Comprehensive Immunohistochemical Analysis of Mesonephric Marker Expression in Low-grade Endometrial Endometrioid Carcinoma

Abstract

Immunohistochemical markers shown to be useful in identifying/confirming mesonephric/mesonephric-like differentiation (MLD markers) include thyroid transcription factor (TTF1), GATA-binding protein 3 (GATA3), and cluster of differentiation 10 (CD10). Only a few studies have examined the expression levels of MLD markers in endometrial endometrioid carcinomas (EECs). This study aimed to analyze the frequency and pattern of MLD marker expression in low-grade EECs. We performed immunostaining for the detection of TTF1, GATA3, and CD10 expression in 50 low-grade EEC tissue samples and evaluated their staining proportion and intensity. Nine tumors (18.0%) expressed at least one MLD marker in varying proportions and intensities, and 2 of these tumors were positive for 2 MLD markers (TTF1/GATA3 and GATA3/CD10, respectively). Three (6.0%) tumors showed moderate-to-strong nuclear TTF1 immunoreactivity in ≤5% of the tumor cells. Five tumors (10.0%) had at least moderate nuclear GATA3 staining, and three of them displayed a staining proportion of ≥15%. Three tumors (6.0%) were focal (mean proportion, 15%) but strongly positive for CD10. Our findings indicate that a subset of EEC can express one or more MLD markers with varying staining proportions and intensities. Given that a diagnosis of uterine mesonephric-like adenocarcinoma should be established based on a combination of characteristic histologic features, unique immunophenotypes, and confirmed molecular findings, pathologists should not exclude EEC based only on the presence of focal immunoreactivity for MLD markers. Awareness of the atypical expression patterns of MLD markers in EEC helps pathologists avoid misdiagnosing EEC as a uterine mesonephric-like adenocarcinoma.

Endometrial carcinoma (EC) is one of the most common malignant tumors of the female genital tract. The incidence and mortality rates of EC are steadily increasing in developed countries ^1,2. The clinical behavior and outcomes of EC vary depending on the histologic type and pathologic stage. Endometrial endometrioid carcinoma (EEC) is the most common histologic type of EC, which constitutes ~80% of all EC cases. The histologic grade of EEC was determined according to the International Federation of Gynecology and Obstetrics (FIGO) grading system, which adopts a binary scheme that combines FIGO grades 1 and 2 as low grade and FIGO grade 3 as high grade ^3,4. The recognition of typical endometrioid morphology and clinical significance is usually straightforward. Low-grade EEC tends to be present at an early stage and have a favorable prognosis. However, the clinicopathologic diversity of high-grade EEC often poses a diagnostic challenge, and some of their histologic and immunohistochemical features are associated with aggressive clinical behavior ².

Mesonephric-like adenocarcinoma (MLA) arising in the upper female genital tract is a rare but distinct gynecologic malignancy with a characteristic morphology and unique immunophenotypes ⁵. MLA of the uterine corpus was just recently introduced in the 2020 World Health Organization (WHO) Classification of Female Genital Tumors ⁴. Uterine MLA exhibits a variety of architectural patterns within the same tumor; as a result, it can be misdiagnosed as low-grade EEC, serous carcinoma, or carcinosarcoma ⁶. The clinical manifestations, histologic features, immunostaining results, and molecular characteristics of uterine MLA, as well as the difficulty in distinguishing them from other histologic types of EC, have been consistently reported in the literature ^2,7–18. Uterine MLA exhibits more aggressive clinical behavior and worse prognoses than EEC ^5,9,13. Within a large retrospective cohort study of EC, we demonstrated in a large that compared with EEC, uterine MLA showed larger tumor size, deeper myometrial invasion, more advanced stage, more frequent lymphovascular space invasion, post-treatment recurrence, distant metastasis, and lower patient survival ².

