Abstract
Background/Aim: Ovarian endometrioid carcinoma (EC) and high-grade serous carcinoma (HGSC) may exhibit various growth patterns and mimic mesonephric-like adenocarcinoma (MLA). We investigated the clinicopathological and molecular features of ovarian carcinomas with mesonephric-like differentiation (MLD). Patients and Methods: We analyzed the electronic medical records and pathology slides of two EC-MLD and three HGSC-MLD patients, and conducted immunostaining and targeted sequencing of their samples. Results: All cases showed architectural diversity, compactly aggregated small tubules and ducts, and eosinophilic intraluminal secretions, indicating the possibility of an ovarian MLA. However, the following histological and immunophenotypical features confirmed the diagnoses of EC-MLD and HGSC-MLD: squamous, tubal, and sertoliform differentiation; serous tubal intraepithelial carcinoma; solid, endometrioid, transitional (SET) feature; solid, transitional, endometrioid, mucinous-like (STEM) feature; diffuse expression of hormone receptors and Wilms tumor 1; mutant p53 immunostaining pattern; and wild-type v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog gene. Conclusion: A subset of ovarian ECs and HGSCs can display MLD and mimic an MLA. A thorough histological examination combined with ancillary tests is crucial to differentiate between these ovarian neoplastic entities.
- Ovary
- endometrioid carcinoma
- high-grade serous carcinoma
- mesonephric-like differentiation
- mesonephric-like adenocarcinoma
Ovarian carcinoma is the most lethal malignancy of the female genital tract (1, 2), and is characteristically diagnosed at an advanced stage with extensive peritoneal involvement at initial presentation (3). Advances in diagnostic methods and therapeutic strategies for the treatment of ovarian carcinoma have significantly improved the survival rates, although the mortality rate remains high (4). Ovarian carcinoma comprises a morphologically heterogeneous group of entities with different epidemiological, clinicopathological, and molecular characteristics. The five main histological subtypes are high-grade serous carcinoma (HGSC; 70%), endometrioid carcinoma (EC; 10%), clear cell carcinoma (10%), mucinous carcinoma (<5%), and low-grade serous carcinoma (<5%) (5). Among these subtypes, HGSC and EC are distinct entities, as indicated by differences in their clinical presentation, genetic risk factors, precursor lesions, patterns of spread, mutational profiles, and prognosis (6).
Mesonephric adenocarcinoma (MA) is a rare malignant tumor arising from the embryonal remnants of the mesonephric ducts and tubules (7). MA of the uterine cervix or vagina typically arises in association with the mesonephric remnants (MNRs), and is characterized by various histological features, including small tubular, glandular, papillary, solid, retiform, and sex cord-like (8-11). Moreover, several cases of MA involving the uterine corpus and the ovaries have been reported (9, 10, 12-17). As its association with MNRs has not been definitively established, the MA of the upper female genital tract is called mesonephric-like adenocarcinoma (MLA). Both MA and MLA share a unique immunophenotype: immunoreactivities for GATA-binding protein 3 (GATA3) and paired box 2 (PAX2), lack of hormone receptor expression, and wild-type p53 immunostaining pattern. Furthermore, the majority of MAs characteristically harbor pathogenic mutations in the v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) gene (18).
We recently diagnosed several cases of ovarian ECs and HGSCs that exhibited histological features and immunophenotypes that were similar to those of ovarian MLA, which was our primary consideration in each case. However, we diagnosed these cases as ovarian EC or HGSC with mesonephric-like differentiation (EC-MLD or HGSC-MLD, respectively) because the morphological features, immunostaining results, and mutational profiles were insufficient to confirm a diagnosis of MLA. There are neither established diagnostic criteria for MLD nor reports of the clinical significance of MLD co-occurring with ovarian EC or HGSC. No study has differentiated the clinicopathological and genetic features of ovarian MLA from those of EC-MLD and HGSC-MLD.
In this study, we aimed to comprehensively analyze the clinical, pathological, and molecular characteristics of ovarian EC-MLDs and HGSC-MLDs and to characterize their differences with regard to MLA to facilitate the accurate recognition and pathological diagnosis of these rare but notable entities.
Patients and Methods
Case selection. We retrospectively reviewed the hematoxylin and eosin (H&E)-stained slides and electronic medical records of ovarian EC or HGSC patients who underwent surgical staging at the Samsung Medical Center (Seoul, Republic of Korea) between August 2019 and April 2021. Patients who met the following criteria were included in this study: histologically confirmed primary ovarian EC or HGSC; presence of histological features resembling MLA (diverse architectural patterns with distinct areas of compact proliferation of small tubules containing hyaline-like eosinophilic intraluminal secretions) (7-11, 16-27); treatment with curative-intent debulking surgery without a grossly visible residual tumor; no history of other malignancies; and the availability of viable tumor tissue for immunostaining and targeted sequencing. We identified two EC-MLD and three HGSC-MLD cases based on the abovementioned inclusion criteria. The study protocol (2020-11-001) was reviewed and approved by the Institutional Review Board of the Samsung Medical Center (Seoul, Republic of Korea). We conducted this study in accordance with the Ethical Principles for Medical Research involving Human Subjects outlined in the Declaration of Helsinki (28).
