Identifying mesonephric-like adenocarcinoma of the endometrium by combining SOX17 and PAX8 immunohistochemistry.

Seventeen endometrial and ovarian MLAs were collected from multiple institutions and further reviewed to confirm the diagnosis. All 17 MLAs showed diffuse strong staining for PAX8, variable staining for GATA3 and TTF1, luminal staining for CD10, patchy staining for p16, wild‐type staining for p53 and negative staining for ER and PR, consistent with previous findings. Surprisingly, all 17 MLAs showed either complete negative ( n  = 10) or focal weak/moderate ( n  = 7) staining for SOX17, which is almost always more diffuse and stronger than PAX8 in other subtypes of endometrial carcinomas. Additionally, TRPS1 was negative in all cases (Table  2 ; Figure  1 ). SOX17 and PAX8 results in 17 confirmed MLA cases from four institutions The numbers in the table represent percentages of stained tumour cells. ER, oestrogen receptor; NA, not available; PR, progesterone receptor; WT, wild‐type. Two representative mesonephric‐like adenocarcinomas (MLA) with multiple immunohistochemistry (IHC). A–F , One MLA shows negative SOX17 ( B ), diffuse positive PAX8 ( C ), focal strong TTF1 ( D ), negative GATA3 ( E ) and focal positive CD10 ( F ). ( G – L ) Another MLA shows very focal weak SOX17 ( H ), diffuse strong positive PAX8 ( I ), positive TTF1 ( J ), positive GATA3 ( K ) and positive CD10 ( L ). [Color figure can be viewed at wileyonlinelibrary.com ] This result encouraged us to screen TMAs with 652 endometrial carcinomas diagnosed before the MLA era at The Ohio State University using SOX17 and PAX8. Most cases showed positive staining for both SOX17 and PAX8 (88.7%, 578 of 652), 45 cases (6.9%) showed positive SOX17 and negative PAX8 staining (SOX17+/PAX8−), 14 cases (2.1%) showed positive PAX8 and negative SOX17 staining (SOX17−/PAX8+) and 15 cases (2.3%) showed negative staining for both SOX17 and PAX8 (SOX17−/PAX8−). Among 45 SOX17+/PAX8− cases, 38 had been diagnosed as endometrioid, six as malignant mixed Müllerian tumour (MMMT) and one as mixed carcinoma. Among 15 SOX17−/PAX8− cases, there were two endometrioid carcinomas, one serous carcinoma, six MMMT, one mixed carcinoma, four undifferentiated carcinomas and one other. Among 14 SOX17−/PAX8+ cases, eight were diagnosed as endometrioid, one as clear cell carcinoma, one as MMMT, three as mixed carcinomas and one as other (Table  3 ). SOX17 and PAX8 results in 652 endometrial carcinomas diagnosed before mesonephric‐like adenocarcinoma definition MMMT, malignant mixed Müllerian tumour. We further studied 14 SOX17−/PAX8+ cases by examining the morphology and performing additional IHCs (TTF1, GATA3, ER and CD10) on whole tissue sections. Seven (50%) cases were reclassified as MLA based on morphological (typical MLA morphological features) and immunostain results (positive CD10, TTF1 and/or GATA3 staining). Five of the seven cases demonstrated aggressive clinical outcomes with advanced disease (distant metastasis). Additionally, these seven MLA cases showed strong PAX8 staining, while the non‐MLA cases (cases 8, 9, 10, 12, 14 and 14 in Table  4 ) showed only weak to moderate PAX8 staining, except one case (case 11 in Table  4 ). This case (case 11 in Table  4 ) showed strong staining for PAX8 and ER and negative staining for SOX17 on TMA slides, but strong staining for both PAX8 and SOX17 on whole tissue sections, caused by intratumoural heterogeneity (Table  4 ). Immunohistochemical results and diagnosis of 14 SOX17‐negative/PAX8‐positive endometrial carcinoma cases MLA, mesonephric‐like adenocarcinoma; MMMT, malignant mixed Müllerian tumour; ER, oestrogen receptor. After diagnosing these seven cases as MLA, we further compared the expression of SOX17, PAX8 and ER in different types of endometrial carcinomas including endometrioid, serous, clear cell, MMMT, mixed carcinoma, undifferentiated carcinoma and MLAs. Overall, SOX17 