Only a few studies have examined the expression levels of immunohistochemical markers shown to be useful in identifying/confirming mesonephric/mesonephric-like differentiation (MLD markers), including thyroid transcription factor 1 (TTF1), GATA-binding protein 3 (GATA3), and cluster of differentiation 10 (CD10), in EECs ^7,10,19. The clinicopathologic characteristics of MLD marker-expressing EECs have not yet been fully understood, and the frequency and pattern of MLD marker expression in EEC remain unclear. We recently performed a comprehensive immunostaining for the detection of endometrioid, serous, and MLD markers in uterine MLA tissue samples and reported their expression profiles and clinicopathologic significance ¹⁶. We noted that a small subset of uterine MLA displayed atypical intense immunoreactivity for estrogen receptor and progesterone receptor, both of which are known to be uniformly expressed in low-grade EEC, but are negative or focally positive with weak staining intensity in uterine MLA. In this study, conversely, we aimed to investigate the expression status of MLD markers in EEC and describe their clinicopathologic significance and possible diagnostic pitfalls. Understanding the atypical expression patterns of MLD markers in EEC will help pathologists distinguish uterine MLA from EEC and establish an accurate diagnosis.

MATERIALS AND METHODS

This study was approved by the Institutional Review Board of the Samsung Medical Center (2021-06-190). Between February 2021 and January 2022, we found 71 consecutive cases of primary EEC from the institutional databases. Two board-certified gynecologic pathologists reviewed all available hematoxylin and eosin-stained slides to confirm the pathologic diagnosis of EEC and the presence of sufficient tumor tissue for immunostaining. The clinicopathologic information obtained from the electronic medical records and pathology reports were also reviewed. Four patients whose tumor tissues were insufficient to perform immunostaining, 1 patient whose serous carcinoma was initially misdiagnosed as EEC as confirmed during the slide review, 4 patients who were diagnosed with grade 3 EEC, 5 patients who were preoperatively treated with high-dose progestin for grade 1 EEC, and 1 patient who was postoperatively treated with tamoxifen for invasive breast carcinoma, were excluded. We assessed the results of phosphatase and tensin homolog deleted on chromosome 10 (PTEN; prediluted, clone SP218, Ventana Medical Systems) immunostaining, which had been performed during the initial diagnosis before conducting this study. Six patients who had unavailable data on the PTEN expression status were excluded. Finally, 50 cases of grade 1 to 2 EEC showing typical endometrioid morphology were included in the study. Figure 1 shows the flowchart of the participant selection process.

FIG 1.

Flowchart of the patient selection process. CD10 indicates cluster of differentiation 10; EEC, endometrial endometrioid carcinoma; IHC, immunohistochemistry; PTEN, phosphatase and tensin homolog deleted on chromosome 10; TTF1, thyroid transcription factor 1.

Immunostaining was performed using whole tissue sections obtained from either total hysterectomy (27 cases) or endometrial curettage (23 cases) specimens, as described previously ^5,16,20–24. Seventeen of the 23 patients had been referred from outside hospitals and submitted unstained slides obtained from the curettage specimens only. The remaining 6 patients did not undergo surgical treatment. Briefly, 4-µm-thick, formalin-fixed, paraffin-embedded tissue slices were deparaffinized and rehydrated using a xylene and alcohol solution. Immunostaining was performed using an automated immunostainer (BOND-MAX, Leica Biosystems). After antigen retrieval, the slices were incubated with primary antibodies against TTF1 (dilution 1:100, clone 8G7G3/1, Dako, Agilent Technologies), GATA3 (prediluted, clone L50-823, Ventana Medical Systems), and CD10 (dilution 1:100, clone 56C6, Novocastra, Leica Biosystems). After chromogenic visualization, the slices were counterstained with hematoxylin. The appropriate controls were stained concurrently. Positive controls included papillary thyroid carcinoma and invasive lung adenocarcinoma for TTF1, luminal A-type invasive breast carcinoma and papillary urothelial carcinoma for GATA3, normal proliferative-phase endometrial stroma and low-grade endometrial stromal sarcoma for CD10. The negative controls were prepared by substituting the nonimmune serum with primary antibodies, which resulted in undetectable staining. The staining intensities of TTF1, GATA3, and CD10 were designated as negative (score 0), weak (score 1+), moderate (score 2+), or strong (score 3+). The staining proportions were determined in increments of 5% across a 0 to 100% range. CD10 expression was specifically assessed only in nonsquamous areas.

