Endometrioid Intraepithelial Neoplasia in the Secretory Phase: Morphologic and Biomarker Diagnostic Features in a Common, Overlooked Setting.

After approval from the UT Southwestern IRB for a HIPAA-waived retrospective study, text searches identified cases accessioned between 2015 and 2025 at the UT Southwestern William P. Clements Jr. University Hospital and related clinics, and the Parkland Memorial Health and Hospital System, both in Dallas, TX. All cases were internally accessioned; no outside consultation cases were included. Exclusion criteria were (1) a prior history of endometrial or ovarian cancer and (2) intrauterine device use, hormonal therapy including oral contraceptives, or pregnancy at any time in the 1-year interval before the index endometrial sample. A diagnosis of cancer or precancer in the cervix or a nongynecologic organ system was not an exclusion criterion. The following terms were used in text searches: “endometrioid intraepithelial neoplasia,” “atypical,” “hyperplasia,” and “secretory.” The original H&E slides were reviewed to confirm (1) definitive background secretory or ovulatory menstrual endometrium; (2) absence of overt changes suggesting hormone therapy; and (3) histologic features consistent with EIN. Patient electronic records for all cases were systematically reviewed for any prior and follow-up endometrial tissue samples (biopsies and hysterectomies within the above health care systems) relative to the index case and to confirm that the exclusion criteria were met. PAX2, PTEN, and β-catenin immunohistochemistry (IHC) was performed if not done at the time of diagnosis. Ki-67 immunostaining was performed on 17 randomly selected cases. Limited data, including LMP and age, were recorded. For PAX2, PTEN, β-catenin, and Ki-67, staining protocols previously validated for clinical testing were performed on 4 μm sections in a clinical immunohistochemistry laboratory using a DAKO Autostainer Link 48 instrument. The following primary antibodies were used: β-catenin (clone β-catenin-1, ref. IR702; Agilent), PAX2 (clone EP235, ref. BSB2567; Cancer Diagnostics, Durham, NC), PTEN (clone 6H2.1, ref. PM278AA; BioCare Medical, Pacheco, CA), and Ki-67 (clone MIB-1, ref. IR626; Agilent). Antigen retrieval was performed at low pH (6.0) for β-catenin and Ki-67, and at high pH (9.0) in a Tris/EDTA solution (Agilent) for PAX2 and PTEN, at 97 °C for 20 minutes. FLEX peroxidase block was performed for 10 minutes for β-catenin and 5 minutes for other markers. The primary antibody incubation time was 20 minutes for β-catenin, 40 minutes for PAX2 and PTEN, and 30 minutes for Ki-67. Incubation with Mouse Linker (Agilent) for β-catenin was performed for 15 minutes, and incubation with Rabbit Linker (Agilent) for PAX2 was performed for 10 minutes. Secondary antibody (Envision/HRP) incubation time was 20 minutes for PTEN and β-catenin, and 30 minutes for PAX2 and Ki-67. For all antibodies, the enzymatic conversion of the 3,3′-diaminobenzidine tetrahydrochloride chromogen was performed for 10 minutes at room temperature. Noyes criteria 19 were applied to date secretory endometrium as early (day 17 to 18), mid (day 19 to 22), late (day 23 to 28), or ovulatory menstrual. Mitotic activity is abnormal after day 17, and any mitotic figures beyond this point were scored as positive. Because some day 16 to 17 secretory endometria may show residual mitotic activity, cases were scored as positive when mitoses were clearly increased in EIN relative to the background endometrium. Aberrancy for the 3 EIN markers was assessed per prior studies. Aberrancy for a single marker is considered supportive of the diagnosis of EIN, since each of the markers is aberrant in different subsets of EIN. 20 – 23 β-catenin aberrancy is defined by inappropriate nuclear accumulation rather than its normal membranous and cytoplasmic localization, typically accompanied by strongly increased overall expression. 24 – 26 Strong nuclear β-catenin is scored as aberrant, even if focal. Evaluation is restricted to glandular epithelium and excludes obvious morules, which consistently demonstrate nuclear β-catenin and therefore are not diagnostically informative. Low-level nuclear staining may be seen in non-neoplastic endometrium; nuclear staining is considered aberrant only when it is clearly more intense than staining along the lateral cell membranes. PAX2 is a nuclear transcription factor whose expression within the uterus is restricted to the endometrial epithelium. For diagnostic purposes, PAX2 is considered aberrant only when there is a uniform absence of nuclear staining across the entire glandular profiles in cross-section; partial or reduced staining