Mesonephric-like Adenocarcinoma (MLA) Diagnostic Criteria and Controversies: Perspectives and Guidance From Pathologists in the MLA Consortium

Abstract

Mesonephric-like adenocarcinoma (MLA) is a rare and aggressive gynecologic malignancy that has only been recognized in the last decade. It arises in the endometrium, ovaries, and other extrauterine sites (often in association with endometriosis) and closely mimics a variety of other tumor types that occur in these locations. While it shows significant morphologic, immunohistochemical, and molecular homology with cervical mesonephric adenocarcinoma, there are many clinicopathologic features that suggest müllerian derivation, and this is now well established. As research on MLA has accumulated, questions have emerged about optimal practices for the diagnosis of these challenging tumors. In 2022, faculty at M.D. Anderson Cancer Center convened the Mesonephric-like Adenocarcinoma (MLA) Consortium, comprised of international pathologists, gynecologic oncologists, medical oncologists, radiation oncologists, and basic science investigators with expertise in MLA, with the goals to enhance understanding of these tumors, refine diagnostic criteria, improve treatment options, and facilitate research collaborations. An initial review from the consortium was published in 2025, and included diagnostic recommendations from the group’s pathologists. Controversies remain, however, about the morphologic, immunohistochemical, and molecular criteria that should be used to establish a diagnosis of MLA. Herein, the pathologists from the MLA Consortium provide a comprehensive evaluation of the literature on MLA diagnostic criteria, address ongoing controversies in this area, and provide practical guidance for pathologists considering this entity.

THE DISCOVERY OF MESONEPHRIC-LIKE ADENOCARCINOMA

MLA was first described in 2016 in a study coauthored by one of us (W.G.M.); in this study, 12 cases were reported (7 uterine corpus, 5 ovary).¹ The term MLA was recommended since, although the morphologic features and immunophenotype strongly suggested a mesonephric neoplasm, there were other features which suggested a Müllerian derivation. These included the absence of mesonephric remnants associated with the neoplasm and the observation that in all of the uterine cases, the tumor extensively involved the endometrium and appeared to arise there with subsequent invasion into the myometrium; this would not be expected with a true mesonephric adenocarcinoma (MA) arising from mesonephric remnants in the outer aspects of the uterine corpus. As such, the term MLA was recommended until the histogenesis was firmly established. The cases in this series had originally mostly been reported as endometrioid carcinomas, but several of the ovarian neoplasms were diagnosed as malignant struma ovarii (papillary-type thyroid carcinoma), given the nuclear features and the presence of diffuse TTF1 immunoreactivity (discussed below). The discovery of MLA highlights the important point that “new” pathologic entities can still be identified through careful observational studies.

Although this represented the first publication of MLA, in hindsight, the original observations of this neoplasm were several years earlier, when one of us (W.G.M.) reported several cases where the patients had a history of endometrioid carcinoma of the uterine corpus and were subsequently discovered to have a pulmonary tumor.² The pulmonary tumors were TTF1-positive and estrogen receptor (ER)-negative, understandably resulting in designation as an independent lung primary. However, the clinical scenario was thought to be more in keeping with metastasis from the tumor within the uterine corpus and this was supported by the subsequent demonstration of PAX8 positivity within the lung neoplasms. The uterine corpus neoplasms were then stained with TTF1 and were found to be positive. It was speculated whether the propensity for pulmonary metastasis of these TTF1-positive endometrial carcinomas might in some way be due to a “homing” mechanism whereby TTF1-positive tumors had a tendency to metastasize to the lungs.² In the ensuing years, additional studies brought MLA into focus as a distinctive gynecologic primary with aggressive behavior, frequent expression of TTF1 and GATA3, absent to minimal hormone receptor expression, and morphologic features mimicking a variety of other tumor types.^3–5 MLA was then included in the 2020 World Health Organization (WHO) Classification of Endometrial and Ovarian Carcinomas.

MESONEPHRIC-LIKE ADENOCARCINOMA MORPHOLOGY

MLA is so-named because it has morphologic similarity to cervical MA.^1,2,5–12 However, it lacks an association with mesonephric remnants; MLAs are endometrial-based lesions in the uterus and are strongly associated with endometriosis in extrauterine sites.^13,14 Its rarity and histologic resemblance to more commonly encountered carcinomas make MLA susceptible to under-recognition, although there is also a risk of overdiagnosis. While a diagnosis of MLA requires incorporation of histologic and immunohistochemical features, sometimes with molecular data, appreciation of the characteristic morphologic features is foundational to the diagnosis.

Similar to MA, MLA exhibits a wide variety of architectural patterns including ductal, glandular (or pseudoendometrioid), tubular, papillary, corded/trabecular, nested, glomeruloid, retiform, solid and sieve-like; these mostly coexist in various combinations of 2 or more patterns with pseudoendometrioid, tubular and papillary patterns predominating (Fig. 1). Eosinophilic colloid-like secretions are commonly identified within tubular and/or glandular lumina and may also be seen within dilated glandular spaces lined by flattened epithelium resembling thyroid tissue. Uncommon but well-documented patterns include prominent solid growth, spindled morphology, and papillae with dense hyaline cores (Fig. 2). The admixture and merging of different architectural patterns, present in most MLAs, is a frequent enough finding that it can provide an initial clue to the diagnosis. It is important to note that much of the diagnostic confusion and overlap with other carcinomas (eg, endometrioid) is at the architectural level, particularly when ductal/glandular or papillary patterns predominate. Therefore, a strict adherence to the presence of characteristic cytologic features (discussed below) is required for diagnosis. Distinction from other carcinoma histotypes is discussed in further detail under the “Differential diagnosis” section.

FIGURE 1.

Common MLA morphologies. MLAs can demonstrate a variety of architectural patterns, including glandular growth, often with eosinophilic intraluminal material (A), elongate papillary structures (B), glomeruloid forms (C), and sieve-like patterns (D). Nuclei show mild to moderate atypia, often with nuclear clearing reminiscent of papillary thyroid carcinoma (E). Although marked nuclear atypia is absent, mitotic figures may be readily identifiable (F).

FIGURE 2.

Uncommon MLA morphologies. Some examples of MLA will show confluent solid growth (A), which may prompt confusion with dedifferentiation. Others will prompt a prominent spindled stromal response (B), which can generate confusion with carcinosarcoma. Occasional examples bear dense hyaline cores (C), which can mimic clear cell carcinoma or, in extrauterine settings, low-grade serous carcinoma.

Cytologically, MLA is typically composed of cuboidal or columnar cells with a high nuclear to cytoplasmic ratio and angulated nuclei showing open or vesicular chromatin (Fig. 1). There is usually some degree of nuclear overlap, and often nuclear grooves are at least focally present, though they may not be prominent. Cytologic atypia is typically moderate, and mitoses are usually conspicuous. Occasional nuclear hobnailing may be present, which can simulate clear cell carcinoma. Marked nuclear pleomorphism and/or atypical mitoses are seen in occasional cases, but in general are not features strongly associated with MLA, which is most often characterized by a monomorphic population of cells. The presence of metaplasia (ciliated, mucinous, squamous) is considered by some as exclusionary of MLA, although these have been reported in occasional cases.^2,6,8,10,15 When present, such metaplastic changes strongly raise the possibility of either an alternate diagnosis or MLA mixed with another histology.

MESONEPHRIC-LIKE ADENOCARCINOMA MIXED WITH OTHER HISTOTYPES

Although described in both uterine and extrauterine MLA, mixed histology is more common in extrauterine MLA; in one study, this was reported in ∼40% of cases,⁸ though the authors’ experience suggests that it is lower. The proportions of each component vary, and in some cases MLA represents a minor component. In endometrial tumors, MLA is usually mixed with atypical endometrial hyperplasia and/or endometrioid adenocarcinoma.¹⁶ Components reported to be admixed with MLA in extrauterine (mainly ovarian) tumors are much more varied and include endometrioid neoplasia (borderline and carcinoma), low-grade serous neoplasia (borderline and carcinoma), high-grade serous carcinoma, mucinous borderline tumor, clear cell carcinoma, neuroendocrine carcinoma, and germ cell tumors (teratoma, mixed germ cell tumor).^{2,6,7,13,16,17} In addition to an association with mixed histology, extrauterine MLA is also strongly associated with endometriosis (∼60% of cases),^6,8,13 supporting inclusion of MLA on the list of endometriosis-associated cancers.