Pearson χ² test and Fisher exact test were performed to analyze the relationship between the expression status of MLD markers and the clinicopathologic characteristics of EEC patients. All statistical analyses were performed using IBM SPSS Statistics for Windows (version 23.0; IBM Corp.). A P-value of <0.05 was considered significant.

RESULTS

Table 1 summarizes the baseline clinicopathologic characteristics of 50 patients with EEC. Their median age was 53 yrs (range: 34–86 yrs). According to the FIGO grading system ³, 28 (56.0%) and 22 (44.0%) patients were diagnosed with FIGO grade 1 and 2 EECs, respectively. Seventy percent of the patients (35/50) had FIGO stage IA tumors. The remaining patients had stage IB (6/50; 12.0%), II (2/50; 4.0%), III (6/50; 12%), and IV tumors (1/50; 2.0%). Areas of mucinous and squamous differentiation were observed in 12 (24.0%) and 11 tumors (22.0%), respectively. The latter manifests as variable-sized morules or solid sheets of eosinophilic cells with occasional keratinization, whereas the former is characterized by the presence of columnar or pseudostratified epithelial cells containing intracytoplasmic mucin. Three (6.0%) patients developed tumors that showed a microcystic, elongated, and fragmented pattern of myometrial invasion, which is characterized by loss of conventional glandular architecture, presence of squamoid or histiocyte-like cells with eosinophilic cytoplasm, formation of microcysts, compressed elongated structures, flattened and fragmented glandular epithelium, and an edematous or fibromyxoid stroma associated with an acute inflammatory response ^25–29.

TABLE 1.

Baseline Clinicopathologic Characteristics and Immunostaining Results

Characteristic	No. cases (%)
Age (yr)
≥53	25 (50.0)
<53	25 (50.0)
Initial FIGO stage
IA	35 (70.0)
IB	6 (12.0)
II	2 (4.0)
IIIA	2 (4.0)
IIIC1	3 (6.0)
IIIC2	1 (2.0)
IV	1 (2.0)
FIGO grade
1	28 (56.0)
2	22 (44.0)
Distinctive histologic feature
Mucinous differentiation	12 (24.0)
Squamous differentiation	11 (22.0)
MELF pattern of stromal invasion	3 (6.0)
TTF1 positivity
Nuclear	3 (6.0)
Cytoplasmic	2 (4.0)
GATA3 positivity
Nuclear	5 (10.0)
Cytoplasmic	7 (14.0)
CD10 positivity
Luminal	2 (4.0)
Luminal, membranous and cytoplasmic	1 (2.0)
Membranous and cytoplasmic	1 (2.0)
MLD marker expression profile
TTF1+/GATA3+/CD10+	13 (26.0)
Nuclear TTF1+/nuclear GATA3+/luminal CD10+	9 (18.0)
PTEN expression
Intact	16 (32.0)
Loss	34 (68.0)

CD10 indicates cluster of differentiation 10; FIGO, International Federation of Gynecology and Obstetrics; GATA3, GATA-binding protein 3; MLD marker, Immunohistochemical markers shown to be useful in identifying/confirming mesonephric/mesonephric-like differentiation; PTEN, phosphatase and tensin homolog deleted on chromosome 10; TTF1, transcription termination factor 1.

