Histologic and Immunohistochemical Analyses of Endometrial Carcinomas. Experiences From Endometrial Biopsies in 358 Consultation Cases

Histologic and Immunohistochemical Analyses of Endometrial Carcinomas Experiences From Endometrial Biopsies in 358 Consultation Cases Jian-Jun Wei, MD; Ajit Paintal, MD; Pacita Keh, MD Context. Uterine serous carcinoma is biologically more aggressive than the endometrioid carcinoma. Because uterine serous carcinoma has a high propensity for lymphovascular invasion and intraperitoneal and extraabdominal spread, accurate diagnosis of this tumor type in endometrial biopsies/curettings is critical for appropriate clinical management. Objective. To share our experience in the evaluation of endometrial biopsy specimens in type I and type II endometrial adenocarcinoma. Design. We retrospectively reviewed 358 biopsies containing endometrial carcinoma during a recent 3 year period of our consultation records. In cases in which our interpretation differed from the submitting diagnosis, a panel of immunostains was performed. The performance characteristics of each antibody in our panel was calculated in this group of challenging cases. Results. Among the endometrial carcinomas we examined, a diagnosis of type I carcinoma accounted for 91% of cases (327 of 358) and type II carcinoma for 9% of cases (31 of 358); 41 cases (11.5%) were ambiguous or discordant (differing from submitted diagnoses and reviewed) based on histology alone. All 41 ambiguous and discordant cases were further evaluated with a battery of immunohistochemical markers. Of the 41 cases, 36 (88%) were ultimately classified (10 cases [24%] were endometrioid carcinoma; 18 cases [44%] were uterine serous carcinoma; 8 cases [20%] resulted in various other outcomes) and 5 cases (12%) remained indeterminate. Conclusions. Making the distinction between type I and II endometrial carcinoma remains a common problem in general practice. Although no one biomarker provides excellent statistical performance, a panel of immunohistochemical markers is often useful in difficult cases. (Arch Pathol Lab Med. 2013;137:1574 1583; doi: 10.5858/arpa.2012-0445-OA) Endometrial cancer is the most common malignancy of the gynecologic tract in the United States. In 2009, more than 42 000 new uterine cancers were diagnosed and 7780 uterine cancer associated deaths occurred in the United States. 1 Endometrial cancers are classified into 2 broad histologic types, defined as type I and type II. Type I consists of endometrioid carcinoma (EMC) and its histologic variants; type II endometrial cancer includes serous carcinoma, clear cell carcinoma, and carcinosarcoma. Uterine serous carcinoma (ESC), type II, accounts for approximately 10% of cases, whereas EMC, type I, encompasses about 80% to 85% of cases. Clear cell carcinoma and other rare forms of endometrial carcinoma comprise the few remaining cases. Type I and type II endometrial carcinomas are considered Accepted for publication December 11, 2012. From the Department of Pathology, Feinberg Medical School, Northwestern University, Chicago, Illinois. Deceased. The authors have no relevant financial interest in the products or companies described in this article. Presented in part at the 101st Annual Meeting of the United States and Canadian Academy of Pathology; March 17 23, 2012; Vancouver, British Columbia, Canada. Reprints: Jian-Jun Wei, MD, Department of Pathology, Feinberg Medical School 7 334, Northwestern University, 251 E Huron St, Chicago, IL 60611 (e-mail: jianjun-wei@northwestern.edu). biologically distinct disease entities, characterized by their divergent histologies, molecular genetics, tumorigenesis, and clinical behavior. Different patterns of molecular alterations are now thought to underlie the pathogenesis and/or progression of type I and type II uterine carcinomas. 2,3 For example, the most common genetic alterations in EMC include microsatellite instability, mutations of PTEN, b-catenin, PIK3CA, and K-ras. 4 In contrast, in the case of ESC, p53 mutations are common 3 and ERBB2 (formerly HER2) overexpression or amplification is also frequently seen. 2 Clinically, ESC has a higher propensity for lymphovascular invasion and intraperitoneal as well as extra-abdominal spread than EMC. 5 During presentation, approximately 40% to 70% of women with ESC will have disease beyond the uterus. 6 Furthermore, although tumor grade and depth of myometrial invasion are not predictive of extrauterine spread for ESC, the incidence of extrauterine spread in endometrioid cancers correlates with both tumor grade and depth of myometrial invasion. Based on data from the Surveillance, Epidemiology and End Results database (National Cancer Institute, Bethesda, Maryland), among all cases of endometrial carcinoma, ESC represents 50% of stage III or IV disease and nearly 40% of all deaths 7 although comprising only 10% of cases by incidence. Because of the high propensity for lymphovascular invasion, intraperitone- 1574 Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al