was the most positive marker and ER was the least (cut‐off = 10%). SOX17 and PAX8 showed similar positivity in most tumour subtypes except FIGO grades 1 and 2 endometrioid carcinomas (SOX17 > PAX8) and MLAs (PAX8 > SOX17). ER showed less positivity than PAX8 or SOX17 in almost all subtypes except MLAs, which were negative or weakly positive for both ER and SOX17. Positive percentage and H scores were compared among these markers, with SOX17 showing the highest positive percentage and H scores and ER showing the lowest in all subtypes except MLAs (Table  5 ; Figure  2 ). Comparison of SOX17, ER and PAX8 expression (> 10%) in different types of endometrial carcinomas MLA, mesonephric‐like adenocarcinoma; MMMT, malignant mixed Müllerian tumour; ER, oestrogen receptor; NS, not significant. SOX17, PAX8 and oestrogen receptor (ER) immunohistochemistry (IHC) in different types of endometrial carcinoma. Row 1: H&E staining of endometrioid type carcinoma ( A ), one endometrioid type carcinoma shows strong SOX17 ( B ), very focal weak PAX8 ( C ) and very focal weak ER ( D ). Row 2: H&E staining of serous carcinoma ( E ), one serous carcinoma shows diffuse strong SOX17 ( F ), positive PAX8 ( G ) and focal positive ER ( H ). Row 3: H&E staining of clear cell carcinoma ( I ), one clear cell carcinoma (CCC) shows strong positive SOX17 ( J ), PAX8 ( K ) and negative ER ( L ). Row 4: H&E staining of mesonephric‐like adenocarcinoma ( M ), one mesonephric‐like adenocarcinoma (MLA) shows negative SOX17 ( N ), strong positive PAX8 ( O ) and negative ER ( P ). Column 1: haematoxylin and eosin (H&E) staining; column 2: SOX17 IHC; column 3: PAX8 IHC; column 4: ER IHC. [Color figure can be viewed at wileyonlinelibrary.com ]

The findings of this study shed light on the diagnostic significance of SOX17 expression in MLAs of the endometrium and ovary. MLAs represent a unique subset of gynaecological malignancies, sharing morphological and molecular characteristics with MAs of the cervix but lacking the association with mesonephric remnants or hyperplasia. 4 , 5 , 6 , 7 , 8 While previous research has identified key immunophenotypical and molecular features of MLAs, 6 , 10 , 11 , 12 the role of SOX17 expression in these tumours has remained unexplored until the present. Our study revealed that SOX17 expression in MLAs is either completely negative or weakly focal, contrasting with its typical diffuse and strong staining observed in other subtypes of endometrial carcinomas. This finding suggests a potential utility of SOX17 IHC in distinguishing MLAs from other endometrial carcinoma subtypes, such as endometrioid and serous carcinomas, which often exhibit positive staining for SOX17. Furthermore, our study utilised a rational combined IHC approach, incorporating SOX17 and PAX8 IHCs, to retrospectively identify MLAs from a cohort of endometrial carcinomas diagnosed before MLA was an established diagnostic category. This approach proved effective in identifying MLAs from cases with SOX17‐negative and PAX8‐strongly positive staining patterns as we were able to accurately diagnose MLAs from cases initially classified as other subtypes of endometrial carcinoma, highlighting the diagnostic utility of SOX17 IHC in this context. In addition, the comparison of SOX17, PAX8, and ER expression across different types of endometrial carcinomas revealed distinct staining patterns characteristic of different subtypes of endometrial carcinoma. Consistent with our and others’ previous findings, 26 , 27 , 28 SOX17 emerged as the most sensitive marker, demonstrating higher positivity rates and H scores compared to PAX8 and ER across most endometrial carcinoma subtypes. Notably, MLAs exhibited negative or weakly positive staining for both SOX17 and ER but strongly positive staining of PAX8, further emphasising the unique immunophenotypical profile of these tumours. The origin of MLAs has been debated, with conflicting evidence regarding their embryological derivation from Müllerian ducts or Wolffian (mesonephric) ducts. 