Two other uncommon mixed histotype settings in which MLA may be encountered include as a component of carcinosarcoma and in dedifferentiated carcinoma. Mesonephric-like carcinosarcoma is rare and should not be mistaken for MLA with a solid or spindled pattern (Fig. 2). Because the solid pattern of MLA can have ovoid to short, spindled cells emanating from a glandular component, it can have a sarcomatoid appearance. For this reason, spindle cells alone are insufficient for a diagnosis of carcinosarcoma (Fig. 3). As in Müllerian carcinosarcoma, the epithelial and sarcomatous components should be distinct. The mesenchymal component of mesonephric-like carcinosarcoma should be overtly malignant, pleomorphic, and is often associated with heterologous elements, usually chondrosarcoma or less frequently rhabdomyosarcoma.^16,18,19 Dedifferentiated MLA is extremely rare, with only 3 cases reporting MLA associated with undifferentiated carcinoma.^16,20,21 The undifferentiated carcinoma component loses the cytologic features associated with MLA, exhibiting instead features previously described in undifferentiated carcinoma, including discohesive cells lacking an architectural pattern, rhabdoid morphology, high-grade nuclei, and macronucleoli (Fig. 4).

FIGURE 3.

MLA as a component of carcinosarcoma. MLA rarely occurs as the epithelial component of a bona fide carcinosarcoma. (A: left side: MLA; right side: malignant mesenchymal component). The malignant mesenchymal component is frequently heterologous, most often with cartilaginous differentiation (B: top right: MLA component; bottom left: malignant mesenchymal component).

FIGURE 4.

MLA as a component of dedifferentiated carcinoma. MLA rarely occurs in association with a dedifferentiated carcinoma. This case (A) showed typical MLA morphology (B) immediately juxtaposed with a dedifferentiated component (C). Both components showed strong cytokeratin and PAX8 expression, but TTF1 was lost in the dedifferentiated portion. The omental recurrence consisted exclusively of the dedifferentiated component (D).

MESONEPHRIC-LIKE ADENOCARCINOMA IMMUNOHISTOCHEMISTRY

Immunohistochemistry is vital for the diagnosis of MLA. When morphologic findings provoke the possibility of MLA, the MLA Consortium recommends a frontline panel of TTF1, GATA3, ER, and PR to help verify the diagnosis, with the classic immunophenotype being TTF1/GATA3-positive and ER/PR-negative (Fig. 5). Integration of p53 and mismatch repair (MMR) protein immunostaining results and additional testing for markers such as CD10, PAX8, PAX2, and SOX17 when other findings are not classical should be considered. This approach stems from literature showing that MLAs typically show some degree of positivity for TTF1 and/or GATA3, while hormone receptor staining should be either absent or limited.^{1,2,4,5,9,10,12,22,23} PAX8 is consistently positive in MLA and in the absence of diffuse positivity with this marker, a diagnosis of MLA should be doubted.^5,9 Luminal CD10 and nuclear PAX2 positivity have also been reported in the majority of MLA and can serve as second-line markers in diagnostically challenging cases in some contexts.^2,5,9,22,23 SOX17 has also emerged as a potentially useful stain with negative SOX17 staining setting MLA apart from other Müllerian tumors; however, SOX17 is not negative in all cases of MLA.^24,25 Finally, in keeping with the molecularly “simple” nature of these tumors, p53 is classically wild-type, MMR proteins are intact, and PTEN is most often preserved.^{2,5,9,10,12,22,23} The literature on these biomarkers in MLA is summarized in Table 1 and discussed in detail below, with emphasis on guidance for utilization and pitfalls in interpretation summarized in Table 2.

FIGURE 5.

MLA immunohistochemistry. This MLA showed large elongated glands (A) with open nuclear chromatin reminiscent of papillary thyroid carcinoma (B). TTF1 was strongly and diffusely positive throughout (C), while GATA3 showed weak to moderate expression in a subset of cells (D). CD10 demonstrated luminal expression (E), and ER was entirely negative with intact internal control staining in the background stroma (F).

TABLE 1.

Immunophenotype in Endometrial and Extrauterine MLA: Summary of Literature Focused on Immunostaining Patterns

Endometrial MLA immunohistochemistry
	McFarland et al.1	Pors et al.4	Kolin et al.9	Euscher et al.2	Kim et al.22	Mills et al.12	Kim et al.23	Total
PAX8	100% (7/7)	NP	100% (2/2)	NP	NP	NP	NP	100% (9/9)
GATA3	14% (1/7)	100% (4/4)	33% (1/3)	94% (15/16)	92% (23/25)	50% (1/2)	86% (6/7)	80% (51/64)
TTF1	85% (6/7)	100% (4/4)	100% (4/4)	69% (11/16)	100% (21/21)	100% (2/2)	57% (4/7)	85% (52/61)
CD10 (luminal)	83% (5/6)	75% (3/4)	100% (3/3)	100% (10/10)	81% (17/21)	0% (2/2)	86% (6/7)	87% (46/53)
Calretinin	50% (1/2)	0% (0/4)	NP	33% (5/15)	NP	NP	NP	29% (6/21)
ER−	100% (7/7)	50% (2/4)	50% (2/4)	71% (15/21)	68% (17/25)	100% (2/2)	86% (6/7)	73% (51/70)
ER+	0% (0/7)	50% (2/4)	50% (2/4)	29% (6/21)	32% (8/25)	0% (0/2)	14% (1/7)	27% (19/70)
≤10%	0% (0/7)	25% (1/4)	NA	14% (3/21)	16% (4/25)	0% (0/2)	14% (1/7)	14% (9/66)
>10%	0% (0/7)	25% (1/4)	NA	14% (3/21)	16% (4/25)	0% (0/2)	0% (0/7)	12% (8/66)
PR−	100% (6/6)	NP	75% (3/4)	93% (14/15)	88% (22/25)	NP	100% (7/7)	91% (52/57)
PR+	0% (0/6)	NP	25% (1/4)	7% (1/15)	12% (3/25)	NP	0% (0/7)	9% (5/57)
≤10%	0% (0/6)	NP	NA	7% (1/15)	8% (2/25)	NP	0% (0/7)	6% (3/53)
>10%	0% (0/6)	NP	NA	0% (0/15)	4% (1/25)	NP	0% (0/7)	2% (1/53)
Extrauterine MLA immunohistochemistry
	McFarland et al.1		Pors et al.4	Koh et al.26		Euscher et al.8		Total
PAX8	100% (5/5)		100% (1/1)	100% (5/5)		96% (22/23)		97% (33/34)
GATA3	60% (3/5)		100% (1/1)	100% (5/5)		94% 31/33		91% (40/44)
TTF1	100% (5/5)		100% (1/1)	80% (4/5)		75% (24/32)		79% (34/43)
CD10 (lum)	67% (2/3)		100% (1/1)	75% (3/4)		92% (12/13)		86% (18/21)
Calretinin	50% (2/4)		0/1 (0%)	NP		86% (12/14)		78% (14/18)
ER−	100% (5/5)		100% (1/1)	40% (2/5)		62.5% (20/32)		65% (28/43)
ER+	0% (0/5)		0% (0/1)	60% (3/5)		37.5% (12/32)		35% (15/43)
≤10%	0% (0/5)		0% (0/5)	NA		NA		0% (0/10)
>10%	0% (0/5)		0% (0/5)	NA		NA		0% (0/10)
PR−	100% (5/5)		NP	80% (4/5)		67% (20/30)		72.5% (29/40)
PR+	0% (0/5)		NP	20% (1/5)		33% (10/30)		27.5% (11/40)
≤10%	0% (0/5)		NP	20% (1/5)		NA		10% (1/10)
>10%	0% (0/5)		NP	0% (0/5)		NA		0% (1/10)

NA indicates results at the 10% threshold not provided; NP, stain not performed.

TABLE 2.