is not interpreted as aberrant. In most cases, admixed or adjacent non-neoplastic endometrial glands or surface epithelium provide an internal positive control. PTEN is broadly expressed in the endometrium, with nuclear and cytoplasmic staining in glandular epithelium, stroma, and resident leukocytes. Genuine loss of PTEN within the glandular epithelium produces a characteristic “punched-out” appearance, in which PTEN-deficient glands contrast sharply with the surrounding PTEN-positive stroma. Intraepithelial leukocytes—sometimes numerous—retain PTEN expression even in cases showing true epithelial loss. Because PTEN expression may be diminished during the secretory phase, only complete absence of glandular staining is scored as aberrant (Fig. 7 C, upper vs. lower part). Focal loss of PAX2 and PTEN in individual glands or small clusters can occur in benign endometrium, comprising up to 5% of the glands but usually much less. 15 , 21 , 22 , 27 In EIN in the secretory phase, complete loss within architecturally and cytologically distinct areas of concern, where ≥25% of glands were null, was the threshold for definitive aberrancy. The extent of loss sometimes exceeded what was suggested by histologic features alone, consistent with well-established observations that progestins partially suppress histologic features of EIN. 10 Pairwise agreement among the 3 immunostaining results (β-catenin, PTEN, and PAX2) and the presence of morules on H&E was assessed using the Cohen κ statistic. For each comparison, P- values testing the null hypothesis of no agreement beyond chance (κ=0) were reported. Analyses were conducted in R Statistical Software (version 4.4.2; R Core Team, 2024).

Cases were identified by text-based searches and subsequently confirmed by medical record review to ensure that inclusion and exclusion criteria were met, as detailed in the methods section. A clinical summary is provided in Table 1 . All patients were of reproductive age, with a mean age of 41±7 years, which was lower than the previously reported average age at EIN diagnosis of 54±7 years. 28 Mean body mass index (BMI) was 34.1 kg/m 2 , greater than the US mean of 29.9 kg/m 2 for US adult females in the 20- to 49-year-old age strata, approximating the study population. 29 These features of the study population are consistent with the inclusion criteria and prior data establishing a highly significant association between higher BMI and earlier age at EIN diagnosis. 28 Self-reported dates for last menstrual period (LMP) were on average 29.7 days prior (Table 1 ), consistent with most specimens being obtained in the secretory phase, with 14/34 (41.2%) patients reporting LMPs >28 days prior, suggesting ovulatory dysfunction or perimenopausal state. 30 The most frequent clinical diagnosis was AUB (not otherwise subclassified), with ovulatory dysfunction (AUB-O) as one of the attributed causes in 10/40 patients (25%) (see Table 1 legend for additional details). All H&E slides were reviewed for each case. The background secretory endometrium was classified as early, middle, late, or ovulatory menstrual. Clinical Summary Clinical diagnoses at the time of the index specimen are shown per the PALM-COEIN classification for abnormal uterine bleeding (AUB), where L, leiomyoma; O, ovulatory dysfunction; P, polyp; A, adenomyosis; and M, malignancy or hyperplasia. AUB with multiple attributed causes is shown with letter combinations. NOS, not otherwise specified (no specific AUB cause attributed clinically). The 2 AUB-M patients had a diagnosis of EIN on a prior biopsy, with subsequent hysterectomies (the index cases) incidentally performed during the secretory phase. These 2, plus the case of uterine prolapse, were the only index case hysterectomy specimens. The patient with an ovarian mass also had elevated serum CA125 and underwent endometrial sampling during clinical workup. Subsequent salpingo-oophorectomy revealed a benign ovarian serous cyst and ovarian torsion. One patient was an age outlier (54 y), with the next 2 oldest patients being 50 years of age. Clinically, this patient was considered to have irregular perimenopausal menses. Total patients, n=40. Age: 22 to 54 years, mean: 41±7 (n=40). BMI: 21.9 to 45.0 kg/m 2 , mean: 34.1±7.1 (n=40). LMP at the time of biopsy: −76 to 0 days, mean: −29.7 (n=34). On the basis of an initial screening and identification of potential diagnostic/distinguishing features, the following histologic features were assessed: distinctive architecture and cytology relative to the secretory background, cytoplasmic features including eosinophilia and extent of secretory differentiation relative to background, presence of morules, nuclear