Table 2 shows immunophenotypical characteristics of 9 patients with EEC showing MLD marker positivity. Three patients (6.0%) had TTF1-positive tumors, with moderate-to-strong staining intensity within the tumor cell nuclei. TTF1-positive tumor cells were observed in only a few microscopic foci, with staining proportions ranging from 2% to 5% (Fig. 2A–F). Unexpectedly, 2 tumors (4.0%) showed strong cytoplasmic TTF1 expression (Fig. 2G–I). Nuclear and cytoplasmic TTF1 immunoreactivity was occasionally observed in tumor cells possessing intracytoplasmic mucin, whereas the areas showing solid growth or squamous differentiation did not express TTF1.

TABLE 2.

Clinicopathologic and Immunophenotypical Characteristics of 9 Patients With Endometrial Endometrioid Carcinoma Showing MLD Marker Positivity

						MLD marker expression
						Staining intensity (proportion)
Case no.	Age (yr)	Initial FIGO stage	FIGO grade	Distinctive histologic feature	Expression profile	Nuclear TTF1	Nuclear GATA3	Luminal CD10	PTEN expression
1	86	IA	2	Mucinous	TTF1+/GATA3+/CD10−	3+ (1%); 2+ (1%)	3+ (50%); 2+ (20%)	0	Intact
2	41	IA	2	None	TTF1+/GATA3−/CD10−	3+ (1%); 2+ (1%)	0	0	Loss
3	44	IA	1	None	TTF1+/GATA3−/CD10−	3+ (2%); 2+ (2%); 1+ (1%)	0	0	Intact
4	73	IB	1	Mucinous	TTF1−/GATA3+/CD10+	0	2+ (1%); 1+ (1%)	3+ (30%)	Loss
5	46	IA	2	None	TTF1−/GATA3+/CD10−	0	3+ (1%)	0	Loss
6	62	IA	1	Mucinous	TTF1−/GATA3+/CD10−	0	3+ (15%)	0	Intact
7	68	II	2	Squamous	TTF1−/GATA3+/CD10−	0	3+ (10%); 2+ (20%)	0	Intact
8	39	II	2	Squamous	TTF1−/GATA3−/CD10+	0	0	3+ (10%)	Intact
9	51	IA	1	Squamous	TTF1−/GATA3−/CD10+	0	0	3+ (5%)	Intact

CD10 indicates cluster of differentiation 10; FIGO, International Federation of Gynecology and Obstetrics; GATA3, GATA-binding protein 3; MLD marker, Immunohistochemical markers shown to be useful in identifying/confirming mesonephric/mesonephric-like differentiation; PTEN, phosphatase and tensin homolog deleted on chromosome 10; TTF1, transcription termination factor 1.

FIG 2.

Thyroid transcription factor 1 (TTF1) expression in endometrial endometrioid carcinoma. Focal (5%) nuclear TTF1 immunoreactivity with moderate-to-strong staining intensity (white arrows) (A–C). Focal (2%) nuclear TTF1 immunoreactivity with weak-to-moderate-to-strong intensity (black arrows) (D–F). Cytoplasmic granular staining for TTF1 (G–I). Staining method: A, D, and G, hematoxylin and eosin staining; B, C, E, F, H, and I, immunostaining using polymer method. Original magnification: A, 40×; B, 10×; C, 200×; D and E, 40×; F, 200×; G, 100×; H, 40×; I, 200×.

Most tumors (45/50; 90.0%) showed a complete absence of nuclear GATA3 immunoreactivity. The remaining 5 tumors (10.0%) displayed nuclear GATA3 immunoreactivity with variable tumor staining intensities (Figs. 3A, B). One of the 5 patients had nuclear GATA3-positive tumors exhibiting diffuse, moderate, or strong expression (staining proportion: 70%; Figs. 3C, D). In 7 tumors (14.0%), scattered tumor cells exhibited aberrant cytoplasmic GATA3 expression (Fig. 3E). Nuclear and cytoplasmic GATA3 immunoreactivity was not limited to any specific architecture or growth pattern but was dispersed randomly in areas showing conventional glandular (endometrioid) architecture, solid growth, and mucinous or squamous differentiation. In 2 tumors (8.0%), scattered lymphocytes located within and around the tumor cell nests and glands occasionally displayed nuclear GATA3 immunoreactivity with moderate-to-strong staining intensity (Fig. 3F).