MLA Immunohistochemistry: Summary of Expected Staining, Ordering Guidance, and Ongoing Challenges and Controversies

IHC stain	Expected staining in MLA	MLA consortium ordering guidance	Challenges/controversies
TTF1	Usually expressed in at least some tumor cells; may be negative, especially if GATA3 is positive. Expression is more common than in cervical mesonephric adenocarcinoma.	Order first-line; should see staining with TTF1 and/or GATA3	There is no requisite minimum extent; some cases which are entirely negative for TTF1 and GATA3 have been reported Occasionally expressed in other histotypes, including endometrioid and serous carcinomas.
GATA3	Often expressed in at least some tumor cells; may be negative, especially if TTF1 is positive. Expression is less common than in cervical mesonephric adenocarcinoma.	Order first-line; should see staining with TTF1 and/or GATA3	There is no requisite minimum extent; some cases which are entirely negative for TTF1 and GATA3 have been reported. Is occasionally expressed in other histotypes, including endometrioid and serous carcinomas.
ER	Absent to limited (≤10%)	Order first-line	Degree of allowable staining is extremely controversial; however, one should doubt a diagnosis of MLA if diffuse (>50%) Recommend expert consultation for cases with >10% expression. Admixture with another histotype should be considered in tumors with ER staining limited to a subset of the tumor.
PR	Absent to limited (≤10%)	Order first-line	Limited data in MLA. Recommend expert consultation for cases with >10% expression. Admixture with another histotype should be considered in tumors with PR staining limited to a subset of the tumor.
CD10	Luminal expression	Order second-line if first-line immunostaining results are inconclusive.
Calretinin	NA	Not recommended	NA
PAX8	Positive	Consider ordering to classify challenging tumors, particularly extrauterine	While positivity is essentially a requisite for MLA, other müllerian-derived tumors will also be positive and therefore it is of no value for discriminating within this family.
PAX2	Positive	Consider ordering to help classify challenging tumors, particularly extrauterine	Limited data in MLA.
SOX17	Negative/limited in most cases	Consider ordering help classify challenging tumors, particularly extrauterine	While MLAs typically show negative/limited expression, other Müllerian-derived will be positive (with the exception of mucinous tumors). A SOX17-/PAX8+ profile can therefore be supportive of MLA in the right morphologic context.
p53	Wild-type expression	Encourage ordering to ensure that results are in concordance with MLA diagnosis if not already performed as part of reflex testing	The existence of rare cases of p53-abnormal MLA remains controversial; recommend expert consultation before rendering this diagnosis. Abnormal p53 limited to a subset of the tumor should prompt consideration for a MLA mixed with another histotype, such as serous carcinoma.
Mismatch repair proteins	Intact expression	Encourage ordering to ensure that results are in concordance with MLA diagnosis if not already performed as part of reflex testing	MLA should not be diagnosed in the setting of MMR deficiency.

GATA3/TTF1

GATA3 and TTF1 have emerged as the 2 most foundational markers in diagnosing MLA and are discussed in tandem due to their complementary nature. Although some degree of GATA3 and/or TTF1 expression is expected for MLA, there is considerable variability in how these markers are expressed in individual tumors.^{2,5,9,10,12,22,23} Importantly, there is no minimum positivity threshold for GATA3 or TTF1, and many cases show only focal expression. When both markers are positive, they sometimes demonstrate inverse staining patterns with areas positive for GATA3 being negative for TTF1 and vice versa. Initial evidence suggested that while GATA3 is the more sensitive marker for MA, TTF1 is more often and more diffusely positive in MLA.^5,9 McFarland et al.¹ found TTF1 expression in 92% of cases from their original description of 12 MLAs of the endometrium (7) and ovary (5), while only 27% stained for GATA3. In their immunohistochemical characterization of both MA and MLA, Pors et al.⁵ found both GATA3 and TTF1 positivity in all 5 MLAs in the series (4 endometrial and 1 ovarian), although notably GATA3 was only focal in 2 of the endometrial cases. Kolin et al.⁹ described 4 endometrial MLAs; all exhibited TTF1 expression, but only one of the 3 cases with available GATA3 expressed this marker and was only focally positive. Kim et al.²² subsequently evaluated 25 MLAs and found that TTF1 was slightly more sensitive than GATA3 (100% vs. 92%). Finally, neither of the MLAs described by Mills et al.¹² exhibited GATA3 expression, but both were diffusely TTF1-positive.

While these studies suggested that TTF1 is more sensitive for MLA than GATA3, this has not been born out in later series. Euscher et al.² characterized 23 endometrial MLAs and found that GATA3 expression was more frequent than TTF1; 94% expressed GATA3, while 69% were positive for TTF1. Furthermore, GATA3 was positive in 4 cases which were TTF1-negative, whereas TTF1 was not expressed in the single GATA3-negative tumor. In the Kim et al.²³ study of 7 endometrial MLAs, 86% expressed GATA3, while only 57% expressed TTF1; the single GATA3-negative case also lacked TTF1. Similarly, Euscher et al.’s⁸ subsequent study of extrauterine MLAs found that more cases expressed GATA3 than TTF1 (94% vs. 75%). Overall, the aggregated literature suggests that there is some variability in the sensitivity of TTF1 and GATA3 for MLA. Whether this reflects differences in laboratories and their assays or the limited number of cases in most studies is unknown. Regardless of the cause, the MLA Consortium concluded that enlisting both biomarkers is prudent to maximize diagnostic sensitivity.

The occasional dual-negative cases also suggest that GATA3 or TTF1 expression may not be strictly required for an MLA diagnosis.² In one study by Euscher et al.,² the single TTF1-negative/GATA3-negative MLA was based on morphology and ER negativity. Additional examples of dual-negative MLA were reported by Pors et al.⁵; they described 6 MLAs (1 endometrial, 5 ovarian) which lacked supportive immunohistochemistry but had appropriate morphology and molecular findings for a MLA diagnosis. In 2022, Kim et al.²³ also described a GATA3 and TTF1-negative MLA; in this case, the diagnosis was substantiated by morphology,PAX2 and luminal CD10 positivity, ER/PR negativity, and a KRAS p.G12V mutation. These cases suggest that occasional TTF1-negative/GATA3-negative MLA exist, although their rarity underscores the importance of thorough morphologic assessment and additional immunohistochemical and/or molecular evidence before rendering a MLA diagnosis in this setting; expert consultation may also be beneficial in this context.

It is also critical to emphasize that neither TTF1 nor GATA3 expression is unique to MLA. Mills et al.¹² reviewed 300 consecutively diagnosed endometrial carcinomas to investigate the frequency of missed MLA in their patient population. They found TTF1 expression in 9 (3%), 2 of which co-expressed GATA3. While 2 of the 9 TTF1-positive tumors (including one of the cases co-expressing GATA3) were ultimately reclassified as MLA due to morphologic, immunohistochemical, and molecular features, the remaining 7 retained diagnoses of serous carcinoma, endometrioid carcinoma, and dedifferentiated carcinoma after morphologic re-review and molecular evaluation. Notably, 4 were ER-negative, highlighting the dangers of diagnosing MLA based on immunohistochemical profile alone. The pitfalls of using GATA3 or TTF1 in isolation to diagnose MLA were further highlighted by Lee et al.²⁷; they assessed GATA3 and TTF1 alongside other biomarkers used in MLA diagnosis in 50 cases of confirmed conventional low-grade endometrioid carcinoma and found GATA3 expression in 10% and TTF1 expression in 6%, again underscoring the imperfect specificity of these stains for a diagnosis of MLA.

CD10

CD10 is also a useful biomarker to aid in MLA diagnosis, although less useful than TTF1 and GATA3. In the McFarland et al.¹ series, 83% of endometrial tumors had at least focal positivity, although only one case exhibited diffuse staining. Luminal CD10 expression was also reported in 3 of 4 (75%) endometrial MLAs in the 2018 Pors and colleagues study, with positivity extent ranging from 10% to 50% of cells.^4,5 In all, 100% of MLAs were reported to be luminal CD10-positive in the Euscher et al.⁸ and Kolin et al.⁹ studies, albeit often with only very focal expression. Kim et al.²² and Kim et al.²³ identified luminal CD10 in the majority (81% and 86%, respectively) of tumors. Neither case in the Mills et al.¹² study was positive for this marker. There are more limited data on CD10 in extrauterine MLAs, but existing evidence suggests that they also show frequent luminal expression. McFarland and colleagues found CD10 positivity in 2 of 3 ovarian MLA. In 2022, Koh et al.²⁶ reported 5 cases of ovarian MLA; of the 4 with CD10 immunohistochemistry results, 3 were focally CD10-positive. Euscher et al.⁸ reported results from 33 examples of ovarian and peritoneal MLA and found that 92% had some degree of luminal CD10 staining.