Table 1. Year EC, No. (%) al and extra abdominal spread, and recurrence and diseaseassociated death, treatment of ESC is more aggressive and often consists of a combination of surgery, chemotherapy, and radiation therapy. The current surgical protocol for women with ESC includes a total abdominal hysterectomy and bilateral salpingo-oophorectomy, pelvic/paraaortic dissection, omentectomy, assessment of the peritoneal cavity including pelvic and diaphragmatic cytology, and resection of gross metastases. Accurate diagnosis of tumor type in endometrial biopsies/ curetting specimens is critical for patient management. In general, most cases of ESC and EMC are readily differentiated from one another based on their distinct histologic features, growth patterns, and clinical presentation. 8 However, in practice, a proportion of endometrial carcinomas show morphologic features of both ESC and EMC. In other cases, the histologic type may be difficult to determine because of a scant biopsy specimen. For ambiguous cases, meticulous attention to morphology and a battery of immunohistochemical studies may be required. 8,9 Although review of submitted diagnostic material is standard protocol in our gyn-oncology practice, the actual value and importance of this practice is not well explored. In this study, we review our experience with endometrial carcinoma diagnosed in endometrial biopsies/curettings in 358 consultation cases. MATERIALS AND METHODS Case Selection Among 358 endometrial biopsies/curettings with endometrial carcinoma from our consultation files during a recent 3 year period (2009 2011), 41 cases (11%) were selected for review because of differing interpretations regarding type I and II endometrial carcinoma made by the referring pathologist and on second review at our institution. Discrepant cases had to meet the following criteria: (1) 2 pathologists or the consensus opinion at the departmental quality assurance conference felt that the histologic features in the biopsies/curettings did not correspond with the submitting diagnosis, and (2) at least 3 of 5 surrogate markers (43%) supported a change from the submitting diagnosis. Thirtytwo of the 41 cases (78%) were followed by total hysterectomy at Northwestern Memorial Hospital (Chicago, Illinois). These cases are summarized in Table 1. To assess the value of surrogate markers in the differential diagnosis of type I and type II endometrial carcinoma, we collected 56 control cases, which were morphologically diagnosed and resected in our institution. These included 30 cases (54%) of EMC and 26 cases (46%) of ESC. Briefly, features that favored endometrioid adenocarcinoma morphologically included a background of hyperplasia, squamous and/or mucinous metaplasia, sharp luminal borders, a cribriform growth pattern, and, usually, lower-grade nuclei. Features favoring a diagnosis of serous carcinoma included a background of atrophy or endometrial polyp, ragged luminal borders, micropapillary tufting, slitlike spaces within solid sheets of tumor cells, and the Summary of 358 Consultation Cases With Endometrial Carcinoma (EC) Disagreement, No. (%N), (%n) EMC/ESC, No. (%) Others, No. (%) Undetermined, No. (%) coexistence of high-grade nuclei with a papillary or glandular growth pattern (cyto-architectural dissociation). Immunohistochemistry Confirmed by TAH, No. (%) 2009 101 (28) 9 (3), (22) 8 (20) 1 (2) 0 (0) 7 (22) 2010 137 (38) 19 (5), (46) 12 (29) 4 (10) 3 (7) 14 (44) 2011 120 (34) 13 (4), (32) 8 (20) 3 (7) 2 (5) 11 (34) Total 358 (100) 41 (11), (100) 28 (68) 8 (20) 5 (12) 32 (100) Abbreviations: EMC, endometrioid carcinoma; ESC, endometrial serous carcinoma; TAH, total abdominal hysterectomy. For all 41 selected endometrial biopsies (100%), when there was sufficient material, immunostains for estrogen receptor (ER), progesterone receptor (PR), p16, p53, and vimentin were performed. In some cases, immunostains for b-catenin, WT1, ERBB-2 (formerly HER2/neu), HMGA2 (high-mobility group A2 gene), and Ki-67 were also performed when appropriate. Formalin-fixed and paraffin-embedded tissue from the selected blocks were sectioned at 4-lm thick. After deparaffinization and antigen retrieval, all immunohistochemical staining was performed on a Ventana Nexus Automated System (Tucson, Arizona). In brief, endogenous peroxidase activity was blocked with 3% hydrogen peroxide. Primary