atypia, epithelial stratification, mitotic activity, and apoptotic bodies. All cases in this cohort exhibited distinctive architecture and cytology, distinguishing areas of EIN from the secretory background (Table 2 ). These overall changes included altered gland density/crowding; distinctive gland architecture, including the shapes and contours of glands; abnormalities in epithelial layer organization, including increased stratification (≥3 layers) and cribriforming (usually subtle); and altered, distinctive cytoplasm. The latter 2 histologic features are notable as these are not characteristic of normal secretory endometrium. Distinguishing Histologic Features of EIN in Secretory Phase (n=40) The last 4 rows represent “soft” features that may be more variable or difficult to reliably assess. The most unexpected and clinically relevant finding in this systematic study is that the majority of EIN in the secretory phase (83%) showed diminished or absent secretory changes relative to the background endometrium; a representative example is shown in Figures 1 A–C. In this case, the background endometrium corresponded to day 17 and showed distinct subnuclear vacuoles, whereas the EIN, in addition to having highly abnormal, variable, and branching gland contours, showed essentially absent secretory features, with no cytoplasmic vacuolization. This observation is diagnostically important because the term “secretory EIN” implies that secretory differentiation is a characteristic or even defining feature of EIN in the secretory phase. However, some EIN in the secretory phase can show striking secretory features (Figs. 1 D–F). A subset of EIN exhibited distinctive cytoplasmic features characterized by eosinophilic cytoplasm, typically with decreased vacuolization, and in some cases, associated with eosinophilic intraluminal secretions (Fig. 2 ). Whether these findings represent diminished and/or altered secretory differentiation remains uncertain. Secretory differentiation is paradoxically diminished or absent in most EIN in the secretory phase. A–C, Case #1: A, Background early secretory-phase endometrium. B, EIN is evident as a large architecturally and cytologically distinct area of glands with abnormal architecture, minimal to no atypia, and loss of secretory differentiation relative to adjacent normal early secretory glands. C, Higher magnification of the inset in (B) showing the absence of cytoplasmic secretory vacuoles and intraluminal secretions. The EIN consists of glands with abnormal, variable contours and subtle cribriforming. Mitoses are absent, but scattered apoptotic bodies are present. D–F, Case #40, counterexample with prominent secretory features: D, Background early secretory-phase endometrium. E, EIN is evident as architecturally and cytologically distinct areas of glands with prominent papillary infoldings. F, Higher magnification showing 2 neoplastic glands, both with secretory vacuoles and nuclear atypia. The lower gland shows papillary infoldings and focal mucinous metaplasia. Arrowheads indicate mitotic figures. Distinctive cytologic features in a subset of EIN in the secretory phase include eosinophilic cytoplasm (case #8). A, Large area of normal midsecretory background endometrium showing linear arrangements of glands generally perpendicular to the surface and with regularly serrated contours. Area to the right of the dashed line is a higher magnification showing a linear “corkscrew” arrangement of 1 normal secretory gland in multiple cross-sections in the plane of section. B, Large focus of architecturally and cytologically distinct EIN relative to background secretory endometrium (EIN demarcated by black lines). C, Separate field showing a large focus of architecturally and cytologically distinct glands corresponding to EIN (demarcated by black lines) set within secretory endometrium. D, Higher magnification of the inset in (C) showing striking gland crowding, distinctive eosinophilic cytoplasm, and intraluminal secretions with the absence of normal secretory vacuoles and differentiation. Atypia is minimal to absent. Morules—strongly associated with CTNNB1 mutations and aberrant β-catenin localization in EIN and endometrial cancers 10 , 24 – 26 , 31 – 35 —are a valuable and reliable diagnostic clue for EIN in the secretory phase because they were present in 38% of cases (Table 2 ) and can be identified with relative ease. We did not identify plasma cells in these cases; thus, morules in secretory endometrium are unrelated to acute or chronic endometritis, consistent with a clonal (neoplastic) origin associated with CTNNB1 mutation. Accordingly, areas of EIN harboring morules were geographically