While the data on CD10 as a sensitive marker for MLA is overall encouraging, discussions among the MLA Consortium highlighted the frequent focal positivity with this marker even when compared with GATA3 and TTF1. Moreover, several members noted frequent variability in performance across laboratories, with some experts' experience suggesting that the literature overestimates its sensitivity in practice. As such, CD10 emerged as a second-line marker for MLA diagnosis. That said, pathologists should have a low threshold for enlisting this biomarker in cases where the initial panel fails to solidify the diagnosis, particularly if they have confidence in its performance in this context in their own laboratory. It is also important to emphasize that CD10 is poorly specific and can be expressed in other Müllerian carcinomas, including a subset of low-grade endometrioid carcinomas, which are commonly in the morphologic differential for MLA. In 2023, Lee et al.²⁷ investigated MLA markers in 50 examples of FIGO grade 1 to 2 endometrial endometrioid carcinoma and found 3 positive cases (6%): 2 had luminal-only staining, while the third had luminal expression combined with cytoplasmic and membranous reactivity.

Calretinin

Calretinin was initially considered a potentially helpful immunostain for MLA, but early studies showed very low sensitivity with 50% of tumors in the McFarland and colleagues series, none of the Pors and colleagues cases, and only 33% of the Euscher and colleagues cases expressing this biomarker.^1,2,5 Given these data, the focal nature of many positive cases, as well as consortium members' personal experience with insensitivity in this setting, the MLA Consortium recommends against its use to aid in MLA diagnosis.

ER/PR

The allowable extent of ER and PR expression in MLA remains one of the most controversial aspects of this diagnosis. While most correctly diagnosed MLAs will have absent or ≤10% ER staining, examples with more extensive staining are reported in the literature and are occasionally encountered in the authors’ experience. All cases in the original McFarland and colleagues' description, which included both endometrial and ovarian tumors, were ER/PR negative, although notably this was a requirement for inclusion.¹ Pors and colleagues described 5 cases of MLA; 3 of these were ER-negative, and 2 were ER-positive. One of the ER-positive cases had expression in <10% of cells, but the other showed 55% of tumor cells expressing ER. Of the 4 cases reported by Kolin et al., 2 were ER-negative while 2 expressed ER; 1 positive case was described as “patchy,” the other “heterogeneous.” This was also one of the only studies to assess PR expression, and found that while 3 were PR-negative, 1 showed “heterogenous” staining. Euscher et al.² found that 71% of cases in their endometrial MLA series were ER-negative; of the 6 ER-positive tumors, 3 had ≤10% staining while the remaining 3 had faint to moderate expression in 15% to 40% of tumor cells. Of the subset with available PR results, 93% of cases were PR-negative, and the single positive case had ≤10% expression. Kim et al.²² described >10% ER expression in 16% of MLA, with staining in 20% to 50% of cells. Notably, only one of these had PR staining, which was limited to 5% of tumor cells. Both MLAs in the 2022 Mills et al.¹² study were ER-negative, as were 86% of MLAs in the Kim et al.²³ study published that year; the single ER-positive case in that study had ≤10% expression.

While most of these data derive from endometrial MLA, some recent studies of MLAs arising outside the endometrium have shown higher rates of ER/PR expression. Koh et al.²⁶ reported ER positivity in 60% of ovarian MLAs, although expression was focal in all cases and only one had PR expression, which was also focal. Euscher et al.⁸ identified ER and PR staining in 37.5% and 33% of extrauterine MLA cases, respectively. While most had low levels of expression, 3 cases with ≥50% ER staining (maximum: 60%) were reported; no cases exceeded 30% PR positivity. Notably, ER expression was more commonly reported in tumors which had admixed components of other Müllerian histotypes, raising questions about the precise delineation between MLA and other components.

Although the upper limit of ER expression allowable in MLA remains controversial, the existing body of literature indicates that diffuse (>50%) expression is very rarely encountered in rigorously classified endometrial MLA, and that cases with >10% expression are uncommon. More widespread ER expression may be more common in extrauterine MLA, but it remains unclear whether mixed morphologies contribute to the reportedly higher hormone receptor positivity in these cases. PR has not been as well-studied in MLA, but when assessed it is consistently negative or low (usually ≤10%), suggesting that it can be a valuable companion for ER in this setting. Given the infrequency of more than focal hormone receptor expression in MLA, pathologists are encouraged to consider expert consultation before rendering a diagnosis of MLA in a tumor with >10% ER staining, particularly if the tumor is also PR-positive. More extensive staining should also prompt consideration of a potential mixed malignancy and trigger careful morphologic review to assess whether areas of ER/PR expression correlate with a region of the tumor that is better classified as a more common histotype. It is not clear whether some examples of more extensively ER-positive tumors represent mixed tumors, as such mixed tumors were not well-recognized in the early years following the first description of MLA. Finally, pathologists should remain attuned to other ER/PR-negative/low-expressing gynecologic adenocarcinomas before making a diagnosis of MLA. HPV-associated and HPV-independent endocervical primaries and uncommon endometrial carcinoma types, such as clear cell carcinoma and gastric/gastrointestinal-type carcinoma, should also be considered in gynecologic gland-forming tumors that have absent to very limited hormone receptor expression.

Immunohistochemical Molecular Surrogates (p53, MMR Proteins, PTEN)

Because TP53 pathogenic variants are not a feature of MLA (discussed below), p53 expression should be wild-type in the vast majority of these tumors.^{1,2,5,9,10,12,22,23} Rare examples of p53-abnormal MLA have been described, but this is extremely uncommon in strictly defined cases (see the “Molecular features” section for further details). Given this, the MLA Consortium discourages pathologists from making a MLA diagnosis in the setting of aberrant p53 immunohistochemistry without expert consultation.¹⁰ Similarly, intact MMR protein expression is considered foundational for a MLA diagnosis based on the molecular profile of these tumors.^{2,5,9,10,12,22,23} Only very occasional purported examples of MMR-deficient pure MLA have been reported (discussed below), and the MLA Consortium emphasizes that pathologists should seek expert review before rendering this diagnosis in the MMR-deficient setting.

PTEN has potential value as an additional biomarker for tumors with challenging morphology and immunoprofiles which have a differential of MLA versus endometrioid carcinoma, as PTEN mutations are uncommon in MLA, but are frequent in endometrioid carcinoma.^{2,5,9,10,12,22,23} However, PTEN immunohistochemistry has not been well-studied in this tumor type, and caution should be exercised when enlisting it in this setting since a subset of MLAs have been shown to demonstrate PTEN alterations (see the “Molecular Features” section for further details). Future studies assessing PTEN expression, particularly in mixed tumors with MLA and endometrioid components, will be of interest.

Other Immunomarkers to Consider (PAX8, PAX2, SOX17)

Several other markersmay be useful for MLA diagnosis as part of a panel approach, particularly in the extrauterine setting where non-Müllerian entities are often in the differential. PAX8 is consistently diffusely positive in MLA, reflecting the Müllerian origin of these tumors.^1,5,9 This marker may be additive for difficult-to-classify tumors, particularly in the extrauterine setting, as negativity would essentially eliminate MLA from the differential. PAX2 has been less well-studied but shows early promise, especially in ovarian tumors with a morphologic differential of endometrioid carcinoma. Köbel and colleagues found that retained PAX2 was highly sensitive and specific for ovarian MLA in cases with a morphologic differential of endometrioid carcinoma (which typically shows loss of PAX2), and suggested its enlistment as a first-line marker alongside GATA3/TTF1 and ER/PR in this setting.²⁸ SOX17 has also emerged as a helpful aid in MLA diagnosis: this marker commonly shows negative to very limited expression in MLA, whereas other Müllerian carcinoma types typically show diffuse expression of this marker.^24,25

Immunohistochemistry in Mixed Tumors

The diagnosis of MLA mixed with another histotype requires immunohistochemical support, with the MLA component showing a distinct immunohistochemical profile from the other histotypes which are present (Fig. 6). While the non-MLA component usually shows the prototypic expression pattern for its given histotype, MLA Consortium members have encountered rare cases that show unexpected immunoprofiles, such as strong expression of MLA-type markers (GATA3/TTF1) within the background non-MLA component (Fig. 7). Such cases highlight the shared origin of these tumors, with the MLA component representing a clonal evolution from a more conventional “parent” Müllerian tumor in mixed cases.