antibodies were detected using standard biotinylated anti-mouse or anti-rabbit secondary antibodies. A quantitative score system for immunointensity of 0 to 3 (0, negative; 1, weak; 2, moderate; 3, strong) was used for the markers HMGA2, 10 WT1, and b-catenin. Immunostains for vimentin, p16, 11 and p53 12 were scored as 0 to 2 (0, negative; 1, focal/patchy; 2, strong/diffuse [.70% tumor cells]). Scores for the percentage of immunopositivity was used for ER, PR, and Ki-67. Statistical Analysis The expression levels of the selected gene products for the entire patient population and controls were summarized by medians and ranges and differences among cases and controls and among histology and clinical groups using the Wilcoxon rank-sum test. The sensitivity and specificity at specific cutpoints for each marker were analyzed. We define sensitivity as the proportion of true positives that were detected using that cutpoint and specificity as the proportion of true negatives that were detected using that cutpoint. Area under the curve is the area under the receiver operating characteristic curve, which plots sensitivity and specificity through the full range of possible cutpoints. P,.05 was considered statistically significant. RESULTS Overview of the Cases From Endometrial Biopsies In reviewing endometrial biopsies from consultation cases, we encountered several morphologically ambiguous cases, which required further evaluation using immunohistochemistry. Correct interpretation of tumor type in endometrial cancers is one of the most common issues encountered in endometrial biopsies/curettings. As mentioned, the accurate diagnosis and distinction of type I and II endometrial carcinomas is critical for proper surgical planning and further clinical management. To evaluate common biases in the diagnosis of histologic subtypes, we reviewed all endometrial biopsies in our consultation records for a recent 3 year period. The 358 biopsies with endometrial carcinoma were received from 49 hospitals, medical centers, and private groups and were reviewed. Patients ages ranged from 19 to 86 years. Of the 358 cases, 41 (11%) were either morphologically ambiguous or discordant with the submitted diagnosis and were reviewed. Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al 1575

Table 2. Histologic and Immunohistochemical Features of Endometrial Serous Carcinoma (ESC) With Original Diagnosis of Endometrioid Carcinoma (EMC) Immunohistochemistry Histology Order P16 P53 ER% PR% VIM Glandular Papillary Solid Case 1 2 2 50 10 1 þ Case 2 2 0 70 0 þ Case 3 2 0 0 0 0 þ Case 4 2 2 100 50 0 þ þ Case 5 2 2 50 0 1 þ a þ Case 6 2 2 50 10 1 þ Case 7 2 2 60 20 0 þ Case 8 50 0 þ Case 9 0 60 0 1 þ a þ Case 10 2 0 80 20 0 þ Case 11 2 2 30 10 0 þ þ Case 12 2 0 60 15 0 þ Case 13 2 2 100 30 0 þ Case 14 2 2 100 0 1 þ a þ Case 15 80 10 1 þ Case 16 60 30 1 þ Case 17 2 0 50 20 0 þ þ Case 18 2 2 60 0 1 þ þ Positivity, % 100 60 94 69 44 Abbreviations: EMC, endometrioid carcinoma; ESC, endometrial serous carcinoma; ER, estrogen receptor; PR, progesterone receptor, 0, negative; 1, focal/patchy; 2, strong/diffuse; þ, present;, absence; VIM, vimentin. a EIC (endometrial intraepithelial carcinoma). As such, these cases were felt to require further investigation (Table 1). A battery of immunohistochemical stains was used to characterize this subset of cases. Of the 41 cases reviewed, 36 (88%) could ultimately be assigned a definite tumor type based on further evaluation of histology and immunohistochemistry (10 cases [24%] were EMC; 18 cases [44%] were ESC; 8 cases [20%] resulted in various other outcomes). A definite tumor type for the remaining 5 cases (12%) could not be determined. Among these 41 cases, 32 cases (78%) were further reviewed in hysterectomy specimens to confirm the diagnoses. Endometrioid Versus Serous Carcinomas Among the 41 discordant and morphologically ambiguous cases, 18 (44%) were originally submitted with a diagnosis of EMC (Table 2). Those cases all had a predominantly glandular/tubular architecture with less-prominent cytologic atypia (on low-powered view). Five of the 18 cases (28%) showed areas of papillary growth, and only 2 cases (12%) had a focal solid-growth pattern (Table 1). Although almost all cases showed a pseudoglandular architecture, mimicking low-grade EMC, close review revealed significantly discordant architecture and cytology characterized by somewhat ragged luminal borders, loss of cell polarity, prominent nucleoli, and nuclear pleomorphism (bizarre enough to prompt a consideration of ESC) (Figure 1, A and B). Some cases contained foci of endometrial intraepithelial carcinoma. A background of scant and atrophic endometrial epithelium could be appreciated