17 , 18 , 19 While morphological, immunohistochemical and molecular profiling studies suggest a mesonephric/Wolffian origin for MLAs, their occurrence primarily within the endometrium and association with Müllerian‐type tumours challenge this hypothesis. 17 , 18 , 19 In contrast to SOX17's diffuse and strong staining typically observed in Müllerian‐derived tumours, MLAs showed negative or weakly positive staining; however, this may be caused by the transdifferentiation phenomenon. Further research utilising advanced molecular techniques and lineage tracing studies is warranted to elucidate the precise cellular origins of MLAs and clarify their embryological lineage. Our study is limited by the small cohort size of MLAs, the study's retrospective nature and the use of TMAs. Future studies with large cohorts from multiple institutions are warranted to ascertain current findings and confirm the diagnostic utility of this combined IHC approach. In conclusion, the differential expression patterns of SOX17 and PAX8 observed in MLAs present an opportunity to improve the accuracy of MLA diagnosis. Specifically, the detection of strong nuclear labelling for PAX8 coupled with negative staining for SOX17 could serve as a reliable indicator of MLA. By incorporating both SOX17 and PAX8 into diagnostic algorithms, pathologists can enhance their ability to distinguish MLAs from other endometrial carcinoma subtypes.

Mesonephric adencarcinomas (MA) of the uterine cervix are rare, aggressive entities and mainly associated with mesonephric remnants and/or hyperplasia. 1 , 2 , 3 However, an intriguing subset of endometrial and ovarian carcinomas, referred to as mesonephric‐like adenocarcinomas (MLA), share morphological, immunophenotypical and molecular attributes with MAs of the cervix. MLAs are characterised by a clinically aggressive course, often presenting at advanced stages with a predilection for pulmonary metastases. 4 , 5 , 6 , 7 , 8 Unlike cervical MAs, MLAs are not associated with mesonephric remnants and/or hyperplasia, but occur in the endometrium or are associated with endometriosis in the ovary. In 2016 McFarland, Quick and McCluggage delineated uterine corpus MLA from cervical MAs in a series including seven corpus and five ovarian examples. 6 The entity has since been included under the ‘other endometrial carcinomas’ in the current World Health Organisation (WHO) classification of female genital tumours. 9 MLAs present a diverse histological pattern, including tubular, glandular, spindled, solid and papillary structures, posing a diagnostic challenge due to morphological overlap with endometrioid carcinoma, serous carcinoma and carcinosarcoma. Immunophenotypically, MLAs exhibit positive staining for GATA3, TTF‐1, CD10 (luminal staining) and calretinin while being negative for oestrogen receptor (ER)‐ and progesterone receptor (PR)‐like MAs of the cervix. 6 , 10 , 11 , 12 Molecularly, these tumours share also key features with MAs of the cervix, including KRAS mutations, microsatellite stability and frequent gain of chromosome 1q, distinguishing them from other endometrial carcinoma types categorised based on molecular profiling. 5 , 7 , 13 , 14 , 15 , 16 Some studies also suggest that a subset of MLAs harbour alterations typically associated with endometrioid carcinoma, such as PTEN and PIK3CA mutations; however, this is controversial. 5 , 15 This molecular diversity has led to debates regarding the cellular origins of MLA. While morphological, immunohistochemical and molecular profiles suggest a mesonephric/Wolffian origin, the distribution of uterine tumours in the endometrium rather than the myometrium, coupled with associations with endometriosis and Müllerian‐type tumours, strongly supports a Müllerian origin. 