FIGURE 6.

Mixed Endometrioid carcinoma and MLA. This tumor showed 2 discrete histologies: an endometrioid component (A) and an MLA component (B). The endometrioid portion of the tumor was strongly ER-positive (C) while the MLA portion was completely ER-negative (D). Conversely, TTF1 was negative in the endometrioid component (E) but diffusely positive within the MLA (F).

FIGURE 7.

Mixed carcinoma with unexpected immunostaining. This tumor showed 2 morphologies: an endometrioid component (A, lower left) and an MLA component (A, upper right). The MLA portion had morphologically distinctive cytology with prominent cytoplasmic clearing (B); Napsin A was negative, arguing against clear cell differentiation. The endometrioid portion was strongly ER-positive (C, lower left) while the MLA was completely ER-negative (C, upper right). Both areas of tumor, however, showed strong and diffuse expression for TTF1 (D).

MESONEPHRIC-LIKE ADENOCARCINOMA MOLECULAR FEATURES

Molecular testing may be helpful in the diagnosis of MLA, although it is not strictly necessary for tumors with the appropriate classical morphology and immunophenotype. MLA shows significant molecular homology with cervical MA, but there are also divergences which substantiate its classification as a distinct entity. At the time of writing, 35 studies addressing molecular findings in MLA exist in the literature, with representation of 262 tumors, including 158 endometrial MLA and 104 extrauterine MLA.^{2,5,7–9,11,12,14–17,22,23,26,29–49} As has been shown with cervical MA, MLAs consistently demonstrate canonical activating abnormalities in KRAS as well as frequent 1q gains. (Table 3) Also, similar to MLA, TP53, POLE, and MMR genes are typically wild-type. In contrast to MA, however, MLA often shows abnormalities in genes that are typically associated with Müllerian (endometrioid) differentiation, such as PIK3CA, PTEN, and CTNNB1. Further details about the molecular characteristics of MLA based on the review of this literature are summarized in Table 3 and discussed below, with attention to similarities to and differences from MA as well as areas of ongoing controversy.

TABLE 3.

Molecular Features of MLA: Summary of Literature on Reported Cases

	Molecular alterations							CNV
Gene/genomic region	KRAS	ARID1A	PIK3CA	PTEN	CTNNB1	POLE	TP53	1q gain
Endometrial	84.3% (129/153)	17.4% (15/86)	19.1% (21/110)	12.6% (13/103)	6.6% (7/106)	2.5% (2/79)	9.0% (6/67)	80.6% (50/62)
Ovary/Extra-uterine	84.5% (71/84)	9.4% (6/64)	29.6% (21/71)	3.1% (2/65)	7.2% (5/69)	1.4% (1/74)	7.8% (6/77)	89.7% (35/39)

KRAS

Canonical activating KRAS mutations are present in the vast majority of MLA cases, regardless of anatomic site.^{9,23,26,30,36,50} While KRAS pathogenic variants are typical of both MLA and MA, rare cases lacking KRAS mutations have been shown to harbor mutations in other RAS/RAF family genes, including NRAS, HRAS, RRAS2, and BRAF.¹⁷

Genes Associated With Endometrioid Tumors: PIK3CA, PTEN, and CTNNB1

Unlike a large majority of MAs, MLAs can sometimes harbor genetic alterations that are frequently found in Müllerian tumors, such as mutations in PIK3CA and PTEN. When mixed morphology is present, KRAS mutations are usually shared across both histologic components, substantiating the concept that MLA are Müllerian in origin, in contrast to MA, which arises from mesonephric (Wolffian) remnants/hyperplasia^{13,16,17,38,39}. Review of the molecular literature on MLA and MA demonstrates that PIK3CA mutations/variants occur in a subset of endometrial and ovarian MLA [19.1% (21/110) and 29.6% (21/71), respectively] but are less common in MA [8% (3/36)]. Similarly, PTEN mutations/variants have been reported in 12.6% (13/103) of endometrial and 3.1% (2/65) of ovarian/extrauterine MLA, but to our knowledge have not been reported in MA.^2,8,17,26,33

TP53

MLAs are classically TP53-wild-type.^{2,7–9,11,12,16,30,31,35,36,50} TP53 variants are infrequently reported in MLA, but have been demonstrated in 9% (6/67) of endometrial MLAs, and 5% (4/77) of ovarian MLAs reported in the reviewed literature. The prevalence of true pathogenic TP53 alterations is likely even lower in that some reported cases provoke questions about whether they truly represent MLA. For example, one of the endometrial MLAs with a reported TP53 mutation was KRAS wild-type and had alterations in EGFR, PIK3CA, ARID1A, CDKN2A, and PPP2R1A. Furthermore, this tumor lacked typical chromosomal gains or losses of 1q, 10, 12, 20.³² Some other reported cases, however, have exhibited classic morphologic and immunohistochemical features of MLA and harbored pathogenic KRAS mutations in addition to pathogenic TP53. For example, in a study by da Silva and colleagues, one metastatic endometrial MLA harbored a pathogenic KRAS G13D mutation co-occurring with a pathogenic TP53 hotspot mutation (I254N).¹⁷ Lin and colleagues reported findings from a patient with a history of cervical MA and a suspected lung metastasis 6 years after diagnosis, demonstrating concordance of the KRAS alteration (G12D) and chromosomal 1q, 2, 10, 12, and 20 gains in both specimens as well as new TP53 alterations (R249S and R280G) in a circulating tumor DNA (ctDNA) liquid biopsy compared with the original sample.³² While these data suggest that late acquisition of TP53 mutations can rarely occur in MLA and MA, given their rarity, the MLA Consortium emphasizes that pathologists should seek expert review before rendering a MLA diagnosis in the setting of a known pathogenic TP53 variant.

POLE

Very rare examples of MLA with POLE mutations/variants are reported in the literature, including in 2.5% (2/79) of endometrial MLA and 1.4% (1/74) of ovarian MLA. The first endometrial case, reported in da Silva et al.¹⁷ harbored a nonexonuclease domain but likely pathogenic splice site mutation in POLE (p.X1577_splice), in addition to KRAS, MAP2K1, TMPRSS2, PIK3CA, FBXW7, ARID1B, and AMER1 mutations. It is difficult to determine whether this tumor had a true ultramutated profile, and by current definitions, this alteration does not qualify as a pathogenic POLE mutation.⁵¹ The second endometrial case with a hotspot POLE mutation harbored no RAS mutations but did have a TP53 mutation, as well as mutations in several other genes. The specific POLE mutation was not reported, and it is difficult to determine whether the tumor had a true ultramutated profile.³⁶ Finally, Mirkovic et al.¹⁶ reported an ovarian carcinosarcoma with a POLE frameshift mutation (C445Sfs*58) was detected in the MLA component only; this mutation was not identified in the sarcomatous component, nor in the bowel metastasis. Despite the frameshift mutation, this tumor did not demonstrate a true ultramutated profile, bearing significant alterations in only five other genes, including a KRAS mutation, which was shared across the MLA component, the sarcomatous component, and the bowel metastasis.

To summarize, convincing evidence of a true POLE-mutated MLA has not been reported. A tumor with a canonical pathogenic POLE exonuclease domain mutation is not likely to be MLA. The MLA Consortium emphasizes that pathologists should seek expert review before rendering this diagnosis in the POLE-mutated setting.