in 5 of the 18 cases (28%). Two cases (11%) were submitted with negative immunostain results for p53 from the original institution. We performed a panel of immunostains (ER, PR, p16, p53, and vimentin) on these cases. Immunostains revealed that all tumors were diffusely positive for p16 (.90% of tumor cells), 60% (9 of 15) of the cases were strongly immunopositive for p53 (.70% of tumor cells), and 40% of cases (6 of 15) showed total absence of immunoreactivity for p53 (Figure 1, D). Most cases had a high percentage of ER and a low percentage of PR expression. Complete absence of vimentin was found in 56% (10 of 18) of cases, and the remainder of cases (8 of 18; 44%) showed patchy immunoreactivity (Figure 1, F). Most often, our morphologic diagnosis was supported by the pattern of immunoreactivity (Table 2). The major factors that may have led to the original erroneous interpretation were (1) predominant glandular architecture, (2) less-prominent cytologic atypia, (3) misinterpretation of a negative p53 immunostain result (suggesting a nonsense mutation), (4) scant material, and (5) lack of ancillary testing in morphologically ambiguous cases. Serous Versus Endometrioid Carcinomas Ten cases (10 of 41; 24%) were originally diagnosed as ESC (Table 3). Of these 10 cases, 4 (40%) had a predominantly papillary architecture (Figure 2, A and B), 1 case (10%) of villoglandular, and 1 case (10%) demonstrated a solid growth pattern (Table 3). Two cases (20%) contained focal high-grade nuclei with abundant cytoplasm (low nuclear to cytoplasmic ratio). Three cases (30%) were submitted with immunohistochemical stains for p53, which demonstrated scattered (,10% 20% cells) weak immunoreactivity (Figure 2, D). We performed a panel of 5 immunomarkers for all cases. As summarized in Table 3, 7 of 9 cases (78%) had patchy immunoreactivity (,50%) for p16 (Figure 2C). Seven cases (70%) were negative for p53; 3 cases (30%) were focally positive for p53 (,20% of tumor cells). Almost all cases (9 of 10; 90%) showed diffuse immunoreactivity for vimentin and ER (Figure 2, E and F), except for case 1 (10%), and had varied PR expression (Table 3). Potential causes of misinterpretation of cases in this group of cases were (1) a predominantly papillary architecture, (2) focal high-grade cytologic atypia, and (3) focal immunoreactivity for p53. Other Variants of EMCs Eight of the 41 cases (20%) involved either a submitting or revised diagnosis of clear cell carcinoma (Figure 3, A 1576 Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al

Figure 1. This endometrial biopsy was from a 72-year-old woman. A, Low-power view shows adenocarcinoma with glandular and microglandular growth patterns in the background of strips of inactive endometrial epithelium. B, Close view presents cribriform and micropapillary growth patterns with high nuclear to cytoplasmic ratio, moderate nuclear atypia, and eosinophilic cytoplasm. Frequent mitosis and apoptosis can be seen. Immunostain results revealed that the tumor cell results were diffusely positive for p16 (C); completely negative for p53 (D), ER (E), and PR (not shown); and focally positive for vimentin (F). A diagnosis of serous carcinoma was rendered (A and B, hematoxylin-eosin, original magnifications 310 [A] and 340 [B through F]). through F), secretory EMC, endocervical adenocarcinoma, or mixed-type adenocarcinoma (Figure 4, A through F). Three of the 8 cases (38%) were read as clear cell carcinoma originally and revised to either EMC or secretory EMC. Three of the 8 cases (38%) were read as EMC but were histologically and immunohistochemically consistent with endocervical adenocarcinoma of the usual type (Table 4). Two endometrial carcinomas were revised to mixed epithelial carcinomas (Figure 4, A through F). Five of the 41 cases (12%) could not be definitively classified, even after thorough morphologic and immunohistochemical analysis. Two of the 5 cases (40%) showed Table 3. Histologic and Immunohistochemical Features of Endometrioid Carcinoma (EMC) With Original Diagnosis of Endometrial Serous Carcinoma (ESC) Immunohistochemistry Histology Order P16 P53 ER% PR% VIM Glandular Papillary Solid Case 1 1 0 100 0 2 þ þ Case 2 2 0 50 30 2 þ Case 3 1 1 100 80 2 þ Case 4 1 0 100 10 2 Villoglandular Case 5 1 0 80 30 2 þ Case 6 1 100 80 2 þ þ Case 7 1 0 100 70 1 þ a Case 8 2 1 100 70 2 þ a Case 9 1 0 90 60 2 þ Case 10 1 0 90 70 2 þ Abbreviations: EMC, endometrioid carcinoma; ESC, endometrial serous carcinoma; ER, estrogen receptor; PR, progesterone receptor, 0, negative; 1, focal/patchy; 2, strong/diffuse; þ, present;, absence; VIM, vimentin. a focal nuclear grade 3. Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al 1577