Electronic medical record review. We reviewed the electronic medical records or contacted the referring gynecologists to collect the following clinical information: age of patients; history of gynecological diseases; presenting symptoms; magnetic resonance imaging (MRI) findings; serum levels of cancer antigen (CA) 125 and CA 19-9; preoperative clinical impressions; surgical procedures; postoperative treatments; postoperative recurrences; the interval between surgery and the first postoperative recurrence (disease-free survival); current status; mortality information; and interval between surgery and the last follow-up (overall survival).
Slide review. Three board-certified gynecological pathologists examined all of the available H&E-stained slides by using light microscopy. We collected the following pathological information: the location and greatest dimensions of the tumors; tumor extension into the ovarian surface, salpinx, or uterus; any lymphovascular space invasion; metastasis to the peritoneum, lymph node, or distant organs that was detected intraoperatively; any pleural effusion; final pathological diagnosis; and the International Federation of Gynecology and Obstetrics (FIGO) stage (29). Moreover, we investigated the presence of eosinophilic intraluminal secretions, architectural diversity (e.g., tubular, ductal, papillary, transitional, solid, cystic, and sex cord-like patterns), serous tubal intraepithelial carcinoma (STIC) (5), endometriosis; squamous, tubal, or sertoliform differentiation; nuclear pleomorphism, conspicuous nucleoli, and atypical mitotic figures. Mitotic counts (per 10 high-power fields) were also evaluated. For each case, the most representative block was selected for immunostaining and targeted sequencing.
Immunostaining. Briefly, 4-μm-thick, formalin-fixed, paraffin-embedded (FFPE) slices were deparaffinized and rehydrated using a xylene and alcohol solution. Immunostaining was performed using the Bond Polymer Intense Detection System (Vision Biosystems, Mount Waverly, Victoria, Australia) (2, 30-37). After antigen retrieval, the slices were incubated with the following primary antibodies: Wilms tumor 1 (WT1; 1:800, clone 6F-H2, Cell Marque, Rocklin, CA, USA); p53 (1:300, clone DO-7, Novocastra, Leica Biosystems, Newcastle Upon Tyne, UK); p16 (prediluted, clone E6H4, Ventana Medical Systems, Oro Valley, AZ, USA); estrogen receptor (ER; 1:150, clone 6F11, Novocastra), progesterone receptor (PR; 1:100, clone 16, Novocastra), phosphatase and tensin homolog deleted on chromosome 10 (PTEN; prediluted, clone SP218, Ventana Medical Systems); PAX2 (1:100, polyclonal, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA); and GATA3 (1:400, clone L50-823, Cell Marque). After chromogenic visualization, the slices were counterstained with hematoxylin. Appropriate positive and negative controls were concurrently stained to validate the method. For positive controls, we used normal salpinx (for WT1), endometrial SC (for p16 and p53) and EC (for ER and PR), normal proliferative endometrium (for PTEN), uterine MLA (for PAX2), and invasive breast carcinoma of no specific type (for GATA3). We prepared negative controls by substituting non-immune serum for primary antibodies, which resulted in no detectable staining. The nuclear (for WT1, ER, PR, PAX2, and GATA3) and cytoplasmic (for PTEN) staining intensity (weak, moderate, or strong) and proportion (focal or diffuse) were evaluated. The p53 immunostaining pattern was interpreted as a mutant pattern when one of the following patterns was observed: diffuse and strong nuclear immunoreactivity in 75% or more of the tumor cells (over-expression pattern); no nuclear immunoreactivity in any of the tumor cells (complete absence pattern); and an unequivocal diffuse cytoplasmic staining (cytoplasmic pattern) (38). In contrast, p53 expression was interpreted as a wild-type pattern if a variable proportion of tumor cell nuclei expressed p53 protein with mild-to-moderate staining intensity (39, 40). The p16 immunostaining pattern was interpreted as diffuse and strong positive when p16 expression uniformly and intensely involved the nuclei or the nuclei plus cytoplasm. All other p16 immunostaining patterns – described as focal nuclear staining or blob-like, puddled, or scattered cytoplasmic staining – were interpreted as patchy positive staining (7, 41-43).