separate and distinct from background secretory endometrium (see 4 examples in Fig. 3 ). However, the EIN glands showed aberrant and branching contours, merging and cribriforming with each other and the associated morules, making them architecturally and cytologically distinct. In all EIN with morules, the glandular component showed decreased or absent secretory differentiation. In summary, morules are a common feature of EIN in the secretory phase (38% of cases, Table 2 ), and their presence is a “red flag,” necessitating further scrutiny and consideration of EIN. Morules are a reliable diagnostic feature of EIN in the secretory phase. Each label represents a different case. A, Case #2: the left half shows background late secretory endometrium. On the right half is a large focus of EIN defined by a large architecturally and cytologically distinctive area of endometrial glands intimately associated with numerous morules (†). B, Case #6: architecturally abnormal and distinct glands in intimate association with numerous morules. Background endometrium was late secretory (not shown). C, Case #18: midsecretory endometrium on the left. On the right is a separate and distinctive large area of EIN characterized by numerous morules (†). D, Case #31: ovulatory menstrual background endometrium (not shown) with stromal condensation in areas of EIN. EIN was evident in multiple fields due to the presence of architecturally and cytologically distinct glands with abnormal shapes and contours, and subtle cribriforming merging with large and multiple morules. The neoplastic glands in EIN harboring morules show diminished to absent secretory differentiation, as illustrated by these 4 cases. Other histologic features systematically assessed on H&E slides included nuclear atypia, epithelial stratification, mitotic activity, and intraepithelial apoptotic bodies (Table 2 ). We consider these as “soft” features because they varied within zones of EIN, were often subtle or focal, and were absent in many cases. Progestin treatment of EIN reverses nuclear atypia and enlargement 10 , 11 , 36 ; concordantly, nuclear atypia was neither prominent nor common in secretory phase EIN, with moderate to severe nuclear atypia identified in 2 of 40 cases (5%) (Fig. 4 A). Nuclear stratification, defined as ≥3 nuclear layers, was a feature of a subset of cases (Fig. 4 B), as secretory-phase endometrial glands normally exhibit a well-defined epithelial monolayer even when glands become serrated in the late phase. Mitotic figures, which are physiological in interval (day 16) endometrium and can persist through days 16 to 17, were identified in 28% of cases, but were typically sparse and required deliberate searching (Fig. 4 B). Intraepithelial apoptotic bodies were also variable but could be conspicuous when present, making them a potentially helpful diagnostic clue (Fig. 4 C); prior studies implicated apoptosis as an early consequence of progestin therapy in EIN. 37 For these soft features, comparison with background endometrium is recommended, given histologic variation across the secretory phase. “Soft” histologic features of EIN in the secretory phase. A, Case #7 with nuclear atypia characterized by vesicular nuclei with prominent nucleoli (inset, higher magnification of boxed area). This case had the most severe nuclear atypia among all the cases in the study. Significant atypia was absent in the large majority of cases; thus, it is not a characteristic or reliable diagnostic feature. B, Case #9 showing stratification of ≥3 cell layers and mitotic activity (arrowhead). Background midsecretory-phase endometrium was devoid of mitotic activity. C, Case #6 (also shown in Fig. 3 B) with numerous intraepithelial apoptotic bodies. Next, we assessed aberrancy for established EIN markers. β-catenin had the highest incidence of aberrancy (51%), followed by PTEN (46%) and PAX2 (36%). In a prior study of n=71 benign secretory endometria, loss of PAX2 or PTEN was observed only in rare scattered glands (below thresholds for aberrancy), and no sample met criteria for aberrancy for any of the 3 markers. 15 The difference in marker aberrancy between secretory-phase EIN (38/40) vs. previously reported non-neoplastic controls (0/71) is highly significant (2-tailed Fisher exact test, P = 1.03 × 10 –27 ). As previously noted for EIN, not all β-catenin aberrant cases harbored morules, demonstrating that β-catenin has diagnostic value in cases without morules. 