Figure 2. This endometrial biopsy was from a 73-year-old woman. A, Low-power view shows an adenocarcinoma with a mostly papillary growth pattern. B, Close view presented hierarchical papillary fronds with mild and occasional, moderate cytologic atypia and eosinophilic cytoplasm. Immunostain results revealed that the tumor cells were patchy positive for p16 (C), scattered and weakly positive for p53 (D), strongly and diffusely positive for ER (E) and PR (not shown), and positive for vimentin (F). A diagnosis of endometrioid carcinoma was rendered (original magnifications 310 [A, hematoxylin-eosin]), 340 [B, hematoxylin-eosin], and 320 [C through F]). features of ESC and 3 (60%) resembled EMC histologically, but none fit a typical immunoprofile using the 5 selected markers. Three cases (60%) showed deep myometrial invasion (.50% depth of invasion) and 1 (20%) had extrauterine metastasis. To highlight this small fraction of endometrial adenocarcinomas with low-grade histology and aggressive behavior, we summarize 2 cases below. Case 1. A 75-year-old women presented with postmenopausal bleeding, and a biopsy was performed at the submitting institution and read as EMC. The biopsy showed a predominantly glandular and tubular growth pattern with focal areas of papillary growth. High magnification revealed a typical endometrioid-like morphology characterized by areas of cribriform growth, punched-out lumens, sharp luminal borders, abundant pale and eosinophilic cytoplasm, and mostly low-grade nuclei (Figure 5, A and B). On hysterectomy, the tumor was similar in appearance to the biopsy specimen but deeply infiltrated through almost the full thickness of the myometrium. Foci of less-defined architecture, high nuclear to cytoplasmic ratio, and highgrade nuclei could be appreciated. Immunostains revealed that the tumor cells were diffusely immunoreactive for p16 (Figure 5, C and E) and p53 (Figure 5, D and F) with 90% ER and 30% PR expression (data not shown). Focal positivity for vimentin was seen (data not shown). Case 2. A 55-year-old woman presented with a submitted biopsy diagnosis of EMC with possible serous differentiation. The tumor had a mostly glandular/tubular growth pattern with ragged luminal borders, loss of cell polarity, abundant cytoplasm, and areas with a hobnail appearance. Hysterectomy revealed low-grade architecture, moderate cytologic atypia, and deep myometrial invasion (Figure 6, A and B). Immunostain results showed that the tumor cells were focally positive for p16 (,10%, Figure 6, C) and p53 (,10%; Figure 6, D) and were negative for ER (Figure 6, E), PR (data not shown), and vimentin (Figure 6, F). Although serous carcinoma was favored, lack of diffuse positivity for p16 and wild-type p53 expression were felt to be confounding factors precluding a definitive diagnosis. A Useful Panel of Markers in Ambiguous Cases As discussed above, some endometrial carcinomas may not be definitively subclassified based on histologic evaluation alone and may benefit from immunohistochemistry as a second step in their evaluation. To further investigate the value and the reliability of the selected markers in the differential diagnosis of ESC and EMC, we first examined the expression patterns of the selected markers in 56 wellcharacterized endometrial carcinomas (30 EMC and 26 ESC). Based on the scoring methodology summarized in the 1578 Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al

Figure 3. This endometrial biopsy was from a 55-year-old woman. A, Low-power view shows adenocarcinoma with a mostly papillary/glandular growth pattern with irregular luminal borders. B, Close view revealed that tumor cells had mild to moderate cytologic atypia, eosinophilic cytoplasm, and occasional hobnail cells. Immunostain results revealed that the tumor cells were patchy positive for p16 (C), strongly positive for p53 (D), negative for ER (E) and PR (not shown), and focally positive for vimentin (F). A diagnosis of clear cell carcinoma was rendered (original magnifications 310 [A, hematoxylin-eosin], 320 [B, hematoxylin-eosin], and 340 [C through F]). Materials and Methods, we calculated the sensitivity and specificity of each of 5 markers in these cases. As shown in Table 4, none of the individual markers demonstrates both high sensitivity and specificity. For example, diffuse immunoreactivity for p16 provides 65% sensitivity and 82% specificity for the diagnosis of ESC; diffuse immunoreactivity for p53 provides 55% sensitivity and 95% specificity of diagnosis of ESC. Therefore, we tended to render a diagnosis of endometrioid or serous type differentiation based on support from at least 3 of 5 (60%) of selected markers. COMMENT A significant subset of endometrial