Targeted sequencing. DNA and RNA were isolated from 10-μm-thick slices of FFPE tumor tissue using a sterile 26-gauge needle and RecoverAll Multi-Sample RNA/DNA Isolation Workflow (Thermo Fisher Scientific). Tissues obtained by manual microdissection were subjected to DNA and RNA extraction for the preparation of the library. Normal tissue was obtained from the adjacent non-neoplastic area. DNA and RNA were quantified using the Qubit 2.0 Fluorometer (Thermo Fisher Scientific), and DNA libraries were prepared using previously described methods (44, 45). These DNA libraries were generated from 20 ng of DNA per sample using an Ion AmpliSeq Library Kit 2.0 and the Oncomine Comprehensive Assay (OCA) v1 panel (both from Thermo Fisher Scientific) (45-47). RNA libraries were generated from 15 ng of RNA per sample using the Ion AmpliSeq RNA Library Kit, and libraries were quantified using the Ion Library Universal Quantification Kit (both from Thermo Fisher Scientific). The OCA v1 panel (Thermo Fisher Scientific) included 143 genes, of which 73 oncogenes were evaluated for mutational hotspots and 26 tumor-suppressor genes were interrogated for all exons. The panel facilitated the detection of copy number variations (CNVs) in 49 genes and fusion drivers in 22 genes. The gene list is available at https://www.thermofisher.com/kr/ko/home/clinical/preclinical-companion-diagnostic-development/oncomine-oncology/oncomine-cancer-research-panel-workflow.html. Consecutively, a 60 pmol/l pool of DNA:RNA libraries constituted at a 4:1 ratio was used to prepare the templated Ion Sphere Particle (Thermo Fisher Scientific). Sequencing was performed using the Ion 540 Kit-Chef (Thermo Fisher Scientific) and Ion S5 system (Thermo Fisher Scientific). The sequencing data of approximately 200 base pair reads were generated after 500 flow runs.
Bioinformatics and data analysis pipeline. The analysis of the sequencing data was performed using Torrent Suite Software v5.2.2 (Thermo Fisher Scientific). This workflow was created by adding a custom hotspots Browser Extensible Data file to report mutations of interest and a custom CNV baseline (described below) using the manufacturer’s default workflow as described previously (44, 45). The pipeline included signal processing, base calling, quality-score assignment, adapter trimming, read mapping to the human genome assembly GRCh37, quality control of mapping, coverage analysis with down-sampling, and variant calling. The variants were identified using the Torrent Variant Caller plug-in and Ion Reporter Software v5.2 (Thermo Fisher Scientific). Coverage maps were generated using the Coverage Analysis plug-in (Thermo Fisher Scientific). Additionally, ANNOtate VARiation (ANNOVAR; http://annovar.openbioinformatics.org/) was used for the functional annotation of the identified single-nucleotide polymorphisms (SNPs) to investigate their genomic locations and variation (48). To eliminate artifact errors, sequencing data were visually confirmed using the Integrative Genomics Viewer (Broad Institute, Cambridge, MA, USA). This workflow could report SNPs and indels in as low as 1% of the variant allele fraction. Based on the results of a feasibility study, the variant allele fraction threshold was established at 5%. Copy number analysis was performed using the copy number module within the aforementioned workflow of the Ion Reporter Software v5.2 (Thermo Fisher Scientific). Copy numbers ≥4 were considered concordant if the orthogonal assay also reported a copy number ≥4 for target genes. Fusions were detected using the fusion detection module within the Ion Reporter Software (Thermo Fisher Scientific) workflow. This pipeline only reported fusions that were annotated previously, as defined in a reference file that was preloaded into the workflow (44, 45).
Results
Clinical characteristics. Table I summarizes the clinical characteristics. The mean age of the five patients was 50 years (range=33-59 years), and two patients were postmenopausal. Three patients had a history of uterine leiomyoma, but none had a history of any malignancy. The patients presented with an abdominal mass, abdominal distension, vaginal discharge, or vaginal bleeding. MRI findings were available for all patients. The mean size of the ovarian masses was 7.9 cm (range=6.0-10.5 cm) on imaging. Four tumors appeared as mixed solid and cystic lesions, and one was a purely solid mass. Two EC-MLDs (cases 1 and 2) and one HGSC-MLD (case 3) did not have lymph node metastasis or peritoneal seeding, whereas two patients with HGSC-MLDs had multiple enlarged pelvic and para-aortic lymph nodes (case 4), peritoneal seeding (cases 4 and 5), and ascites (case 4). The preoperative serum CA-125 levels of three HGSC-MLD patients were elevated up to 4,125.1, 1,339.2, and 96.5 U/ml (case 4, before neoadjuvant chemotherapy; case 5; and case 3, respectively). Preoperative biopsy was unavailable in all the patients. Four patients underwent primary debulking surgery for clinically diagnosed ovarian cancer. One patient (case 4) underwent neoadjuvant chemotherapy followed by an interval debulking surgery.