FIG 3.

GATA-binding protein 3 (GATA3) expression in an endometrial endometrioid carcinoma. Variable GATA3 staining intensity in the tumor cell nuclei (A, B). Areas of mucinous differentiation showing diffuse and strong GATA3 immunoreactivity (C, D). Aberrant cytoplasmic GATA3 expression (E). Nuclear GATA3 immunoreactivity in the stromal and intraepithelial lymphocytes (F). Staining method: A, C, and E, hematoxylin and eosin staining; B, D, and F, immunostaining using polymer method. Original magnification: A–D, 40×; E, 100×; F, 200×.

Positive CD10 immunoreactivity was observed in four tumors (8.0%). Of these, 2 tumors exhibited luminal staining (Figs. 4A, B), 1 displayed CD10 expression in the membrane and cytoplasm of the tumor cells, and the remaining showed both luminal and membranous/cytoplasmic staining. Although the staining intensity was strong in all CD10-positive tumors, the extent of CD10 staining varied from a few scattered microscopic foci (<5%) to approximately one-third of the entire tumor area (30%). The intervening endometrial stroma also exhibited diffuse membrane/cytoplasmic CD10 immunoreactivity with at least moderate staining intensity (Figs. 4C, D).

FIG 4.

Cluster of differentiation 10 (CD10) expression in an endometrial endometrioid carcinoma. Luminal CD10 immunoreactivity with strong staining intensity (A, B). Moderate-to-strong CD10 immunoreactivity in the intervening endometrial stroma (C, D). Staining method: A and C, hematoxylin and eosin staining; B and D, immunostaining using polymer method. Original magnification: A–B, 100×; C–D, 40×.

Eleven cases (22.0%) had atypical hyperplasia/endometrioid intraepithelial neoplasia adjacent to the EEC. Of these, 8 tumors were completely negative for all three proteins, the remaining three tumors exhibited patchy and weak cytoplasmic GATA3 expression. No significant difference was observed in the MLD marker expression between the microcystic, elongated, and fragmented and conventional invasion patterns.

No significant relationship was observed between TTF1, GATA3, and CD10 expression. As shown in Table 3, the differences in age, initial FIGO stage, FIGO grade, and PTEN expression status according to the MLD marker expression status were not significant. All EECs expressing at least 1 of the 3 MLD markers were early-stage disease, and all advanced-stage tumors were negative for MLD markers; however, the difference was not significant. Similar to the reported frequency of PTEN loss in EEC ^30,31, the tumor samples of 34 patients (68.0%) showed a loss of PTEN expression. The frequency of PTEN loss in the MLD marker-positive EECs (4/9; 44.4%) was lower than that in the MLD marker-negative EECs (30/41; 73.1%), but the difference was not significant.

TABLE 3.

Association Between MLD Marker Expression and Clinicopathologic Characteristics

	No. cases (%)
Characteristic	Mesonephric marker-negative (n=41)	Mesonephric marker-positive (n=9)	P
Age (yr)
≥53	21 (51.2)	4 (44.4)	1.000
<53	20 (48.8)	5 (55.6)
Initial FIGO stage
I–II	34 (82.9)	9 (100.0)	0.324
III–IV	7 (17.1)	0 (0.0)
FIGO grade
1	24 (58.5)	4 (44.4)	0.482
2	17 (41.5)	5 (55.6)
PTEN expression
Intact	11 (26.8)	5 (55.6)	0.124
Loss	30 (73.1)	4 (44.4)

FIGO indicates International Federation of Gynecology and Obstetrics; MLD marker, immunohistochemical markers shown to be useful in identifying/confirming mesonephric/mesonephric-like differentiation; PTEN, phosphatase and tensin homolog deleted on chromosome 10.

