Endometrioid Versus Seromucinous Borderline Ovarian Tumors: Divergent Molecular Signatures and a Shared Role as Precursors to Endometrioid Carcinoma

Case Selection The study was performed after approval from the institutional review board (IRB) of UT Southwestern Medical Center. We identified 11 cases with the diagnosis of EBT and 10 cases of SMBT between August 2015 and September 2025 at 2 UT Southwestern teaching hospitals (Clements University Hospital and Parkland Hospital) using the laboratory information systems and electronic medical records. Pathologic Evaluation Hematoxylin and eosin (H&E) and immunohistochemistry (IHC) slides were reviewed by three gynecologic pathologists (S.N., E.L., and H.C.) to verify the diagnoses and interpret marker expression. Standard histologic criteria were used for the diagnosis of EBT and SMBT. EBTs exhibited 2 growth patterns: adenofibromatous, with closely packed, crowded endometrioid glands of variable size and irregular contours in a vaguely lobular architecture surrounded by fibromatous stroma, and intracystic, with simple papillae protruding into an endometriotic cyst.2,6,7 Both patterns exhibited low to moderate nuclear atypia, nonconfluent architecture, and no evidence of destructive stromal invasion. Morules are defined as whorled nests of ovoid to spindle-shaped cells with ill-defined cell borders, a moderate amount of eosinophilic cytoplasm, and a centrally located round-to-oval nucleus lacking prominent nucleoli.8 SMBTs are characterized by papillae with hierarchical branching architecture and edematous or fibrous stromal cores often containing conspicuous neutrophils, lined by proliferative epithelium composed of a mixture of Müllerian cell types, including endometrioid cells (often with focal squamous or mucinous differentiation), endocervical-type mucinous cells, and ciliated cells.9,10 Microinvasion can be present, but <5 mm in the greatest linear extent. Strong, unequivocal nuclear β-catenin staining in lesional glands, even when focal, was interpreted as positive/abnormal.11–13 Immunohistochemistry (IHC) IHC for β-catenin (prediluted, clone β-catenin-1, #IR70261-2, Agilent, Santa Clara, CA) was performed in the clinical laboratory on a DAKO Autostainer Link 48 instrument, as previously described.11 High pH (50x Tris/EDTA buffer, pH 9) solution was used for antigen retrieval. The antigen retrieval step was performed at 97 °C for 20 min. Primary antibody incubation was 20 min, followed by secondary antibody (EnVision/HRP) incubation for an additional 20 min. DNA and RNA Preparation and Next-generation Sequencing (NGS) Testing was performed at the UT Southwestern Medical Center Clinical NGS Laboratory. For each sample, DNA and RNA were extracted from whole tissue sections of a single formalin-fixed paraffin-embedded (FFPE) block without macro-dissection or micro-dissection, using the AllPrep DNA/RNA FFPE Kit (Qiagen, Germantown, MD) according to the manufacturer’s instructions. Sequencing libraries were prepared using Kapa Biosystems and Illumina chemistry. A custom DNA and RNA probe panel enriched for all exons of over 1425 cancer-related genes (Supplemental Table 1, Supplemental Digital Content 1, https://links.lww.com/IJGP/A231) was sequenced on the Illumina NextSeq. 550 instruments. DNA and RNA sequence analyses were performed using custom germline, somatic, and mRNA bioinformatics pipelines on the UTSW Bio-High Performance Computing cluster, optimized for detecting single-nucleotide variants, indels, and known gene fusions.14,15 The assay achieves a median target exon coverage of 900×, with 94% of exons covered at >100×. The minor allele frequency limit of detection is 5% for single-nucleotide variants. In the 4 EBT cases coexisting with OEC, only tissue blocks containing exclusively the EBT component were selected for NGS to avoid contamination from the carcinoma. Statistical Analysis Two-tailed χ2 or Fisher exact test was used for the comparison of the β-catenin expression or mutation frequency between EBT and SMBT groups. A P<0.05 was considered statistically significant. Statistical analysis was performed using GraphPad Prism 10.0.0 software (San Diego, CA).