carcinomas (in our series accounting for 11% of endometrial carcinoma cases) are difficult to classify based on histology alone (Table 1). A recent review article by Bartosch et al 8 provided an excellent, comprehensive review of the updated morphologic and immunohistochemical features of the various endometrial carcinoma subtypes. Interpretation of tumor type in endometrial carcinoma can be difficult especially when biopsy material is scant or there is abundant necrosis or poorly preserved architectural and cytologic detail. In addition, some cases are morphologically ambiguous, even in a generous, well-preserved biopsy specimen. In such circumstances, immunostains are useful. In our experience, the value of any single immunohistochemical marker is limited because of relatively low sensitivities and moderate specificities when used in isolation. In particular, the distinction between type I and II endometrial carcinoma cannot reliably be based on examination of p53 or p16 alone. Our findings are consistent with 2 recent studies, 11,13 which showed that diffuse immunoreactivity for p16 was seen in 70% to 90% of ESC and 36% to 38% of EMC. Although there are many immunomarkers that have been reported to have utility in the differential diagnosis of EMC and ESC, our experience is that nearly 90% (36 of 41) of the difficult cases can be accurately classified based on 5 immunomarkers (p16, p53, ER, PR and vimentin). When used as a panel, the specificity for the diagnosis of serous carcinoma can be significantly improved. Vimentin is a sensitive marker for endometrioid adenocarcinoma in the context of an endometrial tumor. Many studies, mostly dealing with the differential diagnostic scenario of endometrioid versus endocervical adenocarcinoma, have demonstrated that vimentin immunoreactivity is present in greater than 90% of endometrioid adenocarcinomas and maintains a sensitivity of 80% even in poorly differentiated (International Federation of Gynecology and Obstetrics [FIGO] grade 3) tumors. 14,15 Precluding its use as a single marker, vimentin is nonspecific among endometrial Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al 1579

Figure 4. This endometrial biopsy was from a 68-year-old woman. A and B, Tumor shows adenocarcinoma with glandular (lower left) and papillary (upper right) growth patterns. Glandular component had well-defined luminal border and low-grade nuclei. Papillary component showed irregular luminal border, high nuclear to cytoplasmic ratio, and marked nuclear pleomorphism. Immunostain results for p16 (C), estrogen receptor (E), and vimentin (F) revealed 2 distinct but characteristic endometrioid and serous differentiations; p53 (D) was diffusely positive for both components (original magnifications 320 [A through F, hematoxylin-eosin, A and B]). tumors and has been found to label up to 81.3% of ESCs as well as almost all clear cell carcinomas. 16 Hormone receptors ER and PR are frequently used to distinguish endometrioid and serous carcinomas and generally show strong diffuse expression in approximately 80% of FIGO 1 and 2 endometrioid tumors. The reported degree of immunoreactivity in FIGO 3 endometrioid tumor is markedly less, ranging from 15% to 50%. These frequencies overlap with those seen in ESC (5% 54%). In this setting, PR expression is thought to be somewhat more specific than ER for endometrioid adenocarcinoma. With a high-grade tumor, the pattern of immunoreactivity for ER and PR is important because, even in high-grade endometrioid tumors, the number of positive cells and the degree of Table 4. Sensitivity and Specificity for the Diagnosis of Endometrial Serous Carcinoma (ESC) by Each of the Selected Immunomarkers True False Markers Stain Patterns Positive Negative Positive Negative Sensitivity, % Specificity, % p16 Diffuse 9 14 3 5 64.29 82.35 p53 Diffuse 10 17 1 8 55.56 94.44 ER,10% 1 16 3 17 5.56 84.21 ER 50% 7 16 3 11 38.89 84.21 ER 80% 15 11 8 3 83.33 57.90 PR,10% 5 13 3 11 31.25 81.25 PR 30% 15 11 5 1 93.75 68.75 PR 70% 16 3 13 0 100.00 18.75 VIM Negative 10 18 2 8 55.56 90.00 VIM Focal 18 10 10 0 100.00 50.00 Abbreviations: ER, estrogen receptor; PR, progesterone receptor; VIM, vimentin. 1580 Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al