DISCUSSION

EEC can histologically manifest a mesonephric-like morphology such as small tubules possessing eosinophilic intraluminal hyaline-like secretions, and diverse architectural patterns ⁵. Therefore, it is important to distinguish EEC from uterine MLA, as the latter exhibit a more aggressive clinical behavior ², including advanced stage, early recurrence, and distant metastasis ¹³. Considering that a small subset of EECs with histological features similar to those of uterine MLA exists, but are not sufficient for a definite diagnosis, it is necessary to verify the usefulness of MLD markers for differentiating between EEC and MLA. However, the expression profiles of multiple MLD markers have not yet been examined in EEC. In this study, we performed a comprehensive immunohistochemical analysis of 3 well-known MLD markers in 50 primary EEC tissue samples.

TTF1 is considered a relatively sensitive and specific marker of uterine MLA ^10,32. The incidence of TTF1-positive EEC ranged from 0.9% to 18.8% ^10,32–34. Siami et al. ³³ reported the highest positive rate of nuclear TTF1 expression in EEC (6/32; 18.8%), with staining proportions ranging from <5% to >75%. However, they did not provide any information about the staining intensity. Moritz et al. ³⁴ investigated the expression of TTF1 and neuroendocrine markers in non-neuroendocrine ECs, including 26 EECs. They found that TTF1 expression was observed in 11.5% (3/26) of EECs, even though the frequency was lower than that of nonendometrioid carcinoma (31%). The TTF1 expression pattern in non-neuroendocrine ECs was predominantly focal (13/17; 76.5%) with moderate-to-strong staining intensity (16/17; 94.1%), in consistent with our findings. Mills et al. ³² conducted a TTF1 immunostaining using tissue microarrays (TMAs) and found that 3 of 240 EECs (1.3%) exhibited nuclear TTF1 expression with strong (1/3) or weak (2/3) staining intensity. Pors et al. ¹⁰ who also used TMAs, reported nuclear TTF1 positivity in three of 323 EECs (0.9%). They did not describe the staining intensity or the proportion of TTF1 expression. In this study, we evaluated the intensity and proportion of nuclear TTF1 staining. We observed that TTF1 was focally expressed (staining proportion, 2%–5%), with moderate-to-strong staining intensity. TTF1 expression was observed in FIGO grade 1 areas but not in those showing a solid growth pattern or squamous differentiation, and was also occasionally noted in areas of mucinous differentiation. The expression rates of TTF1 in previous TMA-based studies conducted by Pors et al. (0.9%) ¹⁰ and Mills et al. (1.3%) ³² were lower than that observed in this study using whole tissue sections (3/50; 6.0%). This finding may be attributed to the fact that the tissue core samples in TMA blocks did not represent the entire tumor tissue in each case; therefore, the TTF1-positive areas were not properly collected using the TMA technique. Meanwhile, since the number of cases in our study was smaller than that in previous studies, we cannot exclude the possibility that the TTF1-positive rate may decrease if the number of cases increases.

Mills et al. ³² reported that Kirsten rat sarcoma viral oncogene homolog (KRAS) mutations were identified in 3 cases with TTF1-positive EECs, including 1 with grade 3 EEC harboring a tumor protein 53 mutation, 1 with mismatch repair-deficient EEC, and 1 with EEC showing mucinous differentiation. In the latter case, mucinous differentiation combined with KRAS mutations might drive TTF1 expression. Consistent with this finding, we observed moderate nuclear TTF1 immunoreactivity in 1 patient showing mucinous differentiation. However, we did not classify our TTF1-positive EEC cases molecularly or perform ancillary tests to detect KRAS mutations. Hence, further investigations in larger-scale cohorts with molecular analysis are necessary to clarify the relationship between nuclear TTF1 expression and mucinous differentiation, and the mechanism of TTF1 expression according to the KRAS mutational status.