Mismatch Repair (MMR) Genes

To our knowledge, there is only one report of a purported MMR-deficient endometrial MLA in the literature.⁴¹ This was in the context of a mixed tumor purported to demonstrate three distinct neoplastic components: (i) endometrioid carcinoma, FIGO grade 2; (ii) undifferentiated carcinoma; and (iii) MLA (which was reported to comprise ~50% of overall tumor volume). The putative MLA component was morphologically and immunohistochemically distinct from the endometrioid component, exhibiting positive TTF1/GATA3 expression and negative ER/PR; p53 was wild-type throughout, and MLH1 and PMS2 were lost throughout. A KRAS mutation (p.Q16L) was identified in the MLA component, but not in regions with other histology. Notably, KRAS Q16L is not a known pathogenic mutation, though it is likely this was a typographical error in the manuscript, and the actual mutation identified resulted in p.Q61L in KRAS. This particular detail has not been resolved to date. There is also a single reported example of a microsatellite unstable putative MA; however, this tumor lacked RAS mutations and the study did not include photographs, precluding morphologic review.⁵² Given the lack of well-validated MMR-deficiency/high-level microsatellite instability in morphologically vetted MLA, tumors bearing this molecular abnormality are highly unlikely to represent bona fide examples of these entities. The MLA Consortium encountered no convincing case of MMRd/MSI MLA among the hundreds reported in the literature, nor in routine and consultative clinical practice; as such, we strongly emphasize that pathologists should seek expert review before rendering a diagnosis of MLA in the MMR-deficient setting.

1q Gain and Other Copy Number Variants (CNV) in MLA

Similar to MA, the majority of MLA exhibit chromosome 1q gain (~75%–95%) and frequent chromosome 10 and 12 gains.^{11,17,30,32,35} Less frequent recurrent alterations in MLA include gains of chromosome 2 (~30%–50%), loss of chromosome 1p (~30%–45%) and loss of chromosome 9 (~5%–25%).^17,20,32,35 de Silva et al.¹⁷ found that chromosome 12 gains were significantly more frequent in ovarian MLA (10/15, 67%) than in MA (2/8, 25%) or endometrial MLA (2/13, 15%; P=0.013, Fisher exact test). In a study by Kommoss et al.³⁵, loss of chromosome 1p together with chromosome 1q gain was also identified in both MA and low-grade serous carcinoma (LGSC), while chromosome 2 gain was almost exclusively identified in MLA and MA groups.

Kommoss et al.³⁵ also investigated CNV burden in a series of 19 MLAs (5 endometrial and 14 ovarian) and 7 MAs in the context of a relatively large cohort of various gynecologic carcinoma types. MLA and MA showed higher global CN changes compared with POLE, MMRd, and no specific molecular profile (NSMP) endometrioid carcinomas, and ovarian LGSC, as well as lower CNV burden compared with the copy number high (p53 abnormal) tumors. Hence, MLAs harboured moderate levels of genomic instability. Similarly, copy number analysis by da Silva et al.¹⁷ revealed moderate levels of genomic instability in both MA and MLA with no recurrent amplifications or homozygous deletions.

Epigenetic Studies of MLA

Kommoss et al.³⁵ investigated DNA methylation patterns in a series of 19 MLAs (5 endometrial and 14 ovarian) and 7 cervical MAs in the context of a relatively large cohort of various gynecologic carcinoma types. This included 50 endometrioid endometrial carcinomas comprising all 4 TCGA groups, 28 uterine serous carcinomas, 21 uterine clear cell carcinomas, 10 ovarian LGSCs, and 11 high-grade serous carcinomas. Unsupervised hierarchical clustering analysis identified a shared DNA methylation cluster of MLA and MA, which was distinct from the other groups. Therefore, this study revealed a close epigenetic relationship between MLA and MA, which was significantly different from clusters of other endometrial and ovarian carcinomas.

Molecular Features of Mixed Tumors

When all components of a mixed tumor have been sequenced separately, whether in the endometrium or extrauterine sites, they have usually demonstrated clonality between the MLA and the non-MLA components, although there have been rare exceptions, and the number of tumors studied is small. The shared alterations frequently include but are not limited to KRAS and other genes in the RAS family.^{7,16,17,34,37–40,44,45}

MESONEPHRIC-LIKE ADENOCARCINOMA DIFFERENTIAL DIAGNOSIS

The protean morphologic appearance of MLA results in overlap with a host of other tumors at uterine and extrauterine sites. Prior to the description of MLA as a specific tumor type, these neoplasms were likely diagnosed as a wide variety of other histotypes. More recently, increased awareness of MLA also means that there is a risk of overdiagnosis, particularly when more common entities express MLA-associated markers such as TTF1/GATA3. It is therefore critical for pathologists to have a robust understanding of the differential diagnosis of MLA in both uterine and extrauterine settings.

Cervical Mesonephric Adenocarcinoma (MA)

Occasionally, there is a diagnostic issue as to whether a tumor represents an endometrial MLA or a cervical MA. This may be a consideration in a small biopsy specimen, especially if clinically and radiologically it is not clear whether a tumor is arising from the cervix or the endometrium or where the biopsy has been taken from. This may also be an issue in a resection specimen if tumor involves both the uterine corpus and the cervix. Usually in such cases, the distribution of the tumor will help determine the site of origin, but sometimes this is not the case and a constellation of pathologic features, including immunohistochemistry and occasionally molecular testing, may be necessary to distinguish between the 2 tumor types. Similarities and differences between MLA and MA are detailed in Table 4 and discussed below.

TABLE 4.

Mesonephric Adenocarcinoma (MCA) Versus MLA

	Mesonephric-like adenocarcinoma	Mesonephric adenocarcinoma
Primary tumor site	Endometrium and ovary; rarely extraovarian (arising from endometriosis)	Cervix; rarely vagina
Pathology	Variety of architectural patterns; absence of squamous and mucinous differentiation; angulated overlapping clear vesicular nuclei	Variety of architectural patterns; absence of squamous and mucinous differentiation
Immunophenotype	Often TTF1, GATA3, PAX8, CD10 positive; p53 wild-type; p16 focal; MMR proficient; ER/ PR negative or sometimes limited positivity, especially with ER	Often GATA3, PAX8, CD10 positive; typically TTF1 negative but may be positive; p53 wild-type; p16 focal; MMR proficient; ER/PR negative
Associated findings	Endometriosis; other Mullerian lesions; lack of mesonephric remnants	Mesonephric remnants
Molecular features	KRAS, NRAS, PIK3CA, PTEN, ARID1A mutations	KRAS/NRAS mutations

MLA within the uterine corpus arises from the endometrium, and within the ovary or, less commonly, at extrauterine and extraovarian sites, mostly from endometriosis; these neoplasms are not associated with mesonephric remnants. They are not “centered” in the myometrium with limited or no endometrial involvement, as would be expected with a true MA arising from mesonephric remnants in the uterine corpus. True MAs arise from the cervix or rarely from the vagina, often in association with benign mesonephric remnants, although these are not always seen. Although there have been reports of primary uterine corpus MAs, it is likely that most or all of these actually represent MLAs.³⁰ In support of this is the observation that, although it is often stated that mesonephric remnants may occur in the outer aspects of the myometrium, this is extremely rare.

The morphologic features are similar between MLA and MA. Both neoplasms typically exhibit a variety of architectural patterns, often admixed within a single tumor. However, MLAs are characterized by clear angulated vesicular nuclei, sometimes with overlapping and grooves. These nuclear features are not a particular feature of MA, and the nuclei in these neoplasms are often more hyperchromatic. Admixture with another tumor type may be a clue to an MLA diagnosis since MAs almost always occur as pure neoplasms.^16,17

The immunophenotypes of MLA and MA are similar, but not identical. Both tumor types are typically diffusely positive with PAX8 and CK7, exhibit wild-type p53 expression, non-diffuse p16 expression, are negative for WT1, and are MMR-intact. They are both often positive with variable distribution for TTF1, GATA3, and CD10. However, MLAs are more likely than MAs to be positive with TTF1.^5,53 ER and PR staining may be useful in distinguishing between these neoplasms as MA are almost always entirely negative, while MLAs may show limited stainingwith both markers in some cases.

From a molecular standpoint, both MLA and MA commonly exhibit KRAS and, to a lesser extent, NRAS mutations.¹⁶ However, as discussed, MLAs more commonly exhibit additional mutations which are characteristic of Müllerian carcinomas, such as PIK3CA and PTEN, whereas MAs usually lack such alterations.