Clinical features.
Table II summarizes the postoperative clinical course and status of the patients. All patients received postoperative platinum and taxane-based combination chemotherapy. Two EC-MLDs (cases 1 and 2) and two HGSC-MLDs (cases 3 and 5) patients completed six cycles. One HGSC-MLD patient (case 4) is currently receiving the third cycle of adjuvant chemotherapy. Three patients (cases 1, 2, and 3) were alive without evidence of recurrent disease at the time of this analysis. One HGSC-MLD patient (case 5) with FIGO stage IVA disease developed recurrence 8 months after postoperative chemotherapy and subsequently received four cycles of second-line chemotherapy for multiple recurrences and distant metastases; thereafter, the patient refused further treatment and was lost to follow-up.
Postoperative clinical course.
Pathological characteristics. Table III summarizes the pathological features used for staging. The tumor dimension at the greatest diameter ranged 5.0-10.5 cm (mean, 7.1 cm). All tumors involved the ovarian surfaces. Three HGSC-MLDs extended into the uni- or bilateral salpinges. Lymphovascular space invasion was detected in two HGSC-MLDs; however, lymph node metastasis was observed in only one of the two cases. Multiple pelvic and extrapelvic peritoneal metastases were histologically confirmed in two HGSC-MLDs. One HGSC-MLD patient (case 5) had malignant cells in the pleural effusion. Both EC-MLDs were staged as IC (ovarian surface extension). In contrast, the initial pathological FIGO stages of the three HGSC-MLDs were IIA (salpingeal involvement), IIIC (extrapelvic peritoneal involvement), and IVA (positive pleural fluid cytology).
Pathological features for staging.
Table IV summarizes the morphological features used for the determination of histological subtypes. We evaluated the detailed histological characteristics of EC-MLD and HGSC-MLD based on the parameters that correspond to the MLA. Representative photomicrographs of each case are shown in the order of appearance (i.e., case 1 in Figure 1; case 2 in Figure 2; case 3 in Figure 3; case 4 in Figure 4; and case 5 in Figure 5). All tumors appeared deeply basophilic at scanning view due to nuclear hyperchromasia with a high nuclear-to-cytoplasmic ratio. In all tumors, tubules and ducts contained hyaline-like eosinophilic intraluminal secretions, which were readily detectable at low-power magnifications. In addition to the secretions, intraluminal fibrin, histiocytes, inflammatory cells, and necrotic debris were noted in many areas of all tumors except one. All tumors were architecturally heterogeneous, with various combinations of tubular, ductal, papillary, transitional, solid, cystic, and sex cord-like patterns, which frequently merged with each other. The two most dominant architectural patterns were tubular and ductal in two EC-MLDs and two HGSC-MLDs. In particular, in all cases, the tubular pattern constituted at least 25% of the entire tumor area. One HGSC-MLD (case 4) showed areas of complex glandular proliferation with cribriform architecture that was nearly identical to that of low-grade EC, which occupied more than half of the tumor. In the other HGSC-MLD (case 3), papillary (30%) and tubular (30%) patterns were the most dominant patterns, followed by solid (15%), ductal (10%), transitional (10%), and cystic (5%) patterns. The architectural grade of an EC-MLD (case 2) was 2 (15% of solid growth), but the final FIGO grade was upgraded to 3 based on the severe nuclear pleomorphism observed in more than half of the tumor cells, and the brisk mitotic activity (32 mitotic figures/10 high-power fields). The other EC-MLD (case 1) was graded as FIGO 2 because it displayed 10% solid growth, mild-to-moderate nuclear pleomorphism, and moderate mitotic activity (8 mitotic figures/10 high-power fields). All three HGSC-MLDs were classified as grade 3 based on the predominantly severe nuclear pleomorphism, high mitotic rates (range=32-41 mitotic figures/10 high-power fields), and easily identifiable atypical mitoses.
Pathological features for determining the histological subtype.