GATA3 is the best overall marker for diagnosing mesonephric lesions in the female genital tract, given its high sensitivity and specificity (91% and 94%, respectively) ^10,35. Similar to TTF1, the reported frequencies and patterns of GATA3 expression in EECs vary among studies. Pors et al. ¹⁰ reported that 1.9% (6/323) of EECs expressed GATA3 without any information about the staining intensity or proportion. Terzic et al. ¹⁹ and Howitt et al. ⁷ found nuclear GATA3 expression with varying staining intensities and proportions in 3.4% (1/29) and 1.3% (2/160) of EECs, respectively. In our study, moderate-to-strong nuclear GATA3 positivity was identified in ≤50% of the tumor cells in 5 EECs (10.0%). Based on our observation that the patchy and heterogeneous expression pattern of GATA3 was similar to that of TTF1, we assumed that the higher GATA3-positive rate in this study may be attributed to the differences in sampling methodology (i.e. whole tissue section vs. TMA) and the number of examined cases. Nuclear GATA3 immunoreactivity was observed regardless of the histologic growth patterns, including well-formed glands, solid sheets, and squamous morules. To avoid misdiagnosing EEC as MLA, pathologists should be aware that a small subset of EEC can show nuclear GATA3 staining with focal and patchy patterns. In addition, we found aberrant cytoplasmic expression of GATA3 in 7 EECs (14.0%), and many stromal and intraepithelial lymphocytes expressed GATA3 in their nuclei. Special attention is required for the assessment of GATA3 immunoreactivity in the cytoplasm of tumor and immune cells, as it can cause potential diagnostic pitfalls.

Luminal CD10 immunoreactivity was found in 3 EECs (6.0%). The incidence of CD10 positivity was lower than those of TTF1 (8.0%) and GATA3 (10.0%). Although McCluggage et al. ³⁶ documented that 11.1% (1/9) of EECs showed CD10 expression in the glandular lumina, they did not provide detailed information on the staining intensity and proportion. In this study, luminal CD10 staining was observed in 30% of tumor cells. Staining intensity was strong in all three CD10-positive EECs. Other patterns of CD10 staining observed in the 2 tumors included cytoplasmic or cytoplasmic and membranous immunoreactivity in tumor cells. The nonluminal CD10 expression is more indicative of EEC than of MLA.

We found that all patients (9/9; 100.0%) whose tumors expressed at least one of the 3 MLD markers were diagnosed with early-stage (i.e. uterine-confined) tumors, whereas 17.1% (7/41) of patients with MLD marker-negative EEC had stage III–IV tumors. The tendency of MLD marker expression with the lower stage raises the concern that the lack of awareness of MLD marker positivity in EEC may lead to the misdiagnosis of early-stage EEC as MLA, an aggressive type that requires more intensive treatment. The PTEN expression status of PTEN was not significantly associated with MLD marker expression However, PTEN expression was lost in half of the patients with MLD marker-expressing EECs. Despite the limited diagnostic implications, the loss of PTEN immunoreactivity still helps distinguish EEC from MLA as PTEN expression is preserved in most MLA cases ^6,9,13.

None of the 9 EECs concurrently expressed any of the 3 MLD markers. Seven tumors expressed 1 marker, and the remaining 2 expressed 2 markers. In our recent study, all uterine MLAs expressed 2 or more MLD markers, most of which expressed all 3 markers and diffusely and strongly expressed at least 1 marker ¹⁶. Since a few EEC cases also exhibited strong immunoreactivity for GATA3, CD10 (diffuse), and TTF1 (focal), the use of multiple MLD markers and hormone receptors can aid pathologists in preventing misdiagnosis.

