found that 8/9 tumors were NSMP; such a finding could be expected, given the similarities between SCLEC and conventional low-grade EC. No case fell into the MMR-deficient or POLE-mutant group; in this regard, these molecular groups are less common in ovarian carcinoma than in endometrial carcinoma [9]. One case in our series was p53-abnormal; this case also showed diffuse p16 expression and a high proliferation index (Ki67 = 90%). A subsequent molecular analysis performed for treatment purposes (and not par of our study) showed BRCA mutation, leading us to consider a diagnosis of high-grade serous carcinoma; however, this was excluded based on the low-grade nuclear atypia, the presence of squamous and morular differentiation, and the endometrioid-type immunophenotype (negativity for WT1, loss of PTEN expression, nuclear β-catenin accumulation) [31]. This case was therefore considered a bona fide p53-abnormal SCLEC, highlighting the molecular heterogeneity of this entity. In our previous paper, we also described a case of a BRCA-mutant high-grade serous carcinoma with a sex cord-like pattern [32]; these cases suggest a possible association between BRCA alterations and a sex cord-like architecture in ovarian carcinomas.

Limitations of our study mainly include the low number of tested genes and the relatively small sample size, which preclude to draw conclusions about the molecular background and TCGA classification of SCLEC.