Endometrioid Carcinoma

MLA and endometrioid carcinoma, particularly FIGO grade 1 to 2 tumors, may closely resemble one another, and before the description of MLA most were likely diagnosed as low-grade endometrioid carcinoma^{2,5,9,10,12,22,23} (Table 5). Both neoplasms may be composed of glandular structures of varying degrees of complexity lined by moderately atypical cells. While both may have other architectural patterns, including cribriform and papillary growth, well-formed glomeruloid structures are much more typical of MLA, although they are not seen in all cases. Intraluminal eosinophilic secretions may also be present in both MLA and endometrioid carcinoma, although MLA-associated secretions typically have a denser and brighter appearance than the secretions typical of endometrioid carcinomas. Some MLAs have dense hyaline stromal material that may prompt consideration for the corded and hyalinized variant of endometrioid carcinoma.⁵⁴ Finally, mucinous and squamous metaplasia are common in endometrioid carcinomas, but are not a feature of MLA.

TABLE 5.

MLA Versus Endometrioid Carcinoma and Serous Carcinomas

Histologic feature	MLA	Endometrioid carcinoma	Serous carcinomas
Nuclear features	Generally moderate nuclear atypia with clear to vesicular chromatin and small nucleoli; angulated and hyperchromatic nuclei can also be seen; may show papillary thyroid carcinoma-like nuclear clearing and grooves; occasional hobnailing; mitoses may be brisk but generally not atypical	Mild to moderate nuclear atypia with oval to round enlarged nuclei; often vesicular; variably prominent basophilic nucleoli; mitoses may be brisk but usually are not atypical forms	Low-grade: mild to moderate nuclear atypia without pleomorphism High-grade/uterine serous: marked nuclear atypia and pleomorphism with prominent eosinophilic nucleoli; atypical mitotic forms can be present
Architectural patterns	Wide range of patterns, including tubular, papillary, sieve-like, glandular, and solid spindled growth; glomeruloid forms may be seen	Varying proportions of glandular and solid growth can show villoglandular architecture	Low-grade: typically papillary, often with a borderline background High-grade/uterine serous: typically papillary, forming slit-like spaces; can be solid and glandular; discohesive floating cells common
Extracellular material	Often have dense pink (“colloid-like”) eosinophilic luminal material; can show dense hyaline matrix material; Psammoma bodies are uncommon	Eosinophilic luminal material is common, typically pale; it can show dense hyaline matrix material (particularly in the corded and hyalinized variant)	Low-grade: psammomatous calcifications are typically abundant; lack dense hyaline matrix material High-grade: often show scattered psammoma bodies; lack dense hyaline matrix material
Metaplasias	Should be absent	Often present, can include squamous, tubal/ciliated, and mucinous metaplasia	Low-grade: ER/PR typically positive; p53 wild-type; TTF1/GATA3-negative; WT1 positive High-grade: ER/PR typically positive, but may be limited; p53 abnormal; focal TTF1/GATA3 expression can occasionally be seen; WT1 positive in tubo-ovarian HGSC
Immunohistochemical staining patterns	ER/PR absent to minimal TTF1 and/or GATA3+ p53 wild-type, MMR-intact, PTEN variable WT1 negative	Usually ER/PR-positive unless high-grade TTF1 and GATA3 expression is uncommon p53 is usually wild-type but may be aberrant ~30% will show loss of one or more MMR proteins ~70%–80% PTEN loss	Low-grade: KRAS, BRAF, NRAS mutations common High-grade: TP53 PVs are canonical
Molecular features	Classically show KRAS PVs impacting G12 (G12V, G12D); some show chromosome 1q gains Lack PTEN, MMR PVs; not associated with MLH1hm Typically lack TP53 PVs though data is evolving	PTEN PVs seen in the majority; PIK3CA, ARID1A PVs common; subset shows TP53 PVs; ~25% MLH1hm; ~5% MMR gene PVs (may be germline or somatic)	Low-grade: no clear associations with clinical risk factors or heritable cancer syndromes High-grade: a subset are associated with germline syndromes impacting homologous recombination genes (BRCA1/2, BRIP1, RAD51, etc.)
Clinical associations	No known associations with clinical risk factors or heritable cancer syndromes; extrauterine cases may arise in endometriosis	Commonly associated with poorly opposed estrogen states, including elevated BMI; ~2%–5% have underlying Lynch syndrome; rare cases associated with Cowden syndrome; extrauterine cases may arise in endometriosis	Low-grade: slow-growing, propensity for local spread, including carcinomatosis High-grade/uterine: aggressive, often present with carcinomatosis, but also prone to metastasis
Clinical behavior	Typically aggressive, propensity for early hematogenous spread with pulmonary metastases Behavior independent of nuclear grade, thus FIGO grading is not applied	Varies with grade and molecular subtype; initial spread is typically lymphatic

ER and PR are among the most valuable ancillary markers in this setting, as negative/very low staining is uncommon in low-grade endometrioid carcinoma and should prompt an expanded differential that includes MLA. Indeed, ensuring that MLA and other rare tumors are not missed is among the arguments for reflex ER/PR testing. GATA3, TTF1, and CD10 are most often negative in endometrioid carcinomas, although exceptions occur as highlighted in the Immunohistochemistry section.^12,27 PTEN has potential value in this context, with loss of expression providing some support for an endometroid carcinoma diagnosis though notablysome endometrioid carcinomas lack PTEN pathogenic variants and some MLAs will bear these mutations.

Serous Carcinoma

Although MLA lacks the marked nuclear atypia that is characteristic of uterine serous carcinoma, it typically exhibits moderate nuclear atypia which could provoke concern for an serous carcinoma diagnosis in this location. More often, MLA mimics serous neoplasia in the tube/ovary, where it can convincingly masquerade as a low-grade or high-grade serous carcinoma due to its cytologic features and predilection for papillary growth (Fig. 8, Table 5). The paucity of psammomatous calcifications should be a clue to consider MLA in this context, as should brisk mitoses without associated severe nuclear atypia.

FIGURE 8.

MLA mimicking serous carcinoma. This omental mass in a patient with an ovarian tumor and carcinomatosis was originally interpreted as high-grade serous carcinoma (A). Histology showed papillary structures with areas of background hyalinization (B) with areas of glomeruloid-like architecture (C). Although mitotic activity was robust, cytologic atypia was less prominent than is typical of high-grade serous carcinoma (D). Tumor board review led to additional work-up and revealed wild-type p53 staining (E), which then prompted consideration for low-grade serous carcinoma; however, the lack of psammoma bodies and completely negative ER (F) confounded this. TTF1 and GATA3 were subsequently performed to assess for mesonephric-like differentiation and showed strong diffuse (G) and moderate patchy (F) staining, respectively, supporting an MLA diagnosis.

Immunostaining can also be very valuable in this differential. Mutation-type p53 staining would generally support an endometrial serous or tubo-ovarian high-grade serous carcinoma, while MLAs are classically p53 wild-type (with rare exceptions discussed in the “Molecular” section). TTF1 and GATA3 expression are uncommon in serous carcinomas, and tubo-ovarian high-grade serous carcinoma and low-grade serous carcinomas will typically demonstrate diffuse WT1, which MLAs will lack. Moreover, complete negativity for ER/PR is fairly uncommon among both high and low-grade serous carcinomas, although it can occur in occasional high-grade serous carcinomas.

Clear Cell Carcinoma

MLAs sometimes demonstrate nuclear hobnailing and/or a dense hyalinized stroma reminiscent of some clear cell carcinomas.^2,5 This overlap is immunohistochemically complicated by the fact that these entities both lack significant ER/PR expression. Moreover, a recent study by Mirkovic et al.⁵⁵ showed that over one-third of MLAs exhibit some Napsin A positivity, underscoring the importance of a broad immunopanel (including TTF1 and GATA3) when addressing this differential diagnosis.

Carcinosarcoma

Occasional examples of MLA show prominent spindled growth, which can mimic carcinosarcoma.^2,5 In such cases, it is key to assess whether the spindled morphology qualifies as true mesenchymal differentiation consistent with a sarcomatous element. Immunostaining may be of value here, with significant cytokeratin expression within the spindled component supporting a carcinomatous contribution. This differential is further confounded by the recent description of MLAs bearing bona fide heterologous (typically chondroid but sometimes rhabdomyosarcomatous) differentiation consistent with a mesonephric-like carcinosarcoma diagnosis;¹⁶ such cases are discussed in further detail in the “Morphology” section.