Patient Characteristics All patients were staged according to the International Federation of Gynecology and Obstetrics (FIGO) 2014 system. Patient ages at diagnosis ranged from 22 to 77 yr (median, 46 yr). Follow-up data were available for 19 patients, with a median follow-up duration of 37 mo (range, 8–116 mo). Background endometriosis or a history of pelvic endometriosis was identified in 73% (8/11) of patients with EBT and 60% (6/10) of those with SMBT. Table 1 summarizes patient demographics and clinicopathologic characteristics. | No. | Age | Coexisting OEC | Unilateral (U) vs. Bilateral (B) | FIGO Stage | F/U (mo) | Endometriosis* | Clinical Outcomes | | |---|---|---|---|---|---|---|---|---| | EBT | 1 | 64 | No | U | IA | 116 | Yes | NED | | 2 | 77 | Yes | U | IA | 8 | No | NED | | | 3 | 22 | No | B | IB | 92 | Yes | NED | | | 4 | 29 | No | U | IA | 15 | No | NED | | | 5 | 46 | No | U | IA | 14 | Yes | NED | | | 6 | 45 | No | U | IA | 66 | Yes | NED | | | 7 | 54 | No | U | IA | 102 | Yes | NED | | | 8 | 67 | Yes | U | IIIA (OEC) | 28 | No | Recurrent OEC | | | 9 | 76 | Yes | U | IIIC (OEC) | 42 | Yes | DOD (OEC) | | | 10 | 61 | Yes | U | IA | 37 | Yes | NED | | | 11 | 48 | No | U | IA | 96 | Yes | NED | | | SMBT | 1 | 29 | No | B | IB | 8 | Yes | Radiologic concern for recurrence/ loss of F/U | | 2 | 46 | No | U | IA | 68 | Yes | NED | | | 3 | 34 | No | U | IA | 18 | No | NED | | | 4 | 27 | No | U | IA | NA | Yes | NA | | | 5 | 35 | No | U | IA | 29 | No | NED | | | 6 | 45 | No | B | IB | 69 | No | NED | | | 7 | 27 | No | U | IA | 27 | Yes | NED | | | 8 | 41 | No | B | IB | 102 | No | NED | | | 9 | 61 | No | U | IA | NA | Yes | NA | | | 10 | 57 | No | U | IA | 13 | Yes | NED | Histologic and Immunohistochemical Evaluation of EBT and SMBT Among the 11 EBT cases, 7 (64%) exhibited an adenofibromatous growth pattern, whereas 4 (36%) demonstrated an intracystic growth pattern. Four of 11 (36%) of cases showed coexisting OEC. In all cases of EBT coexisting with OEC, the EBT component demonstrated an adenofibromatous growth pattern, while the OEC component was FIGO grade 1 endometrioid carcinoma. Aberrant β-catenin nuclear staining was observed in 8 of 11 EBT cases (73%), which was significantly higher than that in SMBT (0/10, 0%, P=0.001). Morules were identified in 4/11 (36%) EBTs, all of which also demonstrated aberrant β-catenin nuclear expression. In contrast, none of the SMBTs (0/10, 0%) showed aberrant β-catenin nuclear staining or morules. Representative images of EBTs and SMBTs, along with the distribution of cases according to the growth pattern, aberrant β-catenin expression, and presence of morules, are shown in Figures 1 and 2. All cases with aberrant nuclear β-catenin expression harbored CTNNB1 mutations on NGS analysis, and conversely, all cases with CTNNB1 mutations demonstrated aberrant nuclear β-catenin expression. Comparison of the Mutational Landscape Between EBT and SMBT In our cohort, CTNNB1 mutations were identified in 8 of 11 EBTs (73%), which is comparable to previously reported frequencies (87.5%, 7 of 8 cases).2 The most common alteration was p.D32Y (3/8, 37.5%), followed by other exon 3 hotspot mutations, including p.A43T, p.S37F, p.T41A, and p.S37Y. Other recurrently mutated genes included KRAS (36%), ARID1A (27%), ATR (27%), KMT2D (27%), PIK3CA (18%), PIK3R1 (18%), PTEN (18%), AKT1 (18%), TP53 (18%), KMT2C (18%), BRCA1 (18%), FBXW7 (18%), BCL11B (18%), CANT1 (18%), EP400 (18%), ERCC6 (18%), KIT (18%), MED12 (18%), RB1 (18%), and PDGFRA (18%). In contrast, no CTNNB1 mutations were identified in SMBTs (0%), a frequency significantly lower than that observed in EBTs (P<0.0001). In our cohort, KRAS or BRAF mutations were detected in 9 of 10 cases (90%). KRAS mutations were present in 6 of 9 cases (66.7%), most commonly p.G12D (4/6, 66.7%) and less frequently p.G12V (2/6, 33.3%). BRAF mutations were identified in 3 of 9 cases (33.3%), all of which were p.V600E. Overall, the majority of SMBTs (60%) harbored KRAS mutations, followed by BRAF (30%). KRAS and BRAF somatic variants were mutually exclusive, supporting activation of a common pathway for oncogenesis. Additional recurrently mutated genes included PIK3CA (20%), PIK3R1 (20%), PTEN (20%), ATM (20%), ZFHX3 (20%), AUTS2 (20%), CIC (20%), FAT1 (20%), and PLAT (20%). Notably, 2 cases (20%) harbored concurrent KRAS/PIK3CA mutations. The mutation profiles of all cases are summarized in Figure 3 [mutated genes in commonly affected pathways, class-defining mutated genes, and other frequently mutated genes in endometrial endometrioid carcinoma (EEC)], Figure 4 (additional shared mutated genes between EBTs and SMBTs), and Supplemental Table 2, Supplemental Digital Content 2, https://links.lww.com/IJGP/A232 (differentially mutated genes in EBTs and SMBTs). Detailed mutation profiles of EBTs and SMBTs were listed in Supplemental Table 3, Supplemental Digital Content 3, https://links.lww.com/IJGP/A233. No gene fusions were identified on RNA sequencing analysis. Comparison of Affected Key Signaling Pathways in EBT and SMBT The most frequently affected signaling pathway in EBTs was the WNT/β-catenin canonical pathway (82%). Other commonly altered pathways included the PI3K-PTEN-AKT-mTOR pathway (55%), RAS-MEK-ERK pathway (36%), and the SWI/SNF chromatin remodeling complex (36%). Additional pathways affected were the ATM/ATR DNA damage response pathway (27%) and the NRG/ERBB signaling pathway (18%). Several of the identified mutations involved genes associated with chromatin remodeling and DNA repair mechanisms (64% and 27%, respectively). In contrast, the most frequently altered pathway in SMBTs was the RAS-MEK-ERK, observed in 90% of cases, which was significantly higher than the frequency seen in EBTs (36%, P=0.024). Other commonly altered pathways in SMBTs included the WNT/β-catenin pathway (40%), PI3K-PTEN-AKT-mTOR pathway (40%), and the SWI/SNF chromatin remodeling complex (20%). Alterations in the ATM/ATR DNA damage response pathway were identified in 20% of SMBT cases. NRG/ERBB signaling pathway and chromatin remodeling (other than SWI/SNF complex) or DNA repair genes were not or minimally affected in SMBT. It is also worth noting that, although SMBTs lacked CTNNB1 mutations and aberrant β-catenin expression, a subset harbored mutations in other genes within the canonical WNT/β-catenin pathway, including APC and AXIN1 (components of the destruction complex, WNT “off” state), WNT10A (ligand), and LRP1B (co-receptor). The frequently affected pathways in EBTs and SMBTs are illustrated in Figure 5.

Endometrioid borderline tumors (EBTs) and seromucinous borderline tumors (SMBTs) are rare ovarian neoplasms regarded as potential precursor lesions of ovarian endometrioid carcinoma. The molecular alterations of endometrioid carcinomas are now well defined through studies of endometrial tumors.16–19 Based on The Cancer Genome Atlas (TCGA) analyses, all endometrial carcinomas are classified into 4 molecular groups: (1) POLEmut (ultramutated), (2) MSI-high (hypermutated), (3) copy-number low (no specific molecular profile), and (4) copy-number high/serous-like. Although not absolute, groups 1 to 3 largely correspond to endometrioid carcinoma, whereas group 4 corresponds primarily to serous carcinoma, with occasional inclusion of other high-grade carcinomas. The first three groups frequently harbor mutations in PTEN, CTNNB1, PIK3CA, ARID1A, and KRAS, affecting key signaling pathways such as PI3K-PTEN-AKT-mTOR, RAS-MEK-ERK, WNT-β-catenin, and the SWI/SNF chromatin remodeling complex.16 Other commonly altered pathways include the ATM/ATR DNA damage response pathway,16,20 along with recurrent mutations in FBXW7 and ZFHX3.16,21 In contrast, copy-number high/serous-like tumors are dominated by TP53 mutations, with relatively few alterations in these pathways. Furthermore, recent genome-wide studies of atypical hyperplasia/endometrioid intraepithelial neoplasia (AH/EIN), a well-established precursor of EEC, have shown highly similar mutational profiles, with frequent alterations in PTEN, CTNNB1, PIK3CA, ARID1A, and KRAS.22–26 Several studies investigating the molecular landscape of OEC have shown that the frequently mutated genes (PTEN, CTNNB1, PIK3CA, KMT2D, KMT2B, PIK3R1, ARID1A, TP53, and others) occur at similar frequencies in EEC,27–29 involving key signaling pathways such as PI3K-PTEN-AKT-mTOR, RAS-MEK-ERK, and WNT-β-catenin, similar