Figure 5. This endometrial biopsy was from a 75-year-old woman. A and B, The tumor showed endometrioid morphology, characterized by a cribriform glandular growth pattern, flat luminal borders, and orderly arranged low grade nuclei. Immunostain results revealed the tumor cells were diffusely positive for p16 (C and E), p53 (D and F), and varied for estrogen receptor and progesterone receptor (30% 90%, not shown). and focally positive for vimentin (not shown). This tumor was clinically aggressive with nearly full thickness of myometrial invasion (original magnifications 340 [A and C through F, hematoxylin-eosin, A and B] and 320 [B]). staining intensity are greater than what is generally seen in ESC (focal, weak expression). 15 Aberrant p53 expression is one of the hallmarks of type 2 endometrial cancer. The pattern of expression is important when evaluating immunoreactivity for this antibody as normal tissue and tumors with intact p53 expression display a variable degree of positivity. Using a cutoff of greater than 70% expression and 0% expression (representing a nonsense mutation of p53), an abnormal pattern of p53 expression will be seen in approximately 80% to 90% cases of serous carcinoma. 12,16 18 An important caveat to overreliance on this single antibody is that an abnormal pattern of p53 expression can be seen in a few endometrioid adenocarcinomas (~10%), predominantly in FIGO 3 cases. 19,20 Of note, in cases with ambiguous morphology, where a definitive histologic subtype cannot be established, aberrant p53 expression has been correlated with adverse clinical outcomes. 19 p16 plays a role in cell-cycle regulation and is diffusely positive in a number of different tumor types from many anatomic locations. Characteristically, diffuse strong p16 expression (.90% strong immunoreactivity) is seen in many cases (60% 90%) of ESC and is sensitive in this context. 11,13,15 Patchy weak expression is nonspecific and may be seen in other tumor types as well as in normal tissue. Endometrioid adenocarcinoma can uncommonly show a similar strong diffuse pattern of expression. Notably, FIGO 3 endometrioid adenocarcinoma may show strong diffuse p16 positivity in up to 25% of cases. 13 Endocervical adenocarcinomas typically show a similar pattern of immunoreactivity. None of the immunohistochemical markers commonly used in the distinction of endometrioid and serous carcinoma are specific enough to be individually relied on to make that distinction. In particular, a substantial degree of immunohistochemical overlap exists between FIGO 3 endometrioid adenocarcinoma and ESC. As such, in difficult cases, a panel of immunohistochemical markers and close attention to morphologic detail should be used in conjunction to arrive at a diagnosis. In our review, we encountered a few virulent cases of endometrial carcinoma that had either a low-grade endometrioid-like appearance or mixed features. These ambiguous cases tended to have aggressive behavior and inconclusive patterns of immunoreactivity (Figures 5 and 6). These cases were not associated with loss of MMR genes or b-catenin mutations (data not shown). Further molecular investigation may help to disclose the underlying biology of these tumors. That will require identifying new surrogate markers in cases with atypical histology and clinical presentation. Our knowledge of accurate subclassification of endometrial tumors in small biopsies is undergoing rapid refine- Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al 1581

Figure 6. This endometrial biopsy was from a 55-year-old woman. A and B, The tumor had a mostly glandular/tubular growth pattern, ragged luminal borders, loss of cell polarity, abundant cytoplasm, and areas of hobnailing. Hysterectomy revealed low-grade architecture, moderate cytologic atypia and deep myometrial invasion. Immunostain results showed tumor cells were focally positive for p16 (C), scattered positive for p53 (D), and negative for ER (E), PR (data not shown), and vimentin (F) (A and B, hematoxylin-eosin, original magnifications 320 [A], 340 [B], and 320 [C through F]). ment. In a previous study, 10 we found that HMGA2 can be a potential surrogate marker to differentiate ovarian serous and EMCs. McCluggage et al 20 examined HMGA2 expression in a large cohort of endometrial carcinomas and found that 91% of ESC (30 of 33) and 37% (14/38) of EMC showed immunoreactivity for HMGA2. In our study, we found that 68% of ESC (30 of 44) had immunoreactivity for HMGA2. The lower rate of HMGA2 in our cases can be attributed to our counting HMGA2 as positive in those cases with moderate immunoreactivity. Nevertheless, HMGA2 may not be a good marker for the distinction of EMC and ESC because of overlapping patterns of immunoreactivity. In summary, because of the relatively low individual sensitivities and specificities for each of the listed markers used in the differential diagnosis of type I and type II endometrial adenocarcinoma, at present a battery of stains provides the best tool for accurate classification in morphologically ambiguous cases. Tumor classification in endometrial adenocarcinoma based on histology and immunohistochemistry has some