Case 1 (endometrioid carcinoma with mesonephric-like, squamous, and sertoliform differentiation): (A) The tumor predominantly consists of compact aggregates of tubular (green asterisk) and ductal (blue asterisks) structures with variable amounts of intervening stroma. Note some large, cystically dilated glands (left lower corner). (B) The tubules and microcysts contain eosinophilic intraluminal secretions. (C) The ductal proliferation closely resembles glandular crowding and the back-to-back arrangement of endometrioid carcinoma. Densely eosinophilic intraluminal hyaline-like substances are easily identifiable. (D) A few microscopic foci show mature squamous morules (purple arrows). (E) A benign endometriotic cyst (yellow asterisk) is located adjacent to the tumor (blue asterisk). (F) Some areas of sertoliform differentiation display aggregates of polygonal cells with abundant eosinophilic cytoplasm (orange arrows); their round and relatively regular nuclei are smaller than those of the adjacent neoplastic glandular epithelium (lower half). (G) In addition to the luteinized cells resembling the Leydig cells observed in (F), the small tubules lined by cells with oval nuclei and eosinophilic cytoplasm resemble those of a Sertoli cell tumor (orange arrows). (H–K) Immunostaining revealed (H) uniform and strong nuclear immunoreactivity for estrogen receptor, (I) focal and weak nuclear positivity for paired box 2 (small gray arrows), (J) lack of nuclear GATA-binding protein 3 expression, and (K) wild-type p53 expression pattern. Staining method: (A–G) Hematoxylin and eosin staining; (H–K) immunostaining using polymer method. Original magnification: (A) ×40; (B–D) ×100; (E) ×40; (F) ×100; (G) ×200; (H) ×100; and (I–K) ×200.
Case 2 (high-grade endometrioid carcinoma with mesonephric-like differentiation): (A) The tumor displays ductal and tubular structures infiltrating destructively into the hyalinized stroma. (B) Some of the compactly aggregated and fused small tubules contain lightly eosinophilic intraluminal secretions. (C) Ovoid or stellate-shaped ductal lumina possess variable amounts of hyaline-like eosinophilic materials (green arrows). (D) Areas resembling endometrioid carcinoma show a complex proliferation of stratified columnar epithelium with cribriform architecture. (E) Some dilated glands have intraluminal papillary projections and anastomosing tumor cell cords. A few foci show high-grade nuclear atypia characterized by bizarre pleomorphism and marked hyperchromasia (blue arrows). (F) Some tumor cell nuclei are more than four times the size of the adjacent nuclei (purple arrows), and have atypical mitoses (yellow arrow). (G–J) Immunostaining revealed a complete absence of immunoreactivities for (G) estrogen receptor, (H) progesterone receptor, and (I) Wilms tumor 1 and (J) diffuse and strong nuclear p53 positivity (over-expression pattern). Staining method: (A–F) Hematoxylin and eosin staining; (G–J) immunostaining using the polymer method. Original magnification: (A) ×40; (B and C) ×200; (D) ×150; (E) ×100; (F–H) ×200; (I) ×100; and (J) ×200.
Case 3 [high-grade serous carcinoma (HGSC) with mesonephric-like differentiation and solid, endometrioid, transitional (SET) feature]: (A–D) Various architectural patterns including (A) tubular and ductal, (B) transitional, (C) papillary, and (D) solid patterns are evident. These patterns frequently merged with each other. (E) Although densely eosinophilic intraluminal substance (green arrows) indicates the possibility of mesonephric-like adenocarcinoma (MLA), the severe nuclear pleomorphism observed in almost all of the tumor cells in this image does not support a diagnosis of MLA, but instead, indicates an HGSC. (F) Paired box 2 (PAX2) immunostaining reveals that approximately half of the tumor cells show weak cytoplasmic PAX2 immunoreactivity, and some of them also showed intranuclear protein expression. (G) A weak-to-moderate staining intensity is noted for the nuclear PAX2 expression. (H) Note a diffuse cytoplasmic GATA-binding 3 (GATA3) expression. (I) In some areas, GATA3 immunoreactivity is concentrated in the perinuclear regions of the positive cells. Staining method: (A–E) Hematoxylin and eosin staining; (F–I) immunostaining using the polymer method. Original magnification: (A) ×40; (B–D) ×100; (E) ×200; (F) ×100; (G) ×400; (H) ×100; and (I) ×200.