21 Taken together, 38/40 (95%) of EIN in the secretory phase were aberrant for at least 1 immunohistochemical (IHC) marker. Both cases that were nonaberrant for the 3 markers did not harbor morules (Fig. 5 A). Most EIN in secretory phase were aberrant for only 1 IHC marker but 38% were aberrant for 2 or 3 (Fig. 5 B). Analysis of all pairwise associations showed only 1 significant concordance, that between morules and β-catenin ( P <0.00001; κ=0.66) and only 1 discordance, between PTEN and β-catenin ( P =0.01; κ=-0.39) (Fig. 5 C). Although these quantitative results are distinct from unselected EIN (most strikingly for PAX2, see Discussion), they demonstrate that all 3 EIN markers plus morules are together especially useful in the secretory phase given the often subtle histologic features of EIN in this context. A case matrix is provided in Figure 5 D. Immunophenotypic features of EIN in the secretory phase. A, Incidence of aberrancy for 3 EIN biomarkers and morules in the study set. The 2/40 cases nonaberrant for all 3 markers also did not harbor morules. B, Pie chart showing distribution of cases with aberrancy for 0, 1, 2, or 3 markers. C, Heat map of all pairwise Cohen κ correlations. κ and P- values shown only for associations statistically significant at P <0.05. D, Case matrix of all 40 cases. Colored rectangle indicates presence of features (IHC marker aberrancy, decreased secretory features, etc.); see key for dating of secretory endometrium. Illustrative examples for each of the markers are presented in turn (see Methods for aberrancy scoring criteria). For PAX2, aberrancy was evident as widespread loss across areas of histologic distinctiveness, with surface epithelium or entrapped/adjacent glands serving as an internal positive control (see 3 examples in Fig. 6 ). For PTEN, aberrancy was evident as complete loss of nuclear and cytoplasmic PTEN in endometrial glands, with retention of positivity in surrounding stroma and scattered intraepithelial leukocytes (the latter not readily evident at the magnifications shown) (see 3 examples in Fig. 7 ). In cases where glands did not show decreased secretory features (as in Fig. 7 A), entrapped normal glands (indicated by asterisks) showed decreased secretory differentiation, pointing to abnormalities in the micro-environment. In the normal epithelium of secretory endometria, especially in the late phase, PTEN expression can be decreased. However, the epithelium retains some cytoplasmic and nuclear expression when contrasted with the striking “punched-out” appearance of PTEN-null EIN glands (Fig. 7 C). For β-catenin, aberrancy was evident as distinct regions of overexpression and nuclear localization relative to the more delicate membranous and cytoplasmic expression in adjacent non-neoplastic epithelium (see 4 examples in Fig. 8 , including 1 case without morules in label C). PAX2 aberrancy in secretory phase EIN. Three cases are shown with paired H&E and IHC images. A, Case #29: large multiple fragments of EIN with densely packed glands around a central fragment of background late secretory endometrium. PAX2 is uniformly lost in the multiple fragments of EIN, with the normal fragment serving as an internal positive control for PAX2. B, Case #20: architecturally abnormal glands with complex branching patterns characterized this EIN, which also showed extensive diffuse PAX2 loss. PAX2 expression (shown at lower magnification to better show the extent of loss) is retained in the surface epithelium and a few scattered normal background glands. The case harbored morules and was also β-catenin aberrant; recurrence occurred 5 years later with persistent aberrancy of PAX2 and β-catenin. C, Case #28: architecturally and cytologically distinct focus of EIN with a small cluster of background normal secretory glands in the lower right corner. The EIN was PAX2-null (aberrant) with the cluster of normal glands serving as an internal positive control. Other fragments in the specimen outside this field harbored similar foci of EIN that were also PAX2-null. Note the striking loss of secretory features evident even at low magnification. PTEN aberrancy in secretory-phase EIN. Three cases are shown. A, Case #11: this was a subtle EIN with a minor degree of gland crowding, but the diagnosis was supported by the architectural and cytologic distinctiveness of the foci, which exhibited abnormal eosinophilic cytoplasm. PTEN showed a complete “punched-out” loss across very large areas (only 1 small area is shown here). Entrapped normal glands (*) in secretory-phase EIN (in this case, confirmed by their retained PTEN expression) can show minimal secretory differentiation. B, Case #13: EIN in the bottom half of the specimen is PTEN-null (aberrant). C, Case #16: the upper part shows PTEN immunostaining of background serrated late secretory endometrial glands with weak but definite normal/nonaberrant PTEN nuclear and