Uncommon Ovarian/Adnexal Tumors

In addition to the carcinoma histotypes discussed above, extrauterine MLAs may provoke an expanded differential that includes rare and easy-to-miss entities arising in and around the ovary (Fig. 9). For example, MLAs with tubular growth may be morphologically suggestive of Sertoli or Sertoli-Leydig cell tumors. While recognition of a Leydig cell component is helpful for supporting sex cord differentiation, other ovarian tumors, including MLA, may sometimes induce adjacent steroid cell hyperplasia. Fortunately, this differential can typically be resolved with SF1 or inhibin staining, which should be positive in sex cord tumors and negative in MLA.

FIGURE 9.

Uncommon extrauterine mimickers of MLA. A variety of rare neoplasms may masquerade as MLA in the ovary and adnexa. Well to moderately differentiated Sertoli Leydig tumors show tubular architecture (A), which is reminiscent of the tubules formed by some examples of MLA (B). Sieve-like growth and compressed glands may be seen with Wolffian tumors (C) as well as MLA (D). STK11 adnexal tumor (E) can also show similar architecture and cytomorphology to MLA (F). Well-differentiated peritoneal mesotheliomas with papillary architecture (G) can mimic MLAs showing similar growth (H).

The relatively “low-grade” appearance and tubular to sieve-like growth of MLA can also mimic a Wolffian tumor [female adnexal tumor of Wolffian origin (FATWO)], particularly when the tumor is centered around the adnexa. FATWOs can also express luminal CD10,⁵⁶ prompting further confusion with MLA. However, most FATWOs are fairly mitotically inactive, whereas MLA typically exhibits brisk mitotic activity. Therefore, the identification of prominent mitoses should prompt a pause before rendering a FATWO diagnosis, and conversely, low mitotic activity should prompt reconsideration of MLA. PAX8 expression can also be useful for supporting MLA in this setting, as FATWOs are almost never positive with this marker.

The recently described STK11 adnexal tumor may also result in confusion with MLA.^57,58 Immunohistochemistry can be very helpful in addressing this diagnosis, as STK11 adnexal tumors will typically display some degree of sex cord marker expression and are usually hormone receptor and WT1-positive. As with Wolffian tumors, PAX8 is typically negative in STK11 adnexal tumors, and can be another helpful discriminatory biomarker. While occasional GATA3 staining is reported in STK11 adnexal tumors, TTF1 is consistently negative in this entity.⁵⁷ Molecular testing will also be of value in this distinction.

Finally, the clinical presentation of peritoneal malignant mesothelioma may sometimes suggest an ovarian primary, and these tumors can have considerable histologic similarities with MLA. In particular, the papillary and tubular growth that characterizes some peritoneal malignant mesotheliomas can closely resemble MLA. Although its utilization is not recommended by the MLA Consortium, one could imagine how calretinin immunostaining might cause further confusion at this diagnostic interface since it may be positive in both tumors, although it is typically more widespread in malignant mesothelioma. WT1, conversely, can be of value in this setting as mesothelioma should be WT1-positive, whereas MLA is consistently negative. BAP1 staining is lost in some peritoneal malignant mesotheliomas but would be expected to be retained in MLA. As discussed, some MLAs are focally positive with hormone receptors, while malignant mesotheliomas are negative.

CLINICAL IMPLICATIONS AND FUTURE DIRECTIONS

Endometrial MLA

Clinically, the identification of MLA as a histologic subtype of endometrial cancer has allowed for further prognostication of these high-risk tumors from the general no specific molecular profile (NSMP) molecular cohort. Uterine MLA accounts for only 0.7% to 3% of endometrial cancer cases in unselected patient populations; however, retrospective series show that patients are more likely to present with stage III/IV disease and more likely to have metastatic disease to regional lymph nodes.^{2,5,6,8,10,12,23} Importantly, in a retrospective cohort study comparing uterine MLA to endometrioid and serous cases, 71% of uterine MLA cases had a documented recurrence compared with 7.8% of endometrioid and 29% of serous cases, with a propensity for distant metastatic disease to the lungs.² When considering adjuvant therapy options and known prognostic factors for uterine cancer, MLA presents with higher stage disease, is twice as likely to exhibit lymphovascular space invasion, and should be considered high-risk disease for clinical management. The consideration of uterine MLA as a high-risk disease highlights the critical clinical implications that hinge on the proper pathologic classification of these tumors, where adjuvant chemotherapy and/or radiation are the standard of care. In review of cases sent for expert consultation, ∼80% of MLA cases were originally diagnosed as a histology other than MLA, with the majority being classified as FIGO grade 1-2 endometrioid carcinoma, which may be considered low-risk or intermediate-risk disease based on other pathologic factors, where adjuvant chemotherapy is not standard of care.² In addition, post-treatment surveillance of uterine cancer does not routinely include imaging. However, given the increased risk for pulmonary metastases, specifically in MLA, the proper classification of these tumors allows for consideration of periodic chest imaging in surveillance of these patients.

Extrauterine MLA

Extrauterine MLA has been estimated to account for 0.6% to 0.9% of epithelial ovarian cancer. Experience is limited by the fact that the total number of reported clinical cases in the literature is <100, and clinical management is currently based on the standard of care for epithelial ovarian cancer with platinum-based chemotherapy. However, the proper classification of MLA again is key in the decision for adjuvant therapy, as MLA is diagnosed with stage I/II disease in 64% of patients, where adjuvant therapy may be de-escalated for other histologic subtypes such as endometrioid tumors of the ovary.⁵ Surveillance of extrauterine MLA follows the current standard of care for epithelial ovarian cancer.

Future Directions

While the identification of MLA has allowed for improved prognostication, many questions remain in the clinical management of these tumors. In consideration of uterine MLA, the impact of chemotherapy and/or radiation on the risk of recurrence is largely unknown, as the largest cohort studies to date do not include adjuvant treatment data. In addition, the impact of radiation in the treatment of high-risk, advanced-stage uterine cancer is debated among gynecologic oncology and radiation oncology experts, with practice variation across centers. The MLA Consortium is actively accruing a tumor registry in an effort to further evaluate response to therapies in a larger patient cohort. However, the prospective evaluation of chemotherapy and radiation is needed in this rare patient population. Similarly, the impact of platinum-based chemotherapy on the risk of recurrence in uterine MLA is largely unknown.

The molecular underpinnings of MLA do allow for consideration of exciting novel therapeutics, including small molecule KRAS or NRAS inhibitors, other targeted therapies capitalizing on the Ras/Raf/MEK pathway, vaccine therapies, and cellular therapies, with several early phase studies open to accrual for patients. The first is a single-arm phase II study of avutometinib (VS-6766) and defactinib in advanced or recurrent mesonephric gynecologic cancer. This study capitalizes on the success of the study regimen in low-grade serous ovarian cancer (LGSOC), with recent FDA approval for the regimen in patients with KRAS-mutated LGSOC (NCT05787561). In addition, patients may be eligible for ongoing early-phase basket trials of PAN-KRAS or PAN-RAS inhibitors (NCT04111458, NCT05379985). Lastly, cellular therapy is an interesting treatment strategy with an actively accruing phase I/II trial of TROP2-CAR/IL-15-transduced CB-NK cells delivered intraperitoneally in tumors with at least 1+ TOP2 expression (NCT05922930) in addition to T-cell therapy targeting mutant KRAS proteins specifically (NCT06043713). All these drugs remain investigational, and there are no current Food & Drug Administration approvals for their use in this setting.

CONCLUSIONS

MLA is an uncommon, underrecognized, and potentially overdiagnosed malignancy in the endometrium and extrauterine sites. Its proper identification is critical due to its aggressive behavior and the future potential for unique treatment avenues. The MLA Consortium has been convened to better understand and approach these challenging tumors, and herein provides practical diagnostic guidance for pathologists who are likely to encounter this entity. In particular, we emphasize situations in which expert consultation is desirable, including cases with challenging morphology (such as mixed tumors, including cases with dedifferentiated and carcinosarcomtaous components), unexpected immunohistochemical patterns (such as cases with TTF1 and GATA3 negativity or more than limited hormone receptor staining) and nonclassical molecular profiles (such as cases with TP53 abnormalities or MMR deficiency). Pathologists interpreting gynecologic samples as well as specimens from potential metastatic sites—including lung—should be familiar with MLA and the criteria for its diagnosis.