limitations, particularly in lesions with mixed epithelial differentiation or vague morphology. Moving forward, development of molecular biomarkers may be the next step in refining our classification of these tumors in general and on a case-by-case basis. This article is dedicated to Pacita Keh, MD, a good teacher and outstanding gynecologic pathologist. We thank Irene Helenowski, PhD, and Denise Scholtens, PhD, for the statistical analysis. We are grateful to the Gynecologic Oncologists of Northwestern Prentice Women s Hospital who referred cases to us for review and to the other pathologists of Northwestern Memorial Hospital who provided their opinions in the diagnosis of many of these cases. References 1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, 2009. CA Cancer J Clin. 2009;59(4):225 249. 2. Lax SF. Molecular genetic pathways in various types of endometrial carcinoma: from a phenotypical to a molecular-based classification. Virchows Arch. 2004;444(3):213 223. 3. Zheng W, Xiang L, Fadare O, Kong B. A proposed model for endometrial serous carcinogenesis. Am J Surg Pathol. 2011;35(1):e1 e14. doi:10.1097/pas. 0b013e318202772e. 4. Prat J, Gallardo A, Cuatrecasas M, Catasús L. Endometrial carcinoma: pathology and genetics. Pathology. 2007;39(1):72 87. 5. Hendrickson M, Ross J, Eifel P, Martinez A, Kempson R. Uterine papillary serous carcinoma: a highly malignant form of endometrial adenocarcinoma. Am J Surg Pathol. 1982;6(2):93 108. 6. Slomovitz BM, Burke TW, Eifel PJ, et al. Uterine papillary serous carcinoma (UPSC): a single institution review of 129 cases. Gynecol Oncol. 2003;91(3):463 469. 7. Ueda SM, Kapp DS, Cheung MK, et al. Trends in demographic and clinical characteristics in women diagnosed with corpus cancer and their potential impact on the increasing number of deaths. Am J Obstet Gynecol. 2008;198(2): 218.e1 218.e6. doi:10.1016/j.ajog.2007.08.075. 8. Bartosch C. Manuel Lopes J, Oliva E. Endometrial carcinomas: a review emphasizing overlapping and distinctive morphological and immunohistochemical features. Adv Anat Pathol. 2011;18(6):415 437. 1582 Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al

9. McCluggage WG. Immunohistochemical and functional biomarkers of value in female genital tract lesions. Int J Gynecol Pathol. 2006;25(2):101 120. 10. Mahajan A, Liu Z, Gellert L, et al. HMGA2: a biomarker significantly overexpressed in high-grade ovarian serous carcinoma. Mod Pathol. 2010;23(5): 673 681. 11. Netzer IM, Kerner H, Litwin L, Lowenstein L, Amit A. Diagnostic implications of p16 expression in serous papillary endometrial cancer. Int J Gynecol Cancer. 2011;21(8):1441 1445. 12. Yemelyanova A, Vang R, Kshirsagar M, et al. Immunohistochemical staining patterns of p53 can serve as a surrogate marker for TP53 mutations in ovarian carcinoma: an immunohistochemical and nucleotide sequencing analysis. Mod Pathol. 2011;24(9):1248 1253. 13. Yemelyanova A, Ji H, Shih Ie M, Wang TL, Wu LS, Ronnett BM. Utility of p16 expression for distinction of uterine serous carcinomas from endometrial endometrioid and endocervical adenocarcinomas: immunohistochemical analysis of 201 cases. Am J Surg Pathol. 2009;33(10):1504 1514. 14. Alkushi A, Irving J, Hsu F, et al. Immunoprofile of cervical and endometrial adenocarcinomas using a tissue microarray. Virchows Arch. 2003;442(3):271 277. 15. Reid-Nicholson M, Iyengar P, Hummer AJ, Linkov I, Asher M, Soslow RA. Immunophenotypic diversity of endometrial adenocarcinomas: implications for differential diagnosis. Mod Pathol. 2006;19(8):1091 1100. 16. Darvishian F, Hummer AJ, Thaler HT, et al. Serous endometrial cancers that mimic endometrioid adenocarcinomas: a clinicopathologic and immunohistochemical study of a group of problematic cases. Am J Surg Pathol. 2004;28(12): 1568 1578. 17. Gilks CB, Clarke BA, Han G, et al. Letter to the editor regarding Roh MH, Lassin Y, Miron A et al. High-grade fimbrial-ovarian carcinomas are unified by p53, PTEN and PAX2 expression. Mod Pathol. 2011;24(9):1281 1282; author reply 1282 1283. 18. Egan JA, Ionescu MC, Eapen E, Jones JG, Marshall DS. Differential expression of WT1 and p53 in serous and endometrioid carcinomas of the endometrium. Int J Gynecol Pathol. 2004;23(2):119 122. 19. Garg K, Leitao MM, Jr., Wynveen CA, et al. p53 overexpression in morphologically ambiguous endometrial carcinomas correlates with adverse clinical outcomes. Mod Pathol. 2010;23(1):80 92. 20. McCluggage WG, Connolly LE, McBride HA, Kalloger S, Gilks CB. HMGA2 is commonly expressed in uterine serous carcinomas and is a useful adjunct to diagnosis. Histopathology. Jan 17, 2012. Arch Pathol Lab Med Vol 137, November 2013 Analysis of Endometrial Carcinoma in 358 Biopsies Wei et al 1583