cytoplasmic expression, whereas EIN in the lower part shows definitive loss with “punched-out” appearance (aberrant) and no epithelial signal other than scattered leukocytes. Strong PTEN expression is observed across the endometrial stromal compartment in all cases. β-catenin aberrancy in secretory-phase EIN. A, Case #2 (also shown in Fig. 3 A) with β-catenin overexpression relative to background normal epithelium on the left side of the image. More overt gland crowding was present in other fields. B, Case #18 (also shown in Fig. 3 C). β-catenin clearly delineates a distinctive area of EIN with morules. C, Case #32: no morules were present. Even though morules and β-catenin aberrancy are strongly associated, β-catenin is sometimes aberrant in cases without morules, and thus is a potentially useful marker even in their absence. The (*) indicate background normal glands with membranous localization. D, Case #17: EIN characterized by strong overexpression of β-catenin with scattered nuclear localization; morules are focally present (top middle). The inset indicates normal background endometrium with delicate membranous expression of β-catenin, serving as an internal control. In the lateral part, areas of EIN with abnormal gland contours but significantly less crowding also exhibit aberrant β-catenin in incipient morules and scattered isolated cells. Finally, we assessed Ki-67 labeling in a subset of 17 cases that, after being randomly selected, were ascertained to represent diverse features of the study cases (morules, mitotic activity, different stages of the secretory phase, etc.). In most cases, Ki-67 indices were very low in the EIN epithelium, consistent with potent endogenous progestin-mediated suppression of epithelial proliferation. Labeling indices were higher in the stroma, but positive cells could be found in non-neoplastic epithelium, features that limit the utility of Ki-67 as a diagnostic tool. In cases with mitotic activity, elevated Ki-67 positivity could be evident in some neoplastic glands, but this was highly variable and did not clearly delineate the EIN (Supplemental Figs. 1A–D, Supplemental Digital Content 1, http://links.lww.com/PAS/C277 ). We conclude that Ki-67 interpretation is challenging and does not add significant diagnostic value relative to the detection of mitoses by H&E and the application of the 3 IHC markers. Patients were initially treated by hysterectomy or conservatively with oral progestins or a progestin-secreting intrauterine device (IUD). Although follow-up was limited and not all patients had subsequent endometrial sampling (ie, conservative treatment was initiated but no subsequent specimen had yet been obtained), 26/40 patients (65%) had confirmation of EIN or carcinoma in a different specimen than the index specimen in the secretory phase. For 24 of these 26 patients, the diagnosis was EIN. In 1 of the other 2 patients, who was treated by hysterectomy, the diagnosis was EIN bordering on FIGO 1 adenocarcinoma. The other patient was started on oral megestrol acetate, had 2 negative follow-up biopsies, EIN in a biopsy 1 year later, and then FIGO 1 adenocarcinoma in a curettage 2 years later. In another 3/40 patients, the most significant findings in another specimen were suspicious/atypical but indeterminate. In 1 of these 3 patients, the index biopsy showed EIN with morules, and a curettage 2 months later revealed isolated morules but no EIN. Oral megestrol acetate was initiated, followed by negative biopsies over the next 5 years; current treatment continues with a levonorgestrel IUD. In 1 patient, treatment with a levonorgestrel IUD was initiated, and curettage 3 months after the index specimen showed crowded glands, a nest of cells suggestive of a morule, and focally aberrant β-catenin, with the findings falling short of definitive EIN. The patient had adenomyosis, leiomyomata, and bothersome abnormal uterine bleeding despite the IUD, and a hysterectomy 6 months after the original diagnosis showed decidualized endometrium negative for EIN. In the last of these 3 patients, a curettage performed at the time of IUD insertion 1 month later showed architectural disorder concerning for EIN, while a biopsy 7 months after the index case showed decidualized endometrium and no EIN. These observations confirm that the diagnosis of EIN in the secretory phase can be made with confidence in selected cases and that the finding has clinical significance. The data further indicate that, as a group, these lesions represent bona fide EIN. It is also reassuring that these lesions are not, in aggregate, well-differentiated adenocarcinomas with diagnostic features masked by endogenous progestins.