Case 4 (high-grade serous carcinoma with mesonephric-like differentiation): (A) The most dominant patterns of this tumor are ductal and tubular. Densely eosinophilic intraluminal substances (gray arrows) are readily identifiable even at the low-power magnification. (B) Another low-power magnification image closely resembles a low-grade endometrioid carcinoma because of the predominance of the ductal pattern. (C) Compactly aggregated small tubules have inter-anastomosing slit-like lumina. (D) Some endometrioid-like glands are filled with variable amounts of hyaline-like secretions. (E) Despite the presence of intraluminal secretions (green arrow), the severe-to-marked nuclear pleomorphism in the same tubules and the adjacent poorly formed tubules (blue asterisks) support a diagnosis of high-grade serous carcinoma. (F) A few microscopic areas show microcysts filled with light basophilic or amphophilic, mucin-like substances. (G) Preoperative neoadjuvant chemotherapy led to histiocytic aggregates (orange arrows) and highly pleomorphic tumor cells with bizarre nuclei, deeply eosinophilic cytoplasm, and a low nuclear-to-cytoplasmic ratio. (H–K) Immunostaining revealed (H) diffuse and strong Wilms tumor 1 positivity, (I) wild-type p53 immunostaining pattern, and (J) focal but strong nuclear paired box 2 (PAX2) immunoreactivity (purple arrows). (K) In addition to diffuse and weak cytoplasmic expression, the nuclear immunoreactivity for PAX2 has a patchy distribution. The nuclei highlighted intensely by PAX2 protein are randomly distributed in each neoplastic gland. Staining method: (A–G) Hematoxylin and eosin staining; (H–K) immunostaining using the polymer method. Original magnification: (A and B) ×40; (C) ×100; (D) ×200; (E and F) ×400; (G) ×200; (H) ×100; (I) ×200; (J) ×40; and (K) ×200.
Case 5 [high-grade serous carcinoma (HGSC) with mesonephric-like differentiation and solid, transitional, endometrioid, mucinous-like (STEM) feature]: (A–E) Architectural diversity, consisting of (A) tubular, (B) ductal, (C) papillary, (D) solid and transitional, and (E) sex cord-like patterns, indicates the possibility of mesonephric-like adenocarcinoma (MLA). (F) However, the presence of serous tubal intraepithelial carcinoma (STIC) rules out MLA and supports a diagnosis of HGSC. Note the transition from the normal tubal epithelium (orange arrow) to STIC (green arrows) that is characterized by epithelial stratification, loss of polarity, papillary configuration with surface fragmentation, and a high nuclear-to-cytoplasmic ratio. (G and H) Several microscopic areas display the formation of microcystic and cribriform structures associated with mucin production. (I–K) Immunostaining revealed (I) uniform nuclear Wilms tumor 1 expression, (J) lack of GATA-binding protein 3, and (K) p53 over-expression. (L–K) Mucin staining in three different microscopic areas reveals that the intraluminal mucin and the cytoplasmic vacuoles of mucin-producing tumor cells are positive for (L) mucicarmine, (M) Alcian blue, and (N) periodic acid-Schiff with diastase. These histological features are compatible with an HGSC with the STEM feature. Staining method: (A–H) Hematoxylin and eosin staining; (I–K) immunostaining using the polymer method; (L) mucicarmine; (M) Alcian blue; and (N) periodic acid-Schiff with diastase. Original magnification: (A–C) ×40; (D) ×100; (E) ×40; (F) ×100; (G and H) ×300; (I) ×40; (J and K) ×100; and (L–N) ×200.
Despite the presence of morphological findings suggestive of MLA, three of the five tumors (case 1, 3, and 5) had histological features compatible with the criteria for either EC or HGSC. One EC-MLD (case 1) exhibited foci of squamous and tubal differentiation. Mature squamous morules were identified in some areas showing ductal patterns. Furthermore, this tumor showed sertoliform differentiation comprising aggregates of luteinized cells with abundant eosinophilic cytoplasm and small tubules lined by cells with oval nuclei and eosinophilic cytoplasm that resembled a Sertoli cell tumor. Moreover, we found a benign endometriotic cyst, which was located adjacent to the tumor and strongly supported a diagnosis of EC. In one HGSC-MLD (case 3), the tumor cells were predominantly arranged in sheets (solid) or back-to-back nests with microlumina (endometrioid) and formed broad papillae with thick, stratified epithelium (transitional), compatible with the solid, endometrioid, transitional (SET) feature (49). The other HGSC-MLD (case 5) had histological features similar to those of case 3. Most of the tumor comprised a SET pattern, whereas there was limited conventional HGSC morphology (papillary and micropapillary patterns). In addition, we noted that some tumor cells possessed intracytoplasmic mucin and formed microcystic structures containing intraluminal mucin. Mucicarmine, Alcian blue, and periodic acid-Schiff with diastase were used to highlight the intracytoplasmic and intraluminal mucin. These histological features were compatible with solid, transitional, endometrioid, mucinous-like (STEM) features (50). The abovementioned morphological variations, including squamous, tubal, and sertoliform differentiation; SET feature; and STEM feature, argued against a diagnosis of MLA.