17 , 18 , 19 SOX17 (SRY‐box transcription factor 17), a key player in embryonic development, is widely expressed in endometrial tissue and various visceral organs. 20 , 21 , 22 As a transcriptional regulator, recent studies link SOX17 to epithelial ovarian carcinoma, sharing an expression pattern with PAX8 and jointly influencing downstream genes related to the cell cycle and tissue morphogenesis. 23 , 24 Beyond ovarian carcinoma, SOX17 is implicated in cervical and endometrial carcinogenesis and is proposed as a tumour suppressor for endometrial adenocarcinoma. 23 , 24 , 25 Notably, research on pathological specimens, including studies by our group, identifies SOX17 as a sensitive and specific marker for gynaecological carcinomas. 26 , 27 , 28 However, SOX17 expression has not been explored in MLAs. In this study, we aimed to investigate SOX17 expression in MLAs together with other IHCs to differentiate MLAs from other endometrial carcinomas and then use a rational combined IHC approach to retrospectively identify MLAs from a study cohort harbouring endometrial carcinomas diagnosed prior to MLA definition.

None to declare.

The study cohort included 17 endometrial and ovarian MLAs from The Ohio State University (OSU), Brown University, The Johns Hopkins University and University of Texas MD Anderson Cancer Center during a study period from 2021 to 2023, and 652 endometrial carcinomas from OSU between 2012 and 2015. The pathology database at the OSU Wexner Medical Center was searched to retrieve 652 endometrial carcinomas with hysterectomies between April 2012 and January 2015. 29 , 30 The corresponding clinicopathological findings were collected. A formalin‐fixed paraffin‐embedded tissue block representative of the tumour was collected from each hysterectomy. Tissue microarrays (TMA) with triplicate 1‐mm cores for each tumour were constructed at the OSU pathology core facility. Immunohistochemical staining was performed on a Leica Bond III autostainer system (Leica Biosystems). Formalin‐fixed paraffin‐embedded tissue sections were deparaffinised/rehydrated and antigen retrieval was performed with Bond ER1 (Leica Biosystems, Richmond, VA, USA; equivalent to citrate buffer, pH 6.0) or Bond ER2 [Leica Biosystems; equivalent to ethylenediamine tetraacetic acid (EDTA) buffer, pH 8.0] at 100°C for 20 min. The primary antibody was incubated for 15 min at room temperature; it was detected using the Bond Polymer Refine Detection kit (cat. no. DS9800; Leica Biosystems) and diaminobenzidine chromogen. The tissues were then counterstained using Leica haematoxylin, provided as part of the Leica Bond Polymer Refine Detection kit. Normal endometrial tissues were used as a positive control. The primary antibodies used in this study are summarised in Table  1 . Each IHC was reviewed by two pathologists initially (M.T. and Z.L.), and in difficult cases additional pathologists were consulted and consensus was reached. For SOX17, PAX8, ER, PR and TRPS1 (TRPS1 is a recently identified marker for breast carcinoma, but was also found expressed in a small portion of gynaecological tumours), the positivity was defined using a cut‐off value of 10% of positive tumour cells with staining. The 10% was used to eliminate some non‐specific weak staining to increase specificity. The H ‐score was calculated by multiplying staining percentage (0–100) by intensity (1–3) to obtain a value from 0 to 300. Non‐homogenous staining within fewer than 50% of targeted cells was described as focal staining pattern. Antibodies used in the study RTU, ready‐to‐use. Statistical analysis was performed using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA). Categorical data (IHC positivity) were summarised as frequency and percentage, and continuous variables ( H ‐scores) as medians and ranges. Fisher's exact test was used to compare each variable between different groups. An unpaired t ‐test was used to analyse continuous variables. An adjusted P ‐value of ≤ 0.05 was considered significant.