This study, to our knowledge, is the first to systematically characterize the histologic and immunophenotypic features of EIN in the secretory phase. A key methodological strength was the rigorous exclusion of patients whose background endometrium was not in the secretory phase and of those receiving exogenous hormone therapy that could induce treatment effects, suppress EIN, or otherwise confound interpretation of secretory-phase histology. A paradoxical but central finding is that most EIN in the secretory phase do not exhibit overt secretory differentiation, a feature that contributes to their histologic distinctiveness. Diminished secretory differentiation may be rationalized by EIN’s status as a clonal neoplasm exhibiting abnormal cellular differentiation. 32 , 38 , 39 Yet other secretory-phase EINs exhibit distinctive eosinophilic cytoplasm without the usual secretory features, a feature shared by some progestin-treated EINs. 10 The findings argue against the use of the term “secretory EIN,” which conflates biologically disparate entities and implies that bona fide EINs in the secretory phase should always exhibit prominent secretory changes. We therefore recommend the use of the term “EIN in the secretory phase” to designate this distinct and likely underrecognized clinicopathologic entity. Additional contributions of this study include the identification of eosinophilic cytoplasm and other distinguishing histomorphologic features, including, in a minority of cases, increased stratification, apoptotic bodies, or mitotic figures. Our findings extend and clarify prior studies of so-called secretory EIN, which described loss of secretory features or “eosinophilic metaplasia” in some cases. 10 , 11 As noted above, the normal physiology of the secretory phase endometrium, including variable and often markedly elevated gland-stromal ratios well above 50% preclude application of the gland area criterion in this physiological context. In addition, nuclear features are suppressed, limiting the utility of assessing “atypia.” Nonetheless, a central definitional feature of EIN—cytologic and architectural distinctiveness—remains applicable and permits identification of EIN (EAH/EIN) in the secretory phase, with the 3 biomarkers providing important supportive evidence. There are other well-known clinicopathologic contexts that require assessment of EIN without strict adherence to the area criterion. In EIN or endometrial adenocarcinomas with extensive morules or squamo-morular differentiation, lesional glands can be scant and largely replaced by morules. 40 In progestin-treated EIN, lesions falling far short of the area criterion can be reliably diagnosed as EIN. 10 , 22 , 23 , 36 , 41 The overall aberrancy rate for at least 1 marker in secretory-phase EIN of ~90% is comparable to rates reported in prior studies of EIN overall. 21 , 42 Interestingly, however, aberrancy rates for individual markers are quite different. The most striking difference is that the PAX2 aberrancy rate in secretory-phase EIN (this study) is 36%, much lower than the PAX2 aberrancy rate in overall EIN of 81% ( P <10 −6 per 2-tailed Fisher exact test), 21 which tracks closely with the aberrancy rate of ~80% in endometrioid adenocarcinomas. 27 , 39 It is possible that secretory-phase EIN with PAX2 gene silencing alone (a critical step in the initiation of most endometrial cancers) 39 are especially sensitive to progestin suppression, making them more difficult to detect or diagnose in the luteal phase. Consistent with this possibility, β-catenin alterations have been implicated in initial and acquired resistance to progestin therapy. Progesterone exerts antiproliferative effects through suppression of Wnt/β-catenin signaling, and experimental activation of this pathway attenuates progesterone responsiveness. 43 In progestin-treated cohorts, altered expression of Wnt-pathway genes and enrichment of CTNNB1 -mutant tumors have been observed among nonresponders, suggesting pathway-level association with treatment resistance. 44 , 45 Responders have higher baseline rates of PAX2 aberrancy, while nonresponders have higher rates of PTEN and β-catenin aberrancy by IHC. 46 However, further work is needed to elucidate the biological basis of progestin therapy responses in EIN and endometrial cancer. 47 Assessment of proliferative activity by Ki-67 in these early neoplastic lesions is conceptually appealing. This and previous studies reporting elevated labeling indices by quantitative methods prompted us to reevaluate the issue together with our other findings. 16 – 18 However, we found that several factors greatly limited the utility of Ki-67 in practice: (1) much higher labeling indices in stroma than epithelium; (2) difficulty in distinguishing epithelial versus stromal cell identity for some of the Ki-67 + cells; and (3) very low or less commonly, highly variable labeling indices in the EIN epithelium (Supplemental Figs. 1A–D, Supplemental Digital Content 1, http://links.lww.com/PAS/C277 ). Also, outlier dyssynchronous glands, which are seen in some secretory endometria and are not believed to be neoplastic, can have strikingly elevated Ki-67 indices. 