Immunostaining results (Table V). Tumor tissue samples for immunostaining were available in all cases. Diffuse expressions of both ER and PR with strong staining intensity confirmed the endometrioid subtype in grade 2 EC-MLD (case 1). However, grade 3 EC-MLD (case 2) demonstrated a complete absence of hormone receptor expression. Instead, uniform and intense nuclear p53 immunoreactivity in almost all of the tumor cells (p53 over-expression) correlated with its high-grade histology. Diffuse and strong nuclear WT1 immunoreactivity confirmed the serous subtype in all HGSC-MLD patients (cases 3, 4, and 5). Two of the three cases showed mutant p53 immunostaining pattern, with either complete absence (case 3) or over-expression (case 5). In contrast, wild-type p53 expression pattern, that is, patchy nuclear p53 staining with weak-to-moderate intensity, was observed in one HGSC-MLD (case 4). One EC-MLD and two HGSC-MLDs exhibited focal nuclear PAX2 immunoreactivity with variable staining intensity. In particular, one HGSC-MLD (case 5) had multifocal microscopic areas of strong nuclear PAX2 expression. Cases 1, 3, 4, and 5 displayed weak cytoplasmic PAX2 immunoreactivity with variable staining proportion. None of the cases showed nuclear GATA3 immunoreactivity, although cases 1, 3, and 5 demonstrated focal cytoplasmic GATA3 expression. Of note, case 3 (HGSC-MLD) had a single microscopic area of strong cytoplasmic GATA3 positivity with a dotted pattern and perinuclear concentration.
Results of immunostaining and targeted sequencing.
Targeted sequencing results (Table V). Tumoral tissue samples for targeted sequencing were available in all cases, and none harbored a pathogenic KRAS mutation. One EC-MLD (case 1) harbored pathogenic mutations in phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA; c3140A>G), β-catenin (CTNNB1; c.110C>T), and v-Raf murine sarcoma viral oncogene homolog B1 (BRAF; c.1742A>T). The other EC-MLD (case 2) had a pathogenic missense mutation in tumor protein 53 (TP53; c.743G>A), which is concordant with p53 protein over-expression. Nonsense (c.438G>A; case 3) and missense (c.404G>A; case 5) TP53 mutations were observed in two HGSC-MLDs, concordant to the mutant p53 expression patterns (complete absence in case 3 and over-expression in case 5, respectively). The remaining HGSC-MLD case (case 4), demonstrating wild-type p53 immunostaining pattern, harbored a pathogenic splice site mutation in TP53 (c.919+1G>A).
Discussion
Ovarian MLA is a rare histological subtype of the female genital tract tumors that is poorly recognized and listed as a distinct entity in the latest version of World Health Organization Classification of Female Genital Tumors (5). MLA displays areas of relatively well-formed tubular and glandular structures, which are closely aggregated, and back-to-back small tubules and ducts that are lined by cuboidal cells. Their lumina contain densely eosinophilic secretions (27). Additional architectural patterns include papillary, solid, retiform, sex cord-like, spindle, and glomeruloid features (7, 27). These histological findings can easily be mistaken for EC, HGSC, and other epithelial or mesenchymal tumors by inexperienced pathologists. Furthermore, a small subset of MLAs harbor mutations in PIK3CA, PTEN, CTNNB1, ARID1A, and TP53, similar to ovarian EC or HGSC (11, 25-27, 40). These morphological and genetic overlaps between MLA and the more common subtypes of ovarian carcinoma may evoke the question of whether MLA is a separate entity from or a subtype of Mullerian-origin carcinoma. Thus, in lesions with the abovementioned features, MLA should be included in the differential diagnosis.
In this study, we described five cases of ovarian EC and HGSC that mimic MLA. Each tumor demonstrated areas showing classic morphological features of MLA, which coexist with findings supportive of an EC (squamous, tubal, and sertoliform differentiation) and HGSC (severe-to-marked nuclear pleomorphism observed throughout the tumor, STIC, SET feature, and STEM feature) diagnosis. The immunophenotype of each tumor further supported the final diagnoses of EC-MLD and HGSC-MLD. Diffuse and strong expression for hormone receptors indicated EC, and those for WT1 indicated HGSC. Additionally, in one EC-MLD, we found a benign endometriotic cyst adjacent to the neoplastic glands within the tumor. The presence of endometriosis in association or in close proximity to the tumor indicates an endometrioid subtype. However, several cases of ovarian MLA have recently been documented to coexist with various Mullerian lesions including endometriosis, EC, serous cystadenoma, serous adenofibroma, serous borderline tumor, and low-grade serous carcinoma (51-55). These data suggest that the presence of an endometriotic cyst is not specific for EC and support the possibility that at least some cases of ovarian MLA are of Mullerian origin and transdifferentiate along mesonephric lines (53).