to those in EEC. While the molecular alterations in endometrioid carcinomas and their endometrial precursors are well established from studies of endometrial lesions, the genetic landscape of EBTs and SMBTs, considered precursors of OEC, remains incompletely characterized due to their rarity. In this study, we investigated the genetic profiles of these lesions. Using a 1425-gene NGS panel, we demonstrated that EBTs are primarily defined by CTNNB1 mutations and aberrant β-catenin expression, whereas SMBTs exhibit a KRAS/BRAF-driven molecular profile with frequent co-mutations in PIK3CA and other signaling components. Several investigators have previously examined genomic alterations in EBTs and SMBTs. In the study by Oliva et al.2, genetic analysis of 8 EBTs revealed β-catenin mutations as a nearly constant alteration, present in 7 of 8 cases (87.5%). By contrast, PTEN and KRAS mutations, as well as microsatellite instability (MSI), all commonly observed in endometrioid carcinoma, were rare, with PTEN mutations detected in only 1 case (12.5%), and KRAS mutations and MSI were entirely absent. The findings of a high prevalence of β-catenin and a low frequency of PTEN mutations are consistent with our findings (87.5% and 18%, respectively). However, in contrast to our detection of KRAS mutations in 36% of EBT cases, none were identified in that study. The mutational profile of EBTs in the current study shows substantial overlap with that reported in OEC. Hollis et al.29 performed whole-exome sequencing on 112 OEC cases and identified frequent mutations in CTNNB1 (43%), PIK3CA (43%), ARID1A (36%), PTEN (29%), KRAS (26%), TP53 (26%), and SOX8 (19%). Similar mutational patterns in OEC have also been reported by other authors, including Cybulska et al.,30 Pierson et al.,27 and de Nonneville et al.28 Overall, the spectrum of altered genes in OEC closely parallels that observed in EBTs in this study, with the notable exception of CTNNB1, which was altered at a substantially higher frequency in EBTs. Interestingly, other investigators also demonstrated that CTNNB1 mutations are likely mutually exclusive with TP53 mutations,28,29 a finding that aligns with our observations in EBTs. In our study, 2 EBT cases harbored TP53 mutations (18%). Specifically, case #9 harbored mutations in KRAS, PTEN, and TP53, among others, while case #10 showed mutations in KRAS, ARID1A, and TP53, among others. Notably, neither case demonstrated a CTNNB1 mutation. In addition, Hollis and colleagues reported that tumors with TP53 mutations were more frequently diagnosed at advanced FIGO stage (III/IV; 48%) and were associated with worse survival. In our cohort, the presence of a TP53 mutation may have contributed, at least in part, to the aggressive clinical course observed in case #9, in which the patient ultimately died of the disease. Several groups have also investigated SMBTs, with variable results. Wu et al.3 analyzed 28 cases and identified somatic KRAS mutations in all tumors (100%), followed by PIK3CA (60.7%) and ARID1A (14.3%) mutations. Similarly, Kim et al.4 reported KRAS mutations in exon 1, codon 12 in 12 of 17 SMBTs (69%), with no PTEN mutations detected. In contrast, Sasamori et al.,5 studying 23 Japanese cases, found much lower frequencies of KRAS (4.3%), BRAF (8.6%), PIK3CA (8.6%), and ERBB2 (17.3%) mutations. In our series, KRAS mutations were detected in 60% of SMBTs, a finding similar to that of Kim and colleagues. It is important to note that all prior studies employed PCR-based Sanger sequencing, which assessed only a limited panel of genes. As a result, mutations in other oncogenic pathways may have been overlooked, leaving the broader mutational landscape of SMBTs incompletely defined. Clinically, all patients with EBT not coexisting with OEC (n=7) presented with low-stage disease (FIGO stage IA or IB) and remained free of disease during follow-up. Among the 4 patients with EBT coexisting with OEC, 2 had low-stage disease (FIGO stage IA) and were also free of disease at follow-up. The remaining 2 patients presented with advanced-stage disease (FIGO stages IIIA and IIIC); one experienced recurrent OEC, and the other died of OEC. All patients with seromucinous