48 Thus, while mitotic activity is of some value and should be assessed on H&Es (and compared with the background), we conclude that Ki-67 is of low diagnostic value. Interobserver variability in the diagnosis of EAH/EIN is well recognized and reflects the inherent challenges of applying architectural and cytologic criteria in endometrial biopsies. Prior studies have demonstrated only moderate agreement among pathologists in borderline or subdiagnostic lesions. The present study was not designed to formally evaluate diagnostic reproducibility; however, the morphologic and immunophenotypic features described here provide additional context for the evaluation of these challenging lesions. 49 – 51 Several benign histologic patterns encountered in the secretory phase may elicit concerns for EIN. Dysynchronous glands (also termed mixed secretory and proliferative endometrium) can give the impression of a clonal process. Although their etiology is not well understood, there is no compelling evidence that they represent a neoplastic process. Consistent with this, PAX2 and PTEN expression are retained, 48 and such glands do not meet criteria for aberrancy for any of the 3 EIN markers 22 , 23 (Supplemental Figs. 2A–D, Supplemental Digital Content 2, http://links.lww.com/PAS/C278 ). Basalis is typically not represented in routine endometrial biopsies but is visualized in hysterectomy specimens and may be encountered in hysteroscopic morcellations. The “cytologic distinctiveness” of basalis glands—reflecting their intrinsically diminished progesterone responsiveness—may invite consideration of EIN, but their intimate association with myometrium excludes this diagnosis. Gland area increases over the secretory phase and may reach 80% to 90% in the late phase, exceeding 90% in some areas. While such crowding is often diffuse in the late phase, it may occasionally appear focal. The presence of a conventional serrated glandular pattern and the absence of true architectural or cytologic distinctiveness should mitigate concern for EIN. Additional factors, such as submucosal leiomyomata, can be associated with gland crowding and mild architectural distortion in overlying secretory endometrium; however, these changes lack other defining features of EIN in the secretory phase. Finally, artifactual gland crowding related to stromal collapse in menstrual endometrium can raise concern for EIN or even carcinoma, but these alterations are typically diffuse and unlikely to cause diagnostic confusion when menstrual endometrium is recognized (Supplemental Figs. 3A–E, Supplemental Digital Content 3, http://links.lww.com/PAS/C279 ). Ovulation superimposed on prior anovulation (oligo-ovulatory state) can produce a pattern resembling disordered proliferative endometrium but in the secretory phase, with cystically dilated glands across the specimen (Supplemental Fig. 3F, Supplemental Digital Content 3, http://links.lww.com/PAS/C279 ). Cystically dilated glands were not a prominent feature in most EIN in the secretory phase. Nevertheless, such architectural disorder falls in a histologic spectrum with EIN, and specimens with suspicious but indeterminate findings will be encountered. In these situations, diagnostic terminology such as “suspicious for EIN,” or “secretory endometrium with architectural disorder” may be used, accompanied by a comment that endogenous progestins during the luteal phase can obscure diagnostic features of EIN and that continued follow-up or repeat sampling is recommended. In some cases, aberrancy for 1 or more of the 3 EIN markers can support or establish the diagnosis; however, IHC results may themselves be ambiguous or indeterminate because of small lesion size, equivocal expression patterns, or technical factors. The application of validated scoring criteria is therefore essential, and the use of validated monoclonal antibodies is advised. When uncertainty persists, any immunostain should be reported as indeterminate to avoid overinterpretation of equivocal findings. This study delineates the characteristics of EIN in the secretory phase. Recognition of this entity, with selective use of established biomarkers, can improve diagnostic confidence in challenging cases. A diagnostic approach flowchart summarizing these findings is provided in Supplemental Figure 4, Supplemental Digital Content 4, http://links.lww.com/PAS/C280 . The findings are directly applicable to routine surgical pathology practice.

In summary, architectural and cytologic distinctiveness are the key histologic features permitting identification of EIN in the secretory phase. Notably, the majority of cases show diminished or absent secretory differentiation, an important feature for diagnostic assessment. A significant subset of cases showed distinctive eosinophilic cytoplasm. Several additional soft features, while not diagnostic, can aid case evaluation. Gland crowding and nuclear atypia are not essential criteria, although they can be present. Previously established criteria for aberrancy of each of the 3 immunohistochemical markers, where aberrancy for 1 or more support the diagnosis of EIN, remain applicable in the secretory phase and delineate areas of EIN in most cases, supporting their diagnostic utility in this often challenging diagnostic setting.