In this study, 2 (4.0%) and 7 (14.0%) cases displayed cytoplasmic TTF1 and GATA3 staining in the tumor cells, respectively. We thought these findings not to have significant diagnostic values because cytoplasmic immunoreactivities for TTF1 and GATA3 were not associated with any specific architecture or growth pattern of MLA but were dispersed randomly in areas showing either glandular or solid architecture or mucinous differentiation. TTF1 is a DNA-binding protein that is mainly expressed in the thyroid and lung tissues ³⁷. In a previous study examining the staining pattern, incidence, and significance of TTF1 expression in various tumor types ³⁷, 6.4% (23/361) of the cases showed cytoplasmic TTF1 staining. The authors mentioned that this cytoplasmic TTF1 immunoreactivity is a nonspecific finding caused by cross-reacting antigens, which are not alternative splicing products of TTF1 ³⁸. Another study found that a considerable variation in cytoplasmic TTF1 staining seems to exist between commercial antibodies ³⁹. In particular, 2 of the 6 ovarian mucinous tumors exhibited cytoplasmic TTF1 positivity when using the antibody same as that we used (clone 8G7G3/1, Dako, Agilent Technologies). This antibody has been shown to have the highest rate of cytoplasmic TTF1 expression ³⁹. GATA3 is a transcription factor that is critical for the embryonic development of various tissues, including the parathyroid gland and the kidney ⁴⁰. Its application has been recognized in identifying carcinomas of the mammary or urothelial origin ⁴¹. GATA3 typically shows nuclear staining as it is transported from the cytoplasm across the nuclear membrane and then regulates gene expression ^40,42. In a previous study assessing the expression status of GATA3 in ovarian high-grade serous carcinomas, 11 of the 20 (55.0%) cases showed nonspecific cytoplasmic staining in the tumor cells ⁴². Another recent study documented that a faint cytoplasmic GATA3 staining was observed in non-neoplastic gastric glands and goblet cells of respiratory epithelium, most likely representing nonspecific background staining ⁴⁰.

This study has some limitations. First, this study was conducted at a single institution with a relatively small number of sample, which may limit the generalizability of our findings. Further validation in larger independent cohorts is required. Second, patients with EECs showing mesonephric-like differentiation were not included; therefore, we were unable to compare the MLD marker expression status between the areas of mesonephric-like differentiation and those of conventional EEC. Third, we did not perform any molecular analysis or immunostaining for hormone receptors; therefore, we were unable to explore the association between MLD marker expression and the molecular classification of EEC as well as the expression status of estrogen receptor and progesterone receptor. Considering that the molecular subtypes of EEC are associated with distinct clinical behaviors and prognoses, the expression profiles of MLD markers in each subtype should be scrutinized. Lastly, because survival data were unavailable, the prognostic significance of MLD marker expression in EEC could not be determined. Because MLA shows more aggressive clinical behavior and worse prognoses than EEC, the prognostic value of MLD marker immunoreactivity should be investigated in patients with EECs. Despite these limitations, this study provides valuable information regarding the MLD marker expression status in EEC, along with their clinicopathologic characteristics and the potential pitfalls of immunohistochemical interpretation.

In summary, we investigated the immunohistochemical expression of MLD markers, including TTF1, GATA3, and CD10 in 50 low-grade EECs. Most tumors tested negative for these 3 markers. However, nuclear immunoreactivity for TTF1 and GATA3, as well as the luminal staining pattern of CD10, were observed in some EECs. Hence, pathologists should distinguish EEC from uterine MLA because the latter exhibits more aggressive clinical behavior and worse prognosis than the former. Our findings indicate that a subset of EEC can express one or more MLD markers with varying staining intensities and proportions. However, it is not necessary to perform immunostaining on these markers during the daily routine diagnosis. A routine performance of these markers in otherwise typical endometrioid carcinoma can be misleading and lead to a misdiagnosis. A diagnosis of uterine MLA should be established based on a combination of characteristic histological features, unique immunophenotypes, and confirmed molecular findings. For accurate diagnosis and proper management of patients with EC, pathologists should not exclude EEC based only on the presence of focal immunoreactivity for MLD markers.