borderline tumor (SMBT) had low-stage disease (FIGO stages IA or IB). One patient demonstrated radiologic concern for recurrence, whereas the remaining patients were either free of disease or lost to follow-up. Overall, tumor prognosis appears to be closely associated with clinical stage and the presence or absence of coexisting carcinoma. Increased β-catenin Aberrations and Morule Formation in EBTs The frequency of CTNNB1 mutations and associated β-catenin aberrations in EBTs appears higher than that reported in low-grade EEC or non-polyp-associated atypical hyperplasia/endometrial intraepithelial neoplasia (AH/EIN), which is ∼50%.16,31 In contrast, this frequency is more comparable to that observed in polyp-associated AH/EIN (>60%).32. Likewise, morules were identified in 36% of EBT cases, a rate similar to that in polyp-associated AH/EIN (38%)32 and higher than in non–polyp-associated AH/EIN (24%).32 Given that most EBTs arose in an adenofibromatous background with fibrous stroma resembling that of endometrial polyps, it is intriguing to speculate that this microenvironment may be more permissive to the accumulation of CTNNB1 mutations and the formation of morules. An alternative explanation is that, as intermediate precursor lesions of OEC, the progression from CTNNB1-mutant EBT to carcinoma may occur at a slower rate than that of CTNNB1 wild-type counterparts, thereby providing a longer “window” during which these lesions can be detected. These proposed mechanisms remain speculative and warrant further investigation. Finally, this observation may reflect a small-sample-size bias, given the limited number of cases studied to date. Larger-scale studies will be required to further validate and clarify these findings. The Mutational Landscape of SMBT Closely Resembles That of Seromucinous Variant of OEC But Differs From Low-grade Serous Carcinoma (LGSC) SMBTs exhibit morphologic features distinct from EBTs but share some architectural and cytomorphologic features with serous borderline tumors and low-grade serous carcinomas (LGSC). Among SMBT cases, KRAS mutations were most frequent (60%), followed by BRAF (30%), PIK3CA (20%), PIK3R1 (20%), PTEN (20%), ATM (20%), and ZFHX3 (20%). Concurrent KRAS/PIK3CA mutations were detected in 20% of cases. These findings parallel those reported by Rambau et al.,1 who performed targeted NGS on 32 seromucinous variants of OEC cases and identified frequent mutations in KRAS (70%), PIK3CA (37%), PTEN (19%), and ARID1A (16%), with no CTNNB1 mutations detected. Moreover, 30% of these cases harbored KRAS/PIK3CA co-mutations, further supporting the similarity between SMBT and the seromucinous variant of OEC. By contrast, the comprehensive NGS study of LGSC by Cheasley et al.33 (71 cases, 127-gene panel) revealed considerably lower rates of KRAS (22%) and BRAF (16%) mutations. Instead, LGSC frequently exhibited alterations in USP9X (26.7%), MACF1 (11.2%), and CDKN2A (15.5%), none of which were detected in SMBT. Conversely, several genes commonly mutated in SMBT, including PIK3R1, PTEN, ATM, and ZFHX3 (each 20%), were not altered in LGSC, highlighting distinct molecular profiles. This difference is also observed when our results are compared with findings from several other NGS studies of LGSC.34,35 Notably, unlike LGSC, SMBTs showed frequent alterations in the same signaling pathways commonly affected in EEC, including PI3K-PTEN-AKT-mTOR, RAS-MEK-ERK, WNT-β-catenin, SWI/SNF chromatin remodeling, and ATM/ATR DNA damage response. Moreover, mutations across all 4 molecular classes of EEC were also identified in SMBT. Taken together, these observations support the notion that SMBT is a precursor of the seromucinous variant of OEC. In conclusion, our observations expand the characterization of ovarian borderline endometrioid-type neoplasms and demonstrate that EBTs and SMBTs possess distinct genetic features. Despite their shared strong association with endometriosis, our findings indicate that these neoplasms harbor divergent driver events and molecular profiles that likely contribute to their distinct histopathologic features.

Endometrioid borderline tumor; Seromucinous borderline tumor; Molecular landscape; Pathways