Targeted sequencing analysis revealed that none of the cases harbored a pathogenic KRAS mutation, which is the most characteristic molecular alteration of MLA (7). Instead, TP53 mutations were detected in one EC-MLD and three HGSC-MLDs. Specifically, EC-MLD with high-grade histology (case 2) harbored a missense TP53 mutation, three HGSC-MLDs had a nonsense (case 3), splicing (case 4), and missense (case 5) TP53 mutation, respectively. It was notable that one HGSC-MLD (case 4) that exhibited wild-type p53 expression pattern on immunostaining was found to have a splicing TP53 mutation on sequencing. These results are consistent with our previous observations that a small subset of HGSCs showing wild-type p53 immunostaining pattern harbored splice-site TP53 mutations that posed a diagnostic challenge (56, 57). Along with a careful microscopic examination, using the appropriate ancillary test is critical to prevent misclassification.
Ovarian HGSCs with homologous recombination deficiency have been documented to exhibit unique histological features: mixed solid, pseudoendometrioid, and transitional cell carcinoma-like patterns with pushing margins (SET feature) (49, 58). Furthermore, a recent case of ovarian HGSC demonstrating both features of SET and mucinous differentiation has been reported, and the authors diagnosed this tumor as an HGSC with the STEM feature (50). In HGSC cases, when the proportion of solid, endometrioid-like, and transitional architecture overwhelms that of the papillary and micropapillary patterns, especially when combined with mucinous differentiation, the tumor can mimic MLA at first sight. This is the first report to describe that ovarian HGSCs with the SET or STEM feature can resemble MLA.
PAX2 and GATA3 have been used as immunohistochemical markers for determining mesonephric tumor origin. To diagnose MLA that occurs outside the uterine cervix without evidence of an association with MNRs, the immunophenotypical identification of the mesonephric origin using those markers is useful. PAX2 is a protein that is associated with the development of the Wolffian system and is typically expressed in mesonephric tumors (59); GATA3 is considered the best overall marker for the mesonephric lineage with high sensitivity and specificity (11). However, previous studies showed that a small proportion of other benign and malignant Mullerian lesions exhibited immunoreactivities for both proteins with variable staining intensity and proportion (11, 27, 59). In this study, we observed that both EC-MLD and HGSC-MLD had focal nuclear PAX2 immunoreactivities with variable staining intensity. Therefore, we recommend an immunostaining panel that includes WT1, p53, and hormone receptors.
Due to the small sample size, we were unable to examine the clinical significance of MLD by comparing these cases with the more common subtypes of ovarian carcinomas without MLD and MLAs. Furthermore, MLA has been only recently included as a separate entity in the latest version of World Health Organization Classification. As sufficient data on the ovarian MLA have not been accumulated yet, the clinicopathological features of the more common ovarian carcinoma subtypes with MLD have seldom been investigated. From the pathologists’ perspective, one should be aware that certain ovarian tumors may exhibit MLD and the use of an immunostaining panel is inevitable for making the correct diagnosis. Further investigations using a larger cohort are required to understand the nature of MLD.
In summary, we collected five cases of ovarian carcinoma with MLD and thoroughly described the clinical, histological, immunohistochemical, and molecular features. From our cases, we learned that MLA should not be diagnosed solely on the basis of compact tubular structures and intraluminal secretions. Some morphological features can facilitate the correct diagnosis: the presence of squamous, tubal, and sertoliform differentiation and diffuse and strong expression of ER and PR favor EC. Severe-to-marked nuclear pleomorphism observed throughout the tumor, STIC, SET feature, STEM feature, diffuse and strong expression for WT1, pathogenic TP53 mutation, and aberrant p53 expression favor HGSC. Although ovarian EC or HGSC can exhibit various growth patterns and mimic MLA, a thorough histological examination combined with ancillary tests is critical to prevent misdiagnosis. A compact proliferation of small tubules with back-to-back arrangement and densely eosinophilic intraluminal secretions may not be specific for MLA but suggest another variant of morphological differentiation (e.g., MLD) that can be observed in EC and HGSC.
Acknowledgements
This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (NRF-2019R1G1A1100578).
Footnotes
Authors’ Contributions
All Authors made substantial contributions to the conception and design of the study; the acquisition, analysis, and interpretation of the data; drafting of the article; critical revision of the article for important intellectual content; and the final approval of the version to be published.
Conflicts of Interest
The Authors declare that they have no conflicts of interest in relation to this study.
- Received June 20, 2021.
- Revision received July 7, 2021.
- Accepted July 8, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- Mesonephric-like Carcinosarcoma of the Uterine Corpus: Clinicopathological, Molecular and Prognostic Characteristics in Comparison With Uterine Mesonephric-like Adenocarcinoma and Conventional Endometrial Carcinosarcoma
- Mesonephric-like Adenocarcinoma of the Uterine Corpus: Comprehensive Analyses of Clinicopathological, Molecular, and Prognostic Characteristics With Retrospective Review of 237 Endometrial Carcinoma Cases